torchcell.knowledge_graphs.incremental_import.incremental_import_call

incremental_import_call(groups: list[CsvGroup], headers: dict[str, Path], database: str, *, bin_prefix: str = '/var/lib/neo4j/bin/', report_file: str | None = None, threads: int = 8, max_off_heap_memory: str = '16G', bad_tolerance: int = 1000000000) → str

The neo4j-admin database import incremental command for this increment.

Paths are used verbatim: the output directory is generated inside the container at the same path the import runs from (/var/lib/neo4j/biocypher-out/<ts>). The uniqueness constraints are NOT passed here (--schema is unsupported for incremental import); they are created on the online database beforehand. Threads and off-heap memory are bounded because the import runs INSIDE the serving container next to the (stopped-database, still running) Neo4j server.