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 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 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 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 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 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 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 ctxt->myDoc->properties = XML_DOC_HTML; 998 ctxt->myDoc->parseFlags = ctxt->options; 999 if (ctxt->myDoc == NULL) { 1000 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument"); 1001 return; 1002 } 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) 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 } else { 1181 val = (xmlChar *) value; 1182 } 1183 1184 if (val[0] != 0) { 1185 xmlURIPtr uri; 1186 1187 uri = xmlParseURI((const char *)val); 1188 if (uri == NULL) { 1189 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 1190 ctxt->sax->warning(ctxt->userData, 1191 "xmlns: %s not a valid URI\n", val); 1192 } else { 1193 if (uri->scheme == NULL) { 1194 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) 1195 ctxt->sax->warning(ctxt->userData, 1196 "xmlns: URI %s is not absolute\n", val); 1197 } 1198 xmlFreeURI(uri); 1199 } 1200 } 1201 1202 /* a default namespace definition */ 1203 nsret = xmlNewNs(ctxt->node, val, NULL); 1204 1205 #ifdef LIBXML_VALID_ENABLED 1206 /* 1207 * Validate also for namespace decls, they are attributes from 1208 * an XML-1.0 perspective 1209 */ 1210 if (nsret != NULL && ctxt->validate && ctxt->wellFormed && 1211 ctxt->myDoc && ctxt->myDoc->intSubset) 1212 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, 1213 ctxt->node, prefix, nsret, val); 1214 #endif /* LIBXML_VALID_ENABLED */ 1215 if (name != NULL) 1216 xmlFree(name); 1217 if (nval != NULL) 1218 xmlFree(nval); 1219 if (val != value) 1220 xmlFree(val); 1221 return; 1222 } 1223 if ((!ctxt->html) && 1224 (ns != NULL) && (ns[0] == 'x') && (ns[1] == 'm') && (ns[2] == 'l') && 1225 (ns[3] == 'n') && (ns[4] == 's') && (ns[5] == 0)) { 1226 xmlNsPtr nsret; 1227 xmlChar *val; 1228 1229 if (!ctxt->replaceEntities) { 1230 ctxt->depth++; 1231 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF, 1232 0,0,0); 1233 ctxt->depth--; 1234 if (val == NULL) { 1235 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1236 xmlFree(ns); 1237 if (name != NULL) 1238 xmlFree(name); 1239 return; 1240 } 1241 } else { 1242 val = (xmlChar *) value; 1243 } 1244 1245 if (val[0] == 0) { 1246 xmlNsErrMsg(ctxt, XML_NS_ERR_EMPTY, 1247 "Empty namespace name for prefix %s\n", name, NULL); 1248 } 1249 if ((ctxt->pedantic != 0) && (val[0] != 0)) { 1250 xmlURIPtr uri; 1251 1252 uri = xmlParseURI((const char *)val); 1253 if (uri == NULL) { 1254 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI, 1255 "xmlns:%s: %s not a valid URI\n", name, value); 1256 } else { 1257 if (uri->scheme == NULL) { 1258 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI_RELATIVE, 1259 "xmlns:%s: URI %s is not absolute\n", name, value); 1260 } 1261 xmlFreeURI(uri); 1262 } 1263 } 1264 1265 /* a standard namespace definition */ 1266 nsret = xmlNewNs(ctxt->node, val, name); 1267 xmlFree(ns); 1268 #ifdef LIBXML_VALID_ENABLED 1269 /* 1270 * Validate also for namespace decls, they are attributes from 1271 * an XML-1.0 perspective 1272 */ 1273 if (nsret != NULL && ctxt->validate && ctxt->wellFormed && 1274 ctxt->myDoc && ctxt->myDoc->intSubset) 1275 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, 1276 ctxt->node, prefix, nsret, value); 1277 #endif /* LIBXML_VALID_ENABLED */ 1278 if (name != NULL) 1279 xmlFree(name); 1280 if (nval != NULL) 1281 xmlFree(nval); 1282 if (val != value) 1283 xmlFree(val); 1284 return; 1285 } 1286 1287 if (ns != NULL) { 1288 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, ns); 1289 1290 if (namespace == NULL) { 1291 xmlNsErrMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 1292 "Namespace prefix %s of attribute %s is not defined\n", 1293 ns, name); 1294 } else { 1295 xmlAttrPtr prop; 1296 1297 prop = ctxt->node->properties; 1298 while (prop != NULL) { 1299 if (prop->ns != NULL) { 1300 if ((xmlStrEqual(name, prop->name)) && 1301 ((namespace == prop->ns) || 1302 (xmlStrEqual(namespace->href, prop->ns->href)))) { 1303 xmlNsErrMsg(ctxt, XML_ERR_ATTRIBUTE_REDEFINED, 1304 "Attribute %s in %s redefined\n", 1305 name, namespace->href); 1306 ctxt->wellFormed = 0; 1307 if (ctxt->recovery == 0) ctxt->disableSAX = 1; 1308 goto error; 1309 } 1310 } 1311 prop = prop->next; 1312 } 1313 } 1314 } else { 1315 namespace = NULL; 1316 } 1317 1318 /* !!!!!! <a toto:arg="" xmlns:toto="http://toto.com"> */ 1319 ret = xmlNewNsPropEatName(ctxt->node, namespace, name, NULL); 1320 1321 if (ret != NULL) { 1322 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) { 1323 xmlNodePtr tmp; 1324 1325 ret->children = xmlStringGetNodeList(ctxt->myDoc, value); 1326 tmp = ret->children; 1327 while (tmp != NULL) { 1328 tmp->parent = (xmlNodePtr) ret; 1329 if (tmp->next == NULL) 1330 ret->last = tmp; 1331 tmp = tmp->next; 1332 } 1333 } else if (value != NULL) { 1334 ret->children = xmlNewDocText(ctxt->myDoc, value); 1335 ret->last = ret->children; 1336 if (ret->children != NULL) 1337 ret->children->parent = (xmlNodePtr) ret; 1338 } 1339 } 1340 1341 #ifdef LIBXML_VALID_ENABLED 1342 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && 1343 ctxt->myDoc && ctxt->myDoc->intSubset) { 1344 1345 /* 1346 * If we don't substitute entities, the validation should be 1347 * done on a value with replaced entities anyway. 1348 */ 1349 if (!ctxt->replaceEntities) { 1350 xmlChar *val; 1351 1352 ctxt->depth++; 1353 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF, 1354 0,0,0); 1355 ctxt->depth--; 1356 1357 if (val == NULL) 1358 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 1359 ctxt->myDoc, ctxt->node, ret, value); 1360 else { 1361 xmlChar *nvalnorm; 1362 1363 /* 1364 * Do the last stage of the attribute normalization 1365 * It need to be done twice ... it's an extra burden related 1366 * to the ability to keep xmlSAX2References in attributes 1367 */ 1368 nvalnorm = xmlValidNormalizeAttributeValue(ctxt->myDoc, 1369 ctxt->node, fullname, val); 1370 if (nvalnorm != NULL) { 1371 xmlFree(val); 1372 val = nvalnorm; 1373 } 1374 1375 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 1376 ctxt->myDoc, ctxt->node, ret, val); 1377 xmlFree(val); 1378 } 1379 } else { 1380 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, ctxt->myDoc, 1381 ctxt->node, ret, value); 1382 } 1383 } else 1384 #endif /* LIBXML_VALID_ENABLED */ 1385 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && 1386 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) || 1387 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) { 1388 /* 1389 * when validating, the ID registration is done at the attribute 1390 * validation level. Otherwise we have to do specific handling here. 1391 */ 1392 if (xmlStrEqual(fullname, BAD_CAST "xml:id")) { 1393 /* 1394 * Add the xml:id value 1395 * 1396 * Open issue: normalization of the value. 1397 */ 1398 if (xmlValidateNCName(value, 1) != 0) { 1399 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE, 1400 "xml:id : attribute value %s is not an NCName\n", 1401 (const char *) value, NULL); 1402 } 1403 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret); 1404 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret)) 1405 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret); 1406 else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) 1407 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, value, ret); 1408 } 1409 1410 error: 1411 if (nval != NULL) 1412 xmlFree(nval); 1413 if (ns != NULL) 1414 xmlFree(ns); 1415 } 1416 1417 /* 1418 * xmlCheckDefaultedAttributes: 1419 * 1420 * Check defaulted attributes from the DTD 1421 */ 1422 static void 1423 xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt, const xmlChar *name, 1424 const xmlChar *prefix, const xmlChar **atts) { 1425 xmlElementPtr elemDecl; 1426 const xmlChar *att; 1427 int internal = 1; 1428 int i; 1429 1430 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->intSubset, name, prefix); 1431 if (elemDecl == NULL) { 1432 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, name, prefix); 1433 internal = 0; 1434 } 1435 1436 process_external_subset: 1437 1438 if (elemDecl != NULL) { 1439 xmlAttributePtr attr = elemDecl->attributes; 1440 /* 1441 * Check against defaulted attributes from the external subset 1442 * if the document is stamped as standalone 1443 */ 1444 if ((ctxt->myDoc->standalone == 1) && 1445 (ctxt->myDoc->extSubset != NULL) && 1446 (ctxt->validate)) { 1447 while (attr != NULL) { 1448 if ((attr->defaultValue != NULL) && 1449 (xmlGetDtdQAttrDesc(ctxt->myDoc->extSubset, 1450 attr->elem, attr->name, 1451 attr->prefix) == attr) && 1452 (xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset, 1453 attr->elem, attr->name, 1454 attr->prefix) == NULL)) { 1455 xmlChar *fulln; 1456 1457 if (attr->prefix != NULL) { 1458 fulln = xmlStrdup(attr->prefix); 1459 fulln = xmlStrcat(fulln, BAD_CAST ":"); 1460 fulln = xmlStrcat(fulln, attr->name); 1461 } else { 1462 fulln = xmlStrdup(attr->name); 1463 } 1464 if (fulln == NULL) { 1465 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1466 break; 1467 } 1468 1469 /* 1470 * Check that the attribute is not declared in the 1471 * serialization 1472 */ 1473 att = NULL; 1474 if (atts != NULL) { 1475 i = 0; 1476 att = atts[i]; 1477 while (att != NULL) { 1478 if (xmlStrEqual(att, fulln)) 1479 break; 1480 i += 2; 1481 att = atts[i]; 1482 } 1483 } 1484 if (att == NULL) { 1485 xmlErrValid(ctxt, XML_DTD_STANDALONE_DEFAULTED, 1486 "standalone: attribute %s on %s defaulted from external subset\n", 1487 (const char *)fulln, 1488 (const char *)attr->elem); 1489 } 1490 xmlFree(fulln); 1491 } 1492 attr = attr->nexth; 1493 } 1494 } 1495 1496 /* 1497 * Actually insert defaulted values when needed 1498 */ 1499 attr = elemDecl->attributes; 1500 while (attr != NULL) { 1501 /* 1502 * Make sure that attributes redefinition occuring in the 1503 * internal subset are not overriden by definitions in the 1504 * external subset. 1505 */ 1506 if (attr->defaultValue != NULL) { 1507 /* 1508 * the element should be instantiated in the tree if: 1509 * - this is a namespace prefix 1510 * - the user required for completion in the tree 1511 * like XSLT 1512 * - there isn't already an attribute definition 1513 * in the internal subset overriding it. 1514 */ 1515 if (((attr->prefix != NULL) && 1516 (xmlStrEqual(attr->prefix, BAD_CAST "xmlns"))) || 1517 ((attr->prefix == NULL) && 1518 (xmlStrEqual(attr->name, BAD_CAST "xmlns"))) || 1519 (ctxt->loadsubset & XML_COMPLETE_ATTRS)) { 1520 xmlAttributePtr tst; 1521 1522 tst = xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset, 1523 attr->elem, attr->name, 1524 attr->prefix); 1525 if ((tst == attr) || (tst == NULL)) { 1526 xmlChar fn[50]; 1527 xmlChar *fulln; 1528 1529 fulln = xmlBuildQName(attr->name, attr->prefix, fn, 50); 1530 if (fulln == NULL) { 1531 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1532 return; 1533 } 1534 1535 /* 1536 * Check that the attribute is not declared in the 1537 * serialization 1538 */ 1539 att = NULL; 1540 if (atts != NULL) { 1541 i = 0; 1542 att = atts[i]; 1543 while (att != NULL) { 1544 if (xmlStrEqual(att, fulln)) 1545 break; 1546 i += 2; 1547 att = atts[i]; 1548 } 1549 } 1550 if (att == NULL) { 1551 xmlSAX2AttributeInternal(ctxt, fulln, 1552 attr->defaultValue, prefix); 1553 } 1554 if ((fulln != fn) && (fulln != attr->name)) 1555 xmlFree(fulln); 1556 } 1557 } 1558 } 1559 attr = attr->nexth; 1560 } 1561 if (internal == 1) { 1562 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, 1563 name, prefix); 1564 internal = 0; 1565 goto process_external_subset; 1566 } 1567 } 1568 } 1569 1570 /** 1571 * xmlSAX2StartElement: 1572 * @ctx: the user data (XML parser context) 1573 * @fullname: The element name, including namespace prefix 1574 * @atts: An array of name/value attributes pairs, NULL terminated 1575 * 1576 * called when an opening tag has been processed. 1577 */ 1578 void 1579 xmlSAX2StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts) 1580 { 1581 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 1582 xmlNodePtr ret; 1583 xmlNodePtr parent; 1584 xmlNsPtr ns; 1585 xmlChar *name; 1586 xmlChar *prefix; 1587 const xmlChar *att; 1588 const xmlChar *value; 1589 int i; 1590 1591 if ((ctx == NULL) || (fullname == NULL) || (ctxt->myDoc == NULL)) return; 1592 parent = ctxt->node; 1593 #ifdef DEBUG_SAX 1594 xmlGenericError(xmlGenericErrorContext, 1595 "SAX.xmlSAX2StartElement(%s)\n", fullname); 1596 #endif 1597 1598 /* 1599 * First check on validity: 1600 */ 1601 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) && 1602 ((ctxt->myDoc->intSubset == NULL) || 1603 ((ctxt->myDoc->intSubset->notations == NULL) && 1604 (ctxt->myDoc->intSubset->elements == NULL) && 1605 (ctxt->myDoc->intSubset->attributes == NULL) && 1606 (ctxt->myDoc->intSubset->entities == NULL)))) { 1607 xmlErrValid(ctxt, XML_ERR_NO_DTD, 1608 "Validation failed: no DTD found !", NULL, NULL); 1609 ctxt->validate = 0; 1610 } 1611 1612 1613 /* 1614 * Split the full name into a namespace prefix and the tag name 1615 */ 1616 name = xmlSplitQName(ctxt, fullname, &prefix); 1617 1618 1619 /* 1620 * Note : the namespace resolution is deferred until the end of the 1621 * attributes parsing, since local namespace can be defined as 1622 * an attribute at this level. 1623 */ 1624 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, name, NULL); 1625 if (ret == NULL) { 1626 if (prefix != NULL) 1627 xmlFree(prefix); 1628 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement"); 1629 return; 1630 } 1631 if (ctxt->myDoc->children == NULL) { 1632 #ifdef DEBUG_SAX_TREE 1633 xmlGenericError(xmlGenericErrorContext, "Setting %s as root\n", name); 1634 #endif 1635 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 1636 } else if (parent == NULL) { 1637 parent = ctxt->myDoc->children; 1638 } 1639 ctxt->nodemem = -1; 1640 if (ctxt->linenumbers) { 1641 if (ctxt->input != NULL) { 1642 if (ctxt->input->line < 65535) 1643 ret->line = (short) ctxt->input->line; 1644 else 1645 ret->line = 65535; 1646 } 1647 } 1648 1649 /* 1650 * We are parsing a new node. 1651 */ 1652 #ifdef DEBUG_SAX_TREE 1653 xmlGenericError(xmlGenericErrorContext, "pushing(%s)\n", name); 1654 #endif 1655 nodePush(ctxt, ret); 1656 1657 /* 1658 * Link the child element 1659 */ 1660 if (parent != NULL) { 1661 if (parent->type == XML_ELEMENT_NODE) { 1662 #ifdef DEBUG_SAX_TREE 1663 xmlGenericError(xmlGenericErrorContext, 1664 "adding child %s to %s\n", name, parent->name); 1665 #endif 1666 xmlAddChild(parent, ret); 1667 } else { 1668 #ifdef DEBUG_SAX_TREE 1669 xmlGenericError(xmlGenericErrorContext, 1670 "adding sibling %s to ", name); 1671 xmlDebugDumpOneNode(stderr, parent, 0); 1672 #endif 1673 xmlAddSibling(parent, ret); 1674 } 1675 } 1676 1677 /* 1678 * Insert all the defaulted attributes from the DTD especially namespaces 1679 */ 1680 if ((!ctxt->html) && 1681 ((ctxt->myDoc->intSubset != NULL) || 1682 (ctxt->myDoc->extSubset != NULL))) { 1683 xmlCheckDefaultedAttributes(ctxt, name, prefix, atts); 1684 } 1685 1686 /* 1687 * process all the attributes whose name start with "xmlns" 1688 */ 1689 if (atts != NULL) { 1690 i = 0; 1691 att = atts[i++]; 1692 value = atts[i++]; 1693 if (!ctxt->html) { 1694 while ((att != NULL) && (value != NULL)) { 1695 if ((att[0] == 'x') && (att[1] == 'm') && (att[2] == 'l') && 1696 (att[3] == 'n') && (att[4] == 's')) 1697 xmlSAX2AttributeInternal(ctxt, att, value, prefix); 1698 1699 att = atts[i++]; 1700 value = atts[i++]; 1701 } 1702 } 1703 } 1704 1705 /* 1706 * Search the namespace, note that since the attributes have been 1707 * processed, the local namespaces are available. 1708 */ 1709 ns = xmlSearchNs(ctxt->myDoc, ret, prefix); 1710 if ((ns == NULL) && (parent != NULL)) 1711 ns = xmlSearchNs(ctxt->myDoc, parent, prefix); 1712 if ((prefix != NULL) && (ns == NULL)) { 1713 ns = xmlNewNs(ret, NULL, prefix); 1714 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 1715 "Namespace prefix %s is not defined\n", 1716 prefix, NULL); 1717 } 1718 1719 /* 1720 * set the namespace node, making sure that if the default namspace 1721 * is unbound on a parent we simply kee it NULL 1722 */ 1723 if ((ns != NULL) && (ns->href != NULL) && 1724 ((ns->href[0] != 0) || (ns->prefix != NULL))) 1725 xmlSetNs(ret, ns); 1726 1727 /* 1728 * process all the other attributes 1729 */ 1730 if (atts != NULL) { 1731 i = 0; 1732 att = atts[i++]; 1733 value = atts[i++]; 1734 if (ctxt->html) { 1735 while (att != NULL) { 1736 xmlSAX2AttributeInternal(ctxt, att, value, NULL); 1737 att = atts[i++]; 1738 value = atts[i++]; 1739 } 1740 } else { 1741 while ((att != NULL) && (value != NULL)) { 1742 if ((att[0] != 'x') || (att[1] != 'm') || (att[2] != 'l') || 1743 (att[3] != 'n') || (att[4] != 's')) 1744 xmlSAX2AttributeInternal(ctxt, att, value, NULL); 1745 1746 /* 1747 * Next ones 1748 */ 1749 att = atts[i++]; 1750 value = atts[i++]; 1751 } 1752 } 1753 } 1754 1755 #ifdef LIBXML_VALID_ENABLED 1756 /* 1757 * If it's the Document root, finish the DTD validation and 1758 * check the document root element for validity 1759 */ 1760 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) { 1761 int chk; 1762 1763 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc); 1764 if (chk <= 0) 1765 ctxt->valid = 0; 1766 if (chk < 0) 1767 ctxt->wellFormed = 0; 1768 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); 1769 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1; 1770 } 1771 #endif /* LIBXML_VALID_ENABLED */ 1772 1773 if (prefix != NULL) 1774 xmlFree(prefix); 1775 1776 } 1777 1778 /** 1779 * xmlSAX2EndElement: 1780 * @ctx: the user data (XML parser context) 1781 * @name: The element name 1782 * 1783 * called when the end of an element has been detected. 1784 */ 1785 void 1786 xmlSAX2EndElement(void *ctx, const xmlChar *name ATTRIBUTE_UNUSED) 1787 { 1788 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 1789 xmlNodePtr cur; 1790 1791 if (ctx == NULL) return; 1792 cur = ctxt->node; 1793 #ifdef DEBUG_SAX 1794 if (name == NULL) 1795 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(NULL)\n"); 1796 else 1797 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(%s)\n", name); 1798 #endif 1799 1800 /* Capture end position and add node */ 1801 if (cur != NULL && ctxt->record_info) { 1802 ctxt->nodeInfo->end_pos = ctxt->input->cur - ctxt->input->base; 1803 ctxt->nodeInfo->end_line = ctxt->input->line; 1804 ctxt->nodeInfo->node = cur; 1805 xmlParserAddNodeInfo(ctxt, ctxt->nodeInfo); 1806 } 1807 ctxt->nodemem = -1; 1808 1809 #ifdef LIBXML_VALID_ENABLED 1810 if (ctxt->validate && ctxt->wellFormed && 1811 ctxt->myDoc && ctxt->myDoc->intSubset) 1812 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, 1813 cur); 1814 #endif /* LIBXML_VALID_ENABLED */ 1815 1816 1817 /* 1818 * end of parsing of this node. 1819 */ 1820 #ifdef DEBUG_SAX_TREE 1821 xmlGenericError(xmlGenericErrorContext, "popping(%s)\n", cur->name); 1822 #endif 1823 nodePop(ctxt); 1824 } 1825 #endif /* LIBXML_SAX1_ENABLED || LIBXML_HTML_ENABLE */ 1826 1827 /* 1828 * xmlSAX2TextNode: 1829 * @ctxt: the parser context 1830 * @str: the input string 1831 * @len: the string length 1832 * 1833 * Callback for a text node 1834 * 1835 * Returns the newly allocated string or NULL if not needed or error 1836 */ 1837 static xmlNodePtr 1838 xmlSAX2TextNode(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) { 1839 xmlNodePtr ret; 1840 const xmlChar *intern = NULL; 1841 1842 /* 1843 * Allocate 1844 */ 1845 if (ctxt->freeElems != NULL) { 1846 ret = ctxt->freeElems; 1847 ctxt->freeElems = ret->next; 1848 ctxt->freeElemsNr--; 1849 } else { 1850 ret = (xmlNodePtr) xmlMalloc(sizeof(xmlNode)); 1851 } 1852 if (ret == NULL) { 1853 xmlErrMemory(ctxt, "xmlSAX2Characters"); 1854 return(NULL); 1855 } 1856 memset(ret, 0, sizeof(xmlNode)); 1857 /* 1858 * intern the formatting blanks found between tags, or the 1859 * very short strings 1860 */ 1861 if (ctxt->dictNames) { 1862 xmlChar cur = str[len]; 1863 1864 if ((len < (int) (2 * sizeof(void *))) && 1865 (ctxt->options & XML_PARSE_COMPACT)) { 1866 /* store the string in the node overriding properties and nsDef */ 1867 xmlChar *tmp = (xmlChar *) &(ret->properties); 1868 memcpy(tmp, str, len); 1869 tmp[len] = 0; 1870 intern = tmp; 1871 } else if ((len <= 3) && ((cur == '"') || (cur == '\'') || 1872 ((cur == '<') && (str[len + 1] != '!')))) { 1873 intern = xmlDictLookup(ctxt->dict, str, len); 1874 } else if (IS_BLANK_CH(*str) && (len < 60) && (cur == '<') && 1875 (str[len + 1] != '!')) { 1876 int i; 1877 1878 for (i = 1;i < len;i++) { 1879 if (!IS_BLANK_CH(str[i])) goto skip; 1880 } 1881 intern = xmlDictLookup(ctxt->dict, str, len); 1882 } 1883 } 1884 skip: 1885 ret->type = XML_TEXT_NODE; 1886 1887 ret->name = xmlStringText; 1888 if (intern == NULL) { 1889 ret->content = xmlStrndup(str, len); 1890 if (ret->content == NULL) { 1891 xmlSAX2ErrMemory(ctxt, "xmlSAX2TextNode"); 1892 xmlFree(ret); 1893 return(NULL); 1894 } 1895 } else 1896 ret->content = (xmlChar *) intern; 1897 1898 if (ctxt->linenumbers) { 1899 if (ctxt->input != NULL) { 1900 if (ctxt->input->line < 65535) 1901 ret->line = (short) ctxt->input->line; 1902 else { 1903 ret->line = 65535; 1904 if (ctxt->options & XML_PARSE_BIG_LINES) 1905 ret->psvi = (void *) (long) ctxt->input->line; 1906 } 1907 } 1908 } 1909 1910 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) 1911 xmlRegisterNodeDefaultValue(ret); 1912 return(ret); 1913 } 1914 1915 #ifdef LIBXML_VALID_ENABLED 1916 /* 1917 * xmlSAX2DecodeAttrEntities: 1918 * @ctxt: the parser context 1919 * @str: the input string 1920 * @len: the string length 1921 * 1922 * Remove the entities from an attribute value 1923 * 1924 * Returns the newly allocated string or NULL if not needed or error 1925 */ 1926 static xmlChar * 1927 xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, 1928 const xmlChar *end) { 1929 const xmlChar *in; 1930 xmlChar *ret; 1931 1932 in = str; 1933 while (in < end) 1934 if (*in++ == '&') 1935 goto decode; 1936 return(NULL); 1937 decode: 1938 ctxt->depth++; 1939 ret = xmlStringLenDecodeEntities(ctxt, str, end - str, 1940 XML_SUBSTITUTE_REF, 0,0,0); 1941 ctxt->depth--; 1942 return(ret); 1943 } 1944 #endif /* LIBXML_VALID_ENABLED */ 1945 1946 /** 1947 * xmlSAX2AttributeNs: 1948 * @ctx: the user data (XML parser context) 1949 * @localname: the local name of the attribute 1950 * @prefix: the attribute namespace prefix if available 1951 * @URI: the attribute namespace name if available 1952 * @value: Start of the attribute value 1953 * @valueend: end of the attribute value 1954 * 1955 * Handle an attribute that has been read by the parser. 1956 * The default handling is to convert the attribute into an 1957 * DOM subtree and past it in a new xmlAttr element added to 1958 * the element. 1959 */ 1960 static void 1961 xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt, 1962 const xmlChar * localname, 1963 const xmlChar * prefix, 1964 const xmlChar * value, 1965 const xmlChar * valueend) 1966 { 1967 xmlAttrPtr ret; 1968 xmlNsPtr namespace = NULL; 1969 xmlChar *dup = NULL; 1970 1971 /* 1972 * Note: if prefix == NULL, the attribute is not in the default namespace 1973 */ 1974 if (prefix != NULL) 1975 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, prefix); 1976 1977 /* 1978 * allocate the node 1979 */ 1980 if (ctxt->freeAttrs != NULL) { 1981 ret = ctxt->freeAttrs; 1982 ctxt->freeAttrs = ret->next; 1983 ctxt->freeAttrsNr--; 1984 memset(ret, 0, sizeof(xmlAttr)); 1985 ret->type = XML_ATTRIBUTE_NODE; 1986 1987 ret->parent = ctxt->node; 1988 ret->doc = ctxt->myDoc; 1989 ret->ns = namespace; 1990 1991 if (ctxt->dictNames) 1992 ret->name = localname; 1993 else 1994 ret->name = xmlStrdup(localname); 1995 1996 /* link at the end to preserv order, TODO speed up with a last */ 1997 if (ctxt->node->properties == NULL) { 1998 ctxt->node->properties = ret; 1999 } else { 2000 xmlAttrPtr prev = ctxt->node->properties; 2001 2002 while (prev->next != NULL) prev = prev->next; 2003 prev->next = ret; 2004 ret->prev = prev; 2005 } 2006 2007 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) 2008 xmlRegisterNodeDefaultValue((xmlNodePtr)ret); 2009 } else { 2010 if (ctxt->dictNames) 2011 ret = xmlNewNsPropEatName(ctxt->node, namespace, 2012 (xmlChar *) localname, NULL); 2013 else 2014 ret = xmlNewNsProp(ctxt->node, namespace, localname, NULL); 2015 if (ret == NULL) { 2016 xmlErrMemory(ctxt, "xmlSAX2AttributeNs"); 2017 return; 2018 } 2019 } 2020 2021 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) { 2022 xmlNodePtr tmp; 2023 2024 /* 2025 * We know that if there is an entity reference, then 2026 * the string has been dup'ed and terminates with 0 2027 * otherwise with ' or " 2028 */ 2029 if (*valueend != 0) { 2030 tmp = xmlSAX2TextNode(ctxt, value, valueend - value); 2031 ret->children = tmp; 2032 ret->last = tmp; 2033 if (tmp != NULL) { 2034 tmp->doc = ret->doc; 2035 tmp->parent = (xmlNodePtr) ret; 2036 } 2037 } else { 2038 ret->children = xmlStringLenGetNodeList(ctxt->myDoc, value, 2039 valueend - value); 2040 tmp = ret->children; 2041 while (tmp != NULL) { 2042 tmp->doc = ret->doc; 2043 tmp->parent = (xmlNodePtr) ret; 2044 if (tmp->next == NULL) 2045 ret->last = tmp; 2046 tmp = tmp->next; 2047 } 2048 } 2049 } else if (value != NULL) { 2050 xmlNodePtr tmp; 2051 2052 tmp = xmlSAX2TextNode(ctxt, value, valueend - value); 2053 ret->children = tmp; 2054 ret->last = tmp; 2055 if (tmp != NULL) { 2056 tmp->doc = ret->doc; 2057 tmp->parent = (xmlNodePtr) ret; 2058 } 2059 } 2060 2061 #ifdef LIBXML_VALID_ENABLED 2062 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && 2063 ctxt->myDoc && ctxt->myDoc->intSubset) { 2064 /* 2065 * If we don't substitute entities, the validation should be 2066 * done on a value with replaced entities anyway. 2067 */ 2068 if (!ctxt->replaceEntities) { 2069 dup = xmlSAX2DecodeAttrEntities(ctxt, value, valueend); 2070 if (dup == NULL) { 2071 if (*valueend == 0) { 2072 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2073 ctxt->myDoc, ctxt->node, ret, value); 2074 } else { 2075 /* 2076 * That should already be normalized. 2077 * cheaper to finally allocate here than duplicate 2078 * entry points in the full validation code 2079 */ 2080 dup = xmlStrndup(value, valueend - value); 2081 2082 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2083 ctxt->myDoc, ctxt->node, ret, dup); 2084 } 2085 } else { 2086 /* 2087 * dup now contains a string of the flattened attribute 2088 * content with entities substitued. Check if we need to 2089 * apply an extra layer of normalization. 2090 * It need to be done twice ... it's an extra burden related 2091 * to the ability to keep references in attributes 2092 */ 2093 if (ctxt->attsSpecial != NULL) { 2094 xmlChar *nvalnorm; 2095 xmlChar fn[50]; 2096 xmlChar *fullname; 2097 2098 fullname = xmlBuildQName(localname, prefix, fn, 50); 2099 if (fullname != NULL) { 2100 ctxt->vctxt.valid = 1; 2101 nvalnorm = xmlValidCtxtNormalizeAttributeValue( 2102 &ctxt->vctxt, ctxt->myDoc, 2103 ctxt->node, fullname, dup); 2104 if (ctxt->vctxt.valid != 1) 2105 ctxt->valid = 0; 2106 2107 if ((fullname != fn) && (fullname != localname)) 2108 xmlFree(fullname); 2109 if (nvalnorm != NULL) { 2110 xmlFree(dup); 2111 dup = nvalnorm; 2112 } 2113 } 2114 } 2115 2116 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2117 ctxt->myDoc, ctxt->node, ret, dup); 2118 } 2119 } else { 2120 /* 2121 * if entities already have been substitued, then 2122 * the attribute as passed is already normalized 2123 */ 2124 dup = xmlStrndup(value, valueend - value); 2125 2126 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, 2127 ctxt->myDoc, ctxt->node, ret, dup); 2128 } 2129 } else 2130 #endif /* LIBXML_VALID_ENABLED */ 2131 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && 2132 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) || 2133 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) { 2134 /* 2135 * when validating, the ID registration is done at the attribute 2136 * validation level. Otherwise we have to do specific handling here. 2137 */ 2138 if ((prefix == ctxt->str_xml) && 2139 (localname[0] == 'i') && (localname[1] == 'd') && 2140 (localname[2] == 0)) { 2141 /* 2142 * Add the xml:id value 2143 * 2144 * Open issue: normalization of the value. 2145 */ 2146 if (dup == NULL) 2147 dup = xmlStrndup(value, valueend - value); 2148 #ifdef LIBXML_VALID_ENABLED 2149 if (xmlValidateNCName(dup, 1) != 0) { 2150 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE, 2151 "xml:id : attribute value %s is not an NCName\n", 2152 (const char *) dup, NULL); 2153 } 2154 #endif 2155 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret); 2156 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret)) { 2157 /* might be worth duplicate entry points and not copy */ 2158 if (dup == NULL) 2159 dup = xmlStrndup(value, valueend - value); 2160 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret); 2161 } else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) { 2162 if (dup == NULL) 2163 dup = xmlStrndup(value, valueend - value); 2164 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, dup, ret); 2165 } 2166 } 2167 if (dup != NULL) 2168 xmlFree(dup); 2169 } 2170 2171 /** 2172 * xmlSAX2StartElementNs: 2173 * @ctx: the user data (XML parser context) 2174 * @localname: the local name of the element 2175 * @prefix: the element namespace prefix if available 2176 * @URI: the element namespace name if available 2177 * @nb_namespaces: number of namespace definitions on that node 2178 * @namespaces: pointer to the array of prefix/URI pairs namespace definitions 2179 * @nb_attributes: the number of attributes on that node 2180 * @nb_defaulted: the number of defaulted attributes. 2181 * @attributes: pointer to the array of (localname/prefix/URI/value/end) 2182 * attribute values. 2183 * 2184 * SAX2 callback when an element start has been detected by the parser. 2185 * It provides the namespace informations for the element, as well as 2186 * the new namespace declarations on the element. 2187 */ 2188 void 2189 xmlSAX2StartElementNs(void *ctx, 2190 const xmlChar *localname, 2191 const xmlChar *prefix, 2192 const xmlChar *URI, 2193 int nb_namespaces, 2194 const xmlChar **namespaces, 2195 int nb_attributes, 2196 int nb_defaulted, 2197 const xmlChar **attributes) 2198 { 2199 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2200 xmlNodePtr ret; 2201 xmlNodePtr parent; 2202 xmlNsPtr last = NULL, ns; 2203 const xmlChar *uri, *pref; 2204 xmlChar *lname = NULL; 2205 int i, j; 2206 2207 if (ctx == NULL) return; 2208 parent = ctxt->node; 2209 /* 2210 * First check on validity: 2211 */ 2212 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) && 2213 ((ctxt->myDoc->intSubset == NULL) || 2214 ((ctxt->myDoc->intSubset->notations == NULL) && 2215 (ctxt->myDoc->intSubset->elements == NULL) && 2216 (ctxt->myDoc->intSubset->attributes == NULL) && 2217 (ctxt->myDoc->intSubset->entities == NULL)))) { 2218 xmlErrValid(ctxt, XML_DTD_NO_DTD, 2219 "Validation failed: no DTD found !", NULL, NULL); 2220 ctxt->validate = 0; 2221 } 2222 2223 /* 2224 * Take care of the rare case of an undefined namespace prefix 2225 */ 2226 if ((prefix != NULL) && (URI == NULL)) { 2227 if (ctxt->dictNames) { 2228 const xmlChar *fullname; 2229 2230 fullname = xmlDictQLookup(ctxt->dict, prefix, localname); 2231 if (fullname != NULL) 2232 localname = fullname; 2233 } else { 2234 lname = xmlBuildQName(localname, prefix, NULL, 0); 2235 } 2236 } 2237 /* 2238 * allocate the node 2239 */ 2240 if (ctxt->freeElems != NULL) { 2241 ret = ctxt->freeElems; 2242 ctxt->freeElems = ret->next; 2243 ctxt->freeElemsNr--; 2244 memset(ret, 0, sizeof(xmlNode)); 2245 ret->type = XML_ELEMENT_NODE; 2246 2247 if (ctxt->dictNames) 2248 ret->name = localname; 2249 else { 2250 if (lname == NULL) 2251 ret->name = xmlStrdup(localname); 2252 else 2253 ret->name = lname; 2254 if (ret->name == NULL) { 2255 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs"); 2256 return; 2257 } 2258 } 2259 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) 2260 xmlRegisterNodeDefaultValue(ret); 2261 } else { 2262 if (ctxt->dictNames) 2263 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, 2264 (xmlChar *) localname, NULL); 2265 else if (lname == NULL) 2266 ret = xmlNewDocNode(ctxt->myDoc, NULL, localname, NULL); 2267 else 2268 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, 2269 (xmlChar *) lname, NULL); 2270 if (ret == NULL) { 2271 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs"); 2272 return; 2273 } 2274 } 2275 if (ctxt->linenumbers) { 2276 if (ctxt->input != NULL) { 2277 if (ctxt->input->line < 65535) 2278 ret->line = (short) ctxt->input->line; 2279 else 2280 ret->line = 65535; 2281 } 2282 } 2283 2284 if (parent == NULL) { 2285 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 2286 } 2287 /* 2288 * Build the namespace list 2289 */ 2290 for (i = 0,j = 0;j < nb_namespaces;j++) { 2291 pref = namespaces[i++]; 2292 uri = namespaces[i++]; 2293 ns = xmlNewNs(NULL, uri, pref); 2294 if (ns != NULL) { 2295 if (last == NULL) { 2296 ret->nsDef = last = ns; 2297 } else { 2298 last->next = ns; 2299 last = ns; 2300 } 2301 if ((URI != NULL) && (prefix == pref)) 2302 ret->ns = ns; 2303 } else { 2304 /* 2305 * any out of memory error would already have been raised 2306 * but we can't be garanteed it's the actual error due to the 2307 * API, best is to skip in this case 2308 */ 2309 continue; 2310 } 2311 #ifdef LIBXML_VALID_ENABLED 2312 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && 2313 ctxt->myDoc && ctxt->myDoc->intSubset) { 2314 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, 2315 ret, prefix, ns, uri); 2316 } 2317 #endif /* LIBXML_VALID_ENABLED */ 2318 } 2319 ctxt->nodemem = -1; 2320 2321 /* 2322 * We are parsing a new node. 2323 */ 2324 nodePush(ctxt, ret); 2325 2326 /* 2327 * Link the child element 2328 */ 2329 if (parent != NULL) { 2330 if (parent->type == XML_ELEMENT_NODE) { 2331 xmlAddChild(parent, ret); 2332 } else { 2333 xmlAddSibling(parent, ret); 2334 } 2335 } 2336 2337 /* 2338 * Insert the defaulted attributes from the DTD only if requested: 2339 */ 2340 if ((nb_defaulted != 0) && 2341 ((ctxt->loadsubset & XML_COMPLETE_ATTRS) == 0)) 2342 nb_attributes -= nb_defaulted; 2343 2344 /* 2345 * Search the namespace if it wasn't already found 2346 * Note that, if prefix is NULL, this searches for the default Ns 2347 */ 2348 if ((URI != NULL) && (ret->ns == NULL)) { 2349 ret->ns = xmlSearchNs(ctxt->myDoc, parent, prefix); 2350 if ((ret->ns == NULL) && (xmlStrEqual(prefix, BAD_CAST "xml"))) { 2351 ret->ns = xmlSearchNs(ctxt->myDoc, ret, prefix); 2352 } 2353 if (ret->ns == NULL) { 2354 ns = xmlNewNs(ret, NULL, prefix); 2355 if (ns == NULL) { 2356 2357 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs"); 2358 return; 2359 } 2360 if (prefix != NULL) 2361 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 2362 "Namespace prefix %s was not found\n", 2363 prefix, NULL); 2364 else 2365 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, 2366 "Namespace default prefix was not found\n", 2367 NULL, NULL); 2368 } 2369 } 2370 2371 /* 2372 * process all the other attributes 2373 */ 2374 if (nb_attributes > 0) { 2375 for (j = 0,i = 0;i < nb_attributes;i++,j+=5) { 2376 /* 2377 * Handle the rare case of an undefined atribute prefix 2378 */ 2379 if ((attributes[j+1] != NULL) && (attributes[j+2] == NULL)) { 2380 if (ctxt->dictNames) { 2381 const xmlChar *fullname; 2382 2383 fullname = xmlDictQLookup(ctxt->dict, attributes[j+1], 2384 attributes[j]); 2385 if (fullname != NULL) { 2386 xmlSAX2AttributeNs(ctxt, fullname, NULL, 2387 attributes[j+3], attributes[j+4]); 2388 continue; 2389 } 2390 } else { 2391 lname = xmlBuildQName(attributes[j], attributes[j+1], 2392 NULL, 0); 2393 if (lname != NULL) { 2394 xmlSAX2AttributeNs(ctxt, lname, NULL, 2395 attributes[j+3], attributes[j+4]); 2396 xmlFree(lname); 2397 continue; 2398 } 2399 } 2400 } 2401 xmlSAX2AttributeNs(ctxt, attributes[j], attributes[j+1], 2402 attributes[j+3], attributes[j+4]); 2403 } 2404 } 2405 2406 #ifdef LIBXML_VALID_ENABLED 2407 /* 2408 * If it's the Document root, finish the DTD validation and 2409 * check the document root element for validity 2410 */ 2411 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) { 2412 int chk; 2413 2414 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc); 2415 if (chk <= 0) 2416 ctxt->valid = 0; 2417 if (chk < 0) 2418 ctxt->wellFormed = 0; 2419 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); 2420 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1; 2421 } 2422 #endif /* LIBXML_VALID_ENABLED */ 2423 } 2424 2425 /** 2426 * xmlSAX2EndElementNs: 2427 * @ctx: the user data (XML parser context) 2428 * @localname: the local name of the element 2429 * @prefix: the element namespace prefix if available 2430 * @URI: the element namespace name if available 2431 * 2432 * SAX2 callback when an element end has been detected by the parser. 2433 * It provides the namespace informations for the element. 2434 */ 2435 void 2436 xmlSAX2EndElementNs(void *ctx, 2437 const xmlChar * localname ATTRIBUTE_UNUSED, 2438 const xmlChar * prefix ATTRIBUTE_UNUSED, 2439 const xmlChar * URI ATTRIBUTE_UNUSED) 2440 { 2441 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2442 xmlParserNodeInfo node_info; 2443 xmlNodePtr cur; 2444 2445 if (ctx == NULL) return; 2446 cur = ctxt->node; 2447 /* Capture end position and add node */ 2448 if ((ctxt->record_info) && (cur != NULL)) { 2449 node_info.end_pos = ctxt->input->cur - ctxt->input->base; 2450 node_info.end_line = ctxt->input->line; 2451 node_info.node = cur; 2452 xmlParserAddNodeInfo(ctxt, &node_info); 2453 } 2454 ctxt->nodemem = -1; 2455 2456 #ifdef LIBXML_VALID_ENABLED 2457 if (ctxt->validate && ctxt->wellFormed && 2458 ctxt->myDoc && ctxt->myDoc->intSubset) 2459 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, cur); 2460 #endif /* LIBXML_VALID_ENABLED */ 2461 2462 /* 2463 * end of parsing of this node. 2464 */ 2465 nodePop(ctxt); 2466 } 2467 2468 /** 2469 * xmlSAX2Reference: 2470 * @ctx: the user data (XML parser context) 2471 * @name: The entity name 2472 * 2473 * called when an entity xmlSAX2Reference is detected. 2474 */ 2475 void 2476 xmlSAX2Reference(void *ctx, const xmlChar *name) 2477 { 2478 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2479 xmlNodePtr ret; 2480 2481 if (ctx == NULL) return; 2482 #ifdef DEBUG_SAX 2483 xmlGenericError(xmlGenericErrorContext, 2484 "SAX.xmlSAX2Reference(%s)\n", name); 2485 #endif 2486 if (name[0] == '#') 2487 ret = xmlNewCharRef(ctxt->myDoc, name); 2488 else 2489 ret = xmlNewReference(ctxt->myDoc, name); 2490 #ifdef DEBUG_SAX_TREE 2491 xmlGenericError(xmlGenericErrorContext, 2492 "add xmlSAX2Reference %s to %s \n", name, ctxt->node->name); 2493 #endif 2494 if (xmlAddChild(ctxt->node, ret) == NULL) { 2495 xmlFreeNode(ret); 2496 } 2497 } 2498 2499 /** 2500 * xmlSAX2Characters: 2501 * @ctx: the user data (XML parser context) 2502 * @ch: a xmlChar string 2503 * @len: the number of xmlChar 2504 * 2505 * receiving some chars from the parser. 2506 */ 2507 void 2508 xmlSAX2Characters(void *ctx, const xmlChar *ch, int len) 2509 { 2510 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2511 xmlNodePtr lastChild; 2512 2513 if (ctx == NULL) return; 2514 #ifdef DEBUG_SAX 2515 xmlGenericError(xmlGenericErrorContext, 2516 "SAX.xmlSAX2Characters(%.30s, %d)\n", ch, len); 2517 #endif 2518 /* 2519 * Handle the data if any. If there is no child 2520 * add it as content, otherwise if the last child is text, 2521 * concatenate it, else create a new node of type text. 2522 */ 2523 2524 if (ctxt->node == NULL) { 2525 #ifdef DEBUG_SAX_TREE 2526 xmlGenericError(xmlGenericErrorContext, 2527 "add chars: ctxt->node == NULL !\n"); 2528 #endif 2529 return; 2530 } 2531 lastChild = ctxt->node->last; 2532 #ifdef DEBUG_SAX_TREE 2533 xmlGenericError(xmlGenericErrorContext, 2534 "add chars to %s \n", ctxt->node->name); 2535 #endif 2536 2537 /* 2538 * Here we needed an accelerator mechanism in case of very large 2539 * elements. Use an attribute in the structure !!! 2540 */ 2541 if (lastChild == NULL) { 2542 lastChild = xmlSAX2TextNode(ctxt, ch, len); 2543 if (lastChild != NULL) { 2544 ctxt->node->children = lastChild; 2545 ctxt->node->last = lastChild; 2546 lastChild->parent = ctxt->node; 2547 lastChild->doc = ctxt->node->doc; 2548 ctxt->nodelen = len; 2549 ctxt->nodemem = len + 1; 2550 } else { 2551 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters"); 2552 return; 2553 } 2554 } else { 2555 int coalesceText = (lastChild != NULL) && 2556 (lastChild->type == XML_TEXT_NODE) && 2557 (lastChild->name == xmlStringText); 2558 if ((coalesceText) && (ctxt->nodemem != 0)) { 2559 /* 2560 * The whole point of maintaining nodelen and nodemem, 2561 * xmlTextConcat is too costly, i.e. compute length, 2562 * reallocate a new buffer, move data, append ch. Here 2563 * We try to minimaze realloc() uses and avoid copying 2564 * and recomputing length over and over. 2565 */ 2566 if (lastChild->content == (xmlChar *)&(lastChild->properties)) { 2567 lastChild->content = xmlStrdup(lastChild->content); 2568 lastChild->properties = NULL; 2569 } else if ((ctxt->nodemem == ctxt->nodelen + 1) && 2570 (xmlDictOwns(ctxt->dict, lastChild->content))) { 2571 lastChild->content = xmlStrdup(lastChild->content); 2572 } 2573 if (((size_t)ctxt->nodelen + (size_t)len > XML_MAX_TEXT_LENGTH) && 2574 ((ctxt->options & XML_PARSE_HUGE) == 0)) { 2575 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters: huge text node"); 2576 return; 2577 } 2578 if ((size_t)ctxt->nodelen > SIZE_T_MAX - (size_t)len || 2579 (size_t)ctxt->nodemem + (size_t)len > SIZE_T_MAX / 2) { 2580 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters overflow prevented"); 2581 return; 2582 } 2583 if (ctxt->nodelen + len >= ctxt->nodemem) { 2584 xmlChar *newbuf; 2585 size_t size; 2586 2587 size = ctxt->nodemem + len; 2588 size *= 2; 2589 newbuf = (xmlChar *) xmlRealloc(lastChild->content,size); 2590 if (newbuf == NULL) { 2591 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters"); 2592 return; 2593 } 2594 ctxt->nodemem = size; 2595 lastChild->content = newbuf; 2596 } 2597 memcpy(&lastChild->content[ctxt->nodelen], ch, len); 2598 ctxt->nodelen += len; 2599 lastChild->content[ctxt->nodelen] = 0; 2600 } else if (coalesceText) { 2601 if (xmlTextConcat(lastChild, ch, len)) { 2602 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters"); 2603 } 2604 if (ctxt->node->children != NULL) { 2605 ctxt->nodelen = xmlStrlen(lastChild->content); 2606 ctxt->nodemem = ctxt->nodelen + 1; 2607 } 2608 } else { 2609 /* Mixed content, first time */ 2610 lastChild = xmlSAX2TextNode(ctxt, ch, len); 2611 if (lastChild != NULL) { 2612 xmlAddChild(ctxt->node, lastChild); 2613 if (ctxt->node->children != NULL) { 2614 ctxt->nodelen = len; 2615 ctxt->nodemem = len + 1; 2616 } 2617 } 2618 } 2619 } 2620 } 2621 2622 /** 2623 * xmlSAX2IgnorableWhitespace: 2624 * @ctx: the user data (XML parser context) 2625 * @ch: a xmlChar string 2626 * @len: the number of xmlChar 2627 * 2628 * receiving some ignorable whitespaces from the parser. 2629 * UNUSED: by default the DOM building will use xmlSAX2Characters 2630 */ 2631 void 2632 xmlSAX2IgnorableWhitespace(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch ATTRIBUTE_UNUSED, int len ATTRIBUTE_UNUSED) 2633 { 2634 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */ 2635 #ifdef DEBUG_SAX 2636 xmlGenericError(xmlGenericErrorContext, 2637 "SAX.xmlSAX2IgnorableWhitespace(%.30s, %d)\n", ch, len); 2638 #endif 2639 } 2640 2641 /** 2642 * xmlSAX2ProcessingInstruction: 2643 * @ctx: the user data (XML parser context) 2644 * @target: the target name 2645 * @data: the PI data's 2646 * 2647 * A processing instruction has been parsed. 2648 */ 2649 void 2650 xmlSAX2ProcessingInstruction(void *ctx, const xmlChar *target, 2651 const xmlChar *data) 2652 { 2653 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2654 xmlNodePtr ret; 2655 xmlNodePtr parent; 2656 2657 if (ctx == NULL) return; 2658 parent = ctxt->node; 2659 #ifdef DEBUG_SAX 2660 xmlGenericError(xmlGenericErrorContext, 2661 "SAX.xmlSAX2ProcessingInstruction(%s, %s)\n", target, data); 2662 #endif 2663 2664 ret = xmlNewDocPI(ctxt->myDoc, target, data); 2665 if (ret == NULL) return; 2666 2667 if (ctxt->linenumbers) { 2668 if (ctxt->input != NULL) { 2669 if (ctxt->input->line < 65535) 2670 ret->line = (short) ctxt->input->line; 2671 else 2672 ret->line = 65535; 2673 } 2674 } 2675 if (ctxt->inSubset == 1) { 2676 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret); 2677 return; 2678 } else if (ctxt->inSubset == 2) { 2679 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret); 2680 return; 2681 } 2682 if (parent == NULL) { 2683 #ifdef DEBUG_SAX_TREE 2684 xmlGenericError(xmlGenericErrorContext, 2685 "Setting PI %s as root\n", target); 2686 #endif 2687 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 2688 return; 2689 } 2690 if (parent->type == XML_ELEMENT_NODE) { 2691 #ifdef DEBUG_SAX_TREE 2692 xmlGenericError(xmlGenericErrorContext, 2693 "adding PI %s child to %s\n", target, parent->name); 2694 #endif 2695 xmlAddChild(parent, ret); 2696 } else { 2697 #ifdef DEBUG_SAX_TREE 2698 xmlGenericError(xmlGenericErrorContext, 2699 "adding PI %s sibling to ", target); 2700 xmlDebugDumpOneNode(stderr, parent, 0); 2701 #endif 2702 xmlAddSibling(parent, ret); 2703 } 2704 } 2705 2706 /** 2707 * xmlSAX2Comment: 2708 * @ctx: the user data (XML parser context) 2709 * @value: the xmlSAX2Comment content 2710 * 2711 * A xmlSAX2Comment has been parsed. 2712 */ 2713 void 2714 xmlSAX2Comment(void *ctx, const xmlChar *value) 2715 { 2716 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2717 xmlNodePtr ret; 2718 xmlNodePtr parent; 2719 2720 if (ctx == NULL) return; 2721 parent = ctxt->node; 2722 #ifdef DEBUG_SAX 2723 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2Comment(%s)\n", value); 2724 #endif 2725 ret = xmlNewDocComment(ctxt->myDoc, value); 2726 if (ret == NULL) return; 2727 if (ctxt->linenumbers) { 2728 if (ctxt->input != NULL) { 2729 if (ctxt->input->line < 65535) 2730 ret->line = (short) ctxt->input->line; 2731 else 2732 ret->line = 65535; 2733 } 2734 } 2735 2736 if (ctxt->inSubset == 1) { 2737 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret); 2738 return; 2739 } else if (ctxt->inSubset == 2) { 2740 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret); 2741 return; 2742 } 2743 if (parent == NULL) { 2744 #ifdef DEBUG_SAX_TREE 2745 xmlGenericError(xmlGenericErrorContext, 2746 "Setting xmlSAX2Comment as root\n"); 2747 #endif 2748 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret); 2749 return; 2750 } 2751 if (parent->type == XML_ELEMENT_NODE) { 2752 #ifdef DEBUG_SAX_TREE 2753 xmlGenericError(xmlGenericErrorContext, 2754 "adding xmlSAX2Comment child to %s\n", parent->name); 2755 #endif 2756 xmlAddChild(parent, ret); 2757 } else { 2758 #ifdef DEBUG_SAX_TREE 2759 xmlGenericError(xmlGenericErrorContext, 2760 "adding xmlSAX2Comment sibling to "); 2761 xmlDebugDumpOneNode(stderr, parent, 0); 2762 #endif 2763 xmlAddSibling(parent, ret); 2764 } 2765 } 2766 2767 /** 2768 * xmlSAX2CDataBlock: 2769 * @ctx: the user data (XML parser context) 2770 * @value: The pcdata content 2771 * @len: the block length 2772 * 2773 * called when a pcdata block has been parsed 2774 */ 2775 void 2776 xmlSAX2CDataBlock(void *ctx, const xmlChar *value, int len) 2777 { 2778 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; 2779 xmlNodePtr ret, lastChild; 2780 2781 if (ctx == NULL) return; 2782 #ifdef DEBUG_SAX 2783 xmlGenericError(xmlGenericErrorContext, 2784 "SAX.pcdata(%.10s, %d)\n", value, len); 2785 #endif 2786 lastChild = xmlGetLastChild(ctxt->node); 2787 #ifdef DEBUG_SAX_TREE 2788 xmlGenericError(xmlGenericErrorContext, 2789 "add chars to %s \n", ctxt->node->name); 2790 #endif 2791 if ((lastChild != NULL) && 2792 (lastChild->type == XML_CDATA_SECTION_NODE)) { 2793 xmlTextConcat(lastChild, value, len); 2794 } else { 2795 ret = xmlNewCDataBlock(ctxt->myDoc, value, len); 2796 xmlAddChild(ctxt->node, ret); 2797 } 2798 } 2799 2800 static int xmlSAX2DefaultVersionValue = 2; 2801 2802 #ifdef LIBXML_SAX1_ENABLED 2803 /** 2804 * xmlSAXDefaultVersion: 2805 * @version: the version, 1 or 2 2806 * 2807 * Set the default version of SAX used globally by the library. 2808 * By default, during initialization the default is set to 2. 2809 * Note that it is generally a better coding style to use 2810 * xmlSAXVersion() to set up the version explicitly for a given 2811 * parsing context. 2812 * 2813 * Returns the previous value in case of success and -1 in case of error. 2814 */ 2815 int 2816 xmlSAXDefaultVersion(int version) 2817 { 2818 int ret = xmlSAX2DefaultVersionValue; 2819 2820 if ((version != 1) && (version != 2)) 2821 return(-1); 2822 xmlSAX2DefaultVersionValue = version; 2823 return(ret); 2824 } 2825 #endif /* LIBXML_SAX1_ENABLED */ 2826 2827 /** 2828 * xmlSAXVersion: 2829 * @hdlr: the SAX handler 2830 * @version: the version, 1 or 2 2831 * 2832 * Initialize the default XML SAX handler according to the version 2833 * 2834 * Returns 0 in case of success and -1 in case of error. 2835 */ 2836 int 2837 xmlSAXVersion(xmlSAXHandler *hdlr, int version) 2838 { 2839 if (hdlr == NULL) return(-1); 2840 if (version == 2) { 2841 hdlr->startElement = NULL; 2842 hdlr->endElement = NULL; 2843 hdlr->startElementNs = xmlSAX2StartElementNs; 2844 hdlr->endElementNs = xmlSAX2EndElementNs; 2845 hdlr->serror = NULL; 2846 hdlr->initialized = XML_SAX2_MAGIC; 2847 #ifdef LIBXML_SAX1_ENABLED 2848 } else if (version == 1) { 2849 hdlr->startElement = xmlSAX2StartElement; 2850 hdlr->endElement = xmlSAX2EndElement; 2851 hdlr->initialized = 1; 2852 #endif /* LIBXML_SAX1_ENABLED */ 2853 } else 2854 return(-1); 2855 hdlr->internalSubset = xmlSAX2InternalSubset; 2856 hdlr->externalSubset = xmlSAX2ExternalSubset; 2857 hdlr->isStandalone = xmlSAX2IsStandalone; 2858 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset; 2859 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset; 2860 hdlr->resolveEntity = xmlSAX2ResolveEntity; 2861 hdlr->getEntity = xmlSAX2GetEntity; 2862 hdlr->getParameterEntity = xmlSAX2GetParameterEntity; 2863 hdlr->entityDecl = xmlSAX2EntityDecl; 2864 hdlr->attributeDecl = xmlSAX2AttributeDecl; 2865 hdlr->elementDecl = xmlSAX2ElementDecl; 2866 hdlr->notationDecl = xmlSAX2NotationDecl; 2867 hdlr->unparsedEntityDecl = xmlSAX2UnparsedEntityDecl; 2868 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; 2869 hdlr->startDocument = xmlSAX2StartDocument; 2870 hdlr->endDocument = xmlSAX2EndDocument; 2871 hdlr->reference = xmlSAX2Reference; 2872 hdlr->characters = xmlSAX2Characters; 2873 hdlr->cdataBlock = xmlSAX2CDataBlock; 2874 hdlr->ignorableWhitespace = xmlSAX2Characters; 2875 hdlr->processingInstruction = xmlSAX2ProcessingInstruction; 2876 hdlr->comment = xmlSAX2Comment; 2877 hdlr->warning = xmlParserWarning; 2878 hdlr->error = xmlParserError; 2879 hdlr->fatalError = xmlParserError; 2880 2881 return(0); 2882 } 2883 2884 /** 2885 * xmlSAX2InitDefaultSAXHandler: 2886 * @hdlr: the SAX handler 2887 * @warning: flag if non-zero sets the handler warning procedure 2888 * 2889 * Initialize the default XML SAX2 handler 2890 */ 2891 void 2892 xmlSAX2InitDefaultSAXHandler(xmlSAXHandler *hdlr, int warning) 2893 { 2894 if ((hdlr == NULL) || (hdlr->initialized != 0)) 2895 return; 2896 2897 xmlSAXVersion(hdlr, xmlSAX2DefaultVersionValue); 2898 if (warning == 0) 2899 hdlr->warning = NULL; 2900 else 2901 hdlr->warning = xmlParserWarning; 2902 } 2903 2904 /** 2905 * xmlDefaultSAXHandlerInit: 2906 * 2907 * Initialize the default SAX2 handler 2908 */ 2909 void 2910 xmlDefaultSAXHandlerInit(void) 2911 { 2912 #ifdef LIBXML_SAX1_ENABLED 2913 xmlSAXVersion((xmlSAXHandlerPtr) &xmlDefaultSAXHandler, 1); 2914 #endif /* LIBXML_SAX1_ENABLED */ 2915 } 2916 2917 #ifdef LIBXML_HTML_ENABLED 2918 2919 /** 2920 * xmlSAX2InitHtmlDefaultSAXHandler: 2921 * @hdlr: the SAX handler 2922 * 2923 * Initialize the default HTML SAX2 handler 2924 */ 2925 void 2926 xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler *hdlr) 2927 { 2928 if ((hdlr == NULL) || (hdlr->initialized != 0)) 2929 return; 2930 2931 hdlr->internalSubset = xmlSAX2InternalSubset; 2932 hdlr->externalSubset = NULL; 2933 hdlr->isStandalone = NULL; 2934 hdlr->hasInternalSubset = NULL; 2935 hdlr->hasExternalSubset = NULL; 2936 hdlr->resolveEntity = NULL; 2937 hdlr->getEntity = xmlSAX2GetEntity; 2938 hdlr->getParameterEntity = NULL; 2939 hdlr->entityDecl = NULL; 2940 hdlr->attributeDecl = NULL; 2941 hdlr->elementDecl = NULL; 2942 hdlr->notationDecl = NULL; 2943 hdlr->unparsedEntityDecl = NULL; 2944 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; 2945 hdlr->startDocument = xmlSAX2StartDocument; 2946 hdlr->endDocument = xmlSAX2EndDocument; 2947 hdlr->startElement = xmlSAX2StartElement; 2948 hdlr->endElement = xmlSAX2EndElement; 2949 hdlr->reference = NULL; 2950 hdlr->characters = xmlSAX2Characters; 2951 hdlr->cdataBlock = xmlSAX2CDataBlock; 2952 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace; 2953 hdlr->processingInstruction = xmlSAX2ProcessingInstruction; 2954 hdlr->comment = xmlSAX2Comment; 2955 hdlr->warning = xmlParserWarning; 2956 hdlr->error = xmlParserError; 2957 hdlr->fatalError = xmlParserError; 2958 2959 hdlr->initialized = 1; 2960 } 2961 2962 /** 2963 * htmlDefaultSAXHandlerInit: 2964 * 2965 * Initialize the default SAX handler 2966 */ 2967 void 2968 htmlDefaultSAXHandlerInit(void) 2969 { 2970 xmlSAX2InitHtmlDefaultSAXHandler((xmlSAXHandlerPtr) &htmlDefaultSAXHandler); 2971 } 2972 2973 #endif /* LIBXML_HTML_ENABLED */ 2974 2975 #ifdef LIBXML_DOCB_ENABLED 2976 2977 /** 2978 * xmlSAX2InitDocbDefaultSAXHandler: 2979 * @hdlr: the SAX handler 2980 * 2981 * Initialize the default DocBook SAX2 handler 2982 */ 2983 void 2984 xmlSAX2InitDocbDefaultSAXHandler(xmlSAXHandler *hdlr) 2985 { 2986 if ((hdlr == NULL) || (hdlr->initialized != 0)) 2987 return; 2988 2989 hdlr->internalSubset = xmlSAX2InternalSubset; 2990 hdlr->externalSubset = NULL; 2991 hdlr->isStandalone = xmlSAX2IsStandalone; 2992 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset; 2993 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset; 2994 hdlr->resolveEntity = xmlSAX2ResolveEntity; 2995 hdlr->getEntity = xmlSAX2GetEntity; 2996 hdlr->getParameterEntity = NULL; 2997 hdlr->entityDecl = xmlSAX2EntityDecl; 2998 hdlr->attributeDecl = NULL; 2999 hdlr->elementDecl = NULL; 3000 hdlr->notationDecl = NULL; 3001 hdlr->unparsedEntityDecl = NULL; 3002 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; 3003 hdlr->startDocument = xmlSAX2StartDocument; 3004 hdlr->endDocument = xmlSAX2EndDocument; 3005 hdlr->startElement = xmlSAX2StartElement; 3006 hdlr->endElement = xmlSAX2EndElement; 3007 hdlr->reference = xmlSAX2Reference; 3008 hdlr->characters = xmlSAX2Characters; 3009 hdlr->cdataBlock = NULL; 3010 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace; 3011 hdlr->processingInstruction = NULL; 3012 hdlr->comment = xmlSAX2Comment; 3013 hdlr->warning = xmlParserWarning; 3014 hdlr->error = xmlParserError; 3015 hdlr->fatalError = xmlParserError; 3016 3017 hdlr->initialized = 1; 3018 } 3019 3020 /** 3021 * docbDefaultSAXHandlerInit: 3022 * 3023 * Initialize the default SAX handler 3024 */ 3025 void 3026 docbDefaultSAXHandlerInit(void) 3027 { 3028 xmlSAX2InitDocbDefaultSAXHandler((xmlSAXHandlerPtr) &docbDefaultSAXHandler); 3029 } 3030 3031 #endif /* LIBXML_DOCB_ENABLED */ 3032 #define bottom_SAX2 3033 #include "elfgcchack.h" 3034
