Type system parsers

In TypeScript we have Literal Types, Union Types, Template Literal Types, Generics and Conditional Types.

This allows us to build parsers within TypeScript’s type system!

This repository provides a date validator and a JSON parser that are implemented completely within TypeScript’s type system.

Date validator

src/date/IsDateString.ts defines the type IsDateString<T> which equals to type true if T is a string literal type of a valid date in format yyyy-mm-dd:

type T1 = IsDateString<'2022-01-16'>; // true
type T2 = IsDateString<'2022-02-35'>; // false
type T3 = IsDateString<'2024-02-29'>; // true, supports leap years
type T4 = IsDateString<'2023-02-29'>; // false
type T5 = IsDateString<'2022-01-01' | '2022-12-31'>; // true, every alternative is valid
type T6 = IsDateString<'2022-01-00' | '2022-12-32'>; // false, every alternative is invalid
type T7 = IsDateString<'2022-01-01' | '2022-12-32'>; // boolean, there are both valid and invalid alternatives

How it works

1. Build a deterministic finite-state machine which accepts a valid date:

   

2. For each state define a type with a type variable that takes the rest of the input string. The implicit error state is represented by the type false. The state “1st digit even” for example looks like this:

   type FirstDigitEven<T extends string> =
      T extends `${'0' | '4' | '8'}${infer R}`
      ? CenturyDivisibleByFour<R>
      : T extends `${'1' | '2' | '3' | '5' | '6' | '7' | '9'}${infer R}`
      ? CenturyNotDivisibleByFour<R>
      : false;

3. IsDateString is just the initial state:

   export type IsDateString<T extends string> = Empty<T>;

4. Write unit tests, e. g.:

   { const test: EqualTypes<IsDateString<'2022-01-01'>, true> = true; }
   { const test: EqualTypes<IsDateString<'2022-01-00'>, false> = true; }

JSON parser

src/json/JsonParser.ts defines the type JsonParser<T> which implements a JSON parser. You may have a look at src/json/JsonParser.test.ts to see how to use it.

TODO: describe how it works