Postgresql 如何在Postgres中实现清除操作系统软删除记录？

在我的Postgres 9.4数据库中，我有以下触发器/函数，它实现了“软删除”功能：

ALTER TABLE my_schema.my_table
ADD COLUMN delete_ind integer


CREATE OR REPLACE FUNCTION trigger_mytable_soft_delete()  
  RETURNS trigger AS $$
    DECLARE
      command text := ' SET delete_ind = 1 WHERE uuid_col = $1';
    BEGIN
      EXECUTE 'UPDATE "my_schema"."my_table"' || TG_TABLE_NAME || command USING OLD.uuid_col;
      RETURN NULL;
    END;
  $$ LANGUAGE plpgsql;


CREATE TRIGGER my_table_soft_delete_trigger  
  BEFORE DELETE ON "my_schema"."my_table"
  FOR EACH ROW EXECUTE PROCEDURE trigger_mytable_soft_delete();

上面的代码为我提供了“软删除”功能，但它也阻止我实际删除/清除那些已标记为已删除的行

所需的新行为是使用此delete函数检查delete_ind字段的值，如果该字段已设置为1，则实际永久清除该行

---

正确的条件语法是什么，它可以设置delete\u ind.的值，或者根据delete\u ind列的当前值实际删除相关行？

只需对函数进行相对较小的修改即可：

CREATE OR REPLACE FUNCTION trigger_mytable_soft_delete()  
RETURNS trigger AS
$$  
BEGIN
    if OLD.delete_ind = 1 then
        /* Delete proceeds, you just needs to *do nothing* 
           except for returning the OLD row as it were */
        RETURN OLD ;  
    else
        /* instead of deleting, set a flag */
        UPDATE my_schema.my_table 
           SET deleted_ind = 1
         WHERE uuid_col = old.uuid_col ;

        /* This will skip the process of this row.
           It will also avoid subsequent triggers to be fired, and the row will
           not be counted on the rows-affected count. If more triggers need
           to be processed, make sure this is the last in the chain.
        */
        RETURN NULL ;
    end if ;
END;
$$ 
LANGUAGE plpgsql;

（如果您的函数是逐个表使用的，您可以对其进行硬编码，而不需要动态SQL）

---

---

旁注：如果列

delete\u ind

仅用作标志，则最好将其声明为

布尔非空

，而不是

整数

，只需对函数进行相对较小的修改即可：

CREATE OR REPLACE FUNCTION trigger_mytable_soft_delete()  
RETURNS trigger AS
$$  
BEGIN
    if OLD.delete_ind = 1 then
        /* Delete proceeds, you just needs to *do nothing* 
           except for returning the OLD row as it were */
        RETURN OLD ;  
    else
        /* instead of deleting, set a flag */
        UPDATE my_schema.my_table 
           SET deleted_ind = 1
         WHERE uuid_col = old.uuid_col ;

        /* This will skip the process of this row.
           It will also avoid subsequent triggers to be fired, and the row will
           not be counted on the rows-affected count. If more triggers need
           to be processed, make sure this is the last in the chain.
        */
        RETURN NULL ;
    end if ;
END;
$$ 
LANGUAGE plpgsql;

（如果您的函数是逐个表使用的，您可以对其进行硬编码，而不需要动态SQL）

---

旁注：如果列

delete\u ind

仅用作标志，则最好将其声明为

布尔非空

，而不是

整数