oracle 按旬统计并且每月小计行转列 PIVOT函数与分组小计 ROLLUP 函数

数据库版本:oralce 11g

表名WM_TD_WATER_DAILY 结构如下,我把没用到的字段隐藏了,全放出来不太好哈.

MONITOR_ID为BI_TB_MONITOR_BI的ID ,用于关联站点名称(SHORT_NAME).

业务需求 :现有日水量表,每条数据记录着一天的用水量,bi_tb_monitor_bi中有一下九个站点

全部SQL如下,下面有SQL分解,看起来能清晰很多

SELECT MON, CASEWHEN MON IS NULL AND DATETIME IS NULL THEN'累计水量'WHEN MON IS NOT NULL AND DATETIME IS NULL THEN'小计'ELSECASEWHEN INSTR(DATETIME, '1旬') > 0 THENREPLACE(DATETIME, '1旬', '上旬')WHEN INSTR(DATETIME, '2旬') > 0 THENREPLACE(DATETIME, '2旬', '中旬')WHEN INSTR(DATETIME, '3旬') > 0 THENREPLACE(DATETIME, '3旬', '下旬')ENDEND DATETIME,SUM(NVL(车逻洞, 0)) 车逻洞,SUM(NVL(车逻闸, 0)) 车逻闸,SUM(NVL(车逻, 0)) 车逻,SUM(NVL(南关, 0)) 南关,SUM(NVL(头闸, 0)) 头闸,SUM(NVL(周山洞, 0)) 周山洞,SUM(NVL(界首小闸, 0)) 界首小闸,SUM(NVL(子英闸, 0)) 子英闸,SUM(NVL(周山, 0)) 周山,SUM(NVL(车逻洞, 0)) + SUM(NVL(车逻闸, 0)) + SUM(NVL(南关, 0)) +SUM(NVL(头闸, 0)) + SUM(NVL(周山洞, 0)) + SUM(NVL(界首小闸, 0)) +SUM(NVL(子英闸, 0)) + SUM(NVL(周山, 0)) 合计FROM (SELECT TO_CHAR(MONITOR_DATE, 'yyyy') YEAR,TO_CHAR(MONITOR_DATE, 'yyyy') || TO_CHAR(MONITOR_DATE, 'mm') MON,TO_CHAR(MONITOR_DATE, 'mm') || '月' ||DECODE(TRUNC((TO_CHAR(MONITOR_DATE, 'dd') - 1) / 10), 0, '1旬' ，1, '2旬', '3旬') DATETIME,SUM(车逻洞) / 10000 车逻洞,SUM(车逻闸) / 10000 车逻闸,SUM(车逻) / 10000 车逻,SUM(南关) / 10000 南关,SUM(头闸) / 10000 头闸,SUM(周山洞) / 10000 周山洞,SUM(界首小闸) / 10000 界首小闸,SUM(子英闸) / 10000 子英闸,SUM(周山) / 10000 周山FROM (SELECT B.SHORT_NAME, T.MONITOR_DATE, T.WATER_DAILYFROM WM_TD_WATER_DAILY TLEFT JOIN BI_TB_MONITOR_BI BON T.MONITOR_ID = B.IDWHERE T.REGION_ID = '6-100-018'AND T.MONITOR_DATE BETWEEN TO_DATE('2018-10', 'yyyy-MM') ANDTO_DATE('2018-12', 'yyyy-MM'))PIVOT(SUM(WATER_DAILY)FOR SHORT_NAME IN('车逻洞' 车逻洞,'车逻闸' 车逻闸,'车逻' 车逻,'南关' 南关,'头闸' 头闸,'周山洞' 周山洞,'界首小闸' 界首小闸,'子英闸' 子英闸,'周山' 周山))GROUP BY TO_CHAR(MONITOR_DATE, 'yyyy'),TO_CHAR(MONITOR_DATE, 'mm'),DECODE(TRUNC((TO_CHAR(MONITOR_DATE, 'dd') - 1) / 10),0,'1旬' ，1,'2旬','3旬')ORDER BY MON)GROUP BY ROLLUP(MON, DATETIME)

SQL分解

第一层SQL:

只做最基本的查询,并把条件都写好

 SELECT B.SHORT_NAME, T.MONITOR_DATE, T.WATER_DAILYFROM WM_TD_WATER_DAILY TLEFT JOIN BI_TB_MONITOR_BI BON T.MONITOR_ID = B.IDWHERE T.REGION_ID = '6-100-018'AND T.MONITOR_DATE BETWEEN TO_DATE('2018-10', 'yyyy-MM') ANDTO_DATE('2018-12', 'yyyy-MM')

查询结果如下(只插入了两个站点的数据<车逻洞>,<车逻闸>):

第二层SQL:

用decode(trunc((to_char(monitor_date,'dd')-1)/10),0,'1旬'，1,'2旬','3旬') 把旬分组,

行转列用了 PIVOT (sum(water_daily) for short_name in ('车逻洞' 车逻洞, '车逻闸' 车逻闸,'车逻' 车逻,
'南关' 南关,'头闸' 头闸,'周山洞' 周山洞,'界首小闸' 界首小闸,'子英闸' 子英闸,'周山' 周山))

每个sum后除以10000是业务需求

SELECT TO_CHAR(MONITOR_DATE, 'yyyy') YEAR,TO_CHAR(MONITOR_DATE, 'yyyy') || TO_CHAR(MONITOR_DATE, 'mm') MON,TO_CHAR(MONITOR_DATE, 'mm') || '月' ||DECODE(TRUNC((TO_CHAR(MONITOR_DATE,'dd') - 1) / 10), 0, '1旬' ，1, '2旬', '3旬')DATETIME,SUM(车逻洞) / 10000 车逻洞,SUM(车逻闸) / 10000 车逻闸,SUM(车逻) / 10000 车逻,SUM(南关) / 10000 南关,SUM(头闸) / 10000 头闸,SUM(周山洞) / 10000 周山洞,SUM(界首小闸) / 10000 界首小闸,SUM(子英闸) / 10000 子英闸,SUM(周山) / 10000 周山FROM (SELECT B.SHORT_NAME, T.MONITOR_DATE, T.WATER_DAILYFROM WM_TD_WATER_DAILY TLEFT JOIN BI_TB_MONITOR_BI BON T.MONITOR_ID = B.IDWHERE T.REGION_ID = '6-100-018'AND T.MONITOR_DATE BETWEEN TO_DATE('2018-10', 'yyyy-MM') ANDTO_DATE('2018-12', 'yyyy-MM'))PIVOT(SUM(WATER_DAILY)FOR SHORT_NAME IN('车逻洞' 车逻洞,'车逻闸' 车逻闸,'车逻' 车逻,'南关' 南关,'头闸' 头闸,'周山洞' 周山洞,'界首小闸' 界首小闸,'子英闸' 子英闸,'周山' 周山))GROUP BY TO_CHAR(MONITOR_DATE, 'yyyy'),TO_CHAR(MONITOR_DATE, 'mm'),DECODE(TRUNC((TO_CHAR(MONITOR_DATE, 'dd') - 1) / 10),0,'1旬' ，1,'2旬','3旬')ORDER BY MON

查询结果如下:

最后一层SQL:

用了 GROUP BY ROLLUP (MON,DATETIME) 按MON,DATETIME分组合计

SELECT CASEWHEN MON IS NULL AND DATETIME IS NULL THEN'累计水量'WHEN MON IS NOT NULL AND DATETIME IS NULL THEN'小计'ELSECASEWHEN INSTR(DATETIME, '1旬') > 0 THENREPLACE(DATETIME, '1旬', '上旬')WHEN INSTR(DATETIME, '2旬') > 0 THENREPLACE(DATETIME, '2旬', '中旬')WHEN INSTR(DATETIME, '3旬') > 0 THENREPLACE(DATETIME, '3旬', '下旬')ENDEND DATETIME,SUM(NVL(车逻洞, 0)) 车逻洞,SUM(NVL(车逻闸, 0)) 车逻闸,SUM(NVL(车逻, 0)) 车逻,SUM(NVL(南关, 0)) 南关,SUM(NVL(头闸, 0)) 头闸,SUM(NVL(周山洞, 0)) 周山洞,SUM(NVL(界首小闸, 0)) 界首小闸,SUM(NVL(子英闸, 0)) 子英闸,SUM(NVL(周山, 0)) 周山,SUM(NVL(车逻洞, 0)) + SUM(NVL(车逻闸, 0)) + SUM(NVL(南关, 0)) +SUM(NVL(头闸, 0)) + SUM(NVL(周山洞, 0)) + SUM(NVL(界首小闸, 0)) +SUM(NVL(子英闸, 0)) + SUM(NVL(周山, 0)) 合计FROM (SELECT TO_CHAR(MONITOR_DATE, 'yyyy') YEAR,TO_CHAR(MONITOR_DATE, 'yyyy') || TO_CHAR(MONITOR_DATE, 'mm') MON,TO_CHAR(MONITOR_DATE, 'mm') || '月' ||DECODE(TRUNC((TO_CHAR(MONITOR_DATE,'dd') - 1) / 10), 0, '1旬' ，1, '2旬', '3旬')DATETIME,SUM(车逻洞) / 10000 车逻洞,SUM(车逻闸) / 10000 车逻闸,SUM(车逻) / 10000 车逻,SUM(南关) / 10000 南关,SUM(头闸) / 10000 头闸,SUM(周山洞) / 10000 周山洞,SUM(界首小闸) / 10000 界首小闸,SUM(子英闸) / 10000 子英闸,SUM(周山) / 10000 周山FROM (SELECT B.SHORT_NAME, T.MONITOR_DATE, T.WATER_DAILYFROM WM_TD_WATER_DAILY TLEFT JOIN BI_TB_MONITOR_BI BON T.MONITOR_ID = B.IDWHERE T.REGION_ID = '6-100-018'AND T.MONITOR_DATE BETWEEN TO_DATE('2018-10', 'yyyy-MM') ANDTO_DATE('2018-12', 'yyyy-MM'))PIVOT(SUM(WATER_DAILY)FOR SHORT_NAME IN('车逻洞' 车逻洞,'车逻闸' 车逻闸,'车逻' 车逻,'南关' 南关,'头闸' 头闸,'周山洞' 周山洞,'界首小闸' 界首小闸,'子英闸' 子英闸,'周山' 周山))GROUP BY TO_CHAR(MONITOR_DATE, 'yyyy'),TO_CHAR(MONITOR_DATE, 'mm'),DECODE(TRUNC((TO_CHAR(MONITOR_DATE, 'dd') - 1) / 10),0,'1旬' ，1,'2旬','3旬')ORDER BY MON)GROUP BY ROLLUP(MON, DATETIME)

查询结果如下:

select后的不加case when的效果如下

mon用于第二层的排序省略掉了

总结

这段SQL中用到了以下函数:

TO_DATE(),TO_CHAR(),BETWEEN,TRUNC(),DECODE(),NVL(),SUM(),PIVOT( FOR ),CASE WHEN THEN ELSE END,ROLLUP()

欢迎指正并优化SQL!