Question

我对SQL还是很陌生，并且正在尝试编写查询以查找类似以下情况的逻辑：

wk_id | start_date | end_date | status

-----   ----------    ------    ------ 
1       20160101     20160101     SUCCEEDED
2       20160101     20160101     SUCCEEDED
3       20160101     20160101     SUCCEEDED
4       20160101     20160101     SUCCEEDED
5       20160101     20160101     FAILED
6       20160102     20160102     SUCCEEDED
7       20160102     20160102     SUCCEEDED
8       20160102     20160102     FAILED
9       20160102     20160102     FAILED
10      20160101     20160101     FAILED
11      20160101     20160101     FAILED

因此，三天来提交了一些作品，一些失败了，一些成功了。

我想编写一个查询，该查询返回每天的作品总数，成功的作品和失败的作品。

我肯定做错了，但这就是我的想法。

请指出错误并解释其逻辑。

select 
       w1.end_date,
       w1.status,COUNT(*) as total_instances, 
       COUNT(*) as as succeeded_instances,
       COUNT(*) as as failed_instances 
from 
       work_instances w1 , 
       work_instances w2
 where 
       w1.status LIKE 'SUCCEEDED' 
       and w2.status NOT LIKE 'SUCCEEDED' 
       and w1.wk_id = w2.wk_id 
 group by 
       w1.end_date,w1.status;

Answer 1

将聚合函数与大小写一起使用

  select 
  w1.end_date,COUNT(*) AS total_work,
  sum(case when w1.status='SUCCEEDED' then 1 else 0 end ) as SUCCEEDED_task, 
  sum(case when w1.status='FAILED' then 1 else 0 end ) as FAILED_task                  
  from 
  work_instances w1 group by w1.end_date

Answer 2

也许更优雅的方法是在聚合函数调用中使用FILTER：

WITH work_instances(wk_id, start_date, end_date, status) AS (
    VALUES 
        (1, '20160101'::DATE, '20160101'::DATE, 'SUCCEEDED'),
        (2, '20160101', '20160101', 'SUCCEEDED'),
        (3, '20160101', '20160101', 'SUCCEEDED'),
        (4, '20160101', '20160101', 'SUCCEEDED'),
        (5, '20160101', '20160101', 'FAILED'),
        (6, '20160102', '20160102', 'SUCCEEDED'),
        (7, '20160102', '20160102', 'SUCCEEDED'),
        (8, '20160102', '20160102', 'FAILED'),
        (9, '20160102', '20160102', 'FAILED'),
        (10, '20160101', '20160101', 'FAILED'),
        (11, '20160101', '20160101', 'FAILED')      
)
SELECT 
    end_date,
    count(*) AS total_instances,
    count(*) FILTER (WHERE status = 'SUCCEEDED') AS succeeded_instances,
    count(*) FILTER (WHERE status = 'FAILED') AS failed_instances        
FROM 
    work_instances
GROUP BY 
    end_date

Answer 3

听起来像您可以使用条件聚合。

<mat-select *selectedType placeholder="Type" formControlName="type">
   <mat-option *ngFor="let type of typeData" [value]="type.Type_id">
     {{type.type_name}}
  </mat-option>
</mat-select>

如果状态为正确的SELECT end_date, count(*) total_instances, count(CASE WHEN status = 'SUCCEEDED' THEN 1 END) succeeded_instances, count(CASE WHEN status = 'FAILED' THEN 1 END) failed_instances FROM work_instances GROUP BY end_date;将仅返回一个非null值（此处为1，但可能不是null）。而且CASE不会只计算非空值。

无法形成SQL Join查询

3 个答案: