Oracle伪列Level的使用

Oracle伪列Level的使用

1 根据ID的数量按序展开

现有表数据如下：

用户ID	开始日期	结束日期
USER_ID	START_DT	END_DT
U101	2016-11-02	2016-11-06
U102	2015-06-01	2015-06-03

要生成下面的数据：

用户ID	日期
U101	2016-11-02
U101	2016-11-03
U101	2016-11-04
U101	2016-11-05
U101	2016-11-06
U102	2015-06-01
U102	2015-06-02
U102	2015-06-03

    WITH TB AS
     (SELECT 'U101' USER_ID, DATE'2016-11-02' START_DT, DATE'2016-11-06' END_DT  FROM DUAL
      UNION ALL
      SELECT 'U102' , DATE'2015-06-01', DATE'2015-06-03' FROM DUAL
      )
    SELECT TB.USER_ID, TB.START_DT + LEVEL - 1
    FROM TB
    CONNECT BY TB.USER_ID = PRIOR TB.USER_ID
           AND LEVEL <= (TB.END_DT - TB.START_DT + 1)
           AND PRIOR DBMS_RANDOM.VALUE IS NOT NULL;

2 根据ID的数量按序展开

现有表数据如下：

用户ID	手机号	数量
USER_ID	Mobile	CNT
U101	12345	3
U102	33563	4

要生成下面的数据：

用户ID	手机号	序号
USER_ID	MOBILE	LV
U101	12345	1
U101	12345	2
U101	12345	3
U102	33563	1
U102	33563	2
U102	33563	3
U102	33563	4

WITH TB AS
 (SELECT 'U101' USER_ID, '12345' MOBILE, 3 CNT  FROM DUAL
    UNION ALL
    SELECT 'U102' , '33563', 4 FROM DUAL
 )
SELECT A.USER_ID,A.MOBILE,B.LV
    FROM TB A
    LEFT JOIN (
                    SELECT TB.USER_ID, LEVEL LV
                    FROM TB
                    CONNECT BY TB.USER_ID = PRIOR TB.USER_ID
                                 AND LEVEL <= CNT
                                 AND PRIOR DBMS_RANDOM.VALUE IS NOT NULL
                        ) B
        ON A.USER_ID = B.USER_ID

3 生成随机数据

SELECT DBMS_RANDOM.VALUE
  FROM DUAL 
  CONNECT BY LEVEL <= 100;

SELECT SYSDATE - LEVEL
  FROM DUAL 
  CONNECT BY LEVEL <= 100;

4 转置截取字符串

STR
大
家
好

    SELECT SUBSTR('大家好',LEVEL,1)
      FROM DUAL
    CONNECT BY LEVEL<=LENGTH('大家好');

---

USER_ID	STR
U1	哈
U1	喽
U1	,
U1	世
U1	界
U2	你
U2	好
U2	啊
U2	,
U2	地
U2	球

    WITH TB AS (
      SELECT 'U1' USER_ID, '哈喽,世界' STR FROM DUAL
        UNION ALL
        SELECT 'U2', '你好啊,地球' FROM DUAL
    )
    SELECT TB.USER_ID, 
           SUBSTR(STR,LEVEL,1)
      FROM TB
    CONNECT BY TB.USER_ID = PRIOR TB.USER_ID
             AND LEVEL <= LENGTH(STR)
             AND PRIOR DBMS_RANDOM.VALUE IS NOT NULL;

---

RN	STR
a	1
a	2
a	3
b	4
b	5
b	6

WITH tb AS
 (SELECT 'a' rn,         '1,2,3' i_name    FROM dual
  UNION ALL
  SELECT 'b',         '4,5,6'    FROM dual)
  SELECT rn,
       regexp_substr(i_name, '[^,]+', 1, LEVEL)
  FROM tb
CONNECT BY PRIOR dbms_random.value IS NOT NULL
       AND PRIOR rn = rn
       AND LEVEL <= length(i_name) - length(REPLACE(i_name, ',', '')) + 1;