JsExt

Additional functions for JavaScript to build strong applications.

Import

The recommended way is to only import the ones that are needed:

// Node.js, Bun, Deno (JSR) and Cloudflare Workers
import _try from "@ayonli/jsext/try";
import func from "@ayonli/jsext/func";
// ...

// Deno (URL)
import _try from "https://lib.deno.dev/x/ayonli_jsext@latest/try.ts";
import func from "https://lib.deno.dev/x/ayonli_jsext@latest/func.ts";
// ...

// Browsers (ESM)
import _try from "https://ayonli.github.io/jsext/esm/try.js";
import func from "https://ayonli.github.io/jsext/esm/func.js";
// ...

There is also a bundled version that can be loaded via a <script> tag in the browser.

<script src="https://ayonli.github.io/jsext/bundle/jsext.js">
    // this will also include the sub-modules and augmentations
</script>

Major Functions

  • _try Calls a function safely and return errors when captured.
  • func Declares a function along with a defer keyword, inspired by Golang.
  • wrap Wraps a function for decorator pattern but keep its signature.
  • mixin Declares a class that combines all methods from the base classes.
  • throttle Throttles function calls for frequent access.
  • debounce Debounces function calls for frequent access.
  • queue Handles tasks sequentially and prevent concurrency conflicts.
  • lock Provides mutual exclusion for concurrent operations.
  • read Makes any streaming source readable via for await ... of ... syntax.
  • readAll Reads all streaming data at once.
  • chan Creates a channel that transfers data across routines, even across multiple threads, inspired by Golang.
  • parallel Runs functions in parallel threads and take advantage of multi-core CPUs, inspired by Golang.
  • run Runs a script in another thread and abort at any time.
  • deprecate Marks a function as deprecated and emit warnings when it is called.

Categories

Each of these modules includes specific functions and classes for their target categories:

  • array Functions for dealing with arrays.
  • async Functions for async/promise context handling.
    • Historically, this module was named promise, but that name has been deprecated.
  • bytes Functions for dealing with byte arrays (Uint8Array).
    • Historically, this module was named uint8array, but that name has been deprecated.
  • class Functions for dealing with classes.
  • collections Additional collection data types.
  • dialog (Experimental) Asynchronous dialog functions for both browsers and terminals.
  • error Functions for converting errors to/from other types of objects.
  • json Functions for parsing JSONs to specific structures.
  • math Functions for mathematical calculations.
  • number Functions for dealing with numbers.
  • object Functions for dealing with objects.
  • path (Experimental) Platform-independent utility functions for dealing with system paths and URLs.
  • string Functions for dealing with strings.

Types

  • AsyncFunction This is the very constructor/class of all async functions.
  • AsyncFunctionConstructor
  • AsyncGeneratorFunction This is the very constructor/class of all async generator functions.
  • Constructor<T> This interface represents the constructor/class of the given T type.
  • RealArrayLike<T> A real-array-like object is an array-like object with a slice method.
  • TypedArray See https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/TypedArray
  • Optional<T, K extends keyof T> Constructs a type by making the specified keys optional.
  • Ensured<T, K extends keyof T> Constructs a type by ensuring the specified keys are required.

Augmentation

Check augment.md.

API References

_try

declare function _try<E = unknown, R = any, A extends any[] = any[]>(
  fn: (...args: A) => R,
  ...args: A
): [E, R];
declare function _try<E = unknown, R = any, A extends any[] = any[]>(
  fn: (...args: A) => Promise<R>,
  ...args: A
): Promise<[E, R]>;

Invokes a regular function or an async function and renders its result in an [err, res] tuple.

Example (regular function)

import _try from "@ayonli/jsext/try";

const [err, res] = _try(() => {
  // do something that may fail
});

Example (async function)

import _try from "@ayonli/jsext/try";
import axios from "axios";

let [err, res] = await _try(async () => {
  return await axios.get("https://example.org");
});

if (err) {
  res = (err as any)["response"];
}
declare function _try<E = unknown, R = any>(job: Promise<R>): Promise<[E, R]>;

Resolves a promise and renders its result in an [err, res] tuple.

Example (promise)

import _try from "@ayonli/jsext/try";
import axios from "axios";

let [err, res] = await _try(axios.get("https://example.org"));

if (err) {
  res = (err as any)["response"];
}
declare function _try<
  E = unknown,
  T = any,
  A extends any[] = any[],
  TReturn = any,
  TNext = unknown,
>(
  fn: (...args: A) => Generator<T, TReturn, TNext>,
  ...args: A
): Generator<[E, T], [E, TReturn], TNext>;
declare function _try<
  E = unknown,
  T = any,
  A extends any[] = any[],
  TReturn = any,
  TNext = unknown,
>(
  fn: (...args: A) => AsyncGenerator<T, TReturn, TNext>,
  ...args: A
): AsyncGenerator<[E, T], [E, TReturn], TNext>;

Invokes a generator function or an async generator function and renders its yield value and result in an [err, val] tuple.

Example (generator function)

import _try from "@ayonli/jsext/try";

const iter = _try(function* () {
  // do something that may fail
});

for (const [err, val] of iter) {
  if (err) {
    console.error("something went wrong:", err);
  } else {
    console.log("current value:", val);
  }
}

Example (async generator function)

import _try from "@ayonli/jsext/try";

const iter = _try(async function* () {
  // do something that may fail
});

for await (const [err, val] of iter) {
  if (err) {
    console.error("something went wrong:", err);
  } else {
    console.log("current value:", val);
  }
}
declare function _try<E = unknown, T = any, TReturn = any, TNext = unknown>(
  gen: Generator<T, TReturn, TNext>,
): Generator<[E, T], [E, TReturn], TNext>;
declare function _try<E = unknown, T = any, TReturn = any, TNext = unknown>(
  gen: AsyncGenerator<T, TReturn, TNext>,
): AsyncGenerator<[E, T], [E, TReturn], TNext>;

Resolves a generator or an async generator and renders its yield value and result in an [err, val] tuple.

Example (generator)

import _try from "@ayonli/jsext/try";
import { range } from "@ayonli/jsext/number";

const iter = range(1, 10);

for (const [err, val] of _try(iter)) {
  if (err) {
    console.error("something went wrong:", err);
  } else {
    console.log("current value:", val);
  }
}

Example (async generator)

import _try from "@ayonli/jsext/try";

async function* gen() {
  // do something that may fail
}

for await (const [err, val] of _try(gen())) {
  if (err) {
    console.error("something went wrong:", err);
  } else {
    console.log("current value:", val);
  }
}

func

declare function func<T, R = any, A extends any[] = any[]>(
  fn: (this: T, defer: (cb: () => void) => void, ...args: A) => R,
): (this: T, ...args: A) => R;

Inspired by Golang, creates a function that receives a defer keyword which can be used to carry deferred jobs that will be run after the main function is complete.

Multiple calls of the defer keyword is supported, and the callbacks are called in the LIFO order. Callbacks can be async functions if the main function is an async function or an async generator function, and all the running procedures will be awaited.

Example

import func from "@ayonli/jsext/func";
import * as fs from "node:fs/promises";

export const getVersion = func(async (defer) => {
  const file = await fs.open("./package.json", "r");
  defer(() => file.close());

  const content = await file.readFile("utf8");
  const pkg = JSON.parse(content);

  return pkg.version as string;
});

wrap

declare function wrap<T, Fn extends (this: T, ...args: any[]) => any>(
  fn: Fn,
  wrapper: (this: T, fn: Fn, ...args: Parameters<Fn>) => ReturnType<Fn>,
): Fn;

Wraps a function inside another function and returns a new function that copies the original function's name and other properties.

Example

import wrap from "@ayonli/jsext/wrap";

function log(text: string) {
  console.log(text);
}

const show = wrap(log, function (fn, text) {
  return fn.call(this, new Date().toISOString() + " " + text);
});

console.log(show.name); // log
console.log(show.length); // 1
console.assert(show.toString() === log.toString());

mixin

import { Constructor } from "@ayonli/jsext";
import { UnionToIntersection } from "@ayonli/jsext/mixin";

declare function mixin<T extends Constructor<any>, M extends any[]>(
  base: T,
  ...mixins: { [X in keyof M]: Constructor<M[X]> }
): T & Constructor<UnionToIntersection<FlatArray<M, 1>>>;
declare function mixin<T extends Constructor<any>, M extends any[]>(
  base: T,
  ...mixins: M
): T & Constructor<UnionToIntersection<FlatArray<M, 1>>>;

Creates a class that combines all methods from the given base class and mixin classes.

Example

import mixin from "@ayonli/jsext/mixin";
import { isSubclassOf } from "@ayonli/jsext/class";

class Log {
  log(text: string) {
    console.log(text);
  }
}

class View {
  display(data: Record<string, any>[]) {
    console.table(data);
  }
}

class Controller extends mixin(View, Log) {
  constructor(readonly topic: string) {
    super();
  }
}

const ctrl = new Controller("foo");
ctrl.log("something is happening");
ctrl.display([{ topic: ctrl.topic, content: "something is happening" }]);

console.assert(isSubclassOf(Controller, View));
console.assert(!isSubclassOf(Controller, Log));

throttle

declare function throttle<I, Fn extends (this: I, ...args: any[]) => any>(
  handler: Fn,
  duration: number,
): Fn;
declare function throttle<I, Fn extends (this: I, ...args: any[]) => any>(
  handler: Fn,
  options: {
    duration: number;
    /**
     * Use the throttle strategy `for` the given key, this will keep the
     * result in a global cache, binding new `handler` function for the same
     * key will result in the same result as the previous, unless the
     * duration has passed. This mechanism guarantees that both creating the
     * throttled function in function scopes and overwriting the handler are
     * possible.
     */
    for?: any;
    /**
     * When turned on, respond with the last cache (if available)
     * immediately, even if it has expired, and update the cache in the
     * background.
     */
    noWait?: boolean;
  },
): Fn;

Creates a throttled function that will only be run once in a certain amount of time.

If a subsequent call happens within the duration (in milliseconds), the previous result will be returned and the handler function will not be invoked.

Example

import throttle from "@ayonli/jsext/throttle";
import { sleep } from "@ayonli/jsext/async";

const fn = throttle((input: string) => input, 1_000);
console.log(fn("foo")); // foo
console.log(fn("bar")); // foo

await sleep(1_000);
console.log(fn("bar")); // bar

Example (with key)

import throttle from "@ayonli/jsext/throttle";
import { sleep } from "@ayonli/jsext/async";

const out1 = await throttle(() => Promise.resolve("foo"), {
  duration: 1_000,
  for: "example",
})();
console.log(out1); // foo

const out2 = await throttle(() => Promise.resolve("bar"), {
  duration: 1_000,
  for: "example",
})();
console.log(out2); // foo

await sleep(1_000);
const out3 = await throttle(() => Promise.resolve("bar"), {
  duration: 1_000,
  for: "example",
})();
console.log(out3); // bar

debounce

declare function debounce<I, T, R>(
  handler: (this: I, data: T) => R | Promise<R>,
  delay: number,
  reducer?: (prev: T, data: T) => T,
): (this: I, data: T) => Promise<R>;
declare function debounce<I, T, R>(
  handler: (this: I, data: T) => R | Promise<R>,
  options: {
    delay: number;
    /**
     * Use the debounce strategy `for` the given key, this will keep the
     * debounce context in a global registry, binding new `handler` function
     * for the same key will override the previous settings. This mechanism
     * guarantees that both creating the debounced function in function
     * scopes and overwriting the handler are possible.
     */
    for?: any;
  },
  reducer?: (prev: T, data: T) => T,
): (this: I, data: T) => Promise<R>;

Creates a debounced function that delays invoking handler until after delay duration (in milliseconds) have elapsed since the last time the debounced function was invoked.

If a subsequent call happens within the delay duration (in milliseconds), the previous call will be canceled and it will result in the same return value as the new call's.

Optionally, we can provide a reducer function to merge data before processing so multiple calls can be merged into one.

Example

import debounce from "@ayonli/jsext/debounce";
import { sleep } from "@ayonli/jsext/async";

let count = 0;

const fn = debounce((obj: { foo?: string; bar?: string }) => {
  count++;
  return obj;
}, 1_000);

const [res1, res2] = await Promise.all([
  fn({ foo: "hello", bar: "world" }),
  sleep(100).then(() => fn({ foo: "hi" })),
]);

console.log(res1); // { foo: "hi" }
console.log(res2); // { foo: "hi" }
console.log(count); // 1

Example (with reducer)

import debounce from "@ayonli/jsext/debounce";

const fn = debounce(
  (obj: { foo?: string; bar?: string }) => {
    return obj;
  },
  1_000,
  (prev, current) => {
    return { ...prev, ...current };
  },
);

const [res1, res2] = await Promise.all([
  fn({ foo: "hello", bar: "world" }),
  fn({ foo: "hi" }),
]);

console.log(res1); // { foo: "hi", bar: "world" }
console.assert(res2 === res1);

Example (with key)

import debounce from "@ayonli/jsext/debounce";

const key = "unique_key";
let count = 0;

const [res1, res2] = await Promise.all([
  debounce(
    async (obj: { foo?: string; bar?: string }) => {
      count += 1;
      return await Promise.resolve(obj);
    },
    { delay: 5, for: key },
    (prev, data) => {
      return { ...prev, ...data };
    },
  )({ foo: "hello", bar: "world" }),

  debounce(
    async (obj: { foo?: string; bar?: string }) => {
      count += 2;
      return await Promise.resolve(obj);
    },
    { delay: 5, for: key },
    (prev, data) => {
      return { ...prev, ...data };
    },
  )({ foo: "hi" }),
]);

console.log(res1); // { foo: "hi", bar: "world" }
console.assert(res1 === res2);
console.assert(count === 2);

queue

import type { Queue } from "@ayonli/jsext/queue";

declare function queue<T>(
  handler: (data: T) => Promise<void>,
  bufferSize?: number,
): Queue<T>;

Processes data sequentially by the given handler function and prevents concurrency conflicts, it returns a Queue instance that we can push data into.

bufferSize is the maximum capacity of the underlying channel, once reached, the push operation will block until there is new space available. By default, this option is not set and use a non-buffered channel instead.

Example

import queue from "@ayonli/jsext/queue";

const list: string[] = [];
const q = queue(async (str: string) => {
  await Promise.resolve(null);
  list.push(str);
});

q.onError((err) => {
  console.error(err);
});

await q.push("foo");
await q.push("foo");

console.log(list.length);
q.close();
// output:
// 2

lock

import type { Mutex } from "@ayonli/jsext/lock";

declare function lock(key: any): Promise<Mutex.Lock<undefined>>;

Acquires a mutex lock for the given key in order to perform concurrent operations and prevent conflicts.

If the key is currently being locked by other coroutines, this function will block until the lock becomes available again.

Example

import lock from "@ayonli/jsext/lock";
import func from "@ayonli/jsext/func";

const key = "unique_key";

export const concurrentOperation = func(async (defer) => {
  const ctx = await lock(key);
  defer(() => ctx.unlock()); // don't forget to unlock

  // This block will never be run if there are other coroutines holding
  // the lock.
  //
  // Other coroutines trying to lock the same key will also never be run
  // before `unlock()`.
});

Other than using the lock() function, we can also use new Mutex() to create a mutex instance that holds some shared resource which can only be accessed by one coroutine at a time.

Example

import { Mutex } from "@ayonli/jsext/lock";
import func from "@ayonli/jsext/func";
import { random } from "@ayonli/jsext/number";
import { sleep } from "@ayonli/jsext/async";

const mutex = new Mutex(1);

const concurrentOperation = func(async (defer) => {
  const shared = await mutex.lock();
  defer(() => shared.unlock()); // don't forget to unlock

  const value1 = shared.value;

  await otherAsyncOperations();

  shared.value += 1;
  const value2 = shared.value;

  // Without mutex lock, the shared value may have been modified by other
  // calls during `await otherAsyncOperation()`, and the following
  // assertion will fail.
  console.assert(value1 + 1 === value2);
});

async function otherAsyncOperations() {
  await sleep(100 * random(1, 10));
}

await Promise.all([
  concurrentOperation(),
  concurrentOperation(),
  concurrentOperation(),
  concurrentOperation(),
]);

read

declare function read<I extends AsyncIterable<any>>(iterable: I): I;
declare function read<T>(stream: ReadableStream<T>): AsyncIterable<T>;
declare function read(
  es: EventSource,
  options?: { event?: string },
): AsyncIterable<string>;
declare function read<T extends Uint8Array | string>(
  ws: WebSocket,
): AsyncIterable<T>;
declare function read<T>(target: EventTarget, eventMap?: {
  message?: string;
  error?: string;
  close?: string;
}): AsyncIterable<T>;
declare function read<T>(target: NodeJS.EventEmitter, eventMap?: {
  data?: string;
  error?: string;
  close?: string;
}): AsyncIterable<T>;

Wraps a source as an AsyncIterable object that can be used in the for await...of... loop for reading streaming data.

Example (ReadableStream)

import read from "@ayonli/jsext/read";

const res = new Response("Hello, World!");

for await (const chunk of read(res.body!)) {
  console.log("receive chunk:", chunk);
}

Example (EventSource)

import read from "@ayonli/jsext/read";

// listen to the `onmessage`
const sse = new EventSource("/sse/message");

for await (const msg of read(sse)) {
  console.log("receive message:", msg);
}

// listen to a specific event
const channel = new EventSource("/sse/broadcast");

for await (const msg of read(channel, { event: "broadcast" })) {
  console.log("receive message:", msg);
}

Example (WebSocket)

import read from "@ayonli/jsext/read";

const ws = new WebSocket("/ws");

for await (const data of read(ws)) {
  if (typeof data === "string") {
    console.log("receive text message:", data);
  } else {
    console.log("receive binary data:", data);
  }
}

Example (EventTarget)

import read from "@ayonli/jsext/read";

for await (const msg of read(self)) {
  console.log("receive message from the parent window:", msg);
}

Example (EventEmitter)

import read from "@ayonli/jsext/read";

for await (const msg of read(process)) {
  console.log("receive message from the parent process:", msg);
}

readAll

declare function readAll<T>(iterable: AsyncIterable<T>): Promise<T[]>;

Reads all values from the iterable object at once.

Example

import readAll from "@ayonli/jsext/readAll";
import * as fs from "node:fs";

const file = fs.createReadStream("./package.json");
const chunks = await readAll(file);

console.log(chunks);

chan

import type { Channel } from "@ayonli/jsext/chan";

declare function chan<T>(capacity?: number): Channel<T>;

Inspired by Golang, cerates a Channel that can be used to transfer data across routines.

If capacity is not set, a non-buffered channel will be created. For a non-buffered channel, the sender and receiver must be present at the same time (theoretically), otherwise, the channel will block (non-IO aspect).

If capacity is set, a buffered channel will be created. For a buffered channel, data will be queued in the buffer first and then consumed by the receiver in FIFO order. Once the buffer size reaches the capacity limit, no more data will be sent unless there is new space available.

It is possible to set the capacity to Infinity to allow the channel to never block and behave like a message queue.

Unlike EventEmitter or EventTarget, Channel guarantees the data will always be delivered, even if there is no receiver at the moment.

Also, unlike Golang, await channel.recv() does not prevent the program from exiting.

Channels can be used to send and receive streaming data between main thread and worker threads wrapped by parallel(), but once used that way, channel.close() must be explicitly called in order to release the channel for garbage collection.

Example (non-buffered)

import chan from "@ayonli/jsext/chan";

const channel = chan<number>();

(async () => {
  await channel.send(123);
})();

const num = await channel.recv();
console.log(num); // 123

Example (buffered)

import chan from "@ayonli/jsext/chan";

const channel = chan<number>(3);

await channel.send(123);
await channel.send(456);
await channel.send(789);

const num1 = await channel.recv();
const num2 = await channel.recv();
const num3 = await channel.recv();

console.log(num1); // 123
console.log(num2); // 456
console.log(num3); // 789

Example (iterable)

import chan from "@ayonli/jsext/chan";
import { range } from "@ayonli/jsext/number";

const channel = chan<number>();

(async () => {
  for (const num of range(1, 5)) {
    await channel.send(num);
  }

  channel.close();
})();

for await (const num of channel) {
  console.log(num);
}
// output:
// 1
// 2
// 3
// 4
// 5

parallel

import type { ThreadedFunctions } from "@ayonli/jsext/parallel";

declare function parallel<M extends { [x: string]: any }>(
  mod: string | (() => Promise<M>),
): ThreadedFunctions<M>;

Wraps a module so its functions will be run in worker threads.

In Node.js and Bun, the module can be either an ES module or a CommonJS module, node_modules and built-in modules are also supported.

In browsers and Deno, the module can only be an ES module.

Data are cloned and transferred between threads via Structured Clone Algorithm.

Apart from the standard data types supported by the algorithm, Channel can also be used to transfer data between threads. To do so, just passed a channel instance to the threaded function. But be aware, channel can only be used as a parameter, return a channel from the threaded function is not allowed. Once passed, the data can only be transferred into and out-from the function.

The difference between using a channel and a generator function for streaming processing is, for a generator function, next(value) is coupled with a yield value, the process is blocked between next calls, channel doesn't have this limit, we can use it to stream all the data into the function before processing and receiving any result.

The threaded function also supports ArrayBuffers as transferable objects. If an array buffer is presented as an argument or the direct property of an argument (assume it's a plain object), or the array buffer is the return value or the direct property of the return value (assume it's a plain object), it automatically becomes a transferrable object and will be transferred to the other thread instead of being cloned. This strategy allows us to easily compose objects like Request and Response instances into plain objects and pass them between threads without overhead.

NOTE: If the current module is already in a worker thread, use this function won't create another worker thread.

NOTE: Cloning and transferring data between the main thread and worker threads are very heavy and slow, worker threads are only intended to run CPU-intensive tasks or divide tasks among multiple threads, they have no advantage when performing IO-intensive tasks such as handling HTTP requests, always prefer cluster module for that kind of purpose.

NOTE: For error instances, only the following types are guaranteed to be sent and received properly between threads.

  • Error
  • EvalError
  • RangeError
  • ReferenceError
  • SyntaxError
  • TypeError
  • URIError
  • AggregateError (as arguments, return values, thrown values, or shallow object properties)
  • Exception (as arguments, return values, thrown values, or shallow object properties)
  • DOMException (as arguments, return values, thrown values, or shallow object properties)

In order to handle errors properly between threads, throw well-known error types or use Exception (or DOMException) with error names in the threaded function.

Example (regular or async function)

import parallel from "@ayonli/jsext/parallel";

const mod = parallel(() => import("./examples/worker.mjs"));
console.log(await mod.greet("World")); // Hi, World

Example (generator or async generator function)

import parallel from "@ayonli/jsext/parallel";

const mod = parallel(() => import("./examples/worker.mjs"));

for await (const word of mod.sequence(["foo", "bar"])) {
  console.log(word);
}
// output:
// foo
// bar

Example (use channel)

import parallel from "@ayonli/jsext/parallel";
import chan from "@ayonli/jsext/chan";
import { range } from "@ayonli/jsext/number";
import readAll from "@ayonli/jsext/readAll";

const mod = parallel(() => import("./examples/worker.mjs"));

const channel = chan<{ value: number; done: boolean }>();
const length = mod.twoTimesValues(channel);

for (const value of range(0, 9)) {
  await channel.send({ value, done: value === 9 });
}

const results = (await readAll(channel)).map((item) => item.value);
console.log(results); // [0, 2, 4, 6, 8, 10, 12, 14, 16, 18]
console.log(await length); // 10

Example (use transferrable)

import parallel from "@ayonli/jsext/parallel";

const mod = parallel(() => import("./examples/worker.mjs"));

const arr = Uint8Array.from([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]);
const length = await mod.transfer(arr.buffer);

console.log(length); // 10
console.log(arr.length); // 0

NOTE: If the application is to be bundled, use the following syntax to link the module instead, it will prevent the bundler from including the file and rewriting the path.

const mod = parallel<typeof import("./examples/worker.mjs")>(
  "./examples/worker.mjs",
);
namespace parallel {
  /**
   * The maximum number of workers allowed to exist at the same time. If not
   * set, the program by default uses CPU core numbers as the limit.
   */
  export var maxWorkers: number | undefined;

  /**
   * In browsers, by default, the program loads the worker entry directly from
   * GitHub, which could be slow due to poor internet connection, we can copy
   * the entry file `bundle/worker.mjs` to a local path of our website and set
   * this option to that path so that it can be loaded locally.
   *
   * Or, if the code is bundled, the program won't be able to automatically
   * locate the entry file in the file system, in such case, we can also copy
   * the entry file (`bundle/worker.mjs` for Bun, Deno and the browser,
   * `bundle/worker-node.mjs` for Node.js) to a local directory and supply
   * this option instead.
   */
  export var workerEntry: string | undefined;
}

run

declare function run<R, A extends any[] = any[]>(
  script: string,
  args?: A,
  options?: {
    /**
     * If not set, invoke the default function, otherwise invoke the specified
     * function.
     */
    fn?: string;
    /** Automatically abort the task when timeout (in milliseconds). */
    timeout?: number;
    /**
     * Instead of dropping the worker after the task has completed, keep it
     * alive so that it can be reused by other tasks.
     */
    keepAlive?: boolean;
    /**
     * Choose whether to use `worker_threads` or `child_process` for running
     * the script. The default setting is `worker_threads`.
     *
     * In browsers and Deno, this option is ignored and will always use the web
     * worker.
     *
     * @deprecated Always prefer `worker_threads` over `child_process` since it
     * consumes less system resources and `child_process` may not work in
     * Windows. `child_process` support may be removed in the future once
     * considered thoroughly.
     */
    adapter?: "worker_threads" | "child_process";
  },
): Promise<{
  workerId: number;
  /** Retrieves the return value of the function being called. */
  result(): Promise<R>;
  /** Iterates the yield value if the function being called returns a generator. */
  iterate(): AsyncIterable<R>;
  /** Terminates the worker thread and aborts the task. */
  abort(reason?: Error | null): Promise<void>;
}>;

Runs the given script in a worker thread and abort the task at any time.

This function is similar to parallel(), many features applicable to parallel() are also applicable to run(), except the following:

  1. The script can only be a filename, and is relative to the current working directory (or the current URL) if not absolute.
  2. Only one task is allow to run at a time for one worker thread, set run.maxWorkers to allow more tasks to be run at the same time if needed.
  3. By default, the worker thread is dropped after the task settles, set keepAlive option in order to reused it.

Example (result)

import run from "@ayonli/jsext/run";

const job1 = await run<string, [string]>("examples/worker.mjs", ["World"]);
console.log(await job1.result()); // Hello, World

Example (iterate)

import run from "@ayonli/jsext/run";

const job2 = await run<string, [string[]]>(
  "examples/worker.mjs",
  [["foo", "bar"]],
  { fn: "sequence" },
);
for await (const word of job2.iterate()) {
  console.log(word);
}
// output:
// foo
// bar

Example (abort)

import run from "@ayonli/jsext/run";
import _try from "@ayonli/jsext/try";

const job3 = await run<string, [string]>("examples/worker.mjs", ["foobar"], {
  fn: "takeTooLong",
});
await job3.abort();
const [err, res] = await _try(job3.result());
console.assert(err === null);
console.assert(res === undefined);
namespace run {
  /**
   * The maximum number of workers allowed to exist at the same time.
   * If not set, use the same setting as {@link parallel.maxWorkers}.
   */
  export var maxWorkers: number | undefined;
}

deprecate

declare function deprecate<T, Fn extends (this: T, ...args: any[]) => any>(
  fn: Fn,
  tip?: string,
  once?: boolean,
): Fn;

Marks a function as deprecated and returns a wrapped function.

When the wrapped function is called, a deprecation warning will be emitted to the stdout.

NOTE: The original function must have a name.

Example

import deprecate from "@ayonli/jsext/deprecate";

const sum = deprecate(function sum(a: number, b: number) {
  return a + b;
}, "use `a + b` instead");
console.log(sum(1, 2));
// output:
// DeprecationWarning: sum() is deprecated, use `a + b` instead (at <anonymous>:4:13)
// 3
declare function deprecate(
  target: string,
  forFn: Function,
  tip?: string,
  once?: boolean,
): void;

Emits a deprecation warning for the target, usually a parameter, an option, or the function's name, etc.

Example

import deprecate from "@ayonli/jsext/deprecate";

function pow(a: number, b: number) {
  deprecate("pow()", pow, "use `a ** b` instead");
  return a ** b;
}

console.log(pow(2, 3));
// output:
// DeprecationWarning: pow() is deprecated, use `a ** b` instead (at <anonymous>:5:13)
// 8