#!/usr/bin/env perl

# ABSTRACT: Script for building language models
# PODNAME: yali-builder

use strict;
use warnings;

use Lingua::YALI::Builder;
use Lingua::YALI;

use Getopt::Long;
use Pod::Usage;
use Carp;
use File::Basename;

our $VERSION = '0.012'; # VERSION

my $output_file = undef;
my $file_list = undef;
my $input_file = undef;
my $count = undef;
my $ngram = 4;
my $help = 0;

my $result = GetOptions("filelist=s" => \$file_list,
                     "input|i=s"   => \$input_file,
                     "output|o=s"   => \$output_file,
                     "count|c=i" => \$count,
                     "ngram|n=i"  => \$ngram,
                     "help|h"  => \$help
) || pod2usage( -exitval => 105);

if ($help) {
    pod2usage(-exitval => 0);
}

# reading input from STDIN is default
if ( ! defined($input_file) && ! defined($file_list) ) {
    $input_file = "-";
}

# it is incorrect when parameter filelist and input are specified
if ( defined($input_file) && defined($file_list) ) {
    pod2usage(-exitval => 101, -msg     => "Options --filelist and --input can not be specified in the same time.");
}

# output file is required
if ( ! defined($output_file) ) {
    pod2usage(-exitval => 101, -msg     => "Output file --output has to be specified.");
}

# check parameter range
if ( defined($count) && $count < 1 ) {
    pod2usage(-exitval => 101, -msg     => "The number of n-grams --count has to be positive. $count was used.");
}

if ( $ngram < 1 ) {
    pod2usage(-exitval => 101, -msg     => "The n-gram size --ngram has to be positive. $ngram was used.");
}

# construct builder
my $builder = Lingua::YALI::Builder->new(ngrams=>[$ngram]);

# train models from input
if ( defined($input_file) ) {
    if ( $input_file eq "-" ) {
        $builder->train_handle(\*STDIN);
    } else {
        $builder->train_file($input_file);
    }
} elsif ( defined($file_list) ) {
    my $fh_list = undef;
    if ( $file_list eq "-" ) {
        $fh_list = \*STDIN;
    } else {
        open($fh_list, "<", $file_list) or croak $!;
    }
    while ( <$fh_list> ) {
        chomp;
        $builder->train_file($_);
    }
}

# store results
$builder->store($output_file, $ngram, $count);


__END__
=pod

=head1 NAME

yali-builder - Script for building language models

=head1 VERSION

version 0.012

=head1 SYNOPSIS

yali-builder [options]

Train models for language identification with yali-identifier.

Options:

 -i, --input=F         input file. When F is -, read standard input (default -).
     --filelist=F      input files are read from F. When F is -, read them from standard input. 
 -o, --output=F        output file  
 -n, --ngram=N         n-gram size (default 4)
 -c, --count           the number of n-grams
 -h, --help            prints documentation

More examples are available at L<http://search.cpan.org/perldoc?Lingua%3A%3AYALI%3A%3AExamples>

Version: 0.012 (2012-09-20)

=head1 AUTHOR

Martin Majlis <martin@majlis.cz>

=head1 COPYRIGHT AND LICENSE

This software is Copyright (c) 2012 by Martin Majlis.

This is free software, licensed under:

  The (three-clause) BSD License

=cut