#!/usr/bin/perl 
## -CIO

#use POSIX qw(locale_h);
#setlocale(&POSIX::LC_ALL, "pt_PT");
#use locale;

use Lingua::Jspell;
use utf8::all;
use Lingua::PT::ProperNames;
use strict;

my $pt_dict = Lingua::Jspell->new("port");
my $en_dict = Lingua::Jspell->new("eng");

my %pnlist=();
my %pnlistD=();

my $file = shift;
my $tab = do("tagPnTable"); die("$!$@\n") if ($! or $@);

forPN({in=>$file, sep=>'>', 
       out=>"$file.out.txt", 
       encoding=>"utf8",
       out=>"$file.out", 
       t=>"double"}, \&procPn,\&procPn2);

sub procPn{my $a=shift;
    if($a =~ /(\w+)/){
         my $f=$1;
         my @a=$pt_dict->fea($f);
         my %b = Lingua::Jspell::onethat({CAT => "np"}, @a);
         if(%b){if (defined $b{SEM}){"<pn t='$b{SEM}'>$a</pn>"}
                else                {"<pn t='np??'>$a</pn>"}}
         elsif(defined $tab->{$f})  {"<pn t='$tab->{$f}'>$a</pn>"}
         else {
            @a = $en_dict->fea($f);
            %b = Lingua::Jspell::onethat({CAT => "np"}, @a);
            if(%b){if (defined $b{SEM}){"<pn L='en' t='$b{SEM}'>$a</pn>"}
                   else {print STDERR "?? ",n($a)," -- english but no SEM\n";
                         "<pn t='np??'>$a</pn>"}}
            else { print STDERR "?? $a is undefined\n";
               "<pn t='?'>$a</pn>"}}}
    else {"<pn t='??'>$a</pn>"}
}

sub procPn2{my $a=shift;
    if($a =~ /(\w+)/){
         my @a=$pt_dict->fea($1);
         my %b = Lingua::Jspell::onethat({CAT => "np"}, @a);
         if(%b || scalar @a == 0){procPn($a)}
         elsif($a =~ /(\w+\s+)(\S.*)/){ return($1 . procPn($2));}
         else{ return  undef ; }}
    else {"<pn t='???'>$a</pn>"}
}

sub n{ my $a=shift;
 $a =~ s/\s+/ /g;
 $a =~ s/^ | $//g;
 $a;
}