带有 2 个连接和计数的 Sql 查询和分组依据

Question

I want to use an join to list the car colors count, car type, and users name.

I have 3 table

Table 1 Useres

id|username|fullname
1 | test0  | xy xy
2 | test1  | yx yx

Table 2 Car Type

id|car_type|user_id
1 | Ford   | 1
2 | BMW    | 2
3 | Ford   | 1
4 | Skoda  | 1
5 | BMW    | 2

Table 3 Car Color

id| Color  |user_id|car_id
1 | Red    | 1     |1
2 | Blue   | 2     |2
3 | Red    | 2     |5
4 | Red    | 1     |3
5 | Red    | 1     |4
6 | Green  | 1     |4

One car has 2 color

The result should be:

countType | CountColor  | UserName
   3      |    4        | test0
   2      |    2        | test1

I tryed this:

   SELECT 
   test as BlogsPost, 
   test2 as CommenstPost,
   u.name   
   FROM users u 
   LEFT JOIN (
       select COUNT(blogs.user_id) as test FROM blogs GROUP by blogs.user_id)  blogs
    on blogs.user_id=u.id 
   LEFT JOIN (
       select COUNT(comments.user_id) as test2 FROM comments GROUP by comments.user_id) comments
    on comments.user_id=u.id 
   GROUP by users.id

Answer 1

I agree with the comment that states the table design is not really well constructed yet for you to achieve the counts you want you will need to do subqueries like this:

SELECT 
(SELECT count(1) from CarType where user_id=username) as countType,
(SELECT count(1) from CarColor where user_id=username) as countColor,
username from (
SELECT username from Users
) a 

As a suggestion for design:

Table Users

Table Cars

Table Colors

Then you have a Relationship table where you have user_id, car_id, color_id

This would be the proper table design for this structure

Answer 2

If I understand your question correctly with reference to your actual code section what you want is a list of users with how many blogs they have and how many comments they have. Now if you were wanting to count one matching table you could just do this:

SELECT 
   U.NAME
   ,COUNT(1) AS BLOG_COUNT
FROM USERS U
LEFT JOIN BLOGS B
   ON B.USER_ID = U.ID
GROUP BY U.NAME

But since you are wanting to count two tables you have to do it slightly differently. There's a few ways of doing it but the way I like is like this:

SELECT 
   U.NAME
   ,B.BB_COUNT AS BLOG_COUNT
   ,C.CC_COUNT AS COMMENT_COUNT
FROM USERS U
LEFT JOIN 
(
   SELECT 
      BB.USER_ID
      ,COUNT(1) AS BB_COUNT
   FROM BLOGS BB
   GROUP BY BB.USER_ID
) B
   ON B.USER_ID = U.ID

LEFT JOIN 
(
   SELECT 
      CC.USER_ID
      ,COUNT(1) AS CC_COUNT
   FROM COMMENTS CC
   GROUP BY CC.USER_ID
) C
   ON C.USER_ID = U.ID

That may or may not be the most efficient way but in my experience it works pretty well and it's simple to understand. It all depends a lot on the number of rows in the tables and indexes etc. Usually the idea is to narrow down rows returned as fast as possible. In this case you'll have two sub queries but they'll end up with only as many rows as you have users basically.

Another thing to note, this will return all users, period. That may not be what you want. You might want only a subset of users. If so this inner select may not be the most efficient because you're doing calculations on users that may not be in the final result, wasting time. However I may be getting off topic.