WebHACC/Language/HTML.pm

package WebHACC::Language::HTML;
use strict;
require WebHACC::Language::DOM;
push our @ISA, 'WebHACC::Language::DOM';

sub new ($) {
  return bless {}, shift;
} # new

sub generate_syntax_error_section ($) {
  my $self = shift;
  
  require Message::DOM::DOMImplementation;
  require Encode;
  require Whatpm::HTML;
  
  $self->result->layer_uncertain ('charset');

  my $out = $self->output;
  $out->start_section (role => 'parse-errors');
  $out->start_error_list (role => 'parse-errors');
  $self->result->layer_applicable ('syntax');

  my $input = $self->input;
  my $result = $self->result;

  my $onerror = sub {
    $result->add_error (@_, layer => 'syntax');
  };

  my $dom = Message::DOM::DOMImplementation->new;
  my $doc = $dom->create_document;
  my $el;
  my $inner_html_element = $input->{inner_html_element};
  if (defined $inner_html_element and length $inner_html_element) {
    $input->{charset} ||= 'windows-1252'; ## TODO: for now.
    my $t = \($input->{s});
    unless ($input->{is_char_string}) {
      $t = \(Encode::decode ($input->{charset}, $$t));
      $self->result->layer_uncertain ('encode');
    }
    
    $el = $doc->create_element_ns
        ('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
    Whatpm::HTML->set_inner_html ($el, $$t, $onerror);

    $self->{structure} = $el;
  } else {
    if ($input->{is_char_string}) {
      Whatpm::HTML->parse_char_string ($input->{s} => $doc, $onerror);
    } else {
      Whatpm::HTML->parse_byte_string
          ($input->{charset}, $input->{s} => $doc, $onerror);
      $self->result->layer_uncertain ('encode');
    }

    $self->{structure} = $doc;
  }
  $doc->manakai_charset ($input->{official_charset})
      if defined $input->{official_charset};

  $doc->document_uri ($input->{uri});
  $doc->manakai_entity_base_uri ($input->{base_uri});

  $out->end_error_list (role => 'parse-errors');
  $out->end_section;
} # generate_syntax_error_section

sub source_charset ($) {
  my $self = shift;
  return $self->input->{charset} || ($self->{structure}->owner_document || $self->{structure})->input_encoding;
  ## TODO: Can we always use input_encoding?
} # source_charset

1;
1	wakaba	1.1	package WebHACC::Language::HTML;
2			use strict;
3			require WebHACC::Language::DOM;
4			push our @ISA, 'WebHACC::Language::DOM';
5
6			sub new ($) {
7			return bless {}, shift;
8			} # new
9
10			sub generate_syntax_error_section ($) {
11			my $self = shift;
12
13	wakaba	1.4	require Message::DOM::DOMImplementation;
14	wakaba	1.1	require Encode;
15			require Whatpm::HTML;
16
17	wakaba	1.5	$self->result->layer_uncertain ('charset');
18
19	wakaba	1.1	my $out = $self->output;
20	wakaba	1.3	$out->start_section (role => 'parse-errors');
21			$out->start_error_list (role => 'parse-errors');
22	wakaba	1.5	$self->result->layer_applicable ('syntax');
23	wakaba	1.1
24			my $input = $self->input;
25			my $result = $self->result;
26
27			my $onerror = sub {
28	wakaba	1.2	$result->add_error (@_, layer => 'syntax');
29	wakaba	1.1	};
30
31			my $dom = Message::DOM::DOMImplementation->new;
32			my $doc = $dom->create_document;
33			my $el;
34			my $inner_html_element = $input->{inner_html_element};
35			if (defined $inner_html_element and length $inner_html_element) {
36			$input->{charset} \|\|= 'windows-1252'; ## TODO: for now.
37			my $t = \($input->{s});
38			unless ($input->{is_char_string}) {
39			$t = \(Encode::decode ($input->{charset}, $$t));
40	wakaba	1.6	$self->result->layer_uncertain ('encode');
41	wakaba	1.1	}
42
43			$el = $doc->create_element_ns
44			('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
45			Whatpm::HTML->set_inner_html ($el, $$t, $onerror);
46
47			$self->{structure} = $el;
48			} else {
49			if ($input->{is_char_string}) {
50			Whatpm::HTML->parse_char_string ($input->{s} => $doc, $onerror);
51			} else {
52			Whatpm::HTML->parse_byte_string
53			($input->{charset}, $input->{s} => $doc, $onerror);
54	wakaba	1.6	$self->result->layer_uncertain ('encode');
55	wakaba	1.1	}
56
57			$self->{structure} = $doc;
58			}
59			$doc->manakai_charset ($input->{official_charset})
60			if defined $input->{official_charset};
61
62			$doc->document_uri ($input->{uri});
63			$doc->manakai_entity_base_uri ($input->{base_uri});
64
65	wakaba	1.3	$out->end_error_list (role => 'parse-errors');
66	wakaba	1.1	$out->end_section;
67			} # generate_syntax_error_section
68
69			sub source_charset ($) {
70			my $self = shift;
71			return $self->input->{charset} \|\| ($self->{structure}->owner_document \|\| $self->{structure})->input_encoding;
72			## TODO: Can we always use input_encoding?
73			} # source_charset
74
75			1;