Skip to content
/ pigar Public

☕ A tool to generate requirements.txt for Python project, and more than that. (IT IS NOT A PACKAGE MANAGEMENT TOOL)

License

Notifications You must be signed in to change notification settings

damnever/pigar

Repository files navigation

pigar

  • Generating requirements.txt for Python project.
    • Handling the difference between different Python versions.
    • Jupyter notebook (*.ipynb) support.
    • Including the import statements/magic from exec/eval/importlib, doctest of docstring, etc.
  • Searching ditributions(packages) by the top level import/module names.
  • Checking the latest versions of requirements.

NOTE: Pipenv or other tools is recommended for improving your development flow.

Installation

pigar can run on Python 3.7+.

To install it with pip, use:

[sudo] pip install pigar

To install it with conda, use:

conda install -c conda-forge pigar

To get the newest code from GitHub:

pip install git+https://github.com/damnever/pigar.git@[main or other branch] --upgrade

Usage

  • pigar can consider most kinds of complicated situations(see FAQ). For example, pigar v1 has py2_requirements.txt and py3_requirements.txt for different Python versions.

    # Generate requirements.txt for current directory.
    $ pigar generate
    
    # Generating requirements.txt for given directory in given file.
    $ pigar gen -f ../dev-requirements.txt ../
    

    pigar gen --with-referenced-comments can list all files which referenced the package/distribution(the line numbers for Jupyter notebook may be a bit confusing), for example:

    # project/foo.py: 2,3
    # project/bar/baz.py: 2,7,8,9
    foobar == 3.3.3
    

    If the requirements.txt is overwritten, pigar will show the difference between the old and the new, use --dont-show-differences to disable it.

    NOTE, pigar will search the packages/distributions in local environment first, then it will do further analysis and search missing packages/distributions on PyPI.

    See also: EXPERIMENTAL FEATURES.

  • If you do not know the import name that belongs to a specific distribution (more generally, does Import Error: xxx drive you crazy?), such as bs4 which may come from beautifulsoup4 or MySQLdb which could come from mysql-python, try searching for it:

    $ pigar search bs4 MySQLdb
    
  • Checking for the latest version:

    # Specify a requirements file.
    $ pigar check -f ./requirements.txt
    
    # Or, you can let pigar searching all *requirements.txt in the current directory
    # level by itself.
    $ pigar check
    
  • More:

    TIP: pigar accepts a prefix for a command, such as pigar gen, pigar c.

    pigar --help
    

EXPERIMENTAL FEATURES

  • requirement-annotations

    Some packages may require optional packages/distributions to be installed depending on your usage. To make pigar a little bit more useful, use pigar generate --enable-feature requirement-annotations in conjunction with comments following the format below:

    import foo # pigar: required-imports=import_name_bar,import_name_baz
    import foo # pigar: required-packages=package-name-bar,package-name-baz # Extra comments are allowed.
    foo(features=['bar', 'baz']) # pigar: required-distributions=package-name-bar,package-name-baz

    If you find the comment intrusive, you can extract those comments into a separate Python file and delete it as needed, for example, 'pigar_annotations.py'.

FAQ

Is `pigar` a dependency management tool?

No. I've thought about this many times, but there is too much dirty work to be done to make pigar's way reliable.

I like the way pigar does the job, but sadly, pigar does a bad job of managing dependencies, pigar is more like a tool to assist an old project to migrate to a new development workflow.

(1) Why does `pigar` show multiple packages/distributions for the same import name?

(2) Why does pigar generate different packages/distributions for the same import name in different environment?

pigar can not handle those situations gracefully, you may need to remove the duplicate packages in requirements.txt manually, or select one of them when pigar asks you. Install the required packages/distributions(remove others) in local environment should fix it as well.

Related issues: #32, #68, #75.

Why can't `pigar` find the packages/distributions that have not been explicit import?

Some frameworks may use some magic to import the modules for users automatically, and pigar can not handle it, you may need to fix it manually or use the EXPERIMENTAL FEATURES.

Related issues: #33, #103

More

pigar does not use regular expressions in such a violent way. Instead, it uses AST, which is a better method for extracting imported names from arguments of exec/eval/importlib, doctest of docstring, etc. However, pigar can not solve all the tricky problems, see FAQ.

Also, pigar can detect the difference between different Python versions. For example, you can find concurrent.futures from the Python 3.2 standard library, but you will need install futures in earlier versions of Python to get concurrent.futures, this is not a hardcode.

If you have any issues or suggestions, please submit an issue on GitHub. All contributions are appreciated!

LICENSE

The BSD 3-Clause License