* Check errors don't exist as valid words in the aspell dictionary
* Install aspell on Travis
* Add some missing packages
* Remove a virtual package
* Just install the version of aspell-python we need
* Keep flake8 happy
* Switch to warnings and count them, so we can see all the aspell errors in one go
* Handle different encoding of the word and aspell
* Try and fix the encoding conversion
* Find out the encoding type
* Don't assert on number of warnings
* Don't record warnings for now
* Warn on all the encoding options
* pprint the encoding
* More warning work
* Use the actual encoding type
* Correct the logic
* ENH: Multi dict support
* FIX: Fixes after merge
* FIX: Better error check
* FIX: More thorough testing, locations
* FIX: Try newer aspell
* FIX: Move to new dict
* FIX: Move
* FIX: Restore removals from #1181
* FIX: One from #1362
* Add rare chack->check, cheque,
* Minor tidy of some dictionary check code
* Add some more suggestions.
* Fix the whitespace
* Really fix the whitespace
* FIX: Refactor requirement
* Log an error when aspell not found and not required
* Fix the error logging
* Test all variants of present and missing from aspell
* Undo some tuple tidying
* Fix the true/false values used
* Skip some flake8 tests
* Fix the test cases
* Correct the not in aspell test and fix some test cases
* Remove a duplicate test
* Use a test word that isn't a typo
* Set the ideal aspell detection logic for each dictionary
I suspect we'll have to relax this as more obscure words won't be in the aspell dictionary
* Be more realistic given the size of the dictionary
* Fix a flake8 error
* Fix another line length error
* FIX: Move
* FIX: Make visible, simplify
Co-authored-by: Eric Larson <larson.eric.d@gmail.com>
Make it possible to add custom system-wide ignored directories with an
alias in shell rc script, e.g.:
alias codespell="codespell --skip=".git,node_modules"
and allow to extend this list by executing codespell alias with extra
--skip argument.
* Add context options: -A/-B/-C; Fixes#287
* Fix indentation of line continuation.
* Fix 'only before context' case.
* Add some tests for the new context options.
* Refactor context code (pass context tuple, instead of before/after variables)
* Get rid of 'remove temp file code' in context test.
* Fix reintroduced error (regarding 'only before context').
* Add a test for skipping globs with directory and filename in
* Forcing a Travis build
* Tidy the Travis force
* Skip paths with the glob mathing too
* Add a check for new hidden file functionality
* Fix the code to add the new options
* Fix the test syntax
* Fix the flake8 error
* Improve help text
The current code already allows skipping directories, but the full path
must be given, e.g. "./.git" instead of just ".git", as "codespell ." is
the default.
Allow --skip to be used with relative directory names, too.
The current code already allows skipping directories, but the full path
must be given, e.g. "./.git" instead of just ".git", as "codespell ." is
the default.
As the name of the directory is constant for all files within, move the
check for the directory before the loop.
Also skip descending.
A new command line option -r / --regex is implemented to allow the user
to set the regular expression which is used to find the words in the
input lines. This is useful in projects where you want to overwrite the
default expression, for example to split words after an underscore like
in ID_ESTABLISED.
The new regex option cannot be used together with the --write-changes
functionality since it is not guarantied that this will not break code.
The regular expression for splitting a line into words can be compiled
at startup instead for every file again. This gives some minor
performance improvements, especially when parsing a lot of files.
Enhance the --dictionary option so that multiple dictionary files can
be specified at the command line. With this feature projects can use
the standard dictionary and in addition their own custom files.
