Refactor gdrive_downloader.py - Google auth session for private documents

- Used a double-check locking to ensure that multiple threads won't create redundant auth sessions simultaneously.

Author: milovate

daras_ai_v2/asr.py

@@ -2,6 +2,7 @@
 import multiprocessing
 import os.path
 import tempfile
+import threading
 import typing
 from enum import Enum
 
@@ -873,9 +874,6 @@ def _translate_text(
     return result.strip()
 
 
-_session = None
-
-
 def _MinT_translate_one_text(
     text: str, source_language: str, target_language: str
 ) -> str:
@@ -894,18 +892,27 @@ def _MinT_translate_one_text(
     return tanslation.get("translation", text)
 
 
-def get_google_auth_session():
+_session = None
+_session_lock = threading.Lock()
+
+
+def get_google_auth_session(scopes: typing.Optional[list[str]] = None):
     global _session
 
     if _session is None:
-        import google.auth
-        from google.auth.transport.requests import AuthorizedSession
+        with _session_lock:
+            if _session is None:
+                import google.auth
+                from google.auth.transport.requests import AuthorizedSession
 
-        creds, project = google.auth.default(
-            scopes=["https://www.googleapis.com/auth/cloud-platform"]
-        )
-        # takes care of refreshing the token and adding it to request headers
-        _session = AuthorizedSession(credentials=creds), project
+                if not scopes:
+                    scopes = ["https://www.googleapis.com/auth/cloud-platform"]
+
+                creds, project = google.auth.default(
+                    scopes=scopes,
+                )
+                # takes care of refreshing the token and adding it to request headers
+                _session = AuthorizedSession(credentials=creds), project
 
     return _session
 

daras_ai_v2/gdrive_downloader.py

@@ -74,6 +74,7 @@ def gdrive_download(
 ) -> tuple[bytes, str]:
     from googleapiclient import discovery
     from googleapiclient.http import MediaIoBaseDownload
+    from daras_ai_v2.asr import get_google_auth_session
 
     if export_links is None:
         export_links = {}
@@ -87,7 +88,9 @@ def gdrive_download(
         # export google docs to appropriate type
         export_mime_type = DOCS_EXPORT_MIMETYPES.get(mime_type, mime_type)
         if f_url_export := export_links.get(export_mime_type, None):
-            r = requests.get(f_url_export)
+            drive_scopes = ["https://www.googleapis.com/auth/drive.readonly"]
+            session, _ = get_google_auth_session(drive_scopes)
+            r = session.get(f_url_export)
             file_bytes = r.content
             raise_for_status(r, is_user_url=True)
             return file_bytes, export_mime_type

daras_ai_v2/vector_search.py

@@ -373,9 +373,23 @@ def doc_url_to_file_metadata(f_url: str) -> FileMetadata:
             meta = gdrive_metadata(url_to_gdrive_file_id(f))
         except HttpError as e:
             if e.status_code == 404:
+                from google.oauth2.service_account import Credentials
+
+                service_account_client_email = Credentials.from_service_account_file(
+                    settings.service_account_key_path
+                ).service_account_email
+
                 raise UserError(
-                    f"Could not download the google doc at {f_url} "
-                    f"Please make sure to make the document public for viewing."
+                    # language=HTML
+                    f"""<p>This knowledge base Google Doc is not accessible: <a href="{f_url}" target="_blank">{f_url}</a></p>
+                <p>To address this:</p>
+                <ul>
+                  <li>Please make the Google Doc publicly viewable, or</li>
+                  <li>Share the Doc or its parent folder with <br>
+                    <a href="mailto:{service_account_client_email}">{service_account_client_email}</a>
+                    as an authorized viewer.
+                  </li>
+                </ul>"""
                 ) from e
             else:
                 raise