我创建了一个触发器:
CREATE OR REPLACE FUNCTION public.delete_user()
RETURNS TRIGGER AS $func$
BEGIN
EXECUTE format('UPDATE public.user SET %I = %L WHERE state = %L',
"state", "active", "deleted");
END;
$func$ LANGUAGE plpgsql;
CREATE TRIGGER deleted_user
AFTER UPDATE ON public.user
FOR EACH ROW
WHEN (OLD.state IS DISTINCT FROM NEW.state)
EXECUTE PROCEDURE delete_user();
比跑步后:
update public.user set state = 'active' where fullname = '1234'
我收到错误消息:
> ERROR: column "state" does not exist
> LINE 1: ...UPDATE public.user SET %I = %L WHERE state = %L', "state", "...
> ^
> QUERY: SELECT format('UPDATE public.user SET %I = %L WHERE state = %L', "state", "active", "deleted")
> CONTEXT: PL/pgSQL function delete_user() line 3 at EXECUTE SQL state: 42703
这是在用户表中定义状态的方式:
state text COLLATE pg_catalog."default" NOT NULL DEFAULT 'active'::text,
我是Postgres的新手,所以我想这是语法错误?
答案 0 :(得分:0)
这将起作用:
CREATE OR REPLACE FUNCTION public.delete_user()
RETURNS TRIGGER AS
$func$
BEGIN
EXECUTE format('UPDATE public.user SET %I = %L WHERE state = %L',
'state', 'active', 'deleted');
RETURN NULL; -- or OLD (for AFTER trigger)
END
$func$ LANGUAGE plpgsql;
字符串文字的单引号而不是双引号(用于标识符)。为此,您不需要动态SQL,并且可以简化:
CREATE OR REPLACE FUNCTION public.delete_user()
RETURNS TRIGGER AS
$func$
BEGIN
UPDATE public.user SET state = 'active' WHERE state = 'deleted';
RETURN NULL;
END
$func$ LANGUAGE plpgsql;
虽然语法现在可以使用,但它会导致无限循环,并且按原样没有意义。我不确定你要去哪里。
相关:
在旁边:public.user
起作用,因为架构限定消除了歧义,但是裸表名称user
始终需要双引号:"user"
。 从不 使用保留字如user
作为标识符。