Merge pull request #229851 from raphaelr/mknugetsource-without-nuget

{ dotnetPackages, lib, xml2, stdenvNoCC }:

{ lib, python3, stdenvNoCC }:

{ name

, description ? ""

    inherit name;

    meta.description = description;

    nativeBuildInputs = [ dotnetPackages.Nuget xml2 ];

    nativeBuildInputs = [ python3 ];

    buildCommand = ''

      export HOME=$(mktemp -d)

      mkdir -p $out/{lib,share}

      ${lib.concatMapStringsSep "\n" (dep: ''

        nuget init "${dep}" "$out/lib"

      '') deps}

      (

        shopt -s nullglob

        for nupkg in ${lib.concatMapStringsSep " " (dep: "\"${dep}\"/*.nupkg") deps}; do

          cp --no-clobber "$nupkg" $out/lib

        done

      )

      # Generates a list of all licenses' spdx ids, if available.

      # Note that this currently ignores any license provided in plain text (e.g. "LICENSE.txt")

      find "$out/lib" -name "*.nuspec" -exec sh -c \

        "NUSPEC=\$(xml2 < {}) && echo "\$NUSPEC" | grep license/@type=expression | tr -s \  '\n' | grep "license=" | cut -d'=' -f2" \

      \; | sort -u > $out/share/licenses

      python ${./extract-licenses-from-nupkgs.py} $out/lib > $out/share/licenses

    '';

  } // { # We need data from `$out` for `meta`, so we have to use overrides as to not hit infinite recursion.

    meta.licence = let

#!/usr/bin/env python3

"""

Opens each .nupkg file in a directory, and extracts the SPDX license identifiers

from them if they exist. The SPDX license identifier is stored in the

'<license type="expression">...</license>' tag in the .nuspec file.

All found license identifiers will be printed to stdout.

"""

from glob import glob

from pathlib import Path

import sys

import xml.etree.ElementTree as ET

import zipfile

all_licenses = set()

if len(sys.argv) < 2:

    print(f"Usage: {sys.argv[0]} DIRECTORY")

    sys.exit(1)

nupkg_dir = Path(sys.argv[1])

for nupkg_name in glob("*.nupkg", root_dir=nupkg_dir):

    with zipfile.ZipFile(nupkg_dir / nupkg_name) as nupkg:

        for nuspec_name in [name for name in nupkg.namelist() if name.endswith(".nuspec")]:

            with nupkg.open(nuspec_name) as nuspec_stream:

                nuspec = ET.parse(nuspec_stream)

                licenses = nuspec.findall(".//{*}license[@type='expression']")

                all_licenses.update([license.text for license in licenses])

print("\n".join(sorted(all_licenses)))