You can create a "pseudotable" in form of values('pm.25.df', 'http://xyz'),('other', 'http://abc') and then ask the guy to do a merge on the pseudo table in form of:

-- Create testdata
select *
into #sometable
from 
(
    VALUES  (N'PM.23.fd', cast(N'http://dontchange.com' as nvarchar(255)))
    ,   (N'PM.25.df', NULL)
) t (code,link)

-- the script

;merge #sometable as target
using (
    select * from (
        values  ('pm.25.df', 'http://stackoverflow.com') -- here you generate your changes
        ,   ('pm.xyz', 'http://example.com')
        ,   (N'PM.23.fd', N'http://change.com')
        ) x(code, link)
    ) as source
    ON source.code = target.code
when matched and target.link IS NULL then update set link = source.link
when not matched then insert (code, link) values(source.code, source.link)
;

I use merge because you mentioned it, but if you only interested in update, you can just change the query:

UPDATE  target
SET link = source.link
FROM    #sometable target
INNER JOIN (
        VALUES  ('pm.25.df', 'http://stackoverflow.com')
        ,   ('pm.xyz', 'http://example.com')
        ,   (N'PM.23.fd', N'http://change.com')
    )  source (code, link)
    ON  source.code = target.code
WHERE   target.link IS NULL

This only updates codes where link is null.

If you have more than 1000 rows, then values will not work, then you generate a temp table that you insert into