File Coverage

blib/lib/Treex/Block/Read/BaseAlignedReader.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 Treex::Block::Read::BaseAlignedReader;
$Treex::Block::Read::BaseAlignedReader::VERSION = '0.13095';
use strict;
use warnings;
use Moose;
use Treex::Core::Common;
with 'Treex::Core::DocumentReader';
use Treex::Core::Document;

sub next_document {
    my ($self) = @_;
    return log_fatal "method next_document must be overriden in " . ref($self);
}

has selector => ( isa => 'Treex::Type::Selector', is => 'ro', default => '' );

has file_stem => (
    isa           => 'Str',
    is            => 'ro',
    documentation => 'how to name the loaded documents',
);

# private attributes
has _filenames => (
    isa           => 'HashRef[Str]',
    is            => 'rw',
    init_arg      => undef,
    default       => sub { {} },
    documentation => 'mapping zone_label->filenames to be loaded;'
        . ' automatically initialized from constructor arguments',
);

has _files_per_zone => ( is => 'rw', default => 0 );

has _file_number => (
    isa           => 'Int',
    is            => 'rw',
    default       => 0,
    init_arg      => undef,
    documentation => 'Number of n-tuples of input files loaded so far.',
);

#BUILD is needed for processing generic arguments - now only shortcuts of type langcode_selector
sub BUILD {
    my ( $self, $args ) = @_;
    foreach my $arg ( keys %{$args} ) {
        my ( $lang, $sele ) = ( $arg, '' );
        if ( $arg =~ /_/ ) {
            ( $lang, $sele ) = split /_/, $arg;
        }
        if ( is_lang_code($lang) ) {
            my $files = Treex::Core::Files->new({string => $args->{$arg}});
            if ( !$self->_files_per_zone ) {
                $self->_set_files_per_zone( $files->number_of_files );
            }
            elsif ( $files->number_of_files != $self->_files_per_zone ) {
                log_fatal('All zones must have the same number of files: ' . $files->number_of_files . ' != ' . $self->_files_per_zone);
            }
            $self->_filenames->{$arg} = $files;
        }
        elsif ( $arg =~ /selector|language|scenario/ ) { }
        else                                           { log_warn "$arg is not a zone label (e.g. en_src)"; }
    }
    return;
}

sub current_filenames {
    my ($self) = @_;
    my $n = $self->_file_number;
    return if $n == 0 || $n > $self->_files_per_zone;
    my %result = map { $_ => $self->_filenames->{$_}->filenames->[ $n - 1 ] } keys %{ $self->_filenames };
    return \%result;
}

sub next_filenames {
    my ($self) = @_;
    $self->_set_file_number( $self->_file_number + 1 );
    return $self->current_filenames;
}

sub new_document {
    my ( $self, $load_from ) = @_;
    my %filenames = %{$self->current_filenames()};
    log_fatal "next_filenames() must be called before new_document()" if !%filenames;

    my ( $stem, $file_number ) = ( '', '' );
    my ( $volume, $dirs, $file );
    if ( $self->file_stem ) {
        ( $stem, $file_number ) = ( $self->file_stem, undef );
    }
    else {    # Magical heuristics how to choose default name for a document loaded from several files
        foreach my $zone_label ( keys %filenames ) {
            my $filename = $filenames{$zone_label};
            ( $volume, $dirs, $file ) = File::Spec->splitpath($filename);

            # Delete file extension, e.g.
            # file.01.conll -> file.01
            # cs42.treex.gz -> cs42
            $file =~ s/\.[^.]+(\.gz)?$//;

            # Substitute standard input for noname.
            $file =~ s/^-$/noname/;

            # Heuristically delete indication of language&selector from the filename.
            my ( $lang, $sele ) = ( $zone_label, '' );
            if ( $zone_label =~ /_/ ) {
                ( $lang, $sele ) = split /_/, $zone_label;
            }
            $file =~ s/[_-]?($lang|$sele|$zone_label)[_-]?//gi;
            if ( !$file && !$stem ) {
                $file        = 'noname';
                $file_number = undef;
            }
            if ( $stem !~ /$file/ ) {
                if ( $stem ne '' ) {
                    $stem .= '_';
                }
                $stem .= $file;
            }
        }
    }

    $self->_set_doc_number( $self->doc_number + 1 );
    return Treex::Core::Document->new(
        {
            file_stem => $stem,
            loaded_from => join( ',', values %filenames ),
            defined $file_number ? ( file_number => $file_number )    : (),
            defined $dirs        ? ( path        => $volume . $dirs ) : (),
            defined $load_from   ? ( filename    => $load_from )      : (),
        }
    );
}

sub number_of_documents {
    my $self = shift;
    return $self->_files_per_zone;
}

after 'restart' => sub {
    my $self = shift;
    $self->_set_file_number(0);
};

1;

__END__