File Coverage

blib/lib/NLP/Service.pm

Criterion	Covered	Total	%
statement	7	9	77.7
branch			n/a
condition			n/a
subroutine	3	3	100.0
pod			n/a
total	10	12	83.3

package NLP::Service;

use 5.010000;
use feature ':5.10';
use common::sense;
use Carp ();

BEGIN {
    use Exporter();
    our @ISA     = qw(Exporter);
    our $VERSION = '0.02';
    use NLP::StanfordParser;
}

use Dancer qw(:tests); # we do not want the tests exporting the wrong functions.
use Dancer::Plugin::REST;

my %_nlp = ();

prepare_serializer_for_format;

any [qw/get post/] => '/' => sub {

    #TODO: show a UI form based thing for easy use for the end user.
    return 'This is ' . config->{appname} . "\n";
};

any [qw/get post/] => '/nlp/models.:format' => sub {
    return [ keys %_nlp ];
};

any [qw/get post/] => '/nlp/languages.:format' => sub {
    return [qw/en/];
};

any [qw/get post/] => '/nlp/info.:format' => sub {
    return {
        version        => $NLP::Service::VERSION,
        nlplib_name    => 'Stanford Parser',
        nlplib_source  => PARSER_SOURCE_URI,
        nlplib_release => PARSER_RELEASE_DATE,
    };
};

any [qw/get post/] => '/nlp/relations.:format' => sub {
    return NLP::StanfordParser::relations();
};

#Dancer::forward does not forward the parameters, hence we have to explicitly
#forward them.
any [qw/get post/] => '/nlp/parse.:format' => sub {
    my $model = 'en_pcfg';
    my $route = "/nlp/parse/$model." . params->{format};
    debug "Forwarding to $route";
    if ( request->{method} eq 'GET' ) {
        return forward $route,
          {
            format => params->{format},
            model  => $model,
            data   => params->{data}
          };
    } else {

        # HACK inserted until Dancer's forwarding bug gets fixed.
        # https://github.com/sukria/Dancer/pull/545
        #
        my $data = params->{data};
        $data =~ s/^\s+//g;
        $data =~ s/\s+$//g;
        my $data = params->{data}
          or return send_error( { error => "Empty 'data' parameter" }, 500 );
        debug "Data is $data\n";
        if ( defined $_nlp{$model} ) {
            my $str = $_nlp{$model}->parse($data);
            my $aref = eval $str or Carp::carp "Unable to eval $str";
            return defined $aref ? $aref : "$str\n";
        }
        return send_error( { error => "Invalid NLP object for $model" }, 500 );
    }
};

any [qw/get post/] => '/nlp/parse/:model.:format' => sub {
    my $model = params->{model};
    debug "Model is $model";
    return send_error( { error => "Unknown parsing model $model" }, 500 )
      unless defined $_nlp{$model};
    my $data = params->{data};
    $data =~ s/^\s+//g;
    $data =~ s/\s+$//g;
    my $data = params->{data}
      or return send_error( { error => "Empty 'data' parameter" }, 500 );
    debug "Data is $data\n";

    if ( defined $_nlp{$model} ) {
        my $str = $_nlp{$model}->parse($data);
        my $aref = eval $str or Carp::carp "Unable to eval $str";
        return defined $aref ? $aref : "$str\n";
    }
    return send_error( { error => "Invalid NLP object for $model" }, 500 );
};

sub load_models {
    my ( $force, $jarpath ) = @_;
    say 'Forcing loading of all NLP models.' if $force;
    %_nlp = ();
    $_nlp{en_pcfg} = new NLP::StanfordParser( model => MODEL_EN_PCFG )
      or Carp::croak 'Unable to create MODEL_EN_PCFG for NLP::StanfordParser';

    # PCFG load times are reasonable ~ 5 sec. We force load on startup.
    $_nlp{en_pcfg}->parser if $force;
    $_nlp{en_factored} = new NLP::StanfordParser( model => MODEL_EN_FACTORED )
      or Carp::croak
      'Unable to create MODEL_EN_FACTORED for NLP::StanfordParser';

    # Factored load times can be quite slow ~ 30 sec. We force load on startup.
    $_nlp{en_factored}->parser if $force;

    # PCFG WSJ takes ~ 2-3 seconds to load
    $_nlp{en_pcfgwsj} = new NLP::StanfordParser( model => MODEL_EN_PCFG_WSJ )
      or Carp::croak
      'Unable to create MODEL_EN_PCFG_WSJ for NLP::StanfordParser';
    $_nlp{en_pcfgwsj}->parser if $force;
    $_nlp{en_factoredwsj} =
         new NLP::StanfordParser( model => MODEL_EN_FACTORED_WSJ )
      or Carp::croak
      'Unable to create MODEL_EN_FACTORED_WSJ for NLP::StanfordParser';

    # FACTORED WSJ takes ~ 20 seconds to load
    $_nlp{en_factoredwsj}->parser if $force;
    return unless defined wantarray;    # void context returns nothing
    return wantarray ? %_nlp : scalar( keys(%_nlp) );
}

sub run {
    my %args   = @_;
    my $force  = $args{force} if scalar( keys(%args) );
    my $config = $args{config} if scalar( keys(%args) );
    if ( defined $config and ref $config eq 'HASH' ) {
        map { set( $_ => $config->{$_} ) } keys %$config;
    } else {
        set log          => 'error';
        set logger       => 'console';
        set show_errors  => 1;
        set startup_info => 0;
    }
    load_models($force);
    dance;    # invoke Dancer
}

1;
__END__