improve the OCR pipeline processing and outputs formatting

src/ocr_pipeline/watch/watcher.py

@@ -74,7 +74,7 @@ class Watcher:
         if self.running:
             return
 
-        self.run_id = self.db.start_run()
+        self.run_id, _ = self.db.start_run()
         self.running = True
 
         self.handler = ScreenshotHandler(self._process_file)
@@ -112,29 +112,14 @@ class Watcher:
                         logger.error("watch_process_failed", path=str(path), error=str(e))
                         failed += 1
 
-        self.db.complete_run(
-            time.strftime("%Y-%m-%d"),
-            self.run_id,
-            {"total": processed, "successful": processed, "failed": failed, "chunks": 0},
-        )
+        if self.run_id is not None:
+            self.db.complete_run(self.run_id, {"total": processed + failed, "successful": processed, "failed": failed, "chunks": 0})
         logger.info("watcher_stopped", run_id=self.run_id, processed=processed, failed=failed)
 
     def _process_file(self, path: Path) -> None:
-        from ocr_pipeline.ocr.engine import compute_image_hash
-
-        file_hash = compute_image_hash(path)
-        if self.db.is_processed(file_hash, path):
-            logger.debug("file_already_processed", path=str(path))
-            return
-
         logger.info("processing_new_file", path=str(path))
+        # The pipeline owns deduplication and database writes, preventing double marks.
         self.process_callback(path)
-        self.db.mark_processed(
-            file_hash=file_hash,
-            path=str(path),
-            timestamp=time.time(),
-            run_id=self.run_id,
-        )
 
     def stop(self) -> None:
         self.running = False