防止随机行和重复的重复

Posted 2023-03-24

【中文标题】防止随机行和重复的重复

【英文标题】：Preventing duplicates with random rows and duplicates

【发布时间】：2014-09-15 18:31:35

【问题描述】：

我有一个数据集，我需要随机选择满足特定条件的 x 行、满足其他条件的 x 行等，并将结果全部输出到一个表中。我一直在使用以下内容：

SELECT * FROM (SELECT TOP 1000 * FROM dbo.[Client List]
WHERE Source = 'HC' AND Service = 'Service1' AND Provider = 'ProviderName1'
ORDER BY NEWID()) Table1

UNION ALL

SELECT * FROM (SELECT TOP 500 * FROM dbo.[Client List]
WHERE Source = 'HC' AND Service = 'Service2' AND Provider = 'ProviderName2'
ORDER BY NEWID()) Table2

UNION ALL

SELECT * FROM (SELECT TOP 2200 * FROM dbo.[Client List]
WHERE Source = 'BA' AND Service = 'Service3' AND Provider = 'ProviderName3'
ORDER BY NEWID()) Table3

这可行，但有一个问题：它允许重复的客户端标识符 (dbo.[Client List].[ClientID] 被选中。我所有的结果都必须是唯一的客户端。

换句话说，它必须进行第一次随机选择，然后进行第二次随机选择，而不能选择在第一次选择中选择的任何 ClientID，依此类推。 （是的，我意识到这在技术上不是“随机的”。）有没有办法可以在每个后续 SELECT 的 WHERE 语句中添加某种代码，或者我是否需要重新考虑代码的整个结构？谢谢！

【参考方案1】：

这样的事情怎么样？

with Service1 as
(
    SELECT TOP 1000 * 
    FROM dbo.[Client List]
    WHERE Source = 'HC' 
        AND Service = 'Service1' 
        AND Provider = 'ProviderName1'
    ORDER BY NEWID()
)
, Service2 as
(
    SELECT TOP 500 * 
    FROM dbo.[Client List]
    WHERE Source = 'HC' 
        AND Service = 'Service2' 
        AND Provider = 'ProviderName2'
        AND ClientID not in (select ClientID from Service1)
    ORDER BY NEWID()
)
, Service3 as
(
    SELECT TOP 2200 * 
    FROM dbo.[Client List]
    WHERE Source = 'BA' 
        AND Service = 'Service3' 
        AND Provider = 'ProviderName3'
        AND ClientID not in (select ClientID from Service1)
        AND ClientID not in (select ClientID from Service2)
    ORDER BY NEWID()
)

select * 
from Service1

union all

select *
from Service2

union all

select *
from Service3