#!/usr/bin/perl
# test OpenFTS parser and  dictionaries
# Oleg Bartunov, oleg@sai.msu.su

use strict;
use Getopt::Std;
use lib qw (.);

use Search::OpenFTS::Parser;

# add dictionary here
my $dict = [
    { 'Integer-Short' => 'Dict::IntegerDict->new(MAXLEN=>6, REJECTLONG=>0)' }
    ,    # shortened
    { 'Integer-Stop' => 'Dict::IntegerDict->new(MAXLEN=>6, REJECTLONG=>1)' }
    ,    # is_stoplexem works
    {
        'Decimal-Short' =>
          'Dict::DecimalDict->new(MAXLENFRAC=>2, REJECTLONG=>0)'
    },    # shortened
    {
        'Decimal-Stop' => 'Dict::DecimalDict->new(MAXLENFRAC=>2, REJECTLONG=>1)'
    },    # is_stoplexem works
    { 'PorterEng' => 'Search::OpenFTS::Dict::PorterEng->new()' }
    ,     # standard stemmer
    { 'PorterEng-Dict' => 'Dict::EngStem->new()' },    # strict
    {
        'Snowball-Eng' =>
'Search::OpenFTS::Dict::Snowball->new(lang=>"english",stop_file=>"/u/megera/app/fts/test-suite/Dict/english.stop" )'
    },
    {
        'Snowball-EngDict' =>
'Dict::StemSnowball->new(lang=>"english",stop_file=>"/u/megera/app/fts/test-suite/Dict/english.stop" )'
    },
    {
        'Snowball-Rus' =>
'Search::OpenFTS::Dict::Snowball->new(lang=>"russian",stop_file=>"/u/megera/app/fts/test-suite/Dict/russian.stop" )'
    },
    {
        'Snowball-RusDict' =>
'Dict::StemSnowball->new(lang=>"russian",stop_file=>"/u/megera/app/fts/test-suite/Dict/russian.stop" )'
    },
    {
        'ISpell-Eng' =>
'Search::OpenFTS::Morph::ISpell->new(aff_file=>"/usr/local/lib/english.aff",dict_file=>"/usr/local/lib/english.dict",stop_file=>"/u/megera/app/fts/test-suite/Dict/english.stop" )'
    },
    {
        'ISpell-Rus' =>
'Search::OpenFTS::Morph::ISpell->new(aff_file=>"/usr/local/lib/russian.aff",dict_file=>"/usr/local/lib/russian.dict",stop_file=>"/u/megera/app/fts/test-suite/Dict/russian.stop" )'
    },
];

my ( $mod, @dclass ) = ();

my %opts         = ();
my %ignore_types = ();
getopts( 'vphs:', \%opts );

if ( $opts{h} ) {
    print
"Usage: $0 [-p] [-s] [-h]\n\t -p - don't use dictionaries\n\t -s - don't process space symbols\n\t -v - show dictionaries\n\t -h - help, this text\n";
    exit 0;
}

$opts{s} = 12
  if ( defined $opts{s} && $opts{s} eq '' );   # ignore space symbols by default
map { $ignore_types{$_} = 1 } split ",", $opts{s};

my $pa = Search::OpenFTS::Parser->new;

if ( !$opts{p} ) {    # -p,  don't need any dictionaries
    my $odict = 'dict';
    foreach my $d (@$dict) {
        foreach my $dcom ( keys %$d ) {
            my $key = $$d{$dcom};
            if ( $key =~ /^(.*?)->new/ ) {
                $mod = $1;
                print "Requested module: $mod\n" if $opts{v};
            }
            eval( 'use ' . $mod . ';' );
            die "Can't use module $mod: $@" if ($@);
            $odict = eval $key;
            push @dclass, [ $odict, $dcom ];
        }
    }

}
my ( $type, $word );

$pa->start_parser( \*STDIN );
while ( ( ( $type, $word ) = $pa->get_word ) && $type ) {
    print "lexeme:$word:$type:", $pa->type_description($type), "\n"
      if !$ignore_types{$type};

    #      print "lexeme:$word:$type:",$pa->type_description( $type ),"\n";
    if ( $type ne 12 && !$opts{p} ) {    # not a space symbol
        foreach my $d (@dclass) {
            my $dict = $$d[0];
            print $$d[1], ":\t", "$word -> ",
              join( ',', ( $dict->lemms($word) ) ), ":\t",
              join(
                ',',
                (
                    map { ( $dict->is_stoplexem($_) == 1 ) ? "stop" : " ok " }
                      $dict->lemms($word)
                )
              ),
              "\n";
        }
    }
}
$pa->end_parser;
print "-----------------------\n" if ( !$opts{p} );