排名语句中的用户定义变量

Question

我会在此问题上提供任何帮助。我已经花了几个小时没有任何真正的解决方案。 我有一个 SQL

SELECT to_place, rank
FROM
  (SELECT g1.to_place as to_place, g1.pcount as pcount, 
                @rank := IF(@current_to_place = g1.to_place, @rank + 1, 1) AS rank,
                @current_to_place := g1.to_place 
  FROM 
          (select 
          to_place, count(*) as pcount
          from temp_workflows
          group by to_place
          order by to_place,pcount desc) g1
  ORDER BY g1.to_place, g1.pcount DESC) ranked

在表 g1 中，我对数据进行分组以查找最常见的 to_place。然后我想按升序排列这些出现（这样我以后可以选择每个 to_place 类别中最常见的前 3 个。

问题是用户定义的变量是不可预测的（@rank 有时总是 1），这可能与在一个语句中我不应该引用同一个变量（current_to_place）这一事实有关。我读了很多关于使用单独的语句等的内容，但我可以找到一种方法来以不同的方式编写我的语句。如何在其他地方定义 @current_to_place 以使结果相同？ 提前感谢您的帮助。

Answer 1

我认为你应该测试 pcount 以获得排名并且你应该初始化变量

DROP TABLE IF EXISTS T;
CREATE TABLE T
(to_place int);

insert into t values (1),(2),(2),(3),(3),(3);
SELECT to_place, rank
FROM
  (
  SELECT g1.to_place as to_place, g1.pcount as pcount, 
                @rank := IF(@current_to_place <> pcount, @rank + 1, 1) AS rank,
                @current_to_place := pcount 
  FROM 
          (select 
          to_place, count(*) as pcount
          from t
          group by to_place
          order by to_place,pcount desc) g1
 cross join(select @rank:=0,@current_to_place:=0) r
  ORDER BY g1.pcount DESC
  )
   ranked

+----------+------+
| to_place | rank |
+----------+------+
|        3 |    1 |
|        2 |    2 |
|        1 |    3 |
+----------+------+
3 rows in set (0.016 sec)