this assumes your filenames consist only of a-z, A-Z and _ (which is the meaning of \w) and numbers 0-9 (which is the meaning of \d).

oneliner: replaces all orrcurences in "infile" and leaves "infile.bak" with the original contents.

perl -npi.bak -e "s:([\w\d]+\.data):testname/$1:g" infile
[download]

or as a script:

use strict;
$testname="replacement";

open IN, "infile" or die $!;
open OUT, ">outfile" or die $!;

while ( <IN> )
{
    s:([\w\d]+\.data):$testname/$1:g
    print OUT $_;
}

close IN;
close OUT;
[download]

Update:
As borisz /msg-ed me, \w includes \d. thus [\w\d]is equivavelent to [\w].