如何根据列的值对一个记录集的总和进行求和？

Question

这是我的T-SQL代码：

SELECT H.ORDER_DATE 'Order Date', 
     SUM(D.WORK_QTY) 'Work Qty', 
     SUM(TD. TRAN_QTY) 'Transaction Qty', 
     IM.PUMJ_CD 

FROM [dbo].[ORDER_HEADER] H 

    LEFT JOIN ORDER_DETAIL D ON D.ORDER_ID = H.ORDER_ID 
    LEFT JOIN TRAN_HEADER T ON T.ORDER_ID = H.ORDER_ID 
    LEFT JOIN TRAN_DETAIL TD ON TD.TRAN_ID = T.TRAN_ID 
    LEFT JOIN TRAN_TYPE TT ON TT.TRAN_TYPE_ID = TD.TRAN_TYPE_ID 
    LEFT JOIN ITM_MST IM ON IM.ITM_NM = D.ITEM_NAME 

WHERE TD.TRAN_TYPE_ID = '231' 
     AND PUMJ_CD IN ('BF','BR','SHFT','BJ','TJ') 
     AND H.ORDER_DATE BETWEEN '20170403' AND '20170407' 

GROUP BY IM.PUMJ_CD, 
      H.ORDER_DATE 

ORDER BY H.ORDER_DATE 

，并返回记录集：

Order Date | Work Qty | Transaction Qty | PUMJ_CD 
------------------------------------------------- 
20170403 | 10  | 7    | BF 
20170403 | 24  | 12    | BR 
20170403 | 16  | 14    | TJ 
20170404 | 15  | 10    | BF 

是否有可能从记录集求和行，其中： “订购日期”是同样，和 PUMJ_CD IN（ 'BF'， 'BR'） ，并有导致记录集的样子：

Order Date | Work Qty | Transaction Qty | PUMJ_CD 
------------------------------------------------- 
20170403 | 34  | 19    | TOTAL 
20170403 | 16  | 14    | TJ 
20170404 | 15  | 10    | BF 

我一直在试图找出一种方法在一个查询中做到这一点，但由于我的知识有限，我没有取得成功。我尝试过在线查找，但我发现很难用谷歌中的问题来描述我需要的内容。如果任何人能够帮助或指出我正确的方向，那将是非常感谢。

Answer 1

利用联盟在两组数据：

如果你想要一个单独的值你可以做轻微的修改。

1 - 无BR和BF
2 - 只是BR和BF作为总

SELECT H.ORDER_DATE 'Order Date', 
     SUM(D.WORK_QTY) 'Work Qty', 
     SUM(TD. TRAN_QTY) 'Transaction Qty', 
     IM.PUMJ_CD 

FROM [dbo].[ORDER_HEADER] H 

    LEFT JOIN ORDER_DETAIL D ON D.ORDER_ID = H.ORDER_ID 
    LEFT JOIN TRAN_HEADER T ON T.ORDER_ID = H.ORDER_ID 
    LEFT JOIN TRAN_DETAIL TD ON TD.TRAN_ID = T.TRAN_ID 
    LEFT JOIN TRAN_TYPE TT ON TT.TRAN_TYPE_ID = TD.TRAN_TYPE_ID 
    LEFT JOIN ITM_MST IM ON IM.ITM_NM = D.ITEM_NAME 

WHERE TD.TRAN_TYPE_ID = '231' 
     AND PUMJ_CD IN ('SHFT','BJ','TJ') -- NO BR BF 
     AND H.ORDER_DATE BETWEEN '20170403' AND '20170407' 

GROUP BY IM.PUMJ_CD, 
      H.ORDER_DATE 

union 

SELECT H.ORDER_DATE, 
     SUM(D.WORK_QTY), 
     SUM(TD. TRAN_QTY), 
     'TOTAL' 

FROM [dbo].[ORDER_HEADER] H 

    LEFT JOIN ORDER_DETAIL D ON D.ORDER_ID = H.ORDER_ID 
    LEFT JOIN TRAN_HEADER T ON T.ORDER_ID = H.ORDER_ID 
    LEFT JOIN TRAN_DETAIL TD ON TD.TRAN_ID = T.TRAN_ID 
    LEFT JOIN TRAN_TYPE TT ON TT.TRAN_TYPE_ID = TD.TRAN_TYPE_ID 
    LEFT JOIN ITM_MST IM ON IM.ITM_NM = D.ITEM_NAME 

WHERE TD.TRAN_TYPE_ID = '231' 
     AND PUMJ_CD IN ('BF','BR') -- ONLY BF AND BR AS TOTAL 
     AND H.ORDER_DATE BETWEEN '20170403' AND '20170407' 

GROUP BY H.ORDER_DATE 

Answer 2

是的。您可以使用CASE的聚集和在SELECT：

SELECT H.ORDER_DATE as [Order Date], 
     SUM(D.WORK_QTY) as [Work Qty], 
     SUM(TD. TRAN_QTY) as [Transaction Qty], 
     (CASE WHEN IM.PUMJ_CD IN ('BF','BR') THEN 'TOTAL' ELSE IM.PUMJ_CD 
     END) as PUMJ_CD 
FROM . . . 
GROUP BY H.ORDER_DATE, 
     (CASE WHEN IM.PUMJ_CD IN ('BF','BR') THEN 'TOTAL' ELSE IM.PUMJ_CD 
      END) 

这是你想要的东西有点不同。对于这两个值，它总是会返回TOTAL。

SELECT H.ORDER_DATE as [Order Date], 
     SUM(D.WORK_QTY) as [Work Qty], 
     SUM(TD. TRAN_QTY) as [Transaction Qty], 
     (CASE WHEN COUNT(*) = 1 THEN MIN(IM.PUMJ_CD) 
       ELSE (CASE WHEN IM.PUMJ_CD IN ('BF','BR') THEN 'TOTAL' ELSE IM.PUMJ_CD) 
     END) as PUMJ_CD 
FROM . . . 
GROUP BY H.ORDER_DATE, 
     (CASE WHEN IM.PUMJ_CD IN ('BF','BR') THEN 'TOTAL' ELSE IM.PUMJ_CD 
      END)