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 ('encode');
  $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));
    }
    
    $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->{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	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	require Message::DOM::DOMImplementation;
14	require Encode;
15	require Whatpm::HTML;
16
17	$self->result->layer_uncertain ('encode');
18	$self->result->layer_uncertain ('charset');
19
20	my $out = $self->output;
21	$out->start_section (role => 'parse-errors');
22	$out->start_error_list (role => 'parse-errors');
23	$self->result->layer_applicable ('syntax');
24
25	my $input = $self->input;
26	my $result = $self->result;
27
28	my $onerror = sub {
29	$result->add_error (@_, layer => 'syntax');
30	};
31
32	my $dom = Message::DOM::DOMImplementation->new;
33	my $doc = $dom->create_document;
34	my $el;
35	my $inner_html_element = $input->{inner_html_element};
36	if (defined $inner_html_element and length $inner_html_element) {
37	$input->{charset} \|\|= 'windows-1252'; ## TODO: for now.
38	my $t = \($input->{s});
39	unless ($input->{is_char_string}) {
40	$t = \(Encode::decode ($input->{charset}, $$t));
41	}
42
43	$el = $doc->create_element_ns
44	('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
45	Whatpm::HTML->set_inner_html ($el, $$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	}
55
56	$self->{structure} = $doc;
57	}
58	$doc->manakai_charset ($input->{official_charset})
59	if defined $input->{official_charset};
60
61	$doc->document_uri ($input->{uri});
62	$doc->manakai_entity_base_uri ($input->{base_uri});
63
64	$out->end_error_list (role => 'parse-errors');
65	$out->end_section;
66	} # generate_syntax_error_section
67
68	sub source_charset ($) {
69	my $self = shift;
70	return $self->input->{charset} \|\| ($self->{structure}->owner_document \|\| $self->{structure})->input_encoding;
71	## TODO: Can we always use input_encoding?
72	} # source_charset
73
74	1;