如何用 Oracle SQL 函数或存储过程替换嵌套表

Question

在Oracle中，如果我有以下两个表

CREATE TABLE Test_Persons_A (
    PersonID int,
    LastName varchar(255),
    FirstName varchar(255)
);

INSERT INTO Test_Persons_A
    (PersonID,LastName,FirstName)
    values(11,'LN_A1','FN_A1');

INSERT INTO Test_Persons_A
    (PersonID,LastName,FirstName)
    values(12,'LN_A2','FN_A2');

CREATE TABLE Test_Persons_B (
    PersonID int,
    LastName varchar(255),
    FirstName varchar(255)
);

INSERT INTO Test_Persons_B
    (PersonID,LastName,FirstName)
    values(21,'LN_B1','FN_B1');

INSERT INTO Test_Persons_B
    (PersonID,LastName,FirstName)
    values(22,'LN_B2','FN_B2');

commit;

我可以对联合进行如下查询：

with
UNIONED as
(    
    --block A to replaced by calling a function or stored proc---
    select PersonID as PID, LastName as LN, FirstName as FN
        from Test_Persons_A tp
        where tp.LASTNAME like '%1%'

    union

    --block B to replaced by calling a function or stored proc---
    select PersonID as PID, LastName as LN, FirstName as FN
        from Test_Persons_B tp
        where tp.LASTNAME like '%2%'
)

select * from UNIONED; 

我的问题是如何实现以下目标：

with
UNIONED as
(    
    --block A replaced---
    myFuncOrStoredProc('Test_Persons_A', '%1%');

    union

    --block B replaced---
    myFuncOrStoredProc('Test_Persons_B', '%2%');
)

select * from UNIONED; 

有很多关于“如何从函数或存储过程返回表（或多行数据）”的帖子。但是由于答案众多，我感到更加困惑。

• 流水线表函数 (https://stackoverflow.com/a/2830722/2041023)
• 批量收集 (https://stackoverflow.com/a/2832735/2041023)
• 光标 (https://stackoverflow.com/a/46998148/2041023)

对于这样一个小而简单（我认为很常见）的场景和目的，什么是最干净的方法？（特别是，我更喜欢以避免必须为返回的表指定每一列的类型。）

Answer 1

Oracle 设置：

CREATE TABLE Test_Persons_A ( PersonID, LastName, FirstName ) AS
  SELECT 11, 'LN_A1', 'FN_A1' FROM DUAL UNION ALL
  SELECT 12, 'LN_A2', 'FN_A2' FROM DUAL;

CREATE TABLE Test_Persons_B ( PersonID, LastName, FirstName ) AS
 SELECT 21, 'LN_B1', 'FN_B1' FROM DUAL UNION ALL
 SELECT 22, 'LN_B2', 'FN_B2' FROM DUAL;

CREATE TYPE test_obj IS OBJECT(
  PersonID INT,
  FirstName VARCHAR2(255),
  LastName VARCHAR2(255)
);
/

CREATE TYPE test_tbl IS TABLE OF test_obj;
/

CREATE OR REPLACE FUNCTION test_fn(
  table_name VARCHAR2,
  lastNameFilter VARCHAR2
) RETURN test_tbl
IS
  v_tbl test_tbl;
BEGIN
  IF UPPER( table_name ) = 'TEST_PERSONS_A' THEN
    SELECT TEST_OBJ( PersonID, LastName, FirstName )
    BULK COLLECT INTO v_tbl
    FROM   test_persons_a
    WHERE  LastName LIKE lastNameFilter;
  ELSIF UPPER( table_name ) = 'TEST_PERSONS_B' THEN
    SELECT TEST_OBJ( PersonID, LastName, FirstName )
    BULK COLLECT INTO v_tbl
    FROM   test_persons_b
    WHERE  LastName LIKE lastNameFilter;
  END IF;
  RETURN v_tbl;
END;
/

查询 1 在单个表集合表达式中使用 MULTSET UNION [ALL|DISTINCT]：

SELECT *
FROM   TABLE(
         test_fn( 'TEST_PERSONS_A', '%1%' )
         MULTISET UNION ALL
         test_fn( 'TEST_PERSONS_B', '%2%' )
       );

查询 2 使用UNION [ALL] 和多个表集合表达式：

SELECT * FROM TABLE( test_fn( 'TEST_PERSONS_A', '%1%' ) )
UNION ALL
SELECT * FROM TABLE( test_fn( 'TEST_PERSONS_B', '%2%' ) );

两个查询的

输出：

  PERSONID FIRSTNAME LASTNAME
---------- --------- --------
        11 LN_A1     FN_A1
        22 LN_B2     FN_B2