Veggies is an awesome cucumberjs library for API/CLI testing. Great for testing APIs built upon Express, Koa, HAPI, Loopback and others. It's also the perfect companion for testing CLI applications built with commander, meow & Co.

• Requirements
• Installation
• Features
  • API testing
    • Making a simple request
    • Posting data
    • Posting data using fixture file
    • Using values issued by a previous request
    • Using cookies
    • Testing json response
    • Testing response headers
    • Debugging failing API tests
    • Type system
  • CLI testing
    • Running a simple command
    • Testing a command error
  • File System testing
    • Testing file content
  • Snapshot testing
    • API Snapshot testing
    • CLI Snapshot testing
    • File Snapshot testing
• Extensions
  • state install | gherkin expressions | low level API
  • fixtures install | low level API
  • http API install | gherkin expressions | low level API
  • CLI install | gherkin expressions | low level API
  • fileSystem install | gherkin expressions | low level API
  • snapshot install | low level API
• Helpers
  • cast usage | add a type
• Examples

• Node.js >=12.0.0
• cucumber >=7.0.0

Using npm:

npm install @ekino/veggies

Or yarn:

yarn add @ekino/veggies

Then all you have to do is installing the provided extensions:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { state, fixtures, httpApi, cli } = require('@ekino/veggies')

setWorldConstructor(function() {
    state.extendWorld(this)
    fixtures.extendWorld(this)
    httpApi.extendWorld(this)
    cli.extendWorld(this)
})

state.install()
fixtures.install()
httpApi.install({
    baseUrl: 'http://localhost:3000',
})
cli.install()

For full feature list, have a look at available gherkin expressions for the dedicated extension.

Scenario: Using GitHub API
  Given I set User-Agent request header to veggies/1.0
  When I GET https://api.github.com/
  Then response status code should be 200

You can easily issue a POST request using json payload

Scenario: Creating a resource using json payload
  Given I set request json body
    | username | plouc |
    | gender   | male  |
  When I POST https://my-api.io/users
  Then response status code should be 201

You can also use form encoded values, all you have to do is to change json for form

Scenario: Creating a resource using json payload
  Given I set request form body
    | username | plouc |
    | gender   | male  |
  When I POST https://my-api.io/users
  Then response status code should be 201

Putting large data payloads inside your scenarios can reduce legibility, to improve this, you can use the fixtures extension to define it.

# /features/user/fixtures/user.yml

username: plouc
gender:   male

# /features/user/create_user.feature

Scenario: Creating a resource using json payload
  Given I set request form body from user
  When I POST https://my-api.io/users
  Then response status code should be 201

Imagine you want to test a resource creation and then that you're able to fetch this new entity through the API.

If resource id is generated by your API, it will be impossible to make the second call because id is unknown.

To solve this problem you have the ability to collect data from a previous response, store it in the state and inject it at various places using placeholders.

The following example calls the root GitHub API endpoint, extracts the emojis_url value from the json response and stores it in the current state under the emojisUrl key, then it uses this value to make its next request.

Scenario: Using GitHub API
  Given I set User-Agent request header to veggies/1.0
  When I GET https://api.github.com/
  And I pick response json emojis_url as emojisUrl
  And I GET {{emojisUrl}}
  Then response status code should be 200

It's even possible to mix this approach with scenario outline to have more concise tests (at the cost of clarity thought).

The following example will generates 3 scenario at runtime using different response values for second request.

Scenario Outline: Fetching <key> API endpoint from root endpoint
  Given I set User-Agent request header to veggies/1.0
  When I GET https://api.github.com/
  Then response status code should be 200
  When I pick response json <key> as <key>
  And I GET {{<key>}}
  Then response status code should be 200

  Examples:
    | key              |
    | emojis_url       |
    | feeds_url        |
    | public_gists_url |

Cookies are disabled by default, but you've got the ability to enable/disable the feature using a gherkin Given expression. Be warned that cookies do not persist across scenarios in order to ensure they're autonomous. If you really need to keep a cookie for multiple scenarios, you should consider using a custom step definition and/or using the state extention to store it.

Scenario: Enabling cookies
  Given I enable cookies
  # …
  
Scenario: Disabling cookies
  Given I disable cookies
  # …  

See definitions for all available cookies related gherkin expressions.

veggies gives you the ability to check json responses, the corresponding gherkin expression is:

/^(?:I )?json response should (fully )?match$/

Checking json response properties equal value:

Scenario: Fetching some json response from the internets
    When I GET http://whatever.io/things/1
    Then json response should match
      | field           | matcher | value |
      | name            | equal   | thing |
      | address.country | equal   | Japan |

Checking json response properties contain value:

Scenario: Fetching some json response from the internets
    When I GET http://whatever.io/things/1
    Then json response should match
      | field           | matcher | value |
      | name            | contain | ing   |
      | address.country | contain | Jap   |

Checking json response properties match value:

Scenario: Fetching some json response from the internets
    When I GET http://whatever.io/things/1
    Then json response should match
      | field           | matcher | value     |
      | name            | match   | ^(.+)ing$ |
      | address.country | match   | ^Jap(.+)$ |

Checking json response properties equalRelativeDate value:

Scenario: Fetching some json response from the internets
    When I GET http://whatever.io/things/1
    Then json response should match
      | field           | matcher            | value                                        |
      | endDate         | equalRelativeDate  | 2,days,fr,dddd                               |
      | beginDate       | equalRelativeDate  | -1,week,fr,[Aujourd'hui] YYYY-MM-DD hh[h]mm  |

By default this assertion does not check for full match. Properties not listed will just be ignored, if you want a full match:

Scenario: Fetching some json response from the internets
    When I GET http://whatever.io/things/1
    Then json response should fully match
      | field           | matcher | value     |
      | name            | match   | ^(.+)ing$ |
      | address.country | equal   | Japan     |

Now if the json contains extra properties, the test will fail.

Available matchers are:

Any of these matchers can be negated when preceded by these : !, not, does not, doesn't, is not and isn't.

In order to check response headers, you have the following gherkin expression available:

/^response header (.+) should (not )?(equal|contain|match) (.+)$/

It supports multiple combinations to verify header value conforms to what you expect. This example illustrates its different features:

Scenario: Testing header related expectations
    When I GET http://whatever.io/
    Then response header X-Whatever-A should equal whatever
    And response header X-Whatever-B should not equal whatever
    And response header X-Whatever-C should contain part
    And response header X-Whatever-D should not contain part
    And response header X-Whatever-E should match ^(thing|other)$
    And response header X-Whatever-F should not match ^(thing|other)$

If the header does not exist, the test will fail.

When testing APIs using cucumber, we often faced situations were we didn't understand why a given test were failing. The dirty fix was to add some nasty console.log() everywhere, that's why veggies provides helpers to dump response properties.

Scenario: Fetching something from the internets
    When I GET http://whatever.io/things
    And dump response body
    And dump response headers
    And dump response cookies

You'll now have the response body/headers/cookies dumped in your terminal. You should disable those steps when the test is fixed as it can be noisy enough.

When testing json based APIs, which is a standard nowadays, you have to be aware of data types for sending payloads or making assertions on received responses, that's why veggies provides a lightweight type systems.

The following directives are available:

You can now use those directive for most of step definitions accepting data tables.

For example you can use it to post typed json data:

Scenario: Creating a resource using typed json payload
  Given I set request json body
    | username  | plouc((string))          |
    | team_id   | 1((number))              |
    | is_active | true((boolean))          |
    | hobbies   | drawing,hacking((array)) |
  When I POST https://my-api.io/users
  Then response status code should be 201

which will generate the following payload:

{
  "username": "plouc",
  "team_id": 1,
  "is_active": true,
  "hobbies": [
    "drawing",
    "hacking"
  ]
}

For full feature list, have a look at available gherkin expressions for the dedicated extension.

Scenario: Getting info about installed yarn version
  When I run command yarn --version
  Then exit code should be 0

Scenario: Running an invalid command
  When I run command yarn invalid
  Then exit code should be 1
  And stderr should contain error Command "invalid" not found.

In order to check file content, you have the following gherkin expression available:

/^file (.+) content (not )?(equal|contain|match) (.+)$/

It supports multiple combinations to verify file content conforms to what you expect. This example illustrates its different features:

Scenario: Testing file content related expectations
    Then file sample_A.text content should equal whatever
    And file sample_B.text content should not equal whatever
    And file sample_C.text content should contain part
    And file sample_D.text content should not contain part
    And file sample_E.text content should match ^(thing|other)$
    And file sample_F.text content should not match ^(thing|other)$

If the file does not exist, the test will fail.

Snapshot testing test a response / content against a saved snapshot. Snapshots are stored in a file with same name as the feature file with the extension .snap in a folder snapshots in the same folder as the feature file.

⚠️ Snapshots files should be versioned to be compared while running tests

Folder tree should look like : support/ features/ feature_with_snapshot.feature feature_without_snapshot.feature snapshots/ feature_with_snapshot.feature.snap …

In a snapshot file, snapshot name follow the pattern: SNAPSHOT_NAME NUMBER_OF_TIME_THIS_NAME_WAS_ENCOUNTERED_IN_CURRENT_FILE.NUMBER_OF_TIME_WE_HAD_A_SNAPSHOT_IN_THIS_SCENARIO. For example, this would give: Scnenario 1 1.1

If a snapshot doesn't exists, it will be created the first time.

To update snapshot use the cucumber command line option '-u'. If you narrowed the tests with tags, only the snapshots related to the tagged scenarios will be updated.

In case you need to remove unused snapshots, you can use the option --cleanSnapshots. :warning: You shouldn't use this option with tags. It may result in used snapshots removed. :information_source: Snapshot files related to feature files with no snapshots anymore won't get removed. You need to do it manually.

Sometimes, it could be useful to prevent the creation of snapshots, for instance in a CI environment. To do this, you can use the --preventSnapshotsCreation flag. An error will be thrown if the snapshot is missing and this option is present.

In order to check an api response against a snapshot, you have the following gherkin expression available:

/^response body should match snapshot$/

This example illustrates it:

Scenario: Creating a resource using typed json payload
  Given I set request json body
    | username  | plouc((string))          |
    | team_id   | 1((number))              |
    | is_active | true((boolean))          |
    | hobbies   | drawing,hacking((array)) |
  When I POST https://my-api.io/users
  Then response status code should be 201
  And response body should match snapshot

It is sometimes useful to ignore some fields in a response when comparing with the snapshot. In this case and if it's json you can then use with a table:

/^response json body should match snapshot$/

This examples illustrates it:

Scenario: Creating a resource using typed json payload
  Given I set request json body
    | username  | plouc((string))          |
    | team_id   | 1((number))              |
    | is_active | true((boolean))          |
    | hobbies   | drawing,hacking((array)) |
  When I POST https://my-api.io/users
  Then response status code should be 201
  And response json body should match snapshot
    | field  | matcher | value  |
    | url    | type    | string |

The table supports anything defined in Testing json response

In order to check a CLI output against a snapshot, you have the following gherkin expression available:

/^(stderr|stdout) output should match snapshot$/

This example illustrates it:

Scenario: Getting info about installed yarn version
  When I run command yarn --version
  Then exit code should be 0
  And stdout output should match snapshot
  And stderr output should match snapshot

It is sometimes useful to ignore some fields in a cli json formatted output when comparing with the snapshot. In this case and if it's json you can then use with a table:

/^(stderr|stdout) json output should match snapshot$/

This examples illustrates it:

Scenario: Snapshot testing on a json file
  Given I set cwd to examples/features/snapshot/fixtures
  Then json file file2.json content should match snapshot
    | field           | matcher | value      |
    | gender          | type    | string     |
    | id              | type    | number     |

The table supports anything defined in Testing json response

In order to check a file content against a snapshot, you have the following gherkin expression available:

/^file (.+) should match snapshot$/

This example illustrates it:

Scenario: Testing file content related expectations
    Then file sample_1.text should match snapshot

It is sometimes useful to ignore some fields in a json file when comparing with the snapshot. In this case and if it's json you can then use with a table:

/^json file (.+) content should match snapshot$/

This examples illustrates it:

Scenario: Creating a resource using typed json payload
  Then json file sample_1.text content should match snapshot
    | field  | matcher | value  |
    | url    | type    | string |

The table supports anything defined in Testing json response

This module is composed of several extensions.

state | fixtures | http API | CLI | file system | snapshot

The state extension is a simple helper used to persist state between steps & eventually scenarios (but you should try to avoid coupling scenarios).

It's involved for example when you want to collect values issued by a previous request when using the http API extension.

To install the extension, you should add the following snippet to your world file:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { state } = require('@ekino/veggies')

setWorldConstructor(function() {
    state.extendWorld(this)
})

state.install()

Given:
  - /^(?:I )?set state (.+) to (.+)$/

When:
  - /^(?:I )?clear state$/
  - /^(?:I )?dump state$/

Then:
  # No definitions

When installed, you can access it from the global cucumber context in your own step definitions. For available methods on the state, please refer to its own documentation.

const { When } = require('@cucumber/cucumber')
    
When(/^I do something useful$/, function() {
    const stateValue = this.state.get('whatever')
    // …
})

The fixtures extension can be used to load data from files during testing.

It supports the following file extensions:

• .yaml, .yml - loads and parses a yaml file, result can be Object or Array
• .txt - loads text content, result is a string
• .json - loads json, result is and Object
• .js - loads a javascript module, the module must exports load function via module.exports, result can be whatever type the function returns

To install the extension, you should add the following snippet to your world file:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { fixtures } = require('@ekino/veggies')

setWorldConstructor(function() {
    fixtures.extendWorld(this)
})

fixtures.install()

When installed, you can access it from the global cucumber context in your own step definitions. For available methods on the fixtures loader, please refer to its own documentation.

const { When } = require('@cucumber/cucumber')

When(/^I do something useful with fixtures$/, function() {
    return this.fixtures.load('whatever')
        .then(fixture => {
            // …
        })
})

The http API extension relies on the state & fixtures extensions, so make sure they're registered prior to installation.

To install the extension, you should add the following snippet to your world file:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { state, fixtures, httpApi } = require('@ekino/veggies')

setWorldConstructor(function() {
    state.extendWorld(this)
    fixtures.extendWorld(this)
    httpApi.extendWorld(this)
})

state.install()
httpApi.install({
    baseUrl: 'http://localhost:3000',
})

Given:
  - /^(?:I )?set request headers$/
  - /^(?:I )?do not follow redirect$/
  - /^(?:I )?follow redirect$/
  - /^(?:I )?assign request headers$/
  - /^(?:I )?set ([a-zA-Z0-9-_]+) request header to (.+)$/
  - /^(?:I )?clear request headers/
  - /^(?:I )?set request json body$/
  - /^(?:I )?set request json body from (.+)$/
  - /^(?:I )?set request form body$/
  - /^(?:I )?set request form body from (.+)$/
  - /^(?:I )?set request multipart body from (.+)$/
  - /^(?:I )?clear request body$/
  - /^(?:I )?set request query$/
  - /^(?:I )?pick response json (.+) as (.+)$/
  - /^(?:I )?enable cookies$/
  - /^(?:I )?disable cookies$/
  - /^(?:I )?set cookie from (.+)$/
  - /^(?:I )?clear request cookies$/

When:
  - /^(?:I )?reset http client$/
  - /^(?:I )?(GET|POST|PUT|DELETE) (.+)$/
  - /^(?:I )?dump response body$/
  - /^(?:I )?dump response headers$/
  - /^(?:I )?dump response cookies$/

Then:
  - /^response status code should be ([1-5][0-9][0-9])$/
  - /^response status should be (.+)$/
  - /^response should (not )?have an? (.+) cookie$/
  - /^response (.+) cookie should (not )?be secure$/
  - /^response (.+) cookie should (not )?be http only$/
  - /^response (.+) cookie domain should (not )?be (.+)$/
  - /^(?:I )?json response should (fully )?match$/
  - /^(?:I )?should receive a collection of ([0-9]+) items?(?: for path )?(.+)?$/
  - /^response should match fixture (.+)$/
  - /^response header (.+) should (not )?(equal|contain|match) (.+)$/

When installed, you can access its client from the global cucumber context in your own step definitions. For available methods on the client, please refer to its own documentation.

const { When } = require('@cucumber/cucumber')

When(/^I do something useful$/, function() {
    return this.httpApiClient.makeRequest(/* … */)
})

The CLI extension relies on the state & fixtures extensions, so make sure they're registered prior to installation.

To install the extension, you should add the following snippet to your world file:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { state, fixtures, cli } = require('@ekino/veggies')

setWorldConstructor(function() {
    state.extendWorld(this)
    fixtures.extendWorld(this)
    cli.extendWorld(this)
})

state.install()
fixtures.install()
cli.install()

Given:
  - /^(?:I )?set (?:working directory|cwd) to (.+)$/
  - /^(?:I )?set ([^ ]+) (?:env|environment) (?:var|variable) to (.+)$/
  - /^(?:I )?set (?:env|environment) (?:vars|variables)$/
  - /^(?:I )?kill the process with ([^ ]+) in (\d+)(ms|s)/

When:
  - /^(?:I )?run command (.+)$/
  - /^(?:I )?dump (stderr|stdout)$/

Then:
  - /^(?:the )?(?:command )?exit code should be (\d+)$/
  - /^(stderr|stdout) should be empty$/
  - /^(stderr|stdout) should contain (.+)$/
  - /^(stderr|stdout) should not contain (.+)$/
  - /^(stderr|stdout) should match (.+)$/
  - /^(stderr|stdout) should not match (.+)$/

When installed, you can access it from the global cucumber context in your own step definitions. For available methods on the client, please refer to its own documentation.

const { When } = require('@cucumber/cucumber')

Then(/^I check something from the CLI output$/, function() {
    const out = this.cli.getOutput()
    // …
})

The fileSystem extension relies on the cli extension, so make sure it's registered prior to installation.

To install the extension, you should add the following snippet to your world file:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { state, fixtures, cli, fileSystem } = require('@ekino/veggies')

setWorldConstructor(function() {
    state.extendWorld(this)
    fixtures.extendWorld(this)
    cli.extendWorld(this)
    fileSystem.extendWorld(this)
})

state.install()
fixtures.install()
cli.install()
fileSystem.install()

Given:
  - /^(?:I )?create directory (.+)$/
  - /^(?:I )?remove (?:file|directory) (.+)$/

When:
  # No definitions

Then:
  - /^(file|directory) (.+) should (not )?exist$/
  - /^file (.+) content should (not )?(equal|contain|match) (.+)$/

When installed, you can access it from the global cucumber context in your own step definitions. For available methods on the fileSystem, please refer to its own documentation.

const { Then } = require('@cucumber/cucumber')

Then(/^I check something using file system$/, function() {
    return this.fileSystem.getFileContent('whatever')
        .then(content => {
            // …    
        })
})

The snapshot extension add capabilities to api, cli and file extensions, so you will need these extensions if you want to use snapshot related gherkin definitions.

To install the extension, you should add the following snippet to your world file:

// /support/world.js

const { setWorldConstructor } = require('@cucumber/cucumber')
const { state, fixtures, cli, fileSystem, snapshot } = require('@ekino/veggies')

setWorldConstructor(function() {
    state.extendWorld(this)
    fixtures.extendWorld(this)
    cli.extendWorld(this)
    fileSystem.extendWorld(this)
    snapshot.extendWorld(this)
})

state.install()
fixtures.install()
cli.install()
fileSystem.install()
snapshot.install()

When installed, you can access it from the global cucumber context in your own step definitions. For available methods on the snapshot, please refer to its own documentation.

const { Then } = require('@cucumber/cucumber')

Then(/^Some content should match snapshot$/, function() {
    this.snapshot.expectToMatch('whatever')
})

Cast helper can be used to cast values for custom gherkin rules. To find more about casting see Type System.

This must be used on gherkin arrays. Based on your array type you have to use:

• step.hashes() -> Cast.objects(step.hashes())
• step.rows() -> Cast.array(step.rows())
• step.raw() -> Cast.array(step.raw())
• step.rowsHash() -> Cast.objects(step.rowsHash())

For example:

const { cast } = require('@ekino/veggies')
const { Given, When, Then } = require('@cucumber/cucumber')

Then(/^User data should be$/, (step) => {
    const userData = this.userData
    const expectedData = Cast.objects(step.rowsHash())
    expect(userData).to.be.deep.equal(expectedData)
})

You can provide your own type. For example:

Cast.addType('newType', value => value === 'true')

Can be used on:

  Given I get user id1 profile
  Then I should receive
    | id              | id1             |
    | age             | 1((number))     |
    | name            | veggies         |
    | isPublic        | true((newType)) |

This repository comes with few examples, in order to run them, invoke the following script:

yarn run examples

If you want to only run certain examples, you can use tags, for example to run cli extension examples:

yarn run examples -- --tags @cli

There is a special tag which only runs examples not requiring network access:

yarn run examples -- --tags @offline

Due to public API rate limit (eg. GitHub API), this tag is used when running on CI.