Question

我正在使用的查询是

select count(*) as count,
       TO_CHAR(firstTable1.firstTable_date, 'DD-MON-YY') as  dateAndTime
from  (
  select *
  from   firstTable
  where  firstTable_status = 'Open'
  and    firstTable_date   > sysdate-10
) firstTable1
group by TO_CHAR(firstTable1.firstTable_date, 'DD-MON-YY')

这将使我在11-may-17上获得第1名

并为09-may-17计数1

样本表数据

firstTable

                    id        firstTable_status  firstTable_date

                   A-123         Open                11-may-17
                   B-123         Open                09-may-17

secondTable

  secondTable_Id    firstTable_id      secondTable_Action  secondTable_date

       XX1       A-123            started                 11-may-17
       XX2       A-123            Assigned                11-may-17
       XX3       A-123            action2                 11-may-17
       XX4       A-123            action3                 11-may-17
       XX5       A-123            action4                 11-may-17
       XX6       A-123            action5                 11-may-17
       XX7       A-123            closed                  11-may-17

       MN1       B-123            started                09-may-17
       MN1       B-123            action1                10-may-17
       MN1       B-123            closed                 11-may-17

是否可以修改此查询，例如

如果secondTable_Action在同一日期关闭同一个id，那么查询不应返回该id的计数

例如，firstTable中的id A-123在11-may-17上打开并在相同时关闭在第二个表中为此查询输入的日期应返回计数0表示11- 可-17 而对于身份B-123在不同的日子关闭，所以它应该返回计算在09-may-17

Answer 1

SELECT dt,
       SUM( open_closed ) AS "count"
FROM   (
  SELECT f.id,
         TRUNC( f.firstTable_date ) as dt,
         COUNT( CASE s.secondtable_action WHEN 'started' THEN 1 END )
           - COUNT( CASE s.secondtable_action WHEN 'closed' THEN 1 END ) AS open_closed
  FROM   firstTable f
         INNER JOIN secondtable s
         ON ( f.id = s.firsttable_id )
  WHERE  firstTable_date   > SYSDATE - INTERVAL '10' DAY
  GROUP BY f.id,
         TRUNC( f.firstTable_date )
  HAVING COUNT( CASE s.secondtable_action WHEN 'started' THEN 1 END ) > 0
)
GROUP BY dt

Answer 2

您可以使用not exists或类似内容删除关闭：

select count(*) as count,
       to_char(t.firstTable_date, 'DD-MON-YY') as dateAndTime
from firstTable t
where t.firstTable_status = 'Open' and
      t.firstTable_date > sysdate - 10 and
      not exists (select 1
                  from secondTable t2
                  where t2.id = t.id and
                        t2.firstTable_date = t.firstTable_date and
                        t2.firstTable_status = 'Closed'
                 )
group by to_char(t.firstTable_date, 'DD-MON-YY')
order by min(firstTable_date);

注意：

如果firstTable_date有时间组件，则相关子句应考虑到这一点：trunc(t2.firstTable_date) = trunc(t.firstTable_date)。
您可以使用trunc()代替to_char()。如果您打算使用to_char()，我建议您使用YYYY-MM-DD格式。这是ISO标准，并且可以进行各种操作。
我添加了order by，因此日期将按顺序排列。
不需要子查询。

我们如何在oracle中过滤子查询？

2 个答案: