#!perl

=head2

Create control number file.

Takes field extraction output file as input.
Extracts control numbers from each line.
Outputs one control number per line.

=cut 

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

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

open (OUT, ">$exportfile") || die ("can't open out");
#if using MacPerl, set creator and type to BBEdit and Text
if ($^O eq 'MacOS') {
MacPerl::SetFileInfo('R*ch', 'TEXT', $exportfile);
}


my $linecount = 0;
my $controlnocount = 0;
%counts;
while (my $line = <IN>) {
chomp $line;
#remove ending tabs
$line =~ s/\t*$//;
#remove extra spaces between tabs
$line =~ s/\t\s{7}\@/\t\@/g;
#two tabs separate control numbers from headings
my @linearray = split ("\t\t", $line);
#if heading line had multiple controlnos, split each into its own array position
my @controlnoarray = split ("\t", $linearray[1]);

foreach (my $i = 0; $i <= $#controlnoarray; $i++) {
#remove extra space from end of each controlno
$controlnoarray[$i] =~ s/\s$//;
#remove beginning spaces from control nos after first
$controlnoarray[$i] =~ s/^\s// if ($i>0);

print OUT "$controlnoarray[$i]\n";

} 
$linecount++;
$controlnocount += scalar @controlnoarray;
}
close IN;

close OUT;
print "$linecount headings identified\n with $controlnocount control numbers\n";
print "Press Enter to quit";
<>;


=head1 LICENSE

This code may be distributed under the same terms as Perl itself. 

Please note that this code is not a product of or supported by the 
employers of the various contributors to the code.

=head1 AUTHOR

Bryan Baldus
eija [at] inwave [dot] com

Copyright (c) 2003-2004

=cut