Question

我有一个名为1 Main Contacts（“Main”）的表，以及其他几个表。它们都有名为Contact_ID（或referral_ID）的列，它们将它们挂钩。所有设置方式的一个问题是“Main”中的记录可以链接到引用表中的多个记录，因此在运行查询以通过“Contact_Source”列提取联系人后，我得到重复记录推荐表。

我创建了一个视图，我可以从网站运行查询时从中选择，这样我就可以从与正确情况相关的数据中提取。我也通过他们的“联系人来源”运行这个查询。我之前发布了2个问题（here和here），以避免重复记录。我有那个工作。

这是我让我无法获得重复记录的最终代码：

ALTER VIEW dbo.v_angelview AS

WITH    q AS
        (
        SELECT  dbo.[1_MAIN - Contacts].Contact_ID, dbo.[1_MAIN - Contacts].Date_entered_into_Database, dbo.[1_MAIN - Contacts].Date_of_Initial_Contact, 
                      dbo.[1_MAIN - Contacts].[Company_ Name], dbo.[1_MAIN - Contacts].Key_Contact_Title, dbo.[1_MAIN - Contacts].Key_Contact_First_Name, 
                      dbo.[1_MAIN - Contacts].Key_Contact_Middle, dbo.[1_MAIN - Contacts].Key_Contact_Last_Name, dbo.[1_MAIN - Contacts].Key_Credential, 
                      dbo.[1_MAIN - Contacts].Key_Contact_Occupation, dbo.[1_MAIN - Contacts].Key_Degree_1, dbo.[1_MAIN - Contacts].Key_Degree_2, 
                      dbo.[1_MAIN - Contacts].Key_Degree_3, dbo.[1_MAIN - Contacts].Date_of_Highest_Degree, dbo.[1_MAIN - Contacts].Work_Setting, 
                      dbo.[1_MAIN - Contacts].Website_Address, dbo.[1_MAIN - Contacts].Email_1_Key_Contact, dbo.[1_MAIN - Contacts].Email_2, 
                      dbo.[1_MAIN - Contacts].Email_3, dbo.[1_MAIN - Contacts].Day_Time_Phone_Number, dbo.[1_MAIN - Contacts].Extension, 
                      dbo.[1_MAIN - Contacts].Mobile_Phone_Number, dbo.[1_MAIN - Contacts].Bus_Fax_Number, dbo.[1_MAIN - Contacts].Home_Phone_Number, 
                      dbo.[1_MAIN - Contacts].Home_Fax_Number, dbo.[1_MAIN - Contacts].Mailing_Street_1, dbo.[1_MAIN - Contacts].Mailing_Street_2, 
                      dbo.[1_MAIN - Contacts].Mailing_City, dbo.[1_MAIN - Contacts].Mailing_State, dbo.[1_MAIN - Contacts].[Mailing_Zip/Postal], 
                      dbo.[1_MAIN - Contacts].Mailing_Country, dbo.[1_MAIN - Contacts].[Bad_Address?], dbo.[1_MAIN - Contacts].[PROV/REG?], 
                      dbo.[1_MAIN - Contacts].status_flag, dbo.[1_MAIN - Contacts].status_flag AS status_flag2, dbo.Providers.Referral_Source, dbo.Referral.Contact_Source, 
                      dbo.Resource_Center.cert_start_date, dbo.Resource_Center.cert_exp_date, dbo.prov_training_records.Contact_ID AS Expr2, 
                      dbo.prov_training_records.date_reg_email_sent, dbo.Resource_Center.access, dbo.Providers.Contact_ID AS Expr1,
                ROW_NUMBER() OVER (PARTITION BY dbo.[1_MAIN - Contacts].Contact_ID ORDER BY dbo.[1_MAIN - Contacts].Contact_ID) AS rn
        FROM    dbo.[1_MAIN - Contacts]
        INNER JOIN
                dbo.Referral
        ON      dbo.[1_MAIN - Contacts].Contact_ID = dbo.Referral.Referral_ID
        INNER JOIN
                dbo.prov_training_records
        ON      dbo.[1_MAIN - Contacts].Contact_ID = dbo.prov_training_records.Contact_ID
        LEFT OUTER JOIN
                dbo.Resource_Center
        ON      dbo.[1_MAIN - Contacts].Contact_ID = dbo.Resource_Center.Contact_ID
        FULL OUTER JOIN
                dbo.Providers
        ON      dbo.[1_MAIN - Contacts].Contact_ID = dbo.Providers.Contact_ID
WHERE (dbo.[1_MAIN - Contacts].Mailing_State = N'AL') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'FL') OR

                      (dbo.[1_MAIN - Contacts].Mailing_State = N'GA') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'KY') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'MS') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'NC') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'SC') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'TN') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'PR') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'CO') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'MT') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'ND') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'SD') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'UT') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'WY') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'AR') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'LA') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'NM') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'OK') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'TX') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'AZ') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'CA') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'HI') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'ID') OR
                      (dbo.[1_MAIN - Contacts].Mailing_State = N'NV')
        )
SELECT  *
FROM    q
WHERE   rn = 1

现在我遇到的问题是我需要设置一个优先级，保留我保留的重复记录以及删除哪些记录。

这是从最先例开始的优先级层次结构：

Contact_Source ='PROVIDER'
Contact_Source LIKE'RG _％'
Contact_Source LIKE'IN _％'
Contact_Source LIKE'LD _％'

记录可以链接到其中一个或全部。因此，例如，如果记录有PROVIDER和RG_Train，我想用PROVIDER保留记录。等等列表。同样，所有记录都有一个Contact_ID，这就是我可以告诉它有重复的信息。

有没有办法修改现有的SQL来执行此操作，还是需要新方法？如果是这样，我如何根据我的优先级列表删除重复记录？

我正在使用SQL Server 2005。

提前致谢！

Answer 1

...试

rn = ROW_NUMBER() 
OVER
(
  PARTITION BY dbo.[1_MAIN - Contacts].Contact_ID
  ORDER BY 
  CASE 
    WHEN Contact_Source = 'PROVIDER'
    THEN 1

    WHEN Contact_Source LIKE 'RG_%'
    THEN 2

    WHEN Contact_Source LIKE 'IN_%'
    THEN 3

    WHEN Contact_Source LIKE 'LD_%'
    THEN 4

    ELSE 5
  END
)

Answer 2

这并没有利用你对ROW_NUMBER函数的使用，但我相信这会有效：

with q as (
    -- <query>
),
cp as (
    select  1 as Precedence, 'PROVIDER' ContactSourcePattern
    union all   select 2, 'RG_%'
    union all   select 3, 'IN_%'
    union all   select 4, 'LD_%'
)
select q.*
from q
inner join cp on q.Contact_Source like cp.ContactSourcePattern
where
    -- filter out duplicate records with the same `Contact_ID` that have a lower precedence than other records
    not exists (
        select 1
        from q as q2 inner join cp as cp2 on q2.Contact_Source like cp2.ContactSourcePattern
        where
            q2.Contact_ID = q.Contact_ID -- q2 is a duplicate of q if `Contact_ID` matches
            and cp2.Precedence < cp.Precedence -- q2/cp2 is higher precedence than q/cp if `Precedence` is a smaller number
    )

按层次结构筛选重复记录？

2 个答案: