复杂（ish）加入问题：

Question

我在 mySQL 数据库中有下表，（注意：这些是总结的，因此它们与这个问题相关，一些列被省略了）。

author (id, username, password etc.)
thread (id, title, content)
tag (id, name)
reply (id, content)
thread_replies (thread_id, reply_id)
author_replies (author_id, reply_id)
thread_tags (thread_id, tag_id)
author_threads (author_id, thread_id)

现在要获取某个作者的主题，我通常会这样做：

SELECT thread.title, thread.id AS thread_id, thread.content, author.username, author.id AS author_id
FROM thread
JOIN author_threads ON thread.id = author_threads.thread_id
JOIN author ON author_threads.author_id = author.id
WHERE author.id = '12'

这很好用，但是当我尝试获取与这些线程关联的标签时：

SELECT thread.title, thread.id AS thread_id, thread.content, author.username, author.id AS author_id, GROUP_CONCAT( DISTINCT tag.name
ORDER BY tag.name DESC
SEPARATOR ',' ) AS tags
FROM thread
JOIN thread_tags ON thread.id = thread_tags.thread_id
JOIN tag ON thread_tags.tag_id = tag.id
JOIN author_threads ON thread.id = author_threads.thread_id
JOIN author ON author_threads.author_id = author.id
WHERE author.id = '12'
LIMIT 0 , 30

它只在一个列中显示第一个线程，以及与该作者关联的所有标签。

我在这里做错了什么？

另外，如果我想计算作者对每个线程的回复数，该怎么做？

Answer 1

因为您在查询中使用聚合 (GROUP_CONCAT)，所以您的查询正在分组。由于您没有 group by 子句，因此您的组是整个结果集（因此可以看到作者使用的每个标签）。因为 MySQL 允许在分组语句中使用非分组列，所以您不会收到错误消息，但不会得到您想要的查询。

为了检索正确的结果，您需要将查询分组到 thread.id。

select
    thread.title, 
    thread.id as thread_id, 
    thread.content, 
    author.username, 
    author.id as author_id, 
    group_concat(distinct tag.name order by tag.name desc separator ',') as tags

from thread

join thread_tags ON thread.id = thread_tags.thread_id
join tag ON thread_tags.tag_id = tag.id
join author_threads ON thread.id = author_threads.thread_id
join author ON author_threads.author_id = author.id

where author.id = '12'

group by thread.id

limit 0 , 30

这应该在 MySQL 中工作，尽管它不符合 ANSI 的 SQL，因为您在 select 子句中使用非分组列而没有任何聚合。您可以保持原样，或者您可以编写更合规的 SQL 并在除thread.id 之外的所有列周围使用类似max 的内容。这看起来不那么漂亮，但它会合规。

SELECT 
    max(thread.title) as title, 
    thread.id as thread_id, 
    max(thread.content) as content, 
    max(author.username) as username, 
    max(author.id) as author_id, 
    group_concat(distinct tag.name order by tag.name desc separator ',') as tags

from thread

join thread_tags ON thread.id = thread_tags.thread_id
join tag ON thread_tags.tag_id = tag.id
join author_threads ON thread.id = author_threads.thread_id
join author ON author_threads.author_id = author.id

where author.id = '12'

group by thread.id

LIMIT 0 , 30

回复数

上述查询（连同您的原始查询）仅适用于检索标签列表。您可以编写一个等效的查询来检索回复计数（假设回复不是嵌套的，在这种情况下，您必须使用 MySQL 提供的任何递归查询功能，我不熟悉），但是要检索两者在单个查询中需要子查询：

select
    thread.title, 
    thread.id as thread_id, 
    thread.content, 
    author.username, 
    author.id, 
    (select group_concat(distinct tag.name order by tag.name separator ',')

    from thread_tags

    join tag on tag.id = thread_tags.tag_id 

    where thread_tags.thread_id = thread.id) as tags,
    (select count(1) from thread_replies where thread_id = thread.id) as reply_count

from thread

join author_threads ON thread.id = author_threads.thread_id
join author ON author_threads.author_id = author.id

where author.id = '12'

LIMIT 0 , 30

我已经从这个查询中删除了group by，因为我们的聚合已经被移动到一个子选择中，这意味着外部查询不再被分组。

Answer 2

我会先只在里面预查询线程和标签信息...然后，这将已经有作者和有效线程加入您需要的任何其他内容...

如果您确实想应用限制，请将其放在内部“PREQUERY”上，因为这将成为连接外部级别表的基础......否则，您将获得 100 或 1000 的内部查询加上条目加入其他表并切割成 30 条记录...让 IT 在 30 处停止，您就完成了..

在限制返回的条目时，您可能还希望按大多数当前线程排序。

select STRAIGHT_JOIN
      PreQuery.*,
      Author.username,
      Thread.title,
      Thread.Content

   from 
      ( select STRAIGHT_JOIN
              author_threads.author_id,
              author_threads.thread_id,
              group_concat(distinct tag.name order by tag.name desc separator ',') as tags
           from 
              author_threads
                 join thread_tags
                    on author_threads.thread_id = thread_tags.thread_id
                    join tag
                       on thread_tags.tag_id = tag.id
           where
              author_threads.author_id = '12'
           group by
              author_threads.author_id,
              author_threads.thread_id
           limit 0, 30 ) PreQuery

      join author
         on PreQuery.Author_ID = author.id

      join thread
         on PreQuery.Thread_id = thread.id