Formatters and Reporters

behave provides 2 different concepts for reporting results of a test run:

  • formatters
  • reporters

A slightly different interface is provided for each “formatter” concept. The Formatter is informed about each step that is taken. The Reporter has a more coarse-grained API.


The following reporters are currently supported:

Name Description
junit Provides JUnit XML-like output.
summary Provides a summary of the test run.


The following formatters are currently supported:

Name Mode Description
help normal Shows all registered formatters.
json normal JSON dump of test run
json.pretty normal JSON dump of test run (human readable)
plain normal Very basic formatter with maximum compatibility
pretty normal Standard colourised pretty formatter
progress normal Shows dotted progress for each executed scenario.
progress2 normal Shows dotted progress for each executed step.
progress3 normal Shows detailed progress for each step of a scenario.
rerun normal Emits scenario file locations of failing scenarios
sphinx.steps dry-run Generate sphinx-based documentation for step definitions.
steps dry-run Shows step definitions (step implementations).
steps.doc dry-run Shows documentation for step definitions.
steps.usage dry-run Shows how step definitions are used by steps (in feature files).
tags dry-run Shows tags (and how often they are used).
tags.location dry-run Shows tags and the location where they are used.


You can use more than one formatter during a test run. But in general you have only one formatter that writes to stdout.

The “Mode” column indicates if a formatter is intended to be used in dry-run (--dry-run command-line option) or normal mode.

User-Defined Formatters

Behave allows you to provide your own formatter (class):

# -- USE: Formatter class "Json2Formatter" in python module ""
# NOTE: Formatter must be importable from python search path.
behave -f ...

The usage of a user-defined formatter can be simplified by providing an alias name for it in the configuration file:

# -- FILE: behave.ini
# ALIAS SUPPORTS: behave -f json2 ...
# NOTE: Formatter aliases may override builtin formatters.
json2 =

If your formatter can be configured, you should use the userdata concept to provide them. The formatter should use the attribute schema:

# -- FILE: behave.ini
behave.formatter.json2.use_pretty = true

#    behave -f json2 -D behave.formatter.json2.use_pretty ...

More Formatters

The following contributed formatters are currently known:

Name Description
allure allure-behave, an Allure formatter for behave.
html behave-html-formatter, a simple HTML formatter for behave.
teamcity behave-teamcity, a formatter for JetBrains TeamCity CI testruns with behave.

The usage of a custom formatter can be simplified if a formatter alias is defined for.


# -- FILE: behave.ini
# FORMATTER ALIASES: "behave -f allure" and others...
allure = allure_behave.formatter:AllureFormatter
html = behave_html_formatter:HTMLFormatter
teamcity = behave_teamcity:TeamcityFormatter

Embedding Screenshots / Data in Reports

Hint 1:Only supported by JSON formatter
Hint 2:Binary attachments may require base64 encoding.

You can embed data in reports with the Context method attach(), if you have configured a formatter that supports it. Currently only the JSON formatter supports embedding data.

For example:

# -- FILE: features/steps/
from behave import fiven, when
from behave4example.web_browser.util import take_screenshot_and_attach_to_scenario

@given(u'I open the Google webpage')
@when(u'I open the Google webpage')
def step_open_google_webpage(ctx):
# -- FILE: behave4example/web_browser/
#   * BROWSER-SPECIFIC: Implementation may depend on browser driver.
def take_screenshot_and_attach_to_scenario(ctx):
    # -- HINT: SELENIUM WITH CHROME: ctx.browser.get_screenshot_as_base64()
    screenshot_image = ctx.browser.get_full_page_screenshot_as_png()
    ctx.attach("image/png", screenshot_image)
# -- FILE: features/
# EXAMPLE REQUIRES: This browser driver setup code (or something similar).
from selenium import webdriver

def before_all(ctx):
    ctx.browser = webdriver.Firefox()