From 125e5eaab12c906869aa9aa3cd9dc74f2807db09 Mon Sep 17 00:00:00 2001
From: pablodanswer <pablo@danswer.ai>
Date: Sat, 1 Feb 2025 15:45:52 -0800
Subject: [PATCH 1/3] various mypy improvements

---
 backend/alembic/__init__.py                               | 0
 backend/onyx/auth/users.py                                | 4 ++++
 backend/onyx/background/celery/tasks/indexing/tasks.py    | 3 ++-
 backend/onyx/background/celery/tasks/indexing/utils.py    | 1 +
 backend/onyx/background/celery/tasks/monitoring/tasks.py  | 1 +
 backend/onyx/background/celery/tasks/shared/tasks.py      | 1 +
 backend/onyx/background/celery/tasks/vespa/tasks.py       | 1 +
 backend/onyx/background/indexing/run_indexing.py          | 1 +
 backend/onyx/chat/process_message.py                      | 4 +++-
 backend/onyx/connectors/airtable/airtable_connector.py    | 3 ++-
 backend/onyx/connectors/salesforce/doc_conversion.py      | 1 +
 backend/onyx/db/document_set.py                           | 1 +
 backend/onyx/file_processing/extract_file_text.py         | 2 +-
 backend/onyx/indexing/chunker.py                          | 2 ++
 backend/onyx/onyxbot/slack/utils.py                       | 2 +-
 backend/onyx/redis/redis_connector_doc_perm_sync.py       | 3 ++-
 backend/onyx/redis/redis_connector_ext_group_sync.py      | 3 ++-
 backend/onyx/redis/redis_connector_index.py               | 3 ++-
 backend/onyx/server/manage/users.py                       | 2 ++
 backend/scripts/chat_feedback_dump.py                     | 1 +
 .../daily/connectors/confluence/test_confluence_basic.py  | 8 ++++++++
 .../tests/daily/connectors/gmail/test_gmail_connector.py  | 2 ++
 backend/tests/integration/common_utils/managers/user.py   | 1 +
 23 files changed, 42 insertions(+), 8 deletions(-)
 create mode 100644 backend/alembic/__init__.py

diff --git a/backend/alembic/__init__.py b/backend/alembic/__init__.py
new file mode 100644
index 000000000..e69de29bb
diff --git a/backend/onyx/auth/users.py b/backend/onyx/auth/users.py
index 0d8fa5474..da9865550 100644
--- a/backend/onyx/auth/users.py
+++ b/backend/onyx/auth/users.py
@@ -245,6 +245,8 @@ class UserManager(UUIDIDMixin, BaseUserManager[User, uuid.UUID]):
             referral_source=referral_source,
             request=request,
         )
+        user: User
+
         async with get_async_session_with_tenant(tenant_id) as db_session:
             token = CURRENT_TENANT_ID_CONTEXTVAR.set(tenant_id)
             verify_email_is_invited(user_create.email)
@@ -368,6 +370,8 @@ class UserManager(UUIDIDMixin, BaseUserManager[User, uuid.UUID]):
                 "refresh_token": refresh_token,
             }
 
+            user: User
+
             try:
                 # Attempt to get user by OAuth account
                 user = await self.get_by_oauth_account(oauth_name, account_id)
diff --git a/backend/onyx/background/celery/tasks/indexing/tasks.py b/backend/onyx/background/celery/tasks/indexing/tasks.py
index 2e3c0e3ae..b1451a051 100644
--- a/backend/onyx/background/celery/tasks/indexing/tasks.py
+++ b/backend/onyx/background/celery/tasks/indexing/tasks.py
@@ -586,11 +586,12 @@ def connector_indexing_proxy_task(
 
         # if the job is done, clean up and break
         if job.done():
+            exit_code: int | None
             try:
                 if job.status == "error":
                     ignore_exitcode = False
 
-                    exit_code: int | None = None
+                    exit_code = None
                     if job.process:
                         exit_code = job.process.exitcode
 
diff --git a/backend/onyx/background/celery/tasks/indexing/utils.py b/backend/onyx/background/celery/tasks/indexing/utils.py
index e14e79b5f..00cbc3632 100644
--- a/backend/onyx/background/celery/tasks/indexing/utils.py
+++ b/backend/onyx/background/celery/tasks/indexing/utils.py
@@ -438,6 +438,7 @@ def try_creating_indexing_task(
     if not acquired:
         return None
 
+    redis_connector_index: RedisConnectorIndex
     try:
         redis_connector = RedisConnector(tenant_id, cc_pair.id)
         redis_connector_index = redis_connector.new_index(search_settings.id)
diff --git a/backend/onyx/background/celery/tasks/monitoring/tasks.py b/backend/onyx/background/celery/tasks/monitoring/tasks.py
index 037560841..244b4abd2 100644
--- a/backend/onyx/background/celery/tasks/monitoring/tasks.py
+++ b/backend/onyx/background/celery/tasks/monitoring/tasks.py
@@ -747,6 +747,7 @@ def cloud_check_alembic() -> bool | None:
     revision_counts: dict[str, int] = {}
     out_of_date_tenants: dict[str, str | None] = {}
     top_revision: str = ""
+    tenant_ids: list[str] = []
 
     try:
         # map each tenant_id to its revision
diff --git a/backend/onyx/background/celery/tasks/shared/tasks.py b/backend/onyx/background/celery/tasks/shared/tasks.py
index 5530d9eeb..a1bc12252 100644
--- a/backend/onyx/background/celery/tasks/shared/tasks.py
+++ b/backend/onyx/background/celery/tasks/shared/tasks.py
@@ -247,6 +247,7 @@ def cloud_beat_task_generator(
         return None
 
     last_lock_time = time.monotonic()
+    tenant_ids: list[str] | list[None] = []
 
     try:
         tenant_ids = get_all_tenant_ids()
diff --git a/backend/onyx/background/celery/tasks/vespa/tasks.py b/backend/onyx/background/celery/tasks/vespa/tasks.py
index c695f92a8..4033b2503 100644
--- a/backend/onyx/background/celery/tasks/vespa/tasks.py
+++ b/backend/onyx/background/celery/tasks/vespa/tasks.py
@@ -1084,6 +1084,7 @@ def vespa_metadata_sync_task(
         task_logger.info(f"SoftTimeLimitExceeded exception. doc={document_id}")
         return False
     except Exception as ex:
+        e: Exception | None = None
         if isinstance(ex, RetryError):
             task_logger.warning(
                 f"Tenacity retry failed: num_attempts={ex.last_attempt.attempt_number}"
diff --git a/backend/onyx/background/indexing/run_indexing.py b/backend/onyx/background/indexing/run_indexing.py
index 94f8cb01b..fd0ab0b35 100644
--- a/backend/onyx/background/indexing/run_indexing.py
+++ b/backend/onyx/background/indexing/run_indexing.py
@@ -239,6 +239,7 @@ def _run_indexing(
         callback=callback,
     )
 
+    tracer: OnyxTracer
     if INDEXING_TRACER_INTERVAL > 0:
         logger.debug(f"Memory tracer starting: interval={INDEXING_TRACER_INTERVAL}")
         tracer = OnyxTracer()
diff --git a/backend/onyx/chat/process_message.py b/backend/onyx/chat/process_message.py
index 8c1de70ca..f87de5f48 100644
--- a/backend/onyx/chat/process_message.py
+++ b/backend/onyx/chat/process_message.py
@@ -87,6 +87,7 @@ from onyx.file_store.utils import save_files
 from onyx.llm.exceptions import GenAIDisabledException
 from onyx.llm.factory import get_llms_for_persona
 from onyx.llm.factory import get_main_llm_from_tuple
+from onyx.llm.interfaces import LLM
 from onyx.llm.models import PreviousMessage
 from onyx.llm.utils import litellm_exception_to_error_msg
 from onyx.natural_language_processing.utils import get_tokenizer
@@ -349,7 +350,8 @@ def stream_chat_message_objects(
     new_msg_req.chunks_above = 0
     new_msg_req.chunks_below = 0
 
-    llm = None
+    llm: LLM
+
     try:
         user_id = user.id if user is not None else None
 
diff --git a/backend/onyx/connectors/airtable/airtable_connector.py b/backend/onyx/connectors/airtable/airtable_connector.py
index 37b84623f..c4d0e63d9 100644
--- a/backend/onyx/connectors/airtable/airtable_connector.py
+++ b/backend/onyx/connectors/airtable/airtable_connector.py
@@ -369,11 +369,12 @@ class AirtableConnector(LoadConnector):
         # Process records in parallel batches using ThreadPoolExecutor
         PARALLEL_BATCH_SIZE = 8
         max_workers = min(PARALLEL_BATCH_SIZE, len(records))
+        record_documents: list[Document] = []
 
         # Process records in batches
         for i in range(0, len(records), PARALLEL_BATCH_SIZE):
             batch_records = records[i : i + PARALLEL_BATCH_SIZE]
-            record_documents: list[Document] = []
+            record_documents = []
 
             with ThreadPoolExecutor(max_workers=max_workers) as executor:
                 # Submit batch tasks
diff --git a/backend/onyx/connectors/salesforce/doc_conversion.py b/backend/onyx/connectors/salesforce/doc_conversion.py
index e6acaf2e0..c9f48a9ec 100644
--- a/backend/onyx/connectors/salesforce/doc_conversion.py
+++ b/backend/onyx/connectors/salesforce/doc_conversion.py
@@ -59,6 +59,7 @@ def _clean_salesforce_dict(data: dict | list) -> dict | list:
     elif isinstance(data, list):
         filtered_list = []
         for item in data:
+            filtered_item: dict | list
             if isinstance(item, (dict, list)):
                 filtered_item = _clean_salesforce_dict(item)
                 # Only add non-empty dictionaries or lists
diff --git a/backend/onyx/db/document_set.py b/backend/onyx/db/document_set.py
index 6383a6c02..bf9d6d38d 100644
--- a/backend/onyx/db/document_set.py
+++ b/backend/onyx/db/document_set.py
@@ -221,6 +221,7 @@ def insert_document_set(
             group_ids=document_set_creation_request.groups or [],
         )
 
+    new_document_set_row: DocumentSetDBModel
     try:
         new_document_set_row = DocumentSetDBModel(
             name=document_set_creation_request.name,
diff --git a/backend/onyx/file_processing/extract_file_text.py b/backend/onyx/file_processing/extract_file_text.py
index efdeeb546..cebd52b73 100644
--- a/backend/onyx/file_processing/extract_file_text.py
+++ b/backend/onyx/file_processing/extract_file_text.py
@@ -365,7 +365,7 @@ def extract_file_text(
                     f"Failed to process with Unstructured: {str(unstructured_error)}. Falling back to normal processing."
                 )
                 # Fall through to normal processing
-
+        final_extension: str
         if file_name or extension:
             if extension is not None:
                 final_extension = extension
diff --git a/backend/onyx/indexing/chunker.py b/backend/onyx/indexing/chunker.py
index f95fe86e3..e42dafb29 100644
--- a/backend/onyx/indexing/chunker.py
+++ b/backend/onyx/indexing/chunker.py
@@ -223,6 +223,8 @@ class Chunker:
                 large_chunk_id=None,
             )
 
+        section_link_text: str
+
         for section_idx, section in enumerate(document.sections):
             section_text = clean_text(section.text)
             section_link_text = section.link or ""
diff --git a/backend/onyx/onyxbot/slack/utils.py b/backend/onyx/onyxbot/slack/utils.py
index f8f3ad845..4a940c37a 100644
--- a/backend/onyx/onyxbot/slack/utils.py
+++ b/backend/onyx/onyxbot/slack/utils.py
@@ -518,7 +518,7 @@ def read_slack_thread(
             message_type = MessageType.USER
         else:
             self_slack_bot_id = get_onyx_bot_slack_bot_id(client)
-
+            blocks: Any
             if reply.get("user") == self_slack_bot_id:
                 # OnyxBot response
                 message_type = MessageType.ASSISTANT
diff --git a/backend/onyx/redis/redis_connector_doc_perm_sync.py b/backend/onyx/redis/redis_connector_doc_perm_sync.py
index 7e5873620..538552ed8 100644
--- a/backend/onyx/redis/redis_connector_doc_perm_sync.py
+++ b/backend/onyx/redis/redis_connector_doc_perm_sync.py
@@ -1,5 +1,6 @@
 import time
 from datetime import datetime
+from typing import Any
 from typing import cast
 from uuid import uuid4
 
@@ -95,7 +96,7 @@ class RedisConnectorPermissionSync:
     @property
     def payload(self) -> RedisConnectorPermissionSyncPayload | None:
         # read related data and evaluate/print task progress
-        fence_bytes = cast(bytes, self.redis.get(self.fence_key))
+        fence_bytes = cast(Any, self.redis.get(self.fence_key))
         if fence_bytes is None:
             return None
 
diff --git a/backend/onyx/redis/redis_connector_ext_group_sync.py b/backend/onyx/redis/redis_connector_ext_group_sync.py
index 2f0783dbc..54654bb71 100644
--- a/backend/onyx/redis/redis_connector_ext_group_sync.py
+++ b/backend/onyx/redis/redis_connector_ext_group_sync.py
@@ -1,4 +1,5 @@
 from datetime import datetime
+from typing import Any
 from typing import cast
 
 import redis
@@ -82,7 +83,7 @@ class RedisConnectorExternalGroupSync:
     @property
     def payload(self) -> RedisConnectorExternalGroupSyncPayload | None:
         # read related data and evaluate/print task progress
-        fence_bytes = cast(bytes, self.redis.get(self.fence_key))
+        fence_bytes = cast(Any, self.redis.get(self.fence_key))
         if fence_bytes is None:
             return None
 
diff --git a/backend/onyx/redis/redis_connector_index.py b/backend/onyx/redis/redis_connector_index.py
index 215468f35..b8d72049a 100644
--- a/backend/onyx/redis/redis_connector_index.py
+++ b/backend/onyx/redis/redis_connector_index.py
@@ -1,4 +1,5 @@
 from datetime import datetime
+from typing import Any
 from typing import cast
 from uuid import uuid4
 
@@ -89,7 +90,7 @@ class RedisConnectorIndex:
     @property
     def payload(self) -> RedisConnectorIndexPayload | None:
         # read related data and evaluate/print task progress
-        fence_bytes = cast(bytes, self.redis.get(self.fence_key))
+        fence_bytes = cast(Any, self.redis.get(self.fence_key))
         if fence_bytes is None:
             return None
 
diff --git a/backend/onyx/server/manage/users.py b/backend/onyx/server/manage/users.py
index 3ef0faf39..7e3dd1f6d 100644
--- a/backend/onyx/server/manage/users.py
+++ b/backend/onyx/server/manage/users.py
@@ -271,6 +271,8 @@ def bulk_invite_users(
 
     tenant_id = CURRENT_TENANT_ID_CONTEXTVAR.get()
     new_invited_emails = []
+    email: str
+
     try:
         for email in emails:
             email_info = validate_email(email)
diff --git a/backend/scripts/chat_feedback_dump.py b/backend/scripts/chat_feedback_dump.py
index 6fe9cbc7d..98d9237f1 100644
--- a/backend/scripts/chat_feedback_dump.py
+++ b/backend/scripts/chat_feedback_dump.py
@@ -198,6 +198,7 @@ def process_all_chat_feedback(onyx_url: str, api_key: str | None) -> None:
         r_sessions = get_chat_sessions(onyx_url, headers, user_id)
         logger.info(f"user={user_id} num_sessions={len(r_sessions.sessions)}")
         for session in r_sessions.sessions:
+            s: ChatSessionSnapshot
             try:
                 s = get_session_history(onyx_url, headers, session.id)
             except requests.exceptions.HTTPError:
diff --git a/backend/tests/daily/connectors/confluence/test_confluence_basic.py b/backend/tests/daily/connectors/confluence/test_confluence_basic.py
index 5462a6617..26cfbcf54 100644
--- a/backend/tests/daily/connectors/confluence/test_confluence_basic.py
+++ b/backend/tests/daily/connectors/confluence/test_confluence_basic.py
@@ -6,6 +6,7 @@ from unittest.mock import patch
 import pytest
 
 from onyx.connectors.confluence.connector import ConfluenceConnector
+from onyx.connectors.models import Document
 
 
 @pytest.fixture
@@ -41,6 +42,10 @@ def test_confluence_connector_basic(
 
     assert len(doc_batch) == 3
 
+    page_within_a_page_doc: Document | None = None
+    page_doc: Document | None = None
+    txt_doc: Document | None = None
+
     for doc in doc_batch:
         if doc.semantic_identifier == "DailyConnectorTestSpace Home":
             page_doc = doc
@@ -49,6 +54,7 @@ def test_confluence_connector_basic(
         elif doc.semantic_identifier == "Page Within A Page":
             page_within_a_page_doc = doc
 
+    assert page_within_a_page_doc is not None
     assert page_within_a_page_doc.semantic_identifier == "Page Within A Page"
     assert page_within_a_page_doc.primary_owners
     assert page_within_a_page_doc.primary_owners[0].email == "hagen@danswer.ai"
@@ -62,6 +68,7 @@ def test_confluence_connector_basic(
         == "https://danswerai.atlassian.net/wiki/spaces/DailyConne/pages/200769540/Page+Within+A+Page"
     )
 
+    assert page_doc is not None
     assert page_doc.semantic_identifier == "DailyConnectorTestSpace Home"
     assert page_doc.metadata["labels"] == ["testlabel"]
     assert page_doc.primary_owners
@@ -75,6 +82,7 @@ def test_confluence_connector_basic(
         == "https://danswerai.atlassian.net/wiki/spaces/DailyConne/overview"
     )
 
+    assert txt_doc is not None
     assert txt_doc.semantic_identifier == "small-file.txt"
     assert len(txt_doc.sections) == 1
     assert txt_doc.sections[0].text == "small"
diff --git a/backend/tests/daily/connectors/gmail/test_gmail_connector.py b/backend/tests/daily/connectors/gmail/test_gmail_connector.py
index d746f86bd..1d58344e6 100644
--- a/backend/tests/daily/connectors/gmail/test_gmail_connector.py
+++ b/backend/tests/daily/connectors/gmail/test_gmail_connector.py
@@ -110,6 +110,8 @@ def test_docs_retrieval(
 
     for doc in retrieved_docs:
         id = doc.id
+        retrieved_primary_owner_emails: set[str | None] = set()
+        retrieved_secondary_owner_emails: set[str | None] = set()
         if doc.primary_owners:
             retrieved_primary_owner_emails = set(
                 [owner.email for owner in doc.primary_owners]
diff --git a/backend/tests/integration/common_utils/managers/user.py b/backend/tests/integration/common_utils/managers/user.py
index 63a6c887d..03539ce56 100644
--- a/backend/tests/integration/common_utils/managers/user.py
+++ b/backend/tests/integration/common_utils/managers/user.py
@@ -165,6 +165,7 @@ class UserManager:
         target_status: bool,
         user_performing_action: DATestUser,
     ) -> DATestUser:
+        url_substring: str
         if target_status is True:
             url_substring = "activate"
         elif target_status is False:

From 42a0f45a96d3d917478937093c6c9ff6a29a1651 Mon Sep 17 00:00:00 2001
From: pablodanswer <pablo@danswer.ai>
Date: Sat, 1 Feb 2025 16:02:49 -0800
Subject: [PATCH 2/3] update

---
 backend/ee/onyx/server/oauth.py                             | 2 ++
 backend/onyx/auth/users.py                                  | 2 ++
 backend/onyx/background/celery/tasks/monitoring/tasks.py    | 2 +-
 backend/onyx/background/celery/tasks/shared/tasks.py        | 1 +
 backend/onyx/background/indexing/run_indexing.py            | 4 +++-
 backend/onyx/connectors/google_utils/resources.py           | 1 +
 backend/onyx/db/document_set.py                             | 1 +
 backend/onyx/onyxbot/slack/listener.py                      | 6 +++++-
 .../google/test_google_drive_permission_sync.py             | 1 +
 9 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/backend/ee/onyx/server/oauth.py b/backend/ee/onyx/server/oauth.py
index 4b79750f0..de239f65a 100644
--- a/backend/ee/onyx/server/oauth.py
+++ b/backend/ee/onyx/server/oauth.py
@@ -286,6 +286,7 @@ def prepare_authorization_request(
     oauth_state = (
         base64.urlsafe_b64encode(oauth_uuid.bytes).rstrip(b"=").decode("utf-8")
     )
+    session: str
 
     if connector == DocumentSource.SLACK:
         oauth_url = SlackOAuth.generate_oauth_url(oauth_state)
@@ -554,6 +555,7 @@ def handle_google_drive_oauth_callback(
         )
 
     session_json = session_json_bytes.decode("utf-8")
+    session: GoogleDriveOAuth.OAuthSession
     try:
         session = GoogleDriveOAuth.parse_session(session_json)
 
diff --git a/backend/onyx/auth/users.py b/backend/onyx/auth/users.py
index da9865550..62c3e9e80 100644
--- a/backend/onyx/auth/users.py
+++ b/backend/onyx/auth/users.py
@@ -1047,6 +1047,8 @@ async def api_key_dep(
     if AUTH_TYPE == AuthType.DISABLED:
         return None
 
+    user: User | None = None
+
     hashed_api_key = get_hashed_api_key_from_request(request)
     if not hashed_api_key:
         raise HTTPException(status_code=401, detail="Missing API key")
diff --git a/backend/onyx/background/celery/tasks/monitoring/tasks.py b/backend/onyx/background/celery/tasks/monitoring/tasks.py
index 244b4abd2..fe25c1d80 100644
--- a/backend/onyx/background/celery/tasks/monitoring/tasks.py
+++ b/backend/onyx/background/celery/tasks/monitoring/tasks.py
@@ -747,7 +747,7 @@ def cloud_check_alembic() -> bool | None:
     revision_counts: dict[str, int] = {}
     out_of_date_tenants: dict[str, str | None] = {}
     top_revision: str = ""
-    tenant_ids: list[str] = []
+    tenant_ids: list[str] | list[None] = []
 
     try:
         # map each tenant_id to its revision
diff --git a/backend/onyx/background/celery/tasks/shared/tasks.py b/backend/onyx/background/celery/tasks/shared/tasks.py
index a1bc12252..84f7e12c6 100644
--- a/backend/onyx/background/celery/tasks/shared/tasks.py
+++ b/backend/onyx/background/celery/tasks/shared/tasks.py
@@ -168,6 +168,7 @@ def document_by_cc_pair_cleanup_task(
         task_logger.info(f"SoftTimeLimitExceeded exception. doc={document_id}")
         return False
     except Exception as ex:
+        e: Exception | None = None
         if isinstance(ex, RetryError):
             task_logger.warning(
                 f"Tenacity retry failed: num_attempts={ex.last_attempt.attempt_number}"
diff --git a/backend/onyx/background/indexing/run_indexing.py b/backend/onyx/background/indexing/run_indexing.py
index fd0ab0b35..72fd8e4fe 100644
--- a/backend/onyx/background/indexing/run_indexing.py
+++ b/backend/onyx/background/indexing/run_indexing.py
@@ -256,6 +256,8 @@ def _run_indexing(
     document_count = 0
     chunk_count = 0
     run_end_dt = None
+    tracer_counter: int
+
     for ind, (window_start, window_end) in enumerate(
         get_time_windows_for_index_attempt(
             last_successful_run=datetime.fromtimestamp(
@@ -266,6 +268,7 @@ def _run_indexing(
     ):
         cc_pair_loop: ConnectorCredentialPair | None = None
         index_attempt_loop: IndexAttempt | None = None
+        tracer_counter = 0
 
         try:
             window_start = max(
@@ -290,7 +293,6 @@ def _run_indexing(
                     tenant_id=tenant_id,
                 )
 
-            tracer_counter = 0
             if INDEXING_TRACER_INTERVAL > 0:
                 tracer.snap()
             for doc_batch in connector_runner.run():
diff --git a/backend/onyx/connectors/google_utils/resources.py b/backend/onyx/connectors/google_utils/resources.py
index c93c58097..48bd981c2 100644
--- a/backend/onyx/connectors/google_utils/resources.py
+++ b/backend/onyx/connectors/google_utils/resources.py
@@ -26,6 +26,7 @@ def _get_google_service(
     creds: ServiceAccountCredentials | OAuthCredentials,
     user_email: str | None = None,
 ) -> GoogleDriveService | GoogleDocsService | AdminService | GmailService:
+    service: Resource
     if isinstance(creds, ServiceAccountCredentials):
         creds = creds.with_subject(user_email)
         service = build(service_name, service_version, credentials=creds)
diff --git a/backend/onyx/db/document_set.py b/backend/onyx/db/document_set.py
index bf9d6d38d..0a5397eb8 100644
--- a/backend/onyx/db/document_set.py
+++ b/backend/onyx/db/document_set.py
@@ -222,6 +222,7 @@ def insert_document_set(
         )
 
     new_document_set_row: DocumentSetDBModel
+    ds_cc_pairs: list[DocumentSet__ConnectorCredentialPair]
     try:
         new_document_set_row = DocumentSetDBModel(
             name=document_set_creation_request.name,
diff --git a/backend/onyx/onyxbot/slack/listener.py b/backend/onyx/onyxbot/slack/listener.py
index e8b26f006..ae6274adb 100644
--- a/backend/onyx/onyxbot/slack/listener.py
+++ b/backend/onyx/onyxbot/slack/listener.py
@@ -5,6 +5,7 @@ import sys
 import threading
 import time
 from collections.abc import Callable
+from contextvars import Token
 from threading import Event
 from types import FrameType
 from typing import Any
@@ -250,6 +251,8 @@ class SlackbotHandler:
         """
         all_tenants = get_all_tenant_ids()
 
+        token: Token[str]
+
         # 1) Try to acquire locks for new tenants
         for tenant_id in all_tenants:
             if (
@@ -771,6 +774,7 @@ def process_message(
         client=client.web_client, channel_id=channel
     )
 
+    token: Token[str] | None = None
     # Set the current tenant ID at the beginning for all DB calls within this thread
     if client.tenant_id:
         logger.info(f"Setting tenant ID to {client.tenant_id}")
@@ -825,7 +829,7 @@ def process_message(
                 if notify_no_answer:
                     apologize_for_fail(details, client)
     finally:
-        if client.tenant_id:
+        if token:
             CURRENT_TENANT_ID_CONTEXTVAR.reset(token)
 
 
diff --git a/backend/tests/integration/connector_job_tests/google/test_google_drive_permission_sync.py b/backend/tests/integration/connector_job_tests/google/test_google_drive_permission_sync.py
index 58f58a6c5..30283097c 100644
--- a/backend/tests/integration/connector_job_tests/google/test_google_drive_permission_sync.py
+++ b/backend/tests/integration/connector_job_tests/google/test_google_drive_permission_sync.py
@@ -54,6 +54,7 @@ def google_drive_test_env_setup() -> (
 
     service_account_key = os.environ["FULL_CONTROL_DRIVE_SERVICE_ACCOUNT"]
     drive_id: str | None = None
+    drive_service: GoogleDriveService | None = None
 
     try:
         credentials = {

From ec0c655c8d994992a5bed2dd13ec3f9fb476edc1 Mon Sep 17 00:00:00 2001
From: pablodanswer <pablo@danswer.ai>
Date: Tue, 4 Feb 2025 12:05:37 -0800
Subject: [PATCH 3/3] misc improvement

---
 backend/onyx/connectors/asana/asana_api.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/backend/onyx/connectors/asana/asana_api.py b/backend/onyx/connectors/asana/asana_api.py
index 2270ee36b..72c5c715a 100755
--- a/backend/onyx/connectors/asana/asana_api.py
+++ b/backend/onyx/connectors/asana/asana_api.py
@@ -99,18 +99,18 @@ class AsanaAPI:
         project = self.project_api.get_project(project_gid, opts={})
         if project["archived"]:
             logger.info(f"Skipping archived project: {project['name']} ({project_gid})")
-            return []
+            yield from []
         if not project["team"] or not project["team"]["gid"]:
             logger.info(
                 f"Skipping project without a team: {project['name']} ({project_gid})"
             )
-            return []
+            yield from []
         if project["privacy_setting"] == "private":
             if self.team_gid and project["team"]["gid"] != self.team_gid:
                 logger.info(
                     f"Skipping private project not in configured team: {project['name']} ({project_gid})"
                 )
-                return []
+                yield from []
             else:
                 logger.info(
                     f"Processing private project in configured team: {project['name']} ({project_gid})"