CSV of course immediately triggers "use a csv module", so that part is easy: Text::xSV or one of the CSV modules.

"most common occurence" should trigger "hash for uniqueness". Consider:

my @items = qw(foo poot bar foo baz poot foo);
my %counts;

++$counts{$_} for @items;

print "$_: $counts{$_}\n" for sort keys %counts;
[download]

The trick then is to figure out how to process groups coming at you a row at a time. There are a bundle of ways of doing that. A simple way is to perform the processing for the previous group each time you see a new group header (unless there was no previous group):

use strict;

my @rowTypes = (1, 2, 3, 3, 3, 1, 2, 3, 3, 1, 2, 1, 3);
my $count;

for my $rowType (@rowTypes) {
    if ($rowType == 1) {
        # New group
        next unless defined $count; # No processing for first or empty
+ group
        print "$count ";
        $count = 0; # Reset count. Set to undef to skip empty group
        next;
    }
    
    ++$count if $rowType == 3;
}

print "$count " if defined $count; # Need to process the last group
[download]