#!perl

=head2 Name and Description

Extract File Cleaning (outdated)
This has been replaced by cleanexport010andgreater.pl
and controlfieldscleaning.txt.

Asks for file for cleaning (output from field extraction script works best).
Asks for output filename.
Removes count of field, tag numbering and indicators, and first subfield a code.
Fails if first subfield is not subfield a.
To do: clean control numbers in some way (remove spaces that shouldn't be there).
To do: Sort lines by $a data.


=cut

print "Input file:";
my $extractedinputfile = <>;
$extractedinputfile =~ s/^\"(.*)\"$/$1/;
chomp $extractedinputfile;
print "Output file:";
my $extractedoutfile = <>;
chomp $extractedoutfile;
$extractedoutfile =~ s/^\"(.*)\"$/$1/;
#read line from file
open (EXTRACTEDFILE, "<$extractedinputfile") or die "can't open controlnofile";
open (OUTPUTFILE, ">$extractedoutfile") or die "can't open output";
while (my $line = <EXTRACTEDFILE>){

$line =~ s/^[\d]+?\t[\d]{3}?\s[ \d][ \d]\s\@a\t//;
$line =~ s/\t\s{7}\@/\t\@/g;
print OUTPUTFILE "$line";
} #while
print "done";

print "\n\nPress Enter to quit";
<>;