Home

Awesome

TypeType · Build Status Coverage Status

TypeType is designed to generate complex typescript type with ease.

Usage

> npm i -D @mistlog/typetype

CLI

example: typetype-examples/package.json

typetype build <dir>: build all *.type files in <dir>
typetype build -w <dir>: watch all *.type files in <dir>
typetype clean <dir>: remove all generated *.ts files in <dir>
typetype debug <file>: build <file> in debug mode(backtrace will be available)

API

example: typetype-examples/index.ts

import { transform } from "@mistlog/typetype";

const input = `
    type function TypeName = (T) => ^{
        if(T extends string) {
            return "string"
        } else {
            return "number"
        }
    }
`;
const output = transform(input).code;
console.log(output);
// output: type TypeName<T> = T extends string ? "string" : "number";

Debug mode:

const output = transform(input, { debug: true }).code;

when debug is true, backtrace will be available:

Expected end of input but ";" found.
x 1:11-1:11 MultiLineComment
| type a = 1;
|           ^
o 1:11-1:11 _
| type a = 1;
|           ^
x 1:11-1:11 TypeFunctionDeclaration
| type a = 1;
...
|/ /
| |
|/
o 1:1-1:11 TypeFile
  type a = 1;

Examples

In the url-parser example, function parseURL will be translated to generic type parseURL<text> in typescript:

// input
type function parseURL = (text) => ^{
    if (parseProtocol<text> extends [infer protocol, infer rest]) {
        return {
            protocol,
            rest
        }
    } else {
        return never
    }
}

// output
type parseURL<text> = parseProtocol<text> extends [infer protocol, infer rest]
  ? {
      protocol: protocol;
      rest: rest;
    }
  : never;

Conditional type is presented in this way:

^{ if ... else ...}

It can be nested so that the logic is clear:

type function _isNumberString = (text) => ^{
    if(text extends "") {
        return true
    } else if(text extends `${infer digit}${infer rest}`) {
        return ^{
            if(digit extends Digit) {
                return _isNumberString<rest>
            } else {
                return false
            }
        }
    } else {
        return false
    }
}

we can use js to create types:

type tuple = ["tesla", "model 3", "model X", "model Y"]

type result = ''' "use js"
   return $.use("tuple")
      .tupleToObject()
      .omit(key => !key.startsWith("model"))
      .type();
'''

generated:

type tuple = ["tesla", "model 3", "model X", "model Y"];
type result = {
  "model 3": "model 3";
  "model X": "model X";
  "model Y": "model Y";
};

Syntax

Basic type

type a = never
type b = number
type c = string

type value = 1
type bool = true
type tuple = [1, 2, 3]
type array = string[][]

type str = "abc"
type template = `value is: ${value}`

type obj = { a: 1, b: "abc", c: [1, 2] }
type valueDeep = obj["c"][1]

type keys = keyof { readonly a?: 1, b: 2 }

Union and Intersection

We use union [...] or | [...] to denote union type.

type u1 = union [0, 1, 2]
type u2 = | [0, 1, 2]

Because an intersection type combines multiple types into one, we use combine [...] or & [...] for intersection type:

type i1 = combine [{ a: 1 }, { b: 2 }]
type i2 = & [{ a: 1 }, { b: 2 }]

Function type

type f1 = type () => void
type f2 = type (a:number, b:string) => number
type f3 = type () => type (a:number, b:string) => void

Conditional type

/*
  type conditional = 1 extends string ? "string" : "number"
*/
type conditional = ^{
    if(1 extends string) {
        return "string"
    } else {
        return "number"
    }
}

nested:

/*
  type conditional2 = 1 extends string ? "string" : 1 extends 1 ? "is 1" : "not 1";
*/
type conditional2 = ^{
    if(1 extends string) {
        return "string"
    } else {
        return ^{
            if(1 extends 1) {
                return "is 1"
            } else {
                return "not 1"
            }
        }
    }
}

Mapped type

/* type mapped1 = { [K in Keys]: boolean } */
type mapped1 = ^{
    for(K in Keys) {
        return {
            key: K,
            value: boolean
        }
    }
}

/* type mapped2 = { [K in Keys as `get${K}`]: () => string } */
type mapped2 = ^{
    for(K in Keys) {
        return {
            key: `get${K}`,
            value: type () => string
        }
    }
}

Generic

/* export type Foo<T> = T extends { a: infer U; b: infer U; } ? U : never */
type function Foo = (T) => ^{
    if(T extends {a: infer U, b: infer U}) {
        return U
    } else {
        return never
    }
}

With constraint:

/* export type MyPick<T, Keys extends keyof T> = { [K in Keys]: T[K] } */
export type function MyPick = (T, Keys extends keyof T) => ^{
    for(K in Keys) {
        return {
            key: K,
            value: T[K]
        }
    }
}

Object spread

Object spread syntax can be used, and it will be translated to object$assign<{}, [...]>:

export type function parseURL = (text) => ^{
    if (parseProtocol<text> extends [infer protocol, infer rest]) {
        return {
            protocol,
            ...parseAuthority<rest>
        }
    } else {
        return never
    }
}

as long as object$assign is available globally, this works fine:

export type parseURL<text> = parseProtocol<text> extends [infer protocol, infer rest] ? object$assign<{}, [{
  protocol: protocol;
}, parseAuthority<rest>]> : never;

you can polyfill it using type lib such as ts-toolbelt, for example: polyfill/global.d.ts.

How it works?

It's AST -> AST transformation.

We use react-peg to write parser, as you can see in ./src/parser/expression, generator is even simpler than parser, in ./src/generator/generator, typetype AST is used to generate corresponding babel AST.

License

This project is MIT licensed.