Specification of a data format (and schema) designed for APIs, aiming to replace JSON.

The KSON schema can be used to describe JSON files.

KSON tries solving all the issues of JSON:

JSON schema file using regex-like (*, +, [1-3]) syntax to express

/* description */
{
  /* comment */
  prop1: string,
  prop2: numeric,
  list1*: string
  list2+: {
    itemProp1: {}
    itemProp2?: string
  }
  listOfString: [string],
  listOfString2[]: string,
  yearDigits[2-4]: numeric,
  listOfObjects: [{
  }]
  prop3: {
    .?: string /* any key, optional, type string */
  },
  prop4: {
    .*: * /* 0 or N keys, any type each */
  },
  prop5: {
    .*: * /* 0 or N keys, any type each */
  },
  prop6: ** /* can be anything, any depth */,
  
  /* equivalent to 'prop7: *': */
  prop7,
  prop8: customType
  
  $types:{
    customType: numeric
  }
}

• /**/ : comment

• . : a(ny) key
• * : 0 or N
• + : 1 or N
• ? : optional (0 or 1)
• ``

• (type) : json primitive type
• ** : anything / any depth

KSON:

  prop1: string,
  prop2: numeric,
  list1*: string
  list2+: {
    itemProp1: {}
    itemProp2?: string
  }
  yearDigits: [2, 0, 1, 9]
}

• KSON is NOT JSON, i.e. can't be parsed by a JSON parser

• popularized because of JS front-ends
• quite intuitive syntax for Java/C programmers: {} = map/object, [] = list/array
• can be minified

• everything must be quoted
• loose primitive types: null, boolean, number, string (and object/array for structures).
• no comment possible inside JSON
• no (widely-used) schema (JSON schema is too verbose anyway)

• namespaces?

• verbose (repetition of start/end tags)

TODO