real-nock

Create stubbed HTTP servers that you can modify on the fly.

Sample usage

Say you have a program that queries a backend system, and multiplies its response by 2. You might want to do some black-box testing, spinning up an actual HTTP server so see how it reacts.

Stub = require 'real-nock'
 
describe 'my program', ->
 
  backend = new Stub(port: 6789)
 
  before (done) -> backend.start(done)
  after  (done) -> backend.stop(done)
  beforeEach    -> backend.reset()
 
  it 'multiplies the backend response by 2', (done) ->
    backend.stub.get('/value').reply(200, value: 4)
    program.multiply (err, val) ->
      val.should.eql 8
      done()
 
  it 'also works for large numbers', (done) ->
    backend.stub.get('/value').reply(200, value: 10000)
    program.multiply (err, val) ->
      val.should.eql 20000
      done()
 
  it 'fails gracefully when the backend is slow', (done) ->
    backend.stub.get('/value').delayConnection(1000).reply('slow')
    program.multiply (err, val) ->
      err.message.should.eql 'Failed to call backend'
      done()

Why black-box testing?

In many cases, mocking outbound HTTP calls is a great option. However, sometimes you might need to rely on a real backend server:

• if you want to test the actual HTTP connection
• if you want to write tests that are completely independant of the implementation
• if the program you're testing isn't written in Node

That's great, what type of stubs can I set up?

real-nock uses nock behind the scenes, so you should refer to their documentation for all possible operations. For example:

# reply with hello world after 1 second 
backend.stub
  .get('/')
  .delayConnection(1000)
  .reply(200, 'Hello world')
 
# reply with the contents of a file 
# if the request payload matches 
backend.stub
  .post('/users', name: 'Bob')
  .replyWithFile(200, __dirname + '/bob.json');

Note that stubs are consumed as soon as they are called. Any subsequent call will be considered an unknown route, and trigger the default behaviour (see below).

This allows you to define two stubs in series, and get the corresponding responses in that order:

backend.stub.get('/message').reply(200, 'hello')
backend.stub.get('/message').reply(200, 'goodbye')

You can also configure them to apply more than once:

backend.stub
  .get('/value')
  .times(5)
  .reply(200, 'hello world')

Default behaviour

By default, the HTTP server will ignore any route that wasn't stubbed explicitly, or where the stub has been consumed. The corresponding request will get ETIMEDOUT.

You can also configure the following:

# default behaviour (ETIMEDOUT) 
new Stub(port: 6789, default: 'timeout')
 
# request will get ECONNRESET 
new Stub(port: 6789, default: 'reset')
 
# request will get a custom status code 
new Stub(port: 6789, default: 404)
 
# apply a custom (req, res) function 
new Stub(port: 6789, default: myHandler)

This behaviour can be changed at runtime by setting the default property.

backend.default = 'reset'

Troubleshooting

For debugging, you can log most events on the stub server to stdout.

backend = new Stub(port: 9000, debug: true)

which prints

[localhost:8001] Starting
[localhost:8001] Started
[localhost:8001] GET /users/1 (not stubbed)
[localhost:8001] GET /users/2 (HTTP 200)
[localhost:8001] Stopping
[localhost:8001] Stopped