Google Drive handle invalid PDFs (#838)

backend/danswer/connectors/cross_connector_utils/file_utils.py

@@ -9,6 +9,7 @@ from typing import IO
 
 import chardet
 from pypdf import PdfReader
+from pypdf.errors import PdfStreamError
 
 from danswer.utils.logger import setup_logger
 
@@ -37,29 +38,34 @@ def extract_metadata(line: str) -> dict | None:
 
 
 def read_pdf_file(file: IO[Any], file_name: str, pdf_pass: str | None = None) -> str:
-    pdf_reader = PdfReader(file)
-
-    # if marked as encrypted and a password is provided, try to decrypt
-    if pdf_reader.is_encrypted and pdf_pass is not None:
-        decrypt_success = False
-        if pdf_pass is not None:
-            try:
-                decrypt_success = pdf_reader.decrypt(pdf_pass) != 0
-            except Exception:
-                logger.error(f"Unable to decrypt pdf {file_name}")
-        else:
-            logger.info(f"No Password available to to decrypt pdf {file_name}")
-
-        if not decrypt_success:
-            # By user request, keep files that are unreadable just so they
-            # can be discoverable by title.
-            return ""
-
     try:
+        pdf_reader = PdfReader(file)
+
+        # If marked as encrypted and a password is provided, try to decrypt
+        if pdf_reader.is_encrypted and pdf_pass is not None:
+            decrypt_success = False
+            if pdf_pass is not None:
+                try:
+                    decrypt_success = pdf_reader.decrypt(pdf_pass) != 0
+                except Exception:
+                    logger.error(f"Unable to decrypt pdf {file_name}")
+            else:
+                logger.info(f"No Password available to to decrypt pdf {file_name}")
+
+            if not decrypt_success:
+                # By user request, keep files that are unreadable just so they
+                # can be discoverable by title.
+                return ""
+
         return "\n".join(page.extract_text() for page in pdf_reader.pages)
+    except PdfStreamError:
+        logger.exception(f"PDF file {file_name} is not a valid PDF")
     except Exception:
         logger.exception(f"Failed to read PDF {file_name}")
-        return ""
+
+    # File is still discoverable by title
+    # but the contents are not included as they cannot be parsed
+    return ""
 
 
 def is_macos_resource_fork_file(file_name: str) -> bool: