Question

I have table Users which contains columns Login and Phone and I have table GeoPhone which contains columns Mask and Region.

Phone is stored as varchar (“[phone]”). Mask is stored as varchar (“+38098”). Masks are overlapped, for instance, +380 and +38098.

I need a View that joins Users and GeoPhone tables and assigns Region for each Login by longest matched Mask.

I wrote SQL statement below but result contains duplicated Logins and I do not know how to avoid this:

SELECT TOP (5000) 
    Users.Login, Users.Phone, GeoPhone.Region, LEN(GeoPhone.Mask) AS MaskLen
FROM   Users 
INNER JOIN GeoPhone ON LEFT(Users.Phone, LEN(GeoPhone.Mask)) = GeoPhone.Mask
ORDER BY Users.Login, MaskLen DESC

Please help to improve it.

Answer 1

One way to do this is to use Cross Apply in SQL 2005 or later.

SELECT  
    TOP (5000) Users.Login, 
    Users.Phone, 
    GeoPhone.Region, 
    LEN(GeoPhone.Mask) AS MaskLen 
FROM    Users
    CROSS APPLY
    (
        SELECT  TOP 1 Region,Mask
        FROM    geophone
        WHERE   LEFT (users.phone, Len(geophone.mask)) = geophone.mask 
        ORDER BY
            Len(geophone.mask) DESC
        ) geophone

You should note that it only returns one record from geophone. If there are two records in geophone.mask that tie for the length of mask you won’t be able to determine which record it will return

Here’s a solution that uses MAX() and its realtively portable

SELECT TOP (5000) users.LOGIN, 
                  users.phone, 
                  geophone.region, 
                  Len(geophone.mask) AS masklen 
FROM   users 
       INNER JOIN geophone 
         ON LEFT(users.phone, Len(geophone.mask)) = geophone.mask 
       INNER JOIN (SELECT MAX(Len(geophone.mask)) maxlength, 
                          users.phone 
                   FROM   users 
                          INNER JOIN geophone 
                            ON LEFT(users.phone, Len(geophone.mask)) = 
                               geophone.mask 
                   GROUP  BY users.phone) t 
         ON t.phone = users.phone 
            AND t.maxlength = Len(geophone.mask) 

However if two rows in geophone Tie for longest length both results will be returned. I don’t know if this is better or worse for you.

You can also use ROW_NUMBER in SQL 2005 or later