Complex SQL join 根据给定结构查询

Complex SQL join Query according to given structure

我有两个 table postscategory_relationships。我需要根据以下逻辑从中得到有点复杂的结果集。

个帖子table

id   |      post  
-----|------------------------------|
1000 | Lorem ipsum dolor sit amet   | 
1001 | consectetur adipiscing elit  | 
1002 | sed do eiusmod tempor  ut    | 
1004 | abore et dolore magna aliqua | 

category_relationships table

post_id    cat_id  
---------|---------|
   1000  |   201   | 
   1000  |   202   | 
   1000  |   211   | 
   1001  |   201   | 
   1001  |   211   | 
   1002  |   202   | 
   1002  |   212   | 

首先我将尝试解释我的类别结构。我有如下的三层结构。 (为简单起见,我不包括任何类别table)

flight [level 1] [ID : 100]
    - class      [level 2] [ID : 200]
        -- economy  [level 3] [ID : 201]
        -- business [level 3] [ID : 202]
        -- first    [level 3] [ID : 203]
    - alliance   [level 2] [ID : 210]
        -- star     [level 3] [ID : 211]
        -- oneworld [level 3] [ID : 212]
        -- skyteam  [level 3] [ID : 213]

现在算法:

我需要根据以下规则将所有帖子标记为 flight 类别或任何 child/children。

我需要排除标记为 economy (ID:201);

的帖子
  1. 但是,如果它的兄弟姐妹之一(businessfirst)被标记,它应该仍然在结果集中。
  2. 不应考虑 alliance 或其 child/children 被标记的帖子 IF economy 也被标记在相同的帖子

请注意,我可以根据结构获取类别 ID 并在查询中使用它们。

我目前的做法:

SELECT posts.ID FROM posts  
LEFT JOIN category_relationships AS tt1 ON (posts.ID = tt1.post_id) 
WHERE tt1.cat_id IN (100,200,201,202,203,210,211,212,213) 
AND posts.ID NOT IN ( SELECT post_id FROM category_relationships WHERE cat_id IN (201) ) 

但这里的问题是它正在删除所有标记为 economy 的帖子。但是,它不满足规则编号 1。

理想的结果集如下所示;

1000 - rule number 1
1002 - anyway no `economy` tagged

不包括:

1001 - rule number 2
1004 - no tagged

希望您对这个问题有清晰的认识,如有任何帮助,我们将不胜感激。

将您的条件从 where 子句移动到 ON 子句

SELECT posts.ID FROM posts  
LEFT JOIN category_relationships AS tt1 ON (posts.ID = tt1.post_id) 
and tt1.cat_id IN (100,200,201,202,203,210,211,212,213) 
AND posts.ID NOT IN ( SELECT post_id FROM category_relationships WHERE cat_id IN (201) ) 

所以你的条件之一是"It has not category 201 OR it has category 202 or 203"。您缺少 OR 条件 OR tt1.cat_id IN (202, 203):

SELECT DISTINCT posts.ID FROM posts  
JOIN category_relationships AS tt1 ON (posts.ID = tt1.post_id) 
WHERE tt1.cat_id IN (100,200,201,202,203,210,211,212,213) 
AND (
    posts.ID NOT IN ( SELECT post_id FROM category_relationships WHERE cat_id IN (201) ) 
  OR
    tt1.cat_id IN (202, 203)
  )

请注意,您的 LEFT JOIN 没有任何意义,将由引擎转换为 INNER JOIN。

但是 - 我将按以下方式编写查询:

SELECT posts.ID
FROM posts  
JOIN category_relationships AS tt1 ON posts.ID = tt1.post_id
WHERE tt1.cat_id IN (100,200,201,202,203,210,211,212,213) 
GROUP BY posts.ID
HAVING SUM(tt1.cat_id = 201) = 0
    OR SUM(tt1.cat_id = 202) > 0
    OR SUM(tt1.cat_id = 203) > 0

这是 group byhaving 的良好候选者:

SELECT cr.post_id
FROM category_relationships cr
GROUP BY cr.post_id
HAVING SUM(cr.tag_id = 100) > 0 AND        -- flight
       (SUM(cr.tag_id = 201) > 0 OR        -- economy  
        SUM(cr.tag_id IN (202, 203)) > 0   -- business/first
       ) AND
       NOT (SUM(cr.tag_id = 201) > 0 OR    -- economy 
            SUM(cr.tag_id IN (210, 211, 212, 213) = 0  -- alliance
           );

使用标记可能更容易理解:

SELECT cr.post_id
FROM (SELECT cr.*,
             (cr.tag_id = 100) as is_flight,
             (cr.tag_id = 201) as is_economy,
             (cr.tag_id in (202, 203)) as is_first_business,
             (cr.tag_id IN (210, 211, 212, 213)) as is_alliance
      FROM category_relationships cr
     ) cr
GROUP BY cr.post_id
HAVING SUM(is_flight) > 0 AND
       (SUM(is_economy) > 0 OR  
        SUM(is_business_first) > 0
       ) AND
       NOT (SUM(is_economy) > 0 OR  
            SUM(is_alliance) > 0 
           );