Your question is a bit wide, as you don't tell us how your data is structured. I recommend http://www.activestate.com and their Komodo GUI, which has a regex explorer to become accustomed to regular expressions.

For the sake of the node, let's assume that your data is CSV data (see Text::xSV, also on CPAN), and that you don't want (for some obscure reason) to use Text::XSV.

What I'm going to do now is to construct a regular expression that skips n semicolons and then tries to match =nnn; :

  $line =~ m/(?:      # Collect heading fields
               [^;]+  # delimited by semicolons
               ;
             ){1}     # In your case, there is only one field before t
+he interesting field
             =        # followed by a "="
             (\d+)    # and followed by our numbers, stuff them into $
+1
             ;        # and delimited by another semicolon
            /x;
[download]

Another good approach would have been to split() the line on ";" and then munge the second element of that list afterwards.