#!perl

=head2

Given fieldextraction result (for field 042) as input ,
Identifies and outputs lines without either
pcc, lcac, or lccopycat.


=cut 

print "Input file: ";
$inputfile = <>;
chomp $inputfile;
$inputfile =~ s/^\"(.*)\"$/$1/;
print "Export file: ";
$exportfile = <>;
chomp $exportfile;
$exportfile =~ s/^\"(.*)\"$/$1/;

open (IN, "<$inputfile") || die ("can't open in");

open (OUT, ">$exportfile") || die ("can't open out");
my $linecount = 0;
while (my $line = <IN>) {
chomp $line;
#remove extra spaces between tabs
$line =~ s/\t\s{7}\@/\t\@/g;

#$line =~ s/^\d+?\t\d{3}\s[ \d][ \d]\s\@\w\t//;

my @linearray = split ("\t\t", $line);
my @controlnoarray = split ("\t", $linearray[1]);
if ($linearray[0] !~ /(pcc)|(lcac)|(lccopycat)/) {
print "$linearray[0] is first element, with ", scalar @controlnoarray/2, " elements\n";
print OUT "$line\n";
$linecount++;
}
}


close IN;

close OUT;
print "$linecount lines identified\n";
print "Press Enter to quit";
<>;