#!perl

=head2 NAME

Find bad 043 -- looks through output file for lines without 7-character 043 subfields.

=head2

Input file should be result (export file) from fieldextraction.pl or similar file.
Output file is list of lines with subfields longer or shorter than 7 characters.

=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;
#remove count, tag, and first subfield code
$line =~ s/^\d+?\t\d{3}\s[ \d][ \d]\s\@\w\t//;
my @linearray = split ("\t\t", $line);
#my @controls = $linearray[1];
my @linesubfields = split (\t, $linearray[0]);

foreach $linesubfield (@linesubfields) {
#subfield code--make sure it's subfield 'a'
if (($linesubfield =~ /\@/) && ($linesubfield !~ /\@a/)) 
{print OUT "$line\n";
$linecount++;
print "im in codes\n";
}
#subfield data
else {
unless ($linesubfield =~ /[\w-]{7}/) {
print OUT "$line\n";
print "Im in data\n";
$linecount++;
}
}
} #foreach subfield and subfield code
}
close IN;
close OUT;
print "$linecount lines identified\n";
print "Press Enter to quit";
<>;