資料操作語言（DML）陳述式

BigQuery 入門

Matt Forrest

Field CTO

BigQuery 中的資料操作概觀

INSERT：新增資料列。
UPDATE：修改列中的既有值。
DELETE：從資料表移除不需要的資料。
MERGE：將 INSERT、UPDATE、DELETE 合併成一個陳述式。
CREATE TABLE AS：由查詢結果建立新資料表。

注意事項與效能

能合併的 DML 盡量一起執行，不要逐一執行。
執行 UPDATE 時必須使用 WHERE 條件。
可考慮使用資料表分割與叢集。

¹ https://cloud.google.com/bigquery/docs/reference/standard-sql/data-manipulation-language

INSERT

新增資料到資料表

-- Define the columns in the parentheses 
INSERT INTO customers (customer_id, name, email)

-- Each value is a row to be inserted
VALUES (1, "John Doe", "[email protected]"),
(2, "Jane Doe", "[email protected]"),
(3, "Alice Smith", "[email protected]");

UPDATE

依條件變更資料

UPDATE customers
-- Set one column for each SET statement
SET email = "[email protected]"
-- Make sure to include where otherwise all
-- rows will be updated
WHERE customer_id = 1;

UPDATE 可搭配子查詢或聯結

UPDATE customers c
SET c.email = e.email
FROM emails e
WHERE c.customer_id = 1;

DELETE

DELETE 會「永久」刪除紀錄，且「無法還原」

DELETE FROM customers

-- Include WHERE to ensure only specific rows are deleted
WHERE customer_id = 3;

DELETE FROM customers c
JOIN emails e USING (customer_id)
WHERE email = '[email protected]'

MERGE

在單一作業中結合 INSERT、UPDATE、DELETE

-- Sets the customers table as the target
MERGE customers AS target

-- The source is set to new_customers
USING new_customers AS source

-- Matching condition
ON target.customer_id = source.customer_id

-- If the emails do not match, update the email
WHEN MATCHED AND target.email != source.email THEN
  UPDATE SET email = source.email

-- If the match is not met, insert the record
WHEN NOT MATCHED THEN
  INSERT (customer_id, name, email) VALUES 
  (source.customer_id, source.name, source.email);

CREATE TABLE

由查詢建立新資料表

CREATE TABLE active_customers AS
SELECT customer_id, name, email 
FROM customers
WHERE last_active_date > DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY);

一起來練習吧！

BigQuery 入門