我正在尝试建立一个变化缓慢的维表,它跟踪所有记录的历史记录。该表的架构如下:
CREATE TABLE test.dim
(id text,
column1 text,
column2 text,
begin_date timestamp without time zone,
is_current boolean,
end_date timestamp without time zone)
我定义了一个触发函数,并在每次插入动作之前将其触发:
CREATE OR REPLACE FUNCTION test.slow_change_func()
RETURNS trigger AS
$BODY$
DECLARE
BEGIN
IF ( NOT EXISTS ( SELECT 1 FROM yang_test.dim
WHERE id= NEW.id
AND(column1 = NEW.column1 OR (column1 is null AND NEW.column1 is null))
AND (column2 = NEW.column2 OR (column2 is null AND NEW.column2 is null))
AND is_current
)
)
THEN UPDATE yang_test.dim
SET (end_date, is_current) = (now(), FALSE)
WHERE id = NEW.id
AND is_current;
INSERT INTO test.dim (id, column1, column2, begin_date, is_current, end_date)
VALUES ( NEW.id, NEW.column1, NEW.column2, now(), TRUE, 'infinity'::timestamp );
END IF;
RETURN NULL;
END;
$BODY$
LANGUAGE plpgsql VOLATILE
COST 100;
CREATE TRIGGER slow_change_trigger
BEFORE INSERT
ON test.dim
FOR EACH ROW
EXECUTE PROCEDURE test.slow_change_func();
当我尝试测试时,
INSERT INTO test.dim (id, column1, column2, begin_date, is_current, end_date)
VALUES ( 1, 'hello', 'world', now(), TRUE, 'infinity'::timestamp )
它将引发错误:超出堆栈深度限制。该函数似乎正在运行一个循环。有什么建议吗?
答案 0 :(得分:0)
我想我已经弄清楚了,这符合我的要求:
CREATE OR REPLACE FUNCTION yang_test.slow_change_func()
RETURNS trigger AS
$BODY$
DECLARE
BEGIN
IF ( NOT EXISTS ( SELECT 1 FROM yang_test.dim
WHERE id= NEW.id
AND(column1 = NEW.column1 OR (column1 is null AND NEW.column1 is null))
AND (column2 = NEW.column2 OR (column2 is null AND NEW.column2 is null))
AND is_current
)
)
THEN UPDATE yang_test.dim
SET (end_date, is_current) = (now(), FALSE)
WHERE id = NEW.id
AND is_current;
ELSE RETURN null;
END IF;
RETURN NEW;
END;
$BODY$
LANGUAGE plpgsql VOLATILE
COST 100;