A fully configurable and extensible language-agnostic Hook Management System for GIT
Auto checks your code before you ship it. Works with any programmning language. If not, let me know.
Production version is available from Pypi, and development branch is also published by Travis-CI to Pypi TestServer. Both can be downloaded and installed via the pip command.
Here. There is lots of quick information, and as well other githooks management approaches.
These instructions will show you how to install and use the application.
Supported OSs are Linux, MAC and Windows. However, I was not able to make it work cmd.exe
(like if cmd.exe even works...). If you are using Windows, use it inside GitBash. DO NOT, I repeat, do not use it on cmd.exe
.
pip install hooks4git --user
Depending on your setup, you might want to use pip3
instead of pip
. Sometimes, during execution, Python2.7 complains about not finding module configparser
. Using Python3x this doesn't happen.
Please, keep in mind that --user
folder might not be on your PATH environment var. Usually you can find it under ~/.local/bin
. If you fix your $PATH
now, it will be automatically fixed for any other python tool you might eventually install inside your user context.
Then, a script called hooks4git
will be available all the time, to hook any project you are currently in. By running with the --init
argument, hooks will be applied (i.e replace all your sample hook files).
Please note you need to manually keep upgrading your system tools, like you do for other tools, like pip itself.
Those are currently built-in scripts, some implemented, some planned:
check_branch_name
: Written in bash, receives a regex as parameter to match your branch name.get_staged_files
: Written in Python, will fix Issue#21 ... perhaps. It is still under testing.print_leftovers
: Planned, to find print statements. A few can be legit, so a little planning is required here.
If you want to use, just follow the exemple on the default .ini file, on sub-section 'checkbranch'. This is the way to trigger built-in scripts, prefixing them with 'h4g/'. Make sure you wrap parameters properly with double quotes.
What else do you hate people often push to the repo? Have an idea? Open an issue and let's talk about it. Those IDE .ini files? Yeah, I hate that too.
After installation, your repo needs to be hooked for all events. Prior version used YAML for configuration management, but that caused PyYAML to be a dependency, and things went a little wrong when running it as a tool. So I choose .ini files over .json files (both have Python native parsers) because it looked less ugly.
Inside your git repository, just type:
hooks4git --init
And get all your regular non-sense-hard-to-use-and-hard-to-maintain-and-hard-to-share hook scripts updated.
Then, you just need to open .hooks4git.ini file on the root of your project and configure it the way you want.
This first example section is meant for Python, but you can use any tool you want, at any given git hook event.
Example section for pre-commit, for Python:
[scripts]
flake8 = flake8 --max-line-length=119 --exclude .git,build,dist,.env,.venv
nosetests = nosetests --with-coverage
[hooks.pre-commit.scripts]
check = flake8
It also could be for NodeJS:
[scripts]
eslint = eslint -f checkstyle index.js > checkstyle-result.xml
jshint = jshint *.js
[hooks.pre-commit.scripts]
check_a = eslint
check_b = jslint
Note: All scripts you add here need to be available on your PATH for execution. So you need to make all of them depedencies on your current project, no matter the language it is written with. Per default, the available hooks are only echo
commands, which will always pass!
When running inside CI, if you manage to have hooks4git
package available, you can force trigger a hook this way:
hooks4git -t <hook> --ci
This will run the very same set of scrips you ran on your development workstation prior to the commit. Please note that <hook>
is any valid entry on .hooks4git.ini
file, not only necessarily a git-hook. See below about "Custom Hooks".
The --ci
parameter tells hooks4git to not print in nice colors, just plain strings. But first check if your CI output handle colors or not. For instance, Bitbucket Pipelines handle it nicely, while Jenkins doesn't.
Hooks have those static names because they are automatically triggered by GIT. However, you can create others inside .hooks4git.ini
file. And you can trigger them using the -t
parameter.
So, if you like check_branch_name
feature, you might think running it inside CI wouldn't be a great idea. How to solve it?
[hooks.ci-develop.scripts]
check = flake8
tests = tests_with_report
As said, there is no "ci-develop" git hook. But due to internal hooks4git
mechanics, using -t
flag, hooks4git
will try to find and run that configuration.
So, it would be a matter of adding this to your CI script:
- pip install hooks4git
- hooks4git -t ci-develop
And since you were using flake8 and tests already on your commit and push hooks, you guarantee to run the same tools with the same parameters on CI, with a nice output, colored or not.
Disclaimer: This feature was never intended to exist, and happened to work by accident. Since it is kind of cool and doesn't break the law, I decided to document it.
Here is a sample output for a Python configuration, with Flake8 (black and white... it has actually a full colored output if --ci parameter is not issued):
———————————————————————————————————————————————————————————————————————————————
hooks4git v0.4.x :: Pre-Commit :: hook triggered
———————————————————————————————————————————————————————————————————————————————
STEP | $ flake8 --max-line-length=119 --exclude .git,__pycache__,build,dist
OUT | None
PASS | 'flake8' step executed successfully
———————————————————————————————————————————————————————————————————————————————
STEPS| 1 were executed
TIME | Execution took 0:00:00.684762
PASS | All green! Good!
———————————————————————————————————————————————————————————————————————————————
If you are willing to code something on this project, it is quite simple. You first need to fork it directly on GitHub, so you can get a copy on your computer that you can push to. Therefore, you would be able to open a Pull Request to the original repository.
> git clone [email protected]:<super_cool_developer>/hooks4git.git
> cd hooks4git
> pipenv install --dev
> pip uninstall hooks4git # just in case, it may fail
> pip install -e . --user
> hooks4git --init # OF COURSE!!!
> git checkout -b feature/super_cool_feature
The above will install hooks4git linked to the folder you cloned the repository to, instead of the module you normally download from Pypi. This way, every change you make on the source code will affect your environment, makeing it easy to use. Of course there are several other ways, like using virtualenv, for instance. That was only a suggestion and affects all repos you have. This is the way I usually test develop versions for a few days prior to a release.
This project is licensed under MIT license. See the LICENSE file for details
See list of contributors who participated in this project.
- Marco Lovato
- Collins Abitekaniza (where I forked from)
- Major rework on classes and dependencies usage
- Added more tests
- Fixed Issue#57 - Subprocess Call Error
- Fixed Issue#54 - Pipenv usage
- Fixed Issue#47 - Colorama usage
- Major rework on how strings are printed out
- Added --ci parameter, so no color will be printed out (Idea from Fernando Espíndola)
- Auto create hooks folder (inside .git) if it is missing (Idea from Édouard Lopez)
- Support for Windows with GitBash
- Added docker scripts for quick clean machine testing environment
- Better exception handling when user configures duplicate sections by mistake
- FIXED: Changed default max line length example to 119 instead of 120
- Replaced copying code to .git/hooks with a safe bash caller
- Replaced '_' folder (or 'scripts' folder) with 'h4g' folder for internal scripts
- FIXED: Script order inside a hook definition was random
- Standard Error Output was not being handled and printed accordingly
- Initial release