fix: process_document behavior when exception is raised (#298)

Author: alvarobartt

CHANGELOG.md

@@ -6,6 +6,8 @@
 
 ### Fixes
 
+* Fix `process_document` file cleaning on failure
+
 ## 0.4.16
 
 ### Enhancements

unstructured/ingest/doc_processor/generalized.py

@@ -1,9 +1,12 @@
 """Process aribritrary files with the Unstructured library"""
 
-import logging
+from typing import Any, Dict, List, Optional
 
 from unstructured_inference.models.detectron2 import MODEL_TYPES
 
+from unstructured.ingest.interfaces import BaseIngestDoc as IngestDoc
+from unstructured.logger import logger
+
 
 def initialize():
     """Download models (avoids subprocesses all doing the same)"""
@@ -14,7 +17,7 @@ def initialize():
     MODEL_TYPES[None]["config_path"]
 
 
-def process_document(doc):
+def process_document(doc: "IngestDoc") -> Optional[List[Dict[str, Any]]]:
     """Process any IngestDoc-like class of document with Unstructured's auto partition logic."""
     isd_elems_no_filename = None
     try:
@@ -28,11 +31,9 @@ def process_document(doc):
         # the results. Instead, the MainProcess (caller) may work with the aggregate
         # results across all docs in memory.
         doc.write_result()
-
     except Exception:
         # TODO(crag) save the exception instead of print?
-        logging.error(f"Failed to process {doc}", exc_info=True)
-    else:
-        doc.cleanup_file()
+        logger.error(f"Failed to process {doc}", exc_info=True)
     finally:
+        doc.cleanup_file()
         return isd_elems_no_filename