Skip to main content

Python packaging operations, including PEP-517 support, for use by a setup.py script.

Project description

Overview

The pipcl package provides Python build backend operations (including PEP 517 support), for use by a setup.py script.

  • Designed to help build complex Python extension packages.

  • Can also be used to build simple pure-python packages.

  • Works on Linux, Windows, MacOS and OpenBSD.

  • Is a python module, not a framework, so does not impose any restrictions on usage.

  • Can be used with any external build system by running commands, for example with subprocess.run() or enhanced wrapper pipcl.run().

The intention is to allow a setup.py script to use the full power of Python to do everything that is specific to the package, without having to worry about generic Python packaging issues such wheel formats etc (see https://packaging.python.org/en/latest/specifications/).

Basic usage

To build a pure-python package:

  • Create a setup.py file that imports pipcl and uses it to specify and build a package:

    import pipcl
    
    def build():
        # Creates a module called `mymodule`.
        return [
                ('src/foo.py', 'mymodule/__init__.py'),
                ]
    def sdist():
        return pipcl.git_items('.')
    p = pipcl.Package(
            'mypackage',    # Name of package.
            '1.2.3',        # Package version.
            pure=True,      # Pure python package.
            summary = 'My package',
            description = 'README.md',
            fn_build = build,
            fn_sdist = sdist,
            )
    # PEP 517 support.
    build_wheel = p.build_wheel
    build_sdist = p.build_sdist
  • Create a pyproject.toml that tells pip (and other build frontends) to use setup.py:

    [build-system]
        requires = ['pipcl']
        build-backend = 'setup' # Use setup.py
        backend-path = ['.']    # in top-level of the checkout.

To build a package containing a SWIG extension module, the build() function can be modified to use pipcl.build_extension():

import pipcl

def build():
    so_leaf = pipcl.build_extension(
            name = 'mymodule',      # Name of extension module.
            path_i = 'src/foo.i',   # SWIG input file.
            outdir = 'build',
            )
    # <so_leaf> will be '_mymodule.so' or similar.
    return [
            ('build/foo.py', 'mymodule/__init__.py'),
            (f'build/{so_leaf}', 'mymodule/'),
            ]
def sdist():
    return pipcl.git_items('.')
p = pipcl.Package(
        'mypackage',    # Name of package.
        '1.2.3',        # Package version.
        summary = 'My package',
        description = 'README.md',
        fn_build = build,
        fn_sdist = sdist,
        )
# PEP 517 support.
build_wheel = p.build_wheel
build_sdist = p.build_sdist

For more details, see doctest examples in file:pipcl.py.

Documentation

  • Detailed documentation is in doc comments in Python code.

  • Convert README.rst to html with:

    pip install docutils
    docutils -gdst --halt=3 --pep-references README.rst README.rst.html

    View at: file:README.rst.html

API overview

The pipcl.Package class

This is similar in approach to setuptools/distutils in that one creates an instance of the class, passing the package name, version etc.

  • Most of the constructor arguments correspond exactly to metadata items in https://packaging.python.org/specifications/core-metadata/.

  • An explicit function fn_build() should be provided for building the package. This should return a list of files to be included in a wheel or install, typically as (from, to) pairs, where <from> is the path to a file and <to> is the path within a wheel or install.

  • An explicit function fn_sdist() can be provided for building an sdist, returning a list of files to include.

  • Unlike setuptools, pipcl does not use heuristics for finding files, instead everything must be specified exactly.

Other functions and classes

  • pipcl.build_extension() - build a Python extension using swig. Typically used by pipcl.Package’s fn_build() callback.

  • pipcl.git_get(): Create/update a clean git checkout for a particular branch, tag or sha of a remote repository.

  • pipcl.git_items(): Get list of files within a git checkout.

  • pipcl.git_info*(): Get git information such as sha, comment, diff, branch name, author etc.

  • pipcl.macos_add_brew_path(): on MacOS, support for adding package binaries to PATH.

  • pipcl.macos_patch(): patching of MacOS shared libraries to avoid use of absolute paths.

  • Class pipcl.NewFiles: Support for simple detection of changed files, for example around running of an external command.

    This can be useful when building a wheel with pip, as the leafname of the wheel is not known in advance.

  • pipcl.run(): enhances subprocess.run(), providing dynamic output along with capture of output, prefixing of output, timeouts etc.

    • Also allows commands to be specified as multiple lines, which improves readability in logs.

  • pipcl.run_if(): simple dependency checking to only run commands if prerequisites are newer than the output.

    • This can use information in Makefile-style .d dependency files.

  • pipcl.swig*(): build and use swig from source; this can be useful on MacOS where some swig versions appear to generate incorrect code.

  • Class pipcl.wdev.WindowsVS: on Windows, search for specific/latest versions of Visual Studio compiler (cl.exe) and linker (link.exe).

    • Has a command string that includes running of an appropriate vcvars, which can be used directly in compile and link command.

Other

  • Support for Pyodide.

  • Experimental support for Graal.

Changelog

Version 3:

  • Avoid unhelpful assert failures on Windows if paths differ only in upper/lower case.

Version 2 (2026-04-18):

  • Fixed bug in zip file generation on python<3.13.

  • Moved Python code into src/.

  • Don’t attempt to be usable in raw checkout.

  • Fixed doctest’s to work on Windows.

  • Added python_version_tuple().

  • Avoid spurious differences between wheels built on different systems:

    • Sort lines in generated RECORD file.

    • Use --global core.autocrlf input when running git clone.

Version 1 (2026-04-16):

  • First release to pypi.org

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pipcl-3.tar.gz (69.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

pipcl-3-py3-none-any.whl (50.9 kB view details)

Uploaded Python 3

File details

Details for the file pipcl-3.tar.gz.

File metadata

  • Download URL: pipcl-3.tar.gz
  • Upload date:
  • Size: 69.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.12

File hashes

Hashes for pipcl-3.tar.gz
Algorithm Hash digest
SHA256 3299e50034b22d2ac15fbfaad8f7290c87d9db501c2fab6c2d85633d06a7fc4e
MD5 77a4777fdd7955cf755314a464617f16
BLAKE2b-256 d57c776b1f4c4081343f4ab496a373ef8a675ad86faf58ce5118f901d6a0f9de

See more details on using hashes here.

File details

Details for the file pipcl-3-py3-none-any.whl.

File metadata

  • Download URL: pipcl-3-py3-none-any.whl
  • Upload date:
  • Size: 50.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.12

File hashes

Hashes for pipcl-3-py3-none-any.whl
Algorithm Hash digest
SHA256 a63cd7019af08d4af2f8fa3b107b2f46441300fdeca1be04e39cea55e7f477d4
MD5 ef5c5faf5ac23aa0b9eda85eaf0826d1
BLAKE2b-256 cc5c05d19ce05967c32b4936b8e88da831aa7bddc490a9acf9ab948c01a982c7

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page