File Coverage

blib/lib/HTML/TableContentParser.pm

Criterion	Covered	Total	%
statement	97	101	96.0
branch	42	48	87.5
condition	6	7	85.7
subroutine	9	10	90.0
pod	6	6	100.0
total	160	172	93.0

package HTML::TableContentParser;

use strict;
use warnings;

use parent qw{ HTML::Parser };

our $VERSION = '0.305';

our $CLASSIC = 0;
our $DEBUG = 0;

my @stacked = qw{ current_table current_row current_element };

sub new
{
	my ( $class, %arg ) = @_;
	my $classic = delete $arg{classic};
	my $self = $class->SUPER::new( %arg );
	$self->{ATTR}{classic} =  defined $classic ? $classic : $CLASSIC;
	return $self;
}

sub classic
{
	my ( $self ) = @_;
	return $self->{ATTR}{classic};
}

sub start
{
#	my ($self, $tag, $attr, $attrseq, $origtext) = @_;
	my ($self, $tag, $attr, undef, $origtext) = @_;

	$tag = lc($tag);

# Store the incoming details in the current 'object'.
	if ($tag eq 'table') {
		my $table = $attr;
		push @{ $self->{STORE}{stack} }, {
			map { $_ => $self->{STORE}{$_} } @stacked };
		push @{$self->{STORE}->{tables}}, $table;
		$self->{STORE}->{current_table} = $table;
		$self->{STORE}->{current_row} = undef;
		$self->{STORE}->{current_element} = undef;

	} elsif ($tag eq 'th') {
		my $th = $attr;
		push @{$self->{STORE}->{current_table}->{headers}}, $th;
		unless ( $self->{ATTR}{classic} ) {
			push @{$self->{STORE}->{current_row}->{cells}}, undef;
			push @{$self->{STORE}->{current_row}->{headers}}, $th;
		}
		$self->{STORE}->{current_element} = $th;

	} elsif ($tag eq 'tr') {
		my $tr = $attr;
		push @{$self->{STORE}->{current_table}->{rows}}, $tr;
		$self->{STORE}->{current_row} = $tr;
		$self->{STORE}->{current_element} = $tr;

	} elsif ($tag eq 'td') {
		my $td = $attr;
		push @{$self->{STORE}->{current_row}->{cells}}, $td;
		unless ( $self->{ATTR}{classic} ) {
			push @{$self->{STORE}->{current_row}->{headers}}, undef;
		}
		$self->{STORE}->{current_element} = $td;

	} elsif ($tag eq 'caption') {
		my $cap = $attr;
		$self->{STORE}->{current_table}->{caption} = $cap;
		$self->{STORE}->{current_element} = $cap;

	} else {
## Found a non-table related tag. Push it into the currently-defined td
## or th (if one exists).
		my $elem = $self->{STORE}->{current_element};
		if ($elem) {
			$self->_debug('TEXT(tag) = ', $origtext) if $DEBUG;
			$elem->{data} .= $origtext;
		}

	}

	$self->_debug($origtext) if $DEBUG;

	return;
}



sub text
{
	my ($self, $text) = @_;
	my $elem = $self->{STORE}->{current_element};
	if (!$elem) {
		return;
	}

	$self->_debug('TEXT = ', $text) if $DEBUG;
	$elem->{data} .= $text;

	return;
}



sub end
{
	my ($self, $tag, $origtext) = @_;
	$tag = lc($tag);

# Turn off the current object
	if ($tag eq 'table') {
		my $prev = pop @{ $self->{STORE}{stack} } || [];
		$self->{STORE}{$_} = $prev->{$_} for @stacked;

	} elsif ($tag eq 'th') {
		$self->{STORE}->{current_element} = undef;
	} elsif ($tag eq 'tr') {
		for my $key ( 'cells', $self->{ATTR}{classic} ? () : 'headers' ) {
			my $data = $self->{STORE}{current_row}{$key} || [];
			pop @{ $data } while @{ $data } && !  $data->[-1];
			delete $self->{STORE}{current_row}{$key}
				unless @{ $data };
		}
		$self->{STORE}->{current_row} = undef;
		$self->{STORE}->{current_element} = undef;

	} elsif ($tag eq 'td') {
		$self->{STORE}->{current_element} = undef;

	} elsif ($tag eq 'caption') {
		$self->{STORE}->{current_element} = undef;

	} else {
## Found a non-table related close tag. Push it into the currently-defined
## td or th (if one exists).
		my $elem = $self->{STORE}->{current_element};
		if ($elem) {
			$self->_debug('TEXT(tag) = ', $origtext) if $DEBUG;
			$elem->{data} .= $origtext;
		}

	}

	$self->_debug($origtext) if $DEBUG;

	return;
}


sub parse
{
	my ($self, $data) = @_;

	unless ( defined $data ) {	# RT 7262
	    require Carp;
	    Carp::croak( 'Argument must be defined' );
	}

	$self->{STORE} = {
	    stack	=> [],
	};

	$self->SUPER::parse($data);

	my $tables = $self->{STORE}{tables};
	delete $self->{STORE};

	return $tables;
}




sub _debug
{
	my ( $self, @args ) = @_;
	my $class = ref($self);
	warn "$class: ", join( '', @args ), "\n";
	return;
}


1;


__END__