Implement PEP 639: support for Metadata-Version: 2.4; Part II

make_wheels.py

@@ -2,11 +2,13 @@
 import logging
 import io
 import os
+import re
 import json
 import hashlib
 import tarfile
+from warnings import warn
 import urllib.request
-from pathlib import Path
+from pathlib import Path, PurePath
 from email.message import EmailMessage
 from wheel.wheelfile import WheelFile
 from zipfile import ZipFile, ZipInfo, ZIP_DEFLATED
@@ -28,6 +30,8 @@
     'powerpc64le-linux':  'manylinux_2_17_ppc64le.manylinux2014_ppc64le.musllinux_1_1_ppc64le',
 }
 
+_YELLOW = "\033[93m"
+
 
 class ReproducibleWheelFile(WheelFile):
     def writestr(self, zinfo_or_arcname, data, *args, **kwargs):
@@ -70,13 +74,17 @@ def write_wheel_file(filename, contents):
 def write_wheel(out_dir, *, name, version, tag, metadata, description, contents):
     wheel_name = f'{name}-{version}-{tag}.whl'
     dist_info  = f'{name}-{version}.dist-info'
+    filtered_metadata = []
+    for header, value in metadata:
+        filtered_metadata.append((header, value))
+
     return write_wheel_file(os.path.join(out_dir, wheel_name), {
         **contents,
         f'{dist_info}/METADATA': make_message([
             ('Metadata-Version', '2.4'),
             ('Name', name),
             ('Version', version),
-            *metadata,
+            *filtered_metadata,
         ], description),
         f'{dist_info}/WHEEL': make_message([
             ('Wheel-Version', '1.0'),
@@ -107,17 +115,75 @@ def write_ziglang_wheel(out_dir, *, version, platform, archive):
     contents = {}
     contents['ziglang/__init__.py'] = b''
 
+    license_files = {}
+    found_license_files = set()
+    potential_extra_licenses = set()
+
+    # A bunch of standard license file patterns. If a file matches any of
+    # these, we need to add them to required_license_paths and metadata.
+    license_patterns = [
+        r'COPYING.*',
+        r'COPYRIGHT.*',
+        r'COPYLEFT.*',
+        r'LICEN[CS]E.*',
+        r'LICEN[CS]E-.*',
+        r'LICEN[CS]E\..*',
+        r'PATENTS.*',
+        r'NOTICE.*',
+        r'LEGAL.*',
+        r'AUTHORS.*',
+        r'RIGHT*',
+        r'PERMISSION*',
+        r'THIRD[-_]PARTY[-_]LICENSES?.*',
+        r'EULA*',
+        r'MIT*',
+        r'GPL*',
+        r'AGPL*',
+        r'LGPL*',
+        r'APACHE*',
+    ]
+    license_regex = re.compile('|'.join(f'^{pattern}$' for pattern in license_patterns), re.IGNORECASE)
+
+    # These file paths MUST remain in sync with the paths in the official
+    # Zig tarballs and with the ones defined below in the metadata. The
+    # script will raise an error if any of these files are not found in
+    # the archive.
+    required_license_paths = [
+        'LICENSE',
+        'lib/libc/glibc/LICENSES',
+        'lib/libc/mingw/COPYING',
+        'lib/libc/musl/COPYRIGHT',
+        'lib/libc/wasi/LICENSE',
+        'lib/libc/wasi/LICENSE-APACHE',
+        'lib/libc/wasi/LICENSE-APACHE-LLVM',
+        'lib/libc/wasi/LICENSE-MIT',
+        'lib/libc/wasi/libc-bottom-half/cloudlibc/LICENSE',
+        'lib/libc/wasi/libc-top-half/musl/COPYRIGHT',
+        'lib/libcxx/LICENSE.TXT',
+        'lib/libcxxabi/LICENSE.TXT',
+        'lib/libunwind/LICENSE.TXT',
+    ]
+
     for entry_name, entry_mode, entry_data in iter_archive_contents(archive):
         entry_name = '/'.join(entry_name.split('/')[1:])
         if not entry_name:
             continue
         if entry_name.startswith('doc/'):
             continue
 
+        # Check for additional license-like files
+        potential_license_filename = PurePath(entry_name).name
+        if license_regex.match(potential_license_filename):
+            potential_extra_licenses.add(entry_name)
+
         zip_info = ZipInfo(f'ziglang/{entry_name}')
         zip_info.external_attr = (entry_mode & 0xFFFF) << 16
         contents[zip_info] = entry_data
 
+        if entry_name in required_license_paths:
+            license_files[entry_name] = entry_data
+            found_license_files.add(entry_name)
+
         if entry_name.startswith('zig'):
             contents['ziglang/__main__.py'] = f'''\
 import os, sys
@@ -128,25 +194,54 @@ def write_ziglang_wheel(out_dir, *, version, platform, archive):
     import subprocess; sys.exit(subprocess.call(argv))
 '''.encode('ascii')
 
+    # 1. Check for missing required licenses paths
+    missing_licenses = set(required_license_paths) - found_license_files
+    if missing_licenses:
+        error_message = (
+            f"{_YELLOW}The following required license files were not found in the Zig archive: {', '.join(sorted(missing_licenses))} "
+            f"\nThis may indicate a change in Zig's license file structure or an error in the listing of license files and/or paths.{_YELLOW}"
+        )
+        raise RuntimeError(error_message)
+
+    # 2. Check for potentially missing license files
+    extra_licenses = potential_extra_licenses - set(required_license_paths)
+    if extra_licenses:
+        error_message = (
+            f"{_YELLOW}Found additional potential license files in the Zig archive but not included in the metadata: {', '.join(sorted(extra_licenses))} "
+            f"\nPlease consider adding these to the license paths if they should be included.{_YELLOW}"
+        )
+        raise RuntimeError(error_message)
+
     with open('README.pypi.md') as f:
         description = f.read()
 
+    dist_info = f'ziglang-{version}.dist-info'
+    for license_path, license_data in license_files.items():
+        contents[f'{dist_info}/licenses/ziglang/{license_path}'] = license_data
+
     return write_wheel(out_dir,
         name='ziglang',
         version=version,
         tag=f'py3-none-{platform}',
         metadata=[
             ('Summary', 'Zig is a general-purpose programming language and toolchain for maintaining robust, optimal, and reusable software.'),
             ('Description-Content-Type', "'text/markdown'; charset=UTF-8; variant=GFM"),
+            # The license expression and the file paths MUST remain in sync
+            # with the paths in the official Zig tarballs and with the ones
+            # defined above in the contents. The difference is that these
+            # are prefixed with "ziglang/" to match the paths in the wheel
+            # for metadata compliance.
             ('License-Expression', 'MIT'),
-            ('License-File', 'LICENSE'),
+            ('License-File', 'ziglang/LICENSE'),
             ('License-File', 'ziglang/lib/libc/glibc/LICENSES'),
             ('License-File', 'ziglang/lib/libc/mingw/COPYING'),
             ('License-File', 'ziglang/lib/libc/musl/COPYRIGHT'),
             ('License-File', 'ziglang/lib/libc/wasi/LICENSE'),
             ('License-File', 'ziglang/lib/libc/wasi/LICENSE-APACHE'),
             ('License-File', 'ziglang/lib/libc/wasi/LICENSE-APACHE-LLVM'),
             ('License-File', 'ziglang/lib/libc/wasi/LICENSE-MIT'),
+            ('License-File', 'ziglang/lib/libc/wasi/libc-bottom-half/cloudlibc/LICENSE'),
+            ('License-File', 'ziglang/lib/libc/wasi/libc-top-half/musl/COPYRIGHT'),
             ('License-File', 'ziglang/lib/libcxx/LICENSE.TXT'),
             ('License-File', 'ziglang/lib/libcxxabi/LICENSE.TXT'),
             ('License-File', 'ziglang/lib/libunwind/LICENSE.TXT'),