一个表中的MYSQL SUM（）并与其他2个表中的JOIN

Question

I have 3 tables: items , purchases , and collaborators . A user can own an item, purchase an item, or be a collaborator on an item. Additionally, items that are purchased can be rated up, +1 , or down, -1 . An owner or collaborator can't purchase their own item.

I'd like to get all items for a given user and also display the ratings on each item.

Here's my tables:

      items           |           purchases           |     collaborators
i_id  item_id user_id | p_id item_id  user_id  rating |c_id item_id user_id
  1      1       11   |  1      1         13     -1   |  1     1        12 
  2      2       12   |  2      2         11      1   |  2     2        13
  3      3       13   |  3      3         12     NULL |    
                      |  4      1         14     -1   |

Here's my MYSQL query so far:

select *, count(p_id) as tots, sum(rating=1) as yes, sum(rating= '-1') as no
from items
left join purchases
on items.item_id=purchases.item_id
left join collaborators
on items.item_id=collaborators.item_id
where items.user_id=13 or purchases.user_id=13 or collaborators.user_id=13
group by items.item_id

Here's my expected results for user_id=11 (changing each user_id in the WHERE clause):

item_id    tots  yes no
  1         2     0   2
  2         1     1   0
// notice how user_id=11 doesn't have anything to do with item_id=3

Here's my expected results for user_id=12 :

item_id    tots  yes no
  1         2     0   2
  2         1     1   0    
  3         1     1   0  

Here's my expected results for user_id=13 :

item_id    tots  yes no
  1         2     0   2
  2         1     1   0    
  3         1     1   0   
//notice user_id=13 should have same results as user_id=12. Although, their 
relation to each of the 3 items is different, they still either purchased, 
own, or collaboratored on each of them.

Unfortunately, I get the first two results but not the correct one for user_id=13 . For user_id=13, item_id=1 the tots=1 and not tots=2 for some reason I can't understand.

Any thoughts, such as, "its better to separate this into 2 queries", would be greatly appreciated,

Answer 1

I'm still not entirly sure I understand you correct but you could try following statement and let us work from there.

Edit

Following statement returns the expected results.
You can verify this (using SQL Server) here .

The gist of this is to

• select all possible user_id and item_id combinations from your three tables
• select the counts/ratings for each item
• combine the results

SQL Statement

SELECT u.user_id, pt.item_id, pt.cnt, pt.yes, pt.no
FROM   (
        SELECT user_id, item_id, title FROM items 
        UNION SELECT user_id, item_id, NULL FROM purchases
        UNION SELECT user_id, item_id, NULL FROM collaborators      
       ) u INNER JOIN (
        SELECT COUNT(*) AS cnt
               , SUM(CASE WHEN ISNULL(rating, 1) = 1 THEN 1 ELSE 0 END) AS yes
               , SUM(CASE WHEN rating =-1 THEN 1 ELSE 0 END) AS no
               , item_id
        FROM   purchases
        GROUP BY
               item_id
      ) pt ON pt.item_id = u.item_id    

MYSQL statement

SELECT u.user_id, pt.item_id, pt.cnt, pt.yes, pt.no, u.title
FROM   (
    SELECT user_id, item_id, title FROM items where user_id=13
    UNION SELECT user_id, item_id, NULL FROM purchases where user_id=13
    UNION SELECT user_id, item_id, NULL FROM collaborators where user_id=13       
   ) u INNER JOIN (
    SELECT COUNT(*) AS cnt
           , SUM(rating=1) AS yes
           , SUM(rating =-1) AS no
           , item_id
    FROM   purchases 
    GROUP BY
           item_id
  ) pt ON pt.item_id = u.item_id