You'll get lots of "Use XYZ::CSV module" replies despite that they don't help with your main problems, and add unnecessary complexity.

This is how I would tackle the problem described:

#! perl -sl
use strict;

while( <DATA> ) {
    my( $name, $number, $type, $text ) = m[^([^,]+),(\d+),(\d+|"[^"]+"
+),(\d+)?$]; #"

    $type =~ s["([^"]+)"][$1]; #"

    printf "%s layernumber %d datatype %s%s%s\n",
        $name, $number,
        ( $type =~ m[,] ? join( ' datatype ', split ',', $type ) : $ty
+pe ),
        ( defined( $text ) ? ' text ' . $text : '' ),
        ( $name eq 'NWER' ? ';' : ',' )
    ;
}

__DATA__
NPLUS,32,0,
NW,41,0,1
NWER,51,"0,1,2",12
[download]

Produces:

C:\test>junk44
NPLUS layernumber 32 datatype 0,
NW layernumber 41 datatype 0 text 1,
NWER layernumber 51 datatype 0 datatype 1 datatype 2 text 12;
[download]