Btukfyl

I'm working with Behave in Python 3.6 - it's great, totally, love it, however, I have to use junit reports. The --junit switch works fine, but the output is really heavy because it includes all the feature gherkin definitions. I'd rather just eliminate the whole CDATA field, or at least stop behave from including the feature definitions.

Basically - how do you stop Behave from copying the feature file into the junit reports?:

   <system-out>

   <![ CDATA[

   @scenario.begin



   Scenario: User fills zipcode on form

   Given user is on a visitor site page ... passed in 0.749s

   """

   /urlpath/

   """

   When a user enters a zip code in a form ... passed in 0.193s

    | action_method   | action_params   |

    | move_to_element | #lead-zip    |

    | send_keys       | 27610           |

    | move_to_element | #lead-submit |

    | click           |                 |

    Then a call is sent for a unique field value pair ... passed in 4.027s

    | unique_field | unique_value                      |

    | description  | Service Request Flow entry button |

   And the segment call contains parameters ... failed in 0.000s

    | prop_key               | prop_value                        |

    | activityLocation       | Visitor : Piffle                  |

    | categoryId             | 1002                              |

    | description            | Service Request Flow entry button |

    | marketId               | 100001                            |

    | userId                 |                                   |

    | userSelectedZipCode    | 27610                             |

    | visitorPageCategory    | Rofing                            |

    | visitorPageGeo         | RALEIGH/DURHAM                    |

    | visitorPageGeoCategory | RALEIGH/DURHAM - ROOFING          |



   @scenario.end

  --------------------------------------------------------------------------------

  ]]>

I found a solution that works for me, though it seems like you should still be able to squelch the default output (which basically copies the feature file into the CDATA -- kind of useless)

If you want more control over what goes into the CDATA section, here is a solution:
Copy junit.py from env/lib/python3.6/site-packages/behave/reporter/junit.py to a directory you will maintain as a custom reporter. I put mine in a folder called reporters

(the above path assumes you have a virtualenv called 'env' and you are using python3.6 -- adjust based on your environment)

In that file, find the method, def CDATA(text=None):
modify the text variable as you wish - apply filters or just set text = '' if you don't need it.

Then in your environments.py, add this to the imports area:
from reporters.junit import JUnitReporter

Finally, in the before_all(context) method, add this:

junit_reporter = JUnitReporter(context.config)

context.config.reporters.append(junit_reporter)

Now you have your very own custom JUnit xml report generator!