http-encoding/README.md

Everything you need to handle HTTP message body content-encoding

# http-encoding [![Build Status](https://github.com/httptoolkit/http-encoding/workflows/CI/badge.svg)](https://github.com/httptoolkit/http-encoding/actions) [![Available on NPM](https://img.shields.io/npm/v/http-encoding.svg)](https://npmjs.com/package/http-encoding)

> _Part of [HTTP Toolkit](https://httptoolkit.com): powerful tools for building, testing & debugging HTTP(S)_

**Everything you need to handle HTTP message body content-encoding**

This package includes methods to decode & encode all commonly used HTTP content encodings, in a consistent format, usable in a wide range of Node.js versions and browsers.

The supported codecs are:

* Gzip
* Raw deflate (with or without a zlib wrapper)
* Brotli
* Zstandard
* Base64

All encoding names are case-insensitive (although lowercase is generally standard). The 'identity', 'amz-1.0', 'none', 'text', 'binary', 'utf8' and 'utf-8' encodings are all supported as no-op encodings, passed through with no en/decoding at all. Only 'identity' is standard, but the others are all in common use regardless.

Found a codec used in real-world HTTP that isn't supported? Open an issue!

## API

The library includes two general methods:

### `decodeBuffer(body, encoding)`

Takes an encoded body buffer and encoding (in the format of a standard HTTP [content-encoding header](https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Encoding)) and returns a promise for a decoded buffer, using the zero to many buffers specified in the header.

The input buffer can be any Uint8Array including a Node Buffer (a subclass of Uint8Array). A node-compatible buffer is always returned.

If any encoding is unrecognized or unavailable then this method will throw an exception.

A `decodeBufferSync` method is also available for some use cases, but not recommended, as it's less performant and cannot support some encodings (Brotli or Zstandard).

### `encodeBuffer(body, encoding, { level })`

Takes a raw body buffer and a single encoding (a valid HTTP [content-encoding](https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Encoding) name) and returns a promise for an encoded buffer, using the zero to many buffers specified in the header.

The input buffer can be any Uint8Array (including a Node Buffer, which is a Uint8Array subclass) or an ArrayBuffer. A node-compatible buffer is always returned.

If any encoding is unrecognized or unavailable then this method will throw an exception.

## Per-codec methods

This library also exports consistent async methods to compress and decompress each of the codecs directly:

* `gzip`
* `gunzip`
* `deflate`
* `inflate`
* `inflateRaw`
* `brotliCompress`
* `brotliDecompress`
* `zstdCompress`
* `zstdDecompress`
* `encodeBase64`
* `decodeBase64`

Each method accepts a buffer and returns a promise for a buffer.

## Browser usage

To use this in a browser, you'll need to use a bundler (e.g. Webpack) that can include standard Node.js polyfill packages, you may need to install those polyfill packages, and your bundler needs to support bundling WebAssembly (e.g. Webpack v4+).

In Webpack v4 this should all work automatically. In Webpack v5 this will require explicit dependencies and configuration. See this package's own [test webpack config](./karma.conf.js#L14-L44) and [dev dependencies](./package.json) for a working example.

Brotli and Zstandard are only supported in runtime environments that support WebAssembly. All WebAssembly packages are loaded on-demand and only when native methods (e.g. Node's `zlib.brotli*`) are not available.