Skip to main content

PyWinGUI GUI Automation Library for Windows Forms (Binary Distribution)

Project description

build_tools — README

This document explains how to build, test, and publish Windows Nuitka-compiled wheels for the pywinGUI project using the repository's build scripts and GitHub Actions.

Checklist

  • Create PyPI API token and add it to GitHub Secrets (PYPI_API_TOKEN).
  • Verify make_nuitka_wheel.ps1 and push.ps1 are configured correctly.
  • Test a local build for each target Python interpreter (32-bit and 64-bit).
  • (Optional) Run the GitHub Actions workflow manually to inspect artifacts.
  • Tag and push when ready to publish to PyPI (workflow will run and upload wheels).

Overview

  • make_nuitka_wheel.ps1 — Builds a wheel for a specific Python interpreter using Nuitka. It compiles selected modules into extension modules and packages them into a wheel.
  • push.ps1 — Uploads the built wheel(s) to PyPI using twine, reading the token from environment variables (PYPI_API_TOKEN or TWINE_PASSWORD).
  • GitHub Actions workflow (.github/workflows/build-windows-wheels.yml) — Builds Windows wheels across a matrix of Python versions and architectures (x86/x64) and publishes them to PyPI when a tag is pushed and PYPI_API_TOKEN is set as a secret.

Security model

  • For releasing protected/compiled code, build and publish per-ABI wheels (win32 and win_amd64) and do not include source fallback files in release wheels. To do that, run the build script without the -IncludeSourceFallbacks switch.
  • If you need debug/compatibility builds that include Python source fallbacks, run the build script with -IncludeSourceFallbacks. Do not publish fallback-enabled wheels to PyPI if you want to protect source code.

Local usage

  1. Build a secure wheel (no source fallbacks)

Open PowerShell in the repository root and run:

# Build for a target Python interpreter (replace path with target python.exe)
.\build_tools\make_nuitka_wheel.ps1 -PythonExe "C:\Path\To\Target\python.exe"
  1. Build a compatibility/debug wheel (includes _source.py fallbacks)
# Include source fallbacks for debugging (not recommended for protected releases)
.\build_tools\make_nuitka_wheel.ps1 -PythonExe "C:\Path\To\Target\python.exe" -IncludeSourceFallbacks
  1. Inspect the produced wheel

After the build completes, the wheel(s) are placed under dist\. To list the wheel files:

Get-ChildItem .\dist\*.whl | Sort-Object LastWriteTime -Descending | Format-Table Name,FullName -AutoSize

To list the contents of a wheel (using Python):

python - <<'PY'
import zipfile
w = r".\dist\your-built-wheel.whl"  # replace with the actual file
z = zipfile.ZipFile(w)
print('\n'.join(z.namelist()))
PY
  1. Test install in a fresh venv (use the same interpreter you built for)
"C:\Path\To\Target\python.exe" -m venv .\test_env
.\test_env\Scripts\python.exe -m pip install --upgrade pip
.\test_env\Scripts\python.exe -m pip install .\dist\your-built-wheel.whl
.\test_env\Scripts\python.exe -c "import pywinGUI; import pywinGUI.controller as controller; print('OK', controller)"

Publishing to PyPI (local)

  1. Set token in environment (temporary for current PowerShell session):
$env:PYPI_API_TOKEN = 'pypi-<your-token-here>'
# or the less-preferred variable used by older scripts
$env:TWINE_PASSWORD = 'pypi-<your-token-here>'
  1. Run the push script (it will install twine into the project's venv and upload):
.\build_tools\push.ps1

GitHub Actions (automated build & publish)

What the workflow does

  • For each Python version and architecture in the matrix it:
    • Checks out the repo
    • Sets up Python (x86/x64)
    • Installs build dependencies (nuitka, wheel, setuptools)
    • Runs make_nuitka_wheel.ps1 to build a wheel
    • Uploads the built wheel(s) as an artifact for inspection
  • If the workflow run is triggered by a pushed tag (e.g. v1.2.3) and PYPI_API_TOKEN is set as a repository secret, the workflow will run twine to upload dist/* to PyPI.

One-time setup on GitHub

  1. Create a PyPI API token at https://pypi.org (Account → API tokens).
  2. In GitHub, go to the repository: Settings → Secrets and variables → Actions → New repository secret
    • Name: PYPI_API_TOKEN
    • Value: (the token you copied from PyPI)

Triggering a publish

  • To test the workflow without publishing, run the workflow manually from Actions → your workflow → Run workflow.
  • To publish to PyPI, push a tag. Example:
git add .
git commit -m "release prep"
git tag v1.0.0
git push origin main
git push origin v1.0.0

Versioning

The build scripts set a SMART_VERSION environment variable that setup_build_pkg.py reads and uses as the package version when building a wheel. By default make_nuitka_wheel.ps1 generates a timestamp-based PEP-440-friendly version to keep builds unique (useful for automated builds).

You can control the version in three ways:

  1. Let the build produce a timestamped version (default)
  • This happens automatically when you run make_nuitka_wheel.ps1 without -Version. The script sets SMART_VERSION to a value like 2026.1.2.10.5+2026.01.02.10.05 (format keeps it PEP-440 compatible while preserving a raw timestamp suffix).
  1. Provide an explicit version when running the build script
  • Use the -Version parameter to set a specific PEP-440 compliant version string. Example:
# Set the build version explicitly
.\build_tools\make_nuitka_wheel.ps1 -PythonExe "C:\Path\To\python.exe" -Version "1.2.3"
  • The provided -Version value overrides the generated timestamp and is exported as SMART_VERSION for the wheel build.
  1. In CI (recommended for releases): use the git tag as the version
  • When publishing releases from CI you typically want the wheel version to match the git tag (for example v1.2.3 or 1.2.3). You can pass the tag name into the build script by supplying -Version or by exporting SMART_VERSION in the runner environment.

Example (GitHub Actions step) — pass the tag name into the build script:

- name: Build wheel (Nuitka)
  shell: pwsh
  run: |
    $py = (Get-Command python).Source
    $ver = '${{ github.ref_name }}'   # e.g. 'v1.2.3' or '1.2.3'
    Write-Host "Building wheel for version: $ver"
    .\build_tools\make_nuitka_wheel.ps1 -PythonExe $py -Version $ver

Notes on version format

  • Use a PEP-440 compliant version string (for example: 1.2.3, 1.2.3.post1, 1.2.3rc1). Avoid spaces or characters that are illegal in wheel metadata.
  • If you use git tags prefixed with v (v1.2.3) and prefer to drop the v, strip it in CI before passing to -Version (for example trim the leading v in the script or set -Version ${{ github.ref_name | replace('v','') }}).

How setup_build_pkg.py reads the version

  • setup_build_pkg.py uses os.environ.get('SMART_VERSION', '0.0.0') as the package version. The make_nuitka_wheel.ps1 script sets SMART_VERSION before invoking setup_build_pkg.py, so the wheel will carry the SMART_VERSION value.

Troubleshooting

  • If you see 0.0.0 in the built wheel metadata, SMART_VERSION was not set in the environment at package-build time. Confirm you invoked make_nuitka_wheel.ps1 (it sets SMART_VERSION) and that setup_build_pkg.py is executed in the same process/venv context.
  • If a wheel is already published with an identical version, PyPI rejects the upload; ensure you increment the version for each published release.

Notes & troubleshooting

  • Make sure you build wheels on Windows for Windows wheels. The GitHub Actions workflow uses runs-on: windows-latest.
  • Ensure you build one wheel per ABI. For Windows you typically need both win32 and win_amd64 if you support 32-bit and 64-bit Python users.
  • If a user reports "No module named 'pywinGUI.controller'" after installing from PyPI:
    1. Ask them to run python -c "import platform,sys; print(sys.executable, platform.python_version(), platform.architecture())" and python -m pip show -f pywinGUI.
    2. Verify they installed the wheel matching their Python bitness. If not, provide the correct wheel or publish both ABI wheels.
    3. If the installed package has an incorrectly named file (e.g. _init_.py), remove the package folder and reinstall.

CI considerations

  • If Nuitka or compiled code requires MSVC or other build tools on the runner, add the installation steps in the workflow (Chocolatey or Visual Studio Build Tools) before running the build script.
  • The workflow currently publishes only on tag pushes. You can adjust it to publish on a release or add manual approvals.

Contact

  • If a build fails or a wheel raises import errors on a target interpreter, collect the following and open an issue:
    • The wheel filename
    • The output of python -c "import platform,sys; print(sys.executable, platform.python_version(), platform.architecture())" on the failing machine
    • The output of python -m pip show -f pywinGUI
    • Any traceback from attempting to import pywinGUI.controller

This README is intentionally minimal and focused on build and release flow. If you want, I can:

  • Add a short RELEASE.md with step-by-step release checklist and gating rules.
  • Create a GitHub Actions job that requires manual approval before upload.
  • Add cibuildwheel usage for cross-platform future support.

Which of these would you like next?

Project details


Download files

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

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

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

pywingui-4.2.0-cp313-none-win_amd64.whl (614.3 kB view details)

Uploaded CPython 3.13Windows x86-64

pywingui-4.2.0-cp313-none-win32.whl (522.6 kB view details)

Uploaded CPython 3.13Windows x86

pywingui-4.2.0-cp312-none-win_amd64.whl (621.2 kB view details)

Uploaded CPython 3.12Windows x86-64

pywingui-4.2.0-cp312-none-win32.whl (531.8 kB view details)

Uploaded CPython 3.12Windows x86

File details

Details for the file pywingui-4.2.0-cp313-none-win_amd64.whl.

File metadata

  • Download URL: pywingui-4.2.0-cp313-none-win_amd64.whl
  • Upload date:
  • Size: 614.3 kB
  • Tags: CPython 3.13, Windows x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.11

File hashes

Hashes for pywingui-4.2.0-cp313-none-win_amd64.whl
Algorithm Hash digest
SHA256 11a71e19ff4e21d977e572b99469ac27c22e0a5711e89b120df8579de91371eb
MD5 8c67645e10d9bb2ac5bdaffed7622af4
BLAKE2b-256 26894f479ce29acf65177e738b71d7d42431a35de939ec1f62b4c5b02d35c503

See more details on using hashes here.

File details

Details for the file pywingui-4.2.0-cp313-none-win32.whl.

File metadata

  • Download URL: pywingui-4.2.0-cp313-none-win32.whl
  • Upload date:
  • Size: 522.6 kB
  • Tags: CPython 3.13, Windows x86
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.11

File hashes

Hashes for pywingui-4.2.0-cp313-none-win32.whl
Algorithm Hash digest
SHA256 b9afbfe94ad70f229e1144ff9c10762ad89938ebf433928844acf7d48a3e4e29
MD5 cb05e1ff298821b8d4d49ad03888f566
BLAKE2b-256 51526530ecbb0703779801b13c0115ebc9862e6d3e13dc257804d4e1e7bbf180

See more details on using hashes here.

File details

Details for the file pywingui-4.2.0-cp312-none-win_amd64.whl.

File metadata

  • Download URL: pywingui-4.2.0-cp312-none-win_amd64.whl
  • Upload date:
  • Size: 621.2 kB
  • Tags: CPython 3.12, Windows x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.10

File hashes

Hashes for pywingui-4.2.0-cp312-none-win_amd64.whl
Algorithm Hash digest
SHA256 9140f10f99e1148b0f554fa4a6e485568264c29281ab3eb0090ff837cc658ac0
MD5 901b48a20164e0bb0ef0de5b55c8cb61
BLAKE2b-256 45c3194dfada28bb0dbdf73a7be54af1f6c8b056c7b2d77f8fc09424971ca9bf

See more details on using hashes here.

File details

Details for the file pywingui-4.2.0-cp312-none-win32.whl.

File metadata

  • Download URL: pywingui-4.2.0-cp312-none-win32.whl
  • Upload date:
  • Size: 531.8 kB
  • Tags: CPython 3.12, Windows x86
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.10

File hashes

Hashes for pywingui-4.2.0-cp312-none-win32.whl
Algorithm Hash digest
SHA256 834789004059cd13b41f2136c2256e30086a31276423e7464dc4ea2fd04cad36
MD5 1e3249e2cc3394020d1a995a05f0ef29
BLAKE2b-256 984401c2c4bfd0629b1ba90986cd6eab844aa64083f596607d892c382b779f5e

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