Question

我正在使用Firebird数据库v2.5，我无法从查询中删除重复的行：

SELECT  DISTINCT u.id_user, 
        (SELECT LIST(g.id_user) 
         from gen g 
         where g.id_user=u.id_user 
         GROUP BY id_user) as list_g_user
FROM users u 
where  u.id_user = 1 
  INNER JOIN ...

这是我的结果：

 id_user  |  list_g_user 
===================================
1         |  437,499,718,739,835,865
1         |  437,499,718,739,835,865
1         |  437,499,718,739,835,865
1         |  437,499,718,739,835,865

第二列结果list_g_user是BLOB类型。

相同的结果和多行。有人可以解释原因吗？

这是功能参考：Firebird Documentation: LIST()

Answer 1

正如JNevill distinct已经指出blob does not behave as expected in Firebird：它会比较blob id（指向blob的'指针'），而不是Blob的内容，{ {3}}生成一个blob。

要解决此问题，有几种可能的解决方案：

将list()的结果投放到varchar，以便可以正确比较，例如：
```
SELECT cast(LIST(g.id_user) as varchar(100)) ...
```
但这确实意味着列表的字符长度不应超过varchar长度（本例中为100）。

您在查询中使用inner join而不使用联接表中的任何内容，似乎建议您仅使用联接作为存在检查。用exists支票替换该联接：

SELECT u.id_user, 
        (SELECT LIST(g.id_user) 
         from gen g 
         where g.id_user=u.id_user 
         GROUP BY id_user) as list_g_user
FROM users u 
where  u.id_user = 1 
and exists (select * from <currently joined table> x where x.id_user = u.id_user)

正如JNevill在评论中所建议的那样，您可以尝试在顶级查询中使用group by id_user，但这可能并不总是有效，因为它依赖于相邻的blob ID。在无法使用group by的情况下，它不是一个可行的解决方案：
```
SELECT  DISTINCT u.id_user, 
        (SELECT LIST(g.id_user) 
         from gen g 
         where g.id_user=u.id_user 
         GROUP BY id_user) as list_g_user
FROM users u 
where  u.id_user = 1 
  INNER JOIN ...
group by u.id_user
```

Answer 2

SELECT  DISTINCT u.id_user,
        CAST(SUBSTRING((SELECT LIST(g.id_user)
         from gen g
         where g.id_user=u.id_user
         GROUP BY id_user) FROM 1 FOR 1000) AS VARCHAR(1000)) as list_g_user
FROM users u
where  u.id_user = 1
  INNER JOIN ...

Firebird删除SELECT LIST子查询

2 个答案: