PyCuDAL

Parametric acceptance limits for USP <905> Content Uniformity and USP <711> Dissolution — a Python re-implementation of the SAS programs CALCUSP1/2, CALDISP1/2, EVCUSP1/2, EVDISP1/2, SMPCUSP1/2, SMPDISP1/2, with a CLI and two GUIs.

Features

• Computations (cudal package)
  • Content Uniformity – Sampling Plans 1 & 2 (cusp1, cusp2)
  • Dissolution – Sampling Plans 1 & 2 (disp1, disp2)
  • Three analysis modes each: acceptance-limit table, probability of passing, sample probability
• CLI – python -m cudal.cli (or installed command cudal)
• Tkinter GUI (cudal-gui, stdlib-only → tiny install): 4 tabs × 3 modes, background threads with progress, scrollable parameters, live validation, sortable/colored results table, CSV export, modal matplotlib plots (lines + cubic spline, heatmap with 80/90 % contours), SAS-style PDF listings (Courier, column wrapping, repeated page headers), XLSX export, settings persistence
• Optional PySide6 GUI (extras/cudal_gui_pyside6.py) – modern Qt theme, kept outside the package so the pip install stays small

Repository layout

cudal/            library: computations + CLI + Tk GUI (+ logo package data)
extras/           optional PySide6 GUI (not installed)
tests/            pytest smoke tests
scripts/          local exe build helper
.github/          CI, PyPI publish, exe release workflows

Installation

Requires Python ≥ 3.9.

# from GitHub
pip install git+https://github.com/moazelessawey/cudal.git

# from a source checkout
pip install .            # core (numpy/pandas/scipy) + CLI + Tk GUI
pip install .[all]       # + matplotlib, openpyxl, reportlab (plots/exports)
pip install .[qt]        # + PySide6 for the optional Qt GUI
pip install .[dev]       # + pytest, pyinstaller

Quick start

Python API

from cudal import cusp1

table = cusp1.acceptance_limit_table(10, 100.0, 95.0, 95.0, 85.1, 114.9, 0.5)
print(table)

CLI

cudal --help                 # installed console script
python -m cudal.cli --help   # module form
python -m cudal --help       # package __main__

The CLI covers the same four scenarios × three modes as the GUIs (see --help for the exact option names).

Tkinter GUI (included, stdlib-only)

cudal-gui                 # after pip install
python cudal_gui.py       # from a source checkout

Optional PySide6 GUI (extra file, not installed)

pip install .[qt]
python extras/cudal_gui_pyside6.py

Result exports

• CSV – one click per table
• XLSX – File → Export all results (one sheet per tab)
• PDF – SAS-listing style: Courier, centered title block, Plan‑1 row-wrapped blocks, Plan‑2 SE×SM LL/UL matrix, 2 dp right-aligned numbers, * for NaN, title + table headers repeated on every page
• Plots – lines + cubic spline, or heatmap with 80/90 % threshold contours; Save PNG

Building standalone executables

Locally (Windows):

scripts\build_exe.bat

or manually:

pip install .[all,qt] pyinstaller
pyinstaller --onefile --noconsole --name CuDAL    --add-data "cudal\logo.png;cudal" cudal_gui.py
pyinstaller --onefile --noconsole --name CuDAL-Qt --add-data "cudal\logo.png;cudal" extras\cudal_gui_pyside6.py

On every GitHub Release, the build-exe.yml workflow builds CuDAL.exe and CuDAL-Qt.exe and attaches them to the release automatically.

Releasing & publishing

1. Bump cudal/__version__ and commit.
2. Create a tag & GitHub release: git tag v1.0.0 && git push origin v1.0.0 (or use Releases → Draft new release in the UI).
3. Actions then:
   • build & attach the Windows executables to the release,
   • build the wheel/sdist and publish to PyPI (publish.yml).

Development

pip install -e .[all,dev]
pytest -q
python -m cudal.gui --selftest

License & disclaimer

MIT — see LICENSE.

This software is provided for research/informational use. It does not constitute regulatory advice; validate any output against your own reference implementation before use in a GMP environment.