Aersol's plan was very much on the mark (++), and while he wanted you to work out the rest, I thought I would suggest one alternative that might simplify things a bit, and allow you to use multiple entries for the same perl file in the "changes.txt" listing. Where aersol had this:

    # is it a range or a single line?
    if ($info =~ /(\d+)-(\d+)/) {
        $changes{$file} = [$1, $2];
    } else {
        $changes{$file} = [$info];
    }
[download]

you could have this instead:

    if ($info =~ /(\d+)-(\d+)/) {
        $changes{$file} .= " ".join(" ",($1 .. $2))." ";
    } else {
        $changes{$file} .= " $info ";
    }
[download]

This way, each value in %changes is a string of space-separated line numbers, including every line that has changed in a given file, and by using ".=", the string just gets longer each time the same file is mentioned again in changes.txt. Now, when you check this against the "data.txt" listing, instead of aersol's two-way test, you have a single test for all cases:

    ...
    if (defined $changes{$file}) {
        # skip if this $line isn't mentioned in $changes
        next unless ($changes{$file} =~ / $line /);
    }
    ...
[download]