Apache Airflow を ClickHouse に接続

pip install apache-airflow-providers-clickhousedb

pip install 'apache-airflow-providers-common-sql[pandas,polars]'

export AIRFLOW_CONN_CLICKHOUSE_DEFAULT='clickhouse://default:password@localhost:8123/my_database'

export AIRFLOW_CONN_CLICKHOUSE_DEFAULT='{
    "conn_type": "clickhouse",
    "host": "abc123.clickhouse.cloud",
    "port": 8443,
    "login": "default",
    "password": "secret",
    "schema": "my_database",
    "extra": {
        "secure": true,
        "session_settings": {
            "max_execution_time": 300,
            "max_memory_usage": 10000000000
        }
    }
}'

from datetime import datetime

from airflow import DAG
from airflow.providers.common.sql.hooks.sql import fetch_all_handler
from airflow.providers.common.sql.operators.sql import SQLExecuteQueryOperator

CLICKHOUSE_CONN_ID = "clickhouse_default"
CLICKHOUSE_TABLE = "airflow_example"

with DAG(
    dag_id="example_clickhouse",
    start_date=datetime(2021, 1, 1),
    default_args={"conn_id": CLICKHOUSE_CONN_ID},
    schedule="@once",
    catchup=False,
) as dag:
    create_table = SQLExecuteQueryOperator(
        task_id="create_table",
        sql=f"""
            CREATE TABLE IF NOT EXISTS {CLICKHOUSE_TABLE} (
                id   UInt32,
                name String,
                ts   DateTime DEFAULT now()
            ) ENGINE = MergeTree()
            ORDER BY id
        """,
    )

    insert_rows = SQLExecuteQueryOperator(
        task_id="insert_rows",
        sql=f"""
            INSERT INTO {CLICKHOUSE_TABLE} (id, name) VALUES
                (1, 'Alice'),
                (2, 'Bob'),
                (3, 'Charlie')
        """,
    )

    read_rows = SQLExecuteQueryOperator(
        task_id="read_rows",
        sql=f"SELECT id, name FROM {CLICKHOUSE_TABLE} ORDER BY id",
        handler=fetch_all_handler,
    )

    drop_table = SQLExecuteQueryOperator(
        task_id="drop_table",
        sql=f"DROP TABLE IF EXISTS {CLICKHOUSE_TABLE}",
    )

    create_table >> insert_rows >> read_rows >> drop_table

read_rows = SQLExecuteQueryOperator(
    task_id="read_rows",
    conn_id=CLICKHOUSE_CONN_ID,
    sql="SELECT count() FROM events",
    hook_params={"database": "analytics"},
)

from airflow.providers.clickhousedb.hooks.clickhouse import ClickHouseHook

hook = ClickHouseHook(clickhouse_conn_id="clickhouse_default")

hook.bulk_insert_rows(
    table="events",
    rows=[("user1", "click"), ("user2", "view")],
    column_names=["user_id", "action"],
    batch_size=1000,
)

client = hook.get_client()
total = client.query("SELECT count() FROM events").result_rows[0][0]

hook = ClickHouseHook(
    clickhouse_conn_id="clickhouse_default",
    session_settings={"max_execution_time": 60, "max_threads": 4},
)