whatpm/Whatpm/ContentChecker.pm

package Whatpm::ContentChecker;
use strict;
our $VERSION=do{my @r=(q$Revision: 1.46 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};

require Whatpm::URIChecker;

## ISSUE: How XML and XML Namespaces conformance can (or cannot)
## be applied to an in-memory representation (i.e. DOM)?

my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>;

my $Namespace = {
  q<http://www.w3.org/2005/Atom> => {module => 'Whatpm::ContentChecker::Atom'},
  $HTML_NS => {module => 'Whatpm::ContentChecker::HTML'},
  $XML_NS => {loaded => 1},
  $XMLNS_NS => {loaded => 1},
};

our $AttrChecker = {
  $XML_NS => {
    space => sub {
      my ($self, $attr) = @_;
      my $value = $attr->value;
      if ($value eq 'default' or $value eq 'preserve') {
        #
      } else {
        ## NOTE: An XML "error"
        $self->{onerror}->(node => $attr, level => 'error',
                           type => 'invalid attribute value');
      }
    },
    lang => sub {
      my ($self, $attr) = @_;
      my $value = $attr->value;
      if ($value eq '') {
        #
      } else {
        require Whatpm::LangTag;
        Whatpm::LangTag->check_rfc3066_language_tag ($value, sub {
          my %opt = @_;
          my $type = 'LangTag:'.$opt{type};
          $type .= ':' . $opt{subtag} if defined $opt{subtag};
          $self->{onerror}->(node => $attr, type => $type,
                             value => $opt{value}, level => $opt{level});
        });
      }

      ## NOTE: "The values of the attribute are language identifiers
      ## as defined by [IETF RFC 3066], Tags for the Identification
      ## of Languages, or its successor; in addition, the empty string
      ## may be specified." ("may" in lower case)
      ## NOTE: Is an RFC 3066-valid (but RFC 4647-invalid) language tag
      ## allowed today?

      ## TODO: test data

      if ($attr->owner_document->manakai_is_html) { # MUST NOT
        $self->{onerror}->(node => $attr, type => 'in HTML:xml:lang');
## TODO: Test data...
      }
    },
    base => sub {
      my ($self, $attr) = @_;
      my $value = $attr->value;
      if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters?
        $self->{onerror}->(node => $attr,
                           type => 'invalid attribute value');
      }
      ## NOTE: Conformance to URI standard is not checked since there is
      ## no author requirement on conformance in the XML Base specification.
    },
    id => sub {
      my ($self, $attr) = @_;
      my $value = $attr->value;
      $value =~ s/[\x09\x0A\x0D\x20]+/ /g;
      $value =~ s/^\x20//;
      $value =~ s/\x20$//;
      ## TODO: NCName in XML 1.0 or 1.1
      ## TODO: declared type is ID?
      if ($self->{id}->{$value}) { ## NOTE: An xml:id error
        $self->{onerror}->(node => $attr, level => 'error', 
                           type => 'duplicate ID');
        push @{$self->{id}->{$value}}, $attr;
      } else {
        $self->{id}->{$value} = [$attr];
      }
    },
  },
  $XMLNS_NS => {
    '' => sub {
      my ($self, $attr) = @_;
      my $ln = $attr->manakai_local_name;
      my $value = $attr->value;
      if ($value eq $XML_NS and $ln ne 'xml') {
        $self->{onerror}
          ->(node => $attr, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:=xml');
      } elsif ($value eq $XMLNS_NS) {
        $self->{onerror}
          ->(node => $attr, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:=xmlns');
      }
      if ($ln eq 'xml' and $value ne $XML_NS) {
        $self->{onerror}
          ->(node => $attr, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:xmlns:xml=');
      } elsif ($ln eq 'xmlns') {
        $self->{onerror}
          ->(node => $attr, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:xmlns:xmlns=');
      }
      ## TODO: If XML 1.0 and empty
    },
    xmlns => sub {
      my ($self, $attr) = @_;
      ## TODO: In XML 1.0, URI reference [RFC 3986] or an empty string
      ## TODO: In XML 1.1, IRI reference [RFC 3987] or an empty string
      ## TODO: relative references are deprecated
      my $value = $attr->value;
      if ($value eq $XML_NS) {
        $self->{onerror}
          ->(node => $attr, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:=xml');
      } elsif ($value eq $XMLNS_NS) {
        $self->{onerror}
          ->(node => $attr, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:=xmlns');
      }
    },
  },
};

## ISSUE: Should we really allow these attributes?
$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space};
$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang};
$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base};
$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id};

## ANY
our $AnyChecker = sub {
  my ($self, $todo) = @_;
  my $el = $todo->{node};
  my $new_todos = [];
  my @nodes = (@{$el->child_nodes});
  while (@nodes) {
    my $node = shift @nodes;
    $self->_remove_minuses ($node) and next if ref $node eq 'HASH';

    my $nt = $node->node_type;
    if ($nt == 1) {
      my $node_ns = $node->namespace_uri;
      $node_ns = '' unless defined $node_ns;
      my $node_ln = $node->manakai_local_name;
      if ($self->{minuses}->{$node_ns}->{$node_ln}) {
        $self->{onerror}->(node => $node, type => 'element not allowed');
      }
      push @$new_todos, {type => 'element', node => $node};
    } elsif ($nt == 5) {
      unshift @nodes, @{$node->child_nodes};
    }
  }
  return ($new_todos);
}; # $AnyChecker

our $ElementDefault = {
  checker => sub {
    my ($self, $todo) = @_;
    $self->{onerror}->(node => $todo->{node}, level => 'unsupported',
                       type => 'element');
    return $AnyChecker->($self, $todo);
  },
  attrs_checker => sub {
    my ($self, $todo) = @_;
    for my $attr (@{$todo->{node}->attributes}) {
      my $attr_ns = $attr->namespace_uri;
      $attr_ns = '' unless defined $attr_ns;
      my $attr_ln = $attr->manakai_local_name;
      my $checker = $AttrChecker->{$attr_ns}->{$attr_ln}
        || $AttrChecker->{$attr_ns}->{''};
      if ($checker) {
        $checker->($self, $attr);
      } else {
        $self->{onerror}->(node => $attr, level => 'unsupported',
                           type => 'attribute');
      }
    }
  },
};

my $HTMLTransparentElements = {
  $HTML_NS => {qw/ins 1 font 1 noscript 1/},
  ## NOTE: |html:noscript| is transparent if scripting is disabled
  ## and not in |head|.
};

our $Element = {};

sub check_document ($$$) {
  my ($self, $doc, $onerror) = @_;
  $self = bless {}, $self unless ref $self;
  $self->{onerror} = $onerror;

  my $docel = $doc->document_element;
  unless (defined $docel) {
    ## ISSUE: Should we check content of Document node?
    $onerror->(node => $doc, type => 'no document element');
    ## ISSUE: Is this non-conforming (to what spec)?  Or just a warning?
    return {
            class => {},
            id => {}, table => [], term => {},
           };
  }

  ## ISSUE: Unexpanded entity references and HTML5 conformance
  
  my $docel_nsuri = $docel->namespace_uri;
  $docel_nsuri = '' unless defined $docel_nsuri;
  unless ($Namespace->{$docel_nsuri}->{loaded}) {
    if ($Namespace->{$docel_nsuri}->{module}) {
      eval qq{ require $Namespace->{$docel_nsuri}->{module} } or die $@;
    } else {
      $Namespace->{$docel_nsuri}->{loaded} = 1;
    }
  }
  my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} ||
    $Element->{$docel_nsuri}->{''} ||
    $ElementDefault;
  if ($docel_def->{is_root}) {
    #
  } else {
    $onerror->(node => $docel, type => 'element not allowed');
  }

  ## TODO: Check for other items other than document element
  ## (second (errorous) element, text nodes, PI nodes, doctype nodes)

  return $self->check_element ($docel, $onerror);
} # check_document

sub check_element ($$$) {
  my ($self, $el, $onerror) = @_;
  $self = bless {}, $self unless ref $self;
  $self->{onerror} = $onerror;

  $self->{minuses} = {};
  $self->{id} = {};
  $self->{term} = {};
  $self->{usemap} = [];
  $self->{contextmenu} = [];
  $self->{map} = {};
  $self->{menu} = {};
  $self->{has_link_type} = {};
  #$self->{has_uri_attr};
  #$self->{has_hyperlink_element};
  $self->{return} = {
    class => {},
    id => $self->{id}, table => [], term => $self->{term},
  };

  my @todo = ({type => 'element', node => $el});
  while (@todo) {
    my $todo = shift @todo;
    if ($todo->{type} eq 'element') {
      my $prefix = $todo->{node}->prefix;
      if (defined $prefix and $prefix eq 'xmlns') {
        $self->{onerror}
          ->(node => $todo->{node}, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
      }
      my $nsuri = $todo->{node}->namespace_uri;
      $nsuri = '' unless defined $nsuri;
      unless ($Namespace->{$nsuri}->{loaded}) {
        if ($Namespace->{$nsuri}->{module}) {
          eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
        } else {
          $Namespace->{$nsuri}->{loaded} = 1;
        }
      }
      my $ln = $todo->{node}->manakai_local_name;
      my $eldef = $Element->{$nsuri}->{$ln} ||
        $Element->{$nsuri}->{''} ||
          $ElementDefault;
      $eldef->{attrs_checker}->($self, $todo);
      my ($new_todos) = $eldef->{checker}->($self, $todo);
      unshift @todo, @$new_todos;
    } elsif ($todo->{type} eq 'element-attributes') {
      my $prefix = $todo->{node}->prefix;
      if (defined $prefix and $prefix eq 'xmlns') {
        $self->{onerror}
          ->(node => $todo->{node}, level => 'NC',
             type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
      }
      my $nsuri = $todo->{node}->namespace_uri;
      $nsuri = '' unless defined $nsuri;
      unless ($Namespace->{$nsuri}->{loaded}) {
        if ($Namespace->{$nsuri}->{module}) {
          eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
        } else {
          $Namespace->{$nsuri}->{loaded} = 1;
        }
      }
      my $ln = $todo->{node}->manakai_local_name;
      my $eldef = $Element->{$nsuri}->{$ln} ||
        $Element->{$nsuri}->{''} ||
          $ElementDefault;
      $eldef->{attrs_checker}->($self, $todo);
    } elsif ($todo->{type} eq 'plus') {
      $self->_remove_minuses ($todo);
    } elsif ($todo->{type} eq 'code') {
      $todo->{code}->();
    } else {
      die "$0: Internal error: Unsupported checking action type |$todo->{type}|";
    }
  }

  for (@{$self->{usemap}}) {
    unless ($self->{map}->{$_->[0]}) {
      $self->{onerror}->(node => $_->[1], type => 'no referenced map');
    }
  }

  for (@{$self->{contextmenu}}) {
    unless ($self->{menu}->{$_->[0]}) {
      $self->{onerror}->(node => $_->[1], type => 'no referenced menu');
    }
  }

  delete $self->{minuses};
  delete $self->{onerror};
  delete $self->{id};
  delete $self->{usemap};
  delete $self->{map};
  return $self->{return};
} # check_element

sub _add_minuses ($@) {
  my $self = shift;
  my $r = {};
  for my $list (@_) {
    for my $ns (keys %$list) {
      for my $ln (keys %{$list->{$ns}}) {
        unless ($self->{minuses}->{$ns}->{$ln}) {
          $self->{minuses}->{$ns}->{$ln} = 1;
          $r->{$ns}->{$ln} = 1;
        }
      }
    }
  }
  return {type => 'plus', list => $r};
} # _add_minuses

sub _remove_minuses ($$) {
  my ($self, $todo) = @_;
  for my $ns (keys %{$todo->{list}}) {
    for my $ln (keys %{$todo->{list}->{$ns}}) {
      delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
    }
  }
  1;
} # _remove_minuses

sub _check_get_children ($$$) {
  my ($self, $node, $parent_todo) = @_;
  my $new_todos = [];
  my $sib = [];
  TP: {
    my $node_ns = $node->namespace_uri;
    $node_ns = '' unless defined $node_ns;
    my $node_ln = $node->manakai_local_name;
    if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
      if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') {
        if ($parent_todo->{flag}->{in_head}) {
          #
        } else {
          my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
          push @$sib, $end;
          
          unshift @$sib, @{$node->child_nodes};
          push @$new_todos, {type => 'element-attributes', node => $node};
          last TP;
        }
      } else {
        unshift @$sib, @{$node->child_nodes};
        push @$new_todos, {type => 'element-attributes', node => $node};
        last TP;
      }
    }
    if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) {
      if ($node->has_attribute_ns (undef, 'src')) {
        unshift @$sib, @{$node->child_nodes};
        push @$new_todos, {type => 'element-attributes', node => $node};
        last TP;
      } else {
        my @cn = @{$node->child_nodes};
        CN: while (@cn) {
          my $cn = shift @cn;
          my $cnt = $cn->node_type;
          if ($cnt == 1) {
            my $cn_nsuri = $cn->namespace_uri;
            $cn_nsuri = '' unless defined $cn_nsuri;
            if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') {
              #
            } else {
              last CN;
            }
          } elsif ($cnt == 3 or $cnt == 4) {
            if ($cn->data =~ /[^\x09-\x0D\x20]/) {
              last CN;
            }
          }
        } # CN
        unshift @$sib, @cn;
      }
    }
    push @$new_todos, {type => 'element', node => $node};
  } # TP
  
  for my $new_todo (@$new_todos) {
    $new_todo->{flag} = {%{$parent_todo->{flag} or {}}};
  }
  
  return ($sib, $new_todos);
} # _check_get_children

=head1 LICENSE

Copyright 2007 Wakaba <w@suika.fam.cx>

This library is free software; you can redistribute it
and/or modify it under the same terms as Perl itself.

=cut

1;
# $Date: 2007/08/17 11:53:52 $
1	wakaba	1.1	package Whatpm::ContentChecker;
2			use strict;
3	wakaba	1.47	our $VERSION=do{my @r=(q$Revision: 1.46 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
4	wakaba	1.1
5	wakaba	1.18	require Whatpm::URIChecker;
6
7	wakaba	1.13	## ISSUE: How XML and XML Namespaces conformance can (or cannot)
8			## be applied to an in-memory representation (i.e. DOM)?
9
10	wakaba	1.42	my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
11	wakaba	1.9	my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
12			my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>;
13
14	wakaba	1.42	my $Namespace = {
15	wakaba	1.43	q<http://www.w3.org/2005/Atom> => {module => 'Whatpm::ContentChecker::Atom'},
16	wakaba	1.42	$HTML_NS => {module => 'Whatpm::ContentChecker::HTML'},
17			$XML_NS => {loaded => 1},
18			$XMLNS_NS => {loaded => 1},
19			};
20
21			our $AttrChecker = {
22	wakaba	1.9	$XML_NS => {
23	wakaba	1.13	space => sub {
24			my ($self, $attr) = @_;
25			my $value = $attr->value;
26			if ($value eq 'default' or $value eq 'preserve') {
27			#
28			} else {
29			## NOTE: An XML "error"
30	wakaba	1.33	$self->{onerror}->(node => $attr, level => 'error',
31			type => 'invalid attribute value');
32	wakaba	1.13	}
33			},
34			lang => sub {
35	wakaba	1.35	my ($self, $attr) = @_;
36	wakaba	1.47	my $value = $attr->value;
37			if ($value eq '') {
38			#
39			} else {
40			require Whatpm::LangTag;
41			Whatpm::LangTag->check_rfc3066_language_tag ($value, sub {
42			my %opt = @_;
43			my $type = 'LangTag:'.$opt{type};
44			$type .= ':' . $opt{subtag} if defined $opt{subtag};
45			$self->{onerror}->(node => $attr, type => $type,
46			value => $opt{value}, level => $opt{level});
47			});
48			}
49
50	wakaba	1.13	## NOTE: "The values of the attribute are language identifiers
51			## as defined by [IETF RFC 3066], Tags for the Identification
52			## of Languages, or its successor; in addition, the empty string
53			## may be specified." ("may" in lower case)
54	wakaba	1.47	## NOTE: Is an RFC 3066-valid (but RFC 4647-invalid) language tag
55			## allowed today?
56
57			## TODO: test data
58
59	wakaba	1.35	if ($attr->owner_document->manakai_is_html) { # MUST NOT
60	wakaba	1.36	$self->{onerror}->(node => $attr, type => 'in HTML:xml:lang');
61	wakaba	1.35	## TODO: Test data...
62			}
63	wakaba	1.13	},
64			base => sub {
65			my ($self, $attr) = @_;
66			my $value = $attr->value;
67			if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters?
68			$self->{onerror}->(node => $attr,
69	wakaba	1.33	type => 'invalid attribute value');
70	wakaba	1.13	}
71	wakaba	1.18	## NOTE: Conformance to URI standard is not checked since there is
72			## no author requirement on conformance in the XML Base specification.
73	wakaba	1.13	},
74			id => sub {
75			my ($self, $attr) = @_;
76			my $value = $attr->value;
77			$value =~ s/[\x09\x0A\x0D\x20]+/ /g;
78			$value =~ s/^\x20//;
79			$value =~ s/\x20$//;
80			## TODO: NCName in XML 1.0 or 1.1
81			## TODO: declared type is ID?
82	wakaba	1.33	if ($self->{id}->{$value}) { ## NOTE: An xml:id error
83			$self->{onerror}->(node => $attr, level => 'error',
84			type => 'duplicate ID');
85	wakaba	1.37	push @{$self->{id}->{$value}}, $attr;
86	wakaba	1.13	} else {
87	wakaba	1.37	$self->{id}->{$value} = [$attr];
88	wakaba	1.13	}
89			},
90	wakaba	1.9	},
91			$XMLNS_NS => {
92	wakaba	1.13	'' => sub {
93			my ($self, $attr) = @_;
94			my $ln = $attr->manakai_local_name;
95			my $value = $attr->value;
96			if ($value eq $XML_NS and $ln ne 'xml') {
97			$self->{onerror}
98	wakaba	1.33	->(node => $attr, level => 'NC',
99			type => 'Reserved Prefixes and Namespace Names:=xml');
100	wakaba	1.13	} elsif ($value eq $XMLNS_NS) {
101			$self->{onerror}
102	wakaba	1.33	->(node => $attr, level => 'NC',
103			type => 'Reserved Prefixes and Namespace Names:=xmlns');
104	wakaba	1.13	}
105			if ($ln eq 'xml' and $value ne $XML_NS) {
106			$self->{onerror}
107	wakaba	1.33	->(node => $attr, level => 'NC',
108			type => 'Reserved Prefixes and Namespace Names:xmlns:xml=');
109	wakaba	1.13	} elsif ($ln eq 'xmlns') {
110			$self->{onerror}
111	wakaba	1.33	->(node => $attr, level => 'NC',
112			type => 'Reserved Prefixes and Namespace Names:xmlns:xmlns=');
113	wakaba	1.13	}
114			## TODO: If XML 1.0 and empty
115			},
116			xmlns => sub {
117			my ($self, $attr) = @_;
118			## TODO: In XML 1.0, URI reference [RFC 3986] or an empty string
119			## TODO: In XML 1.1, IRI reference [RFC 3987] or an empty string
120	wakaba	1.18	## TODO: relative references are deprecated
121	wakaba	1.13	my $value = $attr->value;
122			if ($value eq $XML_NS) {
123			$self->{onerror}
124	wakaba	1.33	->(node => $attr, level => 'NC',
125			type => 'Reserved Prefixes and Namespace Names:=xml');
126	wakaba	1.13	} elsif ($value eq $XMLNS_NS) {
127			$self->{onerror}
128	wakaba	1.33	->(node => $attr, level => 'NC',
129			type => 'Reserved Prefixes and Namespace Names:=xmlns');
130	wakaba	1.13	}
131			},
132	wakaba	1.9	},
133			};
134
135	wakaba	1.14	## ISSUE: Should we really allow these attributes?
136	wakaba	1.13	$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space};
137			$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang};
138			$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base};
139			$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id};
140
141	wakaba	1.3	## ANY
142	wakaba	1.42	our $AnyChecker = sub {
143	wakaba	1.4	my ($self, $todo) = @_;
144			my $el = $todo->{node};
145			my $new_todos = [];
146	wakaba	1.3	my @nodes = (@{$el->child_nodes});
147			while (@nodes) {
148			my $node = shift @nodes;
149			$self->_remove_minuses ($node) and next if ref $node eq 'HASH';
150
151			my $nt = $node->node_type;
152			if ($nt == 1) {
153			my $node_ns = $node->namespace_uri;
154			$node_ns = '' unless defined $node_ns;
155			my $node_ln = $node->manakai_local_name;
156			if ($self->{minuses}->{$node_ns}->{$node_ln}) {
157			$self->{onerror}->(node => $node, type => 'element not allowed');
158			}
159	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
160	wakaba	1.3	} elsif ($nt == 5) {
161			unshift @nodes, @{$node->child_nodes};
162			}
163			}
164	wakaba	1.4	return ($new_todos);
165	wakaba	1.3	}; # $AnyChecker
166
167	wakaba	1.42	our $ElementDefault = {
168	wakaba	1.1	checker => sub {
169	wakaba	1.4	my ($self, $todo) = @_;
170	wakaba	1.33	$self->{onerror}->(node => $todo->{node}, level => 'unsupported',
171			type => 'element');
172	wakaba	1.4	return $AnyChecker->($self, $todo);
173	wakaba	1.1	},
174	wakaba	1.9	attrs_checker => sub {
175			my ($self, $todo) = @_;
176			for my $attr (@{$todo->{node}->attributes}) {
177			my $attr_ns = $attr->namespace_uri;
178			$attr_ns = '' unless defined $attr_ns;
179			my $attr_ln = $attr->manakai_local_name;
180			my $checker = $AttrChecker->{$attr_ns}->{$attr_ln}
181			\|\| $AttrChecker->{$attr_ns}->{''};
182			if ($checker) {
183			$checker->($self, $attr);
184	wakaba	1.17	} else {
185	wakaba	1.33	$self->{onerror}->(node => $attr, level => 'unsupported',
186			type => 'attribute');
187	wakaba	1.9	}
188			}
189			},
190	wakaba	1.1	};
191
192	wakaba	1.7	my $HTMLTransparentElements = {
193			$HTML_NS => {qw/ins 1 font 1 noscript 1/},
194	wakaba	1.29	## NOTE: \|html:noscript\| is transparent if scripting is disabled
195			## and not in \|head\|.
196	wakaba	1.7	};
197
198	wakaba	1.42	our $Element = {};
199	wakaba	1.7
200	wakaba	1.42	sub check_document ($$$) {
201			my ($self, $doc, $onerror) = @_;
202			$self = bless {}, $self unless ref $self;
203			$self->{onerror} = $onerror;
204	wakaba	1.1
205	wakaba	1.42	my $docel = $doc->document_element;
206			unless (defined $docel) {
207			## ISSUE: Should we check content of Document node?
208			$onerror->(node => $doc, type => 'no document element');
209			## ISSUE: Is this non-conforming (to what spec)? Or just a warning?
210			return {
211			class => {},
212			id => {}, table => [], term => {},
213			};
214	wakaba	1.1	}
215
216	wakaba	1.42	## ISSUE: Unexpanded entity references and HTML5 conformance
217	wakaba	1.1
218	wakaba	1.42	my $docel_nsuri = $docel->namespace_uri;
219			$docel_nsuri = '' unless defined $docel_nsuri;
220	wakaba	1.43	unless ($Namespace->{$docel_nsuri}->{loaded}) {
221			if ($Namespace->{$docel_nsuri}->{module}) {
222			eval qq{ require $Namespace->{$docel_nsuri}->{module} } or die $@;
223			} else {
224			$Namespace->{$docel_nsuri}->{loaded} = 1;
225			}
226			}
227	wakaba	1.42	my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} \|\|
228			$Element->{$docel_nsuri}->{''} \|\|
229			$ElementDefault;
230			if ($docel_def->{is_root}) {
231			#
232			} else {
233			$onerror->(node => $docel, type => 'element not allowed');
234	wakaba	1.1	}
235
236	wakaba	1.42	## TODO: Check for other items other than document element
237			## (second (errorous) element, text nodes, PI nodes, doctype nodes)
238	wakaba	1.2
239	wakaba	1.42	return $self->check_element ($docel, $onerror);
240			} # check_document
241	wakaba	1.1
242	wakaba	1.42	sub check_element ($$$) {
243			my ($self, $el, $onerror) = @_;
244			$self = bless {}, $self unless ref $self;
245			$self->{onerror} = $onerror;
246	wakaba	1.2
247	wakaba	1.42	$self->{minuses} = {};
248			$self->{id} = {};
249			$self->{term} = {};
250			$self->{usemap} = [];
251			$self->{contextmenu} = [];
252			$self->{map} = {};
253			$self->{menu} = {};
254			$self->{has_link_type} = {};
255	wakaba	1.46	#$self->{has_uri_attr};
256			#$self->{has_hyperlink_element};
257	wakaba	1.42	$self->{return} = {
258			class => {},
259			id => $self->{id}, table => [], term => $self->{term},
260			};
261	wakaba	1.4
262	wakaba	1.42	my @todo = ({type => 'element', node => $el});
263			while (@todo) {
264			my $todo = shift @todo;
265			if ($todo->{type} eq 'element') {
266			my $prefix = $todo->{node}->prefix;
267			if (defined $prefix and $prefix eq 'xmlns') {
268			$self->{onerror}
269			->(node => $todo->{node}, level => 'NC',
270			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
271	wakaba	1.7	}
272	wakaba	1.42	my $nsuri = $todo->{node}->namespace_uri;
273			$nsuri = '' unless defined $nsuri;
274			unless ($Namespace->{$nsuri}->{loaded}) {
275			if ($Namespace->{$nsuri}->{module}) {
276			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
277			} else {
278			$Namespace->{$nsuri}->{loaded} = 1;
279	wakaba	1.1	}
280			}
281	wakaba	1.42	my $ln = $todo->{node}->manakai_local_name;
282			my $eldef = $Element->{$nsuri}->{$ln} \|\|
283			$Element->{$nsuri}->{''} \|\|
284			$ElementDefault;
285			$eldef->{attrs_checker}->($self, $todo);
286			my ($new_todos) = $eldef->{checker}->($self, $todo);
287			unshift @todo, @$new_todos;
288			} elsif ($todo->{type} eq 'element-attributes') {
289			my $prefix = $todo->{node}->prefix;
290			if (defined $prefix and $prefix eq 'xmlns') {
291			$self->{onerror}
292			->(node => $todo->{node}, level => 'NC',
293			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
294			}
295			my $nsuri = $todo->{node}->namespace_uri;
296			$nsuri = '' unless defined $nsuri;
297			unless ($Namespace->{$nsuri}->{loaded}) {
298			if ($Namespace->{$nsuri}->{module}) {
299			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
300	wakaba	1.1	} else {
301	wakaba	1.42	$Namespace->{$nsuri}->{loaded} = 1;
302	wakaba	1.1	}
303			}
304	wakaba	1.9	my $ln = $todo->{node}->manakai_local_name;
305			my $eldef = $Element->{$nsuri}->{$ln} \|\|
306			$Element->{$nsuri}->{''} \|\|
307			$ElementDefault;
308			$eldef->{attrs_checker}->($self, $todo);
309	wakaba	1.4	} elsif ($todo->{type} eq 'plus') {
310			$self->_remove_minuses ($todo);
311	wakaba	1.30	} elsif ($todo->{type} eq 'code') {
312			$todo->{code}->();
313			} else {
314			die "$0: Internal error: Unsupported checking action type \|$todo->{type}\|";
315	wakaba	1.4	}
316	wakaba	1.1	}
317	wakaba	1.17
318			for (@{$self->{usemap}}) {
319			unless ($self->{map}->{$_->[0]}) {
320			$self->{onerror}->(node => $_->[1], type => 'no referenced map');
321			}
322			}
323
324	wakaba	1.32	for (@{$self->{contextmenu}}) {
325			unless ($self->{menu}->{$_->[0]}) {
326			$self->{onerror}->(node => $_->[1], type => 'no referenced menu');
327			}
328			}
329
330	wakaba	1.17	delete $self->{minuses};
331			delete $self->{onerror};
332			delete $self->{id};
333			delete $self->{usemap};
334			delete $self->{map};
335	wakaba	1.33	return $self->{return};
336	wakaba	1.1	} # check_element
337
338	wakaba	1.2	sub _add_minuses ($@) {
339			my $self = shift;
340			my $r = {};
341			for my $list (@_) {
342			for my $ns (keys %$list) {
343			for my $ln (keys %{$list->{$ns}}) {
344			unless ($self->{minuses}->{$ns}->{$ln}) {
345			$self->{minuses}->{$ns}->{$ln} = 1;
346			$r->{$ns}->{$ln} = 1;
347			}
348			}
349			}
350			}
351	wakaba	1.4	return {type => 'plus', list => $r};
352	wakaba	1.2	} # _add_minuses
353
354			sub _remove_minuses ($$) {
355	wakaba	1.4	my ($self, $todo) = @_;
356			for my $ns (keys %{$todo->{list}}) {
357			for my $ln (keys %{$todo->{list}->{$ns}}) {
358			delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
359	wakaba	1.2	}
360			}
361			1;
362			} # _remove_minuses
363
364	wakaba	1.30	sub _check_get_children ($$$) {
365			my ($self, $node, $parent_todo) = @_;
366	wakaba	1.4	my $new_todos = [];
367	wakaba	1.2	my $sib = [];
368			TP: {
369			my $node_ns = $node->namespace_uri;
370			$node_ns = '' unless defined $node_ns;
371			my $node_ln = $node->manakai_local_name;
372	wakaba	1.45	if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
373			if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') {
374			if ($parent_todo->{flag}->{in_head}) {
375			#
376			} else {
377			my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
378			push @$sib, $end;
379
380			unshift @$sib, @{$node->child_nodes};
381			push @$new_todos, {type => 'element-attributes', node => $node};
382			last TP;
383			}
384			} else {
385			unshift @$sib, @{$node->child_nodes};
386			push @$new_todos, {type => 'element-attributes', node => $node};
387			last TP;
388	wakaba	1.2	}
389			}
390	wakaba	1.8	if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) {
391	wakaba	1.2	if ($node->has_attribute_ns (undef, 'src')) {
392			unshift @$sib, @{$node->child_nodes};
393	wakaba	1.9	push @$new_todos, {type => 'element-attributes', node => $node};
394	wakaba	1.2	last TP;
395			} else {
396			my @cn = @{$node->child_nodes};
397			CN: while (@cn) {
398			my $cn = shift @cn;
399			my $cnt = $cn->node_type;
400			if ($cnt == 1) {
401	wakaba	1.8	my $cn_nsuri = $cn->namespace_uri;
402			$cn_nsuri = '' unless defined $cn_nsuri;
403			if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') {
404	wakaba	1.2	#
405			} else {
406			last CN;
407			}
408			} elsif ($cnt == 3 or $cnt == 4) {
409			if ($cn->data =~ /[^\x09-\x0D\x20]/) {
410			last CN;
411			}
412			}
413			} # CN
414			unshift @$sib, @cn;
415			}
416			}
417	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
418	wakaba	1.2	} # TP
419	wakaba	1.30
420			for my $new_todo (@$new_todos) {
421			$new_todo->{flag} = {%{$parent_todo->{flag} or {}}};
422			}
423
424	wakaba	1.4	return ($sib, $new_todos);
425	wakaba	1.2	} # _check_get_children
426
427	wakaba	1.44	=head1 LICENSE
428
429			Copyright 2007 Wakaba <w@suika.fam.cx>
430
431			This library is free software; you can redistribute it
432			and/or modify it under the same terms as Perl itself.
433
434			=cut
435
436	wakaba	1.1	1;
437	wakaba	1.47	# $Date: 2007/08/17 11:53:52 $