0

我试图想出一种方法，我只能用最近的 ACCOUNT 获得 16 行（这是基于 SERV_ACCT 的 16 条不同的记录）。我正在使用 Distinct 和 group by，我认为这可以解决问题，但我一直得到 20 行。

宣布

 @CustomerID int = 8,
 @UtilityCompanyID int = 1


    SELECT DISTINCT SERV_ACCT, ACCOUNT, MAX(INV_DATE)
    FROM tblAPSData

    WHERE SERV_ACCT NOT IN (SELECT ServiceAccount
                FROM tblMEP_Meters
                JOIN tblMEP_Sites
                ON tblMEP_Meters.SiteID = tblMEP_Sites.ID

                JOIN tblMEP_Projects
                ON tblMEP_Projects.ID = tblMEP_Sites.ProjectID

                WHERE CustomerID = 8
                AND Type = 1
                )

    AND ACCOUNT IN (SELECT AccountNumber
                    FROM tblMEP_CustomerAccounts
                    WHERE CustomerID = @CustomerID
                    AND UtilityCompanyID = @UtilityCompanyID)

    AND INV_DATE > DATEADD(month, -6, getdate())
    GROUP BY SERV_ACCT, ACCOUNT
    ORDER BY SERV_ACCT

结果如下：

SERV_ACCT   ACCOUNT (No column name)
0289S61288  117512280   2013-06-12 00:00:00.000
0492S90281  117512280   2013-06-12 00:00:00.000
0492S90281  651412281   2013-04-08 00:00:00.000
1303S90280  117512280   2013-06-12 00:00:00.000
1435S01282  117512280   2013-06-12 00:00:00.000
1440S13289  312937281   2013-06-24 00:00:00.000
1548S00286  117512280   2013-06-12 00:00:00.000
1548S00286  308710287   2013-04-08 00:00:00.000
2498S21288  117512280   2013-06-12 00:00:00.000
5384S92284  979437282   2013-06-24 00:00:00.000
5538S21284  117512280   2013-06-12 00:00:00.000
6109S12286  117512280   2013-06-12 00:00:00.000
7358S00281  117512280   2013-06-12 00:00:00.000
7358S00281  659710281   2013-04-08 00:00:00.000
7488S22289  117512280   2013-06-12 00:00:00.000
8058S12287  117512280   2013-06-12 00:00:00.000
9058S00288  117512280   2013-06-12 00:00:00.000
9168S00282  117512280   2013-06-12 00:00:00.000
9168S00282  570810282   2013-04-08 00:00:00.000
9645S21281  117512280   2013-06-12 00:00:00.000
4

2 回答 2

2

如果您至少使用 SQL-Server 2005，则可以ROW_NUMBER在 cte 中使用排名函数：

WITH CTE AS(
  SELECT [SERV_ACCT], [ACCOUNT], [INV_DATE],
     RN = ROW_NUMBER() OVER (PARTITION BY SERV_ACCT ORDER BY INV_DATE DESC)
  FROM dbo.Table1
)
SELECT [SERV_ACCT], [ACCOUNT], [INV_DATE]
FROM CTE
WHERE RN = 1

演示

于 2013-07-15T22:17:12.537 回答
0

首先，我会将您的 where 子句语句组合到 JOIN 中以获得更好的性能。然后使用分区函数按日期对行进行排名。

排名 (Transact-SQL)

SELECT * FROM
(
SELECT 
aps.SERV_ACCT
, aps.ACCOUNT
, ROW_NUMBER() OVER (PARTITION BY aps.SERV_ACCT ORDER BY aps.INV_DATE DESC) [row]
    FROM tblAPSData aps
    INNER JOIN tblMEP_Meters mep
      ON aps.SERV_ACCT = mep.ServiceAccount
    INNER JOIN tblMEP_Sites site
      ON site.SiteID = mep.ID
    INNER JOIN tbl_MEP_Projects proj
      ON proj.ID = site.ProjectID
      AND proj.CustomerID = 8 -- Move this to appropriate table
      AND proj.Type = 1 --Move this to appropriate table assumed it was on proj
    INNER JOIN tblMEP_CustomerAccounts custAcc
      ON custAcc.AccountNumber = aps.Account
      AND custAcc.CustomerID = @CustomerID
      AND custAcc.UtilityCompanyID = @UtilityCompanyID
WHERE aps.INV_DATE > DATEADD(month, -6, getdate())
) data
WHERE
  data.row = 1
于 2013-07-15T22:24:21.620 回答