Add option --exclude to audit-translations

PythonScripts/audit_translations/README.md

@@ -77,6 +77,7 @@ uv run --project PythonScripts audit-translations --list
   * Region variants are shown as `lang-region` (e.g., `zz-aa`) based on subdirectories under `Rules/Languages/<lang>`.
 * `--source`: Sets the source/reference language. Defaults to `en`.
 * `--file`: Audits a single specific file instead of the whole directory.
+* `--exclude`: Exclude one or more files from the audit.
 * `--rules-dir`: Override the Rules/Languages directory path.
 * `--only`: Filter issue types (comma-separated): `missing`, `untranslated`, `extra`, `diffs`, `all`.
 * `--verbose`: Show detailed output including source/target snippets for rule differences.
@@ -100,9 +101,17 @@ uv run audit-translations de
 # Compare Norwegian Bokmal against Swedish instead of English
 uv run audit-translations nb --source sv
 
-# Audit only a specific file
+# Audit only a specific file (note: --file is incompatible with --exclude)
+uv run audit-translations es --file ClearSpeak_Rules.yaml
 uv run audit-translations es --file SharedRules/default.yaml
 
+# Exclude a list of files from the audit, note that if you use this option before specifying the
+# target language you'll need to use the option terminator (--).
+# (note: --exclude is incompatible with --file)
+uv run audit-translations es --exclude unicode-full.yaml
+uv run audit-translations es --exclude unicode-full.yaml unicode.yaml
+uv run audit-translations --exclude unicode-full.yaml -- es
+
 # Audit a regional variant (merges Rules/Languages/de and Rules/Languages/de/CH)
 uv run audit-translations de-CH
 

PythonScripts/audit_translations/auditor.py

@@ -46,24 +46,31 @@ def is_definitions_file(file_path: str | Path) -> bool:
     return Path(file_path).name == "definitions.yaml"
 
 
-def get_yaml_files(lang_dir: Path, region_dir: Path | None = None) -> list[Path]:
+def get_yaml_files(lang_dir: Path, region_dir: Path | None = None, excluded_files: list[str] | None = None) -> list[Path]:
     """Get all YAML files to audit for a language, including region overrides."""
     files: set[Path] = set()
 
-    def collect_from(directory: Path, root: Path) -> None:
+    def collect_from(directory: Path, root: Path, excluded: list[str] | None) -> None:
         if not directory.exists():
             return
-        for f in directory.glob("*.yaml"):
-            if f.name != "prefs.yaml":
-                files.add(f.relative_to(root))
-        shared_dir = directory / "SharedRules"
-        if shared_dir.exists():
-            for f in shared_dir.glob("*.yaml"):
-                files.add(f.relative_to(root))
-
-    collect_from(lang_dir, lang_dir)
+
+        excluded_paths = {Path(path) for path in excluded or []}
+
+        candidates = {f for f in directory.glob("*.yaml") if f.name != "prefs.yaml"}
+        candidates.update((directory / "SharedRules").glob("*.yaml"))
+
+        relative_candidates = {f.relative_to(directory): f for f in candidates}
+
+        for path in excluded_paths - relative_candidates.keys():
+            console.print(f"\n[yellow]⚠ Warning:[/] File to exclude {path} does not exist.")
+
+        files.update(
+            f.relative_to(root) for relative_path, f in relative_candidates.items() if relative_path not in excluded_paths
+        )
+
+    collect_from(lang_dir, lang_dir, excluded_files)
     if region_dir:
-        collect_from(region_dir, region_dir)
+        collect_from(region_dir, region_dir, excluded_files)
 
     return sorted(files)
 
@@ -215,6 +222,7 @@ def merge_definitions(
 def audit_language(
     language: str,
     specific_file: str | None = None,
+    excluded_files: list[str] | None = None,
     rules_dir: str | None = None,
     issue_filter: set[str] | None = None,
     verbose: bool = False,
@@ -248,7 +256,7 @@ def audit_language(
         raise AuditError(f"Target region directory not found: {translated_region_dir}")
 
     # Get list of files to audit
-    files = [specific_file] if specific_file else get_yaml_files(source_dir, source_region_dir)
+    files = [specific_file] if specific_file else get_yaml_files(source_dir, source_region_dir, excluded_files)
 
     print_audit_header(language, len(files), source_language)
 

PythonScripts/audit_translations/cli.py

@@ -30,7 +30,18 @@ def main() -> None:
 
     parser.add_argument("language", nargs="?", help="Language code to audit (e.g., 'es', 'de', 'fi')")
     parser.add_argument("--source", default="en", help="Source/reference language code (default: 'en')")
-    parser.add_argument("--file", dest="specific_file", help="Audit only a specific file (e.g., 'SharedRules/default.yaml')")
+    file_group = parser.add_mutually_exclusive_group()
+    file_group.add_argument(
+        "--file",
+        dest="specific_file",
+        help="Audit only a specific file (e.g., 'SharedRules/default.yaml')",
+    )
+    file_group.add_argument(
+        "--exclude",
+        nargs="+",
+        dest="excluded_files",
+        help="Exclude a list of files from the audit.",
+    )
     parser.add_argument("--list", action="store_true", help="List available languages")
     parser.add_argument("--rules-dir", help="Override Rules/Languages directory path")
     parser.add_argument(
@@ -67,6 +78,7 @@ def main() -> None:
             audit_language(
                 args.language,
                 args.specific_file,
+                args.excluded_files,
                 args.rules_dir,
                 issue_filter,
                 args.verbose,

PythonScripts/audit_translations/tests/test_cli_end_to_end.py

@@ -127,6 +127,80 @@ def test_cli_main_rich_only_filters_issue_groups(capsys, monkeypatch) -> None:
     assert "Structure Differences" not in output
 
 
+@pytest.mark.parametrize(
+    ("excluded_file", "expected_file", "expected_rule", "unexpected_rule"),
+    [
+        (
+            "default.yaml",
+            "SharedRules/default.yaml",
+            "shared-only",
+            "root-only",
+        ),
+        (
+            "SharedRules/default.yaml",
+            "default.yaml",
+            "root-only",
+            "shared-only",
+        ),
+    ],
+    ids=["exclude-root", "exclude-shared-rules"],
+)
+def test_cli_main_exclude_uses_relative_paths(
+    tmp_path,
+    capsys,
+    monkeypatch,
+    excluded_file,
+    expected_file,
+    expected_rule,
+    unexpected_rule,
+) -> None:
+    """Ensure --exclude distinguishes root files from SharedRules files."""
+    rules_dir = tmp_path / "Rules" / "Languages"
+    source_dir = rules_dir / "en"
+    target_dir = rules_dir / "de"
+
+    (source_dir / "SharedRules").mkdir(parents=True)
+    (target_dir / "SharedRules").mkdir(parents=True)
+
+    (source_dir / "default.yaml").write_text(
+        '- name: root-only\n  tag: mo\n  match: ".//m:mi"\n  replace:\n    - t: "root"\n',
+        encoding="utf-8",
+    )
+    (source_dir / "SharedRules" / "default.yaml").write_text(
+        '- name: shared-only\n  tag: mo\n  match: ".//m:mi"\n  replace:\n    - t: "shared"\n',
+        encoding="utf-8",
+    )
+
+    target_rule = '- name: target-only\n  tag: mo\n  match: ".//m:mi"\n  replace:\n    - t: "target"\n'
+    (target_dir / "default.yaml").write_text(
+        target_rule,
+        encoding="utf-8",
+    )
+    (target_dir / "SharedRules" / "default.yaml").write_text(
+        target_rule,
+        encoding="utf-8",
+    )
+
+    args = [
+        "de",
+        "--rules-dir",
+        str(rules_dir),
+        "--exclude",
+        excluded_file,
+        "--only",
+        "missing",
+    ]
+    monkeypatch.setattr(sys, "argv", ["audit_translations", *args])
+
+    audit_cli.main()
+    output = strip_ansi(capsys.readouterr().out)
+
+    assert "Files to check: 1" in output
+    assert expected_file in output
+    assert expected_rule in output
+    assert unexpected_rule not in output
+
+
 def test_cli_main_accepts_source_language(capsys, monkeypatch) -> None:
     """
     Ensure --source changes the reference language without changing target semantics.