* Fix inverted POSIX check, refactor os_compat, and add tests for it
* Fix missing parentheses and remove skip_win on test that is passing in Windows now
* Fix expected quotes in quoted args
* Make output clearer on failing test
* Bringing skip_win back to test whose failure is a bit more complicated than expected
* Introduce -reminisce, -month, -day, and -year
* Update expected_args in parse_args tests
* Add check before creating compare_d
* Misc changes
* Implement testing for -month, -day, -year, and -reminisce
* Compress tests into one Scenario Outline
* Fix failing tests by updating dates_similar journal
* Create 'we set current date and time to' step
* Use time.parse in reminisce
* Update dates_similar journal
* Make 'Searching in a journal' test shorter
* Lint
* Implement reminiscing test
* Add combination tests
* Finalize tests
* Finalize pytests
* Simplify reminisce tests
* Change reminsice help (since it also shows today's entries)
* Re-do tests; use various tests
* Remove old test data
* Better scenario description
* Standardize format for compare_d
* Rename -reminisce to -today-in-history
* Create datecount plugin
* Fix plugin import
* Update datecount format
* Create datecount test
* Remove outdated comment
* Remove unnecessary datecount export condition
* Update test config
* Move get_date_counts into DateCountExporter; misc changes
* Use --format instead of --export in datecount test
* Update datecount test to include configs
* Better datecount test
* Remove old tests
* Change 'datecounts' to 'dates'
As a beginner in Python it wasn't clear that `poetry install` needed to be run before `make test` for instance. That threw a sort of obscure error until I figured it out what the solution was. This is extremely minor, but make sure installing dependencies is the very first instruction can reduce some friction when contributing.
* Moving configuration values and methods from install.py to config.py -- everything is broken right now
* Using context to store config path - still lots broken
* Use mocks and context.config_path to store test configs - many tests still broken though
* Update changelog [ci skip]
* Fix jrnl --ls crash
* Fix crash when no editor configured
* Attempt to patch config path with test data - doesn't appear to be working
* Properly use patched config path and add config given to scenario that wasn't using it
* Fix copypasta
* Fix editor test that needed patched config and trapping for system exit
* Add exception and handling for when configuration directory is actually a file
* Remove extraneous comment
* Use more generic JrnlError with messaging switchboard
* Format code a bit nicer
* Remove unnecessary given in diagnostic test
* Ensure full error message is output
* Remove unnecessary whitespace characters
* Create tests and steps for temporary filename suffix
* Have temporary filename suffix be -{template_filename} or .jrnl
* Finalize extension_editor_file
* Make suffix local variable in get_text_from_editor
* Fixes for search on docs site
We previously didn't include the search results page in our CI testing,
so we missed some issues on that page. This ensures that page is part of
regular testing, and also includes fixes for the issues present.
* fix sidebar contrast
* add background-color as fallback for gradients on docs site
This is both for old browser support, and for accessibility. Some
accessibility tools turn off gradients since they can make text hard to
read for those with visual impairment.
* change mobile header color for docs site
The mobile header was using a light-purple to dark-purple gradient, even
though every other gradient on the site is mid-purple to dark-purple.
This commit changes that both for consistency on the site, and because
light-purple causes an accessibility issue with the white text in the
header (even though the accessibility tools don't notice because of the
gradient).
* Fix version header detection in changelog
This updates the regex to find the version header in our changelog.
There was a bug that affected patch, alpha, and beta versions. For
example, if we were looking for "v2.5" we might find "v2.5.1" or
"v2.5.1-alpha" or other versions like that.
This led to the changelog not being cleared properly on update, and old
versions sticking around longer than they should.
* re-add option to include author link for merged PRs in changelog
This option was accidentally left out when migrating the changelog
generator tool to github actions.
* fix conditional in changelog workflow
In github actions, booleans always end up as strings. So, it's not
enough to test `env.FULL_RELEASE` because both 'true' and 'false' will
evaluate as true. This fixes the conditional to check for the string
value 'true' instead of relying on a boolean.
