BDD Kickstart San Francisco

Step Definitions

TODO Migrate docs from, especially states (successful, undefined etc.)

When Cucumber executes a Step in a Scenario it will look for a matching Step Definition to execute.

A Step Definition is a small piece of code with a pattern attached to it. The pattern is used to link the step definition to all the matching Steps, and the code is what Cucumber will execute when it sees a Gherkin Step.

Step Definitions sources must be placed in a file where Cucumber can find them. The location depends on what Cucumber implementation you are using. See Glue Code for details.

To understand how Step Definitions work, consider the following Scenario:

Scenario: Some cukes
  Given I have 48 cukes in my belly

The I have 48 cukes in my belly part of the step (the text following the Given keyword) will match the Step Definition below.

@Given("I have (\\d+) cukes in my belly")
public void I_have_cukes_in_my_belly(int cukes) {
    // Do something with the cukes

Lists can also be specified:

Given I am available on "Tuesday,Friday,Sunday"
@Given("I am available on \"(.+)\"")
public void I_am_available_on(List<String> days) {
    // Do something with the days
Given(/I have (\d+) cukes in my belly/) do |cukes|
  # Do something with the cukes
Given(~'^I have (\\d+) cukes in my belly') { int cukes ->
    // Do something with the cukes
Given(/^I have (\d+) cukes in my belly$/, function (cukes) {
    // Do something with the cukes
(Given #"^I have (\d+) cukes in my belly$" [cukes]
  (eat (Integer. cukes)))
GIVEN("^I have (\\d+) cukes in my belly$") {
    REGEX_PARAM(int, cukes);
    USING_CONTEXT(MyAppCtx, context);
    // Do something with the cukes
[When(@"^I have (\d+) cukes in my belly$")]
public void CukesInTheBelly(int cukes)
    // Do something with the cukes
let [<Given>] ``^I have (\d+) cukes in my belly$``(cukes:int) =
  (* Do something with the cukes *)
Given("^I have (%d+) cukes in my belly$", function (cukes)
    -- Do something with the cukes
@Given('^I have (\d+) cukes in my belly$')
def I_have_cukes_in_my_belly(self, cukes):
  # Do something with the cukes
Given("""^I have (\d+) cukes in my belly$"""){ (cukes:Int) =>
  // Do something with the cukes

When Cucumber matches a Step against a regular expression in a Step Definition, it passes the value of all the capture groups to the Step Definition’s arguments. Capture groups are strings (even when they match digits like \d+). For statically typed languages, Cucumber will automatically transform those strings into the appropriate type. For dynamically typed languages, no transformation happens by default, as there is no type information.

Cucumber does not differentiate between Given, When and Then. It is up to you what you do inside the Step Definition’s body.

Doc Strings

Doc Strings are handy for specifying a larger piece of text. This is inspired from Python’s Docstring syntax.

The text should be offset by delimiters consisting of three double-quote marks on lines of their own:

Given a blog post named "Random" with Markdown body
  Some Title, Eh?
  Here is the first paragraph of my blog post. Lorem ipsum dolor sit amet,
  consectetur adipiscing elit.

In your step definition, there’s no need to find this text and match it in your Regexp. It will automatically be passed as the last parameter in the step definition. For example:

@Given("^a blog post named \"([^\"]*)\" with Markdown body$")
public void a_blog_post_named_something_with_markdown_body(String title, String markdown) {
    // Save it in the database
Given /^a blog post named "([^"]*)" with Markdown body$/ do |title, markdown|
  Post.create!(:title => title, :body => markdown)

Indentation of the opening """ is unimportant, although common practice is two spaces in from the enclosing step. The indentation inside the triple quotes, however, is significant. Each line of the string passed to the step definition’s block will be de-indented according to the opening """. Indentation beyond the column of the opening """ will therefore be preserved.

Data Tables

Data Tables are handy for specifying a larger piece of data:

Given the following users exist:
  | name  | email           | phone |
  | Aslak | [email protected] | 123   |
  | Matt  | [email protected]  | 234   |
  | Joe   | [email protected]   | 456   |

Just like Doc Strings, they will be passed to the Step Definition as the last argument:

@Given("^the following users exist$")
public void the_following_users_exist(DataTable users) {
    // Save them in the database
See the DataTable API docs for details about how to access data in the table.
Given /^the following users exist$/ do |users|
See the DataTable API docs for details about how to access data in the table.

Substitution in Scenario Outlines

If you use a DocString or DataTable argument in steps in Scenario Outlines, any < > delimited tokens will be substituted with values from the example tables. For example:

Scenario Outline: Email confirmation
  Given I have a user account with my name "Jojo Binks"
  When an Admin grants me <Role> rights
  Then I should receive an email with the body:
    Dear Jojo Binks,
    You have been granted <Role> rights.  You are <details>. Please be responsible.
    -The Admins
    |  Role     | details                                         |
    |  Manager  | now able to manage your employee accounts       |
    |  Admin    | able to manage any user account on the system   |

Data Table diffing

One very powerful feature in Cucumber is comparison of tables. You can compare a table argument to another table that you provide within your step definition. This is something you would typically do in a Then step, and the other table would typically be constructed programmatically from your application’s data.

Beware that the diffing algorithm expects your data to be column-oriented, and that the first row of both tables represents column names. If your tables don’t have some similarity in the first row you will not get very useful results. The column names must be unique for each column – and they must match.

Here is an example of a Data Table that we want to diff against actual results:

Then I should see the following cukes:
  | Latin           | English      |
  | Cucumis sativus | Cucumber     |
  | Cucumis anguria | Burr Gherkin |

A Step Definition can diff the DataTable with data pulled out of your application, for example from a Web page or a Database:

@Then("^I should see the following cukes:$")
public void the_following_users_exist(DataTable expectedCukesTable) {
    // We'd typically pull this out of a database or a web page...
    List<Cuke> actualCukes = new ArrayList();
    actualCukes.add(new Cuke("Cucumis sativus", "Concombre"));
    actualCukes.add(new Cuke("Cucumis anguria", "Burr Gherkin"));


The list passed to diff can be a DataTable, List<YourType>, List<Map> or a List<List<ScalarType>>.

Then /^I should see the following cukes:$/ do |expected_cukes_table|
  # We'd typically pull this out of a database or a web page...
  actual_cukes = [
    ['Latin', 'English'],
    ['Cucumis sativus', 'Concombre'],
    ['Cucumis anguria', 'Burr Gherkin']


The list passed to diff! can be a Cucumber::Ast::Table, Array of Map or an Array of Array of String.

If you are using [Capybara]( and you want to compare a Gherkin DataTable with a HTML table rendered in a Web page you can construct an Array like so:

rows = find("table#selector").all('tr')
table = { |r| r.all('th,td').map { |c| c.text.strip } }

If the tables are different, an exception is thrown, and the diff of the two tables are reported in the Report.

String Transformations

Cucumber provides an API that lets you take control over how strings are converted to other types. This is useful especially for dynamically typed languages, but also for statically typed languages when you need more control over the transformation.

Let’s consider a common example - turning a string into a date:

Given today's date is "10-03-1971"

First of all, this might mean the 10th of March in some countries, and the 3rd of October in others. It’s best to be explicit about how we want this converted. We’ll try to convert it to 10th of March 1971.

Cucumber-JVM knows how to convert strings into various scalar types. A scalar type is a type that can be derived from a single string value. Cucumber-JVM's built-in scalar types are numbers, enums, java.util.Date, java.util.Calendar and arbitrary types that have a single-argument constructor that is either a String or an Object.

Transformation to java.util.Date and java.util.Calendar will work out-of-the-box as long as the string value matches one of the SHORT, MEDIUM, FULL or LONG formats defined by java.util.DateFormat.

It turns out that 10-03-1971 from our example doesn't match any of those formats, so we have to give Cucumber a hint:

@Given("today's date is \"(.*)\"")
public void todays_date_is(@Format("dd-MM-yyyy") Date today) {

Many Java programmers like to use Joda Time. Cucumber-JVM doesn't have any special support for Joda Time, but since Joda's LocalDate has a LocalDate(Object) constructor it is considered a scalar by default.

However, in this case it wouldn't also know how to pass the _format_ string, so you would get an exception when Cucumber instantiates it with new LocalDate("10-03-1971").

A custom formatter gives you full control:

@Given("today's date is \"(.*)\"")
public void todays_date_is(
  LocalDate today) {

The custom transformer looks like this:

public class JodaTransformer extends Transformer<LocalDate> {
    public LocalDate transform(String value) {
        String format = getParameterInfo().getFormat();
        DateTimeFormatter dateTimeFormatter = DateTimeFormat.forStyle(format);
        dateTimeFormatter = dateTimeFormatter.withLocale(getLocale());
        return dateTimeFormatter.parseLocalDate(value);

Of course, you can write transformers for anything, not just dates.

Let's Transform anything that looks like a date into an instance of Date:

require 'date'

Transform(/^(\d\d-\d\d-\d\d\d\d)$/) do |arg|
  Date.strptime(arg, '%d-%m-%Y')

Data Table Transformations

This applies to Cucumber-JVM only

Data Tables can be transformed to a List of various types. We’ll see how the table in the following Scenario can be transformed to different kinds of lists.

Scenario: Some vegetables
  Given I have these vegetables:
    | name     | color |
    | Cucumber | Green |
    | Tomato   | Red   |

List of YourType

The table can be transformed into a list of vegetables:

public class Vegetable {
    public String name;
    public Color color; // Color is an enum

The Step Definition:

@Given("I have these vegetables:")
public void I_have_these_vegetables(List<Vegetable> vegetables) {
    // Do something with the vegetables

The header row is used to name fields in the generic List type.

IMPORTANT: If the generic List type (Vegetable in this case) is a scalar (i.e. it has a String or Object constructor), the header will not be used to name fields in the class. Instead you would get a List that has one Vegetable for each cell (6 in this case). See List of Scalar below.

List of Map

You can also transform a DataTable to a list of maps:

@Given("I have these vegetables:")
public void I_have_these_vegetables(List<Map<String, String>> vegetables) {
    // Do something with the vegetables

The Key and Value generic types of the Map can be any kind of scalar type.

List of List of scalar

You can also convert it to a list of list scalar:

@Given("I have these vegetables:")
public void I_have_these_vegetables(List<List<String>> vegetables) {
    // Do something with the vegetables

This will convert it to a flattened list like this:

[["name", "color"], ["Cucumber", "Green"], ["Tomato", "Red"]]

You can also convert it to a list of scalar:

List of scalar

@Given("I have these vegetables:")
public void I_have_these_vegetables(List<String> vegetables) {
    // Do something with the vegetables

This will convert it to a flattened list like this: ["name", "color", "Cucumber", "Green", "Tomato", "Red"]

Map of scalar

@Given("I have these vegetables:")
public void I_have_these_vegetables(Map<String,String> vegetables) {
    // Do something with the vegetables

This will convert it to a map list like this: {"name" => "color", "Cucumber" => "Green", "Tomato" => "Red"}

Note that this only works for tables with 2 columns. If you’re converting into a Map it is usually adviseable that the Gherkin table doesn’t have a top row naming the columns.


I want to scope a step definition to a scenario or feature

TODO: Summarise info from: