Question

编辑：以下是我正在使用的数据。我写的查询给了我基于OCCPSTAT字段的重复项（最右边的数据列，C或N）。

我需要一个查询来拉取最接近我定义的日期的记录（代码执行此操作），但是，如果有记录的C和N副本，我只需要N记录。

5000    1000    D001    RNT 100 11/1/2014 0:00  C
5000    1000    D001    GAS 200 11/1/2014 0:00  C
5000    1000    D001    ELC 300 11/1/2014 0:00  C
5000    1000    D001    WAT 400 11/1/2014 0:00  C
5000    1000    D001    TAX 500 11/1/2014 0:00  C
5000    1000    D001    INS 600 11/1/2014 0:00  C
5000    1000    D001    RNT 200 11/1/2015 0:00  C
5000    1000    D001    GAS 300 11/1/2015 0:00  C
5000    1000    D001    ELC 400 11/1/2015 0:00  C
5000    1000    D001    WAT 500 11/1/2015 0:00  C
5000    1000    D001    TAX 600 11/1/2015 0:00  C
5000    1000    D001    INS 700 11/1/2015 0:00  C
5000    1000    D001    RNT 300 11/1/2016 0:00  C
5000    1000    D001    GAS 400 11/1/2016 0:00  C
5000    1000    D001    ELC 500 11/1/2016 0:00  C
5000    1000    D001    WAT 600 11/1/2016 0:00  C
5000    1000    D001    TAX 700 11/1/2016 0:00  C
5000    1000    D001    INS 800 11/1/2016 0:00  C
5000    1000    D001    RNT 500 11/1/2017 0:00  C
5000    1000    D001    GAS 600 11/1/2017 0:00  C
5000    1000    D001    ELC 700 11/1/2017 0:00  C
5000    1000    D001    WAT 800 11/1/2017 0:00  C
5000    1000    D001    TAX 900 11/1/2017 0:00  C
5000    1000    D001    INS 100011/1/2017 0:00  C
5000    1000    D001    RNT 100 11/1/2014 0:00  N
5000    1000    D001    GAS 200 11/1/2014 0:00  N
5000    1000    D001    ELC 300 11/1/2014 0:00  N
5000    1000    D001    WAT 400 11/1/2014 0:00  N
5000    1000    D001    TAX 500 11/1/2014 0:00  N
5000    1000    D001    INS 600 11/1/2014 0:00  N
5000    1000    D001    RNT 200 11/1/2015 0:00  N
5000    1000    D001    GAS 300 11/1/2015 0:00  N
5000    1000    D001    ELC 400 11/1/2015 0:00  N
5000    1000    D001    WAT 500 11/1/2015 0:00  N
5000    1000    D001    TAX 600 11/1/2015 0:00  N
5000    1000    D001    INS 700 11/1/2015 0:00  N
5000    1000    D001    RNT 300 11/1/2016 0:00  N
5000    1000    D001    GAS 400 11/1/2016 0:00  N
5000    1000    D001    ELC 500 11/1/2016 0:00  N
5000    1000    D001    WAT 600 11/1/2016 0:00  N
5000    1000    D001    TAX 700 11/1/2016 0:00  N
5000    1000    D001    INS 800 11/1/2016 0:00  N
5000    1000    D001    RNT 500 11/1/2017 0:00  N
5000    1000    D001    GAS 600 11/1/2017 0:00  N
5000    1000    D001    ELC 700 11/1/2017 0:00  N
5000    1000    D001    WAT 800 11/1/2017 0:00  N
5000    1000    D001    TAX 900 11/1/2017 0:00  N
5000    1000    D001    INS 1000    11/1/2017 0:00  N

查询是：

SELECT
    BLDGID,
    LEASID,
    SUITID,
    INCCAT,
    AMOUNT,
    EFFDATE, 
    OCCPSTAT

FROM( 
    SELECT
    CM.BLDGID,
    CM.LEASID,
    LS.SUITID,
    CM.INCCAT,
    CM.AMOUNT,
    CM.EFFDATE,
    LS.OCCPSTAT,
ROW_NUMBER() OVER(PARTITION BY CM.BLDGID, CM.LEASID, LS.SUITID, CM.INCCAT, LS.OCCPSTAT
ORDER BY CM.EFFDATE DESC) AS rowno 

 FROM SQLDATA.DBO.CMRECC CM INNER JOIN SQLDATA.DBO.LEAS LS ON CM.LEASID = LS.LEASID
 WHERE (CM.EFFDATE < '20170601') AND (CM.BLDGID = '85000') AND (LS.OCCPSTAT = 'C' OR LS.OCCPSTAT = 'N')


) AS d 
WHERE rowno = 1

Answer 1

您不需要case。一种方法是只跟踪每个组的occpstat列，然后在where中应用逻辑。

但是，我认为有一个更简单的解决方案。将occpstat从partition by移至order by并首先选择“N”值：

SELECT BLDGID, LEASID, SUITID, INCCAT, AMOUNT, EFFDATE,  OCCPSTAT
FROM (SELECT CM.BLDGID, CM.LEASID, LS.SUITID, CM.INCCAT
             CM.AMOUNT, CM.EFFDATE, LS.OCCPSTAT,
             ROW_NUMBER() OVER (PARTITION BY CM.BLDGID, CM.LEASID, LS.SUITID, CM.INCCAT 
                                ORDER BY LS.OCCPSTAT DESC, CM.EFFDATE DESC) AS rownum
     FROM SQLDATA.DBO.CMRECC CM INNER JOIN
          SQLDATA.DBO.LEAS LS
          ON CM.LEASID = LS.LEASID
     WHERE CM.EFFDATE < '20170601' AND CM.BLDGID = '85000' AND
           LS.OCCPSTAT IN ('C', 'N')
     ) d 
WHERE rownum = 1

需要帮助编写查询。

1 个答案: