Question

Our customer feeds us contracts with duplicate items. We advertise unique products. How could I select a distinct set of upcs with descriptions without doing many sub queries? Horrible example:

/* 
eventID     int
groupID     int            // This field is different between fields
upc_ean     numeric(18,0)
description varchar(512)   // Description is slightly different but same info
size        varchar(512)   // Size is slightly different but same info
*/

select A.eventid, A.upc_ean,

       ( select top 1 description 
         from myTable B 
         where B.eventid = A.eventid and B.upc_ean = A.upc_ean) as description,

       ( select top 1 size
         from myTable B
         where B.eventid = A.eventid and B.upc_ean = A.upc_ean) as size

from ( select distinct eventid, upc_ean from myTable) A

Is there any way to do the same thing without sub-queries, somehow joining the two together that doesn’t breed records or show duplicates using eventid and upc_ean as a PK?

Answer 1

You can add an ORDER BY clause if necessary to the OVER portion of the following if necessary.

Using a CTE:

WITH example AS (
  SELECT a.eventid, 
         a.upc_ean,
         a.description,
         a.size,
         ROW_NUMBER() OVER(PARTITION BY a.eventi, a.upc_ean) AS rank
    FROM YOUR_TABLE a)
SELECT x.eventid,
       x.upc_ean,
       x.description, 
       x.size
  FROM example x
 WHERE x.rank = 1

Without the CTE:

SELECT x.eventid,
       x.upc_ean,
       x.description, 
       x.size
  FROM (SELECT a.eventid, 
               a.upc_ean,
               a.description,
               a.size,
               ROW_NUMBER() OVER(PARTITION BY a.eventi, a.upc_ean) AS rank
          FROM YOUR_TABLE a) x
 WHERE x.rank = 1