Ready to take your JavaScript development to the next level? Meet npm Enterprise - the ultimate in enterprise JavaScript. Learn more »


0.22.0 • Public • Published


Fork of stucumber with a number of improvements:

  • Fix BeforeFeatures hook ignoring returned promises.
  • Use beforeEach and afterEach so hooks don't eat into test timeouts.
  • Support CRLF line endings.
  • Allow create world to return a promise.

Original docs

A Gherkin parser and Cucumber-like implementation for JavaScript. Used by gherkin-jest and stucumber-register.


Gherkin allows you to write tests in sort of plain English:

Feature: calculator
  Scenario: adding
    Given I take the number 5
    And I take the number 3
    When I add them
    Then I will have 8

You define rules to support the tests in JavaScript, like so:

import { cucumber } from 'stucumber';
cucumber.defineCreateWorld(() => []);
cucumber.defineRule('I take the number {int}', (world, number) => {
cucumber.defineRule('I add them', (world) => {
  const a = world.pop();
  const b = world.pop();
  world.push(+ b);
cucumber.defineRule('I will have {int}', (world, number) => {
  expect(world[world.length - 1]).to.equal(number);

Given, When, Then, And and * are all synonyms, and exist only to make your code more readable.


The package consists of several parts:

  • the Cucumber class, which is basically the test framework
  • the DataTable class, for representing tabular data
  • the parse function, which parses gherkin source and turns it into an AST
  • a Transformer abstract class and GenericTransformer implementation, for converting the AST into JavaScript
  • interfaces or representing the AST, e.g. Feature, Scenario, Clause

Roughly speaking, given the following gherkin:

Feature: calculator
  Scenario: adding
    Given I take the number 5
    And I take the number 3
    When I add them
    Then I will have 8

The GenericTransformer will output something like this, depending on the options:

const {cucumber} = require('stucumber');
describe('Feature: calculator', () => {
  beforeAll(() => cucumber.enterFeature([]));
  afterAll(() => cucumber.exitFeature([]));
  it('adding', () => {
    const world = cucumber.createWorld();
    return cucumber
      .enterScenario(world, [])
      .then(() => cucumber.rule('I take the number 5'))
      .then(() => cucumber.rule('I take the number 3'))
      .then(() => cucumber.rule('I add them'))
      .then(() => cucumber.rule('I will have 8'))
      .then(() => cucumber.exitScenario(world, []));

Template strings

You can write your rules using the template string style notation:

cucumber.defineRule('I have numbers {int} and {int}', (world, a, b) => {
  world.a = a;
  world.b = b;

There are 4 types that can be used as placeholders:

  • {int} - matches an integer ([-+]?\d+) and runs parseInt on it before passing into your handler
  • {float} - matches a floating point number ([-+]?\d*(\.\d+)?) and runs parseFloat on it before passing on
  • {word} - matches a bunch of characters up to a whitespace character ([^\s]+)
  • {string} - matches a double-quoted string and returns only the contents of the string ("([^"]+)")

If you give the templates names, then they are passed as an object parameter instead:

cucumber.defineRule('I have numbers {a:int} and {b:int}', (world, params) => {
  world.a = params.a;
  world.b = params.b;


Any rule can return a promise and it will be awaited before processing the next rule.


You can prefix any feature or scenario with any number of annotations, which consist of a keyword prefixed by an @ symbol. The annotations can optionally have arguments.

Feature: annotations
  @anotherAnnotation(1, "a")
  Scenario: some scenario
    * ...


You can register functions to handle various hooks:

  • HookType.BeforeFeatures - runs once at the beginning of each feature
  • HookType.BeforeScenarios - runs at the beginning of each scenario, just after the call to createWorld
  • HookType.AfterFeatures - runs once at the end of each feature
  • HookType.AfterScenarios - runs at the end of each scenario

To register a handler, call cucumber.addHook:

cucumber.addHook(HookType.BeforeFeatures, function (world, annotations) {
  // do some stuff

The handler functions get two parameters:

  • world - the world object returned from createWorld - for BeforeAll and AfterAll this is not relevant and is always null
  • annotations - an array of any annotations defined on the feature and/or scenario (if relevant)

You can use the annotations parameter to do custom setup behaviour depending on annotations set on the test.

The context of the hook handling function (i.e., this) will be the current feature or scenario, depending on the hook type.

Data tables

You can define data tables in your specs like so:

Feature: Using tables
  Scenario: lots of data
    Given lots of data
      | Header 1 | Header 2 | Header 3 |
      | Value 1a | Value 1b | Value 1c |
      | Value 2a | Value 2b | Value 2c |
    When I use 3 key-value pairs
      | Key 1 | Value 1 |
      | Key 2 | Value 2 |
      | Key 3 | Value 3 |
    And I have a list
      | Value 1 |
      | Value 2 |
      | Value 3 |
    Then I can access all that data

And write rules for them like so:

cucumber.defineRule('lots of data', (world, table) => {
  const obj = table.asObjects();
    'Header 1': 'Value 1a',
    'Header 2': 'Value 1b',
    'Header 3': 'Value 1c'
  const raw = table.raw();
     [ 'Header 1', 'Header 2', 'Header 3' ],
     [ 'Value 1a', 'Value 1b', 'Value 1c' ],
     [ 'Value 2a', 'Value 2b', 'Value 2c' ],
cucumber.defineRule(/^I use (\d+) key-value pairs$/, (world, number, table) => {
  const obj = table.asKeyValuePairs();
  expect(obj['Key 2']).toEqual('Value 2');
cucumber.defineRule('I have a list', (world, table) => {
  const list = table.asList();
  expect(list).toEqual(['Value 1', 'Value 2', 'Value 3']);

Leading and trailing spaces are automatically trimmed, which you can get round by enclosing the value in quotes. Because values with a quote character at the start and end will be treated as strings, if you need to keep the quote characters, you must put another pair of escaped quotes inside the string.

| " value with surrounding spaces " |
| "\"value with quotes at start and end\"" |

See the DataTable class for more information.

Scenario outlines

A simlar concept not to be confused with data tables is the scenario outline:

Scenario Outline: acceptable numbers should be accepted (<number>)
  Given I enter the value "<number>"
  When I click next
  Then I should not see errors
  | number |
  | 1      |
  | 123456 |
  | 1.2    |
  | .123   |
  | -1234  |
  | -12.34 |
  | -.1234 |

This allows you to write a kind of templated scenario, with gaps left for data that will be filled in from the "Examples" table. This table is read as a header line defining the variable names, followed by rows with columnar data.

The parser will output a regular scenario for each (non-header) row in the table, with any variable in angle brackets replaced by its value in the table for that row.

This way you can test a load of different inputs with relatively little typing.

Pro-tip: the parser replaces angle bracket substitutions anywhere within the scenario including the title - it's a good idea to put a variable substitution in the title so you can easily see which ones have failed.

Background steps

You can define steps that will run before each scenario, using the Background: keyword:

  Given I log in as
  And I go to the page
  Given I do a thing

The steps under Background: will be prepended to each scenario, and will use the same world as that scenario.

Inline rules

You can define a rule in the feature file itself. This is useful for making short rules out of repeated steps. E.g.:

  Rule: I enter {title:word} {forename:word} {surname:word} as my name
    * I enter "<title>" in title
    * I enter "<forename>" in forename
    * I enter "<surname>" in surname

  Scenario: enter name
    Given I enter Mr Arthur Dent as my name

The rule is defined using the Rule: keyword, and can match arguments using the named template syntax. These arguments are then substituted into the "sub" rules using the angle bracket syntax.

Rules are scoped to the feature file.


Cucumber class

The JavaScript behind the Gherkin, for defining rules etc.

import { Cucumber } from 'stucumber';

The methods below are meant to be called by people writing tests. The other methods on the class are called by translated gherkin tests.

defineCreateWorld(_createWorld: () => any): void

Defines a factory function for creating a "world", which is passed to every rule. This should be some object which holds the context of your test.


  • _createWorld - a function which returns an instance of a world, whatever that might be

defineRule(match: string, handler: RuleHandler): void
defineRule(match: RegExp, handler: RuleHandler): void

Defines a rule. All rules, whether Given, When, Then or And are treated the same way. When a rule matches match, the handler function will be called.


  • match - either a string or regex that defines what the rule will match
  • handler - a function to execute when the rule is matched (see RuleHandler)

The first argument to the handler is always the world instance.

If match is a string, it can contain placeholders for matched arguments (see template strings). The values matched by the placeholders will be passed to the handler in the order they appear in the string.

If match is a RegExp, any capturing groups will be passed as separate arguments to the handler.

addHook(type: HookType, handler: HookHandler): void

Adds a hook, i.e., a function which will run before or after features or scenarios.



import {RuleHandler} from 'stucumber';

Interface representing a handler function for a rule. The first argument is world, the value returned from the createWorld function. Subsequent arguments are the values for the capturing groups (or placeholders) defined for the rule. The final argument is a the data table if defined.

DataTable class

Represents a data table.

raw(): string[][]

Returns the raw table data, as an array of rows of data, themselves arrays of cells.

asObjects(): Hash<string>[]

Treats the first row as a header containing the names of the columns, and returns an array of objects based on those names. For example, given the following table:

| foo | bar |
| 1   | 2   |
| 3   | 4   |

The following value will be returned:

  {foo: '1', bar: '2'},
  {foo: '3', bar: '4'}

asKeyValuePairs(): Hash<string> asKeyValuePairs<T>(mapValue: (value: string, key?: string) => T): Hash<T>

Expects a 2 column table, where the first column represents keys and the second represents values, and returns an object containing that data.

For example, given the following table:

| foo | 1  |
| bar | 2  |
| baz | 3  |

The following value will be returned:

{foo: '1', bar: '2', baz: '3'}

Optionally, you can provide a function to convert the values:

const obj = data.asKeyValuePairs((value) => parseInt(value));

asList(): string[] asList<T>(mapValue: (value: string, i?: number) => T): T[]

Expects a 1 column table and returns the values as an array. Optionally, you can provide a function to convert the values, e.g.:

const list = data.asList((value) => parseInt(value));

Given the following data table:

| 1 |
| 2 |
| 3 |

The following value will be returned:

[1, 2, 3]

Background steps

You can define steps that will run before each scenario, using the Background: keyword:

  Given I log in as
  And I go to the page
  Given I do a thing

The steps under Background: will be prepended to each scenario, and will use the same world as that scenario.


ISC. Do what you like, don't sue me. Please consider sharing improvements you make.


npm i stucumber-ext

Downloadsweekly downloads









last publish


  • avatar
Report a vulnerability