Question

I have a table like this:

create table time_sheet
(
      StatusCode char(1),
      start_time datetime,
      end_time datetime
)

insert into time_sheet values
('W','2012-08-01 10:00:00','2012-08-01 12:00:00'),
('D','2012-08-01 12:00:00','2012-08-01 14:00:00'),
('N','2012-08-01 16:00:00','2012-08-01 18:00:00')

The output should be like this:

StatusCode start_time               end_time
B          2012-08-01 08:00:00.000  2012-08-01 10:00:00.000
W          2012-08-01 10:00:00.000  2012-08-01 12:00:00.000
D          2012-08-01 12:00:00.000  2012-08-01 14:00:00.000
B          2012-08-01 14:00:00.000  2012-08-01 16:00:00.000
N          2012-08-01 16:00:00.000  2012-08-01 18:00:00.000
B          2012-08-01 18:00:00.000  2012-08-01 20:00:00.000

The beging and end of the day are declared as below.

declare @begingOfDay datetime='2012-08-01 08:00:00.000'
declare @endOfDay    datetime='2012-08-01 20:00:00.000'

Basically I want to have missing time range records in the result set between begingOfDay and endOfDay with statusCode B . Please see that in the output there are 3 records added with StatusCode B

Could anybody help with this?

Answer 1

Your sample data (note, Ts added to strings to enforce unambiguous date conversions):

create table time_sheet
(
      StatusCode char(1),
      start_time datetime,
      end_time datetime
)

insert into time_sheet values
('W','2012-08-01T10:00:00','2012-08-01T12:00:00'),
('D','2012-08-01T12:00:00','2012-08-01T14:00:00'),
('N','2012-08-01T16:00:00','2012-08-01T18:00:00')

declare @begingOfDay datetime='2012-08-01T08:00:00.000'
declare @endOfDay    datetime='2012-08-01T20:00:00.000'

And the query:

;with AllDTs as (
    select @begingOfDay as TimePoint
    union
    select @endOfDay
    union
    select start_time from time_sheet
    union
    select end_time from time_sheet
), OrderedDTs as (
    select TimePoint,ROW_NUMBER() OVER (ORDER BY TimePoint) as rn
    from AllDTs
), Periods as (
    select o1.TimePoint as start_time,o2.TimePoint as end_time
    from
        OrderedDTs o1
            inner join
        OrderedDTs o2
            on
                o1.rn = o2.rn-1
)
select
    COALESCE(ts.StatusCode,'B') as StatusCode,
    p.start_time,
    p.end_time
from
    Periods p
        left join
    time_sheet ts
        on
            p.start_time = ts.start_time and
            p.end_time = ts.end_time

Result:

StatusCode start_time              end_time
---------- ----------------------- -----------------------
B          2012-08-01 08:00:00.000 2012-08-01 10:00:00.000
W          2012-08-01 10:00:00.000 2012-08-01 12:00:00.000
D          2012-08-01 12:00:00.000 2012-08-01 14:00:00.000
B          2012-08-01 14:00:00.000 2012-08-01 16:00:00.000
N          2012-08-01 16:00:00.000 2012-08-01 18:00:00.000
B          2012-08-01 18:00:00.000 2012-08-01 20:00:00.000

Note, I’ve proceeded on the assumption that there are no overlapping time periods in the original table. The first CTE (AllDTs) just finds all unique datetime values that are of interest to us. OrderedDTs and Periods than arrange all of these datetime values into successive periods. The final query then takes each of these periods, and attempts to match them back to the original table, if possible. If not, then it’s obviously a B period.