Oracle PL/SQL：如何删除一行，然后使用管道将其输出？

我有一个包裹：

CREATE OR REPLACE PACKAGE pkg_location IS
    TYPE dep_row IS RECORD (id departments.department_id%TYPE, name departments.department_name%TYPE);
    TYPE dep_table IS TABLE OF dep_row;
    FUNCTION drop_departments (v_city_title IN VARCHAR2) RETURN dep_table PIPELINED;
END pkg_location; 

我需要实现

drop\u departments

功能，该功能通过

city

从

departments

表中删除行，并使用管道输出删除的部门数据

但我真的不知道从哪里开始。你能帮帮我吗

表结构：

DEPARTMENTS: DEPARTMENT_ID (NUMBER) | DEPARTMENT_NAME (VARCHAR 40) | LOCATION_ID (NUMBER)
LOCATIONS:   LOCATION_ID (NUMBER)   | CITY (VARCHAR 40)

您需要3个步骤：

• 将要删除的所有记录保存到与函数类型相同的本地表中
• 删除记录
• 返回您保存的集合（不必使用管道连接）

管道函数存在问题-在选择过程中，无法从表中选择*（my_pipe_函数）并在函数内部执行删除（DML）操作

为了避免删除内部函数的问题，您需要更改逻辑，不使用管道，或者需要在

自治事务中删除记录

参见下面的3个示例

insert into departments(id , department_id , department_name , city_title ) values(1, 1, '1', 'City-1');
insert into departments(id , department_id , department_name , city_title )  values(2, 2, '2', 'City-1');
insert into departments(id , department_id , department_name , city_title )  values(3, 2, '3', 'City-2');

带有

管道的示例1

    CREATE OR REPLACE PACKAGE BODY pkg_location IS
    FUNCTION drop_departments (v_city_title IN VARCHAR2) RETURN dep_table PIPELINED
    is
      tblResult dep_table;
    begin

      --delete the records and save deleted records
      delete from departments 
              where city_title = v_city_title
      RETURNING department_id, department_name 
      BULK COLLECT INTO tblResult;

      --return list of deleted records        
      for i in tblResult.FIRST .. tblResult.LAST loop
        pipe row (tblResult(i));
      end loop;
    end;
END pkg_location;

但是你不能像这样使用smth

从表中选择*（包装位置。放置部门（'City-1'））

样本2-我已移除管道

CREATE OR REPLACE PACKAGE BODY pkg_location IS
    FUNCTION drop_departments (v_city_title IN VARCHAR2) RETURN dep_table 
    is
      tblResult dep_table;
    begin

      --delete the records and save deleted records
      delete from departments 
              where city_title = v_city_title
        RETURNING department_id, department_name 
        BULK COLLECT INTO tblResult;

      --return list of deleted records        
      return tblResult;
  end;      
END pkg_location;

现在您可以这样使用该函数：

declare
  tbl pkg_location.dep_table;
begin
  tbl := pkg_location.drop_departments('City-1');

  for i in tbl.first .. tbl.last loop
    dbms_output.put_line(tbl(i).name);
  end loop;
end;

输出将为1和2

示例3-使用

自主交易

CREATE OR REPLACE PACKAGE BODY pkg_location IS

    FUNCTION drop_departments (v_city_title IN VARCHAR2) RETURN dep_table PIPELINED    
    is
    --it let you use the function in select * from table(f)
    PRAGMA AUTONOMOUS_TRANSACTION;
      tblResult dep_table;
    begin
      --delete the records and save deleted records
      delete from departments 
              where city_title = v_city_title
       returning department_id, department_name 
      bulk collect into tblResult;

      --you must do the commit before pipe row ()
      commit;

      --return list of deleted records        
      for i in tblResult.FIRST .. tblResult.LAST loop
        pipe row (tblResult(i));
      end loop;
    end;
END pkg_location;

现在你可以使用

select * from table(pkg_location.drop_departments('City-1'))

结果将是：

1   1   1
2   2   2

---

删除之前的选择是不必要的，并且可能存在并发问题。您可以使用

DELETE departments WHERE city\u title=v\u city\u title RETURNING department\u id，department\u name BULK COLLECT to tblResult将删除的记录输出到集合中流水线版本没有问题，它也可以使用自治事务。您只需要在管道行之前发出commit，因为这会生成一个值，并将函数保留到需要下一条记录为止，并且必须在保留函数作用域之前提交/回滚自治事务。但在这种情况下，返回集合的非管道方法是最好的。这是最少的代码量，您可以为所有已删除的记录分配内存，因此迭代并生成固定集合是没有意义的。