Question

I want to compute and count all trigrams of a string in SQL Server.

For example if the string is hello I want the following output:

Trigram Count
------- -----
hel     1
ell     1
llo     1
lo-     1

Wikipedia page on n-gram

Answer 1

I still don’t know what an n-gram is but based on Ed’s answer is this what you need?

declare @string varchar(max) = 'hello'
declare @n int = 3

set @string = @string + REPLICATE('-',@n - (len(@string) % @n))

;with n as
(
SELECT 1 AS i
UNION ALL
SELECT i+1 
FROM n
WHERE i <= (LEN(@string)-@n)
)
select SUBSTRING(@string, i, @n), COUNT(*)
from n
group by  SUBSTRING(@string, i, @n)
option (maxrecursion 0)