ORACLE SQL返回最近添加的记录详细信息

Question

编辑：解决了Dems的大量帮助，我将发布一个截断版本的查询，现在效果非常好！

            SELECT
  *

FROM

    O_PERSONS

    /*The below left join returns the most recently added ethnicity classification*/
  LEFT JOIN 

  (SELECT
   ROW_NUMBER() OVER (PARTITION BY O_CLASSIFICATIONS.CLA_SUBJECT_ID ORDER BY O_CLASSIFICATIONS.CLA_DATE_NOTIFIED DESC) AS Sequence_ID,
   O_CLASSIFICATIONS.CLA_CAT_ID,
   O_CLASSIFICATIONS.CLA_SUBJECT_ID
   FROM
   O_CLASSIFICATIONS
   WHERE
   O_CLASSIFICATIONS.CLA_SUBJECT_IND = 'P'
   AND
   O_CLASSIFICATIONS.CLA_TOP_CAT_ID   =  'ETHNIC'
   AND
   O_CLASSIFICATIONS.CLA_CAT_ID <> 'DECLINED'
   ) ETHNIC 
   ON ETHNIC.CLA_SUBJECT_ID = O_PERSONS.PER_ID 
   AND ETHNIC.Sequence_ID = 1

    /*The below left join returns the most recently added PCG classification*/
   LEFT JOIN 

  (SELECT
   ROW_NUMBER() OVER (PARTITION BY O_CLASSIFICATIONS.CLA_SUBJECT_ID ORDER BY O_CLASSIFICATIONS.CLA_DATE_NOTIFIED DESC) AS Sequence_ID,
   O_CLASSIFICATIONS.CLA_CAT_ID,
   O_CLASSIFICATIONS.CLA_SUBJECT_ID
   FROM
   O_CLASSIFICATIONS
   WHERE
   O_CLASSIFICATIONS.CLA_SUBJECT_IND = 'P'
   AND
   O_CLASSIFICATIONS.CLA_TOP_CAT_ID   =  'PRIMARY'
   AND
   O_CLASSIFICATIONS.CLA_CAT_ID <> 'DECLINED'
   ) PCG
   ON PCG.CLA_SUBJECT_ID = O_PERSONS.PER_ID 
   AND PCG.Sequence_ID = 1


  WHERE

   O_PERSONS.PER_ID LIKE 'P%'

我目前正在尝试编写一个查询，该查询将返回我们系统上客户端的某些详细信息。虽然客户只有一个出生日期，性别和P ID，但他们可能有多个种族（由于数据质量）以及客户群体（这更合理，因为客户的需求可能会发生变化）。但是，对于这些特定项目，我只对最近添加的种族或客户群感兴趣。下面我写了一个查询，它为种族和客户组分配一个行号，计划只返回等于1的那些，因为这些通常是最新的。但是，我遇到了两个问题，其中一个我可以解决，另一个我不太确定。

首先，一些客户不会记录种族或客户群。运行以下查询时，它们返回非常奇怪的行号（通常为数千）。但是，我知道在我的'where'中我可以指定如果客户端组或种族是空的，则返回一行（因为这些将是需要解决的数据质量案例）。

第二个问题稍微棘手，我将尝试用下表来表示。请注意，为了便于输入和理解，我缩写了表格。

  P ID Ethnicity   PCG RN E RN P
  P1   WB          OV  1    2 
  P1   WI          OV  2    1          

上述客户有两个种族，虽然WB是最新的（因此行号E中的1是正确的）。但是，客户端只记录了一个PCG，但行号在第一行返回2（可以说是我要返回的行）。我不知道为什么虽然我猜是因为ETHNIC.CLA_SUBJECT_ID在O_PERSONS.PER_ID上加入而PER_ID出现两次，这就是为什么它认为该特定字段有两行。但是，即使是这种情况，有没有办法迫使1只出现在一行？或者我可以采用完全不同的方式吗？希望这个查询有道理，如果部分不清楚则道歉。谢谢，

                         SELECT
          O_PERSONS.PER_ID as "P ID",
          olm_bo.get_per_name(O_PERSONS.PER_ID) as "Full Name",
          O_PERSONS.PER_BIRTH_DATE as "Date of Birth",
          case 
            when O_PERSONS.PER_DECEASED_DATE is null then FLOOR(MONTHS_BETWEEN(sysdate,O_PERSONS.PER_BIRTH_DATE)/12) 
            else FLOOR(MONTHS_BETWEEN(O_PERSONS.PER_DECEASED_DATE,O_PERSONS.PER_BIRTH_DATE)/12) 
          end as "Age",
          O_PERSONS.PER_DECEASED_DATE as "Date Deceased",
          olm_bo.get_gender_desc('P', O_PERSONS.PER_GENDER) as "Gender",
          CASE 
            WHEN ETHNIC.CLA_CAT_ID IN ('C1','C2','C3','C4','ABAN','AIND','AOTH','APKN') THEN 'Asian or Asian British'
            ELSE NULL
          End as "Ethnicity - Top" ,
          CASE 
            WHEN ETHNIC.CLA_CAT_ID IN ('BAFR','D2') THEN 'African'
            ELSE NULL
          End as "Ethnicity - Detail" ,
          CASE 
            WHEN PCG.CLA_CAT_ID IN ('ASYLUM','REFUGEE') THEN 'Asylum Seeker/Refugee'
            ELSE NULL
          End as "PCG",
          CASE 
            WHEN PCG.CLA_CAT_ID IN ('ASYLUM','REFUGEE') THEN 'Asylum Seeker/Refugee'
            ELSE NULL
          End as "PCG - Top",
          CASE 
            WHEN PCG.CLA_CAT_ID IN ('ASYLUM','REFUGEE') THEN 'Asylum Seeker/Refugee'
            ELSE NULL
          End as "PCG - DETAIL",

          to_char(row_number() over(PARTITION BY  ETHNIC.CLA_SUBJECT_ID ORDER BY abs(sysdate - ETHNIC.CLA_DATE_NOTIFIED) asc)) as "Row Number E",
          to_char(row_number() over(PARTITION BY  PCG.CLA_SUBJECT_ID ORDER BY abs(sysdate - PCG.CLA_DATE_NOTIFIED) asc))as "Row Number P"

        FROM

          O_PERSONS

          LEFT JOIN O_CLASSIFICATIONS ETHNIC ON ETHNIC.CLA_SUBJECT_ID = O_PERSONS.PER_ID 
          AND ETHNIC.CLA_SUBJECT_IND = 'P'
          AND ETHNIC.CLA_TOP_CAT_ID   =  'ETHNIC'
          AND ETHNIC.CLA_CAT_ID <> 'DECLINED'

          LEFT JOIN O_CLASSIFICATIONS PCG ON PCG.CLA_SUBJECT_ID = O_PERSONS.PER_ID 
          AND PCG.CLA_SUBJECT_IND = 'P'
          AND PCG.CLA_TOP_CAT_ID   =  'PRIMARY'
          AND PCG.CLA_CAT_ID <> 'DECLINED'

          WHERE
          /*Following line excludes any clients whose is less than 18)*/
            O_PERSONS.PER_BIRTH_DATE > trunc(add_months(O_PERSONS.PER_BIRTH_DATE,-216))
            AND  O_PERSONS.PER_ID LIKE 'P%'

Answer 1

您应该在进行加入之前应用ROW_NUMBER() 。

SELECT
  *     -- Your calculations here, '*' used for brevity
FROM
  O_PERSONS
LEFT JOIN
(
  SELECT
    ROW_NUMBER() OVER (PARTITION BY CLA_SUBJECT_ID ORDER BY CLA_DATE_NOTIFIED DESC) AS sequence_id,
    *
  FROM
    O_CLASSIFICATIONS
  WHERE
        CLA_SUBJECT_IND =  'P'
    AND CLA_TOP_CAT_ID  =  'ETHNIC'
    AND CLA_CAT_ID      <> 'DECLINED'
)
  ETHNIC
    ON  ETHNIC.CLA_SUBJECT_ID = O_PERSONS.PER_ID
    AND ETHNIC.sequence_id    = 1
LEFT JOIN
(
  SELECT
    ROW_NUMBER() OVER (PARTITION BY CLA_SUBJECT_ID ORDER BY CLA_DATE_NOTIFIED DESC) AS sequence_id,
    *
  FROM
    O_CLASSIFICATIONS
  WHERE
        CLA_SUBJECT_IND =  'P'
    AND CLA_TOP_CAT_ID  =  'PRIMARY'
    AND CLA_CAT_ID      <> 'DECLINED'
)
  PCG
    ON  PCG.CLA_SUBJECT_ID = O_PERSONS.PER_ID
    AND PCG.sequence_id    = 1
WHERE
     /*Following line excludes any clients whose is less than 18)*/
     O_PERSONS.PER_BIRTH_DATE > trunc(add_months(O_PERSONS.PER_BIRTH_DATE,-216))
AND  O_PERSONS.PER_ID LIKE 'P%'