whatpm/Whatpm/ContentChecker.pm

package Whatpm::ContentChecker;
use strict;
our $VERSION=do{my @r=(q$Revision: 1.56 $=~/\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)?

## TODO: Conformance of an HTML document with non-html root element.

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 => $node, type => 'element not allowed');
      }
      my ($sib, $ch) = $self->_check_get_children ($node, $todo);
      unshift @nodes, @$sib;
      push @$new_todos, @$ch;
    } elsif ($nt == 3 or $nt == 4) {
      if ($node->data =~ /[^\x09-\x0D\x20]/) {
        $todo->{flag}->{has_descendant}->{significant} = 1;
      }
    } elsif ($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 del 1 font 1 noscript 1 canvas 1/},
  ## NOTE: |html:noscript| is transparent if scripting is disabled
  ## and not in |head|.
};

## Semi-transparent: html:video, html:audio, html:object

our $Element = {};

sub check_document ($$$;$) {
  my ($self, $doc, $onerror, $onsubdoc) = @_;
  $self = bless {}, $self unless ref $self;
  $self->{onerror} = $onerror;
  $self->{onsubdoc} = $onsubdoc || sub {
    warn "A subdocument is not conformance-checked";
  };

  $self->{must_level} = 'm';
  $self->{fact_level} = 'f';
  $self->{should_level} = 's';
  $self->{good_level} = 'w';

  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}) {
    #
  } elsif ($docel_def->{is_xml_root}) {
    unless ($doc->manakai_is_html) {
      #
    } else {
      $onerror->(node => $docel, type => 'element not allowed:root:xml');
    }
  } else {
    $onerror->(node => $docel, type => 'element not allowed:root');
  }

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

  my $return = $self->check_element ($docel, $onerror, $onsubdoc);

  ## TODO: Test for these checks are necessary.
  my $charset_name = $doc->input_encoding;
  if (defined $charset_name) {
    require Message::Charset::Info;
    my $charset = $Message::Charset::Info::IANACharset->{$charset_name};

    if ($doc->manakai_is_html and
        not $doc->manakai_has_bom and
        not defined $doc->manakai_charset) {
      unless ($charset->{is_html_ascii_superset}) {
        $onerror->(node => $doc, level => $self->{must_level},
                   type => 'non ascii superset:'.$charset_name);
      }
      
      if (not $self->{has_charset} and
          not $charset->{iana_names}->{'us-ascii'}) {
        $onerror->(node => $doc, level => $self->{must_level},
                   type => 'no character encoding declaration:'.$charset_name);
      }
    }

    if ($charset->{iana_names}->{'utf-8'}) {
      #
    } elsif ($charset->{iana_names}->{'jis_x0212-1990'} or
             $charset->{iana_names}->{'x-jis0208'} or
             $charset->{iana_names}->{'utf-32'} or ## ISSUE: UTF-32BE? UTF-32LE?
             $charset->{is_ebcdic_based}) {
      $onerror->(node => $doc,
                 type => 'character encoding:'.$charset_name,
                 level => $self->{should_level});
    } elsif ($charset->{iana_names}->{'cesu-8'} or
             $charset->{iana_names}->{'utf-8'} or ## ISSUE: UNICODE-1-1-UTF-7?
             $charset->{iana_names}->{'bocu-1'} or
             $charset->{iana_names}->{'scsu'}) {
      $onerror->(node => $doc,
                 type => 'character encoding:'.$charset_name,
                 level => $self->{must_level});
    } else {
      $onerror->(node => $doc,
                 type => 'character encoding:'.$charset_name,
                 level => $self->{good_level});
    }
  } elsif ($doc->manakai_is_html) {
    ## NOTE: MUST and SHOULD requirements above cannot be tested,
    ## since the document has no input charset encoding information.
    $onerror->(node => $doc,
               type => 'character encoding:',
               level => 'unsupported');
  }

  return $return;
} # check_document

sub check_element ($$$;$) {
  my ($self, $el, $onerror, $onsubdoc) = @_;
  $self = bless {}, $self unless ref $self;
  $self->{onerror} = $onerror;
  $self->{onsubdoc} = $onsubdoc || sub {
    warn "A subdocument is not conformance-checked";
  };

  $self->{must_level} = 'm';
  $self->{fact_level} = 'f';
  $self->{should_level} = 's';
  $self->{good_level} = 'w';

  $self->{pluses} = {};
  $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->{has_charset};
  #$self->{has_base};
  $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 'descendant') {
      for my $key (keys %{$todo->{errors}}) {
        unless ($todo->{flag}->{has_descendant}->{$key}) {
          $todo->{errors}->{$key}->($self, $todo);
        }
        for my $key (keys %{$todo->{old_values}}) {
          $todo->{flag}->{has_descendant}->{$key}
              ||= $todo->{old_values}->{$key};
        }
      }
    } elsif ($todo->{type} eq 'plus' or $todo->{type} eq 'minus') {
      $self->_remove_minuses ($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->{pluses};
  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 _add_pluses ($@) {
  my $self = shift;
  my $r = {};
  for my $list (@_) {
    for my $ns (keys %$list) {
      for my $ln (keys %{$list->{$ns}}) {
        unless ($self->{pluses}->{$ns}->{$ln}) {
          $self->{pluses}->{$ns}->{$ln} = 1;
          $r->{$ns}->{$ln} = 1;
        }
      }
    }
  }
  return {type => 'minus', list => $r};
} # _add_pluses

sub _remove_minuses ($$) {
  my ($self, $todo) = @_;
  if ($todo->{type} eq 'minus') {
    for my $ns (keys %{$todo->{list}}) {
      for my $ln (keys %{$todo->{list}->{$ns}}) {
        delete $self->{pluses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
      }
    }
  } elsif ($todo->{type} eq 'plus') {
    for my $ns (keys %{$todo->{list}}) {
      for my $ln (keys %{$todo->{list}->{$ns}}) {
        delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
      }
    }
  } else {
    die "$0: Unknown +- type: $todo->{type}";
  }
  1;
} # _remove_minuses

## NOTE: Priority for "minuses" and "pluses" are currently left
## undefined and implemented inconsistently; it is not a problem for
## now, since no element belongs to both lists.

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 @$sib, @cn;
      }
    } elsif ($node_ns eq $HTML_NS and $node_ln eq 'object') {
      my @cn = @{$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 $cn->manakai_local_name eq 'param') {
            #
          } else {
            last CN;
          }
        } elsif ($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 2007-2008 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: 2008/02/10 04:09:57 $
1	wakaba	1.1	package Whatpm::ContentChecker;
2			use strict;
3	wakaba	1.57	our $VERSION=do{my @r=(q$Revision: 1.56 $=~/\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	1.50	## TODO: Conformance of an HTML document with non-html root element.
11
12	wakaba	1.42	my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
13	wakaba	1.9	my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
14			my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>;
15
16	wakaba	1.42	my $Namespace = {
17	wakaba	1.43	q<http://www.w3.org/2005/Atom> => {module => 'Whatpm::ContentChecker::Atom'},
18	wakaba	1.42	$HTML_NS => {module => 'Whatpm::ContentChecker::HTML'},
19			$XML_NS => {loaded => 1},
20			$XMLNS_NS => {loaded => 1},
21			};
22
23			our $AttrChecker = {
24	wakaba	1.9	$XML_NS => {
25	wakaba	1.13	space => sub {
26			my ($self, $attr) = @_;
27			my $value = $attr->value;
28			if ($value eq 'default' or $value eq 'preserve') {
29			#
30			} else {
31			## NOTE: An XML "error"
32	wakaba	1.33	$self->{onerror}->(node => $attr, level => 'error',
33			type => 'invalid attribute value');
34	wakaba	1.13	}
35			},
36			lang => sub {
37	wakaba	1.35	my ($self, $attr) = @_;
38	wakaba	1.47	my $value = $attr->value;
39			if ($value eq '') {
40			#
41			} else {
42			require Whatpm::LangTag;
43			Whatpm::LangTag->check_rfc3066_language_tag ($value, sub {
44			my %opt = @_;
45			my $type = 'LangTag:'.$opt{type};
46			$type .= ':' . $opt{subtag} if defined $opt{subtag};
47			$self->{onerror}->(node => $attr, type => $type,
48			value => $opt{value}, level => $opt{level});
49			});
50			}
51
52	wakaba	1.13	## NOTE: "The values of the attribute are language identifiers
53			## as defined by [IETF RFC 3066], Tags for the Identification
54			## of Languages, or its successor; in addition, the empty string
55			## may be specified." ("may" in lower case)
56	wakaba	1.47	## NOTE: Is an RFC 3066-valid (but RFC 4647-invalid) language tag
57			## allowed today?
58
59			## TODO: test data
60
61	wakaba	1.35	if ($attr->owner_document->manakai_is_html) { # MUST NOT
62	wakaba	1.36	$self->{onerror}->(node => $attr, type => 'in HTML:xml:lang');
63	wakaba	1.35	## TODO: Test data...
64			}
65	wakaba	1.13	},
66			base => sub {
67			my ($self, $attr) = @_;
68			my $value = $attr->value;
69			if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters?
70			$self->{onerror}->(node => $attr,
71	wakaba	1.33	type => 'invalid attribute value');
72	wakaba	1.13	}
73	wakaba	1.18	## NOTE: Conformance to URI standard is not checked since there is
74			## no author requirement on conformance in the XML Base specification.
75	wakaba	1.13	},
76			id => sub {
77			my ($self, $attr) = @_;
78			my $value = $attr->value;
79			$value =~ s/[\x09\x0A\x0D\x20]+/ /g;
80			$value =~ s/^\x20//;
81			$value =~ s/\x20$//;
82			## TODO: NCName in XML 1.0 or 1.1
83			## TODO: declared type is ID?
84	wakaba	1.33	if ($self->{id}->{$value}) { ## NOTE: An xml:id error
85			$self->{onerror}->(node => $attr, level => 'error',
86			type => 'duplicate ID');
87	wakaba	1.37	push @{$self->{id}->{$value}}, $attr;
88	wakaba	1.13	} else {
89	wakaba	1.37	$self->{id}->{$value} = [$attr];
90	wakaba	1.13	}
91			},
92	wakaba	1.9	},
93			$XMLNS_NS => {
94	wakaba	1.13	'' => sub {
95			my ($self, $attr) = @_;
96			my $ln = $attr->manakai_local_name;
97			my $value = $attr->value;
98			if ($value eq $XML_NS and $ln ne 'xml') {
99			$self->{onerror}
100	wakaba	1.33	->(node => $attr, level => 'NC',
101			type => 'Reserved Prefixes and Namespace Names:=xml');
102	wakaba	1.13	} elsif ($value eq $XMLNS_NS) {
103			$self->{onerror}
104	wakaba	1.33	->(node => $attr, level => 'NC',
105			type => 'Reserved Prefixes and Namespace Names:=xmlns');
106	wakaba	1.13	}
107			if ($ln eq 'xml' and $value ne $XML_NS) {
108			$self->{onerror}
109	wakaba	1.33	->(node => $attr, level => 'NC',
110			type => 'Reserved Prefixes and Namespace Names:xmlns:xml=');
111	wakaba	1.13	} elsif ($ln eq 'xmlns') {
112			$self->{onerror}
113	wakaba	1.33	->(node => $attr, level => 'NC',
114			type => 'Reserved Prefixes and Namespace Names:xmlns:xmlns=');
115	wakaba	1.13	}
116			## TODO: If XML 1.0 and empty
117			},
118			xmlns => sub {
119			my ($self, $attr) = @_;
120			## TODO: In XML 1.0, URI reference [RFC 3986] or an empty string
121			## TODO: In XML 1.1, IRI reference [RFC 3987] or an empty string
122	wakaba	1.18	## TODO: relative references are deprecated
123	wakaba	1.13	my $value = $attr->value;
124			if ($value eq $XML_NS) {
125			$self->{onerror}
126	wakaba	1.33	->(node => $attr, level => 'NC',
127			type => 'Reserved Prefixes and Namespace Names:=xml');
128	wakaba	1.13	} elsif ($value eq $XMLNS_NS) {
129			$self->{onerror}
130	wakaba	1.33	->(node => $attr, level => 'NC',
131			type => 'Reserved Prefixes and Namespace Names:=xmlns');
132	wakaba	1.13	}
133			},
134	wakaba	1.9	},
135			};
136
137	wakaba	1.14	## ISSUE: Should we really allow these attributes?
138	wakaba	1.13	$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space};
139			$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang};
140			$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base};
141			$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id};
142
143	wakaba	1.3	## ANY
144	wakaba	1.42	our $AnyChecker = sub {
145	wakaba	1.4	my ($self, $todo) = @_;
146			my $el = $todo->{node};
147			my $new_todos = [];
148	wakaba	1.3	my @nodes = (@{$el->child_nodes});
149			while (@nodes) {
150			my $node = shift @nodes;
151			$self->_remove_minuses ($node) and next if ref $node eq 'HASH';
152
153			my $nt = $node->node_type;
154			if ($nt == 1) {
155			my $node_ns = $node->namespace_uri;
156			$node_ns = '' unless defined $node_ns;
157			my $node_ln = $node->manakai_local_name;
158			if ($self->{minuses}->{$node_ns}->{$node_ln}) {
159			$self->{onerror}->(node => $node, type => 'element not allowed');
160			}
161	wakaba	1.54	my ($sib, $ch) = $self->_check_get_children ($node, $todo);
162			unshift @nodes, @$sib;
163			push @$new_todos, @$ch;
164			} elsif ($nt == 3 or $nt == 4) {
165			if ($node->data =~ /[^\x09-\x0D\x20]/) {
166			$todo->{flag}->{has_descendant}->{significant} = 1;
167			}
168	wakaba	1.3	} elsif ($nt == 5) {
169			unshift @nodes, @{$node->child_nodes};
170			}
171			}
172	wakaba	1.4	return ($new_todos);
173	wakaba	1.3	}; # $AnyChecker
174
175	wakaba	1.42	our $ElementDefault = {
176	wakaba	1.1	checker => sub {
177	wakaba	1.4	my ($self, $todo) = @_;
178	wakaba	1.33	$self->{onerror}->(node => $todo->{node}, level => 'unsupported',
179			type => 'element');
180	wakaba	1.4	return $AnyChecker->($self, $todo);
181	wakaba	1.1	},
182	wakaba	1.9	attrs_checker => sub {
183			my ($self, $todo) = @_;
184			for my $attr (@{$todo->{node}->attributes}) {
185			my $attr_ns = $attr->namespace_uri;
186			$attr_ns = '' unless defined $attr_ns;
187			my $attr_ln = $attr->manakai_local_name;
188			my $checker = $AttrChecker->{$attr_ns}->{$attr_ln}
189			\|\| $AttrChecker->{$attr_ns}->{''};
190			if ($checker) {
191			$checker->($self, $attr);
192	wakaba	1.17	} else {
193	wakaba	1.33	$self->{onerror}->(node => $attr, level => 'unsupported',
194			type => 'attribute');
195	wakaba	1.9	}
196			}
197			},
198	wakaba	1.1	};
199
200	wakaba	1.7	my $HTMLTransparentElements = {
201	wakaba	1.57	$HTML_NS => {qw/ins 1 del 1 font 1 noscript 1 canvas 1/},
202	wakaba	1.29	## NOTE: \|html:noscript\| is transparent if scripting is disabled
203			## and not in \|head\|.
204	wakaba	1.7	};
205
206	wakaba	1.57	## Semi-transparent: html:video, html:audio, html:object
207
208	wakaba	1.42	our $Element = {};
209	wakaba	1.7
210	wakaba	1.56	sub check_document ($$$;$) {
211			my ($self, $doc, $onerror, $onsubdoc) = @_;
212	wakaba	1.42	$self = bless {}, $self unless ref $self;
213			$self->{onerror} = $onerror;
214	wakaba	1.56	$self->{onsubdoc} = $onsubdoc \|\| sub {
215			warn "A subdocument is not conformance-checked";
216			};
217	wakaba	1.1
218	wakaba	1.48	$self->{must_level} = 'm';
219			$self->{fact_level} = 'f';
220			$self->{should_level} = 's';
221	wakaba	1.51	$self->{good_level} = 'w';
222	wakaba	1.48
223	wakaba	1.42	my $docel = $doc->document_element;
224			unless (defined $docel) {
225			## ISSUE: Should we check content of Document node?
226			$onerror->(node => $doc, type => 'no document element');
227			## ISSUE: Is this non-conforming (to what spec)? Or just a warning?
228			return {
229			class => {},
230			id => {}, table => [], term => {},
231			};
232	wakaba	1.1	}
233
234	wakaba	1.42	## ISSUE: Unexpanded entity references and HTML5 conformance
235	wakaba	1.1
236	wakaba	1.42	my $docel_nsuri = $docel->namespace_uri;
237			$docel_nsuri = '' unless defined $docel_nsuri;
238	wakaba	1.43	unless ($Namespace->{$docel_nsuri}->{loaded}) {
239			if ($Namespace->{$docel_nsuri}->{module}) {
240			eval qq{ require $Namespace->{$docel_nsuri}->{module} } or die $@;
241			} else {
242			$Namespace->{$docel_nsuri}->{loaded} = 1;
243			}
244			}
245	wakaba	1.42	my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} \|\|
246			$Element->{$docel_nsuri}->{''} \|\|
247			$ElementDefault;
248			if ($docel_def->{is_root}) {
249			#
250	wakaba	1.50	} elsif ($docel_def->{is_xml_root}) {
251			unless ($doc->manakai_is_html) {
252			#
253			} else {
254			$onerror->(node => $docel, type => 'element not allowed:root:xml');
255			}
256	wakaba	1.42	} else {
257	wakaba	1.49	$onerror->(node => $docel, type => 'element not allowed:root');
258	wakaba	1.1	}
259
260	wakaba	1.42	## TODO: Check for other items other than document element
261			## (second (errorous) element, text nodes, PI nodes, doctype nodes)
262	wakaba	1.2
263	wakaba	1.56	my $return = $self->check_element ($docel, $onerror, $onsubdoc);
264	wakaba	1.51
265	wakaba	1.52	## TODO: Test for these checks are necessary.
266	wakaba	1.51	my $charset_name = $doc->input_encoding;
267			if (defined $charset_name) {
268			require Message::Charset::Info;
269			my $charset = $Message::Charset::Info::IANACharset->{$charset_name};
270
271			if ($doc->manakai_is_html and
272			not $doc->manakai_has_bom and
273			not defined $doc->manakai_charset) {
274			unless ($charset->{is_html_ascii_superset}) {
275			$onerror->(node => $doc, level => $self->{must_level},
276			type => 'non ascii superset:'.$charset_name);
277			}
278
279			if (not $self->{has_charset} and
280			not $charset->{iana_names}->{'us-ascii'}) {
281			$onerror->(node => $doc, level => $self->{must_level},
282			type => 'no character encoding declaration:'.$charset_name);
283			}
284			}
285
286			if ($charset->{iana_names}->{'utf-8'}) {
287			#
288			} elsif ($charset->{iana_names}->{'jis_x0212-1990'} or
289			$charset->{iana_names}->{'x-jis0208'} or
290			$charset->{iana_names}->{'utf-32'} or ## ISSUE: UTF-32BE? UTF-32LE?
291			$charset->{is_ebcdic_based}) {
292			$onerror->(node => $doc,
293			type => 'character encoding:'.$charset_name,
294			level => $self->{should_level});
295			} elsif ($charset->{iana_names}->{'cesu-8'} or
296			$charset->{iana_names}->{'utf-8'} or ## ISSUE: UNICODE-1-1-UTF-7?
297			$charset->{iana_names}->{'bocu-1'} or
298			$charset->{iana_names}->{'scsu'}) {
299			$onerror->(node => $doc,
300			type => 'character encoding:'.$charset_name,
301			level => $self->{must_level});
302			} else {
303			$onerror->(node => $doc,
304			type => 'character encoding:'.$charset_name,
305			level => $self->{good_level});
306			}
307	wakaba	1.52	} elsif ($doc->manakai_is_html) {
308			## NOTE: MUST and SHOULD requirements above cannot be tested,
309			## since the document has no input charset encoding information.
310			$onerror->(node => $doc,
311			type => 'character encoding:',
312			level => 'unsupported');
313	wakaba	1.51	}
314
315			return $return;
316	wakaba	1.42	} # check_document
317	wakaba	1.1
318	wakaba	1.56	sub check_element ($$$;$) {
319			my ($self, $el, $onerror, $onsubdoc) = @_;
320	wakaba	1.42	$self = bless {}, $self unless ref $self;
321			$self->{onerror} = $onerror;
322	wakaba	1.56	$self->{onsubdoc} = $onsubdoc \|\| sub {
323			warn "A subdocument is not conformance-checked";
324			};
325	wakaba	1.2
326	wakaba	1.48	$self->{must_level} = 'm';
327			$self->{fact_level} = 'f';
328			$self->{should_level} = 's';
329	wakaba	1.51	$self->{good_level} = 'w';
330	wakaba	1.48
331	wakaba	1.50	$self->{pluses} = {};
332	wakaba	1.42	$self->{minuses} = {};
333			$self->{id} = {};
334			$self->{term} = {};
335			$self->{usemap} = [];
336			$self->{contextmenu} = [];
337			$self->{map} = {};
338			$self->{menu} = {};
339			$self->{has_link_type} = {};
340	wakaba	1.46	#$self->{has_uri_attr};
341			#$self->{has_hyperlink_element};
342	wakaba	1.51	#$self->{has_charset};
343	wakaba	1.57	#$self->{has_base};
344	wakaba	1.42	$self->{return} = {
345			class => {},
346			id => $self->{id}, table => [], term => $self->{term},
347			};
348	wakaba	1.4
349	wakaba	1.42	my @todo = ({type => 'element', node => $el});
350			while (@todo) {
351			my $todo = shift @todo;
352			if ($todo->{type} eq 'element') {
353			my $prefix = $todo->{node}->prefix;
354			if (defined $prefix and $prefix eq 'xmlns') {
355			$self->{onerror}
356			->(node => $todo->{node}, level => 'NC',
357			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
358	wakaba	1.7	}
359	wakaba	1.42	my $nsuri = $todo->{node}->namespace_uri;
360			$nsuri = '' unless defined $nsuri;
361			unless ($Namespace->{$nsuri}->{loaded}) {
362			if ($Namespace->{$nsuri}->{module}) {
363			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
364			} else {
365			$Namespace->{$nsuri}->{loaded} = 1;
366	wakaba	1.1	}
367			}
368	wakaba	1.42	my $ln = $todo->{node}->manakai_local_name;
369			my $eldef = $Element->{$nsuri}->{$ln} \|\|
370			$Element->{$nsuri}->{''} \|\|
371			$ElementDefault;
372			$eldef->{attrs_checker}->($self, $todo);
373			my ($new_todos) = $eldef->{checker}->($self, $todo);
374			unshift @todo, @$new_todos;
375			} elsif ($todo->{type} eq 'element-attributes') {
376			my $prefix = $todo->{node}->prefix;
377			if (defined $prefix and $prefix eq 'xmlns') {
378			$self->{onerror}
379			->(node => $todo->{node}, level => 'NC',
380			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
381			}
382			my $nsuri = $todo->{node}->namespace_uri;
383			$nsuri = '' unless defined $nsuri;
384			unless ($Namespace->{$nsuri}->{loaded}) {
385			if ($Namespace->{$nsuri}->{module}) {
386			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
387	wakaba	1.1	} else {
388	wakaba	1.42	$Namespace->{$nsuri}->{loaded} = 1;
389	wakaba	1.1	}
390			}
391	wakaba	1.9	my $ln = $todo->{node}->manakai_local_name;
392			my $eldef = $Element->{$nsuri}->{$ln} \|\|
393			$Element->{$nsuri}->{''} \|\|
394			$ElementDefault;
395			$eldef->{attrs_checker}->($self, $todo);
396	wakaba	1.53	} elsif ($todo->{type} eq 'descendant') {
397			for my $key (keys %{$todo->{errors}}) {
398			unless ($todo->{flag}->{has_descendant}->{$key}) {
399			$todo->{errors}->{$key}->($self, $todo);
400			}
401			for my $key (keys %{$todo->{old_values}}) {
402			$todo->{flag}->{has_descendant}->{$key}
403			\|\|= $todo->{old_values}->{$key};
404			}
405			}
406	wakaba	1.50	} elsif ($todo->{type} eq 'plus' or $todo->{type} eq 'minus') {
407	wakaba	1.4	$self->_remove_minuses ($todo);
408	wakaba	1.30	} elsif ($todo->{type} eq 'code') {
409			$todo->{code}->();
410			} else {
411			die "$0: Internal error: Unsupported checking action type \|$todo->{type}\|";
412	wakaba	1.4	}
413	wakaba	1.1	}
414	wakaba	1.17
415			for (@{$self->{usemap}}) {
416			unless ($self->{map}->{$_->[0]}) {
417			$self->{onerror}->(node => $_->[1], type => 'no referenced map');
418			}
419			}
420
421	wakaba	1.32	for (@{$self->{contextmenu}}) {
422			unless ($self->{menu}->{$_->[0]}) {
423			$self->{onerror}->(node => $_->[1], type => 'no referenced menu');
424			}
425			}
426
427	wakaba	1.50	delete $self->{pluses};
428	wakaba	1.17	delete $self->{minuses};
429			delete $self->{onerror};
430			delete $self->{id};
431			delete $self->{usemap};
432			delete $self->{map};
433	wakaba	1.33	return $self->{return};
434	wakaba	1.1	} # check_element
435
436	wakaba	1.2	sub _add_minuses ($@) {
437			my $self = shift;
438			my $r = {};
439			for my $list (@_) {
440			for my $ns (keys %$list) {
441			for my $ln (keys %{$list->{$ns}}) {
442			unless ($self->{minuses}->{$ns}->{$ln}) {
443			$self->{minuses}->{$ns}->{$ln} = 1;
444			$r->{$ns}->{$ln} = 1;
445			}
446			}
447			}
448			}
449	wakaba	1.4	return {type => 'plus', list => $r};
450	wakaba	1.2	} # _add_minuses
451
452	wakaba	1.50	sub _add_pluses ($@) {
453			my $self = shift;
454			my $r = {};
455			for my $list (@_) {
456			for my $ns (keys %$list) {
457			for my $ln (keys %{$list->{$ns}}) {
458			unless ($self->{pluses}->{$ns}->{$ln}) {
459			$self->{pluses}->{$ns}->{$ln} = 1;
460			$r->{$ns}->{$ln} = 1;
461			}
462			}
463			}
464			}
465			return {type => 'minus', list => $r};
466			} # _add_pluses
467
468	wakaba	1.2	sub _remove_minuses ($$) {
469	wakaba	1.4	my ($self, $todo) = @_;
470	wakaba	1.50	if ($todo->{type} eq 'minus') {
471			for my $ns (keys %{$todo->{list}}) {
472			for my $ln (keys %{$todo->{list}->{$ns}}) {
473			delete $self->{pluses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
474			}
475	wakaba	1.2	}
476	wakaba	1.50	} elsif ($todo->{type} eq 'plus') {
477			for my $ns (keys %{$todo->{list}}) {
478			for my $ln (keys %{$todo->{list}->{$ns}}) {
479			delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
480			}
481			}
482			} else {
483			die "$0: Unknown +- type: $todo->{type}";
484	wakaba	1.2	}
485			1;
486			} # _remove_minuses
487
488	wakaba	1.50	## NOTE: Priority for "minuses" and "pluses" are currently left
489			## undefined and implemented inconsistently; it is not a problem for
490			## now, since no element belongs to both lists.
491
492	wakaba	1.30	sub _check_get_children ($$$) {
493			my ($self, $node, $parent_todo) = @_;
494	wakaba	1.4	my $new_todos = [];
495	wakaba	1.2	my $sib = [];
496			TP: {
497			my $node_ns = $node->namespace_uri;
498			$node_ns = '' unless defined $node_ns;
499			my $node_ln = $node->manakai_local_name;
500	wakaba	1.45	if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
501			if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') {
502			if ($parent_todo->{flag}->{in_head}) {
503			#
504			} else {
505			my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
506			push @$sib, $end;
507
508			unshift @$sib, @{$node->child_nodes};
509			push @$new_todos, {type => 'element-attributes', node => $node};
510			last TP;
511			}
512			} else {
513			unshift @$sib, @{$node->child_nodes};
514			push @$new_todos, {type => 'element-attributes', node => $node};
515			last TP;
516	wakaba	1.2	}
517			}
518	wakaba	1.8	if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) {
519	wakaba	1.2	if ($node->has_attribute_ns (undef, 'src')) {
520			unshift @$sib, @{$node->child_nodes};
521	wakaba	1.9	push @$new_todos, {type => 'element-attributes', node => $node};
522	wakaba	1.2	last TP;
523			} else {
524			my @cn = @{$node->child_nodes};
525			CN: while (@cn) {
526			my $cn = shift @cn;
527			my $cnt = $cn->node_type;
528			if ($cnt == 1) {
529	wakaba	1.8	my $cn_nsuri = $cn->namespace_uri;
530			$cn_nsuri = '' unless defined $cn_nsuri;
531			if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') {
532	wakaba	1.2	#
533			} else {
534			last CN;
535			}
536			} elsif ($cnt == 3 or $cnt == 4) {
537			if ($cn->data =~ /[^\x09-\x0D\x20]/) {
538			last CN;
539			}
540			}
541			} # CN
542			unshift @$sib, @cn;
543			}
544	wakaba	1.57	} elsif ($node_ns eq $HTML_NS and $node_ln eq 'object') {
545			my @cn = @{$node->child_nodes};
546			CN: while (@cn) {
547			my $cn = shift @cn;
548			my $cnt = $cn->node_type;
549			if ($cnt == 1) {
550			my $cn_nsuri = $cn->namespace_uri;
551			$cn_nsuri = '' unless defined $cn_nsuri;
552			if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'param') {
553			#
554			} else {
555			last CN;
556			}
557			} elsif ($cnt == 3 or $cnt == 4) {
558			if ($cn->data =~ /[^\x09-\x0D\x20]/) {
559			last CN;
560			}
561			}
562			} # CN
563			unshift @$sib, @cn;
564	wakaba	1.2	}
565	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
566	wakaba	1.2	} # TP
567	wakaba	1.30
568			for my $new_todo (@$new_todos) {
569			$new_todo->{flag} = {%{$parent_todo->{flag} or {}}};
570			}
571
572	wakaba	1.4	return ($sib, $new_todos);
573	wakaba	1.2	} # _check_get_children
574
575	wakaba	1.44	=head1 LICENSE
576
577	wakaba	1.56	Copyright 2007-2008 Wakaba <w@suika.fam.cx>
578	wakaba	1.44
579			This library is free software; you can redistribute it
580			and/or modify it under the same terms as Perl itself.
581
582			=cut
583
584	wakaba	1.1	1;
585	wakaba	1.57	# $Date: 2008/02/10 04:09:57 $