@adobe/helix-fetch

Light-weight Fetch implementation transparently supporting both HTTP/1(.1) and HTTP/2

Usage no npm install needed!

<script type="module">
  import adobeHelixFetch from 'https://cdn.skypack.dev/@adobe/helix-fetch';
</script>

README

Helix Fetch

Light-weight Fetch API implementation transparently supporting both HTTP/1(.1) and HTTP/2

codecov CircleCI GitHub license GitHub issues LGTM Code Quality Grade: JavaScript Renovate enabled semantic-release Install size Current version


About

helix-fetch in general adheres to the Fetch API Specification, implementing a subset of the API. However, there are some notable deviations:

  • Response.body returns a Node.js Readable stream.
  • Response.blob() is not implemented. Use Response.buffer() instead.
  • Response.formData() is not implemented.
  • Cookies are not stored by default. However, cookies can be extracted and passed by manipulating request and response headers.
  • The following values of the fetch() option cache are supported: 'default' (the implicit default) and 'no-store'. All other values are currently ignored.
  • The following fetch() options are ignored due to the nature of Node.js and since helix-fetch doesn't have the concept of web pages: mode, referrer, referrerPolicy, integrity and credentials.
  • The fetch() option keepalive is not supported. But you can use the h1.keepAlive context option, as demonstrated here.

helix-fetch also supports the following extensions:

  • Response.buffer() returns a Node.js Buffer.
  • Response.url contains the final url when following redirects.
  • The body that can be sent in a Request can also be a Readable Node.js stream, a Buffer, a string or a plain object.
  • There are no forbidden header names.
  • The Response object has an extra property httpVersion which is one of '1.0', '1.1' or '2.0', depending on what was negotiated with the server.
  • The Response object has an extra property fromCache which determines whether the response was retrieved from cache.
  • Response.headers.plain() returns the headers as a plain object.
  • The Fetch option follow allows to limit the number of redirects to follow (default: 20).
  • The Fetch option compress enables transparent gzip/deflate/br content decoding (default: true)

Note that non-standard Fetch options have been aligned with node-fetch where possible.

Features

  • supports reasonable subset of the standard Fetch specification
  • Transparent handling of HTTP/1(.1) and HTTP/2 connections
  • RFC 7234 compliant cache
  • Support gzip/deflate/br content encoding
  • HTTP/2 request and response multiplexing support
  • HTTP/2 Server Push support (transparent caching and explicit listener support)
  • overridable User-Agent
  • low-level HTTP/1.* agent/connect options support (e.g. keepAlive, rejectUnauthorized)

Installation

Note:

As of v2 Node version >= 12 is required.

$ npm install @adobe/helix-fetch

Upgrading

Upgrading from an old version of helix-fetch? Check out the following files:

API

Apart from the standard Fetch API

  • fetch()
  • Request
  • Response
  • Headers
  • Body

helix-fetch exposes the following extensions:

  • context() - creates a new customized API context
  • reset() - resets the current API context, i.e. closes pending sessions/sockets, clears internal caches, etc ...
  • onPush() - registers an HTTP/2 Server Push listener
  • offPush()- deregisters a listener previously registered with onPush()
  • clearCache() - clears the HTTP cache (cached responses)
  • cacheStats() - returns cache statistics
  • createUrl() - creates a URL with query parameters (convenience)
  • timeoutSignal() - ceates a timeout signal (convenience)

Context

An API context allows to customize certain aspects of the implementation and provides isolation of internal structures (session caches, HTTP cache, etc.) per API context.

The following options are supported:

interface ContextOptions {
  /**
   * Value of `user-agent` request header
   * @default 'helix-fetch/<version>'
   */
  userAgent?: string;
  /**
   * The maximum total size of the cached entries (in bytes)
   * @default 100 * 1024 * 1024
   */
  maxCacheSize?: number;
  /**
   * The protocols to be negotiated, in order of preference
   * @default [ALPN_HTTP2, ALPN_HTTP1_1, ALPN_HTTP1_0]
   */
  alpnProtocols?: ReadonlyArray< ALPNProtocol >;
  /**
   * How long (in milliseconds) should ALPN information be cached for a given host?
   * @default 60 * 60 * 1000
   */
  alpnCacheTTL?: number;
  /**
   * Maximum number of ALPN cache entries
   * @default 100
   */
  alpnCacheSize?: number;
  h1?: Http1Options;
  h2?: Http2Options;
};

interface Http1Options {
  /**
   * Keep sockets around in a pool to be used by other requests in the future.
   * @default false
   */
  keepAlive?: boolean;
  /**
   * When using HTTP KeepAlive, how often to send TCP KeepAlive packets over sockets being kept alive.
   * Only relevant if keepAlive is set to true.
   * @default 1000
   */
  keepAliveMsecs?: number;
  /**
   * (HTTPS only)
   * If not false, the server certificate is verified against the list of supplied CAs. An 'error' event is emitted if verification fails; err.code contains the OpenSSL error code.
   * @default true
   */
  rejectUnauthorized?: boolean;
  /**
   * (HTTPS only)
   * Maximum number of TLS cached sessions. Use 0 to disable TLS session caching.
   * @default 100
   */
  maxCachedSessions?: number;
}

interface Http2Options {
  /**
   * Max idle time in milliseconds after which a session will be automatically closed. 
   * @default 5 * 60 * 1000
   */
  idleSessionTimeout?: number;
  /**
   * Enable HTTP/2 Server Push?
   * @default true
   */
  enablePush?: boolean;
  /**
   * Max idle time in milliseconds after which a pushed stream will be automatically closed. 
   * @default 5000
   */
  pushedStreamIdleTimeout?: number;
};

Common Usage Examples

Access Response Headers and other Meta data

  const { fetch } = require('@adobe/helix-fetch');

  const resp = await fetch('https://httpbin.org/get');
  console.log(resp.ok);
  console.log(resp.status);
  console.log(resp.statusText);
  console.log(resp.httpVersion);
  console.log(resp.headers.plain());
  console.log(resp.headers.get('content-type'));

Fetch JSON

  const { fetch } = require('@adobe/helix-fetch');

  const resp = await fetch('https://httpbin.org/json');
  const jsonData = await resp.json();

Fetch text data

  const { fetch } = require('@adobe/helix-fetch');

  const resp = await fetch('https://httpbin.org/');
  const textData = await resp.text();

Fetch binary data

  const { fetch } = require('@adobe/helix-fetch');

  const resp = await fetch('https://httpbin.org//stream-bytes/65535');
  const imageData = await resp.buffer();

Specify a timeout for a fetch operation

Using timeoutSignal(ms) extension:

  const { fetch, timeoutSignal, AbortError } = require('@adobe/helix-fetch');

  try {
    const resp = await fetch('https://httpbin.org/json', { signal: timeoutSignal(1000) });
    const jsonData = await resp.json();
  } catch (err) {
    if (err instanceof AbortError) {
      console.log('fetch timed out after 1s');
    }
  }

Using AbortController:

  const { fetch, AbortController, AbortError } = require('@adobe/helix-fetch');

  const controller = new AbortController();
  setTimeout(() => controller.abort(), 1000);
  const { signal } = controller;

  try {
    const resp = await fetch('https://httpbin.org/json', { signal });
    const jsonData = await resp.json();
  } catch (err) {
    if (err instanceof AbortError) {
      console.log('fetch timed out after 1s');
    }
  }

Stream an image

  const fs = require('fs');
  const { fetch } = require('@adobe/helix-fetch');

  const resp = await fetch('https://httpbin.org/image/jpeg');
  resp.body.pipe(fs.createWriteStream('saved-image.jpg'));

Post JSON

  const { fetch } = require('@adobe/helix-fetch');

  const method = 'POST';
  const body = { foo: 'bar' };
  const resp = await fetch('https://httpbin.org/post', { method, body });

Post JPEG image

  const fs = require('fs');
  const { fetch } = require('@adobe/helix-fetch');

  const method = 'POST';
  const body = fs.createReadStream('some-image.jpg');
  const headers = { 'content-type': 'image/jpeg' };
  const resp = await fetch('https://httpbin.org/post', { method, body, headers });

Post form data

  const fs = require('fs');
  const { FormData, fetch } = require('@adobe/helix-fetch');

  const method = 'POST';
  const body = new FormData();
  body.append('foo', 'bar');
  body.append('data', [ 0x68, 0x65, 0x6c, 0x69, 0x78, 0x2d, 0x66, 0x65, 0x74, 0x63, 0x68 ]);
  body.append('some_file', fs.createReadStream('/foo/bar.jpg'), 'bar.jpg');
  const resp = await fetch('https://httpbin.org/post', { method, body });

GET with query parameters object

const { createUrl, fetch } = require('@adobe/helix-fetch');

const qs = {
  helix: 'dummy',
  foo: 'bar',
  rumple: "stiltskin",
};

const resp = await fetch(createUrl('https://httpbin.org/json', qs));

or using URLSearchParams:

const { fetch } = require('@adobe/helix-fetch');

const body = new URLSearchParams({
  helix: 'dummy',
  foo: 'bar',
  rumple: "stiltskin",
});

const resp = await fetch('https://httpbin.org/json', { body });

Cache

Responses of GET and HEAD requests are by default cached, according to the rules of RFC 7234:

const { fetch } = require('@adobe/helix-fetch');

const url = 'https://httpbin.org/cache/60'; // -> max-age=60 (seconds)
// send initial request, priming cache
let resp = await fetch(url);
assert(resp.ok);
assert(!resp.fromCache);

// re-send request and verify it's served from cache
resp = await fetch(url);
assert(resp.ok);
assert(resp.fromCache);

You can disable caching with the cache: 'no-store' option:

const { fetch } = require('@adobe/helix-fetch');

const resp = await fetch('https://httbin.org/', { cache: 'no-store' });
assert(resp.ok);
assert(!resp.fromCache);

Advanced Usage Examples

HTTP/2 Server Push

Note that pushed resources will be automatically and transparently added to the cache. You can however add a listener which will be notified on every pushed (and cached) resource.

  const { fetch, onPush } = require('@adobe/helix-fetch');

  onPush((url, response) => console.log(`received server push: ${url} status ${response.status}`));

  const resp = await fetch('https://nghttp2.org');
  console.log(`Http version: ${resp.httpVersion}`);

Force HTTP/1(.1) protocol

  const { context, ALPN_HTTP1_1 } = require('@adobe/helix-fetch');
  const { fetch } = context({
    alpnProtocols: [ALPN_HTTP1_1],
  });

  const resp = await fetch('https://nghttp2.org');
  console.log(`Http version: ${resp.httpVersion}`);

HTTP/1.1 Keep-Alive

const { context, ALPN_HTTP1_1 } = require('@adobe/helix-fetch');
const { fetch } = context({
  alpnProtocols: [ALPN_HTTP1_1], // make sure we're talking HTTP/1.1 to the server
  h1: { // http[s].Agent options
    keepAlive: true
  }
});

const resp = await fetch('https://httpbin.org/status/200');
console.log(`Connection: ${resp.headers.get('connection')}`); // -> keep-alive

Self-signed Certificates

const { fetch } = require('@adobe/helix-fetch').context({ rejectUnauthorized: false });

const resp = await fetch('https://localhost:8443/');  // a server using a self-signed certificate

Set cache size limit

  const { fetch, cacheStats } = require('@adobe/helix-fetch').context({
    maxCacheSize: 100 * 1024, // 100kb (Default: 100mb)
  });

  let resp = await fetch('https://httpbin.org/bytes/60000'); // ~60kb response
  resp = await fetch('https://httpbin.org/bytes/50000'); // ~50kb response
  console.log(cacheStats());

Set a custom user agent

  const { fetch } = require('@adobe/helix-fetch').context({
    userAgent: 'custom-fetch'
  });

  const resp = await fetch('https://httpbin.org//user-agent');
  const json = await resp.json();
  console.log(json['user-agent']);

More examples

More example code can be found in the test source files.

Development

Build

$ npm install

Test

$ npm test

Lint

$ npm run lint

Troubleshooting

You can enable helix-fetch low-level debug console output by setting the DEBUG environment variable to helix-fetch*, e.g.:

$ DEBUG=helix-fetch* node test.js

This will produce console outout similar to:

  ...
  helix-fetch:core established TLS connection: #48 (www.nghttp2.org) +2s
  helix-fetch:core www.nghttp2.org -> h2 +0ms
  helix-fetch:h2 reusing socket #48 (www.nghttp2.org) +2s
  helix-fetch:h2 GET www.nghttp2.org/httpbin/user-agent +0ms
  helix-fetch:h2 session https://www.nghttp2.org established +1ms
  helix-fetch:h2 caching session https://www.nghttp2.org +0ms
  helix-fetch:h2 session https://www.nghttp2.org remoteSettings: {"headerTableSize":8192,"enablePush":true,"initialWindowSize":1048576,"maxFrameSize":16384,"maxConcurrentStreams":100,"maxHeaderListSize":4294967295,"maxHeaderSize":4294967295,"enableConnectProtocol":true} +263ms
  helix-fetch:h2 session https://www.nghttp2.org localSettings: {"headerTableSize":4096,"enablePush":true,"initialWindowSize":65535,"maxFrameSize":16384,"maxConcurrentStreams":4294967295,"maxHeaderListSize":4294967295,"maxHeaderSize":4294967295,"enableConnectProtocol":false} +0ms
  helix-fetch:h2 session https://www.nghttp2.org closed +6ms
  helix-fetch:h2 discarding cached session https://www.nghttp2.org +0ms
  ... 

Additionally, you can enable Node.js low-level debug console output by setting the NODE_DEBUG environment variable appropriately, e.g.

$ export NODE_DEBUG=http*,stream*
$ export DEBUG=helix-fetch*

$ node test.js

Note: this will flood the console with highly verbose debug output.

Acknowledgement

Thanks to node-fetch and github/fetch for providing a solid implementation reference.

License

Apache 2.0