File Coverage

blib/lib/Text/DocumentCollection.pm

Criterion	Covered	Total	%
statement	10	12	83.3
branch			n/a
condition			n/a
subroutine	4	4	100.0
pod			n/a
total	14	16	87.5

package Text::DocumentCollection;

use strict;
use warnings;
use Text::Document;

use DB_File;

use v5.6.0;

$Text::DocumentCollection::VERSION = 1.02;

sub new
{
	my $class = shift;
	my %self = @_;
	my $self = \%self;
	bless $self, $class;

	defined( $self->{file} ) or die( __PACKAGE__ . '::new : '
		. "keyword 'file' is mandatory"
	);

	my %persistent;

	(tie %persistent, 'DB_File', $self->{file}) or die( __PACKAGE__
		. '::new : '
		. "Cannot tie persistent hash: $!"
	);

	$self->{pdocs} = \%persistent;

	return $self;
}

sub NewFromDB
{
	my $self = Text::DocumentCollection::new(@_);
	while( my @kv = each %{$self->{pdocs}} ){
		$self->{docs}->{$kv[0]} =
			Text::Document::NewFromString( $kv[1] );
	}
	return $self;
}

sub Add
{
	my $self = shift;
	my ($key,$doc) = @_;

	if( defined( $self->{docs}->{$key} ) ){
		die( __PACKAGE__ . '::Add : '
			. "document `$key' is already in this collection"
		);
	}

	$self->{docs}->{$key} = $doc;

	delete $self->{IDF};

	$self->{pdocs}->{$key} = $doc->WriteToString();

	return $doc;
}

sub Delete
{
	my $self = shift;
	my ($key) = @_;

	if( not defined( $self->{docs}->{$key} ) ){
		return undef;
	}
	delete $self->{docs}->{$key};
	delete $self->{pdocs}->{$key};
	return 1;
}

sub EnumerateV
{
	my $self = shift;
	my ($callback,$rock) = @_;

	my @result = ();
	while( my @kv = each %{$self->{docs}} ){
		my @l = &{$callback}( $self, $kv[0], $kv[1], $rock );
		push @result, @l;
	}
	return @result;
}

sub IDF_Help
{
	my $self = shift;
	my ($key,$doc,$term) = @_;

	my $o = $doc->Occurrences( $term );
	$self->{_idf_n}++;
	if( $o and ($o>0) ){
		$self->{_idf_dt}++;
	}
}

sub IDF
{
	my $self = shift;
	my ($term) = @_;

	defined( $self->{IDF}->{$term} ) and return $self->{IDF}->{$term};
	$self->{_idf_n} = 0;
	$self->{_idf_dt} = 0;
	$self->EnumerateV( \&Text::DocumentCollection::IDF_Help, $term );
	if( $self->{_idf_dt} <= 0 ){
		warn( "term $term does not occur in any document" );
		return $self->{IDF}->{$term} = 0.0;
	}
	$self->{IDF}->{$term} =
		log( $self->{_idf_n} / $self->{_idf_dt} ) / log(2.0);

#	print "IDF($term) = $self->{IDF}->{$term}\n";
	return $self->{IDF}->{$term} ;
}

1;

__END__