How-To Guides
In the following guides you can find tips and best practices how to cope with some common tasks. While searching this page for particular keywords can give you quick results, reading the whole section should help you to learn some of the Dredd’s core concepts and usual ways how to approach problems when testing with Dredd.
Isolation of HTTP Transactions
Requests in the API description usually aren’t sorted in order to comply with logical workflow of the tested application. To get the best results from testing with Dredd, you should ensure each resource action (API Blueprint) or operation (OpenAPI 2) is executed in isolated context. This can be easily achieved using hooks, where you can provide your own setup and teardown code for each HTTP transaction.
You should understand that testing with Dredd is an analogy to unit tests of your application code. In unit tests, each unit should be testable without any dependency on other units or previous tests.
Example
Common case is to solve a situation where we want to test deleting of a resource. Obviously, to test deleting of a resource, we first need to create one. However, the order of HTTP transactions can be pretty much random in the API description.
To solve the situation, it’s recommended to isolate the deletion test by hooks. Providing before
hook, we can ensure the database fixture will be present every time Dredd will try to send the request to delete a category item.
API Blueprint
FORMAT: 1A
# Categories API
## Categories [/categories]
### Create a Category [POST]
+ Response 201
## Category [/category/{id}]
+ Parameters
+ id: 42 (required)
### Delete a Category [DELETE]
+ Response 204
## Category Items [/category/{id}/items]
+ Parameters
+ id: 42 (required)
## Create an Item [POST]
+ Response 201
To have an idea where we can hook our arbitrary code, we should first ask Dredd to list all available transaction names:
$ dredd api-description.apib http://127.0.0.1:3000 --names
info: Categories > Create a category
info: Category > Delete a category
info: Category Items > Create an item
Now we can create a hooks.js
file. The file will contain setup and teardown of the database fixture:
hooks = require('hooks');
db = require('./lib/db');
beforeAll(function() {
db.cleanUp();
});
afterEach(function(transaction) {
db.cleanUp();
});
before('Category > Delete a Category', function() {
db.createCategory({id: 42});
});
before('Category Items > Create an Item', function() {
db.createCategory({id: 42});
});
OpenAPI 2
swagger: "2.0"
info:
version: "0.0.0"
title: Categories API
license:
name: MIT
host: www.example.com
basePath: /
schemes:
- http
consumes:
- application/json
produces:
- application/json
paths:
/categories:
post:
responses:
200:
description: ""
/category/{id}:
delete:
parameters:
- name: id
in: path
required: true
type: string
enum:
- "42"
responses:
200:
description: ""
/category/{id}/items:
post:
parameters:
- name: id
in: path
required: true
type: string
enum:
- "42"
responses:
200:
description: ""
To have an idea where we can hook our arbitrary code, we should first ask Dredd to list all available transaction names:
$ dredd api-description.yml http://127.0.0.1:3000 --names
info: /categories > POST > 200 > application/json
info: /category/{id} > DELETE > 200 > application/json
info: /category/{id}/items > POST > 200 > application/json
Now we can create a hooks.js
file. The file will contain setup and teardown of the database fixture:
hooks = require('hooks');
db = require('./lib/db');
beforeAll(function() {
db.cleanUp();
});
afterEach(function(transaction) {
db.cleanUp();
});
before('/category/{id}', function() {
db.createCategory({id: 42});
});
before('/category/{id}/items', function() {
db.createCategory({id: 42});
});
Testing API Workflows
Often you want to test a sequence of steps, a scenario, rather than just one request-response pair in isolation. Since the API description formats are quite limited in their support of documenting scenarios, Dredd probably isn’t the best tool to provide you with this kind of testing. There are some tricks though, which can help you to work around some of the limitations.
Note
API Blueprint prepares direct support for testing and scenarios. Interested? Check out api-blueprint#21!
To test various scenarios, you will want to write each of them into a separate API description document. To load them during a single test run, use the --path
option.
For workflows to work properly, you’ll also need to keep shared context between individual HTTP transactions. You can use hooks in order to achieve that. See tips on how to pass data between transactions.
API Blueprint Example
Imagine we have a simple workflow described:
FORMAT: 1A
# My Scenario
## POST /login
+ Request (application/json)
{"username": "john", "password": "d0e"}
+ Response 200 (application/json)
{"token": "s3cr3t"}
## GET /cars
+ Response 200 (application/json)
[
{"id": "42", "color": "red"}
]
## PATCH /cars/{id}
+ Parameters
+ id: 42 (string, required)
+ Request (application/json)
{"color": "yellow"}
+ Response 200 (application/json)
{"id": 42, "color": "yellow"}
Writing Hooks
To have an idea where we can hook our arbitrary code, we should first ask Dredd to list all available transaction names:
$ dredd api-description.apib http://127.0.0.1:3000 --names
info: /login > POST
info: /cars > GET
info: /cars/{id} > PATCH
Now we can create a hooks.js
file. The code of the file will use global stash
variable to share data between requests:
hooks = require('hooks');
db = require('./lib/db');
stash = {}
// Stash the token we've got
after('/login > POST', function (transaction) {
stash.token = JSON.parse(transaction.real.body).token;
});
// Add the token to all HTTP transactions
beforeEach(function (transaction) {
if (stash.token) {
transaction.request.headers['X-Api-Key'] = stash.token
};
});
// Stash the car ID we've got
after('/cars > GET', function (transaction) {
stash.carId = JSON.parse(transaction.real.body).id;
});
// Replace car ID in request with the one we've stashed
before('/cars/{id} > PATCH', function (transaction) {
transaction.fullPath = transaction.fullPath.replace('42', stash.carId)
transaction.request.uri = transaction.fullPath
})
OpenAPI 2 Example
Imagine we have a simple workflow described:
swagger: "2.0"
info:
version: "0.0.0"
title: Categories API
license:
name: MIT
host: www.example.com
basePath: /
schemes:
- http
consumes:
- application/json
produces:
- application/json
paths:
/login:
post:
parameters:
- name: body
in: body
required: true
schema:
type: object
properties:
username:
type: string
password:
type: string
responses:
200:
description: ""
schema:
type: object
properties:
token:
type: string
/cars:
get:
responses:
200:
description: ""
schema:
type: array
items:
type: object
properties:
id:
type: string
color:
type: string
/cars/{id}:
patch:
parameters:
- name: id
in: path
required: true
type: string
enum:
- "42"
- name: body
in: body
required: true
schema:
type: object
properties:
color:
type: string
responses:
200:
description: ""
schema:
type: object
properties:
id:
type: string
color:
type: string
Writing Hooks
To have an idea where we can hook our arbitrary code, we should first ask Dredd to list all available transaction names:
$ dredd api-description.yml http://127.0.0.1:3000 --names
info: /login > POST > 200 > application/json
info: /cars > GET > 200 > application/json
info: /cars/{id} > PATCH > 200 > application/json
Now we can create a hooks.js
file. The code of the file will use global stash
variable to share data between requests:
hooks = require('hooks');
db = require('./lib/db');
stash = {}
// Stash the token we've got
after('/login > POST > 200 > application/json', function (transaction) {
stash.token = JSON.parse(transaction.real.body).token;
});
// Add the token to all HTTP transactions
beforeEach(function (transaction) {
if (stash.token) {
transaction.request.headers['X-Api-Key'] = stash.token
};
});
// Stash the car ID we've got
after('/cars > GET > 200 > application/json', function (transaction) {
stash.carId = JSON.parse(transaction.real.body).id;
});
// Replace car ID in request with the one we've stashed
before('/cars/{id} > PATCH > 200 > application/json', function (transaction) {
transaction.fullPath = transaction.fullPath.replace('42', stash.carId)
transaction.request.uri = transaction.fullPath
})
Making Dredd Validation Stricter
API Blueprint or OpenAPI 2 files are usually created primarily with documentation in mind. But what’s enough for documentation doesn’t need to be enough for testing.
That applies to both MSON (a language powering API Blueprint’s Attributes sections) and JSON Schema (a language powering the OpenAPI 2 format and API Blueprint’s Schema sections).
In following sections you can learn about how to deal with common scenarios.
Avoiding Additional Properties
If you describe a JSON body which has attributes name
and size
, the following payload will be considered as correct:
{"name": "Sparta", "size": 300, "luck": false}
It’s because in both MSON and JSON Schema additional properties are not forbidden by default.
In API Blueprint’s Attributes sections you can mark your object with
fixed-type
(spec), which doesn’t allow additional properties.In API Blueprint’s Schema sections and in OpenAPI 2 you can use
additionalProperties: false
(spec) on the objects.
Requiring Properties
If you describe a JSON body which has attributes name
and size
, the following payload will be considered as correct:
{"name": "Sparta"}
It’s because properties are optional by default in both MSON and JSON Schema and you need to explicitly specify them as required.
In API Blueprint’s Attributes section, you can use
required
(spec).In API Blueprint’s Schema sections and in OpenAPI 2 you can use
required
(spec), where you list the required properties. (Note this is true only for the Draft v4 JSON Schema, in older versions therequired
functionality was done differently.)
Validating Structure of Array Items
If you describe an array of items, where each of the items should have a name
property, the following payload will be considered as correct:
[{"name": "Sparta"}, {"title": "Athens"}, "Thebes"]
That’s because in MSON, the default behavior is that you are specifying what may appear in the array.
In API Blueprint’s Attributes sections you can mark your array with
fixed-type
(spec), which doesn’t allow array items of a different structure then specified.In API Blueprint’s Schema sections and in OpenAPI 2 make sure to learn about how validation of arrays exactly works.
Validating Specific Values
If you describe a JSON body which has attributes name
and size
, the following payload will be considered as correct:
{"name": "Sparta", "size": 42}
If the size should be always equal to 300, you need to specify the fact in your API description.
In API Blueprint’s Attributes sections you can mark your property with
fixed
(spec), which turns the sample value into a required value. You can also useenum
(spec) to provide a set of possible values.In API Blueprint’s Schema sections and in OpenAPI 2 you can use
enum
(spec) with one or more possible values.
Integrating Dredd with Your Test Suite
Generally, if you want to add Dredd to your existing test suite, you can just save Dredd configuration in the dredd.yml
file and add call for dredd
command to your task runner.
There are also some packages which make the integration a piece of cake:
To find more, search for dredd
in your favorite language’s package index.
Continuous Integration
It’s a good practice to make Dredd part of your continuous integration workflow. Only that way you can ensure that application code you’ll produce won’t break the contract you provide in your API documentation.
Dredd’s interactive configuration wizard, dredd init
, can help you with setting up dredd.yml
configuration file and with modifying or generating CI configuration files for Travis CI or CircleCI.
If you prefer to add Dredd yourself or you look for inspiration on how to add Dredd to other continuous integration services, see examples below. When testing in CI, always pin your Dredd version to a specific number and upgrade to newer releases manually.
Authenticated APIs
Dredd supports all common authentication schemes:
Basic access authentication
Digest access authentication
OAuth (any version)
CSRF tokens
…
Use user
setting in your configuration file or the --user
option to provide HTTP basic authentication:
--user=user:password
Most of the authentication schemes use HTTP header for carrying the authentication data. If you don’t want to add authentication HTTP header to every request in the API description, you can instruct Dredd to do it for you by the --header
option:
--header="Authorization: Basic YmVuOnBhc3M="
Sending Multipart Requests
FORMAT: 1A
# Testing 'multipart/form-data' Request API
# POST /data
+ Request (multipart/form-data; boundary=CUSTOM-BOUNDARY)
+ Body
--CUSTOM-BOUNDARY
Content-Disposition: form-data; name="text"
Content-Type: text/plain
test equals to 42
--CUSTOM-BOUNDARY
Content-Disposition: form-data; name="json"
Content-Type: application/json
{"test": 42}
--CUSTOM-BOUNDARY--
+ Response 200 (application/json; charset=utf-8)
+ Body
{"test": "OK"}
swagger: '2.0'
info:
title: "Testing 'multipart/form-data' Request API"
version: '1.0'
consumes:
- multipart/form-data; boundary=CUSTOM-BOUNDARY
produces:
- application/json; charset=utf-8
paths:
'/data':
post:
parameters:
- name: text
in: formData
type: string
required: true
x-example: "test equals to 42"
- name: json
in: formData
type: string
required: true
x-example: '{"test": 42}'
responses:
200:
description: 'Test OK'
examples:
application/json; charset=utf-8:
test: 'OK'
Sending Form Data
FORMAT: 1A
# Testing 'application/x-www-form-urlencoded' Request API
# POST /data
+ Request (application/x-www-form-urlencoded)
+ Body
test=42
+ Response 200 (application/json; charset=utf-8)
+ Body
{"test": "OK"}
swagger: '2.0'
info:
title: "Testing 'application/x-www-form-urlencoded' Request API"
version: '1.0'
consumes:
- application/x-www-form-urlencoded
produces:
- application/json; charset=utf-8
paths:
'/data':
post:
parameters:
- name: test
in: formData
type: string
required: true
x-example: "42"
responses:
200:
description: 'Test OK'
examples:
application/json; charset=utf-8:
test: 'OK'
Working with Images and other Binary Bodies
The API description formats generally do not provide a way to describe binary content. The easiest solution is to describe only the media type, to leave out the body, and to handle the rest using Hooks.
Binary Request Body
API Blueprint
FORMAT: 1A
# Images API
## Resource [/image.png]
### Send an Image [PUT]
+ Request (image/png)
+ Response 200 (application/json; charset=utf-8)
+ Body
{"test": "OK"}
OpenAPI 2
swagger: "2.0"
info:
version: "1.0"
title: Images API
schemes:
- http
consumes:
- image/png
produces:
- application/json
paths:
/image.png:
put:
parameters:
- name: binary
in: body
required: true
schema:
type: string
format: binary
responses:
200:
description: 'Test OK'
examples:
application/json; charset=utf-8:
test: 'OK'
Hooks
In hooks, you can populate the request body with real binary data. The data must be in a form of a Base64-encoded string.
const hooks = require('hooks');
const fs = require('fs');
const path = require('path');
hooks.beforeEach((transaction, done) => {
const buffer = fs.readFileSync(path.join(__dirname, '../image.png'));
transaction.request.body = buffer.toString('base64');
transaction.request.bodyEncoding = 'base64';
done();
});
Binary Response Body
API Blueprint
FORMAT: 1A
# Images API
## Resource [/image.png]
### Retrieve Representation [GET]
+ Response 200 (image/png)
OpenAPI 2
swagger: "2.0"
info:
version: "1.0"
title: Images API
schemes:
- http
produces:
- image/png
paths:
/image.png:
get:
responses:
200:
description: Representation
schema:
type: string
format: binary
examples:
"image/png": ""
Note
Do not use the explicit binary
or bytes
formats with response bodies, as Dredd is not able to properly work with those (api-elements.js#269).
Hooks
In hooks, you can either assert the body:
const hooks = require('hooks');
const fs = require('fs');
const path = require('path');
hooks.beforeEachValidation((transaction, done) => {
const bytes = fs.readFileSync(path.join(__dirname, '../image.png'));
transaction.expected.body = bytes.toString('base64');
done();
});
Or you can ignore it:
const hooks = require('hooks');
hooks.beforeEachValidation((transaction, done) => {
transaction.real.body = '';
done();
});
Multiple Requests and Responses
Note
For details on this topic see also How Dredd Works With HTTP Transactions.
API Blueprint
To test multiple requests and responses within one action in Dredd, you need to cluster them into pairs:
FORMAT: 1A
# My API
## Resource [/resource/{id}]
+ Parameters
+ id: 42 (required)
### Update Resource [PATCH]
+ Request (application/json)
{"color": "yellow"}
+ Response 200 (application/json)
{"color": "yellow", "id": 1}
+ Request Edge Case (application/json)
{"weight": 1}
+ Response 400 (application/vnd.error+json)
{"message": "Validation failed"}
Dredd will detect two HTTP transaction examples and will compile following transaction names:
$ dredd api-description.apib http://127.0.0.1 --names
info: Resource > Update Resource > Example 1
info: Resource > Update Resource > Example 2
In case you need to perform particular request with different URI parameters and standard inheritance of URI parameters isn’t working for you, try modifying transaction before its execution in hooks.
OpenAPI 2
When using OpenAPI 2 format, by default Dredd tests only responses with 2xx
status codes. Responses with other codes are marked as skipped and can be activated in hooks:
var hooks = require('hooks');
hooks.before('/resource > GET > 500 > application/json', function (transaction, done) {
transaction.skip = false;
done();
});
Using Apiary Reporter and Apiary Tests
Command-line output of complex HTTP responses and expectations can be hard to read. To tackle the problem, you can use Dredd to send test reports to Apiary. Apiary provides a comfortable interface for browsing complex test reports:
$ dredd apiary.apib http://127.0.0.1 --reporter=apiary
warn: Apiary API Key or API Project Name were not provided. Configure Dredd to be able to save test reports alongside your Apiary API project: https://dredd.org/en/latest/how-to-guides/#using-apiary-reporter-and-apiary-tests
pass: DELETE /honey duration: 884ms
complete: 1 passing, 0 failing, 0 errors, 0 skipped, 1 total
complete: Tests took 1631ms
complete: See results in Apiary at: https://app.apiary.io/public/tests/run/74d20a82-55c5-49bb-aac9-a3a5a7450f06
Saving Test Reports under Your Account in Apiary
As you can see on the screenshot, the test reports are anonymous by default and will expire after some time. However, if you provide Apiary credentials, your test reports will appear on the Tests page of your API Project. This is great especially for introspection of test reports from Continuous Integration.
To get and setup credentials, just follow the tutorial in Apiary:
As you can see, the parameters go like this:
$ dredd -j apiaryApiKey:<Apiary API Key> -j apiaryApiName:<API Project Name>
In addition to using parameters and dredd.yml
, you can also use environment variables:
APIARY_API_KEY=<Apiary API Key>
- Alternative way to pass credentials to Apiary Reporter.APIARY_API_NAME=<API Project Name>
- Alternative way to pass credentials to Apiary Reporter.
When sending test reports to Apiary, Dredd inspects the environment where it was executed and sends some information about it alongside test results. Those are used mainly for detection whether the environment is Continuous Integration and also, they help you to identify individual test reports on the Tests page. You can use the following variables to tell Dredd what to send:
agent (string) -
DREDD_AGENT
or current user in the OShostname (string) -
DREDD_HOSTNAME
or hostname of the OSCI (boolean) - looks for
TRAVIS
,CIRCLE
,CI
,DRONE
,BUILD_ID
, …
Example Values for Request Parameters
While example values are natural part of the API Blueprint format, the OpenAPI 2 specification allows them only for body
request parameters (schema.example
).
However, Dredd needs to know what values to use when testing described API, so it supports x-example
vendor extension property to overcome the OpenAPI 2 limitation:
...
paths:
/cars:
get:
parameters:
- name: limit
in: query
type: number
x-example: 42
The x-example
property is respected for all kinds of request parameters except of body
parameters, where native schema.example
should be used.
Removing Sensitive Data from Test Reports
Sometimes your API sends back sensitive information you don’t want to get disclosed in Apiary Tests or in your CI log. In that case you can use Hooks to do sanitation. Before diving into examples below, do not forget to consider following:
Be sure to read section about security first.
Only the
transaction.test
(docs) object will make it to reporters. You don’t have to care about sanitation of the rest of thetransaction
(docs) object.The
transaction.test.message
and all thetransaction.test.results.body.results.rawData.*.message
properties contain validation error messages. While they’re very useful for learning about what’s wrong on command line, they can contain direct mentions of header names, header values, body properties, body structure, body values, etc., thus it’s recommended their contents are completely removed to prevent unintended leaks of sensitive information.Without the
transaction.test.results.body.results.rawData
property Apiary reporter won’t be able to render green/red difference between payloads.You can use Ultimate ‘afterEach’ Guard to make sure you won’t leak any sensitive data by mistake.
If your hooks crash, Dredd will send an error to reporters, alongside with current contents of the
transaction.test
(docs) object. See the Sanitation of Test Data of Transaction With Secured Erroring Hooks example to learn how to prevent this.
Sanitation of the Entire Request Body
Sanitation of the Entire Response Body
Sanitation of a Request Body Attribute
Sanitation of a Response Body Attribute
Sanitation of Plain Text Response Body by Pattern Matching
Sanitation of Request Headers
Sanitation of Response Headers
Sanitation of URI Parameters by Pattern Matching
Sanitation of Any Content by Pattern Matching
Sanitation of Test Data of Passing Transaction
Sanitation of Test Data When Transaction Is Marked as Failed in ‘before’ Hook
Sanitation of Test Data When Transaction Is Marked as Failed in ‘after’ Hook
Sanitation of Test Data When Transaction Is Marked as Skipped
Ultimate ‘afterEach’ Guard Using Pattern Matching
You can use this guard to make sure you won’t leak any sensitive data by mistake.
Sanitation of Test Data of Transaction With Secured Erroring Hooks
If your hooks crash, Dredd will send an error to reporters, alongside with current contents of the transaction.test
(docs) object. If you want to prevent this, you need to add try/catch
to your hooks, sanitize the test object, and gracefully fail the transaction.