PL/SQL Split, separate a date into new dates according to black out dates!

前端 未结 2 1112
没有蜡笔的小新
没有蜡笔的小新 2021-02-10 12:23

I have lets say a \"travel date\" and black out dates. I will split the travel date into pieces according to the black out dates.

Note: Travel Date can be between

2条回答
  •  礼貌的吻别
    2021-02-10 13:03

    Your tables:

    SQL> create table travel (start_date,end_date)
      2  as
      3  select date '2011-01-04', date '2011-12-11' from dual
      4  /
    
    Table created.
    
    SQL> create table black_out_dates (bo,start_date,end_date)
      2  as
      3  select 'A', date '2010-11-01', date '2011-02-11' from dual union all
      4  select 'B', date '2011-01-20', date '2011-02-15' from dual union all
      5  select 'C', date '2011-03-13', date '2011-04-10' from dual union all
      6  select 'D', date '2011-03-20', date '2011-06-29' from dual
      7  /
    
    Table created.
    

    And the query, which takes into account completely overlapping black out periods:

    SQL> select 'X' || to_char(row_number() over (order by new_start_date)) new_travel
      2       , new_start_date
      3       , new_end_date
      4    from ( select end_date + 1 new_start_date
      5                , lead(start_date - 1, 1, t_end_date) over (order by start_date) new_end_date
      6             from ( select start_date
      7                         , end_date
      8                         , t_end_date
      9                         , row_number() over (order by start_date) rn_start_date
     10                         , row_number() over (order by end_date) rn_end_date
     11                      from ( select bo.start_date
     12                                  , bo.end_date
     13                                  , t.end_date t_end_date
     14                               from black_out_dates bo
     15                                  , travel t
     16                              where t.start_date <= bo.end_date
     17                                and t.end_date >= bo.start_date
     18                              union all
     19                             select start_date - 1
     20                                  , start_date - 1
     21                                  , null
     22                               from travel
     23                           )
     24                  )
     25            where rn_start_date <= rn_end_date
     26         )
     27   where new_start_date <= new_end_date
     28   order by new_start_date
     29  /
    
    NEW_TRAVEL NEW_START_DATE      NEW_END_DATE
    ---------- ------------------- -------------------
    X1         16-02-2011 00:00:00 12-03-2011 00:00:00
    X2         30-06-2011 00:00:00 11-12-2011 00:00:00
    
    2 rows selected.
    

    If your black_out_dates table contains N rows, then there at most N+1 gaps. The query makes up one dummy black out date period at [2011-01-03,2011-01-03], and then uses the analytic function LEAD to determine where the next black out date starts. Completely overlapping periods are removed by the ROW_NUMBER analytic functions, because they mess up the gap periods.


    EDIT 14-3-2011

    With these tables:

    SQL> create table travel (start_date,end_date)
      2  as
      3  select date '2001-01-04', date '2013-12-11' from dual
      4  /
    
    Table created.
    
    SQL> create table black_out_dates (bo,start_date,end_date)
      2  as
      3  select 'A', date '2010-11-01', date '2011-02-11' from dual union all
      4  select 'B', date '2011-01-20', date '2011-02-15' from dual union all
      5  select 'C', date '2011-03-13', date '2011-04-10' from dual union all
      6  select 'D', date '2011-03-20', date '2011-06-29' from dual
      7  /
    
    Table created.
    

    My previous query still did not handle overlapping periods correctly. So here is a revised version, conveniently stored in view v:

    SQL> create view v
      2  as
      3  with t1 as
      4  ( select bo.start_date
      5         , bo.end_date
      6         , t.end_date t_end_date
      7      from black_out_dates bo
      8         , travel t
      9     where bo.start_date <= t.end_date
     10       and bo.end_date >= t.start_date
     11     union all
     12    select start_date - 1
     13         , start_date - 1
     14         , end_date
     15      from travel
     16  )
     17  , t2 as
     18  ( select t1.*
     19         , nvl
     20           ( max(end_date)
     21             over
     22             ( order by start_date,end_date desc
     23               rows between unbounded preceding and 1 preceding
     24             )
     25           , to_date('1','j')
     26           ) max_date
     27      from t1
     28  )
     29  , t3 as
     30  ( select start_date
     31         , end_date
     32         , t_end_date
     33         , sum( case when start_date > max_date then 1 else 0 end )
     34           over
     35           ( order by start_date, end_date desc ) grp
     36      from t2
     37  )
     38  , t4 as
     39  ( select max(end_date) + 1 new_start_date
     40         , lead(min(start_date) - 1, 1, t_end_date) over (order by min(start_date)) new_end_date
     41      from t3
     42     group by t_end_date
     43         , grp
     44  )
     45  select new_start_date
     46       , new_end_date
     47    from t4
     48   where new_start_date <= new_end_date
     49  /
    
    View created.
    

    And the test results:

    SQL> set feedback off
    SQL> remark  Test 1
    SQL> select * from v
      2  /
    
    NEW_START_DATE      NEW_END_DATE
    ------------------- -------------------
    04-01-2001 00:00:00 31-10-2010 00:00:00
    16-02-2011 00:00:00 12-03-2011 00:00:00
    30-06-2011 00:00:00 11-12-2013 00:00:00
    SQL> remark  Test 2
    SQL> delete travel
      2  /
    SQL> delete black_out_dates
      2  /
    SQL> INSERT INTO TRAVEL(  START_DATE, END_DATE )                 VALUES (   DATE '2011-01-01', DATE '2011-12-31' );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'A',  DATE '2011-02-02', DATE '2011-05-06' );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'B',  DATE '2011-03-03', DATE '2011-03-05'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'C',  DATE '2011-07-07', DATE '2011-07-09'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'D',  DATE '2011-07-08', DATE '2011-07-20'  );
    SQL> select * from v
      2  /
    
    NEW_START_DATE      NEW_END_DATE
    ------------------- -------------------
    01-01-2011 00:00:00 01-02-2011 00:00:00
    07-05-2011 00:00:00 06-07-2011 00:00:00
    21-07-2011 00:00:00 31-12-2011 00:00:00
    SQL> remark  Test 3
    SQL> delete travel
      2  /
    SQL> delete black_out_dates
      2  /
    SQL> INSERT INTO TRAVEL(  START_DATE, END_DATE )                 VALUES (   DATE '2011-04-02', DATE '2011-10-20'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'A',  DATE '2011-01-01', DATE '2011-05-03'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'B',  DATE '2011-06-04', DATE '2011-06-20'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'C',  DATE '2011-06-06', DATE '2011-06-08'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'D',  DATE '2011-08-08', DATE '2011-12-30'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'E',  DATE '2011-08-08', DATE '2011-12-30'  );
    SQL> select * from v
      2  /
    
    NEW_START_DATE      NEW_END_DATE
    ------------------- -------------------
    04-05-2011 00:00:00 03-06-2011 00:00:00
    21-06-2011 00:00:00 07-08-2011 00:00:00
    SQL> remark  Test 4
    SQL> delete travel
      2  /
    SQL> delete black_out_dates
      2  /
    SQL> INSERT INTO TRAVEL(  START_DATE, END_DATE )                 VALUES (   DATE '2011-02-10', DATE '2011-05-15'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'A',  DATE '2011-02-02', DATE '2011-02-15'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'B',  DATE '2011-02-10', DATE '2011-02-20'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'C',  DATE '2011-04-10', DATE '2011-04-25'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'D',  DATE '2011-04-15', DATE '2011-04-20'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'E',  DATE '2011-05-10', DATE '2011-05-20'  );
    SQL> select * from v
      2  /
    
    NEW_START_DATE      NEW_END_DATE
    ------------------- -------------------
    21-02-2011 00:00:00 09-04-2011 00:00:00
    26-04-2011 00:00:00 09-05-2011 00:00:00
    SQL> remark  Test 5
    SQL> delete travel
      2  /
    SQL> delete black_out_dates
      2  /
    SQL> INSERT INTO TRAVEL VALUES (   DATE '2011-02-10', DATE '2011-05-17'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'A',  DATE '2011-02-05', DATE '2011-02-20'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'B',  DATE '2011-02-07', DATE '2011-02-09'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'C',  DATE '2011-05-05', DATE '2011-05-20'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'D',  DATE '2011-05-07', DATE '2011-05-15'  );
    SQL> INSERT INTO BLACK_OUT_DATES( BO, START_DATE, END_DATE ) VALUES ( 'E',  DATE '2011-05-09', DATE '2011-05-12'  );
    SQL> select * from v
      2  /
    
    NEW_START_DATE      NEW_END_DATE
    ------------------- -------------------
    21-02-2011 00:00:00 04-05-2011 00:00:00
    SQL> remark  Test 6
    SQL> delete travel
      2  /
    SQL> delete black_out_dates
      2  /
    SQL> INSERT INTO TRAVEL VALUES (DATE '2011-02-10', DATE '2011-09-20' );
    SQL> INSERT INTO BLACK_OUT_DATES VALUES ('A', DATE '2011-01-05', DATE '2011-10-10' );
    SQL> select * from v
      2  /
    

    Regards,
    Rob.

提交回复
热议问题