The parsing of the CSV data as such should not be too hard a problem. Just use Text::CSV_XS or Text::CSV. Now the only thing left to do is to come up with a criterium (or criteria) than make the headers stand out/be different from "normal" data.

my $csv = Text::CSV_XS->new ({ binary => 1, auto_diag => 1});
my @hdr;
while (my $row = $csv->getline ($fh)) {
    # example: header lines start with A-Z, data doesn't
    if (!@hdr or $row->[0] =~ m/^[A-Z]/) {
        @hdr = @{$row};
        next;
        }
    # just an example
    my %hash;
    @hash{@hdr} = @$row;
    }
[download]