dempfi / Ava Playback

Dealing with HTTP requests in the tests can be painful and what is more important error prone, since we have to stub everything by hand. There are few libraries to help with this situation, like node-nock. But their setup is something complicated (I didn't manage to setup node-nock to record requests and play them back) and they require a lot of manual steps to write single test.

📼 ava-playback is here to help. In record mode, when you write your test, you just allow your app to call real APIs and when you ready, you just switch from record to playback mode and it's done 🎉. In background ava-playback will record new requests and use already existing playbacks for the rest.

Installation

First things first

yarn add ava-playback

or

npm i ava-playback --save

Then in you package.json where you store your ava config just add a requirement of ava-playback.

  // ...
  "ava": {
    "require": [
      "ava-playback"
    ]
  }
  // ...

🎉 that's it.

Playbacks location

By default playbacks will be stored in the root of your project in /playbacks folder, if you want to change the location just add playbacks settings in your package.json.

  // ...
  "ava": {
    "require": [
      "ava-playback"
    ],
    "playbacks": "tests/fixtures"
  },
  // ...

Modes

ava-playback uses env variable AVA_PLAYBACK to get information how it should run. If AVA_PLAYBACK isn't set, ava-playback will not do anything.

• AVA_PLAYBACK=record record all new outgoing requests from your tests
• AVA_PLAYBACK=play turn off HTTP/HTTPS connection and use playbacks to reply to the outgoing requests from the tests

Usage

With ava-playback the flow of writing actual test can look like this.

1. You write a new test and don't activate ava-playback in any mode.
2. When the test is ready you run it one more time with AVA_PLAYBACK=record env variable. ava-playback will record only missing playbacks to playbacks location.
3. You edit new playbacks according to your needs (wildcard auth tokens in the body or in the queries).
4. Check all tests with AVA_PLAYBACK=play mode to verify they pass.
5. Done 🚀

To illustrate the flow take a look at this example

# Write new test file
NODE_ENV=test ava --watch 'new-test-file.js'

# Record all playbacks required for 'new-test-file.js'
NODE_ENV=test AVA_PLAYBACK=record ava 'new-test-file.js'

# Check all tests together
NODE_ENV=test AVA_PLAYBACK=play ava

Wildcards and security

By default ava-playback stores and plays back your requests by taking into account all queries and body. However, you can change this option to hide security information, like tokens in queries.

For example, Slack API allows tokens to be in query params, like slack.com/api/users.list?token=xoxb-XXXXXXXXXX-Z7RKNoLIKOXLPKqtxUy5IhJ5. It's totally fine unless you don't want this token to be stored in git or be available in Travis-ci. For those cases, you can use wildcards feature of ava-playback.

After recording the playback, in you playbacks folder you can find the file matching that request and edit a path entry from

"/api/users.list?token=xoxb-XXXXXXXXXX-Z7RKNoLIKOXLPKqtxUy5IhJ5"

to

"/api/users.list?token=*"

so the file will look like this

{
  "body": "",
  "method": "POST",
  "path": "/api/users.list?token=*",
  "scope": "https://slack.com:443",
  "status": 200,
}

and all future requests to slack API with anything in place of the actual token will be caught by ava-playback.

"/api/users.list?token=*"

"/api/users.list?token=34"
"/api/users.list?token=xoxb-XXXXXXXXXX-Z7RKNoLIKOXLPKqtxUy5IhJ5"

"/api/users.list?tokens=78&tokens=*&tokens=some-token"

"/api/users.list?tokens=78&tokens=anything-here&tokens=some-token"