| 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 |
$Date: 2005/12/28 03:50:17 $ |
| 14 |
|
| 15 |
@DefaultFor: ManakaiDOM|ManakaiDOMLatest |
| 16 |
|
| 17 |
@Require: |
| 18 |
@@Module: |
| 19 |
@@@QName: MDOM|DOMLS |
| 20 |
@@@WithFor: ManakaiDOM|ManakaiDOMLatest |
| 21 |
|
| 22 |
Namespace: |
| 23 |
@dis: |
| 24 |
http://suika.fam.cx/~wakaba/archive/2004/8/18/lang#dis-- |
| 25 |
@DOMCore: |
| 26 |
http://suika.fam.cx/~wakaba/archive/2004/8/18/dom-core# |
| 27 |
@DOMMain: |
| 28 |
http://suika.fam.cx/~wakaba/archive/2004/dom/main# |
| 29 |
@dx: |
| 30 |
http://suika.fam.cx/~wakaba/archive/2005/manakai/Util/Error/DOMException# |
| 31 |
@ecore: |
| 32 |
http://suika.fam.cx/~wakaba/archive/2005/manakai/Util/Error/Core/ |
| 33 |
@f: |
| 34 |
http://suika.fam.cx/~wakaba/archive/2004/dom/feature# |
| 35 |
@idl: |
| 36 |
http://suika.fam.cx/~wakaba/archive/2004/dis/IDL# |
| 37 |
@infoset: |
| 38 |
http://www.w3.org/2001/04/infoset# |
| 39 |
@lang: |
| 40 |
http://suika.fam.cx/~wakaba/archive/2004/8/18/lang# |
| 41 |
@license: |
| 42 |
http://suika.fam.cx/~wakaba/archive/2004/8/18/license# |
| 43 |
@LSEV: |
| 44 |
http://www.w3.org/2002/DOMLS |
| 45 |
@ManakaiDOM: |
| 46 |
http://suika.fam.cx/~wakaba/archive/2004/8/18/manakai-dom# |
| 47 |
@ManakaiDOMLS: |
| 48 |
http://suika.fam.cx/~wakaba/archive/2004/mdom-ls# |
| 49 |
@MDOM: |
| 50 |
http://suika.fam.cx/~wakaba/archive/2004/8/18/manakai-dom#ManakaiDOM. |
| 51 |
@MDOMX: |
| 52 |
http://suika.fam.cx/~wakaba/archive/2004/8/4/manakai-dom-exception# |
| 53 |
@rdf: |
| 54 |
http://www.w3.org/1999/02/22-rdf-syntax-ns# |
| 55 |
@rdfs: |
| 56 |
http://www.w3.org/2000/01/rdf-schema# |
| 57 |
@t: |
| 58 |
http://suika.fam.cx/~wakaba/archive/2004/dom/tree# |
| 59 |
@xml: |
| 60 |
http://www.w3.org/XML/1998/namespace |
| 61 |
@xmlns: |
| 62 |
http://www.w3.org/2000/xmlns/ |
| 63 |
@xp: |
| 64 |
http://suika.fam.cx/~wakaba/archive/2004/dom/xml-parser# |
| 65 |
|
| 66 |
## -- Features |
| 67 |
|
| 68 |
ElementTypeBinding: |
| 69 |
@Name: FeatureDef |
| 70 |
@ElementType: |
| 71 |
dis:ResourceDef |
| 72 |
@ShadowContent: |
| 73 |
@@rdf:type: f|Feature |
| 74 |
@@For: =ManakaiDOM|all |
| 75 |
|
| 76 |
ElementTypeBinding: |
| 77 |
@Name: FeatureVerDef |
| 78 |
@ElementType: |
| 79 |
dis:ResourceDef |
| 80 |
@ShadowContent: |
| 81 |
@@rdf:type: f|Feature |
| 82 |
|
| 83 |
ElementTypeBinding: |
| 84 |
@Name: featureQName |
| 85 |
@ElementType: |
| 86 |
f:name |
| 87 |
@ShadowContent: |
| 88 |
@@ContentType: DISCore|QName |
| 89 |
|
| 90 |
ResourceDef: |
| 91 |
@QName: DOMString |
| 92 |
@AliasFor: DOMMain|DOMString |
| 93 |
@For: ManakaiDOM|DOM |
| 94 |
|
| 95 |
ResourceDef: |
| 96 |
@QName: Node |
| 97 |
@AliasFor: t|Node |
| 98 |
@For: ManakaiDOM|DOM |
| 99 |
|
| 100 |
ResourceDef: |
| 101 |
@QName: Element |
| 102 |
@AliasFor: t|Element |
| 103 |
@For: ManakaiDOM|DOM |
| 104 |
|
| 105 |
ResourceDef: |
| 106 |
@QName: Document |
| 107 |
@AliasFor: t|Document |
| 108 |
@For: ManakaiDOM|DOM |
| 109 |
|
| 110 |
ElementTypeBinding: |
| 111 |
@Name: ClsDef |
| 112 |
@ElementType: |
| 113 |
dis:ResourceDef |
| 114 |
@ShadowContent: |
| 115 |
@@rdf:type: |
| 116 |
@@@@: dis|MultipleResource |
| 117 |
@@@ForCheck: !ManakaiDOM|ForIF !ManakaiDOM|ForClass |
| 118 |
@@resourceFor: |
| 119 |
@@@@: ManakaiDOM|ForClass |
| 120 |
@@@ForCheck: ManakaiDOM|ManakaiDOM !=ManakaiDOM|ManakaiDOM |
| 121 |
@@For: ManakaiDOM|DOM3 |
| 122 |
@@For: =ManakaiDOM|ManakaiDOM |
| 123 |
|
| 124 |
@@rdf:type: |
| 125 |
@@@@: DISLang|Class |
| 126 |
@@@ForCheck: ManakaiDOM|ForClass |
| 127 |
|
| 128 |
ElementTypeBinding: |
| 129 |
@Name: ClsQName |
| 130 |
@ElementType: |
| 131 |
dis:QName |
| 132 |
@ShadowContent: |
| 133 |
@@ForCheck: ManakaiDOM|ForClass |
| 134 |
|
| 135 |
ElementTypeBinding: |
| 136 |
@Name: ClsISA |
| 137 |
@ElementType: |
| 138 |
dis:ISA |
| 139 |
@ShadowContent: |
| 140 |
@@ForCheck: ManakaiDOM|ForClass |
| 141 |
|
| 142 |
ElementTypeBinding: |
| 143 |
@Name: nullCase |
| 144 |
@ElementType: |
| 145 |
dis:ResourceDef |
| 146 |
@ShadowContent: |
| 147 |
@@rdf:type: ManakaiDOM|InCase |
| 148 |
@@Value: |
| 149 |
@@@is-null:1 |
| 150 |
|
| 151 |
ResourceDef: |
| 152 |
@QName: LSParser |
| 153 |
@AliasFor: DOMLS|LSParser |
| 154 |
@For: ManakaiDOM|DOM3 |
| 155 |
|
| 156 |
ClsDef: |
| 157 |
@ClsQName: ManakaiXMLParser |
| 158 |
|
| 159 |
@Implement: DOMLS|LSParser |
| 160 |
|
| 161 |
@f:implements: |
| 162 |
@@@: DOMLS|LSFeature30 |
| 163 |
@@For: ManakaiDOM|DOM3 |
| 164 |
|
| 165 |
@DISLang:role: DOMLS|ParserRole |
| 166 |
|
| 167 |
@Attr: |
| 168 |
@@Name: domConfig |
| 169 |
@@enDesc: |
| 170 |
The configuration of the parser. |
| 171 |
|
| 172 |
@@Get: |
| 173 |
@@@Type: DOMCore|DOMConfiguration |
| 174 |
@@@enDesc: The DOM configuration object. |
| 175 |
@@@PerlDef: |
| 176 |
__CODE{DOMCore|getConfigObject:: |
| 177 |
$target => $self, |
| 178 |
$targetHash => $self, |
| 179 |
$targetType => {<IFName::LSParser>}, |
| 180 |
$result => $r, |
| 181 |
}__; |
| 182 |
|
| 183 |
@Method: |
| 184 |
@@ManakaiDOM:isForInternal:1 |
| 185 |
@@ForCheck: ManakaiDOM|ForClass |
| 186 |
@@Operator: DISPerl|NewMethod |
| 187 |
@@enDesc: |
| 188 |
Creates a new instance of the object. |
| 189 |
@@Param: |
| 190 |
@@@Name: impl |
| 191 |
@@@Type: DOMLS|GLSImplementation |
| 192 |
@@@enDesc: |
| 193 |
The implementation from which the parser is created. |
| 194 |
@@Param: |
| 195 |
@@@Name: features |
| 196 |
@@@Type: DOMString |
| 197 |
@@@dis:actualType: f|FeaturesString |
| 198 |
@@@enDesc: |
| 199 |
The set of features requested for the parser. |
| 200 |
@@Return: |
| 201 |
@@@Type: DOMMain|DOMObject |
| 202 |
@@@dis:actualType: LSParser |
| 203 |
@@@enDesc: |
| 204 |
The newly created parser. |
| 205 |
@@@PerlDef: |
| 206 |
$r = bless { |
| 207 |
<H::DOMCore:implementation> => $impl, |
| 208 |
}, $self; |
| 209 |
|
| 210 |
@Method: |
| 211 |
@@Name: parseString |
| 212 |
@@enImplNote: |
| 213 |
Non-standard - to be removed |
| 214 |
|
| 215 |
@@Param: |
| 216 |
@@@Name: sourceText |
| 217 |
@@@Type: DOMString |
| 218 |
@@Return: |
| 219 |
@@@Type: Document |
| 220 |
@@@PerlDef: |
| 221 |
|
| 222 |
$self->{char} = []; |
| 223 |
$self->{token} = []; |
| 224 |
$self->{source} = $sourceText; |
| 225 |
|
| 226 |
__DEEP{ |
| 227 |
$r = $self->_parse_DocumentEntity |
| 228 |
($self->{<H::DOMCore:implementation>}); |
| 229 |
}__; |
| 230 |
|
| 231 |
@Method: |
| 232 |
@@Name: shiftChar |
| 233 |
@@ManakaiDOM:isForInternal:1 |
| 234 |
@@ForCheck: ManakaiDOM|ForClass |
| 235 |
@@enDesc: |
| 236 |
Returns the next character. |
| 237 |
@@Return: |
| 238 |
@@@Type: idl|long||ManakaiDOM|all |
| 239 |
@@@enDesc: |
| 240 |
The code position number of the next character, if any, |
| 241 |
or <CODE::-2>. |
| 242 |
@@@PerlDef: |
| 243 |
if (@{$self->{char}}) { |
| 244 |
$r = shift @{$self->{char}}; |
| 245 |
} else { |
| 246 |
my $char = substr ($self->{source}, pos ($self->{source}), 1); |
| 247 |
pos ($self->{source})++; |
| 248 |
|
| 249 |
if (length $char) { |
| 250 |
$r = ord $char; |
| 251 |
} else { |
| 252 |
$r = -1; |
| 253 |
} |
| 254 |
} |
| 255 |
|
| 256 |
@Method: |
| 257 |
@@ManakaiDOM:isForInternal: 1 |
| 258 |
@@Operator: ManakaiDOM|MUErrorHandler |
| 259 |
@@enDesc: |
| 260 |
When a <IF::ecore|ErrorInterface||ManakaiDOM|Perl> is <Perl::report>ed, |
| 261 |
then this method is invoked. |
| 262 |
|
| 263 |
The method calls the <cfg::DOMCore|error-handler> if the error is of |
| 264 |
<IF::DOMCore|DOMError>. Otherwise, the error is re-thrown so that |
| 265 |
corresponding <Perl::catch> clause, if any, can catch the error. |
| 266 |
@@Param: |
| 267 |
@@@Name: err |
| 268 |
@@@Type: ecore|ErrorInterface||ManakaiDOM|Perl |
| 269 |
@@@enDesc: |
| 270 |
The reported error object. |
| 271 |
@@Return: |
| 272 |
@@@Type: DISPerl|Any |
| 273 |
@@@enDesc: |
| 274 |
If the <P::err> is a <IF::DOMCore|DOMError>, then the return value |
| 275 |
of the error handler. |
| 276 |
|
| 277 |
{NOTE:: If the error is thrown, the method never returns. |
| 278 |
} |
| 279 |
@@@nullCase: |
| 280 |
@@@@enDesc: |
| 281 |
No error handler. |
| 282 |
@@@PerlDef: |
| 283 |
if ($err->isa (<IFName::DOMCore|DOMError||ManakaiDOM|ManakaiDOM>)) { |
| 284 |
__DEEP{ |
| 285 |
A: { |
| 286 |
my $cfg = $self-><AG::LSParser.domConfig>; |
| 287 |
my $h = $cfg-><M::DOMCore|DOMConfiguration.getParameter> |
| 288 |
('error-handler'); |
| 289 |
$r = $h-><M::DOMCore|DOMErrorHandler.handleError> ($err); |
| 290 |
} # A |
| 291 |
}__; |
| 292 |
} else { |
| 293 |
$err-><M::ecore|ErrorInterface||ManakaiDOM|Perl.throw>; |
| 294 |
} |
| 295 |
|
| 296 |
@DISPerl:dpgDef: |
| 297 |
|
| 298 |
/* |
| 299 |
XML Document Entity |
| 300 |
|
| 301 |
document := prolog element *Misc |
| 302 |
- *Char RestrictedChar *Char ;; [1] |
| 303 |
*/ |
| 304 |
rule DocumentEntity ($impl) : standalone { |
| 305 |
my $doc : return; |
| 306 |
|
| 307 |
lang:Perl { |
| 308 |
$doc = $impl-><M::DOMImpl.createDocument>; |
| 309 |
$doc-><AS::Document.strictErrorChecking> (false); |
| 310 |
} |
| 311 |
|
| 312 |
/* |
| 313 |
prolog := XMLDecl? *Misc [doctypedecl *Misc] ;; [22] |
| 314 |
*/ |
| 315 |
?lexmode 'DocumentStart'; |
| 316 |
|
| 317 |
&XMLDeclarationOpt ($doc => $doc); |
| 318 |
|
| 319 |
?lexmode 'DocumentProlog'; |
| 320 |
|
| 321 |
// *Misc |
| 322 |
~* (CDO) { |
| 323 |
&_CommentDeclaration_ ($doc => $doc, $parent => $doc); |
| 324 |
|
| 325 |
~ (MDC) { |
| 326 |
?lexmode DocumentProlog; |
| 327 |
} else { |
| 328 |
?lexmode DocumentProlog; |
| 329 |
} |
| 330 |
} (PIO) { |
| 331 |
&_ProcessingInstruction_ ($doc => $doc, $parent => $doc); |
| 332 |
|
| 333 |
~ (PIC) { |
| 334 |
?lexmode 'DocumentProlog'; |
| 335 |
} else { |
| 336 |
?lexmode DocumentProlog; |
| 337 |
} |
| 338 |
} (S) { |
| 339 |
// |
| 340 |
} |
| 341 |
|
| 342 |
// doctypedecl |
| 343 |
~? (MDO) { |
| 344 |
&_DocumentTypeDeclaration_ ($doc => $doc); |
| 345 |
|
| 346 |
~ (MDC) { } |
| 347 |
} |
| 348 |
|
| 349 |
?lexmode 'DocumentMisc'; |
| 350 |
|
| 351 |
// *Misc |
| 352 |
~* (CDO) { |
| 353 |
&_CommentDeclaration_ ($doc => $doc, $parent => $doc); |
| 354 |
|
| 355 |
~ (MDC) { |
| 356 |
?lexmode DocumentMisc; |
| 357 |
} else { |
| 358 |
?lexmode DocumentMisc; |
| 359 |
} |
| 360 |
} (PIO) { |
| 361 |
&_ProcessingInstruction_ ($doc => $doc, $parent => $doc); |
| 362 |
|
| 363 |
~ (PIC) { |
| 364 |
?lexmode 'DocumentMisc'; |
| 365 |
} else { |
| 366 |
?lexmode DocumentMisc; |
| 367 |
} |
| 368 |
} (S) { |
| 369 |
// |
| 370 |
} |
| 371 |
|
| 372 |
// Document element |
| 373 |
~ (STAGO) { |
| 374 |
&Element_ ($doc => $doc, $parent => $doc) |
| 375 |
: unshift-current-token; |
| 376 |
~ (TAGC) { |
| 377 |
?lexmode DocumentEnd; |
| 378 |
} else { |
| 379 |
?lexmode DocumentEnd; |
| 380 |
} |
| 381 |
} else { |
| 382 |
?lexmode 'DocumentEnd'; |
| 383 |
} |
| 384 |
|
| 385 |
// *Misc |
| 386 |
~* (CDO) { |
| 387 |
&_CommentDeclaration_ ($doc => $doc, $parent => $doc); |
| 388 |
|
| 389 |
~ (MDC) { |
| 390 |
?lexmode DocumentEnd; |
| 391 |
} else { |
| 392 |
?lexmode DocumentEnd; |
| 393 |
} |
| 394 |
} (PIO) { |
| 395 |
&_ProcessingInstruction_ ($doc => $doc, $parent => $doc); |
| 396 |
~ (PIC) { |
| 397 |
?lexmode 'DocumentEnd'; |
| 398 |
} else { |
| 399 |
?lexmode DocumentEnd; |
| 400 |
} |
| 401 |
} (S) { |
| 402 |
// |
| 403 |
} |
| 404 |
|
| 405 |
~ (#EOF) { } |
| 406 |
|
| 407 |
lang:Perl { |
| 408 |
if ($self->{has_error}) { |
| 409 |
__EXCEPTION{DOMLS|PARSE_ERR:: |
| 410 |
}__; |
| 411 |
} |
| 412 |
|
| 413 |
$doc-><AS::Document.strictErrorChecking> (true); |
| 414 |
} |
| 415 |
} // DocumentEntity |
| 416 |
|
| 417 |
/* |
| 418 |
XML Declaration |
| 419 |
|
| 420 |
XMLDecl := '<?xml' VersionInfo |
| 421 |
[EncodingDecl] |
| 422 |
[SDDecl] |
| 423 |
[S] '?>' ;; [23] |
| 424 |
|
| 425 |
NOTE: XML declaration is optional in XML 1.0 |
| 426 |
while it is required in XML 1.1. |
| 427 |
*/ |
| 428 |
|
| 429 |
rule XMLDeclarationOpt ($doc) { |
| 430 |
~? (XDO) { |
| 431 |
&_XMLDeclaration ($doc => $doc); |
| 432 |
} |
| 433 |
} // XMLDeclarationOpt |
| 434 |
|
| 435 |
rule XMLDeclaration ($doc) { |
| 436 |
~ (XDO) { |
| 437 |
&_XMLDeclaration ($doc => $doc); |
| 438 |
} |
| 439 |
} // XMLDeclaration |
| 440 |
|
| 441 |
rule _XMLDeclaration ($doc) { |
| 442 |
?lexmode 'TAG'; |
| 443 |
|
| 444 |
// TODO: implement this |
| 445 |
|
| 446 |
} // _XMLDeclaration |
| 447 |
|
| 448 |
/* |
| 449 |
Document Type Declaration |
| 450 |
*/ |
| 451 |
rule _DocumentTypeDeclaration_ ($doc) { |
| 452 |
?lexmode 'DocumentTypeDeclaration'; |
| 453 |
|
| 454 |
~ (Name == 'DOCTYPE') { } |
| 455 |
|
| 456 |
~ (S) { } |
| 457 |
|
| 458 |
// Document type name |
| 459 |
~ (Name) { |
| 460 |
|
| 461 |
} |
| 462 |
|
| 463 |
// TODO: Implement this |
| 464 |
|
| 465 |
// ~ (MDC) { } |
| 466 |
} // _DocumentTypeDeclaration_ |
| 467 |
|
| 468 |
/* |
| 469 |
Comment Declaration |
| 470 |
|
| 471 |
Comment := '<!--' *(Char - '-' / '-' (Char - '-')) |
| 472 |
'-->' ;; [15] |
| 473 |
*/ |
| 474 |
rule _CommentDeclaration_ ($doc, $parent) { |
| 475 |
?lexmode 'CommentDeclaration'; |
| 476 |
|
| 477 |
~? (STRING) { |
| 478 |
lang:Perl ($data => $token.value) { |
| 479 |
my $com = $doc-><M::Document.createComment> ($data); |
| 480 |
$parent-><M::Node.appendChild> ($com); |
| 481 |
} |
| 482 |
} else { |
| 483 |
lang:Perl { |
| 484 |
my $com = $doc-><M::Document.createComment> (''); |
| 485 |
$parent-><M::Node.appendChild> ($com); |
| 486 |
} |
| 487 |
} |
| 488 |
|
| 489 |
~ (COM) { |
| 490 |
?lexmode MarkupDeclaration; |
| 491 |
} else { |
| 492 |
?lexmode MarkupDeclaration; |
| 493 |
} |
| 494 |
|
| 495 |
// ~ (MDC) { } |
| 496 |
} // _CommentDeclaration_ |
| 497 |
|
| 498 |
/* |
| 499 |
Processing Instruction |
| 500 |
|
| 501 |
PI := '<?' PITarget [S *Char - *Char '?>' *Char] |
| 502 |
'?>' ;; [16] |
| 503 |
*/ |
| 504 |
rule _ProcessingInstruction_ ($doc, $parent) { |
| 505 |
?lexmode 'PIName'; |
| 506 |
|
| 507 |
my $pi; |
| 508 |
|
| 509 |
~ (Name) { |
| 510 |
lang:Perl ($name => $token.value) { |
| 511 |
if (lc $name eq 'xml') { |
| 512 |
## TODO: Well-formedness (syntax) error |
| 513 |
} |
| 514 |
## TODO: Namespace well-formedness |
| 515 |
$pi = $doc-><M::Document.createProcessingInstruction> |
| 516 |
($name); |
| 517 |
} |
| 518 |
} |
| 519 |
|
| 520 |
~ (S) { |
| 521 |
?lexmode 'PIData'; |
| 522 |
|
| 523 |
my $tdata; |
| 524 |
|
| 525 |
~? (DATA) { |
| 526 |
lang:Perl ($data => $token.value) { |
| 527 |
$tdata = $data; |
| 528 |
} |
| 529 |
} else { |
| 530 |
lang:Perl { |
| 531 |
$tdata = ''; |
| 532 |
} |
| 533 |
} |
| 534 |
|
| 535 |
lang:Perl { |
| 536 |
$pi-><AS::Node.nodeValue> ($tdata); |
| 537 |
} |
| 538 |
} |
| 539 |
|
| 540 |
lang:Perl { |
| 541 |
$parent-><M::Node.appendChild> ($pi); |
| 542 |
## TODO: PIs in document type declaration subsets |
| 543 |
} |
| 544 |
|
| 545 |
// ~ (PIC) { } |
| 546 |
} // _ProcessingInstruction_ |
| 547 |
|
| 548 |
/* |
| 549 |
Element content parsing mode |
| 550 |
|
| 551 |
element := EmptyElemTag / |
| 552 |
STag content ETag ;; [39] |
| 553 |
content := (CharData / element / Reference / CDSect / |
| 554 |
PI / Comment) ;; [43] |
| 555 |
*/ |
| 556 |
rule Element_ ($doc, $parent) : standalone { |
| 557 |
?lexmode 'ElementContent'; |
| 558 |
|
| 559 |
my $node; // Current "parent" node |
| 560 |
my $nodes; // Node stack (w/o $current_node) |
| 561 |
my $type; // Current "parent" element type QName |
| 562 |
my $types; // Element type stack (w/o $current_type) |
| 563 |
my $ns; // Current in-scope namespace bindings |
| 564 |
my $nses; // Namespace binding stack (w/o $current_ns) |
| 565 |
|
| 566 |
lang:Perl { |
| 567 |
$node = $parent; |
| 568 |
$nodes = []; |
| 569 |
$type = ''; |
| 570 |
$types = []; |
| 571 |
$ns = { |
| 572 |
xml => <Q::xml:>, |
| 573 |
xmlns => <Q::xmlns:>, |
| 574 |
}; |
| 575 |
$nses = []; |
| 576 |
} |
| 577 |
|
| 578 |
~* : name => CONTENT |
| 579 |
(CharData) { |
| 580 |
// Character data |
| 581 |
lang:Perl ($data => $token.value) { |
| 582 |
if (index ($data, ']]>') > -1) { |
| 583 |
## TODO: Well-formedness (syntax) error |
| 584 |
} |
| 585 |
$node-><M::Node.appendChild> |
| 586 |
($doc-><M::Document.createTextNode> ($data)); |
| 587 |
} |
| 588 |
} (STAGO) { |
| 589 |
// Start tag or empty element tag |
| 590 |
|
| 591 |
?lexmode 'StartTag'; |
| 592 |
|
| 593 |
~ (Name) { |
| 594 |
my $attrs; |
| 595 |
lang:Perl ($name => $token.value) { |
| 596 |
push @{$types}, $type; |
| 597 |
$type = $name; |
| 598 |
$attrs = {}; |
| 599 |
} |
| 600 |
|
| 601 |
~? (S) { |
| 602 |
&AttributeSpecificationList |
| 603 |
($doc => $doc, $attrs => $attrs); |
| 604 |
} |
| 605 |
|
| 606 |
my $el; |
| 607 |
|
| 608 |
lang:Perl { |
| 609 |
push @{$nses}, $ns; |
| 610 |
$ns = {%$ns}; |
| 611 |
|
| 612 |
my %gattr; |
| 613 |
my %lattr; |
| 614 |
for my $atqname (keys %$attrs) { |
| 615 |
my ($pfx, $lname) = split /:/, $atqname; |
| 616 |
if (defined $lname) { ## Global attribute |
| 617 |
## TODO: Namespace well-formedness (lname is NCName) |
| 618 |
if ($pfx eq 'xmlns') { |
| 619 |
my $nsuri = $attrs->{$atqname}->{value}; |
| 620 |
if ($lname eq 'xml' and |
| 621 |
$nsuri ne <Q::xml:>) { |
| 622 |
## TODO: error |
| 623 |
} elsif ($lname eq 'xmlns') { |
| 624 |
## TODO: error |
| 625 |
} |
| 626 |
if ($nsuri eq '') { |
| 627 |
## TODO: error in XML 1.0 |
| 628 |
} elsif ($nsuri eq <Q::xml:> and |
| 629 |
$lname ne 'xml') { |
| 630 |
## TODO: error |
| 631 |
} elsif ($nsuri eq <Q::xmlns:>) { |
| 632 |
## TODO: error |
| 633 |
} |
| 634 |
$ns->{$lname} = $attrs->{$atqname}->{value}; |
| 635 |
delete $ns->{$lname} unless length $ns->{$lname}; |
| 636 |
} elsif ($pfx eq '') { |
| 637 |
## TODO: pfx is not NCName error |
| 638 |
} else { |
| 639 |
if ($gattr{$pfx}->{$lname}) { |
| 640 |
## TODO: Namespace well-formedness error |
| 641 |
} |
| 642 |
} |
| 643 |
$gattr{$pfx}->{$lname} = $attrs->{$atqname}; |
| 644 |
} else { ## Local attribute |
| 645 |
if ($pfx eq 'xmlns') { |
| 646 |
$ns->{''} = $attrs->{xmlns}->{value}; |
| 647 |
delete $ns->{''} unless length $ns->{''}; |
| 648 |
} else { |
| 649 |
$lattr{$pfx} = $attrs->{$atqname}; |
| 650 |
} |
| 651 |
} |
| 652 |
} |
| 653 |
|
| 654 |
my ($pfx, $lname) = split /:/, $type; |
| 655 |
my $nsuri; |
| 656 |
## TODO: lname is NCName? |
| 657 |
if (defined $lname) { ## Prefixed namespace |
| 658 |
if ($pfx eq '') { |
| 659 |
## TODO: pfx is not NCName error |
| 660 |
} |
| 661 |
if (defined $ns->{$pfx}) { |
| 662 |
$nsuri = $ns->{$pfx}; |
| 663 |
} else { |
| 664 |
## TODO: namespace ill-formed |
| 665 |
} |
| 666 |
} else { ## Default namespace |
| 667 |
$nsuri = $ns->{''}; |
| 668 |
} |
| 669 |
|
| 670 |
$el = $doc-><M::Document.createElementNS> |
| 671 |
($nsuri, $type); |
| 672 |
|
| 673 |
if ($attrs->{xmlns}) { |
| 674 |
my $attr = $doc-><M::Document.createAttributeNS> |
| 675 |
(<Q::xmlns:>, 'xmlns'); |
| 676 |
for (@{$attrs->{xmlns}->{nodes}}) { |
| 677 |
$attr-><M::Node.appendChild> ($_); |
| 678 |
} |
| 679 |
$el-><M::Element.setAttributeNodeNS> ($attr); |
| 680 |
} |
| 681 |
|
| 682 |
for my $lname (keys %lattr) { |
| 683 |
my $attr = $doc-><M::Document.createAttributeNS> |
| 684 |
(null, $lname); |
| 685 |
for (@{$lattr{$lname}->{nodes}}) { |
| 686 |
$attr-><M::Node.appendChild> ($_); |
| 687 |
} |
| 688 |
$el-><M::Element.setAttributeNodeNS> ($attr); |
| 689 |
} |
| 690 |
|
| 691 |
for my $pfx (keys %gattr) { |
| 692 |
for my $lname (keys %{$gattr{$pfx}}) { |
| 693 |
my $attr = $doc-><M::Document.createAttributeNS> |
| 694 |
($ns->{$pfx}, $pfx.':'.$lname); |
| 695 |
for (@{$gattr{$pfx}->{$lname}->{nodes}}) { |
| 696 |
$attr-><M::Node.appendChild> ($_); |
| 697 |
} |
| 698 |
$el-><M::Element.setAttributeNodeNS> ($attr); |
| 699 |
} |
| 700 |
} |
| 701 |
|
| 702 |
$node-><M::Node.appendChild> ($el); |
| 703 |
} |
| 704 |
|
| 705 |
~ (TAGC) { |
| 706 |
lang:Perl { |
| 707 |
push @{$nodes}, $node; |
| 708 |
$node = $el; |
| 709 |
} |
| 710 |
?lexmode ElementContent; |
| 711 |
} (MTAGC) { |
| 712 |
lang:Perl { |
| 713 |
$ns = pop @{$nses}; |
| 714 |
$type = pop @{$types}; |
| 715 |
} |
| 716 |
?lexmode ElementContent; |
| 717 |
} else { |
| 718 |
?lexmode ElementContent; |
| 719 |
} |
| 720 |
} else { |
| 721 |
?lexmode ElementContent; |
| 722 |
} |
| 723 |
|
| 724 |
} (ETAGO) { |
| 725 |
// End tag |
| 726 |
|
| 727 |
?lexmode 'EndTag'; |
| 728 |
|
| 729 |
my $is_docel; |
| 730 |
|
| 731 |
~ (Name) { |
| 732 |
lang:Perl ($name => $token.value) { |
| 733 |
if ($name eq $type) { |
| 734 |
$type = pop @{$types}; |
| 735 |
if ($type eq '') { |
| 736 |
$is_docel = true; |
| 737 |
} |
| 738 |
$node = pop @{$nodes}; |
| 739 |
$ns = pop @{$nses}; |
| 740 |
} else { |
| 741 |
## TODO: Element type match well-formedness error |
| 742 |
} |
| 743 |
} |
| 744 |
} |
| 745 |
|
| 746 |
~? (S) { } |
| 747 |
|
| 748 |
if-true ($is_docel) { |
| 749 |
lang:Perl { |
| 750 |
if (@{$types}) { |
| 751 |
## WF error |
| 752 |
} |
| 753 |
} |
| 754 |
return; |
| 755 |
} |
| 756 |
|
| 757 |
~ (TAGC) { |
| 758 |
?lexmode ElementContent; |
| 759 |
} else { |
| 760 |
?lexmode 'ElementContent'; |
| 761 |
} |
| 762 |
|
| 763 |
} (HCRO) { |
| 764 |
&_HexadecimalCharacterReference_ |
| 765 |
($doc => $doc, $parent => $node); |
| 766 |
|
| 767 |
~ (REFC) { |
| 768 |
?lexmode 'ElementContent'; |
| 769 |
} else { |
| 770 |
?lexmode ElementContent; |
| 771 |
} |
| 772 |
} (CRO) { |
| 773 |
&_NumericCharacterReference_ |
| 774 |
($doc => $doc, $parent => $node); |
| 775 |
|
| 776 |
~ (REFC) { |
| 777 |
?lexmode 'ElementContent'; |
| 778 |
} else { |
| 779 |
?lexmode ElementContent; |
| 780 |
} |
| 781 |
} (ERO) { |
| 782 |
&_GeneralEntityReference_ |
| 783 |
($doc => $doc, $parent => $node); |
| 784 |
|
| 785 |
~ (REFC) { |
| 786 |
?lexmode 'ElementContent'; |
| 787 |
} else { |
| 788 |
?lexmode ElementContent; |
| 789 |
} |
| 790 |
} (CDO) { |
| 791 |
&_CommentDeclaration_ ($doc => $doc, $parent => $node); |
| 792 |
|
| 793 |
~ (MDC) { |
| 794 |
?lexmode ElementContent; |
| 795 |
} else { |
| 796 |
?lexmode ElementContent; |
| 797 |
} |
| 798 |
} (CDSO) { |
| 799 |
&_CDATASection_ ($doc => $doc, $parent => $node); |
| 800 |
|
| 801 |
~ (MSE) { |
| 802 |
?lexmode 'ElementContent'; |
| 803 |
} else { |
| 804 |
?lexmode ElementContent; |
| 805 |
} |
| 806 |
} (PIO) { |
| 807 |
&_ProcessingInstruction_ ($doc => $doc, $parent => $node); |
| 808 |
|
| 809 |
~ (PIC) { |
| 810 |
?lexmode 'ElementContent'; |
| 811 |
} else { |
| 812 |
?lexmode ElementContent; |
| 813 |
} |
| 814 |
} |
| 815 |
} // Element_ |
| 816 |
|
| 817 |
rule AttributeSpecificationList ($doc, $attrs) |
| 818 |
: standalone |
| 819 |
{ |
| 820 |
?lexmode 'StartTag'; |
| 821 |
|
| 822 |
my $i; |
| 823 |
lang:Perl { |
| 824 |
$i = 0; |
| 825 |
} |
| 826 |
|
| 827 |
~* (Name) { |
| 828 |
my $atqname; |
| 829 |
lang:Perl ($name => $token.value) { |
| 830 |
$atqname = $name; |
| 831 |
} |
| 832 |
|
| 833 |
~? (S) { } |
| 834 |
~ (VI) { } |
| 835 |
~? (S) { } |
| 836 |
|
| 837 |
my $vals; |
| 838 |
lang:Perl { |
| 839 |
if ($attrs->{$atqname}) { |
| 840 |
## TODO: Unique attr name well-formedness error |
| 841 |
} |
| 842 |
|
| 843 |
$vals = $attrs->{$atqname} = { |
| 844 |
nodes => [], |
| 845 |
value => '', |
| 846 |
index => $i++, |
| 847 |
}; |
| 848 |
} |
| 849 |
|
| 850 |
~ (LIT) { |
| 851 |
&_AttributeValueSpecification_ |
| 852 |
($doc => $doc, $vals => $vals); |
| 853 |
|
| 854 |
~ (LIT) { |
| 855 |
?lexmode StartTag; |
| 856 |
} else { |
| 857 |
?lexmode StartTag; |
| 858 |
} |
| 859 |
} (LITA) { |
| 860 |
&_AttributeValueSpecificationA_ |
| 861 |
($doc => $doc, $vals => $vals); |
| 862 |
|
| 863 |
~ (LITA) { |
| 864 |
?lexmode StartTag; |
| 865 |
} else { |
| 866 |
?lexmode StartTag; |
| 867 |
} |
| 868 |
} |
| 869 |
} (S) : separator : terminator? { } |
| 870 |
} // AttributeSpecificationList |
| 871 |
|
| 872 |
rule _AttributeValueSpecification_ ($doc, $vals) { |
| 873 |
// ~ (LIT) { } |
| 874 |
?lexmode 'AttributeValueLiteral'; |
| 875 |
|
| 876 |
~* (STRING) { |
| 877 |
lang:Perl ($value => $token.value) { |
| 878 |
$value =~ s/[\x09\x0A\x0D]/ /g; |
| 879 |
my $text = $doc-><M::Document.createTextNode> ($value); |
| 880 |
push @{$vals->{nodes}}, $text; |
| 881 |
$vals->{value} .= $value; |
| 882 |
} |
| 883 |
} (HCRO) { |
| 884 |
&_HexadecimalCharacterReferenceV_ |
| 885 |
($doc => $doc, $vals => $vals); |
| 886 |
|
| 887 |
~ (REFC) { |
| 888 |
?lexmode AttributeValueLiteral; |
| 889 |
} else { |
| 890 |
?lexmode AttributeValueLiteral; |
| 891 |
} |
| 892 |
} (CRO) { |
| 893 |
&_NumericCharacterReferenceV_ |
| 894 |
($doc => $doc, $vals => $vals); |
| 895 |
|
| 896 |
~ (REFC) { |
| 897 |
?lexmode AttributeValueLiteral; |
| 898 |
} else { |
| 899 |
?lexmode AttributeValueLiteral; |
| 900 |
} |
| 901 |
} (ERO) { |
| 902 |
// TODO: Attribute value normalization |
| 903 |
&_GeneralEntityReferenceV_ |
| 904 |
($doc => $doc, $vals => $vals); |
| 905 |
|
| 906 |
~ (REFC) { |
| 907 |
?lexmode AttributeValueLiteral; |
| 908 |
} else { |
| 909 |
?lexmode AttributeValueLiteral; |
| 910 |
} |
| 911 |
} |
| 912 |
|
| 913 |
// ~ (LIT) { } (LITA) { } |
| 914 |
} // _AttributeValueSpecification_ |
| 915 |
|
| 916 |
rule _AttributeValueSpecificationA_ ($doc, $vals) { |
| 917 |
// ~ (LITA) { } |
| 918 |
?lexmode 'AttributeValueLiteralA'; |
| 919 |
|
| 920 |
~* (STRING) { |
| 921 |
lang:Perl ($value => $token.value) { |
| 922 |
$value =~ s/[\x09\x0A\x0D]/ /g; |
| 923 |
my $text = $doc-><M::Document.createTextNode> ($value); |
| 924 |
push @{$vals->{nodes}}, $text; |
| 925 |
$vals->{value} .= $value; |
| 926 |
} |
| 927 |
} (HCRO) { |
| 928 |
&_HexadecimalCharacterReferenceV_ |
| 929 |
($doc => $doc, $vals => $vals); |
| 930 |
|
| 931 |
~ (REFC) { |
| 932 |
?lexmode AttributeValueLiteralA; |
| 933 |
} else { |
| 934 |
?lexmode AttributeValueLiteralA; |
| 935 |
} |
| 936 |
} (CRO) { |
| 937 |
&_NumericCharacterReferenceV_ |
| 938 |
($doc => $doc, $vals => $vals); |
| 939 |
|
| 940 |
~ (REFC) { |
| 941 |
?lexmode AttributeValueLiteralA; |
| 942 |
} else { |
| 943 |
?lexmode AttributeValueLiteralA; |
| 944 |
} |
| 945 |
} (ERO) { |
| 946 |
// TODO: Attribute value normalization |
| 947 |
&_GeneralEntityReferenceV_ |
| 948 |
($doc => $doc, $vals => $vals); |
| 949 |
|
| 950 |
~ (REFC) { |
| 951 |
?lexmode AttributeValueLiteralA; |
| 952 |
} else { |
| 953 |
?lexmode AttributeValueLiteralA; |
| 954 |
} |
| 955 |
} |
| 956 |
|
| 957 |
// ~ (LITA) { } |
| 958 |
} // _AttributeValueSpecificationA_ |
| 959 |
|
| 960 |
/* |
| 961 |
CDATA Section Content Parsing Mode |
| 962 |
*/ |
| 963 |
rule _CDATASection_ ($doc, $parent) { |
| 964 |
?lexmode 'CDATASectionContent'; |
| 965 |
|
| 966 |
my $cdata; |
| 967 |
|
| 968 |
~? (CData) { |
| 969 |
lang:Perl ($data => $token.value) { |
| 970 |
$cdata = $data; |
| 971 |
} |
| 972 |
} else { |
| 973 |
lang:Perl { |
| 974 |
$cdata = ''; |
| 975 |
} |
| 976 |
} |
| 977 |
|
| 978 |
lang:Perl { |
| 979 |
my $cdsect = $doc-><M::Document.createCDATASection> |
| 980 |
($cdata); |
| 981 |
$parent-><M::Node.appendChild> ($cdsect); |
| 982 |
} |
| 983 |
|
| 984 |
// ~ (MSE) { } |
| 985 |
} // _CDATASection_ |
| 986 |
|
| 987 |
rule _NumericCharacterReference_ ($doc, $parent) { |
| 988 |
?lexmode 'NumericCharacterReference'; |
| 989 |
|
| 990 |
~ (NUMBER) { |
| 991 |
lang:Perl ($num => $token.value) { |
| 992 |
## TODO: [WFC: Lecal Character] |
| 993 |
my $ncr = $doc-><M::Document.createTextNode> |
| 994 |
(chr (0+$num)); |
| 995 |
$parent-><M::Node.appendChild> ($ncr); |
| 996 |
} |
| 997 |
} |
| 998 |
|
| 999 |
// ~ (REFC) { } |
| 1000 |
} // _NumericCharacterReference_ |
| 1001 |
|
| 1002 |
rule _NumericCharacterReferenceV_ ($doc, $vals) { |
| 1003 |
?lexmode 'NumericCharacterReference'; |
| 1004 |
|
| 1005 |
~ (NUMBER) { |
| 1006 |
lang:Perl ($num => $token.value) { |
| 1007 |
## TODO: [WFC: Lecal Character] |
| 1008 |
my $ncr = $doc-><M::Document.createTextNode> |
| 1009 |
(my $char = chr (0+$num)); |
| 1010 |
push @{$vals->{nodes}}, $ncr; |
| 1011 |
$vals->{value} .= $char; |
| 1012 |
} |
| 1013 |
} |
| 1014 |
|
| 1015 |
// ~ (REFC) { } |
| 1016 |
} // _NumericCharacterReferenceV_ |
| 1017 |
|
| 1018 |
rule _HexadecimalCharacterReference_ ($doc, $parent) { |
| 1019 |
?lexmode 'HexadecimalCharacterReference'; |
| 1020 |
|
| 1021 |
~ (Hex) { |
| 1022 |
lang:Perl ($num => $token.value) { |
| 1023 |
## TODO: [WFC: Lecal Character] |
| 1024 |
my $ncr = $doc-><M::Document.createTextNode> |
| 1025 |
(chr hex $num); |
| 1026 |
$parent-><M::Node.appendChild> ($ncr); |
| 1027 |
} |
| 1028 |
} |
| 1029 |
|
| 1030 |
// ~ (REFC) { } |
| 1031 |
} // _HexadecimalCharacterReference_ |
| 1032 |
|
| 1033 |
rule _HexadecimalCharacterReferenceV_ ($doc, $parent) { |
| 1034 |
?lexmode 'HexadecimalCharacterReference'; |
| 1035 |
|
| 1036 |
~ (Hex) { |
| 1037 |
lang:Perl ($num => $token.value) { |
| 1038 |
## TODO: [WFC: Lecal Character] |
| 1039 |
my $ncr = $doc-><M::Document.createTextNode> |
| 1040 |
(my $char = chr hex $num); |
| 1041 |
push @{$vals->{nodes}}, $ncr; |
| 1042 |
$vals->{value} .= $char; |
| 1043 |
} |
| 1044 |
} |
| 1045 |
|
| 1046 |
// ~ (REFC) { } |
| 1047 |
} // _HexadecimalCharacterReferenceV_ |
| 1048 |
|
| 1049 |
rule _GeneralEntityReference_ ($doc, $parent) { |
| 1050 |
// TODO: Expansion |
| 1051 |
?lexmode 'EntityReference'; |
| 1052 |
|
| 1053 |
~ (Name) { |
| 1054 |
lang:Perl ($name => $token.value) { |
| 1055 |
## TODO: Namespace well-formedness |
| 1056 |
## TODO: Entity declared constraints |
| 1057 |
my $er = $doc-><M::Document.createEntityReference> |
| 1058 |
($name); |
| 1059 |
$parent-><M::Node.appendChild> ($er); |
| 1060 |
} |
| 1061 |
} |
| 1062 |
|
| 1063 |
// ~ (REFC) { } |
| 1064 |
} // _GeneralEntityReference_ |
| 1065 |
|
| 1066 |
rule _GeneralEntityReferenceV_ ($doc, $vals) { |
| 1067 |
// TODO: Expansion |
| 1068 |
?lexmode 'EntityReference'; |
| 1069 |
|
| 1070 |
~ (Name) { |
| 1071 |
lang:Perl ($name => $token.value) { |
| 1072 |
## TODO: Namespace well-formedness |
| 1073 |
## TODO: Entity declared constraints |
| 1074 |
my $er = $doc-><M::Document.createEntityReference> |
| 1075 |
($name); |
| 1076 |
push @{$vals->{nodes}}, $er; |
| 1077 |
} |
| 1078 |
} |
| 1079 |
|
| 1080 |
// ~ (REFC) { } |
| 1081 |
} // _GeneralEntityReferenceV_ |
| 1082 |
|
| 1083 |
|
| 1084 |
/* |
| 1085 |
XML Name |
| 1086 |
*/ |
| 1087 |
lexmode Name { |
| 1088 |
$NameStartChar10 := [ |
| 1089 |
'_' ':' |
| 1090 |
// Letter |
| 1091 |
// BaseChar |
| 1092 |
U+0041..U+005A U+0061..U+007A U+00C0..U+00D6 |
| 1093 |
U+00D8..U+00F6 U+00F8..U+00FF U+0100..U+0131 |
| 1094 |
U+0134..U+013E U+0141..U+0148 U+014A..U+017E |
| 1095 |
U+0180..U+01C3 U+01CD..U+01F0 U+01F4..U+01F5 |
| 1096 |
U+01FA..U+0217 U+0250..U+02A8 U+02BB..U+02C1 |
| 1097 |
U+0386 U+0388..U+038A U+038C U+038E..U+03A1 |
| 1098 |
U+03A3..U+03CE U+03D0..U+03D6 U+03DA U+03DC |
| 1099 |
U+03DE U+03E0 U+03E2..U+03F3 U+0401..U+040C |
| 1100 |
U+040E..U+044F U+0451..U+045C U+045E..U+0481 |
| 1101 |
U+0490..U+04C4 U+04C7..U+04C8 U+04CB..U+04CC |
| 1102 |
U+04D0..U+04EB U+04EE..U+04F5 U+04F8..U+04F9 |
| 1103 |
U+0531..U+0556 U+0559 U+0561..U+0586 |
| 1104 |
U+05D0..U+05EA U+05F0..U+05F2 U+0621..U+063A |
| 1105 |
U+0641..U+064A U+0671..U+06B7 U+06BA..U+06BE |
| 1106 |
U+06C0..U+06CE U+06D0..U+06D3 U+06D5 |
| 1107 |
U+06E5..U+06E6 U+0905..U+0939 U+093D |
| 1108 |
U+0958..U+0961 U+0985..U+098C U+098F..U+0990 |
| 1109 |
U+0993..U+09A8 U+09AA..U+09B0 U+09B2 |
| 1110 |
U+09B6..U+09B9 U+09DC..U+09DD U+09DF..U+09E1 |
| 1111 |
U+09F0..U+09F1 U+0A05..U+0A0A U+0A0F..U+0A10 |
| 1112 |
U+0A13..U+0A28 U+0A2A..U+0A30 U+0A32..U+0A33 |
| 1113 |
U+0A35..U+0A36 U+0A38..U+0A39 U+0A59..U+0A5C |
| 1114 |
U+0A5E U+0A72..U+0A74 U+0A85..U+0A8B U+0A8D |
| 1115 |
U+0A8F..U+0A91 U+0A93..U+0AA8 U+0AAA..U+0AB0 |
| 1116 |
U+0AB2..U+0AB3 U+0AB5..U+0AB9 U+0ABD U+0AE0 |
| 1117 |
U+0B05..U+0B0C U+0B0F..U+0B10 U+0B13..U+0B28 |
| 1118 |
U+0B2A..U+0B30 U+0B32..U+0B33 U+0B36..U+0B39 |
| 1119 |
U+0B3D U+0B5C..U+0B5D U+0B5F..U+0B61 |
| 1120 |
U+0B85..U+0B8A U+0B8E..U+0B90 U+0B92..U+0B95 |
| 1121 |
U+0B99..U+0B9A U+0B9C U+0B9E..U+0B9F |
| 1122 |
U+0BA3..U+0BA4 U+0BA8..U+0BAA U+0BAE..U+0BB5 |
| 1123 |
U+0BB7..U+0BB9 U+0C05..U+0C0C U+0C0E..U+0C10 |
| 1124 |
U+0C12..U+0C28 U+0C2A..U+0C33 U+0C35..U+0C39 |
| 1125 |
U+0C60..U+0C61 U+0C85..U+0C8C U+0C8E..U+0C90 |
| 1126 |
U+0C92..U+0CA8 U+0CAA..U+0CB3 U+0CB5..U+0CB9 |
| 1127 |
U+0CDE U+0CE0..U+0CE1 U+0D05..U+0D0C |
| 1128 |
U+0D0E..U+0D10 U+0D12..U+0D28 U+0D2A..U+0D39 |
| 1129 |
U+0D60..U+0D61 U+0E01..U+0E2E U+0E30 |
| 1130 |
U+0E32..U+0E33 U+0E40..U+0E45 U+0E81..U+0E82 |
| 1131 |
U+0E84 U+0E87..U+0E88 U+0E8A U+0E8D |
| 1132 |
U+0E94..U+0E97 U+0E99..U+0E9F U+0EA1..U+0EA3 |
| 1133 |
U+0EA5 U+0EA7 U+0EAA..U+0EAB U+0EAD..U+0EAE |
| 1134 |
U+0EB0 U+0EB2..U+0EB3 U+0EBD U+0EC0..U+0EC4 |
| 1135 |
U+0F40..U+0F47 U+0F49..U+0F69 U+10A0..U+10C5 |
| 1136 |
U+10D0..U+10F6 U+1100 U+1102..U+1103 |
| 1137 |
U+1105..U+1107 U+1109 U+110B..U+110C |
| 1138 |
U+110E..U+1112 U+113C U+113E U+1140 U+114C |
| 1139 |
U+114E U+1150 U+1154..U+1155 U+1159 |
| 1140 |
U+115F..U+1161 U+1163 U+1165 U+1167 U+1169 |
| 1141 |
U+116D..U+116E U+1172..U+1173 U+1175 U+119E |
| 1142 |
U+11A8 U+11AB U+11AE..U+11AF U+11B7..U+11B8 |
| 1143 |
U+11BA U+11BC..U+11C2 U+11EB U+11F0 U+11F9 |
| 1144 |
U+1E00..U+1E9B U+1EA0..U+1EF9 U+1F00..U+1F15 |
| 1145 |
U+1F18..U+1F1D U+1F20..U+1F45 U+1F48..U+1F4D |
| 1146 |
U+1F50..U+1F57 U+1F59 U+1F5B U+1F5D |
| 1147 |
U+1F5F..U+1F7D U+1F80..U+1FB4 U+1FB6..U+1FBC |
| 1148 |
U+1FBE U+1FC2..U+1FC4 U+1FC6..U+1FCC |
| 1149 |
U+1FD0..U+1FD3 U+1FD6..U+1FDB U+1FE0..U+1FEC |
| 1150 |
U+1FF2..U+1FF4 U+1FF6..U+1FFC U+2126 |
| 1151 |
U+212A..U+212B U+212E U+2180..U+2182 |
| 1152 |
U+3041..U+3094 U+30A1..U+30FA U+3105..U+312C |
| 1153 |
U+AC00..U+D7A3 |
| 1154 |
// Ideographic |
| 1155 |
U+4E00..U+9FA5 U+3007 U+3021..U+3029 |
| 1156 |
]; |
| 1157 |
$NameChar10 := [ |
| 1158 |
'.' '-' '_' ':' |
| 1159 |
// Letter |
| 1160 |
// BaseChar |
| 1161 |
U+0041..U+005A U+0061..U+007A U+00C0..U+00D6 |
| 1162 |
U+00D8..U+00F6 U+00F8..U+00FF U+0100..U+0131 |
| 1163 |
U+0134..U+013E U+0141..U+0148 U+014A..U+017E |
| 1164 |
U+0180..U+01C3 U+01CD..U+01F0 U+01F4..U+01F5 |
| 1165 |
U+01FA..U+0217 U+0250..U+02A8 U+02BB..U+02C1 |
| 1166 |
U+0386 U+0388..U+038A U+038C U+038E..U+03A1 |
| 1167 |
U+03A3..U+03CE U+03D0..U+03D6 U+03DA U+03DC |
| 1168 |
U+03DE U+03E0 U+03E2..U+03F3 U+0401..U+040C |
| 1169 |
U+040E..U+044F U+0451..U+045C U+045E..U+0481 |
| 1170 |
U+0490..U+04C4 U+04C7..U+04C8 U+04CB..U+04CC |
| 1171 |
U+04D0..U+04EB U+04EE..U+04F5 U+04F8..U+04F9 |
| 1172 |
U+0531..U+0556 U+0559 U+0561..U+0586 |
| 1173 |
U+05D0..U+05EA U+05F0..U+05F2 U+0621..U+063A |
| 1174 |
U+0641..U+064A U+0671..U+06B7 U+06BA..U+06BE |
| 1175 |
U+06C0..U+06CE U+06D0..U+06D3 U+06D5 |
| 1176 |
U+06E5..U+06E6 U+0905..U+0939 U+093D |
| 1177 |
U+0958..U+0961 U+0985..U+098C U+098F..U+0990 |
| 1178 |
U+0993..U+09A8 U+09AA..U+09B0 U+09B2 |
| 1179 |
U+09B6..U+09B9 U+09DC..U+09DD U+09DF..U+09E1 |
| 1180 |
U+09F0..U+09F1 U+0A05..U+0A0A U+0A0F..U+0A10 |
| 1181 |
U+0A13..U+0A28 U+0A2A..U+0A30 U+0A32..U+0A33 |
| 1182 |
U+0A35..U+0A36 U+0A38..U+0A39 U+0A59..U+0A5C |
| 1183 |
U+0A5E U+0A72..U+0A74 U+0A85..U+0A8B U+0A8D |
| 1184 |
U+0A8F..U+0A91 U+0A93..U+0AA8 U+0AAA..U+0AB0 |
| 1185 |
U+0AB2..U+0AB3 U+0AB5..U+0AB9 U+0ABD U+0AE0 |
| 1186 |
U+0B05..U+0B0C U+0B0F..U+0B10 U+0B13..U+0B28 |
| 1187 |
U+0B2A..U+0B30 U+0B32..U+0B33 U+0B36..U+0B39 |
| 1188 |
U+0B3D U+0B5C..U+0B5D U+0B5F..U+0B61 |
| 1189 |
U+0B85..U+0B8A U+0B8E..U+0B90 U+0B92..U+0B95 |
| 1190 |
U+0B99..U+0B9A U+0B9C U+0B9E..U+0B9F |
| 1191 |
U+0BA3..U+0BA4 U+0BA8..U+0BAA U+0BAE..U+0BB5 |
| 1192 |
U+0BB7..U+0BB9 U+0C05..U+0C0C U+0C0E..U+0C10 |
| 1193 |
U+0C12..U+0C28 U+0C2A..U+0C33 U+0C35..U+0C39 |
| 1194 |
U+0C60..U+0C61 U+0C85..U+0C8C U+0C8E..U+0C90 |
| 1195 |
U+0C92..U+0CA8 U+0CAA..U+0CB3 U+0CB5..U+0CB9 |
| 1196 |
U+0CDE U+0CE0..U+0CE1 U+0D05..U+0D0C |
| 1197 |
U+0D0E..U+0D10 U+0D12..U+0D28 U+0D2A..U+0D39 |
| 1198 |
U+0D60..U+0D61 U+0E01..U+0E2E U+0E30 |
| 1199 |
U+0E32..U+0E33 U+0E40..U+0E45 U+0E81..U+0E82 |
| 1200 |
U+0E84 U+0E87..U+0E88 U+0E8A U+0E8D |
| 1201 |
U+0E94..U+0E97 U+0E99..U+0E9F U+0EA1..U+0EA3 |
| 1202 |
U+0EA5 U+0EA7 U+0EAA..U+0EAB U+0EAD..U+0EAE |
| 1203 |
U+0EB0 U+0EB2..U+0EB3 U+0EBD U+0EC0..U+0EC4 |
| 1204 |
U+0F40..U+0F47 U+0F49..U+0F69 U+10A0..U+10C5 |
| 1205 |
U+10D0..U+10F6 U+1100 U+1102..U+1103 |
| 1206 |
U+1105..U+1107 U+1109 U+110B..U+110C |
| 1207 |
U+110E..U+1112 U+113C U+113E U+1140 U+114C |
| 1208 |
U+114E U+1150 U+1154..U+1155 U+1159 |
| 1209 |
U+115F..U+1161 U+1163 U+1165 U+1167 U+1169 |
| 1210 |
U+116D..U+116E U+1172..U+1173 U+1175 U+119E |
| 1211 |
U+11A8 U+11AB U+11AE..U+11AF U+11B7..U+11B8 |
| 1212 |
U+11BA U+11BC..U+11C2 U+11EB U+11F0 U+11F9 |
| 1213 |
U+1E00..U+1E9B U+1EA0..U+1EF9 U+1F00..U+1F15 |
| 1214 |
U+1F18..U+1F1D U+1F20..U+1F45 U+1F48..U+1F4D |
| 1215 |
U+1F50..U+1F57 U+1F59 U+1F5B U+1F5D |
| 1216 |
U+1F5F..U+1F7D U+1F80..U+1FB4 U+1FB6..U+1FBC |
| 1217 |
U+1FBE U+1FC2..U+1FC4 U+1FC6..U+1FCC |
| 1218 |
U+1FD0..U+1FD3 U+1FD6..U+1FDB U+1FE0..U+1FEC |
| 1219 |
U+1FF2..U+1FF4 U+1FF6..U+1FFC U+2126 |
| 1220 |
U+212A..U+212B U+212E U+2180..U+2182 |
| 1221 |
U+3041..U+3094 U+30A1..U+30FA U+3105..U+312C |
| 1222 |
U+AC00..U+D7A3 |
| 1223 |
// Ideographic |
| 1224 |
U+4E00..U+9FA5 U+3007 U+3021..U+3029 |
| 1225 |
// Digit |
| 1226 |
U+0030..U+0039 U+0660..U+0669 U+06F0..U+06F9 |
| 1227 |
U+0966..U+096F U+09E6..U+09EF U+0A66..U+0A6F |
| 1228 |
U+0AE6..U+0AEF U+0B66..U+0B6F U+0BE7..U+0BEF |
| 1229 |
U+0C66..U+0C6F U+0CE6..U+0CEF U+0D66..U+0D6F |
| 1230 |
U+0E50..U+0E59 U+0ED0..U+0ED9 U+0F20..U+0F29 |
| 1231 |
// CombiningChar |
| 1232 |
U+0300..U+0345 U+0360..U+0361 U+0483..U+0486 |
| 1233 |
U+0591..U+05A1 U+05A3..U+05B9 U+05BB..U+05BD |
| 1234 |
U+05BF U+05C1..U+05C2 U+05C4 U+064B..U+0652 |
| 1235 |
U+0670 U+06D6..U+06DC U+06DD..U+06DF |
| 1236 |
U+06E0..U+06E4 U+06E7..U+06E8 U+06EA..U+06ED |
| 1237 |
U+0901..U+0903 U+093C U+093E..U+094C U+094D |
| 1238 |
U+0951..U+0954 U+0962..U+0963 U+0981..U+0983 |
| 1239 |
U+09BC U+09BE U+09BF U+09C0..U+09C4 |
| 1240 |
U+09C7..U+09C8 U+09CB..U+09CD U+09D7 |
| 1241 |
U+09E2..U+09E3 U+0A02 U+0A3C U+0A3E U+0A3F |
| 1242 |
U+0A40..U+0A42 U+0A47..U+0A48 U+0A4B..U+0A4D |
| 1243 |
U+0A70..U+0A71 U+0A81..U+0A83 U+0ABC |
| 1244 |
U+0ABE..U+0AC5 U+0AC7..U+0AC9 U+0ACB..U+0ACD |
| 1245 |
U+0B01..U+0B03 U+0B3C U+0B3E..U+0B43 |
| 1246 |
U+0B47..U+0B48 U+0B4B..U+0B4D U+0B56..U+0B57 |
| 1247 |
U+0B82..U+0B83 U+0BBE..U+0BC2 U+0BC6..U+0BC8 |
| 1248 |
U+0BCA..U+0BCD U+0BD7 U+0C01..U+0C03 |
| 1249 |
U+0C3E..U+0C44 U+0C46..U+0C48 U+0C4A..U+0C4D |
| 1250 |
U+0C55..U+0C56 U+0C82..U+0C83 U+0CBE..U+0CC4 |
| 1251 |
U+0CC6..U+0CC8 U+0CCA..U+0CCD U+0CD5..U+0CD6 |
| 1252 |
U+0D02..U+0D03 U+0D3E..U+0D43 U+0D46..U+0D48 |
| 1253 |
U+0D4A..U+0D4D U+0D57 U+0E31 U+0E34..U+0E3A |
| 1254 |
U+0E47..U+0E4E U+0EB1 U+0EB4..U+0EB9 |
| 1255 |
U+0EBB..U+0EBC U+0EC8..U+0ECD U+0F18..U+0F19 |
| 1256 |
U+0F35 U+0F37 U+0F39 U+0F3E U+0F3F |
| 1257 |
U+0F71..U+0F84 U+0F86..U+0F8B U+0F90..U+0F95 |
| 1258 |
U+0F97 U+0F99..U+0FAD U+0FB1..U+0FB7 U+0FB9 |
| 1259 |
U+20D0..U+20DC U+20E1 U+302A..U+302F U+3099 |
| 1260 |
U+309A |
| 1261 |
// Extender |
| 1262 |
U+00B7 U+02D0 U+02D1 U+0387 U+0640 U+0E46 |
| 1263 |
U+0EC6 U+3005 U+3031..U+3035 U+309D..U+309E |
| 1264 |
U+30FC..U+30FE |
| 1265 |
]; |
| 1266 |
|
| 1267 |
$NameStartChar11 := [ |
| 1268 |
':' '_' |
| 1269 |
'A' 'B' 'C' 'D' 'E' 'F' 'G' 'H' 'I' 'J' 'K' 'L' 'M' |
| 1270 |
'N' 'O' 'P' 'Q' 'R' 'S' 'T' 'U' 'V' 'W' 'X' 'Y' 'Z' |
| 1271 |
'a' 'b' 'c' 'd' 'e' 'f' 'g' 'h' 'i' 'j' 'k' 'l' 'm' |
| 1272 |
'n' 'o' 'p' 'q' 'r' 's' 't' 'u' 'v' 'w' 'x' 'y' 'z' |
| 1273 |
U+00C0..U+00D6 U+00D8..U+00F6 U+00F8..U+02FF |
| 1274 |
U+0370..U+037D U+037F..U+1FFF U+200C..U+200D |
| 1275 |
U+2070..U+218F U+2C00..U+2FEF U+3001..U+D7FF |
| 1276 |
U+F900..U+FDCF U+FDF0..U+FFFD U+10000..U+EFFFF |
| 1277 |
]; |
| 1278 |
$NameChar11 := [ |
| 1279 |
'-' '.' '0' '1' '2' '3' '4' '5' '6' '7' '8' '9' |
| 1280 |
U+00B7 U+0300..U+036F U+203F..U+2040 |
| 1281 |
// NameStartChar |
| 1282 |
':' '_' |
| 1283 |
'A' 'B' 'C' 'D' 'E' 'F' 'G' 'H' 'I' 'J' 'K' 'L' 'M' |
| 1284 |
'N' 'O' 'P' 'Q' 'R' 'S' 'T' 'U' 'V' 'W' 'X' 'Y' 'Z' |
| 1285 |
'a' 'b' 'c' 'd' 'e' 'f' 'g' 'h' 'i' 'j' 'k' 'l' 'm' |
| 1286 |
'n' 'o' 'p' 'q' 'r' 's' 't' 'u' 'v' 'w' 'x' 'y' 'z' |
| 1287 |
U+00C0..U+00D6 U+00D8..U+00F6 U+00F8..U+02FF |
| 1288 |
U+0370..U+037D U+037F..U+1FFF U+200C..U+200D |
| 1289 |
U+2070..U+218F U+2C00..U+2FEF U+3001..U+D7FF |
| 1290 |
U+F900..U+FDCF U+FDF0..U+FFFD U+10000..U+EFFFF |
| 1291 |
]; |
| 1292 |
Name : value := $NameStartChar11 $NameChar11*; |
| 1293 |
} // Name |
| 1294 |
|
| 1295 |
/* |
| 1296 |
Space |
| 1297 |
*/ |
| 1298 |
lexmode S { |
| 1299 |
S := [U+0009 U+000A U+000D U+0020]+; |
| 1300 |
} // S |
| 1301 |
|
| 1302 |
/* |
| 1303 |
Document end scanning mode |
| 1304 |
*/ |
| 1305 |
lexmode DocumentEnd |
| 1306 |
: standalone |
| 1307 |
: extends => 'S' |
| 1308 |
{ |
| 1309 |
/* |
| 1310 |
Processing instruction |
| 1311 |
*/ |
| 1312 |
PIO := ['<'] ['?']; |
| 1313 |
|
| 1314 |
/* |
| 1315 |
Comment declaration |
| 1316 |
*/ |
| 1317 |
CDO := ['<'] ['!'] ['-'] ['-']; |
| 1318 |
} // DocumentEnd |
| 1319 |
|
| 1320 |
/* |
| 1321 |
Document misc scanning mode |
| 1322 |
|
| 1323 |
This mode scans |Misc| constructions as well |
| 1324 |
as document element's start tag. |
| 1325 |
*/ |
| 1326 |
lexmode DocumentMisc |
| 1327 |
: standalone |
| 1328 |
: extends => 'DocumentEnd' |
| 1329 |
{ |
| 1330 |
/* |
| 1331 |
Document element start tag |
| 1332 |
*/ |
| 1333 |
STAGO := ['<']; |
| 1334 |
} // DocumentMisc |
| 1335 |
|
| 1336 |
/* |
| 1337 |
Document prolog scanning mode |
| 1338 |
*/ |
| 1339 |
lexmode DocumentProlog |
| 1340 |
: standalone |
| 1341 |
: extends => 'DocumentMisc' |
| 1342 |
{ |
| 1343 |
/* |
| 1344 |
|DOCTYPE| declaration |
| 1345 |
*/ |
| 1346 |
MDO := ['<'] ['!']; |
| 1347 |
} // DocumentProlog |
| 1348 |
|
| 1349 |
/* |
| 1350 |
Document start scanning mode |
| 1351 |
*/ |
| 1352 |
lexmode DocumentStart |
| 1353 |
: initial |
| 1354 |
: standalone |
| 1355 |
: extends => 'DocumentProlog' |
| 1356 |
{ |
| 1357 |
/* |
| 1358 |
XML declaration |
| 1359 |
*/ |
| 1360 |
XDO := ['<'] ['?'] ['x'] ['m'] ['l']; |
| 1361 |
} // DocumentStart |
| 1362 |
|
| 1363 |
/* |
| 1364 |
Markup declaration scanning mode |
| 1365 |
|
| 1366 |
This mode is used to recognize |MDC| that terminates |
| 1367 |
a comment declaration as well as the base |lexmode| |
| 1368 |
for e.g. document type declaration scanning mode. |
| 1369 |
*/ |
| 1370 |
lexmode MarkupDeclaration |
| 1371 |
: standalone |
| 1372 |
: extends => 'Name' |
| 1373 |
{ |
| 1374 |
/* |
| 1375 |
Markup declaration close |
| 1376 |
*/ |
| 1377 |
MDC := ['>']; |
| 1378 |
|
| 1379 |
/* |
| 1380 |
Literal open |
| 1381 |
*/ |
| 1382 |
LIT := ['"']; |
| 1383 |
|
| 1384 |
/* |
| 1385 |
Alternative literal open |
| 1386 |
*/ |
| 1387 |
LITA := [U+0027]; |
| 1388 |
} // MarkupDeclaration |
| 1389 |
|
| 1390 |
lexmode DocumentTypeDeclaration |
| 1391 |
: standalone |
| 1392 |
: extends => 'MarkupDeclaration' |
| 1393 |
{ |
| 1394 |
/* |
| 1395 |
Declaration subset close |
| 1396 |
*/ |
| 1397 |
DSO := ['[']; |
| 1398 |
|
| 1399 |
/* |
| 1400 |
Declaration subset close |
| 1401 |
*/ |
| 1402 |
DSC := [']']; |
| 1403 |
} // DocumentTypeDeclaration |
| 1404 |
|
| 1405 |
/* |
| 1406 |
Comment declaration scanning mode |
| 1407 |
*/ |
| 1408 |
lexmode CommentDeclaration |
| 1409 |
: standalone |
| 1410 |
{ |
| 1411 |
/* |
| 1412 |
Comment close |
| 1413 |
*/ |
| 1414 |
COM := ['-'] ['-']; |
| 1415 |
|
| 1416 |
/* |
| 1417 |
Comment data |
| 1418 |
*/ |
| 1419 |
$string := ['-']? [^'-']; |
| 1420 |
STRING : value := $string+; |
| 1421 |
} // CommentDeclaration |
| 1422 |
|
| 1423 |
/* |
| 1424 |
Processing instruction name and |S| scanning mode |
| 1425 |
*/ |
| 1426 |
lexmode PIName |
| 1427 |
: standalone |
| 1428 |
: extends => 'Name' |
| 1429 |
: extends => 'S' |
| 1430 |
{ |
| 1431 |
/* |
| 1432 |
Processing instruction close |
| 1433 |
*/ |
| 1434 |
PIC := ['?'] ['>']; |
| 1435 |
} // PIName |
| 1436 |
|
| 1437 |
/* |
| 1438 |
Processing instruction data scanning mode |
| 1439 |
*/ |
| 1440 |
lexmode PIData |
| 1441 |
: standalone |
| 1442 |
{ |
| 1443 |
/* |
| 1444 |
Processing instruction close |
| 1445 |
*/ |
| 1446 |
PIC := ['?'] ['>']; |
| 1447 |
|
| 1448 |
/* |
| 1449 |
Processing instruction target data |
| 1450 |
*/ |
| 1451 |
?default-token DATA : value; |
| 1452 |
} // PIData |
| 1453 |
|
| 1454 |
/* |
| 1455 |
Content of element scanning mode |
| 1456 |
*/ |
| 1457 |
lexmode ElementContent |
| 1458 |
: standalone |
| 1459 |
{ |
| 1460 |
/* |
| 1461 |
Start tag open |
| 1462 |
*/ |
| 1463 |
STAGO := ['<']; |
| 1464 |
|
| 1465 |
/* |
| 1466 |
End tag open |
| 1467 |
*/ |
| 1468 |
ETAGO := ['<'] ['/']; |
| 1469 |
|
| 1470 |
/* |
| 1471 |
Hexadecimal character reference open |
| 1472 |
*/ |
| 1473 |
HCRO := ['&'] ['#'] ['x']; |
| 1474 |
|
| 1475 |
/* |
| 1476 |
Numeric character reference open |
| 1477 |
*/ |
| 1478 |
CRO := ['&'] ['#']; |
| 1479 |
|
| 1480 |
/* |
| 1481 |
General entity reference open |
| 1482 |
*/ |
| 1483 |
ERO := ['&']; |
| 1484 |
|
| 1485 |
/* |
| 1486 |
Comment declaration open |
| 1487 |
*/ |
| 1488 |
CDO := ['<'] ['!'] ['-'] ['-']; |
| 1489 |
|
| 1490 |
/* |
| 1491 |
CDATA section open |
| 1492 |
*/ |
| 1493 |
CDSO := ['<'] ['!'] ['['] |
| 1494 |
['C'] ['D'] ['A'] ['T'] ['A'] ['[']; |
| 1495 |
|
| 1496 |
/* |
| 1497 |
Processing instruction open |
| 1498 |
*/ |
| 1499 |
PIO := ['<'] ['?']; |
| 1500 |
|
| 1501 |
/* |
| 1502 |
Markup section end |
| 1503 |
*/ |
| 1504 |
MSE := [']'] [']'] ['>']; |
| 1505 |
|
| 1506 |
/* |
| 1507 |
Character data |
| 1508 |
*/ |
| 1509 |
/* |
| 1510 |
Character data and/or |MSE| |
| 1511 |
*/ |
| 1512 |
?default-token CharData : value; |
| 1513 |
} // ElementContent |
| 1514 |
|
| 1515 |
/* |
| 1516 |
CDATA section content scanning mode |
| 1517 |
*/ |
| 1518 |
lexmode CDATASectionContent |
| 1519 |
: standalone |
| 1520 |
{ |
| 1521 |
/* |
| 1522 |
Markup section end |
| 1523 |
*/ |
| 1524 |
MSE := [']'] [']'] ['>']; |
| 1525 |
|
| 1526 |
/* |
| 1527 |
Character data |
| 1528 |
*/ |
| 1529 |
?default-token CData : value; |
| 1530 |
} // CDATASectionContent |
| 1531 |
|
| 1532 |
lexmode EntityReference |
| 1533 |
: standalone |
| 1534 |
: extends => 'Name' |
| 1535 |
{ |
| 1536 |
/* |
| 1537 |
Reference close |
| 1538 |
*/ |
| 1539 |
REFC := [';']; |
| 1540 |
} // EntityReference |
| 1541 |
|
| 1542 |
lexmode NumericCharacterReference |
| 1543 |
: standalone |
| 1544 |
{ |
| 1545 |
/* |
| 1546 |
Decimal number |
| 1547 |
*/ |
| 1548 |
$digit := ['0' '1' '2' '3' '4' '5' '6' '7' '8' '9']; |
| 1549 |
NUMBER : value := $digit+; |
| 1550 |
|
| 1551 |
/* |
| 1552 |
Reference close |
| 1553 |
*/ |
| 1554 |
REFC := [';']; |
| 1555 |
} // NumericCharacterReference |
| 1556 |
|
| 1557 |
lexmode HexadecimalCharacterReference |
| 1558 |
: standalone |
| 1559 |
{ |
| 1560 |
/* |
| 1561 |
Hexadecimal number |
| 1562 |
*/ |
| 1563 |
$hexdigit := ['0' '1' '2' '3' '4' '5' '6' '7' '8' '9' |
| 1564 |
'A' 'B' 'C' 'D' 'E' 'F' |
| 1565 |
'a' 'b' 'c' 'd' 'e' 'f']; |
| 1566 |
Hex : value := $hexdigit+; |
| 1567 |
|
| 1568 |
/* |
| 1569 |
Reference close |
| 1570 |
*/ |
| 1571 |
REFC := [';']; |
| 1572 |
} // HexadecimalCharacterReference |
| 1573 |
|
| 1574 |
lexmode StartTag |
| 1575 |
: standalone |
| 1576 |
: extends => 'Name' |
| 1577 |
: extends => 'S' |
| 1578 |
{ |
| 1579 |
|
| 1580 |
/* |
| 1581 |
Value indicator |
| 1582 |
*/ |
| 1583 |
VI := ['=']; |
| 1584 |
|
| 1585 |
/* |
| 1586 |
Literal open |
| 1587 |
*/ |
| 1588 |
LIT := ['"']; |
| 1589 |
LITA := [U+0027]; |
| 1590 |
|
| 1591 |
/* |
| 1592 |
Tag close |
| 1593 |
*/ |
| 1594 |
TAGC := ['>']; |
| 1595 |
|
| 1596 |
/* |
| 1597 |
Empty element tag close |
| 1598 |
*/ |
| 1599 |
MTAGC := ['/'] ['>']; |
| 1600 |
} // StartTag |
| 1601 |
|
| 1602 |
lexmode EndTag |
| 1603 |
: standalone |
| 1604 |
: extends => 'Name' |
| 1605 |
: extends => 'S' |
| 1606 |
{ |
| 1607 |
/* |
| 1608 |
Tag close |
| 1609 |
*/ |
| 1610 |
TAGC := ['>']; |
| 1611 |
} // EndTag |
| 1612 |
|
| 1613 |
lexmode AttributeValueLiteral_ { |
| 1614 |
ERO := ['&']; |
| 1615 |
CRO := ['&'] ['#']; |
| 1616 |
HCRO := ['&'] ['#'] ['x']; |
| 1617 |
} // AttributeValueLiteral_ |
| 1618 |
|
| 1619 |
lexmode AttributeValueLiteral |
| 1620 |
: standalone |
| 1621 |
: extends => 'AttributeValueLiteral_' |
| 1622 |
{ |
| 1623 |
LIT := ['"']; |
| 1624 |
STRING : value := [^'"' '&' '<']; |
| 1625 |
} // AttributeValueLiteral |
| 1626 |
|
| 1627 |
lexmode AttributeValueLiteralA |
| 1628 |
: standalone |
| 1629 |
: extends => 'AttributeValueLiteral_' |
| 1630 |
{ |
| 1631 |
LIT := [U+0027]; |
| 1632 |
STRING : value := [^U+0027 '&' '<']; |
| 1633 |
} // AttributeValueLiteralA |
| 1634 |
|
| 1635 |
token-error default : default { |
| 1636 |
lang:Perl { |
| 1637 |
my $location = { |
| 1638 |
utf32_offset => pos ($self->{source}), |
| 1639 |
}; |
| 1640 |
my $continue = __DOMCore:ERROR{xp|bad-token-error:: |
| 1641 |
xp|error-token => {$token}, |
| 1642 |
DOMCore|location => {$location}, |
| 1643 |
xp|source-text => {\($self->{source})}, |
| 1644 |
}__; |
| 1645 |
unless ($continue) { |
| 1646 |
__EXCEPTION{DOMLS|PARSE_ERR:: |
| 1647 |
}__; |
| 1648 |
} |
| 1649 |
$self->{has_error} = true; |
| 1650 |
} |
| 1651 |
} // default |
| 1652 |
##ManakaiXMLParser |
| 1653 |
|
| 1654 |
|
| 1655 |
ElementTypeBinding: |
| 1656 |
@Name: RuleDef |
| 1657 |
@ElementType: |
| 1658 |
dis:ResourceDef |
| 1659 |
@ShadowContent: |
| 1660 |
@@ForCheck: ManakaiDOM|ForClass |
| 1661 |
@@rdf:type: Muf2003|RuleDefClass |
| 1662 |
|
| 1663 |
ElementTypeBinding: |
| 1664 |
@Name: RuleParam |
| 1665 |
@ElementType: |
| 1666 |
dis:ResourceDef |
| 1667 |
@ShadowContent: |
| 1668 |
@@rdf:type: Muf2003|RuleParameter |
| 1669 |
|
| 1670 |
ElementTypeBinding: |
| 1671 |
@Name: enImplNote |
| 1672 |
@ElementType: |
| 1673 |
dis:ImplNote |
| 1674 |
@ShadowContent: |
| 1675 |
@@lang:en |
| 1676 |
|
| 1677 |
ElementTypeBinding: |
| 1678 |
@Name: ErrDef |
| 1679 |
@ElementType: |
| 1680 |
dis:ResourceDef |
| 1681 |
@ShadowContent: |
| 1682 |
@@rdf:type: DOMCore|DOMErrorType |
| 1683 |
@@For: ManakaiDOM|DOM3 |
| 1684 |
@@ecore:textFormatter: ManakaiXMLParserExceptionFormatter |
| 1685 |
|
| 1686 |
ErrDef: |
| 1687 |
@QName: xp|bad-token-error |
| 1688 |
@enDesc: |
| 1689 |
The parser is encountered to a token whose type is not |
| 1690 |
allowed there. |
| 1691 |
@DOMCore:severity: DOMCore|SEVERITY_ERROR |
| 1692 |
@enMufDef: |
| 1693 |
Token |%xp-error-token-type;|%xp-error-token-value |
| 1694 |
(prefix => { (|}, suffix => {|)}); is not |
| 1695 |
allowed %xp-error-lines (prefix => {(|}, suffix => {|)}); |
| 1696 |
|
| 1697 |
PropDef: |
| 1698 |
@QName: xp|error-token |
| 1699 |
@enDesc: |
| 1700 |
The token where the parser found an error. |
| 1701 |
|
| 1702 |
PropDef: |
| 1703 |
@QName: xp|source-text |
| 1704 |
@enDesc: |
| 1705 |
A reference to the original source text, if available. |
| 1706 |
|
| 1707 |
ElementTypeBinding: |
| 1708 |
@Name:enMufDef |
| 1709 |
@ElementType: |
| 1710 |
ecore:defaultMessage |
| 1711 |
@ShadowContent: |
| 1712 |
@@lang:en |
| 1713 |
@@ContentType: |
| 1714 |
lang:muf |
| 1715 |
|
| 1716 |
ResourceDef: |
| 1717 |
@QName: DOMImpl |
| 1718 |
@AliasFor: DOMCore|DOMImplementation |
| 1719 |
@For: ManakaiDOM|DOM |
| 1720 |
|
| 1721 |
ElementTypeBinding: |
| 1722 |
@Name: Attr |
| 1723 |
@ElementType: |
| 1724 |
dis:ResourceDef |
| 1725 |
@ShadowContent: |
| 1726 |
@@rdf:type: DISLang|Attribute |
| 1727 |
@@ForCheck: !=ManakaiDOM|ManakaiDOM |
| 1728 |
|
| 1729 |
ElementTypeBinding: |
| 1730 |
@Name: Get |
| 1731 |
@ElementType: |
| 1732 |
dis:ResourceDef |
| 1733 |
@ShadowContent: |
| 1734 |
@@rdf:type: DISLang|AttributeGet |
| 1735 |
|
| 1736 |
ElementTypeBinding: |
| 1737 |
@Name: Set |
| 1738 |
@ElementType: |
| 1739 |
dis:ResourceDef |
| 1740 |
@ShadowContent: |
| 1741 |
@@rdf:type: DISLang|AttributeSet |
| 1742 |
|
| 1743 |
ElementTypeBinding: |
| 1744 |
@Name: enDesc |
| 1745 |
@ElementType: |
| 1746 |
dis:Description |
| 1747 |
@ShadowContent: |
| 1748 |
@@lang:en |
| 1749 |
|
| 1750 |
ElementTypeBinding: |
| 1751 |
@Name: Method |
| 1752 |
@ElementType: |
| 1753 |
dis:ResourceDef |
| 1754 |
@ShadowContent: |
| 1755 |
@@rdf:type: DISLang|Method |
| 1756 |
@@For: !=ManakaiDOM|ManakaiDOM |
| 1757 |
|
| 1758 |
ElementTypeBinding: |
| 1759 |
@Name: Return |
| 1760 |
@ElementType: |
| 1761 |
dis:ResourceDef |
| 1762 |
@ShadowContent: |
| 1763 |
@@rdf:type: DISLang|MethodReturn |
| 1764 |
|
| 1765 |
ElementTypeBinding: |
| 1766 |
@Name: Param |
| 1767 |
@ElementType: |
| 1768 |
dis:ResourceDef |
| 1769 |
@ShadowContent: |
| 1770 |
@@rdf:type: DISLang|MethodParameter |
| 1771 |
|
| 1772 |
ElementTypeBinding: |
| 1773 |
@Name: PerlDef |
| 1774 |
@ElementType: |
| 1775 |
dis:Def |
| 1776 |
@ShadowContent: |
| 1777 |
@@ContentType: lang|Perl |
| 1778 |
|
| 1779 |
ElementTypeBinding: |
| 1780 |
@Name: PropDef |
| 1781 |
@ElementType: |
| 1782 |
dis:ResourceDef |
| 1783 |
@ShadowContent: |
| 1784 |
@@rdf:type: rdf|Property |
| 1785 |
|
| 1786 |
ClsDef: |
| 1787 |
@ClsQName: ManakaiXMLParserExceptionFormatter |
| 1788 |
|
| 1789 |
@ClsISA: ecore|MUErrorFormatter||ManakaiDOM|Perl |
| 1790 |
|
| 1791 |
@RuleDef: |
| 1792 |
@@Name: xp-error-token-type |
| 1793 |
@@enDesc: |
| 1794 |
The type of the token the parser is encountered. |
| 1795 |
|
| 1796 |
@@Method: |
| 1797 |
@@@Name: after |
| 1798 |
@@@Param: |
| 1799 |
@@@@Name: name |
| 1800 |
@@@@Type: DOMString |
| 1801 |
@@@@enDesc: The name of the method. |
| 1802 |
@@@Param: |
| 1803 |
@@@@Name: p |
| 1804 |
@@@@Type: DISPerl|HASH |
| 1805 |
@@@@enDesc: The set of the parameters to the method. |
| 1806 |
@@@Param: |
| 1807 |
@@@@Name: o |
| 1808 |
@@@@Type: DISPerl|HASH |
| 1809 |
@@@@enDesc: The option value. |
| 1810 |
@@@Return: |
| 1811 |
@@@@PerlDef: |
| 1812 |
$p->{-result} = $o->{<H::xp|error-token>}->{type} |
| 1813 |
if defined $o->{<H::xp|error-token>}->{type}; |
| 1814 |
|
| 1815 |
@RuleDef: |
| 1816 |
@@Name: xp-error-token-value |
| 1817 |
@@enDesc: |
| 1818 |
The value of the token the parser is encountered, if any. |
| 1819 |
|
| 1820 |
@@Method: |
| 1821 |
@@@Name: after |
| 1822 |
@@@Param: |
| 1823 |
@@@@Name: name |
| 1824 |
@@@@Type: DOMString |
| 1825 |
@@@@enDesc: The name of the method. |
| 1826 |
@@@Param: |
| 1827 |
@@@@Name: p |
| 1828 |
@@@@Type: DISPerl|HASH |
| 1829 |
@@@@enDesc: The set of the parameters to the method. |
| 1830 |
@@@Param: |
| 1831 |
@@@@Name: o |
| 1832 |
@@@@Type: DISPerl|HASH |
| 1833 |
@@@@enDesc: The option value. |
| 1834 |
@@@Return: |
| 1835 |
@@@@PerlDef: |
| 1836 |
$p->{-result} = $o->{<H::xp|error-token>}->{value} |
| 1837 |
if defined $o->{<H::xp|error-token>}->{value}; |
| 1838 |
|
| 1839 |
@RuleDef: |
| 1840 |
@@Name: xp-error-lines |
| 1841 |
@@enDesc: |
| 1842 |
A copy of fragment of the source text that contains the line |
| 1843 |
where the error occurred, if available. |
| 1844 |
|
| 1845 |
@@Method: |
| 1846 |
@@@Name: after |
| 1847 |
@@@Param: |
| 1848 |
@@@@Name: name |
| 1849 |
@@@@Type: DOMString |
| 1850 |
@@@@enDesc: The name of the method. |
| 1851 |
@@@Param: |
| 1852 |
@@@@Name: p |
| 1853 |
@@@@Type: DISPerl|HASH |
| 1854 |
@@@@enDesc: The set of the parameters to the method. |
| 1855 |
@@@Param: |
| 1856 |
@@@@Name: o |
| 1857 |
@@@@Type: DISPerl|HASH |
| 1858 |
@@@@enDesc: The option value. |
| 1859 |
@@@Return: |
| 1860 |
@@@@PerlDef: |
| 1861 |
my $pos = $o-><AG::DOMCore|DOMError.location> |
| 1862 |
-><AG::DOMCore|DOMLocator.utf32Offset>; |
| 1863 |
if ($pos > -1) { |
| 1864 |
my $src = $o->{<H::xp|source-text>}; |
| 1865 |
my $start = $pos; |
| 1866 |
$start = rindex ($$src, "\x0A", $start - 1) for 0..2; |
| 1867 |
$start++; |
| 1868 |
my $end = $pos; |
| 1869 |
$end = index ($$src, "\x0A", $end + 1) for 0..2; |
| 1870 |
$end = length $$src if $end < 0; |
| 1871 |
$p->{-result} = substr $$src, $start, $end - $start; |
| 1872 |
} |
| 1873 |
##XMLParserExceptionFormatter |
Parent Directory
| 
Revision Log