Use a regex pattern to check for missing licenses

Co-Authored-By: Catherine <[email protected]>

Author: agriyakhetarpal

make_wheels.py

@@ -107,17 +107,43 @@ def iter_archive_contents(archive):
         raise RuntimeError("Unsupported archive format")
 
 
+import re
+from pathlib import PurePath
+
 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)
 
-    # The paths to these licenses MUST match both the actual files
-    # in the Zig source tarballs and the License-File entries listed
-    # below in the metadata. These are not prefixed with "ziglang/"
-    # since these are the actual paths in the Zig source tarballs.
-    license_paths = [
+    required_license_paths = [
         'LICENSE',
         'lib/libc/glibc/LICENSES',
         'lib/libc/mingw/COPYING',
@@ -126,29 +152,32 @@ def write_ziglang_wheel(out_dir, *, version, platform, archive):
         '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'
+        'lib/libunwind/LICENSE.TXT',
     ]
 
-    found_license_files = set()
-
     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
 
-        # Store license files separately for the dist-info
-        # directory (PEP 639)
-        if entry_name in license_paths:
-            license_files[entry_name] = entry_data  # adding to license files
-            found_license_files.add(entry_name)  # tracking found licenses
+        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'''\
@@ -160,14 +189,17 @@ def write_ziglang_wheel(out_dir, *, version, platform, archive):
     import subprocess; sys.exit(subprocess.call(argv))
 '''.encode('ascii')
 
-    # Check if any license files are missing and warn. This
-    # is useful to detect changes in the Zig archive structure
-    # around licenses that may require updating the metadata.
-    missing_licenses = set(license_paths) - found_license_files
+    # 1. Check for missing required licenses paths
+    missing_licenses = set(required_license_paths) - found_license_files
     if missing_licenses:
-        print(f"\033[93mWarning: the following license files were not found in the Zig archive: {', '.join(sorted(missing_licenses))} "
-              "\nThis may indicate a change in Zig's license file structure or an error in the listing of license files and/or paths.\033[0m"
-            )
+        print(f"\033[93mWarning: the following required license files were not found in the Zig archive: {', '.join(sorted(missing_licenses))} "
+              "\nThis may indicate a change in Zig's license file structure or an error in the listing of license files and/or paths.\033[0m")
+
+    # 2. Check for potentially missing license files
+    extra_licenses = potential_extra_licenses - set(required_license_paths)
+    if extra_licenses:
+        print(f"\033[93mWarning: found additional potential license files in the Zig archive but not included in the metadata: {', '.join(sorted(extra_licenses))} "
+              "\nPlease consider adding these to the license paths if they should be included.\033[0m")
 
     with open('README.pypi.md') as f:
         description = f.read()
@@ -198,6 +230,8 @@ def write_ziglang_wheel(out_dir, *, version, platform, archive):
             ('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'),