mirror of
https://github.com/danswer-ai/danswer.git
synced 2025-09-19 12:03:54 +02:00
Heavy task improvements, logging, and validation (#4058)
This commit is contained in:
@@ -4,6 +4,7 @@ from sqlalchemy.orm import Session
|
||||
from onyx.configs.constants import DocumentSource
|
||||
from onyx.db.connector_credential_pair import get_connector_credential_pair
|
||||
from onyx.db.enums import AccessType
|
||||
from onyx.db.enums import ConnectorCredentialPairStatus
|
||||
from onyx.db.models import Connector
|
||||
from onyx.db.models import ConnectorCredentialPair
|
||||
from onyx.db.models import UserGroup__ConnectorCredentialPair
|
||||
@@ -35,10 +36,11 @@ def _delete_connector_credential_pair_user_groups_relationship__no_commit(
|
||||
def get_cc_pairs_by_source(
|
||||
db_session: Session,
|
||||
source_type: DocumentSource,
|
||||
only_sync: bool,
|
||||
access_type: AccessType | None = None,
|
||||
status: ConnectorCredentialPairStatus | None = None,
|
||||
) -> list[ConnectorCredentialPair]:
|
||||
"""
|
||||
Get all cc_pairs for a given source type (and optionally only sync)
|
||||
Get all cc_pairs for a given source type with optional filtering by access_type and status
|
||||
result is sorted by cc_pair id
|
||||
"""
|
||||
query = (
|
||||
@@ -48,8 +50,11 @@ def get_cc_pairs_by_source(
|
||||
.order_by(ConnectorCredentialPair.id)
|
||||
)
|
||||
|
||||
if only_sync:
|
||||
query = query.filter(ConnectorCredentialPair.access_type == AccessType.SYNC)
|
||||
if access_type is not None:
|
||||
query = query.filter(ConnectorCredentialPair.access_type == access_type)
|
||||
|
||||
if status is not None:
|
||||
query = query.filter(ConnectorCredentialPair.status == status)
|
||||
|
||||
cc_pairs = query.all()
|
||||
return cc_pairs
|
||||
|
@@ -62,12 +62,14 @@ def _fetch_permissions_for_permission_ids(
|
||||
user_email=(owner_email or google_drive_connector.primary_admin_email),
|
||||
)
|
||||
|
||||
# We continue on 404 or 403 because the document may not exist or the user may not have access to it
|
||||
fetched_permissions = execute_paginated_retrieval(
|
||||
retrieval_function=drive_service.permissions().list,
|
||||
list_key="permissions",
|
||||
fileId=doc_id,
|
||||
fields="permissions(id, emailAddress, type, domain)",
|
||||
supportsAllDrives=True,
|
||||
continue_on_404_or_403=True,
|
||||
)
|
||||
|
||||
permissions_for_doc_id = []
|
||||
@@ -104,7 +106,13 @@ def _get_permissions_from_slim_doc(
|
||||
user_emails: set[str] = set()
|
||||
group_emails: set[str] = set()
|
||||
public = False
|
||||
skipped_permissions = 0
|
||||
|
||||
for permission in permissions_list:
|
||||
if not permission:
|
||||
skipped_permissions += 1
|
||||
continue
|
||||
|
||||
permission_type = permission["type"]
|
||||
if permission_type == "user":
|
||||
user_emails.add(permission["emailAddress"])
|
||||
@@ -121,6 +129,11 @@ def _get_permissions_from_slim_doc(
|
||||
elif permission_type == "anyone":
|
||||
public = True
|
||||
|
||||
if skipped_permissions > 0:
|
||||
logger.warning(
|
||||
f"Skipped {skipped_permissions} permissions of {len(permissions_list)} for document {slim_doc.id}"
|
||||
)
|
||||
|
||||
drive_id = permission_info.get("drive_id")
|
||||
group_ids = group_emails | ({drive_id} if drive_id is not None else set())
|
||||
|
||||
|
Reference in New Issue
Block a user