view release on metacpan or  search on metacpan

lib/OBO/Parser/OBOParser.pm  view on Meta::CPAN

#
# Module  : OBOParser.pm
# Purpose : Parse OBO-formatted files.
# License : Copyright (c) 2006-2015 by Erick Antezana. All rights reserved.
#           This program is free software; you can redistribute it and/or
#           modify it under the same terms as Perl itself.
# Contact : Erick Antezana <erick.antezana -@- gmail.com>
#
package OBO::Parser::OBOParser;

use OBO::Core::Dbxref;
use OBO::Core::Instance;
use OBO::Core::Ontology;
use OBO::Core::Relationship;
use OBO::Core::RelationshipType;
use OBO::Core::SubsetDef;
use OBO::Core::SynonymTypeDef;
use OBO::Core::Term;
use OBO::Util::IDspaceSet;
use OBO::Util::Set;
use OBO::XO::OBO_ID;

use Carp;
use Date::Manip qw(ParseDate UnixDate);
use strict;
use warnings;

use open qw(:std :utf8); # Make All I/O Default to UTF-8

$Carp::Verbose = 1;

sub new {
	my $class         = shift;
	my $self          = {};
	
	$self->{OBO_FILE} = undef; # required, (1)
        
	bless ($self, $class);
	return $self;
}

=head2 work

  Usage    - $OBOParser->work($obo_file_path)
  Returns  - the parsed OBO ontology
  Args     - the OBO file to be parsed
  Function - parses an OBO file
  
=cut

sub work {
	my $self = shift;
	if (defined $_[0]) {
		$self->{OBO_FILE} = shift;
	} else {
		croak 'You have to provide an OBO file as input';
	}
	
	open (OBO_FILE, $self->{OBO_FILE}) || croak 'The OBO file (', $self->{OBO_FILE}, ') cannot be opened: ', $!;
	
	$/ = ""; # one paragraph at the time
	chomp(my @chunks = <OBO_FILE>);
	chomp(@chunks);
	close OBO_FILE;

	#
	# Treat OBO file header tags
	#
	my $file_line_number = 0;
	if (defined $chunks[0] && $chunks[0] =~ /^format-version:\s*(.*)/) {

		my @header        = split (/\n/, $chunks[0]);
		$file_line_number = $#header + 2; # amount of lines in the header
		$chunks[0]        = join("\n", @header);

		#
		# format-version
		#
		my $format_version;
		if ($chunks[0] =~ /(^format-version:\s*(.*)\n?)/m) { # required tag
			$format_version = $2;
			$chunks[0]      =~ s/$1//;
		}

		#
		# data_version
		#
		my $data_version;
		if ($chunks[0] =~ /(^data-version:\s*(.*)\n?)/m) {
			$data_version = $2;
			$chunks[0]    =~ s/$1//;
		}
		
		#
		# ontology
		#
		my $ontology_id_space;
		if ($chunks[0] =~ /(^ontology:\s*(.*)\n?)/m) { # as of OBO spec 1.4
			$ontology_id_space = $2;
			$chunks[0]         =~ s/$1//;
		}
		
		#
		# date
		#
		my $date;
		if ($chunks[0] =~ /(^date:\s*(.*)\n?)/m) {
			$date      = $2;
			$chunks[0] =~ s/$1//;
		}
		
		#
		# saved_by
		#
		my $saved_by;
		if ($chunks[0] =~ /(^saved-by:\s*(.*)\n?)/m) {
			$saved_by  = $2;
			$chunks[0] =~ s/$1//;
		}

		#