feat: add clean-up method

Signed-off-by: Aleš Kalfas <kalfas.ales@gmail.com>

Author: aleskalfas

apps/agentstack-server/src/agentstack_server/service_layer/services/files.py

@@ -65,20 +65,26 @@ async def extract_text(self, file_id: UUID, job_id: str):
             CancelledError: If the job is cancelled
             Exception: For any extraction or upload errors
         """
-        error_log = []
-        async with self._uow() as uow:
-            extraction = await uow.files.get_extraction_by_file_id(file_id=file_id)
-            file = await uow.files.get(file_id=file_id)
-            error_log.append(file.model_dump())
-            user = await uow.users.get(user_id=file.created_by)
-
-            extraction.set_started(
-                job_id=job_id,
-                backend=self._extraction_backend.backend_name,
-            )
-            await uow.files.update_extraction(extraction=extraction)
-            await uow.commit()
+        error_log: list[str] = []
+        extraction: TextExtraction | None = None
+        file: File | None = None
+        user: User | None = None
+        uploaded_file_ids: list[UUID] = []
+
         try:
+            async with self._uow() as uow:
+                extraction = await uow.files.get_extraction_by_file_id(file_id=file_id)
+                file = await uow.files.get(file_id=file_id)
+                error_log.append(file.model_dump())
+                user = await uow.users.get(user_id=file.created_by)
+
+                extraction.set_started(
+                    job_id=job_id,
+                    backend=self._extraction_backend.backend_name,
+                )
+                await uow.files.update_extraction(extraction=extraction)
+                await uow.commit()
+
             file_url = await self._object_storage.get_file_url(file_id=file_id)
             error_log.append(f"file url: {file_url}")
 
@@ -95,24 +101,30 @@ async def extract_text(self, file_id: UUID, job_id: str):
                         context_id=file.context_id,
                         parent_file_id=file_id,
                     )
+                    uploaded_file_ids.append(extracted_db_file.id)
                     extracted_files.append(ExtractedFileInfo(file_id=extracted_db_file.id, format=extraction_format))
 
             extraction.set_completed(extracted_files=extracted_files)
             async with self._uow() as uow:
                 await uow.files.update_extraction(extraction=extraction)
                 await uow.commit()
+            uploaded_file_ids.clear()
         except CancelledError:
-            async with self._uow() as uow:
-                extraction.set_cancelled()
-                await uow.files.update_extraction(extraction=extraction)
-                await uow.commit()
+            await self._cleanup_extracted_files(uploaded_file_ids)
+            if extraction:
+                async with self._uow() as uow:
+                    extraction.set_cancelled()
+                    await uow.files.update_extraction(extraction=extraction)
+                    await uow.commit()
             raise
         except Exception as ex:
             error_log.append(str(ex))
-            async with self._uow() as uow:
-                extraction.set_failed("\n".join(str(e) for e in error_log))
-                await uow.files.update_extraction(extraction=extraction)
-                await uow.commit()
+            await self._cleanup_extracted_files(uploaded_file_ids)
+            if extraction:
+                async with self._uow() as uow:
+                    extraction.set_failed("\n".join(str(e) for e in error_log))
+                    await uow.files.update_extraction(extraction=extraction)
+                    await uow.commit()
             raise
 
     async def upload_file(
@@ -202,6 +214,22 @@ async def delete(self, *, file_id: UUID, user: User, context_id: UUID | None = N
                 await self._object_storage.delete_files(file_ids=file_ids_to_delete)
                 await uow.commit()
 
+    async def _cleanup_extracted_files(self, file_ids: list[UUID]) -> None:
+        """Best-effort cleanup for partially uploaded extracted files."""
+        if not file_ids:
+            return
+
+        unique_ids = list(dict.fromkeys(file_ids))
+
+        with suppress(Exception):
+            await self._object_storage.delete_files(file_ids=unique_ids)
+
+        with suppress(Exception):
+            async with self._uow() as uow:
+                for extracted_file_id in unique_ids:
+                    await uow.files.delete(file_id=extracted_file_id)
+                await uow.commit()
+
     async def create_extraction(
         self,
         *,