File Coverage

blib/lib/FileMetadata/Miner/HTML.pm

Criterion	Covered	Total	%
statement	28	28	100.0
branch	10	10	100.0
condition	2	3	66.6
subroutine	8	8	100.0
pod	5	5	100.0
total	53	54	98.1

package FileMetadata::Miner::HTML;

require HTML::Parser;
@ISA = qw(HTML::Parser);
use HTML::Entities ();
use strict;
use utf8;
our $VERSION = '1.0';

sub new {

  my $class = shift;
  my $self = $class->SUPER::new(); # Inheriting

  # Set handlers

  $self->handler (text => 'text', "self, text");
  $self->handler (start => 'start', 'self, tagname, attr');
  $self->handler (end => 'end', 'self, tagname');

  return $self;

}

sub mine {

  my ($self, $path, $meta) = @_;
  $self->{data} = $meta;
  return 0 unless defined $self->parse_file ($path);
  return 1;

}

#
# This method handles text by appending it to the title key of
# $self->{'data'} when the tag is title. We are not interested in
# any other text
#

sub text {

  my ($self, $text) = @_;
  $self->{'data'}->{ref ($self)."::".'title'} .= $text
    if defined $self->{'tag-title'};

}

#
# This method handles start tags. It ignores all but meta and title tags
#

sub start {

  my($self, $tag, $attr) = @_;

  if ($tag eq 'meta' && defined $attr->{'name'}) {
    $self->{'data'}->{ref ($self)."::".$attr->{'name'}} = $attr->{'content'};
  } elsif ($tag eq 'title') {
    # Flag this
    $self->{'tag-title'} = '';
  }

}

sub end {

  my($self, $tag) = @_;

  # Stop processing
  $self->eof() if ($tag eq 'head');

  # Unflag
  delete $self->{'tag-title'};

}

1;

__END__