避免 groupby 子句中的列

Question

这里我需要避免 group by 子句中的这 2 列，它返回的记录数与我的预期值相比更多。

实际上来自st.cnt 的查询结果返回了 3 条记录，但是当我们以SP 运行时，我得到了预期的两倍。

st_cnt.COUNT_FILES，
st_cnt.COUNT_SCANS，

SELECT table1.DA_SEC,
  table1.DA_REG,
  table1.PROJ,
  table1.SOL,
  table1.SCAN_YEAR_MON,
  to_char(TO_DATE(table1.SCAN_YEAR_MON, 'YYYY-MM'), 'YYYY'),
  to_char(TO_DATE(table1.SCAN_YEAR_MON, 'YYYY-MM'), 'MM'),
  COUNT(DISTINCT table1.ISSUE_ID),
  COUNT(DISTINCT table1.USERNAME),
  COUNT(DISTINCT table1.PATH),
  st_cnt.COUNT_FILES,
  st_cnt.COUNT_SCANS,
  sum(unq_Pro.HIGH_CNT),
  sum(unq_Pro.MEDIUM_CNT),
  sum(unq_Pro.LOW_CNT),
  sum(SUPPR_HIGH) - sum(UNSUPPR_HIGH),
  sum(SUPPR_MEDIUM) - sum(UNSUPPR_MEDIUM),
  sum(SUPPR_LOW) - sum(UNSUPPR_LOW),
  CURRENT_DATE
FROM table1
JOIN
  (SELECT PROJ, SOL, SCAN_YEAR, SCAN_MONTH, COUNT(DISTINCT PATH) COUNT_FILES, COUNT(DISTINCT SCAN_ID) COUNT_SCANS FROM table2 GROUP BY PROJ, SOL, SCAN_YEAR, SCAN_MONTH) st_cnt
ON st_cnt.PROJ = table1.PROJ--AND st_cnt.SOL = table1.SOL
GROUP BY table1.DA_SEC, table1.DA_REG, table1.PROJ, table1.SOL, table1.SCAN_YEAR_MON, st_cnt.COUNT_FILES, st_cnt.COUNT_SCAN;

Answer 1

大量时间使用 With Clause 在这些类型的场景中非常有用。这是讨论 With 子句的 ask tom 线程 Ask Tom Link 也是使用 with 子句的子查询分解的有用链接Sub Query Factoring

它还有助于我们了解在这种情况下表 1 的数据是什么样的，以便我们可以测试答案。但是有了你分享的信息，这样的东西可以帮助你

WITH TAB1_DATA AS (
SELECT table1.DA_SEC,
 table1.DA_REG,
 table1.PROJ,
 table1.SOL,
 table1.SCAN_YEAR_MON,
 to_char(TO_DATE(table1.SCAN_YEAR_MON, 'YYYY-MM'), 'YYYY'),
 to_char(TO_DATE(table1.SCAN_YEAR_MON, 'YYYY-MM'), 'MM'),
COUNT(DISTINCT table1.ISSUE_ID),
COUNT(DISTINCT table1.USERNAME),
COUNT(DISTINCT table1.PATH),
sum(unq_Pro.HIGH_CNT),
sum(unq_Pro.MEDIUM_CNT),
sum(unq_Pro.LOW_CNT),
sum(SUPPR_HIGH) - sum(UNSUPPR_HIGH),
sum(SUPPR_MEDIUM) - sum(UNSUPPR_MEDIUM),
sum(SUPPR_LOW) - sum(UNSUPPR_LOW),
CURRENT_DATE
  FROM table1
GROUP BY table1.DA_SEC, table1.DA_REG, table1.PROJ, table1.SOL, table1.SCAN_YEAR_MON
)
 SELECT TD.* ,  st_cnt.COUNT_FILES,st_cnt.COUNT_SCANS FROM TAB1_DATA TD ,
(SELECT PROJ, SOL, SCAN_YEAR, SCAN_MONTH, COUNT(DISTINCT PATH) COUNT_FILES, 
COUNT(DISTINCT SCAN_ID) COUNT_SCANS FROM table2 GROUP BY PROJ, SOL, SCAN_YEAR, 
SCAN_MONTH) st_cnt
WHERE st_cnt.PROJ = TD.PROJ;