jpack v0.0.1

Jpack is on its very early stage, it's not ready for production. Don't make PR for it, only issue will be resolved.

Overview

Jpack should be language neutral, extendable and fast. And it should support event IE6. It is based on the subset of json-schema.

Terms

JsonType

Same as the json.org spec. It represents one of these types: String, Number, Boolean, Object, Array or null.

SimpleValue

It represents one of these types: String, Number, Boolean or null.

SimpleArray

It's an array that contains only SimpleValue or SimpleArray. Such as [1, 'str', [true, null]] is a SimpleArray, but [1, {a: true}] isn't.

SimpleType

It is an SimpleValue or SimpleArray.

Pack Process

0. Serializable to SimpleType

   Such as 10 to 10 or { a: 1, b: { d: 3, e: 'test' }, c: 2 } to [1, [3, 'test'], 2].

1. SimpleType to Spack

   Such as [1, [3, 'test'], 2] to [1|[3|{test}]].

Spack Encoding

Convert SimpleType into a plain string. Spack is simpler and faster than JSON. The result string will be more compact and smaller than JSON. Thus it's not human readable.

Null

\0

Boolean

• true to \1
• false to \2

String:

• string sign: \3

Array

• [ to \4
• ] to \5
• , to \6

Such as: [1, 2, 3] to \41\62\63\5.

Number:

Number will be convert to base 119.

• minus sign: \7
• decimal point: \8
• digit: \9 ~ \127

Spack won't support something like 10E+10.

Hash Sum

Hash sum of the schema will be placed before the pack:

| 1 | 2 | ... | | hash | pack |

The hash is 16bit

It will make sure the decoding schema is on the right version.

API

• genSchema(val)

  Generate simple schema from a sample object. Use it if you feel boring to craft a schema by hand.

  • param: val { JsonType }

  • return: { JsonType }

• pack(val, schema)

  Serialize anything to a data pack.

  • param: val { JsonType }

  • param: schema { JsonType }

  • return: { ArrayBuffer }

• unpack(data, schema)

  Deserialize the data pack to the origin value.

  • param: data { ArrayBuffer }

  • param: schema { JsonType }

  • return: { JsonType }

• types

  Use it to extend the data type that jpack supports. By default it implements the Date type.

  • type: { Object }

    Each type should implement two functions. One is serialize: (val) -> SimpleType. Another one is parse: (val) -> any.

  • example:

    This will let jpack support custom type Size.

    var Size = function Size (w, h) {
        this.w = w
        this.h = h
    }
    Size.prototype.area = function () {
        return this.w * this.h
    }
    
    jpack.types['Size'] = {
        serialize: function (s) {
            return [s.w, s.h]
        },
        parse: function (s) {
            return new Size(s[0], s[1])
        }
    }
    
    console.log(Size.name) // => "Size"
    
    var obj = {
        'a': new Size(1, 2)
    }
    
    var schema = jpack.genSchema(obj)
    
    jpack.pack(obj, schema)