pkgs/build-support/dotnet/make-nuget-source/extract-licenses-from-nupkgs.py

   1 #!/usr/bin/env python3
   2 """
   3 Opens each .nupkg file in a directory, and extracts the SPDX license identifiers
   4 from them if they exist. The SPDX license identifier is stored in the
   5 '<license type="expression">...</license>' tag in the .nuspec file.
   6 All found license identifiers will be printed to stdout.
   7 """
   8
   9 from glob import glob
  10 from pathlib import Path
  11 import sys
  12 import xml.etree.ElementTree as ET
  13 import zipfile
  14
  15 all_licenses = set()
  16
  17 if len(sys.argv) < 2:
  18     print(f"Usage: {sys.argv[0]} DIRECTORY")
  19     sys.exit(1)
  20
  21 nupkg_dir = Path(sys.argv[1])
  22 for nupkg_name in glob("*.nupkg", root_dir=nupkg_dir):
  23     with zipfile.ZipFile(nupkg_dir / nupkg_name) as nupkg:
  24         for nuspec_name in [name for name in nupkg.namelist() if name.endswith(".nuspec")]:
  25             with nupkg.open(nuspec_name) as nuspec_stream:
  26                 nuspec = ET.parse(nuspec_stream)
  27                 licenses = nuspec.findall(".//{*}license[@type='expression']")
  28                 all_licenses.update([license.text for license in licenses])
  29
  30 print("\n".join(sorted(all_licenses)))