I am stuck with this SQL Server query; here is my set of records:
Name Value ---- ---- abc 0.7 xyz 0.4 pqr 0.6 ijk 0.3 fgh 0.1 cde 1.0 uvw 0.8
I want a query which will give me sum of value until it reach 1.0 and greater then continue same with next rows, like
Names Value ----- ----- abc,xyz 1.1 pqr,ijk,fgh 1.0 cde 1.0 uvw 0.8
Please help me with this query; I tried with partitions and recursive CTE, but nothing worked so far.
Advertisement
Answer
SQL tables represent unordered sets.  Your results assume a particular ordering and there is no column for the ordering.  The following assumes that such a column exists.  It is just called id.
This type of problem requires a recursive CTE (or some other iterative processing), as far as I know. Here is one method:
with t as (
      select t.*, row_number() over (order by id) as seqnum
      from mytable t
     ),
     cte as (
      select top (1) seqnum, convert(varchar(max), name) as name, convert(decimal(10, 1), value) as value,
             (case when value >= 1.0 then 1 else 0 end) as is_new
      from t
      order by seqnum
      union all
      select t.seqnum, 
             (case when cte.value >= 1 then t.name else concat(cte.name, ',', t.name) end),
             convert(decimal(10, 1), (case when cte.value > =1 then t.value else t.value + cte.value end)),
             (case when cte.value >= 1 then 1 else 0 end) as is_new
      from cte join
           t
           on t.seqnum = cte.seqnum + 1
     )
select name, value as sum_value
from (select cte.*, lead(name) over (order by seqnum) as nextname
      from cte
     ) cte
where nextname is null or nextname not like name + '%';
Here is a db<>fiddle.