welcome gitea

prune_manifest.py

@@ -0,0 +1,64 @@
+"""
+VocabListGenerator — Manifest cleaner
+----------------------------------------
+Removes entries from vocab_manifest.json whose output files no longer
+exist on disk.  Reads output_dir and manifest filename from batch.yaml
+(falls back to sensible defaults if batch.yaml is missing).
+
+Usage:
+    python prune_manifest.py
+"""
+
+import os
+import sys
+
+import yaml
+
+from manifest_manager import prune_missing_files, print_manifest
+
+
+def main() -> None:
+    # ── Resolve paths from batch.yaml (or defaults) ──────────────────────────
+    script_dir = os.path.dirname(os.path.abspath(__file__))
+
+    batch_yaml = os.path.join(script_dir, "batch.yaml")
+    output_dir    = "output"
+    manifest_file = "vocab_manifest.json"
+
+    if os.path.isfile(batch_yaml):
+        try:
+            with open(batch_yaml, "r", encoding="utf-8") as f:
+                cfg = yaml.safe_load(f) or {}
+            settings      = cfg.get("settings", {})
+            output_dir    = settings.get("output_dir",        output_dir)
+            manifest_file = settings.get("manifest_filename", manifest_file)
+        except yaml.YAMLError as e:
+            print(f"  WARNING: Could not parse batch.yaml ({e}), using defaults.")
+
+    output_dir    = os.path.join(script_dir, output_dir)
+    manifest_path = os.path.join(output_dir, manifest_file)
+
+    # ── Run prune ─────────────────────────────────────────────────────────────
+    print("=" * 50)
+    print("   VocabListGenerator — Manifest Cleaner")
+    print("=" * 50)
+    print(f"  Manifest : {manifest_path}")
+    print(f"  Output   : {output_dir}")
+    print()
+
+    if not os.path.isfile(manifest_path):
+        print("  No manifest file found — nothing to do.")
+        sys.exit(0)
+
+    removed = prune_missing_files(manifest_path, output_dir)
+
+    if removed == 0:
+        print("  Manifest is already clean — no stale entries found.")
+    else:
+        print()
+
+    print_manifest(manifest_path)
+
+
+if __name__ == "__main__":
+    main()