Oracle 10g 从两个不同的行和列聚合成一行

Question

我有一个包含就业开始和结束日期的表格，如下所示：

EEID | EMP_START_DATE | EMP_TERM_DATE 
-----+----------------+---------------
1    | 2014-01-01     | null
1    | null           | 2014-03-30
1    | 2014-06-01     | null
1    | null           | 2014-12-31
1    | 2015-08-01     | null
2    | 2001-07-01     | null
3    | 2010-03-01     | null
3    | null           | 2010-03-30
3    | 2010-04-01     | null
3    | null           | 2010-04-30

我想将其转换如下，按employeeId (EEID) 和工作期限排序：

EEID | EMP_START_DATE | EMP_TERM_DATE  
-----+----------------+---------------
1    | 2014-01-01     | 2014-03-30
1    | 2014-06-01     | 2014-12-31
1    | 2015-08-01     | null
2    | 2001-07-01     | null
3    | 2010-03-01     | 2010-03-30
3    | 2010-04-01     | 2010-04-30

问题

谁能告诉我该怎么做？

Answer 1

一种方法是条件聚合。这需要枚举开始和结束，然后使用此枚举进行聚合：

select eeid, max(emp_start_date) as emp_start_date,
       max(emp_end_date) as emp_end_date
from (select t.*,
             (case when emp_start_date is not null
                   then count(emp_start_date) over (partition by eeid order by emp_start_date)
                   else count(emp_end_date) over (partition by eeid order by emp_send_date)
              end) as seqnum
      from t
     ) t
group by eeid, seqnum;

Answer 2

当当前行的值为空时，您可以使用领先和滞后来获取上一个/下一个日期值：

select eeid,
  emp_start_date as orig_start_date,
  emp_term_date as orig_term_date,
  coalesce(emp_start_date,
    lag(emp_start_date)
      over (partition by eeid order by coalesce(emp_start_date, emp_term_date)))
    as emp_start_date,
  coalesce(emp_term_date,
    lead(emp_term_date)
      over (partition by eeid order by coalesce(emp_term_date, emp_start_date)))
    as emp_term_date
from your_table
order by eeid, emp_start_date, orig_start_date, emp_term_date, orig_term_date;

      EEID ORIG_START_DATE ORIG_TERM_DATE EMP_START_DATE EMP_TERM_DATE
---------- --------------- -------------- -------------- -------------
         1 2014-01-01                     2014-01-01     2014-03-30   
         1                 2014-03-30     2014-01-01     2014-03-30   
         1 2014-06-01                     2014-06-01     2014-12-31   
         1                 2014-12-31     2014-06-01     2014-12-31   
         1 2015-08-01                     2015-08-01                  
         2 2001-07-01                     2001-07-01                  
         3 2010-03-01                     2010-03-01     2010-03-30   
         3                 2010-03-30     2010-03-01     2010-03-30   
         3 2010-04-01                     2010-04-01     2010-04-30   
         3                 2010-04-30     2010-04-01     2010-04-30   

然后消除重复：

select distinct eeid,
  coalesce(emp_start_date,
    lag(emp_start_date)
      over (partition by eeid order by coalesce(emp_start_date, emp_term_date)))
    as emp_start_date,
  coalesce(emp_term_date,
    lead(emp_term_date)
      over (partition by eeid order by coalesce(emp_term_date, emp_start_date)))
    as emp_term_date
from your_table
order by eeid, emp_start_date;

      EEID EMP_START_DATE EMP_TERM_DATE
---------- -------------- -------------
         1 2014-01-01     2014-03-30   
         1 2014-06-01     2014-12-31   
         1 2015-08-01                  
         2 2001-07-01                  
         3 2010-03-01     2010-03-30   
         3 2010-04-01     2010-04-30