A collection of pre-commit
hooks for handling text files.
In particular, hooks for handling unicode characters which may be undesirable in a repository.
To use with pre-commit
, include this repo and the desired hooks in
.pre-commit-config.yaml
:
- repo: https://github.com/sirosen/texthooks
rev: 0.6.6
hooks:
- id: alphabetize-codeowners
- id: fix-smartquotes
- id: fix-ligatures
Each hook is usable as a CLI script. Simply
pip install texthooks
and then invoke, e.g.
fix-smartquotes FILENAME
Hook | Description |
---|---|
alphabetize-codeowners |
Alphabetize names in CODEOWNERS files. |
fix-smartquotes |
Replace curly quotes with ASCII quotes. |
fix-spaces |
Normalize special space markers to ASCII spaces. |
fix-ligatures |
Convert stylistic ligatures to ASCII text. |
forbid-bidi-controls |
Check for bi-directional text. |
macro-expand |
A simple way to write text formatting macros. |
Normalize CODEOWNERS
files to always list people and teams in the same order
by alphabetizing.
The default hook targets CODEOWNERS
, .github/CODEOWNERS
, and
docs/CODEOWNERS
.
alphabetize-codeowners
alphabetizes the lists of owners per path.
It does not alphabetize the lines in the file or otherwise sort them.
Any comment lines or empty lines should be left unmodified by the hook.
On the lines which are modified, the hook will normalize the line to have no leading whitespace, and to separate codeowner names with a single space character.
This fixes copy-paste from some applications which replace double-quotes with curly quotes. It does not convert corner brackets, braile quotation marks, or angle quotation marks. Those characters are not typically the result of copy-paste errors, so they are allowed.
Low quotation marks vary in usage and meaning by language, and some languages use quotation marks which are facing "outwards" (opposite facing from english). For the most part, these and exotic characters (double-prime quotes) are ignored.
In files with the offending marks, they are replaced and the run is marked as failed.
Two options are available for specifying exactly which characters will be replaced. For ease of use, they are specified as hex-encoded unicode codepoints.
Suppose you wanted to avoid replacing the "Heavy single comma quotation
mark ornament" (275C
) and the "Heavy single turned comma quotation mark
ornament" (275B
) characters. You could override the single quote codepoints
as follows:
- repo: https://github.com/sirosen/texthooks
rev: 0.6.6
hooks:
- id: fix-smartquotes
# replace default single quote chars with this set:
# apostrophe, fullwidth apostrophe, left single quote, single high
# reversed-9 quote, right single quote
args: ["--single-quote-codepoints", "0027,FF07,2018,201B,2019"]
Replace various unicode space characters with " "
.
This normalizes No-Break Space and similar characters to ensure that your files render the same way in all contexts. It does not modify newlines, carriage returns, or other whitespace characters outside of the Space Separator category.
An option is available for specifying exactly which characters will be replaced. For ease of use, they are specified as hex-encoded unicode codepoints.
Suppose you wanted to only replace Thin Space (codepoint 2009
).
You could override the space codepoints as follows:
- repo: https://github.com/sirosen/texthooks
rev: 0.6.6
hooks:
- id: fix-spaces
args: ["--separator-codepoints", "2009"]
Automatically find and replace ligature characters with their ascii equivalents.
This replaces liguatures which may be created by programs like LaTeX for
presentation with their strictly-equivalent ASCII counterparts. For example,
fi
and ff
may be ligature-ized.
This hook converts these back into ASCII so that tools like grep
will behave
as expected.
This is checker which forbids the use of unicode bidirectional text control characters.
These are directional formatting characters which can be used to construct text
with unexpected or unclear semantics. For example, in programming languages
which allow bidirectional text in statements, "X" = ייִדיש
can be written
with right-to-left reversal to mean that the variable ייִדיש
is assigned a
value of "X"
.
Replace simple "macro" strings in text. This fixer is a no-op if no macro
arguments are supplied. Add --macro
to arguments to do replacements.
For example, convert issue:NNN
to an issue link in markdown with the
following sample config:
- repo: https://github.com/sirosen/texthooks
rev: 0.6.6
hooks:
- id: macro-expand
args:
- "--macro"
- "issue:"
- '[texthooks#$VALUE](https://github.com/sirosen/texthooks/issues/$VALUE)'
- Bugfix for empty line handling in
alphabetize-codeowners
alphabetize-codeowners
is now case insensitive and normalizes whitespace better. Thanks @kurtmckee for the PR!
- Add support for Gitea and GitLab CODEOWNERS files to
alphabetize-codeowners
. Thanks @adam-moss for the PR!
- Reduce length of hook names for
pre-commit.com
requirements
- Minor whitespace bugfix for
alphabetize-codeowners
- Bugfix for
alphabetize-codeowners
stripping ending newlines
- Add
alphabetize-codeowners
fixer (ported from sirosen/alphabetize-codeowners)
- Fix a bug in fixers when running on Windows which could cause data to be written with the wrong encoding
- Add
-v/--verbose
and-q/--quiet
flags to tune output verbosity
- Add
fix-spaces
fixer
- Minor fixes to docstrings
- Add the macro-expand fixer
- Fix a bug in CLI argument handling for all hooks
- Fix a typo in
forbid-bidi-controls
entrypoint
- Add the
forbid-bidi-controls
hook - Adjust the handling of file encodings. Files will be read with UTF-8 encoding by default in most cases.
- Initial release with
fix-ligatures
andfix-smartquotes
hooks