Full Code of mafintosh/jitson for AI

Repository: mafintosh/jitson
Branch: master
Commit: cd18c20fbe83
Files: 6
Total size: 8.0 KB

Directory structure:
gitextract_x_prs31a/

├── .gitignore
├── LICENSE
├── README.md
├── example.js
├── index.js
└── package.json

================================================
FILE CONTENTS
================================================

================================================
FILE: .gitignore
================================================
node_modules


================================================
FILE: LICENSE
================================================
The MIT License (MIT)

Copyright (c) 2018 Mathias Buus

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.


================================================
FILE: README.md
================================================
# jitson

Just-In-Time JSON.parse compiler

```
npm install jitson
```

Works by schema sampling the incoming data and if the schema is stable, it
will compile a fast parser for it using [turbo-json-parse](https://github.com/mafintosh/turbo-json-parse)

## Usage

``` js
const jitson = require('jitson')

// make an instance
const parse = jitson()

for (var i = 0; i < 10; i++) {
  console.log(parse(JSON.stringify({hello: 'world', number: Math.random()})))
}

// Check if the compiler found a matching schema.
// If so it is using an optimised parser to parse the JSON
console.log(parse.schema)
```

## API

#### `const parse = jitson(opts)`

Create a new JSON parser.

Options include

```js
{
  sampleInterval: 100 // sample the schema everytime we parse 100 objects
}
```

It keeps a small internal cache around of old schemas that is used to produce a better parser.
If the cache is empty it will sample right away as well.

Any additional options are forwarded to [turbo-json-parse](https://github.com/mafintosh/turbo-json-parse)
when it triggers a parser compilation.

It works the best if you try to only pass data to it that has a schema so make
an instance for each of your http endpoints for example

#### `const object = parse(src)`

Similar to JSON.parse. Will schema sample the input once in a while
to check if it has a stable schema. If so it'll optimise the parser.

If the optimised parser fails it will fallback to JSON.parse and resample more data in future. If the optimised parser keeps failing it'll increase the sampling interval, to not waste too much time sampling and compiling.

If you are parsing from Node.js buffers make sure to pass that as the `src`
instead of `toString()`ing it first as that will produce a faster parser
when compiling.

## License

MIT


================================================
FILE: example.js
================================================
const jitson = require('./')

// make an instance
const parse = jitson()

for (var i = 0; i < 10; i++) {
  console.log(parse(JSON.stringify({ hello: 'world', number: Math.random() })))
}

// Check if the compiler found a matching schema.
// If so it is using an optimised parser to parse the JSON
console.log(parse.schema)


================================================
FILE: index.js
================================================
const turbo = require('turbo-json-parse')

module.exports = jitson

function jitson (opts) {
  if (!opts) opts = {}

  const schemas = []
  const inputs = []
  const partial = !!opts.partial

  var sampleInterval = opts.sampleInterval || 100
  var tick = sampleInterval
  var compiled = null
  var failures = 0

  parsePartial.pointer = 0

  const result = partial ? parsePartial : parse
  result.recompiles = 0
  result.schema = null
  result.turbo = null
  return result

  function sample (src) {
    const data = JSON.parse(src)

    if (--tick > 0 && inputs.length === 8 && compiled) return data
    tick = sampleInterval

    if (inputs.length === 8) {
      inputs.shift()
      schemas.shift()
    }

    const sch = getSchema(data)
    if (!sch) return data

    inputs.push(src)
    schemas.push(sch)

    const schema = result.schema = joinSchemas(schemas)
    if (!schema) {
      if (++failures >= 7) {
        sampleInterval *= 2
      }
      return data
    }

    failures = 0
    opts.buffer = inputs.every(isBuffer)
    opts.defaults = false

    result.recompiles++
    compiled = result.turbo = turbo(schema, opts)
    return data
  }

  function parse (src) {
    try {
      if (compiled) return compiled(src, 0)
    } catch (err) {
      compiled = result.compiled = result.turbo = null
    }
    return sample(src)
  }

  function parsePartial (src, ptr) {
    try {
      if (compiled) {
        const data = compiled(src, ptr)
        parse.pointer = compiled.pointer
        return data
      }
    } catch (err) {}
    throw new Error('not impl')
  }
}

function isBuffer (buf) {
  return typeof buf !== 'string'
}

function getSchema (object) {
  const type = Array.isArray(object) ? 'array' : typeof object
  switch (type) {
    case 'number':
    case 'string':
    case 'boolean': {
      return { type }
    }

    case 'object': {
      if (!object) return null
      const res = { type: 'object', ordered: true, properties: {} }
      for (const key of Object.keys(object)) {
        const sch = res.properties[key] = getSchema(object[key])
        if (!sch) return null
      }
      return res
    }

    case 'array': {
      const res = { type: 'array', items: null }
      if (object.length) {
        res.items = getSchema(object[0])
        if (!res.items) return null
      }
      return res
    }
  }

  return null
}

function joinSchemas (samples) {
  if (!samples.length) return null
  const s = samples[0]

  if (samples.length === 1) return s

  for (const sample of samples) {
    if (!sample || sample.type !== s.type) return null
  }

  switch (s.type) {
    case 'number':
    case 'string':
    case 'boolean':
    case 'integer': {
      return { type: s.type }
    }

    case 'object': {
      const isOrdered = create()
      const res = { type: 'object', ordered: true, properties: {} }
      const props = {}

      for (const sample of samples) {
        if (res.ordered && !isOrdered(Object.keys(sample.properties))) {
          res.ordered = false
        }
        for (const prop of Object.keys(sample.properties)) {
          if (!props[prop]) props[prop] = []
          props[prop].push(sample.properties[prop])
        }
      }
      for (const prop of Object.keys(props)) {
        const val = res.properties[prop] = joinSchemas(props[prop])
        if (!val) return null
      }

      return res
    }

    case 'array': {
      const items = []
      for (const sample of samples) {
        if (sample.items) items.push(sample.items)
      }
      const res = { type: 'array', items: joinSchemas(items) }
      return items.length && !res.items ? null : res
    }
  }
}

function upsert (map, key, val) {
  if (map.has(key)) return map.get(key)
  map.set(key, val)
  return val
}

function create () {
  const lts = new Uint32Array(32)
  const gts = new Uint32Array(32)
  const tokens = new Map()

  return isOrdered

  function isOrdered (keys) {
    const ids = new Uint8Array(keys.length)

    for (var i = 0; i < keys.length; i++) {
      const id = upsert(tokens, keys[i], tokens.size)
      if (id >= 32) return false
      ids[i] = id
    }

    var l = 0
    var g = 0

    for (const id of ids) {
      l |= (1 << id)
    }

    for (const id of ids) {
      if (lts[id] & g) return false
      if (gts[id] & l) return false

      lts[id] |= l
      gts[id] |= g

      l &= ~(1 << id)
      g |= (1 << id)
    }

    return true
  }
}


================================================
FILE: package.json
================================================
{
  "name": "jitson",
  "version": "1.0.0",
  "description": "Just-In-Time JSON.parse compiler",
  "main": "index.js",
  "dependencies": {
    "turbo-json-parse": "^2.2.1"
  },
  "devDependencies": {
    "standard": "^12.0.0"
  },
  "scripts": {
    "test": "standard"
  },
  "repository": {
    "type": "git",
    "url": "https://github.com/mafintosh/jitson.git"
  },
  "author": "Mathias Buus (@mafintosh)",
  "license": "MIT",
  "bugs": {
    "url": "https://github.com/mafintosh/jitson/issues"
  },
  "homepage": "https://github.com/mafintosh/jitson"
}