If there is only one data field that can contain commas, I would recommend a slightly different approach.

Split the line up on every comma.
Grab the last M fields from the back.
Grab the first N fields from the front.
Join whatever is left using commas into a single field
Join the three pieces together with tabs.

Here is a quick example to get you up to speed.

#!/usr/bin/perl
use strict;

my $line = 'john,doe,manager,Pepsi Cola, inc.,Detroit,Michigan';
my @fields = split(/,/,$line);      # split on commas

my @last2  = splice(@fields,-2);    # Grab two fields off the back
my @first3 = splice(@fields, 0,3);  # Grab three fields from the front
my @middle = @fields;               # Whatever is left is the difficul
+t "company" data

my $middlefield = join(',',@middle);  # join the middle portion back t
+ogether

print "First:  $_\n" for (@first3);
print "Middle: $_\n" for (@middle);
print "Last:   $_\n" for (@last2);
print "\n";

my $newline = join("\t",@first3,$middlefield,@last2);  # generate new 
+line

print "Old: $line\n";
print "New: $newline\n";
[download]

I didn't quite understand your regex, but if you want to collapse multiple spaces down into a single one you might try s/\s+/ /g.

-Blake