whatpm/Whatpm/ContentChecker.pm

package Whatpm::ContentChecker;
use strict;
our $VERSION=do{my @r=(q$Revision: 1.50 $=~/\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 => $node, type => 'element not allowed');
      }
      push @$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;

  $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);

  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});
    }
  }

  return $return;
} # check_document

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

  $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->{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' 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 @$sib, @cn;
      }
    }
    push @$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: 2007/10/14 09:21:46 $
1	wakaba	1.1	package Whatpm::ContentChecker;
2			use strict;
3	wakaba	1.51	our $VERSION=do{my @r=(q$Revision: 1.50 $=~/\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 => $node, type => 'element not allowed');
160			}
161	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
162	wakaba	1.3	} elsif ($nt == 5) {
163			unshift @nodes, @{$node->child_nodes};
164			}
165			}
166	wakaba	1.4	return ($new_todos);
167	wakaba	1.3	}; # $AnyChecker
168
169	wakaba	1.42	our $ElementDefault = {
170	wakaba	1.1	checker => sub {
171	wakaba	1.4	my ($self, $todo) = @_;
172	wakaba	1.33	$self->{onerror}->(node => $todo->{node}, level => 'unsupported',
173			type => 'element');
174	wakaba	1.4	return $AnyChecker->($self, $todo);
175	wakaba	1.1	},
176	wakaba	1.9	attrs_checker => sub {
177			my ($self, $todo) = @_;
178			for my $attr (@{$todo->{node}->attributes}) {
179			my $attr_ns = $attr->namespace_uri;
180			$attr_ns = '' unless defined $attr_ns;
181			my $attr_ln = $attr->manakai_local_name;
182			my $checker = $AttrChecker->{$attr_ns}->{$attr_ln}
183			\|\| $AttrChecker->{$attr_ns}->{''};
184			if ($checker) {
185			$checker->($self, $attr);
186	wakaba	1.17	} else {
187	wakaba	1.33	$self->{onerror}->(node => $attr, level => 'unsupported',
188			type => 'attribute');
189	wakaba	1.9	}
190			}
191			},
192	wakaba	1.1	};
193
194	wakaba	1.7	my $HTMLTransparentElements = {
195			$HTML_NS => {qw/ins 1 font 1 noscript 1/},
196	wakaba	1.29	## NOTE: \|html:noscript\| is transparent if scripting is disabled
197			## and not in \|head\|.
198	wakaba	1.7	};
199
200	wakaba	1.42	our $Element = {};
201	wakaba	1.7
202	wakaba	1.42	sub check_document ($$$) {
203			my ($self, $doc, $onerror) = @_;
204			$self = bless {}, $self unless ref $self;
205			$self->{onerror} = $onerror;
206	wakaba	1.1
207	wakaba	1.48	$self->{must_level} = 'm';
208			$self->{fact_level} = 'f';
209			$self->{should_level} = 's';
210	wakaba	1.51	$self->{good_level} = 'w';
211	wakaba	1.48
212	wakaba	1.42	my $docel = $doc->document_element;
213			unless (defined $docel) {
214			## ISSUE: Should we check content of Document node?
215			$onerror->(node => $doc, type => 'no document element');
216			## ISSUE: Is this non-conforming (to what spec)? Or just a warning?
217			return {
218			class => {},
219			id => {}, table => [], term => {},
220			};
221	wakaba	1.1	}
222
223	wakaba	1.42	## ISSUE: Unexpanded entity references and HTML5 conformance
224	wakaba	1.1
225	wakaba	1.42	my $docel_nsuri = $docel->namespace_uri;
226			$docel_nsuri = '' unless defined $docel_nsuri;
227	wakaba	1.43	unless ($Namespace->{$docel_nsuri}->{loaded}) {
228			if ($Namespace->{$docel_nsuri}->{module}) {
229			eval qq{ require $Namespace->{$docel_nsuri}->{module} } or die $@;
230			} else {
231			$Namespace->{$docel_nsuri}->{loaded} = 1;
232			}
233			}
234	wakaba	1.42	my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} \|\|
235			$Element->{$docel_nsuri}->{''} \|\|
236			$ElementDefault;
237			if ($docel_def->{is_root}) {
238			#
239	wakaba	1.50	} elsif ($docel_def->{is_xml_root}) {
240			unless ($doc->manakai_is_html) {
241			#
242			} else {
243			$onerror->(node => $docel, type => 'element not allowed:root:xml');
244			}
245	wakaba	1.42	} else {
246	wakaba	1.49	$onerror->(node => $docel, type => 'element not allowed:root');
247	wakaba	1.1	}
248
249	wakaba	1.42	## TODO: Check for other items other than document element
250			## (second (errorous) element, text nodes, PI nodes, doctype nodes)
251	wakaba	1.2
252	wakaba	1.51	my $return = $self->check_element ($docel, $onerror);
253
254			my $charset_name = $doc->input_encoding;
255			if (defined $charset_name) {
256			require Message::Charset::Info;
257			my $charset = $Message::Charset::Info::IANACharset->{$charset_name};
258
259			if ($doc->manakai_is_html and
260			not $doc->manakai_has_bom and
261			not defined $doc->manakai_charset) {
262			unless ($charset->{is_html_ascii_superset}) {
263			$onerror->(node => $doc, level => $self->{must_level},
264			type => 'non ascii superset:'.$charset_name);
265			}
266
267			if (not $self->{has_charset} and
268			not $charset->{iana_names}->{'us-ascii'}) {
269			$onerror->(node => $doc, level => $self->{must_level},
270			type => 'no character encoding declaration:'.$charset_name);
271			}
272			}
273
274			if ($charset->{iana_names}->{'utf-8'}) {
275			#
276			} elsif ($charset->{iana_names}->{'jis_x0212-1990'} or
277			$charset->{iana_names}->{'x-jis0208'} or
278			$charset->{iana_names}->{'utf-32'} or ## ISSUE: UTF-32BE? UTF-32LE?
279			$charset->{is_ebcdic_based}) {
280			$onerror->(node => $doc,
281			type => 'character encoding:'.$charset_name,
282			level => $self->{should_level});
283			} elsif ($charset->{iana_names}->{'cesu-8'} or
284			$charset->{iana_names}->{'utf-8'} or ## ISSUE: UNICODE-1-1-UTF-7?
285			$charset->{iana_names}->{'bocu-1'} or
286			$charset->{iana_names}->{'scsu'}) {
287			$onerror->(node => $doc,
288			type => 'character encoding:'.$charset_name,
289			level => $self->{must_level});
290			} else {
291			$onerror->(node => $doc,
292			type => 'character encoding:'.$charset_name,
293			level => $self->{good_level});
294			}
295			}
296
297			return $return;
298	wakaba	1.42	} # check_document
299	wakaba	1.1
300	wakaba	1.42	sub check_element ($$$) {
301			my ($self, $el, $onerror) = @_;
302			$self = bless {}, $self unless ref $self;
303			$self->{onerror} = $onerror;
304	wakaba	1.2
305	wakaba	1.48	$self->{must_level} = 'm';
306			$self->{fact_level} = 'f';
307			$self->{should_level} = 's';
308	wakaba	1.51	$self->{good_level} = 'w';
309	wakaba	1.48
310	wakaba	1.50	$self->{pluses} = {};
311	wakaba	1.42	$self->{minuses} = {};
312			$self->{id} = {};
313			$self->{term} = {};
314			$self->{usemap} = [];
315			$self->{contextmenu} = [];
316			$self->{map} = {};
317			$self->{menu} = {};
318			$self->{has_link_type} = {};
319	wakaba	1.46	#$self->{has_uri_attr};
320			#$self->{has_hyperlink_element};
321	wakaba	1.51	#$self->{has_charset};
322	wakaba	1.42	$self->{return} = {
323			class => {},
324			id => $self->{id}, table => [], term => $self->{term},
325			};
326	wakaba	1.4
327	wakaba	1.42	my @todo = ({type => 'element', node => $el});
328			while (@todo) {
329			my $todo = shift @todo;
330			if ($todo->{type} eq 'element') {
331			my $prefix = $todo->{node}->prefix;
332			if (defined $prefix and $prefix eq 'xmlns') {
333			$self->{onerror}
334			->(node => $todo->{node}, level => 'NC',
335			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
336	wakaba	1.7	}
337	wakaba	1.42	my $nsuri = $todo->{node}->namespace_uri;
338			$nsuri = '' unless defined $nsuri;
339			unless ($Namespace->{$nsuri}->{loaded}) {
340			if ($Namespace->{$nsuri}->{module}) {
341			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
342			} else {
343			$Namespace->{$nsuri}->{loaded} = 1;
344	wakaba	1.1	}
345			}
346	wakaba	1.42	my $ln = $todo->{node}->manakai_local_name;
347			my $eldef = $Element->{$nsuri}->{$ln} \|\|
348			$Element->{$nsuri}->{''} \|\|
349			$ElementDefault;
350			$eldef->{attrs_checker}->($self, $todo);
351			my ($new_todos) = $eldef->{checker}->($self, $todo);
352			unshift @todo, @$new_todos;
353			} elsif ($todo->{type} eq 'element-attributes') {
354			my $prefix = $todo->{node}->prefix;
355			if (defined $prefix and $prefix eq 'xmlns') {
356			$self->{onerror}
357			->(node => $todo->{node}, level => 'NC',
358			type => 'Reserved Prefixes and Namespace Names:<xmlns:>');
359			}
360			my $nsuri = $todo->{node}->namespace_uri;
361			$nsuri = '' unless defined $nsuri;
362			unless ($Namespace->{$nsuri}->{loaded}) {
363			if ($Namespace->{$nsuri}->{module}) {
364			eval qq{ require $Namespace->{$nsuri}->{module} } or die $@;
365	wakaba	1.1	} else {
366	wakaba	1.42	$Namespace->{$nsuri}->{loaded} = 1;
367	wakaba	1.1	}
368			}
369	wakaba	1.9	my $ln = $todo->{node}->manakai_local_name;
370			my $eldef = $Element->{$nsuri}->{$ln} \|\|
371			$Element->{$nsuri}->{''} \|\|
372			$ElementDefault;
373			$eldef->{attrs_checker}->($self, $todo);
374	wakaba	1.50	} elsif ($todo->{type} eq 'plus' or $todo->{type} eq 'minus') {
375	wakaba	1.4	$self->_remove_minuses ($todo);
376	wakaba	1.30	} elsif ($todo->{type} eq 'code') {
377			$todo->{code}->();
378			} else {
379			die "$0: Internal error: Unsupported checking action type \|$todo->{type}\|";
380	wakaba	1.4	}
381	wakaba	1.1	}
382	wakaba	1.17
383			for (@{$self->{usemap}}) {
384			unless ($self->{map}->{$_->[0]}) {
385			$self->{onerror}->(node => $_->[1], type => 'no referenced map');
386			}
387			}
388
389	wakaba	1.32	for (@{$self->{contextmenu}}) {
390			unless ($self->{menu}->{$_->[0]}) {
391			$self->{onerror}->(node => $_->[1], type => 'no referenced menu');
392			}
393			}
394
395	wakaba	1.50	delete $self->{pluses};
396	wakaba	1.17	delete $self->{minuses};
397			delete $self->{onerror};
398			delete $self->{id};
399			delete $self->{usemap};
400			delete $self->{map};
401	wakaba	1.33	return $self->{return};
402	wakaba	1.1	} # check_element
403
404	wakaba	1.2	sub _add_minuses ($@) {
405			my $self = shift;
406			my $r = {};
407			for my $list (@_) {
408			for my $ns (keys %$list) {
409			for my $ln (keys %{$list->{$ns}}) {
410			unless ($self->{minuses}->{$ns}->{$ln}) {
411			$self->{minuses}->{$ns}->{$ln} = 1;
412			$r->{$ns}->{$ln} = 1;
413			}
414			}
415			}
416			}
417	wakaba	1.4	return {type => 'plus', list => $r};
418	wakaba	1.2	} # _add_minuses
419
420	wakaba	1.50	sub _add_pluses ($@) {
421			my $self = shift;
422			my $r = {};
423			for my $list (@_) {
424			for my $ns (keys %$list) {
425			for my $ln (keys %{$list->{$ns}}) {
426			unless ($self->{pluses}->{$ns}->{$ln}) {
427			$self->{pluses}->{$ns}->{$ln} = 1;
428			$r->{$ns}->{$ln} = 1;
429			}
430			}
431			}
432			}
433			return {type => 'minus', list => $r};
434			} # _add_pluses
435
436	wakaba	1.2	sub _remove_minuses ($$) {
437	wakaba	1.4	my ($self, $todo) = @_;
438	wakaba	1.50	if ($todo->{type} eq 'minus') {
439			for my $ns (keys %{$todo->{list}}) {
440			for my $ln (keys %{$todo->{list}->{$ns}}) {
441			delete $self->{pluses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
442			}
443	wakaba	1.2	}
444	wakaba	1.50	} elsif ($todo->{type} eq 'plus') {
445			for my $ns (keys %{$todo->{list}}) {
446			for my $ln (keys %{$todo->{list}->{$ns}}) {
447			delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
448			}
449			}
450			} else {
451			die "$0: Unknown +- type: $todo->{type}";
452	wakaba	1.2	}
453			1;
454			} # _remove_minuses
455
456	wakaba	1.50	## NOTE: Priority for "minuses" and "pluses" are currently left
457			## undefined and implemented inconsistently; it is not a problem for
458			## now, since no element belongs to both lists.
459
460	wakaba	1.30	sub _check_get_children ($$$) {
461			my ($self, $node, $parent_todo) = @_;
462	wakaba	1.4	my $new_todos = [];
463	wakaba	1.2	my $sib = [];
464			TP: {
465			my $node_ns = $node->namespace_uri;
466			$node_ns = '' unless defined $node_ns;
467			my $node_ln = $node->manakai_local_name;
468	wakaba	1.45	if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
469			if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') {
470			if ($parent_todo->{flag}->{in_head}) {
471			#
472			} else {
473			my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
474			push @$sib, $end;
475
476			unshift @$sib, @{$node->child_nodes};
477			push @$new_todos, {type => 'element-attributes', node => $node};
478			last TP;
479			}
480			} else {
481			unshift @$sib, @{$node->child_nodes};
482			push @$new_todos, {type => 'element-attributes', node => $node};
483			last TP;
484	wakaba	1.2	}
485			}
486	wakaba	1.8	if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) {
487	wakaba	1.2	if ($node->has_attribute_ns (undef, 'src')) {
488			unshift @$sib, @{$node->child_nodes};
489	wakaba	1.9	push @$new_todos, {type => 'element-attributes', node => $node};
490	wakaba	1.2	last TP;
491			} else {
492			my @cn = @{$node->child_nodes};
493			CN: while (@cn) {
494			my $cn = shift @cn;
495			my $cnt = $cn->node_type;
496			if ($cnt == 1) {
497	wakaba	1.8	my $cn_nsuri = $cn->namespace_uri;
498			$cn_nsuri = '' unless defined $cn_nsuri;
499			if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') {
500	wakaba	1.2	#
501			} else {
502			last CN;
503			}
504			} elsif ($cnt == 3 or $cnt == 4) {
505			if ($cn->data =~ /[^\x09-\x0D\x20]/) {
506			last CN;
507			}
508			}
509			} # CN
510			unshift @$sib, @cn;
511			}
512			}
513	wakaba	1.4	push @$new_todos, {type => 'element', node => $node};
514	wakaba	1.2	} # TP
515	wakaba	1.30
516			for my $new_todo (@$new_todos) {
517			$new_todo->{flag} = {%{$parent_todo->{flag} or {}}};
518			}
519
520	wakaba	1.4	return ($sib, $new_todos);
521	wakaba	1.2	} # _check_get_children
522
523	wakaba	1.44	=head1 LICENSE
524
525			Copyright 2007 Wakaba <w@suika.fam.cx>
526
527			This library is free software; you can redistribute it
528			and/or modify it under the same terms as Perl itself.
529
530			=cut
531
532	wakaba	1.1	1;
533	wakaba	1.51	# $Date: 2007/10/14 09:21:46 $