将多行合并为一行，在 SQL Server 中的两列之间交换具有相同值

如何解决将多行合并为一行，在 SQL Server 中的两列之间交换具有相同值

我正在开发基于 API 的简单聊天模块。我正在尝试为特定用户获取聊天对话，但由于具有相同值的 2 列彼此交换导致我的数据被复制。

我想合并具有相同值的行在 2 列之间交换，并且合并行应基于插入到数据库中的最新条目。

数据如下：

Id  To  From   Message        ConversationTime
1   1    2     hello              11:00AM
2   3    1     hi                 12:00PM
3   1    3     how are you?       12:15PM
4   3    1     I am fine.         12:30PM
5   4    5     Hi!                04:30PM
6   5    4     Hello              04:35PM
7   1    5     Hola!              06:30PM

例如，如果用户 ID 为 1，我的结果需要如下所示：

Id  To  From   Message        ConversationTime
1   1    2     hello              11:00AM
4   3    1     I am fine.         12:30PM
7   1    5     Hola!              06:30PM

如果 Id 是 5，那么结果将是这样的：

Id  To  From   Message        ConversationTime
6   5    4     Hello              04:35PM
7   1    5     Hola!              06:30PM

我的结果集如下所示：

Id  To  From   Message        ConversationTime
1   1    2     hello              11:00AM
3   1    3     how are you?       12:15PM
4   3    1     I am fine.         12:30PM
7   1    5     Hola!              06:30PM

任何帮助将不胜感激。提前致谢！

解决方法

想法是与链接的副本Get top 1 row of each group 相同；只需使用 CASE 表达式来获取 other 用户的 ID：

DECLARE @ID int = 1;
WITH RNs AS(
    SELECT ID,[To],--TO is a reserved keyword and should not be used for object names
           [From],--FROM is a reserved keyword and should not be used for object names
           Message,ConversationTime,--I assume this is a time
           ROW_NUMBER() OVER (PARTITION BY CASE [To] WHEN @ID THEN [From] ELSE [To] END ORDER BY ConversationTime DESC) AS RN --TO and FROM are reserved keywords and should not be used for object names
    FROM dbo.YourTable
    WHERE @ID IN ([To],[From])) --TO and FROM are reserved keywords and should not be used for object names
SELECT ID,--TO is a reserved keyword and should not be used for object names
       [From],--FROM is a reserved keyword and should not be used for object names
       Message,ConversationTime --I assume this is a time
FROM RN
WHERE RN = 1;

SQL Server 允许您在不使用 case 表达式的情况下通过反透视数据然后使用窗口函数来执行此操作：

select t.*
from (select t.*,row_number() over (partition by v.user_other order by t.conversationTime desc) as seqnum
      from t cross apply
           (values (t.to,t.from),(t.from,to.to)
           ) v(user,user_other)
      where v.user = 1
     ) t
where seqnum = 1;