Call

async AsyncCogniteClient.raw.rows.__call__(
db_name: str,
table_name: str,
chunk_size: int | None = None,
partitions: int | None = None,
limit: int | None = None,
min_last_updated_time: int | None = None,
max_last_updated_time: int | None = None,
columns: list[str] | None = None,
) → AsyncIterator[Row] | AsyncIterator[RowList]

Iterate over rows.

Fetches rows as they are iterated over, so you keep a limited number of rows in memory.

Note

When iterating using partitions > 1, the memory usage is bounded at 2 x partitions x chunk_size. This is implemented by halting retrieval speed when the callers code can’t keep up.

Parameters:

  • db_name (str) – Name of the database.

  • table_name (str) – Name of the table.

  • chunk_size (int | None) – Number of rows to return in each chunk (may be lower). Defaults to yielding one row at a time. Note: When used together with ‘partitions’ the default is 10000 (matching the API limit) and there’s an implicit minimum of 1000 rows.

  • partitions (int | None) – Retrieve rows in parallel using this number of workers. Defaults to not use concurrency. The setting is capped at global_config.concurrency_settings.raw.read and _can_ be used with a finite limit. To prevent unexpected problems and maximize read throughput, check out concurrency limits in the API documentation.

  • limit (int | None) – Maximum number of rows to return. Can be used with partitions. Defaults to returning all items.

  • min_last_updated_time (int | None) – Rows must have been last updated after this time (exclusive). Milliseconds since epoch.

  • max_last_updated_time (int | None) – Rows must have been last updated before this time (inclusive). Milliseconds since epoch.

  • columns (list[str] | None) – List of column keys. Set to None to retrieving all, use empty list, [], to retrieve only row keys.

Yields:

Row | RowList – An iterator yielding the requested row or rows.