full drive perm sync

backend/onyx/connectors/google_drive/connector.py

@@ -619,7 +619,6 @@ class GoogleDriveConnector(SlimConnector, CheckpointedConnector[GoogleDriveCheck
 
         if checkpoint.completion_stage == DriveRetrievalStage.USER_EMAILS:
             all_org_emails: list[str] = self._get_all_user_emails()
-            if not is_slim:
             checkpoint.user_emails = all_org_emails
             checkpoint.completion_stage = DriveRetrievalStage.DRIVE_IDS
         else:
@@ -730,7 +729,6 @@ class GoogleDriveConnector(SlimConnector, CheckpointedConnector[GoogleDriveCheck
             elif self.include_shared_drives:
                 sorted_drive_ids = sorted(all_drive_ids)
 
-            if not is_slim:
             checkpoint.drive_ids_to_retrieve = sorted_drive_ids
             checkpoint.folder_ids_to_retrieve = sorted_folder_ids
             checkpoint.completion_stage = next_stage
@@ -908,9 +906,6 @@ class GoogleDriveConnector(SlimConnector, CheckpointedConnector[GoogleDriveCheck
             start=start,
             end=end,
         )
-        if is_slim:
-            yield from drive_files
-            return
 
         for file in drive_files:
             logger.debug(
@@ -1146,13 +1141,14 @@ class GoogleDriveConnector(SlimConnector, CheckpointedConnector[GoogleDriveCheck
 
     def _extract_slim_docs_from_google_drive(
         self,
+        checkpoint: GoogleDriveCheckpoint,
         start: SecondsSinceUnixEpoch | None = None,
         end: SecondsSinceUnixEpoch | None = None,
         callback: IndexingHeartbeatInterface | None = None,
     ) -> GenerateSlimDocumentOutput:
         slim_batch = []
         for file in self._fetch_drive_items(
-            checkpoint=self.build_dummy_checkpoint(),
+            checkpoint=checkpoint,
             is_slim=True,
             start=start,
             end=end,
@@ -1179,9 +1175,15 @@ class GoogleDriveConnector(SlimConnector, CheckpointedConnector[GoogleDriveCheck
         callback: IndexingHeartbeatInterface | None = None,
     ) -> GenerateSlimDocumentOutput:
         try:
+            checkpoint = self.build_dummy_checkpoint()
+            while checkpoint.completion_stage != DriveRetrievalStage.DONE:
                 yield from self._extract_slim_docs_from_google_drive(
-                start, end, callback=callback
+                    checkpoint=checkpoint,
+                    start=start,
+                    end=end,
+                    callback=callback,
                 )
+
         except Exception as e:
             if MISSING_SCOPES_ERROR_STR in str(e):
                 raise PermissionError(ONYX_SCOPE_INSTRUCTIONS) from e

backend/onyx/connectors/google_drive/file_retrieval.py

@@ -30,7 +30,7 @@ FILE_FIELDS = (
 )
 SLIM_FILE_FIELDS = (
     f"nextPageToken, files(mimeType, driveId, id, name, {PERMISSION_FULL_DESCRIPTION}, "
-    "permissionIds, webViewLink, owners(emailAddress))"
+    "permissionIds, webViewLink, owners(emailAddress), modifiedTime)"
 )
 FOLDER_FIELDS = "nextPageToken, files(id, name, permissions, modifiedTime, webViewLink, shortcutDetails)"
 

backend/tests/daily/connectors/google_drive/test_drive_perm_sync.py

@@ -154,4 +154,21 @@ def test_gdrive_perm_sync_with_real_data(
                 f"but is accessible to {emails_with_access}. Raw result: {doc_to_raw_result_mapping[doc_id]} "
             )
 
+    # Verify that we checked every file in ACCESS_MAPPING
+    all_expected_files = set()
+    for file_ids in ACCESS_MAPPING.values():
+        all_expected_files.update(file_ids)
+
+    checked_file_ids = {
+        url_to_id_mapping[doc_id]
+        for doc_id in doc_to_email_mapping
+        if doc_id in url_to_id_mapping
+    }
+
+    assert all_expected_files == checked_file_ids, (
+        f"Not all expected files were checked. "
+        f"Missing files: {all_expected_files - checked_file_ids}, "
+        f"Extra files checked: {checked_file_ids - all_expected_files}"
+    )
+
     print(f"Checked permissions for {checked_files} files from drive_id_mapping.json")