18

桌子：

(`post_id`, `forum_id`, `topic_id`, `post_time`) 
(79, 8, 4, '2012-11-19 06:58:08');
(80, 3, 3, '2012-11-19 06:58:42'),
(81, 9, 9, '2012-11-19 06:59:04'),
(82, 11, 6, '2012-11-19 16:05:39'),
(83, 9, 9, '2012-11-19 16:07:46'),
(84, 9, 11, '2012-11-19 16:09:33'),

查询：

SELECT  post_id, forum_id, topic_id FROM posts 
GROUP BY topic_id 
ORDER BY post_time DESC
LIMIT 5

结果：

[0] => [post_id] => 84 [forum_id] => 9 [topic_id] => 11  
[1] => [post_id] => 82 [forum_id] => 11 [topic_id] => 6  
[2] => [post_id] => 81 [forum_id] => 9 [topic_id] => 9  
[3] => [post_id] => 80 [forum_id] => 3 [topic_id] => 3  
[4] => [post_id] => 79 [forum_id] => 8 [topic_id] => 4

问题：

如何重写查询以使其返回 post_id -> 83 而不是 post_id -> 81 ？

他们都有相同的论坛和主题 ID，但 post_id -> 81 的日期比 post_id -> 83 的日期更早。

但似乎 Group By 获得了“第一”记录，而不是“最新”记录。

我尝试将查询更改为

SELECT  post_id, forum_id, topic_id, MAX(post_time)

但这会返回 post_id 81 和 83

4

5 回答 5

27

如果您选择未在 group 子句中使用且不是聚合的属性，则结果未指定。即，您不知道从哪些行中选择了其他属性。（sql标准不允许这样的查询，但是MySQL比较宽松）。

然后应该将查询编写为，例如

SELECT post_id, forum_id, topic_id
FROM posts p
WHERE post_time =
  (SELECT max(post_time) FROM posts p2
   WHERE p2.topic_id = p.topic_id
   AND p2.forum_id = p.forum_id)
GROUP BY forum_id, topic_id, post_id
ORDER BY post_time DESC
LIMIT 5;

或者

SELECT post_id, forum_id, topic_id FROM posts
NATURAL JOIN
(SELECT forum_id, topic_id, max(post_time) AS post_time
 FROM posts
 GROUP BY forum_id, topic_id) p
ORDER BY post_time
LIMIT 5;
于 2012-11-19T19:12:38.447 回答
12

它不是很漂亮，但它有效：

SELECT * FROM (SELECT  post_id, forum_id, topic_id FROM posts
ORDER BY post_time DESC) as temp
GROUP BY topic_id
于 2012-11-19T18:15:27.397 回答
4

尝试类似的东西

SELECT post_id, forum_id, topic_id 
FROM   (SELECT post_id, forum_id, topic_id
        FROM posts
        ORDER BY post_time DESC) 
GROUP BY topic_id 
ORDER BY topic_id desc
LIMIT 0,5

根据需要更改order by和limit。

于 2012-11-19T18:15:36.633 回答
0

也许不是最好的方法，但有时函数group_concat()可以是全用户的，它会返回一个包含所有聚合值的字符串，这些值按您想要的方式排序并用逗号分隔（耦合值用空格分隔）。然后我使用函数 SPLIT_STRING() 来剪切字符串中的第一个 id。

SELECT  
post_id, 
SPLIT_STRING( group_concat( forum_id, post_time ORDER BY post_time DESC ) ,' ',1 )as forum_id, 
SPLIT_STRING( group_concat( topic_id, post_time ORDER BY post_time DESC ) ,' ',1 )as topic_id ,
FROM posts 
GROUP BY topic_id 
ORDER BY post_time DESC
LIMIT 5

所以聚合的 forum_id, post_time 将是这样的：

81 2012-11-19 06:59:04,83 2012-11-19 16:07:46

因此，您需要使用整数和日期时间对的字符串表示形式，每对以逗号分隔，因此我使用此函数来获取第一个 INT ：

CREATE FUNCTION SPLIT_STRING(str VARCHAR(255), delim VARCHAR(12), pos INT)
RETURNS VARCHAR(255)
RETURN REPLACE(SUBSTRING(SUBSTRING_INDEX(str, delim, pos),
       LENGTH(SUBSTRING_INDEX(str, delim, pos-1)) + 1),
       delim, '');

注意：函数 SPLIT_STRING(str, delim, pos) 在这里找到：Equivalent of explode() to work with strings in MySQL

于 2015-05-28T19:06:43.950 回答
0

这也适合你。

SELECT *
FROM (
  SELECT post_id, forum_id, topic_id FROM posts
  ORDER BY post_time DESC
  LIMIT 5
) customeTable
GROUP BY topic_id
于 2015-07-22T06:21:07.963 回答