DAN-136 Fix incorrect num of github docs (#112)

backend/danswer/datastores/typesense/store.py

@@ -87,18 +87,12 @@ def get_typesense_document_whitelists(
 def delete_typesense_doc_chunks(
     document_id: str, collection_name: str, ts_client: typesense.Client
 ) -> bool:
-    search_parameters = {
-        "q": document_id,
-        "query_by": DOCUMENT_ID,
-    }
+    doc_id_filter = {"filter_by": f"{DOCUMENT_ID}:'{document_id}'"}
 
-    # TODO consider race conditions if running multiple processes/threads
-    hits = ts_client.collections[collection_name].documents.search(search_parameters)
-    [
-        ts_client.collections[collection_name].documents[hit["document"]["id"]].delete()
-        for hit in hits["hits"]
-    ]
-    return True if hits else False
+    # Typesense doesn't seem to prioritize individual deletions, problem not seen with this approach
+    # Point to consider if we see instances of number of Typesense and Qdrant docs not matching
+    del_result = ts_client.collections[collection_name].documents.delete(doc_id_filter)
+    return del_result["num_deleted"] != 0
 
 
 def index_typesense_chunks(

backend/danswer/server/manage.py

@@ -1,11 +1,9 @@
-from collections import defaultdict
 from typing import cast
 
 from danswer.auth.schemas import UserRole
 from danswer.auth.users import current_admin_user
 from danswer.auth.users import current_user
 from danswer.configs.app_configs import MASK_CREDENTIAL_PREFIX
-from danswer.configs.constants import DocumentSource
 from danswer.configs.constants import OPENAI_API_KEY_STORAGE_KEY
 from danswer.connectors.file.utils import write_temp_files
 from danswer.connectors.google_drive.connector_auth import DB_CREDENTIALS_DICT_KEY
@@ -21,8 +19,6 @@ from danswer.db.connector import create_connector
 from danswer.db.connector import delete_connector
 from danswer.db.connector import fetch_connector_by_id
 from danswer.db.connector import fetch_connectors
-from danswer.db.connector import fetch_latest_index_attempt_by_connector
-from danswer.db.connector import fetch_latest_index_attempts_by_status
 from danswer.db.connector import get_connector_credential_ids
 from danswer.db.connector import update_connector
 from danswer.db.connector_credential_pair import add_credential_to_connector
@@ -37,9 +33,6 @@ from danswer.db.credentials import update_credential
 from danswer.db.engine import get_session
 from danswer.db.engine import get_sqlalchemy_async_engine
 from danswer.db.index_attempt import create_index_attempt
-from danswer.db.models import Connector
-from danswer.db.models import IndexAttempt
-from danswer.db.models import IndexingStatus
 from danswer.db.models import User
 from danswer.direct_qa.key_validation import check_openai_api_key_is_valid
 from danswer.direct_qa.question_answer import get_openai_api_key
@@ -56,7 +49,6 @@ from danswer.server.models import CredentialSnapshot
 from danswer.server.models import FileUploadResponse
 from danswer.server.models import GDriveCallback
 from danswer.server.models import GoogleAppCredentials
-from danswer.server.models import IndexAttemptSnapshot
 from danswer.server.models import ObjectCreationIdResponse
 from danswer.server.models import RunConnectorRequest
 from danswer.server.models import StatusResponse

backend/danswer/server/models.py

@@ -177,15 +177,5 @@ class CredentialSnapshot(CredentialBase):
     time_updated: datetime
 
 
-class IndexAttemptSnapshot(BaseModel):
-    source: DocumentSource
-    input_type: InputType
-    status: IndexingStatus
-    connector_specific_config: dict[str, Any]
-    docs_indexed: int
-    time_created: datetime
-    time_updated: datetime
-
-
 class ApiKey(BaseModel):
     api_key: str

backend/danswer/utils/indexing_pipeline.py

@@ -38,11 +38,11 @@ def _indexing_pipeline(
     net_doc_count_keyword = keyword_index.index(chunks, user_id)
     chunks_with_embeddings = embedder.embed(chunks)
     net_doc_count_vector = vector_index.index(chunks_with_embeddings, user_id)
-    if net_doc_count_vector != net_doc_count_vector:
-        logger.exception(
-            "Document count change from keyword/vector indices don't align"
-        )
-    return max(net_doc_count_keyword, net_doc_count_vector)
+    if net_doc_count_vector != net_doc_count_keyword:
+        logger.warning("Document count change from keyword/vector indices don't align")
+    net_new_docs = max(net_doc_count_keyword, net_doc_count_vector)
+    logger.info(f"Indexed {net_new_docs} new documents")
+    return net_new_docs
 
 
 def build_indexing_pipeline(