1 /* 2 * SAX2.c : Default SAX2 handler to build a tree. 3 * 4 * See Copyright for the status of this software. 5 * 6 * Daniel Veillard <daniel (at) veillard.com> 7 */ 8 9 10 #define IN_LIBXML 11 #include "libxml.h" 12 #include <stdlib.h> 13 #include <string.h> 14 #include <limits.h> 15 #include <libxml/xmlmemory.h> 16 #include <libxml/tree.h> 17 #include <libxml/parser.h> 18 #include <libxml/parserInternals.h> 19 #include <libxml/valid.h> 20 #include <libxml/entities.h> 21 #include <libxml/xmlerror.h> 22 #include <libxml/debugXML.h> 23 #include <libxml/xmlIO.h> 24 #include <libxml/SAX.h> 25 #include <libxml/uri.h> 26 #include <libxml/valid.h> 27 #include <libxml/HTMLtree.h> 28 #include <libxml/globals.h> 29 30 /* Define SIZE_T_MAX unless defined through <limits.h>. */ 31 #ifndef SIZE_T_MAX 32 # define SIZE_T_MAX ((size_t)-1) 33 #endif /* !SIZE_T_MAX */ 34 35 /* #define DEBUG_SAX2 */ 36 /* #define DEBUG_SAX2_TREE */ 37 38 /** 39 * TODO: 40 * 41 * macro to flag unimplemented blocks 42 * XML_CATALOG_PREFER user env to select between system/public prefered 43 * option. C.f. Richard Tobin <richard (at) cogsci.ed.ac.uk> 44 *> Just FYI, I am using an environment variable XML_CATALOG_PREFER with 45 *> values "system" and "public". I have made the default be "system" to 46 *> match yours. 47 */ 48 #define TODO \ 49 xmlGenericError(xmlGenericErrorContext, \ 50 "Unimplemented block at %s:%d\n", \ 51 __FILE__, __LINE__); 52 53 /* 54 * xmlSAX2ErrMemory: 55 * @ctxt: an XML validation parser context 56 * @msg: a string to accompany the error message 57 */ 58 static void LIBXML_ATTR_FORMAT(2,0) 59 xmlSAX2ErrMemory(xmlParserCtxtPtr ctxt, const char *msg) { 60 xmlStructuredErrorFunc schannel = NULL; 61 const char *str1 = "out of memory\n"; 62 63 if (ctxt != NULL) { 64 ctxt->errNo = XML_ERR_NO_MEMORY; 65 if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC)) 66 schannel = ctxt->sax->serror; 67 __xmlRaiseError(schannel, 68 ctxt->vctxt.error, ctxt->vctxt.userData, 69 ctxt, NULL, XML_FROM_PARSER, XML_ERR_NO_MEMORY, 70 XML_ERR_ERROR, NULL, 0, (const char *) str1, 71 NULL, NULL, 0, 0, 72 msg, (const char *) str1, NULL); 73 ctxt->errNo = XML_ERR_NO_MEMORY; 74 ctxt->instate = XML_PARSER_EOF; 75 ctxt->disableSAX = 1; 76 } else { 77 __xmlRaiseError(schannel, 78 NULL, NULL, 79 ctxt, NULL, XML_FROM_PARSER, XML_ERR_NO_MEMORY, 80 XML_ERR_ERROR, NULL, 0, (const char *) str1, 81 NULL, NULL, 0, 0, 82 msg, (const char *) str1, NULL); 83 } 84 } 85 86 /** 87 * xmlValidError: 88 * @ctxt: an XML validation parser context 89 * @error: the error number 90 * @msg: the error message 91 * @str1: extra data 92 * @str2: extra data 93 * 94 * Handle a validation error 95 */ 96 static void LIBXML_ATTR_FORMAT(3,0) 97 xmlErrValid(xmlParserCtxtPtr ctxt, xmlParserErrors error, 98 const char *msg, const char *str1, const char *str2) 99 { 100 xmlStructuredErrorFunc schannel = NULL; 101 102 if ((ctxt != NULL) && (ctxt->disableSAX != 0) && 103 (ctxt->instate == XML_PARSER_EOF)) 104 return; 105 if (ctxt != NULL) { 106 ctxt->errNo = error; 107 if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC)) 108 schannel = ctxt->sax->serror; 109 __xmlRaiseError(schannel, 110 ctxt->vctxt.error, ctxt->vctxt.userData, 111 ctxt, NULL, XML_FROM_DTD, error, 112 XML_ERR_ERROR, NULL, 0, (const char *) str1, 113 (const char *) str2, NULL, 0, 0, 114 msg, (const char *) str1, (const char *) str2); 115 ctxt->valid = 0; 116 } else { 117 __xmlRaiseError(schannel, 118 NULL, NULL, 119 ctxt, NULL, XML_FROM_DTD, error, 120 XML_ERR_ERROR, NULL, 0, (const char *) str1, 121 (const char *) str2, NULL, 0, 0, 122 msg, (const char *) str1, (const char *) str2); 123 } 124 } 125 126 /** 127 * xmlFatalErrMsg: 128 * @ctxt: an XML parser context 129 * @error: the error number 130 * @msg: the error message 131 * @str1: an error string 132 * @str2: an error string 133 * 134 * Handle a fatal parser error, i.e. violating Well-Formedness constraints 135 */ 136 static void LIBXML_ATTR_FORMAT(3,0) 137 xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, 138 const char *msg, const xmlChar *str1, const xmlChar *str2) 139 { 140 if ((ctxt != NULL) && (ctxt->disableSAX != 0) && 141 (ctxt->instate == XML_PARSER_EOF)) 142 return; 143 if (ctxt != NULL) 144 ctxt->errNo = error; 145 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, 146 XML_ERR_FATAL, NULL, 0, 147 (const char *) str1, (const char *) str2, 148 NULL, 0, 0, msg, str1, str2); 149 if (ctxt != NULL) { 150 ctxt->wellFormed = 0; 151 ctxt->valid = 0; 152 if (ctxt->recovery == 0) 153 ctxt->disableSAX = 1; 154 } 155 } 156 157 /** 158 * xmlWarnMsg: 159 * @ctxt: an XML parser context 160 * @error: the error number 161 * @msg: the error message 162 * @str1: an error string 163 * @str2: an error string 164 * 165 * Handle a parser warning 166 */ 167 static void LIBXML_ATTR_FORMAT(3,0) 168 xmlWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, 169 const char *msg, const xmlChar *str1) 170 { 171 if ((ctxt != NULL) && (ctxt->disableSAX != 0) && 172 (ctxt->instate == XML_PARSER_EOF)) 173 return; 174 if (ctxt != NULL) 175 ctxt->errNo = error; 176 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, 177 XML_ERR_WARNING, NULL, 0, 178 (const char *) str1, NULL, 179 NULL, 0, 0, msg, str1); 180 } 181 182 /** 183 * xmlNsErrMsg: 184 * @ctxt: an XML parser context 185 * @error: the error number 186 * @msg: the error message 187 * @str1: an error string 188 * @str2: an error string 189 * 190 * Handle a namespace error 191 */ 192 static void LIBXML_ATTR_FORMAT(3,0) 193 xmlNsErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, 194 const char *msg, const xmlChar *str1, const xmlChar *str2) 195 { 196 if ((ctxt != NULL) && (ctxt->disableSAX != 0) && 197 (ctxt->instate == XML_PARSER_EOF)) 198 return; 199 if (ctxt != NULL) 200 ctxt->errNo = error; 201 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, 202 XML_ERR_ERROR, NULL, 0, 203 (const char *) str1, (const char *) str2, 204 NULL, 0, 0, msg, str1, str2); 205 } 206 207 /** 208 * xmlNsWarnMsg: 209 * @ctxt: an XML parser context 210 * @error: the error number 211 * @msg: the error message 212 * @str1: an error string 213 * 214 * Handle a namespace warning 215 */ 216 static void LIBXML_ATTR_FORMAT(3,0) 217 xmlNsWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, 218 const char *msg, const xmlChar *str1, const xmlChar *str2) 219 { 220 if ((ctxt != NULL) && (ctxt->disableSAX != 0) && 221 (ctxt->instate == XML_PARSER_EOF)) 222 return; 223 if (ctxt != NULL) 224 ctxt->errNo = error; 225 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, 226 XML_ERR_WARNING, NULL, 0, 227 (const char *) str1, (const char *) str2, 228 NULL, 0, 0, msg, str1, str2); 229 } 230 231 /** 232 * xmlSAX2GetPublicId: 233 * @ctx: the user data (XML parser context) 234 * 235 * Provides the public ID e.g. "-//SGMLSOURCE//DTD DEMO//EN" 236 * 237 * Returns a xmlChar * 238 */ 239 const xmlChar * 240 xmlSAX2GetPublicId(void *ctx ATTRIBUTE_UNUSED) 241 { 242 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */ 243 return(NULL); 244 } 245 246 /** 247 * xmlSAX2GetSystemId: 248 * @ctx: the user data (XML parser context) 249 * 250 * Provides the system ID, basically URL or filename e.g. 251 * http://www.sgmlsource.com/dtds/memo.dtd 252 * 253 * Returns a xmlChar * 254 */ 255 const xmlChar * 256 xmlSAX2GetSystemId(void *ctx) 257 { 258 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 259 if ((ctx == NULL) || (ctxt->input == NULL)) return(NULL); 260 return((const xmlChar *) ctxt->input->filename); 261 } 262 263 /** 264 * xmlSAX2GetLineNumber: 265 * @ctx: the user data (XML parser context) 266 * 267 * Provide the line number of the current parsing point. 268 * 269 * Returns an int 270 */ 271 int 272 xmlSAX2GetLineNumber(void *ctx) 273 { 274 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 275 if ((ctx == NULL) || (ctxt->input == NULL)) return(0); 276 return(ctxt->input->line); 277 } 278 279 /** 280 * xmlSAX2GetColumnNumber: 281 * @ctx: the user data (XML parser context) 282 * 283 * Provide the column number of the current parsing point. 284 * 285 * Returns an int 286 */ 287 int 288 xmlSAX2GetColumnNumber(void *ctx) 289 { 290 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 291 if ((ctx == NULL) || (ctxt->input == NULL)) return(0); 292 return(ctxt->input->col); 293 } 294 295 /** 296 * xmlSAX2IsStandalone: 297 * @ctx: the user data (XML parser context) 298 * 299 * Is this document tagged standalone ? 300 * 301 * Returns 1 if true 302 */ 303 int 304 xmlSAX2IsStandalone(void *ctx) 305 { 306 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 307 if ((ctx == NULL) || (ctxt->myDoc == NULL)) return(0); 308 return(ctxt->myDoc->standalone == 1); 309 } 310 311 /** 312 * xmlSAX2HasInternalSubset: 313 * @ctx: the user data (XML parser context) 314 * 315 * Does this document has an internal subset 316 * 317 * Returns 1 if true 318 */ 319 int 320 xmlSAX2HasInternalSubset(void *ctx) 321 { 322 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 323 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0); 324 return(ctxt->myDoc->intSubset != NULL); 325 } 326 327 /** 328 * xmlSAX2HasExternalSubset: 329 * @ctx: the user data (XML parser context) 330 * 331 * Does this document has an external subset 332 * 333 * Returns 1 if true 334 */ 335 int 336 xmlSAX2HasExternalSubset(void *ctx) 337 { 338 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 339 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0); 340 return(ctxt->myDoc->extSubset != NULL); 341 } 342 343 /** 344 * xmlSAX2InternalSubset: 345 * @ctx: the user data (XML parser context) 346 * @name: the root element name 347 * @ExternalID: the external ID 348 * @SystemID: the SYSTEM ID (e.g. filename or URL) 349 * 350 * Callback on internal subset declaration. 351 */ 352 void 353 xmlSAX2InternalSubset(void *ctx, const xmlChar *name, 354 const xmlChar *ExternalID, const xmlChar *SystemID) 355 { 356 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 357 xmlDtdPtr dtd; 358 if (ctx == NULL) return; 359 #ifdef DEBUG_SAX 360 xmlGenericError(xmlGenericErrorContext, 361 "SAX.xmlSAX2InternalSubset(%s, %s, %s)\n", 362 name, ExternalID, SystemID); 363 #endif 364 365 if (ctxt->myDoc == NULL) 366 return; 367 dtd = xmlGetIntSubset(ctxt->myDoc); 368 if (dtd != NULL) { 369 if (ctxt->html) 370 return; 371 xmlUnlinkNode((xmlNodePtr) dtd); 372 xmlFreeDtd(dtd); 373 ctxt->myDoc->intSubset = NULL; 374 } 375 ctxt->myDoc->intSubset = 376 xmlCreateIntSubset(ctxt->myDoc, name, ExternalID, SystemID); 377 if (ctxt->myDoc->intSubset == NULL) 378 xmlSAX2ErrMemory(ctxt, "xmlSAX2InternalSubset"); 379 } 380 381 /** 382 * xmlSAX2ExternalSubset: 383 * @ctx: the user data (XML parser context) 384 * @name: the root element name 385 * @ExternalID: the external ID 386 * @SystemID: the SYSTEM ID (e.g. filename or URL) 387 * 388 * Callback on external subset declaration. 389 */ 390 void 391 xmlSAX2ExternalSubset(void *ctx, const xmlChar *name, 392 const xmlChar *ExternalID, const xmlChar *SystemID) 393 { 394 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 395 if (ctx == NULL) return; 396 #ifdef DEBUG_SAX 397 xmlGenericError(xmlGenericErrorContext, 398 "SAX.xmlSAX2ExternalSubset(%s, %s, %s)\n", 399 name, ExternalID, SystemID); 400 #endif 401 if (((ExternalID != NULL) || (SystemID != NULL)) && 402 (((ctxt->validate) || (ctxt->loadsubset != 0)) && 403 (ctxt->wellFormed && ctxt->myDoc))) { 404 /* 405 * Try to fetch and parse the external subset. 406 */ 407 xmlParserInputPtr oldinput; 408 int oldinputNr; 409 int oldinputMax; 410 xmlParserInputPtr *oldinputTab; 411 xmlParserInputPtr input = NULL; 412 xmlCharEncoding enc; 413 int oldcharset; 414 const xmlChar *oldencoding; 415 416 /* 417 * Ask the Entity resolver to load the damn thing 418 */ 419 if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) 420 input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, 421 SystemID); 422 if (input == NULL) { 423 return; 424 } 425 426 xmlNewDtd(ctxt->myDoc, name, ExternalID, SystemID); 427 428 /* 429 * make sure we won't destroy the main document context 430 */ 431 oldinput = ctxt->input; 432 oldinputNr = ctxt->inputNr; 433 oldinputMax = ctxt->inputMax; 434 oldinputTab = ctxt->inputTab; 435 oldcharset = ctxt->charset; 436 oldencoding = ctxt->encoding; 437 ctxt->encoding = NULL; 438 439 ctxt->inputTab = (xmlParserInputPtr *) 440 xmlMalloc(5 * sizeof(xmlParserInputPtr)); 441 if (ctxt->inputTab == NULL) { 442 xmlSAX2ErrMemory(ctxt, "xmlSAX2ExternalSubset"); 443 ctxt->input = oldinput; 444 ctxt->inputNr = oldinputNr; 445 ctxt->inputMax = oldinputMax; 446 ctxt->inputTab = oldinputTab; 447 ctxt->charset = oldcharset; 448 ctxt->encoding = oldencoding; 449 return; 450 } 451 ctxt->inputNr = 0; 452 ctxt->inputMax = 5; 453 ctxt->input = NULL; 454 xmlPushInput(ctxt, input); 455 456 /* 457 * On the fly encoding conversion if needed 458 */ 459 if (ctxt->input->length >= 4) { 460 enc = xmlDetectCharEncoding(ctxt->input->cur, 4); 461 xmlSwitchEncoding(ctxt, enc); 462 } 463 464 if (input->filename == NULL) 465 input->filename = (char *) xmlCanonicPath(SystemID); 466 input->line = 1; 467 input->col = 1; 468 input->base = ctxt->input->cur; 469 input->cur = ctxt->input->cur; 470 input->free = NULL; 471 472 /* 473 * let's parse that entity knowing it's an external subset. 474 */ 475 xmlParseExternalSubset(ctxt, ExternalID, SystemID); 476 477 /* 478 * Free up the external entities 479 */ 480 481 while (ctxt->inputNr > 1) 482 xmlPopInput(ctxt); 483 xmlFreeInputStream(ctxt->input); 484 xmlFree(ctxt->inputTab); 485 486 /* 487 * Restore the parsing context of the main entity 488 */ 489 ctxt->input = oldinput; 490 ctxt->inputNr = oldinputNr; 491 ctxt->inputMax = oldinputMax; 492 ctxt->inputTab = oldinputTab; 493 ctxt->charset = oldcharset; 494 if ((ctxt->encoding != NULL) && 495 ((ctxt->dict == NULL) || 496 (!xmlDictOwns(ctxt->dict, ctxt->encoding)))) 497 xmlFree((xmlChar *) ctxt->encoding); 498 ctxt->encoding = oldencoding; 499 /* ctxt->wellFormed = oldwellFormed; */ 500 } 501 } 502 503 /** 504 * xmlSAX2ResolveEntity: 505 * @ctx: the user data (XML parser context) 506 * @publicId: The public ID of the entity 507 * @systemId: The system ID of the entity 508 * 509 * The entity loader, to control the loading of external entities, 510 * the application can either: 511 * - override this xmlSAX2ResolveEntity() callback in the SAX block 512 * - or better use the xmlSetExternalEntityLoader() function to 513 * set up it's own entity resolution routine 514 * 515 * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour. 516 */ 517 xmlParserInputPtr 518 xmlSAX2ResolveEntity(void *ctx, const xmlChar *publicId, const xmlChar *systemId) 519 { 520 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 521 xmlParserInputPtr ret; 522 xmlChar *URI; 523 const char *base = NULL; 524 525 if (ctx == NULL) return(NULL); 526 if (ctxt->input != NULL) 527 base = ctxt->input->filename; 528 if (base == NULL) 529 base = ctxt->directory; 530 531 URI = xmlBuildURI(systemId, (const xmlChar *) base); 532 533 #ifdef DEBUG_SAX 534 xmlGenericError(xmlGenericErrorContext, 535 "SAX.xmlSAX2ResolveEntity(%s, %s)\n", publicId, systemId); 536 #endif 537 538 ret = xmlLoadExternalEntity((const char *) URI, 539 (const char *) publicId, ctxt); 540 if (URI != NULL) 541 xmlFree(URI); 542 return(ret); 543 } 544 545 /** 546 * xmlSAX2GetEntity: 547 * @ctx: the user data (XML parser context) 548 * @name: The entity name 549 * 550 * Get an entity by name 551 * 552 * Returns the xmlEntityPtr if found. 553 */ 554 xmlEntityPtr 555 xmlSAX2GetEntity(void *ctx, const xmlChar *name) 556 { 557 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 558 xmlEntityPtr ret = NULL; 559 560 if (ctx == NULL) return(NULL); 561 #ifdef DEBUG_SAX 562 xmlGenericError(xmlGenericErrorContext, 563 "SAX.xmlSAX2GetEntity(%s)\n", name); 564 #endif 565 566 if (ctxt->inSubset == 0) { 567 ret = xmlGetPredefinedEntity(name); 568 if (ret != NULL) 569 return(ret); 570 } 571 if ((ctxt->myDoc != NULL) && (ctxt->myDoc->standalone == 1)) { 572 if (ctxt->inSubset == 2) { 573 ctxt->myDoc->standalone = 0; 574 ret = xmlGetDocEntity(ctxt->myDoc, name); 575 ctxt->myDoc->standalone = 1; 576 } else { 577 ret = xmlGetDocEntity(ctxt->myDoc, name); 578 if (ret == NULL) { 579 ctxt->myDoc->standalone = 0; 580 ret = xmlGetDocEntity(ctxt->myDoc, name); 581 if (ret != NULL) { 582 xmlFatalErrMsg(ctxt, XML_ERR_NOT_STANDALONE, 583 "Entity(%s) document marked standalone but requires external subset\n", 584 name, NULL); 585 } 586 ctxt->myDoc->standalone = 1; 587 } 588 } 589 } else { 590 ret = xmlGetDocEntity(ctxt->myDoc, name); 591 } 592 if ((ret != NULL) && 593 ((ctxt->validate) || (ctxt->replaceEntities)) && 594 (ret->children == NULL) && 595 (ret->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { 596 int val; 597 598 /* 599 * for validation purposes we really need to fetch and 600 * parse the external entity 601 */ 602 xmlNodePtr children; 603 unsigned long oldnbent = ctxt->nbentities; 604 605 val = xmlParseCtxtExternalEntity(ctxt, ret->URI, 606 ret->ExternalID, &children); 607 if (val == 0) { 608 xmlAddChildList((xmlNodePtr) ret, children); 609 } else { 610 xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_PROCESSING, 611 "Failure to process entity %s\n", name, NULL); 612 ctxt->validate = 0; 613 return(NULL); 614 } 615 ret->owner = 1; 616 if (ret->checked == 0) { 617 ret->checked = (ctxt->nbentities - oldnbent + 1) * 2; 618 if ((ret->content != NULL) && (xmlStrchr(ret->content, '<'))) 619 ret->checked |= 1; 620 } 621 } 622 return(ret); 623 } 624 625 /** 626 * xmlSAX2GetParameterEntity: 627 * @ctx: the user data (XML parser context) 628 * @name: The entity name 629 * 630 * Get a parameter entity by name 631 * 632 * Returns the xmlEntityPtr if found. 633 */ 634 xmlEntityPtr 635 xmlSAX2GetParameterEntity(void *ctx, const xmlChar *name) 636 { 637 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 638 xmlEntityPtr ret; 639 640 if (ctx == NULL) return(NULL); 641 #ifdef DEBUG_SAX 642 xmlGenericError(xmlGenericErrorContext, 643 "SAX.xmlSAX2GetParameterEntity(%s)\n", name); 644 #endif 645 646 ret = xmlGetParameterEntity(ctxt->myDoc, name); 647 return(ret); 648 } 649 650 651 /** 652 * xmlSAX2EntityDecl: 653 * @ctx: the user data (XML parser context) 654 * @name: the entity name 655 * @type: the entity type 656 * @publicId: The public ID of the entity 657 * @systemId: The system ID of the entity 658 * @content: the entity value (without processing). 659 * 660 * An entity definition has been parsed 661 */ 662 void 663 xmlSAX2EntityDecl(void *ctx, const xmlChar *name, int type, 664 const xmlChar *publicId, const xmlChar *systemId, xmlChar *content) 665 { 666 xmlEntityPtr ent; 667 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 668 669 if (ctx == NULL) return; 670 #ifdef DEBUG_SAX 671 xmlGenericError(xmlGenericErrorContext, 672 "SAX.xmlSAX2EntityDecl(%s, %d, %s, %s, %s)\n", 673 name, type, publicId, systemId, content); 674 #endif 675 if (ctxt->inSubset == 1) { 676 ent = xmlAddDocEntity(ctxt->myDoc, name, type, publicId, 677 systemId, content); 678 if ((ent == NULL) && (ctxt->pedantic)) 679 xmlWarnMsg(ctxt, XML_WAR_ENTITY_REDEFINED, 680 "Entity(%s) already defined in the internal subset\n", 681 name); 682 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) { 683 xmlChar *URI; 684 const char *base = NULL; 685 686 if (ctxt->input != NULL) 687 base = ctxt->input->filename; 688 if (base == NULL) 689 base = ctxt->directory; 690 691 URI = xmlBuildURI(systemId, (const xmlChar *) base); 692 ent->URI = URI; 693 } 694 } else if (ctxt->inSubset == 2) { 695 ent = xmlAddDtdEntity(ctxt->myDoc, name, type, publicId, 696 systemId, content); 697 if ((ent == NULL) && (ctxt->pedantic) && 698 (ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 699 ctxt->sax->warning(ctxt->userData, 700 "Entity(%s) already defined in the external subset\n", name); 701 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) { 702 xmlChar *URI; 703 const char *base = NULL; 704 705 if (ctxt->input != NULL) 706 base = ctxt->input->filename; 707 if (base == NULL) 708 base = ctxt->directory; 709 710 URI = xmlBuildURI(systemId, (const xmlChar *) base); 711 ent->URI = URI; 712 } 713 } else { 714 xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_PROCESSING, 715 "SAX.xmlSAX2EntityDecl(%s) called while not in subset\n", 716 name, NULL); 717 } 718 } 719 720 /** 721 * xmlSAX2AttributeDecl: 722 * @ctx: the user data (XML parser context) 723 * @elem: the name of the element 724 * @fullname: the attribute name 725 * @type: the attribute type 726 * @def: the type of default value 727 * @defaultValue: the attribute default value 728 * @tree: the tree of enumerated value set 729 * 730 * An attribute definition has been parsed 731 */ 732 void 733 xmlSAX2AttributeDecl(void *ctx, const xmlChar *elem, const xmlChar *fullname, 734 int type, int def, const xmlChar *defaultValue, 735 xmlEnumerationPtr tree) 736 { 737 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 738 xmlAttributePtr attr; 739 xmlChar *name = NULL, *prefix = NULL; 740 741 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) 742 return; 743 744 #ifdef DEBUG_SAX 745 xmlGenericError(xmlGenericErrorContext, 746 "SAX.xmlSAX2AttributeDecl(%s, %s, %d, %d, %s, ...)\n", 747 elem, fullname, type, def, defaultValue); 748 #endif 749 if ((xmlStrEqual(fullname, BAD_CAST "xml:id")) && 750 (type != XML_ATTRIBUTE_ID)) { 751 /* 752 * Raise the error but keep the validity flag 753 */ 754 int tmp = ctxt->valid; 755 xmlErrValid(ctxt, XML_DTD_XMLID_TYPE, 756 "xml:id : attribute type should be ID\n", NULL, NULL); 757 ctxt->valid = tmp; 758 } 759 /* TODO: optimize name/prefix allocation */ 760 name = xmlSplitQName(ctxt, fullname, &prefix); 761 ctxt->vctxt.valid = 1; 762 if (ctxt->inSubset == 1) 763 attr = xmlAddAttributeDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, elem, 764 name, prefix, (xmlAttributeType) type, 765 (xmlAttributeDefault) def, defaultValue, tree); 766 else if (ctxt->inSubset == 2) 767 attr = xmlAddAttributeDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, elem, 768 name, prefix, (xmlAttributeType) type, 769 (xmlAttributeDefault) def, defaultValue, tree); 770 else { 771 xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, 772 "SAX.xmlSAX2AttributeDecl(%s) called while not in subset\n", 773 name, NULL); 774 xmlFreeEnumeration(tree); 775 return; 776 } 777 #ifdef LIBXML_VALID_ENABLED 778 if (ctxt->vctxt.valid == 0) 779 ctxt->valid = 0; 780 if ((attr != NULL) && (ctxt->validate) && (ctxt->wellFormed) && 781 (ctxt->myDoc->intSubset != NULL)) 782 ctxt->valid &= xmlValidateAttributeDecl(&ctxt->vctxt, ctxt->myDoc, 783 attr); 784 #endif /* LIBXML_VALID_ENABLED */ 785 if (prefix != NULL) 786 xmlFree(prefix); 787 if (name != NULL) 788 xmlFree(name); 789 } 790 791 /** 792 * xmlSAX2ElementDecl: 793 * @ctx: the user data (XML parser context) 794 * @name: the element name 795 * @type: the element type 796 * @content: the element value tree 797 * 798 * An element definition has been parsed 799 */ 800 void 801 xmlSAX2ElementDecl(void *ctx, const xmlChar * name, int type, 802 xmlElementContentPtr content) 803 { 804 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 805 xmlElementPtr elem = NULL; 806 807 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) 808 return; 809 810 #ifdef DEBUG_SAX 811 xmlGenericError(xmlGenericErrorContext, 812 "SAX.xmlSAX2ElementDecl(%s, %d, ...)\n", name, type); 813 #endif 814 815 if (ctxt->inSubset == 1) 816 elem = xmlAddElementDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, 817 name, (xmlElementTypeVal) type, content); 818 else if (ctxt->inSubset == 2) 819 elem = xmlAddElementDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, 820 name, (xmlElementTypeVal) type, content); 821 else { 822 xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, 823 "SAX.xmlSAX2ElementDecl(%s) called while not in subset\n", 824 name, NULL); 825 return; 826 } 827 #ifdef LIBXML_VALID_ENABLED 828 if (elem == NULL) 829 ctxt->valid = 0; 830 if (ctxt->validate && ctxt->wellFormed && 831 ctxt->myDoc && ctxt->myDoc->intSubset) 832 ctxt->valid &= 833 xmlValidateElementDecl(&ctxt->vctxt, ctxt->myDoc, elem); 834 #endif /* LIBXML_VALID_ENABLED */ 835 } 836 837 /** 838 * xmlSAX2NotationDecl: 839 * @ctx: the user data (XML parser context) 840 * @name: The name of the notation 841 * @publicId: The public ID of the entity 842 * @systemId: The system ID of the entity 843 * 844 * What to do when a notation declaration has been parsed. 845 */ 846 void 847 xmlSAX2NotationDecl(void *ctx, const xmlChar *name, 848 const xmlChar *publicId, const xmlChar *systemId) 849 { 850 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 851 xmlNotationPtr nota = NULL; 852 853 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) 854 return; 855 856 #ifdef DEBUG_SAX 857 xmlGenericError(xmlGenericErrorContext, 858 "SAX.xmlSAX2NotationDecl(%s, %s, %s)\n", name, publicId, systemId); 859 #endif 860 861 if ((publicId == NULL) && (systemId == NULL)) { 862 xmlFatalErrMsg(ctxt, XML_ERR_NOTATION_PROCESSING, 863 "SAX.xmlSAX2NotationDecl(%s) externalID or PublicID missing\n", 864 name, NULL); 865 return; 866 } else if (ctxt->inSubset == 1) 867 nota = xmlAddNotationDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, name, 868 publicId, systemId); 869 else if (ctxt->inSubset == 2) 870 nota = xmlAddNotationDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, name, 871 publicId, systemId); 872 else { 873 xmlFatalErrMsg(ctxt, XML_ERR_NOTATION_PROCESSING, 874 "SAX.xmlSAX2NotationDecl(%s) called while not in subset\n", 875 name, NULL); 876 return; 877 } 878 #ifdef LIBXML_VALID_ENABLED 879 if (nota == NULL) ctxt->valid = 0; 880 if ((ctxt->validate) && (ctxt->wellFormed) && 881 (ctxt->myDoc->intSubset != NULL)) 882 ctxt->valid &= xmlValidateNotationDecl(&ctxt->vctxt, ctxt->myDoc, 883 nota); 884 #endif /* LIBXML_VALID_ENABLED */ 885 } 886 887 /** 888 * xmlSAX2UnparsedEntityDecl: 889 * @ctx: the user data (XML parser context) 890 * @name: The name of the entity 891 * @publicId: The public ID of the entity 892 * @systemId: The system ID of the entity 893 * @notationName: the name of the notation 894 * 895 * What to do when an unparsed entity declaration is parsed 896 */ 897 void 898 xmlSAX2UnparsedEntityDecl(void *ctx, const xmlChar *name, 899 const xmlChar *publicId, const xmlChar *systemId, 900 const xmlChar *notationName) 901 { 902 xmlEntityPtr ent; 903 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 904 if (ctx == NULL) return; 905 #ifdef DEBUG_SAX 906 xmlGenericError(xmlGenericErrorContext, 907 "SAX.xmlSAX2UnparsedEntityDecl(%s, %s, %s, %s)\n", 908 name, publicId, systemId, notationName); 909 #endif 910 if (ctxt->inSubset == 1) { 911 ent = xmlAddDocEntity(ctxt->myDoc, name, 912 XML_EXTERNAL_GENERAL_UNPARSED_ENTITY, 913 publicId, systemId, notationName); 914 if ((ent == NULL) && (ctxt->pedantic) && 915 (ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 916 ctxt->sax->warning(ctxt->userData, 917 "Entity(%s) already defined in the internal subset\n", name); 918 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) { 919 xmlChar *URI; 920 const char *base = NULL; 921 922 if (ctxt->input != NULL) 923 base = ctxt->input->filename; 924 if (base == NULL) 925 base = ctxt->directory; 926 927 URI = xmlBuildURI(systemId, (const xmlChar *) base); 928 ent->URI = URI; 929 } 930 } else if (ctxt->inSubset == 2) { 931 ent = xmlAddDtdEntity(ctxt->myDoc, name, 932 XML_EXTERNAL_GENERAL_UNPARSED_ENTITY, 933 publicId, systemId, notationName); 934 if ((ent == NULL) && (ctxt->pedantic) && 935 (ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 936 ctxt->sax->warning(ctxt->userData, 937 "Entity(%s) already defined in the external subset\n", name); 938 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) { 939 xmlChar *URI; 940 const char *base = NULL; 941 942 if (ctxt->input != NULL) 943 base = ctxt->input->filename; 944 if (base == NULL) 945 base = ctxt->directory; 946 947 URI = xmlBuildURI(systemId, (const xmlChar *) base); 948 ent->URI = URI; 949 } 950 } else { 951 xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, 952 "SAX.xmlSAX2UnparsedEntityDecl(%s) called while not in subset\n", 953 name, NULL); 954 } 955 } 956 957 /** 958 * xmlSAX2SetDocumentLocator: 959 * @ctx: the user data (XML parser context) 960 * @loc: A SAX Locator 961 * 962 * Receive the document locator at startup, actually xmlDefaultSAXLocator 963 * Everything is available on the context, so this is useless in our case. 964 */ 965 void 966 xmlSAX2SetDocumentLocator(void *ctx ATTRIBUTE_UNUSED, xmlSAXLocatorPtr loc ATTRIBUTE_UNUSED) 967 { 968 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */ 969 #ifdef DEBUG_SAX 970 xmlGenericError(xmlGenericErrorContext, 971 "SAX.xmlSAX2SetDocumentLocator()\n"); 972 #endif 973 } 974 975 /** 976 * xmlSAX2StartDocument: 977 * @ctx: the user data (XML parser context) 978 * 979 * called when the document start being processed. 980 */ 981 void 982 xmlSAX2StartDocument(void *ctx) 983 { 984 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 985 xmlDocPtr doc; 986 987 if (ctx == NULL) return; 988 989 #ifdef DEBUG_SAX 990 xmlGenericError(xmlGenericErrorContext, 991 "SAX.xmlSAX2StartDocument()\n"); 992 #endif 993 if (ctxt->html) { 994 #ifdef LIBXML_HTML_ENABLED 995 if (ctxt->myDoc == NULL) 996 ctxt->myDoc = htmlNewDocNoDtD(NULL, NULL); 997 if (ctxt->myDoc == NULL) { 998 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument"); 999 return; 1000 } 1001 ctxt->myDoc->properties = XML_DOC_HTML; 1002 ctxt->myDoc->parseFlags = ctxt->options; 1003 #else 1004 xmlGenericError(xmlGenericErrorContext, 1005 "libxml2 built without HTML support\n"); 1006 ctxt->errNo = XML_ERR_INTERNAL_ERROR; 1007 ctxt->instate = XML_PARSER_EOF; 1008 ctxt->disableSAX = 1; 1009 return; 1010 #endif 1011 } else { 1012 doc = ctxt->myDoc = xmlNewDoc(ctxt->version); 1013 if (doc != NULL) { 1014 doc->properties = 0; 1015 if (ctxt->options & XML_PARSE_OLD10) 1016 doc->properties |= XML_DOC_OLD10; 1017 doc->parseFlags = ctxt->options; 1018 if (ctxt->encoding != NULL) 1019 doc->encoding = xmlStrdup(ctxt->encoding); 1020 else 1021 doc->encoding = NULL; 1022 doc->standalone = ctxt->standalone; 1023 } else { 1024 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument"); 1025 return; 1026 } 1027 if ((ctxt->dictNames) && (doc != NULL)) { 1028 doc->dict = ctxt->dict; 1029 xmlDictReference(doc->dict); 1030 } 1031 } 1032 if ((ctxt->myDoc != NULL) && (ctxt->myDoc->URL == NULL) && 1033 (ctxt->input != NULL) && (ctxt->input->filename != NULL)) { 1034 ctxt->myDoc->URL = xmlPathToURI((const xmlChar *)ctxt->input->filename); 1035 if (ctxt->myDoc->URL == NULL) 1036 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument"); 1037 } 1038 } 1039 1040 /** 1041 * xmlSAX2EndDocument: 1042 * @ctx: the user data (XML parser context) 1043 * 1044 * called when the document end has been detected. 1045 */ 1046 void 1047 xmlSAX2EndDocument(void *ctx) 1048 { 1049 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 1050 #ifdef DEBUG_SAX 1051 xmlGenericError(xmlGenericErrorContext, 1052 "SAX.xmlSAX2EndDocument()\n"); 1053 #endif 1054 if (ctx == NULL) return; 1055 #ifdef LIBXML_VALID_ENABLED 1056 if (ctxt->validate && ctxt->wellFormed && 1057 ctxt->myDoc && ctxt->myDoc->intSubset) 1058 ctxt->valid &= xmlValidateDocumentFinal(&ctxt->vctxt, ctxt->myDoc); 1059 #endif /* LIBXML_VALID_ENABLED */ 1060 1061 /* 1062 * Grab the encoding if it was added on-the-fly 1063 */ 1064 if ((ctxt->encoding != NULL) && (ctxt->myDoc != NULL) && 1065 (ctxt->myDoc->encoding == NULL)) { 1066 ctxt->myDoc->encoding = ctxt->encoding; 1067 ctxt->encoding = NULL; 1068 } 1069 if ((ctxt->inputTab != NULL) && 1070 (ctxt->inputNr > 0) && (ctxt->inputTab[0] != NULL) && 1071 (ctxt->inputTab[0]->encoding != NULL) && (ctxt->myDoc != NULL) && 1072 (ctxt->myDoc->encoding == NULL)) { 1073 ctxt->myDoc->encoding = xmlStrdup(ctxt->inputTab[0]->encoding); 1074 } 1075 if ((ctxt->charset != XML_CHAR_ENCODING_NONE) && (ctxt->myDoc != NULL) && 1076 (ctxt->myDoc->charset == XML_CHAR_ENCODING_NONE)) { 1077 ctxt->myDoc->charset = ctxt->charset; 1078 } 1079 } 1080 1081 #if defined(LIBXML_SAX1_ENABLED) || defined(LIBXML_HTML_ENABLED) || defined(LIBXML_WRITER_ENABLED) || defined(LIBXML_DOCB_ENABLED) || defined(LIBXML_LEGACY_ENABLED) 1082 /** 1083 * xmlSAX2AttributeInternal: 1084 * @ctx: the user data (XML parser context) 1085 * @fullname: The attribute name, including namespace prefix 1086 * @value: The attribute value 1087 * @prefix: the prefix on the element node 1088 * 1089 * Handle an attribute that has been read by the parser. 1090 * The default handling is to convert the attribute into an 1091 * DOM subtree and past it in a new xmlAttr element added to 1092 * the element. 1093 */ 1094 static void 1095 xmlSAX2AttributeInternal(void *ctx, const xmlChar *fullname, 1096 const xmlChar *value, const xmlChar *prefix ATTRIBUTE_UNUSED) 1097 { 1098 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 1099 xmlAttrPtr ret; 1100 xmlChar *name; 1101 xmlChar *ns; 1102 xmlChar *nval; 1103 xmlNsPtr namespace; 1104 1105 if (ctxt->html) { 1106 name = xmlStrdup(fullname); 1107 ns = NULL; 1108 namespace = NULL; 1109 } else { 1110 /* 1111 * Split the full name into a namespace prefix and the tag name 1112 */ 1113 name = xmlSplitQName(ctxt, fullname, &ns); 1114 if ((name != NULL) && (name[0] == 0)) { 1115 if (xmlStrEqual(ns, BAD_CAST "xmlns")) { 1116 xmlNsErrMsg(ctxt, XML_ERR_NS_DECL_ERROR, 1117 "invalid namespace declaration '%s'\n", 1118 fullname, NULL); 1119 } else { 1120 xmlNsWarnMsg(ctxt, XML_WAR_NS_COLUMN, 1121 "Avoid attribute ending with ':' like '%s'\n", 1122 fullname, NULL); 1123 } 1124 if (ns != NULL) 1125 xmlFree(ns); 1126 ns = NULL; 1127 xmlFree(name); 1128 name = xmlStrdup(fullname); 1129 } 1130 } 1131 if (name == NULL) { 1132 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1133 if (ns != NULL) 1134 xmlFree(ns); 1135 return; 1136 } 1137 1138 #ifdef LIBXML_HTML_ENABLED 1139 if ((ctxt->html) && 1140 (value == NULL) && (htmlIsBooleanAttr(fullname))) { 1141 nval = xmlStrdup(fullname); 1142 value = (const xmlChar *) nval; 1143 } else 1144 #endif 1145 { 1146 #ifdef LIBXML_VALID_ENABLED 1147 /* 1148 * Do the last stage of the attribute normalization 1149 * Needed for HTML too: 1150 * http://www.w3.org/TR/html4/types.html#h-6.2 1151 */ 1152 ctxt->vctxt.valid = 1; 1153 nval = xmlValidCtxtNormalizeAttributeValue(&ctxt->vctxt, 1154 ctxt->myDoc, ctxt->node, 1155 fullname, value); 1156 if (ctxt->vctxt.valid != 1) { 1157 ctxt->valid = 0; 1158 } 1159 if (nval != NULL) 1160 value = nval; 1161 #else 1162 nval = NULL; 1163 #endif /* LIBXML_VALID_ENABLED */ 1164 } 1165 1166 /* 1167 * Check whether it's a namespace definition 1168 */ 1169 if ((!ctxt->html) && (ns == NULL) && 1170 (name[0] == 'x') && (name[1] == 'm') && (name[2] == 'l') && 1171 (name[3] == 'n') && (name[4] == 's') && (name[5] == 0)) { 1172 xmlNsPtr nsret; 1173 xmlChar *val; 1174 1175 if (!ctxt->replaceEntities) { 1176 ctxt->depth++; 1177 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF, 1178 0,0,0); 1179 ctxt->depth--; 1180 if (val == NULL) { 1181 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1182 if (name != NULL) 1183 xmlFree(name); 1184 return; 1185 } 1186 } else { 1187 val = (xmlChar *) value; 1188 } 1189 1190 if (val[0] != 0) { 1191 xmlURIPtr uri; 1192 1193 uri = xmlParseURI((const char *)val); 1194 if (uri == NULL) { 1195 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 1196 ctxt->sax->warning(ctxt->userData, 1197 "xmlns: %s not a valid URI\n", val); 1198 } else { 1199 if (uri->scheme == NULL) { 1200 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 1201 ctxt->sax->warning(ctxt->userData, 1202 "xmlns: URI %s is not absolute\n", val); 1203 } 1204 xmlFreeURI(uri); 1205 } 1206 } 1207 1208 /* a default namespace definition */ 1209 nsret = xmlNewNs(ctxt->node, val, NULL); 1210 1211 #ifdef LIBXML_VALID_ENABLED 1212 /* 1213 * Validate also for namespace decls, they are attributes from 1214 * an XML-1.0 perspective 1215 */ 1216 if (nsret != NULL && ctxt->validate && ctxt->wellFormed && 1217 ctxt->myDoc && ctxt->myDoc->intSubset) 1218 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, 1219 ctxt->node, prefix, nsret, val); 1220 #endif /* LIBXML_VALID_ENABLED */ 1221 if (name != NULL) 1222 xmlFree(name); 1223 if (nval != NULL) 1224 xmlFree(nval); 1225 if (val != value) 1226 xmlFree(val); 1227 return; 1228 } 1229 if ((!ctxt->html) && 1230 (ns != NULL) && (ns[0] == 'x') && (ns[1] == 'm') && (ns[2] == 'l') && 1231 (ns[3] == 'n') && (ns[4] == 's') && (ns[5] == 0)) { 1232 xmlNsPtr nsret; 1233 xmlChar *val; 1234 1235 if (!ctxt->replaceEntities) { 1236 ctxt->depth++; 1237 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF, 1238 0,0,0); 1239 ctxt->depth--; 1240 if (val == NULL) { 1241 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1242 xmlFree(ns); 1243 if (name != NULL) 1244 xmlFree(name); 1245 return; 1246 } 1247 } else { 1248 val = (xmlChar *) value; 1249 } 1250 1251 if (val[0] == 0) { 1252 xmlNsErrMsg(ctxt, XML_NS_ERR_EMPTY, 1253 "Empty namespace name for prefix %s\n", name, NULL); 1254 } 1255 if ((ctxt->pedantic != 0) && (val[0] != 0)) { 1256 xmlURIPtr uri; 1257 1258 uri = xmlParseURI((const char *)val); 1259 if (uri == NULL) { 1260 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI, 1261 "xmlns:%s: %s not a valid URI\n", name, value); 1262 } else { 1263 if (uri->scheme == NULL) { 1264 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI_RELATIVE, 1265 "xmlns:%s: URI %s is not absolute\n", name, value); 1266 } 1267 xmlFreeURI(uri); 1268 } 1269 } 1270 1271 /* a standard namespace definition */ 1272 nsret = xmlNewNs(ctxt->node, val, name); 1273 xmlFree(ns); 1274 #ifdef LIBXML_VALID_ENABLED 1275 /* 1276 * Validate also for namespace decls, they are attributes from 1277 * an XML-1.0 perspective 1278 */ 1279 if (nsret != NULL && ctxt->validate && ctxt->wellFormed && 1280 ctxt->myDoc && ctxt->myDoc->intSubset) 1281 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, 1282 ctxt->node, prefix, nsret, value); 1283 #endif /* LIBXML_VALID_ENABLED */ 1284 if (name != NULL) 1285 xmlFree(name); 1286 if (nval != NULL) 1287 xmlFree(nval); 1288 if (val != value) 1289 xmlFree(val); 1290 return; 1291 } 1292 1293 if (ns != NULL) { 1294 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, ns); 1295 1296 if (namespace == NULL) { 1297 xmlNsErrMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 1298 "Namespace prefix %s of attribute %s is not defined\n", 1299 ns, name); 1300 } else { 1301 xmlAttrPtr prop; 1302 1303 prop = ctxt->node->properties; 1304 while (prop != NULL) { 1305 if (prop->ns != NULL) { 1306 if ((xmlStrEqual(name, prop->name)) && 1307 ((namespace == prop->ns) || 1308 (xmlStrEqual(namespace->href, prop->ns->href)))) { 1309 xmlNsErrMsg(ctxt, XML_ERR_ATTRIBUTE_REDEFINED, 1310 "Attribute %s in %s redefined\n", 1311 name, namespace->href); 1312 ctxt->wellFormed = 0; 1313 if (ctxt->recovery == 0) ctxt->disableSAX = 1; 1314 goto error; 1315 } 1316 } 1317 prop = prop->next; 1318 } 1319 } 1320 } else { 1321 namespace = NULL; 1322 } 1323 1324 /* !!!!!! <a toto:arg="" xmlns:toto="http://toto.com"> */ 1325 ret = xmlNewNsPropEatName(ctxt->node, namespace, name, NULL); 1326 1327 if (ret != NULL) { 1328 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) { 1329 xmlNodePtr tmp; 1330 1331 ret->children = xmlStringGetNodeList(ctxt->myDoc, value); 1332 tmp = ret->children; 1333 while (tmp != NULL) { 1334 tmp->parent = (xmlNodePtr) ret; 1335 if (tmp->next == NULL) 1336 ret->last = tmp; 1337 tmp = tmp->next; 1338 } 1339 } else if (value != NULL) { 1340 ret->children = xmlNewDocText(ctxt->myDoc, value); 1341 ret->last = ret->children; 1342 if (ret->children != NULL) 1343 ret->children->parent = (xmlNodePtr) ret; 1344 } 1345 } 1346 1347 #ifdef LIBXML_VALID_ENABLED 1348 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && 1349 ctxt->myDoc && ctxt->myDoc->intSubset) { 1350 1351 /* 1352 * If we don't substitute entities, the validation should be 1353 * done on a value with replaced entities anyway. 1354 */ 1355 if (!ctxt->replaceEntities) { 1356 xmlChar *val; 1357 1358 ctxt->depth++; 1359 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF, 1360 0,0,0); 1361 ctxt->depth--; 1362 1363 if (val == NULL) 1364 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 1365 ctxt->myDoc, ctxt->node, ret, value); 1366 else { 1367 xmlChar *nvalnorm; 1368 1369 /* 1370 * Do the last stage of the attribute normalization 1371 * It need to be done twice ... it's an extra burden related 1372 * to the ability to keep xmlSAX2References in attributes 1373 */ 1374 nvalnorm = xmlValidNormalizeAttributeValue(ctxt->myDoc, 1375 ctxt->node, fullname, val); 1376 if (nvalnorm != NULL) { 1377 xmlFree(val); 1378 val = nvalnorm; 1379 } 1380 1381 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 1382 ctxt->myDoc, ctxt->node, ret, val); 1383 xmlFree(val); 1384 } 1385 } else { 1386 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, ctxt->myDoc, 1387 ctxt->node, ret, value); 1388 } 1389 } else 1390 #endif /* LIBXML_VALID_ENABLED */ 1391 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && 1392 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) || 1393 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) { 1394 /* 1395 * when validating, the ID registration is done at the attribute 1396 * validation level. Otherwise we have to do specific handling here. 1397 */ 1398 if (xmlStrEqual(fullname, BAD_CAST "xml:id")) { 1399 /* 1400 * Add the xml:id value 1401 * 1402 * Open issue: normalization of the value. 1403 */ 1404 if (xmlValidateNCName(value, 1) != 0) { 1405 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE, 1406 "xml:id : attribute value %s is not an NCName\n", 1407 (const char *) value, NULL); 1408 } 1409 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret); 1410 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret)) 1411 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret); 1412 else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) 1413 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, value, ret); 1414 } 1415 1416 error: 1417 if (nval != NULL) 1418 xmlFree(nval); 1419 if (ns != NULL) 1420 xmlFree(ns); 1421 } 1422 1423 /* 1424 * xmlCheckDefaultedAttributes: 1425 * 1426 * Check defaulted attributes from the DTD 1427 */ 1428 static void 1429 xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt, const xmlChar *name, 1430 const xmlChar *prefix, const xmlChar **atts) { 1431 xmlElementPtr elemDecl; 1432 const xmlChar *att; 1433 int internal = 1; 1434 int i; 1435 1436 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->intSubset, name, prefix); 1437 if (elemDecl == NULL) { 1438 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, name, prefix); 1439 internal = 0; 1440 } 1441 1442 process_external_subset: 1443 1444 if (elemDecl != NULL) { 1445 xmlAttributePtr attr = elemDecl->attributes; 1446 /* 1447 * Check against defaulted attributes from the external subset 1448 * if the document is stamped as standalone 1449 */ 1450 if ((ctxt->myDoc->standalone == 1) && 1451 (ctxt->myDoc->extSubset != NULL) && 1452 (ctxt->validate)) { 1453 while (attr != NULL) { 1454 if ((attr->defaultValue != NULL) && 1455 (xmlGetDtdQAttrDesc(ctxt->myDoc->extSubset, 1456 attr->elem, attr->name, 1457 attr->prefix) == attr) && 1458 (xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset, 1459 attr->elem, attr->name, 1460 attr->prefix) == NULL)) { 1461 xmlChar *fulln; 1462 1463 if (attr->prefix != NULL) { 1464 fulln = xmlStrdup(attr->prefix); 1465 fulln = xmlStrcat(fulln, BAD_CAST ":"); 1466 fulln = xmlStrcat(fulln, attr->name); 1467 } else { 1468 fulln = xmlStrdup(attr->name); 1469 } 1470 if (fulln == NULL) { 1471 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1472 break; 1473 } 1474 1475 /* 1476 * Check that the attribute is not declared in the 1477 * serialization 1478 */ 1479 att = NULL; 1480 if (atts != NULL) { 1481 i = 0; 1482 att = atts[i]; 1483 while (att != NULL) { 1484 if (xmlStrEqual(att, fulln)) 1485 break; 1486 i += 2; 1487 att = atts[i]; 1488 } 1489 } 1490 if (att == NULL) { 1491 xmlErrValid(ctxt, XML_DTD_STANDALONE_DEFAULTED, 1492 "standalone: attribute %s on %s defaulted from external subset\n", 1493 (const char *)fulln, 1494 (const char *)attr->elem); 1495 } 1496 xmlFree(fulln); 1497 } 1498 attr = attr->nexth; 1499 } 1500 } 1501 1502 /* 1503 * Actually insert defaulted values when needed 1504 */ 1505 attr = elemDecl->attributes; 1506 while (attr != NULL) { 1507 /* 1508 * Make sure that attributes redefinition occuring in the 1509 * internal subset are not overriden by definitions in the 1510 * external subset. 1511 */ 1512 if (attr->defaultValue != NULL) { 1513 /* 1514 * the element should be instantiated in the tree if: 1515 * - this is a namespace prefix 1516 * - the user required for completion in the tree 1517 * like XSLT 1518 * - there isn't already an attribute definition 1519 * in the internal subset overriding it. 1520 */ 1521 if (((attr->prefix != NULL) && 1522 (xmlStrEqual(attr->prefix, BAD_CAST "xmlns"))) || 1523 ((attr->prefix == NULL) && 1524 (xmlStrEqual(attr->name, BAD_CAST "xmlns"))) || 1525 (ctxt->loadsubset & XML_COMPLETE_ATTRS)) { 1526 xmlAttributePtr tst; 1527 1528 tst = xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset, 1529 attr->elem, attr->name, 1530 attr->prefix); 1531 if ((tst == attr) || (tst == NULL)) { 1532 xmlChar fn[50]; 1533 xmlChar *fulln; 1534 1535 fulln = xmlBuildQName(attr->name, attr->prefix, fn, 50); 1536 if (fulln == NULL) { 1537 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1538 return; 1539 } 1540 1541 /* 1542 * Check that the attribute is not declared in the 1543 * serialization 1544 */ 1545 att = NULL; 1546 if (atts != NULL) { 1547 i = 0; 1548 att = atts[i]; 1549 while (att != NULL) { 1550 if (xmlStrEqual(att, fulln)) 1551 break; 1552 i += 2; 1553 att = atts[i]; 1554 } 1555 } 1556 if (att == NULL) { 1557 xmlSAX2AttributeInternal(ctxt, fulln, 1558 attr->defaultValue, prefix); 1559 } 1560 if ((fulln != fn) && (fulln != attr->name)) 1561 xmlFree(fulln); 1562 } 1563 } 1564 } 1565 attr = attr->nexth; 1566 } 1567 if (internal == 1) { 1568 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, 1569 name, prefix); 1570 internal = 0; 1571 goto process_external_subset; 1572 } 1573 } 1574 } 1575 1576 /** 1577 * xmlSAX2StartElement: 1578 * @ctx: the user data (XML parser context) 1579 * @fullname: The element name, including namespace prefix 1580 * @atts: An array of name/value attributes pairs, NULL terminated 1581 * 1582 * called when an opening tag has been processed. 1583 */ 1584 void 1585 xmlSAX2StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts) 1586 { 1587 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 1588 xmlNodePtr ret; 1589 xmlNodePtr parent; 1590 xmlNsPtr ns; 1591 xmlChar *name; 1592 xmlChar *prefix; 1593 const xmlChar *att; 1594 const xmlChar *value; 1595 int i; 1596 1597 if ((ctx == NULL) || (fullname == NULL) || (ctxt->myDoc == NULL)) return; 1598 parent = ctxt->node; 1599 #ifdef DEBUG_SAX 1600 xmlGenericError(xmlGenericErrorContext, 1601 "SAX.xmlSAX2StartElement(%s)\n", fullname); 1602 #endif 1603 1604 /* 1605 * First check on validity: 1606 */ 1607 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) && 1608 ((ctxt->myDoc->intSubset == NULL) || 1609 ((ctxt->myDoc->intSubset->notations == NULL) && 1610 (ctxt->myDoc->intSubset->elements == NULL) && 1611 (ctxt->myDoc->intSubset->attributes == NULL) && 1612 (ctxt->myDoc->intSubset->entities == NULL)))) { 1613 xmlErrValid(ctxt, XML_ERR_NO_DTD, 1614 "Validation failed: no DTD found !", NULL, NULL); 1615 ctxt->validate = 0; 1616 } 1617 1618 1619 /* 1620 * Split the full name into a namespace prefix and the tag name 1621 */ 1622 name = xmlSplitQName(ctxt, fullname, &prefix); 1623 1624 1625 /* 1626 * Note : the namespace resolution is deferred until the end of the 1627 * attributes parsing, since local namespace can be defined as 1628 * an attribute at this level. 1629 */ 1630 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, name, NULL); 1631 if (ret == NULL) { 1632 if (prefix != NULL) 1633 xmlFree(prefix); 1634 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1635 return; 1636 } 1637 if (ctxt->myDoc->children == NULL) { 1638 #ifdef DEBUG_SAX_TREE 1639 xmlGenericError(xmlGenericErrorContext, "Setting %s as root\n", name); 1640 #endif 1641 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 1642 } else if (parent == NULL) { 1643 parent = ctxt->myDoc->children; 1644 } 1645 ctxt->nodemem = -1; 1646 if (ctxt->linenumbers) { 1647 if (ctxt->input != NULL) { 1648 if (ctxt->input->line < 65535) 1649 ret->line = (short) ctxt->input->line; 1650 else 1651 ret->line = 65535; 1652 } 1653 } 1654 1655 /* 1656 * We are parsing a new node. 1657 */ 1658 #ifdef DEBUG_SAX_TREE 1659 xmlGenericError(xmlGenericErrorContext, "pushing(%s)\n", name); 1660 #endif 1661 nodePush(ctxt, ret); 1662 1663 /* 1664 * Link the child element 1665 */ 1666 if (parent != NULL) { 1667 if (parent->type == XML_ELEMENT_NODE) { 1668 #ifdef DEBUG_SAX_TREE 1669 xmlGenericError(xmlGenericErrorContext, 1670 "adding child %s to %s\n", name, parent->name); 1671 #endif 1672 xmlAddChild(parent, ret); 1673 } else { 1674 #ifdef DEBUG_SAX_TREE 1675 xmlGenericError(xmlGenericErrorContext, 1676 "adding sibling %s to ", name); 1677 xmlDebugDumpOneNode(stderr, parent, 0); 1678 #endif 1679 xmlAddSibling(parent, ret); 1680 } 1681 } 1682 1683 /* 1684 * Insert all the defaulted attributes from the DTD especially namespaces 1685 */ 1686 if ((!ctxt->html) && 1687 ((ctxt->myDoc->intSubset != NULL) || 1688 (ctxt->myDoc->extSubset != NULL))) { 1689 xmlCheckDefaultedAttributes(ctxt, name, prefix, atts); 1690 } 1691 1692 /* 1693 * process all the attributes whose name start with "xmlns" 1694 */ 1695 if (atts != NULL) { 1696 i = 0; 1697 att = atts[i++]; 1698 value = atts[i++]; 1699 if (!ctxt->html) { 1700 while ((att != NULL) && (value != NULL)) { 1701 if ((att[0] == 'x') && (att[1] == 'm') && (att[2] == 'l') && 1702 (att[3] == 'n') && (att[4] == 's')) 1703 xmlSAX2AttributeInternal(ctxt, att, value, prefix); 1704 1705 att = atts[i++]; 1706 value = atts[i++]; 1707 } 1708 } 1709 } 1710 1711 /* 1712 * Search the namespace, note that since the attributes have been 1713 * processed, the local namespaces are available. 1714 */ 1715 ns = xmlSearchNs(ctxt->myDoc, ret, prefix); 1716 if ((ns == NULL) && (parent != NULL)) 1717 ns = xmlSearchNs(ctxt->myDoc, parent, prefix); 1718 if ((prefix != NULL) && (ns == NULL)) { 1719 ns = xmlNewNs(ret, NULL, prefix); 1720 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 1721 "Namespace prefix %s is not defined\n", 1722 prefix, NULL); 1723 } 1724 1725 /* 1726 * set the namespace node, making sure that if the default namspace 1727 * is unbound on a parent we simply kee it NULL 1728 */ 1729 if ((ns != NULL) && (ns->href != NULL) && 1730 ((ns->href[0] != 0) || (ns->prefix != NULL))) 1731 xmlSetNs(ret, ns); 1732 1733 /* 1734 * process all the other attributes 1735 */ 1736 if (atts != NULL) { 1737 i = 0; 1738 att = atts[i++]; 1739 value = atts[i++]; 1740 if (ctxt->html) { 1741 while (att != NULL) { 1742 xmlSAX2AttributeInternal(ctxt, att, value, NULL); 1743 att = atts[i++]; 1744 value = atts[i++]; 1745 } 1746 } else { 1747 while ((att != NULL) && (value != NULL)) { 1748 if ((att[0] != 'x') || (att[1] != 'm') || (att[2] != 'l') || 1749 (att[3] != 'n') || (att[4] != 's')) 1750 xmlSAX2AttributeInternal(ctxt, att, value, NULL); 1751 1752 /* 1753 * Next ones 1754 */ 1755 att = atts[i++]; 1756 value = atts[i++]; 1757 } 1758 } 1759 } 1760 1761 #ifdef LIBXML_VALID_ENABLED 1762 /* 1763 * If it's the Document root, finish the DTD validation and 1764 * check the document root element for validity 1765 */ 1766 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) { 1767 int chk; 1768 1769 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc); 1770 if (chk <= 0) 1771 ctxt->valid = 0; 1772 if (chk < 0) 1773 ctxt->wellFormed = 0; 1774 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); 1775 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1; 1776 } 1777 #endif /* LIBXML_VALID_ENABLED */ 1778 1779 if (prefix != NULL) 1780 xmlFree(prefix); 1781 1782 } 1783 1784 /** 1785 * xmlSAX2EndElement: 1786 * @ctx: the user data (XML parser context) 1787 * @name: The element name 1788 * 1789 * called when the end of an element has been detected. 1790 */ 1791 void 1792 xmlSAX2EndElement(void *ctx, const xmlChar *name ATTRIBUTE_UNUSED) 1793 { 1794 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 1795 xmlNodePtr cur; 1796 1797 if (ctx == NULL) return; 1798 cur = ctxt->node; 1799 #ifdef DEBUG_SAX 1800 if (name == NULL) 1801 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(NULL)\n"); 1802 else 1803 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(%s)\n", name); 1804 #endif 1805 1806 /* Capture end position and add node */ 1807 if (cur != NULL && ctxt->record_info) { 1808 ctxt->nodeInfo->end_pos = ctxt->input->cur - ctxt->input->base; 1809 ctxt->nodeInfo->end_line = ctxt->input->line; 1810 ctxt->nodeInfo->node = cur; 1811 xmlParserAddNodeInfo(ctxt, ctxt->nodeInfo); 1812 } 1813 ctxt->nodemem = -1; 1814 1815 #ifdef LIBXML_VALID_ENABLED 1816 if (ctxt->validate && ctxt->wellFormed && 1817 ctxt->myDoc && ctxt->myDoc->intSubset) 1818 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, 1819 cur); 1820 #endif /* LIBXML_VALID_ENABLED */ 1821 1822 1823 /* 1824 * end of parsing of this node. 1825 */ 1826 #ifdef DEBUG_SAX_TREE 1827 xmlGenericError(xmlGenericErrorContext, "popping(%s)\n", cur->name); 1828 #endif 1829 nodePop(ctxt); 1830 } 1831 #endif /* LIBXML_SAX1_ENABLED || LIBXML_HTML_ENABLED || LIBXML_LEGACY_ENABLED */ 1832 1833 /* 1834 * xmlSAX2TextNode: 1835 * @ctxt: the parser context 1836 * @str: the input string 1837 * @len: the string length 1838 * 1839 * Callback for a text node 1840 * 1841 * Returns the newly allocated string or NULL if not needed or error 1842 */ 1843 static xmlNodePtr 1844 xmlSAX2TextNode(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) { 1845 xmlNodePtr ret; 1846 const xmlChar *intern = NULL; 1847 1848 /* 1849 * Allocate 1850 */ 1851 if (ctxt->freeElems != NULL) { 1852 ret = ctxt->freeElems; 1853 ctxt->freeElems = ret->next; 1854 ctxt->freeElemsNr--; 1855 } else { 1856 ret = (xmlNodePtr) xmlMalloc(sizeof(xmlNode)); 1857 } 1858 if (ret == NULL) { 1859 xmlErrMemory(ctxt, "xmlSAX2Characters"); 1860 return(NULL); 1861 } 1862 memset(ret, 0, sizeof(xmlNode)); 1863 /* 1864 * intern the formatting blanks found between tags, or the 1865 * very short strings 1866 */ 1867 if (ctxt->dictNames) { 1868 xmlChar cur = str[len]; 1869 1870 if ((len < (int) (2 * sizeof(void *))) && 1871 (ctxt->options & XML_PARSE_COMPACT)) { 1872 /* store the string in the node overriding properties and nsDef */ 1873 xmlChar *tmp = (xmlChar *) &(ret->properties); 1874 memcpy(tmp, str, len); 1875 tmp[len] = 0; 1876 intern = tmp; 1877 } else if ((len <= 3) && ((cur == '"') || (cur == '\'') || 1878 ((cur == '<') && (str[len + 1] != '!')))) { 1879 intern = xmlDictLookup(ctxt->dict, str, len); 1880 } else if (IS_BLANK_CH(*str) && (len < 60) && (cur == '<') && 1881 (str[len + 1] != '!')) { 1882 int i; 1883 1884 for (i = 1;i < len;i++) { 1885 if (!IS_BLANK_CH(str[i])) goto skip; 1886 } 1887 intern = xmlDictLookup(ctxt->dict, str, len); 1888 } 1889 } 1890 skip: 1891 ret->type = XML_TEXT_NODE; 1892 1893 ret->name = xmlStringText; 1894 if (intern == NULL) { 1895 ret->content = xmlStrndup(str, len); 1896 if (ret->content == NULL) { 1897 xmlSAX2ErrMemory(ctxt, "xmlSAX2TextNode"); 1898 xmlFree(ret); 1899 return(NULL); 1900 } 1901 } else 1902 ret->content = (xmlChar *) intern; 1903 1904 if (ctxt->linenumbers) { 1905 if (ctxt->input != NULL) { 1906 if (ctxt->input->line < 65535) 1907 ret->line = (short) ctxt->input->line; 1908 else { 1909 ret->line = 65535; 1910 if (ctxt->options & XML_PARSE_BIG_LINES) 1911 ret->psvi = (void *) (long) ctxt->input->line; 1912 } 1913 } 1914 } 1915 1916 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) 1917 xmlRegisterNodeDefaultValue(ret); 1918 return(ret); 1919 } 1920 1921 #ifdef LIBXML_VALID_ENABLED 1922 /* 1923 * xmlSAX2DecodeAttrEntities: 1924 * @ctxt: the parser context 1925 * @str: the input string 1926 * @len: the string length 1927 * 1928 * Remove the entities from an attribute value 1929 * 1930 * Returns the newly allocated string or NULL if not needed or error 1931 */ 1932 static xmlChar * 1933 xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, 1934 const xmlChar *end) { 1935 const xmlChar *in; 1936 xmlChar *ret; 1937 1938 in = str; 1939 while (in < end) 1940 if (*in++ == '&') 1941 goto decode; 1942 return(NULL); 1943 decode: 1944 ctxt->depth++; 1945 ret = xmlStringLenDecodeEntities(ctxt, str, end - str, 1946 XML_SUBSTITUTE_REF, 0,0,0); 1947 ctxt->depth--; 1948 return(ret); 1949 } 1950 #endif /* LIBXML_VALID_ENABLED */ 1951 1952 /** 1953 * xmlSAX2AttributeNs: 1954 * @ctx: the user data (XML parser context) 1955 * @localname: the local name of the attribute 1956 * @prefix: the attribute namespace prefix if available 1957 * @URI: the attribute namespace name if available 1958 * @value: Start of the attribute value 1959 * @valueend: end of the attribute value 1960 * 1961 * Handle an attribute that has been read by the parser. 1962 * The default handling is to convert the attribute into an 1963 * DOM subtree and past it in a new xmlAttr element added to 1964 * the element. 1965 */ 1966 static void 1967 xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt, 1968 const xmlChar * localname, 1969 const xmlChar * prefix, 1970 const xmlChar * value, 1971 const xmlChar * valueend) 1972 { 1973 xmlAttrPtr ret; 1974 xmlNsPtr namespace = NULL; 1975 xmlChar *dup = NULL; 1976 1977 /* 1978 * Note: if prefix == NULL, the attribute is not in the default namespace 1979 */ 1980 if (prefix != NULL) 1981 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, prefix); 1982 1983 /* 1984 * allocate the node 1985 */ 1986 if (ctxt->freeAttrs != NULL) { 1987 ret = ctxt->freeAttrs; 1988 ctxt->freeAttrs = ret->next; 1989 ctxt->freeAttrsNr--; 1990 memset(ret, 0, sizeof(xmlAttr)); 1991 ret->type = XML_ATTRIBUTE_NODE; 1992 1993 ret->parent = ctxt->node; 1994 ret->doc = ctxt->myDoc; 1995 ret->ns = namespace; 1996 1997 if (ctxt->dictNames) 1998 ret->name = localname; 1999 else 2000 ret->name = xmlStrdup(localname); 2001 2002 /* link at the end to preserv order, TODO speed up with a last */ 2003 if (ctxt->node->properties == NULL) { 2004 ctxt->node->properties = ret; 2005 } else { 2006 xmlAttrPtr prev = ctxt->node->properties; 2007 2008 while (prev->next != NULL) prev = prev->next; 2009 prev->next = ret; 2010 ret->prev = prev; 2011 } 2012 2013 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) 2014 xmlRegisterNodeDefaultValue((xmlNodePtr)ret); 2015 } else { 2016 if (ctxt->dictNames) 2017 ret = xmlNewNsPropEatName(ctxt->node, namespace, 2018 (xmlChar *) localname, NULL); 2019 else 2020 ret = xmlNewNsProp(ctxt->node, namespace, localname, NULL); 2021 if (ret == NULL) { 2022 xmlErrMemory(ctxt, "xmlSAX2AttributeNs"); 2023 return; 2024 } 2025 } 2026 2027 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) { 2028 xmlNodePtr tmp; 2029 2030 /* 2031 * We know that if there is an entity reference, then 2032 * the string has been dup'ed and terminates with 0 2033 * otherwise with ' or " 2034 */ 2035 if (*valueend != 0) { 2036 tmp = xmlSAX2TextNode(ctxt, value, valueend - value); 2037 ret->children = tmp; 2038 ret->last = tmp; 2039 if (tmp != NULL) { 2040 tmp->doc = ret->doc; 2041 tmp->parent = (xmlNodePtr) ret; 2042 } 2043 } else { 2044 ret->children = xmlStringLenGetNodeList(ctxt->myDoc, value, 2045 valueend - value); 2046 tmp = ret->children; 2047 while (tmp != NULL) { 2048 tmp->doc = ret->doc; 2049 tmp->parent = (xmlNodePtr) ret; 2050 if (tmp->next == NULL) 2051 ret->last = tmp; 2052 tmp = tmp->next; 2053 } 2054 } 2055 } else if (value != NULL) { 2056 xmlNodePtr tmp; 2057 2058 tmp = xmlSAX2TextNode(ctxt, value, valueend - value); 2059 ret->children = tmp; 2060 ret->last = tmp; 2061 if (tmp != NULL) { 2062 tmp->doc = ret->doc; 2063 tmp->parent = (xmlNodePtr) ret; 2064 } 2065 } 2066 2067 #ifdef LIBXML_VALID_ENABLED 2068 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && 2069 ctxt->myDoc && ctxt->myDoc->intSubset) { 2070 /* 2071 * If we don't substitute entities, the validation should be 2072 * done on a value with replaced entities anyway. 2073 */ 2074 if (!ctxt->replaceEntities) { 2075 dup = xmlSAX2DecodeAttrEntities(ctxt, value, valueend); 2076 if (dup == NULL) { 2077 if (*valueend == 0) { 2078 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2079 ctxt->myDoc, ctxt->node, ret, value); 2080 } else { 2081 /* 2082 * That should already be normalized. 2083 * cheaper to finally allocate here than duplicate 2084 * entry points in the full validation code 2085 */ 2086 dup = xmlStrndup(value, valueend - value); 2087 2088 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2089 ctxt->myDoc, ctxt->node, ret, dup); 2090 } 2091 } else { 2092 /* 2093 * dup now contains a string of the flattened attribute 2094 * content with entities substitued. Check if we need to 2095 * apply an extra layer of normalization. 2096 * It need to be done twice ... it's an extra burden related 2097 * to the ability to keep references in attributes 2098 */ 2099 if (ctxt->attsSpecial != NULL) { 2100 xmlChar *nvalnorm; 2101 xmlChar fn[50]; 2102 xmlChar *fullname; 2103 2104 fullname = xmlBuildQName(localname, prefix, fn, 50); 2105 if (fullname != NULL) { 2106 ctxt->vctxt.valid = 1; 2107 nvalnorm = xmlValidCtxtNormalizeAttributeValue( 2108 &ctxt->vctxt, ctxt->myDoc, 2109 ctxt->node, fullname, dup); 2110 if (ctxt->vctxt.valid != 1) 2111 ctxt->valid = 0; 2112 2113 if ((fullname != fn) && (fullname != localname)) 2114 xmlFree(fullname); 2115 if (nvalnorm != NULL) { 2116 xmlFree(dup); 2117 dup = nvalnorm; 2118 } 2119 } 2120 } 2121 2122 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2123 ctxt->myDoc, ctxt->node, ret, dup); 2124 } 2125 } else { 2126 /* 2127 * if entities already have been substitued, then 2128 * the attribute as passed is already normalized 2129 */ 2130 dup = xmlStrndup(value, valueend - value); 2131 2132 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2133 ctxt->myDoc, ctxt->node, ret, dup); 2134 } 2135 } else 2136 #endif /* LIBXML_VALID_ENABLED */ 2137 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && 2138 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) || 2139 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) { 2140 /* 2141 * when validating, the ID registration is done at the attribute 2142 * validation level. Otherwise we have to do specific handling here. 2143 */ 2144 if ((prefix == ctxt->str_xml) && 2145 (localname[0] == 'i') && (localname[1] == 'd') && 2146 (localname[2] == 0)) { 2147 /* 2148 * Add the xml:id value 2149 * 2150 * Open issue: normalization of the value. 2151 */ 2152 if (dup == NULL) 2153 dup = xmlStrndup(value, valueend - value); 2154 #if defined(LIBXML_SAX1_ENABLED) || defined(LIBXML_HTML_ENABLED) || defined(LIBXML_WRITER_ENABLED) || defined(LIBXML_DOCB_ENABLED) || defined(LIBXML_LEGACY_ENABLED) 2155 #ifdef LIBXML_VALID_ENABLED 2156 if (xmlValidateNCName(dup, 1) != 0) { 2157 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE, 2158 "xml:id : attribute value %s is not an NCName\n", 2159 (const char *) dup, NULL); 2160 } 2161 #endif 2162 #endif 2163 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret); 2164 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret)) { 2165 /* might be worth duplicate entry points and not copy */ 2166 if (dup == NULL) 2167 dup = xmlStrndup(value, valueend - value); 2168 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret); 2169 } else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) { 2170 if (dup == NULL) 2171 dup = xmlStrndup(value, valueend - value); 2172 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, dup, ret); 2173 } 2174 } 2175 if (dup != NULL) 2176 xmlFree(dup); 2177 } 2178 2179 /** 2180 * xmlSAX2StartElementNs: 2181 * @ctx: the user data (XML parser context) 2182 * @localname: the local name of the element 2183 * @prefix: the element namespace prefix if available 2184 * @URI: the element namespace name if available 2185 * @nb_namespaces: number of namespace definitions on that node 2186 * @namespaces: pointer to the array of prefix/URI pairs namespace definitions 2187 * @nb_attributes: the number of attributes on that node 2188 * @nb_defaulted: the number of defaulted attributes. 2189 * @attributes: pointer to the array of (localname/prefix/URI/value/end) 2190 * attribute values. 2191 * 2192 * SAX2 callback when an element start has been detected by the parser. 2193 * It provides the namespace informations for the element, as well as 2194 * the new namespace declarations on the element. 2195 */ 2196 void 2197 xmlSAX2StartElementNs(void *ctx, 2198 const xmlChar *localname, 2199 const xmlChar *prefix, 2200 const xmlChar *URI, 2201 int nb_namespaces, 2202 const xmlChar **namespaces, 2203 int nb_attributes, 2204 int nb_defaulted, 2205 const xmlChar **attributes) 2206 { 2207 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2208 xmlNodePtr ret; 2209 xmlNodePtr parent; 2210 xmlNsPtr last = NULL, ns; 2211 const xmlChar *uri, *pref; 2212 xmlChar *lname = NULL; 2213 int i, j; 2214 2215 if (ctx == NULL) return; 2216 parent = ctxt->node; 2217 /* 2218 * First check on validity: 2219 */ 2220 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) && 2221 ((ctxt->myDoc->intSubset == NULL) || 2222 ((ctxt->myDoc->intSubset->notations == NULL) && 2223 (ctxt->myDoc->intSubset->elements == NULL) && 2224 (ctxt->myDoc->intSubset->attributes == NULL) && 2225 (ctxt->myDoc->intSubset->entities == NULL)))) { 2226 xmlErrValid(ctxt, XML_DTD_NO_DTD, 2227 "Validation failed: no DTD found !", NULL, NULL); 2228 ctxt->validate = 0; 2229 } 2230 2231 /* 2232 * Take care of the rare case of an undefined namespace prefix 2233 */ 2234 if ((prefix != NULL) && (URI == NULL)) { 2235 if (ctxt->dictNames) { 2236 const xmlChar *fullname; 2237 2238 fullname = xmlDictQLookup(ctxt->dict, prefix, localname); 2239 if (fullname != NULL) 2240 localname = fullname; 2241 } else { 2242 lname = xmlBuildQName(localname, prefix, NULL, 0); 2243 } 2244 } 2245 /* 2246 * allocate the node 2247 */ 2248 if (ctxt->freeElems != NULL) { 2249 ret = ctxt->freeElems; 2250 ctxt->freeElems = ret->next; 2251 ctxt->freeElemsNr--; 2252 memset(ret, 0, sizeof(xmlNode)); 2253 ret->type = XML_ELEMENT_NODE; 2254 2255 if (ctxt->dictNames) 2256 ret->name = localname; 2257 else { 2258 if (lname == NULL) 2259 ret->name = xmlStrdup(localname); 2260 else 2261 ret->name = lname; 2262 if (ret->name == NULL) { 2263 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs"); 2264 return; 2265 } 2266 } 2267 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) 2268 xmlRegisterNodeDefaultValue(ret); 2269 } else { 2270 if (ctxt->dictNames) 2271 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, 2272 (xmlChar *) localname, NULL); 2273 else if (lname == NULL) 2274 ret = xmlNewDocNode(ctxt->myDoc, NULL, localname, NULL); 2275 else 2276 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, 2277 (xmlChar *) lname, NULL); 2278 if (ret == NULL) { 2279 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs"); 2280 return; 2281 } 2282 } 2283 if (ctxt->linenumbers) { 2284 if (ctxt->input != NULL) { 2285 if (ctxt->input->line < 65535) 2286 ret->line = (short) ctxt->input->line; 2287 else 2288 ret->line = 65535; 2289 } 2290 } 2291 2292 if (parent == NULL) { 2293 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 2294 } 2295 /* 2296 * Build the namespace list 2297 */ 2298 for (i = 0,j = 0;j < nb_namespaces;j++) { 2299 pref = namespaces[i++]; 2300 uri = namespaces[i++]; 2301 ns = xmlNewNs(NULL, uri, pref); 2302 if (ns != NULL) { 2303 if (last == NULL) { 2304 ret->nsDef = last = ns; 2305 } else { 2306 last->next = ns; 2307 last = ns; 2308 } 2309 if ((URI != NULL) && (prefix == pref)) 2310 ret->ns = ns; 2311 } else { 2312 /* 2313 * any out of memory error would already have been raised 2314 * but we can't be garanteed it's the actual error due to the 2315 * API, best is to skip in this case 2316 */ 2317 continue; 2318 } 2319 #ifdef LIBXML_VALID_ENABLED 2320 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && 2321 ctxt->myDoc && ctxt->myDoc->intSubset) { 2322 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, 2323 ret, prefix, ns, uri); 2324 } 2325 #endif /* LIBXML_VALID_ENABLED */ 2326 } 2327 ctxt->nodemem = -1; 2328 2329 /* 2330 * We are parsing a new node. 2331 */ 2332 nodePush(ctxt, ret); 2333 2334 /* 2335 * Link the child element 2336 */ 2337 if (parent != NULL) { 2338 if (parent->type == XML_ELEMENT_NODE) { 2339 xmlAddChild(parent, ret); 2340 } else { 2341 xmlAddSibling(parent, ret); 2342 } 2343 } 2344 2345 /* 2346 * Insert the defaulted attributes from the DTD only if requested: 2347 */ 2348 if ((nb_defaulted != 0) && 2349 ((ctxt->loadsubset & XML_COMPLETE_ATTRS) == 0)) 2350 nb_attributes -= nb_defaulted; 2351 2352 /* 2353 * Search the namespace if it wasn't already found 2354 * Note that, if prefix is NULL, this searches for the default Ns 2355 */ 2356 if ((URI != NULL) && (ret->ns == NULL)) { 2357 ret->ns = xmlSearchNs(ctxt->myDoc, parent, prefix); 2358 if ((ret->ns == NULL) && (xmlStrEqual(prefix, BAD_CAST "xml"))) { 2359 ret->ns = xmlSearchNs(ctxt->myDoc, ret, prefix); 2360 } 2361 if (ret->ns == NULL) { 2362 ns = xmlNewNs(ret, NULL, prefix); 2363 if (ns == NULL) { 2364 2365 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs"); 2366 return; 2367 } 2368 if (prefix != NULL) 2369 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 2370 "Namespace prefix %s was not found\n", 2371 prefix, NULL); 2372 else 2373 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 2374 "Namespace default prefix was not found\n", 2375 NULL, NULL); 2376 } 2377 } 2378 2379 /* 2380 * process all the other attributes 2381 */ 2382 if (nb_attributes > 0) { 2383 for (j = 0,i = 0;i < nb_attributes;i++,j+=5) { 2384 /* 2385 * Handle the rare case of an undefined atribute prefix 2386 */ 2387 if ((attributes[j+1] != NULL) && (attributes[j+2] == NULL)) { 2388 if (ctxt->dictNames) { 2389 const xmlChar *fullname; 2390 2391 fullname = xmlDictQLookup(ctxt->dict, attributes[j+1], 2392 attributes[j]); 2393 if (fullname != NULL) { 2394 xmlSAX2AttributeNs(ctxt, fullname, NULL, 2395 attributes[j+3], attributes[j+4]); 2396 continue; 2397 } 2398 } else { 2399 lname = xmlBuildQName(attributes[j], attributes[j+1], 2400 NULL, 0); 2401 if (lname != NULL) { 2402 xmlSAX2AttributeNs(ctxt, lname, NULL, 2403 attributes[j+3], attributes[j+4]); 2404 xmlFree(lname); 2405 continue; 2406 } 2407 } 2408 } 2409 xmlSAX2AttributeNs(ctxt, attributes[j], attributes[j+1], 2410 attributes[j+3], attributes[j+4]); 2411 } 2412 } 2413 2414 #ifdef LIBXML_VALID_ENABLED 2415 /* 2416 * If it's the Document root, finish the DTD validation and 2417 * check the document root element for validity 2418 */ 2419 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) { 2420 int chk; 2421 2422 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc); 2423 if (chk <= 0) 2424 ctxt->valid = 0; 2425 if (chk < 0) 2426 ctxt->wellFormed = 0; 2427 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); 2428 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1; 2429 } 2430 #endif /* LIBXML_VALID_ENABLED */ 2431 } 2432 2433 /** 2434 * xmlSAX2EndElementNs: 2435 * @ctx: the user data (XML parser context) 2436 * @localname: the local name of the element 2437 * @prefix: the element namespace prefix if available 2438 * @URI: the element namespace name if available 2439 * 2440 * SAX2 callback when an element end has been detected by the parser. 2441 * It provides the namespace informations for the element. 2442 */ 2443 void 2444 xmlSAX2EndElementNs(void *ctx, 2445 const xmlChar * localname ATTRIBUTE_UNUSED, 2446 const xmlChar * prefix ATTRIBUTE_UNUSED, 2447 const xmlChar * URI ATTRIBUTE_UNUSED) 2448 { 2449 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2450 xmlParserNodeInfo node_info; 2451 xmlNodePtr cur; 2452 2453 if (ctx == NULL) return; 2454 cur = ctxt->node; 2455 /* Capture end position and add node */ 2456 if ((ctxt->record_info) && (cur != NULL)) { 2457 node_info.end_pos = ctxt->input->cur - ctxt->input->base; 2458 node_info.end_line = ctxt->input->line; 2459 node_info.node = cur; 2460 xmlParserAddNodeInfo(ctxt, &node_info); 2461 } 2462 ctxt->nodemem = -1; 2463 2464 #ifdef LIBXML_VALID_ENABLED 2465 if (ctxt->validate && ctxt->wellFormed && 2466 ctxt->myDoc && ctxt->myDoc->intSubset) 2467 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, cur); 2468 #endif /* LIBXML_VALID_ENABLED */ 2469 2470 /* 2471 * end of parsing of this node. 2472 */ 2473 nodePop(ctxt); 2474 } 2475 2476 /** 2477 * xmlSAX2Reference: 2478 * @ctx: the user data (XML parser context) 2479 * @name: The entity name 2480 * 2481 * called when an entity xmlSAX2Reference is detected. 2482 */ 2483 void 2484 xmlSAX2Reference(void *ctx, const xmlChar *name) 2485 { 2486 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2487 xmlNodePtr ret; 2488 2489 if (ctx == NULL) return; 2490 #ifdef DEBUG_SAX 2491 xmlGenericError(xmlGenericErrorContext, 2492 "SAX.xmlSAX2Reference(%s)\n", name); 2493 #endif 2494 if (name[0] == '#') 2495 ret = xmlNewCharRef(ctxt->myDoc, name); 2496 else 2497 ret = xmlNewReference(ctxt->myDoc, name); 2498 #ifdef DEBUG_SAX_TREE 2499 xmlGenericError(xmlGenericErrorContext, 2500 "add xmlSAX2Reference %s to %s \n", name, ctxt->node->name); 2501 #endif 2502 if (xmlAddChild(ctxt->node, ret) == NULL) { 2503 xmlFreeNode(ret); 2504 } 2505 } 2506 2507 /** 2508 * xmlSAX2Characters: 2509 * @ctx: the user data (XML parser context) 2510 * @ch: a xmlChar string 2511 * @len: the number of xmlChar 2512 * 2513 * receiving some chars from the parser. 2514 */ 2515 void 2516 xmlSAX2Characters(void *ctx, const xmlChar *ch, int len) 2517 { 2518 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2519 xmlNodePtr lastChild; 2520 2521 if (ctx == NULL) return; 2522 #ifdef DEBUG_SAX 2523 xmlGenericError(xmlGenericErrorContext, 2524 "SAX.xmlSAX2Characters(%.30s, %d)\n", ch, len); 2525 #endif 2526 /* 2527 * Handle the data if any. If there is no child 2528 * add it as content, otherwise if the last child is text, 2529 * concatenate it, else create a new node of type text. 2530 */ 2531 2532 if (ctxt->node == NULL) { 2533 #ifdef DEBUG_SAX_TREE 2534 xmlGenericError(xmlGenericErrorContext, 2535 "add chars: ctxt->node == NULL !\n"); 2536 #endif 2537 return; 2538 } 2539 lastChild = ctxt->node->last; 2540 #ifdef DEBUG_SAX_TREE 2541 xmlGenericError(xmlGenericErrorContext, 2542 "add chars to %s \n", ctxt->node->name); 2543 #endif 2544 2545 /* 2546 * Here we needed an accelerator mechanism in case of very large 2547 * elements. Use an attribute in the structure !!! 2548 */ 2549 if (lastChild == NULL) { 2550 lastChild = xmlSAX2TextNode(ctxt, ch, len); 2551 if (lastChild != NULL) { 2552 ctxt->node->children = lastChild; 2553 ctxt->node->last = lastChild; 2554 lastChild->parent = ctxt->node; 2555 lastChild->doc = ctxt->node->doc; 2556 ctxt->nodelen = len; 2557 ctxt->nodemem = len + 1; 2558 } else { 2559 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters"); 2560 return; 2561 } 2562 } else { 2563 int coalesceText = (lastChild != NULL) && 2564 (lastChild->type == XML_TEXT_NODE) && 2565 (lastChild->name == xmlStringText); 2566 if ((coalesceText) && (ctxt->nodemem != 0)) { 2567 /* 2568 * The whole point of maintaining nodelen and nodemem, 2569 * xmlTextConcat is too costly, i.e. compute length, 2570 * reallocate a new buffer, move data, append ch. Here 2571 * We try to minimaze realloc() uses and avoid copying 2572 * and recomputing length over and over. 2573 */ 2574 if (lastChild->content == (xmlChar *)&(lastChild->properties)) { 2575 lastChild->content = xmlStrdup(lastChild->content); 2576 lastChild->properties = NULL; 2577 } else if ((ctxt->nodemem == ctxt->nodelen + 1) && 2578 (xmlDictOwns(ctxt->dict, lastChild->content))) { 2579 lastChild->content = xmlStrdup(lastChild->content); 2580 } 2581 if (lastChild->content == NULL) { 2582 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters: xmlStrdup returned NULL"); 2583 return; 2584 } 2585 if (((size_t)ctxt->nodelen + (size_t)len > XML_MAX_TEXT_LENGTH) && 2586 ((ctxt->options & XML_PARSE_HUGE) == 0)) { 2587 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters: huge text node"); 2588 return; 2589 } 2590 if ((size_t)ctxt->nodelen > SIZE_T_MAX - (size_t)len || 2591 (size_t)ctxt->nodemem + (size_t)len > SIZE_T_MAX / 2) { 2592 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters overflow prevented"); 2593 return; 2594 } 2595 if (ctxt->nodelen + len >= ctxt->nodemem) { 2596 xmlChar *newbuf; 2597 size_t size; 2598 2599 size = ctxt->nodemem + len; 2600 size *= 2; 2601 newbuf = (xmlChar *) xmlRealloc(lastChild->content,size); 2602 if (newbuf == NULL) { 2603 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters"); 2604 return; 2605 } 2606 ctxt->nodemem = size; 2607 lastChild->content = newbuf; 2608 } 2609 memcpy(&lastChild->content[ctxt->nodelen], ch, len); 2610 ctxt->nodelen += len; 2611 lastChild->content[ctxt->nodelen] = 0; 2612 } else if (coalesceText) { 2613 if (xmlTextConcat(lastChild, ch, len)) { 2614 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters"); 2615 } 2616 if (ctxt->node->children != NULL) { 2617 ctxt->nodelen = xmlStrlen(lastChild->content); 2618 ctxt->nodemem = ctxt->nodelen + 1; 2619 } 2620 } else { 2621 /* Mixed content, first time */ 2622 lastChild = xmlSAX2TextNode(ctxt, ch, len); 2623 if (lastChild != NULL) { 2624 xmlAddChild(ctxt->node, lastChild); 2625 if (ctxt->node->children != NULL) { 2626 ctxt->nodelen = len; 2627 ctxt->nodemem = len + 1; 2628 } 2629 } 2630 } 2631 } 2632 } 2633 2634 /** 2635 * xmlSAX2IgnorableWhitespace: 2636 * @ctx: the user data (XML parser context) 2637 * @ch: a xmlChar string 2638 * @len: the number of xmlChar 2639 * 2640 * receiving some ignorable whitespaces from the parser. 2641 * UNUSED: by default the DOM building will use xmlSAX2Characters 2642 */ 2643 void 2644 xmlSAX2IgnorableWhitespace(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch ATTRIBUTE_UNUSED, int len ATTRIBUTE_UNUSED) 2645 { 2646 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */ 2647 #ifdef DEBUG_SAX 2648 xmlGenericError(xmlGenericErrorContext, 2649 "SAX.xmlSAX2IgnorableWhitespace(%.30s, %d)\n", ch, len); 2650 #endif 2651 } 2652 2653 /** 2654 * xmlSAX2ProcessingInstruction: 2655 * @ctx: the user data (XML parser context) 2656 * @target: the target name 2657 * @data: the PI data's 2658 * 2659 * A processing instruction has been parsed. 2660 */ 2661 void 2662 xmlSAX2ProcessingInstruction(void *ctx, const xmlChar *target, 2663 const xmlChar *data) 2664 { 2665 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2666 xmlNodePtr ret; 2667 xmlNodePtr parent; 2668 2669 if (ctx == NULL) return; 2670 parent = ctxt->node; 2671 #ifdef DEBUG_SAX 2672 xmlGenericError(xmlGenericErrorContext, 2673 "SAX.xmlSAX2ProcessingInstruction(%s, %s)\n", target, data); 2674 #endif 2675 2676 ret = xmlNewDocPI(ctxt->myDoc, target, data); 2677 if (ret == NULL) return; 2678 2679 if (ctxt->linenumbers) { 2680 if (ctxt->input != NULL) { 2681 if (ctxt->input->line < 65535) 2682 ret->line = (short) ctxt->input->line; 2683 else 2684 ret->line = 65535; 2685 } 2686 } 2687 if (ctxt->inSubset == 1) { 2688 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret); 2689 return; 2690 } else if (ctxt->inSubset == 2) { 2691 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret); 2692 return; 2693 } 2694 if (parent == NULL) { 2695 #ifdef DEBUG_SAX_TREE 2696 xmlGenericError(xmlGenericErrorContext, 2697 "Setting PI %s as root\n", target); 2698 #endif 2699 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 2700 return; 2701 } 2702 if (parent->type == XML_ELEMENT_NODE) { 2703 #ifdef DEBUG_SAX_TREE 2704 xmlGenericError(xmlGenericErrorContext, 2705 "adding PI %s child to %s\n", target, parent->name); 2706 #endif 2707 xmlAddChild(parent, ret); 2708 } else { 2709 #ifdef DEBUG_SAX_TREE 2710 xmlGenericError(xmlGenericErrorContext, 2711 "adding PI %s sibling to ", target); 2712 xmlDebugDumpOneNode(stderr, parent, 0); 2713 #endif 2714 xmlAddSibling(parent, ret); 2715 } 2716 } 2717 2718 /** 2719 * xmlSAX2Comment: 2720 * @ctx: the user data (XML parser context) 2721 * @value: the xmlSAX2Comment content 2722 * 2723 * A xmlSAX2Comment has been parsed. 2724 */ 2725 void 2726 xmlSAX2Comment(void *ctx, const xmlChar *value) 2727 { 2728 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2729 xmlNodePtr ret; 2730 xmlNodePtr parent; 2731 2732 if (ctx == NULL) return; 2733 parent = ctxt->node; 2734 #ifdef DEBUG_SAX 2735 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2Comment(%s)\n", value); 2736 #endif 2737 ret = xmlNewDocComment(ctxt->myDoc, value); 2738 if (ret == NULL) return; 2739 if (ctxt->linenumbers) { 2740 if (ctxt->input != NULL) { 2741 if (ctxt->input->line < 65535) 2742 ret->line = (short) ctxt->input->line; 2743 else 2744 ret->line = 65535; 2745 } 2746 } 2747 2748 if (ctxt->inSubset == 1) { 2749 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret); 2750 return; 2751 } else if (ctxt->inSubset == 2) { 2752 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret); 2753 return; 2754 } 2755 if (parent == NULL) { 2756 #ifdef DEBUG_SAX_TREE 2757 xmlGenericError(xmlGenericErrorContext, 2758 "Setting xmlSAX2Comment as root\n"); 2759 #endif 2760 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 2761 return; 2762 } 2763 if (parent->type == XML_ELEMENT_NODE) { 2764 #ifdef DEBUG_SAX_TREE 2765 xmlGenericError(xmlGenericErrorContext, 2766 "adding xmlSAX2Comment child to %s\n", parent->name); 2767 #endif 2768 xmlAddChild(parent, ret); 2769 } else { 2770 #ifdef DEBUG_SAX_TREE 2771 xmlGenericError(xmlGenericErrorContext, 2772 "adding xmlSAX2Comment sibling to "); 2773 xmlDebugDumpOneNode(stderr, parent, 0); 2774 #endif 2775 xmlAddSibling(parent, ret); 2776 } 2777 } 2778 2779 /** 2780 * xmlSAX2CDataBlock: 2781 * @ctx: the user data (XML parser context) 2782 * @value: The pcdata content 2783 * @len: the block length 2784 * 2785 * called when a pcdata block has been parsed 2786 */ 2787 void 2788 xmlSAX2CDataBlock(void *ctx, const xmlChar *value, int len) 2789 { 2790 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2791 xmlNodePtr ret, lastChild; 2792 2793 if (ctx == NULL) return; 2794 #ifdef DEBUG_SAX 2795 xmlGenericError(xmlGenericErrorContext, 2796 "SAX.pcdata(%.10s, %d)\n", value, len); 2797 #endif 2798 lastChild = xmlGetLastChild(ctxt->node); 2799 #ifdef DEBUG_SAX_TREE 2800 xmlGenericError(xmlGenericErrorContext, 2801 "add chars to %s \n", ctxt->node->name); 2802 #endif 2803 if ((lastChild != NULL) && 2804 (lastChild->type == XML_CDATA_SECTION_NODE)) { 2805 xmlTextConcat(lastChild, value, len); 2806 } else { 2807 ret = xmlNewCDataBlock(ctxt->myDoc, value, len); 2808 xmlAddChild(ctxt->node, ret); 2809 } 2810 } 2811 2812 static int xmlSAX2DefaultVersionValue = 2; 2813 2814 #ifdef LIBXML_SAX1_ENABLED 2815 /** 2816 * xmlSAXDefaultVersion: 2817 * @version: the version, 1 or 2 2818 * 2819 * Set the default version of SAX used globally by the library. 2820 * By default, during initialization the default is set to 2. 2821 * Note that it is generally a better coding style to use 2822 * xmlSAXVersion() to set up the version explicitly for a given 2823 * parsing context. 2824 * 2825 * Returns the previous value in case of success and -1 in case of error. 2826 */ 2827 int 2828 xmlSAXDefaultVersion(int version) 2829 { 2830 int ret = xmlSAX2DefaultVersionValue; 2831 2832 if ((version != 1) && (version != 2)) 2833 return(-1); 2834 xmlSAX2DefaultVersionValue = version; 2835 return(ret); 2836 } 2837 #endif /* LIBXML_SAX1_ENABLED */ 2838 2839 /** 2840 * xmlSAXVersion: 2841 * @hdlr: the SAX handler 2842 * @version: the version, 1 or 2 2843 * 2844 * Initialize the default XML SAX handler according to the version 2845 * 2846 * Returns 0 in case of success and -1 in case of error. 2847 */ 2848 int 2849 xmlSAXVersion(xmlSAXHandler *hdlr, int version) 2850 { 2851 if (hdlr == NULL) return(-1); 2852 if (version == 2) { 2853 hdlr->startElement = NULL; 2854 hdlr->endElement = NULL; 2855 hdlr->startElementNs = xmlSAX2StartElementNs; 2856 hdlr->endElementNs = xmlSAX2EndElementNs; 2857 hdlr->serror = NULL; 2858 hdlr->initialized = XML_SAX2_MAGIC; 2859 #ifdef LIBXML_SAX1_ENABLED 2860 } else if (version == 1) { 2861 hdlr->startElement = xmlSAX2StartElement; 2862 hdlr->endElement = xmlSAX2EndElement; 2863 hdlr->initialized = 1; 2864 #endif /* LIBXML_SAX1_ENABLED */ 2865 } else 2866 return(-1); 2867 hdlr->internalSubset = xmlSAX2InternalSubset; 2868 hdlr->externalSubset = xmlSAX2ExternalSubset; 2869 hdlr->isStandalone = xmlSAX2IsStandalone; 2870 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset; 2871 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset; 2872 hdlr->resolveEntity = xmlSAX2ResolveEntity; 2873 hdlr->getEntity = xmlSAX2GetEntity; 2874 hdlr->getParameterEntity = xmlSAX2GetParameterEntity; 2875 hdlr->entityDecl = xmlSAX2EntityDecl; 2876 hdlr->attributeDecl = xmlSAX2AttributeDecl; 2877 hdlr->elementDecl = xmlSAX2ElementDecl; 2878 hdlr->notationDecl = xmlSAX2NotationDecl; 2879 hdlr->unparsedEntityDecl = xmlSAX2UnparsedEntityDecl; 2880 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; 2881 hdlr->startDocument = xmlSAX2StartDocument; 2882 hdlr->endDocument = xmlSAX2EndDocument; 2883 hdlr->reference = xmlSAX2Reference; 2884 hdlr->characters = xmlSAX2Characters; 2885 hdlr->cdataBlock = xmlSAX2CDataBlock; 2886 hdlr->ignorableWhitespace = xmlSAX2Characters; 2887 hdlr->processingInstruction = xmlSAX2ProcessingInstruction; 2888 hdlr->comment = xmlSAX2Comment; 2889 hdlr->warning = xmlParserWarning; 2890 hdlr->error = xmlParserError; 2891 hdlr->fatalError = xmlParserError; 2892 2893 return(0); 2894 } 2895 2896 /** 2897 * xmlSAX2InitDefaultSAXHandler: 2898 * @hdlr: the SAX handler 2899 * @warning: flag if non-zero sets the handler warning procedure 2900 * 2901 * Initialize the default XML SAX2 handler 2902 */ 2903 void 2904 xmlSAX2InitDefaultSAXHandler(xmlSAXHandler *hdlr, int warning) 2905 { 2906 if ((hdlr == NULL) || (hdlr->initialized != 0)) 2907 return; 2908 2909 xmlSAXVersion(hdlr, xmlSAX2DefaultVersionValue); 2910 if (warning == 0) 2911 hdlr->warning = NULL; 2912 else 2913 hdlr->warning = xmlParserWarning; 2914 } 2915 2916 /** 2917 * xmlDefaultSAXHandlerInit: 2918 * 2919 * Initialize the default SAX2 handler 2920 */ 2921 void 2922 xmlDefaultSAXHandlerInit(void) 2923 { 2924 #ifdef LIBXML_SAX1_ENABLED 2925 xmlSAXVersion((xmlSAXHandlerPtr) &xmlDefaultSAXHandler, 1); 2926 #endif /* LIBXML_SAX1_ENABLED */ 2927 } 2928 2929 #ifdef LIBXML_HTML_ENABLED 2930 2931 /** 2932 * xmlSAX2InitHtmlDefaultSAXHandler: 2933 * @hdlr: the SAX handler 2934 * 2935 * Initialize the default HTML SAX2 handler 2936 */ 2937 void 2938 xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler *hdlr) 2939 { 2940 if ((hdlr == NULL) || (hdlr->initialized != 0)) 2941 return; 2942 2943 hdlr->internalSubset = xmlSAX2InternalSubset; 2944 hdlr->externalSubset = NULL; 2945 hdlr->isStandalone = NULL; 2946 hdlr->hasInternalSubset = NULL; 2947 hdlr->hasExternalSubset = NULL; 2948 hdlr->resolveEntity = NULL; 2949 hdlr->getEntity = xmlSAX2GetEntity; 2950 hdlr->getParameterEntity = NULL; 2951 hdlr->entityDecl = NULL; 2952 hdlr->attributeDecl = NULL; 2953 hdlr->elementDecl = NULL; 2954 hdlr->notationDecl = NULL; 2955 hdlr->unparsedEntityDecl = NULL; 2956 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; 2957 hdlr->startDocument = xmlSAX2StartDocument; 2958 hdlr->endDocument = xmlSAX2EndDocument; 2959 hdlr->startElement = xmlSAX2StartElement; 2960 hdlr->endElement = xmlSAX2EndElement; 2961 hdlr->reference = NULL; 2962 hdlr->characters = xmlSAX2Characters; 2963 hdlr->cdataBlock = xmlSAX2CDataBlock; 2964 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace; 2965 hdlr->processingInstruction = xmlSAX2ProcessingInstruction; 2966 hdlr->comment = xmlSAX2Comment; 2967 hdlr->warning = xmlParserWarning; 2968 hdlr->error = xmlParserError; 2969 hdlr->fatalError = xmlParserError; 2970 2971 hdlr->initialized = 1; 2972 } 2973 2974 /** 2975 * htmlDefaultSAXHandlerInit: 2976 * 2977 * Initialize the default SAX handler 2978 */ 2979 void 2980 htmlDefaultSAXHandlerInit(void) 2981 { 2982 xmlSAX2InitHtmlDefaultSAXHandler((xmlSAXHandlerPtr) &htmlDefaultSAXHandler); 2983 } 2984 2985 #endif /* LIBXML_HTML_ENABLED */ 2986 2987 #ifdef LIBXML_DOCB_ENABLED 2988 2989 /** 2990 * xmlSAX2InitDocbDefaultSAXHandler: 2991 * @hdlr: the SAX handler 2992 * 2993 * Initialize the default DocBook SAX2 handler 2994 */ 2995 void 2996 xmlSAX2InitDocbDefaultSAXHandler(xmlSAXHandler *hdlr) 2997 { 2998 if ((hdlr == NULL) || (hdlr->initialized != 0)) 2999 return; 3000 3001 hdlr->internalSubset = xmlSAX2InternalSubset; 3002 hdlr->externalSubset = NULL; 3003 hdlr->isStandalone = xmlSAX2IsStandalone; 3004 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset; 3005 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset; 3006 hdlr->resolveEntity = xmlSAX2ResolveEntity; 3007 hdlr->getEntity = xmlSAX2GetEntity; 3008 hdlr->getParameterEntity = NULL; 3009 hdlr->entityDecl = xmlSAX2EntityDecl; 3010 hdlr->attributeDecl = NULL; 3011 hdlr->elementDecl = NULL; 3012 hdlr->notationDecl = NULL; 3013 hdlr->unparsedEntityDecl = NULL; 3014 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; 3015 hdlr->startDocument = xmlSAX2StartDocument; 3016 hdlr->endDocument = xmlSAX2EndDocument; 3017 hdlr->startElement = xmlSAX2StartElement; 3018 hdlr->endElement = xmlSAX2EndElement; 3019 hdlr->reference = xmlSAX2Reference; 3020 hdlr->characters = xmlSAX2Characters; 3021 hdlr->cdataBlock = NULL; 3022 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace; 3023 hdlr->processingInstruction = NULL; 3024 hdlr->comment = xmlSAX2Comment; 3025 hdlr->warning = xmlParserWarning; 3026 hdlr->error = xmlParserError; 3027 hdlr->fatalError = xmlParserError; 3028 3029 hdlr->initialized = 1; 3030 } 3031 3032 /** 3033 * docbDefaultSAXHandlerInit: 3034 * 3035 * Initialize the default SAX handler 3036 */ 3037 void 3038 docbDefaultSAXHandlerInit(void) 3039 { 3040 xmlSAX2InitDocbDefaultSAXHandler((xmlSAXHandlerPtr) &docbDefaultSAXHandler); 3041 } 3042 3043 #endif /* LIBXML_DOCB_ENABLED */ 3044 #define bottom_SAX2 3045 #include "elfgcchack.h" 3046
