| 1 |
package Whatpm::ContentChecker; |
package Whatpm::ContentChecker; |
| 2 |
use strict; |
use strict; |
| 3 |
|
our $VERSION=do{my @r=(q$Revision$=~/\d+/g);sprintf "%d."."%02d" x $#r,@r}; |
| 4 |
|
|
| 5 |
require Whatpm::URIChecker; |
require Whatpm::URIChecker; |
| 6 |
|
|
| 7 |
## ISSUE: How XML and XML Namespaces conformance can (or cannot) |
## ISSUE: How XML and XML Namespaces conformance can (or cannot) |
| 8 |
## be applied to an in-memory representation (i.e. DOM)? |
## be applied to an in-memory representation (i.e. DOM)? |
| 9 |
|
|
| 10 |
|
## TODO: Conformance of an HTML document with non-html root element. |
| 11 |
|
|
| 12 |
|
## Stability |
| 13 |
|
sub FEATURE_STATUS_REC () { 0b1 } ## Interoperable standard |
| 14 |
|
sub FEATURE_STATUS_CR () { 0b10 } ## Call for implementation |
| 15 |
|
sub FEATURE_STATUS_LC () { 0b100 } ## Last call for comments |
| 16 |
|
sub FEATURE_STATUS_WD () { 0b1000 } ## Working or editor's draft |
| 17 |
|
|
| 18 |
|
## Deprecated |
| 19 |
|
sub FEATURE_DEPRECATED_SHOULD () { 0b100000 } ## SHOULD-level |
| 20 |
|
sub FEATURE_DEPRECATED_INFO () { 0b1000000 } ## Does not affect conformance |
| 21 |
|
|
| 22 |
|
## Conformance |
| 23 |
|
sub FEATURE_ALLOWED () { 0b10000 } |
| 24 |
|
|
| 25 |
|
my $HTML_NS = q<http://www.w3.org/1999/xhtml>; |
| 26 |
my $XML_NS = q<http://www.w3.org/XML/1998/namespace>; |
my $XML_NS = q<http://www.w3.org/XML/1998/namespace>; |
| 27 |
my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>; |
my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>; |
| 28 |
|
|
| 29 |
my $AttrChecker = { |
my $Namespace = { |
| 30 |
|
'' => {loaded => 1}, |
| 31 |
|
q<http://www.w3.org/2005/Atom> => {module => 'Whatpm::ContentChecker::Atom'}, |
| 32 |
|
q<http://purl.org/syndication/history/1.0> |
| 33 |
|
=> {module => 'Whatpm::ContentChecker::Atom'}, |
| 34 |
|
q<http://purl.org/syndication/threading/1.0> |
| 35 |
|
=> {module => 'Whatpm::ContentChecker::Atom'}, |
| 36 |
|
$HTML_NS => {module => 'Whatpm::ContentChecker::HTML'}, |
| 37 |
|
$XML_NS => {loaded => 1}, |
| 38 |
|
$XMLNS_NS => {loaded => 1}, |
| 39 |
|
q<http://www.w3.org/1999/02/22-rdf-syntax-ns#> => {loaded => 1}, |
| 40 |
|
}; |
| 41 |
|
|
| 42 |
|
sub load_ns_module ($) { |
| 43 |
|
my $nsuri = shift; # namespace URI or '' |
| 44 |
|
unless ($Namespace->{$nsuri}->{loaded}) { |
| 45 |
|
if ($Namespace->{$nsuri}->{module}) { |
| 46 |
|
eval qq{ require $Namespace->{$nsuri}->{module} } or die $@; |
| 47 |
|
} else { |
| 48 |
|
$Namespace->{$nsuri}->{loaded} = 1; |
| 49 |
|
} |
| 50 |
|
} |
| 51 |
|
} # load_ns_module |
| 52 |
|
|
| 53 |
|
our $AttrChecker = { |
| 54 |
$XML_NS => { |
$XML_NS => { |
| 55 |
space => sub { |
space => sub { |
| 56 |
my ($self, $attr) = @_; |
my ($self, $attr) = @_; |
| 59 |
# |
# |
| 60 |
} else { |
} else { |
| 61 |
## NOTE: An XML "error" |
## NOTE: An XML "error" |
| 62 |
$self->{onerror}->(node => $attr, level => 'error', |
$self->{onerror}->(node => $attr, level => $self->{level}->{xml_error}, |
| 63 |
type => 'invalid attribute value'); |
type => 'invalid attribute value'); |
| 64 |
} |
} |
| 65 |
}, |
}, |
| 66 |
lang => sub { |
lang => sub { |
| 67 |
my ($self, $attr) = @_; |
my ($self, $attr) = @_; |
| 68 |
|
my $value = $attr->value; |
| 69 |
|
if ($value eq '') { |
| 70 |
|
# |
| 71 |
|
} else { |
| 72 |
|
require Whatpm::LangTag; |
| 73 |
|
Whatpm::LangTag->check_rfc3066_language_tag ($value, sub { |
| 74 |
|
$self->{onerror}->(@_, node => $attr); |
| 75 |
|
}, $self->{level}); |
| 76 |
|
} |
| 77 |
|
|
| 78 |
## NOTE: "The values of the attribute are language identifiers |
## NOTE: "The values of the attribute are language identifiers |
| 79 |
## as defined by [IETF RFC 3066], Tags for the Identification |
## as defined by [IETF RFC 3066], Tags for the Identification |
| 80 |
## of Languages, or its successor; in addition, the empty string |
## of Languages, or its successor; in addition, the empty string |
| 81 |
## may be specified." ("may" in lower case) |
## may be specified." ("may" in lower case) |
| 82 |
$self->{onerror}->(node => $attr, level => 'unsupported', |
## NOTE: Is an RFC 3066-valid (but RFC 4646-invalid) language tag |
| 83 |
type => 'language tag'); |
## allowed today? |
| 84 |
|
|
| 85 |
|
## TODO: test data |
| 86 |
|
|
| 87 |
|
my $nsuri = $attr->owner_element->namespace_uri; |
| 88 |
|
if (defined $nsuri and $nsuri eq $HTML_NS) { |
| 89 |
|
my $lang_attr = $attr->owner_element->get_attribute_node_ns |
| 90 |
|
(undef, 'lang'); |
| 91 |
|
if ($lang_attr) { |
| 92 |
|
my $lang_attr_value = $lang_attr->value; |
| 93 |
|
$lang_attr_value =~ tr/A-Z/a-z/; ## ASCII case-insensitive |
| 94 |
|
my $value = $value; |
| 95 |
|
$value =~ tr/A-Z/a-z/; ## ASCII case-insensitive |
| 96 |
|
if ($lang_attr_value ne $value) { |
| 97 |
|
## NOTE: HTML5 Section "The |lang| and |xml:lang| attributes" |
| 98 |
|
$self->{onerror}->(node => $attr, |
| 99 |
|
type => 'xml:lang ne lang', |
| 100 |
|
level => $self->{level}->{must}); |
| 101 |
|
} |
| 102 |
|
} |
| 103 |
|
} |
| 104 |
|
|
| 105 |
if ($attr->owner_document->manakai_is_html) { # MUST NOT |
if ($attr->owner_document->manakai_is_html) { # MUST NOT |
| 106 |
$self->{onerror}->(node => $attr, type => 'in HTML:xml:lang'); |
$self->{onerror}->(node => $attr, type => 'in HTML:xml:lang', |
| 107 |
|
level => $self->{level}->{must}); |
| 108 |
## TODO: Test data... |
## TODO: Test data... |
| 109 |
} |
} |
| 110 |
}, |
}, |
| 113 |
my $value = $attr->value; |
my $value = $attr->value; |
| 114 |
if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters? |
if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters? |
| 115 |
$self->{onerror}->(node => $attr, |
$self->{onerror}->(node => $attr, |
| 116 |
type => 'invalid attribute value'); |
type => 'invalid attribute value', |
| 117 |
|
level => $self->{level}->{fact}, ## TODO: correct? |
| 118 |
|
); |
| 119 |
} |
} |
| 120 |
## NOTE: Conformance to URI standard is not checked since there is |
## NOTE: Conformance to URI standard is not checked since there is |
| 121 |
## no author requirement on conformance in the XML Base specification. |
## no author requirement on conformance in the XML Base specification. |
| 128 |
$value =~ s/\x20$//; |
$value =~ s/\x20$//; |
| 129 |
## TODO: NCName in XML 1.0 or 1.1 |
## TODO: NCName in XML 1.0 or 1.1 |
| 130 |
## TODO: declared type is ID? |
## TODO: declared type is ID? |
| 131 |
if ($self->{id}->{$value}) { ## NOTE: An xml:id error |
if ($self->{id}->{$value}) { |
| 132 |
$self->{onerror}->(node => $attr, level => 'error', |
$self->{onerror}->(node => $attr, |
| 133 |
type => 'duplicate ID'); |
type => 'duplicate ID', |
| 134 |
|
level => $self->{level}->{xml_id_error}); |
| 135 |
|
push @{$self->{id}->{$value}}, $attr; |
| 136 |
} else { |
} else { |
| 137 |
$self->{id}->{$value} = 1; |
$self->{id}->{$value} = [$attr]; |
| 138 |
} |
} |
| 139 |
}, |
}, |
| 140 |
}, |
}, |
| 145 |
my $value = $attr->value; |
my $value = $attr->value; |
| 146 |
if ($value eq $XML_NS and $ln ne 'xml') { |
if ($value eq $XML_NS and $ln ne 'xml') { |
| 147 |
$self->{onerror} |
$self->{onerror} |
| 148 |
->(node => $attr, level => 'NC', |
->(node => $attr, |
| 149 |
type => 'Reserved Prefixes and Namespace Names:=xml'); |
type => 'Reserved Prefixes and Namespace Names:Name', |
| 150 |
|
text => $value, |
| 151 |
|
level => $self->{level}->{nc}); |
| 152 |
} elsif ($value eq $XMLNS_NS) { |
} elsif ($value eq $XMLNS_NS) { |
| 153 |
$self->{onerror} |
$self->{onerror} |
| 154 |
->(node => $attr, level => 'NC', |
->(node => $attr, |
| 155 |
type => 'Reserved Prefixes and Namespace Names:=xmlns'); |
type => 'Reserved Prefixes and Namespace Names:Name', |
| 156 |
|
text => $value, |
| 157 |
|
level => $self->{level}->{nc}); |
| 158 |
} |
} |
| 159 |
if ($ln eq 'xml' and $value ne $XML_NS) { |
if ($ln eq 'xml' and $value ne $XML_NS) { |
| 160 |
$self->{onerror} |
$self->{onerror} |
| 161 |
->(node => $attr, level => 'NC', |
->(node => $attr, |
| 162 |
type => 'Reserved Prefixes and Namespace Names:xmlns:xml='); |
type => 'Reserved Prefixes and Namespace Names:Prefix', |
| 163 |
|
text => $ln, |
| 164 |
|
level => $self->{level}->{nc}); |
| 165 |
} elsif ($ln eq 'xmlns') { |
} elsif ($ln eq 'xmlns') { |
| 166 |
$self->{onerror} |
$self->{onerror} |
| 167 |
->(node => $attr, level => 'NC', |
->(node => $attr, |
| 168 |
type => 'Reserved Prefixes and Namespace Names:xmlns:xmlns='); |
type => 'Reserved Prefixes and Namespace Names:Prefix', |
| 169 |
|
text => $ln, |
| 170 |
|
level => $self->{level}->{nc}); |
| 171 |
} |
} |
| 172 |
## TODO: If XML 1.0 and empty |
## TODO: If XML 1.0 and empty |
| 173 |
}, |
}, |
| 179 |
my $value = $attr->value; |
my $value = $attr->value; |
| 180 |
if ($value eq $XML_NS) { |
if ($value eq $XML_NS) { |
| 181 |
$self->{onerror} |
$self->{onerror} |
| 182 |
->(node => $attr, level => 'NC', |
->(node => $attr, |
| 183 |
type => 'Reserved Prefixes and Namespace Names:=xml'); |
type => 'Reserved Prefixes and Namespace Names:Name', |
| 184 |
|
text => $value, |
| 185 |
|
level => $self->{level}->{nc}); |
| 186 |
} elsif ($value eq $XMLNS_NS) { |
} elsif ($value eq $XMLNS_NS) { |
| 187 |
$self->{onerror} |
$self->{onerror} |
| 188 |
->(node => $attr, level => 'NC', |
->(node => $attr, |
| 189 |
type => 'Reserved Prefixes and Namespace Names:=xmlns'); |
type => 'Reserved Prefixes and Namespace Names:Name', |
| 190 |
|
text => $value, |
| 191 |
|
level => $self->{level}->{nc}); |
| 192 |
} |
} |
| 193 |
}, |
}, |
| 194 |
}, |
}, |
| 197 |
## ISSUE: Should we really allow these attributes? |
## ISSUE: Should we really allow these attributes? |
| 198 |
$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space}; |
$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space}; |
| 199 |
$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang}; |
$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang}; |
| 200 |
|
## NOTE: Checker for (null, "xml:lang") attribute is shadowed for |
| 201 |
|
## HTML elements in Whatpm::ContentChecker::HTML. |
| 202 |
$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base}; |
$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base}; |
| 203 |
$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id}; |
$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id}; |
| 204 |
|
|
| 205 |
## ANY |
our $AttrStatus; |
|
my $AnyChecker = sub { |
|
|
my ($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 |
|
|
|
|
|
my $ElementDefault = { |
|
|
checker => sub { |
|
|
my ($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 $Element = {}; |
|
|
|
|
|
my $HTML_NS = q<http://www.w3.org/1999/xhtml>; |
|
|
|
|
|
my $HTMLMetadataElements = { |
|
|
$HTML_NS => { |
|
|
qw/link 1 meta 1 style 1 script 1 event-source 1 command 1 base 1 title 1 |
|
|
noscript 1 |
|
|
/, |
|
|
}, |
|
|
}; |
|
|
|
|
|
my $HTMLSectioningElements = { |
|
|
$HTML_NS => {qw/body 1 section 1 nav 1 article 1 blockquote 1 aside 1/}, |
|
|
}; |
|
|
|
|
|
my $HTMLBlockLevelElements = { |
|
|
$HTML_NS => { |
|
|
qw/ |
|
|
section 1 nav 1 article 1 blockquote 1 aside 1 |
|
|
h1 1 h2 1 h3 1 h4 1 h5 1 h6 1 header 1 footer 1 |
|
|
address 1 p 1 hr 1 dialog 1 pre 1 ol 1 ul 1 dl 1 |
|
|
ins 1 del 1 figure 1 map 1 table 1 script 1 noscript 1 |
|
|
event-source 1 details 1 datagrid 1 menu 1 div 1 font 1 |
|
|
/, |
|
|
}, |
|
|
}; |
|
|
|
|
|
my $HTMLStrictlyInlineLevelElements = { |
|
|
$HTML_NS => { |
|
|
qw/ |
|
|
br 1 a 1 q 1 cite 1 em 1 strong 1 small 1 m 1 dfn 1 abbr 1 |
|
|
time 1 meter 1 progress 1 code 1 var 1 samp 1 kbd 1 |
|
|
sub 1 sup 1 span 1 i 1 b 1 bdo 1 ins 1 del 1 img 1 |
|
|
iframe 1 embed 1 object 1 video 1 audio 1 canvas 1 area 1 |
|
|
script 1 noscript 1 event-source 1 command 1 font 1 |
|
|
/, |
|
|
}, |
|
|
}; |
|
|
|
|
|
my $HTMLStructuredInlineLevelElements = { |
|
|
$HTML_NS => {qw/blockquote 1 pre 1 ol 1 ul 1 dl 1 table 1 menu 1/}, |
|
|
}; |
|
|
|
|
|
my $HTMLInteractiveElements = { |
|
|
$HTML_NS => {a => 1, details => 1, datagrid => 1}, |
|
|
}; |
|
|
## NOTE: |html:a| and |html:datagrid| are not allowed as a descendant |
|
|
## of interactive elements |
|
|
|
|
|
my $HTMLTransparentElements = { |
|
|
$HTML_NS => {qw/ins 1 font 1 noscript 1/}, |
|
|
## NOTE: |html:noscript| is transparent if scripting is disabled |
|
|
## and not in |head|. |
|
|
}; |
|
|
|
|
|
#my $HTMLSemiTransparentElements = { |
|
|
# $HTML_NS => {qw/video 1 audio 1/}, |
|
|
#}; |
|
|
|
|
|
my $HTMLEmbededElements = { |
|
|
$HTML_NS => {qw/img 1 iframe 1 embed 1 object 1 video 1 audio 1 canvas 1/}, |
|
|
}; |
|
|
|
|
|
## Empty |
|
|
my $HTMLEmptyChecker = sub { |
|
|
my ($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) { |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}; |
|
|
|
|
|
## Text |
|
|
my $HTMLTextChecker = sub { |
|
|
my ($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) { |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
$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 ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}; |
|
|
|
|
|
## Zero or more |html:style| elements, |
|
|
## followed by zero or more block-level elements |
|
|
my $HTMLStylableBlockChecker = sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $has_non_style; |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'style') { |
|
|
$not_allowed = 1 if $has_non_style or |
|
|
not $node->has_attribute_ns (undef, 'scoped'); |
|
|
} elsif ($HTMLBlockLevelElements->{$node_ns}->{$node_ln}) { |
|
|
$has_non_style = 1; |
|
|
} else { |
|
|
$has_non_style = 1; |
|
|
$not_allowed = 1; |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $not_allowed; |
|
|
my ($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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}; # $HTMLStylableBlockChecker |
|
|
|
|
|
## Zero or more block-level elements |
|
|
my $HTMLBlockChecker = sub { |
|
|
my ($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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
$not_allowed = 1 |
|
|
unless $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $not_allowed; |
|
|
my ($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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}; # $HTMLBlockChecker |
|
|
|
|
|
## Inline-level content |
|
|
my $HTMLInlineChecker = sub { |
|
|
my ($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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} or |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $not_allowed; |
|
|
my ($sib, $ch) = $self->_check_get_children ($node, $todo); |
|
|
unshift @nodes, @$sib; |
|
|
push @$new_todos, @$ch; |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
} |
|
|
return ($new_todos); |
|
|
}; # $HTMLInlineChecker |
|
|
|
|
|
my $HTMLSignificantInlineChecker = $HTMLInlineChecker; |
|
|
## TODO: check significant content |
|
|
|
|
|
## Strictly inline-level content |
|
|
my $HTMLStrictlyInlineChecker = sub { |
|
|
my ($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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $not_allowed; |
|
|
my ($sib, $ch) = $self->_check_get_children ($node, $todo); |
|
|
unshift @nodes, @$sib; |
|
|
push @$new_todos, @$ch; |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
$_->{strictly_inline} = 1; |
|
|
} |
|
|
return ($new_todos); |
|
|
}; # $HTMLStrictlyInlineChecker |
|
|
|
|
|
my $HTMLSignificantStrictlyInlineChecker = $HTMLStrictlyInlineChecker; |
|
|
## TODO: check significant content |
|
|
|
|
|
## Inline-level or strictly inline-kevek content |
|
|
my $HTMLInlineOrStrictlyInlineChecker = sub { |
|
|
my ($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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($todo->{strictly_inline}) { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
} else { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} or |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $not_allowed; |
|
|
my ($sib, $ch) = $self->_check_get_children ($node, $todo); |
|
|
unshift @nodes, @$sib; |
|
|
push @$new_todos, @$ch; |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
$_->{strictly_inline} = 1; |
|
|
} |
|
|
return ($new_todos); |
|
|
}; # $HTMLInlineOrStrictlyInlineChecker |
|
|
|
|
|
my $HTMLSignificantInlineOrStrictlyInlineChecker |
|
|
= $HTMLInlineOrStrictlyInlineChecker; |
|
|
## TODO: check significant content |
|
|
|
|
|
my $HTMLBlockOrInlineChecker = sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $content = 'block-or-inline'; # or 'block' or 'inline' |
|
|
my @block_not_inline; |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($content eq 'block') { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
} elsif ($content eq 'inline') { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} or |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
} else { |
|
|
my $is_block = $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
my $is_inline |
|
|
= $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} || |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
|
|
|
push @block_not_inline, $node |
|
|
if $is_block and not $is_inline and not $not_allowed; |
|
|
unless ($is_block) { |
|
|
$content = 'inline'; |
|
|
for (@block_not_inline) { |
|
|
$self->{onerror}->(node => $_, type => 'element not allowed'); |
|
|
} |
|
|
$not_allowed = 1 unless $is_inline; |
|
|
} |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $not_allowed; |
|
|
my ($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]/) { |
|
|
if ($content eq 'block') { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} else { |
|
|
$content = 'inline'; |
|
|
for (@block_not_inline) { |
|
|
$self->{onerror}->(node => $_, type => 'element not allowed'); |
|
|
} |
|
|
} |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
if ($content eq 'inline') { |
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}; |
|
|
|
|
|
## Zero or more XXX element, then either block-level or inline-level |
|
|
my $GetHTMLZeroOrMoreThenBlockOrInlineChecker = sub ($$) { |
|
|
my ($elnsuri, $ellname) = @_; |
|
|
return sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $has_non_style; |
|
|
my $content = 'block-or-inline'; # or 'block' or 'inline' |
|
|
my @block_not_inline; |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($node_ns eq $elnsuri and $node_ln eq $ellname) { |
|
|
$not_allowed = 1 if $has_non_style; |
|
|
if ($ellname eq 'style' and |
|
|
not $node->has_attribute_ns (undef, 'scoped')) { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
} elsif ($content eq 'block') { |
|
|
$has_non_style = 1; |
|
|
$not_allowed = 1 |
|
|
unless $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
} elsif ($content eq 'inline') { |
|
|
$has_non_style = 1; |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} or |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
} else { |
|
|
$has_non_style = 1; |
|
|
my $is_block = $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
my $is_inline |
|
|
= $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} || |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
|
|
|
push @block_not_inline, $node |
|
|
if $is_block and not $is_inline and not $not_allowed; |
|
|
unless ($is_block) { |
|
|
$content = 'inline'; |
|
|
for (@block_not_inline) { |
|
|
$self->{onerror}->(node => $_, type => 'element not allowed'); |
|
|
} |
|
|
$not_allowed = 1 unless $is_inline; |
|
|
} |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $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]/) { |
|
|
$has_non_style = 1; |
|
|
if ($content eq 'block') { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} else { |
|
|
$content = 'inline'; |
|
|
for (@block_not_inline) { |
|
|
$self->{onerror}->(node => $_, type => 'element not allowed'); |
|
|
} |
|
|
} |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
if ($content eq 'inline') { |
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}; |
|
|
}; # $GetHTMLZeroOrMoreThenBlockOrInlineChecker |
|
|
|
|
|
my $HTMLTransparentChecker = $HTMLBlockOrInlineChecker; |
|
|
|
|
|
my $GetHTMLEnumeratedAttrChecker = sub { |
|
|
my $states = shift; # {value => conforming ? 1 : -1} |
|
|
return sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = lc $attr->value; ## TODO: ASCII case insensitibility? |
|
|
if ($states->{$value} > 0) { |
|
|
# |
|
|
} elsif ($states->{$value}) { |
|
|
$self->{onerror}->(node => $attr, type => 'enumerated:non-conforming'); |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => 'enumerated:invalid'); |
|
|
} |
|
|
}; |
|
|
}; # $GetHTMLEnumeratedAttrChecker |
|
|
|
|
|
my $GetHTMLBooleanAttrChecker = sub { |
|
|
my $local_name = shift; |
|
|
return sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
unless ($value eq $local_name or $value eq '') { |
|
|
$self->{onerror}->(node => $attr, type => 'boolean:invalid'); |
|
|
} |
|
|
}; |
|
|
}; # $GetHTMLBooleanAttrChecker |
|
|
|
|
|
my $HTMLUnorderedSetOfSpaceSeparatedTokensAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
my %word; |
|
|
for my $word (grep {length $_} split /[\x09-\x0D\x20]/, $attr->value) { |
|
|
unless ($word{$word}) { |
|
|
$word{$word} = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => 'duplicate token:'.$word); |
|
|
} |
|
|
} |
|
|
}; # $HTMLUnorderedSetOfSpaceSeparatedTokensAttrChecker |
|
|
|
|
|
## |rel| attribute (unordered set of space separated tokens, |
|
|
## whose allowed values are defined by the section on link types) |
|
|
my $HTMLLinkTypesAttrChecker = sub { |
|
|
my ($a_or_area, $self, $attr) = @_; |
|
|
my %word; |
|
|
for my $word (grep {length $_} split /[\x09-\x0D\x20]/, $attr->value) { |
|
|
unless ($word{$word}) { |
|
|
$word{$word} = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => 'duplicate token:'.$word); |
|
|
} |
|
|
} |
|
|
## NOTE: Case sensitive match (since HTML5 spec does not say link |
|
|
## types are case-insensitive and it says "The value should not |
|
|
## be confusingly similar to any other defined value (e.g. |
|
|
## differing only in case)."). |
|
|
## NOTE: Though there is no explicit "MUST NOT" for undefined values, |
|
|
## "MAY"s and "only ... MAY" restrict non-standard non-registered |
|
|
## values to be used conformingly. |
|
|
require Whatpm::_LinkTypeList; |
|
|
our $LinkType; |
|
|
for my $word (keys %word) { |
|
|
my $def = $LinkType->{$word}; |
|
|
if (defined $def) { |
|
|
if ($def->{status} eq 'accepted') { |
|
|
if (defined $def->{effect}->[$a_or_area]) { |
|
|
# |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'link type:bad context:'.$word); |
|
|
} |
|
|
} elsif ($def->{status} eq 'proposal') { |
|
|
$self->{onerror}->(node => $attr, level => 's', |
|
|
type => 'link type:proposed:'.$word); |
|
|
} else { # rejected or synonym |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'link type:non-conforming:'.$word); |
|
|
} |
|
|
if ($def->{unique}) { |
|
|
unless ($self->{has_link_type}->{$word}) { |
|
|
$self->{has_link_type}->{$word} = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'link type:duplicate:'.$word); |
|
|
} |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'link type:'.$word); |
|
|
} |
|
|
} |
|
|
## TODO: The Pingback 1.0 specification, which is referenced by HTML5, |
|
|
## says that using both X-Pingback: header field and HTML |
|
|
## <link rel=pingback> is deprecated and if both appears they |
|
|
## SHOULD contain exactly the same value. |
|
|
## ISSUE: Pingback 1.0 specification defines the exact representation |
|
|
## of its link element, which cannot be tested by the current arch. |
|
|
## ISSUE: Pingback 1.0 specification says that the document MUST NOT |
|
|
## include any string that matches to the pattern for the rel=pingback link, |
|
|
## which again inpossible to test. |
|
|
## ISSUE: rel=pingback href MUST NOT include entities other than predefined 4. |
|
|
}; # $HTMLLinkTypesAttrChecker |
|
|
|
|
|
## URI (or IRI) |
|
|
my $HTMLURIAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
## ISSUE: Relative references are allowed? (RFC 3987 "IRI" is an absolute reference with optional fragment identifier.) |
|
|
my $value = $attr->value; |
|
|
Whatpm::URIChecker->check_iri_reference ($value, sub { |
|
|
my %opt = @_; |
|
|
$self->{onerror}->(node => $attr, level => $opt{level}, |
|
|
type => 'URI::'.$opt{type}. |
|
|
(defined $opt{position} ? ':'.$opt{position} : '')); |
|
|
}); |
|
|
}; # $HTMLURIAttrChecker |
|
|
|
|
|
## A space separated list of one or more URIs (or IRIs) |
|
|
my $HTMLSpaceURIsAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $i = 0; |
|
|
for my $value (split /[\x09-\x0D\x20]+/, $attr->value) { |
|
|
Whatpm::URIChecker->check_iri_reference ($value, sub { |
|
|
my %opt = @_; |
|
|
$self->{onerror}->(node => $attr, level => $opt{level}, |
|
|
type => 'URIs:'.$i.'::'. |
|
|
$opt{type}. |
|
|
(defined $opt{position} ? ':'.$opt{position} : '')); |
|
|
}); |
|
|
$i++; |
|
|
} |
|
|
## ISSUE: Relative references? |
|
|
## ISSUE: Leading or trailing white spaces are conformant? |
|
|
## ISSUE: A sequence of white space characters are conformant? |
|
|
## ISSUE: A zero-length string is conformant? (It does contain a relative reference, i.e. same as base URI.) |
|
|
## NOTE: Duplication seems not an error. |
|
|
}; # $HTMLSpaceURIsAttrChecker |
|
|
|
|
|
my $HTMLDatetimeAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
## ISSUE: "space", not "space character" (in parsing algorihtm, "space character") |
|
|
if ($value =~ /\A([0-9]{4})-([0-9]{2})-([0-9]{2})(?>[\x09-\x0D\x20]+(?>T[\x09-\x0D\x20]*)?|T[\x09-\x0D\x20]*)([0-9]{2}):([0-9]{2})(?>:([0-9]{2}))?(?>\.([0-9]+))?[\x09-\x0D\x20]*(?>Z|[+-]([0-9]{2}):([0-9]{2}))\z/) { |
|
|
my ($y, $M, $d, $h, $m, $s, $f, $zh, $zm) |
|
|
= ($1, $2, $3, $4, $5, $6, $7, $8, $9); |
|
|
if (0 < $M and $M < 13) { ## ISSUE: This is not explicitly specified (though in parsing algorithm) |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad day') |
|
|
if $d < 1 or |
|
|
$d > [0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]->[$M]; |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad day') |
|
|
if $M == 2 and $d == 29 and |
|
|
not ($y % 400 == 0 or ($y % 4 == 0 and $y % 100 != 0)); |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad month'); |
|
|
} |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad hour') if $h > 23; |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad minute') if $m > 59; |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad second') |
|
|
if defined $s and $s > 59; |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad timezone hour') |
|
|
if $zh > 23; |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:bad timezone minute') |
|
|
if $zm > 59; |
|
|
## ISSUE: Maybe timezone -00:00 should have same semantics as in RFC 3339. |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => 'datetime:syntax error'); |
|
|
} |
|
|
}; # $HTMLDatetimeAttrChecker |
|
|
|
|
|
my $HTMLIntegerAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
unless ($value =~ /\A-?[0-9]+\z/) { |
|
|
$self->{onerror}->(node => $attr, type => 'integer:syntax error'); |
|
|
} |
|
|
}; # $HTMLIntegerAttrChecker |
|
|
|
|
|
my $GetHTMLNonNegativeIntegerAttrChecker = sub { |
|
|
my $range_check = shift; |
|
|
return sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
if ($value =~ /\A[0-9]+\z/) { |
|
|
unless ($range_check->($value + 0)) { |
|
|
$self->{onerror}->(node => $attr, type => 'nninteger:out of range'); |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'nninteger:syntax error'); |
|
|
} |
|
|
}; |
|
|
}; # $GetHTMLNonNegativeIntegerAttrChecker |
|
|
|
|
|
my $GetHTMLFloatingPointNumberAttrChecker = sub { |
|
|
my $range_check = shift; |
|
|
return sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
if ($value =~ /\A-?[0-9.]+\z/ and $value =~ /[0-9]/) { |
|
|
unless ($range_check->($value + 0)) { |
|
|
$self->{onerror}->(node => $attr, type => 'float:out of range'); |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'float:syntax error'); |
|
|
} |
|
|
}; |
|
|
}; # $GetHTMLFloatingPointNumberAttrChecker |
|
|
|
|
|
## "A valid MIME type, optionally with parameters. [RFC 2046]" |
|
|
## ISSUE: RFC 2046 does not define syntax of media types. |
|
|
## ISSUE: The definition of "a valid MIME type" is unknown. |
|
|
## Syntactical correctness? |
|
|
my $HTMLIMTAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
## ISSUE: RFC 2045 Content-Type header field allows insertion |
|
|
## of LWS/comments between tokens. Is it allowed in HTML? Maybe no. |
|
|
## ISSUE: RFC 2231 extension? Maybe no. |
|
|
my $lws0 = qr/(?>(?>\x0D\x0A)?[\x09\x20])*/; |
|
|
my $token = qr/[\x21\x23-\x27\x2A\x2B\x2D\x2E\x30-\x39\x41-\x5A\x5E-\x7E]+/; |
|
|
my $qs = qr/"(?>[\x00-\x0C\x0E-\x21\x23-\x5B\x5D-\x7E]|\x0D\x0A[\x09\x20]|\x5C[\x00-\x7F])*"/; |
|
|
if ($value =~ m#\A$lws0($token)$lws0/$lws0($token)$lws0((?>;$lws0$token$lws0=$lws0(?>$token|$qs)$lws0)*)\z#) { |
|
|
my @type = ($1, $2); |
|
|
my $param = $3; |
|
|
while ($param =~ s/^;$lws0($token)$lws0=$lws0(?>($token)|($qs))$lws0//) { |
|
|
if (defined $2) { |
|
|
push @type, $1 => $2; |
|
|
} else { |
|
|
my $n = $1; |
|
|
my $v = $2; |
|
|
$v =~ s/\\(.)/$1/gs; |
|
|
push @type, $n => $v; |
|
|
} |
|
|
} |
|
|
require Whatpm::IMTChecker; |
|
|
Whatpm::IMTChecker->check_imt (sub { |
|
|
my %opt = @_; |
|
|
$self->{onerror}->(node => $attr, level => $opt{level}, |
|
|
type => 'IMT:'.$opt{type}); |
|
|
}, @type); |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => 'IMT:syntax error'); |
|
|
} |
|
|
}; # $HTMLIMTAttrChecker |
|
|
|
|
|
my $HTMLLanguageTagAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'language tag'); |
|
|
if ($attr->value eq '') { |
|
|
$self->{onerror}->(node => $attr, type => 'language tag:syntax error'); |
|
|
} |
|
|
## TODO: RFC 3066 test |
|
|
## ISSUE: RFC 4646 (3066bis)? |
|
|
}; # $HTMLLanguageTagAttrChecker |
|
|
|
|
|
## "A valid media query [MQ]" |
|
|
my $HTMLMQAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'media query'); |
|
|
## ISSUE: What is "a valid media query"? |
|
|
}; # $HTMLMQAttrChecker |
|
|
|
|
|
my $HTMLEventHandlerAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'event handler'); |
|
|
## TODO: MUST contain valid ECMAScript code matching the |
|
|
## ECMAScript |FunctionBody| production. [ECMA262] |
|
|
## ISSUE: MUST be ES3? E4X? ES4? JS1.x? |
|
|
## ISSUE: Automatic semicolon insertion does not apply? |
|
|
## ISSUE: Other script languages? |
|
|
}; # $HTMLEventHandlerAttrChecker |
|
|
|
|
|
my $HTMLUsemapAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
## MUST be a valid hashed ID reference to a |map| element |
|
|
my $value = $attr->value; |
|
|
if ($value =~ s/^#//) { |
|
|
## ISSUE: Is |usemap="#"| conformant? (c.f. |id=""| is non-conformant.) |
|
|
push @{$self->{usemap}}, [$value => $attr]; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, type => '#idref:syntax error'); |
|
|
} |
|
|
## NOTE: Space characters in hashed ID references are conforming. |
|
|
## ISSUE: UA algorithm for matching is case-insensitive; IDs only different in cases should be reported |
|
|
}; # $HTMLUsemapAttrChecker |
|
|
|
|
|
my $HTMLTargetAttrChecker = sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
if ($value =~ /^_/) { |
|
|
$value = lc $value; ## ISSUE: ASCII case-insentitive? |
|
|
unless ({ |
|
|
_self => 1, _parent => 1, _top => 1, |
|
|
}->{$value}) { |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'reserved browsing context name'); |
|
|
} |
|
|
} else { |
|
|
#$ ISSUE: An empty string is conforming? |
|
|
} |
|
|
}; # $HTMLTargetAttrChecker |
|
| 206 |
|
|
| 207 |
my $HTMLAttrChecker = { |
for (qw/space lang base id/) { |
| 208 |
id => sub { |
$AttrStatus->{$XML_NS}->{$_} = FEATURE_STATUS_REC | FEATURE_ALLOWED; |
| 209 |
## NOTE: |map| has its own variant of |id=""| checker |
$AttrStatus->{''}->{"xml:$_"} = FEATURE_STATUS_REC | FEATURE_ALLOWED; |
| 210 |
my ($self, $attr) = @_; |
## XML 1.0: FEATURE_STATUS_CR |
| 211 |
my $value = $attr->value; |
## XML 1.1: FEATURE_STATUS_REC |
| 212 |
if (length $value > 0) { |
## XML Namespaces 1.0: FEATURE_STATUS_CR |
| 213 |
if ($self->{id}->{$value}) { |
## XML Namespaces 1.1: FEATURE_STATUS_REC |
| 214 |
$self->{onerror}->(node => $attr, type => 'duplicate ID'); |
## XML Base: FEATURE_STATUS_REC |
| 215 |
} else { |
## xml:id: FEATURE_STATUS_REC |
|
$self->{id}->{$value} = 1; |
|
|
} |
|
|
if ($value =~ /[\x09-\x0D\x20]/) { |
|
|
$self->{onerror}->(node => $attr, type => 'space in ID'); |
|
|
} |
|
|
} else { |
|
|
## NOTE: MUST contain at least one character |
|
|
$self->{onerror}->(node => $attr, type => 'empty attribute value'); |
|
|
} |
|
|
}, |
|
|
title => sub {}, ## NOTE: No conformance creteria |
|
|
lang => sub { |
|
|
my ($self, $attr) = @_; |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'language tag'); |
|
|
## TODO: RFC 3066 or empty test |
|
|
## ISSUE: RFC 4646 (3066bis)? |
|
|
unless ($attr->owner_document->manakai_is_html) { |
|
|
$self->{onerror}->(node => $attr, type => 'in XML:lang'); |
|
|
} |
|
|
}, |
|
|
dir => $GetHTMLEnumeratedAttrChecker->({ltr => 1, rtl => 1}), |
|
|
class => $HTMLUnorderedSetOfSpaceSeparatedTokensAttrChecker, |
|
|
contextmenu => sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
push @{$self->{contextmenu}}, [$value => $attr]; |
|
|
## ISSUE: "The value must be the ID of a menu element in the DOM." |
|
|
## What is "in the DOM"? A menu Element node that is not part |
|
|
## of the Document tree is in the DOM? A menu Element node that |
|
|
## belong to another Document tree is in the DOM? |
|
|
}, |
|
|
irrelevant => $GetHTMLBooleanAttrChecker->('irrelevant'), |
|
|
tabindex => $HTMLIntegerAttrChecker, |
|
|
}; |
|
|
|
|
|
for (qw/ |
|
|
onabort onbeforeunload onblur onchange onclick oncontextmenu |
|
|
ondblclick ondrag ondragend ondragenter ondragleave ondragover |
|
|
ondragstart ondrop onerror onfocus onkeydown onkeypress |
|
|
onkeyup onload onmessage onmousedown onmousemove onmouseout |
|
|
onmouseover onmouseup onmousewheel onresize onscroll onselect |
|
|
onsubmit onunload |
|
|
/) { |
|
|
$HTMLAttrChecker->{$_} = $HTMLEventHandlerAttrChecker; |
|
| 216 |
} |
} |
| 217 |
|
|
| 218 |
my $GetHTMLAttrsChecker = sub { |
$AttrStatus->{$XMLNS_NS}->{''} = FEATURE_STATUS_REC | FEATURE_ALLOWED; |
|
my $element_specific_checker = shift; |
|
|
return 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; |
|
|
if ($attr_ns eq '') { |
|
|
$checker = $element_specific_checker->{$attr_ln} |
|
|
|| $HTMLAttrChecker->{$attr_ln}; |
|
|
} |
|
|
$checker ||= $AttrChecker->{$attr_ns}->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{''}; |
|
|
if ($checker) { |
|
|
$checker->($self, $attr, $todo); |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'attribute'); |
|
|
## ISSUE: No comformance createria for unknown attributes in the spec |
|
|
} |
|
|
} |
|
|
}; |
|
|
}; # $GetHTMLAttrsChecker |
|
| 219 |
|
|
| 220 |
$Element->{$HTML_NS}->{''} = { |
## TODO: xsi:schemaLocation for XHTML2 support (very, very low priority) |
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $ElementDefault->{checker}, |
|
|
}; |
|
| 221 |
|
|
| 222 |
$Element->{$HTML_NS}->{html} = { |
our %AnyChecker = ( |
| 223 |
is_root => 1, |
check_start => sub { }, |
| 224 |
attrs_checker => $GetHTMLAttrsChecker->({ |
check_attrs => sub { |
| 225 |
xmlns => sub { |
my ($self, $item, $element_state) = @_; |
| 226 |
my ($self, $attr) = @_; |
for my $attr (@{$item->{node}->attributes}) { |
|
my $value = $attr->value; |
|
|
unless ($value eq $HTML_NS) { |
|
|
$self->{onerror}->(node => $attr, type => 'invalid attribute value'); |
|
|
} |
|
|
unless ($attr->owner_document->manakai_is_html) { |
|
|
$self->{onerror}->(node => $attr, type => 'in XML:xmlns'); |
|
|
## TODO: Test |
|
|
} |
|
|
}, |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $phase = 'before head'; |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($phase eq 'before head') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'head') { |
|
|
$phase = 'after head'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'body') { |
|
|
$self->{onerror}->(node => $node, type => 'ps element missing:head'); |
|
|
$phase = 'after body'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
# before head |
|
|
} |
|
|
} elsif ($phase eq 'after head') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'body') { |
|
|
$phase = 'after body'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
# after head |
|
|
} |
|
|
} else { #elsif ($phase eq 'after body') { |
|
|
$not_allowed = 1; |
|
|
# after body |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
if ($phase eq 'before head') { |
|
|
$self->{onerror}->(node => $el, type => 'child element missing:head'); |
|
|
$self->{onerror}->(node => $el, type => 'child element missing:body'); |
|
|
} elsif ($phase eq 'after head') { |
|
|
$self->{onerror}->(node => $el, type => 'child element missing:body'); |
|
|
} |
|
|
|
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{head} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $has_title; |
|
|
my $phase = 'initial'; # 'after charset', 'after base' |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'title') { |
|
|
$phase = 'after base'; |
|
|
unless ($has_title) { |
|
|
$has_title = 1; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'meta') { |
|
|
if ($node->has_attribute_ns (undef, 'charset')) { |
|
|
if ($phase eq 'initial') { |
|
|
$phase = 'after charset'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
## NOTE: See also |base|'s "contexts" field in the spec |
|
|
} |
|
|
} else { |
|
|
$phase = 'after base'; |
|
|
} |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'base') { |
|
|
if ($phase eq 'initial' or $phase eq 'after charset') { |
|
|
$phase = 'after base'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'style') { |
|
|
$phase = 'after base'; |
|
|
if ($node->has_attribute_ns (undef, 'scoped')) { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
} elsif ($HTMLMetadataElements->{$node_ns}->{$node_ln}) { |
|
|
$phase = 'after base'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
unless ($has_title) { |
|
|
$self->{onerror}->(node => $el, type => 'child element missing:title'); |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{title} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLTextChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{base} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
href => $HTMLURIAttrChecker, |
|
|
target => $HTMLTargetAttrChecker, |
|
|
}), |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{link} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({ |
|
|
href => $HTMLURIAttrChecker, |
|
|
rel => sub { $HTMLLinkTypesAttrChecker->(0, @_) }, |
|
|
media => $HTMLMQAttrChecker, |
|
|
hreflang => $HTMLLanguageTagAttrChecker, |
|
|
type => $HTMLIMTAttrChecker, |
|
|
## NOTE: Though |title| has special semantics, |
|
|
## syntactically same as the |title| as global attribute. |
|
|
})->($self, $todo); |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'href')) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:href'); |
|
|
} |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'rel')) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:rel'); |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{meta} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $name_attr; |
|
|
my $http_equiv_attr; |
|
|
my $charset_attr; |
|
|
my $content_attr; |
|
|
for my $attr (@{$todo->{node}->attributes}) { |
|
| 227 |
my $attr_ns = $attr->namespace_uri; |
my $attr_ns = $attr->namespace_uri; |
| 228 |
$attr_ns = '' unless defined $attr_ns; |
if (defined $attr_ns) { |
| 229 |
my $attr_ln = $attr->manakai_local_name; |
load_ns_module ($attr_ns); |
|
my $checker; |
|
|
if ($attr_ns eq '') { |
|
|
if ($attr_ln eq 'content') { |
|
|
$content_attr = $attr; |
|
|
$checker = 1; |
|
|
} elsif ($attr_ln eq 'name') { |
|
|
$name_attr = $attr; |
|
|
$checker = 1; |
|
|
} elsif ($attr_ln eq 'http-equiv') { |
|
|
$http_equiv_attr = $attr; |
|
|
$checker = 1; |
|
|
} elsif ($attr_ln eq 'charset') { |
|
|
$charset_attr = $attr; |
|
|
$checker = 1; |
|
|
} else { |
|
|
$checker = $HTMLAttrChecker->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{''}; |
|
|
} |
|
|
} else { |
|
|
$checker ||= $AttrChecker->{$attr_ns}->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{''}; |
|
|
} |
|
|
if ($checker) { |
|
|
$checker->($self, $attr) if ref $checker; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'attribute'); |
|
|
## ISSUE: No comformance createria for unknown attributes in the spec |
|
|
} |
|
|
} |
|
|
|
|
|
if (defined $name_attr) { |
|
|
if (defined $http_equiv_attr) { |
|
|
$self->{onerror}->(node => $http_equiv_attr, |
|
|
type => 'attribute not allowed'); |
|
|
} elsif (defined $charset_attr) { |
|
|
$self->{onerror}->(node => $charset_attr, |
|
|
type => 'attribute not allowed'); |
|
|
} |
|
|
my $metadata_name = $name_attr->value; |
|
|
my $metadata_value; |
|
|
if (defined $content_attr) { |
|
|
$metadata_value = $content_attr->value; |
|
|
} else { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:content'); |
|
|
$metadata_value = ''; |
|
|
} |
|
|
} elsif (defined $http_equiv_attr) { |
|
|
if (defined $charset_attr) { |
|
|
$self->{onerror}->(node => $charset_attr, |
|
|
type => 'attribute not allowed'); |
|
|
} |
|
|
unless (defined $content_attr) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:content'); |
|
|
} |
|
|
} elsif (defined $charset_attr) { |
|
|
if (defined $content_attr) { |
|
|
$self->{onerror}->(node => $content_attr, |
|
|
type => 'attribute not allowed'); |
|
|
} |
|
|
} else { |
|
|
if (defined $content_attr) { |
|
|
$self->{onerror}->(node => $content_attr, |
|
|
type => 'attribute not allowed'); |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:name|http-equiv'); |
|
|
} else { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:name|http-equiv|charset'); |
|
|
} |
|
|
} |
|
|
|
|
|
## TODO: metadata conformance |
|
|
|
|
|
## TODO: pragma conformance |
|
|
if (defined $http_equiv_attr) { ## An enumerated attribute |
|
|
my $keyword = lc $http_equiv_attr->value; ## TODO: ascii case? |
|
|
if ({ |
|
|
'refresh' => 1, |
|
|
'default-style' => 1, |
|
|
}->{$keyword}) { |
|
|
# |
|
| 230 |
} else { |
} else { |
| 231 |
$self->{onerror}->(node => $http_equiv_attr, |
$attr_ns = ''; |
|
type => 'enumerated:invalid'); |
|
|
} |
|
|
} |
|
|
|
|
|
if (defined $charset_attr) { |
|
|
unless ($todo->{node}->owner_document->manakai_is_html) { |
|
|
$self->{onerror}->(node => $charset_attr, |
|
|
type => 'in XML:charset'); |
|
|
} |
|
|
## TODO: charset |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{style} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
type => $HTMLIMTAttrChecker, ## TODO: MUST be a styling language |
|
|
media => $HTMLMQAttrChecker, |
|
|
scoped => $GetHTMLBooleanAttrChecker->('scoped'), |
|
|
## NOTE: |title| has special semantics for |style|s, but is syntactically |
|
|
## not different |
|
|
}), |
|
|
checker => sub { |
|
|
## NOTE: |html:style| has no conformance creteria on content model |
|
|
my ($self, $todo) = @_; |
|
|
my $type = $todo->{node}->get_attribute_ns (undef, 'type'); |
|
|
$type = 'text/css' unless defined $type; |
|
|
$self->{onerror}->(node => $todo->{node}, level => 'unsupported', |
|
|
type => 'style:'.$type); ## TODO: $type normalization |
|
|
return $AnyChecker->($self, $todo); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{body} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLBlockChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{section} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStylableBlockChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{nav} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLBlockOrInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{article} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStylableBlockChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{blockquote} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
cite => $HTMLURIAttrChecker, |
|
|
}), |
|
|
checker => $HTMLBlockChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{aside} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $GetHTMLZeroOrMoreThenBlockOrInlineChecker->($HTML_NS, 'style'), |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{h1} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$todo->{flag}->{has_heading}->[0] = 1; |
|
|
return $HTMLSignificantStrictlyInlineChecker->($self, $todo); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{h2} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{h1}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{h3} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{h1}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{h4} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{h1}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{h5} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{h1}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{h6} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{h1}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{header} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $old_flag = $todo->{flag}->{has_heading} || []; |
|
|
my $new_flag = []; |
|
|
local $todo->{flag}->{has_heading} = $new_flag; |
|
|
my $node = $todo->{node}; |
|
|
|
|
|
my $end = $self->_add_minuses |
|
|
({$HTML_NS => {qw/header 1 footer 1/}}, |
|
|
$HTMLSectioningElements); |
|
|
my ($new_todos, $ch) = $HTMLBlockChecker->($self, $todo); |
|
|
push @$new_todos, $end, |
|
|
{type => 'code', code => sub { |
|
|
if ($new_flag->[0]) { |
|
|
$old_flag->[0] = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element missing:hn'); |
|
|
} |
|
|
}}; |
|
|
return ($new_todos, $ch); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{footer} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { ## block -hn -header -footer -sectioning or inline |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $content = 'block-or-inline'; # or 'block' or 'inline' |
|
|
my @block_not_inline; |
|
|
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; |
|
|
my $not_allowed; |
|
|
if ($self->{minuses}->{$node_ns}->{$node_ln}) { |
|
|
$not_allowed = 1; |
|
|
} elsif ($node_ns eq $HTML_NS and |
|
|
{ |
|
|
qw/h1 1 h2 1 h3 1 h4 1 h5 1 h6 1 header 1 footer 1/ |
|
|
}->{$node_ln}) { |
|
|
$not_allowed = 1; |
|
|
} elsif ($HTMLSectioningElements->{$node_ns}->{$node_ln}) { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
if ($content eq 'block') { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
} elsif ($content eq 'inline') { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} or |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
} else { |
|
|
my $is_block = $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
my $is_inline |
|
|
= $HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} || |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}; |
|
|
|
|
|
push @block_not_inline, $node |
|
|
if $is_block and not $is_inline and not $not_allowed; |
|
|
unless ($is_block) { |
|
|
$content = 'inline'; |
|
|
for (@block_not_inline) { |
|
|
$self->{onerror}->(node => $_, type => 'element not allowed'); |
|
|
} |
|
|
$not_allowed = 1 unless $is_inline; |
|
|
} |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $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]/) { |
|
|
if ($content eq 'block') { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} else { |
|
|
$content = 'inline'; |
|
|
for (@block_not_inline) { |
|
|
$self->{onerror}->(node => $_, type => 'element not allowed'); |
|
|
} |
|
|
} |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
my $end = $self->_add_minuses |
|
|
({$HTML_NS => {qw/h1 1 h2 1 h3 1 h4 1 h5 1 h6 1/}}, |
|
|
$HTMLSectioningElements); |
|
|
push @$new_todos, $end; |
|
|
|
|
|
if ($content eq 'inline') { |
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
| 232 |
} |
} |
|
} |
|
|
|
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{address} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{p} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLSignificantInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{hr} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{br} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{dialog} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $phase = 'before dt'; |
|
|
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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
if ($phase eq 'before dt') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'dt') { |
|
|
$phase = 'before dd'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'dd') { |
|
|
$self->{onerror} |
|
|
->(node => $node, type => 'ps element missing:dt'); |
|
|
$phase = 'before dt'; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} else { # before dd |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'dd') { |
|
|
$phase = 'before dt'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'dt') { |
|
|
$self->{onerror} |
|
|
->(node => $node, type => 'ps element missing:dd'); |
|
|
$phase = 'before dd'; |
|
|
} else { |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
if ($phase eq 'before dd') { |
|
|
$self->{onerror}->(node => $el, type => 'ps element missing:dd'); |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{pre} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{ol} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
start => $HTMLIntegerAttrChecker, |
|
|
}), |
|
|
checker => 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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
unless ($node_ns eq $HTML_NS and $node_ln eq 'li') { |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
if ($todo->{inline}) { |
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{ul} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{ol}->{checker}, |
|
|
}; |
|
|
|
|
|
|
|
|
$Element->{$HTML_NS}->{li} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
start => sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $parent = $attr->owner_element->manakai_parent_element; |
|
|
if (defined $parent) { |
|
|
my $parent_ns = $parent->namespace_uri; |
|
|
$parent_ns = '' unless defined $parent_ns; |
|
|
my $parent_ln = $parent->manakai_local_name; |
|
|
unless ($parent_ns eq $HTML_NS and $parent_ln eq 'ol') { |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'attribute'); |
|
|
} |
|
|
} |
|
|
$HTMLIntegerAttrChecker->($self, $attr); |
|
|
}, |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
if ($todo->{inline}) { |
|
|
return $HTMLInlineChecker->($self, $todo); |
|
|
} else { |
|
|
return $HTMLBlockOrInlineChecker->($self, $todo); |
|
|
} |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{dl} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $phase = 'before dt'; |
|
|
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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
if ($phase eq 'in dds') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'dd') { |
|
|
#$phase = 'in dds'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'dt') { |
|
|
$phase = 'in dts'; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} elsif ($phase eq 'in dts') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'dt') { |
|
|
#$phase = 'in dts'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'dd') { |
|
|
$phase = 'in dds'; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} else { # before dt |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'dt') { |
|
|
$phase = 'in dts'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'dd') { |
|
|
$self->{onerror} |
|
|
->(node => $node, type => 'ps element missing:dt'); |
|
|
$phase = 'in dds'; |
|
|
} else { |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
if ($phase eq 'in dts') { |
|
|
$self->{onerror}->(node => $el, type => 'ps element missing:dd'); |
|
|
} |
|
|
|
|
|
if ($todo->{inline}) { |
|
|
for (@$new_todos) { |
|
|
$_->{inline} = 1; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{dt} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{dd} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{li}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{a} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my %attr; |
|
|
for my $attr (@{$todo->{node}->attributes}) { |
|
|
my $attr_ns = $attr->namespace_uri; |
|
|
$attr_ns = '' unless defined $attr_ns; |
|
| 233 |
my $attr_ln = $attr->manakai_local_name; |
my $attr_ln = $attr->manakai_local_name; |
| 234 |
my $checker; |
|
| 235 |
if ($attr_ns eq '') { |
my $checker = $AttrChecker->{$attr_ns}->{$attr_ln} |
| 236 |
$checker = { |
|| $AttrChecker->{$attr_ns}->{''}; |
| 237 |
target => $HTMLTargetAttrChecker, |
my $status = $AttrStatus->{$attr_ns}->{$attr_ln} |
| 238 |
href => $HTMLURIAttrChecker, |
|| $AttrStatus->{$attr_ns}->{''}; |
| 239 |
ping => $HTMLSpaceURIsAttrChecker, |
if (not defined $status) { |
| 240 |
rel => sub { $HTMLLinkTypesAttrChecker->(1, @_) }, |
$status = FEATURE_ALLOWED; |
| 241 |
media => $HTMLMQAttrChecker, |
## NOTE: FEATURE_ALLOWED for all attributes, since the element |
| 242 |
hreflang => $HTMLLanguageTagAttrChecker, |
## is not supported and therefore "attribute not defined" error |
| 243 |
type => $HTMLIMTAttrChecker, |
## should not raised (too verbose) and global attributes should be |
| 244 |
}->{$attr_ln}; |
## allowed anyway (if a global attribute has its specified creteria |
| 245 |
if ($checker) { |
## for where it may be specified, then it should be checked in it's |
| 246 |
$attr{$attr_ln} = $attr; |
## checker function). |
|
} else { |
|
|
$checker = $HTMLAttrChecker->{$attr_ln}; |
|
|
} |
|
|
} |
|
|
$checker ||= $AttrChecker->{$attr_ns}->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{''}; |
|
|
if ($checker) { |
|
|
$checker->($self, $attr) if ref $checker; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'attribute'); |
|
|
## ISSUE: No comformance createria for unknown attributes in the spec |
|
|
} |
|
|
} |
|
|
|
|
|
unless (defined $attr{href}) { |
|
|
for (qw/target ping rel media hreflang type/) { |
|
|
if (defined $attr{$_}) { |
|
|
$self->{onerror}->(node => $attr{$_}, |
|
|
type => 'attribute not allowed'); |
|
|
} |
|
|
} |
|
|
} |
|
|
}, |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
|
|
|
my $end = $self->_add_minuses ($HTMLInteractiveElements); |
|
|
my ($new_todos, $ch) |
|
|
= $HTMLSignificantInlineOrStrictlyInlineChecker->($self, $todo); |
|
|
push @$new_todos, $end; |
|
|
|
|
|
$_->{flag}->{has_a} = 1 for @$new_todos; |
|
|
|
|
|
return ($new_todos, $ch); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{q} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
cite => $HTMLURIAttrChecker, |
|
|
}), |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{cite} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{em} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{strong} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{small} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{m} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{dfn} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
|
|
|
my $end = $self->_add_minuses ({$HTML_NS => {dfn => 1}}); |
|
|
my ($sib, $ch) = $HTMLStrictlyInlineChecker->($self, $todo); |
|
|
push @$sib, $end; |
|
|
|
|
|
my $node = $todo->{node}; |
|
|
my $term = $node->get_attribute_ns (undef, 'title'); |
|
|
unless (defined $term) { |
|
|
for my $child (@{$node->child_nodes}) { |
|
|
if ($child->node_type == 1) { # ELEMENT_NODE |
|
|
if (defined $term) { |
|
|
undef $term; |
|
|
last; |
|
|
} elsif ($child->manakai_local_name eq 'abbr') { |
|
|
my $nsuri = $child->namespace_uri; |
|
|
if (defined $nsuri and $nsuri eq $HTML_NS) { |
|
|
my $attr = $child->get_attribute_node_ns (undef, 'title'); |
|
|
if ($attr) { |
|
|
$term = $attr->value; |
|
|
} |
|
|
} |
|
|
} |
|
|
} elsif ($child->node_type == 3 or $child->node_type == 4) { |
|
|
## TEXT_NODE or CDATA_SECTION_NODE |
|
|
if ($child->data =~ /\A[\x09-\x0D\x20]+\z/) { # Inter-element whitespace |
|
|
next; |
|
|
} |
|
|
undef $term; |
|
|
last; |
|
|
} |
|
|
} |
|
|
unless (defined $term) { |
|
|
$term = $node->text_content; |
|
|
} |
|
|
} |
|
|
if ($self->{term}->{$term}) { |
|
|
$self->{onerror}->(node => $node, type => 'duplicate term'); |
|
|
push @{$self->{term}->{$term}}, $node; |
|
|
} else { |
|
|
$self->{term}->{$term} = [$node]; |
|
|
} |
|
|
|
|
|
return ($sib, $ch); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{abbr} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
## NOTE: |title| has special semantics for |abbr|s, but is syntactically |
|
|
## not different. The spec says that the |title| MAY be omitted |
|
|
## if there is a |dfn| whose defining term is the abbreviation, |
|
|
## but it does not prohibit |abbr| w/o |title| in other cases. |
|
|
}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{time} = { ## TODO: validate content |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), ## TODO: datetime |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{meter} = { ## TODO: "The recommended way of giving the value is to include it as contents of the element" |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
value => $GetHTMLFloatingPointNumberAttrChecker->(sub { 1 }), |
|
|
min => $GetHTMLFloatingPointNumberAttrChecker->(sub { 1 }), |
|
|
low => $GetHTMLFloatingPointNumberAttrChecker->(sub { 1 }), |
|
|
high => $GetHTMLFloatingPointNumberAttrChecker->(sub { 1 }), |
|
|
max => $GetHTMLFloatingPointNumberAttrChecker->(sub { 1 }), |
|
|
optimum => $GetHTMLFloatingPointNumberAttrChecker->(sub { 1 }), |
|
|
}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{progress} = { ## TODO: recommended to use content |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
value => $GetHTMLFloatingPointNumberAttrChecker->(sub { shift >= 0 }), |
|
|
max => $GetHTMLFloatingPointNumberAttrChecker->(sub { shift > 0 }), |
|
|
}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{code} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
## NOTE: Though |title| has special semantics, |
|
|
## syntatically same as the |title| as global attribute. |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{var} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
## NOTE: Though |title| has special semantics, |
|
|
## syntatically same as the |title| as global attribute. |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{samp} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
## NOTE: Though |title| has special semantics, |
|
|
## syntatically same as the |title| as global attribute. |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{kbd} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{sub} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{sup} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{span} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
## NOTE: Though |title| has special semantics, |
|
|
## syntatically same as the |title| as global attribute. |
|
|
checker => $HTMLInlineOrStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{i} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
## NOTE: Though |title| has special semantics, |
|
|
## syntatically same as the |title| as global attribute. |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{b} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{bdo} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({})->($self, $todo); |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'dir')) { |
|
|
$self->{onerror}->(node => $todo->{node}, type => 'attribute missing:dir'); |
|
|
} |
|
|
}, |
|
|
## ISSUE: The spec does not directly say that |dir| is a enumerated attr. |
|
|
checker => $HTMLStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{ins} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
cite => $HTMLURIAttrChecker, |
|
|
datetime => $HTMLDatetimeAttrChecker, |
|
|
}), |
|
|
checker => $HTMLTransparentChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{del} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
cite => $HTMLURIAttrChecker, |
|
|
datetime => $HTMLDatetimeAttrChecker, |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
|
|
|
my $parent = $todo->{node}->manakai_parent_element; |
|
|
if (defined $parent) { |
|
|
my $nsuri = $parent->namespace_uri; |
|
|
$nsuri = '' unless defined $nsuri; |
|
|
my $ln = $parent->manakai_local_name; |
|
|
my $eldef = $Element->{$nsuri}->{$ln} || |
|
|
$Element->{$nsuri}->{''} || |
|
|
$ElementDefault; |
|
|
return $eldef->{checker}->($self, $todo); |
|
|
} else { |
|
|
return $HTMLBlockOrInlineChecker->($self, $todo); |
|
|
} |
|
|
}, |
|
|
}; |
|
|
|
|
|
## TODO: figure |
|
|
|
|
|
$Element->{$HTML_NS}->{img} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({ |
|
|
alt => sub { }, ## NOTE: No syntactical requirement |
|
|
src => $HTMLURIAttrChecker, |
|
|
usemap => $HTMLUsemapAttrChecker, |
|
|
ismap => sub { |
|
|
my ($self, $attr, $parent_todo) = @_; |
|
|
if (not $todo->{flag}->{has_a}) { |
|
|
$self->{onerror}->(node => $attr, type => 'attribute not allowed'); |
|
|
} |
|
|
$GetHTMLBooleanAttrChecker->('ismap')->($self, $attr, $parent_todo); |
|
|
}, |
|
|
## TODO: height |
|
|
## TODO: width |
|
|
})->($self, $todo); |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'alt')) { |
|
|
$self->{onerror}->(node => $todo->{node}, type => 'attribute missing:alt'); |
|
|
} |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'src')) { |
|
|
$self->{onerror}->(node => $todo->{node}, type => 'attribute missing:src'); |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{iframe} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
src => $HTMLURIAttrChecker, |
|
|
}), |
|
|
checker => $HTMLTextChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{embed} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $has_src; |
|
|
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; |
|
|
if ($attr_ns eq '') { |
|
|
if ($attr_ln eq 'src') { |
|
|
$checker = $HTMLURIAttrChecker; |
|
|
$has_src = 1; |
|
|
} elsif ($attr_ln eq 'type') { |
|
|
$checker = $HTMLIMTAttrChecker; |
|
|
} else { |
|
|
## TODO: height |
|
|
## TODO: width |
|
|
$checker = $HTMLAttrChecker->{$attr_ln} |
|
|
|| sub { }; ## NOTE: Any local attribute is ok. |
|
|
} |
|
| 247 |
} |
} |
|
$checker ||= $AttrChecker->{$attr_ns}->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{''}; |
|
| 248 |
if ($checker) { |
if ($checker) { |
| 249 |
$checker->($self, $attr); |
$checker->($self, $attr); |
| 250 |
} else { |
} else { |
| 251 |
$self->{onerror}->(node => $attr, level => 'unsupported', |
$self->{onerror}->(node => $attr, |
| 252 |
type => 'attribute'); |
type => 'unknown attribute', |
| 253 |
## ISSUE: No comformance createria for global attributes in the spec |
level => $self->{level}->{uncertain}); |
|
} |
|
|
} |
|
|
|
|
|
unless ($has_src) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:src'); |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{object} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({ |
|
|
data => $HTMLURIAttrChecker, |
|
|
type => $HTMLIMTAttrChecker, |
|
|
usemap => $HTMLUsemapAttrChecker, |
|
|
## TODO: width |
|
|
## TODO: height |
|
|
})->($self, $todo); |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'data')) { |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'type')) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:data|type'); |
|
|
} |
|
|
} |
|
|
}, |
|
|
checker => $ElementDefault->{checker}, ## TODO |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{param} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({ |
|
|
name => sub { }, |
|
|
value => sub { }, |
|
|
})->($self, $todo); |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'name')) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:name'); |
|
|
} |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'value')) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:value'); |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{video} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
src => $HTMLURIAttrChecker, |
|
|
## TODO: start, loopstart, loopend, end |
|
|
## ISSUE: they MUST be "value time offset"s. Value? |
|
|
## ISSUE: loopcount has no conformance creteria |
|
|
autoplay => $GetHTMLBooleanAttrChecker->('autoplay'), |
|
|
controls => $GetHTMLBooleanAttrChecker->('controls'), |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
|
|
|
if ($todo->{node}->has_attribute_ns (undef, 'src')) { |
|
|
return $HTMLBlockOrInlineChecker->($self, $todo); |
|
|
} else { |
|
|
return $GetHTMLZeroOrMoreThenBlockOrInlineChecker->($HTML_NS, 'source') |
|
|
->($self, $todo); |
|
|
} |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{audio} = { |
|
|
attrs_checker => $Element->{$HTML_NS}->{video}->{attrs_checker}, |
|
|
checker => $Element->{$HTML_NS}->{video}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{source} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({ |
|
|
src => $HTMLURIAttrChecker, |
|
|
type => $HTMLIMTAttrChecker, |
|
|
media => $HTMLMQAttrChecker, |
|
|
})->($self, $todo); |
|
|
unless ($todo->{node}->has_attribute_ns (undef, 'src')) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:src'); |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{canvas} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
height => $GetHTMLNonNegativeIntegerAttrChecker->(sub { 1 }), |
|
|
width => $GetHTMLNonNegativeIntegerAttrChecker->(sub { 1 }), |
|
|
}), |
|
|
checker => $HTMLInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{map} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
id => sub { |
|
|
## NOTE: same as global |id=""|, with |$self->{map}| registeration |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
if (length $value > 0) { |
|
|
if ($self->{id}->{$value}) { |
|
|
$self->{onerror}->(node => $attr, type => 'duplicate ID'); |
|
|
} else { |
|
|
$self->{id}->{$value} = 1; |
|
|
} |
|
|
} else { |
|
|
## NOTE: MUST contain at least one character |
|
|
$self->{onerror}->(node => $attr, type => 'empty attribute value'); |
|
|
} |
|
|
if ($value =~ /[\x09-\x0D\x20]/) { |
|
|
$self->{onerror}->(node => $attr, type => 'space in ID'); |
|
|
} |
|
|
$self->{map}->{$value} ||= $attr; |
|
|
}, |
|
|
}), |
|
|
checker => $HTMLBlockChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{area} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my %attr; |
|
|
my $coords; |
|
|
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; |
|
|
if ($attr_ns eq '') { |
|
|
$checker = { |
|
|
alt => sub { }, |
|
|
## NOTE: |alt| value has no conformance creteria. |
|
|
shape => $GetHTMLEnumeratedAttrChecker->({ |
|
|
circ => -1, circle => 1, |
|
|
default => 1, |
|
|
poly => 1, polygon => -1, |
|
|
rect => 1, rectangle => -1, |
|
|
}), |
|
|
coords => sub { |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
if ($value =~ /\A-?[0-9]+(?>,-?[0-9]+)*\z/) { |
|
|
$coords = [split /,/, $value]; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, |
|
|
type => 'coords:syntax error'); |
|
|
} |
|
|
}, |
|
|
target => $HTMLTargetAttrChecker, |
|
|
href => $HTMLURIAttrChecker, |
|
|
ping => $HTMLSpaceURIsAttrChecker, |
|
|
rel => sub { $HTMLLinkTypesAttrChecker->(1, @_) }, |
|
|
media => $HTMLMQAttrChecker, |
|
|
hreflang => $HTMLLanguageTagAttrChecker, |
|
|
type => $HTMLIMTAttrChecker, |
|
|
}->{$attr_ln}; |
|
|
if ($checker) { |
|
|
$attr{$attr_ln} = $attr; |
|
|
} else { |
|
|
$checker = $HTMLAttrChecker->{$attr_ln}; |
|
|
} |
|
|
} |
|
|
$checker ||= $AttrChecker->{$attr_ns}->{$attr_ln} |
|
|
|| $AttrChecker->{$attr_ns}->{''}; |
|
|
if ($checker) { |
|
|
$checker->($self, $attr) if ref $checker; |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr, level => 'unsupported', |
|
|
type => 'attribute'); |
|
|
## ISSUE: No comformance createria for unknown attributes in the spec |
|
|
} |
|
|
} |
|
|
|
|
|
if (defined $attr{href}) { |
|
|
unless (defined $attr{alt}) { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:alt'); |
|
|
} |
|
|
} else { |
|
|
for (qw/target ping rel media hreflang type alt/) { |
|
|
if (defined $attr{$_}) { |
|
|
$self->{onerror}->(node => $attr{$_}, |
|
|
type => 'attribute not allowed'); |
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
|
my $shape = 'rectangle'; |
|
|
if (defined $attr{shape}) { |
|
|
$shape = { |
|
|
circ => 'circle', circle => 'circle', |
|
|
default => 'default', |
|
|
poly => 'polygon', polygon => 'polygon', |
|
|
rect => 'rectangle', rectangle => 'rectangle', |
|
|
}->{lc $attr{shape}->value} || 'rectangle'; |
|
|
## TODO: ASCII lowercase? |
|
|
} |
|
|
|
|
|
if ($shape eq 'circle') { |
|
|
if (defined $attr{coords}) { |
|
|
if (defined $coords) { |
|
|
if (@$coords == 3) { |
|
|
if ($coords->[2] < 0) { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:out of range:2'); |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:number:3:'.@$coords); |
|
|
} |
|
|
} else { |
|
|
## NOTE: A syntax error has been reported. |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:coords'); |
|
|
} |
|
|
} elsif ($shape eq 'default') { |
|
|
if (defined $attr{coords}) { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'attribute not allowed'); |
|
|
} |
|
|
} elsif ($shape eq 'polygon') { |
|
|
if (defined $attr{coords}) { |
|
|
if (defined $coords) { |
|
|
if (@$coords >= 6) { |
|
|
unless (@$coords % 2 == 0) { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:number:even:'.@$coords); |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:number:>=6:'.@$coords); |
|
|
} |
|
|
} else { |
|
|
## NOTE: A syntax error has been reported. |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:coords'); |
|
|
} |
|
|
} elsif ($shape eq 'rectangle') { |
|
|
if (defined $attr{coords}) { |
|
|
if (defined $coords) { |
|
|
if (@$coords == 4) { |
|
|
unless ($coords->[0] < $coords->[2]) { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:out of range:0'); |
|
|
} |
|
|
unless ($coords->[1] < $coords->[3]) { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:out of range:1'); |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $attr{coords}, |
|
|
type => 'coords:number:4:'.@$coords); |
|
|
} |
|
|
} else { |
|
|
## NOTE: A syntax error has been reported. |
|
|
} |
|
|
} else { |
|
|
$self->{onerror}->(node => $todo->{node}, |
|
|
type => 'attribute missing:coords'); |
|
|
} |
|
|
} |
|
|
}, |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
## TODO: only in map |
|
|
|
|
|
$Element->{$HTML_NS}->{table} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $phase = 'before caption'; |
|
|
my $has_tfoot; |
|
|
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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
if ($phase eq 'in tbodys') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'tbody') { |
|
|
#$phase = 'in tbodys'; |
|
|
} elsif (not $has_tfoot and |
|
|
$node_ns eq $HTML_NS and $node_ln eq 'tfoot') { |
|
|
$phase = 'after tfoot'; |
|
|
$has_tfoot = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} elsif ($phase eq 'in trs') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'tr') { |
|
|
#$phase = 'in trs'; |
|
|
} elsif (not $has_tfoot and |
|
|
$node_ns eq $HTML_NS and $node_ln eq 'tfoot') { |
|
|
$phase = 'after tfoot'; |
|
|
$has_tfoot = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} elsif ($phase eq 'after thead') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'tbody') { |
|
|
$phase = 'in tbodys'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tr') { |
|
|
$phase = 'in trs'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tfoot') { |
|
|
$phase = 'in tbodys'; |
|
|
$has_tfoot = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} elsif ($phase eq 'in colgroup') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'colgroup') { |
|
|
$phase = 'in colgroup'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'thead') { |
|
|
$phase = 'after thead'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tbody') { |
|
|
$phase = 'in tbodys'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tr') { |
|
|
$phase = 'in trs'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tfoot') { |
|
|
$phase = 'in tbodys'; |
|
|
$has_tfoot = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} elsif ($phase eq 'before caption') { |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'caption') { |
|
|
$phase = 'in colgroup'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'colgroup') { |
|
|
$phase = 'in colgroup'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'thead') { |
|
|
$phase = 'after thead'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tbody') { |
|
|
$phase = 'in tbodys'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tr') { |
|
|
$phase = 'in trs'; |
|
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'tfoot') { |
|
|
$phase = 'in tbodys'; |
|
|
$has_tfoot = 1; |
|
|
} else { |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed'); |
|
|
} |
|
|
} else { # after tfoot |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
## Table model errors |
|
|
require Whatpm::HTMLTable; |
|
|
Whatpm::HTMLTable->form_table ($todo->{node}, sub { |
|
|
my %opt = @_; |
|
|
$self->{onerror}->(type => 'table:'.$opt{type}, node => $opt{node}); |
|
|
}); |
|
|
push @{$self->{return}->{table}}, $todo->{node}; |
|
|
|
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{caption} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $HTMLSignificantStrictlyInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{colgroup} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
span => $GetHTMLNonNegativeIntegerAttrChecker->(sub { shift > 0 }), |
|
|
## NOTE: Defined only if "the |colgroup| element contains no |col| elements" |
|
|
## TODO: "attribute not supported" if |col|. |
|
|
## ISSUE: MUST NOT if any |col|? |
|
|
## ISSUE: MUST NOT for |<colgroup span="1"><any><col/></any></colgroup>| (though non-conforming)? |
|
|
}), |
|
|
checker => 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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
unless ($node_ns eq $HTML_NS and $node_ln eq 'col') { |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{col} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
span => $GetHTMLNonNegativeIntegerAttrChecker->(sub { shift > 0 }), |
|
|
}), |
|
|
checker => $HTMLEmptyChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{tbody} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $has_tr; |
|
|
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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'tr') { |
|
|
$has_tr = 1; |
|
|
} else { |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
unless ($has_tr) { |
|
|
$self->{onerror}->(node => $el, type => 'child element missing:tr'); |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{thead} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{tbody}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{tfoot} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $Element->{$HTML_NS}->{tbody}->{checker}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{tr} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $has_td; |
|
|
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; |
|
|
## NOTE: |minuses| list is not checked since redundant |
|
|
if ($node_ns eq $HTML_NS and ($node_ln eq 'td' or $node_ln eq 'th')) { |
|
|
$has_td = 1; |
|
|
} else { |
|
|
$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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
unless ($has_td) { |
|
|
$self->{onerror}->(node => $el, type => 'child element missing:td|th'); |
|
|
} |
|
|
return ($new_todos); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{td} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
colspan => $GetHTMLNonNegativeIntegerAttrChecker->(sub { shift > 0 }), |
|
|
rowspan => $GetHTMLNonNegativeIntegerAttrChecker->(sub { shift > 0 }), |
|
|
}), |
|
|
checker => $HTMLBlockOrInlineChecker, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{th} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
colspan => $GetHTMLNonNegativeIntegerAttrChecker->(sub { shift > 0 }), |
|
|
rowspan => $GetHTMLNonNegativeIntegerAttrChecker->(sub { shift > 0 }), |
|
|
scope => $GetHTMLEnumeratedAttrChecker |
|
|
->({row => 1, col => 1, rowgroup => 1, colgroup => 1}), |
|
|
}), |
|
|
checker => $HTMLBlockOrInlineChecker, |
|
|
}; |
|
|
|
|
|
## TODO: forms |
|
|
|
|
|
$Element->{$HTML_NS}->{script} = { |
|
|
attrs_checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
$GetHTMLAttrsChecker->({ |
|
|
src => $HTMLURIAttrChecker, |
|
|
defer => $GetHTMLBooleanAttrChecker->('defer'), |
|
|
async => $GetHTMLBooleanAttrChecker->('async'), |
|
|
type => $HTMLIMTAttrChecker, |
|
|
})->($self, $todo); |
|
|
if ($todo->{node}->has_attribute_ns (undef, 'defer')) { |
|
|
my $async_attr = $todo->{node}->get_attribute_node_ns (undef, 'async'); |
|
|
if ($async_attr) { |
|
|
$self->{onerror}->(node => $async_attr, |
|
|
type => 'attribute not allowed'); # MUST NOT |
|
| 254 |
} |
} |
| 255 |
|
$self->_attr_status_info ($attr, $status); |
| 256 |
} |
} |
| 257 |
}, |
}, |
| 258 |
checker => sub { |
check_child_element => sub { |
| 259 |
my ($self, $todo) = @_; |
my ($self, $item, $child_el, $child_nsuri, $child_ln, |
| 260 |
|
$child_is_transparent, $element_state) = @_; |
| 261 |
if ($todo->{node}->has_attribute_ns (undef, 'src')) { |
if ($self->{minus_elements}->{$child_nsuri}->{$child_ln}) { |
| 262 |
return $HTMLEmptyChecker->($self, $todo); |
$self->{onerror}->(node => $child_el, |
| 263 |
|
type => 'element not allowed:minus', |
| 264 |
|
level => $self->{level}->{must}); |
| 265 |
|
} elsif ($self->{plus_elements}->{$child_nsuri}->{$child_ln}) { |
| 266 |
|
# |
| 267 |
} else { |
} else { |
| 268 |
## NOTE: No content model conformance in HTML5 spec. |
# |
|
my $type = $todo->{node}->get_attribute_ns (undef, 'type'); |
|
|
my $language = $todo->{node}->get_attribute_ns (undef, 'language'); |
|
|
if ((defined $type and $type eq '') or |
|
|
(defined $language and $language eq '')) { |
|
|
$type = 'text/javascript'; |
|
|
} elsif (defined $type) { |
|
|
# |
|
|
} elsif (defined $language) { |
|
|
$type = 'text/' . $language; |
|
|
} else { |
|
|
$type = 'text/javascript'; |
|
|
} |
|
|
$self->{onerror}->(node => $todo->{node}, level => 'unsupported', |
|
|
type => 'script:'.$type); ## TODO: $type normalization |
|
|
return $AnyChecker->($self, $todo); |
|
| 269 |
} |
} |
| 270 |
}, |
}, |
| 271 |
}; |
check_child_text => sub { }, |
| 272 |
|
check_end => sub { |
| 273 |
## NOTE: When script is disabled. |
my ($self, $item, $element_state) = @_; |
| 274 |
$Element->{$HTML_NS}->{noscript} = { |
## NOTE: There is a modified copy of the code below for |html:ruby|. |
| 275 |
attrs_checker => $GetHTMLAttrsChecker->({}), |
if ($element_state->{has_significant}) { |
| 276 |
checker => sub { |
$item->{real_parent_state}->{has_significant} = 1; |
| 277 |
my ($self, $todo) = @_; |
} |
| 278 |
|
}, |
| 279 |
my $end = $self->_add_minuses ({$HTML_NS => {noscript => 1}}); |
); |
| 280 |
my ($sib, $ch) = $HTMLBlockOrInlineChecker->($self, $todo); |
|
| 281 |
push @$sib, $end; |
our $ElementDefault = { |
| 282 |
return ($sib, $ch); |
%AnyChecker, |
| 283 |
|
status => FEATURE_ALLOWED, |
| 284 |
|
## NOTE: No "element not defined" error - it is not supported anyway. |
| 285 |
|
check_start => sub { |
| 286 |
|
my ($self, $item, $element_state) = @_; |
| 287 |
|
$self->{onerror}->(node => $item->{node}, |
| 288 |
|
type => 'unknown element', |
| 289 |
|
level => $self->{level}->{uncertain}); |
| 290 |
}, |
}, |
| 291 |
}; |
}; |
|
## TODO: noscript in head |
|
|
## TODO: noscript in XHTML |
|
| 292 |
|
|
| 293 |
$Element->{$HTML_NS}->{'event-source'} = { |
our $HTMLEmbeddedContent = { |
| 294 |
attrs_checker => $GetHTMLAttrsChecker->({ |
## NOTE: All embedded content is also phrasing content. |
| 295 |
src => $HTMLURIAttrChecker, |
$HTML_NS => { |
| 296 |
}), |
img => 1, iframe => 1, embed => 1, object => 1, video => 1, audio => 1, |
| 297 |
checker => $HTMLEmptyChecker, |
canvas => 1, |
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{details} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
open => $GetHTMLBooleanAttrChecker->('open'), |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
|
|
|
my $end = $self->_add_minuses ({$HTML_NS => {a => 1, datagrid => 1}}); |
|
|
my ($sib, $ch) |
|
|
= $GetHTMLZeroOrMoreThenBlockOrInlineChecker->($HTML_NS, 'legend') |
|
|
->($self, $todo); |
|
|
push @$sib, $end; |
|
|
return ($sib, $ch); |
|
|
}, |
|
|
}; |
|
|
|
|
|
$Element->{$HTML_NS}->{datagrid} = { |
|
|
attrs_checker => $GetHTMLAttrsChecker->({ |
|
|
disabled => $GetHTMLBooleanAttrChecker->('disabled'), |
|
|
multiple => $GetHTMLBooleanAttrChecker->('multiple'), |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $end = $self->_add_minuses ({$HTML_NS => {a => 1, datagrid => 1}}); |
|
|
|
|
|
## Block-table Block* | table | select | datalist | Empty |
|
|
my $mode = 'any'; |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($mode eq 'block') { |
|
|
$not_allowed = 1 |
|
|
unless $HTMLBlockLevelElements->{$node_ns}->{$node_ln}; |
|
|
} elsif ($mode eq 'any') { |
|
|
if ($node_ns eq $HTML_NS and |
|
|
{table => 1, select => 1, datalist => 1}->{$node_ln}) { |
|
|
$mode = 'none'; |
|
|
} elsif ($HTMLBlockLevelElements->{$node_ns}->{$node_ln}) { |
|
|
$mode = 'block'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $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]/) { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
|
|
|
|
push @$new_todos, $end; |
|
|
return ($new_todos); |
|
| 298 |
}, |
}, |
| 299 |
}; |
q<http://www.w3.org/1998/Math/MathML> => {math => 1}, |
| 300 |
|
q<http://www.w3.org/2000/svg> => {svg => 1}, |
| 301 |
$Element->{$HTML_NS}->{command} = { |
## NOTE: Foreign elements with content (but no metadata) are |
| 302 |
attrs_checker => $GetHTMLAttrsChecker->({ |
## embedded content. |
| 303 |
checked => $GetHTMLBooleanAttrChecker->('checked'), |
}; |
| 304 |
default => $GetHTMLBooleanAttrChecker->('default'), |
|
| 305 |
disabled => $GetHTMLBooleanAttrChecker->('disabled'), |
our $IsInHTMLInteractiveContent = sub { |
| 306 |
hidden => $GetHTMLBooleanAttrChecker->('hidden'), |
my ($el, $nsuri, $ln) = @_; |
| 307 |
icon => $HTMLURIAttrChecker, |
|
| 308 |
label => sub { }, ## NOTE: No conformance creteria |
## NOTE: This CODE returns whether an element that is conditionally |
| 309 |
radiogroup => sub { }, ## NOTE: No conformance creteria |
## categorizzed as an interactive content is currently in that |
| 310 |
## NOTE: |title| has special semantics, but no syntactical difference |
## condition or not. See $HTMLInteractiveContent list defined in |
| 311 |
type => sub { |
## Whatpm::ContentChecler::HTML for the list of all (conditionally |
| 312 |
my ($self, $attr) = @_; |
## or permanently) interactive content. |
| 313 |
my $value = $attr->value; |
|
| 314 |
unless ({command => 1, checkbox => 1, radio => 1}->{$value}) { |
if ($nsuri eq $HTML_NS and ($ln eq 'video' or $ln eq 'audio')) { |
| 315 |
$self->{onerror}->(node => $attr, type => 'attribute value not allowed'); |
return $el->has_attribute ('controls'); |
| 316 |
} |
} elsif ($nsuri eq $HTML_NS and $ln eq 'menu') { |
| 317 |
}, |
my $value = $el->get_attribute ('type'); |
| 318 |
}), |
$value =~ tr/A-Z/a-z/; # ASCII case-insensitive |
| 319 |
checker => $HTMLEmptyChecker, |
return ($value eq 'toolbar'); |
| 320 |
}; |
} else { |
| 321 |
|
return 1; |
| 322 |
$Element->{$HTML_NS}->{menu} = { |
} |
| 323 |
attrs_checker => $GetHTMLAttrsChecker->({ |
}; # $IsInHTMLInteractiveContent |
|
autosubmit => $GetHTMLBooleanAttrChecker->('autosubmit'), |
|
|
id => sub { |
|
|
## NOTE: same as global |id=""|, with |$self->{menu}| registeration |
|
|
my ($self, $attr) = @_; |
|
|
my $value = $attr->value; |
|
|
if (length $value > 0) { |
|
|
if ($self->{id}->{$value}) { |
|
|
$self->{onerror}->(node => $attr, type => 'duplicate ID'); |
|
|
} else { |
|
|
$self->{id}->{$value} = 1; |
|
|
} |
|
|
} else { |
|
|
## NOTE: MUST contain at least one character |
|
|
$self->{onerror}->(node => $attr, type => 'empty attribute value'); |
|
|
} |
|
|
if ($value =~ /[\x09-\x0D\x20]/) { |
|
|
$self->{onerror}->(node => $attr, type => 'space in ID'); |
|
|
} |
|
|
$self->{menu}->{$value} ||= $attr; |
|
|
## ISSUE: <menu id=""><p contextmenu=""> match? |
|
|
}, |
|
|
label => sub { }, ## NOTE: No conformance creteria |
|
|
type => $GetHTMLEnumeratedAttrChecker->({context => 1, toolbar => 1}), |
|
|
}), |
|
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
my $el = $todo->{node}; |
|
|
my $new_todos = []; |
|
|
my @nodes = (@{$el->child_nodes}); |
|
|
|
|
|
my $content = 'li or inline'; |
|
|
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; |
|
|
my $not_allowed = $self->{minuses}->{$node_ns}->{$node_ln}; |
|
|
if ($node_ns eq $HTML_NS and $node_ln eq 'li') { |
|
|
if ($content eq 'inline') { |
|
|
$not_allowed = 1; |
|
|
} elsif ($content eq 'li or inline') { |
|
|
$content = 'li'; |
|
|
} |
|
|
} else { |
|
|
if ($HTMLStrictlyInlineLevelElements->{$node_ns}->{$node_ln} or |
|
|
$HTMLStructuredInlineLevelElements->{$node_ns}->{$node_ln}) { |
|
|
$content = 'inline'; |
|
|
} else { |
|
|
$not_allowed = 1; |
|
|
} |
|
|
} |
|
|
$self->{onerror}->(node => $node, type => 'element not allowed') |
|
|
if $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]/) { |
|
|
if ($content eq 'li') { |
|
|
$self->{onerror}->(node => $node, type => 'character not allowed'); |
|
|
} elsif ($content eq 'li or inline') { |
|
|
$content = 'inline'; |
|
|
} |
|
|
} |
|
|
} elsif ($nt == 5) { |
|
|
unshift @nodes, @{$node->child_nodes}; |
|
|
} |
|
|
} |
|
| 324 |
|
|
| 325 |
for (@$new_todos) { |
my $HTMLTransparentElements = { |
| 326 |
$_->{inline} = 1; |
$HTML_NS => {qw/ins 1 del 1 font 1 noscript 1 canvas 1 a 1/}, |
| 327 |
} |
## NOTE: |html:noscript| is transparent if scripting is disabled |
| 328 |
return ($new_todos); |
## and not in |head|. |
|
}, |
|
| 329 |
}; |
}; |
| 330 |
|
|
| 331 |
$Element->{$HTML_NS}->{legend} = { |
my $HTMLSemiTransparentElements = { |
| 332 |
attrs_checker => $GetHTMLAttrsChecker->({}), |
$HTML_NS => {object => 1, video => 1, audio => 1}, |
|
checker => sub { |
|
|
my ($self, $todo) = @_; |
|
|
|
|
|
my $parent = $todo->{node}->manakai_parent_element; |
|
|
if (defined $parent) { |
|
|
my $nsuri = $parent->namespace_uri; |
|
|
$nsuri = '' unless defined $nsuri; |
|
|
my $ln = $parent->manakai_local_name; |
|
|
if ($nsuri eq $HTML_NS and $ln eq 'figure') { |
|
|
return $HTMLInlineChecker->($self, $todo); |
|
|
} else { |
|
|
return $HTMLSignificantStrictlyInlineChecker->($self, $todo); |
|
|
} |
|
|
} else { |
|
|
return $HTMLInlineChecker->($self, $todo); |
|
|
} |
|
|
|
|
|
## ISSUE: Content model is defined only for fieldset/legend, |
|
|
## details/legend, and figure/legend. |
|
|
}, |
|
| 333 |
}; |
}; |
| 334 |
|
|
| 335 |
$Element->{$HTML_NS}->{div} = { |
our $Element = {}; |
|
attrs_checker => $GetHTMLAttrsChecker->({}), |
|
|
checker => $GetHTMLZeroOrMoreThenBlockOrInlineChecker->($HTML_NS, 'style'), |
|
|
}; |
|
| 336 |
|
|
| 337 |
$Element->{$HTML_NS}->{font} = { |
$Element->{q<http://www.w3.org/1999/02/22-rdf-syntax-ns#>}->{RDF} = { |
| 338 |
attrs_checker => $GetHTMLAttrsChecker->({}), ## TODO |
%AnyChecker, |
| 339 |
checker => $HTMLTransparentChecker, |
status => FEATURE_STATUS_REC | FEATURE_ALLOWED, |
| 340 |
|
is_root => 1, ## ISSUE: Not explicitly allowed for non application/rdf+xml |
| 341 |
|
check_start => sub { |
| 342 |
|
my ($self, $item, $element_state) = @_; |
| 343 |
|
my $triple = []; |
| 344 |
|
push @{$self->{return}->{rdf}}, [$item->{node}, $triple]; |
| 345 |
|
require Whatpm::RDFXML; |
| 346 |
|
my $rdf = Whatpm::RDFXML->new; |
| 347 |
|
## TODO: Should we make bnodeid unique in a document? |
| 348 |
|
$rdf->{onerror} = $self->{onerror}; |
| 349 |
|
$rdf->{level} = $self->{level}; |
| 350 |
|
$rdf->{ontriple} = sub { |
| 351 |
|
my %opt = @_; |
| 352 |
|
push @$triple, |
| 353 |
|
[$opt{node}, $opt{subject}, $opt{predicate}, $opt{object}]; |
| 354 |
|
if (defined $opt{id}) { |
| 355 |
|
push @$triple, |
| 356 |
|
[$opt{node}, |
| 357 |
|
$opt{id}, |
| 358 |
|
{uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#subject>}, |
| 359 |
|
$opt{subject}]; |
| 360 |
|
push @$triple, |
| 361 |
|
[$opt{node}, |
| 362 |
|
$opt{id}, |
| 363 |
|
{uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate>}, |
| 364 |
|
$opt{predicate}]; |
| 365 |
|
push @$triple, |
| 366 |
|
[$opt{node}, |
| 367 |
|
$opt{id}, |
| 368 |
|
{uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#object>}, |
| 369 |
|
$opt{object}]; |
| 370 |
|
push @$triple, |
| 371 |
|
[$opt{node}, |
| 372 |
|
$opt{id}, |
| 373 |
|
{uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>}, |
| 374 |
|
{uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#Statement>}]; |
| 375 |
|
} |
| 376 |
|
}; |
| 377 |
|
$rdf->convert_rdf_element ($item->{node}); |
| 378 |
|
}, |
| 379 |
|
}; |
| 380 |
|
|
| 381 |
|
my $default_error_level = { |
| 382 |
|
must => 'm', |
| 383 |
|
should => 's', |
| 384 |
|
warn => 'w', |
| 385 |
|
good => 'w', |
| 386 |
|
undefined => 'w', |
| 387 |
|
info => 'i', |
| 388 |
|
|
| 389 |
|
uncertain => 'u', |
| 390 |
|
|
| 391 |
|
html4_fact => 'm', |
| 392 |
|
html5_no_may => 'm', |
| 393 |
|
|
| 394 |
|
xml_error => 'm', ## TODO: correct? |
| 395 |
|
xml_id_error => 'm', ## TODO: ? |
| 396 |
|
nc => 'm', ## XML Namespace Constraints ## TODO: correct? |
| 397 |
|
|
| 398 |
|
## |Whatpm::URIChecker| |
| 399 |
|
uri_syntax => 'm', |
| 400 |
|
uri_fact => 'm', |
| 401 |
|
uri_lc_must => 'm', |
| 402 |
|
uri_lc_should => 'w', |
| 403 |
|
|
| 404 |
|
## |Whatpm::IMTChecker| |
| 405 |
|
mime_must => 'm', # lowercase "must" |
| 406 |
|
mime_fact => 'm', |
| 407 |
|
mime_strongly_discouraged => 'w', |
| 408 |
|
mime_discouraged => 'w', |
| 409 |
|
|
| 410 |
|
## |Whatpm::LangTag| |
| 411 |
|
langtag_fact => 'm', |
| 412 |
|
|
| 413 |
|
## |Whatpm::RDFXML| |
| 414 |
|
rdf_fact => 'm', |
| 415 |
|
rdf_grammer => 'm', |
| 416 |
|
rdf_lc_must => 'm', |
| 417 |
|
|
| 418 |
|
## |Message::Charset::Info| and |Whatpm::Charset::DecodeHandle| |
| 419 |
|
charset_variant => 'm', |
| 420 |
|
## An error caused by use of a variant charset that is not conforming |
| 421 |
|
## to the original charset (e.g. use of 0x80 in an ISO-8859-1 document |
| 422 |
|
## which is interpreted as a Windows-1252 document instead). |
| 423 |
|
charset_fact => 'm', |
| 424 |
|
iso_shall => 'm', |
| 425 |
}; |
}; |
| 426 |
|
|
| 427 |
sub check_document ($$$) { |
sub check_document ($$$;$) { |
| 428 |
my ($self, $doc, $onerror) = @_; |
my ($self, $doc, $onerror, $onsubdoc) = @_; |
| 429 |
$self = bless {}, $self unless ref $self; |
$self = bless {}, $self unless ref $self; |
| 430 |
$self->{onerror} = $onerror; |
$self->{onerror} = $onerror; |
| 431 |
|
$self->{onsubdoc} = $onsubdoc || sub { |
| 432 |
|
warn "A subdocument is not conformance-checked"; |
| 433 |
|
}; |
| 434 |
|
|
| 435 |
|
$self->{level} ||= $default_error_level; |
| 436 |
|
|
| 437 |
|
## TODO: If application/rdf+xml, RDF/XML mode should be invoked. |
| 438 |
|
|
| 439 |
my $docel = $doc->document_element; |
my $docel = $doc->document_element; |
| 440 |
unless (defined $docel) { |
unless (defined $docel) { |
| 441 |
## ISSUE: Should we check content of Document node? |
## ISSUE: Should we check content of Document node? |
| 442 |
$onerror->(node => $doc, type => 'no document element'); |
$onerror->(node => $doc, type => 'no document element', |
| 443 |
|
level => $self->{level}->{must}); |
| 444 |
## ISSUE: Is this non-conforming (to what spec)? Or just a warning? |
## ISSUE: Is this non-conforming (to what spec)? Or just a warning? |
| 445 |
return; |
return { |
| 446 |
|
class => {}, |
| 447 |
|
id => {}, table => [], term => {}, |
| 448 |
|
}; |
| 449 |
} |
} |
| 450 |
|
|
| 451 |
## ISSUE: Unexpanded entity references and HTML5 conformance |
## ISSUE: Unexpanded entity references and HTML5 conformance |
| 452 |
|
|
| 453 |
my $docel_nsuri = $docel->namespace_uri; |
my $docel_nsuri = $docel->namespace_uri; |
| 454 |
$docel_nsuri = '' unless defined $docel_nsuri; |
if (defined $docel_nsuri) { |
| 455 |
|
load_ns_module ($docel_nsuri); |
| 456 |
|
} else { |
| 457 |
|
$docel_nsuri = ''; |
| 458 |
|
} |
| 459 |
my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} || |
my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} || |
| 460 |
$Element->{$docel_nsuri}->{''} || |
$Element->{$docel_nsuri}->{''} || |
| 461 |
$ElementDefault; |
$ElementDefault; |
| 462 |
if ($docel_def->{is_root}) { |
if ($docel_def->{is_root}) { |
| 463 |
# |
# |
| 464 |
|
} elsif ($docel_def->{is_xml_root}) { |
| 465 |
|
unless ($doc->manakai_is_html) { |
| 466 |
|
# |
| 467 |
|
} else { |
| 468 |
|
$onerror->(node => $docel, type => 'element not allowed:root:xml', |
| 469 |
|
level => $self->{level}->{must}); |
| 470 |
|
} |
| 471 |
} else { |
} else { |
| 472 |
$onerror->(node => $docel, type => 'element not allowed'); |
$onerror->(node => $docel, type => 'element not allowed:root', |
| 473 |
|
level => $self->{level}->{must}); |
| 474 |
} |
} |
| 475 |
|
|
| 476 |
## TODO: Check for other items other than document element |
## TODO: Check for other items other than document element |
| 477 |
## (second (errorous) element, text nodes, PI nodes, doctype nodes) |
## (second (errorous) element, text nodes, PI nodes, doctype nodes) |
| 478 |
|
|
| 479 |
return $self->check_element ($docel, $onerror); |
my $return = $self->check_element ($docel, $onerror, $onsubdoc); |
| 480 |
|
|
| 481 |
|
## TODO: Test for these checks are necessary. |
| 482 |
|
my $charset_name = $doc->input_encoding; |
| 483 |
|
if (defined $charset_name) { |
| 484 |
|
require Message::Charset::Info; |
| 485 |
|
my $charset = $Message::Charset::Info::IANACharset->{$charset_name}; |
| 486 |
|
|
| 487 |
|
if ($doc->manakai_is_html) { |
| 488 |
|
if (not $doc->manakai_has_bom and |
| 489 |
|
not defined $doc->manakai_charset) { |
| 490 |
|
unless ($charset->{is_html_ascii_superset}) { |
| 491 |
|
$onerror->(node => $doc, |
| 492 |
|
level => $self->{level}->{must}, |
| 493 |
|
type => 'non ascii superset', |
| 494 |
|
text => $charset_name); |
| 495 |
|
} |
| 496 |
|
|
| 497 |
|
if (not $self->{has_charset} and ## TODO: This does not work now. |
| 498 |
|
not $charset->{iana_names}->{'us-ascii'}) { |
| 499 |
|
$onerror->(node => $doc, |
| 500 |
|
level => $self->{level}->{must}, |
| 501 |
|
type => 'no character encoding declaration', |
| 502 |
|
text => $charset_name); |
| 503 |
|
} |
| 504 |
|
} |
| 505 |
|
|
| 506 |
|
if ($charset->{iana_names}->{'utf-8'}) { |
| 507 |
|
# |
| 508 |
|
} elsif ($charset->{iana_names}->{'jis_x0212-1990'} or |
| 509 |
|
$charset->{iana_names}->{'x-jis0208'} or |
| 510 |
|
$charset->{iana_names}->{'utf-32'} or ## ISSUE: UTF-32BE? UTF-32LE? |
| 511 |
|
($charset->{category} & Message::Charset::Info::CHARSET_CATEGORY_EBCDIC ())) { |
| 512 |
|
$onerror->(node => $doc, |
| 513 |
|
type => 'bad character encoding', |
| 514 |
|
text => $charset_name, |
| 515 |
|
level => $self->{level}->{should}, |
| 516 |
|
layer => 'encode'); |
| 517 |
|
} elsif ($charset->{iana_names}->{'cesu-8'} or |
| 518 |
|
$charset->{iana_names}->{'utf-8'} or ## ISSUE: UNICODE-1-1-UTF-7? |
| 519 |
|
$charset->{iana_names}->{'bocu-1'} or |
| 520 |
|
$charset->{iana_names}->{'scsu'}) { |
| 521 |
|
$onerror->(node => $doc, |
| 522 |
|
type => 'disallowed character encoding', |
| 523 |
|
text => $charset_name, |
| 524 |
|
level => $self->{level}->{must}, |
| 525 |
|
layer => 'encode'); |
| 526 |
|
} else { |
| 527 |
|
$onerror->(node => $doc, |
| 528 |
|
type => 'non-utf-8 character encoding', |
| 529 |
|
text => $charset_name, |
| 530 |
|
level => $self->{level}->{good}, |
| 531 |
|
layer => 'encode'); |
| 532 |
|
} |
| 533 |
|
} |
| 534 |
|
} elsif ($doc->manakai_is_html) { |
| 535 |
|
## NOTE: MUST and SHOULD requirements above cannot be tested, |
| 536 |
|
## since the document has no input charset encoding information. |
| 537 |
|
$onerror->(node => $doc, |
| 538 |
|
type => 'character encoding unchecked', |
| 539 |
|
level => $self->{level}->{info}, |
| 540 |
|
layer => 'encode'); |
| 541 |
|
} |
| 542 |
|
|
| 543 |
|
return $return; |
| 544 |
} # check_document |
} # check_document |
| 545 |
|
|
| 546 |
sub check_element ($$$) { |
## Check an element. The element is checked as if it is an orphan node (i.e. |
| 547 |
my ($self, $el, $onerror) = @_; |
## an element without a parent node). |
| 548 |
|
sub check_element ($$$;$) { |
| 549 |
|
my ($self, $el, $onerror, $onsubdoc) = @_; |
| 550 |
$self = bless {}, $self unless ref $self; |
$self = bless {}, $self unless ref $self; |
| 551 |
$self->{onerror} = $onerror; |
$self->{onerror} = $onerror; |
| 552 |
|
$self->{onsubdoc} = $onsubdoc || sub { |
| 553 |
|
warn "A subdocument is not conformance-checked"; |
| 554 |
|
}; |
| 555 |
|
|
| 556 |
$self->{minuses} = {}; |
$self->{level} ||= $default_error_level; |
| 557 |
|
|
| 558 |
|
$self->{plus_elements} = {}; |
| 559 |
|
$self->{minus_elements} = {}; |
| 560 |
$self->{id} = {}; |
$self->{id} = {}; |
| 561 |
|
$self->{form} = {}; |
| 562 |
$self->{term} = {}; |
$self->{term} = {}; |
| 563 |
$self->{usemap} = []; |
$self->{usemap} = []; |
| 564 |
|
$self->{ref} = []; # datetemplate data references |
| 565 |
|
$self->{template} = []; # datatemplate template references |
| 566 |
$self->{contextmenu} = []; |
$self->{contextmenu} = []; |
| 567 |
$self->{map} = {}; |
$self->{map} = {}; |
| 568 |
$self->{menu} = {}; |
$self->{menu} = {}; |
| 569 |
$self->{has_link_type} = {}; |
$self->{has_link_type} = {}; |
| 570 |
|
$self->{flag} = {}; |
| 571 |
|
#$self->{has_uri_attr}; |
| 572 |
|
#$self->{has_hyperlink_element}; |
| 573 |
|
#$self->{has_charset}; |
| 574 |
|
#$self->{has_base}; |
| 575 |
$self->{return} = { |
$self->{return} = { |
| 576 |
table => [], term => $self->{term}, |
class => {}, |
| 577 |
|
id => $self->{id}, |
| 578 |
|
table => [], # table objects returned by Whatpm::HTMLTable |
| 579 |
|
term => $self->{term}, |
| 580 |
|
uri => {}, # URIs other than those in RDF triples |
| 581 |
|
## TODO: xmlns="", SYSTEM "", atom:* src="", xml:base="" |
| 582 |
|
rdf => [], |
| 583 |
}; |
}; |
| 584 |
|
|
| 585 |
my @todo = ({type => 'element', node => $el}); |
my @item = ({type => 'element', node => $el, parent_state => {}}); |
| 586 |
while (@todo) { |
$item[-1]->{real_parent_state} = $item[-1]->{parent_state}; |
| 587 |
my $todo = shift @todo; |
while (@item) { |
| 588 |
if ($todo->{type} eq 'element') { |
my $item = shift @item; |
| 589 |
my $prefix = $todo->{node}->prefix; |
if (ref $item eq 'ARRAY') { |
| 590 |
if (defined $prefix and $prefix eq 'xmlns') { |
my $code = shift @$item; |
| 591 |
$self->{onerror} |
next unless $code;## TODO: temp. |
| 592 |
->(node => $todo->{node}, level => 'NC', |
$code->(@$item); |
| 593 |
type => 'Reserved Prefixes and Namespace Names:<xmlns:>'); |
} elsif ($item->{type} eq 'element') { |
| 594 |
|
my $el_nsuri = $item->{node}->namespace_uri; |
| 595 |
|
if (defined $el_nsuri) { |
| 596 |
|
load_ns_module ($el_nsuri); |
| 597 |
|
} else { |
| 598 |
|
$el_nsuri = ''; |
| 599 |
} |
} |
| 600 |
my $nsuri = $todo->{node}->namespace_uri; |
my $el_ln = $item->{node}->manakai_local_name; |
| 601 |
$nsuri = '' unless defined $nsuri; |
|
| 602 |
my $ln = $todo->{node}->manakai_local_name; |
my $element_state = {}; |
| 603 |
my $eldef = $Element->{$nsuri}->{$ln} || |
my $eldef = $Element->{$el_nsuri}->{$el_ln} || |
| 604 |
$Element->{$nsuri}->{''} || |
$Element->{$el_nsuri}->{''} || |
| 605 |
$ElementDefault; |
$ElementDefault; |
| 606 |
$eldef->{attrs_checker}->($self, $todo); |
my $content_def = $item->{transparent} |
| 607 |
my ($new_todos) = $eldef->{checker}->($self, $todo); |
? $item->{parent_def} || $eldef : $eldef; |
| 608 |
unshift @todo, @$new_todos; |
my $content_state = $item->{transparent} |
| 609 |
} elsif ($todo->{type} eq 'element-attributes') { |
? $item->{parent_def} |
| 610 |
my $prefix = $todo->{node}->prefix; |
? $item->{parent_state} || $element_state : $element_state |
| 611 |
if (defined $prefix and $prefix eq 'xmlns') { |
: $element_state; |
| 612 |
$self->{onerror} |
|
| 613 |
->(node => $todo->{node}, level => 'NC', |
unless ($eldef->{status} & FEATURE_STATUS_REC) { |
| 614 |
type => 'Reserved Prefixes and Namespace Names:<xmlns:>'); |
my $status = $eldef->{status} & FEATURE_STATUS_CR ? 'cr' : |
| 615 |
|
$eldef->{status} & FEATURE_STATUS_LC ? 'lc' : |
| 616 |
|
$eldef->{status} & FEATURE_STATUS_WD ? 'wd' : 'non-standard'; |
| 617 |
|
$self->{onerror}->(node => $item->{node}, |
| 618 |
|
type => 'status:'.$status.':element', |
| 619 |
|
level => $self->{level}->{info}); |
| 620 |
|
} |
| 621 |
|
if (not ($eldef->{status} & FEATURE_ALLOWED)) { |
| 622 |
|
$self->{onerror}->(node => $item->{node}, |
| 623 |
|
type => 'element not defined', |
| 624 |
|
level => $self->{level}->{must}); |
| 625 |
|
} elsif ($eldef->{status} & FEATURE_DEPRECATED_SHOULD) { |
| 626 |
|
$self->{onerror}->(node => $item->{node}, |
| 627 |
|
type => 'deprecated:element', |
| 628 |
|
level => $self->{level}->{should}); |
| 629 |
|
} elsif ($eldef->{status} & FEATURE_DEPRECATED_INFO) { |
| 630 |
|
$self->{onerror}->(node => $item->{node}, |
| 631 |
|
type => 'deprecated:element', |
| 632 |
|
level => $self->{level}->{info}); |
| 633 |
|
} |
| 634 |
|
|
| 635 |
|
my @new_item; |
| 636 |
|
push @new_item, [$eldef->{check_start}, $self, $item, $element_state]; |
| 637 |
|
push @new_item, [$eldef->{check_attrs}, $self, $item, $element_state]; |
| 638 |
|
|
| 639 |
|
my @child = @{$item->{node}->child_nodes}; |
| 640 |
|
while (@child) { |
| 641 |
|
my $child = shift @child; |
| 642 |
|
my $child_nt = $child->node_type; |
| 643 |
|
if ($child_nt == 1) { # ELEMENT_NODE |
| 644 |
|
my $child_nsuri = $child->namespace_uri; |
| 645 |
|
$child_nsuri = '' unless defined $child_nsuri; |
| 646 |
|
my $child_ln = $child->manakai_local_name; |
| 647 |
|
if ($HTMLTransparentElements->{$child_nsuri}->{$child_ln} and |
| 648 |
|
not (($self->{flag}->{in_head} or |
| 649 |
|
($el_nsuri eq $HTML_NS and $el_ln eq 'head')) and |
| 650 |
|
$child_nsuri eq $HTML_NS and $child_ln eq 'noscript')) { |
| 651 |
|
push @new_item, [$content_def->{check_child_element}, |
| 652 |
|
$self, $item, $child, |
| 653 |
|
$child_nsuri, $child_ln, 1, |
| 654 |
|
$content_state, $element_state]; |
| 655 |
|
push @new_item, {type => 'element', node => $child, |
| 656 |
|
parent_state => $content_state, |
| 657 |
|
parent_def => $content_def, |
| 658 |
|
real_parent_state => $element_state, |
| 659 |
|
transparent => 1}; |
| 660 |
|
} else { |
| 661 |
|
if ($item->{parent_def} and # has parent |
| 662 |
|
$el_nsuri eq $HTML_NS) { ## $HTMLSemiTransparentElements |
| 663 |
|
if ($el_ln eq 'object') { |
| 664 |
|
if ($self->{plus_elements}->{$child_nsuri}->{$child_ln}) { |
| 665 |
|
# |
| 666 |
|
} elsif ($child_nsuri eq $HTML_NS and $child_ln eq 'param') { |
| 667 |
|
# |
| 668 |
|
} else { |
| 669 |
|
$content_def = $item->{parent_def} || $content_def; |
| 670 |
|
$content_state = $item->{parent_state} || $content_state; |
| 671 |
|
} |
| 672 |
|
} elsif ($el_ln eq 'video' or $el_ln eq 'audio') { |
| 673 |
|
if ($self->{plus_elements}->{$child_nsuri}->{$child_ln}) { |
| 674 |
|
# |
| 675 |
|
} elsif ($child_nsuri eq $HTML_NS and $child_ln eq 'source') { |
| 676 |
|
$element_state->{has_source} = 1; |
| 677 |
|
} else { |
| 678 |
|
$content_def = $item->{parent_def} || $content_def; |
| 679 |
|
$content_state = $item->{parent_state} || $content_state; |
| 680 |
|
} |
| 681 |
|
} |
| 682 |
|
} |
| 683 |
|
|
| 684 |
|
push @new_item, [$content_def->{check_child_element}, |
| 685 |
|
$self, $item, $child, |
| 686 |
|
$child_nsuri, $child_ln, |
| 687 |
|
$HTMLSemiTransparentElements |
| 688 |
|
->{$child_nsuri}->{$child_ln}, |
| 689 |
|
$content_state, $element_state]; |
| 690 |
|
push @new_item, {type => 'element', node => $child, |
| 691 |
|
parent_def => $content_def, |
| 692 |
|
real_parent_state => $element_state, |
| 693 |
|
parent_state => $content_state}; |
| 694 |
|
} |
| 695 |
|
|
| 696 |
|
if ($HTMLEmbeddedContent->{$child_nsuri}->{$child_ln}) { |
| 697 |
|
$element_state->{has_significant} = 1; |
| 698 |
|
} |
| 699 |
|
} elsif ($child_nt == 3 or # TEXT_NODE |
| 700 |
|
$child_nt == 4) { # CDATA_SECTION_NODE |
| 701 |
|
my $has_significant = ($child->data =~ /[^\x09\x0A\x0C\x0D\x20]/); |
| 702 |
|
push @new_item, [$content_def->{check_child_text}, |
| 703 |
|
$self, $item, $child, $has_significant, |
| 704 |
|
$content_state, $element_state]; |
| 705 |
|
$element_state->{has_significant} ||= $has_significant; |
| 706 |
|
if ($has_significant and |
| 707 |
|
$HTMLSemiTransparentElements->{$el_nsuri}->{$el_ln}) { |
| 708 |
|
$content_def = $item->{parent_def} || $content_def; |
| 709 |
|
} |
| 710 |
|
} elsif ($child_nt == 5) { # ENTITY_REFERENCE_NODE |
| 711 |
|
push @child, @{$child->child_nodes}; |
| 712 |
|
} |
| 713 |
|
## TODO: PI_NODE |
| 714 |
|
## TODO: Unknown node type |
| 715 |
|
} |
| 716 |
|
|
| 717 |
|
push @new_item, [$eldef->{check_end}, $self, $item, $element_state]; |
| 718 |
|
|
| 719 |
|
unshift @item, @new_item; |
| 720 |
|
} else { |
| 721 |
|
die "$0: Internal error: Unsupported checking action type |$item->{type}|"; |
| 722 |
|
} |
| 723 |
|
} |
| 724 |
|
|
| 725 |
|
for (@{$self->{template}}) { |
| 726 |
|
## TODO: If the document is an XML document, ... |
| 727 |
|
## NOTE: If the document is an HTML document: |
| 728 |
|
## ISSUE: We need to percent-decode? |
| 729 |
|
F: { |
| 730 |
|
if ($self->{id}->{$_->[0]}) { |
| 731 |
|
my $el = $self->{id}->{$_->[0]}->[0]->owner_element; |
| 732 |
|
if ($el->node_type == 1 and # ELEMENT_NODE |
| 733 |
|
$el->manakai_local_name eq 'datatemplate') { |
| 734 |
|
my $nsuri = $el->namespace_uri; |
| 735 |
|
if (defined $nsuri and $nsuri eq $HTML_NS) { |
| 736 |
|
if ($el eq $_->[1]->owner_element) { |
| 737 |
|
$self->{onerror}->(node => $_->[1], |
| 738 |
|
type => 'fragment points itself', |
| 739 |
|
level => $self->{level}->{must}); |
| 740 |
|
} |
| 741 |
|
|
| 742 |
|
last F; |
| 743 |
|
} |
| 744 |
|
} |
| 745 |
|
} |
| 746 |
|
## TODO: Should we raise a "fragment points nothing" error instead |
| 747 |
|
## if the fragment identifier identifies no element? |
| 748 |
|
|
| 749 |
|
$self->{onerror}->(node => $_->[1], type => 'template:not template', |
| 750 |
|
level => $self->{level}->{must}); |
| 751 |
|
} # F |
| 752 |
|
} |
| 753 |
|
|
| 754 |
|
for (@{$self->{ref}}) { |
| 755 |
|
## TOOD: If XML |
| 756 |
|
## NOTE: If it is an HTML document: |
| 757 |
|
if ($_->[0] eq '') { |
| 758 |
|
## NOTE: It points the top of the document. |
| 759 |
|
} elsif ($self->{id}->{$_->[0]}) { |
| 760 |
|
if ($self->{id}->{$_->[0]}->[0]->owner_element |
| 761 |
|
eq $_->[1]->owner_element) { |
| 762 |
|
$self->{onerror}->(node => $_->[1], type => 'fragment points itself', |
| 763 |
|
level => $self->{level}->{must}); |
| 764 |
} |
} |
|
my $nsuri = $todo->{node}->namespace_uri; |
|
|
$nsuri = '' unless defined $nsuri; |
|
|
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}->(); |
|
| 765 |
} else { |
} else { |
| 766 |
die "$0: Internal error: Unsupported checking action type |$todo->{type}|"; |
$self->{onerror}->(node => $_->[1], type => 'fragment points nothing', |
| 767 |
|
level => $self->{level}->{must}); |
| 768 |
} |
} |
| 769 |
} |
} |
| 770 |
|
|
| 771 |
|
## TODO: Maybe we should have $document->manakai_get_by_fragment or something |
| 772 |
|
|
| 773 |
for (@{$self->{usemap}}) { |
for (@{$self->{usemap}}) { |
| 774 |
unless ($self->{map}->{$_->[0]}) { |
unless ($self->{map}->{$_->[0]}) { |
| 775 |
$self->{onerror}->(node => $_->[1], type => 'no referenced map'); |
$self->{onerror}->(node => $_->[1], type => 'no referenced map', |
| 776 |
|
level => $self->{level}->{must}); |
| 777 |
} |
} |
| 778 |
} |
} |
| 779 |
|
|
| 780 |
for (@{$self->{contextmenu}}) { |
for (@{$self->{contextmenu}}) { |
| 781 |
unless ($self->{menu}->{$_->[0]}) { |
unless ($self->{menu}->{$_->[0]}) { |
| 782 |
$self->{onerror}->(node => $_->[1], type => 'no referenced menu'); |
$self->{onerror}->(node => $_->[1], type => 'no referenced menu', |
| 783 |
|
level => $self->{level}->{must}); |
| 784 |
} |
} |
| 785 |
} |
} |
| 786 |
|
|
| 787 |
delete $self->{minuses}; |
delete $self->{plus_elements}; |
| 788 |
|
delete $self->{minus_elements}; |
| 789 |
delete $self->{onerror}; |
delete $self->{onerror}; |
| 790 |
delete $self->{id}; |
delete $self->{id}; |
| 791 |
|
delete $self->{form}; |
| 792 |
delete $self->{usemap}; |
delete $self->{usemap}; |
| 793 |
|
delete $self->{ref}; |
| 794 |
|
delete $self->{template}; |
| 795 |
delete $self->{map}; |
delete $self->{map}; |
| 796 |
return $self->{return}; |
return $self->{return}; |
| 797 |
} # check_element |
} # check_element |
| 798 |
|
|
| 799 |
|
sub _add_minus_elements ($$@) { |
| 800 |
|
my $self = shift; |
| 801 |
|
my $element_state = shift; |
| 802 |
|
for my $elements (@_) { |
| 803 |
|
for my $nsuri (keys %$elements) { |
| 804 |
|
for my $ln (keys %{$elements->{$nsuri}}) { |
| 805 |
|
unless ($self->{minus_elements}->{$nsuri}->{$ln}) { |
| 806 |
|
$element_state->{minus_elements_original}->{$nsuri}->{$ln} = 0; |
| 807 |
|
$self->{minus_elements}->{$nsuri}->{$ln} = 1; |
| 808 |
|
} |
| 809 |
|
} |
| 810 |
|
} |
| 811 |
|
} |
| 812 |
|
} # _add_minus_elements |
| 813 |
|
|
| 814 |
|
sub _remove_minus_elements ($$) { |
| 815 |
|
my $self = shift; |
| 816 |
|
my $element_state = shift; |
| 817 |
|
for my $nsuri (keys %{$element_state->{minus_elements_original}}) { |
| 818 |
|
for my $ln (keys %{$element_state->{minus_elements_original}->{$nsuri}}) { |
| 819 |
|
delete $self->{minus_elements}->{$nsuri}->{$ln}; |
| 820 |
|
} |
| 821 |
|
} |
| 822 |
|
} # _remove_minus_elements |
| 823 |
|
|
| 824 |
|
sub _add_plus_elements ($$@) { |
| 825 |
|
my $self = shift; |
| 826 |
|
my $element_state = shift; |
| 827 |
|
for my $elements (@_) { |
| 828 |
|
for my $nsuri (keys %$elements) { |
| 829 |
|
for my $ln (keys %{$elements->{$nsuri}}) { |
| 830 |
|
unless ($self->{plus_elements}->{$nsuri}->{$ln}) { |
| 831 |
|
$element_state->{plus_elements_original}->{$nsuri}->{$ln} = 0; |
| 832 |
|
$self->{plus_elements}->{$nsuri}->{$ln} = 1; |
| 833 |
|
} |
| 834 |
|
} |
| 835 |
|
} |
| 836 |
|
} |
| 837 |
|
} # _add_plus_elements |
| 838 |
|
|
| 839 |
|
sub _remove_plus_elements ($$) { |
| 840 |
|
my $self = shift; |
| 841 |
|
my $element_state = shift; |
| 842 |
|
for my $nsuri (keys %{$element_state->{plus_elements_original}}) { |
| 843 |
|
for my $ln (keys %{$element_state->{plus_elements_original}->{$nsuri}}) { |
| 844 |
|
delete $self->{plus_elements}->{$nsuri}->{$ln}; |
| 845 |
|
} |
| 846 |
|
} |
| 847 |
|
} # _remove_plus_elements |
| 848 |
|
|
| 849 |
|
sub _attr_status_info ($$$) { |
| 850 |
|
my ($self, $attr, $status_code) = @_; |
| 851 |
|
|
| 852 |
|
if (not ($status_code & FEATURE_ALLOWED)) { |
| 853 |
|
$self->{onerror}->(node => $attr, |
| 854 |
|
type => 'attribute not defined', |
| 855 |
|
level => $self->{level}->{must}); |
| 856 |
|
} elsif ($status_code & FEATURE_DEPRECATED_SHOULD) { |
| 857 |
|
$self->{onerror}->(node => $attr, |
| 858 |
|
type => 'deprecated:attr', |
| 859 |
|
level => $self->{level}->{should}); |
| 860 |
|
} elsif ($status_code & FEATURE_DEPRECATED_INFO) { |
| 861 |
|
$self->{onerror}->(node => $attr, |
| 862 |
|
type => 'deprecated:attr', |
| 863 |
|
level => $self->{level}->{info}); |
| 864 |
|
} |
| 865 |
|
|
| 866 |
|
my $status; |
| 867 |
|
if ($status_code & FEATURE_STATUS_REC) { |
| 868 |
|
return; |
| 869 |
|
} elsif ($status_code & FEATURE_STATUS_CR) { |
| 870 |
|
$status = 'cr'; |
| 871 |
|
} elsif ($status_code & FEATURE_STATUS_LC) { |
| 872 |
|
$status = 'lc'; |
| 873 |
|
} elsif ($status_code & FEATURE_STATUS_WD) { |
| 874 |
|
$status = 'wd'; |
| 875 |
|
} else { |
| 876 |
|
$status = 'non-standard'; |
| 877 |
|
} |
| 878 |
|
$self->{onerror}->(node => $attr, |
| 879 |
|
type => 'status:'.$status.':attr', |
| 880 |
|
level => $self->{level}->{info}); |
| 881 |
|
} # _attr_status_info |
| 882 |
|
|
| 883 |
sub _add_minuses ($@) { |
sub _add_minuses ($@) { |
| 884 |
my $self = shift; |
my $self = shift; |
| 885 |
my $r = {}; |
my $r = {}; |
| 896 |
return {type => 'plus', list => $r}; |
return {type => 'plus', list => $r}; |
| 897 |
} # _add_minuses |
} # _add_minuses |
| 898 |
|
|
| 899 |
|
sub _add_pluses ($@) { |
| 900 |
|
my $self = shift; |
| 901 |
|
my $r = {}; |
| 902 |
|
for my $list (@_) { |
| 903 |
|
for my $ns (keys %$list) { |
| 904 |
|
for my $ln (keys %{$list->{$ns}}) { |
| 905 |
|
unless ($self->{pluses}->{$ns}->{$ln}) { |
| 906 |
|
$self->{pluses}->{$ns}->{$ln} = 1; |
| 907 |
|
$r->{$ns}->{$ln} = 1; |
| 908 |
|
} |
| 909 |
|
} |
| 910 |
|
} |
| 911 |
|
} |
| 912 |
|
return {type => 'minus', list => $r}; |
| 913 |
|
} # _add_pluses |
| 914 |
|
|
| 915 |
sub _remove_minuses ($$) { |
sub _remove_minuses ($$) { |
| 916 |
my ($self, $todo) = @_; |
my ($self, $todo) = @_; |
| 917 |
for my $ns (keys %{$todo->{list}}) { |
if ($todo->{type} eq 'minus') { |
| 918 |
for my $ln (keys %{$todo->{list}->{$ns}}) { |
for my $ns (keys %{$todo->{list}}) { |
| 919 |
delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln}; |
for my $ln (keys %{$todo->{list}->{$ns}}) { |
| 920 |
|
delete $self->{pluses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln}; |
| 921 |
|
} |
| 922 |
|
} |
| 923 |
|
} elsif ($todo->{type} eq 'plus') { |
| 924 |
|
for my $ns (keys %{$todo->{list}}) { |
| 925 |
|
for my $ln (keys %{$todo->{list}->{$ns}}) { |
| 926 |
|
delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln}; |
| 927 |
|
} |
| 928 |
} |
} |
| 929 |
|
} else { |
| 930 |
|
die "$0: Unknown +- type: $todo->{type}"; |
| 931 |
} |
} |
| 932 |
1; |
1; |
| 933 |
} # _remove_minuses |
} # _remove_minuses |
| 934 |
|
|
| 935 |
|
## NOTE: Priority for "minuses" and "pluses" are currently left |
| 936 |
|
## undefined and implemented inconsistently; it is not a problem for |
| 937 |
|
## now, since no element belongs to both lists. |
| 938 |
|
|
| 939 |
sub _check_get_children ($$$) { |
sub _check_get_children ($$$) { |
| 940 |
my ($self, $node, $parent_todo) = @_; |
my ($self, $node, $parent_todo) = @_; |
| 941 |
my $new_todos = []; |
my $new_todos = []; |
| 944 |
my $node_ns = $node->namespace_uri; |
my $node_ns = $node->namespace_uri; |
| 945 |
$node_ns = '' unless defined $node_ns; |
$node_ns = '' unless defined $node_ns; |
| 946 |
my $node_ln = $node->manakai_local_name; |
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, $end; |
|
|
} |
|
|
} |
|
|
## TODO: |noscript| is not a transparent element in |head|. |
|
| 947 |
if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) { |
if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) { |
| 948 |
unshift @$sib, @{$node->child_nodes}; |
if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') { |
| 949 |
push @$new_todos, {type => 'element-attributes', node => $node}; |
if ($parent_todo->{flag}->{in_head}) { |
| 950 |
last TP; |
# |
| 951 |
|
} else { |
| 952 |
|
my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}}); |
| 953 |
|
push @$sib, $end; |
| 954 |
|
|
| 955 |
|
unshift @$sib, @{$node->child_nodes}; |
| 956 |
|
push @$new_todos, {type => 'element-attributes', node => $node}; |
| 957 |
|
last TP; |
| 958 |
|
} |
| 959 |
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'del') { |
| 960 |
|
my $sig_flag = $parent_todo->{flag}->{has_descendant}->{significant}; |
| 961 |
|
unshift @$sib, @{$node->child_nodes}; |
| 962 |
|
push @$new_todos, {type => 'element-attributes', node => $node}; |
| 963 |
|
push @$new_todos, |
| 964 |
|
{type => 'code', |
| 965 |
|
code => sub { |
| 966 |
|
$parent_todo->{flag}->{has_descendant}->{significant} = 0 |
| 967 |
|
if not $sig_flag; |
| 968 |
|
}}; |
| 969 |
|
last TP; |
| 970 |
|
} else { |
| 971 |
|
unshift @$sib, @{$node->child_nodes}; |
| 972 |
|
push @$new_todos, {type => 'element-attributes', node => $node}; |
| 973 |
|
last TP; |
| 974 |
|
} |
| 975 |
} |
} |
| 976 |
if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) { |
if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) { |
| 977 |
if ($node->has_attribute_ns (undef, 'src')) { |
if ($node->has_attribute_ns (undef, 'src')) { |
| 992 |
last CN; |
last CN; |
| 993 |
} |
} |
| 994 |
} elsif ($cnt == 3 or $cnt == 4) { |
} elsif ($cnt == 3 or $cnt == 4) { |
| 995 |
if ($cn->data =~ /[^\x09-\x0D\x20]/) { |
if ($cn->data =~ /[^\x09\x0A\x0C\x0D\x20]/) { |
| 996 |
last CN; |
last CN; |
| 997 |
} |
} |
| 998 |
} |
} |
| 999 |
} # CN |
} # CN |
| 1000 |
unshift @$sib, @cn; |
unshift @$sib, @cn; |
| 1001 |
} |
} |
| 1002 |
|
} elsif ($node_ns eq $HTML_NS and $node_ln eq 'object') { |
| 1003 |
|
my @cn = @{$node->child_nodes}; |
| 1004 |
|
CN: while (@cn) { |
| 1005 |
|
my $cn = shift @cn; |
| 1006 |
|
my $cnt = $cn->node_type; |
| 1007 |
|
if ($cnt == 1) { |
| 1008 |
|
my $cn_nsuri = $cn->namespace_uri; |
| 1009 |
|
$cn_nsuri = '' unless defined $cn_nsuri; |
| 1010 |
|
if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'param') { |
| 1011 |
|
# |
| 1012 |
|
} else { |
| 1013 |
|
last CN; |
| 1014 |
|
} |
| 1015 |
|
} elsif ($cnt == 3 or $cnt == 4) { |
| 1016 |
|
if ($cn->data =~ /[^\x09\x0A\x0C\x0D\x20]/) { |
| 1017 |
|
last CN; |
| 1018 |
|
} |
| 1019 |
|
} |
| 1020 |
|
} # CN |
| 1021 |
|
unshift @$sib, @cn; |
| 1022 |
} |
} |
| 1023 |
push @$new_todos, {type => 'element', node => $node}; |
push @$new_todos, {type => 'element', node => $node}; |
| 1024 |
} # TP |
} # TP |
| 1030 |
return ($sib, $new_todos); |
return ($sib, $new_todos); |
| 1031 |
} # _check_get_children |
} # _check_get_children |
| 1032 |
|
|
| 1033 |
|
=head1 LICENSE |
| 1034 |
|
|
| 1035 |
|
Copyright 2007-2008 Wakaba <[email protected]> |
| 1036 |
|
|
| 1037 |
|
This library is free software; you can redistribute it |
| 1038 |
|
and/or modify it under the same terms as Perl itself. |
| 1039 |
|
|
| 1040 |
|
=cut |
| 1041 |
|
|
| 1042 |
1; |
1; |
| 1043 |
# $Date$ |
# $Date$ |