Arff::Util - ARFF files processing utilities.

  use ARFF::Util;

  my $arff_object = ARFF::Util->new();
  # load .arff formatted file into the buffer, and return pointer to buffer
  $arff_hash = $arff_object->load_arff($file_address);

  # save given buffer into the .arff formatted file
  $arff_object->save_arff($arff_hash, $file_address);

 This is a buffer hash
 Structure of hash:

 relation -> {
		attributes => [
				{
					attribute_name => x1,
					attribute_type => x2,
				},...
			      ]
		records    => [
				{
					attribute_name1 => value1,
					attribute_name2 => value2,...
				},...
			      ]
		data_record_count => x
	      }
=cut

 - Ondrej Dusek

    perldoc Arff::Util

package Arff::Util;

use Moose;

use Data::Dumper;
use String::Util ':all';
use Text::CSV;

our $VERSION = '1.2';

has relation => (
	is => 'rw',
	isa => 'HashRef',
	default => sub{{
		attributes => [],
		records => [],
		data_record_count => 0
	}}
);

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

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

sub load_arff {
	my ($self, $arff_file) = @_;
	my $status = q/normal/;
	if($self->debug_mode){
 		print "Loading $arff_file ...\n";
	}
	my $record_count = 0;
	my $attribute_count = 0;
	my $line_counter = 1;
	my $relation = $self->relation;
        my $line_parser = Text::CSV->new({binary => 1, quote_char => '\'', escape_char => '\\', allow_loose_escapes => 1}) or die "Cannot use CSV Parser: ".Text::CSV->error_diag ();
    
	local *FILE;
	open(FILE, $arff_file)  or die $!;;

	while (my $current_line = <FILE>)
	{
		$current_line = trim($current_line);
		#Check for comments
		if ($current_line =~ /^\s*%/i)
		{
			$status = q/comment/;
		}
		elsif ($current_line =~ /^\s*\@RELATION\s+(\S*)/i )
		{
			$relation->{relation_name} = $1;
		}
		elsif ($current_line =~ /^\s*\@ATTRIBUTE\s+(\S*)\s+(\S*)/i )
		{
			if(!$relation->{attributes}){
				$relation->{attributes} = [];
			}
			my $attribute = { "attribute_name" => $1, "attribute_type" => $2};
			my $attributes = $relation -> {"attributes"};
 			#log_msg(Dumper $attribute);

			push  @$attributes , $attribute;
			$attribute_count ++;
		}
		elsif ($current_line =~ /^\s*\@DATA(\.*)/i )
		{
			$status = q/data/;
		}
		elsif($status eq q/data/ && $current_line ne q//){
			if(!$relation -> {"records"}){
				$relation -> {"records"} = [];
			}
			#log_msg("extracting data $current_line");
                        $line_parser->parse($current_line);                        
			my @data_parts = $line_parser->fields(); # split(/,/, $current_line);

			my $attributes = $relation -> {"attributes"};
 			my $records = $relation -> {"records"};

			my $cur_record={};

 			#log_msg("DATA PARTS".Dumper @data_parts);
			if(scalar @data_parts == $attribute_count){
				for(my $i=0;$i<=$#data_parts;$i++)
				{
					$cur_record->{$$attributes[$i]->{"attribute_name"}} = trim($data_parts[$i]);
				}
				#log_msg("parts: ".$#data_parts);
				$record_count ++;
				push  @$records , $cur_record;
			}else
			{
				if($self->debug_mode){
					print "Line $line_counter : Invalid data record: $current_line Contains ".(scalar @data_parts)." Expected $attribute_count\n";
				}
				$self->error_count($self->error_count+1);
			}
			
		}
		$line_counter ++;
	}
	
	$relation -> {"data_record_count"} = $record_count;
 	$relation -> {"attribute_count"} = $attribute_count;

	if($self->debug_mode){
		eval("use Devel::Size qw(size total_size)");

		print "$arff_file loaded with ".$self->error_count." error(s).\n";
	}
	$self->relation($relation);
	return $relation;
}

sub save_arff {
	my ($self, $buffer, $arff_file) = @_;
	
	local *FILE;
	
	open(FILE, ">$arff_file");
	
	my $record_count = 0;
 	
	if($self->debug_mode){
		print "Writing buffer to $arff_file ...\n";
	}
	
	if($buffer->{relation_name})
	{
		print FILE q/@RELATION / . $buffer->{relation_name} . "\n";
	}
	print FILE "\n";
	print FILE "\n";

	if($buffer->{attributes}){
		foreach my $attribute (@{$buffer -> {"attributes"}}){
			print FILE q/@ATTRIBUTE / . $attribute->{attribute_name} . q/ / . $attribute->{attribute_type} . "\n";
		}
		print FILE "\n";
		print FILE "\n";
	
		if($buffer->{records}){
			print FILE "\@DATA\n";
			print FILE "\n";

			foreach my $record (@{$buffer -> {"records"}}){
				my $record_string = q//;
				foreach my $attribute (@{$buffer -> {"attributes"}}){
					if($record->{$attribute->{attribute_name}}){
						$record_string .= $record->{$attribute->{attribute_name}} . q/,/;
					}else{
						if($self->debug_mode){
							print "Invali buffer passed, ".$attribute->{attribute_name}." is not defined for record... write UNKNOWN\n";
						}
						$record_string .= q/UNKNOWN,/;
						$self->error_count($self->error_count+1);
					}
				}

				$record_string =~ s/,$//;

				print FILE $record_string . "\n";
				$record_count ++;
			}
		}
	}

	if($self->debug_mode){
		eval("use Devel::Size qw(size total_size)");
		
		print "Buffer saved to $arff_file with ".$self->error_count." error(s).\n";
		print "Data rows count: ".$record_count."\n";
	}

	return 1;
}

1; # End of Arff::Util