T-Sql count string sequences over multiple rows - sql

Question

How can I find subsets of data over multiple rows in sql?
I want to count the number of occurrences of a string (or number) before another string is found and then count the number of times this string occurs before another one is found.
All these strings can be in random order.
This is what I want to achieve:
I have one table with one column (columnx) with data like this:
A
A
B
C
A
B
B
The result I want from the query should be like this:
2 A
1 B
1 C
1 A
2 B
Is this even possible in sql or would it be easier just to write a little C# app to do this?

Answer 1

Since, as per your comment, you can add a column that will unambiguously define the order in which the columnx values go, you can try the following query (provided the SQL product you are using supports CTEs and ranking functions):
WITH marked AS (
SELECT
columnx,
sortcolumn,
grp = ROW_NUMBER() OVER ( ORDER BY sortcolumn)
- ROW_NUMBER() OVER (PARTITION BY columnx ORDER BY sortcolumn)
FROM data
)
SELECT
columnx,
COUNT(*)
FROM marked
GROUP BY
columnx,
grp
ORDER BY
MIN(sortcolumn)
;
You can see the method in work on SQL Fiddle.
If sortcolumn is an auto-increment integer column that is guaranteed to have no gaps, you can replace the first ROW_NUMBER() expression with just sortcolumn. But, I guess, that cannot be guaranteed in general. Besides, you might indeed want to sort on a timestamp instead of an integer.

Answer 2

I dont think you can do it with a single select.
You can use AdventureWorks cursor:
create table my_Strings
(
my_string varchar(50)
)
insert into my_strings values('A'),('A'),('B'),('C'),('A'),('B'),('B') -- this method will only work on SQL Server 2008
--select my_String from my_strings
declare #temp_result table(
string varchar(50),
nr int)
declare #myString varchar(50)
declare #myLastString varchar(50)
declare #nr int
set #myLastString='A' --set this with the value of your FIRST string on the table
set #nr=0
DECLARE string_cursor CURSOR
FOR
SELECT my_string as aux_column FROM my_strings
OPEN string_cursor
FETCH NEXT FROM string_cursor into #myString
WHILE ##FETCH_STATUS = 0 BEGIN
if (#myString = #myLastString) begin
set #nr=#nr+1
set #myLastString=#myString
end else begin
insert into #temp_result values (#myLastString, #nr)
set #myLastString=#myString
set #nr=1
end
FETCH NEXT FROM string_cursor into #myString
END
insert into #temp_result values (#myLastString, #nr)
CLOSE string_cursor;
DEALLOCATE string_cursor;
select * from #temp_result
Result:
A 2
B 1
C 1
A 1
B 2

Answer 3

Try this :
;with sample as (
select 'A' as columnx
union all
select 'A'
union all
select 'B'
union all
select 'C'
union all
select 'A'
union all
select 'B'
union all
select 'B'
), data
as (
select columnx,
Row_Number() over(order by (select 0)) id
from sample
) , CTE as (
select * ,
Row_Number() over(order by (select 0)) rno from data
) , result as (
SELECT d.*
, ( SELECT MAX(ID)
FROM CTE c
WHERE NOT EXISTS (SELECT * FROM CTE
WHERE rno = c.rno-1 and columnx = c.columnx)
AND c.ID <= d.ID) AS g
FROM data d
)
SELECT columnx,
COUNT(1) cnt
FROM result
GROUP BY columnx,
g
Result :
columnx cnt
A 2
B 1
C 1
A 1
B 2