[ { "path": "machine-learning-ex1/ex1/computeCost.m", "content": "function J = computeCost(X, y, theta)\n%COMPUTECOST Compute cost for linear regression\n% J = COMPUTECOST(X, y, theta) computes the cost of using theta as the\n% parameter for linear regression to fit the data points in X and y\n\n% Initialize some useful values\nm = length(y); % number of training examples\n\n% You need to return the following variables correctly \nJ = 0;\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Compute the cost of a particular choice of theta\n% You should set J to the cost.\n\ntemp1=X*theta-y;\ntemp2=sum(temp1.^2);\nJ=1/(2*m)*temp2;\n\n%sprintf('%f',J);\n\n% =========================================================================\n\nend\n" }, { "path": "machine-learning-ex1/ex1/computeCostMulti.m", "content": "function J = computeCostMulti(X, y, theta)\n%COMPUTECOSTMULTI Compute cost for linear regression with multiple variables\n% J = COMPUTECOSTMULTI(X, y, theta) computes the cost of using theta as the\n% parameter for linear regression to fit the data points in X and y\n\n% Initialize some useful values\nm = length(y); % number of training examples\n\n% You need to return the following variables correctly \nJ = 0;\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Compute the cost of a particular choice of theta\n% You should set J to the cost.\ntemp=(X*theta-y)'*(X*theta-y);\nJ=1/(2*m)*temp;\n\n\n% =========================================================================\n\nend\n" }, { "path": "machine-learning-ex1/ex1/ex1.m", "content": "%% Machine Learning Online Class - Exercise 1: Linear Regression\n\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% linear exercise. You will need to complete the following functions \n% in this exericse:\n%\n% warmUpExercise.m\n% plotData.m\n% gradientDescent.m\n% computeCost.m\n% gradientDescentMulti.m\n% computeCostMulti.m\n% featureNormalize.m\n% normalEqn.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n% x refers to the population size in 10,000s\n% y refers to the profit in $10,000s\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% ==================== Part 1: Basic Function ====================\n% Complete warmUpExercise.m \nfprintf('Running warmUpExercise ... \\n');\nfprintf('5x5 Identity Matrix: \\n');\nwarmUpExercise()\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ======================= Part 2: Plotting =======================\nfprintf('Plotting Data ...\\n')\ndata = load('ex1data1.txt');\nX = data(:, 1); y = data(:, 2);\nm = length(y); % number of training examples\n\n% Plot Data\n% Note: You have to complete the code in plotData.m\nplotData(X, y);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =================== Part 3: Gradient descent ===================\nfprintf('Running Gradient Descent ...\\n')\n\nX = [ones(m, 1), data(:,1)]; % Add a column of ones to x\ntheta = zeros(2, 1); % initialize fitting parameters\n\n% Some gradient descent settings\niterations = 1500;\nalpha = 0.01;\n\n% compute and display initial cost\ncomputeCost(X, y, theta)\n\n% run gradient descent\ntheta = gradientDescent(X, y, theta, alpha, iterations);\n\n% print theta to screen\nfprintf('Theta found by gradient descent: ');\nfprintf('%f %f \\n', theta(1), theta(2));\n\n% Plot the linear fit\nhold on; % keep previous plot visible\nplot(X(:,2), X*theta, '-')\nlegend('Training data', 'Linear regression')\nhold off % don't overlay any more plots on this figure\n\n% Predict values for population sizes of 35,000 and 70,000\npredict1 = [1, 3.5] *theta;\nfprintf('For population = 35,000, we predict a profit of %f\\n',...\n predict1*10000);\npredict2 = [1, 7] * theta;\nfprintf('For population = 70,000, we predict a profit of %f\\n',...\n predict2*10000);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ============= Part 4: Visualizing J(theta_0, theta_1) =============\nfprintf('Visualizing J(theta_0, theta_1) ...\\n')\n\n% Grid over which we will calculate J\ntheta0_vals = linspace(-10, 10, 100);\ntheta1_vals = linspace(-1, 4, 100);\n\n% initialize J_vals to a matrix of 0's\nJ_vals = zeros(length(theta0_vals), length(theta1_vals));\n\n% Fill out J_vals\nfor i = 1:length(theta0_vals)\n for j = 1:length(theta1_vals)\n\t t = [theta0_vals(i); theta1_vals(j)]; \n\t J_vals(i,j) = computeCost(X, y, t);\n end\nend\n\n\n% Because of the way meshgrids work in the surf command, we need to \n% transpose J_vals before calling surf, or else the axes will be flipped\nJ_vals = J_vals';\n% Surface plot\nfigure;\nsurf(theta0_vals, theta1_vals, J_vals)\nxlabel('\\theta_0'); ylabel('\\theta_1');\n\n% Contour plot\nfigure;\n% Plot J_vals as 15 contours spaced logarithmically between 0.01 and 100\ncontour(theta0_vals, theta1_vals, J_vals, logspace(-2, 3, 20))\nxlabel('\\theta_0'); ylabel('\\theta_1');\nhold on;\nplot(theta(1), theta(2), 'rx', 'MarkerSize', 10, 'LineWidth', 2);\n" }, { "path": "machine-learning-ex1/ex1/ex1_multi.m", "content": "%% Machine Learning Online Class\n% Exercise 1: Linear regression with multiple variables\n%\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% linear regression exercise. \n%\n% You will need to complete the following functions in this \n% exericse:\n%\n% warmUpExercise.m\n% plotData.m\n% gradientDescent.m\n% computeCost.m\n% gradientDescentMulti.m\n% computeCostMulti.m\n% featureNormalize.m\n% normalEqn.m\n%\n% For this part of the exercise, you will need to change some\n% parts of the code below for various experiments (e.g., changing\n% learning rates).\n%\n\n%% Initialization\n\n%% ================ Part 1: Feature Normalization ================\n\n%% Clear and Close Figures\nclear ; close all; clc\n\nfprintf('Loading data ...\\n');\n\n%% Load Data\ndata = load('ex1data2.txt');\nX = data(:, 1:2);\ny = data(:, 3);\nm = length(y);\n\n% Print out some data points\nfprintf('First 10 examples from the dataset: \\n');\nfprintf(' x = [%.0f %.0f], y = %.0f \\n', [X(1:10,:) y(1:10,:)]');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n% Scale features and set them to zero mean\nfprintf('Normalizing Features ...\\n');\n\n[X mu sigma] = featureNormalize(X);%ע⣬XӦX_norm\n\n% Add intercept term to X\nX = [ones(m, 1) X];\n\n\n%% ================ Part 2: Gradient Descent ================\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: We have provided you with the following starter\n% code that runs gradient descent with a particular\n% learning rate (alpha). \n%\n% Your task is to first make sure that your functions - \n% computeCost and gradientDescent already work with \n% this starter code and support multiple variables.\n%\n% After that, try running gradient descent with \n% different values of alpha and see which one gives\n% you the best result.\n%\n% Finally, you should complete the code at the end\n% to predict the price of a 1650 sq-ft, 3 br house.\n%\n% Hint: By using the 'hold on' command, you can plot multiple\n% graphs on the same figure.\n%\n% Hint: At prediction, make sure you do the same feature normalization.\n%\n\nfprintf('Running gradient descent ...\\n');\n\n% Choose some alpha value\nalpha = 0.01;\nnum_iters = 400;\n\n% Init Theta and Run Gradient Descent \ntheta = zeros(3, 1);\n[theta, J_history] = gradientDescentMulti(X, y, theta, alpha, num_iters);\n\n% Plot the convergence graph\nfigure;\nplot(1:numel(J_history), J_history, '-b', 'LineWidth', 2);\nxlabel('Number of iterations');\nylabel('Cost J');\n\n% Display gradient descent's result\nfprintf('Theta computed from gradient descent: \\n');\nfprintf(' %f \\n', theta);\nfprintf('\\n');\n\n% Estimate the price of a 1650 sq-ft, 3 br house\n% ====================== YOUR CODE HERE ======================\n% Recall that the first column of X is all-ones. Thus, it does\n% not need to be normalized.\nprice = 0; % You should change this\n\npredict1=[1650 3];\nnorm_predict=(predict1-mu)./sigma;\nfinal_predict=[1 norm_predict];\n\nprice=final_predict*theta;\n% ============================================================\n\nfprintf(['Predicted price of a 1650 sq-ft, 3 br house ' ...\n '(using gradient descent):\\n $%f\\n'], price);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ================ Part 3: Normal Equations ================\n\nfprintf('Solving with normal equations...\\n');\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: The following code computes the closed form \n% solution for linear regression using the normal\n% equations. You should complete the code in \n% normalEqn.m\n%\n% After doing so, you should complete this code \n% to predict the price of a 1650 sq-ft, 3 br house.\n%\n\n%% Load Data\ndata = csvread('ex1data2.txt');\nX = data(:, 1:2);\ny = data(:, 3);\nm = length(y);\n\n% Add intercept term to X\nX = [ones(m, 1) X];\n\n% Calculate the parameters from the normal equation\ntheta = normalEqn(X, y);\n\n% Display normal equation's result\nfprintf('Theta computed from the normal equations: \\n');\nfprintf(' %f \\n', theta);\nfprintf('\\n');\n\n\n% Estimate the price of a 1650 sq-ft, 3 br house\n% ====================== YOUR CODE HERE ======================\nprice = 0; % You should change this\n\npredict2=[1 1650 3];\nprice=predict2*theta;\n\n% ============================================================\n\nfprintf(['Predicted price of a 1650 sq-ft, 3 br house ' ...\n '(using normal equations):\\n $%f\\n'], price);\n\n" }, { "path": "machine-learning-ex1/ex1/ex1data1.txt", "content": "6.1101,17.592\n5.5277,9.1302\n8.5186,13.662\n7.0032,11.854\n5.8598,6.8233\n8.3829,11.886\n7.4764,4.3483\n8.5781,12\n6.4862,6.5987\n5.0546,3.8166\n5.7107,3.2522\n14.164,15.505\n5.734,3.1551\n8.4084,7.2258\n5.6407,0.71618\n5.3794,3.5129\n6.3654,5.3048\n5.1301,0.56077\n6.4296,3.6518\n7.0708,5.3893\n6.1891,3.1386\n20.27,21.767\n5.4901,4.263\n6.3261,5.1875\n5.5649,3.0825\n18.945,22.638\n12.828,13.501\n10.957,7.0467\n13.176,14.692\n22.203,24.147\n5.2524,-1.22\n6.5894,5.9966\n9.2482,12.134\n5.8918,1.8495\n8.2111,6.5426\n7.9334,4.5623\n8.0959,4.1164\n5.6063,3.3928\n12.836,10.117\n6.3534,5.4974\n5.4069,0.55657\n6.8825,3.9115\n11.708,5.3854\n5.7737,2.4406\n7.8247,6.7318\n7.0931,1.0463\n5.0702,5.1337\n5.8014,1.844\n11.7,8.0043\n5.5416,1.0179\n7.5402,6.7504\n5.3077,1.8396\n7.4239,4.2885\n7.6031,4.9981\n6.3328,1.4233\n6.3589,-1.4211\n6.2742,2.4756\n5.6397,4.6042\n9.3102,3.9624\n9.4536,5.4141\n8.8254,5.1694\n5.1793,-0.74279\n21.279,17.929\n14.908,12.054\n18.959,17.054\n7.2182,4.8852\n8.2951,5.7442\n10.236,7.7754\n5.4994,1.0173\n20.341,20.992\n10.136,6.6799\n7.3345,4.0259\n6.0062,1.2784\n7.2259,3.3411\n5.0269,-2.6807\n6.5479,0.29678\n7.5386,3.8845\n5.0365,5.7014\n10.274,6.7526\n5.1077,2.0576\n5.7292,0.47953\n5.1884,0.20421\n6.3557,0.67861\n9.7687,7.5435\n6.5159,5.3436\n8.5172,4.2415\n9.1802,6.7981\n6.002,0.92695\n5.5204,0.152\n5.0594,2.8214\n5.7077,1.8451\n7.6366,4.2959\n5.8707,7.2029\n5.3054,1.9869\n8.2934,0.14454\n13.394,9.0551\n5.4369,0.61705\n" }, { "path": "machine-learning-ex1/ex1/ex1data2.txt", "content": "2104,3,399900\n1600,3,329900\n2400,3,369000\n1416,2,232000\n3000,4,539900\n1985,4,299900\n1534,3,314900\n1427,3,198999\n1380,3,212000\n1494,3,242500\n1940,4,239999\n2000,3,347000\n1890,3,329999\n4478,5,699900\n1268,3,259900\n2300,4,449900\n1320,2,299900\n1236,3,199900\n2609,4,499998\n3031,4,599000\n1767,3,252900\n1888,2,255000\n1604,3,242900\n1962,4,259900\n3890,3,573900\n1100,3,249900\n1458,3,464500\n2526,3,469000\n2200,3,475000\n2637,3,299900\n1839,2,349900\n1000,1,169900\n2040,4,314900\n3137,3,579900\n1811,4,285900\n1437,3,249900\n1239,3,229900\n2132,4,345000\n4215,4,549000\n2162,4,287000\n1664,2,368500\n2238,3,329900\n2567,4,314000\n1200,3,299000\n852,2,179900\n1852,4,299900\n1203,3,239500\n" }, { "path": "machine-learning-ex1/ex1/featureNormalize.m", "content": "function [X_norm, mu, sigma] = featureNormalize(X)\n%FEATURENORMALIZE Normalizes the features in X \n% FEATURENORMALIZE(X) returns a normalized version of X where\n% the mean value of each feature is 0 and the standard deviation\n% is 1. This is often a good preprocessing step to do when\n% working with learning algorithms.\n\n% You need to set these values correctly\nX_norm = X;\nmu = zeros(1, size(X, 2));\nsigma = zeros(1, size(X, 2));\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: First, for each feature dimension, compute the mean\n% of the feature and subtract it from the dataset,\n% storing the mean value in mu. Next, compute the \n% standard deviation of each feature and divide\n% each feature by it's standard deviation, storing\n% the standard deviation in sigma. \n%\n% Note that X is a matrix where each column is a \n% feature and each row is an example. You need \n% to perform the normalization separately for \n% each feature. \n%\n% Hint: You might find the 'mean' and 'std' functions useful.\n% \nmu=mean(X_norm);\nsigma=std(X_norm);\nfor i=1:size(X,2)\n X_norm(:,i)=X_norm(:,i)-mu(i);\n X_norm(:,i)=X_norm(:,i)./sigma(i);\nend\n\n\n\n% ============================================================\n\nend\n" }, { "path": "machine-learning-ex1/ex1/gradientDescent.m", "content": "function [theta, J_history] = gradientDescent(X, y, theta, alpha, num_iters)\n%GRADIENTDESCENT Performs gradient descent to learn theta\n% theta = GRADIENTDESENT(X, y, theta, alpha, num_iters) updates theta by \n% taking num_iters gradient steps with learning rate alpha\n\n% Initialize some useful values\nm = length(y); % number of training examples\nJ_history = zeros(num_iters, 1);\n\nfor iter = 1:num_iters\n\n % ====================== YOUR CODE HERE ======================\n % Instructions: Perform a single gradient step on the parameter vector\n % theta. \n %\n % Hint: While debugging, it can be useful to print out the values\n % of the cost function (computeCost) and gradient here.\n %\n \n temp1=1/m*alpha*sum(X*theta-y);\n temp2=1/m*alpha*sum((X*theta-y).*X(:,2));\n theta(1)=theta(1)-temp1;\n theta(2)=theta(2)-temp2;\n\n\n % ============================================================\n\n % Save the cost J in every iteration \n J_history(iter) = computeCost(X, y, theta);\n\nend\n\nend\n" }, { "path": "machine-learning-ex1/ex1/gradientDescentMulti.m", "content": "function [theta, J_history] = gradientDescentMulti(X, y, theta, alpha, num_iters)\n%GRADIENTDESCENTMULTI Performs gradient descent to learn theta\n% theta = GRADIENTDESCENTMULTI(x, y, theta, alpha, num_iters) updates theta by\n% taking num_iters gradient steps with learning rate alpha\n\n% Initialize some useful values\nm = length(y); % number of training examples\nn = length(theta);\ntemp = zeros(n, num_iters);\nJ_history = zeros(num_iters, 1);\n\nfor iter = 1:num_iters\n\n % ====================== YOUR CODE HERE ======================\n % Instructions: Perform a single gradient step on the parameter vector\n % theta. \n %\n % Hint: While debugging, it can be useful to print out the values\n % of the cost function (computeCostMulti) and gradient here.\n %\n h = X*theta;\n temp(:,iter) = theta - ((alpha/m)*(X'*(h-y)));\n theta = temp(:,iter);\n \n %temp1=1/m*alpha*sum(X*theta-y);\n %temp2=1/m*alpha*sum((X*theta-y).*X(:,2));\n %temp3=1/m*alpha*sum((X*theta-y).*X(:,3));\n %theta(1)=theta(1)-temp1;\n %theta(2)=theta(2)-temp2;\n %theta(3)=theta(3)-temp3;\n\n % ============================================================\n\n % Save the cost J in every iteration \n J_history(iter) = computeCostMulti(X, y, theta);\n\nend\n\nend\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex1/ex1/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex2/ex2/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i3 matrix, where the first column is all-ones\n\n% Plot Data\nplotData(X(:,2:3), y);\nhold on\n\nif size(X, 2) <= 3\n % Only need 2 points to define a line, so choose two endpoints\n plot_x = [min(X(:,2))-2, max(X(:,2))+2];\n\n % Calculate the decision boundary line\n plot_y = (-1./theta(3)).*(theta(2).*plot_x + theta(1)); % plot_xΪX1plot_yΪX2\n\n % Plot, and adjust axes for better viewing\n plot(plot_x, plot_y)\n \n % Legend, specific for the exercise\n legend('Admitted', 'Not admitted', 'Decision Boundary')\n axis([30, 100, 30, 100])\nelse\n % Here is the grid range\n u = linspace(-1, 1.5, 50);\n v = linspace(-1, 1.5, 50);\n\n z = zeros(length(u), length(v));\n % Evaluate z = theta*x over the grid\n for i = 1:length(u)\n for j = 1:length(v)\n z(i,j) = mapFeature(u(i), v(j))*theta;\n end\n end\n z = z'; % important to transpose z before calling contour\n\n % Plot z = 0\n % Notice you need to specify the range [0, 0]\n contour(u, v, z, [0, 0], 'LineWidth', 2) %ֵΪ0ĵֵ\nend\nhold off\n\nend\n" }, { "path": "machine-learning-ex2/ex2/predict.m", "content": "function p = predict(theta, X)\n%PREDICT Predict whether the label is 0 or 1 using learned logistic \n%regression parameters theta\n% p = PREDICT(theta, X) computes the predictions for X using a \n% threshold at 0.5 (i.e., if sigmoid(theta'*x) >= 0.5, predict 1)\n\nm = size(X, 1); % Number of training examples\n\n% You need to return the following variables correctly\np = zeros(m, 1);\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Complete the following code to make predictions using\n% your learned logistic regression parameters. \n% You should set p to a vector of 0's and 1's\n%\n\n\np = sigmoid(X*theta);\n\nfor i=1:m\n if p(i)>0.5\n p(i)=1;\n else\n p(i)=0;\n end\nend\n\n\n\n\n% =========================================================================\n\n\nend\n" }, { "path": "machine-learning-ex2/ex2/sigmoid.m", "content": "function g = sigmoid(z)\n%SIGMOID Compute sigmoid functoon\n% J = SIGMOID(z) computes the sigmoid of z.\n\n% You need to return the following variables correctly \ng = zeros(size(z));\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Compute the sigmoid of each value of z (z can be a matrix,\n% vector or scalar).\n\ng = 1./(1+exp(-z));\n\n\n\n% =============================================================\n\nend\n" }, { "path": "machine-learning-ex2/ex2/submit.m", "content": "function submit()\n addpath('./lib');\n\n conf.assignmentSlug = 'logistic-regression';\n conf.itemName = 'Logistic Regression';\n conf.partArrays = { ...\n { ...\n '1', ...\n { 'sigmoid.m' }, ...\n 'Sigmoid Function', ...\n }, ...\n { ...\n '2', ...\n { 'costFunction.m' }, ...\n 'Logistic Regression Cost', ...\n }, ...\n { ...\n '3', ...\n { 'costFunction.m' }, ...\n 'Logistic Regression Gradient', ...\n }, ...\n { ...\n '4', ...\n { 'predict.m' }, ...\n 'Predict', ...\n }, ...\n { ...\n '5', ...\n { 'costFunctionReg.m' }, ...\n 'Regularized Logistic Regression Cost', ...\n }, ...\n { ...\n '6', ...\n { 'costFunctionReg.m' }, ...\n 'Regularized Logistic Regression Gradient', ...\n }, ...\n };\n conf.output = @output;\n\n submitWithConfiguration(conf);\nend\n\nfunction out = output(partId, auxstring)\n % Random Test Cases\n X = [ones(20,1) (exp(1) * sin(1:1:20))' (exp(0.5) * cos(1:1:20))'];\n y = sin(X(:,1) + X(:,2)) > 0;\n if partId == '1'\n out = sprintf('%0.5f ', sigmoid(X));\n elseif partId == '2'\n out = sprintf('%0.5f ', costFunction([0.25 0.5 -0.5]', X, y));\n elseif partId == '3'\n [cost, grad] = costFunction([0.25 0.5 -0.5]', X, y);\n out = sprintf('%0.5f ', grad);\n elseif partId == '4'\n out = sprintf('%0.5f ', predict([0.25 0.5 -0.5]', X));\n elseif partId == '5'\n out = sprintf('%0.5f ', costFunctionReg([0.25 0.5 -0.5]', X, y, 0.1));\n elseif partId == '6'\n [cost, grad] = costFunctionReg([0.25 0.5 -0.5]', X, y, 0.1);\n out = sprintf('%0.5f ', grad);\n end \nend\n" }, { "path": "machine-learning-ex3/ex3/displayData.m", "content": "function [h, display_array] = displayData(X, example_width)\n%DISPLAYDATA Display 2D data in a nice grid\n% [h, display_array] = DISPLAYDATA(X, example_width) displays 2D data\n% stored in X in a nice grid. It returns the figure handle h and the \n% displayed array if requested.\n\n% Set example_width automatically if not passed in\nif ~exist('example_width', 'var') || isempty(example_width) \n\texample_width = round(sqrt(size(X, 2)));\nend\n\n% Gray Image\ncolormap(gray);\n\n% Compute rows, cols\n[m n] = size(X);\nexample_height = (n / example_width);\n\n% Compute number of items to display\ndisplay_rows = floor(sqrt(m));\ndisplay_cols = ceil(m / display_rows);\n\n% Between images padding\npad = 1;\n\n% Setup blank display\ndisplay_array = - ones(pad + display_rows * (example_height + pad), ...\n pad + display_cols * (example_width + pad));\n\n% Copy each example into a patch on the display array\ncurr_ex = 1;\nfor j = 1:display_rows\n\tfor i = 1:display_cols\n\t\tif curr_ex > m, \n\t\t\tbreak; \n\t\tend\n\t\t% Copy the patch\n\t\t\n\t\t% Get the max value of the patch\n\t\tmax_val = max(abs(X(curr_ex, :)));\n\t\tdisplay_array(pad + (j - 1) * (example_height + pad) + (1:example_height), ...\n\t\t pad + (i - 1) * (example_width + pad) + (1:example_width)) = ...\n\t\t\t\t\t\treshape(X(curr_ex, :), example_height, example_width) / max_val;\n\t\tcurr_ex = curr_ex + 1;\n\tend\n\tif curr_ex > m, \n\t\tbreak; \n\tend\nend\n\n% Display Image\nh = imagesc(display_array, [-1 1]);\n\n% Do not show axis\naxis image off\n\ndrawnow;\n\nend\n" }, { "path": "machine-learning-ex3/ex3/ex3.m", "content": "%% Machine Learning Online Class - Exercise 3 | Part 1: One-vs-all\n\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% linear exercise. You will need to complete the following functions \n% in this exericse:\n%\n% lrCostFunction.m (logistic regression cost function)\n% oneVsAll.m\n% predictOneVsAll.m\n% predict.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% Setup the parameters you will use for this part of the exercise\ninput_layer_size = 400; % 20x20 Input Images of Digits\nnum_labels = 10; % 10 labels, from 1 to 10 \n % (note that we have mapped \"0\" to label 10)\n\n%% =========== Part 1: Loading and Visualizing Data =============\n% We start the exercise by first loading and visualizing the dataset. \n% You will be working with a dataset that contains handwritten digits.\n%\n\n% Load Training Data\nfprintf('Loading and Visualizing Data ...\\n')\n\nload('ex3data1.mat'); % training data stored in arrays X, y\nm = size(X, 1);\n\n% Randomly select 100 data points to display\nrand_indices = randperm(m);\nsel = X(rand_indices(1:100), :);\n\ndisplayData(sel);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ============ Part 2: Vectorize Logistic Regression ============\n% In this part of the exercise, you will reuse your logistic regression\n% code from the last exercise. You task here is to make sure that your\n% regularized logistic regression implementation is vectorized. After\n% that, you will implement one-vs-all classification for the handwritten\n% digit dataset.\n%\n\nfprintf('\\nTraining One-vs-All Logistic Regression...\\n')\n\nlambda = 0.1;\n[all_theta] = oneVsAll(X, y, num_labels, lambda);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ================ Part 3: Predict for One-Vs-All ================\n% After ...\npred = predictOneVsAll(all_theta, X);\n\nfprintf('\\nTraining Set Accuracy: %f\\n', mean(double(pred == y)) * 100);\n\n\n" }, { "path": "machine-learning-ex3/ex3/ex3_nn.m", "content": "%% Machine Learning Online Class - Exercise 3 | Part 2: Neural Networks\n\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% linear exercise. You will need to complete the following functions \n% in this exericse:\n%\n% lrCostFunction.m (logistic regression cost function)\n% oneVsAll.m\n% predictOneVsAll.m\n% predict.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% Setup the parameters you will use for this exercise\ninput_layer_size = 400; % 20x20 Input Images of Digits\nhidden_layer_size = 25; % 25 hidden units\nnum_labels = 10; % 10 labels, from 1 to 10 \n % (note that we have mapped \"0\" to label 10)\n\n%% =========== Part 1: Loading and Visualizing Data =============\n% We start the exercise by first loading and visualizing the dataset. \n% You will be working with a dataset that contains handwritten digits.\n%\n\n% Load Training Data\nfprintf('Loading and Visualizing Data ...\\n')\n\nload('ex3data1.mat');\nm = size(X, 1);\n\n% Randomly select 100 data points to display\nsel = randperm(size(X, 1));\nsel = sel(1:100);\n\ndisplayData(X(sel, :));\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ================ Part 2: Loading Pameters ================\n% In this part of the exercise, we load some pre-initialized \n% neural network parameters.\n\nfprintf('\\nLoading Saved Neural Network Parameters ...\\n')\n\n% Load the weights into variables Theta1 and Theta2\nload('ex3weights.mat');\n\n%% ================= Part 3: Implement Predict =================\n% After training the neural network, we would like to use it to predict\n% the labels. You will now implement the \"predict\" function to use the\n% neural network to predict the labels of the training set. This lets\n% you compute the training set accuracy.\n\npred = predict(Theta1, Theta2, X);\n\nfprintf('\\nTraining Set Accuracy: %f\\n', mean(double(pred == y)) * 100);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n% To give you an idea of the network's output, you can also run\n% through the examples one at the a time to see what it is predicting.\n\n% Randomly permute examples\nrp = randperm(m);\n\nfor i = 1:m\n % Display \n fprintf('\\nDisplaying Example Image\\n');\n displayData(X(rp(i), :));\n\n pred = predict(Theta1, Theta2, X(rp(i),:));\n fprintf('\\nNeural Network Prediction: %d (digit %d)\\n', pred, mod(pred, 10));%mod࣬Ϊ0ӳ10\n \n % Pause\n fprintf('Program paused. Press enter to continue.\\n');\n pause;\nend\n\n" }, { "path": "machine-learning-ex3/ex3/fmincg.m", "content": "function [X, fX, i] = fmincg(f, X, options, P1, P2, P3, P4, P5)\n% Minimize a continuous differentialble multivariate function. Starting point\n% is given by \"X\" (D by 1), and the function named in the string \"f\", must\n% return a function value and a vector of partial derivatives. The Polack-\n% Ribiere flavour of conjugate gradients is used to compute search directions,\n% and a line search using quadratic and cubic polynomial approximations and the\n% Wolfe-Powell stopping criteria is used together with the slope ratio method\n% for guessing initial step sizes. Additionally a bunch of checks are made to\n% make sure that exploration is taking place and that extrapolation will not\n% be unboundedly large. The \"length\" gives the length of the run: if it is\n% positive, it gives the maximum number of line searches, if negative its\n% absolute gives the maximum allowed number of function evaluations. You can\n% (optionally) give \"length\" a second component, which will indicate the\n% reduction in function value to be expected in the first line-search (defaults\n% to 1.0). The function returns when either its length is up, or if no further\n% progress can be made (ie, we are at a minimum, or so close that due to\n% numerical problems, we cannot get any closer). If the function terminates\n% within a few iterations, it could be an indication that the function value\n% and derivatives are not consistent (ie, there may be a bug in the\n% implementation of your \"f\" function). The function returns the found\n% solution \"X\", a vector of function values \"fX\" indicating the progress made\n% and \"i\" the number of iterations (line searches or function evaluations,\n% depending on the sign of \"length\") used.\n%\n% Usage: [X, fX, i] = fmincg(f, X, options, P1, P2, P3, P4, P5)\n%\n% See also: checkgrad \n%\n% Copyright (C) 2001 and 2002 by Carl Edward Rasmussen. Date 2002-02-13\n%\n%\n% (C) Copyright 1999, 2000 & 2001, Carl Edward Rasmussen\n% \n% Permission is granted for anyone to copy, use, or modify these\n% programs and accompanying documents for purposes of research or\n% education, provided this copyright notice is retained, and note is\n% made of any changes that have been made.\n% \n% These programs and documents are distributed without any warranty,\n% express or implied. As the programs were written for research\n% purposes only, they have not been tested to the degree that would be\n% advisable in any important application. All use of these programs is\n% entirely at the user's own risk.\n%\n% [ml-class] Changes Made:\n% 1) Function name and argument specifications\n% 2) Output display\n%\n\n% Read options\nif exist('options', 'var') && ~isempty(options) && isfield(options, 'MaxIter')\n length = options.MaxIter;\nelse\n length = 100;\nend\n\n\nRHO = 0.01; % a bunch of constants for line searches\nSIG = 0.5; % RHO and SIG are the constants in the Wolfe-Powell conditions\nINT = 0.1; % don't reevaluate within 0.1 of the limit of the current bracket\nEXT = 3.0; % extrapolate maximum 3 times the current bracket\nMAX = 20; % max 20 function evaluations per line search\nRATIO = 100; % maximum allowed slope ratio\n\nargstr = ['feval(f, X']; % compose string used to call function\nfor i = 1:(nargin - 3)\n argstr = [argstr, ',P', int2str(i)];\nend\nargstr = [argstr, ')'];\n\nif max(size(length)) == 2, red=length(2); length=length(1); else red=1; end\nS=['Iteration '];\n\ni = 0; % zero the run length counter\nls_failed = 0; % no previous line search has failed\nfX = [];\n[f1 df1] = eval(argstr); % get function value and gradient\ni = i + (length<0); % count epochs?!\ns = -df1; % search direction is steepest\nd1 = -s'*s; % this is the slope\nz1 = red/(1-d1); % initial step is red/(|s|+1)\n\nwhile i < abs(length) % while not finished\n i = i + (length>0); % count iterations?!\n\n X0 = X; f0 = f1; df0 = df1; % make a copy of current values\n X = X + z1*s; % begin line search\n [f2 df2] = eval(argstr);\n i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n f3 = f1; d3 = d1; z3 = -z1; % initialize point 3 equal to point 1\n if length>0, M = MAX; else M = min(MAX, -length-i); end\n success = 0; limit = -1; % initialize quanteties\n while 1\n while ((f2 > f1+z1*RHO*d1) || (d2 > -SIG*d1)) && (M > 0) \n limit = z1; % tighten the bracket\n if f2 > f1\n z2 = z3 - (0.5*d3*z3*z3)/(d3*z3+f2-f3); % quadratic fit\n else\n A = 6*(f2-f3)/z3+3*(d2+d3); % cubic fit\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = (sqrt(B*B-A*d2*z3*z3)-B)/A; % numerical error possible - ok!\n end\n if isnan(z2) || isinf(z2)\n z2 = z3/2; % if we had a numerical problem then bisect\n end\n z2 = max(min(z2, INT*z3),(1-INT)*z3); % don't accept too close to limits\n z1 = z1 + z2; % update the step\n X = X + z2*s;\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n z3 = z3-z2; % z3 is now relative to the location of z2\n end\n if f2 > f1+z1*RHO*d1 || d2 > -SIG*d1\n break; % this is a failure\n elseif d2 > SIG*d1\n success = 1; break; % success\n elseif M == 0\n break; % failure\n end\n A = 6*(f2-f3)/z3+3*(d2+d3); % make cubic extrapolation\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = -d2*z3*z3/(B+sqrt(B*B-A*d2*z3*z3)); % num. error possible - ok!\n if ~isreal(z2) || isnan(z2) || isinf(z2) || z2 < 0 % num prob or wrong sign?\n if limit < -0.5 % if we have no upper limit\n z2 = z1 * (EXT-1); % the extrapolate the maximum amount\n else\n z2 = (limit-z1)/2; % otherwise bisect\n end\n elseif (limit > -0.5) && (z2+z1 > limit) % extraplation beyond max?\n z2 = (limit-z1)/2; % bisect\n elseif (limit < -0.5) && (z2+z1 > z1*EXT) % extrapolation beyond limit\n z2 = z1*(EXT-1.0); % set to extrapolation limit\n elseif z2 < -z3*INT\n z2 = -z3*INT;\n elseif (limit > -0.5) && (z2 < (limit-z1)*(1.0-INT)) % too close to limit?\n z2 = (limit-z1)*(1.0-INT);\n end\n f3 = f2; d3 = d2; z3 = -z2; % set point 3 equal to point 2\n z1 = z1 + z2; X = X + z2*s; % update current estimates\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n end % end of line search\n\n if success % if line search succeeded\n f1 = f2; fX = [fX' f1]';\n fprintf('%s %4i | Cost: %4.6e\\r', S, i, f1);\n s = (df2'*df2-df1'*df2)/(df1'*df1)*s - df2; % Polack-Ribiere direction\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n d2 = df1'*s;\n if d2 > 0 % new slope must be negative\n s = -df1; % otherwise use steepest direction\n d2 = -s'*s; \n end\n z1 = z1 * min(RATIO, d1/(d2-realmin)); % slope ratio but max RATIO\n d1 = d2;\n ls_failed = 0; % this line search did not fail\n else\n X = X0; f1 = f0; df1 = df0; % restore point from before failed line search\n if ls_failed || i > abs(length) % line search failed twice in a row\n break; % or we ran out of time, so we give up\n end\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n s = -df1; % try steepest\n d1 = -s'*s;\n z1 = 1/(1-d1); \n ls_failed = 1; % this line search failed\n end\n if exist('OCTAVE_VERSION')\n fflush(stdout);\n end\nend\nfprintf('\\n');\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex3/ex3/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i 0;\n Xm = [ -1 -1 ; -1 -2 ; -2 -1 ; -2 -2 ; ...\n 1 1 ; 1 2 ; 2 1 ; 2 2 ; ...\n -1 1 ; -1 2 ; -2 1 ; -2 2 ; ...\n 1 -1 ; 1 -2 ; -2 -1 ; -2 -2 ];\n ym = [ 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4 ]';\n t1 = sin(reshape(1:2:24, 4, 3));\n t2 = cos(reshape(1:2:40, 4, 5));\n\n if partId == '1'\n [J, grad] = lrCostFunction([0.25 0.5 -0.5]', X, y, 0.1);\n out = sprintf('%0.5f ', J);\n out = [out sprintf('%0.5f ', grad)];\n elseif partId == '2'\n out = sprintf('%0.5f ', oneVsAll(Xm, ym, 4, 0.1));\n elseif partId == '3'\n out = sprintf('%0.5f ', predictOneVsAll(t1, Xm));\n elseif partId == '4'\n out = sprintf('%0.5f ', predict(t1, t2, Xm));\n end \nend\n" }, { "path": "machine-learning-ex4/ex4/checkNNGradients.m", "content": "function checkNNGradients(lambda)\n%CHECKNNGRADIENTS Creates a small neural network to check the\n%backpropagation gradients\n% CHECKNNGRADIENTS(lambda) Creates a small neural network to check the\n% backpropagation gradients, it will output the analytical gradients\n% produced by your backprop code and the numerical gradients (computed\n% using computeNumericalGradient). These two gradient computations should\n% result in very similar values.\n%\n\nif ~exist('lambda', 'var') || isempty(lambda)\n lambda = 0;\nend\n\ninput_layer_size = 3;\nhidden_layer_size = 5;\nnum_labels = 3;\nm = 5;\n\n% We generate some 'random' test data\nTheta1 = debugInitializeWeights(hidden_layer_size, input_layer_size);\nTheta2 = debugInitializeWeights(num_labels, hidden_layer_size);\n% Reusing debugInitializeWeights to generate X\nX = debugInitializeWeights(m, input_layer_size - 1);\ny = 1 + mod(1:m, num_labels)';\n\n% Unroll parameters\nnn_params = [Theta1(:) ; Theta2(:)];\n\n% Short hand for cost function\ncostFunc = @(p) nnCostFunction(p, input_layer_size, hidden_layer_size, ...\n num_labels, X, y, lambda);\n\n[cost, grad] = costFunc(nn_params);\nnumgrad = computeNumericalGradient(costFunc, nn_params);\n\n% Visually examine the two gradient computations. The two columns\n% you get should be very similar. \ndisp([numgrad grad]);\nfprintf(['The above two columns you get should be very similar.\\n' ...\n '(Left-Your Numerical Gradient, Right-Analytical Gradient)\\n\\n']);\n\n% Evaluate the norm of the difference between two solutions. \n% If you have a correct implementation, and assuming you used EPSILON = 0.0001 \n% in computeNumericalGradient.m, then diff below should be less than 1e-9\ndiff = norm(numgrad-grad)/norm(numgrad+grad);\n\nfprintf(['If your backpropagation implementation is correct, then \\n' ...\n 'the relative difference will be small (less than 1e-9). \\n' ...\n '\\nRelative Difference: %g\\n'], diff);\n\nend\n" }, { "path": "machine-learning-ex4/ex4/computeNumericalGradient.m", "content": "function numgrad = computeNumericalGradient(J, theta)\n%COMPUTENUMERICALGRADIENT Computes the gradient using \"finite differences\"\n%and gives us a numerical estimate of the gradient.\n% numgrad = COMPUTENUMERICALGRADIENT(J, theta) computes the numerical\n% gradient of the function J around theta. Calling y = J(theta) should\n% return the function value at theta.\n\n% Notes: The following code implements numerical gradient checking, and \n% returns the numerical gradient.It sets numgrad(i) to (a numerical \n% approximation of) the partial derivative of J with respect to the \n% i-th input argument, evaluated at theta. (i.e., numgrad(i) should \n% be the (approximately) the partial derivative of J with respect \n% to theta(i).)\n% \n\nnumgrad = zeros(size(theta));\nperturb = zeros(size(theta));\ne = 1e-4;\nfor p = 1:numel(theta)\n % Set perturbation vector\n perturb(p) = e;\n loss1 = J(theta - perturb);\n loss2 = J(theta + perturb);\n % Compute Numerical Gradient\n numgrad(p) = (loss2 - loss1) / (2*e);\n perturb(p) = 0;\nend\n\nend\n" }, { "path": "machine-learning-ex4/ex4/debugInitializeWeights.m", "content": "function W = debugInitializeWeights(fan_out, fan_in)\n%DEBUGINITIALIZEWEIGHTS Initialize the weights of a layer with fan_in\n%incoming connections and fan_out outgoing connections using a fixed\n%strategy, this will help you later in debugging\n% W = DEBUGINITIALIZEWEIGHTS(fan_in, fan_out) initializes the weights \n% of a layer with fan_in incoming connections and fan_out outgoing \n% connections using a fix set of values\n%\n% Note that W should be set to a matrix of size(1 + fan_in, fan_out) as\n% the first row of W handles the \"bias\" terms\n%\n\n% Set W to zeros\nW = zeros(fan_out, 1 + fan_in);\n\n% Initialize W using \"sin\", this ensures that W is always of the same\n% values and will be useful for debugging\nW = reshape(sin(1:numel(W)), size(W)) / 10;\n\n% =========================================================================\n\nend\n" }, { "path": "machine-learning-ex4/ex4/displayData.m", "content": "function [h, display_array] = displayData(X, example_width)\n%DISPLAYDATA Display 2D data in a nice grid\n% [h, display_array] = DISPLAYDATA(X, example_width) displays 2D data\n% stored in X in a nice grid. It returns the figure handle h and the \n% displayed array if requested.\n\n% Set example_width automatically if not passed in\nif ~exist('example_width', 'var') || isempty(example_width) \n\texample_width = round(sqrt(size(X, 2)));\nend\n\n% Gray Image\ncolormap(gray);\n\n% Compute rows, cols\n[m n] = size(X);\nexample_height = (n / example_width);\n\n% Compute number of items to display\ndisplay_rows = floor(sqrt(m));\ndisplay_cols = ceil(m / display_rows);\n\n% Between images padding\npad = 1;\n\n% Setup blank display\ndisplay_array = - ones(pad + display_rows * (example_height + pad), ...\n pad + display_cols * (example_width + pad));\n\n% Copy each example into a patch on the display array\ncurr_ex = 1;\nfor j = 1:display_rows\n\tfor i = 1:display_cols\n\t\tif curr_ex > m, \n\t\t\tbreak; \n\t\tend\n\t\t% Copy the patch\n\t\t\n\t\t% Get the max value of the patch\n\t\tmax_val = max(abs(X(curr_ex, :)));\n\t\tdisplay_array(pad + (j - 1) * (example_height + pad) + (1:example_height), ...\n\t\t pad + (i - 1) * (example_width + pad) + (1:example_width)) = ...\n\t\t\t\t\t\treshape(X(curr_ex, :), example_height, example_width) / max_val;\n\t\tcurr_ex = curr_ex + 1;\n\tend\n\tif curr_ex > m, \n\t\tbreak; \n\tend\nend\n\n% Display Image\nh = imagesc(display_array, [-1 1]);\n\n% Do not show axis\naxis image off\n\ndrawnow;\n\nend\n" }, { "path": "machine-learning-ex4/ex4/ex4.m", "content": "%% Machine Learning Online Class - Exercise 4 Neural Network Learning\n\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% linear exercise. You will need to complete the following functions \n% in this exericse:\n%\n% sigmoidGradient.m\n% randInitializeWeights.m\n% nnCostFunction.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% Setup the parameters you will use for this exercise\ninput_layer_size = 400; % 20x20 Input Images of Digits\nhidden_layer_size = 25; % 25 hidden units\nnum_labels = 10; % 10 labels, from 1 to 10 \n % (note that we have mapped \"0\" to label 10)\n\n%% =========== Part 1: Loading and Visualizing Data =============\n% We start the exercise by first loading and visualizing the dataset. \n% You will be working with a dataset that contains handwritten digits.\n%\n\n% Load Training Data\nfprintf('Loading and Visualizing Data ...\\n')\n\nload('ex4data1.mat');\nm = size(X, 1);\n\n% Randomly select 100 data points to display\nsel = randperm(size(X, 1));\nsel = sel(1:100);\n\ndisplayData(X(sel, :));\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ================ Part 2: Loading Parameters ================\n% In this part of the exercise, we load some pre-initialized \n% neural network parameters.\n\nfprintf('\\nLoading Saved Neural Network Parameters ...\\n')\n\n% Load the weights into variables Theta1 and Theta2\nload('ex4weights.mat');\n\n% Unroll parameters \nnn_params = [Theta1(:) ; Theta2(:)];\n\n%% ================ Part 3: Compute Cost (Feedforward) ================\n% To the neural network, you should first start by implementing the\n% feedforward part of the neural network that returns the cost only. You\n% should complete the code in nnCostFunction.m to return cost. After\n% implementing the feedforward to compute the cost, you can verify that\n% your implementation is correct by verifying that you get the same cost\n% as us for the fixed debugging parameters.\n%\n% We suggest implementing the feedforward cost *without* regularization\n% first so that it will be easier for you to debug. Later, in part 4, you\n% will get to implement the regularized cost.\n%\nfprintf('\\nFeedforward Using Neural Network ...\\n')\n\n% Weight regularization parameter (we set this to 0 here).\nlambda = 0.1;\n\nJ = nnCostFunction(nn_params, input_layer_size, hidden_layer_size, ...\n num_labels, X, y, lambda);\n\nfprintf(['Cost at parameters (loaded from ex4weights): %f '...\n '\\n(this value should be about 0.287629)\\n'], J);\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n%% =============== Part 4: Implement Regularization ===============\n% Once your cost function implementation is correct, you should now\n% continue to implement the regularization with the cost.\n%\n\nfprintf('\\nChecking Cost Function (w/ Regularization) ... \\n')\n\n% Weight regularization parameter (we set this to 1 here).\nlambda = 1;\n\nJ = nnCostFunction(nn_params, input_layer_size, hidden_layer_size, ...\n num_labels, X, y, lambda);\n\nfprintf(['Cost at parameters (loaded from ex4weights): %f '...\n '\\n(this value should be about 0.383770)\\n'], J);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ================ Part 5: Sigmoid Gradient ================\n% Before you start implementing the neural network, you will first\n% implement the gradient for the sigmoid function. You should complete the\n% code in the sigmoidGradient.m file.\n%\n\nfprintf('\\nEvaluating sigmoid gradient...\\n')\n\ng = sigmoidGradient([1 -0.5 0 0.5 1]);\nfprintf('Sigmoid gradient evaluated at [1 -0.5 0 0.5 1]:\\n ');\nfprintf('%f ', g);\nfprintf('\\n\\n');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ================ Part 6: Initializing Pameters ================\n% In this part of the exercise, you will be starting to implment a two\n% layer neural network that classifies digits. You will start by\n% implementing a function to initialize the weights of the neural network\n% (randInitializeWeights.m)\n\nfprintf('\\nInitializing Neural Network Parameters ...\\n')\n\ninitial_Theta1 = randInitializeWeights(input_layer_size, hidden_layer_size);\ninitial_Theta2 = randInitializeWeights(hidden_layer_size, num_labels);\n\n% Unroll parameters\ninitial_nn_params = [initial_Theta1(:) ; initial_Theta2(:)];\n\n\n%% =============== Part 7: Implement Backpropagation ===============\n% Once your cost matches up with ours, you should proceed to implement the\n% backpropagation algorithm for the neural network. You should add to the\n% code you've written in nnCostFunction.m to return the partial\n% derivatives of the parameters.\n%\nfprintf('\\nChecking Backpropagation... \\n');\n\n% Check gradients by running checkNNGradients\ncheckNNGradients;\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n\n%% =============== Part 8: Implement Regularization ===============\n% Once your backpropagation implementation is correct, you should now\n% continue to implement the regularization with the cost and gradient.\n%\n\nfprintf('\\nChecking Backpropagation (w/ Regularization) ... \\n')\n\n% Check gradients by running checkNNGradients\nlambda = 3;\ncheckNNGradients(lambda);\n\n% Also output the costFunction debugging values\ndebug_J = nnCostFunction(nn_params, input_layer_size, ...\n hidden_layer_size, num_labels, X, y, lambda);\n\nfprintf(['\\n\\nCost at (fixed) debugging parameters (w/ lambda = 10): %f ' ...\n '\\n(this value should be about 0.576051)\\n\\n'], debug_J);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% =================== Part 8: Training NN ===================\n% You have now implemented all the code necessary to train a neural \n% network. To train your neural network, we will now use \"fmincg\", which\n% is a function which works similarly to \"fminunc\". Recall that these\n% advanced optimizers are able to train our cost functions efficiently as\n% long as we provide them with the gradient computations.\n%\nfprintf('\\nTraining Neural Network... \\n')\n\n% After you have completed the assignment, change the MaxIter to a larger\n% value to see how more training helps.\noptions = optimset('MaxIter', 50);\n\n% You should also try different values of lambda\nlambda = 1;\n\n% Create \"short hand\" for the cost function to be minimized\ncostFunction = @(p) nnCostFunction(p, ...\n input_layer_size, ...\n hidden_layer_size, ...\n num_labels, X, y, lambda);\n\n% Now, costFunction is a function that takes in only one argument (the\n% neural network parameters)\n[nn_params, cost] = fmincg(costFunction, initial_nn_params, options);\n\n% Obtain Theta1 and Theta2 back from nn_params\nTheta1 = reshape(nn_params(1:hidden_layer_size * (input_layer_size + 1)), ...\n hidden_layer_size, (input_layer_size + 1));\n\nTheta2 = reshape(nn_params((1 + (hidden_layer_size * (input_layer_size + 1))):end), ...\n num_labels, (hidden_layer_size + 1));\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ================= Part 9: Visualize Weights =================\n% You can now \"visualize\" what the neural network is learning by \n% displaying the hidden units to see what features they are capturing in \n% the data.\n\nfprintf('\\nVisualizing Neural Network... \\n')\n\ndisplayData(Theta1(:, 2:end));\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n%% ================= Part 10: Implement Predict =================\n% After training the neural network, we would like to use it to predict\n% the labels. You will now implement the \"predict\" function to use the\n% neural network to predict the labels of the training set. This lets\n% you compute the training set accuracy.\n\npred = predict(Theta1, Theta2, X);\n\nfprintf('\\nTraining Set Accuracy: %f\\n', mean(double(pred == y)) * 100);\n\n\n" }, { "path": "machine-learning-ex4/ex4/fmincg.m", "content": "function [X, fX, i] = fmincg(f, X, options, P1, P2, P3, P4, P5)\n% Minimize a continuous differentialble multivariate function. Starting point\n% is given by \"X\" (D by 1), and the function named in the string \"f\", must\n% return a function value and a vector of partial derivatives. The Polack-\n% Ribiere flavour of conjugate gradients is used to compute search directions,\n% and a line search using quadratic and cubic polynomial approximations and the\n% Wolfe-Powell stopping criteria is used together with the slope ratio method\n% for guessing initial step sizes. Additionally a bunch of checks are made to\n% make sure that exploration is taking place and that extrapolation will not\n% be unboundedly large. The \"length\" gives the length of the run: if it is\n% positive, it gives the maximum number of line searches, if negative its\n% absolute gives the maximum allowed number of function evaluations. You can\n% (optionally) give \"length\" a second component, which will indicate the\n% reduction in function value to be expected in the first line-search (defaults\n% to 1.0). The function returns when either its length is up, or if no further\n% progress can be made (ie, we are at a minimum, or so close that due to\n% numerical problems, we cannot get any closer). If the function terminates\n% within a few iterations, it could be an indication that the function value\n% and derivatives are not consistent (ie, there may be a bug in the\n% implementation of your \"f\" function). The function returns the found\n% solution \"X\", a vector of function values \"fX\" indicating the progress made\n% and \"i\" the number of iterations (line searches or function evaluations,\n% depending on the sign of \"length\") used.\n%\n% Usage: [X, fX, i] = fmincg(f, X, options, P1, P2, P3, P4, P5)\n%\n% See also: checkgrad \n%\n% Copyright (C) 2001 and 2002 by Carl Edward Rasmussen. Date 2002-02-13\n%\n%\n% (C) Copyright 1999, 2000 & 2001, Carl Edward Rasmussen\n% \n% Permission is granted for anyone to copy, use, or modify these\n% programs and accompanying documents for purposes of research or\n% education, provided this copyright notice is retained, and note is\n% made of any changes that have been made.\n% \n% These programs and documents are distributed without any warranty,\n% express or implied. As the programs were written for research\n% purposes only, they have not been tested to the degree that would be\n% advisable in any important application. All use of these programs is\n% entirely at the user's own risk.\n%\n% [ml-class] Changes Made:\n% 1) Function name and argument specifications\n% 2) Output display\n%\n\n% Read options\nif exist('options', 'var') && ~isempty(options) && isfield(options, 'MaxIter')\n length = options.MaxIter;\nelse\n length = 100;\nend\n\n\nRHO = 0.01; % a bunch of constants for line searches\nSIG = 0.5; % RHO and SIG are the constants in the Wolfe-Powell conditions\nINT = 0.1; % don't reevaluate within 0.1 of the limit of the current bracket\nEXT = 3.0; % extrapolate maximum 3 times the current bracket\nMAX = 20; % max 20 function evaluations per line search\nRATIO = 100; % maximum allowed slope ratio\n\nargstr = ['feval(f, X']; % compose string used to call function\nfor i = 1:(nargin - 3)\n argstr = [argstr, ',P', int2str(i)];\nend\nargstr = [argstr, ')'];\n\nif max(size(length)) == 2, red=length(2); length=length(1); else red=1; end\nS=['Iteration '];\n\ni = 0; % zero the run length counter\nls_failed = 0; % no previous line search has failed\nfX = [];\n[f1 df1] = eval(argstr); % get function value and gradient\ni = i + (length<0); % count epochs?!\ns = -df1; % search direction is steepest\nd1 = -s'*s; % this is the slope\nz1 = red/(1-d1); % initial step is red/(|s|+1)\n\nwhile i < abs(length) % while not finished\n i = i + (length>0); % count iterations?!\n\n X0 = X; f0 = f1; df0 = df1; % make a copy of current values\n X = X + z1*s; % begin line search\n [f2 df2] = eval(argstr);\n i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n f3 = f1; d3 = d1; z3 = -z1; % initialize point 3 equal to point 1\n if length>0, M = MAX; else M = min(MAX, -length-i); end\n success = 0; limit = -1; % initialize quanteties\n while 1\n while ((f2 > f1+z1*RHO*d1) || (d2 > -SIG*d1)) && (M > 0) \n limit = z1; % tighten the bracket\n if f2 > f1\n z2 = z3 - (0.5*d3*z3*z3)/(d3*z3+f2-f3); % quadratic fit\n else\n A = 6*(f2-f3)/z3+3*(d2+d3); % cubic fit\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = (sqrt(B*B-A*d2*z3*z3)-B)/A; % numerical error possible - ok!\n end\n if isnan(z2) || isinf(z2)\n z2 = z3/2; % if we had a numerical problem then bisect\n end\n z2 = max(min(z2, INT*z3),(1-INT)*z3); % don't accept too close to limits\n z1 = z1 + z2; % update the step\n X = X + z2*s;\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n z3 = z3-z2; % z3 is now relative to the location of z2\n end\n if f2 > f1+z1*RHO*d1 || d2 > -SIG*d1\n break; % this is a failure\n elseif d2 > SIG*d1\n success = 1; break; % success\n elseif M == 0\n break; % failure\n end\n A = 6*(f2-f3)/z3+3*(d2+d3); % make cubic extrapolation\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = -d2*z3*z3/(B+sqrt(B*B-A*d2*z3*z3)); % num. error possible - ok!\n if ~isreal(z2) || isnan(z2) || isinf(z2) || z2 < 0 % num prob or wrong sign?\n if limit < -0.5 % if we have no upper limit\n z2 = z1 * (EXT-1); % the extrapolate the maximum amount\n else\n z2 = (limit-z1)/2; % otherwise bisect\n end\n elseif (limit > -0.5) && (z2+z1 > limit) % extraplation beyond max?\n z2 = (limit-z1)/2; % bisect\n elseif (limit < -0.5) && (z2+z1 > z1*EXT) % extrapolation beyond limit\n z2 = z1*(EXT-1.0); % set to extrapolation limit\n elseif z2 < -z3*INT\n z2 = -z3*INT;\n elseif (limit > -0.5) && (z2 < (limit-z1)*(1.0-INT)) % too close to limit?\n z2 = (limit-z1)*(1.0-INT);\n end\n f3 = f2; d3 = d2; z3 = -z2; % set point 3 equal to point 2\n z1 = z1 + z2; X = X + z2*s; % update current estimates\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n end % end of line search\n\n if success % if line search succeeded\n f1 = f2; fX = [fX' f1]';\n fprintf('%s %4i | Cost: %4.6e\\r', S, i, f1);\n s = (df2'*df2-df1'*df2)/(df1'*df1)*s - df2; % Polack-Ribiere direction\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n d2 = df1'*s;\n if d2 > 0 % new slope must be negative\n s = -df1; % otherwise use steepest direction\n d2 = -s'*s; \n end\n z1 = z1 * min(RATIO, d1/(d2-realmin)); % slope ratio but max RATIO\n d1 = d2;\n ls_failed = 0; % this line search did not fail\n else\n X = X0; f1 = f0; df1 = df0; % restore point from before failed line search\n if ls_failed || i > abs(length) % line search failed twice in a row\n break; % or we ran out of time, so we give up\n end\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n s = -df1; % try steepest\n d1 = -s'*s;\n z1 = 1/(1-d1); \n ls_failed = 1; % this line search failed\n end\n if exist('OCTAVE_VERSION')\n fflush(stdout);\n end\nend\nfprintf('\\n');\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex4/ex4/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i0); % count iterations?!\n\n X0 = X; f0 = f1; df0 = df1; % make a copy of current values\n X = X + z1*s; % begin line search\n [f2 df2] = eval(argstr);\n i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n f3 = f1; d3 = d1; z3 = -z1; % initialize point 3 equal to point 1\n if length>0, M = MAX; else M = min(MAX, -length-i); end\n success = 0; limit = -1; % initialize quanteties\n while 1\n while ((f2 > f1+z1*RHO*d1) || (d2 > -SIG*d1)) && (M > 0) \n limit = z1; % tighten the bracket\n if f2 > f1\n z2 = z3 - (0.5*d3*z3*z3)/(d3*z3+f2-f3); % quadratic fit\n else\n A = 6*(f2-f3)/z3+3*(d2+d3); % cubic fit\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = (sqrt(B*B-A*d2*z3*z3)-B)/A; % numerical error possible - ok!\n end\n if isnan(z2) || isinf(z2)\n z2 = z3/2; % if we had a numerical problem then bisect\n end\n z2 = max(min(z2, INT*z3),(1-INT)*z3); % don't accept too close to limits\n z1 = z1 + z2; % update the step\n X = X + z2*s;\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n z3 = z3-z2; % z3 is now relative to the location of z2\n end\n if f2 > f1+z1*RHO*d1 || d2 > -SIG*d1\n break; % this is a failure\n elseif d2 > SIG*d1\n success = 1; break; % success\n elseif M == 0\n break; % failure\n end\n A = 6*(f2-f3)/z3+3*(d2+d3); % make cubic extrapolation\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = -d2*z3*z3/(B+sqrt(B*B-A*d2*z3*z3)); % num. error possible - ok!\n if ~isreal(z2) || isnan(z2) || isinf(z2) || z2 < 0 % num prob or wrong sign?\n if limit < -0.5 % if we have no upper limit\n z2 = z1 * (EXT-1); % the extrapolate the maximum amount\n else\n z2 = (limit-z1)/2; % otherwise bisect\n end\n elseif (limit > -0.5) && (z2+z1 > limit) % extraplation beyond max?\n z2 = (limit-z1)/2; % bisect\n elseif (limit < -0.5) && (z2+z1 > z1*EXT) % extrapolation beyond limit\n z2 = z1*(EXT-1.0); % set to extrapolation limit\n elseif z2 < -z3*INT\n z2 = -z3*INT;\n elseif (limit > -0.5) && (z2 < (limit-z1)*(1.0-INT)) % too close to limit?\n z2 = (limit-z1)*(1.0-INT);\n end\n f3 = f2; d3 = d2; z3 = -z2; % set point 3 equal to point 2\n z1 = z1 + z2; X = X + z2*s; % update current estimates\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n end % end of line search\n\n if success % if line search succeeded\n f1 = f2; fX = [fX' f1]';\n fprintf('%s %4i | Cost: %4.6e\\r', S, i, f1);\n s = (df2'*df2-df1'*df2)/(df1'*df1)*s - df2; % Polack-Ribiere direction\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n d2 = df1'*s;\n if d2 > 0 % new slope must be negative\n s = -df1; % otherwise use steepest direction\n d2 = -s'*s; \n end\n z1 = z1 * min(RATIO, d1/(d2-realmin)); % slope ratio but max RATIO\n d1 = d2;\n ls_failed = 0; % this line search did not fail\n else\n X = X0; f1 = f0; df1 = df0; % restore point from before failed line search\n if ls_failed || i > abs(length) % line search failed twice in a row\n break; % or we ran out of time, so we give up\n end\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n s = -df1; % try steepest\n d1 = -s'*s;\n z1 = 1/(1-d1); \n ls_failed = 1; % this line search failed\n end\n if exist('OCTAVE_VERSION')\n fflush(stdout);\n end\nend\nfprintf('\\n');\n" }, { "path": "machine-learning-ex5/ex5/learningCurve.m", "content": "function [error_train, error_val] = ...\n learningCurve(X, y, Xval, yval, lambda)\n%LEARNINGCURVE Generates the train and cross validation set errors needed \n%to plot a learning curve\n% [error_train, error_val] = ...\n% LEARNINGCURVE(X, y, Xval, yval, lambda) returns the train and\n% cross validation set errors for a learning curve. In particular, \n% it returns two vectors of the same length - error_train and \n% error_val. Then, error_train(i) contains the training error for\n% i examples (and similarly for error_val(i)).\n%\n% In this function, you will compute the train and test errors for\n% dataset sizes from 1 up to m. In practice, when working with larger\n% datasets, you might want to do this in larger intervals.\n%\n\n% Number of training examples\nm = size(X, 1);\n\n% You need to return these values correctly\nerror_train = zeros(m, 1);\nerror_val = zeros(m, 1);\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Fill in this function to return training errors in \n% error_train and the cross validation errors in error_val. \n% i.e., error_train(i) and \n% error_val(i) should give you the errors\n% obtained after training on i examples.\n%\n% Note: You should evaluate the training error on the first i training\n% examples (i.e., X(1:i, :) and y(1:i)).\n%\n% For the cross-validation error, you should instead evaluate on\n% the _entire_ cross validation set (Xval and yval).\n%\n% Note: If you are using your cost function (linearRegCostFunction)\n% to compute the training and cross validation error, you should \n% call the function with the lambda argument set to 0. \n% Do note that you will still need to use lambda when running\n% the training to obtain the theta parameters.\n%\n% Hint: You can loop over the examples with the following:\n%\n% for i = 1:m\n% % Compute train/cross validation errors using training examples \n% % X(1:i, :) and y(1:i), storing the result in \n% % error_train(i) and error_val(i)\n% ....\n% \n% end\n%\n\n% ---------------------- Sample Solution ----------------------\n\nfor i = 1:m\n [theta] = trainLinearReg(X(1:i,:),y(1:i),lambda);\n h_train(1:i,i) = X(1:i,:)*theta;\n h_val(:,i) = Xval*theta;\n error_train(i) = (((h_train(1:i,i)-y(1:i))'*(h_train(1:i,i)-y(1:i)))/(2*i));\n error_val(i) = (((h_val(:,i)-yval)'*(h_val(:,i)-yval))/(2*length(yval)));\nend\n\n\n\n\n% -------------------------------------------------------------\n\n% =========================================================================\n\nend\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex5/ex5/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i Anyone knows how much it costs to host a web portal ?\n>\nWell, it depends on how many visitors you're expecting.\nThis can be anywhere from less than 10 bucks a month to a couple of $100. \nYou should checkout http://www.rackspace.com/ or perhaps Amazon EC2 \nif youre running something big..\n\nTo unsubscribe yourself from this mailing list, send an email to:\ngroupname-unsubscribe@egroups.com\n\n" }, { "path": "machine-learning-ex6/ex6/emailSample2.txt", "content": "Folks,\n \nmy first time posting - have a bit of Unix experience, but am new to Linux.\n\n \nJust got a new PC at home - Dell box with Windows XP. Added a second hard disk\nfor Linux. Partitioned the disk and have installed Suse 7.2 from CD, which went\nfine except it didn't pick up my monitor.\n \nI have a Dell branded E151FPp 15\" LCD flat panel monitor and a nVidia GeForce4\nTi4200 video card, both of which are probably too new to feature in Suse's default\nset. I downloaded a driver from the nVidia website and installed it using RPM.\nThen I ran Sax2 (as was recommended in some postings I found on the net), but\nit still doesn't feature my video card in the available list. What next?\n \nAnother problem. I have a Dell branded keyboard and if I hit Caps-Lock twice,\nthe whole machine crashes (in Linux, not Windows) - even the on/off switch is\ninactive, leaving me to reach for the power cable instead.\n \nIf anyone can help me in any way with these probs., I'd be really grateful -\nI've searched the 'net but have run out of ideas.\n \nOr should I be going for a different version of Linux such as RedHat? Opinions\nwelcome.\n \nThanks a lot,\nPeter\n\n-- \nIrish Linux Users' Group: ilug@linux.ie\nhttp://www.linux.ie/mailman/listinfo/ilug for (un)subscription information.\nList maintainer: listmaster@linux.ie\n\n\n" }, { "path": "machine-learning-ex6/ex6/ex6.m", "content": "%% Machine Learning Online Class\n% Exercise 6 | Support Vector Machines\n%\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% exercise. You will need to complete the following functions:\n%\n% gaussianKernel.m\n% dataset3Params.m\n% processEmail.m\n% emailFeatures.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% =============== Part 1: Loading and Visualizing Data ================\n% We start the exercise by first loading and visualizing the dataset. \n% The following code will load the dataset into your environment and plot\n% the data.\n%\n\nfprintf('Loading and Visualizing Data ...\\n')\n\n% Load from ex6data1: \n% You will have X, y in your environment\nload('ex6data1.mat');\n\n% Plot training data\nplotData(X, y);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ==================== Part 2: Training Linear SVM ====================\n% The following code will train a linear SVM on the dataset and plot the\n% decision boundary learned.\n%\n\n% Load from ex6data1: \n% You will have X, y in your environment\nload('ex6data1.mat');\n\nfprintf('\\nTraining Linear SVM ...\\n')\n\n% You should try to change the C value below and see how the decision\n% boundary varies (e.g., try C = 1000)\nC = 1;\nmodel = svmTrain(X, y, C, @linearKernel, 1e-3, 20);\nvisualizeBoundaryLinear(X, y, model);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =============== Part 3: Implementing Gaussian Kernel ===============\n% You will now implement the Gaussian kernel to use\n% with the SVM. You should complete the code in gaussianKernel.m\n%\nfprintf('\\nEvaluating the Gaussian Kernel ...\\n')\n\nx1 = [1 2 1]; x2 = [0 4 -1]; sigma = 2;\nsim = gaussianKernel(x1, x2, sigma);\n\nfprintf(['Gaussian Kernel between x1 = [1; 2; 1], x2 = [0; 4; -1], sigma = 0.5 :' ...\n '\\n\\t%f\\n(this value should be about 0.324652)\\n'], sim);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =============== Part 4: Visualizing Dataset 2 ================\n% The following code will load the next dataset into your environment and \n% plot the data. \n%\n\nfprintf('Loading and Visualizing Data ...\\n')\n\n% Load from ex6data2: \n% You will have X, y in your environment\nload('ex6data2.mat');\n\n% Plot training data\nplotData(X, y);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ========== Part 5: Training SVM with RBF Kernel (Dataset 2) ==========\n% After you have implemented the kernel, we can now use it to train the \n% SVM classifier.\n% \nfprintf('\\nTraining SVM with RBF Kernel (this may take 1 to 2 minutes) ...\\n');\n\n% Load from ex6data2: \n% You will have X, y in your environment\nload('ex6data2.mat');\n\n% SVM Parameters\nC = 1; sigma = 0.1;\n\n% We set the tolerance and max_passes lower here so that the code will run\n% faster. However, in practice, you will want to run the training to\n% convergence.\nmodel= svmTrain(X, y, C, @(x1, x2) gaussianKernel(x1, x2, sigma)); \nvisualizeBoundary(X, y, model);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =============== Part 6: Visualizing Dataset 3 ================\n% The following code will load the next dataset into your environment and \n% plot the data. \n%\n\nfprintf('Loading and Visualizing Data ...\\n')\n\n% Load from ex6data3: \n% You will have X, y in your environment\nload('ex6data3.mat');\n\n% Plot training data\nplotData(X, y);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ========== Part 7: Training SVM with RBF Kernel (Dataset 3) ==========\n\n% This is a different dataset that you can use to experiment with. Try\n% different values of C and sigma here.\n% \n\n% Load from ex6data3: \n% You will have X, y in your environment\nload('ex6data3.mat');\n\n% Try different SVM Parameters here\n[C, sigma] = dataset3Params(X, y, Xval, yval);\n\n% Train the SVM\nmodel= svmTrain(X, y, C, @(x1, x2) gaussianKernel(x1, x2, sigma));\nvisualizeBoundary(X, y, model);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n" }, { "path": "machine-learning-ex6/ex6/ex6_spam.m", "content": "%% Machine Learning Online Class\n% Exercise 6 | Spam Classification with SVMs\n%\n% Instructions\n% ------------\n% \n% This file contains code that helps you get started on the\n% exercise. You will need to complete the following functions:\n%\n% gaussianKernel.m\n% dataset3Params.m\n% processEmail.m\n% emailFeatures.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% ==================== Part 1: Email Preprocessing ====================\n% To use an SVM to classify emails into Spam v.s. Non-Spam, you first need\n% to convert each email into a vector of features. In this part, you will\n% implement the preprocessing steps for each email. You should\n% complete the code in processEmail.m to produce a word indices vector\n% for a given email.\n\nfprintf('\\nPreprocessing sample email (emailSample1.txt)\\n');\n\n% Extract Features\nfile_contents = readFile('emailSample1.txt');\nword_indices = processEmail(file_contents);\n\n% Print Stats\nfprintf('Word Indices: \\n');\nfprintf(' %d', word_indices);\nfprintf('\\n\\n');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ==================== Part 2: Feature Extraction ====================\n% Now, you will convert each email into a vector of features in R^n. \n% You should complete the code in emailFeatures.m to produce a feature\n% vector for a given email.\n\nfprintf('\\nExtracting features from sample email (emailSample1.txt)\\n');\n\n% Extract Features\nfile_contents = readFile('emailSample1.txt');\nword_indices = processEmail(file_contents);\nfeatures = emailFeatures(word_indices);\n\n% Print Stats\nfprintf('Length of feature vector: %d\\n', length(features));\nfprintf('Number of non-zero entries: %d\\n', sum(features > 0));\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =========== Part 3: Train Linear SVM for Spam Classification ========\n% In this section, you will train a linear classifier to determine if an\n% email is Spam or Not-Spam.\n\n% Load the Spam Email dataset\n% You will have X, y in your environment\nload('spamTrain.mat');\n\nfprintf('\\nTraining Linear SVM (Spam Classification)\\n')\nfprintf('(this may take 1 to 2 minutes) ...\\n')\n\nC = 0.1;\nmodel = svmTrain(X, y, C, @linearKernel);\n\np = svmPredict(model, X);\n\nfprintf('Training Accuracy: %f\\n', mean(double(p == y)) * 100);\n\n%% =================== Part 4: Test Spam Classification ================\n% After training the classifier, we can evaluate it on a test set. We have\n% included a test set in spamTest.mat\n\n% Load the test dataset\n% You will have Xtest, ytest in your environment\nload('spamTest.mat');\n\nfprintf('\\nEvaluating the trained Linear SVM on a test set ...\\n')\n\np = svmPredict(model, Xtest);\n\nfprintf('Test Accuracy: %f\\n', mean(double(p == ytest)) * 100);\npause;\n\n\n%% ================= Part 5: Top Predictors of Spam ====================\n% Since the model we are training is a linear SVM, we can inspect the\n% weights learned by the model to understand better how it is determining\n% whether an email is spam or not. The following code finds the words with\n% the highest weights in the classifier. Informally, the classifier\n% 'thinks' that these words are the most likely indicators of spam.\n%\n\n% Sort the weights and obtin the vocabulary list\n[weight, idx] = sort(model.w, 'descend');\nvocabList = getVocabList();\n\nfprintf('\\nTop predictors of spam: \\n');\nfor i = 1:15\n fprintf(' %-15s (%f) \\n', vocabList{idx(i)}, weight(i));\nend\n\nfprintf('\\n\\n');\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n%% =================== Part 6: Try Your Own Emails =====================\n% Now that you've trained the spam classifier, you can use it on your own\n% emails! In the starter code, we have included spamSample1.txt,\n% spamSample2.txt, emailSample1.txt and emailSample2.txt as examples. \n% The following code reads in one of these emails and then uses your \n% learned SVM classifier to determine whether the email is Spam or \n% Not Spam\n\n% Set the file to be read in (change this to spamSample2.txt,\n% emailSample1.txt or emailSample2.txt to see different predictions on\n% different emails types). Try your own emails as well!\nfilename = 'spamSample1.txt';\n\n% Read and predict\nfile_contents = readFile(filename);\nword_indices = processEmail(file_contents);\nx = emailFeatures(word_indices);\np = svmPredict(model, x);\n\nfprintf('\\nProcessed %s\\n\\nSpam Classification: %d\\n', filename, p);\nfprintf('(1 indicates spam, 0 indicates not spam)\\n\\n');\n\n" }, { "path": "machine-learning-ex6/ex6/gaussianKernel.m", "content": "function sim = gaussianKernel(x1, x2, sigma)\n%RBFKERNEL returns a radial basis function kernel between x1 and x2\n% sim = gaussianKernel(x1, x2) returns a gaussian kernel between x1 and x2\n% and returns the value in sim\n\n% Ensure that x1 and x2 are column vectors\nx1 = x1(:); x2 = x2(:);\n\n% You need to return the following variables correctly.\nsim = 0;\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Fill in this function to return the similarity between x1\n% and x2 computed using a Gaussian kernel with bandwidth\n% sigma\n%\n%\ntemp = (x1-x2)'*(x1-x2);\n\nsim = exp(-temp/(2*(sigma^2)));\n\n\n\n\n% =============================================================\n \nend\n" }, { "path": "machine-learning-ex6/ex6/getVocabList.m", "content": "function vocabList = getVocabList()\n%GETVOCABLIST reads the fixed vocabulary list in vocab.txt and returns a\n%cell array of the words\n% vocabList = GETVOCABLIST() reads the fixed vocabulary list in vocab.txt \n% and returns a cell array of the words in vocabList.\n\n\n%% Read the fixed vocabulary list\nfid = fopen('vocab.txt');\n\n% Store all dictionary words in cell array vocab{}\nn = 1899; % Total number of words in the dictionary\n\n% For ease of implementation, we use a struct to map the strings => integers\n% In practice, you'll want to use some form of hashmap\nvocabList = cell(n, 1);\nfor i = 1:n\n % Word Index (can ignore since it will be = i)\n fscanf(fid, '%d', 1);\n % Actual Word\n vocabList{i} = fscanf(fid, '%s', 1);\nend\nfclose(fid);\n\nend\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex6/ex6/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i 2\n % Output displays per step are commented out.\n %disp(sprintf('Word to stem: %s', b));\n x = step1ab(b, k, k0);\n %disp(sprintf('Steps 1A and B yield: %s', x{1}));\n x = step1c(x{1}, x{2}, k0);\n %disp(sprintf('Step 1C yields: %s', x{1}));\n x = step2(x{1}, x{2}, k0);\n %disp(sprintf('Step 2 yields: %s', x{1}));\n x = step3(x{1}, x{2}, k0);\n %disp(sprintf('Step 3 yields: %s', x{1}));\n x = step4(x{1}, x{2}, k0);\n %disp(sprintf('Step 4 yields: %s', x{1}));\n x = step5(x{1}, x{2}, k0);\n %disp(sprintf('Step 5 yields: %s', x{1}));\n stem = x{1};\nend\n\n% cons(j) is TRUE <=> b[j] is a consonant.\nfunction c = cons(i, b, k0)\nc = true;\nswitch(b(i))\n case {'a', 'e', 'i', 'o', 'u'}\n c = false;\n case 'y'\n if i == k0\n c = true;\n else\n c = ~cons(i - 1, b, k0);\n end\nend\n\n% mseq() measures the number of consonant sequences between k0 and j. If\n% c is a consonant sequence and v a vowel sequence, and <..> indicates\n% arbitrary presence,\n\n% gives 0\n% vc gives 1\n% vcvc gives 2\n% vcvcvc gives 3\n% ....\nfunction n = measure(b, k0)\nglobal j;\nn = 0;\ni = k0;\nwhile true\n if i > j\n return\n end\n if ~cons(i, b, k0)\n break;\n end\n i = i + 1;\nend\ni = i + 1;\nwhile true\n while true\n if i > j\n return\n end\n if cons(i, b, k0)\n break;\n end\n i = i + 1;\n end\n i = i + 1;\n n = n + 1;\n while true\n if i > j\n return\n end\n if ~cons(i, b, k0)\n break;\n end\n i = i + 1;\n end\n i = i + 1;\nend\n\n\n% vowelinstem() is TRUE <=> k0,...j contains a vowel\nfunction vis = vowelinstem(b, k0)\nglobal j;\nfor i = k0:j,\n if ~cons(i, b, k0)\n vis = true;\n return\n end\nend\nvis = false;\n\n%doublec(i) is TRUE <=> i,(i-1) contain a double consonant.\nfunction dc = doublec(i, b, k0)\nif i < k0+1\n dc = false;\n return\nend\nif b(i) ~= b(i-1)\n dc = false;\n return\nend\ndc = cons(i, b, k0);\n\n\n% cvc(j) is TRUE <=> j-2,j-1,j has the form consonant - vowel - consonant\n% and also if the second c is not w,x or y. this is used when trying to\n% restore an e at the end of a short word. e.g.\n%\n% cav(e), lov(e), hop(e), crim(e), but\n% snow, box, tray.\n\nfunction c1 = cvc(i, b, k0)\nif ((i < (k0+2)) || ~cons(i, b, k0) || cons(i-1, b, k0) || ~cons(i-2, b, k0))\n c1 = false;\nelse\n if (b(i) == 'w' || b(i) == 'x' || b(i) == 'y')\n c1 = false;\n return\n end\n c1 = true;\nend\n\n% ends(s) is TRUE <=> k0,...k ends with the string s.\nfunction s = ends(str, b, k)\nglobal j;\nif (str(length(str)) ~= b(k))\n s = false;\n return\nend % tiny speed-up\nif (length(str) > k)\n s = false;\n return\nend\nif strcmp(b(k-length(str)+1:k), str)\n s = true;\n j = k - length(str);\n return\nelse\n s = false;\nend\n\n% setto(s) sets (j+1),...k to the characters in the string s, readjusting\n% k accordingly.\n\nfunction so = setto(s, b, k)\nglobal j;\nfor i = j+1:(j+length(s))\n b(i) = s(i-j);\nend\nif k > j+length(s)\n b((j+length(s)+1):k) = '';\nend\nk = length(b);\nso = {b, k};\n\n% rs(s) is used further down.\n% [Note: possible null/value for r if rs is called]\nfunction r = rs(str, b, k, k0)\nr = {b, k};\nif measure(b, k0) > 0\n r = setto(str, b, k);\nend\n\n% step1ab() gets rid of plurals and -ed or -ing. e.g.\n\n% caresses -> caress\n% ponies -> poni\n% ties -> ti\n% caress -> caress\n% cats -> cat\n\n% feed -> feed\n% agreed -> agree\n% disabled -> disable\n\n% matting -> mat\n% mating -> mate\n% meeting -> meet\n% milling -> mill\n% messing -> mess\n\n% meetings -> meet\n\nfunction s1ab = step1ab(b, k, k0)\nglobal j;\nif b(k) == 's'\n if ends('sses', b, k)\n k = k-2;\n elseif ends('ies', b, k)\n retVal = setto('i', b, k);\n b = retVal{1};\n k = retVal{2};\n elseif (b(k-1) ~= 's')\n k = k-1;\n end\nend\nif ends('eed', b, k)\n if measure(b, k0) > 0;\n k = k-1;\n end\nelseif (ends('ed', b, k) || ends('ing', b, k)) && vowelinstem(b, k0)\n k = j;\n retVal = {b, k};\n if ends('at', b, k)\n retVal = setto('ate', b(k0:k), k);\n elseif ends('bl', b, k)\n retVal = setto('ble', b(k0:k), k);\n elseif ends('iz', b, k)\n retVal = setto('ize', b(k0:k), k);\n elseif doublec(k, b, k0)\n retVal = {b, k-1};\n if b(retVal{2}) == 'l' || b(retVal{2}) == 's' || ...\n b(retVal{2}) == 'z'\n retVal = {retVal{1}, retVal{2}+1};\n end\n elseif measure(b, k0) == 1 && cvc(k, b, k0)\n retVal = setto('e', b(k0:k), k);\n end\n k = retVal{2};\n b = retVal{1}(k0:k);\nend\nj = k;\ns1ab = {b(k0:k), k};\n\n% step1c() turns terminal y to i when there is another vowel in the stem.\nfunction s1c = step1c(b, k, k0)\nglobal j;\nif ends('y', b, k) && vowelinstem(b, k0)\n b(k) = 'i';\nend\nj = k;\ns1c = {b, k};\n\n% step2() maps double suffices to single ones. so -ization ( = -ize plus\n% -ation) maps to -ize etc. note that the string before the suffix must give\n% m() > 0.\nfunction s2 = step2(b, k, k0)\nglobal j;\ns2 = {b, k};\nswitch b(k-1)\n case {'a'}\n if ends('ational', b, k) s2 = rs('ate', b, k, k0);\n elseif ends('tional', b, k) s2 = rs('tion', b, k, k0); end;\n case {'c'}\n if ends('enci', b, k) s2 = rs('ence', b, k, k0);\n elseif ends('anci', b, k) s2 = rs('ance', b, k, k0); end;\n case {'e'}\n if ends('izer', b, k) s2 = rs('ize', b, k, k0); end;\n case {'l'}\n if ends('bli', b, k) s2 = rs('ble', b, k, k0);\n elseif ends('alli', b, k) s2 = rs('al', b, k, k0);\n elseif ends('entli', b, k) s2 = rs('ent', b, k, k0);\n elseif ends('eli', b, k) s2 = rs('e', b, k, k0);\n elseif ends('ousli', b, k) s2 = rs('ous', b, k, k0); end;\n case {'o'}\n if ends('ization', b, k) s2 = rs('ize', b, k, k0);\n elseif ends('ation', b, k) s2 = rs('ate', b, k, k0);\n elseif ends('ator', b, k) s2 = rs('ate', b, k, k0); end;\n case {'s'}\n if ends('alism', b, k) s2 = rs('al', b, k, k0);\n elseif ends('iveness', b, k) s2 = rs('ive', b, k, k0);\n elseif ends('fulness', b, k) s2 = rs('ful', b, k, k0);\n elseif ends('ousness', b, k) s2 = rs('ous', b, k, k0); end;\n case {'t'}\n if ends('aliti', b, k) s2 = rs('al', b, k, k0);\n elseif ends('iviti', b, k) s2 = rs('ive', b, k, k0);\n elseif ends('biliti', b, k) s2 = rs('ble', b, k, k0); end;\n case {'g'}\n if ends('logi', b, k) s2 = rs('log', b, k, k0); end;\nend\nj = s2{2};\n\n% step3() deals with -ic-, -full, -ness etc. similar strategy to step2.\nfunction s3 = step3(b, k, k0)\nglobal j;\ns3 = {b, k};\nswitch b(k)\n case {'e'}\n if ends('icate', b, k) s3 = rs('ic', b, k, k0);\n elseif ends('ative', b, k) s3 = rs('', b, k, k0);\n elseif ends('alize', b, k) s3 = rs('al', b, k, k0); end;\n case {'i'}\n if ends('iciti', b, k) s3 = rs('ic', b, k, k0); end;\n case {'l'}\n if ends('ical', b, k) s3 = rs('ic', b, k, k0);\n elseif ends('ful', b, k) s3 = rs('', b, k, k0); end;\n case {'s'}\n if ends('ness', b, k) s3 = rs('', b, k, k0); end;\nend\nj = s3{2};\n\n% step4() takes off -ant, -ence etc., in context vcvc.\nfunction s4 = step4(b, k, k0)\nglobal j;\nswitch b(k-1)\n case {'a'}\n if ends('al', b, k) end;\n case {'c'}\n if ends('ance', b, k)\n elseif ends('ence', b, k) end;\n case {'e'}\n if ends('er', b, k) end;\n case {'i'}\n if ends('ic', b, k) end;\n case {'l'}\n if ends('able', b, k)\n elseif ends('ible', b, k) end;\n case {'n'}\n if ends('ant', b, k)\n elseif ends('ement', b, k)\n elseif ends('ment', b, k)\n elseif ends('ent', b, k) end;\n case {'o'}\n if ends('ion', b, k)\n if j == 0\n elseif ~(strcmp(b(j),'s') || strcmp(b(j),'t'))\n j = k;\n end\n elseif ends('ou', b, k) end;\n case {'s'}\n if ends('ism', b, k) end;\n case {'t'}\n if ends('ate', b, k)\n elseif ends('iti', b, k) end;\n case {'u'}\n if ends('ous', b, k) end;\n case {'v'}\n if ends('ive', b, k) end;\n case {'z'}\n if ends('ize', b, k) end;\nend\nif measure(b, k0) > 1\n s4 = {b(k0:j), j};\nelse\n s4 = {b(k0:k), k};\nend\n\n% step5() removes a final -e if m() > 1, and changes -ll to -l if m() > 1.\nfunction s5 = step5(b, k, k0)\nglobal j;\nj = k;\nif b(k) == 'e'\n a = measure(b, k0);\n if (a > 1) || ((a == 1) && ~cvc(k-1, b, k0))\n k = k-1;\n end\nend\nif (b(k) == 'l') && doublec(k, b, k0) && (measure(b, k0) > 1)\n k = k-1;\nend\ns5 = {b(k0:k), k};\n" }, { "path": "machine-learning-ex6/ex6/processEmail.m", "content": "function word_indices = processEmail(email_contents)\n%PROCESSEMAIL preprocesses a the body of an email and\n%returns a list of word_indices \n% word_indices = PROCESSEMAIL(email_contents) preprocesses \n% the body of an email and returns a list of indices of the \n% words contained in the email. \n%\n\n% Load Vocabulary\nvocabList = getVocabList();\n\n% Init return value\nword_indices = [];\n\n% ========================== Preprocess Email ===========================\n\n% Find the Headers ( \\n\\n and remove )\n% Uncomment the following lines if you are working with raw emails with the\n% full headers\n\n% hdrstart = strfind(email_contents, ([char(10) char(10)]));\n% email_contents = email_contents(hdrstart(1):end);\n\n% Lower case\nemail_contents = lower(email_contents);\n\n% Strip all HTML\n% Looks for any expression that starts with < and ends with > and replace\n% and does not have any < or > in the tag it with a space\nemail_contents = regexprep(email_contents, '<[^<>]+>', ' ');\n\n% Handle Numbers\n% Look for one or more characters between 0-9\nemail_contents = regexprep(email_contents, '[0-9]+', 'number');\n\n% Handle URLS\n% Look for strings starting with http:// or https://\nemail_contents = regexprep(email_contents, ...\n '(http|https)://[^\\s]*', 'httpaddr');\n\n% Handle Email Addresses\n% Look for strings with @ in the middle\nemail_contents = regexprep(email_contents, '[^\\s]+@[^\\s]+', 'emailaddr');\n\n% Handle $ sign\nemail_contents = regexprep(email_contents, '[$]+', 'dollar');\n\n\n% ========================== Tokenize Email ===========================\n\n% Output the email to screen as well\nfprintf('\\n==== Processed Email ====\\n\\n');\n\n% Process file\nl = 0;\n\nwhile ~isempty(email_contents)\n\n % Tokenize and also get rid of any punctuation\n [str, email_contents] = ...\n strtok(email_contents, ...\n [' @$/#.-:&*+=[]?!(){},''\">_<;%' char(10) char(13)]);\n \n % Remove any non alphanumeric characters\n str = regexprep(str, '[^a-zA-Z0-9]', '');\n\n % Stem the word \n % (the porterStemmer sometimes has issues, so we use a try catch block)\n try str = porterStemmer(strtrim(str)); \n catch str = ''; continue;\n end;\n\n % Skip the word if it is too short\n if length(str) < 1\n continue;\n end\n\n % Look up the word in the dictionary and add to word_indices if\n % found\n % ====================== YOUR CODE HERE ======================\n % Instructions: Fill in this function to add the index of str to\n % word_indices if it is in the vocabulary. At this point\n % of the code, you have a stemmed word from the email in\n % the variable str. You should look up str in the\n % vocabulary list (vocabList). If a match exists, you\n % should add the index of the word to the word_indices\n % vector. Concretely, if str = 'action', then you should\n % look up the vocabulary list to find where in vocabList\n % 'action' appears. For example, if vocabList{18} =\n % 'action', then, you should add 18 to the word_indices \n % vector (e.g., word_indices = [word_indices ; 18]; ).\n % \n % Note: vocabList{idx} returns a the word with index idx in the\n % vocabulary list.\n % \n % Note: You can use strcmp(str1, str2) to compare two strings (str1 and\n % str2). It will return 1 only if the two strings are equivalent.\n %\n\n for i = 1:size(vocabList)\n if strcmp(str,vocabList{i}) == 1\n word_indices = [word_indices;i];\n end\n end\n\n\n\n\n\n\n\n\n % =============================================================\n\n\n % Print to screen, ensuring that the output lines are not too long\n if (l + length(str) + 1) > 78\n fprintf('\\n');\n l = 0;\n end\n fprintf('%s ', str);\n l = l + length(str) + 1;\n\nend\n\n% Print footer\nfprintf('\\n\\n=========================\\n');\n\nend\n" }, { "path": "machine-learning-ex6/ex6/readFile.m", "content": "function file_contents = readFile(filename)\n%READFILE reads a file and returns its entire contents \n% file_contents = READFILE(filename) reads a file and returns its entire\n% contents in file_contents\n%\n\n% Load File\nfid = fopen(filename);\nif fid\n file_contents = fscanf(fid, '%c', inf);\n fclose(fid);\nelse\n file_contents = '';\n fprintf('Unable to open %s\\n', filename);\nend\n\nend\n\n" }, { "path": "machine-learning-ex6/ex6/spamSample1.txt", "content": "Do You Want To Make $1000 Or More Per Week?\n\n \n\nIf you are a motivated and qualified individual - I \nwill personally demonstrate to you a system that will \nmake you $1,000 per week or more! This is NOT mlm.\n\n \n\nCall our 24 hour pre-recorded number to get the \ndetails. \n\n \n\n000-456-789\n\n \n\nI need people who want to make serious money. Make \nthe call and get the facts. \n\nInvest 2 minutes in yourself now!\n\n \n\n000-456-789\n\n \n\nLooking forward to your call and I will introduce you \nto people like yourself who\nare currently making $10,000 plus per week!\n\n \n\n000-456-789\n\n\n\n3484lJGv6-241lEaN9080lRmS6-271WxHo7524qiyT5-438rjUv5615hQcf0-662eiDB9057dMtVl72\n\n" }, { "path": "machine-learning-ex6/ex6/spamSample2.txt", "content": "Best Buy Viagra Generic Online\n\nViagra 100mg x 60 Pills $125, Free Pills & Reorder Discount, Top Selling 100% Quality & Satisfaction guaranteed!\n\nWe accept VISA, Master & E-Check Payments, 90000+ Satisfied Customers!\nhttp://medphysitcstech.ru\n\n\n" }, { "path": "machine-learning-ex6/ex6/submit.m", "content": "function submit()\n addpath('./lib');\n\n conf.assignmentSlug = 'support-vector-machines';\n conf.itemName = 'Support Vector Machines';\n conf.partArrays = { ...\n { ...\n '1', ...\n { 'gaussianKernel.m' }, ...\n 'Gaussian Kernel', ...\n }, ...\n { ...\n '2', ...\n { 'dataset3Params.m' }, ...\n 'Parameters (C, sigma) for Dataset 3', ...\n }, ...\n { ...\n '3', ...\n { 'processEmail.m' }, ...\n 'Email Preprocessing', ...\n }, ...\n { ...\n '4', ...\n { 'emailFeatures.m' }, ...\n 'Email Feature Extraction', ...\n }, ...\n };\n conf.output = @output;\n\n submitWithConfiguration(conf);\nend\n\nfunction out = output(partId, auxstring)\n % Random Test Cases\n x1 = sin(1:10)';\n x2 = cos(1:10)';\n ec = 'the quick brown fox jumped over the lazy dog';\n wi = 1 + abs(round(x1 * 1863));\n wi = [wi ; wi];\n if partId == '1'\n sim = gaussianKernel(x1, x2, 2);\n out = sprintf('%0.5f ', sim);\n elseif partId == '2'\n load('ex6data3.mat');\n [C, sigma] = dataset3Params(X, y, Xval, yval);\n out = sprintf('%0.5f ', C);\n out = [out sprintf('%0.5f ', sigma)];\n elseif partId == '3'\n word_indices = processEmail(ec);\n out = sprintf('%d ', word_indices);\n elseif partId == '4'\n x = emailFeatures(wi);\n out = sprintf('%d ', x);\n end \nend\n" }, { "path": "machine-learning-ex6/ex6/svmPredict.m", "content": "function pred = svmPredict(model, X)\n%SVMPREDICT returns a vector of predictions using a trained SVM model\n%(svmTrain). \n% pred = SVMPREDICT(model, X) returns a vector of predictions using a \n% trained SVM model (svmTrain). X is a mxn matrix where there each \n% example is a row. model is a svm model returned from svmTrain.\n% predictions pred is a m x 1 column of predictions of {0, 1} values.\n%\n\n% Check if we are getting a column vector, if so, then assume that we only\n% need to do prediction for a single example\nif (size(X, 2) == 1)\n % Examples should be in rows\n X = X';\nend\n\n% Dataset \nm = size(X, 1);\np = zeros(m, 1);\npred = zeros(m, 1);\n\nif strcmp(func2str(model.kernelFunction), 'linearKernel')\n % We can use the weights and bias directly if working with the \n % linear kernel\n p = X * model.w + model.b;\nelseif strfind(func2str(model.kernelFunction), 'gaussianKernel')\n % Vectorized RBF Kernel\n % This is equivalent to computing the kernel on every pair of examples\n X1 = sum(X.^2, 2);\n X2 = sum(model.X.^2, 2)';\n K = bsxfun(@plus, X1, bsxfun(@plus, X2, - 2 * X * model.X'));\n K = model.kernelFunction(1, 0) .^ K;\n K = bsxfun(@times, model.y', K);\n K = bsxfun(@times, model.alphas', K);\n p = sum(K, 2);\nelse\n % Other Non-linear kernel\n for i = 1:m\n prediction = 0;\n for j = 1:size(model.X, 1)\n prediction = prediction + ...\n model.alphas(j) * model.y(j) * ...\n model.kernelFunction(X(i,:)', model.X(j,:)');\n end\n p(i) = prediction + model.b;\n end\nend\n\n% Convert predictions into 0 / 1\npred(p >= 0) = 1;\npred(p < 0) = 0;\n\nend\n\n" }, { "path": "machine-learning-ex6/ex6/svmTrain.m", "content": "function [model] = svmTrain(X, Y, C, kernelFunction, ...\n tol, max_passes)\n%SVMTRAIN Trains an SVM classifier using a simplified version of the SMO \n%algorithm. \n% [model] = SVMTRAIN(X, Y, C, kernelFunction, tol, max_passes) trains an\n% SVM classifier and returns trained model. X is the matrix of training \n% examples. Each row is a training example, and the jth column holds the \n% jth feature. Y is a column matrix containing 1 for positive examples \n% and 0 for negative examples. C is the standard SVM regularization \n% parameter. tol is a tolerance value used for determining equality of \n% floating point numbers. max_passes controls the number of iterations\n% over the dataset (without changes to alpha) before the algorithm quits.\n%\n% Note: This is a simplified version of the SMO algorithm for training\n% SVMs. In practice, if you want to train an SVM classifier, we\n% recommend using an optimized package such as: \n%\n% LIBSVM (http://www.csie.ntu.edu.tw/~cjlin/libsvm/)\n% SVMLight (http://svmlight.joachims.org/)\n%\n%\n\nif ~exist('tol', 'var') || isempty(tol)\n tol = 1e-3;\nend\n\nif ~exist('max_passes', 'var') || isempty(max_passes)\n max_passes = 5;\nend\n\n% Data parameters\nm = size(X, 1);\nn = size(X, 2);\n\n% Map 0 to -1\nY(Y==0) = -1;\n\n% Variables\nalphas = zeros(m, 1);\nb = 0;\nE = zeros(m, 1);\npasses = 0;\neta = 0;\nL = 0;\nH = 0;\n\n% Pre-compute the Kernel Matrix since our dataset is small\n% (in practice, optimized SVM packages that handle large datasets\n% gracefully will _not_ do this)\n% \n% We have implemented optimized vectorized version of the Kernels here so\n% that the svm training will run faster.\nif strcmp(func2str(kernelFunction), 'linearKernel')\n % Vectorized computation for the Linear Kernel\n % This is equivalent to computing the kernel on every pair of examples\n K = X*X';\nelseif strfind(func2str(kernelFunction), 'gaussianKernel')\n % Vectorized RBF Kernel\n % This is equivalent to computing the kernel on every pair of examples\n X2 = sum(X.^2, 2);\n K = bsxfun(@plus, X2, bsxfun(@plus, X2', - 2 * (X * X')));\n K = kernelFunction(1, 0) .^ K;\nelse\n % Pre-compute the Kernel Matrix\n % The following can be slow due to the lack of vectorization\n K = zeros(m);\n for i = 1:m\n for j = i:m\n K(i,j) = kernelFunction(X(i,:)', X(j,:)');\n K(j,i) = K(i,j); %the matrix is symmetric\n end\n end\nend\n\n% Train\nfprintf('\\nTraining ...');\ndots = 12;\nwhile passes < max_passes,\n \n num_changed_alphas = 0;\n for i = 1:m,\n \n % Calculate Ei = f(x(i)) - y(i) using (2). \n % E(i) = b + sum (X(i, :) * (repmat(alphas.*Y,1,n).*X)') - Y(i);\n E(i) = b + sum (alphas.*Y.*K(:,i)) - Y(i);\n \n if ((Y(i)*E(i) < -tol && alphas(i) < C) || (Y(i)*E(i) > tol && alphas(i) > 0)),\n \n % In practice, there are many heuristics one can use to select\n % the i and j. In this simplified code, we select them randomly.\n j = ceil(m * rand());\n while j == i, % Make sure i \\neq j\n j = ceil(m * rand());\n end\n\n % Calculate Ej = f(x(j)) - y(j) using (2).\n E(j) = b + sum (alphas.*Y.*K(:,j)) - Y(j);\n\n % Save old alphas\n alpha_i_old = alphas(i);\n alpha_j_old = alphas(j);\n \n % Compute L and H by (10) or (11). \n if (Y(i) == Y(j)),\n L = max(0, alphas(j) + alphas(i) - C);\n H = min(C, alphas(j) + alphas(i));\n else\n L = max(0, alphas(j) - alphas(i));\n H = min(C, C + alphas(j) - alphas(i));\n end\n \n if (L == H),\n % continue to next i. \n continue;\n end\n\n % Compute eta by (14).\n eta = 2 * K(i,j) - K(i,i) - K(j,j);\n if (eta >= 0),\n % continue to next i. \n continue;\n end\n \n % Compute and clip new value for alpha j using (12) and (15).\n alphas(j) = alphas(j) - (Y(j) * (E(i) - E(j))) / eta;\n \n % Clip\n alphas(j) = min (H, alphas(j));\n alphas(j) = max (L, alphas(j));\n \n % Check if change in alpha is significant\n if (abs(alphas(j) - alpha_j_old) < tol),\n % continue to next i. \n % replace anyway\n alphas(j) = alpha_j_old;\n continue;\n end\n \n % Determine value for alpha i using (16). \n alphas(i) = alphas(i) + Y(i)*Y(j)*(alpha_j_old - alphas(j));\n \n % Compute b1 and b2 using (17) and (18) respectively. \n b1 = b - E(i) ...\n - Y(i) * (alphas(i) - alpha_i_old) * K(i,j)' ...\n - Y(j) * (alphas(j) - alpha_j_old) * K(i,j)';\n b2 = b - E(j) ...\n - Y(i) * (alphas(i) - alpha_i_old) * K(i,j)' ...\n - Y(j) * (alphas(j) - alpha_j_old) * K(j,j)';\n\n % Compute b by (19). \n if (0 < alphas(i) && alphas(i) < C),\n b = b1;\n elseif (0 < alphas(j) && alphas(j) < C),\n b = b2;\n else\n b = (b1+b2)/2;\n end\n\n num_changed_alphas = num_changed_alphas + 1;\n\n end\n \n end\n \n if (num_changed_alphas == 0),\n passes = passes + 1;\n else\n passes = 0;\n end\n\n fprintf('.');\n dots = dots + 1;\n if dots > 78\n dots = 0;\n fprintf('\\n');\n end\n if exist('OCTAVE_VERSION')\n fflush(stdout);\n end\nend\nfprintf(' Done! \\n\\n');\n\n% Save the model\nidx = alphas > 0;\nmodel.X= X(idx,:);\nmodel.y= Y(idx);\nmodel.kernelFunction = kernelFunction;\nmodel.b= b;\nmodel.alphas= alphas(idx);\nmodel.w = ((alphas.*Y)'*X)';\n\nend\n" }, { "path": "machine-learning-ex6/ex6/visualizeBoundary.m", "content": "function visualizeBoundary(X, y, model, varargin)\n%VISUALIZEBOUNDARY plots a non-linear decision boundary learned by the SVM\n% VISUALIZEBOUNDARYLINEAR(X, y, model) plots a non-linear decision \n% boundary learned by the SVM and overlays the data on it\n\n% Plot the training data on top of the boundary\nplotData(X, y)\n\n% Make classification predictions over a grid of values\nx1plot = linspace(min(X(:,1)), max(X(:,1)), 100)';\nx2plot = linspace(min(X(:,2)), max(X(:,2)), 100)';\n[X1, X2] = meshgrid(x1plot, x2plot);\nvals = zeros(size(X1));\nfor i = 1:size(X1, 2)\n this_X = [X1(:, i), X2(:, i)];\n vals(:, i) = svmPredict(model, this_X);\nend\n\n% Plot the SVM boundary\nhold on\ncontour(X1, X2, vals, [0 1], 'Color', 'b'); %ԭһ󣬷ΧӦ[0 1]\nhold off;\n\nend\n" }, { "path": "machine-learning-ex6/ex6/visualizeBoundaryLinear.m", "content": "function visualizeBoundaryLinear(X, y, model)\n%VISUALIZEBOUNDARYLINEAR plots a linear decision boundary learned by the\n%SVM\n% VISUALIZEBOUNDARYLINEAR(X, y, model) plots a linear decision boundary \n% learned by the SVM and overlays the data on it\n\nw = model.w;\nb = model.b;\nxp = linspace(min(X(:,1)), max(X(:,1)), 100);\nyp = - (w(1)*xp + b)/w(2);\nplotData(X, y);\nhold on;\nplot(xp, yp, '-b'); \nhold off\n\nend\n" }, { "path": "machine-learning-ex6/ex6/vocab.txt", "content": "1\taa\n2\tab\n3\tabil\n4\tabl\n5\tabout\n6\tabov\n7\tabsolut\n8\tabus\n9\tac\n10\taccept\n11\taccess\n12\taccord\n13\taccount\n14\tachiev\n15\tacquir\n16\tacross\n17\tact\n18\taction\n19\tactiv\n20\tactual\n21\tad\n22\tadam\n23\tadd\n24\taddit\n25\taddress\n26\tadministr\n27\tadult\n28\tadvanc\n29\tadvantag\n30\tadvertis\n31\tadvic\n32\tadvis\n33\tae\n34\taf\n35\taffect\n36\taffili\n37\tafford\n38\tafrica\n39\tafter\n40\tag\n41\tagain\n42\tagainst\n43\tagenc\n44\tagent\n45\tago\n46\tagre\n47\tagreement\n48\taid\n49\tair\n50\tal\n51\talb\n52\talign\n53\tall\n54\tallow\n55\talmost\n56\talon\n57\talong\n58\talreadi\n59\talsa\n60\talso\n61\taltern\n62\talthough\n63\talwai\n64\tam\n65\tamaz\n66\tamerica\n67\tamerican\n68\tamong\n69\tamount\n70\tamp\n71\tan\n72\tanalysi\n73\tanalyst\n74\tand\n75\tani\n76\tanim\n77\tannounc\n78\tannual\n79\tannuiti\n80\tanoth\n81\tanswer\n82\tanti\n83\tanumb\n84\tanybodi\n85\tanymor\n86\tanyon\n87\tanyth\n88\tanywai\n89\tanywher\n90\taol\n91\tap\n92\tapolog\n93\tapp\n94\tappar\n95\tappear\n96\tappl\n97\tappli\n98\tapplic\n99\tappreci\n100\tapproach\n101\tapprov\n102\tapt\n103\tar\n104\tarchiv\n105\tarea\n106\taren\n107\targument\n108\tarial\n109\tarm\n110\taround\n111\tarrai\n112\tarriv\n113\tart\n114\tarticl\n115\tartist\n116\tas\n117\tascii\n118\task\n119\tasset\n120\tassist\n121\tassoci\n122\tassum\n123\tassur\n124\tat\n125\tatol\n126\tattach\n127\tattack\n128\tattempt\n129\tattent\n130\tattornei\n131\tattract\n132\taudio\n133\taug\n134\taugust\n135\tauthor\n136\tauto\n137\tautom\n138\tautomat\n139\tavail\n140\taverag\n141\tavoid\n142\tawai\n143\tawar\n144\taward\n145\tba\n146\tbabi\n147\tback\n148\tbackground\n149\tbackup\n150\tbad\n151\tbalanc\n152\tban\n153\tbank\n154\tbar\n155\tbase\n156\tbasenumb\n157\tbasi\n158\tbasic\n159\tbb\n160\tbc\n161\tbd\n162\tbe\n163\tbeat\n164\tbeberg\n165\tbecaus\n166\tbecom\n167\tbeen\n168\tbefor\n169\tbegin\n170\tbehalf\n171\tbehavior\n172\tbehind\n173\tbeliev\n174\tbelow\n175\tbenefit\n176\tbest\n177\tbeta\n178\tbetter\n179\tbetween\n180\tbf\n181\tbig\n182\tbill\n183\tbillion\n184\tbin\n185\tbinari\n186\tbit\n187\tblack\n188\tblank\n189\tblock\n190\tblog\n191\tblood\n192\tblue\n193\tbnumber\n194\tboard\n195\tbodi\n196\tboi\n197\tbonu\n198\tbook\n199\tboot\n200\tborder\n201\tboss\n202\tboston\n203\tbotan\n204\tboth\n205\tbottl\n206\tbottom\n207\tboundari\n208\tbox\n209\tbrain\n210\tbrand\n211\tbreak\n212\tbrian\n213\tbring\n214\tbroadcast\n215\tbroker\n216\tbrowser\n217\tbug\n218\tbui\n219\tbuild\n220\tbuilt\n221\tbulk\n222\tburn\n223\tbush\n224\tbusi\n225\tbut\n226\tbutton\n227\tby\n228\tbyte\n229\tca\n230\tcabl\n231\tcach\n232\tcalcul\n233\tcalifornia\n234\tcall\n235\tcame\n236\tcamera\n237\tcampaign\n238\tcan\n239\tcanada\n240\tcannot\n241\tcanon\n242\tcapabl\n243\tcapillari\n244\tcapit\n245\tcar\n246\tcard\n247\tcare\n248\tcareer\n249\tcarri\n250\tcartridg\n251\tcase\n252\tcash\n253\tcat\n254\tcatch\n255\tcategori\n256\tcaus\n257\tcb\n258\tcc\n259\tcd\n260\tce\n261\tcell\n262\tcent\n263\tcenter\n264\tcentral\n265\tcenturi\n266\tceo\n267\tcertain\n268\tcertainli\n269\tcf\n270\tchalleng\n271\tchanc\n272\tchang\n273\tchannel\n274\tchar\n275\tcharact\n276\tcharg\n277\tcharset\n278\tchat\n279\tcheap\n280\tcheck\n281\tcheer\n282\tchief\n283\tchildren\n284\tchina\n285\tchip\n286\tchoic\n287\tchoos\n288\tchri\n289\tciti\n290\tcitizen\n291\tcivil\n292\tclaim\n293\tclass\n294\tclassifi\n295\tclean\n296\tclear\n297\tclearli\n298\tclick\n299\tclient\n300\tclose\n301\tclue\n302\tcnet\n303\tcnumber\n304\tco\n305\tcode\n306\tcollect\n307\tcolleg\n308\tcolor\n309\tcom\n310\tcombin\n311\tcome\n312\tcomfort\n313\tcommand\n314\tcomment\n315\tcommentari\n316\tcommerci\n317\tcommiss\n318\tcommit\n319\tcommon\n320\tcommun\n321\tcompani\n322\tcompar\n323\tcomparison\n324\tcompat\n325\tcompet\n326\tcompetit\n327\tcompil\n328\tcomplet\n329\tcomprehens\n330\tcomput\n331\tconcentr\n332\tconcept\n333\tconcern\n334\tcondit\n335\tconf\n336\tconfer\n337\tconfid\n338\tconfidenti\n339\tconfig\n340\tconfigur\n341\tconfirm\n342\tconflict\n343\tconfus\n344\tcongress\n345\tconnect\n346\tconsid\n347\tconsolid\n348\tconstitut\n349\tconstruct\n350\tconsult\n351\tconsum\n352\tcontact\n353\tcontain\n354\tcontent\n355\tcontinu\n356\tcontract\n357\tcontribut\n358\tcontrol\n359\tconveni\n360\tconvers\n361\tconvert\n362\tcool\n363\tcooper\n364\tcopi\n365\tcopyright\n366\tcore\n367\tcorpor\n368\tcorrect\n369\tcorrespond\n370\tcost\n371\tcould\n372\tcouldn\n373\tcount\n374\tcountri\n375\tcoupl\n376\tcours\n377\tcourt\n378\tcover\n379\tcoverag\n380\tcrash\n381\tcreat\n382\tcreativ\n383\tcredit\n384\tcritic\n385\tcross\n386\tcultur\n387\tcurrent\n388\tcustom\n389\tcut\n390\tcv\n391\tda\n392\tdagga\n393\tdai\n394\tdaili\n395\tdan\n396\tdanger\n397\tdark\n398\tdata\n399\tdatabas\n400\tdatapow\n401\tdate\n402\tdave\n403\tdavid\n404\tdc\n405\tde\n406\tdead\n407\tdeal\n408\tdear\n409\tdeath\n410\tdebt\n411\tdecad\n412\tdecid\n413\tdecis\n414\tdeclar\n415\tdeclin\n416\tdecor\n417\tdefault\n418\tdefend\n419\tdefens\n420\tdefin\n421\tdefinit\n422\tdegre\n423\tdelai\n424\tdelet\n425\tdeliv\n426\tdeliveri\n427\tdell\n428\tdemand\n429\tdemocrat\n430\tdepart\n431\tdepend\n432\tdeposit\n433\tdescrib\n434\tdescript\n435\tdeserv\n436\tdesign\n437\tdesir\n438\tdesktop\n439\tdespit\n440\tdetail\n441\tdetect\n442\tdetermin\n443\tdev\n444\tdevel\n445\tdevelop\n446\tdevic\n447\tdi\n448\tdial\n449\tdid\n450\tdidn\n451\tdiet\n452\tdiffer\n453\tdifficult\n454\tdigit\n455\tdirect\n456\tdirectli\n457\tdirector\n458\tdirectori\n459\tdisabl\n460\tdiscount\n461\tdiscov\n462\tdiscoveri\n463\tdiscuss\n464\tdisk\n465\tdisplai\n466\tdisposit\n467\tdistanc\n468\tdistribut\n469\tdn\n470\tdnumber\n471\tdo\n472\tdoc\n473\tdocument\n474\tdoe\n475\tdoer\n476\tdoesn\n477\tdollar\n478\tdollarac\n479\tdollarnumb\n480\tdomain\n481\tdon\n482\tdone\n483\tdont\n484\tdoubl\n485\tdoubt\n486\tdown\n487\tdownload\n488\tdr\n489\tdraw\n490\tdream\n491\tdrive\n492\tdriver\n493\tdrop\n494\tdrug\n495\tdue\n496\tdure\n497\tdvd\n498\tdw\n499\tdynam\n500\tea\n501\teach\n502\tearli\n503\tearlier\n504\tearn\n505\tearth\n506\teasi\n507\teasier\n508\teasili\n509\teat\n510\teb\n511\tebai\n512\tec\n513\techo\n514\teconom\n515\teconomi\n516\ted\n517\tedg\n518\tedit\n519\teditor\n520\teduc\n521\teff\n522\teffect\n523\teffici\n524\teffort\n525\teither\n526\tel\n527\telectron\n528\telimin\n529\tels\n530\temail\n531\temailaddr\n532\temerg\n533\tempir\n534\temploy\n535\temploye\n536\ten\n537\tenabl\n538\tencod\n539\tencourag\n540\tend\n541\tenemi\n542\tenenkio\n543\tenergi\n544\tengin\n545\tenglish\n546\tenhanc\n547\tenjoi\n548\tenough\n549\tensur\n550\tenter\n551\tenterpris\n552\tentertain\n553\tentir\n554\tentri\n555\tenumb\n556\tenviron\n557\tequal\n558\tequip\n559\tequival\n560\terror\n561\tespeci\n562\tessenti\n563\testablish\n564\testat\n565\testim\n566\tet\n567\tetc\n568\teuro\n569\teurop\n570\teuropean\n571\teven\n572\tevent\n573\teventu\n574\tever\n575\teveri\n576\teveryon\n577\teveryth\n578\tevid\n579\tevil\n580\texactli\n581\texampl\n582\texcel\n583\texcept\n584\texchang\n585\texcit\n586\texclus\n587\texecut\n588\texercis\n589\texist\n590\texmh\n591\texpand\n592\texpect\n593\texpens\n594\texperi\n595\texpert\n596\texpir\n597\texplain\n598\texplor\n599\texpress\n600\textend\n601\textens\n602\textra\n603\textract\n604\textrem\n605\tey\n606\tfa\n607\tface\n608\tfact\n609\tfactor\n610\tfail\n611\tfair\n612\tfall\n613\tfals\n614\tfamili\n615\tfaq\n616\tfar\n617\tfast\n618\tfaster\n619\tfastest\n620\tfat\n621\tfather\n622\tfavorit\n623\tfax\n624\tfb\n625\tfd\n626\tfeatur\n627\tfeder\n628\tfee\n629\tfeed\n630\tfeedback\n631\tfeel\n632\tfemal\n633\tfew\n634\tffffff\n635\tffnumber\n636\tfield\n637\tfight\n638\tfigur\n639\tfile\n640\tfill\n641\tfilm\n642\tfilter\n643\tfinal\n644\tfinanc\n645\tfinanci\n646\tfind\n647\tfine\n648\tfinish\n649\tfire\n650\tfirewal\n651\tfirm\n652\tfirst\n653\tfit\n654\tfive\n655\tfix\n656\tflag\n657\tflash\n658\tflow\n659\tfnumber\n660\tfocu\n661\tfolder\n662\tfolk\n663\tfollow\n664\tfont\n665\tfood\n666\tfor\n667\tforc\n668\tforeign\n669\tforev\n670\tforget\n671\tfork\n672\tform\n673\tformat\n674\tformer\n675\tfortun\n676\tforward\n677\tfound\n678\tfoundat\n679\tfour\n680\tfranc\n681\tfree\n682\tfreedom\n683\tfrench\n684\tfreshrpm\n685\tfri\n686\tfridai\n687\tfriend\n688\tfrom\n689\tfront\n690\tftoc\n691\tftp\n692\tfull\n693\tfulli\n694\tfun\n695\tfunction\n696\tfund\n697\tfurther\n698\tfutur\n699\tga\n700\tgain\n701\tgame\n702\tgari\n703\tgarrigu\n704\tgave\n705\tgcc\n706\tgeek\n707\tgener\n708\tget\n709\tgif\n710\tgift\n711\tgirl\n712\tgive\n713\tgiven\n714\tglobal\n715\tgnome\n716\tgnu\n717\tgnupg\n718\tgo\n719\tgoal\n720\tgod\n721\tgoe\n722\tgold\n723\tgone\n724\tgood\n725\tgoogl\n726\tgot\n727\tgovern\n728\tgpl\n729\tgrand\n730\tgrant\n731\tgraphic\n732\tgreat\n733\tgreater\n734\tground\n735\tgroup\n736\tgrow\n737\tgrowth\n738\tgt\n739\tguarante\n740\tguess\n741\tgui\n742\tguid\n743\tha\n744\thack\n745\thad\n746\thalf\n747\tham\n748\thand\n749\thandl\n750\thappen\n751\thappi\n752\thard\n753\thardwar\n754\that\n755\thate\n756\thave\n757\thaven\n758\the\n759\thead\n760\theader\n761\theadlin\n762\thealth\n763\thear\n764\theard\n765\theart\n766\theaven\n767\thei\n768\theight\n769\theld\n770\thello\n771\thelp\n772\thelvetica\n773\ther\n774\therba\n775\there\n776\thermio\n777\thettinga\n778\thi\n779\thigh\n780\thigher\n781\thighli\n782\thighlight\n783\thim\n784\thistori\n785\thit\n786\thold\n787\thome\n788\thonor\n789\thope\n790\thost\n791\thot\n792\thour\n793\thous\n794\thow\n795\thowev\n796\thp\n797\thtml\n798\thttp\n799\thttpaddr\n800\thuge\n801\thuman\n802\thundr\n803\tibm\n804\tid\n805\tidea\n806\tident\n807\tidentifi\n808\tidnumb\n809\tie\n810\tif\n811\tignor\n812\tii\n813\tiii\n814\tiiiiiiihnumberjnumberhnumberjnumberhnumb\n815\tilleg\n816\tim\n817\timag\n818\timagin\n819\timmedi\n820\timpact\n821\timplement\n822\timport\n823\timpress\n824\timprov\n825\tin\n826\tinc\n827\tinclud\n828\tincom\n829\tincreas\n830\tincred\n831\tinde\n832\tindepend\n833\tindex\n834\tindia\n835\tindian\n836\tindic\n837\tindividu\n838\tindustri\n839\tinfo\n840\tinform\n841\tiniti\n842\tinlin\n843\tinnov\n844\tinput\n845\tinsert\n846\tinsid\n847\tinstal\n848\tinstanc\n849\tinstant\n850\tinstead\n851\tinstitut\n852\tinstruct\n853\tinsur\n854\tint\n855\tintegr\n856\tintel\n857\tintellig\n858\tintend\n859\tinteract\n860\tinterest\n861\tinterfac\n862\tintern\n863\tinternet\n864\tinterview\n865\tinto\n866\tintro\n867\tintroduc\n868\tinumb\n869\tinvest\n870\tinvestig\n871\tinvestor\n872\tinvok\n873\tinvolv\n874\tip\n875\tireland\n876\tirish\n877\tis\n878\tisland\n879\tisn\n880\tiso\n881\tisp\n882\tissu\n883\tit\n884\titem\n885\titself\n886\tjabber\n887\tjame\n888\tjava\n889\tjim\n890\tjnumberiiiiiiihepihepihf\n891\tjob\n892\tjoe\n893\tjohn\n894\tjoin\n895\tjournal\n896\tjudg\n897\tjudgment\n898\tjul\n899\tjuli\n900\tjump\n901\tjune\n902\tjust\n903\tjustin\n904\tkeep\n905\tkei\n906\tkept\n907\tkernel\n908\tkevin\n909\tkeyboard\n910\tkid\n911\tkill\n912\tkind\n913\tking\n914\tkingdom\n915\tknew\n916\tknow\n917\tknowledg\n918\tknown\n919\tla\n920\tlack\n921\tland\n922\tlanguag\n923\tlaptop\n924\tlarg\n925\tlarger\n926\tlargest\n927\tlaser\n928\tlast\n929\tlate\n930\tlater\n931\tlatest\n932\tlaunch\n933\tlaw\n934\tlawrenc\n935\tle\n936\tlead\n937\tleader\n938\tlearn\n939\tleast\n940\tleav\n941\tleft\n942\tlegal\n943\tlender\n944\tlength\n945\tless\n946\tlesson\n947\tlet\n948\tletter\n949\tlevel\n950\tlib\n951\tlibrari\n952\tlicens\n953\tlife\n954\tlifetim\n955\tlight\n956\tlike\n957\tlimit\n958\tline\n959\tlink\n960\tlinux\n961\tlist\n962\tlisten\n963\tlittl\n964\tlive\n965\tll\n966\tlo\n967\tload\n968\tloan\n969\tlocal\n970\tlocat\n971\tlock\n972\tlockergnom\n973\tlog\n974\tlong\n975\tlonger\n976\tlook\n977\tlose\n978\tloss\n979\tlost\n980\tlot\n981\tlove\n982\tlow\n983\tlower\n984\tlowest\n985\tlt\n986\tma\n987\tmac\n988\tmachin\n989\tmade\n990\tmagazin\n991\tmai\n992\tmail\n993\tmailer\n994\tmain\n995\tmaintain\n996\tmajor\n997\tmake\n998\tmaker\n999\tmale\n1000\tman\n1001\tmanag\n1002\tmani\n1003\tmanual\n1004\tmanufactur\n1005\tmap\n1006\tmarch\n1007\tmargin\n1008\tmark\n1009\tmarket\n1010\tmarshal\n1011\tmass\n1012\tmaster\n1013\tmatch\n1014\tmateri\n1015\tmatter\n1016\tmatthia\n1017\tmayb\n1018\tme\n1019\tmean\n1020\tmeasur\n1021\tmechan\n1022\tmedia\n1023\tmedic\n1024\tmeet\n1025\tmember\n1026\tmembership\n1027\tmemori\n1028\tmen\n1029\tmention\n1030\tmenu\n1031\tmerchant\n1032\tmessag\n1033\tmethod\n1034\tmh\n1035\tmichael\n1036\tmicrosoft\n1037\tmiddl\n1038\tmight\n1039\tmike\n1040\tmile\n1041\tmilitari\n1042\tmillion\n1043\tmime\n1044\tmind\n1045\tmine\n1046\tmini\n1047\tminimum\n1048\tminut\n1049\tmiss\n1050\tmistak\n1051\tmobil\n1052\tmode\n1053\tmodel\n1054\tmodem\n1055\tmodifi\n1056\tmodul\n1057\tmoment\n1058\tmon\n1059\tmondai\n1060\tmonei\n1061\tmonitor\n1062\tmonth\n1063\tmonthli\n1064\tmore\n1065\tmorn\n1066\tmortgag\n1067\tmost\n1068\tmostli\n1069\tmother\n1070\tmotiv\n1071\tmove\n1072\tmovi\n1073\tmpnumber\n1074\tmr\n1075\tms\n1076\tmsg\n1077\tmuch\n1078\tmulti\n1079\tmultipart\n1080\tmultipl\n1081\tmurphi\n1082\tmusic\n1083\tmust\n1084\tmy\n1085\tmyself\n1086\tname\n1087\tnation\n1088\tnatur\n1089\tnbsp\n1090\tnear\n1091\tnearli\n1092\tnecessari\n1093\tneed\n1094\tneg\n1095\tnet\n1096\tnetscap\n1097\tnetwork\n1098\tnever\n1099\tnew\n1100\tnewslett\n1101\tnext\n1102\tnextpart\n1103\tnice\n1104\tnigeria\n1105\tnight\n1106\tno\n1107\tnobodi\n1108\tnon\n1109\tnone\n1110\tnor\n1111\tnormal\n1112\tnorth\n1113\tnot\n1114\tnote\n1115\tnoth\n1116\tnotic\n1117\tnow\n1118\tnt\n1119\tnull\n1120\tnumber\n1121\tnumbera\n1122\tnumberam\n1123\tnumberanumb\n1124\tnumberb\n1125\tnumberbit\n1126\tnumberc\n1127\tnumbercb\n1128\tnumbercbr\n1129\tnumbercfont\n1130\tnumbercli\n1131\tnumbercnumb\n1132\tnumbercp\n1133\tnumberctd\n1134\tnumberd\n1135\tnumberdari\n1136\tnumberdnumb\n1137\tnumberenumb\n1138\tnumberf\n1139\tnumberfb\n1140\tnumberff\n1141\tnumberffont\n1142\tnumberfp\n1143\tnumberftd\n1144\tnumberk\n1145\tnumberm\n1146\tnumbermb\n1147\tnumberp\n1148\tnumberpd\n1149\tnumberpm\n1150\tnumberpx\n1151\tnumberst\n1152\tnumberth\n1153\tnumbertnumb\n1154\tnumberx\n1155\tobject\n1156\toblig\n1157\tobtain\n1158\tobvious\n1159\toccur\n1160\toct\n1161\toctob\n1162\tof\n1163\toff\n1164\toffer\n1165\toffic\n1166\toffici\n1167\toften\n1168\toh\n1169\tok\n1170\told\n1171\ton\n1172\tonc\n1173\tonli\n1174\tonlin\n1175\topen\n1176\toper\n1177\topinion\n1178\topportun\n1179\topt\n1180\toptim\n1181\toption\n1182\tor\n1183\torder\n1184\torg\n1185\torgan\n1186\torigin\n1187\tos\n1188\tosdn\n1189\tother\n1190\totherwis\n1191\tour\n1192\tout\n1193\toutlook\n1194\toutput\n1195\toutsid\n1196\tover\n1197\town\n1198\towner\n1199\toz\n1200\tpacif\n1201\tpack\n1202\tpackag\n1203\tpage\n1204\tpai\n1205\tpaid\n1206\tpain\n1207\tpalm\n1208\tpanel\n1209\tpaper\n1210\tparagraph\n1211\tparent\n1212\tpart\n1213\tparti\n1214\tparticip\n1215\tparticular\n1216\tparticularli\n1217\tpartit\n1218\tpartner\n1219\tpass\n1220\tpassword\n1221\tpast\n1222\tpatch\n1223\tpatent\n1224\tpath\n1225\tpattern\n1226\tpaul\n1227\tpayment\n1228\tpc\n1229\tpeac\n1230\tpeopl\n1231\tper\n1232\tpercent\n1233\tpercentag\n1234\tperfect\n1235\tperfectli\n1236\tperform\n1237\tperhap\n1238\tperiod\n1239\tperl\n1240\tperman\n1241\tpermiss\n1242\tperson\n1243\tpgp\n1244\tphone\n1245\tphoto\n1246\tphp\n1247\tphrase\n1248\tphysic\n1249\tpick\n1250\tpictur\n1251\tpiec\n1252\tpiiiiiiii\n1253\tpipe\n1254\tpjnumber\n1255\tplace\n1256\tplai\n1257\tplain\n1258\tplan\n1259\tplanet\n1260\tplant\n1261\tplanta\n1262\tplatform\n1263\tplayer\n1264\tpleas\n1265\tplu\n1266\tplug\n1267\tpm\n1268\tpocket\n1269\tpoint\n1270\tpolic\n1271\tpolici\n1272\tpolit\n1273\tpoor\n1274\tpop\n1275\tpopul\n1276\tpopular\n1277\tport\n1278\tposit\n1279\tpossibl\n1280\tpost\n1281\tpotenti\n1282\tpound\n1283\tpowel\n1284\tpower\n1285\tpowershot\n1286\tpractic\n1287\tpre\n1288\tpredict\n1289\tprefer\n1290\tpremium\n1291\tprepar\n1292\tpresent\n1293\tpresid\n1294\tpress\n1295\tpretti\n1296\tprevent\n1297\tpreviou\n1298\tprevious\n1299\tprice\n1300\tprincipl\n1301\tprint\n1302\tprintabl\n1303\tprinter\n1304\tprivaci\n1305\tprivat\n1306\tprize\n1307\tpro\n1308\tprobabl\n1309\tproblem\n1310\tprocedur\n1311\tprocess\n1312\tprocessor\n1313\tprocmail\n1314\tproduc\n1315\tproduct\n1316\tprofession\n1317\tprofil\n1318\tprofit\n1319\tprogram\n1320\tprogramm\n1321\tprogress\n1322\tproject\n1323\tpromis\n1324\tpromot\n1325\tprompt\n1326\tproperti\n1327\tpropos\n1328\tproprietari\n1329\tprospect\n1330\tprotect\n1331\tprotocol\n1332\tprove\n1333\tproven\n1334\tprovid\n1335\tproxi\n1336\tpub\n1337\tpublic\n1338\tpublish\n1339\tpudg\n1340\tpull\n1341\tpurchas\n1342\tpurpos\n1343\tput\n1344\tpython\n1345\tqnumber\n1346\tqualifi\n1347\tqualiti\n1348\tquarter\n1349\tquestion\n1350\tquick\n1351\tquickli\n1352\tquit\n1353\tquot\n1354\tradio\n1355\tragga\n1356\trais\n1357\trandom\n1358\trang\n1359\trate\n1360\trather\n1361\tratio\n1362\trazor\n1363\trazornumb\n1364\tre\n1365\treach\n1366\tread\n1367\treader\n1368\treadi\n1369\treal\n1370\trealiz\n1371\trealli\n1372\treason\n1373\treceiv\n1374\trecent\n1375\trecipi\n1376\trecommend\n1377\trecord\n1378\tred\n1379\tredhat\n1380\treduc\n1381\trefer\n1382\trefin\n1383\treg\n1384\tregard\n1385\tregion\n1386\tregist\n1387\tregul\n1388\tregular\n1389\trel\n1390\trelat\n1391\trelationship\n1392\treleas\n1393\trelev\n1394\treliabl\n1395\tremain\n1396\trememb\n1397\tremot\n1398\tremov\n1399\treplac\n1400\trepli\n1401\treport\n1402\trepositori\n1403\trepres\n1404\trepubl\n1405\trequest\n1406\trequir\n1407\tresearch\n1408\treserv\n1409\tresid\n1410\tresourc\n1411\trespect\n1412\trespond\n1413\trespons\n1414\trest\n1415\tresult\n1416\tretail\n1417\treturn\n1418\treveal\n1419\trevenu\n1420\trevers\n1421\treview\n1422\trevok\n1423\trh\n1424\trich\n1425\tright\n1426\trisk\n1427\troad\n1428\trobert\n1429\trock\n1430\trole\n1431\troll\n1432\trom\n1433\troman\n1434\troom\n1435\troot\n1436\tround\n1437\trpm\n1438\trss\n1439\trule\n1440\trun\n1441\tsa\n1442\tsafe\n1443\tsai\n1444\tsaid\n1445\tsale\n1446\tsame\n1447\tsampl\n1448\tsan\n1449\tsaou\n1450\tsat\n1451\tsatellit\n1452\tsave\n1453\tsaw\n1454\tscan\n1455\tschedul\n1456\tschool\n1457\tscienc\n1458\tscore\n1459\tscreen\n1460\tscript\n1461\tse\n1462\tsearch\n1463\tseason\n1464\tsecond\n1465\tsecret\n1466\tsection\n1467\tsecur\n1468\tsee\n1469\tseed\n1470\tseek\n1471\tseem\n1472\tseen\n1473\tselect\n1474\tself\n1475\tsell\n1476\tseminar\n1477\tsend\n1478\tsender\n1479\tsendmail\n1480\tsenior\n1481\tsens\n1482\tsensit\n1483\tsent\n1484\tsep\n1485\tsepar\n1486\tseptemb\n1487\tsequenc\n1488\tseri\n1489\tserif\n1490\tseriou\n1491\tserv\n1492\tserver\n1493\tservic\n1494\tset\n1495\tsetup\n1496\tseven\n1497\tseventh\n1498\tsever\n1499\tsex\n1500\tsexual\n1501\tsf\n1502\tshape\n1503\tshare\n1504\tshe\n1505\tshell\n1506\tship\n1507\tshop\n1508\tshort\n1509\tshot\n1510\tshould\n1511\tshow\n1512\tside\n1513\tsign\n1514\tsignatur\n1515\tsignific\n1516\tsimilar\n1517\tsimpl\n1518\tsimpli\n1519\tsinc\n1520\tsincer\n1521\tsingl\n1522\tsit\n1523\tsite\n1524\tsituat\n1525\tsix\n1526\tsize\n1527\tskeptic\n1528\tskill\n1529\tskin\n1530\tskip\n1531\tsleep\n1532\tslow\n1533\tsmall\n1534\tsmart\n1535\tsmoke\n1536\tsmtp\n1537\tsnumber\n1538\tso\n1539\tsocial\n1540\tsocieti\n1541\tsoftwar\n1542\tsold\n1543\tsolut\n1544\tsolv\n1545\tsome\n1546\tsomeon\n1547\tsometh\n1548\tsometim\n1549\tson\n1550\tsong\n1551\tsoni\n1552\tsoon\n1553\tsorri\n1554\tsort\n1555\tsound\n1556\tsourc\n1557\tsouth\n1558\tspace\n1559\tspain\n1560\tspam\n1561\tspamassassin\n1562\tspamd\n1563\tspammer\n1564\tspeak\n1565\tspec\n1566\tspecial\n1567\tspecif\n1568\tspecifi\n1569\tspeech\n1570\tspeed\n1571\tspend\n1572\tsponsor\n1573\tsport\n1574\tspot\n1575\tsrc\n1576\tssh\n1577\tst\n1578\tstabl\n1579\tstaff\n1580\tstai\n1581\tstand\n1582\tstandard\n1583\tstar\n1584\tstart\n1585\tstate\n1586\tstatement\n1587\tstatu\n1588\tstep\n1589\tsteve\n1590\tstill\n1591\tstock\n1592\tstop\n1593\tstorag\n1594\tstore\n1595\tstori\n1596\tstrategi\n1597\tstream\n1598\tstreet\n1599\tstring\n1600\tstrip\n1601\tstrong\n1602\tstructur\n1603\tstudi\n1604\tstuff\n1605\tstupid\n1606\tstyle\n1607\tsubject\n1608\tsubmit\n1609\tsubscrib\n1610\tsubscript\n1611\tsubstanti\n1612\tsuccess\n1613\tsuch\n1614\tsuffer\n1615\tsuggest\n1616\tsuit\n1617\tsum\n1618\tsummari\n1619\tsummer\n1620\tsun\n1621\tsuper\n1622\tsuppli\n1623\tsupport\n1624\tsuppos\n1625\tsure\n1626\tsurpris\n1627\tsuse\n1628\tsuspect\n1629\tsweet\n1630\tswitch\n1631\tsystem\n1632\ttab\n1633\ttabl\n1634\ttablet\n1635\ttag\n1636\ttake\n1637\ttaken\n1638\ttalk\n1639\ttape\n1640\ttarget\n1641\ttask\n1642\ttax\n1643\tteach\n1644\tteam\n1645\ttech\n1646\ttechnic\n1647\ttechniqu\n1648\ttechnolog\n1649\ttel\n1650\ttelecom\n1651\ttelephon\n1652\ttell\n1653\ttemperatur\n1654\ttempl\n1655\tten\n1656\tterm\n1657\ttermin\n1658\tterror\n1659\tterrorist\n1660\ttest\n1661\ttexa\n1662\ttext\n1663\tthan\n1664\tthank\n1665\tthat\n1666\tthe\n1667\tthei\n1668\ttheir\n1669\tthem\n1670\tthemselv\n1671\tthen\n1672\ttheori\n1673\tthere\n1674\ttherefor\n1675\tthese\n1676\tthi\n1677\tthing\n1678\tthink\n1679\tthinkgeek\n1680\tthird\n1681\tthose\n1682\tthough\n1683\tthought\n1684\tthousand\n1685\tthread\n1686\tthreat\n1687\tthree\n1688\tthrough\n1689\tthu\n1690\tthursdai\n1691\tti\n1692\tticket\n1693\ttim\n1694\ttime\n1695\ttip\n1696\ttire\n1697\ttitl\n1698\ttm\n1699\tto\n1700\ttodai\n1701\ttogeth\n1702\ttoken\n1703\ttold\n1704\ttoll\n1705\ttom\n1706\ttoner\n1707\ttoni\n1708\ttoo\n1709\ttook\n1710\ttool\n1711\ttop\n1712\ttopic\n1713\ttotal\n1714\ttouch\n1715\ttoward\n1716\ttrack\n1717\ttrade\n1718\ttradit\n1719\ttraffic\n1720\ttrain\n1721\ttransact\n1722\ttransfer\n1723\ttravel\n1724\ttreat\n1725\ttree\n1726\ttri\n1727\ttrial\n1728\ttrick\n1729\ttrip\n1730\ttroubl\n1731\ttrue\n1732\ttruli\n1733\ttrust\n1734\ttruth\n1735\ttry\n1736\ttue\n1737\ttuesdai\n1738\tturn\n1739\ttv\n1740\ttwo\n1741\ttype\n1742\tuk\n1743\tultim\n1744\tun\n1745\tunder\n1746\tunderstand\n1747\tunfortun\n1748\tuniqu\n1749\tunison\n1750\tunit\n1751\tunivers\n1752\tunix\n1753\tunless\n1754\tunlik\n1755\tunlimit\n1756\tunseen\n1757\tunsolicit\n1758\tunsubscrib\n1759\tuntil\n1760\tup\n1761\tupdat\n1762\tupgrad\n1763\tupon\n1764\turgent\n1765\turl\n1766\tus\n1767\tusa\n1768\tusag\n1769\tusb\n1770\tusd\n1771\tusdollarnumb\n1772\tuseless\n1773\tuser\n1774\tusr\n1775\tusual\n1776\tutil\n1777\tvacat\n1778\tvalid\n1779\tvalu\n1780\tvaluabl\n1781\tvar\n1782\tvariabl\n1783\tvarieti\n1784\tvariou\n1785\tve\n1786\tvendor\n1787\tventur\n1788\tveri\n1789\tverifi\n1790\tversion\n1791\tvia\n1792\tvideo\n1793\tview\n1794\tvirtual\n1795\tvisa\n1796\tvisit\n1797\tvisual\n1798\tvnumber\n1799\tvoic\n1800\tvote\n1801\tvs\n1802\tvulner\n1803\twa\n1804\twai\n1805\twait\n1806\twake\n1807\twalk\n1808\twall\n1809\twant\n1810\twar\n1811\twarm\n1812\twarn\n1813\twarranti\n1814\twashington\n1815\twasn\n1816\twast\n1817\twatch\n1818\twater\n1819\twe\n1820\twealth\n1821\tweapon\n1822\tweb\n1823\tweblog\n1824\twebsit\n1825\twed\n1826\twednesdai\n1827\tweek\n1828\tweekli\n1829\tweight\n1830\twelcom\n1831\twell\n1832\twent\n1833\twere\n1834\twest\n1835\twhat\n1836\twhatev\n1837\twhen\n1838\twhere\n1839\twhether\n1840\twhich\n1841\twhile\n1842\twhite\n1843\twhitelist\n1844\twho\n1845\twhole\n1846\twhose\n1847\twhy\n1848\twi\n1849\twide\n1850\twidth\n1851\twife\n1852\twill\n1853\twilliam\n1854\twin\n1855\twindow\n1856\twing\n1857\twinner\n1858\twireless\n1859\twish\n1860\twith\n1861\twithin\n1862\twithout\n1863\twnumberp\n1864\twoman\n1865\twomen\n1866\twon\n1867\twonder\n1868\tword\n1869\twork\n1870\tworker\n1871\tworld\n1872\tworldwid\n1873\tworri\n1874\tworst\n1875\tworth\n1876\twould\n1877\twouldn\n1878\twrite\n1879\twritten\n1880\twrong\n1881\twrote\n1882\twww\n1883\tximian\n1884\txml\n1885\txp\n1886\tyahoo\n1887\tye\n1888\tyeah\n1889\tyear\n1890\tyesterdai\n1891\tyet\n1892\tyork\n1893\tyou\n1894\tyoung\n1895\tyour\n1896\tyourself\n1897\tzdnet\n1898\tzero\n1899\tzip\n" }, { "path": "machine-learning-ex7/ex7/computeCentroids.m", "content": "function centroids = computeCentroids(X, idx, K)\n%COMPUTECENTROIDS returs the new centroids by computing the means of the \n%data points assigned to each centroid.\n% centroids = COMPUTECENTROIDS(X, idx, K) returns the new centroids by \n% computing the means of the data points assigned to each centroid. It is\n% given a dataset X where each row is a single data point, a vector\n% idx of centroid assignments (i.e. each entry in range [1..K]) for each\n% example, and K, the number of centroids. You should return a matrix\n% centroids, where each row of centroids is the mean of the data points\n% assigned to it.\n%\n\n% Useful variables\n[m n] = size(X);\n\n% You need to return the following variables correctly.\ncentroids = zeros(K, n);\n\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Go over every centroid and compute mean of all points that\n% belong to it. Concretely, the row vector centroids(i, :)\n% should contain the mean of the data points assigned to\n% centroid i.\n%\n% Note: You can use a for-loop over the centroids to compute this.\n%\n\n\nfor i = 1:K\n centroids(i,:) = mean(X(idx==i,:),1);\nend\n\n\n\n\n\n% =============================================================\n\n\nend\n\n" }, { "path": "machine-learning-ex7/ex7/displayData.m", "content": "function [h, display_array] = displayData(X, example_width)\n%DISPLAYDATA Display 2D data in a nice grid\n% [h, display_array] = DISPLAYDATA(X, example_width) displays 2D data\n% stored in X in a nice grid. It returns the figure handle h and the \n% displayed array if requested.\n\n% Set example_width automatically if not passed in\nif ~exist('example_width', 'var') || isempty(example_width) \n\texample_width = round(sqrt(size(X, 2)));\nend\n\n% Gray Image\ncolormap(gray);\n\n% Compute rows, cols\n[m n] = size(X);\nexample_height = (n / example_width);\n\n% Compute number of items to display\ndisplay_rows = floor(sqrt(m));\ndisplay_cols = ceil(m / display_rows);\n\n% Between images padding\npad = 1;\n\n% Setup blank display\ndisplay_array = - ones(pad + display_rows * (example_height + pad), ...\n pad + display_cols * (example_width + pad));\n\n% Copy each example into a patch on the display array\ncurr_ex = 1;\nfor j = 1:display_rows\n\tfor i = 1:display_cols\n\t\tif curr_ex > m, \n\t\t\tbreak; \n\t\tend\n\t\t% Copy the patch\n\t\t\n\t\t% Get the max value of the patch\n\t\tmax_val = max(abs(X(curr_ex, :)));\n\t\tdisplay_array(pad + (j - 1) * (example_height + pad) + (1:example_height), ...\n\t\t pad + (i - 1) * (example_width + pad) + (1:example_width)) = ...\n\t\t\t\t\t\treshape(X(curr_ex, :), example_height, example_width) / max_val;\n\t\tcurr_ex = curr_ex + 1;\n\tend\n\tif curr_ex > m, \n\t\tbreak; \n\tend\nend\n\n% Display Image\nh = imagesc(display_array, [-1 1]);\n\n% Do not show axis\naxis image off\n\ndrawnow;\n\nend\n" }, { "path": "machine-learning-ex7/ex7/drawLine.m", "content": "function drawLine(p1, p2, varargin)\n%DRAWLINE Draws a line from point p1 to point p2\n% DRAWLINE(p1, p2) Draws a line from point p1 to point p2 and holds the\n% current figure\n\nplot([p1(1) p2(1)], [p1(2) p2(2)], varargin{:});\n\nend" }, { "path": "machine-learning-ex7/ex7/ex7.m", "content": "%% Machine Learning Online Class\n% Exercise 7 | Principle Component Analysis and K-Means Clustering\n%\n% Instructions\n% ------------\n%\n% This file contains code that helps you get started on the\n% exercise. You will need to complete the following functions:\n%\n% pca.m\n% projectData.m\n% recoverData.m\n% computeCentroids.m\n% findClosestCentroids.m\n% kMeansInitCentroids.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% ================= Part 1: Find Closest Centroids ====================\n% To help you implement K-Means, we have divided the learning algorithm \n% into two functions -- findClosestCentroids and computeCentroids. In this\n% part, you shoudl complete the code in the findClosestCentroids function. \n%\nfprintf('Finding closest centroids.\\n\\n');\n\n% Load an example dataset that we will be using\nload('ex7data2.mat');\n\n% Select an initial set of centroids\nK = 3; % 3 Centroids\ninitial_centroids = [3 3; 6 2; 8 5];\n\n% Find the closest centroids for the examples using the\n% initial_centroids\nidx = findClosestCentroids(X, initial_centroids);\n\nfprintf('Closest centroids for the first 3 examples: \\n')\nfprintf(' %d', idx(1:3));\nfprintf('\\n(the closest centroids should be 1, 3, 2 respectively)\\n');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ===================== Part 2: Compute Means =========================\n% After implementing the closest centroids function, you should now\n% complete the computeCentroids function.\n%\nfprintf('\\nComputing centroids means.\\n\\n');\n\n% Compute means based on the closest centroids found in the previous part.\ncentroids = computeCentroids(X, idx, K);\n\nfprintf('Centroids computed after initial finding of closest centroids: \\n')\nfprintf(' %f %f \\n' , centroids');\nfprintf('\\n(the centroids should be\\n');\nfprintf(' [ 2.428301 3.157924 ]\\n');\nfprintf(' [ 5.813503 2.633656 ]\\n');\nfprintf(' [ 7.119387 3.616684 ]\\n\\n');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% =================== Part 3: K-Means Clustering ======================\n% After you have completed the two functions computeCentroids and\n% findClosestCentroids, you have all the necessary pieces to run the\n% kMeans algorithm. In this part, you will run the K-Means algorithm on\n% the example dataset we have provided. \n%\nfprintf('\\nRunning K-Means clustering on example dataset.\\n\\n');\n\n% Load an example dataset\nload('ex7data2.mat');\n\n% Settings for running K-Means\nK = 3;\nmax_iters = 10;\n\n% For consistency, here we set centroids to specific values\n% but in practice you want to generate them automatically, such as by\n% settings them to be random examples (as can be seen in\n% kMeansInitCentroids).\ninitial_centroids = [3 3; 6 2; 8 5];\n\n% Run K-Means algorithm. The 'true' at the end tells our function to plot\n% the progress of K-Means\n[centroids, idx] = runkMeans(X, initial_centroids, max_iters, true);\nfprintf('\\nK-Means Done.\\n\\n');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ============= Part 4: K-Means Clustering on Pixels ===============\n% In this exercise, you will use K-Means to compress an image. To do this,\n% you will first run K-Means on the colors of the pixels in the image and\n% then you will map each pixel on to it's closest centroid.\n% \n% You should now complete the code in kMeansInitCentroids.m\n%\n\nfprintf('\\nRunning K-Means clustering on pixels from an image.\\n\\n');\n\n% Load an image of a bird\nA = double(imread('bird_small.png'));\n\n% If imread does not work for you, you can try instead\n% load ('bird_small.mat');\n\nA = A / 255; % Divide by 255 so that all values are in the range 0 - 1\n\n% Size of the image\nimg_size = size(A);\n\n% Reshape the image into an Nx3 matrix where N = number of pixels.\n% Each row will contain the Red, Green and Blue pixel values\n% This gives us our dataset matrix X that we will use K-Means on.\nX = reshape(A, img_size(1) * img_size(2), 3);\n\n% Run your K-Means algorithm on this data\n% You should try different values of K and max_iters here\nK = 16; \nmax_iters = 10;\n\n% When using K-Means, it is important the initialize the centroids\n% randomly. \n% You should complete the code in kMeansInitCentroids.m before proceeding\ninitial_centroids = kMeansInitCentroids(X, K);\n\n% Run K-Means\n[centroids, idx] = runkMeans(X, initial_centroids, max_iters);\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ================= Part 5: Image Compression ======================\n% In this part of the exercise, you will use the clusters of K-Means to\n% compress an image. To do this, we first find the closest clusters for\n% each example. After that, we \n\nfprintf('\\nApplying K-Means to compress an image.\\n\\n');\n\n% Find closest cluster members\nidx = findClosestCentroids(X, centroids);\n\n% Essentially, now we have represented the image X as in terms of the\n% indices in idx. \n\n% We can now recover the image from the indices (idx) by mapping each pixel\n% (specified by it's index in idx) to the centroid value\nX_recovered = centroids(idx,:);\n\n% Reshape the recovered image into proper dimensions\nX_recovered = reshape(X_recovered, img_size(1), img_size(2), 3);\n\n% Display the original image \nsubplot(1, 2, 1);\nimagesc(A); \ntitle('Original');\n\n% Display compressed image side by side\nsubplot(1, 2, 2);\nimagesc(X_recovered)\ntitle(sprintf('Compressed, with %d colors.', K));\n\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n" }, { "path": "machine-learning-ex7/ex7/ex7_pca.m", "content": "%% Machine Learning Online Class\n% Exercise 7 | Principle Component Analysis and K-Means Clustering\n%\n% Instructions\n% ------------\n%\n% This file contains code that helps you get started on the\n% exercise. You will need to complete the following functions:\n%\n% pca.m\n% projectData.m\n% recoverData.m\n% computeCentroids.m\n% findClosestCentroids.m\n% kMeansInitCentroids.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% ================== Part 1: Load Example Dataset ===================\n% We start this exercise by using a small dataset that is easily to\n% visualize\n%\nfprintf('Visualizing example dataset for PCA.\\n\\n');\n\n% The following command loads the dataset. You should now have the \n% variable X in your environment\nload ('ex7data1.mat');\n\n% Visualize the example dataset\nplot(X(:, 1), X(:, 2), 'bo');\naxis([0.5 6.5 2 8]); axis square;\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% =============== Part 2: Principal Component Analysis ===============\n% You should now implement PCA, a dimension reduction technique. You\n% should complete the code in pca.m\n%\nfprintf('\\nRunning PCA on example dataset.\\n\\n');\n\n% Before running PCA, it is important to first normalize X\n[X_norm, mu, sigma] = featureNormalize(X);\n\n% Run PCA\n[U, S] = pca(X_norm);\n\n% Compute mu, the mean of the each feature\n\n% Draw the eigenvectors centered at mean of data. These lines show the\n% directions of maximum variations in the dataset.\nhold on;\ndrawLine(mu, mu + 1.5 * S(1,1) * U(:,1)', '-k', 'LineWidth', 2);\ndrawLine(mu, mu + 1.5 * S(2,2) * U(:,2)', '-k', 'LineWidth', 2);\nhold off;\n\nfprintf('Top eigenvector: \\n');\nfprintf(' U(:,1) = %f %f \\n', U(1,1), U(2,1));\nfprintf('\\n(you should expect to see -0.707107 -0.707107)\\n');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% =================== Part 3: Dimension Reduction ===================\n% You should now implement the projection step to map the data onto the \n% first k eigenvectors. The code will then plot the data in this reduced \n% dimensional space. This will show you what the data looks like when \n% using only the corresponding eigenvectors to reconstruct it.\n%\n% You should complete the code in projectData.m\n%\nfprintf('\\nDimension reduction on example dataset.\\n\\n');\n\n% Plot the normalized dataset (returned from pca)\nplot(X_norm(:, 1), X_norm(:, 2), 'bo');\naxis([-4 3 -4 3]); axis square\n\n% Project the data onto K = 1 dimension\nK = 1;\nZ = projectData(X_norm, U, K);\nfprintf('Projection of the first example: %f\\n', Z(1));\nfprintf('\\n(this value should be about 1.481274)\\n\\n');\n\nX_rec = recoverData(Z, U, K);\nfprintf('Approximation of the first example: %f %f\\n', X_rec(1, 1), X_rec(1, 2));\nfprintf('\\n(this value should be about -1.047419 -1.047419)\\n\\n');\n\n% Draw lines connecting the projected points to the original points\nhold on;\nplot(X_rec(:, 1), X_rec(:, 2), 'ro');\nfor i = 1:size(X_norm, 1)\n drawLine(X_norm(i,:), X_rec(i,:), '--k', 'LineWidth', 1);\nend\nhold off\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =============== Part 4: Loading and Visualizing Face Data =============\n% We start the exercise by first loading and visualizing the dataset.\n% The following code will load the dataset into your environment\n%\nfprintf('\\nLoading face dataset.\\n\\n');\n\n% Load Face dataset\nload ('ex7faces.mat')\n\n% Display the first 100 faces in the dataset\ndisplayData(X(1:100, :));\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% =========== Part 5: PCA on Face Data: Eigenfaces ===================\n% Run PCA and visualize the eigenvectors which are in this case eigenfaces\n% We display the first 36 eigenfaces.\n%\nfprintf(['\\nRunning PCA on face dataset.\\n' ...\n '(this mght take a minute or two ...)\\n\\n']);\n\n% Before running PCA, it is important to first normalize X by subtracting \n% the mean value from each feature\n[X_norm, mu, sigma] = featureNormalize(X);\n\n% Run PCA\n[U, S] = pca(X_norm);\n\n% Visualize the top 36 eigenvectors found\ndisplayData(U(:, 1:36)');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% ============= Part 6: Dimension Reduction for Faces =================\n% Project images to the eigen space using the top k eigenvectors \n% If you are applying a machine learning algorithm \nfprintf('\\nDimension reduction for face dataset.\\n\\n');\n\nK = 100;\nZ = projectData(X_norm, U, K);\n\nfprintf('The projected data Z has a size of: ')\nfprintf('%d ', size(Z));\n\nfprintf('\\n\\nProgram paused. Press enter to continue.\\n');\npause;\n\n%% ==== Part 7: Visualization of Faces after PCA Dimension Reduction ====\n% Project images to the eigen space using the top K eigen vectors and \n% visualize only using those K dimensions\n% Compare to the original input, which is also displayed\n\nfprintf('\\nVisualizing the projected (reduced dimension) faces.\\n\\n');\n\nK = 100;\nX_rec = recoverData(Z, U, K);\n\n% Display normalized data\nsubplot(1, 2, 1);\ndisplayData(X_norm(1:100,:));\ntitle('Original faces');\naxis square;\n\n% Display reconstructed data from only k eigenfaces\nsubplot(1, 2, 2);\ndisplayData(X_rec(1:100,:));\ntitle('Recovered faces');\naxis square;\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n\n%% === Part 8(a): Optional (ungraded) Exercise: PCA for Visualization ===\n% One useful application of PCA is to use it to visualize high-dimensional\n% data. In the last K-Means exercise you ran K-Means on 3-dimensional \n% pixel colors of an image. We first visualize this output in 3D, and then\n% apply PCA to obtain a visualization in 2D.\n\nclose all; close all; clc\n\n% Re-load the image from the previous exercise and run K-Means on it\n% For this to work, you need to complete the K-Means assignment first\nA = double(imread('bird_small.png'));\n\n% If imread does not work for you, you can try instead\n% load ('bird_small.mat');\n\nA = A / 255;\nimg_size = size(A);\nX = reshape(A, img_size(1) * img_size(2), 3);\nK = 16; \nmax_iters = 10;\ninitial_centroids = kMeansInitCentroids(X, K);\n[centroids, idx] = runkMeans(X, initial_centroids, max_iters);\n\n% Sample 1000 random indexes (since working with all the data is\n% too expensive. If you have a fast computer, you may increase this.\nsel = floor(rand(1000, 1) * size(X, 1)) + 1;\n\n% Setup Color Palette\npalette = hsv(K);\ncolors = palette(idx(sel), :);\n\n% Visualize the data and centroid memberships in 3D\nfigure;\nscatter3(X(sel, 1), X(sel, 2), X(sel, 3), 10, colors);\ntitle('Pixel dataset plotted in 3D. Color shows centroid memberships');\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% === Part 8(b): Optional (ungraded) Exercise: PCA for Visualization ===\n% Use PCA to project this cloud to 2D for visualization\n\n% Subtract the mean to use PCA\n[X_norm, mu, sigma] = featureNormalize(X);\n\n% PCA and project the data to 2D\n[U, S] = pca(X_norm);\nZ = projectData(X_norm, U, 2);\n\n% Plot in 2D\nfigure;\nplotDataPoints(Z(sel, :), idx(sel), K);\ntitle('Pixel dataset plotted in 2D, using PCA for dimensionality reduction');\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n" }, { "path": "machine-learning-ex7/ex7/featureNormalize.m", "content": "function [X_norm, mu, sigma] = featureNormalize(X)\n%FEATURENORMALIZE Normalizes the features in X \n% FEATURENORMALIZE(X) returns a normalized version of X where\n% the mean value of each feature is 0 and the standard deviation\n% is 1. This is often a good preprocessing step to do when\n% working with learning algorithms.\n\nmu = mean(X);\nX_norm = bsxfun(@minus, X, mu);\n\nsigma = std(X_norm);\nX_norm = bsxfun(@rdivide, X_norm, sigma);\n\n\n% ============================================================\n\nend\n" }, { "path": "machine-learning-ex7/ex7/findClosestCentroids.m", "content": "function idx = findClosestCentroids(X, centroids)\n%FINDCLOSESTCENTROIDS computes the centroid memberships for every example\n% idx = FINDCLOSESTCENTROIDS (X, centroids) returns the closest centroids\n% in idx for a dataset X where each row is a single example. idx = m x 1 \n% vector of centroid assignments (i.e. each entry in range [1..K])\n%\n\n% Set K\nK = size(centroids, 1);\n\n% You need to return the following variables correctly.\nidx = zeros(size(X,1), 1);\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Go over every example, find its closest centroid, and store\n% the index inside idx at the appropriate location.\n% Concretely, idx(i) should contain the index of the centroid\n% closest to example i. Hence, it should be a value in the \n% range 1..K\n%\n% Note: You can use a for-loop over the examples to compute this.\n%\n\nm = size(X,1);\n\nfor i = 1:m \n for j = 1:K\n idx(i,j) = (X(i,:)-centroids(j,:))*(X(i,:)-centroids(j,:))';\n end\nend\n[dummy idx] = min(idx,[],2);\n\n\n\n\n\n% =============================================================\n\nend\n\n" }, { "path": "machine-learning-ex7/ex7/kMeansInitCentroids.m", "content": "function centroids = kMeansInitCentroids(X, K)\n%KMEANSINITCENTROIDS This function initializes K centroids that are to be \n%used in K-Means on the dataset X\n% centroids = KMEANSINITCENTROIDS(X, K) returns K initial centroids to be\n% used with the K-Means on the dataset X\n%\n\n% You should return this values correctly\ncentroids = zeros(K, size(X, 2));\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: You should set centroids to randomly chosen examples from\n% the dataset X\n%\n\n\nrandidx = randperm(size(X, 1));\n\ncentroids = X(randidx(1:K), :);\n\n\n\n\n\n% =============================================================\n\nend\n\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex7/ex7/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i 0.5) = 0;\nR = zeros(size(Y));\nR(Y ~= 0) = 1;\n\n%% Run Gradient Checking\nX = randn(size(X_t));\nTheta = randn(size(Theta_t));\nnum_users = size(Y, 2);\nnum_movies = size(Y, 1);\nnum_features = size(Theta_t, 2);\n\nnumgrad = computeNumericalGradient( ...\n @(t) cofiCostFunc(t, Y, R, num_users, num_movies, ...\n num_features, lambda), [X(:); Theta(:)]);\n\n[cost, grad] = cofiCostFunc([X(:); Theta(:)], Y, R, num_users, ...\n num_movies, num_features, lambda);\n\ndisp([numgrad grad]);\nfprintf(['The above two columns you get should be very similar.\\n' ...\n '(Left-Your Numerical Gradient, Right-Analytical Gradient)\\n\\n']);\n\ndiff = norm(numgrad-grad)/norm(numgrad+grad);\nfprintf(['If your backpropagation implementation is correct, then \\n' ...\n 'the relative difference will be small (less than 1e-9). \\n' ...\n '\\nRelative Difference: %g\\n'], diff);\n\nend" }, { "path": "machine-learning-ex8/ex8/cofiCostFunc.m", "content": "function [J, grad] = cofiCostFunc(params, Y, R, num_users, num_movies, ...\n num_features, lambda)\n%COFICOSTFUNC Collaborative filtering cost function\n% [J, grad] = COFICOSTFUNC(params, Y, R, num_users, num_movies, ...\n% num_features, lambda) returns the cost and gradient for the\n% collaborative filtering problem.\n%\n\n% Unfold the U and W matrices from params\nX = reshape(params(1:num_movies*num_features), num_movies, num_features);\nTheta = reshape(params(num_movies*num_features+1:end), ...\n num_users, num_features);\n\n \n% You need to return the following values correctly\nJ = 0;\nX_grad = zeros(size(X));\nTheta_grad = zeros(size(Theta));\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Compute the cost function and gradient for collaborative\n% filtering. Concretely, you should first implement the cost\n% function (without regularization) and make sure it is\n% matches our costs. After that, you should implement the \n% gradient and use the checkCostFunction routine to check\n% that the gradient is correct. Finally, you should implement\n% regularization.\n%\n% Notes: X - num_movies x num_features matrix of movie features\n% Theta - num_users x num_features matrix of user features\n% Y - num_movies x num_users matrix of user ratings of movies\n% R - num_movies x num_users matrix, where R(i, j) = 1 if the \n% i-th movie was rated by the j-th user\n%\n% You should set the following variables correctly:\n%\n% X_grad - num_movies x num_features matrix, containing the \n% partial derivatives w.r.t. to each element of X\n% Theta_grad - num_users x num_features matrix, containing the \n% partial derivatives w.r.t. to each element of Theta\n%\n\nterm = X*Theta'.*R-Y.*R;\nJ = (term(:)'*term(:)+lambda*Theta(:)'*Theta(:)+lambda*X(:)'*X(:))/2;\n\n\n\nX_grad = term*Theta+lambda*X;\nTheta_grad = term'*X+lambda*Theta;\n\n\n\n\n\n\n\n\n\n% =============================================================\n\ngrad = [X_grad(:); Theta_grad(:)];\n\nend\n" }, { "path": "machine-learning-ex8/ex8/computeNumericalGradient.m", "content": "function numgrad = computeNumericalGradient(J, theta)\n%COMPUTENUMERICALGRADIENT Computes the gradient using \"finite differences\"\n%and gives us a numerical estimate of the gradient.\n% numgrad = COMPUTENUMERICALGRADIENT(J, theta) computes the numerical\n% gradient of the function J around theta. Calling y = J(theta) should\n% return the function value at theta.\n\n% Notes: The following code implements numerical gradient checking, and \n% returns the numerical gradient.It sets numgrad(i) to (a numerical \n% approximation of) the partial derivative of J with respect to the \n% i-th input argument, evaluated at theta. (i.e., numgrad(i) should \n% be the (approximately) the partial derivative of J with respect \n% to theta(i).)\n% \n\nnumgrad = zeros(size(theta));\nperturb = zeros(size(theta));\ne = 1e-4;\nfor p = 1:numel(theta)\n % Set perturbation vector\n perturb(p) = e;\n loss1 = J(theta - perturb);\n loss2 = J(theta + perturb);\n % Compute Numerical Gradient\n numgrad(p) = (loss2 - loss1) / (2*e);\n perturb(p) = 0;\nend\n\nend\n" }, { "path": "machine-learning-ex8/ex8/estimateGaussian.m", "content": "function [mu sigma2] = estimateGaussian(X)\n%ESTIMATEGAUSSIAN This function estimates the parameters of a \n%Gaussian distribution using the data in X\n% [mu sigma2] = estimateGaussian(X), \n% The input X is the dataset with each n-dimensional data point in one row\n% The output is an n-dimensional vector mu, the mean of the data set\n% and the variances sigma^2, an n x 1 vector\n% \n\n% Useful variables\n[m, n] = size(X);\n\n% You should return these values correctly\nmu = zeros(n, 1);\nsigma2 = zeros(n, 1);\n\n% ====================== YOUR CODE HERE ======================\n% Instructions: Compute the mean of the data and the variances\n% In particular, mu(i) should contain the mean of\n% the data for the i-th feature and sigma2(i)\n% should contain variance of the i-th feature.\n%\n\nmu = mean(X,1)'; %ÿеƽֵתΪ\nsigma2 = ((m-1)*var(X)/m)'; %var(X)ÿеķ,ҪעmĹϵ\n\n\n\n\n\n\n\n\n% =============================================================\n\n\nend\n" }, { "path": "machine-learning-ex8/ex8/ex8.m", "content": "%% Machine Learning Online Class\n% Exercise 8 | Anomaly Detection and Collaborative Filtering\n%\n% Instructions\n% ------------\n%\n% This file contains code that helps you get started on the\n% exercise. You will need to complete the following functions:\n%\n% estimateGaussian.m\n% selectThreshold.m\n% cofiCostFunc.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% Initialization\nclear ; close all; clc\n\n%% ================== Part 1: Load Example Dataset ===================\n% We start this exercise by using a small dataset that is easy to\n% visualize.\n%\n% Our example case consists of 2 network server statistics across\n% several machines: the latency and throughput of each machine.\n% This exercise will help us find possibly faulty (or very fast) machines.\n%\n\nfprintf('Visualizing example dataset for outlier detection.\\n\\n');\n\n% The following command loads the dataset. You should now have the\n% variables X, Xval, yval in your environment\nload('ex8data1.mat');\n\n% Visualize the example dataset\nplot(X(:, 1), X(:, 2), 'bx');\naxis([0 30 0 30]);\nxlabel('Latency (ms)');\nylabel('Throughput (mb/s)');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause\n\n\n%% ================== Part 2: Estimate the dataset statistics ===================\n% For this exercise, we assume a Gaussian distribution for the dataset.\n%\n% We first estimate the parameters of our assumed Gaussian distribution, \n% then compute the probabilities for each of the points and then visualize \n% both the overall distribution and where each of the points falls in \n% terms of that distribution.\n%\nfprintf('Visualizing Gaussian fit.\\n\\n');\n\n% Estimate my and sigma2\n[mu sigma2] = estimateGaussian(X);\n\n% Returns the density of the multivariate normal at each data point (row) \n% of X\np = multivariateGaussian(X, mu, sigma2);\n\n% Visualize the fit\nvisualizeFit(X, mu, sigma2);\nxlabel('Latency (ms)');\nylabel('Throughput (mb/s)');\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ================== Part 3: Find Outliers ===================\n% Now you will find a good epsilon threshold using a cross-validation set\n% probabilities given the estimated Gaussian distribution\n% \n\npval = multivariateGaussian(Xval, mu, sigma2);\n\n[epsilon F1] = selectThreshold(yval, pval);\nfprintf('Best epsilon found using cross-validation: %e\\n', epsilon);\nfprintf('Best F1 on Cross Validation Set: %f\\n', F1);\nfprintf(' (you should see a value epsilon of about 8.99e-05)\\n\\n');\n\n% Find the outliers in the training set and plot the\noutliers = find(p < epsilon);\n\n% Draw a red circle around those outliers\nhold on\nplot(X(outliers, 1), X(outliers, 2), 'ro', 'LineWidth', 2, 'MarkerSize', 10);\nhold off\n\nfprintf('Program paused. Press enter to continue.\\n');\npause;\n\n%% ================== Part 4: Multidimensional Outliers ===================\n% We will now use the code from the previous part and apply it to a \n% harder problem in which more features describe each datapoint and only \n% some features indicate whether a point is an outlier.\n%\n\n% Loads the second dataset. You should now have the\n% variables X, Xval, yval in your environment\nload('ex8data2.mat');\n\n% Apply the same steps to the larger dataset\n[mu sigma2] = estimateGaussian(X);\n\n% Training set \np = multivariateGaussian(X, mu, sigma2);\n\n% Cross-validation set\npval = multivariateGaussian(Xval, mu, sigma2);\n\n% Find the best threshold\n[epsilon F1] = selectThreshold(yval, pval);\n\nfprintf('Best epsilon found using cross-validation: %e\\n', epsilon);\nfprintf('Best F1 on Cross Validation Set: %f\\n', F1);\nfprintf('# Outliers found: %d\\n', sum(p < epsilon));\nfprintf(' (you should see a value epsilon of about 1.38e-18)\\n\\n');\npause\n\n\n\n" }, { "path": "machine-learning-ex8/ex8/ex8_cofi.m", "content": "%% Machine Learning Online Class\n% Exercise 8 | Anomaly Detection and Collaborative Filtering\n%\n% Instructions\n% ------------\n%\n% This file contains code that helps you get started on the\n% exercise. You will need to complete the following functions:\n%\n% estimateGaussian.m\n% selectThreshold.m\n% cofiCostFunc.m\n%\n% For this exercise, you will not need to change any code in this file,\n% or any other files other than those mentioned above.\n%\n\n%% =============== Part 1: Loading movie ratings dataset ================\n% You will start by loading the movie ratings dataset to understand the\n% structure of the data.\n% \nfprintf('Loading movie ratings dataset.\\n\\n');\n\n% Load data\nload ('ex8_movies.mat');\n\n% Y is a 1682x943 matrix, containing ratings (1-5) of 1682 movies on \n% 943 users\n%\n% R is a 1682x943 matrix, where R(i,j) = 1 if and only if user j gave a\n% rating to movie i\n\n% From the matrix, we can compute statistics like average rating.\nfprintf('Average rating for movie 1 (Toy Story): %f / 5\\n\\n', ...\n mean(Y(1, R(1, :))));\n\n% We can \"visualize\" the ratings matrix by plotting it with imagesc\nimagesc(Y);\nylabel('Movies');\nxlabel('Users');\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n%% ============ Part 2: Collaborative Filtering Cost Function ===========\n% You will now implement the cost function for collaborative filtering.\n% To help you debug your cost function, we have included set of weights\n% that we trained on that. Specifically, you should complete the code in \n% cofiCostFunc.m to return J.\n\n% Load pre-trained weights (X, Theta, num_users, num_movies, num_features)\nload ('ex8_movieParams.mat');\n\n% Reduce the data set size so that this runs faster\nnum_users = 4; num_movies = 5; num_features = 3;\nX = X(1:num_movies, 1:num_features);\nTheta = Theta(1:num_users, 1:num_features);\nY = Y(1:num_movies, 1:num_users);\nR = R(1:num_movies, 1:num_users);\n\n% Evaluate cost function\nJ = cofiCostFunc([X(:) ; Theta(:)], Y, R, num_users, num_movies, ...\n num_features, 0);\n \nfprintf(['Cost at loaded parameters: %f '...\n '\\n(this value should be about 22.22)\\n'], J);\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n\n%% ============== Part 3: Collaborative Filtering Gradient ==============\n% Once your cost function matches up with ours, you should now implement \n% the collaborative filtering gradient function. Specifically, you should \n% complete the code in cofiCostFunc.m to return the grad argument.\n% \nfprintf('\\nChecking Gradients (without regularization) ... \\n');\n\n% Check gradients by running checkNNGradients\ncheckCostFunction;\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n\n%% ========= Part 4: Collaborative Filtering Cost Regularization ========\n% Now, you should implement regularization for the cost function for \n% collaborative filtering. You can implement it by adding the cost of\n% regularization to the original cost computation.\n% \n\n% Evaluate cost function\nJ = cofiCostFunc([X(:) ; Theta(:)], Y, R, num_users, num_movies, ...\n num_features, 1.5);\n \nfprintf(['Cost at loaded parameters (lambda = 1.5): %f '...\n '\\n(this value should be about 31.34)\\n'], J);\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n\n%% ======= Part 5: Collaborative Filtering Gradient Regularization ======\n% Once your cost matches up with ours, you should proceed to implement \n% regularization for the gradient. \n%\n\n% \nfprintf('\\nChecking Gradients (with regularization) ... \\n');\n\n% Check gradients by running checkNNGradients\ncheckCostFunction(1.5);\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n\n%% ============== Part 6: Entering ratings for a new user ===============\n% Before we will train the collaborative filtering model, we will first\n% add ratings that correspond to a new user that we just observed. This\n% part of the code will also allow you to put in your own ratings for the\n% movies in our dataset!\n%\nmovieList = loadMovieList();\n\n% Initialize my ratings\nmy_ratings = zeros(1682, 1);\n\n% Check the file movie_idx.txt for id of each movie in our dataset\n% For example, Toy Story (1995) has ID 1, so to rate it \"4\", you can set\nmy_ratings(1) = 4;\n\n% Or suppose did not enjoy Silence of the Lambs (1991), you can set\nmy_ratings(98) = 2;\n\n% We have selected a few movies we liked / did not like and the ratings we\n% gave are as follows:\nmy_ratings(7) = 3;\nmy_ratings(12)= 5;\nmy_ratings(54) = 4;\nmy_ratings(64)= 5;\nmy_ratings(66)= 3;\nmy_ratings(69) = 5;\nmy_ratings(183) = 4;\nmy_ratings(226) = 5;\nmy_ratings(355)= 5;\n\nfprintf('\\n\\nNew user ratings:\\n');\nfor i = 1:length(my_ratings)\n if my_ratings(i) > 0 \n fprintf('Rated %d for %s\\n', my_ratings(i), ...\n movieList{i});\n end\nend\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n\n%% ================== Part 7: Learning Movie Ratings ====================\n% Now, you will train the collaborative filtering model on a movie rating \n% dataset of 1682 movies and 943 users\n%\n\nfprintf('\\nTraining collaborative filtering...\\n');\n\n% Load data\nload('ex8_movies.mat');\n\n% Y is a 1682x943 matrix, containing ratings (1-5) of 1682 movies by \n% 943 users\n%\n% R is a 1682x943 matrix, where R(i,j) = 1 if and only if user j gave a\n% rating to movie i\n\n% Add our own ratings to the data matrix\nY = [my_ratings Y];\nR = [(my_ratings ~= 0) R];\n\n% Normalize Ratings\n[Ynorm, Ymean] = normalizeRatings(Y, R);\n\n% Useful Values\nnum_users = size(Y, 2);\nnum_movies = size(Y, 1);\nnum_features = 10;\n\n% Set Initial Parameters (Theta, X)\nX = randn(num_movies, num_features);\nTheta = randn(num_users, num_features);\n\ninitial_parameters = [X(:); Theta(:)];\n\n% Set options for fmincg\noptions = optimset('GradObj', 'on', 'MaxIter', 100);\n\n% Set Regularization\nlambda = 10;\ntheta = fmincg (@(t)(cofiCostFunc(t, Y, R, num_users, num_movies, ...\n num_features, lambda)), ...\n initial_parameters, options);\n\n% Unfold the returned theta back into U and W\nX = reshape(theta(1:num_movies*num_features), num_movies, num_features);\nTheta = reshape(theta(num_movies*num_features+1:end), ...\n num_users, num_features);\n\nfprintf('Recommender system learning completed.\\n');\n\nfprintf('\\nProgram paused. Press enter to continue.\\n');\npause;\n\n%% ================== Part 8: Recommendation for you ====================\n% After training the model, you can now make recommendations by computing\n% the predictions matrix.\n%\n\np = X * Theta';\nmy_predictions = p(:,1) + Ymean;\n\nmovieList = loadMovieList();\n\n[r, ix] = sort(my_predictions, 'descend');\nfprintf('\\nTop recommendations for you:\\n');\nfor i=1:10\n j = ix(i);\n fprintf('Predicting rating %.1f for movie %s\\n', my_predictions(j), ...\n movieList{j});\nend\n\nfprintf('\\n\\nOriginal ratings provided:\\n');\nfor i = 1:length(my_ratings)\n if my_ratings(i) > 0 \n fprintf('Rated %d for %s\\n', my_ratings(i), ...\n movieList{i});\n end\nend\n" }, { "path": "machine-learning-ex8/ex8/fmincg.m", "content": "function [X, fX, i] = fmincg(f, X, options, P1, P2, P3, P4, P5)\n% Minimize a continuous differentialble multivariate function. Starting point\n% is given by \"X\" (D by 1), and the function named in the string \"f\", must\n% return a function value and a vector of partial derivatives. The Polack-\n% Ribiere flavour of conjugate gradients is used to compute search directions,\n% and a line search using quadratic and cubic polynomial approximations and the\n% Wolfe-Powell stopping criteria is used together with the slope ratio method\n% for guessing initial step sizes. Additionally a bunch of checks are made to\n% make sure that exploration is taking place and that extrapolation will not\n% be unboundedly large. The \"length\" gives the length of the run: if it is\n% positive, it gives the maximum number of line searches, if negative its\n% absolute gives the maximum allowed number of function evaluations. You can\n% (optionally) give \"length\" a second component, which will indicate the\n% reduction in function value to be expected in the first line-search (defaults\n% to 1.0). The function returns when either its length is up, or if no further\n% progress can be made (ie, we are at a minimum, or so close that due to\n% numerical problems, we cannot get any closer). If the function terminates\n% within a few iterations, it could be an indication that the function value\n% and derivatives are not consistent (ie, there may be a bug in the\n% implementation of your \"f\" function). The function returns the found\n% solution \"X\", a vector of function values \"fX\" indicating the progress made\n% and \"i\" the number of iterations (line searches or function evaluations,\n% depending on the sign of \"length\") used.\n%\n% Usage: [X, fX, i] = fmincg(f, X, options, P1, P2, P3, P4, P5)\n%\n% See also: checkgrad \n%\n% Copyright (C) 2001 and 2002 by Carl Edward Rasmussen. Date 2002-02-13\n%\n%\n% (C) Copyright 1999, 2000 & 2001, Carl Edward Rasmussen\n% \n% Permission is granted for anyone to copy, use, or modify these\n% programs and accompanying documents for purposes of research or\n% education, provided this copyright notice is retained, and note is\n% made of any changes that have been made.\n% \n% These programs and documents are distributed without any warranty,\n% express or implied. As the programs were written for research\n% purposes only, they have not been tested to the degree that would be\n% advisable in any important application. All use of these programs is\n% entirely at the user's own risk.\n%\n% [ml-class] Changes Made:\n% 1) Function name and argument specifications\n% 2) Output display\n%\n\n% Read options\nif exist('options', 'var') && ~isempty(options) && isfield(options, 'MaxIter')\n length = options.MaxIter;\nelse\n length = 100;\nend\n\n\nRHO = 0.01; % a bunch of constants for line searches\nSIG = 0.5; % RHO and SIG are the constants in the Wolfe-Powell conditions\nINT = 0.1; % don't reevaluate within 0.1 of the limit of the current bracket\nEXT = 3.0; % extrapolate maximum 3 times the current bracket\nMAX = 20; % max 20 function evaluations per line search\nRATIO = 100; % maximum allowed slope ratio\n\nargstr = ['feval(f, X']; % compose string used to call function\nfor i = 1:(nargin - 3)\n argstr = [argstr, ',P', int2str(i)];\nend\nargstr = [argstr, ')'];\n\nif max(size(length)) == 2, red=length(2); length=length(1); else red=1; end\nS=['Iteration '];\n\ni = 0; % zero the run length counter\nls_failed = 0; % no previous line search has failed\nfX = [];\n[f1 df1] = eval(argstr); % get function value and gradient\ni = i + (length<0); % count epochs?!\ns = -df1; % search direction is steepest\nd1 = -s'*s; % this is the slope\nz1 = red/(1-d1); % initial step is red/(|s|+1)\n\nwhile i < abs(length) % while not finished\n i = i + (length>0); % count iterations?!\n\n X0 = X; f0 = f1; df0 = df1; % make a copy of current values\n X = X + z1*s; % begin line search\n [f2 df2] = eval(argstr);\n i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n f3 = f1; d3 = d1; z3 = -z1; % initialize point 3 equal to point 1\n if length>0, M = MAX; else M = min(MAX, -length-i); end\n success = 0; limit = -1; % initialize quanteties\n while 1\n while ((f2 > f1+z1*RHO*d1) || (d2 > -SIG*d1)) && (M > 0) \n limit = z1; % tighten the bracket\n if f2 > f1\n z2 = z3 - (0.5*d3*z3*z3)/(d3*z3+f2-f3); % quadratic fit\n else\n A = 6*(f2-f3)/z3+3*(d2+d3); % cubic fit\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = (sqrt(B*B-A*d2*z3*z3)-B)/A; % numerical error possible - ok!\n end\n if isnan(z2) || isinf(z2)\n z2 = z3/2; % if we had a numerical problem then bisect\n end\n z2 = max(min(z2, INT*z3),(1-INT)*z3); % don't accept too close to limits\n z1 = z1 + z2; % update the step\n X = X + z2*s;\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n z3 = z3-z2; % z3 is now relative to the location of z2\n end\n if f2 > f1+z1*RHO*d1 || d2 > -SIG*d1\n break; % this is a failure\n elseif d2 > SIG*d1\n success = 1; break; % success\n elseif M == 0\n break; % failure\n end\n A = 6*(f2-f3)/z3+3*(d2+d3); % make cubic extrapolation\n B = 3*(f3-f2)-z3*(d3+2*d2);\n z2 = -d2*z3*z3/(B+sqrt(B*B-A*d2*z3*z3)); % num. error possible - ok!\n if ~isreal(z2) || isnan(z2) || isinf(z2) || z2 < 0 % num prob or wrong sign?\n if limit < -0.5 % if we have no upper limit\n z2 = z1 * (EXT-1); % the extrapolate the maximum amount\n else\n z2 = (limit-z1)/2; % otherwise bisect\n end\n elseif (limit > -0.5) && (z2+z1 > limit) % extraplation beyond max?\n z2 = (limit-z1)/2; % bisect\n elseif (limit < -0.5) && (z2+z1 > z1*EXT) % extrapolation beyond limit\n z2 = z1*(EXT-1.0); % set to extrapolation limit\n elseif z2 < -z3*INT\n z2 = -z3*INT;\n elseif (limit > -0.5) && (z2 < (limit-z1)*(1.0-INT)) % too close to limit?\n z2 = (limit-z1)*(1.0-INT);\n end\n f3 = f2; d3 = d2; z3 = -z2; % set point 3 equal to point 2\n z1 = z1 + z2; X = X + z2*s; % update current estimates\n [f2 df2] = eval(argstr);\n M = M - 1; i = i + (length<0); % count epochs?!\n d2 = df2'*s;\n end % end of line search\n\n if success % if line search succeeded\n f1 = f2; fX = [fX' f1]';\n fprintf('%s %4i | Cost: %4.6e\\r', S, i, f1);\n s = (df2'*df2-df1'*df2)/(df1'*df1)*s - df2; % Polack-Ribiere direction\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n d2 = df1'*s;\n if d2 > 0 % new slope must be negative\n s = -df1; % otherwise use steepest direction\n d2 = -s'*s; \n end\n z1 = z1 * min(RATIO, d1/(d2-realmin)); % slope ratio but max RATIO\n d1 = d2;\n ls_failed = 0; % this line search did not fail\n else\n X = X0; f1 = f0; df1 = df0; % restore point from before failed line search\n if ls_failed || i > abs(length) % line search failed twice in a row\n break; % or we ran out of time, so we give up\n end\n tmp = df1; df1 = df2; df2 = tmp; % swap derivatives\n s = -df1; % try steepest\n d1 = -s'*s;\n z1 = 1/(1-d1); \n ls_failed = 1; % this line search failed\n end\n if exist('OCTAVE_VERSION')\n fflush(stdout);\n end\nend\nfprintf('\\n');\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/AUTHORS.txt", "content": "The author of \"jsonlab\" toolbox is Qianqian Fang. Qianqian\nis currently an Assistant Professor at Massachusetts General Hospital, \nHarvard Medical School.\n\nAddress: Martinos Center for Biomedical Imaging, \n Massachusetts General Hospital, \n Harvard Medical School\n Bldg 149, 13th St, Charlestown, MA 02129, USA\nURL: http://nmr.mgh.harvard.edu/~fangq/\nEmail: or \n\n\nThe script loadjson.m was built upon previous works by\n\n- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n date: 2009/11/02\n- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n date: 2009/03/22\n- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565\n date: 2008/07/03\n\n\nThis toolbox contains patches submitted by the following contributors:\n\n- Blake Johnson \n part of revision 341\n\n- Niclas Borlin \n various fixes in revision 394, including\n - loadjson crashes for all-zero sparse matrix.\n - loadjson crashes for empty sparse matrix.\n - Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.\n - loadjson crashes for sparse real column vector.\n - loadjson crashes for sparse complex column vector.\n - Data is corrupted by savejson for sparse real row vector.\n - savejson crashes for sparse complex row vector. \n\n- Yul Kang \n patches for svn revision 415.\n - savejson saves an empty cell array as [] instead of null\n - loadjson differentiates an empty struct from an empty array\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/ChangeLog.txt", "content": "============================================================================\n\n JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave\n\n----------------------------------------------------------------------------\n\nJSONlab ChangeLog (key features marked by *):\n\n== JSONlab 1.0 (codename: Optimus - Final), FangQ ==\n\n 2015/01/02 polish help info for all major functions, update examples, finalize 1.0\n 2014/12/19 fix a bug to strictly respect NoRowBracket in savejson\n\n== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ ==\n\n 2014/11/22 show progress bar in loadjson ('ShowProgress') \n 2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')\n 2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels\n 2014/09/18 start official github mirror: https://github.com/fangq/jsonlab\n\n== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ ==\n\n 2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8\n 2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson\n 2014/08/04 escape special characters in a JSON string\n 2014/02/16 fix a bug when saving ubjson files\n\n== JSONlab 0.9.9 (codename: Optimus - beta), FangQ ==\n\n 2014/01/22 use binary read and write in saveubjson and loadubjson\n\n== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ ==\n\n 2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)\n\n== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ ==\n 2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson\n\n== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ ==\n 2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)\n\n== JSONlab 0.9.0 (codename: Rodimus), FangQ ==\n\n 2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson\n 2012/06/01 support JSONP in savejson\n 2012/05/25 fix the empty cell bug (reported by Cyril Davin)\n 2012/04/05 savejson can save to a file (suggested by Patrick Rapin)\n\n== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ ==\n\n 2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS\n 2012/01/25 patch to handle root-less objects, contributed by Blake Johnson\n\n== JSONlab 0.8.0 (codename: Sentiel), FangQ ==\n\n 2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab\n 2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer\n 2011/12/22 *accept sequence of 'param',value input in savejson and loadjson\n 2011/11/18 fix struct array bug reported by Mykel Kochenderfer\n\n== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ ==\n\n 2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration\n 2011/10/20 loadjson supports JSON collections - concatenated JSON objects\n\n== JSONlab 0.5.0 (codename: Nexus), FangQ ==\n\n 2011/10/16 package and release jsonlab 0.5.0\n 2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug\n 2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level\n 2011/10/10 create jsonlab project, start jsonlab website, add online documentation\n 2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support\n 2011/10/06 *savejson works for structs, cells and arrays\n 2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/LICENSE_BSD.txt", "content": "Copyright 2011-2015 Qianqian Fang . All rights reserved.\n\nRedistribution and use in source and binary forms, with or without modification, are\npermitted provided that the following conditions are met:\n\n 1. Redistributions of source code must retain the above copyright notice, this list of\n conditions and the following disclaimer.\n\n 2. Redistributions in binary form must reproduce the above copyright notice, this list\n of conditions and the following disclaimer in the documentation and/or other materials\n provided with the distribution.\n\nTHIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS \nOR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR\nCONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR\nSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON\nANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING\nNEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF\nADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nThe views and conclusions contained in the software and documentation are those of the\nauthors and should not be interpreted as representing official policies, either expressed\nor implied, of the copyright holders.\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/README.txt", "content": "===============================================================================\n= JSONLab =\n= An open-source MATLAB/Octave JSON encoder and decoder =\n===============================================================================\n\n*Copyright (C) 2011-2015 Qianqian Fang \n*License: BSD License, see License_BSD.txt for details\n*Version: 1.0 (Optimus - Final)\n\n-------------------------------------------------------------------------------\n\nTable of Content:\n\nI. Introduction\nII. Installation\nIII.Using JSONLab\nIV. Known Issues and TODOs\nV. Contribution and feedback\n\n-------------------------------------------------------------------------------\n\nI. Introduction\n\nJSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable, \nhuman-readable and \"[http://en.wikipedia.org/wiki/JSON fat-free]\" text format \nto represent complex and hierarchical data. It is as powerful as \n[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely \nused for data-exchange in applications, and is essential for the wild success \nof [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and \n[http://en.wikipedia.org/wiki/Web_2.0 Web2.0]. \n\nUBJSON (Universal Binary JSON) is a binary JSON format, specifically \noptimized for compact file size and better performance while keeping\nthe semantics as simple as the text-based JSON format. Using the UBJSON\nformat allows to wrap complex binary data in a flexible and extensible\nstructure, making it possible to process complex and large dataset \nwithout accuracy loss due to text conversions.\n\nWe envision that both JSON and its binary version will serve as part of \nthe mainstream data-exchange formats for scientific research in the future. \nIt will provide the flexibility and generality achieved by other popular \ngeneral-purpose file specifications, such as\n[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly \nreduced complexity and enhanced performance.\n\nJSONLab is a free and open-source implementation of a JSON/UBJSON encoder \nand a decoder in the native MATLAB language. It can be used to convert a MATLAB \ndata structure (array, struct, cell, struct array and cell array) into \nJSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB \ndata structure. JSONLab supports both MATLAB and \n[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).\n\n-------------------------------------------------------------------------------\n\nII. Installation\n\nThe installation of JSONLab is no different than any other simple\nMATLAB toolbox. You only need to download/unzip the JSONLab package\nto a folder, and add the folder's path to MATLAB/Octave's path list\nby using the following command:\n\n addpath('/path/to/jsonlab');\n\nIf you want to add this path permanently, you need to type \"pathtool\", \nbrowse to the jsonlab root folder and add to the list, then click \"Save\".\nThen, run \"rehash\" in MATLAB, and type \"which loadjson\", if you see an \noutput, that means JSONLab is installed for MATLAB/Octave.\n\n-------------------------------------------------------------------------------\n\nIII.Using JSONLab\n\nJSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder, \nand savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and \ntwo equivallent functions -- loadubjson and saveubjson for the binary \nJSON. The detailed help info for the four functions can be found below:\n\n=== loadjson.m ===\n
\n  data=loadjson(fname,opt)\n     or\n  data=loadjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09, including previous works from \n \n          Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n             created on 2009/11/02\n          Franois Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n             created on  2009/03/22\n          Joel Feenstra:\n          http://www.mathworks.com/matlabcentral/fileexchange/20565\n             created on 2008/07/03\n \n  $Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a JSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n                          speed-optimized array parser when loading an \n                          array object. The fast array parser may \n                          collapse block arrays into a single large\n                          array similar to rules defined in cell2mat; 0 to \n                          use a legacy parser; if set to a larger-than-1\n                          value, this option will specify the minimum\n                          dimension to enable the fast array parser. For\n                          example, if the input is a 3D array, setting\n                          FastArrayParser to 1 will return a 3D array;\n                          setting to 2 will return a cell array of 2D\n                          arrays; setting to 3 will return to a 2D cell\n                          array of 1D vectors; setting to 4 will return a\n                          3D cell array.\n            opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n       dat=loadjson(['examples' filesep 'example1.json'])\n       dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n
\n\n=== savejson.m ===\n\n
\n  json=savejson(rootname,obj,filename)\n     or\n  json=savejson(rootname,obj,opt)\n  json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n  Object Notation) string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2011/09/09\n \n  $Id: savejson.m 458 2014-12-19 22:17:17Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array).\n       filename: a string for the file name to save the output JSON data.\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n                          of a 1D/2D array;\n         opt.ArrayIndent [1|0]: if 1, output explicit data array with\n                          precedent indentation; if 0, no indentation\n         opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n                          to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n                          and $1 represents the sign. For those who want to use\n                          1e999 to represent Inf, they can set opt.Inf to '$11e999'\n         opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n                          to represent NaN\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSONP='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n         opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n         opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a string in the JSON format (see http://json.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       savejson('jmesh',jsonmesh)\n       savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n
\n\n=== loadubjson.m ===\n\n
\n  data=loadubjson(fname,opt)\n     or\n  data=loadubjson(fname,'param1',value1,'param2',value2,...)\n \n  parse a JSON (JavaScript Object Notation) file or string\n \n  authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/01\n \n  $Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $\n \n  input:\n       fname: input file name, if fname contains \"{}\" or \"[]\", fname\n              will be interpreted as a UBJSON string\n       opt: a struct to store parsing options, opt can be replaced by \n            a list of ('param',value) pairs - the param string is equivallent\n            to a field in opt. opt can have the following \n            fields (first in [.|.] is the default)\n \n            opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n                          for each element of the JSON data, and group \n                          arrays based on the cell2mat rules.\n            opt.IntEndian [B|L]: specify the endianness of the integer fields\n                          in the UBJSON input data. B - Big-Endian format for \n                          integers (as required in the UBJSON specification); \n                          L - input integer fields are in Little-Endian order.\n \n  output:\n       dat: a cell array, where {...} blocks are converted into cell arrays,\n            and [...] are converted to arrays\n \n  examples:\n       obj=struct('string','value','array',[1 2 3]);\n       ubjdata=saveubjson('obj',obj);\n       dat=loadubjson(ubjdata)\n       dat=loadubjson(['examples' filesep 'example1.ubj'])\n       dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n
\n\n=== saveubjson.m ===\n\n
\n  json=saveubjson(rootname,obj,filename)\n     or\n  json=saveubjson(rootname,obj,opt)\n  json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n \n  convert a MATLAB object (cell, struct or array) into a Universal \n  Binary JSON (UBJSON) binary string\n \n  author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n  created on 2013/08/17\n \n  $Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $\n \n  input:\n       rootname: the name of the root-object, when set to '', the root name\n         is ignored, however, when opt.ForceRootName is set to 1 (see below),\n         the MATLAB variable name will be used as the root name.\n       obj: a MATLAB object (array, cell, cell array, struct, struct array)\n       filename: a string for the file name to save the output UBJSON data\n       opt: a struct for additional options, ignore to use default values.\n         opt can have the following fields (first in [.|.] is the default)\n \n         opt.FileName [''|string]: a file name to save the output JSON data\n         opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n                          array in JSON array format; if sets to 1, an\n                          array will be shown as a struct with fields\n                          \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n                          sparse arrays, the non-zero elements will be\n                          saved to _ArrayData_ field in triplet-format i.e.\n                          (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n                          with a value of 1; for a complex array, the \n                          _ArrayData_ array will include two columns \n                          (4 for sparse) to record the real and imaginary \n                          parts, and also \"_ArrayIsComplex_\":1 is added. \n         opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n                          will use true/false rather than 1/0.\n         opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n                          numerical element will be shown without a square\n                          bracket, unless it is the root object; if 0, square\n                          brackets are forced for any numerical arrays.\n         opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n                          will use the name of the passed obj variable as the \n                          root object name; if obj is an expression and \n                          does not have a name, 'root' will be used; if this \n                          is set to 0 and rootname is empty, the root level \n                          will be merged down to the lower level.\n         opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n                          for example, if opt.JSON='foo', the JSON data is\n                          wrapped inside a function call as 'foo(...);'\n         opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n                          back to the string form\n \n         opt can be replaced by a list of ('param',value) pairs. The param \n         string is equivallent to a field in opt and is case sensitive.\n  output:\n       json: a binary string in the UBJSON format (see http://ubjson.org)\n \n  examples:\n       jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n                'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n                'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n                           2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n                'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n                'SpecialData',[nan, inf, -inf]);\n       saveubjson('jsonmesh',jsonmesh)\n       saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n
\n\n\n=== examples ===\n\nUnder the \"examples\" folder, you can find several scripts to demonstrate the\nbasic utilities of JSONLab. Running the \"demo_jsonlab_basic.m\" script, you \nwill see the conversions from MATLAB data structure to JSON text and backward.\nIn \"jsonlab_selftest.m\", we load complex JSON files downloaded from the Internet\nand validate the loadjson/savejson functions for regression testing purposes.\nSimilarly, a \"demo_ubjson_basic.m\" script is provided to test the saveubjson\nand loadubjson pairs for various matlab data structures.\n\nPlease run these examples and understand how JSONLab works before you use\nit to process your data.\n\n-------------------------------------------------------------------------------\n\nIV. Known Issues and TODOs\n\nJSONLab has several known limitations. We are striving to make it more general\nand robust. Hopefully in a few future releases, the limitations become less.\n\nHere are the known issues:\n\n# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;\n# When processing names containing multi-byte characters, Octave and MATLAB \\\ncan give different field-names; you can use feature('DefaultCharacterSet','latin1') \\\nin MATLAB to get consistant results\n# savejson can not handle class and dataset.\n# saveubjson converts a logical array into a uint8 ([U]) array\n# an unofficial N-D array count syntax is implemented in saveubjson. We are \\\nactively communicating with the UBJSON spec maintainer to investigate the \\\npossibility of making it upstream\n# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \\\nfiles, however, it can parse all UBJSON files produced by saveubjson.\n\n-------------------------------------------------------------------------------\n\nV. Contribution and feedback\n\nJSONLab is an open-source project. This means you can not only use it and modify\nit as you wish, but also you can contribute your changes back to JSONLab so\nthat everyone else can enjoy the improvement. For anyone who want to contribute,\nplease download JSONLab source code from it's subversion repository by using the\nfollowing command:\n\n svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab\n\nYou can make changes to the files as needed. Once you are satisfied with your\nchanges, and ready to share it with others, please cd the root directory of \nJSONLab, and type\n\n svn diff > yourname_featurename.patch\n\nYou then email the .patch file to JSONLab's maintainer, Qianqian Fang, at\nthe email address shown in the beginning of this file. Qianqian will review \nthe changes and commit it to the subversion if they are satisfactory.\n\nWe appreciate any suggestions and feedbacks from you. Please use iso2mesh's\nmailing list to report any questions you may have with JSONLab:\n\nhttp://groups.google.com/group/iso2mesh-users?hl=en&pli=1\n\n(Subscription to the mailing list is needed in order to post messages).\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/jsonopt.m", "content": "function val=jsonopt(key,default,varargin)\n%\n% val=jsonopt(key,default,optstruct)\n%\n% setting options based on a struct. The struct can be produced\n% by varargin2struct from a list of 'param','value' pairs\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n%\n% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $\n%\n% input:\n% key: a string with which one look up a value from a struct\n% default: if the key does not exist, return default\n% optstruct: a struct where each sub-field is a key \n%\n% output:\n% val: if key exists, val=optstruct.key; otherwise val=default\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n% \n\nval=default;\nif(nargin<=2) return; end\nopt=varargin{1};\nif(isstruct(opt) && isfield(opt,key))\n val=getfield(opt,key);\nend\n\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/loadjson.m", "content": "function data = loadjson(fname,varargin)\n%\n% data=loadjson(fname,opt)\n% or\n% data=loadjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09, including previous works from \n%\n% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713\n% created on 2009/11/02\n% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393\n% created on 2009/03/22\n% Joel Feenstra:\n% http://www.mathworks.com/matlabcentral/fileexchange/20565\n% created on 2008/07/03\n%\n% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a JSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.FastArrayParser [1|0 or integer]: if set to 1, use a\n% speed-optimized array parser when loading an \n% array object. The fast array parser may \n% collapse block arrays into a single large\n% array similar to rules defined in cell2mat; 0 to \n% use a legacy parser; if set to a larger-than-1\n% value, this option will specify the minimum\n% dimension to enable the fast array parser. For\n% example, if the input is a 3D array, setting\n% FastArrayParser to 1 will return a 3D array;\n% setting to 2 will return a cell array of 2D\n% arrays; setting to 3 will return to a 2D cell\n% array of 1D vectors; setting to 4 will return a\n% 3D cell array.\n% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% dat=loadjson('{\"obj\":{\"string\":\"value\",\"array\":[1,2,3]}}')\n% dat=loadjson(['examples' filesep 'example1.json'])\n% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\n\nif(jsonopt('ShowProgress',0,opt)==1)\n opt.progressbar_=waitbar(0,'loading ...');\nend\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\nif(isfield(opt,'progressbar_'))\n close(opt.progressbar_);\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=data(j).x0x5F_ArraySize_;\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n if next_char ~= '}'\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n parse_char(':');\n val = parse_value(varargin{:});\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}'\n break;\n end\n parse_char(',');\n end\n end\n parse_char('}');\n\n%%-------------------------------------------------------------------------\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim2=[];\n arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});\n pbar=jsonopt('progressbar_',-1,varargin{:});\n\n if next_char ~= ']'\n\tif(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))\n [endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);\n arraystr=['[' inStr(pos:endpos)];\n arraystr=regexprep(arraystr,'\"_NaN_\"','NaN');\n arraystr=regexprep(arraystr,'\"([-+]*)_Inf_\"','$1Inf');\n arraystr(arraystr==sprintf('\\n'))=[];\n arraystr(arraystr==sprintf('\\r'))=[];\n %arraystr=regexprep(arraystr,'\\s*,',','); % this is slow,sometimes needed\n if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D\n \tastr=inStr((e1l+1):(e1r-1));\n \tastr=regexprep(astr,'\"_NaN_\"','NaN');\n \tastr=regexprep(astr,'\"([-+]*)_Inf_\"','$1Inf');\n \tastr(astr==sprintf('\\n'))=[];\n \tastr(astr==sprintf('\\r'))=[];\n \tastr(astr==' ')='';\n \tif(isempty(find(astr=='[', 1))) % array is 2D\n dim2=length(sscanf(astr,'%f,',[1 inf]));\n \tend\n else % array is 1D\n \tastr=arraystr(2:end-1);\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);\n \tif(nextidx>=length(astr)-1)\n object=obj;\n pos=endpos;\n parse_char(']');\n return;\n \tend\n end\n if(~isempty(dim2))\n \tastr=arraystr;\n \tastr(astr=='[')='';\n \tastr(astr==']')='';\n \tastr(astr==' ')='';\n \t[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);\n \tif(nextidx>=length(astr)-1)\n object=reshape(obj,dim2,numel(obj)/dim2)';\n pos=endpos;\n parse_char(']');\n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n return;\n \tend\n end\n arraystr=regexprep(arraystr,'\\]\\s*,','];');\n\telse\n arraystr='[';\n\tend\n try\n if(isoct && regexp(arraystr,'\"','once'))\n error('Octave eval can produce empty cells for JSON-like input');\n end\n object=eval(arraystr);\n pos=endpos;\n catch\n while 1\n newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);\n val = parse_value(newopt);\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n parse_char(',');\n end\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n parse_char(']');\n \n if(pbar>0)\n waitbar(pos/length(inStr),pbar,'loading ...');\n end\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr len esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n if inStr(pos) ~= '\"'\n error_pos('String starting with \" expected at position %d');\n else\n pos = pos + 1;\n end\n str = '';\n while pos <= len\n while index_esc <= len_esc && esc(index_esc) < pos\n index_esc = index_esc + 1;\n end\n if index_esc > len_esc\n str = [str inStr(pos:len)];\n pos = len + 1;\n break;\n else\n str = [str inStr(pos:esc(index_esc)-1)];\n pos = esc(index_esc);\n end\n nstr = length(str); switch inStr(pos)\n case '\"'\n pos = pos + 1;\n if(~isempty(str))\n if(strcmp(str,'_Inf_'))\n str=Inf;\n elseif(strcmp(str,'-_Inf_'))\n str=-Inf;\n elseif(strcmp(str,'_NaN_'))\n str=NaN;\n end\n end\n return;\n case '\\'\n if pos+1 > len\n error_pos('End of file reached right after escape character');\n end\n pos = pos + 1;\n switch inStr(pos)\n case {'\"' '\\' '/'}\n str(nstr+1) = inStr(pos);\n pos = pos + 1;\n case {'b' 'f' 'n' 'r' 't'}\n str(nstr+1) = sprintf(['\\' inStr(pos)]);\n pos = pos + 1;\n case 'u'\n if pos+4 > len\n error_pos('End of file reached in escaped unicode character');\n end\n str(nstr+(1:6)) = inStr(pos-1:pos+4);\n pos = pos + 5;\n end\n otherwise % should never happen\n str(nstr+1) = inStr(pos), keyboard\n pos = pos + 1;\n end\n end\n error_pos('End of file while expecting end of inStr');\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct\n currstr=inStr(pos:end);\n numstr=0;\n if(isoct~=0)\n numstr=regexp(currstr,'^\\s*-?(?:0|[1-9]\\d*)(?:\\.\\d+)?(?:[eE][+\\-]?\\d+)?','end');\n [num, one] = sscanf(currstr, '%f', 1);\n delta=numstr+1;\n else\n [num, one, err, delta] = sscanf(currstr, '%f', 1);\n if ~isempty(err)\n error_pos('Error reading number at position %d');\n end\n end\n pos = pos + delta-1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n \n pbar=jsonopt('progressbar_',-1,varargin{:});\n if(pbar>0)\n waitbar(pos/len,pbar,'loading ...');\n end\n \n switch(inStr(pos))\n case '\"'\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'-','0','1','2','3','4','5','6','7','8','9'}\n val = parse_number(varargin{:});\n return;\n case 't'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')\n val = true;\n pos = pos + 4;\n return;\n end\n case 'f'\n if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')\n val = false;\n pos = pos + 5;\n return;\n end\n case 'n'\n if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')\n val = [];\n pos = pos + 4;\n return;\n end\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/loadubjson.m", "content": "function data = loadubjson(fname,varargin)\n%\n% data=loadubjson(fname,opt)\n% or\n% data=loadubjson(fname,'param1',value1,'param2',value2,...)\n%\n% parse a JSON (JavaScript Object Notation) file or string\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/01\n%\n% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% fname: input file name, if fname contains \"{}\" or \"[]\", fname\n% will be interpreted as a UBJSON string\n% opt: a struct to store parsing options, opt can be replaced by \n% a list of ('param',value) pairs - the param string is equivallent\n% to a field in opt. opt can have the following \n% fields (first in [.|.] is the default)\n%\n% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat\n% for each element of the JSON data, and group \n% arrays based on the cell2mat rules.\n% opt.IntEndian [B|L]: specify the endianness of the integer fields\n% in the UBJSON input data. B - Big-Endian format for \n% integers (as required in the UBJSON specification); \n% L - input integer fields are in Little-Endian order.\n%\n% output:\n% dat: a cell array, where {...} blocks are converted into cell arrays,\n% and [...] are converted to arrays\n%\n% examples:\n% obj=struct('string','value','array',[1 2 3]);\n% ubjdata=saveubjson('obj',obj);\n% dat=loadubjson(ubjdata)\n% dat=loadubjson(['examples' filesep 'example1.ubj'])\n% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nglobal pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian\n\nif(regexp(fname,'[\\{\\}\\]\\[]','once'))\n string=fname;\nelseif(exist(fname,'file'))\n fid = fopen(fname,'rb');\n string = fread(fid,inf,'uint8=>char')';\n fclose(fid);\nelse\n error('input file does not exist');\nend\n\npos = 1; len = length(string); inStr = string;\nisoct=exist('OCTAVE_VERSION','builtin');\narraytoken=find(inStr=='[' | inStr==']' | inStr=='\"');\njstr=regexprep(inStr,'\\\\\\\\',' ');\nescquote=regexp(jstr,'\\\\\"');\narraytoken=sort([arraytoken escquote]);\n\n% String delimiters and escape chars identified to improve speed:\nesc = find(inStr=='\"' | inStr=='\\' ); % comparable to: regexp(inStr, '[\"\\\\]');\nindex_esc = 1; len_esc = length(esc);\n\nopt=varargin2struct(varargin{:});\nfileendian=upper(jsonopt('IntEndian','B',opt));\n[os,maxelem,systemendian]=computer;\n\njsoncount=1;\nwhile pos <= len\n switch(next_char)\n case '{'\n data{jsoncount} = parse_object(opt);\n case '['\n data{jsoncount} = parse_array(opt);\n otherwise\n error_pos('Outer level structure must be an object or an array');\n end\n jsoncount=jsoncount+1;\nend % while\n\njsoncount=length(data);\nif(jsoncount==1 && iscell(data))\n data=data{1};\nend\n\nif(~isempty(data))\n if(isstruct(data)) % data can be a struct array\n data=jstruct2array(data);\n elseif(iscell(data))\n data=jcell2array(data);\n end\nend\n\n\n%%\nfunction newdata=parse_collection(id,data,obj)\n\nif(jsoncount>0 && exist('data','var')) \n if(~iscell(data))\n newdata=cell(1);\n newdata{1}=data;\n data=newdata;\n end\nend\n\n%%\nfunction newdata=jcell2array(data)\nlen=length(data);\nnewdata=data;\nfor i=1:len\n if(isstruct(data{i}))\n newdata{i}=jstruct2array(data{i});\n elseif(iscell(data{i}))\n newdata{i}=jcell2array(data{i});\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newdata=jstruct2array(data)\nfn=fieldnames(data);\nnewdata=data;\nlen=length(data);\nfor i=1:length(fn) % depth-first\n for j=1:len\n if(isstruct(getfield(data(j),fn{i})))\n newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));\n end\n end\nend\nif(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))\n newdata=cell(len,1);\n for j=1:len\n ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);\n iscpx=0;\n if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))\n if(data(j).x0x5F_ArrayIsComplex_)\n iscpx=1;\n end\n end\n if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))\n if(data(j).x0x5F_ArrayIsSparse_)\n if(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n dim=double(data(j).x0x5F_ArraySize_);\n if(iscpx && size(ndata,2)==4-any(dim==1))\n ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));\n end\n if isempty(ndata)\n % All-zeros sparse\n ndata=sparse(dim(1),prod(dim(2:end)));\n elseif dim(1)==1\n % Sparse row vector\n ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));\n elseif dim(2)==1\n % Sparse column vector\n ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));\n else\n % Generic sparse array.\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));\n end\n else\n if(iscpx && size(ndata,2)==4)\n ndata(:,3)=complex(ndata(:,3),ndata(:,4));\n end\n ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));\n end\n end\n elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))\n if(iscpx && size(ndata,2)==2)\n ndata=complex(ndata(:,1),ndata(:,2));\n end\n ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);\n end\n newdata{j}=ndata;\n end\n if(len==1)\n newdata=newdata{1};\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction object = parse_object(varargin)\n parse_char('{');\n object = [];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1); % TODO\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n count=double(parse_number());\n end\n if next_char ~= '}'\n num=0;\n while 1\n str = parseStr(varargin{:});\n if isempty(str)\n error_pos('Name of value at position %d cannot be empty');\n end\n %parse_char(':');\n val = parse_value(varargin{:});\n num=num+1;\n eval( sprintf( 'object.%s = val;', valid_field(str) ) );\n if next_char == '}' || (count>=0 && num>=count)\n break;\n end\n %parse_char(',');\n end\n end\n if(count==-1)\n parse_char('}');\n end\n\n%%-------------------------------------------------------------------------\nfunction [cid,len]=elem_info(type)\nid=strfind('iUIlLdD',type);\ndataclass={'int8','uint8','int16','int32','int64','single','double'};\nbytelen=[1,1,2,4,8,4,8];\nif(id>0)\n cid=dataclass{id};\n len=bytelen(id);\nelse\n error_pos('unsupported type at position %d');\nend\n%%-------------------------------------------------------------------------\n\n\nfunction [data adv]=parse_block(type,count,varargin)\nglobal pos inStr isoct fileendian systemendian\n[cid,len]=elem_info(type);\ndatastr=inStr(pos:pos+len*count-1);\nif(isoct)\n newdata=int8(datastr);\nelse\n newdata=uint8(datastr);\nend\nid=strfind('iUIlLdD',type);\nif(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,cid));\nend\ndata=typecast(newdata,cid);\nadv=double(len*count);\n\n%%-------------------------------------------------------------------------\n\n\nfunction object = parse_array(varargin) % JSON array is written in row-major order\nglobal pos inStr isoct\n parse_char('[');\n object = cell(0, 1);\n dim=[];\n type='';\n count=-1;\n if(next_char == '$')\n type=inStr(pos+1);\n pos=pos+2;\n end\n if(next_char == '#')\n pos=pos+1;\n if(next_char=='[')\n dim=parse_array(varargin{:});\n count=prod(double(dim));\n else\n count=double(parse_number());\n end\n end\n if(~isempty(type))\n if(count>=0)\n [object adv]=parse_block(type,count,varargin{:});\n if(~isempty(dim))\n object=reshape(object,dim);\n end\n pos=pos+adv;\n return;\n else\n endpos=matching_bracket(inStr,pos);\n [cid,len]=elem_info(type);\n count=(endpos-pos)/len;\n [object adv]=parse_block(type,count,varargin{:});\n pos=pos+adv;\n parse_char(']');\n return;\n end\n end\n if next_char ~= ']'\n while 1\n val = parse_value(varargin{:});\n object{end+1} = val;\n if next_char == ']'\n break;\n end\n %parse_char(',');\n end\n end\n if(jsonopt('SimplifyCell',0,varargin{:})==1)\n try\n oldobj=object;\n object=cell2mat(object')';\n if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)\n object=oldobj;\n elseif(size(object,1)>1 && ndims(object)==2)\n object=object';\n end\n catch\n end\n end\n if(count==-1)\n parse_char(']');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction parse_char(c)\n global pos inStr len\n skip_whitespace;\n if pos > len || inStr(pos) ~= c\n error_pos(sprintf('Expected %c at position %%d', c));\n else\n pos = pos + 1;\n skip_whitespace;\n end\n\n%%-------------------------------------------------------------------------\n\nfunction c = next_char\n global pos inStr len\n skip_whitespace;\n if pos > len\n c = [];\n else\n c = inStr(pos);\n end\n\n%%-------------------------------------------------------------------------\n\nfunction skip_whitespace\n global pos inStr len\n while pos <= len && isspace(inStr(pos))\n pos = pos + 1;\n end\n\n%%-------------------------------------------------------------------------\nfunction str = parseStr(varargin)\n global pos inStr esc index_esc len_esc\n % len, ns = length(inStr), keyboard\n type=inStr(pos);\n if type ~= 'S' && type ~= 'C' && type ~= 'H'\n error_pos('String starting with S expected at position %d');\n else\n pos = pos + 1;\n end\n if(type == 'C')\n str=inStr(pos);\n pos=pos+1;\n return;\n end\n bytelen=double(parse_number());\n if(length(inStr)>=pos+bytelen-1)\n str=inStr(pos:pos+bytelen-1);\n pos=pos+bytelen;\n else\n error_pos('End of file while expecting end of inStr');\n end\n\n%%-------------------------------------------------------------------------\n\nfunction num = parse_number(varargin)\n global pos inStr len isoct fileendian systemendian\n id=strfind('iUIlLdD',inStr(pos));\n if(isempty(id))\n error_pos('expecting a number at position %d');\n end\n type={'int8','uint8','int16','int32','int64','single','double'};\n bytelen=[1,1,2,4,8,4,8];\n datastr=inStr(pos+1:pos+bytelen(id));\n if(isoct)\n newdata=int8(datastr);\n else\n newdata=uint8(datastr);\n end\n if(id<=5 && fileendian~=systemendian)\n newdata=swapbytes(typecast(newdata,type{id}));\n end\n num=typecast(newdata,type{id});\n pos = pos + bytelen(id)+1;\n\n%%-------------------------------------------------------------------------\n\nfunction val = parse_value(varargin)\n global pos inStr len\n true = 1; false = 0;\n\n switch(inStr(pos))\n case {'S','C','H'}\n val = parseStr(varargin{:});\n return;\n case '['\n val = parse_array(varargin{:});\n return;\n case '{'\n val = parse_object(varargin{:});\n if isstruct(val)\n if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))\n val=jstruct2array(val);\n end\n elseif isempty(val)\n val = struct;\n end\n return;\n case {'i','U','I','l','L','d','D'}\n val = parse_number(varargin{:});\n return;\n case 'T'\n val = true;\n pos = pos + 1;\n return;\n case 'F'\n val = false;\n pos = pos + 1;\n return;\n case {'Z','N'}\n val = [];\n pos = pos + 1;\n return;\n end\n error_pos('Value expected at position %d');\n%%-------------------------------------------------------------------------\n\nfunction error_pos(msg)\n global pos inStr len\n poShow = max(min([pos-15 pos-1 pos pos+20],len),1);\n if poShow(3) == poShow(2)\n poShow(3:4) = poShow(2)+[0 -1]; % display nothing after\n end\n msg = [sprintf(msg, pos) ': ' ...\n inStr(poShow(1):poShow(2)) '' inStr(poShow(3):poShow(4)) ];\n error( ['JSONparser:invalidFormat: ' msg] );\n\n%%-------------------------------------------------------------------------\n\nfunction str = valid_field(str)\nglobal isoct\n% From MATLAB doc: field names must begin with a letter, which may be\n% followed by any combination of letters, digits, and underscores.\n% Invalid characters will be converted to underscores, and the prefix\n% \"x0x[Hex code]_\" will be added if the first character is not a letter.\n pos=regexp(str,'^[^A-Za-z]','once');\n if(~isempty(pos))\n if(~isoct)\n str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');\n else\n str=sprintf('x0x%X_%s',char(str(1)),str(2:end));\n end\n end\n if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end\n if(~isoct)\n str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');\n else\n pos=regexp(str,'[^0-9A-Za-z_]');\n if(isempty(pos)) return; end\n str0=str;\n pos0=[0 pos(:)' length(str)];\n str='';\n for i=1:length(pos)\n str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];\n end\n if(pos(end)~=length(str))\n str=[str str0(pos0(end-1)+1:pos0(end))];\n end\n end\n %str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';\n\n%%-------------------------------------------------------------------------\nfunction endpos = matching_quote(str,pos)\nlen=length(str);\nwhile(pos1 && str(pos-1)=='\\'))\n endpos=pos;\n return;\n end \n end\n pos=pos+1;\nend\nerror('unmatched quotation mark');\n%%-------------------------------------------------------------------------\nfunction [endpos e1l e1r maxlevel] = matching_bracket(str,pos)\nglobal arraytoken\nlevel=1;\nmaxlevel=level;\nendpos=0;\nbpos=arraytoken(arraytoken>=pos);\ntokens=str(bpos);\nlen=length(tokens);\npos=1;\ne1l=[];\ne1r=[];\nwhile(pos<=len)\n c=tokens(pos);\n if(c==']')\n level=level-1;\n if(isempty(e1r)) e1r=bpos(pos); end\n if(level==0)\n endpos=bpos(pos);\n return\n end\n end\n if(c=='[')\n if(isempty(e1l)) e1l=bpos(pos); end\n level=level+1;\n maxlevel=max(maxlevel,level);\n end\n if(c=='\"')\n pos=matching_quote(tokens,pos+1);\n end\n pos=pos+1;\nend\nif(endpos==0) \n error('unmatched \"]\"');\nend\n\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/mergestruct.m", "content": "function s=mergestruct(s1,s2)\n%\n% s=mergestruct(s1,s2)\n%\n% merge two struct objects into one\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% s1,s2: a struct object, s1 and s2 can not be arrays\n%\n% output:\n% s: the merged struct object. fields in s1 and s2 will be combined in s.\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(~isstruct(s1) || ~isstruct(s2))\n error('input parameters contain non-struct');\nend\nif(length(s1)>1 || length(s2)>1)\n error('can not merge struct arrays');\nend\nfn=fieldnames(s2);\ns=s1;\nfor i=1:length(fn) \n s=setfield(s,fn{i},getfield(s2,fn{i}));\nend\n\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/savejson.m", "content": "function json=savejson(rootname,obj,varargin)\n%\n% json=savejson(rootname,obj,filename)\n% or\n% json=savejson(rootname,obj,opt)\n% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript\n% Object Notation) string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2011/09/09\n%\n% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array).\n% filename: a string for the file name to save the output JSON data.\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.FloatFormat ['%.10g'|string]: format to show each numeric element\n% of a 1D/2D array;\n% opt.ArrayIndent [1|0]: if 1, output explicit data array with\n% precedent indentation; if 0, no indentation\n% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [0|1]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.Inf ['\"$1_Inf_\"'|string]: a customized regular expression pattern\n% to represent +/-Inf. The matched pattern is '([-+]*)Inf'\n% and $1 represents the sign. For those who want to use\n% 1e999 to represent Inf, they can set opt.Inf to '$11e999'\n% opt.NaN ['\"_NaN_\"'|string]: a customized regular expression pattern\n% to represent NaN\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSONP='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.\n% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a string in the JSON format (see http://json.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% savejson('jmesh',jsonmesh)\n% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\\t%.5g')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\n\nwhitespaces=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nif(jsonopt('Compact',0,opt)==1)\n whitespaces=struct('tab','','newline','','sep',',');\nend\nif(~isfield(opt,'whitespaces_'))\n opt.whitespaces_=whitespaces;\nend\n\nnl=whitespaces.newline;\n\njson=obj2json(rootname,obj,rootlevel,opt);\nif(rootisarray)\n json=sprintf('%s%s',json,nl);\nelse\n json=sprintf('{%s%s%s}\\n',nl,json,nl);\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=sprintf('%s(%s);%s',jsonp,json,nl);\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n if(jsonopt('SaveBinary',0,opt)==1)\n\t fid = fopen(opt.FileName, 'wb');\n\t fwrite(fid,json);\n else\n\t fid = fopen(opt.FileName, 'wt');\n\t fwrite(fid,json,'char');\n end\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2json(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2json(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2json(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2json(name,item,level,varargin{:});\nelse\n txt=mat2json(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2json(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=jsonopt('whitespaces_',struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n')),varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nif(len>1)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": [%s',padding0, checkname(name,varargin{:}),nl); name=''; \n else\n txt=sprintf('%s[%s',padding0,nl); \n end\nelseif(len==0)\n if(~isempty(name))\n txt=sprintf('%s\"%s\": []',padding0, checkname(name,varargin{:})); name=''; \n else\n txt=sprintf('%s[]',padding0); \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));\n if(i1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2json(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding0=repmat(ws.tab,1,level);\npadding2=repmat(ws.tab,1,level+1);\npadding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));\nnl=ws.newline;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding0,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding0,nl); end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=sprintf('%s%s\"%s\": {%s',txt,padding1, checkname(name,varargin{:}),nl); \n else\n txt=sprintf('%s%s{%s',txt,padding1,nl); \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));\n if(e1) txt=sprintf('%s%s%s]',txt,nl,padding2); end\n if(j1) txt=sprintf('%s%s%s]',txt,nl,padding0); end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2json(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(~isempty(name)) \n if(len>1) txt=sprintf('%s\"%s\": [%s',padding1,checkname(name,varargin{:}),nl); end\nelse\n if(len>1) txt=sprintf('%s[%s',padding1,nl); end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n if(isoct)\n val=regexprep(item(e,:),'\\\\','\\\\');\n val=regexprep(val,'\"','\\\"');\n val=regexprep(val,'^\"','\\\"');\n else\n val=regexprep(item(e,:),'\\\\','\\\\\\\\');\n val=regexprep(val,'\"','\\\\\"');\n val=regexprep(val,'^\"','\\\\\"');\n end\n val=escapejsonstring(val);\n if(len==1)\n obj=['\"' checkname(name,varargin{:}) '\": ' '\"',val,'\"'];\n\tif(isempty(name)) obj=['\"',val,'\"']; end\n txt=sprintf('%s%s%s%s',txt,padding1,obj);\n else\n txt=sprintf('%s%s%s%s',txt,padding0,['\"',val,'\"']);\n end\n if(e==len) sep=''; end\n txt=sprintf('%s%s',txt,sep);\nend\nif(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2json(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\npadding1=repmat(ws.tab,1,level);\npadding0=repmat(ws.tab,1,level+1);\nnl=ws.newline;\nsep=ws.sep;\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))\n if(isempty(name))\n \ttxt=sprintf('%s{%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n else\n \ttxt=sprintf('%s\"%s\": {%s%s\"_ArrayType_\": \"%s\",%s%s\"_ArraySize_\": %s,%s',...\n padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\\s+',','),nl);\n end\nelse\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)\n numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\\[',''),']','');\n else\n numtxt=matdata2json(item,level+1,varargin{:});\n end\n if(isempty(name))\n \ttxt=sprintf('%s%s',padding1,numtxt);\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n \ttxt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n else\n \t txt=sprintf('%s\"%s\": %s',padding1,checkname(name,varargin{:}),numtxt);\n end\n end\n return;\nend\ndataformat='%s%s%s%s%s';\n\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n end\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsSparse_\": ','1', sep);\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([iy(:),data'],level+2,varargin{:}), nl);\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,data],level+2,varargin{:}), nl);\n else\n % General case, store row and column indices.\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([ix,iy,data],level+2,varargin{:}), nl);\n end\nelse\n if(isreal(item))\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json(item(:)',level+2,varargin{:}), nl);\n else\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayIsComplex_\": ','1', sep);\n txt=sprintf(dataformat,txt,padding0,'\"_ArrayData_\": ',...\n matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);\n end\nend\ntxt=sprintf('%s%s%s',txt,padding1,'}');\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2json(mat,level,varargin)\n\nws=struct('tab',sprintf('\\t'),'newline',sprintf('\\n'),'sep',sprintf(',\\n'));\nws=jsonopt('whitespaces_',ws,varargin{:});\ntab=ws.tab;\nnl=ws.newline;\n\nif(size(mat,1)==1)\n pre='';\n post='';\n level=level-1;\nelse\n pre=sprintf('[%s',nl);\n post=sprintf('%s%s]',nl,repmat(tab,1,level-1));\nend\n\nif(isempty(mat))\n txt='null';\n return;\nend\nfloatformat=jsonopt('FloatFormat','%.10g',varargin{:});\n%if(numel(mat)>1)\n formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];\n%else\n% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\\n')]];\n%end\n\nif(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)\n formatstr=[repmat(tab,1,level) formatstr];\nend\n\ntxt=sprintf(formatstr,mat');\ntxt(end-length(nl):end)=[];\nif(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)\n txt=regexprep(txt,'1','true');\n txt=regexprep(txt,'0','false');\nend\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],\\n['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\ntxt=[pre txt post];\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n\n%%-------------------------------------------------------------------------\nfunction newstr=escapejsonstring(str)\nnewstr=str;\nisoct=exist('OCTAVE_VERSION','builtin');\nif(isoct)\n vv=sscanf(OCTAVE_VERSION,'%f');\n if(vv(1)>=3.8) isoct=0; end\nend\nif(isoct)\n escapechars={'\\a','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},escapechars{i});\n end\nelse\n escapechars={'\\a','\\b','\\f','\\n','\\r','\\t','\\v'};\n for i=1:length(escapechars);\n newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\\\','\\\\\\\\'));\n end\nend\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/saveubjson.m", "content": "function json=saveubjson(rootname,obj,varargin)\n%\n% json=saveubjson(rootname,obj,filename)\n% or\n% json=saveubjson(rootname,obj,opt)\n% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)\n%\n% convert a MATLAB object (cell, struct or array) into a Universal \n% Binary JSON (UBJSON) binary string\n%\n% author: Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% created on 2013/08/17\n%\n% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $\n%\n% input:\n% rootname: the name of the root-object, when set to '', the root name\n% is ignored, however, when opt.ForceRootName is set to 1 (see below),\n% the MATLAB variable name will be used as the root name.\n% obj: a MATLAB object (array, cell, cell array, struct, struct array)\n% filename: a string for the file name to save the output UBJSON data\n% opt: a struct for additional options, ignore to use default values.\n% opt can have the following fields (first in [.|.] is the default)\n%\n% opt.FileName [''|string]: a file name to save the output JSON data\n% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D\n% array in JSON array format; if sets to 1, an\n% array will be shown as a struct with fields\n% \"_ArrayType_\", \"_ArraySize_\" and \"_ArrayData_\"; for\n% sparse arrays, the non-zero elements will be\n% saved to _ArrayData_ field in triplet-format i.e.\n% (ix,iy,val) and \"_ArrayIsSparse_\" will be added\n% with a value of 1; for a complex array, the \n% _ArrayData_ array will include two columns \n% (4 for sparse) to record the real and imaginary \n% parts, and also \"_ArrayIsComplex_\":1 is added. \n% opt.ParseLogical [1|0]: if this is set to 1, logical array elem\n% will use true/false rather than 1/0.\n% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single\n% numerical element will be shown without a square\n% bracket, unless it is the root object; if 0, square\n% brackets are forced for any numerical arrays.\n% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson\n% will use the name of the passed obj variable as the \n% root object name; if obj is an expression and \n% does not have a name, 'root' will be used; if this \n% is set to 0 and rootname is empty, the root level \n% will be merged down to the lower level.\n% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),\n% for example, if opt.JSON='foo', the JSON data is\n% wrapped inside a function call as 'foo(...);'\n% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson \n% back to the string form\n%\n% opt can be replaced by a list of ('param',value) pairs. The param \n% string is equivallent to a field in opt and is case sensitive.\n% output:\n% json: a binary string in the UBJSON format (see http://ubjson.org)\n%\n% examples:\n% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],... \n% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...\n% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...\n% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...\n% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...\n% 'SpecialData',[nan, inf, -inf]);\n% saveubjson('jsonmesh',jsonmesh)\n% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details\n%\n% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nif(nargin==1)\n varname=inputname(1);\n obj=rootname;\n if(isempty(varname)) \n varname='root';\n end\n rootname=varname;\nelse\n varname=inputname(2);\nend\nif(length(varargin)==1 && ischar(varargin{1}))\n opt=struct('FileName',varargin{1});\nelse\n opt=varargin2struct(varargin{:});\nend\nopt.IsOctave=exist('OCTAVE_VERSION','builtin');\nrootisarray=0;\nrootlevel=1;\nforceroot=jsonopt('ForceRootName',0,opt);\nif((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)\n rootisarray=1;\n rootlevel=0;\nelse\n if(isempty(rootname))\n rootname=varname;\n end\nend\nif((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)\n rootname='root';\nend\njson=obj2ubjson(rootname,obj,rootlevel,opt);\nif(~rootisarray)\n json=['{' json '}'];\nend\n\njsonp=jsonopt('JSONP','',opt);\nif(~isempty(jsonp))\n json=[jsonp '(' json ')'];\nend\n\n% save to a file if FileName is set, suggested by Patrick Rapin\nif(~isempty(jsonopt('FileName','',opt)))\n fid = fopen(opt.FileName, 'wb');\n fwrite(fid,json);\n fclose(fid);\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=obj2ubjson(name,item,level,varargin)\n\nif(iscell(item))\n txt=cell2ubjson(name,item,level,varargin{:});\nelseif(isstruct(item))\n txt=struct2ubjson(name,item,level,varargin{:});\nelseif(ischar(item))\n txt=str2ubjson(name,item,level,varargin{:});\nelse\n txt=mat2ubjson(name,item,level,varargin{:});\nend\n\n%%-------------------------------------------------------------------------\nfunction txt=cell2ubjson(name,item,level,varargin)\ntxt='';\nif(~iscell(item))\n error('input is not a cell');\nend\n\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item); % let's handle 1D cell first\nif(len>1) \n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) '[']; name=''; \n else\n txt='['; \n end\nelseif(len==0)\n if(~isempty(name))\n txt=[S_(checkname(name,varargin{:})) 'Z']; name=''; \n else\n txt='Z'; \n end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=struct2ubjson(name,item,level,varargin)\ntxt='';\nif(~isstruct(item))\n\terror('input is not a struct');\nend\ndim=size(item);\nif(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now\n item=reshape(item,dim(1),numel(item)/dim(1));\n dim=size(item);\nend\nlen=numel(item);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nfor j=1:dim(2)\n if(dim(1)>1) txt=[txt '[']; end\n for i=1:dim(1)\n names = fieldnames(item(i,j));\n if(~isempty(name) && len==1)\n txt=[txt S_(checkname(name,varargin{:})) '{']; \n else\n txt=[txt '{']; \n end\n if(~isempty(names))\n for e=1:length(names)\n\t txt=[txt obj2ubjson(names{e},getfield(item(i,j),...\n names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];\n end\n end\n txt=[txt '}'];\n end\n if(dim(1)>1) txt=[txt ']']; end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=str2ubjson(name,item,level,varargin)\ntxt='';\nif(~ischar(item))\n error('input is not a string');\nend\nitem=reshape(item, max(size(item),[1 0]));\nlen=size(item,1);\n\nif(~isempty(name)) \n if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end\nelse\n if(len>1) txt='['; end\nend\nisoct=jsonopt('IsOctave',0,varargin{:});\nfor e=1:len\n val=item(e,:);\n if(len==1)\n obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];\n\tif(isempty(name)) obj=['',S_(val),'']; end\n txt=[txt,'',obj];\n else\n txt=[txt,'',['',S_(val),'']];\n end\nend\nif(len>1) txt=[txt ']']; end\n\n%%-------------------------------------------------------------------------\nfunction txt=mat2ubjson(name,item,level,varargin)\nif(~isnumeric(item) && ~islogical(item))\n error('input is not an array');\nend\n\nif(length(size(item))>2 || issparse(item) || ~isreal(item) || ...\n isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))\n cid=I_(uint32(max(size(item))));\n if(isempty(name))\n \ttxt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];\n else\n if(isempty(item))\n txt=[S_(checkname(name,varargin{:})),'Z'];\n return;\n else\n \t txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];\n end\n end\nelse\n if(isempty(name))\n \ttxt=matdata2ubjson(item,level+1,varargin{:});\n else\n if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)\n numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\\[',''),']','');\n \ttxt=[S_(checkname(name,varargin{:})) numtxt];\n else\n \t txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];\n end\n end\n return;\nend\nif(issparse(item))\n [ix,iy]=find(item);\n data=full(item(find(item)));\n if(~isreal(item))\n data=[real(data(:)),imag(data(:))];\n if(size(item,1)==1)\n % Kludge to have data's 'transposedness' match item's.\n % (Necessary for complex row vector handling below.)\n data=data';\n end\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n end\n txt=[txt,S_('_ArrayIsSparse_'),'T'];\n if(size(item,1)==1)\n % Row vector, store only column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([iy(:),data'],level+2,varargin{:})];\n elseif(size(item,2)==1)\n % Column vector, store only row indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,data],level+2,varargin{:})];\n else\n % General case, store row and column indices.\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([ix,iy,data],level+2,varargin{:})];\n end\nelse\n if(isreal(item))\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson(item(:)',level+2,varargin{:})];\n else\n txt=[txt,S_('_ArrayIsComplex_'),'T'];\n txt=[txt,S_('_ArrayData_'),...\n matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];\n end\nend\ntxt=[txt,'}'];\n\n%%-------------------------------------------------------------------------\nfunction txt=matdata2ubjson(mat,level,varargin)\nif(isempty(mat))\n txt='Z';\n return;\nend\nif(size(mat,1)==1)\n level=level-1;\nend\ntype='';\nhasnegtive=(mat<0);\nif(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))\n if(isempty(hasnegtive))\n if(max(mat(:))<=2^8)\n type='U';\n end\n end\n if(isempty(type))\n % todo - need to consider negative ones separately\n id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);\n if(isempty(find(id)))\n error('high-precision data is not yet supported');\n end\n key='iIlL';\n\ttype=key(find(id));\n end\n txt=[I_a(mat(:),type,size(mat))];\nelseif(islogical(mat))\n logicalval='FT';\n if(numel(mat)==1)\n txt=logicalval(mat+1);\n else\n txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];\n end\nelse\n if(numel(mat)==1)\n txt=['[' D_(mat) ']'];\n else\n txt=D_a(mat(:),'D',size(mat));\n end\nend\n\n%txt=regexprep(mat2str(mat),'\\s+',',');\n%txt=regexprep(txt,';',sprintf('],['));\n% if(nargin>=2 && size(mat,1)>1)\n% txt=regexprep(txt,'\\[',[repmat(sprintf('\\t'),1,level) '[']);\n% end\nif(any(isinf(mat(:))))\n txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','\"$1_Inf_\"',varargin{:}));\nend\nif(any(isnan(mat(:))))\n txt=regexprep(txt,'NaN',jsonopt('NaN','\"_NaN_\"',varargin{:}));\nend\n\n%%-------------------------------------------------------------------------\nfunction newname=checkname(name,varargin)\nisunpack=jsonopt('UnpackHex',1,varargin{:});\nnewname=name;\nif(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))\n return\nend\nif(isunpack)\n isoct=jsonopt('IsOctave',0,varargin{:});\n if(~isoct)\n newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');\n else\n pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');\n pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');\n if(isempty(pos)) return; end\n str0=name;\n pos0=[0 pend(:)' length(name)];\n newname='';\n for i=1:length(pos)\n newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];\n end\n if(pos(end)~=length(name))\n newname=[newname str0(pos0(end-1)+1:pos0(end))];\n end\n end\nend\n%%-------------------------------------------------------------------------\nfunction val=S_(str)\nif(length(str)==1)\n val=['C' str];\nelse\n val=['S' I_(int32(length(str))) str];\nend\n%%-------------------------------------------------------------------------\nfunction val=I_(num)\nif(~isinteger(num))\n error('input is not an integer');\nend\nif(num>=0 && num<255)\n val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];\n return;\nend\nkey='iIlL';\ncid={'int8','int16','int32','int64'};\nfor i=1:4\n if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))\n val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];\n return;\n end\nend\nerror('unsupported integer');\n\n%%-------------------------------------------------------------------------\nfunction val=D_(num)\nif(~isfloat(num))\n error('input is not a float');\nend\n\nif(isa(num,'single'))\n val=['d' data2byte(num,'uint8')];\nelse\n val=['D' data2byte(num,'uint8')];\nend\n%%-------------------------------------------------------------------------\nfunction data=I_a(num,type,dim,format)\nid=find(ismember('iUIlL',type));\n\nif(id==0)\n error('unsupported integer array');\nend\n\n% based on UBJSON specs, all integer types are stored in big endian format\n\nif(id==1)\n data=data2byte(swapbytes(int8(num)),'uint8');\n blen=1;\nelseif(id==2)\n data=data2byte(swapbytes(uint8(num)),'uint8');\n blen=1;\nelseif(id==3)\n data=data2byte(swapbytes(int16(num)),'uint8');\n blen=2;\nelseif(id==4)\n data=data2byte(swapbytes(int32(num)),'uint8');\n blen=4;\nelseif(id==5)\n data=data2byte(swapbytes(int64(num)),'uint8');\n blen=8;\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];\n end\n data=['[' data(:)'];\nelse\n data=reshape(data,blen,numel(data)/blen);\n data(2:blen+1,:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction data=D_a(num,type,dim,format)\nid=find(ismember('dD',type));\n\nif(id==0)\n error('unsupported float array');\nend\n\nif(id==1)\n data=data2byte(single(num),'uint8');\nelseif(id==2)\n data=data2byte(double(num),'uint8');\nend\n\nif(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))\n format='opt';\nend\nif((nargin<4 || strcmp(format,'opt')) && numel(num)>1)\n if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))\n cid=I_(uint32(max(dim)));\n data=['$' type '#' I_a(dim,cid(1)) data(:)'];\n else\n data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];\n end\n data=['[' data];\nelse\n data=reshape(data,(id*4),length(data)/(id*4));\n data(2:(id*4+1),:)=data;\n data(1,:)=type;\n data=data(:)';\n data=['[' data(:)' ']'];\nend\n%%-------------------------------------------------------------------------\nfunction bytes=data2byte(varargin)\nbytes=typecast(varargin{:});\nbytes=bytes(:)';\n" }, { "path": "machine-learning-ex8/ex8/lib/jsonlab/varargin2struct.m", "content": "function opt=varargin2struct(varargin)\n%\n% opt=varargin2struct('param1',value1,'param2',value2,...)\n% or\n% opt=varargin2struct(...,optstruct,...)\n%\n% convert a series of input parameters into a structure\n%\n% authors:Qianqian Fang (fangq nmr.mgh.harvard.edu)\n% date: 2012/12/22\n%\n% input:\n% 'param', value: the input parameters should be pairs of a string and a value\n% optstruct: if a parameter is a struct, the fields will be merged to the output struct\n%\n% output:\n% opt: a struct where opt.param1=value1, opt.param2=value2 ...\n%\n% license:\n% BSD, see LICENSE_BSD.txt files for details \n%\n% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)\n%\n\nlen=length(varargin);\nopt=struct;\nif(len==0) return; end\ni=1;\nwhile(i<=len)\n if(isstruct(varargin{i}))\n opt=mergestruct(opt,varargin{i});\n elseif(ischar(varargin{i}) && i bestF1\n bestF1 = F1;\n bestEpsilon = epsilon;\n end\nend\n\nend\n" }, { "path": "machine-learning-ex8/ex8/submit.m", "content": "function submit()\n addpath('./lib');\n\n conf.assignmentSlug = 'anomaly-detection-and-recommender-systems';\n conf.itemName = 'Anomaly Detection and Recommender Systems';\n conf.partArrays = { ...\n { ...\n '1', ...\n { 'estimateGaussian.m' }, ...\n 'Estimate Gaussian Parameters', ...\n }, ...\n { ...\n '2', ...\n { 'selectThreshold.m' }, ...\n 'Select Threshold', ...\n }, ...\n { ...\n '3', ...\n { 'cofiCostFunc.m' }, ...\n 'Collaborative Filtering Cost', ...\n }, ...\n { ...\n '4', ...\n { 'cofiCostFunc.m' }, ...\n 'Collaborative Filtering Gradient', ...\n }, ...\n { ...\n '5', ...\n { 'cofiCostFunc.m' }, ...\n 'Regularized Cost', ...\n }, ...\n { ...\n '6', ...\n { 'cofiCostFunc.m' }, ...\n 'Regularized Gradient', ...\n }, ...\n };\n conf.output = @output;\n\n submitWithConfiguration(conf);\nend\n\nfunction out = output(partId, auxstring)\n % Random Test Cases\n n_u = 3; n_m = 4; n = 5;\n X = reshape(sin(1:n_m*n), n_m, n);\n Theta = reshape(cos(1:n_u*n), n_u, n);\n Y = reshape(sin(1:2:2*n_m*n_u), n_m, n_u);\n R = Y > 0.5;\n pval = [abs(Y(:)) ; 0.001; 1];\n yval = [R(:) ; 1; 0];\n params = [X(:); Theta(:)];\n if partId == '1'\n [mu sigma2] = estimateGaussian(X);\n out = sprintf('%0.5f ', [mu(:); sigma2(:)]);\n elseif partId == '2'\n [bestEpsilon bestF1] = selectThreshold(yval, pval);\n out = sprintf('%0.5f ', [bestEpsilon(:); bestF1(:)]);\n elseif partId == '3'\n [J] = cofiCostFunc(params, Y, R, n_u, n_m, ...\n n, 0);\n out = sprintf('%0.5f ', J(:));\n elseif partId == '4'\n [J, grad] = cofiCostFunc(params, Y, R, n_u, n_m, ...\n n, 0);\n out = sprintf('%0.5f ', grad(:));\n elseif partId == '5'\n [J] = cofiCostFunc(params, Y, R, n_u, n_m, ...\n n, 1.5);\n out = sprintf('%0.5f ', J(:));\n elseif partId == '6'\n [J, grad] = cofiCostFunc(params, Y, R, n_u, n_m, ...\n n, 1.5);\n out = sprintf('%0.5f ', grad(:));\n end \nend\n" }, { "path": "machine-learning-ex8/ex8/visualizeFit.m", "content": "function visualizeFit(X, mu, sigma2)\n%VISUALIZEFIT Visualize the dataset and its estimated distribution.\n% VISUALIZEFIT(X, p, mu, sigma2) This visualization shows you the \n% probability density function of the Gaussian distribution. Each example\n% has a location (x1, x2) that depends on its feature values.\n%\n\n[X1,X2] = meshgrid(0:.5:35); \nZ = multivariateGaussian([X1(:) X2(:)],mu,sigma2);\nZ = reshape(Z,size(X1));\n\nplot(X(:, 1), X(:, 2),'bx');\nhold on;\n% Do not plot if there are infinities\nif (sum(isinf(Z)) == 0)\n contour(X1, X2, Z, 10.^(-20:3:0)');\nend\nhold off;\n\nend" }, { "path": "readme.md", "content": "Machine Learning Exercises\n======\nCoursera上斯坦福大学的机器学习编程作业\n\n### 1. [machine-learning-ex1][1]\n - 线性回归\n - 梯度下降\n - 正规方程\n![machine-learning-ex1][1.1]\n\n\n### 2. [machine-learning-ex2][2]\n - 逻辑回归\n - 正则化\n![machine-learning-ex2][2.1]\n\n### 3. [machine-learning-ex3][3]\n - 多分类逻辑回归\n - 神经网络预测(权值weights已知)\n![machine-learning-ex3][3.1]\n\n### 4. [machine-learning-ex4][4]\n - bp神经网络识别手写数字\n![machine-learning-ex4][4.1]\n \n### 5. [machine-learning-ex5][5]\n - 学习曲线 \n - 验证集 \n![machine-learning-ex5][5.1]\n\n### 6. [machine-learning-ex6][6]\n - 支持向量SVM \n - 高斯核函数\n - 线性核函数\n - 垃圾邮件分类\n![machine-learning-ex6][6.1]\n \n### 7. [machine-learning-ex7][7]\n - k-means\n - 降维PCA（主成分分析）\n \n![machine-learning-ex7][7.1]\n\n### 8. [machine-learning-ex8][8]\n - 异常检测（高斯分布）\n - 协同过滤\n \n![machine-learning-ex8][8.1]\n\n \n [1]:https://github.com/lawlite19/MachineLearningEx/tree/master/machine-learning-ex1\n [1.1]: ./images/machine-learning-ex1_01.png \"machine-learning-ex1_01.png\"\n \n [2]:https://github.com/lawlite19/MachineLearningEx/tree/master/machine-learning-ex2\n [2.1]: ./images/machine-learning-ex2_01.png \"machine-learning-ex2_01.png\"\n \n [3]:https://github.com/lawlite19/MachineLearningEx/blob/master/machine-learning-ex3\n [3.1]: ./images/machine-learning-ex3_01.png \"machine-learning-ex3_01.png\"\n\n [4]:https://github.com/lawlite19/MachineLearningEx/blob/master/machine-learning-ex4\n [4.1]: ./images/machine-learning-ex4_01.png \"machine-learning-ex4_01.png\"\n \n [5]:https://github.com/lawlite19/MachineLearningEx/blob/master/machine-learning-ex5\n [5.1]: ./images/machine-learning-ex5_01.png \"machine-learning-ex5_01.png\"\n \n [6]:https://github.com/lawlite19/MachineLearningEx/blob/master/machine-learning-ex6\n [6.1]: ./images/machine-learning-ex6_01.png \"machine-learning-ex6_01.png\"\n \n [7]:https://github.com/lawlite19/MachineLearningEx/blob/master/machine-learning-ex7\n [7.1]: ./images/machine-learning-ex7_01.png \"machine-learning-ex7_01.png\"\n \n [8]:https://github.com/lawlite19/MachineLearningEx/blob/master/machine-learning-ex8\n [8.1]: ./images/machine-learning-ex8_01.png \"machine-learning-ex8_01.png\"\n" } ]