feat(bulk-ingest): Add --ignored Flag to Exclude Specific Files and Directories During Ingestion (#1432)

.gitignore

@@ -1,4 +1,6 @@
 .venv
+.env
+venv
 
 settings-me.yaml
 

Makefile

@@ -56,3 +56,20 @@ wipe:
 
 setup:
 	poetry run python scripts/setup
+
+list:
+	@echo "Available commands:"
+	@echo "  test            : Run tests using pytest"
+	@echo "  test-coverage   : Run tests with coverage report"
+	@echo "  black           : Check code format with black"
+	@echo "  ruff            : Check code with ruff"
+	@echo "  format          : Format code with black and ruff"
+	@echo "  mypy            : Run mypy for type checking"
+	@echo "  check           : Run format and mypy commands"
+	@echo "  run             : Run the application"
+	@echo "  dev-windows     : Run the application in development mode on Windows"
+	@echo "  dev             : Run the application in development mode"
+	@echo "  api-docs        : Generate API documentation"
+	@echo "  ingest          : Ingest data using specified script"
+	@echo "  wipe            : Wipe data using specified script"
+	@echo "  setup           : Setup the application"

scripts/ingest_folder.py

@@ -20,20 +20,20 @@ class LocalIngestWorker:
 
         self._files_under_root_folder: list[Path] = list()
 
-    def _find_all_files_in_folder(self, root_path: Path) -> None:
+    def _find_all_files_in_folder(self, root_path: Path, ignored: list[str]) -> None:
         """Search all files under the root folder recursively.
         Count them at the same time
         """
         for file_path in root_path.iterdir():
-            if file_path.is_file():
+            if file_path.is_file() and file_path.name not in ignored:
                 self.total_documents += 1
                 self._files_under_root_folder.append(file_path)
-            elif file_path.is_dir():
+            elif file_path.is_dir() and file_path.name not in ignored:
-                self._find_all_files_in_folder(file_path)
+                self._find_all_files_in_folder(file_path, ignored)
 
-    def ingest_folder(self, folder_path: Path) -> None:
+    def ingest_folder(self, folder_path: Path, ignored: list[str]) -> None:
         # Count total documents before ingestion
-        self._find_all_files_in_folder(folder_path)
+        self._find_all_files_in_folder(folder_path, ignored)
         self._ingest_all(self._files_under_root_folder)
 
     def _ingest_all(self, files_to_ingest: list[Path]) -> None:
@@ -64,12 +64,19 @@ parser.add_argument(
     action=argparse.BooleanOptionalAction,
     default=False,
 )
+parser.add_argument(
+    "--ignored",
+    nargs="*",
+    help="List of files/directories to ignore",
+    default=[],
+)
 parser.add_argument(
     "--log-file",
     help="Optional path to a log file. If provided, logs will be written to this file.",
     type=str,
     default=None,
 )
+
 args = parser.parse_args()
 
 # Set up logging to a file if a path is provided
@@ -91,9 +98,17 @@ if __name__ == "__main__":
 
     ingest_service = global_injector.get(IngestService)
     worker = LocalIngestWorker(ingest_service)
-    worker.ingest_folder(root_path)
+    worker.ingest_folder(root_path, args.ignored)
+
+    if args.ignored:
+        logger.info(f"Skipping following files and directories: {args.ignored}")
 
     if args.watch:
         logger.info(f"Watching {args.folder} for changes, press Ctrl+C to stop...")
+        directories_to_watch = [
+            dir
+            for dir in root_path.iterdir()
+            if dir.is_dir() and dir.name not in args.ignored
+        ]
         watcher = IngestWatcher(args.folder, worker.ingest_on_watch)
         watcher.start()