/[suikacvs]/markup/html/whatpm/Whatpm/ContentChecker.pm
Suika

Contents of /markup/html/whatpm/Whatpm/ContentChecker.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.84 - (show annotations) (download)
Fri Aug 15 14:13:42 2008 UTC (18 years ago) by wakaba
Branch: MAIN
Changes since 1.83: +8 -3 lines
++ whatpm/Whatpm/ChangeLog	15 Aug 2008 14:12:36 -0000
	* ContentChecker.pm: Support for RDF/XML error levels.

	* HTMLTable.pm, RDFXML.pm: Support for new style of error level
	specifying.  Error types are revised.

2008-08-15  Wakaba  <wakaba@suika.fam.cx>

++ whatpm/Whatpm/ContentChecker/ChangeLog	15 Aug 2008 14:13:40 -0000
	* HTML.pm: Mark HTML4's "fact"-level errors as such.  Typo fixed.

2008-08-15  Wakaba  <wakaba@suika.fam.cx>

1 package Whatpm::ContentChecker;
2 use strict;
3 our $VERSION=do{my @r=(q$Revision: 1.83 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
4
5 require Whatpm::URIChecker;
6
7 ## ISSUE: How XML and XML Namespaces conformance can (or cannot)
8 ## 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>;
27 my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>;
28
29 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 $XML_NS => {
55 space => sub {
56 my ($self, $attr) = @_;
57 my $value = $attr->value;
58 if ($value eq 'default' or $value eq 'preserve') {
59 #
60 } else {
61 ## NOTE: An XML "error"
62 $self->{onerror}->(node => $attr, level => $self->{level}->{xml_error},
63 type => 'invalid attribute value');
64 }
65 },
66 lang => sub {
67 my ($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 => $attr);
75 });
76 }
77
78 ## NOTE: "The values of the attribute are language identifiers
79 ## as defined by [IETF RFC 3066], Tags for the Identification
80 ## of Languages, or its successor; in addition, the empty string
81 ## may be specified." ("may" in lower case)
82 ## NOTE: Is an RFC 3066-valid (but RFC 4647-invalid) language tag
83 ## allowed today?
84
85 ## TODO: test data
86
87 if ($attr->owner_document->manakai_is_html) { # MUST NOT
88 $self->{onerror}->(node => $attr, type => 'in HTML:xml:lang',
89 level => $self->{level}->{must});
90 ## TODO: Test data...
91 }
92 },
93 base => sub {
94 my ($self, $attr) = @_;
95 my $value = $attr->value;
96 if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters?
97 $self->{onerror}->(node => $attr,
98 type => 'invalid attribute value',
99 level => $self->{level}->{fact}, ## TODO: correct?
100 );
101 }
102 ## NOTE: Conformance to URI standard is not checked since there is
103 ## no author requirement on conformance in the XML Base specification.
104 },
105 id => sub {
106 my ($self, $attr) = @_;
107 my $value = $attr->value;
108 $value =~ s/[\x09\x0A\x0D\x20]+/ /g;
109 $value =~ s/^\x20//;
110 $value =~ s/\x20$//;
111 ## TODO: NCName in XML 1.0 or 1.1
112 ## TODO: declared type is ID?
113 if ($self->{id}->{$value}) {
114 $self->{onerror}->(node => $attr,
115 type => 'duplicate ID',
116 level => $self->{level}->{xml_id_error});
117 push @{$self->{id}->{$value}}, $attr;
118 } else {
119 $self->{id}->{$value} = [$attr];
120 }
121 },
122 },
123 $XMLNS_NS => {
124 '' => sub {
125 my ($self, $attr) = @_;
126 my $ln = $attr->manakai_local_name;
127 my $value = $attr->value;
128 if ($value eq $XML_NS and $ln ne 'xml') {
129 $self->{onerror}
130 ->(node => $attr,
131 type => 'Reserved Prefixes and Namespace Names:Name',
132 text => $value,
133 level => $self->{level}->{nc});
134 } elsif ($value eq $XMLNS_NS) {
135 $self->{onerror}
136 ->(node => $attr,
137 type => 'Reserved Prefixes and Namespace Names:Name',
138 text => $value,
139 level => $self->{level}->{nc});
140 }
141 if ($ln eq 'xml' and $value ne $XML_NS) {
142 $self->{onerror}
143 ->(node => $attr,
144 type => 'Reserved Prefixes and Namespace Names:Prefix',
145 text => $ln,
146 level => $self->{level}->{nc});
147 } elsif ($ln eq 'xmlns') {
148 $self->{onerror}
149 ->(node => $attr,
150 type => 'Reserved Prefixes and Namespace Names:Prefix',
151 text => $ln,
152 level => $self->{level}->{nc});
153 }
154 ## TODO: If XML 1.0 and empty
155 },
156 xmlns => sub {
157 my ($self, $attr) = @_;
158 ## TODO: In XML 1.0, URI reference [RFC 3986] or an empty string
159 ## TODO: In XML 1.1, IRI reference [RFC 3987] or an empty string
160 ## TODO: relative references are deprecated
161 my $value = $attr->value;
162 if ($value eq $XML_NS) {
163 $self->{onerror}
164 ->(node => $attr,
165 type => 'Reserved Prefixes and Namespace Names:Name',
166 text => $value,
167 level => $self->{level}->{nc});
168 } elsif ($value eq $XMLNS_NS) {
169 $self->{onerror}
170 ->(node => $attr,
171 type => 'Reserved Prefixes and Namespace Names:Name',
172 text => $value,
173 level => $self->{level}->{nc});
174 }
175 },
176 },
177 };
178
179 ## ISSUE: Should we really allow these attributes?
180 $AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space};
181 $AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang};
182 $AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base};
183 $AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id};
184
185 our $AttrStatus;
186
187 for (qw/space lang base id/) {
188 $AttrStatus->{$XML_NS}->{$_} = FEATURE_STATUS_REC | FEATURE_ALLOWED;
189 $AttrStatus->{''}->{"xml:$_"} = FEATURE_STATUS_REC | FEATURE_ALLOWED;
190 ## XML 1.0: FEATURE_STATUS_CR
191 ## XML 1.1: FEATURE_STATUS_REC
192 ## XML Namespaces 1.0: FEATURE_STATUS_CR
193 ## XML Namespaces 1.1: FEATURE_STATUS_REC
194 ## XML Base: FEATURE_STATUS_REC
195 ## xml:id: FEATURE_STATUS_REC
196 }
197
198 $AttrStatus->{$XMLNS_NS}->{''} = FEATURE_STATUS_REC | FEATURE_ALLOWED;
199
200 ## TODO: xsi:schemaLocation for XHTML2 support (very, very low priority)
201
202 our %AnyChecker = (
203 check_start => sub { },
204 check_attrs => sub {
205 my ($self, $item, $element_state) = @_;
206 for my $attr (@{$item->{node}->attributes}) {
207 my $attr_ns = $attr->namespace_uri;
208 $attr_ns = '' unless defined $attr_ns;
209 my $attr_ln = $attr->manakai_local_name;
210
211 load_ns_module ($attr_ns);
212
213 my $checker = $AttrChecker->{$attr_ns}->{$attr_ln}
214 || $AttrChecker->{$attr_ns}->{''};
215 my $status = $AttrStatus->{$attr_ns}->{$attr_ln}
216 || $AttrStatus->{$attr_ns}->{''};
217 if (not defined $status) {
218 $status = FEATURE_ALLOWED;
219 ## NOTE: FEATURE_ALLOWED for all attributes, since the element
220 ## is not supported and therefore "attribute not defined" error
221 ## should not raised (too verbose) and global attributes should be
222 ## allowed anyway (if a global attribute has its specified creteria
223 ## for where it may be specified, then it should be checked in it's
224 ## checker function).
225 }
226 if ($checker) {
227 $checker->($self, $attr);
228 } else {
229 $self->{onerror}->(node => $attr,
230 type => 'unknown attribute',
231 level => $self->{level}->{uncertain});
232 }
233 $self->_attr_status_info ($attr, $status);
234 }
235 },
236 check_child_element => sub {
237 my ($self, $item, $child_el, $child_nsuri, $child_ln,
238 $child_is_transparent, $element_state) = @_;
239 if ($self->{minus_elements}->{$child_nsuri}->{$child_ln}) {
240 $self->{onerror}->(node => $child_el,
241 type => 'element not allowed:minus',
242 level => $self->{level}->{must});
243 } elsif ($self->{plus_elements}->{$child_nsuri}->{$child_ln}) {
244 #
245 } else {
246 #
247 }
248 },
249 check_child_text => sub { },
250 check_end => sub {
251 my ($self, $item, $element_state) = @_;
252 ## NOTE: There is a modified copy of the code below for |html:ruby|.
253 if ($element_state->{has_significant}) {
254 $item->{real_parent_state}->{has_significant} = 1;
255 }
256 },
257 );
258
259 our $ElementDefault = {
260 %AnyChecker,
261 status => FEATURE_ALLOWED,
262 ## NOTE: No "element not defined" error - it is not supported anyway.
263 check_start => sub {
264 my ($self, $item, $element_state) = @_;
265 $self->{onerror}->(node => $item->{node},
266 type => 'unknown element',
267 level => $self->{level}->{uncertain});
268 },
269 };
270
271 our $HTMLEmbeddedContent = {
272 ## NOTE: All embedded content is also phrasing content.
273 $HTML_NS => {
274 img => 1, iframe => 1, embed => 1, object => 1, video => 1, audio => 1,
275 canvas => 1,
276 },
277 q<http://www.w3.org/1998/Math/MathML> => {math => 1},
278 q<http://www.w3.org/2000/svg> => {svg => 1},
279 ## NOTE: Foreign elements with content (but no metadata) are
280 ## embedded content.
281 };
282
283 my $HTMLTransparentElements = {
284 $HTML_NS => {qw/ins 1 del 1 font 1 noscript 1 canvas 1/},
285 ## NOTE: |html:noscript| is transparent if scripting is disabled
286 ## and not in |head|.
287 };
288
289 my $HTMLSemiTransparentElements = {
290 $HTML_NS => {object => 1, video => 1, audio => 1},
291 };
292
293 our $Element = {};
294
295 $Element->{q<http://www.w3.org/1999/02/22-rdf-syntax-ns#>}->{RDF} = {
296 %AnyChecker,
297 status => FEATURE_STATUS_REC | FEATURE_ALLOWED,
298 is_root => 1, ## ISSUE: Not explicitly allowed for non application/rdf+xml
299 check_start => sub {
300 my ($self, $item, $element_state) = @_;
301 my $triple = [];
302 push @{$self->{return}->{rdf}}, [$item->{node}, $triple];
303 require Whatpm::RDFXML;
304 my $rdf = Whatpm::RDFXML->new;
305 ## TODO: Should we make bnodeid unique in a document?
306 $rdf->{onerror} = $self->{onerror};
307 $rdf->{level} = $self->{level};
308 $rdf->{ontriple} = sub {
309 my %opt = @_;
310 push @$triple,
311 [$opt{node}, $opt{subject}, $opt{predicate}, $opt{object}];
312 if (defined $opt{id}) {
313 push @$triple,
314 [$opt{node},
315 $opt{id},
316 {uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#subject>},
317 $opt{subject}];
318 push @$triple,
319 [$opt{node},
320 $opt{id},
321 {uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate>},
322 $opt{predicate}];
323 push @$triple,
324 [$opt{node},
325 $opt{id},
326 {uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#object>},
327 $opt{object}];
328 push @$triple,
329 [$opt{node},
330 $opt{id},
331 {uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>},
332 {uri => q<http://www.w3.org/1999/02/22-rdf-syntax-ns#Statement>}];
333 }
334 };
335 $rdf->convert_rdf_element ($item->{node});
336 },
337 };
338
339 my $default_error_level = {
340 must => 'm',
341 should => 's',
342 warn => 'w',
343 good => 'w',
344 info => 'i',
345 uncertain => 'u',
346
347 html4_fact => 'm',
348 xml_error => 'm', ## TODO: correct?
349 nc => 'm', ## XML Namespace Constraints ## TODO: correct?
350
351 rdf_fact => 'm',
352 rdf_grammer => 'm',
353 rdf_lc_must => 'm',
354 };
355
356 sub check_document ($$$;$) {
357 my ($self, $doc, $onerror, $onsubdoc) = @_;
358 $self = bless {}, $self unless ref $self;
359 $self->{onerror} = $onerror;
360 $self->{onsubdoc} = $onsubdoc || sub {
361 warn "A subdocument is not conformance-checked";
362 };
363
364 $self->{level} ||= $default_error_level;
365
366 ## TODO: If application/rdf+xml, RDF/XML mode should be invoked.
367
368 my $docel = $doc->document_element;
369 unless (defined $docel) {
370 ## ISSUE: Should we check content of Document node?
371 $onerror->(node => $doc, type => 'no document element',
372 level => $self->{level}->{must});
373 ## ISSUE: Is this non-conforming (to what spec)? Or just a warning?
374 return {
375 class => {},
376 id => {}, table => [], term => {},
377 };
378 }
379
380 ## ISSUE: Unexpanded entity references and HTML5 conformance
381
382 my $docel_nsuri = $docel->namespace_uri;
383 $docel_nsuri = '' unless defined $docel_nsuri;
384 load_ns_module ($docel_nsuri);
385 my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} ||
386 $Element->{$docel_nsuri}->{''} ||
387 $ElementDefault;
388 if ($docel_def->{is_root}) {
389 #
390 } elsif ($docel_def->{is_xml_root}) {
391 unless ($doc->manakai_is_html) {
392 #
393 } else {
394 $onerror->(node => $docel, type => 'element not allowed:root:xml',
395 level => $self->{level}->{must});
396 }
397 } else {
398 $onerror->(node => $docel, type => 'element not allowed:root',
399 level => $self->{level}->{must});
400 }
401
402 ## TODO: Check for other items other than document element
403 ## (second (errorous) element, text nodes, PI nodes, doctype nodes)
404
405 my $return = $self->check_element ($docel, $onerror, $onsubdoc);
406
407 ## TODO: Test for these checks are necessary.
408 my $charset_name = $doc->input_encoding;
409 if (defined $charset_name) {
410 require Message::Charset::Info;
411 my $charset = $Message::Charset::Info::IANACharset->{$charset_name};
412
413 if ($doc->manakai_is_html) {
414 if (not $doc->manakai_has_bom and
415 not defined $doc->manakai_charset) {
416 unless ($charset->{is_html_ascii_superset}) {
417 $onerror->(node => $doc, level => $self->{level}->{must},
418 type => 'non ascii superset',
419 text => $charset_name);
420 }
421
422 if (not $self->{has_charset} and ## TODO: This does not work now.
423 not $charset->{iana_names}->{'us-ascii'}) {
424 $onerror->(node => $doc, level => $self->{level}->{must},
425 type => 'no character encoding declaration',
426 text => $charset_name);
427 }
428 }
429
430 if ($charset->{iana_names}->{'utf-8'}) {
431 #
432 } elsif ($charset->{iana_names}->{'jis_x0212-1990'} or
433 $charset->{iana_names}->{'x-jis0208'} or
434 $charset->{iana_names}->{'utf-32'} or ## ISSUE: UTF-32BE? UTF-32LE?
435 $charset->{is_ebcdic_based}) {
436 $onerror->(node => $doc,
437 type => 'bad character encoding',
438 text => $charset_name,
439 level => $self->{level}->{should},
440 layer => 'encode');
441 } elsif ($charset->{iana_names}->{'cesu-8'} or
442 $charset->{iana_names}->{'utf-8'} or ## ISSUE: UNICODE-1-1-UTF-7?
443 $charset->{iana_names}->{'bocu-1'} or
444 $charset->{iana_names}->{'scsu'}) {
445 $onerror->(node => $doc,
446 type => 'disallowed character encoding',
447 text => $charset_name,
448 level => $self->{level}->{must},
449 layer => 'encode');
450 } else {
451 $onerror->(node => $doc,
452 type => 'non-utf-8 character encoding',
453 text => $charset_name,
454 level => $self->{level}->{good},
455 layer => 'encode');
456 }
457 }
458 } elsif ($doc->manakai_is_html) {
459 ## NOTE: MUST and SHOULD requirements above cannot be tested,
460 ## since the document has no input charset encoding information.
461 $onerror->(node => $doc,
462 type => 'character encoding unchecked',
463 level => $self->{level}->{info},
464 layer => 'encode');
465 }
466
467 return $return;
468 } # check_document
469
470 ## Check an element. The element is checked as if it is an orphan node (i.e.
471 ## an element without a parent node).
472 sub check_element ($$$;$) {
473 my ($self, $el, $onerror, $onsubdoc) = @_;
474 $self = bless {}, $self unless ref $self;
475 $self->{onerror} = $onerror;
476 $self->{onsubdoc} = $onsubdoc || sub {
477 warn "A subdocument is not conformance-checked";
478 };
479
480 $self->{level} ||= $default_error_level;
481
482 $self->{plus_elements} = {};
483 $self->{minus_elements} = {};
484 $self->{id} = {};
485 $self->{term} = {};
486 $self->{usemap} = [];
487 $self->{ref} = []; # datetemplate data references
488 $self->{template} = []; # datatemplate template references
489 $self->{contextmenu} = [];
490 $self->{map} = {};
491 $self->{menu} = {};
492 $self->{has_link_type} = {};
493 $self->{flag} = {};
494 #$self->{has_uri_attr};
495 #$self->{has_hyperlink_element};
496 #$self->{has_charset};
497 #$self->{has_base};
498 $self->{return} = {
499 class => {},
500 id => $self->{id},
501 table => [], # table objects returned by Whatpm::HTMLTable
502 term => $self->{term},
503 uri => {}, # URIs other than those in RDF triples
504 ## TODO: xmlns="", SYSTEM "", atom:* src="", xml:base=""
505 rdf => [],
506 };
507
508 my @item = ({type => 'element', node => $el, parent_state => {}});
509 $item[-1]->{real_parent_state} = $item[-1]->{parent_state};
510 while (@item) {
511 my $item = shift @item;
512 if (ref $item eq 'ARRAY') {
513 my $code = shift @$item;
514 next unless $code;## TODO: temp.
515 $code->(@$item);
516 } elsif ($item->{type} eq 'element') {
517 my $el_nsuri = $item->{node}->namespace_uri;
518 $el_nsuri = '' unless defined $el_nsuri;
519 my $el_ln = $item->{node}->manakai_local_name;
520
521 load_ns_module ($el_nsuri);
522
523 my $element_state = {};
524 my $eldef = $Element->{$el_nsuri}->{$el_ln} ||
525 $Element->{$el_nsuri}->{''} ||
526 $ElementDefault;
527 my $content_def = $item->{transparent}
528 ? $item->{parent_def} || $eldef : $eldef;
529 my $content_state = $item->{transparent}
530 ? $item->{parent_def}
531 ? $item->{parent_state} || $element_state : $element_state
532 : $element_state;
533
534 unless ($eldef->{status} & FEATURE_STATUS_REC) {
535 my $status = $eldef->{status} & FEATURE_STATUS_CR ? 'cr' :
536 $eldef->{status} & FEATURE_STATUS_LC ? 'lc' :
537 $eldef->{status} & FEATURE_STATUS_WD ? 'wd' : 'non-standard';
538 $self->{onerror}->(node => $item->{node},
539 type => 'status:'.$status.':element',
540 level => $self->{level}->{info});
541 }
542 if (not ($eldef->{status} & FEATURE_ALLOWED)) {
543 $self->{onerror}->(node => $item->{node},
544 type => 'element not defined',
545 level => $self->{level}->{must});
546 } elsif ($eldef->{status} & FEATURE_DEPRECATED_SHOULD) {
547 $self->{onerror}->(node => $item->{node},
548 type => 'deprecated:element',
549 level => $self->{level}->{should});
550 } elsif ($eldef->{status} & FEATURE_DEPRECATED_INFO) {
551 $self->{onerror}->(node => $item->{node},
552 type => 'deprecated:element',
553 level => $self->{level}->{info});
554 }
555
556 my @new_item;
557 push @new_item, [$eldef->{check_start}, $self, $item, $element_state];
558 push @new_item, [$eldef->{check_attrs}, $self, $item, $element_state];
559
560 my @child = @{$item->{node}->child_nodes};
561 while (@child) {
562 my $child = shift @child;
563 my $child_nt = $child->node_type;
564 if ($child_nt == 1) { # ELEMENT_NODE
565 my $child_nsuri = $child->namespace_uri;
566 $child_nsuri = '' unless defined $child_nsuri;
567 my $child_ln = $child->manakai_local_name;
568 if ($HTMLTransparentElements->{$child_nsuri}->{$child_ln} and
569 not (($self->{flag}->{in_head} or
570 ($el_nsuri eq $HTML_NS and $el_ln eq 'head')) and
571 $child_nsuri eq $HTML_NS and $child_ln eq 'noscript')) {
572 push @new_item, [$content_def->{check_child_element},
573 $self, $item, $child,
574 $child_nsuri, $child_ln, 1,
575 $content_state, $element_state];
576 push @new_item, {type => 'element', node => $child,
577 parent_state => $content_state,
578 parent_def => $content_def,
579 real_parent_state => $element_state,
580 transparent => 1};
581 } else {
582 if ($item->{parent_def} and # has parent
583 $el_nsuri eq $HTML_NS) { ## $HTMLSemiTransparentElements
584 if ($el_ln eq 'object') {
585 if ($self->{plus_elements}->{$child_nsuri}->{$child_ln}) {
586 #
587 } elsif ($child_nsuri eq $HTML_NS and $child_ln eq 'param') {
588 #
589 } else {
590 $content_def = $item->{parent_def} || $content_def;
591 $content_state = $item->{parent_state} || $content_state;
592 }
593 } elsif ($el_ln eq 'video' or $el_ln eq 'audio') {
594 if ($self->{plus_elements}->{$child_nsuri}->{$child_ln}) {
595 #
596 } elsif ($child_nsuri eq $HTML_NS and $child_ln eq 'source') {
597 $element_state->{has_source} = 1;
598 } else {
599 $content_def = $item->{parent_def} || $content_def;
600 $content_state = $item->{parent_state} || $content_state;
601 }
602 }
603 }
604
605 push @new_item, [$content_def->{check_child_element},
606 $self, $item, $child,
607 $child_nsuri, $child_ln,
608 $HTMLSemiTransparentElements
609 ->{$child_nsuri}->{$child_ln},
610 $content_state, $element_state];
611 push @new_item, {type => 'element', node => $child,
612 parent_def => $content_def,
613 real_parent_state => $element_state,
614 parent_state => $content_state};
615 }
616
617 if ($HTMLEmbeddedContent->{$child_nsuri}->{$child_ln}) {
618 $element_state->{has_significant} = 1;
619 }
620 } elsif ($child_nt == 3 or # TEXT_NODE
621 $child_nt == 4) { # CDATA_SECTION_NODE
622 my $has_significant = ($child->data =~ /[^\x09-\x0D\x20]/);
623 push @new_item, [$content_def->{check_child_text},
624 $self, $item, $child, $has_significant,
625 $content_state, $element_state];
626 $element_state->{has_significant} ||= $has_significant;
627 if ($has_significant and
628 $HTMLSemiTransparentElements->{$el_nsuri}->{$el_ln}) {
629 $content_def = $item->{parent_def} || $content_def;
630 }
631 } elsif ($child_nt == 5) { # ENTITY_REFERENCE_NODE
632 push @child, @{$child->child_nodes};
633 }
634 ## TODO: PI_NODE
635 ## TODO: Unknown node type
636 }
637
638 push @new_item, [$eldef->{check_end}, $self, $item, $element_state];
639
640 unshift @item, @new_item;
641 } else {
642 die "$0: Internal error: Unsupported checking action type |$item->{type}|";
643 }
644 }
645
646 for (@{$self->{template}}) {
647 ## TODO: If the document is an XML document, ...
648 ## NOTE: If the document is an HTML document:
649 ## ISSUE: We need to percent-decode?
650 F: {
651 if ($self->{id}->{$_->[0]}) {
652 my $el = $self->{id}->{$_->[0]}->[0]->owner_element;
653 if ($el->node_type == 1 and # ELEMENT_NODE
654 $el->manakai_local_name eq 'datatemplate') {
655 my $nsuri = $el->namespace_uri;
656 if (defined $nsuri and $nsuri eq $HTML_NS) {
657 if ($el eq $_->[1]->owner_element) {
658 $self->{onerror}->(node => $_->[1],
659 type => 'fragment points itself',
660 level => $self->{level}->{must});
661 }
662
663 last F;
664 }
665 }
666 }
667 ## TODO: Should we raise a "fragment points nothing" error instead
668 ## if the fragment identifier identifies no element?
669
670 $self->{onerror}->(node => $_->[1], type => 'template:not template',
671 level => $self->{level}->{must});
672 } # F
673 }
674
675 for (@{$self->{ref}}) {
676 ## TOOD: If XML
677 ## NOTE: If it is an HTML document:
678 if ($_->[0] eq '') {
679 ## NOTE: It points the top of the document.
680 } elsif ($self->{id}->{$_->[0]}) {
681 if ($self->{id}->{$_->[0]}->[0]->owner_element
682 eq $_->[1]->owner_element) {
683 $self->{onerror}->(node => $_->[1], type => 'fragment points itself',
684 level => $self->{level}->{must});
685 }
686 } else {
687 $self->{onerror}->(node => $_->[1], type => 'fragment points nothing',
688 level => $self->{level}->{must});
689 }
690 }
691
692 ## TODO: Maybe we should have $document->manakai_get_by_fragment or something
693
694 for (@{$self->{usemap}}) {
695 unless ($self->{map}->{$_->[0]}) {
696 $self->{onerror}->(node => $_->[1], type => 'no referenced map',
697 level => $self->{level}->{must});
698 }
699 }
700
701 for (@{$self->{contextmenu}}) {
702 unless ($self->{menu}->{$_->[0]}) {
703 $self->{onerror}->(node => $_->[1], type => 'no referenced menu',
704 level => $self->{level}->{must});
705 }
706 }
707
708 delete $self->{plus_elements};
709 delete $self->{minus_elements};
710 delete $self->{onerror};
711 delete $self->{id};
712 delete $self->{usemap};
713 delete $self->{ref};
714 delete $self->{template};
715 delete $self->{map};
716 return $self->{return};
717 } # check_element
718
719 sub _add_minus_elements ($$@) {
720 my $self = shift;
721 my $element_state = shift;
722 for my $elements (@_) {
723 for my $nsuri (keys %$elements) {
724 for my $ln (keys %{$elements->{$nsuri}}) {
725 unless ($self->{minus_elements}->{$nsuri}->{$ln}) {
726 $element_state->{minus_elements_original}->{$nsuri}->{$ln} = 0;
727 $self->{minus_elements}->{$nsuri}->{$ln} = 1;
728 }
729 }
730 }
731 }
732 } # _add_minus_elements
733
734 sub _remove_minus_elements ($$) {
735 my $self = shift;
736 my $element_state = shift;
737 for my $nsuri (keys %{$element_state->{minus_elements_original}}) {
738 for my $ln (keys %{$element_state->{minus_elements_original}->{$nsuri}}) {
739 delete $self->{minus_elements}->{$nsuri}->{$ln};
740 }
741 }
742 } # _remove_minus_elements
743
744 sub _add_plus_elements ($$@) {
745 my $self = shift;
746 my $element_state = shift;
747 for my $elements (@_) {
748 for my $nsuri (keys %$elements) {
749 for my $ln (keys %{$elements->{$nsuri}}) {
750 unless ($self->{plus_elements}->{$nsuri}->{$ln}) {
751 $element_state->{plus_elements_original}->{$nsuri}->{$ln} = 0;
752 $self->{plus_elements}->{$nsuri}->{$ln} = 1;
753 }
754 }
755 }
756 }
757 } # _add_plus_elements
758
759 sub _remove_plus_elements ($$) {
760 my $self = shift;
761 my $element_state = shift;
762 for my $nsuri (keys %{$element_state->{plus_elements_original}}) {
763 for my $ln (keys %{$element_state->{plus_elements_original}->{$nsuri}}) {
764 delete $self->{plus_elements}->{$nsuri}->{$ln};
765 }
766 }
767 } # _remove_plus_elements
768
769 sub _attr_status_info ($$$) {
770 my ($self, $attr, $status_code) = @_;
771
772 if (not ($status_code & FEATURE_ALLOWED)) {
773 $self->{onerror}->(node => $attr,
774 type => 'attribute not defined',
775 level => $self->{level}->{must});
776 } elsif ($status_code & FEATURE_DEPRECATED_SHOULD) {
777 $self->{onerror}->(node => $attr,
778 type => 'deprecated:attr',
779 level => $self->{level}->{should});
780 } elsif ($status_code & FEATURE_DEPRECATED_INFO) {
781 $self->{onerror}->(node => $attr,
782 type => 'deprecated:attr',
783 level => $self->{level}->{info});
784 }
785
786 my $status;
787 if ($status_code & FEATURE_STATUS_REC) {
788 return;
789 } elsif ($status_code & FEATURE_STATUS_CR) {
790 $status = 'cr';
791 } elsif ($status_code & FEATURE_STATUS_LC) {
792 $status = 'lc';
793 } elsif ($status_code & FEATURE_STATUS_WD) {
794 $status = 'wd';
795 } else {
796 $status = 'non-standard';
797 }
798 $self->{onerror}->(node => $attr,
799 type => 'status:'.$status.':attr',
800 level => $self->{level}->{info});
801 } # _attr_status_info
802
803 sub _add_minuses ($@) {
804 my $self = shift;
805 my $r = {};
806 for my $list (@_) {
807 for my $ns (keys %$list) {
808 for my $ln (keys %{$list->{$ns}}) {
809 unless ($self->{minuses}->{$ns}->{$ln}) {
810 $self->{minuses}->{$ns}->{$ln} = 1;
811 $r->{$ns}->{$ln} = 1;
812 }
813 }
814 }
815 }
816 return {type => 'plus', list => $r};
817 } # _add_minuses
818
819 sub _add_pluses ($@) {
820 my $self = shift;
821 my $r = {};
822 for my $list (@_) {
823 for my $ns (keys %$list) {
824 for my $ln (keys %{$list->{$ns}}) {
825 unless ($self->{pluses}->{$ns}->{$ln}) {
826 $self->{pluses}->{$ns}->{$ln} = 1;
827 $r->{$ns}->{$ln} = 1;
828 }
829 }
830 }
831 }
832 return {type => 'minus', list => $r};
833 } # _add_pluses
834
835 sub _remove_minuses ($$) {
836 my ($self, $todo) = @_;
837 if ($todo->{type} eq 'minus') {
838 for my $ns (keys %{$todo->{list}}) {
839 for my $ln (keys %{$todo->{list}->{$ns}}) {
840 delete $self->{pluses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
841 }
842 }
843 } elsif ($todo->{type} eq 'plus') {
844 for my $ns (keys %{$todo->{list}}) {
845 for my $ln (keys %{$todo->{list}->{$ns}}) {
846 delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln};
847 }
848 }
849 } else {
850 die "$0: Unknown +- type: $todo->{type}";
851 }
852 1;
853 } # _remove_minuses
854
855 ## NOTE: Priority for "minuses" and "pluses" are currently left
856 ## undefined and implemented inconsistently; it is not a problem for
857 ## now, since no element belongs to both lists.
858
859 sub _check_get_children ($$$) {
860 my ($self, $node, $parent_todo) = @_;
861 my $new_todos = [];
862 my $sib = [];
863 TP: {
864 my $node_ns = $node->namespace_uri;
865 $node_ns = '' unless defined $node_ns;
866 my $node_ln = $node->manakai_local_name;
867 if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) {
868 if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') {
869 if ($parent_todo->{flag}->{in_head}) {
870 #
871 } else {
872 my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}});
873 push @$sib, $end;
874
875 unshift @$sib, @{$node->child_nodes};
876 push @$new_todos, {type => 'element-attributes', node => $node};
877 last TP;
878 }
879 } elsif ($node_ns eq $HTML_NS and $node_ln eq 'del') {
880 my $sig_flag = $parent_todo->{flag}->{has_descendant}->{significant};
881 unshift @$sib, @{$node->child_nodes};
882 push @$new_todos, {type => 'element-attributes', node => $node};
883 push @$new_todos,
884 {type => 'code',
885 code => sub {
886 $parent_todo->{flag}->{has_descendant}->{significant} = 0
887 if not $sig_flag;
888 }};
889 last TP;
890 } else {
891 unshift @$sib, @{$node->child_nodes};
892 push @$new_todos, {type => 'element-attributes', node => $node};
893 last TP;
894 }
895 }
896 if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) {
897 if ($node->has_attribute_ns (undef, 'src')) {
898 unshift @$sib, @{$node->child_nodes};
899 push @$new_todos, {type => 'element-attributes', node => $node};
900 last TP;
901 } else {
902 my @cn = @{$node->child_nodes};
903 CN: while (@cn) {
904 my $cn = shift @cn;
905 my $cnt = $cn->node_type;
906 if ($cnt == 1) {
907 my $cn_nsuri = $cn->namespace_uri;
908 $cn_nsuri = '' unless defined $cn_nsuri;
909 if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') {
910 #
911 } else {
912 last CN;
913 }
914 } elsif ($cnt == 3 or $cnt == 4) {
915 if ($cn->data =~ /[^\x09-\x0D\x20]/) {
916 last CN;
917 }
918 }
919 } # CN
920 unshift @$sib, @cn;
921 }
922 } elsif ($node_ns eq $HTML_NS and $node_ln eq 'object') {
923 my @cn = @{$node->child_nodes};
924 CN: while (@cn) {
925 my $cn = shift @cn;
926 my $cnt = $cn->node_type;
927 if ($cnt == 1) {
928 my $cn_nsuri = $cn->namespace_uri;
929 $cn_nsuri = '' unless defined $cn_nsuri;
930 if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'param') {
931 #
932 } else {
933 last CN;
934 }
935 } elsif ($cnt == 3 or $cnt == 4) {
936 if ($cn->data =~ /[^\x09-\x0D\x20]/) {
937 last CN;
938 }
939 }
940 } # CN
941 unshift @$sib, @cn;
942 }
943 push @$new_todos, {type => 'element', node => $node};
944 } # TP
945
946 for my $new_todo (@$new_todos) {
947 $new_todo->{flag} = {%{$parent_todo->{flag} or {}}};
948 }
949
950 return ($sib, $new_todos);
951 } # _check_get_children
952
953 =head1 LICENSE
954
955 Copyright 2007-2008 Wakaba <[email protected]>
956
957 This library is free software; you can redistribute it
958 and/or modify it under the same terms as Perl itself.
959
960 =cut
961
962 1;
963 # $Date: 2008/08/15 12:46:44 $

[email protected]
ViewVC Help
Powered by ViewVC 1.1.24