reuse: add option to ignore paths for copyright check

pre-commit-action/action.yml

@@ -40,6 +40,11 @@ inputs:
       Name of the reuse Jinja2 template in .reuse/templates/ (without .jinja2 suffix). Consumer repos can provide their own template. Defaults to "opensovd".
     required: false
     default: 'opensovd'
+  ignore-paths:
+    description: >
+      Comma-separated list of file patterns to ignore during REUSE checks (e.g., "*.md,docs/**,*.txt"). Files matching these patterns will be skipped by reuse annotate.
+    required: false
+    default: ''
 
 runs:
   using: 'composite'
@@ -99,4 +104,5 @@ runs:
           --hook-script="${GITHUB_ACTION_PATH}/reuse-annotate-hook.py" \
           --copyright="${{ inputs.copyright-text }}" \
           --license="${{ inputs.license }}" \
-          --template="${{ inputs.reuse-template }}"
+          --template="${{ inputs.reuse-template }}" \
+          --ignore-paths="${{ inputs.ignore-paths }}"

pre-commit-action/reuse-annotate-hook.py

@@ -51,6 +51,7 @@
 DEFAULT_COPYRIGHT = "The Contributors to Eclipse OpenSOVD (see CONTRIBUTORS)"
 DEFAULT_LICENSE = "Apache-2.0"
 DEFAULT_TEMPLATE = "opensovd"
+DEFAULT_IGNORE_PATHS = ""
 STYLES_CONFIG = ".reuse/styles.toml"
 
 
@@ -163,10 +164,39 @@ def fix_wrong_copyright(filepath: str, copyright_text: str) -> None:
     path.write_text("".join(lines))
 
 
+def should_ignore(filepath: str, ignore_patterns: list[str]) -> bool:
+    """Check if a file should be ignored based on ignore patterns."""
+    if not ignore_patterns:
+        return False
+
+    # Normalize the filepath
+    filepath = filepath.replace("\\", "/")
+
+    for pattern in ignore_patterns:
+        pattern = pattern.strip()
+        if not pattern:
+            continue
+
+        # Support both simple glob patterns and path-based patterns
+        if fnmatch(filepath, pattern):
+            return True
+
+        # Also check just the filename
+        basename = os.path.basename(filepath)
+        if fnmatch(basename, pattern):
+            return True
+
+    return False
+
+
 def main() -> int:
     copyright_text = os.environ.get("REUSE_COPYRIGHT", DEFAULT_COPYRIGHT)
     license_id = os.environ.get("REUSE_LICENSE", DEFAULT_LICENSE)
     template = os.environ.get("REUSE_TEMPLATE", DEFAULT_TEMPLATE)
+    ignore_paths_str = os.environ.get("REUSE_IGNORE_PATHS", DEFAULT_IGNORE_PATHS)
+
+    # Parse ignore patterns from comma-separated string
+    ignore_patterns = [p.strip() for p in ignore_paths_str.split(",") if p.strip()]
 
     files = sys.argv[1:]
     if not files:
@@ -181,6 +211,10 @@ def main() -> int:
     styles = load_styles()
 
     for filepath in files:
+        # Skip ignored files
+        if should_ignore(filepath, ignore_patterns):
+            continue
+
         # Never annotate license text files in the LICENSES/ directory
         if filepath.startswith("LICENSES/") or "/LICENSES/" in filepath:
             continue

run_checks.py

@@ -70,7 +70,7 @@ def patch_config(config_content, *, fix_mode):
 
 
 def patch_hook_script(
-    script_content, *, copyright_text, license_id, template, fix_mode
+    script_content, *, copyright_text, license_id, template, ignore_paths, fix_mode
 ):
     """Patch the reuse-annotate hook script with configured values.
 
@@ -89,6 +89,10 @@ def patch_hook_script(
         'DEFAULT_TEMPLATE = "opensovd"',
         f'DEFAULT_TEMPLATE = "{template}"',
     )
+    script_content = script_content.replace(
+        'DEFAULT_IGNORE_PATHS = ""',
+        f'DEFAULT_IGNORE_PATHS = "{ignore_paths}"',
+    )
 
     return script_content
 
@@ -181,6 +185,11 @@ def main():
         action="store_true",
         help="Run in check-only mode (no auto-fix). Used in CI to report issues without modifying files.",
     )
+    parser.add_argument(
+        "--ignore-paths",
+        default="",
+        help="Comma-separated list of file patterns to ignore during REUSE checks (e.g., '*.md,docs/**,*.txt')",
+    )
 
     args = parser.parse_args()
 
@@ -249,6 +258,7 @@ def main():
             copyright_text=args.copyright,
             license_id=args.license,
             template=args.template,
+            ignore_paths=args.ignore_paths,
             fix_mode=fix_mode,
         )