whatpm/Whatpm/ContentChecker.pm

package Whatpm::ContentChecker;
use strict;
our $VERSION=do{my @r=(q$Revision: 1.53 $=~/\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) = @_;
      ## 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)
      $self->{onerror}->(node => $attr, level => 'unsupported',
                         type => 'language tag');
      if ($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->{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 = $node->manakai_local_name;
    if ($node_ns eq $HTML_NS) {
      if ($node_ln eq 'noscript') {
        my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
        push @$sib, $end;
      }
    }
    ## TODO: |noscript| is not a transparent element in |head|.
    if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
      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: 2007/08/05 04:50:57 $
1	wakaba	1.1	package Whatpm::ContentChecker;
2			use strict;
3	wakaba	1.44	our $VERSION=do{my @r=(q$Revision: 1.53 $=~/\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	1.13	## NOTE: "The values of the attribute are language identifiers
37			## as defined by [IETF RFC 3066], Tags for the Identification
38			## of Languages, or its successor; in addition, the empty string
39			## may be specified." ("may" in lower case)
40	wakaba	1.36	$self->{onerror}->(node => $attr, level => 'unsupported',
41			type => 'language tag');
42	wakaba	1.35	if ($attr->owner_document->manakai_is_html) { # MUST NOT
43	wakaba	1.36	$self->{onerror}->(node => $attr, type => 'in HTML:xml:lang');
44	wakaba	1.35	## TODO: Test data...
45			}
46	wakaba	1.13	},
47			base => sub {
48			my ($self, $attr) = @_;
49			my $value = $attr->value;
50			if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters?
51			$self->{onerror}->(node => $attr,
52	wakaba	1.33	type => 'invalid attribute value');
53	wakaba	1.13	}
54	wakaba	1.18	## NOTE: Conformance to URI standard is not checked since there is
55			## no author requirement on conformance in the XML Base specification.
56	wakaba	1.13	},
57			id => sub {
58			my ($self, $attr) = @_;
59			my $value = $attr->value;
60			$value =~ s/[\x09\x0A\x0D\x20]+/ /g;
61			$value =~ s/^\x20//;
62			$value =~ s/\x20$//;
63			## TODO: NCName in XML 1.0 or 1.1
64			## TODO: declared type is ID?
65	wakaba	1.33	if ($self->{id}->{$value}) { ## NOTE: An xml:id error
66			$self->{onerror}->(node => $attr, level => 'error',
67			type => 'duplicate ID');
68	wakaba	1.37	push @{$self->{id}->{$value}}, $attr;
69	wakaba	1.13	} else {
70	wakaba	1.37	$self->{id}->{$value} = [$attr];
71	wakaba	1.13	}
72			},
73	wakaba	1.9	},
74			$XMLNS_NS => {
75	wakaba	1.13	'' => sub {
76			my ($self, $attr) = @_;
77			my $ln = $attr->manakai_local_name;
78			my $value = $attr->value;
79			if ($value eq $XML_NS and $ln ne 'xml') {
80			$self->{onerror}
81	wakaba	1.33	->(node => $attr, level => 'NC',
82			type => 'Reserved Prefixes and Namespace Names:=xml');
83	wakaba	1.13	} elsif ($value eq $XMLNS_NS) {
84			$self->{onerror}
85	wakaba	1.33	->(node => $attr, level => 'NC',
86			type => 'Reserved Prefixes and Namespace Names:=xmlns');
87	wakaba	1.13	}
88			if ($ln eq 'xml' and $value ne $XML_NS) {
89			$self->{onerror}
90	wakaba	1.33	->(node => $attr, level => 'NC',
91			type => 'Reserved Prefixes and Namespace Names:xmlns:xml=');
92	wakaba	1.13	} elsif ($ln eq 'xmlns') {
93			$self->{onerror}
94	wakaba	1.33	->(node => $attr, level => 'NC',
95			type => 'Reserved Prefixes and Namespace Names:xmlns:xmlns=');
96	wakaba	1.13	}
97			## TODO: If XML 1.0 and empty
98			},
99			xmlns => sub {
100			my ($self, $attr) = @_;
101			## TODO: In XML 1.0, URI reference [RFC 3986] or an empty string
102			## TODO: In XML 1.1, IRI reference [RFC 3987] or an empty string
103	wakaba	1.18	## TODO: relative references are deprecated
104	wakaba	1.13	my $value = $attr->value;
105			if ($value eq $XML_NS) {
106			$self->{onerror}
107	wakaba	1.33	->(node => $attr, level => 'NC',
108			type => 'Reserved Prefixes and Namespace Names:=xml');
109	wakaba	1.13	} elsif ($value eq $XMLNS_NS) {
110			$self->{onerror}
111	wakaba	1.33	->(node => $attr, level => 'NC',
112			type => 'Reserved Prefixes and Namespace Names:=xmlns');
113	wakaba	1.13	}
114			},
115	wakaba	1.9	},
116			};
117
118	wakaba	1.14	## ISSUE: Should we really allow these attributes?
119	wakaba	1.13	$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space};
120			$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang};
121			$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base};
122			$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id};
123
124	wakaba	1.3	## ANY
125	wakaba	1.42	our $AnyChecker = sub {
126	wakaba	1.4	my ($self, $todo) = @_;
127			my $el = $todo->{node};
128			my $new_todos = [];
129	wakaba	1.3	my @nodes = (@{$el->child_nodes});
130			while (@nodes) {
131			my $node = shift @nodes;
132			$self->_remove_minuses ($node) and next if ref $node eq 'HASH';
133
134			my $nt = $node->node_type;
135			if ($nt == 1) {
136			my $node_ns = $node->namespace_uri;
137			$node_ns = '' unless defined $node_ns;
138			my $node_ln = $node->manakai_local_name;
139			if ($self->{minuses}->{$node_ns}->{$node_ln}) {
140			$self->{onerror}->(node => $node, type => 'element not allowed');
141			}
142	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
143	wakaba	1.3	} elsif ($nt == 5) {
144			unshift @nodes, @{$node->child_nodes};
145			}
146			}
147	wakaba	1.4	return ($new_todos);
148	wakaba	1.3	}; # $AnyChecker
149
150	wakaba	1.42	our $ElementDefault = {
151	wakaba	1.1	checker => sub {
152	wakaba	1.4	my ($self, $todo) = @_;
153	wakaba	1.33	$self->{onerror}->(node => $todo->{node}, level => 'unsupported',
154			type => 'element');
155	wakaba	1.4	return $AnyChecker->($self, $todo);
156	wakaba	1.1	},
157	wakaba	1.9	attrs_checker => sub {
158			my ($self, $todo) = @_;
159			for my $attr (@{$todo->{node}->attributes}) {
160			my $attr_ns = $attr->namespace_uri;
161			$attr_ns = '' unless defined $attr_ns;
162			my $attr_ln = $attr->manakai_local_name;
163			my $checker = $AttrChecker->{$attr_ns}->{$attr_ln}
164			\|\| $AttrChecker->{$attr_ns}->{''};
165			if ($checker) {
166			$checker->($self, $attr);
167	wakaba	1.17	} else {
168	wakaba	1.33	$self->{onerror}->(node => $attr, level => 'unsupported',
169			type => 'attribute');
170	wakaba	1.9	}
171			}
172			},
173	wakaba	1.1	};
174
175	wakaba	1.7	my $HTMLTransparentElements = {
176			$HTML_NS => {qw/ins 1 font 1 noscript 1/},
177	wakaba	1.29	## NOTE: \|html:noscript\| is transparent if scripting is disabled
178			## and not in \|head\|.
179	wakaba	1.7	};
180
181	wakaba	1.42	our $Element = {};
182	wakaba	1.7
183	wakaba	1.42	sub check_document ($$$) {
184			my ($self, $doc, $onerror) = @_;
185			$self = bless {}, $self unless ref $self;
186			$self->{onerror} = $onerror;
187	wakaba	1.1
188	wakaba	1.42	my $docel = $doc->document_element;
189			unless (defined $docel) {
190			## ISSUE: Should we check content of Document node?
191			$onerror->(node => $doc, type => 'no document element');
192			## ISSUE: Is this non-conforming (to what spec)? Or just a warning?
193			return {
194			class => {},
195			id => {}, table => [], term => {},
196			};
197	wakaba	1.1	}
198
199	wakaba	1.42	## ISSUE: Unexpanded entity references and HTML5 conformance
200	wakaba	1.1
201	wakaba	1.42	my $docel_nsuri = $docel->namespace_uri;
202			$docel_nsuri = '' unless defined $docel_nsuri;
203	wakaba	1.43	unless ($Namespace->{$docel_nsuri}->{loaded}) {
204			if ($Namespace->{$docel_nsuri}->{module}) {
205			eval qq{ require $Namespace->{$docel_nsuri}->{module} } or die $@;
206			} else {
207			$Namespace->{$docel_nsuri}->{loaded} = 1;
208			}
209			}
210	wakaba	1.42	my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} \|\|
211			$Element->{$docel_nsuri}->{''} \|\|
212			$ElementDefault;
213			if ($docel_def->{is_root}) {
214			#
215			} else {
216			$onerror->(node => $docel, type => 'element not allowed');
217	wakaba	1.1	}
218
219	wakaba	1.42	## TODO: Check for other items other than document element
220			## (second (errorous) element, text nodes, PI nodes, doctype nodes)
221	wakaba	1.2
222	wakaba	1.42	return $self->check_element ($docel, $onerror);
223			} # check_document
224	wakaba	1.1
225	wakaba	1.42	sub check_element ($$$) {
226			my ($self, $el, $onerror) = @_;
227			$self = bless {}, $self unless ref $self;
228			$self->{onerror} = $onerror;
229	wakaba	1.2
230	wakaba	1.42	$self->{minuses} = {};
231			$self->{id} = {};
232			$self->{term} = {};
233			$self->{usemap} = [];
234			$self->{contextmenu} = [];
235			$self->{map} = {};
236			$self->{menu} = {};
237			$self->{has_link_type} = {};
238			$self->{return} = {
239			class => {},
240			id => $self->{id}, table => [], term => $self->{term},
241			};
242	wakaba	1.4
243	wakaba	1.42	my @todo = ({type => 'element', node => $el});
244			while (@todo) {
245			my $todo = shift @todo;
246			if ($todo->{type} eq 'element') {
247			my $prefix = $todo->{node}->prefix;
248			if (defined $prefix and $prefix eq 'xmlns') {
249			$self->{onerror}
250			->(node => $todo->{node}, level => 'NC',
251			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
252	wakaba	1.7	}
253	wakaba	1.42	my $nsuri = $todo->{node}->namespace_uri;
254			$nsuri = '' unless defined $nsuri;
255			unless ($Namespace->{$nsuri}->{loaded}) {
256			if ($Namespace->{$nsuri}->{module}) {
257			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
258			} else {
259			$Namespace->{$nsuri}->{loaded} = 1;
260	wakaba	1.1	}
261			}
262	wakaba	1.42	my $ln = $todo->{node}->manakai_local_name;
263			my $eldef = $Element->{$nsuri}->{$ln} \|\|
264			$Element->{$nsuri}->{''} \|\|
265			$ElementDefault;
266			$eldef->{attrs_checker}->($self, $todo);
267			my ($new_todos) = $eldef->{checker}->($self, $todo);
268			unshift @todo, @$new_todos;
269			} elsif ($todo->{type} eq 'element-attributes') {
270			my $prefix = $todo->{node}->prefix;
271			if (defined $prefix and $prefix eq 'xmlns') {
272			$self->{onerror}
273			->(node => $todo->{node}, level => 'NC',
274			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
275			}
276			my $nsuri = $todo->{node}->namespace_uri;
277			$nsuri = '' unless defined $nsuri;
278			unless ($Namespace->{$nsuri}->{loaded}) {
279			if ($Namespace->{$nsuri}->{module}) {
280			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
281	wakaba	1.1	} else {
282	wakaba	1.42	$Namespace->{$nsuri}->{loaded} = 1;
283	wakaba	1.1	}
284			}
285	wakaba	1.9	my $ln = $todo->{node}->manakai_local_name;
286			my $eldef = $Element->{$nsuri}->{$ln} \|\|
287			$Element->{$nsuri}->{''} \|\|
288			$ElementDefault;
289			$eldef->{attrs_checker}->($self, $todo);
290	wakaba	1.4	} elsif ($todo->{type} eq 'plus') {
291			$self->_remove_minuses ($todo);
292	wakaba	1.30	} elsif ($todo->{type} eq 'code') {
293			$todo->{code}->();
294			} else {
295			die "$0: Internal error: Unsupported checking action type \|$todo->{type}\|";
296	wakaba	1.4	}
297	wakaba	1.1	}
298	wakaba	1.17
299			for (@{$self->{usemap}}) {
300			unless ($self->{map}->{$_->[0]}) {
301			$self->{onerror}->(node => $_->[1], type => 'no referenced map');
302			}
303			}
304
305	wakaba	1.32	for (@{$self->{contextmenu}}) {
306			unless ($self->{menu}->{$_->[0]}) {
307			$self->{onerror}->(node => $_->[1], type => 'no referenced menu');
308			}
309			}
310
311	wakaba	1.17	delete $self->{minuses};
312			delete $self->{onerror};
313			delete $self->{id};
314			delete $self->{usemap};
315			delete $self->{map};
316	wakaba	1.33	return $self->{return};
317	wakaba	1.1	} # check_element
318
319	wakaba	1.2	sub _add_minuses ($@) {
320			my $self = shift;
321			my $r = {};
322			for my $list (@_) {
323			for my $ns (keys %$list) {
324			for my $ln (keys %{$list->{$ns}}) {
325			unless ($self->{minuses}->{$ns}->{$ln}) {
326			$self->{minuses}->{$ns}->{$ln} = 1;
327			$r->{$ns}->{$ln} = 1;
328			}
329			}
330			}
331			}
332	wakaba	1.4	return {type => 'plus', list => $r};
333	wakaba	1.2	} # _add_minuses
334
335			sub _remove_minuses ($$) {
336	wakaba	1.4	my ($self, $todo) = @_;
337			for my $ns (keys %{$todo->{list}}) {
338			for my $ln (keys %{$todo->{list}->{$ns}}) {
339			delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
340	wakaba	1.2	}
341			}
342			1;
343			} # _remove_minuses
344
345	wakaba	1.30	sub _check_get_children ($$$) {
346			my ($self, $node, $parent_todo) = @_;
347	wakaba	1.4	my $new_todos = [];
348	wakaba	1.2	my $sib = [];
349			TP: {
350			my $node_ns = $node->namespace_uri;
351			$node_ns = '' unless defined $node_ns;
352			my $node_ln = $node->manakai_local_name;
353			if ($node_ns eq $HTML_NS) {
354			if ($node_ln eq 'noscript') {
355			my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
356			push @$sib, $end;
357			}
358			}
359	wakaba	1.31	## TODO: \|noscript\| is not a transparent element in \|head\|.
360	wakaba	1.7	if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
361			unshift @$sib, @{$node->child_nodes};
362	wakaba	1.9	push @$new_todos, {type => 'element-attributes', node => $node};
363	wakaba	1.7	last TP;
364	wakaba	1.2	}
365	wakaba	1.8	if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) {
366	wakaba	1.2	if ($node->has_attribute_ns (undef, 'src')) {
367			unshift @$sib, @{$node->child_nodes};
368	wakaba	1.9	push @$new_todos, {type => 'element-attributes', node => $node};
369	wakaba	1.2	last TP;
370			} else {
371			my @cn = @{$node->child_nodes};
372			CN: while (@cn) {
373			my $cn = shift @cn;
374			my $cnt = $cn->node_type;
375			if ($cnt == 1) {
376	wakaba	1.8	my $cn_nsuri = $cn->namespace_uri;
377			$cn_nsuri = '' unless defined $cn_nsuri;
378			if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') {
379	wakaba	1.2	#
380			} else {
381			last CN;
382			}
383			} elsif ($cnt == 3 or $cnt == 4) {
384			if ($cn->data =~ /[^\x09-\x0D\x20]/) {
385			last CN;
386			}
387			}
388			} # CN
389			unshift @$sib, @cn;
390			}
391			}
392	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
393	wakaba	1.2	} # TP
394	wakaba	1.30
395			for my $new_todo (@$new_todos) {
396			$new_todo->{flag} = {%{$parent_todo->{flag} or {}}};
397			}
398
399	wakaba	1.4	return ($sib, $new_todos);
400	wakaba	1.2	} # _check_get_children
401
402	wakaba	1.44	=head1 LICENSE
403
404			Copyright 2007 Wakaba <w@suika.fam.cx>
405
406			This library is free software; you can redistribute it
407			and/or modify it under the same terms as Perl itself.
408
409			=cut
410
411	wakaba	1.1	1;
412	wakaba	1.44	# $Date: 2007/08/05 04:50:57 $