* Encapsulate all multiline strings in triple-quotes in Gherkin files
Since pytest-bdd v8.0.0 uses the official Gherkin parser, multiline
strings must now be encapsulated by triple-quotes.
See:
- https://pytest-bdd.readthedocs.io/en/stable/#id2
- https://pytest-bdd.readthedocs.io/en/stable/#docstrings
* Remove comments in Gherkin files causing test breakage
These comments break the step matching.
* Fix compatibility of step-functions matching on multiple lines
In pytest-bdd v8.0.0 it is no longer possible to match based on multiple
lines, which breaks essentially all steps that support docstrings. Solve
this by adding a wrapper-function for each of these instances, that
matches the docstring step, and calls the original function.
So, what used to be:
@then(parse("the output should match\n{regex}"))
@then(parse('the output should match "{regex}"'))
def output_should_match(regex, cli_run):
...
Is now:
@then(parse("the output should match"))
def output_should_match_docstring(cli_run, docstring):
output_should_match(docstring, cli_run)
@then(parse('the output should match "{regex}"'))
def output_should_match(regex, cli_run):
...
There is possibly a way around this that is much better than what I've
done here, but this is a start at least.
* Update version requirement of pytest-bdd to >=8.0
* Update tox config to match poetry config
---------
Co-authored-by: Micah Jerome Ellison <micah.jerome.ellison@gmail.com>
* Support pytest_bdd 7.1.2 and later
Fix#1875.
* Prevent use of pytest 8.1 for now due to unrelated error
* poetry lock --no-update
---------
Co-authored-by: Micah Jerome Ellison <micah.jerome.ellison@gmail.com>
* Stop ignoring parsedatetime warning flag style warning
* Use new VERSION_CONTEXT_STYLE to bypass parsedatetime flag style warning
* Replace "flag" and its magic numbers with contextual booleans
* Add ruff
* Add ruff config
* Add ruff rules that look useful and are already passing
* Add more ruff rules after talking with Jonathan
* Add line length exception for acceptably long indented line
* Resolve ruff line length 88 rule in args. Changing small lines but adding a noqa ignore directive to longer lines that look best as they are. Their dedented length is still less than 88
* poe format
* Resolve all remaining ruff line length errors
* Replace flake* and isort with ruff calls
* Add black --check as final lint step. ruff catches most but not all black formatting issues
* Remove unneeded flakeheaven setting
* Remove flake* and isort now that ruff is handling all their business
* Update pyproject, lockfile with latest version of ruff
* Document each ruff rule with comment
* Add black --version call before black --check
* Remove extraneous period
* Move path concerns to path.py and template concerns to editor.py -- BDD tests are failing
* Move path-related constants from config.py to path.py
* Mock get_templates_path in its new calling file
* Mediate template arg vs. config in controller then read template text in editor and unify those two use cases.
Some tests still failing
* Fix test whose message had changed
* poe format
* Refactor for easier unit testing and add unit tests
* Use path strings instead of Path objects in return values to prevent side effects that caused unit tests to fail on some platforms
* poe format
* Attempt to bypass getcwd errors in CI with patch
* Consistently use strings for paths instead of some strings and some pathlib.Path
* Keep pathlib within a function for readability
* fix for ruamel.yaml versions >=0.17.22
* Run poe format
---------
Co-authored-by: Jonathan Wren <jonathan@nowandwren.com>
* update pytest-bdd to 6.0
* update lock file
* fix first test (inject command fixture to request)
* fix some more tests
* fix cli_run fixture
* fix password fixture
* Remove unused import
* Fix greedy should_or_should_not parsing problems while also consolidating its parse/transformation-to-bool code
* Prevent greedy matching in "we run" by using regular expression lookahead
* Add missing "Outline" in scenario outlines with examples
* Split "we use the config" and "we use no config" so pytest won't try to consume config_file as a fixture
* Fix missing ShouldOrShouldNot
* Formatting
* fix get_fixture function
* change output of failing test to be a little more useful
* update lock file
* update type builder to for should/should not to be in it's own file, rename some vars for readability
* add parse-type new dev/testing dependency
* update lock file
---------
Co-authored-by: Jonathan Wren <jonathan@nowandwren.com>
* Use latest mkdocs version 1.4.2
* Add docs_theme to mkdocs watch so it automatically reloads localhost when modifying theme
* Bring in ReadTheDocs breadcrumb theme and fix display issues around "Docs" root breadcrumb
* The problem is CSS, not the breadcrumbs template. Tightly scope offending CSS and remove breadcrumbs template override
* Bring mkdocs ReadTheDocs breadcrumbs theme back in
* Fix pa11y validation issue by replacing breadcrumb a alt attribute with aria-label
* Attribute source of new breadcrumbs file
* Use mkdocs 1.4 or greater with no upper bound
* run format
* rename cli.py to main.py
* rename jrnl.py to controller.py
* move journal class files into journals dir
* rename start -> run in controller.py
* add black and isort to flakeheaven
* update lock file
* clean up poe config
* run formta on python blocks in markdown file
* disable code for black being confused about markdown
* add cache timeout for flakeheaven
See: https://github.com/flakeheaven/flakeheaven/issues/71
* Drop Python 3.9 and use Python 3.11 release instead of pre-release
* Run poetry lock
* Fix need for quoted 3.10 string
* Use 3.11.0-rc.2 for Python 3.11 for now
* Update minimum Python version in installation docs
* Change Python 3.11 RC version to 3.11
* Run docs and release workflows on Python 3.11
* Including Python 3.12 in allowable versions
* Restore poetry.lock from develop
* Run poetry lock --no-update
* Retrieve poetry.lock from develop and lock with --no-update
* poetry lock --no-update
* replace flake8 with flakeheaven
* update pyproject.toml for new flake tool
* update lock file
* change flake8 to flakeheaven in tests
* undo fix for poetry
* remove unused plugin marker for flakeheaven
* remove unused import in markdown file
* Attempt to exclude .venv from flakeheaven runs in CI
* Exclude more dirs from flakeheaven
Co-authored-by: Micah Jerome Ellison <micah.jerome.ellison@gmail.com>
* update docs pipeline to use poe task runner
* move npm dependency into package.json file instead of hardcoding into pipeline
* replace nix-specific call with python version (for windows compat)
* update commands to use python for win compat
* update lock file
* change deps to take out xq and use xmltodict instead
* move script tasks from pyproject into new tasks.py
* run linting on new tasks.py file
* fix typo
* clean up pyproject to make valid toml
* update docs with correct task runner command
* use npx in case node_modules/.bin isn't in the path
* Run pa11y-ci in cross-platform manner (works on Windows)
* Add node_modules to .gitignore
* clean up poe alias
Co-authored-by: Micah Jerome Ellison <micah.jerome.ellison@gmail.com>
* Clean up pyproject file
This reduces the differences between local tasks and tasks run in the CI
pipelines
* remove linting step (it's part of test now)
* remove useless arg
* add xdist for parallel test execution in pytest
* Clean up pyproject file
This reduces the differences between local tasks and tasks run in the CI
pipelines
* remove linting step (it's part of test now)
* remove useless arg
* Get rid of `make` in favor of `poe`
This moves the tasks that were previously in Makefile, into
pyproject.toml (with all the other config)
This is also more inclusive of Windows developers since they only need
Python, and no longer have to install make separately.
* update docs
* don't make code blocks also links
* implement tox for testing
* update command to use new task runner
