torchcell.knowledge_graphs.incremental_import.query_existing_edges

query_existing_edges(group: CsvGroup, rows: list[tuple[str, str]], session: Any, batch_size: int = 5000) → set[tuple[str, str]]

Which (start, end) pairs already carry a relationship of this group’s type.

Uses the Entity.id uniqueness index on both endpoints, so a batch of a few thousand lookups takes well under a second on the served store.