How can I manipulate a string a T-SQL? - sql

Question

business_id
open_day
open_time
close_day
close_time
1
FRIDAY
08:00
FRIDAY
12:00
1
FRIDAY
13:00
FRIDAY
17:00
1
MONDAY
08:00
MONDAY
17:00
2
SATURDAY
08:00
SATURDAY
16:00
2
SUNDAY
08:00
SUNDAY
16:00
3
MONDAY
08:00
MONDAY
16:00
I have a task to create and format opening hours for a business, I'm required to group these by days in a single string. However, it is repeating the day. Would it be possible to search for a particular value if it occurs twice within a string? I have gotten this far:
create table open_times (business_id int, open_day varchar(10), open_time varchar(10), close_day varchar(10), close_time varchar(10) )
insert into open_times (business_id, open_day, open_time, close_day, close_time) values (1, 'FRIDAY', '08:00', 'FRIDAY', '12:00')
insert into open_times (business_id, open_day, open_time, close_day, close_time) values (1, 'FRIDAY', '13:00', 'FRIDAY', '17:00')
insert into open_times (business_id, open_day, open_time, close_day, close_time) values (1, 'MONDAY', '08:00', 'MONDAY', '17:00')
insert into open_times (business_id, open_day, open_time, close_day, close_time) values (2, 'SATURDAY', '08:00', 'SATURDAY', '16:00')
insert into open_times (business_id, open_day, open_time, close_day, close_time) values (2, 'SUNDAY', '08:00', 'SUNDAY', '16:00')
insert into open_times (business_id, open_day, open_time, close_day, close_time) values (3, 'MONDAY', '08:00', 'MONDAY', '16:00')
drop table open_times
I appreciate your help.
select
business_id
,left(name_values, LEN(name_values)-1) as opening_hours
from
(
select
results.business_id
,STUFF((
select
( case when open_day = 'FRIDAY' then 'Fr' when open_day = 'MONDAY' then 'Mo' when open_day = 'TUESDAY' then 'Tu' when open_day = 'WEDNESDAY' then 'We' When open_day = 'THURSDAY' then 'Th' when open_day = 'SATURDAY' then 'Sa' else 'Su' end )
+ ' ' + open_time + '-' + close_time + '; '
from open_times
where business_id = results.business_id
for xml path(''),type).value('(./text())[1]','VARCHAR(MAX)'),1,0, '') as name_values
from open_times results
group by business_id
) innerquery
Current Output for Business 1: 'Fr 08:00-12:00; Fr 13:00-17:00; Mo 08:00-17:00'
Desired Output For Business 1: 'Fr 08:00-12:00, 13:00-17:00; Mo 08:00-17:00'

Answer 1

You can use STRING_AGG here. You just need to do two levels of grouping, once per day, then again for the whole business_id
SELECT
ot.business_id,
Times = STRING_AGG(CONCAT(
UPPER(LEFT(ot.open_day, 1)),
LOWER(SUBSTRING(ot.open_day, 2, 1)),
' ',
ot.Times)
, '; ')
FROM (
SELECT
ot.business_id,
ot.open_day,
Times = STRING_AGG(CONCAT(
ot.open_time,
'-',
ot.close_time),
', ')
FROM open_times ot
GROUP BY ot.business_id, ot.open_day
) ot
GROUP BY ot.business_id;
db<>fiddle

Answer 2

You can do this with one level of aggregation by incorporating window functions into your logic:
select ot.business_id,
stuff((select (case when seqnum = 1 then '; ' else ', ' end) +
(case when seqnum = 1 and ot2.open_day = 'FRIDAY' then 'Fr '
when seqnum = 1 and ot2.open_day = 'MONDAY' then 'Mo '
when seqnum = 1 and ot2.open_day = 'TUESDAY' then 'Tu '
when seqnum = 1 and ot2.open_day = 'WEDNESDAY' then 'We '
when seqnum = 1 and ot2.open_day = 'THURSDAY' then 'Th '
when seqnum = 1 and ot2.open_day = 'SATURDAY' then 'Sa '
when seqnum = 1 then 'Su '
else ''
end ) +
ot2.open_time + '-' + ot2.close_time
from (select ot2.*,
row_number() over (partition by ot2.open_day order by ot2.open_time) as seqnum
from open_times ot2
where ot2.business_id = ot.business_id
) ot2
order by ot2.open_day, ot2.open_time
for xml path(''),type
).value('(./text())[1]', 'VARCHAR(MAX)'), 1, 2, ''
) as name_values
from open_times ot
group by ot.business_id;
Here is a db<>fiddle.
Your version of the query is actually incorrect, because you do not have an order by in the XML subquery. The results can be in any order.
In addition, you don't just want to remove the duplicate names, but you also want to change the delimiter from ; to ,.
The idea in the above query is that the times are enumerated on each day. The delimiter is ; for the first time and , for the subsequent ones. The abbreviation is used only for the first one. I also added qualifications to all the column references, a highly recommended practice.
Note that you can simplify the case logic to:
left(ot2.open_day, 1) + lower(substring(ot2.open_day, 2, 1))

Answer 3

For older versions use FOR XML PATH aggregation trick. Assuming no more then 2 open intervals a day
with t as (
select business_id, left(open_day,2) + ' ' + min(open_time + '-' + close_time) +
case when min(open_time) = max(open_time) then ''
else ', ' + max(open_time + '-' + close_time) end ots
from open_times
group by business_id, open_day
)
select business_id, stuff(
(select '; ' + ots
from t t2
where t2.business_id = t1.business_id
order by left(ots,2)
for xml path(''))
, 1, 2, '') opening_hours
from t t1
group by business_id