Using AdbcHook¶

Use AdbcHook to interact with any database that has an ADBC driver. The hook extends DbApiHook, so it works transparently with SQLExecuteQueryOperator and the other operators from the common.sql provider — just pass conn_id pointing at an adbc connection.

For direct hook usage (e.g. inside a @dag.task function), the example below demonstrates creating a table, bulk-inserting rows via Arrow-native transfer, querying the results, and cleaning up — all against a SQLite ADBC connection.

tests/system/apache/arrow/example_adbc.py[source]

    @dag.task
    def create_table():
        from airflow.providers.apache.arrow.hooks.adbc import AdbcHook

        hook = AdbcHook(adbc_conn_id=CONN_ID)
        hook.run("CREATE TABLE IF NOT EXISTS users (  id   INTEGER PRIMARY KEY,  name TEXT NOT NULL)")

    @dag.task
    def insert_rows():
        from airflow.providers.apache.arrow.hooks.adbc import AdbcHook

        hook = AdbcHook(adbc_conn_id=CONN_ID)
        rows = [(1, "Alice"), (2, "Bob"), (3, "Carol")]
        hook.insert_rows(table="users", rows=rows, target_fields=["id", "name"])

    @dag.task
    def query_rows():
        from airflow.providers.apache.arrow.hooks.adbc import AdbcHook

        hook = AdbcHook(adbc_conn_id=CONN_ID)
        records = hook.get_records("SELECT id, name FROM users ORDER BY id")
        assert len(records) == 3, f"Expected 3 rows, got {len(records)}"

    @dag.task
    def drop_table():
        from airflow.providers.apache.arrow.hooks.adbc import AdbcHook

        hook = AdbcHook(adbc_conn_id=CONN_ID)
        hook.run("DROP TABLE IF EXISTS users")

    create_table() >> insert_rows() >> query_rows() >> drop_table()

See Connection types for how to configure the adbc connection.