JSONTokenizer is a project design to demonstrate parsing JSON using Swift's awesome regular expression support. It might also be useful if you wish to handle broken JSON when you don't have a choice. JSONTokenizer leaves as much as it can for you to deal with yourself. Good Luck!

Examples

Normal object

{
  "Hello" : "World"
}

.object([(key: .string("Hello"), value: .string("World"))])

Weird object

JSONTokenizer doesn't care that an object's key is a quoted string, just that it's a valid JSONToken. Therefore the following is valid:

{
  ["Hello"] : "World"
}

.object([(key: .array([.string("Hello")]), value: .string("World"))])

Null, booleans and numbers

JSONTokenizer doesn't handle null, booleans and numbers, it just captures them as barewords for you to handle as you wish. This avoids any preprocessing Swift might do (ie, treat 0 and 0.0 as identical)

[
  null,
  true,
  10e-30,
  0,
  0.0
]

.array([.bareword("null"), .bareword("true"), .bareword("10e-30"), .bareword("0"), .bareword("0.0")])

Inside a regular expression

JSONTokenizer can also be used directly in a regular expression as it conforms to RegexComponent.

let token = Reference(JSONToken.self)
let regex = Regex {
    "Hello"
    OneOrMore(.whitespace)
    Capture(as: token) {
        JSONTokenizer()
    }
    OneOrMore(.whitespace)
    "World"
}
let input = """
Hello ["small", "big"] World
"""
let result = try regex.wholeMatch(in: input)
print(result![token])

.array([.string("small"), .string("big")])