按名称和日期范围对数据进行分组

时间:2017-10-02 15:15:05

标签: sql database oracle oracle12c

我的oracle表中有数据,我的名字和日期范围如下:

Name                From        To
Lopes, Janine       07-Jun-17   16-Jul-17
Lopes, Janine       17-Jul-17   23-Jul-17
Lopes, Janine       24-Jul-17   31-Aug-17
Baptista, Maria     23-Dec-16   19-Feb-17
Deyak,Sr, Thomas    22-Jan-17   18-Apr-17
Deyak,Sr, Thomas    27-Apr-17   14-May-17
Deyak,Sr, Thomas    15-May-17   21-May-17
Deyak,Sr, Thomas    22-May-17   28-May-17
Deyak,Sr, Thomas    29-May-17   31-May-17
Serrentino, Joyce   18-Mar-17   30-Apr-17
More, Cathleen      30-Jul-17   13-Aug-17
More, Cathleen      14-Aug-17   20-Aug-17
More, Cathleen      21-Aug-17   27-Aug-17
More, Cathleen      28-Aug-17   03-Sep-17
More, Cathleen      04-Sep-17   10-Sep-17
More, Cathleen      11-Sep-17   24-Sep-17
Barrows, Michael    30-Jan-17   19-Mar-17
Barrows, Michael    20-Mar-17   26-Mar-17
Barrows, Michael    27-Mar-17   02-Apr-17
Barrows, Michael    03-Apr-17   07-Apr-17

对于一个用户来说,迄今为止的日期比起日期大一个并且是连续的,但在某些情况下会破坏数据,因此我的输出应如下所示:

Name                From            To
Lopes, Janine       07-Jun-17   31-Aug-17
Baptista, Maria     23-Dec-16   19-Feb-17
Deyak,Sr, Thomas    22-Jan-17   18-Apr-17
Deyak,Sr, Thomas    27-Apr-17   31-May-17
Serrentino, Joyce   18-Mar-17   30-Apr-17
More, Cathleen      30-Jul-17   24-Sep-17
Barrows, Michael    30-Jan-17   07-Apr-17

如果我做min(from)和max(to),我会放弃像Thomas这样的记录。 我应该如何编写sql以获取我需要的数据。

2 个答案:

答案 0 :(得分:5)

这可以通过Tabibitosan方法很好地解决。

的制备:将

alter session set nls_date_format = 'dd-Mon-rr';

Session altered.
;

查询(为方便起见,包括模拟输入):

with
     inputs ( name, date_fr, date_to ) as (
       select 'Lopes, Janine'    , to_date('07-Jun-17'), to_date('16-Jul-17') from dual union all
       select 'Lopes, Janine'    , to_date('17-Jul-17'), to_date('23-Jul-17') from dual union all
       select 'Lopes, Janine'    , to_date('24-Jul-17'), to_date('31-Aug-17') from dual union all
       select 'Baptista, Maria'  , to_date('23-Dec-16'), to_date('19-Feb-17') from dual union all
       select 'Deyak,Sr, Thomas' , to_date('22-Jan-17'), to_date('18-Apr-17') from dual union all
       select 'Deyak,Sr, Thomas' , to_date('27-Apr-17'), to_date('14-May-17') from dual union all
       select 'Deyak,Sr, Thomas' , to_date('15-May-17'), to_date('21-May-17') from dual union all
       select 'Deyak,Sr, Thomas' , to_date('22-May-17'), to_date('28-May-17') from dual union all
       select 'Deyak,Sr, Thomas' , to_date('29-May-17'), to_date('31-May-17') from dual union all
       select 'Serrentino, Joyce', to_date('18-Mar-17'), to_date('30-Apr-17') from dual union all
       select 'More, Cathleen'   , to_date('30-Jul-17'), to_date('13-Aug-17') from dual union all
       select 'More, Cathleen'   , to_date('14-Aug-17'), to_date('20-Aug-17') from dual union all
       select 'More, Cathleen'   , to_date('21-Aug-17'), to_date('27-Aug-17') from dual union all
       select 'More, Cathleen'   , to_date('28-Aug-17'), to_date('03-Sep-17') from dual union all
       select 'More, Cathleen'   , to_date('04-Sep-17'), to_date('10-Sep-17') from dual union all
       select 'More, Cathleen'   , to_date('11-Sep-17'), to_date('24-Sep-17') from dual union all
       select 'Barrows, Michael' , to_date('30-Jan-17'), to_date('19-Mar-17') from dual union all
       select 'Barrows, Michael' , to_date('20-Mar-17'), to_date('26-Mar-17') from dual union all
       select 'Barrows, Michael' , to_date('27-Mar-17'), to_date('02-Apr-17') from dual union all
       select 'Barrows, Michael' , to_date('03-Apr-17'), to_date('07-Apr-17') from dual
     )
-- End of simulated inputs (for testing only, not part of the solution).
-- SQL query begins BELOW THIS LINE. Use your actual table and column names.
select name, min(date_fr) as date_fr, max(date_to) as date_to
from   ( select name, date_fr, date_to,
                date_to - sum( date_to - date_fr + 1 ) over (partition by name 
                                                            order by date_fr) as gr
         from   inputs
       )
group by name, gr
order by name, date_fr
;

输出:

NAME              DATE_FR   DATE_TO 
----------------- --------- ---------
Baptista, Maria   23-Dec-16 19-Feb-17
Barrows, Michael  30-Jan-17 07-Apr-17
Deyak,Sr, Thomas  22-Jan-17 18-Apr-17
Deyak,Sr, Thomas  27-Apr-17 31-May-17
Lopes, Janine     07-Jun-17 31-Aug-17
More, Cathleen    30-Jul-17 24-Sep-17
Serrentino, Joyce 18-Mar-17 30-Apr-17

 7 rows selected 

答案 1 :(得分:5)

在Oracle 12.1及更高版本中,the MATCH_RECOGNIZE子句可以快速完成此类要求。我在其他答案中使用相同的设置和模拟数据(WITH子句),输出也相同。

select name, date_fr, date_to
from   inputs
match_recognize(
  partition by name
  order by date_fr
  measures a.date_fr     as date_fr,
           last(date_to) as date_to
  pattern ( a b* )
  define b as date_fr = prev(date_to) + 1
)
;