torrefy

torrefy

An ESM package that uses Web Streams API to create v1, v2 or hybrid torrents in your web browser.

🏗This package is under active development.🏗

Install

npm i torrefy # or yarn add torrefy

Basic usage

import { create, encode, decode } from "torrefy";

// create a test file
const testFile = new File(
  ["Hello world. This is the test file content."],
  "testfile.txt"
);

// calculate (hash) the meta info of the test file
const metaInfo = await create([testFile]);

// bencode meta info into a readable stream
const torrentStream = encode(metaInfo);

// tee the readable stream into two readable streams
const [torrentStream1, torrentStream2] = torrentStream.tee();

// consume the first readable stream as an array buffer
const torrentBinary = await new Response(torrentStream1).arrayBuffer();

// decode the second readable stream into meta info
const decodedMetaInfo = await decode(torrentStream2);

Features

Supports Creating V1, V2 or Hybrid Torrents

This package supports creating V1, V2 (introduction blog) or Hybrid (introduction blog) torrents.

Covers Various Web File APIs

This package can handle input files or directories acquired from File API, File and Directory Entries API or File System Access API.

Supports Comprehensive Options

TBD

Supports Handling Progress

TBD

Exposes Stream-Based APIs

The create function consumes an iterable of input files as ReadableStreams with options and populates a MetaInfo object. This function internally uses several TransformStreams to chop the files into pieces and hash them.

The encode function consumes any bcodec friendly entity (e.g. MetaInfo object) and bencodes it into a ReadableStream.

The decode function consumes any bcodec friendly ReadableStream (e.g. torrent ReadableStream) and bdecodes it into the corresponding entity. This function internally uses a TransformStream called Tokenizer to tokenize the input ReadableStream and then calls parse function to parse the Tokens.

All TransformStreams used in this package are also exported.

Supports a Comprehensive Set of Bcodec Friendly Javascript Types

Bcodec friendly Javascript types includes (for the time being):

Bcodec Type \ Javascript Type	Strict	Loose (& Strict)
ByteString	string	ArrayBuffer
Integer	number bigint	boolean
List	Strict[]	Loose[]
Dictionary	{[key: string]: Strict}	{[key: string]: Loose} Map<string | ArrayBuffer, Loose>
ignored	-	undefined null

encode function supports all Loose type inputs and decode function always returns Strict type results.

Supports Hooks in Bencoding

⚠ The terminology may change in the future.

You can register hooks when using the encode function. A common use case is extracting the bencoded info dictionary and calculating the infohash. (This package doesn't provide an out-of-box function to calculate infohash for now)

To use hooks, you will have to install the peer dependency @sec-ant/trie-map, which allows you to register hook handlers with iterable paths as keys. You can learn more about this package in its README.

This package provides 3 helper functions to help you register hook handlers on hooks and consume their results as you please: useUint8ArrayStreamHook, useArrayBufferPromiseHook, useTextPromiseHook.

Here is how you should use this feature:

import { encode, EncoderHooks, useArrayBufferPromiseHook } from "torrefy";
import { TrieMap } from "@sec-ant/trie-map";

// create a dummy object to encode
const dummyObject = {
  a: "b",
  c: 1,
  info: {
    foo: "bar",
  },
  s: ["t"],
};

// initialize hooks for registration of hook handlers
const hooks: EncoderHooks = new TrieMap();

// register an array buffer promise hook under `dummyObject.info`
const [infoArrayBufferPromise] = useArrayBufferPromiseHook(["info"], hooks);

// provide the hooks as an input argument into the encode function
const bencodedReadableStream = encode(dummyObject, hooks);

// await the hook result
const infoArrayBuffer = await infoArrayBufferPromise; // => ArrayBuffer(12)