Parent Directory
| 
Revision Log
++ manakai/lib/Message/DOM/ChangeLog 15 Feb 2006 13:28:44 -0000 2006-02-15 Wakaba <wakaba@suika.fam.cx> * XMLParser.dis: Issues |xp:error-internal-predefined-entity| error if a predefined entity declaration references an external entity. Issues |xp:error-malformed-predefined-entity| error if a predefined entity declaration defines different thing than one as in XML specification. Issies |xp:warning-entity-declaration-ignored| warning if more than one entity declarations for an entity is found. (WFErrDef): Now all well-formedness errors are marked as |SEVERITY_FATAL_ERROR|. * XDoctype.dis (DocumentXDoctype.createDocumentTypeDefinition): Creates predefined entity declarations. (DocumentXDoctype.createNotation): New method. (DocumentTypeDefinition.notations): New attribute. (DocumentTypeDefinition.getNotationNode): New method. (DocumentTypeDefinition.setNotationNode): New method. (newNotationForTest): New code fragment. * Tree.dis (ManakaiDOMXML:ManakaiDOMNotations): Removed. (c:ManakaiDOMNotationMap): New class. (c:ManakaiDOMNotationMapArray): New class. (Document.allDeclarationsProcessed): New attribute. * DOMXML.dis (DocumentType.notations): Implemented. (Notation): Revised. (Notation.ownerDocumentTypeDefinition): New attribute.
| 1 | wakaba | 1.1 | Module: |
| 2 | @QName: MDOM|XMLParser | ||
| 3 | @Namespace: | ||
| 4 | http://suika.fam.cx/~wakaba/archive/2004/dom/xml-parser# | ||
| 5 | |||
| 6 | @FullName: | ||
| 7 | @@lang:en | ||
| 8 | @@@: XML Parser | ||
| 9 | |||
| 10 | @DISCore:author: DISCore|Wakaba | ||
| 11 | @License: license|Perl+MPL | ||
| 12 | @Date: | ||
| 13 | wakaba | 1.24 | $Date: 2006/02/13 17:03:22 $ |
| 14 | wakaba | 1.1 | |
| 15 | @DefaultFor: ManakaiDOM|ManakaiDOMLatest | ||
| 16 | |||
| 17 | @Require: | ||
| 18 | @@Module: | ||
| 19 | @@@QName: MDOM|DOMLS | ||
| 20 | @@@WithFor: ManakaiDOM|ManakaiDOMLatest | ||
| 21 | wakaba | 1.6 | @@Module: |
| 22 | @@@QName: MDOM|XDoctype | ||
| 23 | @@@WithFor: ManakaiDOM|ManakaiDOMLatest | ||
| 24 | wakaba | 1.1 | |
| 25 | Namespace: | ||
| 26 | wakaba | 1.8 | @c: |
| 27 | http://suika.fam.cx/~wakaba/archive/2004/8/18/dom-core# | ||
| 28 | wakaba | 1.11 | @cfg: |
| 29 | http://suika.fam.cx/www/2006/dom-config/ | ||
| 30 | wakaba | 1.6 | @d: |
| 31 | http://suika.fam.cx/~wakaba/archive/2004/dom/xdt# | ||
| 32 | wakaba | 1.1 | @dis: |
| 33 | http://suika.fam.cx/~wakaba/archive/2004/8/18/lang#dis-- | ||
| 34 | wakaba | 1.5 | @dtest: |
| 35 | http://suika.fam.cx/~wakaba/archive/2005/manakai/Util/DIS#Test/ | ||
| 36 | wakaba | 1.1 | @dx: |
| 37 | http://suika.fam.cx/~wakaba/archive/2005/manakai/Util/Error/DOMException# | ||
| 38 | @ecore: | ||
| 39 | http://suika.fam.cx/~wakaba/archive/2005/manakai/Util/Error/Core/ | ||
| 40 | @f: | ||
| 41 | http://suika.fam.cx/~wakaba/archive/2004/dom/feature# | ||
| 42 | wakaba | 1.11 | @fe: |
| 43 | http://suika.fam.cx/www/2006/feature/ | ||
| 44 | @gls: | ||
| 45 | http://suika.fam.cx/~wakaba/archive/2004/dom/gls# | ||
| 46 | wakaba | 1.1 | @idl: |
| 47 | http://suika.fam.cx/~wakaba/archive/2004/dis/IDL# | ||
| 48 | @infoset: | ||
| 49 | http://www.w3.org/2001/04/infoset# | ||
| 50 | @lang: | ||
| 51 | http://suika.fam.cx/~wakaba/archive/2004/8/18/lang# | ||
| 52 | @license: | ||
| 53 | http://suika.fam.cx/~wakaba/archive/2004/8/18/license# | ||
| 54 | @LSEV: | ||
| 55 | http://www.w3.org/2002/DOMLS | ||
| 56 | @ManakaiDOM: | ||
| 57 | http://suika.fam.cx/~wakaba/archive/2004/8/18/manakai-dom# | ||
| 58 | @ManakaiDOMLS: | ||
| 59 | http://suika.fam.cx/~wakaba/archive/2004/mdom-ls# | ||
| 60 | @MDOM: | ||
| 61 | http://suika.fam.cx/~wakaba/archive/2004/8/18/manakai-dom#ManakaiDOM. | ||
| 62 | @MDOMX: | ||
| 63 | http://suika.fam.cx/~wakaba/archive/2004/8/4/manakai-dom-exception# | ||
| 64 | wakaba | 1.11 | @mn: |
| 65 | http://suika.fam.cx/~wakaba/archive/2005/manakai/Util/ManakaiNode# | ||
| 66 | wakaba | 1.1 | @rdf: |
| 67 | http://www.w3.org/1999/02/22-rdf-syntax-ns# | ||
| 68 | @rdfs: | ||
| 69 | http://www.w3.org/2000/01/rdf-schema# | ||
| 70 | @t: | ||
| 71 | http://suika.fam.cx/~wakaba/archive/2004/dom/tree# | ||
| 72 | wakaba | 1.5 | @test: |
| 73 | http://suika.fam.cx/~wakaba/archive/2004/dis/Test# | ||
| 74 | wakaba | 1.8 | @x: |
| 75 | http://suika.fam.cx/~wakaba/archive/2004/dom/xml# | ||
| 76 | wakaba | 1.1 | @xml: |
| 77 | http://www.w3.org/XML/1998/namespace | ||
| 78 | wakaba | 1.9 | @xml-dtd: |
| 79 | http://www.w3.org/TR/REC-xml | ||
| 80 | wakaba | 1.1 | @xmlns: |
| 81 | http://www.w3.org/2000/xmlns/ | ||
| 82 | @xp: | ||
| 83 | http://suika.fam.cx/~wakaba/archive/2004/dom/xml-parser# | ||
| 84 | |||
| 85 | ## -- Features | ||
| 86 | |||
| 87 | ElementTypeBinding: | ||
| 88 | @Name: FeatureDef | ||
| 89 | @ElementType: | ||
| 90 | dis:ResourceDef | ||
| 91 | @ShadowContent: | ||
| 92 | @@rdf:type: f|Feature | ||
| 93 | @@For: =ManakaiDOM|all | ||
| 94 | |||
| 95 | ElementTypeBinding: | ||
| 96 | @Name: FeatureVerDef | ||
| 97 | @ElementType: | ||
| 98 | dis:ResourceDef | ||
| 99 | @ShadowContent: | ||
| 100 | @@rdf:type: f|Feature | ||
| 101 | |||
| 102 | ElementTypeBinding: | ||
| 103 | @Name: featureQName | ||
| 104 | @ElementType: | ||
| 105 | f:name | ||
| 106 | @ShadowContent: | ||
| 107 | @@ContentType: DISCore|QName | ||
| 108 | |||
| 109 | ResourceDef: | ||
| 110 | @QName: DOMString | ||
| 111 | @AliasFor: DOMMain|DOMString | ||
| 112 | @For: ManakaiDOM|DOM | ||
| 113 | |||
| 114 | ResourceDef: | ||
| 115 | @QName: Node | ||
| 116 | @AliasFor: t|Node | ||
| 117 | @For: ManakaiDOM|DOM | ||
| 118 | |||
| 119 | ResourceDef: | ||
| 120 | @QName: Element | ||
| 121 | @AliasFor: t|Element | ||
| 122 | @For: ManakaiDOM|DOM | ||
| 123 | |||
| 124 | ResourceDef: | ||
| 125 | @QName: Document | ||
| 126 | @AliasFor: t|Document | ||
| 127 | @For: ManakaiDOM|DOM | ||
| 128 | |||
| 129 | wakaba | 1.6 | ResourceDef: |
| 130 | @QName: DocumentXDoctype | ||
| 131 | @AliasFor: d|DocumentXDoctype | ||
| 132 | @For: ManakaiDOM|DOM | ||
| 133 | |||
| 134 | wakaba | 1.8 | ResourceDef: |
| 135 | @QName: DTDef | ||
| 136 | @AliasFor: d|DocumentTypeDefinition | ||
| 137 | @For: ManakaiDOM|DOM | ||
| 138 | |||
| 139 | ResourceDef: | ||
| 140 | @QName: ETDef | ||
| 141 | @AliasFor: d|ElementTypeDefinition | ||
| 142 | @For: ManakaiDOM|DOM | ||
| 143 | |||
| 144 | ResourceDef: | ||
| 145 | @QName: ATDef | ||
| 146 | @AliasFor: d|AttributeDefinition | ||
| 147 | @For: ManakaiDOM|DOM | ||
| 148 | |||
| 149 | wakaba | 1.1 | ElementTypeBinding: |
| 150 | @Name: ClsDef | ||
| 151 | @ElementType: | ||
| 152 | dis:ResourceDef | ||
| 153 | @ShadowContent: | ||
| 154 | @@rdf:type: | ||
| 155 | @@@@: dis|MultipleResource | ||
| 156 | @@@ForCheck: !ManakaiDOM|ForIF !ManakaiDOM|ForClass | ||
| 157 | @@resourceFor: | ||
| 158 | @@@@: ManakaiDOM|ForClass | ||
| 159 | @@@ForCheck: ManakaiDOM|ManakaiDOM !=ManakaiDOM|ManakaiDOM | ||
| 160 | @@For: ManakaiDOM|DOM3 | ||
| 161 | @@For: =ManakaiDOM|ManakaiDOM | ||
| 162 | |||
| 163 | @@rdf:type: | ||
| 164 | @@@@: DISLang|Class | ||
| 165 | @@@ForCheck: ManakaiDOM|ForClass | ||
| 166 | |||
| 167 | ElementTypeBinding: | ||
| 168 | @Name: ClsQName | ||
| 169 | @ElementType: | ||
| 170 | dis:QName | ||
| 171 | @ShadowContent: | ||
| 172 | @@ForCheck: ManakaiDOM|ForClass | ||
| 173 | |||
| 174 | ElementTypeBinding: | ||
| 175 | @Name: ClsISA | ||
| 176 | @ElementType: | ||
| 177 | dis:ISA | ||
| 178 | @ShadowContent: | ||
| 179 | @@ForCheck: ManakaiDOM|ForClass | ||
| 180 | |||
| 181 | ElementTypeBinding: | ||
| 182 | @Name: nullCase | ||
| 183 | @ElementType: | ||
| 184 | dis:ResourceDef | ||
| 185 | @ShadowContent: | ||
| 186 | @@rdf:type: ManakaiDOM|InCase | ||
| 187 | @@Value: | ||
| 188 | @@@is-null:1 | ||
| 189 | |||
| 190 | ResourceDef: | ||
| 191 | @QName: LSParser | ||
| 192 | @AliasFor: DOMLS|LSParser | ||
| 193 | @For: ManakaiDOM|DOM3 | ||
| 194 | |||
| 195 | ClsDef: | ||
| 196 | @ClsQName: ManakaiXMLParser | ||
| 197 | |||
| 198 | @Implement: DOMLS|LSParser | ||
| 199 | |||
| 200 | @f:implements: | ||
| 201 | @@@: DOMLS|LSFeature30 | ||
| 202 | @@For: ManakaiDOM|DOM3 | ||
| 203 | |||
| 204 | wakaba | 1.11 | @DISLang:role: gls|ParserRole |
| 205 | wakaba | 1.1 | |
| 206 | wakaba | 1.3 | @enDesc: |
| 207 | Note that the <Class::ManakaiXMLParser> reports any XML errors | ||
| 208 | (syntax errors and / or well-formedness constraint errors) | ||
| 209 | via the <IF::DOMCore:error-handler> registered to | ||
| 210 | the <A::DOMLS:LSParser.domConfig> object. Each error has | ||
| 211 | its <A::DOMCore:DOMError.severity>, either <C::DOMCore:SEVERITY_ERROR> | ||
| 212 | or <C::DOMCore:SEVERITY_FATAL_ERROR>. However, their semantics | ||
| 213 | are slight different from the ones of <QUOTE::error> and | ||
| 214 | <QUOTE::fatal error> in XML; in this implemenetation, | ||
| 215 | <C::DOMCore:SEVERITY_ERROR> implies that the parsing process | ||
| 216 | can effectively be continued to detect more errors while | ||
| 217 | <C::DOMCore:SEVERITY_FATAL_ERROR> implies that the error | ||
| 218 | is serious so that the result document tree and any errors | ||
| 219 | might be far from the ones obtained when the error would not | ||
| 220 | be found. | ||
| 221 | |||
| 222 | wakaba | 1.1 | @Attr: |
| 223 | @@Name: domConfig | ||
| 224 | @@enDesc: | ||
| 225 | The configuration of the parser. | ||
| 226 | |||
| 227 | @@Get: | ||
| 228 | @@@Type: DOMCore|DOMConfiguration | ||
| 229 | @@@enDesc: The DOM configuration object. | ||
| 230 | @@@PerlDef: | ||
| 231 | __CODE{DOMCore|getConfigObject:: | ||
| 232 | $target => $self, | ||
| 233 | $targetHash => $self, | ||
| 234 | $targetType => {<IFName::LSParser>}, | ||
| 235 | $result => $r, | ||
| 236 | }__; | ||
| 237 | |||
| 238 | @Method: | ||
| 239 | @@ManakaiDOM:isForInternal:1 | ||
| 240 | @@ForCheck: ManakaiDOM|ForClass | ||
| 241 | @@Operator: DISPerl|NewMethod | ||
| 242 | @@enDesc: | ||
| 243 | Creates a new instance of the object. | ||
| 244 | @@Param: | ||
| 245 | @@@Name: impl | ||
| 246 | wakaba | 1.11 | @@@Type: gls|GLSImplementation |
| 247 | wakaba | 1.1 | @@@enDesc: |
| 248 | The implementation from which the parser is created. | ||
| 249 | @@Param: | ||
| 250 | @@@Name: features | ||
| 251 | @@@Type: DOMString | ||
| 252 | @@@dis:actualType: f|FeaturesString | ||
| 253 | @@@enDesc: | ||
| 254 | The set of features requested for the parser. | ||
| 255 | @@Return: | ||
| 256 | @@@Type: DOMMain|DOMObject | ||
| 257 | @@@dis:actualType: LSParser | ||
| 258 | @@@enDesc: | ||
| 259 | The newly created parser. | ||
| 260 | @@@PerlDef: | ||
| 261 | $r = bless { | ||
| 262 | <H::DOMCore:implementation> => $impl, | ||
| 263 | }, $self; | ||
| 264 | |||
| 265 | @Method: | ||
| 266 | @@Name: parseString | ||
| 267 | @@enImplNote: | ||
| 268 | Non-standard - to be removed | ||
| 269 | |||
| 270 | @@Param: | ||
| 271 | @@@Name: sourceText | ||
| 272 | @@@Type: DOMString | ||
| 273 | @@Return: | ||
| 274 | @@@Type: Document | ||
| 275 | @@@PerlDef: | ||
| 276 | |||
| 277 | $self->{char} = []; | ||
| 278 | $self->{token} = []; | ||
| 279 | wakaba | 1.6 | $self->{entity} = [{ |
| 280 | reptxt => \$sourceText, | ||
| 281 | line => 1, | ||
| 282 | column => 1, | ||
| 283 | pos => 0, | ||
| 284 | }]; | ||
| 285 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 286 | wakaba | 1.6 | $self->{entity_char} = []; |
| 287 | $self->{entity_token} = []; | ||
| 288 | wakaba | 1.3 | $self->{xml_version} = '1.0'; |
| 289 | $self->{standalone} = false; | ||
| 290 | wakaba | 1.19 | $self->{has_xref} = false; |
| 291 | wakaba | 1.11 | $self->{general_entity} = { |
| 292 | lt => {is_predefined => true}, | ||
| 293 | gt => {is_predefined => true}, | ||
| 294 | amp => {is_predefined => true}, | ||
| 295 | quot => {is_predefined => true}, | ||
| 296 | apos => {is_predefined => true}, | ||
| 297 | }; | ||
| 298 | wakaba | 1.6 | $self->{param_entity} = {}; |
| 299 | wakaba | 1.9 | $self->{attr} = {}; |
| 300 | # $self->{attr}->{$element_type_name}->{$attr_name} = $attr_def | ||
| 301 | wakaba | 1.17 | |
| 302 | wakaba | 1.3 | ## Well-formedness constraint Entity Declared takes effect? |
| 303 | wakaba | 1.1 | |
| 304 | __DEEP{ | ||
| 305 | wakaba | 1.11 | $r = $self->{<H::c|implementation>}-><M::DOMImpl.createDocument>; |
| 306 | $self->{doc} = $r; | ||
| 307 | wakaba | 1.21 | $self->{doc_cfg} = $r-><AG::Document.domConfig>; |
| 308 | wakaba | 1.11 | CORE::delete $self->{docx}; |
| 309 | CORE::delete $self->{dtdef}; | ||
| 310 | wakaba | 1.17 | CORE::delete $self->{dtdecl}; |
| 311 | wakaba | 1.21 | |
| 312 | $r-><AS::Document.strictErrorChecking> (false); | ||
| 313 | ## NOTE: Any checks such as |Name| validity done by | ||
| 314 | ## methods on DOM nodes are disabled. It might result | ||
| 315 | ## in creating an ill-formed DOM tree when parser | ||
| 316 | ## errors are traped by |error-handler|. | ||
| 317 | $self->{doc_cfg}-><M::c|DOMConfiguration.setParameter> | ||
| 318 | (<Q::cfg|entity-reference-read-only> => false); | ||
| 319 | ## NOTE: Turn this configuration parameter off is required | ||
| 320 | ## because entity reference subtree is created by | ||
| 321 | ## the parser rather than |createEntityReference|'s | ||
| 322 | ## cloning. Parser's |cfg:entity-reference-read-only| | ||
| 323 | ## configuration parameter controls whether generated | ||
| 324 | ## subtree is then marked as read-only or not. | ||
| 325 | $self->{doc_cfg}-><M::c|DOMConfiguration.setParameter> | ||
| 326 | (<Q::cfg|clone-entity-reference-subtree> => true); | ||
| 327 | ## NOTE: Turn this configuration parameter on makes | ||
| 328 | ## entity reference subtrees in attribute default values | ||
| 329 | ## cloned as is into default attribute node subtrees. | ||
| 330 | wakaba | 1.11 | |
| 331 | ## Document entity -> |Document| node | ||
| 332 | $self->_parse_DocumentEntity; | ||
| 333 | |||
| 334 | ## Replacement tree for general |Entity| nodes | ||
| 335 | wakaba | 1.20 | my @ent = values %{$self->{general_entity}}; |
| 336 | for my $ent (@ent) { | ||
| 337 | if (exists $ent->{has_replacement_text}) { | ||
| 338 | wakaba | 1.11 | my $ent_name = $ent->{name}; |
| 339 | $self->{entity} = [<Code::getCopyOfEntityState:: | ||
| 340 | $entity_type = 'general_entity', | ||
| 341 | $entity_name = $ent_name>]; | ||
| 342 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 343 | wakaba | 1.11 | $self->{token} = []; |
| 344 | $self->{char} = []; | ||
| 345 | wakaba | 1.23 | local $ent->{is_opened} = true; |
| 346 | wakaba | 1.11 | ## TODO: External entity support |
| 347 | $self->_parse_InternalGeneralParsedEntity ($ent->{node}); | ||
| 348 | $ent->{node}-><AS::x|Entity.hasReplacementTree> (true); | ||
| 349 | wakaba | 1.21 | ## ISSUE: Should errors detected by this phase result |
| 350 | ## in |DOMLS:PARSE_ERR| exception thrown? | ||
| 351 | wakaba | 1.20 | } |
| 352 | wakaba | 1.11 | ## TODO: Read-only if configuration (for all entities) |
| 353 | } | ||
| 354 | |||
| 355 | wakaba | 1.21 | ## Turns flags to their default value |
| 356 | $self->{doc_cfg}-><M::c|DOMConfiguration.setParameter> | ||
| 357 | (<Q::cfg|entity-reference-read-only> => null); | ||
| 358 | $self->{doc_cfg}-><M::c|DOMConfiguration.setParameter> | ||
| 359 | (<Q::cfg|clone-entity-reference-subtree> => null); | ||
| 360 | wakaba | 1.11 | $r-><AS::Document.strictErrorChecking> (true); |
| 361 | wakaba | 1.1 | }__; |
| 362 | |||
| 363 | @Method: | ||
| 364 | @@Name: shiftChar | ||
| 365 | @@ManakaiDOM:isForInternal:1 | ||
| 366 | @@ForCheck: ManakaiDOM|ForClass | ||
| 367 | @@enDesc: | ||
| 368 | Returns the next character. | ||
| 369 | @@Return: | ||
| 370 | @@@Type: idl|long||ManakaiDOM|all | ||
| 371 | @@@enDesc: | ||
| 372 | The code position number of the next character, if any, | ||
| 373 | wakaba | 1.6 | or <CODE::-1>. |
| 374 | wakaba | 1.1 | @@@PerlDef: |
| 375 | if (@{$self->{char}}) { | ||
| 376 | $r = shift @{$self->{char}}; | ||
| 377 | } else { | ||
| 378 | wakaba | 1.7 | no warnings 'substr'; |
| 379 | ## substr outside length warnings at the end of the string | ||
| 380 | wakaba | 1.6 | GETCHAR: { |
| 381 | my $char = substr (${$self->{entity}->[-1]->{reptxt}}, | ||
| 382 | $self->{entity}->[-1]->{pos}, 1); | ||
| 383 | |||
| 384 | wakaba | 1.7 | if (defined $char and length $char) { |
| 385 | wakaba | 1.13 | $self->{entity}->[-1]->{pos}++; |
| 386 | wakaba | 1.6 | $r = ord $char; |
| 387 | if ($r == 0x000A) { | ||
| 388 | $self->{entity}->[-1]->{line}++; | ||
| 389 | $self->{entity}->[-1]->{column} = 1; | ||
| 390 | } elsif ($r == 0x000D) { | ||
| 391 | my $next_char = substr (${$self->{entity}->[-1]->{reptxt}}, | ||
| 392 | $self->{entity}->[-1]->{pos}, 1); | ||
| 393 | if ($next_char eq "\x0A") { | ||
| 394 | $self->{entity}->[-1]->{pos}++; | ||
| 395 | wakaba | 1.13 | $self->{entity}->[-1]->{line}++; |
| 396 | wakaba | 1.6 | $self->{entity}->[-1]->{column} = 1; |
| 397 | } elsif ($next_char eq "\x85") { | ||
| 398 | if ($self->{xml_version} eq '1.1') { | ||
| 399 | wakaba | 1.16 | if ($self->{entity}->[-1]->{no_xml11_eol}) { |
| 400 | my $location = { | ||
| 401 | utf32_offset => $self->{entity}->[-1]->{pos} - 1, | ||
| 402 | line_number => $self->{entity}->[-1]->{line}, | ||
| 403 | column_number => $self->{entity}->[-1]->{column}, | ||
| 404 | }; | ||
| 405 | my $continue = __DOMCore:ERROR{ | ||
| 406 | xp|fatal-xml11-end-of-line-in-xml-declaration:: | ||
| 407 | DOMCore|location => {$location}, | ||
| 408 | xp|character-number => {$r}, | ||
| 409 | }__; | ||
| 410 | unless ($continue) { | ||
| 411 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 412 | } | ||
| 413 | } # no_xml11_eol | ||
| 414 | wakaba | 1.6 | $self->{entity}->[-1]->{pos}++; |
| 415 | wakaba | 1.13 | $self->{entity}->[-1]->{line}++; |
| 416 | wakaba | 1.6 | $self->{entity}->[-1]->{column} = 1; |
| 417 | } else { | ||
| 418 | wakaba | 1.16 | ## |no_xml11_eol| will be tested later. |
| 419 | wakaba | 1.13 | $self->{entity}->[-1]->{column}++; |
| 420 | ## Line number will be increased by next |shiftChar| call. | ||
| 421 | wakaba | 1.6 | } |
| 422 | wakaba | 1.3 | } else { |
| 423 | wakaba | 1.13 | $self->{entity}->[-1]->{line}++; |
| 424 | wakaba | 1.6 | $self->{entity}->[-1]->{column} = 1; |
| 425 | } | ||
| 426 | $r = 0x000A; | ||
| 427 | } elsif ( | ||
| 428 | not ((0x0020 <= $r and $r <= 0x007E) or | ||
| 429 | (0x00A0 <= $r and $r <= 0xD7FF) or | ||
| 430 | (0xE000 <= $r and $r <= 0xFFFD) or | ||
| 431 | (0x10000 <= $r and $r <= 0x10FFFF)) and | ||
| 432 | $r != 0x0009 and $r != 0x0085 and | ||
| 433 | not ($self->{xml_version} eq '1.0' and | ||
| 434 | (0x007F <= $r and $r <= 0x009F)) | ||
| 435 | ) { | ||
| 436 | my $location = { | ||
| 437 | wakaba | 1.13 | utf32_offset => $self->{entity}->[-1]->{pos} - 1, |
| 438 | wakaba | 1.6 | line_number => $self->{entity}->[-1]->{line}, |
| 439 | column_number => $self->{entity}->[-1]->{column}, | ||
| 440 | }; | ||
| 441 | my $continue = __DOMCore:ERROR{xp|wf-legal-literal-character:: | ||
| 442 | DOMCore|location => {$location}, | ||
| 443 | xp|character-number => {$r}, | ||
| 444 | }__; | ||
| 445 | unless ($continue) { | ||
| 446 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 447 | wakaba | 1.3 | } |
| 448 | wakaba | 1.6 | $self->{entity}->[-1]->{column}++; |
| 449 | } elsif ($r == 0x0085 or $r == 0x2028) { | ||
| 450 | wakaba | 1.16 | if ($self->{entity}->[-1]->{no_xml11_eol}) { |
| 451 | my $location = { | ||
| 452 | utf32_offset => $self->{entity}->[-1]->{pos} - 1, | ||
| 453 | line_number => $self->{entity}->[-1]->{line}, | ||
| 454 | column_number => $self->{entity}->[-1]->{column}, | ||
| 455 | }; | ||
| 456 | my $continue = __DOMCore:ERROR{ | ||
| 457 | xp|fatal-xml11-end-of-line-in-xml-declaration:: | ||
| 458 | DOMCore|location => {$location}, | ||
| 459 | xp|character-number => {$r}, | ||
| 460 | }__; | ||
| 461 | unless ($continue) { | ||
| 462 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 463 | } | ||
| 464 | } # no_xml11_eol | ||
| 465 | wakaba | 1.6 | $r = 0x000A if $self->{xml_version} eq '1.1'; |
| 466 | wakaba | 1.13 | ## Even in XML 1.0 it increases the line number. |
| 467 | wakaba | 1.6 | $self->{entity}->[-1]->{line}++; |
| 468 | $self->{entity}->[-1]->{column} = 1; | ||
| 469 | wakaba | 1.3 | } else { |
| 470 | wakaba | 1.6 | $self->{entity}->[-1]->{column}++; |
| 471 | wakaba | 1.3 | } |
| 472 | } else { | ||
| 473 | wakaba | 1.6 | $r = -1; |
| 474 | wakaba | 1.3 | } |
| 475 | wakaba | 1.6 | } # GETCHAR |
| 476 | wakaba | 1.1 | } |
| 477 | |||
| 478 | @Method: | ||
| 479 | @@ManakaiDOM:isForInternal: 1 | ||
| 480 | @@Operator: ManakaiDOM|MUErrorHandler | ||
| 481 | @@enDesc: | ||
| 482 | When a <IF::ecore|ErrorInterface||ManakaiDOM|Perl> is <Perl::report>ed, | ||
| 483 | then this method is invoked. | ||
| 484 | |||
| 485 | The method calls the <cfg::DOMCore|error-handler> if the error is of | ||
| 486 | <IF::DOMCore|DOMError>. Otherwise, the error is re-thrown so that | ||
| 487 | corresponding <Perl::catch> clause, if any, can catch the error. | ||
| 488 | @@Param: | ||
| 489 | @@@Name: err | ||
| 490 | @@@Type: ecore|ErrorInterface||ManakaiDOM|Perl | ||
| 491 | @@@enDesc: | ||
| 492 | The reported error object. | ||
| 493 | @@Return: | ||
| 494 | @@@Type: DISPerl|Any | ||
| 495 | @@@enDesc: | ||
| 496 | If the <P::err> is a <IF::DOMCore|DOMError>, then the return value | ||
| 497 | of the error handler. | ||
| 498 | |||
| 499 | {NOTE:: If the error is thrown, the method never returns. | ||
| 500 | } | ||
| 501 | @@@nullCase: | ||
| 502 | @@@@enDesc: | ||
| 503 | No error handler. | ||
| 504 | @@@PerlDef: | ||
| 505 | if ($err->isa (<IFName::DOMCore|DOMError||ManakaiDOM|ManakaiDOM>)) { | ||
| 506 | __DEEP{ | ||
| 507 | A: { | ||
| 508 | my $cfg = $self-><AG::LSParser.domConfig>; | ||
| 509 | my $h = $cfg-><M::DOMCore|DOMConfiguration.getParameter> | ||
| 510 | ('error-handler'); | ||
| 511 | $r = $h-><M::DOMCore|DOMErrorHandler.handleError> ($err); | ||
| 512 | } # A | ||
| 513 | }__; | ||
| 514 | } else { | ||
| 515 | $err-><M::ecore|ErrorInterface||ManakaiDOM|Perl.throw>; | ||
| 516 | } | ||
| 517 | |||
| 518 | @DISPerl:dpgDef: | ||
| 519 | |||
| 520 | /* | ||
| 521 | XML Document Entity | ||
| 522 | |||
| 523 | document := prolog element *Misc | ||
| 524 | - *Char RestrictedChar *Char ;; [1] | ||
| 525 | */ | ||
| 526 | wakaba | 1.11 | rule DocumentEntity : standalone { |
| 527 | wakaba | 1.1 | /* |
| 528 | prolog := XMLDecl? *Misc [doctypedecl *Misc] ;; [22] | ||
| 529 | */ | ||
| 530 | wakaba | 1.16 | ?lexmode DocumentStart; |
| 531 | wakaba | 1.11 | |
| 532 | my $doc; | ||
| 533 | lang:Perl { | ||
| 534 | $doc = $self->{doc}; | ||
| 535 | } | ||
| 536 | wakaba | 1.1 | |
| 537 | wakaba | 1.3 | ~? (XDO) { |
| 538 | wakaba | 1.16 | ?lexmode XMLDeclarationOrPI; |
| 539 | lang:Perl { | ||
| 540 | $self->{entity}->[-1]->{no_xml11_eol} = true; | ||
| 541 | } | ||
| 542 | ~ (S) { // XML declaration | ||
| 543 | &_XMLDeclaration_ ($doc => $doc); | ||
| 544 | ~ (PIC2) { | ||
| 545 | ?lexmode DocumentProlog; | ||
| 546 | } else { | ||
| 547 | ?lexmode DocumentProlog; | ||
| 548 | } | ||
| 549 | } (NameChar) { // Processing instruction whose target begins with |xml| | ||
| 550 | my $target_token; | ||
| 551 | lang:Perl ($v => $token.value) { | ||
| 552 | $target_token = $token; | ||
| 553 | $target_token->{value} = 'xml'.$target_token->{value}; | ||
| 554 | $self->{entity}->[-1]->{no_xml11_eol} = false; | ||
| 555 | } | ||
| 556 | ?lexmode PINmtoken; | ||
| 557 | ~? (Nmtoken) { | ||
| 558 | lang:Perl ($v => $token.value) { | ||
| 559 | $target_token->{value} .= $v; | ||
| 560 | } | ||
| 561 | ?lexmode PIName; | ||
| 562 | } | ||
| 563 | lang:Perl { | ||
| 564 | $target_token->{type} = 'Name'; | ||
| 565 | $target_token->{location_d} += 3; # 'xml' | ||
| 566 | unshift @{$self->{token}}, $token; | ||
| 567 | $token = $target_token; | ||
| 568 | } | ||
| 569 | &_ProcessingInstruction_ ($doc => $doc, $parent => $doc); | ||
| 570 | ~ (PIC) { | ||
| 571 | ?lexmode DocumentProlog; | ||
| 572 | } else { | ||
| 573 | ?lexmode DocumentProlog; | ||
| 574 | } | ||
| 575 | wakaba | 1.3 | } else { |
| 576 | wakaba | 1.16 | lang:Perl { |
| 577 | $self->{entity}->[-1]->{no_xml11_eol} = false; | ||
| 578 | } | ||
| 579 | wakaba | 1.3 | } |
| 580 | } else { | ||
| 581 | ?lexmode 'DocumentProlog'; | ||
| 582 | } | ||
| 583 | wakaba | 1.1 | |
| 584 | // *Misc | ||
| 585 | wakaba | 1.2 | ~* (CDO) { |
| 586 | wakaba | 1.1 | &_CommentDeclaration_ ($doc => $doc, $parent => $doc); |
| 587 | |||
| 588 | ~ (MDC) { | ||
| 589 | ?lexmode DocumentProlog; | ||
| 590 | } else { | ||
| 591 | ?lexmode DocumentProlog; | ||
| 592 | } | ||
| 593 | } (PIO) { | ||
| 594 | &_ProcessingInstruction_ ($doc => $doc, $parent => $doc); | ||
| 595 | |||
| 596 | ~ (PIC) { | ||
| 597 | ?lexmode 'DocumentProlog'; | ||
| 598 | } else { | ||
| 599 | ?lexmode DocumentProlog; | ||
| 600 | } | ||
| 601 | } (S) { | ||
| 602 | // | ||
| 603 | } | ||
| 604 | |||
| 605 | // doctypedecl | ||
| 606 | ~? (MDO) { | ||
| 607 | &_DocumentTypeDeclaration_ ($doc => $doc); | ||
| 608 | |||
| 609 | wakaba | 1.3 | ~ (MDC) { |
| 610 | ?lexmode DocumentMisc; | ||
| 611 | } else { | ||
| 612 | ?lexmode DocumentMisc; | ||
| 613 | } | ||
| 614 | } else { | ||
| 615 | ?lexmode DocumentMisc; | ||
| 616 | wakaba | 1.1 | } |
| 617 | |||
| 618 | // *Misc | ||
| 619 | wakaba | 1.2 | ~* (CDO) { |
| 620 | wakaba | 1.1 | &_CommentDeclaration_ ($doc => $doc, $parent => $doc); |
| 621 | |||
| 622 | ~ (MDC) { | ||
| 623 | ?lexmode DocumentMisc; | ||
| 624 | } else { | ||
| 625 | ?lexmode DocumentMisc; | ||
| 626 | } | ||
| 627 | } (PIO) { | ||
| 628 | &_ProcessingInstruction_ ($doc => $doc, $parent => $doc); | ||
| 629 | |||
| 630 | ~ (PIC) { | ||
| 631 | ?lexmode 'DocumentMisc'; | ||
| 632 | } else { | ||
| 633 | ?lexmode DocumentMisc; | ||
| 634 | } | ||
| 635 | } (S) { | ||
| 636 | // | ||
| 637 | } | ||
| 638 | |||
| 639 | // Document element | ||
| 640 | ~ (STAGO) { | ||
| 641 | &Element_ ($doc => $doc, $parent => $doc) | ||
| 642 | : unshift-current-token; | ||
| 643 | ~ (TAGC) { | ||
| 644 | ?lexmode DocumentEnd; | ||
| 645 | } else { | ||
| 646 | ?lexmode DocumentEnd; | ||
| 647 | } | ||
| 648 | } else { | ||
| 649 | ?lexmode 'DocumentEnd'; | ||
| 650 | } | ||
| 651 | |||
| 652 | // *Misc | ||
| 653 | wakaba | 1.2 | ~* (CDO) { |
| 654 | wakaba | 1.1 | &_CommentDeclaration_ ($doc => $doc, $parent => $doc); |
| 655 | |||
| 656 | ~ (MDC) { | ||
| 657 | ?lexmode DocumentEnd; | ||
| 658 | } else { | ||
| 659 | ?lexmode DocumentEnd; | ||
| 660 | } | ||
| 661 | } (PIO) { | ||
| 662 | &_ProcessingInstruction_ ($doc => $doc, $parent => $doc); | ||
| 663 | ~ (PIC) { | ||
| 664 | ?lexmode 'DocumentEnd'; | ||
| 665 | } else { | ||
| 666 | ?lexmode DocumentEnd; | ||
| 667 | } | ||
| 668 | } (S) { | ||
| 669 | // | ||
| 670 | } | ||
| 671 | |||
| 672 | ~ (#EOF) { } | ||
| 673 | } // DocumentEntity | ||
| 674 | wakaba | 1.11 | |
| 675 | /* | ||
| 676 | Internal General Parsed Entity | ||
| 677 | |||
| 678 | An internal general parsed entity is well-formed if its | ||
| 679 | replacement text matches the production labeled |content|. | ||
| 680 | */ | ||
| 681 | rule InternalGeneralParsedEntity ($ent) : standalone { | ||
| 682 | ?lexmode ElementContent; | ||
| 683 | |||
| 684 | my $ns; | ||
| 685 | lang:Perl { | ||
| 686 | $ns = { | ||
| 687 | xml => <Q::xml:>, | ||
| 688 | xmlns => <Q::xmlns:>, | ||
| 689 | }; | ||
| 690 | } | ||
| 691 | |||
| 692 | &content ($parent => $ent, $ns => $ns); | ||
| 693 | |||
| 694 | ~ (#EOF) { } | ||
| 695 | } // InternalGeneralParsedEntity | ||
| 696 | |||
| 697 | wakaba | 1.1 | /* |
| 698 | XML Declaration | ||
| 699 | |||
| 700 | XMLDecl := '<?xml' VersionInfo | ||
| 701 | [EncodingDecl] | ||
| 702 | [SDDecl] | ||
| 703 | [S] '?>' ;; [23] | ||
| 704 | |||
| 705 | NOTE: XML declaration is optional in XML 1.0 | ||
| 706 | while it is required in XML 1.1. | ||
| 707 | */ | ||
| 708 | wakaba | 1.16 | rule _XMLDeclaration_ { |
| 709 | // ~ (XDO) { } | ||
| 710 | wakaba | 1.3 | ?lexmode XMLDeclaration; |
| 711 | wakaba | 1.16 | //lang:Perl { |
| 712 | // $self->{entity}->[-1]->{no_xml11_eol} = true; | ||
| 713 | //} | ||
| 714 | //~ (S) { } | ||
| 715 | wakaba | 1.3 | |
| 716 | ~ (Name == 'version') { | ||
| 717 | ~? (S) { } | ||
| 718 | ~ (VI) { } | ||
| 719 | ~? (S) { } | ||
| 720 | wakaba | 1.1 | |
| 721 | wakaba | 1.3 | my $ver; |
| 722 | my $bad_token; | ||
| 723 | |||
| 724 | ~ (LIT) { | ||
| 725 | ?lexmode AttributeValueLiteral; | ||
| 726 | |||
| 727 | ~ (STRING) { | ||
| 728 | lang:Perl ($version => $token.value) { | ||
| 729 | $ver = $version; | ||
| 730 | $bad_token = $token; | ||
| 731 | } | ||
| 732 | } | ||
| 733 | |||
| 734 | ~ (LIT) { | ||
| 735 | ?lexmode XMLDeclaration; | ||
| 736 | } | ||
| 737 | } (LITA) { | ||
| 738 | ?lexmode AttributeValueLiteralA; | ||
| 739 | |||
| 740 | ~ (STRING) { | ||
| 741 | lang:Perl ($version => $token.value) { | ||
| 742 | $ver = $version; | ||
| 743 | $bad_token = $token; | ||
| 744 | } | ||
| 745 | } | ||
| 746 | |||
| 747 | ~ (LITA) { | ||
| 748 | ?lexmode XMLDeclaration; | ||
| 749 | } | ||
| 750 | } | ||
| 751 | |||
| 752 | lang:Perl : has-error { | ||
| 753 | unless ($ver eq '1.0' or $ver eq '1.1') { | ||
| 754 | my $location; | ||
| 755 | __CODE{xp|get-location-from-token:: | ||
| 756 | $token => {$bad_token}, | ||
| 757 | $result => {$location}, | ||
| 758 | }__; | ||
| 759 | my $continue = __DOMCore:ERROR{xp|wf-unsupported-xml-version:: | ||
| 760 | DOMCore|location => {$location}, | ||
| 761 | wakaba | 1.16 | xp|parent => {$self->{doc}}, |
| 762 | wakaba | 1.3 | infoset|version => {$ver}, |
| 763 | xp|error-token => {$bad_token}, | ||
| 764 | }__; | ||
| 765 | unless ($continue) { | ||
| 766 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 767 | } | ||
| 768 | } | ||
| 769 | wakaba | 1.16 | $self->{doc}-><AS::Document.xmlVersion> ($ver); |
| 770 | wakaba | 1.3 | $self->{xml_version} = $ver; |
| 771 | } | ||
| 772 | |||
| 773 | ~? (S) { } | ||
| 774 | wakaba | 1.1 | } |
| 775 | |||
| 776 | wakaba | 1.3 | ~? (Name == 'encoding') { |
| 777 | ~? (S) { } | ||
| 778 | ~ (VI) { } | ||
| 779 | ~? (S) { } | ||
| 780 | |||
| 781 | my $ver; | ||
| 782 | my $bad_token; | ||
| 783 | |||
| 784 | ~ (LIT) { | ||
| 785 | ?lexmode AttributeValueLiteral; | ||
| 786 | |||
| 787 | ~ (STRING) { | ||
| 788 | lang:Perl ($version => $token.value) { | ||
| 789 | $ver = $version; | ||
| 790 | $bad_token = $token; | ||
| 791 | } | ||
| 792 | } | ||
| 793 | |||
| 794 | ~ (LIT) { | ||
| 795 | ?lexmode XMLDeclaration; | ||
| 796 | } | ||
| 797 | } (LITA) { | ||
| 798 | ?lexmode AttributeValueLiteralA; | ||
| 799 | |||
| 800 | ~ (STRING) { | ||
| 801 | lang:Perl ($version => $token.value) { | ||
| 802 | $ver = $version; | ||
| 803 | $bad_token = $token; | ||
| 804 | } | ||
| 805 | } | ||
| 806 | |||
| 807 | ~ (LITA) { | ||
| 808 | ?lexmode XMLDeclaration; | ||
| 809 | } | ||
| 810 | } | ||
| 811 | |||
| 812 | lang:Perl : has-error { | ||
| 813 | unless ($ver =~ /\A[A-Za-z][A-Za-z0-9._-]*\z/) { | ||
| 814 | my $location; | ||
| 815 | __CODE{xp|get-location-from-token:: | ||
| 816 | $token => {$bad_token}, | ||
| 817 | $result => {$location}, | ||
| 818 | }__; | ||
| 819 | my $continue = __DOMCore:ERROR{xp|wf-malformed-enc-name:: | ||
| 820 | DOMCore|location => {$location}, | ||
| 821 | wakaba | 1.16 | xp|parent => {$self->{doc}}, |
| 822 | wakaba | 1.3 | xp|name => {$ver}, |
| 823 | xp|error-token => {$bad_token}, | ||
| 824 | }__; | ||
| 825 | unless ($continue) { | ||
| 826 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 827 | } | ||
| 828 | } | ||
| 829 | wakaba | 1.16 | $self->{doc}-><AS::Document.xmlEncoding> ($ver); |
| 830 | wakaba | 1.3 | } |
| 831 | |||
| 832 | ~? (S) { } | ||
| 833 | wakaba | 1.1 | } |
| 834 | wakaba | 1.3 | |
| 835 | ~? (Name == 'standalone') { | ||
| 836 | ~? (S) { } | ||
| 837 | ~ (VI) { } | ||
| 838 | ~? (S) { } | ||
| 839 | |||
| 840 | my $ver; | ||
| 841 | my $bad_token; | ||
| 842 | |||
| 843 | ~ (LIT) { | ||
| 844 | ?lexmode AttributeValueLiteral; | ||
| 845 | |||
| 846 | ~ (STRING) { | ||
| 847 | lang:Perl ($version => $token.value) { | ||
| 848 | $ver = $version; | ||
| 849 | $bad_token = $token; | ||
| 850 | } | ||
| 851 | } | ||
| 852 | |||
| 853 | ~ (LIT) { | ||
| 854 | ?lexmode XMLDeclaration; | ||
| 855 | } | ||
| 856 | } (LITA) { | ||
| 857 | ?lexmode AttributeValueLiteralA; | ||
| 858 | |||
| 859 | ~ (STRING) { | ||
| 860 | lang:Perl ($version => $token.value) { | ||
| 861 | $ver = $version; | ||
| 862 | $bad_token = $token; | ||
| 863 | } | ||
| 864 | } | ||
| 865 | wakaba | 1.1 | |
| 866 | wakaba | 1.3 | ~ (LITA) { |
| 867 | ?lexmode XMLDeclaration; | ||
| 868 | } | ||
| 869 | } | ||
| 870 | |||
| 871 | lang:Perl : has-error { | ||
| 872 | unless ($ver eq 'yes' or $ver eq 'no') { | ||
| 873 | my $location; | ||
| 874 | __CODE{xp|get-location-from-token:: | ||
| 875 | $token => {$bad_token}, | ||
| 876 | $result => {$location}, | ||
| 877 | }__; | ||
| 878 | my $continue = __DOMCore:ERROR{xp|wf-malformed-xml-standalone:: | ||
| 879 | DOMCore|location => {$location}, | ||
| 880 | wakaba | 1.16 | xp|parent => {$self->{doc}}, |
| 881 | wakaba | 1.3 | xp|name => {$ver}, |
| 882 | xp|error-token => {$bad_token}, | ||
| 883 | }__; | ||
| 884 | unless ($continue) { | ||
| 885 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 886 | } | ||
| 887 | } | ||
| 888 | wakaba | 1.13 | if ($ver eq 'yes') { |
| 889 | wakaba | 1.16 | $self->{doc}-><AS::Document.xmlStandalone> (true); |
| 890 | wakaba | 1.13 | $self->{standalone} = true; |
| 891 | } | ||
| 892 | wakaba | 1.3 | } |
| 893 | |||
| 894 | ~? (S) { } | ||
| 895 | } | ||
| 896 | |||
| 897 | wakaba | 1.16 | ~ (PIC1) { |
| 898 | lang:Perl { | ||
| 899 | $self->{entity}->[-1]->{no_xml11_eol} = false; | ||
| 900 | } | ||
| 901 | } else { | ||
| 902 | lang:Perl { | ||
| 903 | $self->{entity}->[-1]->{no_xml11_eol} = false; | ||
| 904 | } | ||
| 905 | } | ||
| 906 | |||
| 907 | // ~ (PIC2) { } | ||
| 908 | wakaba | 1.3 | } // _XMLDeclaration_ |
| 909 | wakaba | 1.1 | |
| 910 | /* | ||
| 911 | Comment Declaration | ||
| 912 | |||
| 913 | Comment := '<!--' *(Char - '-' / '-' (Char - '-')) | ||
| 914 | '-->' ;; [15] | ||
| 915 | wakaba | 1.4 | */ |
| 916 | wakaba | 1.1 | rule _CommentDeclaration_ ($doc, $parent) { |
| 917 | ?lexmode 'CommentDeclaration'; | ||
| 918 | |||
| 919 | ~? (STRING) { | ||
| 920 | lang:Perl ($data => $token.value) { | ||
| 921 | my $com = $doc-><M::Document.createComment> ($data); | ||
| 922 | $parent-><M::Node.appendChild> ($com); | ||
| 923 | } | ||
| 924 | } else { | ||
| 925 | lang:Perl { | ||
| 926 | my $com = $doc-><M::Document.createComment> (''); | ||
| 927 | $parent-><M::Node.appendChild> ($com); | ||
| 928 | } | ||
| 929 | } | ||
| 930 | |||
| 931 | ~ (COM) { | ||
| 932 | ?lexmode MarkupDeclaration; | ||
| 933 | } else { | ||
| 934 | ?lexmode MarkupDeclaration; | ||
| 935 | } | ||
| 936 | |||
| 937 | // ~ (MDC) { } | ||
| 938 | wakaba | 1.4 | } // _CommentDeclaration |
| 939 | wakaba | 1.15 | |
| 940 | rule _CommentDeclarationDTD { | ||
| 941 | ?lexmode CommentDeclaration; | ||
| 942 | wakaba | 1.4 | |
| 943 | ~? (STRING) { | ||
| 944 | // | ||
| 945 | } | ||
| 946 | |||
| 947 | ~ (COM) { | ||
| 948 | ?lexmode MarkupDeclaration; | ||
| 949 | } else { | ||
| 950 | ?lexmode MarkupDeclaration; | ||
| 951 | } | ||
| 952 | |||
| 953 | ~ (MDC) { | ||
| 954 | ?lexmode DTD; | ||
| 955 | } else { | ||
| 956 | ?lexmode DTD; | ||
| 957 | } | ||
| 958 | } // _CommentDeclarationDTD | ||
| 959 | |||
| 960 | wakaba | 1.1 | /* |
| 961 | Processing Instruction | ||
| 962 | |||
| 963 | PI := '<?' PITarget [S *Char - *Char '?>' *Char] | ||
| 964 | '?>' ;; [16] | ||
| 965 | */ | ||
| 966 | rule _ProcessingInstruction_ ($doc, $parent) { | ||
| 967 | ?lexmode 'PIName'; | ||
| 968 | |||
| 969 | my $pi; | ||
| 970 | |||
| 971 | ~ (Name) { | ||
| 972 | wakaba | 1.3 | lang:Perl ($name => $token.value) : has-error { |
| 973 | wakaba | 1.1 | if (lc $name eq 'xml') { |
| 974 | wakaba | 1.3 | my $location; |
| 975 | __CODE{xp|get-location-from-token:: | ||
| 976 | $token => {$token}, | ||
| 977 | $result => {$location}, | ||
| 978 | }__; | ||
| 979 | my $continue = __DOMCore:ERROR{xp|wf-pi-target-is-xml:: | ||
| 980 | xp|name => {$name}, | ||
| 981 | DOMCore|location => {$location}, | ||
| 982 | xp|parent => {$parent}, | ||
| 983 | }__; | ||
| 984 | unless ($continue) { | ||
| 985 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 986 | }__; | ||
| 987 | } | ||
| 988 | wakaba | 1.1 | } |
| 989 | ## TODO: Namespace well-formedness | ||
| 990 | $pi = $doc-><M::Document.createProcessingInstruction> | ||
| 991 | ($name); | ||
| 992 | } | ||
| 993 | wakaba | 1.14 | } else { |
| 994 | lang:Perl { | ||
| 995 | $pi = $doc-><M::Document.createProcessingInstruction> ('#INVALID'); | ||
| 996 | } | ||
| 997 | wakaba | 1.1 | } |
| 998 | |||
| 999 | wakaba | 1.14 | ~? (S) { |
| 1000 | wakaba | 1.1 | ?lexmode 'PIData'; |
| 1001 | |||
| 1002 | my $tdata; | ||
| 1003 | |||
| 1004 | ~? (DATA) { | ||
| 1005 | lang:Perl ($data => $token.value) { | ||
| 1006 | $tdata = $data; | ||
| 1007 | } | ||
| 1008 | } else { | ||
| 1009 | lang:Perl { | ||
| 1010 | $tdata = ''; | ||
| 1011 | } | ||
| 1012 | } | ||
| 1013 | |||
| 1014 | lang:Perl { | ||
| 1015 | $pi-><AS::Node.nodeValue> ($tdata); | ||
| 1016 | } | ||
| 1017 | } | ||
| 1018 | |||
| 1019 | lang:Perl { | ||
| 1020 | $parent-><M::Node.appendChild> ($pi); | ||
| 1021 | } | ||
| 1022 | |||
| 1023 | // ~ (PIC) { } | ||
| 1024 | } // _ProcessingInstruction_ | ||
| 1025 | wakaba | 1.4 | |
| 1026 | wakaba | 1.6 | /* |
| 1027 | Processing instruction in DTD | ||
| 1028 | */ | ||
| 1029 | rule _ProcessingInstructionDTD ($doc, $doctype) { | ||
| 1030 | wakaba | 1.4 | ?lexmode 'PIName'; |
| 1031 | wakaba | 1.6 | |
| 1032 | my $pi; | ||
| 1033 | wakaba | 1.4 | |
| 1034 | ~ (Name) { | ||
| 1035 | wakaba | 1.6 | lang:Perl ($name => $token.value) : has-error { |
| 1036 | if (lc $name eq 'xml') { | ||
| 1037 | my $location; | ||
| 1038 | __CODE{xp|get-location-from-token:: | ||
| 1039 | $token => {$token}, | ||
| 1040 | $result => {$location}, | ||
| 1041 | }__; | ||
| 1042 | my $continue = __DOMCore:ERROR{xp|wf-pi-target-is-xml:: | ||
| 1043 | xp|name => {$name}, | ||
| 1044 | DOMCore|location => {$location}, | ||
| 1045 | xp|parent => {$doctype}, | ||
| 1046 | }__; | ||
| 1047 | unless ($continue) { | ||
| 1048 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 1049 | }__; | ||
| 1050 | } | ||
| 1051 | } | ||
| 1052 | ## TODO: Namespace well-formedness | ||
| 1053 | $pi = $doc-><M::Document.createProcessingInstruction> | ||
| 1054 | ($name); | ||
| 1055 | } | ||
| 1056 | wakaba | 1.14 | } else { |
| 1057 | lang:Perl { | ||
| 1058 | $pi = $doc-><M::Document.createProcessingInstruction> ('#INVALID'); | ||
| 1059 | } | ||
| 1060 | wakaba | 1.4 | } |
| 1061 | |||
| 1062 | wakaba | 1.14 | ~? (S) { |
| 1063 | wakaba | 1.4 | ?lexmode 'PIData'; |
| 1064 | |||
| 1065 | wakaba | 1.6 | my $tdata; |
| 1066 | |||
| 1067 | wakaba | 1.4 | ~? (DATA) { |
| 1068 | wakaba | 1.6 | lang:Perl ($data => $token.value) { |
| 1069 | $tdata = $data; | ||
| 1070 | } | ||
| 1071 | } else { | ||
| 1072 | lang:Perl { | ||
| 1073 | $tdata = ''; | ||
| 1074 | } | ||
| 1075 | } | ||
| 1076 | |||
| 1077 | lang:Perl { | ||
| 1078 | $pi-><AS::Node.nodeValue> ($tdata); | ||
| 1079 | wakaba | 1.4 | } |
| 1080 | } | ||
| 1081 | |||
| 1082 | wakaba | 1.6 | lang:Perl { |
| 1083 | $doctype-><M::Node.appendChild> ($pi); | ||
| 1084 | } | ||
| 1085 | |||
| 1086 | wakaba | 1.4 | ~ (PIC) { |
| 1087 | ?lexmode DTD; | ||
| 1088 | } else { | ||
| 1089 | ?lexmode DTD; | ||
| 1090 | } | ||
| 1091 | } // _ProcessingInstructionDTD | ||
| 1092 | wakaba | 1.1 | |
| 1093 | /* | ||
| 1094 | Element content parsing mode | ||
| 1095 | |||
| 1096 | element := EmptyElemTag / | ||
| 1097 | STag content ETag ;; [39] | ||
| 1098 | content := (CharData / element / Reference / CDSect / | ||
| 1099 | PI / Comment) ;; [43] | ||
| 1100 | */ | ||
| 1101 | wakaba | 1.6 | rule Element_ ($doc, $parent, $ns) : standalone { |
| 1102 | wakaba | 1.1 | ?lexmode 'ElementContent'; |
| 1103 | |||
| 1104 | my $node; // Current "parent" node | ||
| 1105 | my $nodes; // Node stack (w/o $current_node) | ||
| 1106 | my $type; // Current "parent" element type QName | ||
| 1107 | my $types; // Element type stack (w/o $current_type) | ||
| 1108 | wakaba | 1.6 | //my $ns; // Current in-scope namespace bindings |
| 1109 | wakaba | 1.1 | my $nses; // Namespace binding stack (w/o $current_ns) |
| 1110 | |||
| 1111 | lang:Perl { | ||
| 1112 | $node = $parent; | ||
| 1113 | $nodes = []; | ||
| 1114 | $type = ''; | ||
| 1115 | $types = []; | ||
| 1116 | wakaba | 1.6 | $ns ||= { |
| 1117 | wakaba | 1.1 | xml => <Q::xml:>, |
| 1118 | xmlns => <Q::xmlns:>, | ||
| 1119 | }; | ||
| 1120 | $nses = []; | ||
| 1121 | } | ||
| 1122 | |||
| 1123 | ~* : name => CONTENT | ||
| 1124 | (CharData) { | ||
| 1125 | // Character data | ||
| 1126 | lang:Perl ($data => $token.value) { | ||
| 1127 | $node-><M::Node.appendChild> | ||
| 1128 | ($doc-><M::Document.createTextNode> ($data)); | ||
| 1129 | } | ||
| 1130 | } (STAGO) { | ||
| 1131 | // Start tag or empty element tag | ||
| 1132 | |||
| 1133 | ?lexmode 'StartTag'; | ||
| 1134 | |||
| 1135 | ~ (Name) { | ||
| 1136 | my $attrs; | ||
| 1137 | lang:Perl ($name => $token.value) { | ||
| 1138 | push @{$types}, $type; | ||
| 1139 | $type = $name; | ||
| 1140 | $attrs = {}; | ||
| 1141 | } | ||
| 1142 | |||
| 1143 | ~? (S) { | ||
| 1144 | &AttributeSpecificationList | ||
| 1145 | ($doc => $doc, $attrs => $attrs); | ||
| 1146 | } | ||
| 1147 | |||
| 1148 | my $el; | ||
| 1149 | wakaba | 1.9 | |
| 1150 | /* | ||
| 1151 | Note that this implementation does not perform | ||
| 1152 | attribute value tokenization (but does white space | ||
| 1153 | normalization common to attribute types) and | ||
| 1154 | construct the tree as is. DOM Level 3 Core spec | ||
| 1155 | is unclear on this point. With tokenization, | ||
| 1156 | entity references cannot be preserved. | ||
| 1157 | |||
| 1158 | The manakai Node.nodeValue and Attr.value attributes | ||
| 1159 | do tokenization according to attribute types. | ||
| 1160 | */ | ||
| 1161 | |||
| 1162 | /* | ||
| 1163 | ISSUE: Should |xml:id| attribute be typed? | ||
| 1164 | */ | ||
| 1165 | wakaba | 1.1 | |
| 1166 | lang:Perl { | ||
| 1167 | push @{$nses}, $ns; | ||
| 1168 | $ns = {%$ns}; | ||
| 1169 | |||
| 1170 | my %gattr; | ||
| 1171 | my %lattr; | ||
| 1172 | for my $atqname (keys %$attrs) { | ||
| 1173 | my ($pfx, $lname) = split /:/, $atqname; | ||
| 1174 | wakaba | 1.9 | $attrs->{$atqname}->{def} = $self->{attr}->{$type}->{$atqname}; |
| 1175 | wakaba | 1.1 | if (defined $lname) { ## Global attribute |
| 1176 | ## TODO: Namespace well-formedness (lname is NCName) | ||
| 1177 | if ($pfx eq 'xmlns') { | ||
| 1178 | my $nsuri = $attrs->{$atqname}->{value}; | ||
| 1179 | if ($lname eq 'xml' and | ||
| 1180 | $nsuri ne <Q::xml:>) { | ||
| 1181 | ## TODO: error | ||
| 1182 | } elsif ($lname eq 'xmlns') { | ||
| 1183 | ## TODO: error | ||
| 1184 | } | ||
| 1185 | if ($nsuri eq '') { | ||
| 1186 | ## TODO: error in XML 1.0 | ||
| 1187 | } elsif ($nsuri eq <Q::xml:> and | ||
| 1188 | $lname ne 'xml') { | ||
| 1189 | ## TODO: error | ||
| 1190 | } elsif ($nsuri eq <Q::xmlns:>) { | ||
| 1191 | ## TODO: error | ||
| 1192 | } | ||
| 1193 | $ns->{$lname} = $attrs->{$atqname}->{value}; | ||
| 1194 | wakaba | 1.9 | if ($attrs->{$atqname}->{def}) { |
| 1195 | my $dt = $attrs->{$atqname}->{def}-><AG::ATDef.declaredType>; | ||
| 1196 | if ({ | ||
| 1197 | <C::ATDef.ID_ATTR> => true, | ||
| 1198 | <C::ATDef.IDREF_ATTR> => true, | ||
| 1199 | <C::ATDef.IDREFS_ATTR> => true, | ||
| 1200 | <C::ATDef.ENTITY_ATTR> => true, | ||
| 1201 | <C::ATDef.ENTITIES_ATTR> => true, | ||
| 1202 | <C::ATDef.NMTOKEN_ATTR> => true, | ||
| 1203 | <C::ATDef.NMTOKENS_ATTR> => true, | ||
| 1204 | <C::ATDef.NOTATION_ATTR> => true, | ||
| 1205 | <C::ATDef.ENUMERATION_ATTR> => true, | ||
| 1206 | }->{$dt}) { | ||
| 1207 | ## Tokenization (XML 1 3.3.3) | ||
| 1208 | for ($ns->{$lname}) { | ||
| 1209 | s/^\x20+//; | ||
| 1210 | s/\x20+\z//; | ||
| 1211 | s/\x20+/ /g; | ||
| 1212 | } | ||
| 1213 | } | ||
| 1214 | } | ||
| 1215 | wakaba | 1.1 | delete $ns->{$lname} unless length $ns->{$lname}; |
| 1216 | } elsif ($pfx eq '') { | ||
| 1217 | ## TODO: pfx is not NCName error | ||
| 1218 | } else { | ||
| 1219 | if ($gattr{$pfx}->{$lname}) { | ||
| 1220 | ## TODO: Namespace well-formedness error | ||
| 1221 | } | ||
| 1222 | } | ||
| 1223 | $gattr{$pfx}->{$lname} = $attrs->{$atqname}; | ||
| 1224 | } else { ## Local attribute | ||
| 1225 | if ($pfx eq 'xmlns') { | ||
| 1226 | $ns->{''} = $attrs->{xmlns}->{value}; | ||
| 1227 | wakaba | 1.9 | if ($attrs->{$atqname}->{def}) { |
| 1228 | my $dt = $attrs->{$atqname}->{def}-><AG::ATDef.declaredType>; | ||
| 1229 | if ({ | ||
| 1230 | <C::ATDef.ID_ATTR> => true, | ||
| 1231 | <C::ATDef.IDREF_ATTR> => true, | ||
| 1232 | <C::ATDef.IDREFS_ATTR> => true, | ||
| 1233 | <C::ATDef.ENTITY_ATTR> => true, | ||
| 1234 | <C::ATDef.ENTITIES_ATTR> => true, | ||
| 1235 | <C::ATDef.NMTOKEN_ATTR> => true, | ||
| 1236 | <C::ATDef.NMTOKENS_ATTR> => true, | ||
| 1237 | <C::ATDef.NOTATION_ATTR> => true, | ||
| 1238 | <C::ATDef.ENUMERATION_ATTR> => true, | ||
| 1239 | }->{$dt}) { | ||
| 1240 | ## Tokenization (XML 1 3.3.3) | ||
| 1241 | for ($ns->{''}) { | ||
| 1242 | s/^\x20+//; | ||
| 1243 | s/\x20+\z//; | ||
| 1244 | s/\x20+/ /g; | ||
| 1245 | } | ||
| 1246 | } | ||
| 1247 | } | ||
| 1248 | wakaba | 1.1 | delete $ns->{''} unless length $ns->{''}; |
| 1249 | } else { | ||
| 1250 | $lattr{$pfx} = $attrs->{$atqname}; | ||
| 1251 | } | ||
| 1252 | } | ||
| 1253 | } | ||
| 1254 | |||
| 1255 | my ($pfx, $lname) = split /:/, $type; | ||
| 1256 | my $nsuri; | ||
| 1257 | ## TODO: lname is NCName? | ||
| 1258 | if (defined $lname) { ## Prefixed namespace | ||
| 1259 | if ($pfx eq '') { | ||
| 1260 | ## TODO: pfx is not NCName error | ||
| 1261 | } | ||
| 1262 | if (defined $ns->{$pfx}) { | ||
| 1263 | $nsuri = $ns->{$pfx}; | ||
| 1264 | } else { | ||
| 1265 | ## TODO: namespace ill-formed | ||
| 1266 | } | ||
| 1267 | } else { ## Default namespace | ||
| 1268 | $nsuri = $ns->{''}; | ||
| 1269 | } | ||
| 1270 | |||
| 1271 | $el = $doc-><M::Document.createElementNS> | ||
| 1272 | ($nsuri, $type); | ||
| 1273 | |||
| 1274 | if ($attrs->{xmlns}) { | ||
| 1275 | my $attr = $doc-><M::Document.createAttributeNS> | ||
| 1276 | (<Q::xmlns:>, 'xmlns'); | ||
| 1277 | for (@{$attrs->{xmlns}->{nodes}}) { | ||
| 1278 | $attr-><M::Node.appendChild> ($_); | ||
| 1279 | } | ||
| 1280 | wakaba | 1.9 | if ($attrs->{xmlns}->{def}) { |
| 1281 | __CODE{t|setAttrType:: | ||
| 1282 | $attr => $attr, | ||
| 1283 | $type => {$attrs->{xmlns}->{def}-><AG::ATDef.declaredType>}, | ||
| 1284 | }__; | ||
| 1285 | } | ||
| 1286 | wakaba | 1.1 | $el-><M::Element.setAttributeNodeNS> ($attr); |
| 1287 | } | ||
| 1288 | |||
| 1289 | for my $lname (keys %lattr) { | ||
| 1290 | my $attr = $doc-><M::Document.createAttributeNS> | ||
| 1291 | (null, $lname); | ||
| 1292 | for (@{$lattr{$lname}->{nodes}}) { | ||
| 1293 | $attr-><M::Node.appendChild> ($_); | ||
| 1294 | } | ||
| 1295 | wakaba | 1.9 | if ($attrs->{$lname}->{def}) { |
| 1296 | __CODE{t|setAttrType:: | ||
| 1297 | $attr => $attr, | ||
| 1298 | $type => {$attrs->{$lname}->{def}-><AG::ATDef.declaredType>}, | ||
| 1299 | }__; | ||
| 1300 | } | ||
| 1301 | wakaba | 1.1 | $el-><M::Element.setAttributeNodeNS> ($attr); |
| 1302 | } | ||
| 1303 | |||
| 1304 | for my $pfx (keys %gattr) { | ||
| 1305 | for my $lname (keys %{$gattr{$pfx}}) { | ||
| 1306 | my $attr = $doc-><M::Document.createAttributeNS> | ||
| 1307 | ($ns->{$pfx}, $pfx.':'.$lname); | ||
| 1308 | for (@{$gattr{$pfx}->{$lname}->{nodes}}) { | ||
| 1309 | $attr-><M::Node.appendChild> ($_); | ||
| 1310 | } | ||
| 1311 | wakaba | 1.9 | if ($attrs->{$pfx}->{$lname}->{def}) { |
| 1312 | __CODE{t|setAttrType:: | ||
| 1313 | $attr => $attr, | ||
| 1314 | $type => {$attrs->{$pfx}->{$lname} | ||
| 1315 | ->{def}-><AG::ATDef.declaredType>}, | ||
| 1316 | }__; | ||
| 1317 | } | ||
| 1318 | wakaba | 1.1 | $el-><M::Element.setAttributeNodeNS> ($attr); |
| 1319 | } | ||
| 1320 | } | ||
| 1321 | |||
| 1322 | $node-><M::Node.appendChild> ($el); | ||
| 1323 | } | ||
| 1324 | |||
| 1325 | ~ (TAGC) { | ||
| 1326 | lang:Perl { | ||
| 1327 | push @{$nodes}, $node; | ||
| 1328 | $node = $el; | ||
| 1329 | } | ||
| 1330 | ?lexmode ElementContent; | ||
| 1331 | wakaba | 1.6 | } (NESTC) { |
| 1332 | my $is_docel; | ||
| 1333 | wakaba | 1.1 | lang:Perl { |
| 1334 | $ns = pop @{$nses}; | ||
| 1335 | $type = pop @{$types}; | ||
| 1336 | wakaba | 1.6 | $is_docel = (@{$types} == 0); |
| 1337 | } | ||
| 1338 | |||
| 1339 | if-true ($is_docel) { | ||
| 1340 | return; | ||
| 1341 | } | ||
| 1342 | |||
| 1343 | ~ (TAGC) { | ||
| 1344 | ?lexmode ElementContent; | ||
| 1345 | } else { | ||
| 1346 | ?lexmode ElementContent; | ||
| 1347 | wakaba | 1.1 | } |
| 1348 | } else { | ||
| 1349 | ?lexmode ElementContent; | ||
| 1350 | } | ||
| 1351 | } else { | ||
| 1352 | ?lexmode ElementContent; | ||
| 1353 | } | ||
| 1354 | |||
| 1355 | } (ETAGO) { | ||
| 1356 | // End tag | ||
| 1357 | |||
| 1358 | ?lexmode 'EndTag'; | ||
| 1359 | |||
| 1360 | my $is_docel; | ||
| 1361 | |||
| 1362 | ~ (Name) { | ||
| 1363 | wakaba | 1.3 | lang:Perl ($name => $token.value) : has-error { |
| 1364 | wakaba | 1.1 | if ($name eq $type) { |
| 1365 | $type = pop @{$types}; | ||
| 1366 | if ($type eq '') { | ||
| 1367 | $is_docel = true; | ||
| 1368 | } | ||
| 1369 | $node = pop @{$nodes}; | ||
| 1370 | $ns = pop @{$nses}; | ||
| 1371 | } else { | ||
| 1372 | wakaba | 1.3 | my $location; |
| 1373 | __CODE{xp|get-location-from-token:: | ||
| 1374 | $token => $token, | ||
| 1375 | $result => $location, | ||
| 1376 | }__; | ||
| 1377 | my $continue = __DOMCore:ERROR{xp|wf-element-type-match:: | ||
| 1378 | DOMCore:location => {$location}, | ||
| 1379 | xp|token => {$token}, | ||
| 1380 | xp|expected-element-type => {$type}, | ||
| 1381 | xp|actual-element-type => {$name}, | ||
| 1382 | xp|node => {$node}, | ||
| 1383 | }__; | ||
| 1384 | unless ($continue) { | ||
| 1385 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1386 | } | ||
| 1387 | wakaba | 1.1 | } |
| 1388 | } | ||
| 1389 | } | ||
| 1390 | |||
| 1391 | ~? (S) { } | ||
| 1392 | |||
| 1393 | if-true ($is_docel) { | ||
| 1394 | wakaba | 1.3 | lang:Perl : has-error { |
| 1395 | wakaba | 1.1 | if (@{$types}) { |
| 1396 | wakaba | 1.3 | my $location; |
| 1397 | __CODE{xp|get-location-from-token:: | ||
| 1398 | $token => $token, | ||
| 1399 | $result => $location, | ||
| 1400 | }__; | ||
| 1401 | for my $type (reverse @{$types}) { | ||
| 1402 | my $continue = __DOMCore:ERROR{xp|wf-no-end-tag:: | ||
| 1403 | DOMCore:location => {$location}, | ||
| 1404 | xp|token => {$token}, | ||
| 1405 | xp|expected-element-type => {$type}, | ||
| 1406 | xp|node => {$node}, | ||
| 1407 | }__; | ||
| 1408 | unless ($continue) { | ||
| 1409 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1410 | } | ||
| 1411 | $node = shift @{$nodes}; | ||
| 1412 | } | ||
| 1413 | wakaba | 1.1 | } |
| 1414 | } | ||
| 1415 | return; | ||
| 1416 | } | ||
| 1417 | |||
| 1418 | ~ (TAGC) { | ||
| 1419 | ?lexmode ElementContent; | ||
| 1420 | } else { | ||
| 1421 | ?lexmode 'ElementContent'; | ||
| 1422 | } | ||
| 1423 | |||
| 1424 | } (HCRO) { | ||
| 1425 | &_HexadecimalCharacterReference_ | ||
| 1426 | ($doc => $doc, $parent => $node); | ||
| 1427 | |||
| 1428 | ~ (REFC) { | ||
| 1429 | ?lexmode 'ElementContent'; | ||
| 1430 | } else { | ||
| 1431 | ?lexmode ElementContent; | ||
| 1432 | } | ||
| 1433 | } (CRO) { | ||
| 1434 | &_NumericCharacterReference_ | ||
| 1435 | ($doc => $doc, $parent => $node); | ||
| 1436 | |||
| 1437 | ~ (REFC) { | ||
| 1438 | ?lexmode 'ElementContent'; | ||
| 1439 | } else { | ||
| 1440 | ?lexmode ElementContent; | ||
| 1441 | } | ||
| 1442 | } (ERO) { | ||
| 1443 | wakaba | 1.6 | &_GeneralEntityReferenceEC |
| 1444 | ($doc => $doc, $parent => $node, $ns => $ns); | ||
| 1445 | wakaba | 1.1 | } (CDO) { |
| 1446 | &_CommentDeclaration_ ($doc => $doc, $parent => $node); | ||
| 1447 | |||
| 1448 | ~ (MDC) { | ||
| 1449 | ?lexmode ElementContent; | ||
| 1450 | } else { | ||
| 1451 | ?lexmode ElementContent; | ||
| 1452 | } | ||
| 1453 | } (CDSO) { | ||
| 1454 | &_CDATASection_ ($doc => $doc, $parent => $node); | ||
| 1455 | |||
| 1456 | ~ (MSE) { | ||
| 1457 | ?lexmode 'ElementContent'; | ||
| 1458 | } else { | ||
| 1459 | ?lexmode ElementContent; | ||
| 1460 | } | ||
| 1461 | } (PIO) { | ||
| 1462 | &_ProcessingInstruction_ ($doc => $doc, $parent => $node); | ||
| 1463 | |||
| 1464 | ~ (PIC) { | ||
| 1465 | ?lexmode 'ElementContent'; | ||
| 1466 | } else { | ||
| 1467 | ?lexmode ElementContent; | ||
| 1468 | } | ||
| 1469 | } | ||
| 1470 | wakaba | 1.3 | |
| 1471 | ~ (#NONE) { } | ||
| 1472 | wakaba | 1.1 | } // Element_ |
| 1473 | wakaba | 1.11 | |
| 1474 | /* | ||
| 1475 | content := [CharData] | ||
| 1476 | *((element / Reference / CDSect / PI / Comment) [CharData]) | ||
| 1477 | */ | ||
| 1478 | rule content ($parent, $ns) { | ||
| 1479 | // ?lexmode ElementContent; | ||
| 1480 | |||
| 1481 | my $doc; | ||
| 1482 | lang:Perl { | ||
| 1483 | $doc = $self->{doc}; | ||
| 1484 | } | ||
| 1485 | |||
| 1486 | ~* (CharData) { | ||
| 1487 | lang:Perl ($data => $token.value) { | ||
| 1488 | $parent-><M::Node.appendChild> | ||
| 1489 | ($self->{doc}-><M::Document.createTextNode> ($data)); | ||
| 1490 | } | ||
| 1491 | } (STAGO) { | ||
| 1492 | &Element_ ($doc => $doc, $parent => $parent, $ns => $ns) | ||
| 1493 | : unshift-current-token; | ||
| 1494 | ~ (TAGC) { | ||
| 1495 | ?lexmode ElementContent; | ||
| 1496 | } else { | ||
| 1497 | ?lexmode ElementContent; | ||
| 1498 | } | ||
| 1499 | } (HCRO) { | ||
| 1500 | &_HexadecimalCharacterReference_ | ||
| 1501 | ($doc => $doc, $parent => $parent); | ||
| 1502 | |||
| 1503 | ~ (REFC) { | ||
| 1504 | ?lexmode 'ElementContent'; | ||
| 1505 | } else { | ||
| 1506 | ?lexmode ElementContent; | ||
| 1507 | } | ||
| 1508 | } (CRO) { | ||
| 1509 | &_NumericCharacterReference_ | ||
| 1510 | ($doc => $doc, $parent => $parent); | ||
| 1511 | |||
| 1512 | ~ (REFC) { | ||
| 1513 | ?lexmode 'ElementContent'; | ||
| 1514 | } else { | ||
| 1515 | ?lexmode ElementContent; | ||
| 1516 | } | ||
| 1517 | } (ERO) { | ||
| 1518 | &_GeneralEntityReferenceEC | ||
| 1519 | ($doc => $doc, $parent => $parent, $ns => $ns); | ||
| 1520 | } (CDO) { | ||
| 1521 | &_CommentDeclaration_ ($doc => $doc, $parent => $parent); | ||
| 1522 | |||
| 1523 | ~ (MDC) { | ||
| 1524 | ?lexmode ElementContent; | ||
| 1525 | } else { | ||
| 1526 | ?lexmode ElementContent; | ||
| 1527 | } | ||
| 1528 | } (CDSO) { | ||
| 1529 | &_CDATASection_ ($doc => $doc, $parent => $parent); | ||
| 1530 | |||
| 1531 | ~ (MSE) { | ||
| 1532 | ?lexmode 'ElementContent'; | ||
| 1533 | } else { | ||
| 1534 | ?lexmode ElementContent; | ||
| 1535 | } | ||
| 1536 | } (PIO) { | ||
| 1537 | &_ProcessingInstruction_ ($doc => $doc, $parent => $parent); | ||
| 1538 | |||
| 1539 | ~ (PIC) { | ||
| 1540 | ?lexmode 'ElementContent'; | ||
| 1541 | } else { | ||
| 1542 | ?lexmode ElementContent; | ||
| 1543 | } | ||
| 1544 | } | ||
| 1545 | } // content | ||
| 1546 | wakaba | 1.1 | |
| 1547 | rule AttributeSpecificationList ($doc, $attrs) | ||
| 1548 | : standalone | ||
| 1549 | { | ||
| 1550 | ?lexmode 'StartTag'; | ||
| 1551 | |||
| 1552 | my $i; | ||
| 1553 | lang:Perl { | ||
| 1554 | $i = 0; | ||
| 1555 | } | ||
| 1556 | |||
| 1557 | ~* (Name) { | ||
| 1558 | my $atqname; | ||
| 1559 | lang:Perl ($name => $token.value) { | ||
| 1560 | $atqname = $name; | ||
| 1561 | } | ||
| 1562 | |||
| 1563 | my $vals; | ||
| 1564 | wakaba | 1.18 | lang:Perl : has-error { |
| 1565 | wakaba | 1.1 | if ($attrs->{$atqname}) { |
| 1566 | wakaba | 1.3 | my $location; |
| 1567 | __CODE{xp|get-location-from-token:: | ||
| 1568 | $token => $token, | ||
| 1569 | $result => $location, | ||
| 1570 | }__; | ||
| 1571 | my $continue = __DOMCore:ERROR{xp|wf-unique-att-spec:: | ||
| 1572 | DOMCore:location => {$location}, | ||
| 1573 | xp|token => {$token}, | ||
| 1574 | xp|name => {$atqname}, | ||
| 1575 | }__; | ||
| 1576 | unless ($continue) { | ||
| 1577 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1578 | } | ||
| 1579 | wakaba | 1.1 | } |
| 1580 | |||
| 1581 | $vals = $attrs->{$atqname} = { | ||
| 1582 | nodes => [], | ||
| 1583 | value => '', | ||
| 1584 | index => $i++, | ||
| 1585 | }; | ||
| 1586 | } | ||
| 1587 | wakaba | 1.3 | |
| 1588 | ~? (S) { } | ||
| 1589 | ~ (VI) { } | ||
| 1590 | ~? (S) { } | ||
| 1591 | wakaba | 1.1 | |
| 1592 | ~ (LIT) { | ||
| 1593 | &_AttributeValueSpecification_ | ||
| 1594 | ($doc => $doc, $vals => $vals); | ||
| 1595 | |||
| 1596 | ~ (LIT) { | ||
| 1597 | ?lexmode StartTag; | ||
| 1598 | } else { | ||
| 1599 | ?lexmode StartTag; | ||
| 1600 | } | ||
| 1601 | } (LITA) { | ||
| 1602 | &_AttributeValueSpecificationA_ | ||
| 1603 | ($doc => $doc, $vals => $vals); | ||
| 1604 | |||
| 1605 | ~ (LITA) { | ||
| 1606 | ?lexmode StartTag; | ||
| 1607 | } else { | ||
| 1608 | ?lexmode StartTag; | ||
| 1609 | } | ||
| 1610 | } | ||
| 1611 | } (S) : separator : terminator? { } | ||
| 1612 | } // AttributeSpecificationList | ||
| 1613 | |||
| 1614 | rule _AttributeValueSpecification_ ($doc, $vals) { | ||
| 1615 | // ~ (LIT) { } | ||
| 1616 | ?lexmode 'AttributeValueLiteral'; | ||
| 1617 | |||
| 1618 | ~* (STRING) { | ||
| 1619 | lang:Perl ($value => $token.value) { | ||
| 1620 | $value =~ s/[\x09\x0A\x0D]/ /g; | ||
| 1621 | my $text = $doc-><M::Document.createTextNode> ($value); | ||
| 1622 | push @{$vals->{nodes}}, $text; | ||
| 1623 | $vals->{value} .= $value; | ||
| 1624 | } | ||
| 1625 | } (HCRO) { | ||
| 1626 | &_HexadecimalCharacterReferenceV_ | ||
| 1627 | ($doc => $doc, $vals => $vals); | ||
| 1628 | |||
| 1629 | ~ (REFC) { | ||
| 1630 | ?lexmode AttributeValueLiteral; | ||
| 1631 | } else { | ||
| 1632 | ?lexmode AttributeValueLiteral; | ||
| 1633 | } | ||
| 1634 | } (CRO) { | ||
| 1635 | &_NumericCharacterReferenceV_ | ||
| 1636 | ($doc => $doc, $vals => $vals); | ||
| 1637 | |||
| 1638 | ~ (REFC) { | ||
| 1639 | ?lexmode AttributeValueLiteral; | ||
| 1640 | } else { | ||
| 1641 | ?lexmode AttributeValueLiteral; | ||
| 1642 | } | ||
| 1643 | } (ERO) { | ||
| 1644 | // TODO: Attribute value normalization | ||
| 1645 | &_GeneralEntityReferenceV_ | ||
| 1646 | ($doc => $doc, $vals => $vals); | ||
| 1647 | |||
| 1648 | ~ (REFC) { | ||
| 1649 | ?lexmode AttributeValueLiteral; | ||
| 1650 | } else { | ||
| 1651 | ?lexmode AttributeValueLiteral; | ||
| 1652 | } | ||
| 1653 | } | ||
| 1654 | |||
| 1655 | // ~ (LIT) { } (LITA) { } | ||
| 1656 | } // _AttributeValueSpecification_ | ||
| 1657 | |||
| 1658 | rule _AttributeValueSpecificationA_ ($doc, $vals) { | ||
| 1659 | // ~ (LITA) { } | ||
| 1660 | ?lexmode 'AttributeValueLiteralA'; | ||
| 1661 | |||
| 1662 | ~* (STRING) { | ||
| 1663 | lang:Perl ($value => $token.value) { | ||
| 1664 | $value =~ s/[\x09\x0A\x0D]/ /g; | ||
| 1665 | my $text = $doc-><M::Document.createTextNode> ($value); | ||
| 1666 | push @{$vals->{nodes}}, $text; | ||
| 1667 | $vals->{value} .= $value; | ||
| 1668 | } | ||
| 1669 | } (HCRO) { | ||
| 1670 | &_HexadecimalCharacterReferenceV_ | ||
| 1671 | ($doc => $doc, $vals => $vals); | ||
| 1672 | |||
| 1673 | ~ (REFC) { | ||
| 1674 | ?lexmode AttributeValueLiteralA; | ||
| 1675 | } else { | ||
| 1676 | ?lexmode AttributeValueLiteralA; | ||
| 1677 | } | ||
| 1678 | } (CRO) { | ||
| 1679 | &_NumericCharacterReferenceV_ | ||
| 1680 | ($doc => $doc, $vals => $vals); | ||
| 1681 | |||
| 1682 | ~ (REFC) { | ||
| 1683 | ?lexmode AttributeValueLiteralA; | ||
| 1684 | } else { | ||
| 1685 | ?lexmode AttributeValueLiteralA; | ||
| 1686 | } | ||
| 1687 | } (ERO) { | ||
| 1688 | // TODO: Attribute value normalization | ||
| 1689 | &_GeneralEntityReferenceV_ | ||
| 1690 | ($doc => $doc, $vals => $vals); | ||
| 1691 | |||
| 1692 | ~ (REFC) { | ||
| 1693 | ?lexmode AttributeValueLiteralA; | ||
| 1694 | } else { | ||
| 1695 | ?lexmode AttributeValueLiteralA; | ||
| 1696 | } | ||
| 1697 | } | ||
| 1698 | |||
| 1699 | // ~ (LITA) { } | ||
| 1700 | } // _AttributeValueSpecificationA_ | ||
| 1701 | |||
| 1702 | /* | ||
| 1703 | wakaba | 1.11 | Parsing replacement text of an entity referenced in |
| 1704 | an attribute value specification. | ||
| 1705 | */ | ||
| 1706 | rule AttributeValueLiteralE_ ($parent, $vals) | ||
| 1707 | : recursive | ||
| 1708 | { | ||
| 1709 | // ?lexmode AttributeValueLiteralE; | ||
| 1710 | |||
| 1711 | ~* (STRING) { | ||
| 1712 | lang:Perl ($value => $token.value) { | ||
| 1713 | $value =~ s/[\x09\x0A\x0D]/ /g; | ||
| 1714 | my $text = $self->{doc}-><M::Document.createTextNode> ($value); | ||
| 1715 | $parent-><M::Node.appendChild> ($text); | ||
| 1716 | $vals->{value} .= $value; | ||
| 1717 | } | ||
| 1718 | } (HCRO) { | ||
| 1719 | &_HexadecimalCharacterReferenceAE_ | ||
| 1720 | ($parent => $parent, $vals => $vals); | ||
| 1721 | |||
| 1722 | ~ (REFC) { | ||
| 1723 | ?lexmode AttributeValueLiteralE; | ||
| 1724 | } else { | ||
| 1725 | ?lexmode AttributeValueLiteralE; | ||
| 1726 | } | ||
| 1727 | } (CRO) { | ||
| 1728 | &_NumericCharacterReferenceAE_ | ||
| 1729 | ($parent => $parent, $vals => $vals); | ||
| 1730 | |||
| 1731 | ~ (REFC) { | ||
| 1732 | ?lexmode AttributeValueLiteralE; | ||
| 1733 | } else { | ||
| 1734 | ?lexmode AttributeValueLiteralE; | ||
| 1735 | } | ||
| 1736 | } (ERO) { | ||
| 1737 | &_GeneralEntityReferenceAE_ ($parent => $parent, $vals => $vals); | ||
| 1738 | |||
| 1739 | ~ (REFC) { | ||
| 1740 | ?lexmode AttributeValueLiteralE; | ||
| 1741 | } else { | ||
| 1742 | ?lexmode AttributeValueLiteralE; | ||
| 1743 | } | ||
| 1744 | wakaba | 1.18 | } (#INVALID == '<') { |
| 1745 | lang:Perl : has-error { | ||
| 1746 | my $location; | ||
| 1747 | __CODE{xp|get-location-from-token:: | ||
| 1748 | $token => $token, | ||
| 1749 | $result => $location, | ||
| 1750 | }__; | ||
| 1751 | my $continue = __DOMCore:ERROR{xp|wf-no-lt-in-attribute-values:: | ||
| 1752 | c|location => {$location}, | ||
| 1753 | xp|token => {$token}, | ||
| 1754 | }__; | ||
| 1755 | unless ($continue) { | ||
| 1756 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1757 | } | ||
| 1758 | |||
| 1759 | my $text = $self->{doc}-><M::Document.createTextNode> ('<'); | ||
| 1760 | $parent-><M::Node.appendChild> ($text); | ||
| 1761 | $vals->{value} .= '<'; | ||
| 1762 | } | ||
| 1763 | wakaba | 1.11 | } |
| 1764 | |||
| 1765 | // ~ (#EOF) { } | ||
| 1766 | } // AttributeValueLiteralE_ | ||
| 1767 | |||
| 1768 | /* | ||
| 1769 | wakaba | 1.1 | CDATA Section Content Parsing Mode |
| 1770 | */ | ||
| 1771 | rule _CDATASection_ ($doc, $parent) { | ||
| 1772 | ?lexmode 'CDATASectionContent'; | ||
| 1773 | |||
| 1774 | my $cdata; | ||
| 1775 | |||
| 1776 | wakaba | 1.2 | ~? (CData) { |
| 1777 | wakaba | 1.1 | lang:Perl ($data => $token.value) { |
| 1778 | $cdata = $data; | ||
| 1779 | } | ||
| 1780 | } else { | ||
| 1781 | lang:Perl { | ||
| 1782 | $cdata = ''; | ||
| 1783 | } | ||
| 1784 | } | ||
| 1785 | |||
| 1786 | lang:Perl { | ||
| 1787 | my $cdsect = $doc-><M::Document.createCDATASection> | ||
| 1788 | ($cdata); | ||
| 1789 | $parent-><M::Node.appendChild> ($cdsect); | ||
| 1790 | } | ||
| 1791 | |||
| 1792 | // ~ (MSE) { } | ||
| 1793 | } // _CDATASection_ | ||
| 1794 | |||
| 1795 | rule _NumericCharacterReference_ ($doc, $parent) { | ||
| 1796 | ?lexmode 'NumericCharacterReference'; | ||
| 1797 | |||
| 1798 | ~ (NUMBER) { | ||
| 1799 | wakaba | 1.3 | lang:Perl ($num => $token.value) : has-error { |
| 1800 | $num += 0; | ||
| 1801 | unless ( | ||
| 1802 | ($self->{xml_version} eq '1.0' and | ||
| 1803 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 1804 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1805 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 1806 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 1807 | ($self->{xml_version} eq '1.1' and | ||
| 1808 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 1809 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1810 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 1811 | ) { | ||
| 1812 | my $location; | ||
| 1813 | __CODE{xp|get-location-from-token:: | ||
| 1814 | $token => $token, | ||
| 1815 | $result => $location, | ||
| 1816 | }__; | ||
| 1817 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 1818 | DOMCore:location => {$location}, | ||
| 1819 | xp|token => {$token}, | ||
| 1820 | xp|character-number => {$num}, | ||
| 1821 | xp|parent => {$parent}, | ||
| 1822 | }__; | ||
| 1823 | unless ($continue) { | ||
| 1824 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1825 | } | ||
| 1826 | } | ||
| 1827 | my $ncr = $doc-><M::Document.createTextNode> (chr $num); | ||
| 1828 | wakaba | 1.1 | $parent-><M::Node.appendChild> ($ncr); |
| 1829 | } | ||
| 1830 | } | ||
| 1831 | |||
| 1832 | // ~ (REFC) { } | ||
| 1833 | } // _NumericCharacterReference_ | ||
| 1834 | |||
| 1835 | rule _NumericCharacterReferenceV_ ($doc, $vals) { | ||
| 1836 | ?lexmode 'NumericCharacterReference'; | ||
| 1837 | |||
| 1838 | ~ (NUMBER) { | ||
| 1839 | wakaba | 1.3 | lang:Perl ($num => $token.value) : has-error { |
| 1840 | $num += 0; | ||
| 1841 | unless ( | ||
| 1842 | ($self->{xml_version} eq '1.0' and | ||
| 1843 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 1844 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1845 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 1846 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 1847 | ($self->{xml_version} eq '1.1' and | ||
| 1848 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 1849 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1850 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 1851 | ) { | ||
| 1852 | my $location; | ||
| 1853 | __CODE{xp|get-location-from-token:: | ||
| 1854 | $token => $token, | ||
| 1855 | $result => $location, | ||
| 1856 | }__; | ||
| 1857 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 1858 | DOMCore:location => {$location}, | ||
| 1859 | xp|token => {$token}, | ||
| 1860 | xp|character-number => {$num}, | ||
| 1861 | }__; | ||
| 1862 | unless ($continue) { | ||
| 1863 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1864 | } | ||
| 1865 | } | ||
| 1866 | wakaba | 1.1 | my $ncr = $doc-><M::Document.createTextNode> |
| 1867 | wakaba | 1.9 | (my $char = chr $num); |
| 1868 | wakaba | 1.1 | push @{$vals->{nodes}}, $ncr; |
| 1869 | $vals->{value} .= $char; | ||
| 1870 | } | ||
| 1871 | } | ||
| 1872 | |||
| 1873 | // ~ (REFC) { } | ||
| 1874 | } // _NumericCharacterReferenceV_ | ||
| 1875 | |||
| 1876 | wakaba | 1.11 | /* |
| 1877 | Numeric character reference in the replacement text | ||
| 1878 | of the entity referenced in an attribute value specification | ||
| 1879 | */ | ||
| 1880 | rule _NumericCharacterReferenceAE_ ($parent, $vals) | ||
| 1881 | { | ||
| 1882 | ?lexmode NumericCharacterReference; | ||
| 1883 | |||
| 1884 | ~ (NUMBER) { | ||
| 1885 | lang:Perl ($num => $token.value) : has-error { | ||
| 1886 | $num += 0; | ||
| 1887 | unless ( | ||
| 1888 | ($self->{xml_version} eq '1.0' and | ||
| 1889 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 1890 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1891 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 1892 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 1893 | ($self->{xml_version} eq '1.1' and | ||
| 1894 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 1895 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1896 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 1897 | ) { | ||
| 1898 | my $location; | ||
| 1899 | __CODE{xp|get-location-from-token:: | ||
| 1900 | $token => $token, | ||
| 1901 | $result => $location, | ||
| 1902 | }__; | ||
| 1903 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 1904 | DOMCore:location => {$location}, | ||
| 1905 | xp|token => {$token}, | ||
| 1906 | xp|character-number => {$num}, | ||
| 1907 | }__; | ||
| 1908 | unless ($continue) { | ||
| 1909 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1910 | } | ||
| 1911 | } | ||
| 1912 | my $ncr = $self->{doc}-><M::Document.createTextNode> | ||
| 1913 | (my $char = chr $num); | ||
| 1914 | $parent-><M::Node.appendChild> ($ncr); | ||
| 1915 | $vals->{value} .= $char; | ||
| 1916 | } | ||
| 1917 | } | ||
| 1918 | |||
| 1919 | // ~ (REFC) { } | ||
| 1920 | } // _NumericCharacterReferenceAE_ | ||
| 1921 | |||
| 1922 | rule _NumericCharacterReferenceEV_ ($vals) | ||
| 1923 | { | ||
| 1924 | ?lexmode 'NumericCharacterReference'; | ||
| 1925 | |||
| 1926 | ~ (NUMBER) { | ||
| 1927 | lang:Perl ($num => $token.value) : has-error { | ||
| 1928 | $num += 0; | ||
| 1929 | unless ( | ||
| 1930 | ($self->{xml_version} eq '1.0' and | ||
| 1931 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 1932 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1933 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 1934 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 1935 | ($self->{xml_version} eq '1.1' and | ||
| 1936 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 1937 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1938 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 1939 | ) { | ||
| 1940 | my $location; | ||
| 1941 | __CODE{xp|get-location-from-token:: | ||
| 1942 | $token => $token, | ||
| 1943 | $result => $location, | ||
| 1944 | }__; | ||
| 1945 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 1946 | DOMCore:location => {$location}, | ||
| 1947 | xp|token => {$token}, | ||
| 1948 | xp|character-number => {$num}, | ||
| 1949 | }__; | ||
| 1950 | unless ($continue) { | ||
| 1951 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1952 | } | ||
| 1953 | } | ||
| 1954 | push @{$vals}, chr $num; | ||
| 1955 | } | ||
| 1956 | } | ||
| 1957 | |||
| 1958 | // ~ (REFC) { } | ||
| 1959 | } // _NumericCharacterReferenceEV_ | ||
| 1960 | |||
| 1961 | wakaba | 1.1 | rule _HexadecimalCharacterReference_ ($doc, $parent) { |
| 1962 | ?lexmode 'HexadecimalCharacterReference'; | ||
| 1963 | |||
| 1964 | ~ (Hex) { | ||
| 1965 | wakaba | 1.9 | lang:Perl ($v => $token.value) : has-error { |
| 1966 | my $num = hex $v; | ||
| 1967 | wakaba | 1.3 | unless ( |
| 1968 | ($self->{xml_version} eq '1.0' and | ||
| 1969 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 1970 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1971 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 1972 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 1973 | ($self->{xml_version} eq '1.1' and | ||
| 1974 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 1975 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 1976 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 1977 | ) { | ||
| 1978 | my $location; | ||
| 1979 | __CODE{xp|get-location-from-token:: | ||
| 1980 | $token => $token, | ||
| 1981 | $result => $location, | ||
| 1982 | }__; | ||
| 1983 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 1984 | DOMCore:location => {$location}, | ||
| 1985 | xp|token => {$token}, | ||
| 1986 | xp|character-number => {$num}, | ||
| 1987 | xp|parent => {$parent}, | ||
| 1988 | }__; | ||
| 1989 | unless ($continue) { | ||
| 1990 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 1991 | } | ||
| 1992 | } | ||
| 1993 | wakaba | 1.1 | my $ncr = $doc-><M::Document.createTextNode> |
| 1994 | wakaba | 1.9 | (chr $num); |
| 1995 | wakaba | 1.1 | $parent-><M::Node.appendChild> ($ncr); |
| 1996 | } | ||
| 1997 | } | ||
| 1998 | |||
| 1999 | // ~ (REFC) { } | ||
| 2000 | } // _HexadecimalCharacterReference_ | ||
| 2001 | |||
| 2002 | wakaba | 1.3 | rule _HexadecimalCharacterReferenceV_ ($doc, $vals) { |
| 2003 | wakaba | 1.1 | ?lexmode 'HexadecimalCharacterReference'; |
| 2004 | |||
| 2005 | ~ (Hex) { | ||
| 2006 | wakaba | 1.9 | lang:Perl ($v => $token.value) : has-error { |
| 2007 | my $num = hex $v; | ||
| 2008 | wakaba | 1.3 | unless ( |
| 2009 | ($self->{xml_version} eq '1.0' and | ||
| 2010 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 2011 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 2012 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 2013 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 2014 | ($self->{xml_version} eq '1.1' and | ||
| 2015 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 2016 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 2017 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 2018 | ) { | ||
| 2019 | my $location; | ||
| 2020 | __CODE{xp|get-location-from-token:: | ||
| 2021 | $token => $token, | ||
| 2022 | $result => $location, | ||
| 2023 | }__; | ||
| 2024 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 2025 | DOMCore:location => {$location}, | ||
| 2026 | xp|token => {$token}, | ||
| 2027 | xp|character-number => {$num}, | ||
| 2028 | }__; | ||
| 2029 | unless ($continue) { | ||
| 2030 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2031 | } | ||
| 2032 | } | ||
| 2033 | wakaba | 1.1 | my $ncr = $doc-><M::Document.createTextNode> |
| 2034 | wakaba | 1.9 | (my $char = chr $num); |
| 2035 | wakaba | 1.1 | push @{$vals->{nodes}}, $ncr; |
| 2036 | $vals->{value} .= $char; | ||
| 2037 | } | ||
| 2038 | } | ||
| 2039 | |||
| 2040 | // ~ (REFC) { } | ||
| 2041 | wakaba | 1.11 | } // _HexadecimalCharacterReferenceV_ |
| 2042 | |||
| 2043 | /* | ||
| 2044 | Hexadecimal character reference in the replacement text | ||
| 2045 | of the entity referenced in an attribute value literal | ||
| 2046 | */ | ||
| 2047 | rule _HexadecimalCharacterReferenceAE_ ($parent, $vals) | ||
| 2048 | { | ||
| 2049 | ?lexmode HexadecimalCharacterReference; | ||
| 2050 | |||
| 2051 | ~ (Hex) { | ||
| 2052 | lang:Perl ($v => $token.value) : has-error { | ||
| 2053 | my $num = hex $v; | ||
| 2054 | unless ( | ||
| 2055 | ($self->{xml_version} eq '1.0' and | ||
| 2056 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 2057 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 2058 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 2059 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 2060 | ($self->{xml_version} eq '1.1' and | ||
| 2061 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 2062 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 2063 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 2064 | ) { | ||
| 2065 | my $location; | ||
| 2066 | __CODE{xp|get-location-from-token:: | ||
| 2067 | $token => $token, | ||
| 2068 | $result => $location, | ||
| 2069 | }__; | ||
| 2070 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 2071 | DOMCore:location => {$location}, | ||
| 2072 | xp|token => {$token}, | ||
| 2073 | xp|character-number => {$num}, | ||
| 2074 | }__; | ||
| 2075 | unless ($continue) { | ||
| 2076 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2077 | } | ||
| 2078 | } | ||
| 2079 | my $ncr = $self->{doc}-><M::Document.createTextNode> | ||
| 2080 | (my $char = chr $num); | ||
| 2081 | $parent-><M::Node.appendChild> ($ncr); | ||
| 2082 | $vals->{value} .= $char; | ||
| 2083 | } | ||
| 2084 | } | ||
| 2085 | |||
| 2086 | // ~ (REFC) { } | ||
| 2087 | } // _HexadecimalCharacterReferenceAE_ | ||
| 2088 | |||
| 2089 | rule _HexadecimalCharacterReferenceEV_ ($vals) { | ||
| 2090 | ?lexmode HexadecimalCharacterReference; | ||
| 2091 | |||
| 2092 | ~ (Hex) { | ||
| 2093 | lang:Perl ($v => $token.value) : has-error { | ||
| 2094 | my $num = hex $v; | ||
| 2095 | unless ( | ||
| 2096 | ($self->{xml_version} eq '1.0' and | ||
| 2097 | ((0x0020 <= $num and $num <= 0xD7FF) or | ||
| 2098 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 2099 | (0x10000 <= $num and $num <= 0x10FFFF) or | ||
| 2100 | $num == 0x9 or $num == 0xA or $num == 0xD)) or | ||
| 2101 | ($self->{xml_version} eq '1.1' and | ||
| 2102 | ((0x0001 <= $num and $num <= 0xD7FF) or | ||
| 2103 | (0xE000 <= $num and $num <= 0xFFFD) or | ||
| 2104 | (0x10000 <= $num and $num <= 0x10FFFF))) | ||
| 2105 | ) { | ||
| 2106 | my $location; | ||
| 2107 | __CODE{xp|get-location-from-token:: | ||
| 2108 | $token => $token, | ||
| 2109 | $result => $location, | ||
| 2110 | }__; | ||
| 2111 | my $continue = __DOMCore:ERROR{xp|wf-legal-character:: | ||
| 2112 | DOMCore:location => {$location}, | ||
| 2113 | xp|token => {$token}, | ||
| 2114 | xp|character-number => {$num}, | ||
| 2115 | }__; | ||
| 2116 | unless ($continue) { | ||
| 2117 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2118 | } | ||
| 2119 | } | ||
| 2120 | push @{$vals}, chr $num; | ||
| 2121 | } | ||
| 2122 | } | ||
| 2123 | |||
| 2124 | // ~ (REFC) { } | ||
| 2125 | } // _HexadecimalCharacterReferenceEV_ | ||
| 2126 | |||
| 2127 | /* | ||
| 2128 | General entity reference in element's content | ||
| 2129 | */ | ||
| 2130 | rule _GeneralEntityReferenceEC ($doc, $parent, $ns) | ||
| 2131 | : recursive | ||
| 2132 | { | ||
| 2133 | ?lexmode 'EntityReference'; | ||
| 2134 | |||
| 2135 | ~ (Name == 'lt') { | ||
| 2136 | lang:Perl { | ||
| 2137 | $parent-><M::Node.appendChild> | ||
| 2138 | ($self->{doc}-><M::Document.createTextNode> ('<')); | ||
| 2139 | } | ||
| 2140 | } (Name == 'gt') { | ||
| 2141 | lang:Perl { | ||
| 2142 | $parent-><M::Node.appendChild> | ||
| 2143 | ($self->{doc}-><M::Document.createTextNode> ('>')); | ||
| 2144 | } | ||
| 2145 | } (Name == 'amp') { | ||
| 2146 | lang:Perl { | ||
| 2147 | $parent-><M::Node.appendChild> | ||
| 2148 | ($self->{doc}-><M::Document.createTextNode> ('&')); | ||
| 2149 | } | ||
| 2150 | } (Name == 'quot') { | ||
| 2151 | lang:Perl { | ||
| 2152 | $parent-><M::Node.appendChild> | ||
| 2153 | ($self->{doc}-><M::Document.createTextNode> ('"')); | ||
| 2154 | } | ||
| 2155 | } (Name == 'apos') { | ||
| 2156 | lang:Perl { | ||
| 2157 | $parent-><M::Node.appendChild> | ||
| 2158 | ($self->{doc}-><M::Document.createTextNode> ("'")); | ||
| 2159 | } | ||
| 2160 | } (Name) { | ||
| 2161 | my $er; | ||
| 2162 | wakaba | 1.19 | lang:Perl ($name => $token.value) : has-error { |
| 2163 | wakaba | 1.11 | ## TODO: Namespace well-formedness |
| 2164 | $er = $doc-><M::Document.createEntityReference> | ||
| 2165 | ($name); | ||
| 2166 | wakaba | 1.20 | $er-><AS::Node.textContent> (''); |
| 2167 | ## NOTE: When document entity (and entities referenced directly | ||
| 2168 | ## or indirectly from it) is parsed, no general entity | ||
| 2169 | ## node have its replacement tree. During general | ||
| 2170 | ## entity node collection construction, however, | ||
| 2171 | ## some entity node has replacement tree. | ||
| 2172 | wakaba | 1.11 | $parent-><M::Node.appendChild> ($er); |
| 2173 | wakaba | 1.19 | |
| 2174 | my $ent = $self->{general_entity}->{$name}; | ||
| 2175 | if (not $ent) { # no entity declaration | ||
| 2176 | if ($self->{standalone} or not $self->{has_xref}) { # WFC error | ||
| 2177 | my $location; | ||
| 2178 | __CODE{xp|get-location-from-token:: | ||
| 2179 | $token => {$token}, | ||
| 2180 | $result => {$location}, | ||
| 2181 | }__; | ||
| 2182 | my $continue = __DOMCore:ERROR{xp|wf-entity-declared:: | ||
| 2183 | xp|error-token => {$token}, | ||
| 2184 | c|location => {$location}, | ||
| 2185 | xp|name => {$name}, | ||
| 2186 | }__; | ||
| 2187 | unless ($continue) { | ||
| 2188 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2189 | } | ||
| 2190 | } # Otherwise VC error | ||
| 2191 | wakaba | 1.23 | push @{$self->{entity}}, <Code::getEmptyEntityState:: |
| 2192 | $name = $name>; | ||
| 2193 | wakaba | 1.20 | $er-><AS::x|EntityReference.isExpanded> (false); |
| 2194 | wakaba | 1.19 | |
| 2195 | } else { # there IS entity declaration | ||
| 2196 | if (($self->{standalone} or not $self->{has_xref}) and | ||
| 2197 | $ent->{is_externally_declared} and | ||
| 2198 | not $self->{entity}->[-1]->{is_externally_declared}) { | ||
| 2199 | my $location; | ||
| 2200 | __CODE{xp|get-location-from-token:: | ||
| 2201 | $token => {$token}, | ||
| 2202 | $result => {$location}, | ||
| 2203 | }__; | ||
| 2204 | my $continue = __DOMCore:ERROR{xp|wf-entity-declared:: | ||
| 2205 | xp|error-token => {$token}, | ||
| 2206 | c|location => {$location}, | ||
| 2207 | xp|name => {$name}, | ||
| 2208 | }__; | ||
| 2209 | unless ($continue) { | ||
| 2210 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2211 | } | ||
| 2212 | } | ||
| 2213 | $er-><AS::x|EntityReference.isExpanded> | ||
| 2214 | ($ent->{has_replacement_text}); | ||
| 2215 | push @{$self->{entity}}, <Code::getCopyOfEntityState:: | ||
| 2216 | wakaba | 1.11 | $entity_type = 'general_entity', |
| 2217 | $entity_name = $name>; | ||
| 2218 | wakaba | 1.23 | |
| 2219 | if (defined $ent->{notation}) { | ||
| 2220 | my $location; | ||
| 2221 | __CODE{xp|get-location-from-token:: | ||
| 2222 | $token => {$token}, | ||
| 2223 | $result => {$location}, | ||
| 2224 | }__; | ||
| 2225 | my $continue = __c|ERROR{xp|wf-parsed-entity:: | ||
| 2226 | xp|error-token => {$token}, | ||
| 2227 | c|location => {$location}, | ||
| 2228 | xp|name => {$name}, | ||
| 2229 | }__; | ||
| 2230 | unless ($continue) { | ||
| 2231 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2232 | }__; | ||
| 2233 | } | ||
| 2234 | } # if unparsed entity | ||
| 2235 | |||
| 2236 | if ($ent->{is_opened}) { | ||
| 2237 | my $location; | ||
| 2238 | __CODE{xp|get-location-from-token:: | ||
| 2239 | $token => {$token}, | ||
| 2240 | $result => {$location}, | ||
| 2241 | }__; | ||
| 2242 | my $continue = __c|ERROR{xp|wf-no-recursion:: | ||
| 2243 | xp|error-token => {$token}, | ||
| 2244 | c|location => {$location}, | ||
| 2245 | xp|name => {$name}, | ||
| 2246 | }__; | ||
| 2247 | unless ($continue) { | ||
| 2248 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2249 | }__; | ||
| 2250 | } | ||
| 2251 | $self->{entity}->[-1]->{reptxt} = \''; | ||
| 2252 | $er-><AS::x|EntityReference.isExpanded> (false); | ||
| 2253 | } # recursive | ||
| 2254 | $ent->{is_opened} = true; | ||
| 2255 | } # entity declared | ||
| 2256 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 2257 | wakaba | 1.11 | push @{$self->{entity_token}}, $self->{token}; |
| 2258 | $self->{token} = []; | ||
| 2259 | push @{$self->{entity_char}}, $self->{char}; | ||
| 2260 | $self->{char} = []; | ||
| 2261 | } | ||
| 2262 | |||
| 2263 | ?lexmode ElementContent; | ||
| 2264 | &content ($doc => $doc, $parent => $er, $ns => $ns); | ||
| 2265 | ~ (#EOF) { } | ||
| 2266 | lang:Perl { | ||
| 2267 | wakaba | 1.23 | $self->{general_entity}->{$self->{entity}->[-1]->{name}} |
| 2268 | ->{is_opened} = false | ||
| 2269 | if $self->{general_entity}->{$self->{entity}->[-1]->{name}}; | ||
| 2270 | wakaba | 1.11 | $self->{token} = pop @{$self->{entity_token}}; |
| 2271 | $self->{char} = pop @{$self->{entity_char}}; | ||
| 2272 | pop @{$self->{entity}}; | ||
| 2273 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 2274 | wakaba | 1.11 | } |
| 2275 | |||
| 2276 | // TODO: Set read-only flag | ||
| 2277 | |||
| 2278 | ?lexmode EntityReference; | ||
| 2279 | ?requires-next-token; | ||
| 2280 | } | ||
| 2281 | |||
| 2282 | ~ (REFC) { | ||
| 2283 | ?lexmode ElementContent; | ||
| 2284 | } else { | ||
| 2285 | ?lexmode ElementContent; | ||
| 2286 | } | ||
| 2287 | } // _GeneralEntityReferenceEC | ||
| 2288 | |||
| 2289 | /* | ||
| 2290 | General entity reference in an attribute value literal | ||
| 2291 | */ | ||
| 2292 | rule _GeneralEntityReferenceV_ ($vals) { | ||
| 2293 | ?lexmode EntityReference; | ||
| 2294 | |||
| 2295 | ~ (Name == 'lt') { | ||
| 2296 | lang:Perl { | ||
| 2297 | push @{$vals->{nodes}}, $self->{doc}-><M::Document.createTextNode> | ||
| 2298 | ('<'); | ||
| 2299 | $vals->{value} .= '<'; | ||
| 2300 | } | ||
| 2301 | } (Name == 'gt') { | ||
| 2302 | lang:Perl { | ||
| 2303 | push @{$vals->{nodes}}, $self->{doc}-><M::Document.createTextNode> | ||
| 2304 | ('>'); | ||
| 2305 | $vals->{value} .= '>'; | ||
| 2306 | } | ||
| 2307 | } (Name == 'amp') { | ||
| 2308 | lang:Perl { | ||
| 2309 | push @{$vals->{nodes}}, $self->{doc}-><M::Document.createTextNode> | ||
| 2310 | ('&'); | ||
| 2311 | $vals->{value} .= '&'; | ||
| 2312 | } | ||
| 2313 | } (Name == 'quot') { | ||
| 2314 | lang:Perl { | ||
| 2315 | push @{$vals->{nodes}}, $self->{doc}-><M::Document.createTextNode> | ||
| 2316 | ('"'); | ||
| 2317 | $vals->{value} .= '"'; | ||
| 2318 | } | ||
| 2319 | } (Name == 'apos') { | ||
| 2320 | lang:Perl { | ||
| 2321 | push @{$vals->{nodes}}, $self->{doc}-><M::Document.createTextNode> | ||
| 2322 | ("'"); | ||
| 2323 | $vals->{value} .= "'"; | ||
| 2324 | } | ||
| 2325 | } (Name) { | ||
| 2326 | my $er; | ||
| 2327 | wakaba | 1.18 | lang:Perl ($name => $token.value) : has-error { |
| 2328 | wakaba | 1.11 | ## TODO: Namespace well-formedness |
| 2329 | wakaba | 1.18 | |
| 2330 | wakaba | 1.11 | $er = $self->{doc}-><M::Document.createEntityReference> ($name); |
| 2331 | wakaba | 1.20 | $er-><AS::Node.textContent> (''); |
| 2332 | ## NOTE: When document entity (and entities referenced directly | ||
| 2333 | ## or indirectly from it) is parsed, no general entity | ||
| 2334 | ## node have its replacement tree. During general | ||
| 2335 | ## entity node collection construction, however, | ||
| 2336 | ## some entity node has replacement tree. | ||
| 2337 | wakaba | 1.19 | my $ent = $self->{general_entity}->{$name}; |
| 2338 | wakaba | 1.20 | if (not $ent) { # no entity declaration |
| 2339 | if ($self->{standalone} or not $self->{has_xref}) { # WFC error | ||
| 2340 | my $location; | ||
| 2341 | __CODE{xp|get-location-from-token:: | ||
| 2342 | $token => {$token}, | ||
| 2343 | $result => {$location}, | ||
| 2344 | }__; | ||
| 2345 | my $continue = __DOMCore:ERROR{xp|wf-entity-declared:: | ||
| 2346 | xp|error-token => {$token}, | ||
| 2347 | c|location => {$location}, | ||
| 2348 | xp|name => {$name}, | ||
| 2349 | }__; | ||
| 2350 | unless ($continue) { | ||
| 2351 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2352 | } | ||
| 2353 | } # Otherwise VC error | ||
| 2354 | wakaba | 1.23 | push @{$self->{entity}}, <Code::getEmptyEntityState:: |
| 2355 | $name = $name>; | ||
| 2356 | wakaba | 1.20 | $er-><AS::x|EntityReference.isExpanded> (false); |
| 2357 | |||
| 2358 | } else { # there IS entity declaration | ||
| 2359 | if (($self->{standalone} or not $self->{has_xref}) and | ||
| 2360 | $ent->{is_externally_declared} and | ||
| 2361 | not $self->{entity}->[-1]->{is_externally_declared}) { | ||
| 2362 | my $location; | ||
| 2363 | __CODE{xp|get-location-from-token:: | ||
| 2364 | $token => {$token}, | ||
| 2365 | $result => {$location}, | ||
| 2366 | }__; | ||
| 2367 | my $continue = __DOMCore:ERROR{xp|wf-entity-declared:: | ||
| 2368 | xp|error-token => {$token}, | ||
| 2369 | c|location => {$location}, | ||
| 2370 | xp|name => {$name}, | ||
| 2371 | }__; | ||
| 2372 | unless ($continue) { | ||
| 2373 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2374 | } | ||
| 2375 | } | ||
| 2376 | |||
| 2377 | if ($ent->{is_external_entity}) { | ||
| 2378 | my $location; | ||
| 2379 | __CODE{xp|get-location-from-token:: | ||
| 2380 | $token => {$token}, | ||
| 2381 | $result => {$location}, | ||
| 2382 | }__; | ||
| 2383 | my $continue = __c|ERROR{xp|wf-no-external-entity-references:: | ||
| 2384 | xp|error-token => {$token}, | ||
| 2385 | c|location => {$location}, | ||
| 2386 | xp|name => {$name}, | ||
| 2387 | }__; | ||
| 2388 | unless ($continue) { | ||
| 2389 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2390 | }__; | ||
| 2391 | } | ||
| 2392 | wakaba | 1.23 | ## NOTE: |xp:wf-parsed-entity| is not checked |
| 2393 | wakaba | 1.20 | } # if external entity |
| 2394 | |||
| 2395 | $er-><AS::x|EntityReference.isExpanded> | ||
| 2396 | wakaba | 1.19 | ($ent->{has_replacement_text}); |
| 2397 | wakaba | 1.20 | push @{$self->{entity}}, <Code::getCopyOfEntityState:: |
| 2398 | wakaba | 1.11 | $entity_type = 'general_entity', |
| 2399 | $entity_name = $name>; | ||
| 2400 | wakaba | 1.23 | |
| 2401 | if ($ent->{is_opened}) { | ||
| 2402 | my $location; | ||
| 2403 | __CODE{xp|get-location-from-token:: | ||
| 2404 | $token => {$token}, | ||
| 2405 | $result => {$location}, | ||
| 2406 | }__; | ||
| 2407 | my $continue = __c|ERROR{xp|wf-no-recursion:: | ||
| 2408 | xp|error-token => {$token}, | ||
| 2409 | c|location => {$location}, | ||
| 2410 | xp|name => {$name}, | ||
| 2411 | }__; | ||
| 2412 | unless ($continue) { | ||
| 2413 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2414 | }__; | ||
| 2415 | } | ||
| 2416 | $self->{entity}->[-1]->{reptxt} = \''; | ||
| 2417 | $er-><AS::x|EntityReference.isExpanded> (false); | ||
| 2418 | } # recursive | ||
| 2419 | $ent->{is_opened} = true; | ||
| 2420 | wakaba | 1.20 | } |
| 2421 | push @{$vals->{nodes}}, $er; | ||
| 2422 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 2423 | wakaba | 1.11 | push @{$self->{entity_token}}, $self->{token}; |
| 2424 | $self->{token} = []; | ||
| 2425 | push @{$self->{entity_char}}, $self->{char}; | ||
| 2426 | $self->{char} = []; | ||
| 2427 | } | ||
| 2428 | |||
| 2429 | ?lexmode AttributeValueLiteralE; | ||
| 2430 | &AttributeValueLiteralE_ ($parent => $er, $vals => $vals); | ||
| 2431 | ~ (#EOF) { } | ||
| 2432 | lang:Perl { | ||
| 2433 | wakaba | 1.23 | $self->{general_entity}->{$self->{entity}->[-1]->{name}} |
| 2434 | ->{is_opened} = false | ||
| 2435 | if $self->{general_entity}->{$self->{entity}->[-1]->{name}}; | ||
| 2436 | wakaba | 1.11 | $self->{token} = pop @{$self->{entity_token}}; |
| 2437 | $self->{char} = pop @{$self->{entity_char}}; | ||
| 2438 | pop @{$self->{entity}}; | ||
| 2439 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 2440 | wakaba | 1.11 | } |
| 2441 | |||
| 2442 | // TODO: Set read-only flag | ||
| 2443 | |||
| 2444 | ?lexmode EntityReference; | ||
| 2445 | ?requires-next-token; | ||
| 2446 | } | ||
| 2447 | |||
| 2448 | // ~ (REFC) { } | ||
| 2449 | } // _GeneralEntityReferenceV_ | ||
| 2450 | wakaba | 1.1 | |
| 2451 | wakaba | 1.6 | /* |
| 2452 | wakaba | 1.11 | General entity reference in the replacement text |
| 2453 | of the entity referenced in an attribute value literal | ||
| 2454 | wakaba | 1.6 | */ |
| 2455 | wakaba | 1.11 | rule _GeneralEntityReferenceAE_ ($parent, $vals) |
| 2456 | wakaba | 1.6 | { |
| 2457 | wakaba | 1.11 | ?lexmode EntityReference; |
| 2458 | wakaba | 1.1 | |
| 2459 | wakaba | 1.11 | ~ (Name == 'lt') { |
| 2460 | lang:Perl { | ||
| 2461 | $parent-><M::Node.appendChild> | ||
| 2462 | ($self->{doc}-><M::Document.createTextNode> ('<')); | ||
| 2463 | $vals->{value} .= '<'; | ||
| 2464 | } | ||
| 2465 | } (Name == 'gt') { | ||
| 2466 | lang:Perl { | ||
| 2467 | $parent-><M::Node.appendChild> | ||
| 2468 | ($self->{doc}-><M::Document.createTextNode> ('>')); | ||
| 2469 | $vals->{value} .= '>'; | ||
| 2470 | } | ||
| 2471 | } (Name == 'amp') { | ||
| 2472 | lang:Perl { | ||
| 2473 | $parent-><M::Node.appendChild> | ||
| 2474 | ($self->{doc}-><M::Document.createTextNode> ('&')); | ||
| 2475 | $vals->{value} .= '&'; | ||
| 2476 | } | ||
| 2477 | } (Name == 'quot') { | ||
| 2478 | lang:Perl { | ||
| 2479 | $parent-><M::Node.appendChild> | ||
| 2480 | ($self->{doc}-><M::Document.createTextNode> ('"')); | ||
| 2481 | $vals->{value} .= '"'; | ||
| 2482 | } | ||
| 2483 | } (Name == 'apos') { | ||
| 2484 | lang:Perl { | ||
| 2485 | $parent-><M::Node.appendChild> | ||
| 2486 | ($self->{doc}-><M::Document.createTextNode> ("'")); | ||
| 2487 | $vals->{value} .= "'"; | ||
| 2488 | } | ||
| 2489 | } (Name) { | ||
| 2490 | wakaba | 1.6 | my $er; |
| 2491 | wakaba | 1.1 | lang:Perl ($name => $token.value) { |
| 2492 | ## TODO: Namespace well-formedness | ||
| 2493 | wakaba | 1.18 | |
| 2494 | wakaba | 1.11 | $er = $self->{doc}-><M::Document.createEntityReference> ($name); |
| 2495 | wakaba | 1.20 | $er-><AS::Node.textContent> (''); |
| 2496 | ## NOTE: When document entity (and entities referenced directly | ||
| 2497 | ## or indirectly from it) is parsed, no general entity | ||
| 2498 | ## node have its replacement tree. During general | ||
| 2499 | ## entity node collection construction, however, | ||
| 2500 | ## some entity node has replacement tree. | ||
| 2501 | wakaba | 1.19 | my $ent = $self->{general_entity}->{$name}; |
| 2502 | wakaba | 1.22 | if (not $ent) { # no entity declaration |
| 2503 | if ($self->{standalone} or not $self->{has_xref}) { # WFC error | ||
| 2504 | my $location; | ||
| 2505 | __CODE{xp|get-location-from-token:: | ||
| 2506 | $token => {$token}, | ||
| 2507 | $result => {$location}, | ||
| 2508 | }__; | ||
| 2509 | my $continue = __DOMCore:ERROR{xp|wf-entity-declared:: | ||
| 2510 | xp|error-token => {$token}, | ||
| 2511 | c|location => {$location}, | ||
| 2512 | xp|name => {$name}, | ||
| 2513 | }__; | ||
| 2514 | unless ($continue) { | ||
| 2515 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2516 | } | ||
| 2517 | } # Otherwise VC error | ||
| 2518 | wakaba | 1.23 | push @{$self->{entity}}, <Code::getEmptyEntityState:: |
| 2519 | $name = $name>; | ||
| 2520 | wakaba | 1.22 | $er-><AS::x|EntityReference.isExpanded> (false); |
| 2521 | |||
| 2522 | } else { # there IS entity declaration | ||
| 2523 | if (($self->{standalone} or not $self->{has_xref}) and | ||
| 2524 | $ent->{is_externally_declared} and | ||
| 2525 | not $self->{entity}->[-1]->{is_externally_declared}) { | ||
| 2526 | my $location; | ||
| 2527 | __CODE{xp|get-location-from-token:: | ||
| 2528 | $token => {$token}, | ||
| 2529 | $result => {$location}, | ||
| 2530 | }__; | ||
| 2531 | my $continue = __DOMCore:ERROR{xp|wf-entity-declared:: | ||
| 2532 | xp|error-token => {$token}, | ||
| 2533 | c|location => {$location}, | ||
| 2534 | xp|name => {$name}, | ||
| 2535 | }__; | ||
| 2536 | unless ($continue) { | ||
| 2537 | __EXCEPTION{DOMLS|PARSE_ERR}__; | ||
| 2538 | } | ||
| 2539 | } | ||
| 2540 | |||
| 2541 | if ($ent->{is_external_entity}) { | ||
| 2542 | my $location; | ||
| 2543 | __CODE{xp|get-location-from-token:: | ||
| 2544 | $token => {$token}, | ||
| 2545 | $result => {$location}, | ||
| 2546 | }__; | ||
| 2547 | my $continue = __c|ERROR{xp|wf-no-external-entity-references:: | ||
| 2548 | xp|error-token => {$token}, | ||
| 2549 | c|location => {$location}, | ||
| 2550 | xp|name => {$name}, | ||
| 2551 | }__; | ||
| 2552 | unless ($continue) { | ||
| 2553 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2554 | }__; | ||
| 2555 | } | ||
| 2556 | wakaba | 1.23 | ## NOTE: |xp:wf-parsed-entity| is not checked |
| 2557 | wakaba | 1.22 | } # if external entity |
| 2558 | |||
| 2559 | $er-><AS::x|EntityReference.isExpanded> | ||
| 2560 | wakaba | 1.19 | ($ent->{has_replacement_text}); |
| 2561 | wakaba | 1.22 | push @{$self->{entity}}, <Code::getCopyOfEntityState:: |
| 2562 | wakaba | 1.6 | $entity_type = 'general_entity', |
| 2563 | $entity_name = $name>; | ||
| 2564 | wakaba | 1.23 | |
| 2565 | if ($ent->{is_opened}) { | ||
| 2566 | my $location; | ||
| 2567 | __CODE{xp|get-location-from-token:: | ||
| 2568 | $token => {$token}, | ||
| 2569 | $result => {$location}, | ||
| 2570 | }__; | ||
| 2571 | my $continue = __c|ERROR{xp|wf-no-recursion:: | ||
| 2572 | xp|error-token => {$token}, | ||
| 2573 | c|location => {$location}, | ||
| 2574 | xp|name => {$name}, | ||
| 2575 | }__; | ||
| 2576 | unless ($continue) { | ||
| 2577 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2578 | }__; | ||
| 2579 | } | ||
| 2580 | $self->{entity}->[-1]->{reptxt} = \''; | ||
| 2581 | $er-><AS::x|EntityReference.isExpanded> (false); | ||
| 2582 | } # recursive | ||
| 2583 | $ent->{is_opened} = true; | ||
| 2584 | wakaba | 1.22 | } |
| 2585 | $parent-><M::Node.appendChild> ($er); | ||
| 2586 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 2587 | wakaba | 1.6 | push @{$self->{entity_token}}, $self->{token}; |
| 2588 | $self->{token} = []; | ||
| 2589 | push @{$self->{entity_char}}, $self->{char}; | ||
| 2590 | $self->{char} = []; | ||
| 2591 | } | ||
| 2592 | |||
| 2593 | wakaba | 1.11 | ?lexmode AttributeValueLiteralE; |
| 2594 | &AttributeValueLiteralE_ ($parent => $er, $vals => $vals); | ||
| 2595 | wakaba | 1.6 | ~ (#EOF) { } |
| 2596 | lang:Perl { | ||
| 2597 | wakaba | 1.23 | $self->{general_entity}->{$self->{entity}->[-1]->{name}} |
| 2598 | ->{is_opened} = false | ||
| 2599 | if $self->{general_entity}->{$self->{entity}->[-1]->{name}}; | ||
| 2600 | wakaba | 1.6 | $self->{token} = pop @{$self->{entity_token}}; |
| 2601 | $self->{char} = pop @{$self->{entity_char}}; | ||
| 2602 | pop @{$self->{entity}}; | ||
| 2603 | wakaba | 1.14 | $self->{location} = $self->{entity}->[-1]; |
| 2604 | wakaba | 1.6 | } |
| 2605 | |||
| 2606 | wakaba | 1.11 | // TODO: Set read-only flag |
| 2607 | |||
| 2608 | wakaba | 1.6 | ?lexmode EntityReference; |
| 2609 | ?requires-next-token; | ||
| 2610 | wakaba | 1.1 | } |
| 2611 | |||
| 2612 | // ~ (REFC) { } | ||
| 2613 | wakaba | 1.11 | } // _GeneralEntityReferenceAE_ |
| 2614 | wakaba | 1.6 | |
| 2615 | /* | ||
| 2616 | wakaba | 1.24 | General Entity Reference in Literal Entity Value in Entity Declaration |
| 2617 | wakaba | 1.6 | */ |
| 2618 | wakaba | 1.11 | rule _GeneralEntityReferenceEV_ ($vals) { |
| 2619 | wakaba | 1.6 | ?lexmode 'EntityReference'; |
| 2620 | |||
| 2621 | ~ (Name) { | ||
| 2622 | lang:Perl ($name => $token.value) { | ||
| 2623 | ## TODO: Namespace well-formedness | ||
| 2624 | push @$vals, $name; | ||
| 2625 | } | ||
| 2626 | wakaba | 1.24 | /* |
| 2627 | WFCs such as Entity Declared are not checked | ||
| 2628 | <http://www.w3.org/TR/xml11/#bypass> | ||
| 2629 | |||
| 2630 | Note that violation to the Parsed Entity WFC is an error | ||
| 2631 | rather than fatal error here. | ||
| 2632 | */ | ||
| 2633 | wakaba | 1.6 | } |
| 2634 | wakaba | 1.11 | |
| 2635 | // No expansion | ||
| 2636 | wakaba | 1.6 | |
| 2637 | // ~ (REFC) { } | ||
| 2638 | } // _GeneralEntityReferenceEV_ | ||
| 2639 | wakaba | 1.1 | |
| 2640 | /* | ||
| 2641 | wakaba | 1.4 | Document Type Declaration |
| 2642 | wakaba | 1.1 | */ |
| 2643 | wakaba | 1.4 | rule _DocumentTypeDeclaration_ ($doc) { |
| 2644 | ?lexmode MarkupDeclaration; | ||
| 2645 | |||
| 2646 | ~ (Name == 'DOCTYPE') { } | ||
| 2647 | |||
| 2648 | ~ (S) { } | ||
| 2649 | |||
| 2650 | // Document type name | ||
| 2651 | wakaba | 1.6 | my $name; |
| 2652 | wakaba | 1.4 | ~ (Name) { |
| 2653 | wakaba | 1.6 | lang:Perl ($v => $token.value) { |
| 2654 | $name = $v; | ||
| 2655 | } | ||
| 2656 | } | ||
| 2657 | wakaba | 1.17 | |
| 2658 | my $node; | ||
| 2659 | my $decl; | ||
| 2660 | wakaba | 1.6 | lang:Perl { |
| 2661 | wakaba | 1.11 | $self->{docx} = $self->{doc} |
| 2662 | -><M::Node.getFeature> (<Q::fe:XDoctype>, '3.0'); | ||
| 2663 | $node = $self->{docx}-><M::DocumentXDoctype.createDocumentTypeDefinition> | ||
| 2664 | ($name); | ||
| 2665 | wakaba | 1.17 | $decl = $self->{dtdecl} = $node-><M::Node.getFeature> |
| 2666 | (<Q::fe:XDoctypeDeclaration>, '3.0'); | ||
| 2667 | wakaba | 1.4 | } |
| 2668 | |||
| 2669 | ~? (S) { | ||
| 2670 | ~? (Name == 'PUBLIC') { | ||
| 2671 | ~ (S) { } | ||
| 2672 | |||
| 2673 | wakaba | 1.17 | &PubidLiteral ($decl => $decl); |
| 2674 | wakaba | 1.4 | |
| 2675 | ~ (S) { } | ||
| 2676 | |||
| 2677 | wakaba | 1.17 | &SystemLiteral ($decl => $decl); |
| 2678 | wakaba | 1.4 | |
| 2679 | ~? (S) { } | ||
| 2680 | wakaba | 1.19 | |
| 2681 | lang:Perl { | ||
| 2682 | $self->{has_xref} = true; | ||
| 2683 | } | ||
| 2684 | wakaba | 1.4 | } (Name == 'SYSTEM') { |
| 2685 | ~ (S) { } | ||
| 2686 | |||
| 2687 | wakaba | 1.17 | &SystemLiteral ($decl => $decl); |
| 2688 | wakaba | 1.4 | |
| 2689 | ~? (S) { } | ||
| 2690 | wakaba | 1.19 | |
| 2691 | lang:Perl { | ||
| 2692 | $self->{has_xref} = true; | ||
| 2693 | } | ||
| 2694 | wakaba | 1.4 | } |
| 2695 | } | ||
| 2696 | wakaba | 1.17 | |
| 2697 | wakaba | 1.11 | lang:Perl { |
| 2698 | $self->{dtdef} = $node; | ||
| 2699 | wakaba | 1.21 | $self->{doc}-><M::Node.appendChild> ($node); |
| 2700 | $self->{doc_cfg}-><M::c|DOMConfiguration.setParameter> | ||
| 2701 | ('schema-type' => <Q::xml-dtd:>); | ||
| 2702 | ## ISSUE: Should default schema language information be | ||
| 2703 | ## preserved by some other flag? | ||
| 2704 | wakaba | 1.11 | } |
| 2705 | |||
| 2706 | wakaba | 1.4 | ~? (DSO) { |
| 2707 | wakaba | 1.6 | &InternalSubset ($doc => $doc, $doctype => $node); |
| 2708 | wakaba | 1.4 | |
| 2709 | ~ (DSC) { | ||
| 2710 | ?lexmode MarkupDeclaration; | ||
| 2711 | } else { | ||
| 2712 | ?lexmode MarkupDeclaration; | ||
| 2713 | } | ||
| 2714 | |||
| 2715 | ~? (S) { } | ||
| 2716 | } | ||
| 2717 | |||
| 2718 | // ~ (MDC) { } | ||
| 2719 | } // _DocumentTypeDeclaration_ | ||
| 2720 | |||
| 2721 | wakaba | 1.17 | /* |
| 2722 | Public Identifier Literal | ||
| 2723 | */ | ||
| 2724 | rule PubidLiteral ($decl) { | ||
| 2725 | my $pubid; | ||
| 2726 | my $pubid_token; | ||
| 2727 | wakaba | 1.4 | ~ (LIT) { |
| 2728 | ?lexmode SystemLiteral; | ||
| 2729 | |||
| 2730 | ~? (STRING) { | ||
| 2731 | wakaba | 1.17 | lang:Perl ($v => $token.value) { |
| 2732 | $pubid = $v; | ||
| 2733 | $pubid_token = $token; | ||
| 2734 | } | ||
| 2735 | wakaba | 1.4 | } |
| 2736 | |||
| 2737 | ~ (LIT) { | ||
| 2738 | ?lexmode MarkupDeclaration; | ||
| 2739 | } else { | ||
| 2740 | ?lexmode MarkupDeclaration; | ||
| 2741 | } | ||
| 2742 | } (LITA) { | ||
| 2743 | ?lexmode SystemLiteralA; | ||
| 2744 | |||
| 2745 | ~? (STRING) { | ||
| 2746 | wakaba | 1.17 | lang:Perl ($v => $token.value) { |
| 2747 | $pubid = $v; | ||
| 2748 | $pubid_token = $token; | ||
| 2749 | } | ||
| 2750 | wakaba | 1.4 | } |
| 2751 | |||
| 2752 | ~ (LITA) { | ||
| 2753 | ?lexmode MarkupDeclaration; | ||
| 2754 | } else { | ||
| 2755 | ?lexmode MarkupDeclaration; | ||
| 2756 | } | ||
| 2757 | } | ||
| 2758 | wakaba | 1.17 | |
| 2759 | if-true ($pubid_token) { | ||
| 2760 | lang:Perl : has-error { | ||
| 2761 | if ($pubid =~ m{[^\x20\x0D\x0Aa-zA-Z0-9'()+,./:=?;!*#\@\$_%-]}s) { | ||
| 2762 | my $location; | ||
| 2763 | __CODE{xp|get-location-from-token:: | ||
| 2764 | $token => {$pubid_token}, | ||
| 2765 | $result => {$location}, | ||
| 2766 | }__; | ||
| 2767 | ## Should this be other (new) error type? | ||
| 2768 | my $continue = __DOMCore:ERROR{xp|wf-syntax-error:: | ||
| 2769 | xp|error-token => {$token}, | ||
| 2770 | c|location => {$location}, | ||
| 2771 | }__; | ||
| 2772 | unless ($continue) { | ||
| 2773 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 2774 | }__; | ||
| 2775 | } | ||
| 2776 | } | ||
| 2777 | |||
| 2778 | ## NOTE: U+0009 is syntactically illegal. | ||
| 2779 | wakaba | 1.23 | $pubid =~ s/[\x09\x0A\x0D\x20]+/\x20/gs; |
| 2780 | wakaba | 1.17 | $pubid =~ s/\A\x20//s; |
| 2781 | $pubid =~ s/\x20\z//s; | ||
| 2782 | ## NOTE: Bare attribute name is written. | ||
| 2783 | $decl->public_id ($pubid); | ||
| 2784 | } | ||
| 2785 | } | ||
| 2786 | wakaba | 1.4 | } // PubidLiteral |
| 2787 | |||
| 2788 | wakaba | 1.17 | /* |
| 2789 | System Identifier Literal | ||
| 2790 | */ | ||
| 2791 | rule SystemLiteral ($decl) { | ||
| 2792 | wakaba | 1.4 | ~ (LIT) { |
| 2793 | ?lexmode SystemLiteral; | ||
| 2794 | wakaba | 1.17 | &_SystemLiteral ($decl => $decl); |
| 2795 | wakaba | 1.4 | } (LITA) { |
| 2796 | ?lexmode SystemLiteralA; | ||
| 2797 | wakaba | 1.17 | &_SystemLiteral ($decl => $decl); |
| 2798 | wakaba | 1.4 | } |
| 2799 | } // SystemLiteral | ||
| 2800 | |||
| 2801 | wakaba | 1.17 | /* |
| 2802 | System Identifier Literal Content | ||
| 2803 | */ | ||
| 2804 | rule _SystemLiteral ($decl) { | ||
| 2805 | // LIT or LITA | ||
| 2806 | |||
| 2807 | wakaba | 1.4 | ~? (STRING) { |
| 2808 | wakaba | 1.17 | lang:Perl ($v => $token.value) { |
| 2809 | ## NOTE: Bare attribute name is written. | ||
| 2810 | $decl->system_id ($v); | ||
| 2811 | } | ||
| 2812 | wakaba | 1.4 | } |
| 2813 | |||
| 2814 | ~ (LIT) { | ||
| 2815 | ?lexmode MarkupDeclaration; | ||
| 2816 | } (LITA) { | ||
| 2817 | ?lexmode MarkupDeclaration; | ||
| 2818 | } else { | ||
| 2819 | ?lexmode MarkupDeclaration; | ||
| 2820 | } | ||
| 2821 | } // _SystemLiteral | ||
| 2822 | |||
| 2823 | /* | ||
| 2824 | DTD Internal Subset | ||
| 2825 | |||
| 2826 | intSubset := *(markupdecl / DeclSep) ;; [28b] | ||
| 2827 | */ | ||
| 2828 | wakaba | 1.6 | rule InternalSubset ($doc, $doctype) { |
| 2829 | wakaba | 1.4 | ?lexmode DTD; |
| 2830 | |||
| 2831 | ~* (MDO) { | ||
| 2832 | ?lexmode MarkupDeclaration; | ||
| 2833 | |||
| 2834 | ~ (Name == 'ELEMENT') { | ||
| 2835 | &_ElementDeclaration ($doc => $doc); | ||
| 2836 | } (Name == 'ATTLIST') { | ||
| 2837 | wakaba | 1.8 | &_AttlistDeclaration ($doc => $doc, $doctype => $doctype); |
| 2838 | wakaba | 1.4 | } (Name == 'ENTITY') { |
| 2839 | &_EntityDeclaration ($doc => $doc); | ||
| 2840 | } (Name == 'NOTATION') { | ||
| 2841 | &_NotationDeclaration ($doc => $doc); | ||
| 2842 | } | ||
| 2843 | } (S) { | ||
| 2844 | // | ||
| 2845 | } (CDO) { | ||
| 2846 | &_CommentDeclarationDTD ($doc => $doc); | ||
| 2847 | } (PIO) { | ||
| 2848 | wakaba | 1.6 | &_ProcessingInstructionDTD ($doc => $doc, $doctype => $doctype); |
| 2849 | wakaba | 1.4 | } (PERO) { |
| 2850 | ?lexmode EntityReference; | ||
| 2851 | |||
| 2852 | wakaba | 1.19 | lang:Perl { |
| 2853 | $self->{has_xref} = true; | ||
| 2854 | } | ||
| 2855 | |||
| 2856 | wakaba | 1.4 | ~ (Name) { |
| 2857 | |||
| 2858 | } | ||
| 2859 | |||
| 2860 | ~ (REFC) { | ||
| 2861 | ?lexmode DTD; | ||
| 2862 | } else { | ||
| 2863 | ?lexmode DTD; | ||
| 2864 | } | ||
| 2865 | } | ||
| 2866 | } // InternalSubset | ||
| 2867 | |||
| 2868 | rule _ElementDeclaration ($doc) { | ||
| 2869 | // ~ (MDO) { } | ||
| 2870 | // ?lexmode MarkupDeclaration | ||
| 2871 | // ~ (Name == 'Element') { } | ||
| 2872 | |||
| 2873 | ~ (S) { } | ||
| 2874 | |||
| 2875 | ~ (Name) { | ||
| 2876 | |||
| 2877 | } | ||
| 2878 | |||
| 2879 | ?lexmode ElementDeclaration; | ||
| 2880 | |||
| 2881 | ~ (S) { } | ||
| 2882 | |||
| 2883 | // contentspec | ||
| 2884 | ~ (MGO) { | ||
| 2885 | &_ContentModel ($doc => $doc); | ||
| 2886 | } (Name == 'EMPTY') { | ||
| 2887 | |||
| 2888 | } (Name == 'ANY') { | ||
| 2889 | |||
| 2890 | } | ||
| 2891 | |||
| 2892 | ~? (S) { } | ||
| 2893 | |||
| 2894 | ~ (MDC) { | ||
| 2895 | ?lexmode DTD; | ||
| 2896 | } else { | ||
| 2897 | ?lexmode DTD; | ||
| 2898 | } | ||
| 2899 | } // _ElementDeclaration | ||
| 2900 | |||
| 2901 | rule _ContentModel ($doc) { | ||
| 2902 | // ~ (MGO) { } | ||
| 2903 | |||
| 2904 | ~? (S) { } | ||
| 2905 | |||
| 2906 | ~ (Name) { | ||
| 2907 | // Element content model | ||
| 2908 | &_ModelGroup ($doc => $doc) | ||
| 2909 | : unshift-current-token; | ||
| 2910 | |||
| 2911 | ~? (OPT) { | ||
| 2912 | |||
| 2913 | } (REP) { | ||
| 2914 | |||
| 2915 | } (PLUS) { | ||
| 2916 | |||
| 2917 | } | ||
| 2918 | |||
| 2919 | } (MDO) { | ||
| 2920 | &_ModelGroup ($doc => $doc) | ||
| 2921 | : unshift-current-token; | ||
| 2922 | |||
| 2923 | ~? (OPT) { | ||
| 2924 | |||
| 2925 | } (REP) { | ||
| 2926 | |||
| 2927 | } (PLUS) { | ||
| 2928 | |||
| 2929 | } | ||
| 2930 | |||
| 2931 | } (PCDATA) { | ||
| 2932 | // Mixed content declaration | ||
| 2933 | ~? (S) { } | ||
| 2934 | |||
| 2935 | ~* (OR) { | ||
| 2936 | ~? (S) { } | ||
| 2937 | |||
| 2938 | ~ (Name) { | ||
| 2939 | |||
| 2940 | } | ||
| 2941 | |||
| 2942 | ~? (S) { } | ||
| 2943 | } | ||
| 2944 | |||
| 2945 | ~ (MGC) { } | ||
| 2946 | |||
| 2947 | ~? (REP) { | ||
| 2948 | |||
| 2949 | } else { | ||
| 2950 | // TODO: error if |Name| | ||
| 2951 | } | ||
| 2952 | |||
| 2953 | |||
| 2954 | } | ||
| 2955 | } // _ContentModel | ||
| 2956 | |||
| 2957 | rule _ModelGroup ($doc) | ||
| 2958 | : standalone | ||
| 2959 | : recursive | ||
| 2960 | { | ||
| 2961 | // ~ (MGO) { } | ||
| 2962 | // ~? (S) { } | ||
| 2963 | |||
| 2964 | &ContentParticle ($doc => $doc); | ||
| 2965 | |||
| 2966 | ~? (S) { } | ||
| 2967 | |||
| 2968 | ~? (OR) { | ||
| 2969 | ~? (S) { } | ||
| 2970 | |||
| 2971 | &ContentParticle ($doc => $doc); | ||
| 2972 | |||
| 2973 | ~? (S) { } | ||
| 2974 | |||
| 2975 | ~* (OR) { | ||
| 2976 | ~? (S) { } | ||
| 2977 | |||
| 2978 | &ContentParticle ($doc => $doc); | ||
| 2979 | |||
| 2980 | ~? (S) { } | ||
| 2981 | } | ||
| 2982 | |||
| 2983 | } (SEQ) { | ||
| 2984 | ~? (S) { } | ||
| 2985 | |||
| 2986 | &ContentParticle ($doc => $doc); | ||
| 2987 | |||
| 2988 | ~? (S) { } | ||
| 2989 | |||
| 2990 | ~* (SEQ) { | ||
| 2991 | ~? (S) { } | ||
| 2992 | |||
| 2993 | &ContentParticle ($doc => $doc); | ||
| 2994 | |||
| 2995 | ~? (S) { } | ||
| 2996 | } | ||
| 2997 | |||
| 2998 | } | ||
| 2999 | |||
| 3000 | ~ (MGC) { } | ||
| 3001 | } // _ModelGroup | ||
| 3002 | |||
| 3003 | rule ContentParticle ($doc) { | ||
| 3004 | ~ (Name) { | ||
| 3005 | |||
| 3006 | } (MGO) { | ||
| 3007 | ~? (S) { } | ||
| 3008 | |||
| 3009 | &_ModelGroup ($doc => $doc); | ||
| 3010 | } | ||
| 3011 | |||
| 3012 | ~? (OPT) { | ||
| 3013 | |||
| 3014 | } (REP) { | ||
| 3015 | |||
| 3016 | } (PLUS) { | ||
| 3017 | |||
| 3018 | } | ||
| 3019 | } // ContentParticle | ||
| 3020 | |||
| 3021 | wakaba | 1.8 | /* |
| 3022 | Attribute list declaration | ||
| 3023 | */ | ||
| 3024 | rule _AttlistDeclaration ($doc, $doctype) { | ||
| 3025 | wakaba | 1.4 | // ~ (MDO) { } |
| 3026 | // ~ (Name == 'ATTLIST') { } | ||
| 3027 | |||
| 3028 | ?lexmode AttlistDeclaration; | ||
| 3029 | |||
| 3030 | ~ (S) { } | ||
| 3031 | |||
| 3032 | wakaba | 1.8 | my $name; |
| 3033 | wakaba | 1.4 | ~ (Name) { |
| 3034 | wakaba | 1.8 | lang:Perl ($v => $token.value) { |
| 3035 | $name = $v; | ||
| 3036 | } | ||
| 3037 | } else { | ||
| 3038 | lang:Perl { $name = '#ILLEGAL' } | ||
| 3039 | } | ||
| 3040 | wakaba | 1.4 | |
| 3041 | wakaba | 1.8 | my $docxd; |
| 3042 | my $et; | ||
| 3043 | lang:Perl { | ||
| 3044 | $et = $doctype-><M::DTDef.getElementTypeDefinitionNode> ($name); | ||
| 3045 | unless ($et) { | ||
| 3046 | wakaba | 1.11 | $docxd = $doc-><M::Node.getFeature> (<Q::fe:XDoctype>, '3.0'); |
| 3047 | wakaba | 1.8 | $et = $docxd-><M::DocumentXDoctype.createElementTypeDefinition> |
| 3048 | ($name); | ||
| 3049 | $doctype-><M::DTDef.setElementTypeDefinitionNode> ($et) | ||
| 3050 | unless $name eq '#ILLEGAL'; | ||
| 3051 | } | ||
| 3052 | wakaba | 1.4 | } |
| 3053 | |||
| 3054 | ~? (S) { } | ||
| 3055 | |||
| 3056 | ~* (Name) { | ||
| 3057 | wakaba | 1.8 | my $at; |
| 3058 | lang:Perl ($v => $token.value) { | ||
| 3059 | $docxd ||= $doc-><M::Node.getFeature> | ||
| 3060 | wakaba | 1.11 | (<Q::fe:XDoctype>, '3.0'); |
| 3061 | wakaba | 1.8 | $at = $docxd-><M::DocumentXDoctype.createAttributeDefinition> ($v); |
| 3062 | wakaba | 1.9 | unless (exists $et-><AG::ETDef.attributeDefinitions>->{$v}) { |
| 3063 | $et-><M::ETDef.setAttributeDefinitionNode> ($at); | ||
| 3064 | $self->{attr}->{$name}->{$v} = $at; | ||
| 3065 | } | ||
| 3066 | wakaba | 1.8 | } |
| 3067 | wakaba | 1.4 | |
| 3068 | ~ (S) { } | ||
| 3069 | |||
| 3070 | // AttType | ||
| 3071 | ~ (Name == 'NOTATION') { | ||
| 3072 | ~ (S) { } | ||
| 3073 | |||
| 3074 | wakaba | 1.8 | my $kwd; |
| 3075 | lang:Perl { | ||
| 3076 | $at-><AS::ATDef.declaredType> (<C::ATDef.NOTATION_ATTR>); | ||
| 3077 | $kwd = $at-><AG::ATDef.allowedTokens>; | ||
| 3078 | } | ||
| 3079 | |||
| 3080 | wakaba | 1.4 | ~ (EGO) { |
| 3081 | ~? (S) { } | ||
| 3082 | |||
| 3083 | wakaba | 1.8 | ~+ (Name) { |
| 3084 | lang:Perl ($v => $token.value) { | ||
| 3085 | push @$kwd, $v; | ||
| 3086 | } | ||
| 3087 | wakaba | 1.4 | |
| 3088 | ~? (S) { } | ||
| 3089 | } (OR) : separator { | ||
| 3090 | ~? (S) { } | ||
| 3091 | } | ||
| 3092 | |||
| 3093 | ~ (EGC) { } | ||
| 3094 | } | ||
| 3095 | wakaba | 1.8 | |
| 3096 | wakaba | 1.4 | } (Name) { |
| 3097 | wakaba | 1.8 | my $type; |
| 3098 | lang:Perl ($v => $token.value) : has-error { | ||
| 3099 | my $map = { | ||
| 3100 | CDATA => <C::ATDef.CDATA_ATTR>, | ||
| 3101 | ID => <C::ATDef.ID_ATTR>, | ||
| 3102 | IDREF => <C::ATDef.IDREF_ATTR>, | ||
| 3103 | IDREFS => <C::ATDef.IDREFS_ATTR>, | ||
| 3104 | ENTITY => <C::ATDef.ENTITY_ATTR>, | ||
| 3105 | ENTITIES => <C::ATDef.ENTITIES_ATTR>, | ||
| 3106 | NMTOKEN => <C::ATDef.NMTOKEN_ATTR>, | ||
| 3107 | NMTOKENS => <C::ATDef.NMTOKENS_ATTR>, | ||
| 3108 | }; | ||
| 3109 | if ($map->{$v}) { | ||
| 3110 | $at-><AS::ATDef.declaredType> ($map->{$v}); | ||
| 3111 | } else { | ||
| 3112 | ## TODO: Exception | ||
| 3113 | } | ||
| 3114 | } | ||
| 3115 | wakaba | 1.4 | |
| 3116 | } (EGO) { | ||
| 3117 | ?lexmode Enumeration; | ||
| 3118 | |||
| 3119 | ~? (S) { } | ||
| 3120 | |||
| 3121 | wakaba | 1.8 | my $kwd; |
| 3122 | lang:Perl { | ||
| 3123 | $at-><AS::ATDef.declaredType> (<C::ATDef.ENUMERATION_ATTR>); | ||
| 3124 | $kwd = $at-><AG::ATDef.allowedTokens>; | ||
| 3125 | } | ||
| 3126 | |||
| 3127 | ~+ (Nmtoken) { | ||
| 3128 | lang:Perl ($v => $token.value) { | ||
| 3129 | push @$kwd, $v; | ||
| 3130 | } | ||
| 3131 | wakaba | 1.4 | |
| 3132 | ~? (S) { } | ||
| 3133 | } (OR) : separator { | ||
| 3134 | ~? (S) { } | ||
| 3135 | } | ||
| 3136 | |||
| 3137 | ~ (EGC) { | ||
| 3138 | ?lexmode AttlistDeclaration; | ||
| 3139 | } else { | ||
| 3140 | ?lexmode AttlistDeclaration; | ||
| 3141 | } | ||
| 3142 | |||
| 3143 | } | ||
| 3144 | |||
| 3145 | ~ (S) { } | ||
| 3146 | |||
| 3147 | // DefaultDecl | ||
| 3148 | ~ (RNI) { | ||
| 3149 | ~ (Name == 'REQUIRED') { | ||
| 3150 | wakaba | 1.8 | lang:Perl { |
| 3151 | $at-><AS::ATDef.defaultType> (<C::ATDef.REQUIRED_DEFAULT>); | ||
| 3152 | } | ||
| 3153 | wakaba | 1.4 | } (Name == 'IMPLIED') { |
| 3154 | wakaba | 1.8 | lang:Perl { |
| 3155 | $at-><AS::ATDef.defaultType> (<C::ATDef.IMPLIED_DEFAULT>); | ||
| 3156 | } | ||
| 3157 | wakaba | 1.4 | } (Name == 'FIXED') { |
| 3158 | ~ (S) { } | ||
| 3159 | wakaba | 1.8 | |
| 3160 | lang:Perl { | ||
| 3161 | $at-><AS::ATDef.defaultType> (<C::ATDef.FIXED_DEFAULT>); | ||
| 3162 | } | ||
| 3163 | wakaba | 1.4 | |
| 3164 | ~ (LIT) { | ||
| 3165 | my $vals; | ||
| 3166 | lang:Perl { | ||
| 3167 | wakaba | 1.8 | $vals = {nodes => [], value => ''}; |
| 3168 | wakaba | 1.4 | } |
| 3169 | |||
| 3170 | &_AttributeValueSpecification_ ($doc => $doc, $vals => $vals); | ||
| 3171 | |||
| 3172 | wakaba | 1.8 | lang:Perl { |
| 3173 | for (@{$vals->{nodes}}) { | ||
| 3174 | $at-><M::Node.appendChild> ($_); | ||
| 3175 | } | ||
| 3176 | } | ||
| 3177 | |||
| 3178 | wakaba | 1.4 | ~ (LIT) { |
| 3179 | ?lexmode AttlistDeclaration; | ||
| 3180 | } else { | ||
| 3181 | ?lexmode AttlistDeclaration; | ||
| 3182 | } | ||
| 3183 | } (LITA) { | ||
| 3184 | my $vals; | ||
| 3185 | lang:Perl { | ||
| 3186 | wakaba | 1.8 | $vals = {nodes => [], value => ''}; |
| 3187 | wakaba | 1.4 | } |
| 3188 | |||
| 3189 | &_AttributeValueSpecificationA_ ($doc => $doc, $vals => $vals); | ||
| 3190 | |||
| 3191 | wakaba | 1.8 | lang:Perl { |
| 3192 | for (@{$vals->{nodes}}) { | ||
| 3193 | $at-><M::Node.appendChild> ($_); | ||
| 3194 | } | ||
| 3195 | } | ||
| 3196 | |||
| 3197 | ~ (LITA) { | ||
| 3198 | wakaba | 1.4 | ?lexmode AttlistDeclaration; |
| 3199 | } else { | ||
| 3200 | ?lexmode AttlistDeclaration; | ||
| 3201 | } | ||
| 3202 | } | ||
| 3203 | } | ||
| 3204 | |||
| 3205 | } (LIT) { | ||
| 3206 | my $vals; | ||
| 3207 | lang:Perl { | ||
| 3208 | wakaba | 1.8 | $at-><AS::ATDef.defaultType> (<C::ATDef.EXPLICIT_DEFAULT>); |
| 3209 | $vals = {nodes => [], value => ''}; | ||
| 3210 | wakaba | 1.4 | } |
| 3211 | |||
| 3212 | &_AttributeValueSpecification_ ($doc => $doc, $vals => $vals); | ||
| 3213 | |||
| 3214 | wakaba | 1.8 | lang:Perl { |
| 3215 | for (@{$vals->{nodes}}) { | ||
| 3216 | $at-><M::Node.appendChild> ($_); | ||
| 3217 | } | ||
| 3218 | } | ||
| 3219 | |||
| 3220 | wakaba | 1.4 | ~ (LIT) { |
| 3221 | ?lexmode AttlistDeclaration; | ||
| 3222 | } else { | ||
| 3223 | ?lexmode AttlistDeclaration; | ||
| 3224 | } | ||
| 3225 | } (LITA) { | ||
| 3226 | my $vals; | ||
| 3227 | lang:Perl { | ||
| 3228 | wakaba | 1.8 | $at-><AS::ATDef.defaultType> (<C::ATDef.EXPLICIT_DEFAULT>); |
| 3229 | $vals = {nodes => [], value => ''}; | ||
| 3230 | wakaba | 1.4 | } |
| 3231 | |||
| 3232 | &_AttributeValueSpecificationA_ ($doc => $doc, $vals => $vals); | ||
| 3233 | |||
| 3234 | wakaba | 1.8 | lang:Perl { |
| 3235 | for (@{$vals->{nodes}}) { | ||
| 3236 | $at-><M::Node.appendChild> ($_); | ||
| 3237 | } | ||
| 3238 | } | ||
| 3239 | |||
| 3240 | wakaba | 1.4 | ~ (LITA) { |
| 3241 | ?lexmode AttlistDeclaration; | ||
| 3242 | } else { | ||
| 3243 | ?lexmode AttlistDeclaration; | ||
| 3244 | } | ||
| 3245 | } | ||
| 3246 | |||
| 3247 | } (S) : separator : terminator? { | ||
| 3248 | // | ||
| 3249 | } | ||
| 3250 | |||
| 3251 | ~ (MDC) { | ||
| 3252 | ?lexmode DTD; | ||
| 3253 | } else { | ||
| 3254 | ?lexmode DTD; | ||
| 3255 | } | ||
| 3256 | } // _AttlistDeclaration | ||
| 3257 | |||
| 3258 | wakaba | 1.11 | /* |
| 3259 | Entity declaration | ||
| 3260 | */ | ||
| 3261 | wakaba | 1.4 | rule _EntityDeclaration ($doc) { |
| 3262 | // ~ (MDO) { } | ||
| 3263 | // ~ (Name == ENTITY) { } | ||
| 3264 | |||
| 3265 | ~ (S) { } | ||
| 3266 | |||
| 3267 | wakaba | 1.23 | my $is_param_entity; |
| 3268 | wakaba | 1.6 | my $decl; |
| 3269 | wakaba | 1.18 | lang:Perl { $decl = { name => '#ILLEGAL', reptxt => \'' }; } |
| 3270 | wakaba | 1.6 | |
| 3271 | wakaba | 1.4 | ~? (PERO) { |
| 3272 | ~ (S) { } | ||
| 3273 | wakaba | 1.6 | lang:Perl { |
| 3274 | wakaba | 1.23 | $is_param_entity = $decl->{is_param_entity} = true; |
| 3275 | wakaba | 1.6 | } |
| 3276 | wakaba | 1.4 | } |
| 3277 | |||
| 3278 | wakaba | 1.19 | /* |
| 3279 | TODO: Set |$decl->{is_externally_declared}| flag | ||
| 3280 | and |Entity.isExternallyDeclared| attribute | ||
| 3281 | for entity declarations in external subset or | ||
| 3282 | in parameter entity. | ||
| 3283 | */ | ||
| 3284 | |||
| 3285 | wakaba | 1.4 | ~ (Name) { |
| 3286 | wakaba | 1.6 | lang:Perl ($v => $token.value) { |
| 3287 | $decl->{name} = $v; | ||
| 3288 | } | ||
| 3289 | wakaba | 1.4 | } |
| 3290 | |||
| 3291 | wakaba | 1.18 | my $node; |
| 3292 | wakaba | 1.11 | lang:Perl { |
| 3293 | wakaba | 1.18 | $node = |
| 3294 | wakaba | 1.11 | $decl->{node} = $self->{docx}-><M::DocumentXDoctype.createGeneralEntity> |
| 3295 | ($decl->{name}); | ||
| 3296 | ## TODO: Parameter entity... | ||
| 3297 | } | ||
| 3298 | |||
| 3299 | wakaba | 1.4 | ~ (S) { } |
| 3300 | |||
| 3301 | ~ (LIT) { | ||
| 3302 | wakaba | 1.6 | &_EntityValue ($doc => $doc, $decl => $decl); |
| 3303 | wakaba | 1.4 | } (LITA) { |
| 3304 | wakaba | 1.6 | &_EntityValueA ($doc => $doc, $decl => $decl); |
| 3305 | wakaba | 1.4 | } (Name == 'PUBLIC') { |
| 3306 | ~ (S) { } | ||
| 3307 | |||
| 3308 | wakaba | 1.18 | &PubidLiteral ($decl => $node); |
| 3309 | wakaba | 1.4 | |
| 3310 | ~ (S) { } | ||
| 3311 | |||
| 3312 | wakaba | 1.18 | &SystemLiteral ($decl => $node); |
| 3313 | |||
| 3314 | lang:Perl { | ||
| 3315 | $decl->{is_external_entity} = true; | ||
| 3316 | } | ||
| 3317 | wakaba | 1.4 | } (Name == 'SYSTEM') { |
| 3318 | ~ (S) { } | ||
| 3319 | |||
| 3320 | wakaba | 1.18 | &SystemLiteral ($decl => $node); |
| 3321 | |||
| 3322 | lang:Perl { | ||
| 3323 | $decl->{is_external_entity} = true; | ||
| 3324 | } | ||
| 3325 | wakaba | 1.4 | } |
| 3326 | |||
| 3327 | ~? (S) { | ||
| 3328 | ~? (Name == 'NDATA') { | ||
| 3329 | wakaba | 1.23 | if-true ($is_param_entity) { |
| 3330 | lang:Perl : has-error { | ||
| 3331 | my $location; | ||
| 3332 | __CODE{xp|get-location-from-token:: | ||
| 3333 | $token => {$token}, | ||
| 3334 | $result => {$location}, | ||
| 3335 | }__; | ||
| 3336 | my $continue = __DOMCore:ERROR{xp|wf-syntax-error:: | ||
| 3337 | xp|error-token => {$token}, | ||
| 3338 | c|location => {$location}, | ||
| 3339 | }__; | ||
| 3340 | unless ($continue) { | ||
| 3341 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 3342 | }__; | ||
| 3343 | } | ||
| 3344 | } | ||
| 3345 | } | ||
| 3346 | wakaba | 1.4 | |
| 3347 | ~ (S) { } | ||
| 3348 | |||
| 3349 | ~ (Name) { | ||
| 3350 | wakaba | 1.6 | lang:Perl ($v => $token.value) { |
| 3351 | $decl->{notation} = $v; | ||
| 3352 | wakaba | 1.11 | $decl->{node}-><AG::x|Entity.notationName> ($v); |
| 3353 | wakaba | 1.6 | } |
| 3354 | wakaba | 1.4 | } |
| 3355 | |||
| 3356 | ~? (S) { } | ||
| 3357 | } | ||
| 3358 | } | ||
| 3359 | |||
| 3360 | wakaba | 1.24 | lang:Perl : has-error { |
| 3361 | wakaba | 1.6 | if ($self->{$decl->{is_param_entity} ? 'param_entity' : 'general_entity'} |
| 3362 | ->{$decl->{name}}) { | ||
| 3363 | wakaba | 1.24 | ## Predefined entity |
| 3364 | if (not $decl->{is_param_entity} and | ||
| 3365 | {lt => true, gt => true, amp => true, | ||
| 3366 | quot => true, apos => true}->{$decl->{name}}) { | ||
| 3367 | if ($decl->{is_external_entity}) { | ||
| 3368 | my $location; | ||
| 3369 | __CODE{xp|get-location-from-token:: | ||
| 3370 | $token => {$token}, | ||
| 3371 | $result => {$location}, | ||
| 3372 | }__; | ||
| 3373 | my $continue = __c|ERROR{xp|error-internal-predefined-entity:: | ||
| 3374 | xp|error-token => {$token}, | ||
| 3375 | c|location => {$location}, | ||
| 3376 | xp|name => {$decl->{name}}, | ||
| 3377 | }__; | ||
| 3378 | unless ($continue) { | ||
| 3379 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 3380 | }__; | ||
| 3381 | } | ||
| 3382 | } elsif ( | ||
| 3383 | not ({gt => true, apos => true, quot => true}->{$decl->{name}} and | ||
| 3384 | ${$decl->{reptxt}} eq { | ||
| 3385 | gt => '>', apos => "'", quot => '"', | ||
| 3386 | }->{$decl->{name}}) and | ||
| 3387 | not (${$decl->{reptxt}} =~ /\A�*@{[{ | ||
| 3388 | lt => '60', gt => '62', amp => '38', | ||
| 3389 | apos => '39', quot => '34', | ||
| 3390 | }->{$decl->{name}}]};\z/) and | ||
| 3391 | not (${$decl->{reptxt}} =~ /\A�*(?:(?i)@{[{ | ||
| 3392 | lt => '3C', gt => '3E', amp => '26', | ||
| 3393 | apos => '27', quot => '22', | ||
| 3394 | }->{$decl->{name}}]});\z/) | ||
| 3395 | ) { | ||
| 3396 | ## NOTE: See "SuikaWiki - Predefined Entities" | ||
| 3397 | ## "http://suika.fam.cx/gate/2005/sw/%E5%AE%9A%E7%BE%A9%E6%B8%88%E5%AE%9F%E4%BD%93". | ||
| 3398 | my $location; | ||
| 3399 | __CODE{xp|get-location-from-token:: | ||
| 3400 | $token => {$token}, | ||
| 3401 | $result => {$location}, | ||
| 3402 | }__; | ||
| 3403 | my $continue = __c|ERROR{xp|error-malformed-predefined-entity:: | ||
| 3404 | xp|error-token => {$token}, | ||
| 3405 | c|location => {$location}, | ||
| 3406 | xp|name => {$decl->{name}}, | ||
| 3407 | xp|replacement-text => {${$decl->{reptxt}}}, | ||
| 3408 | }__; | ||
| 3409 | unless ($continue) { | ||
| 3410 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 3411 | }__; | ||
| 3412 | } | ||
| 3413 | } | ||
| 3414 | ## TODO: | ||
| 3415 | } else { ## Dupulicating declaration | ||
| 3416 | my $location; | ||
| 3417 | __CODE{xp|get-location-from-token:: | ||
| 3418 | $token => {$token}, | ||
| 3419 | $result => {$location}, | ||
| 3420 | }__; | ||
| 3421 | my $continue = __c|ERROR{xp|warning-entity-declaration-ignored:: | ||
| 3422 | xp|error-token => {$token}, | ||
| 3423 | c|location => {$location}, | ||
| 3424 | xp|name => {$decl->{name}}, | ||
| 3425 | }__; | ||
| 3426 | unless ($continue) { | ||
| 3427 | __EXCEPTION{DOMLS|PARSE_ERR:: | ||
| 3428 | }__; | ||
| 3429 | } | ||
| 3430 | } | ||
| 3431 | wakaba | 1.6 | } else { |
| 3432 | $self->{$decl->{is_param_entity} ? 'param_entity' : 'general_entity'} | ||
| 3433 | ->{$decl->{name}} = $decl; | ||
| 3434 | wakaba | 1.11 | $self->{dtdef}-><M::DTDef.setGeneralEntityNode> ($decl->{node}) |
| 3435 | unless $decl->{is_param_entity}; | ||
| 3436 | wakaba | 1.6 | } |
| 3437 | } | ||
| 3438 | |||
| 3439 | wakaba | 1.4 | ~ (MDC) { |
| 3440 | ?lexmode DTD; | ||
| 3441 | } else { | ||
| 3442 | ?lexmode DTD; | ||
| 3443 | } | ||
| 3444 | } // _EntityDeclaration | ||
| 3445 | |||
| 3446 | rule _NotationDeclaration ($doc) { | ||
| 3447 | // ~ (MDO) { } | ||
| 3448 | // ~ (Name == NOTATION) { } | ||
| 3449 | |||
| 3450 | ~ (S) { } | ||
| 3451 | |||
| 3452 | ~ (Name) { | ||
| 3453 | |||
| 3454 | } | ||
| 3455 | |||
| 3456 | ~ (S) { } | ||
| 3457 | |||
| 3458 | wakaba | 1.17 | my $decl; |
| 3459 | lang:Perl { | ||
| 3460 | $decl = $self->{docx}-><M::DocumentXDoctype.createGeneralEntity> | ||
| 3461 | ($decl->{name}); | ||
| 3462 | ## Temporary: to avoid error. | ||
| 3463 | } | ||
| 3464 | |||
| 3465 | wakaba | 1.4 | ~ (Name == 'PUBLIC') { |
| 3466 | ~ (S) { } | ||
| 3467 | |||
| 3468 | wakaba | 1.17 | &PubidLiteral ($decl => $decl); |
| 3469 | wakaba | 1.4 | |
| 3470 | ~? (S) { | ||
| 3471 | ~? (LIT) { | ||
| 3472 | ?lexmode SystemLiteral; | ||
| 3473 | wakaba | 1.17 | &_SystemLiteral ($decl => $decl); |
| 3474 | wakaba | 1.4 | |
| 3475 | ~? (S) { } | ||
| 3476 | } (LITA) { | ||
| 3477 | ?lexmode SystemLiteralA; | ||
| 3478 | wakaba | 1.17 | &_SystemLiteral ($decl => $decl); |
| 3479 | wakaba | 1.4 | |
| 3480 | ~? (S) { } | ||
| 3481 | } | ||
| 3482 | } | ||
| 3483 | } (Name == 'SYSTEM') { | ||
| 3484 | ~ (S) { } | ||
| 3485 | |||
| 3486 | wakaba | 1.17 | &SystemLiteral ($decl => $decl); |
| 3487 | wakaba | 1.4 | |
| 3488 | ~? (S) { } | ||
| 3489 | } | ||
| 3490 | |||
| 3491 | ~ (MDC) { | ||
| 3492 | ?lexmode DTD; | ||
| 3493 | } else { | ||
| 3494 | ?lexmode DTD; | ||
| 3495 | } | ||
| 3496 | } // _NotationDeclaration | ||
| 3497 | |||
| 3498 | wakaba | 1.11 | rule _EntityValue ($decl) { |
| 3499 | wakaba | 1.4 | ?lexmode EntityValue; |
| 3500 | |||
| 3501 | my $vals; | ||
| 3502 | wakaba | 1.6 | my $reptxt; |
| 3503 | wakaba | 1.4 | lang:Perl { |
| 3504 | $vals = []; | ||
| 3505 | wakaba | 1.6 | $reptxt = ''; |
| 3506 | wakaba | 1.4 | } |
| 3507 | |||
| 3508 | ~* (STRING) { | ||
| 3509 | wakaba | 1.6 | lang:Perl ($v => $token.value) { |
| 3510 | $reptxt .= $v; | ||
| 3511 | } | ||
| 3512 | wakaba | 1.4 | } (PERO) { |
| 3513 | ?lexmode EntityDeclaration; | ||
| 3514 | wakaba | 1.19 | |
| 3515 | lang:Perl { | ||
| 3516 | $self->{has_xref} = true; | ||
| 3517 | } | ||
| 3518 | wakaba | 1.6 | |
| 3519 | // TODO: Expand or wferror if internal subset | ||
| 3520 | wakaba | 1.4 | |
| 3521 | ~ (Name) { | ||
| 3522 | |||
| 3523 | } | ||
| 3524 | |||
| 3525 | ~ (REFC) { | ||
| 3526 | ?lexmode EntityValue; | ||
| 3527 | } else { | ||
| 3528 | ?lexmode EntityValue; | ||
| 3529 | } | ||
| 3530 | } (HCRO) { | ||
| 3531 | wakaba | 1.11 | &_HexadecimalCharacterReferenceEV_ ($vals => $vals); |
| 3532 | wakaba | 1.4 | |
| 3533 | wakaba | 1.6 | lang:Perl { |
| 3534 | wakaba | 1.11 | $reptxt .= $vals->[-1]; |
| 3535 | wakaba | 1.6 | } |
| 3536 | |||
| 3537 | wakaba | 1.4 | ~ (REFC) { |
| 3538 | ?lexmode EntityValue; | ||
| 3539 | } else { | ||
| 3540 | ?lexmode EntityValue; | ||
| 3541 | } | ||
| 3542 | } (CRO) { | ||
| 3543 | wakaba | 1.11 | &_NumericCharacterReferenceEV_ ($vals => $vals); |
| 3544 | wakaba | 1.6 | |
| 3545 | lang:Perl { | ||
| 3546 | wakaba | 1.11 | $reptxt .= $vals->[-1]; |
| 3547 | wakaba | 1.6 | } |
| 3548 | wakaba | 1.4 | |
| 3549 | ~ (REFC) { | ||
| 3550 | ?lexmode EntityValue; | ||
| 3551 | } else { | ||
| 3552 | ?lexmode EntityValue; | ||
| 3553 | } | ||
| 3554 | } (ERO) { | ||
| 3555 | wakaba | 1.11 | &_GeneralEntityReferenceEV_ ($vals => $vals); |
| 3556 | wakaba | 1.4 | |
| 3557 | wakaba | 1.6 | lang:Perl { |
| 3558 | wakaba | 1.11 | $reptxt .= '&' . $vals->[-1] . ';'; |
| 3559 | wakaba | 1.6 | } |
| 3560 | |||
| 3561 | wakaba | 1.4 | ~ (REFC) { |
| 3562 | ?lexmode EntityValue; | ||
| 3563 | } else { | ||
| 3564 | ?lexmode EntityValue; | ||
| 3565 | } | ||
| 3566 | } | ||
| 3567 | |||
| 3568 | ~ (LIT) { | ||
| 3569 | ?lexmode MarkupDeclaration; | ||
| 3570 | } (LITA) { | ||
| 3571 | ?lexmode MarkupDeclaration; | ||
| 3572 | } else { | ||
| 3573 | ?lexmode MarkupDeclaration; | ||
| 3574 | } | ||
| 3575 | wakaba | 1.6 | |
| 3576 | lang:Perl { | ||
| 3577 | $decl->{reptxt} = \$reptxt; | ||
| 3578 | wakaba | 1.19 | $decl->{has_replacement_text} = true; |
| 3579 | wakaba | 1.6 | } |
| 3580 | wakaba | 1.4 | } // _EntityValue |
| 3581 | |||
| 3582 | wakaba | 1.11 | rule _EntityValueA ($decl) { |
| 3583 | wakaba | 1.4 | ?lexmode EntityValueA; |
| 3584 | |||
| 3585 | my $vals; | ||
| 3586 | wakaba | 1.6 | my $reptxt; |
| 3587 | wakaba | 1.4 | lang:Perl { |
| 3588 | $vals = []; | ||
| 3589 | wakaba | 1.6 | $reptxt = ''; |
| 3590 | wakaba | 1.4 | } |
| 3591 | |||
| 3592 | ~* (STRING) { | ||
| 3593 | wakaba | 1.6 | lang:Perl ($v => $token.value) { |
| 3594 | $reptxt .= $v; | ||
| 3595 | } | ||
| 3596 | wakaba | 1.4 | } (PERO) { |
| 3597 | ?lexmode EntityDeclaration; | ||
| 3598 | |||
| 3599 | wakaba | 1.19 | lang:Perl { |
| 3600 | $self->{has_xref} = true; | ||
| 3601 | } | ||
| 3602 | |||
| 3603 | wakaba | 1.6 | // TODO: implement this |
| 3604 | wakaba | 1.4 | ~ (Name) { |
| 3605 | |||
| 3606 | } | ||
| 3607 | |||
| 3608 | ~ (REFC) { | ||
| 3609 | ?lexmode EntityValueA; | ||
| 3610 | } else { | ||
| 3611 | ?lexmode EntityValueA; | ||
| 3612 | } | ||
| 3613 | } (HCRO) { | ||
| 3614 | wakaba | 1.11 | &_HexadecimalCharacterReferenceEV_ ($vals => $vals); |
| 3615 | wakaba | 1.4 | |
| 3616 | wakaba | 1.6 | lang:Perl { |
| 3617 | wakaba | 1.11 | $reptxt .= $vals->[-1]; |
| 3618 | wakaba | 1.6 | } |
| 3619 | |||
| 3620 | |||
| 3621 | wakaba | 1.4 | ~ (REFC) { |
| 3622 | ?lexmode EntityValueA; | ||
| 3623 | } else { | ||
| 3624 | ?lexmode EntityValueA; | ||
| 3625 | } | ||
| 3626 | } (CRO) { | ||
| 3627 | wakaba | 1.11 | &_NumericCharacterReferenceEV_ ($vals => $vals); |
| 3628 | wakaba | 1.4 | |
| 3629 | wakaba | 1.6 | lang:Perl { |
| 3630 | wakaba | 1.11 | $reptxt .= $vals->[-1]; |
| 3631 | wakaba | 1.6 | } |
| 3632 | |||
| 3633 | wakaba | 1.4 | ~ (REFC) { |
| 3634 | ?lexmode EntityValueA; | ||
| 3635 | } else { | ||
| 3636 | ?lexmode EntityValueA; | ||
| 3637 | } | ||
| 3638 | } (ERO) { | ||
| 3639 | wakaba | 1.11 | &_GeneralEntityReferenceEV_ ($vals => $vals); |
| 3640 | wakaba | 1.4 | |
| 3641 | wakaba | 1.6 | lang:Perl { |
| 3642 | wakaba | 1.11 | $reptxt .= '&' . $vals->[-1] . ';'; |
| 3643 | wakaba | 1.6 | } |
| 3644 | |||
| 3645 | wakaba | 1.4 | ~ (REFC) { |
| 3646 | ?lexmode EntityValueA; | ||
| 3647 | } else { | ||
| 3648 | ?lexmode EntityValueA; | ||
| 3649 | } | ||
| 3650 | } | ||
| 3651 | |||
| 3652 | ~ (LITA) { | ||
| 3653 | ?lexmode MarkupDeclaration; | ||
| 3654 | } else { | ||
| 3655 | ?lexmode MarkupDeclaration; | ||
| 3656 | } | ||
| 3657 | wakaba | 1.6 | |
| 3658 | lang:Perl { | ||
| 3659 | $decl->{reptxt} = \$reptxt; | ||
| 3660 | wakaba | 1.19 | $decl->{has_replacement_text} = true; |
| 3661 | wakaba | 1.6 | } |
| 3662 | wakaba | 1.4 | } // _EntityValueA |
| 3663 | |||
| 3664 | |||
| 3665 | /* | ||
| 3666 | XML Name | ||
| 3667 | */ | ||
| 3668 | lexmode NameChar { | ||
| 3669 | $NameStartChar10 := [ | ||
| 3670 | '_' ':' | ||
| 3671 | // Letter | ||
| 3672 | // BaseChar | ||
| 3673 | U+0041..U+005A U+0061..U+007A U+00C0..U+00D6 | ||
| 3674 | U+00D8..U+00F6 U+00F8..U+00FF U+0100..U+0131 | ||
| 3675 | U+0134..U+013E U+0141..U+0148 U+014A..U+017E | ||
| 3676 | U+0180..U+01C3 U+01CD..U+01F0 U+01F4..U+01F5 | ||
| 3677 | U+01FA..U+0217 U+0250..U+02A8 U+02BB..U+02C1 | ||
| 3678 | U+0386 U+0388..U+038A U+038C U+038E..U+03A1 | ||
| 3679 | U+03A3..U+03CE U+03D0..U+03D6 U+03DA U+03DC | ||
| 3680 | U+03DE U+03E0 U+03E2..U+03F3 U+0401..U+040C | ||
| 3681 | U+040E..U+044F U+0451..U+045C U+045E..U+0481 | ||
| 3682 | U+0490..U+04C4 U+04C7..U+04C8 U+04CB..U+04CC | ||
| 3683 | U+04D0..U+04EB U+04EE..U+04F5 U+04F8..U+04F9 | ||
| 3684 | U+0531..U+0556 U+0559 U+0561..U+0586 | ||
| 3685 | U+05D0..U+05EA U+05F0..U+05F2 U+0621..U+063A | ||
| 3686 | U+0641..U+064A U+0671..U+06B7 U+06BA..U+06BE | ||
| 3687 | U+06C0..U+06CE U+06D0..U+06D3 U+06D5 | ||
| 3688 | U+06E5..U+06E6 U+0905..U+0939 U+093D | ||
| 3689 | U+0958..U+0961 U+0985..U+098C U+098F..U+0990 | ||
| 3690 | U+0993..U+09A8 U+09AA..U+09B0 U+09B2 | ||
| 3691 | U+09B6..U+09B9 U+09DC..U+09DD U+09DF..U+09E1 | ||
| 3692 | U+09F0..U+09F1 U+0A05..U+0A0A U+0A0F..U+0A10 | ||
| 3693 | U+0A13..U+0A28 U+0A2A..U+0A30 U+0A32..U+0A33 | ||
| 3694 | U+0A35..U+0A36 U+0A38..U+0A39 U+0A59..U+0A5C | ||
| 3695 | U+0A5E U+0A72..U+0A74 U+0A85..U+0A8B U+0A8D | ||
| 3696 | U+0A8F..U+0A91 U+0A93..U+0AA8 U+0AAA..U+0AB0 | ||
| 3697 | U+0AB2..U+0AB3 U+0AB5..U+0AB9 U+0ABD U+0AE0 | ||
| 3698 | U+0B05..U+0B0C U+0B0F..U+0B10 U+0B13..U+0B28 | ||
| 3699 | U+0B2A..U+0B30 U+0B32..U+0B33 U+0B36..U+0B39 | ||
| 3700 | U+0B3D U+0B5C..U+0B5D U+0B5F..U+0B61 | ||
| 3701 | U+0B85..U+0B8A U+0B8E..U+0B90 U+0B92..U+0B95 | ||
| 3702 | U+0B99..U+0B9A U+0B9C U+0B9E..U+0B9F | ||
| 3703 | U+0BA3..U+0BA4 U+0BA8..U+0BAA U+0BAE..U+0BB5 | ||
| 3704 | U+0BB7..U+0BB9 U+0C05..U+0C0C U+0C0E..U+0C10 | ||
| 3705 | U+0C12..U+0C28 U+0C2A..U+0C33 U+0C35..U+0C39 | ||
| 3706 | U+0C60..U+0C61 U+0C85..U+0C8C U+0C8E..U+0C90 | ||
| 3707 | U+0C92..U+0CA8 U+0CAA..U+0CB3 U+0CB5..U+0CB9 | ||
| 3708 | U+0CDE U+0CE0..U+0CE1 U+0D05..U+0D0C | ||
| 3709 | U+0D0E..U+0D10 U+0D12..U+0D28 U+0D2A..U+0D39 | ||
| 3710 | U+0D60..U+0D61 U+0E01..U+0E2E U+0E30 | ||
| 3711 | U+0E32..U+0E33 U+0E40..U+0E45 U+0E81..U+0E82 | ||
| 3712 | U+0E84 U+0E87..U+0E88 U+0E8A U+0E8D | ||
| 3713 | U+0E94..U+0E97 U+0E99..U+0E9F U+0EA1..U+0EA3 | ||
| 3714 | U+0EA5 U+0EA7 U+0EAA..U+0EAB U+0EAD..U+0EAE | ||
| 3715 | U+0EB0 U+0EB2..U+0EB3 U+0EBD U+0EC0..U+0EC4 | ||
| 3716 | U+0F40..U+0F47 U+0F49..U+0F69 U+10A0..U+10C5 | ||
| 3717 | U+10D0..U+10F6 U+1100 U+1102..U+1103 | ||
| 3718 | U+1105..U+1107 U+1109 U+110B..U+110C | ||
| 3719 | U+110E..U+1112 U+113C U+113E U+1140 U+114C | ||
| 3720 | U+114E U+1150 U+1154..U+1155 U+1159 | ||
| 3721 | U+115F..U+1161 U+1163 U+1165 U+1167 U+1169 | ||
| 3722 | U+116D..U+116E U+1172..U+1173 U+1175 U+119E | ||
| 3723 | U+11A8 U+11AB U+11AE..U+11AF U+11B7..U+11B8 | ||
| 3724 | U+11BA U+11BC..U+11C2 U+11EB U+11F0 U+11F9 | ||
| 3725 | U+1E00..U+1E9B U+1EA0..U+1EF9 U+1F00..U+1F15 | ||
| 3726 | U+1F18..U+1F1D U+1F20..U+1F45 U+1F48..U+1F4D | ||
| 3727 | U+1F50..U+1F57 U+1F59 U+1F5B U+1F5D | ||
| 3728 | U+1F5F..U+1F7D U+1F80..U+1FB4 U+1FB6..U+1FBC | ||
| 3729 | U+1FBE U+1FC2..U+1FC4 U+1FC6..U+1FCC | ||
| 3730 | U+1FD0..U+1FD3 U+1FD6..U+1FDB U+1FE0..U+1FEC | ||
| 3731 | U+1FF2..U+1FF4 U+1FF6..U+1FFC U+2126 | ||
| 3732 | U+212A..U+212B U+212E U+2180..U+2182 | ||
| 3733 | U+3041..U+3094 U+30A1..U+30FA U+3105..U+312C | ||
| 3734 | U+AC00..U+D7A3 | ||
| 3735 | // Ideographic | ||
| 3736 | U+4E00..U+9FA5 U+3007 U+3021..U+3029 | ||
| 3737 | ]; | ||
| 3738 | $NameChar10 := [ | ||
| 3739 | '.' '-' '_' ':' | ||
| 3740 | // Letter | ||
| 3741 | // BaseChar | ||
| 3742 | U+0041..U+005A U+0061..U+007A U+00C0..U+00D6 | ||
| 3743 | U+00D8..U+00F6 U+00F8..U+00FF U+0100..U+0131 | ||
| 3744 | U+0134..U+013E U+0141..U+0148 U+014A..U+017E | ||
| 3745 | U+0180..U+01C3 U+01CD..U+01F0 U+01F4..U+01F5 | ||
| 3746 | U+01FA..U+0217 U+0250..U+02A8 U+02BB..U+02C1 | ||
| 3747 | U+0386 U+0388..U+038A U+038C U+038E..U+03A1 | ||
| 3748 | U+03A3..U+03CE U+03D0..U+03D6 U+03DA U+03DC | ||
| 3749 | U+03DE U+03E0 U+03E2..U+03F3 U+0401..U+040C | ||
| 3750 | U+040E..U+044F U+0451..U+045C U+045E..U+0481 | ||
| 3751 | U+0490..U+04C4 U+04C7..U+04C8 U+04CB..U+04CC | ||
| 3752 | U+04D0..U+04EB U+04EE..U+04F5 U+04F8..U+04F9 | ||
| 3753 | U+0531..U+0556 U+0559 U+0561..U+0586 | ||
| 3754 | U+05D0..U+05EA U+05F0..U+05F2 U+0621..U+063A | ||
| 3755 | U+0641..U+064A U+0671..U+06B7 U+06BA..U+06BE | ||
| 3756 | U+06C0..U+06CE U+06D0..U+06D3 U+06D5 | ||
| 3757 | U+06E5..U+06E6 U+0905..U+0939 U+093D | ||
| 3758 | U+0958..U+0961 U+0985..U+098C U+098F..U+0990 | ||
| 3759 | U+0993..U+09A8 U+09AA..U+09B0 U+09B2 | ||
| 3760 | U+09B6..U+09B9 U+09DC..U+09DD U+09DF..U+09E1 | ||
| 3761 | U+09F0..U+09F1 U+0A05..U+0A0A U+0A0F..U+0A10 | ||
| 3762 | U+0A13..U+0A28 U+0A2A..U+0A30 U+0A32..U+0A33 | ||
| 3763 | U+0A35..U+0A36 U+0A38..U+0A39 U+0A59..U+0A5C | ||
| 3764 | U+0A5E U+0A72..U+0A74 U+0A85..U+0A8B U+0A8D | ||
| 3765 | U+0A8F..U+0A91 U+0A93..U+0AA8 U+0AAA..U+0AB0 | ||
| 3766 | U+0AB2..U+0AB3 U+0AB5..U+0AB9 U+0ABD U+0AE0 | ||
| 3767 | U+0B05..U+0B0C U+0B0F..U+0B10 U+0B13..U+0B28 | ||
| 3768 | U+0B2A..U+0B30 U+0B32..U+0B33 U+0B36..U+0B39 | ||
| 3769 | U+0B3D U+0B5C..U+0B5D U+0B5F..U+0B61 | ||
| 3770 | U+0B85..U+0B8A U+0B8E..U+0B90 U+0B92..U+0B95 | ||
| 3771 | U+0B99..U+0B9A U+0B9C U+0B9E..U+0B9F | ||
| 3772 | U+0BA3..U+0BA4 U+0BA8..U+0BAA U+0BAE..U+0BB5 | ||
| 3773 | U+0BB7..U+0BB9 U+0C05..U+0C0C U+0C0E..U+0C10 | ||
| 3774 | U+0C12..U+0C28 U+0C2A..U+0C33 U+0C35..U+0C39 | ||
| 3775 | U+0C60..U+0C61 U+0C85..U+0C8C U+0C8E..U+0C90 | ||
| 3776 | U+0C92..U+0CA8 U+0CAA..U+0CB3 U+0CB5..U+0CB9 | ||
| 3777 | U+0CDE U+0CE0..U+0CE1 U+0D05..U+0D0C | ||
| 3778 | U+0D0E..U+0D10 U+0D12..U+0D28 U+0D2A..U+0D39 | ||
| 3779 | U+0D60..U+0D61 U+0E01..U+0E2E U+0E30 | ||
| 3780 | U+0E32..U+0E33 U+0E40..U+0E45 U+0E81..U+0E82 | ||
| 3781 | U+0E84 U+0E87..U+0E88 U+0E8A U+0E8D | ||
| 3782 | U+0E94..U+0E97 U+0E99..U+0E9F U+0EA1..U+0EA3 | ||
| 3783 | U+0EA5 U+0EA7 U+0EAA..U+0EAB U+0EAD..U+0EAE | ||
| 3784 | U+0EB0 U+0EB2..U+0EB3 U+0EBD U+0EC0..U+0EC4 | ||
| 3785 | U+0F40..U+0F47 U+0F49..U+0F69 U+10A0..U+10C5 | ||
| 3786 | U+10D0..U+10F6 U+1100 U+1102..U+1103 | ||
| 3787 | U+1105..U+1107 U+1109 U+110B..U+110C | ||
| 3788 | U+110E..U+1112 U+113C U+113E U+1140 U+114C | ||
| 3789 | U+114E U+1150 U+1154..U+1155 U+1159 | ||
| 3790 | U+115F..U+1161 U+1163 U+1165 U+1167 U+1169 | ||
| 3791 | U+116D..U+116E U+1172..U+1173 U+1175 U+119E | ||
| 3792 | U+11A8 U+11AB U+11AE..U+11AF U+11B7..U+11B8 | ||
| 3793 | U+11BA U+11BC..U+11C2 U+11EB U+11F0 U+11F9 | ||
| 3794 | U+1E00..U+1E9B U+1EA0..U+1EF9 U+1F00..U+1F15 | ||
| 3795 | U+1F18..U+1F1D U+1F20..U+1F45 U+1F48..U+1F4D | ||
| 3796 | U+1F50..U+1F57 U+1F59 U+1F5B U+1F5D | ||
| 3797 | U+1F5F..U+1F7D U+1F80..U+1FB4 U+1FB6..U+1FBC | ||
| 3798 | U+1FBE U+1FC2..U+1FC4 U+1FC6..U+1FCC | ||
| 3799 | U+1FD0..U+1FD3 U+1FD6..U+1FDB U+1FE0..U+1FEC | ||
| 3800 | U+1FF2..U+1FF4 U+1FF6..U+1FFC U+2126 | ||
| 3801 | U+212A..U+212B U+212E U+2180..U+2182 | ||
| 3802 | U+3041..U+3094 U+30A1..U+30FA U+3105..U+312C | ||
| 3803 | U+AC00..U+D7A3 | ||
| 3804 | // Ideographic | ||
| 3805 | U+4E00..U+9FA5 U+3007 U+3021..U+3029 | ||
| 3806 | // Digit | ||
| 3807 | U+0030..U+0039 U+0660..U+0669 U+06F0..U+06F9 | ||
| 3808 | U+0966..U+096F U+09E6..U+09EF U+0A66..U+0A6F | ||
| 3809 | U+0AE6..U+0AEF U+0B66..U+0B6F U+0BE7..U+0BEF | ||
| 3810 | U+0C66..U+0C6F U+0CE6..U+0CEF U+0D66..U+0D6F | ||
| 3811 | U+0E50..U+0E59 U+0ED0..U+0ED9 U+0F20..U+0F29 | ||
| 3812 | // CombiningChar | ||
| 3813 | U+0300..U+0345 U+0360..U+0361 U+0483..U+0486 | ||
| 3814 | U+0591..U+05A1 U+05A3..U+05B9 U+05BB..U+05BD | ||
| 3815 | U+05BF U+05C1..U+05C2 U+05C4 U+064B..U+0652 | ||
| 3816 | U+0670 U+06D6..U+06DC U+06DD..U+06DF | ||
| 3817 | U+06E0..U+06E4 U+06E7..U+06E8 U+06EA..U+06ED | ||
| 3818 | U+0901..U+0903 U+093C U+093E..U+094C U+094D | ||
| 3819 | U+0951..U+0954 U+0962..U+0963 U+0981..U+0983 | ||
| 3820 | U+09BC U+09BE U+09BF U+09C0..U+09C4 | ||
| 3821 | U+09C7..U+09C8 U+09CB..U+09CD U+09D7 | ||
| 3822 | U+09E2..U+09E3 U+0A02 U+0A3C U+0A3E U+0A3F | ||
| 3823 | U+0A40..U+0A42 U+0A47..U+0A48 U+0A4B..U+0A4D | ||
| 3824 | U+0A70..U+0A71 U+0A81..U+0A83 U+0ABC | ||
| 3825 | U+0ABE..U+0AC5 U+0AC7..U+0AC9 U+0ACB..U+0ACD | ||
| 3826 | U+0B01..U+0B03 U+0B3C U+0B3E..U+0B43 | ||
| 3827 | U+0B47..U+0B48 U+0B4B..U+0B4D U+0B56..U+0B57 | ||
| 3828 | U+0B82..U+0B83 U+0BBE..U+0BC2 U+0BC6..U+0BC8 | ||
| 3829 | U+0BCA..U+0BCD U+0BD7 U+0C01..U+0C03 | ||
| 3830 | U+0C3E..U+0C44 U+0C46..U+0C48 U+0C4A..U+0C4D | ||
| 3831 | U+0C55..U+0C56 U+0C82..U+0C83 U+0CBE..U+0CC4 | ||
| 3832 | U+0CC6..U+0CC8 U+0CCA..U+0CCD U+0CD5..U+0CD6 | ||
| 3833 | U+0D02..U+0D03 U+0D3E..U+0D43 U+0D46..U+0D48 | ||
| 3834 | U+0D4A..U+0D4D U+0D57 U+0E31 U+0E34..U+0E3A | ||
| 3835 | U+0E47..U+0E4E U+0EB1 U+0EB4..U+0EB9 | ||
| 3836 | U+0EBB..U+0EBC U+0EC8..U+0ECD U+0F18..U+0F19 | ||
| 3837 | U+0F35 U+0F37 U+0F39 U+0F3E U+0F3F | ||
| 3838 | U+0F71..U+0F84 U+0F86..U+0F8B U+0F90..U+0F95 | ||
| 3839 | U+0F97 U+0F99..U+0FAD U+0FB1..U+0FB7 U+0FB9 | ||
| 3840 | U+20D0..U+20DC U+20E1 U+302A..U+302F U+3099 | ||
| 3841 | U+309A | ||
| 3842 | // Extender | ||
| 3843 | U+00B7 U+02D0 U+02D1 U+0387 U+0640 U+0E46 | ||
| 3844 | U+0EC6 U+3005 U+3031..U+3035 U+309D..U+309E | ||
| 3845 | U+30FC..U+30FE | ||
| 3846 | ]; | ||
| 3847 | wakaba | 1.1 | |
| 3848 | $NameStartChar11 := [ | ||
| 3849 | ':' '_' | ||
| 3850 | 'A' 'B' 'C' 'D' 'E' 'F' 'G' 'H' 'I' 'J' 'K' 'L' 'M' | ||
| 3851 | 'N' 'O' 'P' 'Q' 'R' 'S' 'T' 'U' 'V' 'W' 'X' 'Y' 'Z' | ||
| 3852 | 'a' 'b' 'c' 'd' 'e' 'f' 'g' 'h' 'i' 'j' 'k' 'l' 'm' | ||
| 3853 | 'n' 'o' 'p' 'q' 'r' 's' 't' 'u' 'v' 'w' 'x' 'y' 'z' | ||
| 3854 | U+00C0..U+00D6 U+00D8..U+00F6 U+00F8..U+02FF | ||
| 3855 | U+0370..U+037D U+037F..U+1FFF U+200C..U+200D | ||
| 3856 | U+2070..U+218F U+2C00..U+2FEF U+3001..U+D7FF | ||
| 3857 | U+F900..U+FDCF U+FDF0..U+FFFD U+10000..U+EFFFF | ||
| 3858 | ]; | ||
| 3859 | $NameChar11 := [ | ||
| 3860 | '-' '.' '0' '1' '2' '3' '4' '5' '6' '7' '8' '9' | ||
| 3861 | U+00B7 U+0300..U+036F U+203F..U+2040 | ||
| 3862 | // NameStartChar | ||
| 3863 | ':' '_' | ||
| 3864 | 'A' 'B' 'C' 'D' 'E' 'F' 'G' 'H' 'I' 'J' 'K' 'L' 'M' | ||
| 3865 | 'N' 'O' 'P' 'Q' 'R' 'S' 'T' 'U' 'V' 'W' 'X' 'Y' 'Z' | ||
| 3866 | 'a' 'b' 'c' 'd' 'e' 'f' 'g' 'h' 'i' 'j' 'k' 'l' 'm' | ||
| 3867 | 'n' 'o' 'p' 'q' 'r' 's' 't' 'u' 'v' 'w' 'x' 'y' 'z' | ||
| 3868 | U+00C0..U+00D6 U+00D8..U+00F6 U+00F8..U+02FF | ||
| 3869 | U+0370..U+037D U+037F..U+1FFF U+200C..U+200D | ||
| 3870 | U+2070..U+218F U+2C00..U+2FEF U+3001..U+D7FF | ||
| 3871 | U+F900..U+FDCF U+FDF0..U+FFFD U+10000..U+EFFFF | ||
| 3872 | ]; | ||
| 3873 | wakaba | 1.4 | } // NameChar |
| 3874 | |||
| 3875 | lexmode Name | ||
| 3876 | : extends => 'NameChar' | ||
| 3877 | { | ||
| 3878 | wakaba | 1.1 | Name : value := $NameStartChar11 $NameChar11*; |
| 3879 | } // Name | ||
| 3880 | wakaba | 1.4 | |
| 3881 | lexmode Nmtoken | ||
| 3882 | : extends => 'NameChar' | ||
| 3883 | { | ||
| 3884 | Nmtoken : value := $NameChar11*; | ||
| 3885 | } // Nmtoken | ||
| 3886 | wakaba | 1.1 | |
| 3887 | /* | ||
| 3888 | wakaba | 1.14 | S |
| 3889 | wakaba | 1.1 | */ |
| 3890 | lexmode S { | ||
| 3891 | wakaba | 1.14 | $s := [U+0009 U+000A U+000D U+0020]; |
| 3892 | S := $s+; | ||
| 3893 | wakaba | 1.1 | } // S |
| 3894 | |||
| 3895 | /* | ||
| 3896 | Document end scanning mode | ||
| 3897 | */ | ||
| 3898 | lexmode DocumentEnd | ||
| 3899 | : standalone | ||
| 3900 | : extends => 'S' | ||
| 3901 | { | ||
| 3902 | /* | ||
| 3903 | Processing instruction | ||
| 3904 | */ | ||
| 3905 | PIO := ['<'] ['?']; | ||
| 3906 | |||
| 3907 | /* | ||
| 3908 | Comment declaration | ||
| 3909 | */ | ||
| 3910 | CDO := ['<'] ['!'] ['-'] ['-']; | ||
| 3911 | } // DocumentEnd | ||
| 3912 | |||
| 3913 | /* | ||
| 3914 | Document misc scanning mode | ||
| 3915 | |||
| 3916 | This mode scans |Misc| constructions as well | ||
| 3917 | as document element's start tag. | ||
| 3918 | */ | ||
| 3919 | lexmode DocumentMisc | ||
| 3920 | : standalone | ||
| 3921 | : extends => 'DocumentEnd' | ||
| 3922 | { | ||
| 3923 | /* | ||
| 3924 | Document element start tag | ||
| 3925 | */ | ||
| 3926 | STAGO := ['<']; | ||
| 3927 | } // DocumentMisc | ||
| 3928 | |||
| 3929 | /* | ||
| 3930 | Document prolog scanning mode | ||
| 3931 | */ | ||
| 3932 | lexmode DocumentProlog | ||
| 3933 | : standalone | ||
| 3934 | : extends => 'DocumentMisc' | ||
| 3935 | { | ||
| 3936 | /* | ||
| 3937 | |DOCTYPE| declaration | ||
| 3938 | */ | ||
| 3939 | MDO := ['<'] ['!']; | ||
| 3940 | } // DocumentProlog | ||
| 3941 | wakaba | 1.16 | |
| 3942 | wakaba | 1.1 | /* |
| 3943 | wakaba | 1.16 | Document Start Scanning Mode |
| 3944 | wakaba | 1.1 | */ |
| 3945 | lexmode DocumentStart | ||
| 3946 | : standalone | ||
| 3947 | : extends => 'DocumentProlog' | ||
| 3948 | { | ||
| 3949 | /* | ||
| 3950 | XML declaration | ||
| 3951 | */ | ||
| 3952 | wakaba | 1.16 | XDO := ['<'] ['?'] ['x'] ['m'] ['l']; |
| 3953 | wakaba | 1.1 | } // DocumentStart |
| 3954 | |||
| 3955 | /* | ||
| 3956 | Comment declaration scanning mode | ||
| 3957 | */ | ||
| 3958 | lexmode CommentDeclaration | ||
| 3959 | : standalone | ||
| 3960 | { | ||
| 3961 | /* | ||
| 3962 | Comment close | ||
| 3963 | */ | ||
| 3964 | COM := ['-'] ['-']; | ||
| 3965 | |||
| 3966 | /* | ||
| 3967 | Comment data | ||
| 3968 | */ | ||
| 3969 | wakaba | 1.15 | ?default-token STRING : value; |
| 3970 | wakaba | 1.1 | } // CommentDeclaration |
| 3971 | |||
| 3972 | /* | ||
| 3973 | Processing instruction name and |S| scanning mode | ||
| 3974 | */ | ||
| 3975 | lexmode PIName | ||
| 3976 | : standalone | ||
| 3977 | : extends => 'Name' | ||
| 3978 | : extends => 'S' | ||
| 3979 | { | ||
| 3980 | /* | ||
| 3981 | Processing instruction close | ||
| 3982 | */ | ||
| 3983 | PIC := ['?'] ['>']; | ||
| 3984 | } // PIName | ||
| 3985 | |||
| 3986 | /* | ||
| 3987 | wakaba | 1.16 | Processing instruction name and |S| scanning mode |
| 3988 | */ | ||
| 3989 | lexmode PINmtoken | ||
| 3990 | : standalone | ||
| 3991 | : extends => 'Nmtoken' | ||
| 3992 | : extends => 'S' | ||
| 3993 | { | ||
| 3994 | /* | ||
| 3995 | Processing instruction close | ||
| 3996 | */ | ||
| 3997 | PIC := ['?'] ['>']; | ||
| 3998 | } // PINmtoken | ||
| 3999 | |||
| 4000 | /* | ||
| 4001 | wakaba | 1.1 | Processing instruction data scanning mode |
| 4002 | */ | ||
| 4003 | lexmode PIData | ||
| 4004 | : standalone | ||
| 4005 | { | ||
| 4006 | /* | ||
| 4007 | Processing instruction close | ||
| 4008 | */ | ||
| 4009 | PIC := ['?'] ['>']; | ||
| 4010 | |||
| 4011 | /* | ||
| 4012 | Processing instruction target data | ||
| 4013 | */ | ||
| 4014 | wakaba | 1.2 | ?default-token DATA : value; |
| 4015 | wakaba | 1.1 | } // PIData |
| 4016 | |||
| 4017 | /* | ||
| 4018 | Content of element scanning mode | ||
| 4019 | */ | ||
| 4020 | lexmode ElementContent | ||
| 4021 | : standalone | ||
| 4022 | { | ||
| 4023 | /* | ||
| 4024 | Start tag open | ||
| 4025 | */ | ||
| 4026 | STAGO := ['<']; | ||
| 4027 | |||
| 4028 | /* | ||
| 4029 | End tag open | ||
| 4030 | */ | ||
| 4031 | ETAGO := ['<'] ['/']; | ||
| 4032 | |||
| 4033 | /* | ||
| 4034 | Hexadecimal character reference open | ||
| 4035 | */ | ||
| 4036 | HCRO := ['&'] ['#'] ['x']; | ||
| 4037 | |||
| 4038 | /* | ||
| 4039 | Numeric character reference open | ||
| 4040 | */ | ||
| 4041 | CRO := ['&'] ['#']; | ||
| 4042 | |||
| 4043 | /* | ||
| 4044 | General entity reference open | ||
| 4045 | */ | ||
| 4046 | ERO := ['&']; | ||
| 4047 | |||
| 4048 | /* | ||
| 4049 | Comment declaration open | ||
| 4050 | */ | ||
| 4051 | CDO := ['<'] ['!'] ['-'] ['-']; | ||
| 4052 | |||
| 4053 | /* | ||
| 4054 | CDATA section open | ||
| 4055 | */ | ||
| 4056 | CDSO := ['<'] ['!'] ['['] | ||
| 4057 | ['C'] ['D'] ['A'] ['T'] ['A'] ['[']; | ||
| 4058 | |||
| 4059 | /* | ||
| 4060 | Processing instruction open | ||
| 4061 | */ | ||
| 4062 | PIO := ['<'] ['?']; | ||
| 4063 | wakaba | 1.2 | |
| 4064 | /* | ||
| 4065 | Markup section end | ||
| 4066 | */ | ||
| 4067 | MSE := [']'] [']'] ['>']; | ||
| 4068 | |||
| 4069 | /* | ||
| 4070 | Character data | ||
| 4071 | */ | ||
| 4072 | /* | ||
| 4073 | Character data and/or |MSE| | ||
| 4074 | */ | ||
| 4075 | ?default-token CharData : value; | ||
| 4076 | wakaba | 1.1 | } // ElementContent |
| 4077 | |||
| 4078 | /* | ||
| 4079 | CDATA section content scanning mode | ||
| 4080 | */ | ||
| 4081 | lexmode CDATASectionContent | ||
| 4082 | : standalone | ||
| 4083 | { | ||
| 4084 | /* | ||
| 4085 | Markup section end | ||
| 4086 | */ | ||
| 4087 | MSE := [']'] [']'] ['>']; | ||
| 4088 | |||
| 4089 | /* | ||
| 4090 | Character data | ||
| 4091 | wakaba | 1.2 | */ |
| 4092 | ?default-token CData : value; | ||
| 4093 | wakaba | 1.1 | } // CDATASectionContent |
| 4094 | |||
| 4095 | lexmode EntityReference | ||
| 4096 | : standalone | ||
| 4097 | : extends => 'Name' | ||
| 4098 | { | ||
| 4099 | /* | ||
| 4100 | Reference close | ||
| 4101 | */ | ||
| 4102 | REFC := [';']; | ||
| 4103 | } // EntityReference | ||
| 4104 | |||
| 4105 | lexmode NumericCharacterReference | ||
| 4106 | : standalone | ||
| 4107 | { | ||
| 4108 | /* | ||
| 4109 | Decimal number | ||
| 4110 | */ | ||
| 4111 | $digit := ['0' '1' '2' '3' '4' '5' '6' '7' '8' '9']; | ||
| 4112 | NUMBER : value := $digit+; | ||
| 4113 | |||
| 4114 | /* | ||
| 4115 | Reference close | ||
| 4116 | */ | ||
| 4117 | REFC := [';']; | ||
| 4118 | } // NumericCharacterReference | ||
| 4119 | |||
| 4120 | lexmode HexadecimalCharacterReference | ||
| 4121 | : standalone | ||
| 4122 | { | ||
| 4123 | /* | ||
| 4124 | Hexadecimal number | ||
| 4125 | */ | ||
| 4126 | $hexdigit := ['0' '1' '2' '3' '4' '5' '6' '7' '8' '9' | ||
| 4127 | 'A' 'B' 'C' 'D' 'E' 'F' | ||
| 4128 | 'a' 'b' 'c' 'd' 'e' 'f']; | ||
| 4129 | Hex : value := $hexdigit+; | ||
| 4130 | |||
| 4131 | /* | ||
| 4132 | Reference close | ||
| 4133 | */ | ||
| 4134 | REFC := [';']; | ||
| 4135 | } // HexadecimalCharacterReference | ||
| 4136 | |||
| 4137 | wakaba | 1.16 | /* |
| 4138 | The XML Declaration and Text Declaration Scanning Mode | ||
| 4139 | */ | ||
| 4140 | wakaba | 1.3 | lexmode XMLDeclaration |
| 4141 | : standalone | ||
| 4142 | : extends => 'Name' | ||
| 4143 | : extends => 'S' | ||
| 4144 | { | ||
| 4145 | |||
| 4146 | /* | ||
| 4147 | Value indicator | ||
| 4148 | */ | ||
| 4149 | VI := ['=']; | ||
| 4150 | |||
| 4151 | /* | ||
| 4152 | Literal open | ||
| 4153 | */ | ||
| 4154 | LIT := ['"']; | ||
| 4155 | LITA := [U+0027]; | ||
| 4156 | |||
| 4157 | /* | ||
| 4158 | Processing instruction close | ||
| 4159 | wakaba | 1.16 | |
| 4160 | The |PIC|, i.e. |?>|, is split into two tokens |PIC1| and |PIC2| | ||
| 4161 | in this scanning mode to set |no_xml11_eol| flag in the | ||
| 4162 | |_XMLDeclaration_| |rule|. | ||
| 4163 | wakaba | 1.3 | */ |
| 4164 | wakaba | 1.16 | PIC1 := ['?']; |
| 4165 | PIC2 := ['>']; | ||
| 4166 | wakaba | 1.3 | } // XMLDeclaration |
| 4167 | wakaba | 1.16 | |
| 4168 | /* | ||
| 4169 | The XML Declaration or Processing Instruction Scanning Mode | ||
| 4170 | |||
| 4171 | '<?xml' S ... | ||
| 4172 | (DocumentStart) (XMLDeclarationOrPI) (XMLDeclaration) | ||
| 4173 | '<?xml' '1' S ... | ||
| 4174 | (DocumentStart) (XMLDeclarationOrPI) (PINmtoken)(PIName) (PITarget) | ||
| 4175 | '<?xml' '-' 'stylesheet' S ... | ||
| 4176 | (DocumentStart) (XMLDeclarationOrPI) (PINmtoken) (PIName) (PITarget) | ||
| 4177 | */ | ||
| 4178 | lexmode XMLDeclarationOrPI | ||
| 4179 | : standalone | ||
| 4180 | : extends => 'NameChar' | ||
| 4181 | : extends => 'S' | ||
| 4182 | { | ||
| 4183 | NameChar : value := $NameChar11; | ||
| 4184 | } // XMLDeclarationOrPI | ||
| 4185 | wakaba | 1.3 | |
| 4186 | wakaba | 1.1 | lexmode StartTag |
| 4187 | : standalone | ||
| 4188 | : extends => 'Name' | ||
| 4189 | : extends => 'S' | ||
| 4190 | { | ||
| 4191 | |||
| 4192 | /* | ||
| 4193 | Value indicator | ||
| 4194 | */ | ||
| 4195 | VI := ['=']; | ||
| 4196 | |||
| 4197 | /* | ||
| 4198 | Literal open | ||
| 4199 | */ | ||
| 4200 | LIT := ['"']; | ||
| 4201 | LITA := [U+0027]; | ||
| 4202 | |||
| 4203 | /* | ||
| 4204 | Tag close | ||
| 4205 | */ | ||
| 4206 | TAGC := ['>']; | ||
| 4207 | |||
| 4208 | /* | ||
| 4209 | wakaba | 1.6 | Null end-tag enabled start-tag close |
| 4210 | wakaba | 1.1 | */ |
| 4211 | wakaba | 1.6 | NESTC := ['/']; |
| 4212 | wakaba | 1.1 | } // StartTag |
| 4213 | |||
| 4214 | lexmode EndTag | ||
| 4215 | : standalone | ||
| 4216 | : extends => 'Name' | ||
| 4217 | : extends => 'S' | ||
| 4218 | { | ||
| 4219 | /* | ||
| 4220 | Tag close | ||
| 4221 | */ | ||
| 4222 | TAGC := ['>']; | ||
| 4223 | } // EndTag | ||
| 4224 | |||
| 4225 | lexmode AttributeValueLiteral_ { | ||
| 4226 | ERO := ['&']; | ||
| 4227 | CRO := ['&'] ['#']; | ||
| 4228 | HCRO := ['&'] ['#'] ['x']; | ||
| 4229 | } // AttributeValueLiteral_ | ||
| 4230 | |||
| 4231 | lexmode AttributeValueLiteral | ||
| 4232 | : standalone | ||
| 4233 | : extends => 'AttributeValueLiteral_' | ||
| 4234 | { | ||
| 4235 | LIT := ['"']; | ||
| 4236 | wakaba | 1.3 | STRING : value := [^'"' '&' '<']+; |
| 4237 | wakaba | 1.1 | } // AttributeValueLiteral |
| 4238 | |||
| 4239 | lexmode AttributeValueLiteralA | ||
| 4240 | : standalone | ||
| 4241 | : extends => 'AttributeValueLiteral_' | ||
| 4242 | { | ||
| 4243 | wakaba | 1.4 | LITA := [U+0027]; |
| 4244 | wakaba | 1.3 | STRING : value := [^U+0027 '&' '<']+; |
| 4245 | wakaba | 1.1 | } // AttributeValueLiteralA |
| 4246 | wakaba | 1.11 | |
| 4247 | lexmode AttributeValueLiteralE | ||
| 4248 | : standalone | ||
| 4249 | : extends => 'AttributeValueLiteral_' | ||
| 4250 | { | ||
| 4251 | STRING : value := [^'&' '<']+; | ||
| 4252 | } // AttributeValueLiteralE | ||
| 4253 | wakaba | 1.4 | |
| 4254 | lexmode EntityValue_ { | ||
| 4255 | PERO := ['%']; | ||
| 4256 | ERO := ['&']; | ||
| 4257 | CRO := ['&'] ['#']; | ||
| 4258 | HCRO := ['&'] ['#'] ['x']; | ||
| 4259 | } // EntityValue_ | ||
| 4260 | |||
| 4261 | lexmode EntityValue | ||
| 4262 | : standalone | ||
| 4263 | : extends => 'EntityValue_' | ||
| 4264 | { | ||
| 4265 | LIT := ['"']; | ||
| 4266 | STRING : value := [^'"' '&' '%']+; | ||
| 4267 | } // EntityValue | ||
| 4268 |