Add a batch size for cleanup (#10948)

Add pagination to indexing cleanup to deal with large numbers of
documents that need to be deleted.

libs/langchain/langchain/indexes/_api.py

@@ -171,6 +171,7 @@ def index(
     batch_size: int = 100,
     cleanup: Literal["incremental", "full", None] = None,
     source_id_key: Union[str, Callable[[Document], str], None] = None,
+    cleanup_batch_size: int = 1_000,
 ) -> IndexingResult:
     """Index data from the loader into the vector store.
 
@@ -208,6 +209,7 @@ def index(
             - None: Do not delete any documents.
         source_id_key: Optional key that helps identify the original source
             of the document.
+        cleanup_batch_size: Batch size to use when cleaning up documents.
 
     Returns:
         Indexing result which contains information about how many documents
@@ -329,14 +331,14 @@ def index(
                 num_deleted += len(uids_to_delete)
 
     if cleanup == "full":
-        uids_to_delete = record_manager.list_keys(before=index_start_dt)
+        while uids_to_delete := record_manager.list_keys(
-
+            before=index_start_dt, limit=cleanup_batch_size
-        if uids_to_delete:
+        ):
             # First delete from record store.
             vector_store.delete(uids_to_delete)
             # Then delete from record manager.
             record_manager.delete_keys(uids_to_delete)
-            num_deleted = len(uids_to_delete)
+            num_deleted += len(uids_to_delete)
 
     return {
         "num_added": num_added,

libs/langchain/langchain/indexes/_sql_record_manager.py

@@ -259,6 +259,7 @@ class SQLRecordManager(RecordManager):
         before: Optional[float] = None,
         after: Optional[float] = None,
         group_ids: Optional[Sequence[str]] = None,
+        limit: Optional[int] = None,
     ) -> List[str]:
         """List records in the SQLite database based on the provided date range."""
         with self._make_session() as session:
@@ -279,6 +280,9 @@ class SQLRecordManager(RecordManager):
                 query = query.filter(  # type: ignore[attr-defined]
                     UpsertionRecord.group_id.in_(group_ids)
                 )
+
+            if limit:
+                query = query.limit(limit)  # type: ignore[attr-defined]
             records = query.all()  # type: ignore[attr-defined]
         return [r.key for r in records]
 

libs/langchain/langchain/indexes/base.py

@@ -74,6 +74,7 @@ class RecordManager(ABC):
         before: Optional[float] = None,
         after: Optional[float] = None,
         group_ids: Optional[Sequence[str]] = None,
+        limit: Optional[int] = None,
     ) -> List[str]:
         """List records in the database based on the provided filters.
 
@@ -81,6 +82,7 @@ class RecordManager(ABC):
             before: Filter to list records updated before this time.
             after: Filter to list records updated after this time.
             group_ids: Filter to list records with specific group IDs.
+            limit: optional limit on the number of records to return.
 
         Returns:
             A list of keys for the matching records.

libs/langchain/tests/unit_tests/indexes/test_indexing.py

@@ -474,6 +474,43 @@ def test_deduplication(
     }
 
 
+def test_cleanup_with_different_batchsize(
+    record_manager: SQLRecordManager, vector_store: VectorStore
+) -> None:
+    """Check that we can clean up with different batch size."""
+    docs = [
+        Document(
+            page_content="This is a test document.",
+            metadata={"source": str(d)},
+        )
+        for d in range(1000)
+    ]
+
+    assert index(docs, record_manager, vector_store, cleanup="full") == {
+        "num_added": 1000,
+        "num_deleted": 0,
+        "num_skipped": 0,
+        "num_updated": 0,
+    }
+
+    docs = [
+        Document(
+            page_content="Different doc",
+            metadata={"source": str(d)},
+        )
+        for d in range(1001)
+    ]
+
+    assert index(
+        docs, record_manager, vector_store, cleanup="full", cleanup_batch_size=17
+    ) == {
+        "num_added": 1001,
+        "num_deleted": 1000,
+        "num_skipped": 0,
+        "num_updated": 0,
+    }
+
+
 def test_deduplication_v2(
     record_manager: SQLRecordManager, vector_store: VectorStore
 ) -> None:
@@ -497,7 +534,6 @@ def test_deduplication_v2(
         ),
     ]
 
-    # Should result in only a single document being added
     assert index(docs, record_manager, vector_store, cleanup="full") == {
         "num_added": 3,
         "num_deleted": 0,