Skip to main content

Data Driven

CSV data feeds​

Since v1.0.0

CSV data files can be passed into the interpreter to perform feature execution over multiple data sets. The first row in a CSV data file must be a list of column names for the contained data that follows. The values in each record will be bound to attributes in the feature scope having these same names. Feature steps can reference the bound data using these names.

Say you have a users.csv file containing following user records.

CSV

File: data/users.csv

FIRST_NAME,LAST_NAME,JOB
Gwen,Stefani,Pop Singer
Gwyneth,Paltrow,Actress
Gwen,Cooper,Torchwood Agent

The first row in this file specifies a comma separated list of column names. The remaining rows specify the comma separated lists of values for each record. Now say you wanted to submit the details for each user to a web page through a single submit-user.feature file that accepts all the user data as input one at a time. You can do this in the feature by referencing CSV data values by their column names as follows:

Feature

File: features/submit-user.feature

Feature: Submit user ${FIRST_NAME} ${LAST_NAME}

  Scenario: submit user details
    Given I navigate to the user details page
     When I type FIRST_NAME in the first name field
      And I type LAST_NAME in the last name field
      And I type JOB in the job field
      And I click the save button
     Then the alert msg should be "User ${FIRST_NAME} saved"

Note: Interpolation of CSV fields ${FIRST_NAME} ${LAST_NAME} in the feature name and description is supported since v3.12.1

The current CSV record number (starting at 1) is available as an implicit attribute named data record number and can be accessed by reference or interpolation.

You can execute this feature for all users in the CSV file by launching Gwen with the -i|--input-data CLI option.

Examples​

yarn gwen -b -i gwen/data/users.csv gwen/features/submit-user.feature

or 

yarn gwen -b --input-data gwen/data/users.csv gwen/features/submit-user.feature

Gwen will read in the CSV file, and for each record will:

  • Bind the data to the feature
  • And execute the feature

A feature detail report will be generated for each data record. If you specify the --parallel CLI option, the feature will execute all records in parallel.

@Examples annotation​

Since v2.10.0

Consider the following scenario outline with an examples table

Example

Scenario outline with inlined data

  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item
    
    Examples:
      | Item         | Status  |
      | Walk the dog | pending |
      | Get the milk | done    |
      | Feed the cat | pending |

The data table in the examples clause can be externalised to a CSV file and imported using the @Examples annotation as follows:

Example

CSV File: data/items.csv

Item,Status
Walk the dog,pending
Get the milk,done
Take out trash,pending

Scenario outline with externalised data

  @Examples("data/items.csv")
  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item

This will resolve to an identical outline when expanded at runtime:

  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item
    
    Examples: Data file: data/items.csv
      | Item         | Status  |
      | Walk the dog | pending |
      | Get the milk | done    |
      | Feed the cat | pending |

Where filter​

Since v3.27.0

The data can also be filtered using a where JavaScript expression filter. For example, to get all the records where the data in the Status column matches pending:

Example

Scenario outline with externalised and filtered data

  @Examples(file="data/items.csv",where="'${Status}'=='pending'")
  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item

Will resolve to the following containing pending Status records only:

  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item
    
    Examples: Data file: data/items.csv, where: '${Status}'=='pending'
      | Item         | Status  |
      | Walk the dog | pending |
      | Feed the cat | pending |

String interpolation syntax can also be used to reference a value in the CSV record or current scope. For example, to get all the records where the Status column matches the value bound to an in-memory attribute named SelectedStatus:

Example

Scenario outline with externalised and filtered data

  @Examples(file="data/items.csv",where="'${Status}'=='${SelectedStatus}'")
  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item

Name prefix​

Since v3.27.0

A prefix="string" attribute can be added to prepend a prefix to each column name in the CSV heaader record. This is useful for avoiding any potential name clashes with existing data in the current scope.

Example

Scenario outline sourced from CSV file having column names Status, Item

  @Examples(file="data/items.csv",prefix="todo.",where="'${todo.Status}'=='pending'")
  Scenario Outline: I load items from CSV files
    When I add a <todo.Status> "<todo.Item>" item

Will result in the column names todo.Status, todo.Item when todo. prefix is applied at runtime

  Scenario Outline: I load items from CSV files
    When I add a <todo.Status> "<todo.Item>" item
    
    Examples: Data file: data/items.csv, prefix: todo., where: '${todo.Status}'=='pending'
      | todo.Item    | todo.Status |
      | Walk the dog | pending     |
      | Feed the cat | pending     |

Required flag​

Since v3.20.0

A required=true attribute can be added to raise an error if no data is returned by the where clause filter.

Example

Scenario outline which raises an error if externalised and filtered data is empty

  @Examples(file="data/items.csv",where="'${Status}'=='${SelectedStatus}'",required=true)
  Scenario Outline: I load items from CSV files
    When I add a <Status> "<Item>" item