Bugfix/slack stop 2 (#3916)

* use callback in slim doc functions

* more callbacks

---------

Co-authored-by: Richard Kuo (Danswer) <rkuo@onyx.app>
This commit is contained in:
rkuo-danswer
2025-02-08 15:45:41 -08:00
committed by GitHub
parent a222fae7c8
commit 4c184bb7f0
13 changed files with 81 additions and 15 deletions

View File

@@ -365,7 +365,9 @@ def confluence_doc_sync(
slim_docs = []
logger.debug("Fetching all slim documents from confluence")
for doc_batch in confluence_connector.retrieve_all_slim_documents():
for doc_batch in confluence_connector.retrieve_all_slim_documents(
callback=callback
):
logger.debug(f"Got {len(doc_batch)} slim documents from confluence")
if callback:
if callback.should_stop():

View File

@@ -15,6 +15,7 @@ logger = setup_logger()
def _get_slim_doc_generator(
cc_pair: ConnectorCredentialPair,
gmail_connector: GmailConnector,
callback: IndexingHeartbeatInterface | None = None,
) -> GenerateSlimDocumentOutput:
current_time = datetime.now(timezone.utc)
start_time = (
@@ -24,7 +25,9 @@ def _get_slim_doc_generator(
)
return gmail_connector.retrieve_all_slim_documents(
start=start_time, end=current_time.timestamp()
start=start_time,
end=current_time.timestamp(),
callback=callback,
)
@@ -40,7 +43,9 @@ def gmail_doc_sync(
gmail_connector = GmailConnector(**cc_pair.connector.connector_specific_config)
gmail_connector.load_credentials(cc_pair.credential.credential_json)
slim_doc_generator = _get_slim_doc_generator(cc_pair, gmail_connector)
slim_doc_generator = _get_slim_doc_generator(
cc_pair, gmail_connector, callback=callback
)
document_external_access: list[DocExternalAccess] = []
for slim_doc_batch in slim_doc_generator:

View File

@@ -21,6 +21,7 @@ _PERMISSION_ID_PERMISSION_MAP: dict[str, dict[str, Any]] = {}
def _get_slim_doc_generator(
cc_pair: ConnectorCredentialPair,
google_drive_connector: GoogleDriveConnector,
callback: IndexingHeartbeatInterface | None = None,
) -> GenerateSlimDocumentOutput:
current_time = datetime.now(timezone.utc)
start_time = (
@@ -30,7 +31,9 @@ def _get_slim_doc_generator(
)
return google_drive_connector.retrieve_all_slim_documents(
start=start_time, end=current_time.timestamp()
start=start_time,
end=current_time.timestamp(),
callback=callback,
)

View File

@@ -20,19 +20,11 @@ def _get_slack_document_ids_and_channels(
slack_connector = SlackPollConnector(**cc_pair.connector.connector_specific_config)
slack_connector.load_credentials(cc_pair.credential.credential_json)
slim_doc_generator = slack_connector.retrieve_all_slim_documents()
slim_doc_generator = slack_connector.retrieve_all_slim_documents(callback=callback)
channel_doc_map: dict[str, list[str]] = {}
for doc_metadata_batch in slim_doc_generator:
for doc_metadata in doc_metadata_batch:
if callback:
if callback.should_stop():
raise RuntimeError(
"_get_slack_document_ids_and_channels: Stop signal detected"
)
callback.progress("_get_slack_document_ids_and_channels", 1)
if doc_metadata.perm_sync_data is None:
continue
channel_id = doc_metadata.perm_sync_data["channel_id"]
@@ -40,6 +32,14 @@ def _get_slack_document_ids_and_channels(
channel_doc_map[channel_id] = []
channel_doc_map[channel_id].append(doc_metadata.id)
if callback:
if callback.should_stop():
raise RuntimeError(
"_get_slack_document_ids_and_channels: Stop signal detected"
)
callback.progress("_get_slack_document_ids_and_channels", 1)
return channel_doc_map