use Lingua::TH::Wordbreak;
use Getopt::Std;
use Encode::Registry;
use IO::File;
use Benchmark;

getopts('a:c:d:m:p:stu:z:');

unless ($opt_t || defined $ARGV[1])
{
    die <<'EOT';
    thaisplit [-a file] [-c hex] [-m mapping] [-s] [-u file] infile outfile
Word breaks a text according to the dictionary given.

    -a file     Additional dictionary entries
    -c hex      byte to use for wordbreak [/]
    -m mapping  Mapping identifier to use
    -s          skip SFMs
    -u file     List of unknown words encountered
EOT
}

$time_start = new Benchmark if ($opt_z & 2);

$opt_m ||= 'thai-upc';
$enc = find_encoding($opt_m) || die "Can't find $opt_m encoding";
if ($opt_c)
{ $opt_c = $enc->decode(pack('C', hex($opt_c))); }
else
{ $opt_c = '/'; }

if ($opt_a)
{
    my (@words);
    $fh = IO::File->new("< $opt_a") || die "Can't open $opt_a for reading";
    while (<$fh>)
    {
        chomp;
        next unless $_;
        push (@words, $enc->decode($_));
    }
    $fh->close();
    Lingua::TH::Wordbreak::add_words(@words);
}

unless ($opt_t)
{
    $fhin = IO::File->new("< $ARGV[0]") || die "Can't open $ARGV[0]";
    $fhout = IO::File->new("> $ARGV[1]") || die "Can't write to $ARGV[1]";
    $trees = $opt_p ? {} : undef;
    
    $count = 0;
    
    while (<$fhin>)
    {
        use utf8;
        
        print STDERR "." if (++$count % 100 == 0);
        $res = '';
        chomp;
        $_ = $enc->decode($_);
        if ($opt_s)
        {
            if (s/^\\v\s+([0-9-]+)\s*//o)
            { $res = "\\v $1 "; }
            elsif (s/^\\(\S+)\s*//o)
            { $res = "\\$1 "; }
        }
        foreach $w (split(/\s+/, $_))
        { $res .= Lingua::TH::Wordbreak::wordbreak($w, $opt_c, undef, undef, $trees) . ($opt_c ? "$opt_c " : " "); }
        $res =~ s/.$//o;
#        $fhout->print($enc->encode($res) . "\n");
        $fhout->print("$res\n");
        if ($opt_p)
        {
            foreach $w (keys %{$trees})
            { delete $trees->{$w} unless (length($w) <= $opt_p * 3); }
        }
    }
    
    print STDERR "Trees are: " . scalar(%{$trees}) . "\n" if ($opt_p && $opt_z & 1);
    $fhin->close();
    $fhout->close();
}
else
{
    $test = 'อานุุภาพของพระวิญญาณบริสุทธิ์แก‹เหล‹าอัครทูต';
    print Lingua::TH::Wordbreak::wordbreak($enc->decode($test), $opt_c);
}
if ($opt_u && scalar keys %unk)
{
    $fh = IO::File->new("> $opt_u") || die "Can't write $opt_u";
    foreach $w (sort keys %unk)
    { $fh->print($enc->encode($w) . "\n"); }
    $fh->close();
}

if ($opt_z & 2)
{
    $time_end = new Benchmark;
    print STDERR "Timings: " . timestr(timediff($time_end, $time_start)) . "\n";
}