feat: Add method to get extractions by status and implement user info retrieval in extraction service

app/services/extraction_processor.py

@@ -35,6 +35,9 @@ class ExtractionProcessor:
             logger.warning("Extraction processor is already running")
             return
 
+        # Reset any stuck extractions from previous runs
+        await self._reset_stuck_extractions()
+
         self.shutdown_event.clear()
         self.processor_task = asyncio.create_task(self._process_queue())
         logger.info("Started extraction processor")
@@ -179,6 +182,46 @@ class ExtractionProcessor:
                 self.max_concurrent,
             )
 
+    async def _reset_stuck_extractions(self) -> None:
+        """Reset any extractions stuck in 'processing' status back to 'pending'."""
+        try:
+            async with AsyncSession(engine) as session:
+                extraction_service = ExtractionService(session)
+
+                # Get all extractions stuck in processing status
+                stuck_extractions = (
+                    await extraction_service.extraction_repo.get_by_status("processing")
+                )
+
+                if not stuck_extractions:
+                    logger.info("No stuck extractions found to reset")
+                    return
+
+                reset_count = 0
+                for extraction in stuck_extractions:
+                    try:
+                        await extraction_service.extraction_repo.update(
+                            extraction, {"status": "pending", "error": None}
+                        )
+                        reset_count += 1
+                        logger.info(
+                            "Reset stuck extraction %d from processing to pending",
+                            extraction.id,
+                        )
+                    except Exception:
+                        logger.exception(
+                            "Failed to reset extraction %d", extraction.id
+                        )
+
+                await session.commit()
+                logger.info(
+                    "Successfully reset %d stuck extractions from processing to pending",
+                    reset_count,
+                )
+
+        except Exception:
+            logger.exception("Failed to reset stuck extractions")
+
     def get_status(self) -> dict:
         """Get the current status of the extraction processor."""
         return {