如何使用多个select查询优化一个UNION mysql查询？

Question

我有一个要求，以显示商品详细页的同类产品以这样的方式，上市的产品应低于

1. 大小，颜色，类别，公司给出的分级方式符合目前产品性能应与当前的产品
2. 大小，颜色，类别应与当前的产品
3. 大小，颜色应与当前的产品
4. 大小应与当前的产品

如下面给出

我的SQL查询：

(SELECT pd.product_id, pd.name, p.price 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.color_id = '2' AND 
      pd.category_id = '3' AND 
      pd.company_id = '1' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
      ORDER BY RAND() LIMIT 10 
    ) 
     UNION 
    (SELECT pd.product_id, pd.name, p.price 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.color_id = '2' AND 
      pd.category_id = '3' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
      ORDER BY RAND() LIMIT 10 
    ) 
     UNION 
    (SELECT pd.product_id, pd.name, p.price 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.color_id = '2' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
      ORDER BY RAND() LIMIT 10 
    ) 
     UNION 
    (SELECT pd.product_id, pd.name, p.price 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
      ORDER BY RAND() LIMIT 10 
    ) 

53 - 当前的产品ID和状态表示可用！

有没有什么办法来优化上述查询？

注输出要求：我们需要10个相似的产品。如果存在4个匹配条件1的产品，则需要先按随机顺序列出。同样，我们需要列出与其下面的其他条件匹配的产品。

在此先感谢！

Answer 1

保持使用ORDER BY RAND（）（这是不高效的），那么一个非常基本的改进就是为每个子句添加一个优先级，将order子句移动到最后，因此只需要一次。

(SELECT pd.product_id, pd.name, p.price , 1 AS recpriority 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.color_id = '2' AND 
      pd.category_id = '3' AND 
      pd.company_id = '1' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
    ) 
     UNION 
    (SELECT pd.product_id, pd.name, p.price , 2 AS recpriority 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.color_id = '2' AND 
      pd.category_id = '3' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
    ) 
     UNION 
    (SELECT pd.product_id, pd.name, p.price , 3 AS recpriority 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.color_id = '2' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
    ) 
     UNION 
    (SELECT pd.product_id, pd.name, p.price , 4 AS recpriority 
     FROM mg_product_description pd 
      JOIN `mg_product` p ON p.product_id = pd.product_id 
     WHERE 
      pd.size_id = '33' AND 
      pd.product_id != '53' AND 
      p.status = '1' 
    ) 
    ORDER BY recpriority, RAND() LIMIT 10 

这可能再被做这样的事情，而不需要做的工会： -

SELECT pd.product_id, 
    pd.name, 
    p.price , 
    CASE 
     WHEN pd.color_id = '2' AND pd.category_id = '3' AND pd.company_id = '1' THEN 1 
     WHEN pd.color_id = '2' AND pd.category_id = '3' THEN 2 
     WHEN pd.color_id = '2' THEN 3 
     ELSE 4 
    END AS recpriority 
FROM mg_product_description pd 
JOIN `mg_product` p ON p.product_id = pd.product_id 
WHERE pd.size_id = '33' 
AND pd.product_id != '53' 
AND p.status = '1' 
ORDER BY recpriority, RAND() 
LIMIT 10