CGI does not support parsing multipart data that way with new().

Final update: (I erased a whole lot of tomfoolery with the CGI module. Forget it.) What you need is some MIME:: module, not CGI.pm, in fact.

You will probably need to do something like my($content_type) = grep /^Content-type: /, split /\r?\n/, $r->headers; in there too because that's where the boundary code (in your case, 200672192830548) lives.