Documentation and samples for your Node.js RESTful API
supersamples
is a Mocha reporter that understands Supertest to generate reliable and up-to-date API samples. In a nutshell:
- define concrete request/response examples in your test suite
- if you need to, use mocks to make sure you fully control the API responses
- add a few explanations in Markdown
- choose from a few output formats
- get high-level API documentation that's always up-to-date!
Works with any Node.js http.Server
, like Express or Restify
Nothing special! Simply use supertest
in your test suite, and supersamples
will generate the request/response documentation for you!
it(`
# Get list of sports
- list is ordered alphabetically
- doesn't return sports with no active competitions
`, function (done) {
request(server)
.get('/sports')
.set('Accept', 'application/json')
.expect(200)
.expect('Content-Type', /json/)
.expect(
sports: [
{ id: 1, name: 'Soccer' }
{ id: 2, name: 'Tennis' }
]
)
.end(done)
})
supersamples
comes with several renderers built-in:
html
generates a multi-page static HTML websitemarkdown
to generate a single Markdown page you can easily upload to Githubjson
to generate JSON metadata you can process laterpostman
to generate a postman collection for your API
See a live example of the HTML output over here.
npm install supersamples --save-dev
Have a look at the example folder to get started. You can add tests to the usual test
folder, or keep them separate if you want. Simply run Mocha with the provided reporter:
./node_modules/.bin/mocha --reporter supersamples path/to/tests
You also need to specify documentation options in a supersamples.opts file at the root. This file has to be valid JSON
, but also supports comments:
{
// Base URL for the API
"baseUrl": "http://my-api.com",
// Mocha reporter to display test results
// e.g. Dot, TAP, Spec...
"reporter": "Dot",
// One or more rendering modes
// And their associated options
"renderers": {
"<name>": { ... }
"<name>": { ... }
}
}
See each renderer for the set of available options:
Well it depends on the renderer you choose, but they all work off the same set of data:
The hierarchy
The nested suite of describe()
statements that lead to your test becomes the hierarchy / breadcrumbs. In the HTML renderer, the first 2 levels make up the navigation sidebar.
Your markdown content
The it()
statements can contain valid Markdown, which make up the description of each example.
A name for each sample
By default, the content of the it
also becomes your sample name. This is used in the JSON
renderer to help you identify samples. You can also override the name with
it('gets a list of sports', function () {
this.supersamples = { name: 'valid list' }
request(server)
.get('/sports')
.end(done)
})
Ignoring Requests
Supersamples instruments every request that goes through supertest by default. If you are making multiple requests per it
, sometimes the results can be a bit problematic as request and responses get merged.
You can explicitly ignore certain requests from being captured by setting the following header
it('gets a single sport', function (done) {
request(server)
.get('/sports')
.set('SupersamplesIgnore', 'true')
.end(function (err, response) {
request(server)
.get(`/sports/${response.body[0].id}`)
.end(done)
})
})
Ignoring whole tests
Perhaps you include you docs specs in the same tests as other integration tests that you don't want appearing in your docs
You can exclude them with the following
it('backdoor entrance to get a password', function (done) {
this.supersamples = { ignore: true }
request(server)
.get('/supersecretthing')
.end(done)
})
The requests
- The request headers, including custom ones. However it excludes typically irrelevant headers for the context of documentation (
accept-encoding: gzip, deflate
,host: http://localhost:1234
...). - The request payload & file attachments.
The responses
- The response status code, regardless of any
expect()
. - The response headers, but only if they were mentioned in
expect()
. The reason is that many frameworks will add dozens of default headers, which could seriously clutter the docs. - The actually response body, regardless of any
expect()
. Note that even if they don't affect the docs, expectations are checked during the generation process. We 100% recommend that you add some to give extra confidence that the HTTP response are correct.
supersamples
DOES NOT provide a way to describe every path or query string parameter. It's meant to give you reliable but low-cost API samples. If you want a very detailed API description, you might like other tools better:
- tools like Apiary or ApiDoc let you document your API in text-format (for example Markdown or JavaScript comments). Just remember to keep these up to date!
- tools like Swagger provide a JavaScript API to define your routes. It can generate docs that are always up-to-date, if you don't mind using their syntax instead of vanilla Express or Restify.
In our current project, we actually use swagger
and supersamples
together to generate formal API specs as well as request/response examples, and display both side by side in our API portal.