find-new-titles

#!/usr/local/bin/perl
#
# CC-BY Stefan Müller (HU Berlin)
#


#use strict;
use warnings;

use Text::BibTeX;
 
#use utf8;


my $bibfile1 = Text::BibTeX::File->new(shift @ARGV);


#my $checkedfile = Text::BibTeX::File->new(">checked.bib");
my $new_file = Text::BibTeX::File->new(">new.bib");



#binmode(STDOUT, ":utf8");
 
#my entry;

#@entries = ();

my %seen;

while ($entry1 = Text::BibTeX::Entry->new($bibfile1))
{
   next unless $entry1->parse_ok;

   #print $entry->get('title'); print "\n";
 
   #push(@entries, ($title));

if ($entry1->exists ('title')) {
   $title = normalize_title($entry1->get('title'));
   $seen{$title}++;
} 

}

my $bibfile2 = Text::BibTeX::File->new(shift @ARGV);


while ($entry2 = Text::BibTeX::Entry->new($bibfile2))
{
   next unless $entry2->parse_ok;

   #print $entry->get('title'); print "\n";
 
   #push(@entries, ($title));

   if ($entry2->exists ('title')) {
       $title = normalize_title($entry2->get('title'));
     if (!$seen{$title}) {
      $entry2->set_key($entry2->key . "hpsg-bib"); #new name space
      $entry2->write ($new_file);
     }
 }


}


sub normalize_title {
  local($string) = @_;

  $string =~ s/\\emph//g;

  $string =~ s/\\textsc//g;

  $string =~ s/"=/-/g;

  $string =~ s/-//g;

  $string =~ s/ //g;

  $string =~ s/[\{\}\$#@~!&*()\[\];.,:?^ `\\\/]+//g;

  return lc $string;
}