Update to 1.15 - compatible with behave version 1.3.0

[modehnal]

Fix behave version to 1.3.0.
Fix before scenario step handling (new behave calls methods in reverse).
Add default background to unhandled scenario status.
Support late registration of formatter in before_scenario. Closes #94
Fix dependency on external tools in tests. Closes #96

[modehnal]

@jenisys Would you know why the tests are failing like this? https://github.com/behave-contrib/behave-html-pretty-formatter/actions/runs/16989669598?pr=97

KeyError: "'__name__' not in globals"

I am trying to figure out where the problems is but no luck for now. Perhaps you also dealt with in behave?

[modehnal]

I mean, I can add to all files that the problem is in:

if "__name__" not in globals():
    __name__ = "__main__"

but something is weird here, how did the main got removed from globals?

[modehnal]

@jenisys Alright, it seems somewhere between 1.2.6 and 1.3.0 something got broken with globals in tests/acceptance. Do you have any idea where the issue might be? The 1.3.1 fails, the 1.3.0 fails, the 1.2.6 is alright, as can be seen here.

Reverting back to 1.3.1

[jenisys]

Mmh, what should I say:

• Never ever put a Python package in the steps directory, meaning: behave4cmd0
• A Python package or step-library belongs somewhere, where it is in the Python search path.
• Use PYTHONPATH=$PWD environment variable and place it into the root-directory of the repo or similar (or install them globally or in a virtual-environment).
• For example, use direnv to ensure that the environment is set up when you enter the repo directory.

WHAT IS THE PROBLEM (the failures in the CI test run):

• Relative imports, like from . import command_steps, can only be used within a Python package, but not in Python module or Python scripts.
• Because you placed behave4cmd0 under the steps directory (and @bittner merged a pull-request that recursively imports step-modules below the steps directory), this constraint is violated now. Steps are basically loaded like Python module/script and not a Python package.

ADDITIONAL FIXES THAT I NEEDED:

# -- FILE: tests/acceptance/features/help-screen.feature
...
     Then it should pass
     And the command output should contain:
       """
-      Available formatters:
+      AVAILABLE FORMATTERS:
+        captured       Inspect captured output.
         html-pretty    Pretty HTML Formatter
         json           JSON dump of test run
       """

# -- FILE: tests/acceptance/features/html-output.feature
# NEEDED: feature-file, steps-directory even in dry-run mode
...
+    And a file named "features/steps/use_behave4cmd0_steps.py" with:
+      """
+      from behave4cmd0 import passing_steps
+      """
+    And a file named "features/passing.feature" with:
+      """
+      Feature: Passing
+        Scenario: One
+          Given a step passes
+          When another step passes
+      """
     When I run "behave --format html-pretty --dry-run"
     Then it should pass
     And the command output should contain:

# -- FILE: tests/formatter_features/behave.ini
+[behave]
+default_format = pretty
+# -- ENSURE: At least one html-pretty formatter is used.
+format = html-pretty
+output = report.html
...

[modehnal]

@jenisys thank you, these hints were very useful.

Since I did not setup the behave4cmd0, it did not occur to me it could be the reason (this kind of error is a first for me, next time I will know what to focus on). Moving the behave4cmd0 from steps and some minor changes were enough to fix it.

[fpokryvk]

LGTM.