一种更好的方式来处理下面的SQL查询过滤条件

Question

考虑下面的SQL查询

INSERT INTO [dbo].[PartnerCommissionData] 
SELECT X.* 
FROM 
(
     SELECT 
       cs.partner_id 
       ,cs.quarter 
       ,cs.year 
       ,cs.partner_currency_amount 
       ,p.partner_email       
       ,cs.report_status_id  
       ,Creation_dt = GETDATE()  
     FROM [dbo].[CommissionSummary] cs WITH(NOLOCK) 
     INNER JOIN [dbo].[Partner] p WITH(NOLOCK) ON p.partner_id = cs.partner_id 
     WHERE 
       cs.year = YEAR(@FirstDayOfQuarter) 
     AND  cs.quarter = @Quarter 
     AND  cs.report_status_id IN (1,2) 
    )X 
WHERE 
    X.partner_id NOT IN (SELECT [Partner_Id] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 
AND X.quarter NOT IN (SELECT [Quarter] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 
AND X.year NOT IN (SELECT [Year] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 
AND X.report_status_id NOT IN (SELECT [CommissionStatus_id] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 

指定过滤条件表明，如果已经存在于PartnerCommissionData表中的记录，那么它不应该被进一步插入。但按照我的看法，这是一种恶劣的实施。

WHERE 
    X.partner_id NOT IN (SELECT [Partner_Id] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 
AND X.quarter NOT IN (SELECT [Quarter] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 
AND X.year NOT IN (SELECT [Year] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 
AND X.report_status_id NOT IN (SELECT [CommissionStatus_id] FROM [dbo].[PartnerCommissionData] WITH (NOLOCK)) 

我们怎样才能以更好的方式改写它？

我们可以应用合并或任何其他方式....？

预先感谢

Answer 1

你能做这样的事吗？不需要离开连接（扫描另一张桌子）。 承担PARTNER_ID，季，年，report_status_id需要转换为nvarchar

INSERT INTO [dbo].[PartnerCommissionData] 
    SELECT X.* 
    FROM 
    (
      SELECT 
        cs.partner_id 
        ,cs.quarter 
        ,cs.year 
        ,cs.partner_currency_amount 
        ,p.partner_email       
        ,cs.report_status_id  
        ,Creation_dt = GETDATE()  
      FROM [dbo].[CommissionSummary] cs WITH(NOLOCK) 
      INNER JOIN [dbo].[Partner] p WITH(NOLOCK) ON p.partner_id = cs.partner_id 
      WHERE 
        cs.year = YEAR(@FirstDayOfQuarter) 
      AND  cs.quarter = @Quarter 
      AND  cs.report_status_id IN (1,2) 
     )X 
    WHERE 
    (x.partner_id + x.quarter + x.year + x.report_status_id) <> 
    (partner_id + x.quarter + x.year + x.report_status_id) 

Answer 2

执行连接，而不是内选择。这是非常混乱的SQL。考虑一套而不是程序。

Answer 3

所以，如果我的理解是正确的，你避免将可能有重复的数据，我改变了你的SQL的东西如下：

INSERT INTO [dbo].[PartnerCommissionData] 
SELECT X.* 
FROM 
(
     SELECT 
       cs.partner_id 
       ,cs.quarter 
       ,cs.year 
       ,cs.partner_currency_amount 
       ,p.partner_email       
       ,cs.report_status_id  
       ,Creation_dt = GETDATE()  
     FROM [dbo].[CommissionSummary] cs WITH(NOLOCK) 
     INNER JOIN [dbo].[Partner] p WITH(NOLOCK) ON p.partner_id = cs.partner_id 
     WHERE 
       cs.year = YEAR(@FirstDayOfQuarter) 
     AND  cs.quarter = @Quarter 
     AND  cs.report_status_id IN (1,2) 
    )X 
    LEFT JOIN [PartnerCommissionData] PCD WITH (NOLOCK)) 
      ON X.partner_id = PCD.[Partner_Id] OR X.quarter = PCD.[Quarter] 
       OR X.year = PCD.[Year] OR X.report_status_id = PCD.[CommissionStatus_id] 
WHERE PCD.ID IS NULL 

我想在这里做的事情是第一个LEFT JOIN该表并尝试抓取任何数据（如果它们存在），然后使用IS NULL检查来过滤掉任何现有数据，以便我们不需要NOT IN，其性能不佳并且不被推荐。