Question

我正在尝试设置一个可能包含1000条记录的视图。某些字段需要子查询，这些子查询返回完全相同的数据。我想知道我是否可以在视图中查询一次，而不是每次都运行它。

以下是一些示例表/数据：

DROP VIEW MYVIEW;
DROP TABLE MYTABLE;
DROP TABLE MYTABLE_PARENT;

-- TABLES FOR VIEW
CREATE TABLE MYTABLE_PARENT ( PRIMARY_KEY NUMBER PRIMARY KEY );
CREATE TABLE MYTABLE ( PRIMARY_KEY NUMBER, MYVAL VARCHAR(255), PARENT_PRIMARY_KEY NUMBER);
INSERT INTO MYTABLE_PARENT VALUES (1);
INSERT INTO MYTABLE_PARENT VALUES (2);
INSERT INTO MYTABLE VALUES (1, 'MYVAL1-1', 1);
INSERT INTO MYTABLE VALUES (2, 'MYVAL1-2', 1);
INSERT INTO MYTABLE VALUES (3, 'MYVAL2-1', 2);

-- VIEW
CREATE OR REPLACE FORCE VIEW MYVIEW AS 
  SELECT CS.PRIMARY_KEY AS PARENT_PK,

  -- THE BELOW STATEMENTS REUSE THE SAME INNER SUBQUERY, IF THE 
  -- VIEW CONTAINS 1000 ROWS, THE INNER SUBQUERY WILL BE EXECUTED 
  -- 1000 TIMES, RETURNING THE SAME DATA EACH TIME.
  (SELECT PRIMARY_KEY FROM MYTABLE WHERE MYVAL = 'MYVAL1-1' AND 
  PARENT_PRIMARY_KEY = (SELECT PRIMARY_KEY FROM MYTABLE_PARENT 
  WHERE PRIMARY_KEY = CS.PRIMARY_KEY)) AS OUTPUT_VAL_1,

  (SELECT PRIMARY_KEY FROM MYTABLE WHERE MYVAL = 'MYVAL1-2' AND 
  PARENT_PRIMARY_KEY = (SELECT PRIMARY_KEY FROM MYTABLE_PARENT 
  WHERE PRIMARY_KEY = CS.PRIMARY_KEY)) AS OUTPUT_VAL_2
-- DEFINE CS
  FROM MYTABLE_PARENT CS;

SELECT * FROM MYVIEW;

select语句的结果如下：

PARENT_PK              OUTPUT_VAL_1           OUTPUT_VAL_2           
---------------------- ---------------------- ---------------------- 
1                      1                      2                      
2

因此，在上面的查询中，我想对视图中的每一行只运行一次以下查询：

(SELECT PRIMARY_KEY FROM MYTABLE_PARENT WHERE PRIMARY_KEY = CS.PRIMARY_KEY)

有没有办法在视图中优化内部子查询？

Answer 1

要对mytable和mytable_parent进行单独访问，请尝试：

CREATE OR REPLACE FORCE VIEW MYVIEW AS 
  SELECT CS.PRIMARY_KEY AS PARENT_PK,
         MAX(DECODE(MT.MYVAL, 'MYVAL1-1',MT.PRIMARY_KEY, TO_NUMBER(NULL))
                        AS OUTPUT_VAL_1,
         MAX(DECODE(MT.MYVAL, 'MYVAL1-2',MT.PRIMARY_KEY, TO_NUMBER(NULL))
                        AS OUTPUT_VAL_2
  FROM MYTABLE_PARENT CS
  LEFT JOIN MYTABLE MT ON MT.PARENT_PRIMARY_KEY = CS.PRIMARY_KEY AND
                          MT.MYVAL IN ('MYVAL1-1', 'MYVAL1-2')
  GROUP BY CS.PRIMARY_KEY

Answer 2

我现在无法对此进行测试，但请在视图的SELECT部分尝试此操作。我认为它至少会切断你的嵌套子查询。

SELECT CS.PRIMARY_KEY AS PARENT_PK, mv1.PRIMARY_KEY AS OUTPUT_VAL_1, mv2.PRIMARY_KEY AS OUTPUT_VAL_2
FROM MYTABLE_PARENT CS
LEFT JOIN MYTABLE mv1 ON mv1.PARENT_PRIMARY_KEY = CS.PRIMARY_KEY
LEFT JOIN MYTABLE mv2 ON mv2.PARENT_PRIMARY_KEY = CS.PRIMARY_KEY
WHERE mv1.MYVAL = 'MYVAL1-1'
AND mv2.MYVAL = 'MYVAL1-2';

Answer 3

子查询显示：

(SELECT PRIMARY_KEY FROM MYTABLE_PARENT WHERE PRIMARY_KEY = CS.PRIMARY_KEY)

但别名CS也指向MYTABLE_PARENT：

MYTABLE_PARENT CS;

所以整个子查询都是多余的，你可以用CS.PRIMARY_KEY替换它吗？

然后可以重写整个查询，如：

select    cs.primary_key as parent_pk
,         mt1.primary_key
,         mt2.primary_key
from      mytable_parent cs
join      mytable mt1
on        mt.parent_primary_key = cs.primary_key
          and mt1.myval = 'myval1-1'
join      mytable mt2
on        mt2.parent_primary_key = cs.primary_key
          and mt2.myval = 'myval1-2'

VIEW中的Oracle SQL Reuse子查询

3 个答案: