Question

我在构建查询时遇到了麻烦，我觉得这很简单......

我有一个简单的表格，我在那里做一个简单的计数，当我用日期过滤结果时，我希望零没有行，而不是没有结果。

例如：

select APP_ID as APPLI, count(ID) as NB 
from ARCHIVE 
where GROUP_ID = 'Sample'
GROUP BY APP_ID

返回

和

select APP_ID as APPLI, count(ID) as NB 
from ARCHIVE 
where GROUP_ID = 'Sample' and TIMESTAMP > TO_DATE('01/01/2014','dd/mm/yyyy') 
GROUP BY APP_ID

返回

B   40
D   82
E   4

我想要

一些示例数据：

ID              APP_ID  TIMESTAMP
ARCH_1/20/216       A   2010-12-16
ARCH_1/20/218       A   2010-12-16
ARCH_1/20/219       A   2010-12-16
ARCH_1/40/28        A   2011-02-09
ARCH_1/40/29        A   2011-02-09
ARCH_1/40/30        A   2011-02-09
ARCH_1/260/9012     B   2014-04-19
ARCH_1/260/7093     B   2014-04-12
ARCH_1/260/8996     B   2014-04-19
ARCH_1/44/61        C   2011-09-12
ARCH_1/44/62        C   2011-09-13
ARCH_1/260/8382     D   2014-04-16
ARCH_1/260/6872     D   2014-04-11
ARCH_1/260/6271     D   2014-04-08
ARCH_1/260/8607     D   2014-04-17
ARCH_1/260/8998     E   2014-04-19
ARCH_1/260/9011     E   2014-04-19

编辑：我使用的是Oracle SQL

Answer 1

为了完成这项工作，您首先需要一个（明确的）所有 ID列表。在这种情况下，首选子查询：

SELECT Archive_List.app_id, COALESCE(COUNT(), 0) as NB
FROM (SELECT DISTINCT app_id
      FROM Archive) Archive_List
LEFT JOIN Archive
       ON Archive.app_id = Archive_List.app_id
          AND Archive.group_id = 'Sample'
          AND Archive.timestamp >= CAST('2014-01-01' as DATE)
GROUP BY Archive_List.app_id

如果timestamp实际指示的类型（顺便说一下，这是列的可怕的名称），那么您应该使用{{1因此，您可以在午夜设置任何行该语句适用于大多数/所有RDBMS

Answer 2

试试这个：

select A1.APP_ID as APPLI, NVL(count(A2.ID),0) as NB 
from ARCHIVE A1 LEFT JOIN
     ARCHIVE A2 on A1.APP_ID=A2.APP_ID
where A1.GROUP_ID = 'Sample' and A1.TIMESTAMP > TO_DATE('01/01/2014','dd/mm/yyyy') 
GROUP BY A1.APP_ID

Answer 3

select 
A2.APP_ID as APPLI,
NVL(count(A1.ID),0) as NB 
from ARCHIVE A1
RIGHT JOIN ARCHIVE A2 on A1.APP_ID = A2.APP_ID
where A1.GROUP_ID = 'Sample' and A1.TIMESTAMP > TO_DATE('01/01/2014','dd/mm/yyyy') 
GROUP BY A2.APP_ID

这个怎么样？

WITH X AS(
SELECT DISTINCT
APP_ID
from
ARCHIVE)

SELECT
X.APP_ID
,COALESCE(count(A.ID),0)
from
X
left join ARCHIVE A on X.APP_ID = A.APP_ID
where
A1.GROUP_ID = 'Sample' and A1.TIMESTAMP > TO_DATE('01/01/2014','dd/mm/yyyy') 
group by
X.APP_ID

如果没有结果则返回零

3 个答案: