heroku-client

A wrapper for the Heroku v3 API

heroku-client

A wrapper around the v3 Heroku API.

$ npm install heroku-client --save

Docs are auto-generated and live in the docs directory.

heroku-client works by providing functions that return proxy objects for interacting with different resources through the Heroku API.

To begin, require the Heroku module and create a client, passing in an API token:

var Heroku = require('heroku-client'),
    heroku = new Heroku({ token: process.env.HEROKU_API_TOKEN });

The simplest example is listing a user's apps. First, we call heroku.apps(), which returns a proxy object to the /apps endpoint, then we call list() to actually perform the API call:

heroku.apps().list(function (errapps) {
  // `apps` is a parsed JSON response from the API 
});

The advantage of using proxy objects is that they are reusable. Let's get the info for the user's app "my-app", get the dynos for the app, and remove a collaborator:

var app = heroku.apps('my-app');
 
app.info(function (errapp) {
  // Details about the `app` 
});
 
app.dynos().list(function (errdynos) {
  // List of the app's `dynos` 
});
 
app.collaborators('user@example.com').delete(function (errcollaborator) {
  // The `collaborator` has been removed unless `err` 
});

Requests that require a body are easy, as well. Let's add a collaborator to the user's app "another-app":

var app  = heroku.apps('another-app'),
    user = { email: 'new-user@example.com' };
 
app.collaborators().create({ user: user }, function (errcollaborator) {
  // `collaborator` is the newly added collaborator unless `err` 
});

heroku-client has get, post, patch, and delete functions which can make requests with the specified HTTP method to any endpoint:

heroku.get('/apps', function (errapps) {
});
 
// Request body is optional on both `post` and `patch` 
heroku.post('/apps', function (errapp) {
});
 
heroku.post('/apps', { name: 'my-new-app' }, function (errapp) {
});
 
heroku.patch('/apps/my-app', { name: 'my-renamed-app' }, function (errapp) {
});
 
heroku.delete('/apps/my-old-app', function (errapp) {
});

There is also an even more generic request function that can accept many more options:

heroku.request({
  method: 'GET',
  path: '/apps',
  headers: {
    'Foo': 'Bar'
  },
  parseJSON: false
}, function (errresponseBody) {
});

heroku-client works with Node-style callbacks, but also implements promises with the Q library.

var q = require('q');
 
// Fetches dynos for all of my apps. 
heroku.apps().list().then(function (apps) {
 
  return q.all(apps.map(function (app) {
    return heroku.apps(app.name).dynos().list();
  }));
 
}).then(function (dynos) {
 
  console.log(dynos);
 
});

It's easy to get heroku-client working with generators. In this example, I'll use the co library to wrap a function that will get the list of all of my apps, and then get the dynos for each of those apps:

let co     = require('co');
let heroku = require('heroku-client');
let hk     = heroku.createClient({ token: process.env.HEROKU_API_KEY });
 
let main = function* () {
  let apps  = yield hk.apps().list();
  let dynos = yield apps.map(getDynos);
 
  console.log(dynos);
 
  function getDynos(app) {
    return hk.apps(app.name).dynos().list();
  }
};
 
co(main)();

As long as you're using Node >= 0.11, you can run this script with:

$ node --harmony --use-strict file.js

Hooray, no callbacks or promises in sight!

If you'd like to make requests through an HTTP proxy, set the HEROKU_HTTP_PROXY_HOST environment variable with your proxy host, and HEROKU_HTTP_PROXY_PORT with the desired port (defaults to 8080). heroku-client will then make requests through this proxy instead of directly to api.heroku.com.

heroku-client can optionally perform caching of API requests.

heroku-client will cache any response from the Heroku API that comes with an ETag header, and each response is cached individually (i.e. even though the client might make multiple calls for a user's apps and then aggregate them into a single JSON array, each required API call is individually cached). For each API request it performs, heroku-client sends an If-None-Match header if there is a cached response for the API request. If API returns a 304 response code, heroku-client returns the cached response. Otherwise, it writes the new API response to the cache and returns that.

To tell heroku-client to perform caching, call the configure function.

Caching requires an encryption key to encrypt the results prior to caching. This must be set in the environment variable HEROKU_CLIENT_ENCRYPTION_SECRET. HEROKU_CLIENT_ENCRYPTION_SECRET should be a long, random string of characters. heroku-client includes bin/secret as one way of generating values for this variable. Do not publish this secret or commit it to source control. If it's compromised, flush your memcache and generate a new encryption secret.

If cache is the boolean value true then heroku-client will use memjs for caching.

Example:

var Heroku = require('heroku').configure({ cache: true });

This requires a MEMCACHIER_SERVERS environment variable, as well as a HEROKU_CLIENT_ENCRYPTION_SECRET environment variable that heroku-client uses to build cache keys and encrypt cache contents.

MEMCACHIER_SERVERS can be a single hostname:port memache address, or a comma-separated list of memcache addresses, e.g. example.com:11211,example.net:11211. Note that while the environment variable that memjs looks for is named for the MemCachier service it was originally built for, it will work with any memcache server that speaks the binary protocol.

Alternatively you can specify a custom cache implementation. Your custom implementation must define get(key, cb(err, value)) and set(key, value) functions.

Here's a sample implementation that uses Redis to cache API responses for 5-minutes each:

var redis        = require('redis');
var client       = redis.createClient();
var cacheTtlSecs = 5 * 60; // 5 minutes 
 
var redisCache = {
  getfunction(keycb) {
    // Namespace the keys: 
    var redisKey = 'heroku:api:' + key;
    client.GET(redisKey, cb);
  },
 
  setfunction(keyvalue) {
    // Namespace the keys: 
    var redisKey = 'heroku:api:' + key;
    client.SETEX(redisKey, cacheTtlSecs, value, function(err) {
      // ignore errors on set 
    });
  }
};
 
var Heroku = require('heroku-client');
Heroku.configure({
  cache: redisCache
});

To fetch the latest schema, generate documentation, and run the tests:

$ bin/update

Inspect your changes, and bump the version number accordingly when cutting a release.

Documentation for heroku-client is auto-generated from the API schema.

Docs are generated like so:

$ bin/docs

Generating docs also runs a cursory test, ensuring that every documented function is a function that can be called.

heroku-client uses jasmine-node for tests:

$ npm test