MySQL-按计数分组

Question

My MySQL query is :

SELECT 
CASE 
    WHEN count(`image`.`context_uuid`) = 0 THEN 'No image' 
    WHEN count(`image`.`context_uuid`) = 1 THEN '1 image' 
    WHEN count(`image`.`context_uuid`) = 2 THEN '2 images' 
    WHEN count(`image`.`context_uuid`) = 3 THEN '3 images' ELSE '4 and + images' 
END AS `Informations`, count(`exchange_product`.`exchange_product_id`) AS `total`
FROM `exchange_product`
INNER JOIN `product` ON `product`.`product_id` = `exchange_product`.`product_id`
INNER JOIN `image` ON `image`.`context_uuid` = `product`.`product_uuid`
GROUP BY 
CASE 
    WHEN count(`image`.`context_uuid`) = 0 THEN 'No image' 
    WHEN count(`image`.`context_uuid`) = 1 THEN '1 image' 
    WHEN count(`image`.`context_uuid`) = 2 THEN '2 images' 
    WHEN count(`image`.`context_uuid`) = 3 THEN '3 images' ELSE '4 and + images' 
END

In my DB, I have a table :

image

| image ID     | url         | context_type | context_uuid  |
|--------------|-------------|--------------|---------------|
| 1            | www.az.com  | user         | 19            |
| 2            | www.az.com  | product      | 27            |
| 3            | www.az.com  | product      | 27            |
| 4            | www.az.com  | product      | 28            |

exchange_product

| exchange_product_id | owner_id | product_id | receiver_id |
|---------------------|----------|------------|-------------|
| 1                   | 23       | 27         | 19          |
| 2                   | 38       | 28         | 19          |
| 3                   | 94       | 92         | 90          |

product

| product_id   | user_id     | name       | product_uuid |
|--------------|-------------|------------|--------------|
| 1            | 23          | something  | 28           |
| 2            | 38          | something  | 12           |
| 3            | 94          | something  | 23           |

I want to have a result like this but I can't group by count of image...

| Informations | total |
|--------------|-------|
| No image     | 2563  |
| 1 image      | 1029  |
| 2 images     | 567   |
| 3 images     | 180   |
| 4 + images   | 1928  |

I want to know the count of exchanges of product that have :

• 0 image
• 1 image
• 2 images
• 3 images
• 4+ images

My current output when I GROUP BY image.context_uuid is :

| Informations | total |
|--------------|-------|
| No image     | 2563  |
| 1 image      | 1     |
| 1 image      | 3     |
| 1 image      | 1     |
| 1 image      | 2     |

Answer 1

• Use Left Join on image table, in order to consider those cases also, where there is no image (no matching row).
• You need to first count the number of images for every product_id specifically.
• Now, use these count results as a Derived Table , and Count the product_id(s) depending on which category they lie in ('No image' to '4 and + images')
• Note that in MySQL , the Alias defined in the Select clause can be used as it is, in the Group By , Having and Order By clauses.

Try the following:

SELECT 
  CASE 
    WHEN dt.images_count = 0 THEN 'No image' 
    WHEN dt.images_count = 1 THEN '1 image' 
    WHEN dt.images_count = 2 THEN '2 images' 
    WHEN dt.images_count = 3 THEN '3 images' 
    ELSE '4 and + images' 
  END AS `Informations`, 

  COUNT(dt.product_id) AS `total` 
FROM 
(
  SELECT 
    `exchange_product`.`product_id`, 
    COUNT(`image`.`context_uuid`) AS images_count 
  FROM `exchange_product`
  LEFT JOIN `product` ON `product`.`product_id` = `exchange_product`.`product_id`
  LEFT JOIN `image` ON `image`.`context_uuid` = `product`.`product_uuid`
  GROUP BY 
    `exchange_product`.`product_id`
) AS dt 
GROUP BY `Informations`

Answer 2

I'd use an inline view (derived table) to get a count of images for each product_id. (Run the query inside the parens to see the result that is returned. That should be each distinct product_id, with the funky count of images related to the product.

Join the result of the inline view to the exchange_product table. And the outer query, we can do the GROUP BY on the funky count.

SELECT q.image_cnt                          AS `Informations`
     , COUNT(e.product_id)                  AS `total`
  FROM ( -- count of images by product_id
         SELECT p.product_id
              , CASE COUNT(i.context_uuid)
                WHEN 0 THEN 'No image'
                WHEN 1 THEN '1 image'
                WHEN 2 THEN '2 images'
                WHEN 3 THEN '3 images'
                ELSE '4 and + images'
                END AS image_cnt
           FROM `product` p
           LEFT
           JOIN `image` i
             ON i.context_uuid = p.product_uuid
          GROUP
             BY p.product_id
       ) q
  JOIN `exchange_product` e
    ON e.product_id = q.product_id
 GROUP
    BY q.image_cnt
 ORDER
    BY q.image_cnt+0