Collect WHEEL files for installed python wheels #4214

[kaokab33]

Summary

Fixes #4214

Parse the WHEEL file in .dist-info directories to extract wheel tags needed
to reconstruct the original wheel filename for building more detailed PURLs.

A WHEEL file contains a Tag field that encodes the Python version, ABI, and
platform information:

Wheel-Version: 1.0
Generator: bdist_wheel (0.37.1)
Root-Is-Purelib: false
Tag: cp310-cp310-manylinux_2_17_x86_64

This tag is essential to reconstruct the original wheel filename
(e.g., numpy-1.23.0-cp310-cp310-manylinux_2_17_x86_64.whl) which is needed
for constructing proper detailed PURLs.

Changes

New Functions in src/packagedcode/pypi.py:

1. parse_wheel_tags(dist_info_path)

   • Parses .dist-info/WHEEL file
   • Extracts Tag, Wheel-Version, Generator, Root-Is-Purelib
   • Handles multiple Tag: lines (e.g., py2-none-any + py3-none-any)
   • Returns empty dict if no WHEEL file found

2. reconstruct_wheel_filename(name, version, tag)

   • Rebuilds wheel filename from package name, version, and tag string
   • Handles hyphenated names by converting to underscores per wheel naming spec
   • Example: numpy-1.23.0-cp310-cp310-manylinux_2_17_x86_64.whl

Modified Function:

3. parse_metadata()
   • Now reads .dist-info/WHEEL file when available
   • Stores wheel tags and metadata in extra_data

Bug Fix:

4. src/scancode/interrupt.py
   • Fix signal-based interrupt on non-main threads
   • Add a check to detect if the current thread is the main thread before
     attempting to set up signal handlers
   • On Python 3.14+, signal handlers can only be set up in the main thread
     of the main interpreter
   • If not in the main thread, fall back to non-interruptible execution
   • Fixes ValueError: signal only works in main thread of the main interpreter

Before & After

Before:

{
  "name": "numpy",
  "version": "1.23.0",
  "purl": "pkg:pypi/numpy@1.23.0",
  "extra_data": {}
}

After:

  "name": "numpy",
  "version": "1.23.0",
  "purl": "pkg:pypi/numpy@1.23.0",
  "extra_data": {
    "wheel_tags": ["cp310-cp310-manylinux_2_17_x86_64"],
    "wheel_version": "1.0",
    "wheel_generator": "bdist_wheel (0.37.1)",
    "root_is_purelib": false
  }
}

Tests Added

New TestWheelTagCollection class with 11 tests:

Test	Description
test_parse_wheel_tags_with_platform_specific_tag	Platform-specific wheel (cp310-cp310-manylinux)
test_parse_wheel_tags_with_multi_tags	Multiple Tag: lines in WHEEL file
test_parse_wheel_tags_with_pure_wheel	Pure wheel (py3-none-any)
test_parse_wheel_tags_with_no_wheel_file	Edge case: no WHEEL file present
test_reconstruct_wheel_filename_platform_specific	Platform-specific filename reconstruction
test_reconstruct_wheel_filename_pure_wheel	Pure wheel filename reconstruction
test_reconstruct_wheel_filename_with_hyphenated_name	Hyphen-to-underscore conversion
test_parse_metadata_with_platform_wheel_tag	Full integration: platform wheel
test_parse_metadata_with_multi_wheel_tags	Full integration: multi-tag wheel
test_parse_metadata_with_pure_wheel_tag	Full integration: pure wheel
test_parse_metadata_without_wheel_file	Full integration: missing WHEEL file

Test Fixtures Added:

• numpy-1.23.0.dist-info — Platform-specific wheel (cp310-cp310-manylinux_2_17_x86_64)
• cryptography-41.0.0.dist-info — Multi-tag wheel (two Tag: lines)
• requests-2.28.0.dist-info — Pure wheel (py3-none-any)
• no-wheel-1.0.0.dist-info — No WHEEL file (edge case)

Updated Expected JSONs:

Existing test expected JSON files updated to include wheel tag data from
WHEEL files that were already present in test fixtures.

Checklist

• Unit tests passing
• CHANGELOG.rst updated
• AUTHORS.rst updated