Question

可能重复：
How to execute an UPDATE only if one row would be affected?

我在SQL Server 2005中有一个更新查询

update custom_graphics_files 
set actual_file_Name = @CorrectName 
where actual_file_Name = @FileName

现在，如果有多个actual_file_name，我想跳过更新查询

Answer 1

update t
set t.actual_file_Name = @CorrectName 
FROM custom_graphics_files t
INNER JOIN
(
   SELECT actual_file_Name, COUNT(*) TheCount
   FROM custom_graphics_files 
   GROUP BY actual_file_Name
) t2 ON t.actual_file_Name = t2.actual_file_Name AND TheCount = 1
where t.actual_file_Name = @FileName;

Answer 2

我喜欢为此目的使用窗口函数：

with toupdate as (
      select cgf.*, COUNT(*) over (PARTITION by actual_file_name) as ActCnt
      from custom_graphics_files
     )
update toupdate
    set actual_file_Name = @CorrectName 
    where actual_file_Name = @FileName and ActCnt = 1

在大型桌面上，这可能不是最有效的解决方案，具体取决于actual_file_Name = @FileName的选择性。

Answer 3

这是您可以获得的最“可读”的查询：

update custom_graphics_files 
set actual_file_Name = @CorrectName 
where actual_file_Name = @FileName
and (select count(1) from custom_graphics_files where actual_file_Name = @FileName) = 1

如果更新查询影响多行，则跳过更新查询

3 个答案: