Question

我有一个邮件列表，有许多条目是重复的。我试图找到重复项，以便删除它们。当我运行下面的完整查询时，我得到表中的所有项目（142,000 +）。当我运行子查询时，我只得到5768行。我需要每个重复行的所有列来决定删除哪一行。我做错了什么导致完整查询返回所有行？

select * from Mailinglist
where exists
(select count(*), mailaddress, CenterName
from Mailinglist
group by MailAddress, CenterName
having count(*)>1)

Answer 1

你必须这样做：

select t1.*, t2.cnt 
from Mailinglist t1
join (
  select count(*) as cnt, mailaddress, CenterName
  from Mailinglist
  group by MailAddress, CenterName
  having count(*)>1
) t2 ON t1.MailAddress = t2.MailAddress and t1.CenterName = t2.CenterName

使用EXISTS只检查记录是否存在：如果子查询返回一个或多个记录，则EXISTS计算为true。

Answer 2

如果后面的查询中只有一条记录，则

EXISTS返回true。您正在寻找的是

select * from Mailinglist
where mailaddress IN 
(
    select mailaddress
    from Mailinglist
    group by MailAddress, CenterName
    having count(*)>1
)

Answer 3

这是因为如果子查询返回一行，EXISTS将返回true。你的子查询，返回一行或多行，从而为TRUE条件返回EXISTS。

要获得带有重复项的MailingList，您只需运行子查询：

SELECT
    COUNT(*),
    mailaddress,
    CenterName
FROM Mailinglist
GROUP BY
    MailAddress, CenterName
HAVING COUNT(*) > 1

要删除重复项，您可以使用ROW_NUMBER：

WITH Cte AS(
    SELECT *,
        rn = ROW_NUMBER() OVER(PARTITION BY MailAddress, Centername ORDER BY (SELECT NULL))
    FROM MailingList
)
DELETE FROM Cte WHERE rn > 1

只需替换ORDER BY子句，具体取决于您要保留的重复项的哪一行。

Answer 4

子查询中没有过滤（在哪里），所以它总会产生一些结果。

SELECT        *
FROM            Mailinglist AS ML
WHERE        EXISTS
        (SELECT        COUNT(*) AS Expr1, mailaddress, CenterName
        FROM            Mailinglist AS CNT
        WHERE        (ML.MailAddress = CNT.MailAddress) AND (ML.CenterName = CNT.CenterName)
        GROUP BY mailaddress, CenterName
        HAVING         (COUNT(*) > 1))

SQL exists返回所有行

4 个答案: