Question

I am doing a row_number() over some data.

My problem is that I do not wnat to take into account data that is 0.

Here’s my data sample:

+-----------+-----+
| Alex      |   1 |
| Liza      |   2 |
| Harry     |   0 |
| Marge     |  24 |
| Bla       |   0 |
| Something | 234 |
+-----------+-----+

Here’s what I want:

+-----------+--------+------------+
|   name    | number | row_number |
+-----------+--------+------------+
| Harry     |      0 |          0 |
| Bla       |      0 |          0 |
| Something |    234 |          1 |
| Marge     |     24 |          2 |
| Liza      |      2 |          3 |
| Alex      |      1 |          4 |
+-----------+--------+------------+

as you can see the third column is the row_number()

this is what I have so far:

select name, number, row_number() over (partition by name order by name,number)
from myTable

How do I get the query above to return 0 for all 0's in the source data and not count the 0’s at all towards the row_number sequence?

Answer 1

Try this SQL Fiddle:

SELECT name, number, 0 as [row_number]
FROM myTable
WHERE number = 0
UNION ALL
SELECT name, number, row_number() OVER (ORDER BY number DESC)
FROM myTable
WHERE number <> 0