api-test

API Test

API testing made simple

Install

npm install api-test --save-dev

Usage

Create a test file 'test/api-test/sample.md' to test the 'item/get' endpoint, like this:

# item/get
## Setup
### item in items
	name: 'Salad'
	price: 500
## Invalid request
### Post
	randomId()
### Out 400
	error:
		code: 200
## Valid request
### Post
	item.id
### Out
	name: item.name
	price: item.price

And in your mocha testing code:

require('api-test')('test/api-test', {
	mongoUri: 'mongodb://localhost:27017/api_test',
	baseUrl: 'http://localhost:8000/'
})

Concepts

Testing is, at the same time:

• very great because it lets you trust code is ready for production!
• extremely boring to write, because test code is dumb and repetitive

This module tries to solve this by making testing code more concise and unifying testing and documentation.

Markdown was choosen because it's easy to write/read and it's not code!

Test structure

A test is divided in two parts:

Test setup

This is an optional section called 'Setup' that lets you insert documents, declare variables and clear mongo collections to prepare the database before the test cases run.

Inserting documents

The syntax is simply:

### _docName_ in _collection_
	_docDescription_

At the first insertion in a collection, it will be cleared. This is important to make every test isolated. You may refer to this object by its docName.

The syntax for docDescription is described bellow

Clearing collections

The syntax is simply:

### Clear _collection_

Use this only when you won't insert any document in that collection, but want it to be cleared.

All documents in that collection will be removed, indexes will be kept

Declaring variables

You can declare and define a variable to use in test cases, db insertions and more:

### _varName_ is
	_variableContent_

This will make varName available to every following object block.

Test cases

A test case has three optional sections:

• Post: the JSON body to send by POST. Must start with a header like ### Post [_url_]. Default: empty JSON object {}. The url is optional and defaults to the test name
• Out: the expected JSON output. Must start with a header like ### Out [_statusCode_]. Default: no output checking. The statusCode is optional and defaults to 200
• Finds: optional DB assertions. Must start with a header like ### Find in _collection_

In all cases, the syntax is described bellow

Skipping test cases

By appending (skip) to a test case name (see an example) it will be simply ignored. This puts them in a pending state, and is favoured over removing tests which you may forget to add back again.

You can also append (skip) to a test file header (see an example) to skip the whole file.

Object syntax

The syntax was designed to be concise and expressive. The values will be eval'ed as normal JS with a context with special variables (see default context bellow).

The object can be a simple JS value, like:

new Date

Or an object with one property by line and tabs used to declare sub-objects:

user:
	name:
		first: 'Happy'
		last: 'Customer'
	age: 37 + 2
	country: 'cm'.toUpperCase()

Or mixins, like:

user with name.first: 'Unhappy'

Learn more about the syntax in the file doc-syntax.md

Default context

• ObjectId(): the mongo object id constructor
• randomId(): return a random mongo-id as a 24-hex-char string
• randomStr([len=7], [alphabet=a-zA-Z0-9+/])
• randomHex([len=7])
• randomCode([len=7])
• randomEmail([domain='example.com'])
• random([min=0],[max=1])
• randomInt([min=0],[max=100])
• randomBool()
• randomOf(...values): return one of its arguments
• empty: the empty object {}
• post: the request body of the current test case
• out: the response body of the current test case
• prev: an object wity keys:
  • post: the request body of the previous request
  • out: the response body of the previous request

Options

• mongoUri: the mongo uri to connect to. The hostname SHOULD be 'localhost' and the db name SHOULD contains 'test'. If not, the code will ask for confirmation. This protects one from dropping production data, since the tests automatically clear collections, before inserting docs.
• baseUrl: the base API url. Every request url will be composed from this base and the test name.
• describe, it, before: (optional) the mocha interface. Defaults to global mocha functions
• context: (optional) define your own variables/functions accessible to object definitions
• recursive: (optional) whether to look for *.md files inside subfolders (default: false)
• strict: (optional) whether the output check should be strict and complain about unexpected keys (default: true)
• filterFile: (optional) a function that will be called for every file and should return true if this file should be parsed
• preParse: (optional) a function that will be called with all pre-parsed tokens (Header and Obj). This lets you do crazy stuff with the parsing if you want

Type checking

If you don't know the exact value for an API response or field in the collection, you can check for its type:

## Testing types
### Post
	...
### Out
	token: String
### Find in users
	password: String
	lastLogin: Date

The valid values for ### Out are: String, Number, Boolean, Object, Array.

For ### Find in _coll_ are: Number, String, Boolean, Date, RegExp

Custom context

You can use custom context to help writing tests. All default context variables and methods will still be accessible (unless overwritten).

For example: if all endpoints return errors like this: {error: {code: _code_, message: _aDebugString_}}, you can pass as context:

options.context = {
	error: function (code) {
		return {
			error: {
				code: code,
				message: String
			}
		}
	}
}

And then write a test case like this:

## Invalid email should give error 200
### Post
	user:
		email: randomEmail()
### Out
	error(200)

Instead of repeating yourself with:

	error:
		code: 200
		message: String

Examples

See more test examples in the folder test/api-test

Run test

Run npm test in the project root folder.

TODO

• Make keys less restrictive