issue5063: Fixes for building RPM on CentOS plus misc .spec file enhancements.
[python.git] / Modules / expat / xmlparse.c
blobe04426d0cc1591e5d45be5b0f0dcb335226af500
1 /* Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd
2 See the file COPYING for copying permission.
3 */
5 #define XML_BUILDING_EXPAT 1
7 #ifdef COMPILED_FROM_DSP
8 #include "winconfig.h"
9 #elif defined(MACOS_CLASSIC)
10 #include "macconfig.h"
11 #elif defined(__amigaos4__)
12 #include "amigaconfig.h"
13 #elif defined(HAVE_EXPAT_CONFIG_H)
14 #include <expat_config.h>
15 #endif /* ndef COMPILED_FROM_DSP */
17 #include <stddef.h>
18 #include <string.h> /* memset(), memcpy() */
19 #include <assert.h>
21 #include "expat.h"
23 #ifdef XML_UNICODE
24 #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX
25 #define XmlConvert XmlUtf16Convert
26 #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding
27 #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS
28 #define XmlEncode XmlUtf16Encode
29 #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1))
30 typedef unsigned short ICHAR;
31 #else
32 #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX
33 #define XmlConvert XmlUtf8Convert
34 #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding
35 #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS
36 #define XmlEncode XmlUtf8Encode
37 #define MUST_CONVERT(enc, s) (!(enc)->isUtf8)
38 typedef char ICHAR;
39 #endif
42 #ifndef XML_NS
44 #define XmlInitEncodingNS XmlInitEncoding
45 #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding
46 #undef XmlGetInternalEncodingNS
47 #define XmlGetInternalEncodingNS XmlGetInternalEncoding
48 #define XmlParseXmlDeclNS XmlParseXmlDecl
50 #endif
52 #ifdef XML_UNICODE
54 #ifdef XML_UNICODE_WCHAR_T
55 #define XML_T(x) (const wchar_t)x
56 #define XML_L(x) L ## x
57 #else
58 #define XML_T(x) (const unsigned short)x
59 #define XML_L(x) x
60 #endif
62 #else
64 #define XML_T(x) x
65 #define XML_L(x) x
67 #endif
69 /* Round up n to be a multiple of sz, where sz is a power of 2. */
70 #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1))
72 /* Handle the case where memmove() doesn't exist. */
73 #ifndef HAVE_MEMMOVE
74 #ifdef HAVE_BCOPY
75 #define memmove(d,s,l) bcopy((s),(d),(l))
76 #else
77 #error memmove does not exist on this platform, nor is a substitute available
78 #endif /* HAVE_BCOPY */
79 #endif /* HAVE_MEMMOVE */
81 #include "internal.h"
82 #include "xmltok.h"
83 #include "xmlrole.h"
85 typedef const XML_Char *KEY;
87 typedef struct {
88 KEY name;
89 } NAMED;
91 typedef struct {
92 NAMED **v;
93 unsigned char power;
94 size_t size;
95 size_t used;
96 const XML_Memory_Handling_Suite *mem;
97 } HASH_TABLE;
99 /* Basic character hash algorithm, taken from Python's string hash:
100 h = h * 1000003 ^ character, the constant being a prime number.
103 #ifdef XML_UNICODE
104 #define CHAR_HASH(h, c) \
105 (((h) * 0xF4243) ^ (unsigned short)(c))
106 #else
107 #define CHAR_HASH(h, c) \
108 (((h) * 0xF4243) ^ (unsigned char)(c))
109 #endif
111 /* For probing (after a collision) we need a step size relative prime
112 to the hash table size, which is a power of 2. We use double-hashing,
113 since we can calculate a second hash value cheaply by taking those bits
114 of the first hash value that were discarded (masked out) when the table
115 index was calculated: index = hash & mask, where mask = table->size - 1.
116 We limit the maximum step size to table->size / 4 (mask >> 2) and make
117 it odd, since odd numbers are always relative prime to a power of 2.
119 #define SECOND_HASH(hash, mask, power) \
120 ((((hash) & ~(mask)) >> ((power) - 1)) & ((mask) >> 2))
121 #define PROBE_STEP(hash, mask, power) \
122 ((unsigned char)((SECOND_HASH(hash, mask, power)) | 1))
124 typedef struct {
125 NAMED **p;
126 NAMED **end;
127 } HASH_TABLE_ITER;
129 #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */
130 #define INIT_DATA_BUF_SIZE 1024
131 #define INIT_ATTS_SIZE 16
132 #define INIT_ATTS_VERSION 0xFFFFFFFF
133 #define INIT_BLOCK_SIZE 1024
134 #define INIT_BUFFER_SIZE 1024
136 #define EXPAND_SPARE 24
138 typedef struct binding {
139 struct prefix *prefix;
140 struct binding *nextTagBinding;
141 struct binding *prevPrefixBinding;
142 const struct attribute_id *attId;
143 XML_Char *uri;
144 int uriLen;
145 int uriAlloc;
146 } BINDING;
148 typedef struct prefix {
149 const XML_Char *name;
150 BINDING *binding;
151 } PREFIX;
153 typedef struct {
154 const XML_Char *str;
155 const XML_Char *localPart;
156 const XML_Char *prefix;
157 int strLen;
158 int uriLen;
159 int prefixLen;
160 } TAG_NAME;
162 /* TAG represents an open element.
163 The name of the element is stored in both the document and API
164 encodings. The memory buffer 'buf' is a separately-allocated
165 memory area which stores the name. During the XML_Parse()/
166 XMLParseBuffer() when the element is open, the memory for the 'raw'
167 version of the name (in the document encoding) is shared with the
168 document buffer. If the element is open across calls to
169 XML_Parse()/XML_ParseBuffer(), the buffer is re-allocated to
170 contain the 'raw' name as well.
172 A parser re-uses these structures, maintaining a list of allocated
173 TAG objects in a free list.
175 typedef struct tag {
176 struct tag *parent; /* parent of this element */
177 const char *rawName; /* tagName in the original encoding */
178 int rawNameLength;
179 TAG_NAME name; /* tagName in the API encoding */
180 char *buf; /* buffer for name components */
181 char *bufEnd; /* end of the buffer */
182 BINDING *bindings;
183 } TAG;
185 typedef struct {
186 const XML_Char *name;
187 const XML_Char *textPtr;
188 int textLen; /* length in XML_Chars */
189 int processed; /* # of processed bytes - when suspended */
190 const XML_Char *systemId;
191 const XML_Char *base;
192 const XML_Char *publicId;
193 const XML_Char *notation;
194 XML_Bool open;
195 XML_Bool is_param;
196 XML_Bool is_internal; /* true if declared in internal subset outside PE */
197 } ENTITY;
199 typedef struct {
200 enum XML_Content_Type type;
201 enum XML_Content_Quant quant;
202 const XML_Char * name;
203 int firstchild;
204 int lastchild;
205 int childcnt;
206 int nextsib;
207 } CONTENT_SCAFFOLD;
209 #define INIT_SCAFFOLD_ELEMENTS 32
211 typedef struct block {
212 struct block *next;
213 int size;
214 XML_Char s[1];
215 } BLOCK;
217 typedef struct {
218 BLOCK *blocks;
219 BLOCK *freeBlocks;
220 const XML_Char *end;
221 XML_Char *ptr;
222 XML_Char *start;
223 const XML_Memory_Handling_Suite *mem;
224 } STRING_POOL;
226 /* The XML_Char before the name is used to determine whether
227 an attribute has been specified. */
228 typedef struct attribute_id {
229 XML_Char *name;
230 PREFIX *prefix;
231 XML_Bool maybeTokenized;
232 XML_Bool xmlns;
233 } ATTRIBUTE_ID;
235 typedef struct {
236 const ATTRIBUTE_ID *id;
237 XML_Bool isCdata;
238 const XML_Char *value;
239 } DEFAULT_ATTRIBUTE;
241 typedef struct {
242 unsigned long version;
243 unsigned long hash;
244 const XML_Char *uriName;
245 } NS_ATT;
247 typedef struct {
248 const XML_Char *name;
249 PREFIX *prefix;
250 const ATTRIBUTE_ID *idAtt;
251 int nDefaultAtts;
252 int allocDefaultAtts;
253 DEFAULT_ATTRIBUTE *defaultAtts;
254 } ELEMENT_TYPE;
256 typedef struct {
257 HASH_TABLE generalEntities;
258 HASH_TABLE elementTypes;
259 HASH_TABLE attributeIds;
260 HASH_TABLE prefixes;
261 STRING_POOL pool;
262 STRING_POOL entityValuePool;
263 /* false once a parameter entity reference has been skipped */
264 XML_Bool keepProcessing;
265 /* true once an internal or external PE reference has been encountered;
266 this includes the reference to an external subset */
267 XML_Bool hasParamEntityRefs;
268 XML_Bool standalone;
269 #ifdef XML_DTD
270 /* indicates if external PE has been read */
271 XML_Bool paramEntityRead;
272 HASH_TABLE paramEntities;
273 #endif /* XML_DTD */
274 PREFIX defaultPrefix;
275 /* === scaffolding for building content model === */
276 XML_Bool in_eldecl;
277 CONTENT_SCAFFOLD *scaffold;
278 unsigned contentStringLen;
279 unsigned scaffSize;
280 unsigned scaffCount;
281 int scaffLevel;
282 int *scaffIndex;
283 } DTD;
285 typedef struct open_internal_entity {
286 const char *internalEventPtr;
287 const char *internalEventEndPtr;
288 struct open_internal_entity *next;
289 ENTITY *entity;
290 int startTagLevel;
291 XML_Bool betweenDecl; /* WFC: PE Between Declarations */
292 } OPEN_INTERNAL_ENTITY;
294 typedef enum XML_Error PTRCALL Processor(XML_Parser parser,
295 const char *start,
296 const char *end,
297 const char **endPtr);
299 static Processor prologProcessor;
300 static Processor prologInitProcessor;
301 static Processor contentProcessor;
302 static Processor cdataSectionProcessor;
303 #ifdef XML_DTD
304 static Processor ignoreSectionProcessor;
305 static Processor externalParEntProcessor;
306 static Processor externalParEntInitProcessor;
307 static Processor entityValueProcessor;
308 static Processor entityValueInitProcessor;
309 #endif /* XML_DTD */
310 static Processor epilogProcessor;
311 static Processor errorProcessor;
312 static Processor externalEntityInitProcessor;
313 static Processor externalEntityInitProcessor2;
314 static Processor externalEntityInitProcessor3;
315 static Processor externalEntityContentProcessor;
316 static Processor internalEntityProcessor;
318 static enum XML_Error
319 handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName);
320 static enum XML_Error
321 processXmlDecl(XML_Parser parser, int isGeneralTextEntity,
322 const char *s, const char *next);
323 static enum XML_Error
324 initializeEncoding(XML_Parser parser);
325 static enum XML_Error
326 doProlog(XML_Parser parser, const ENCODING *enc, const char *s,
327 const char *end, int tok, const char *next, const char **nextPtr,
328 XML_Bool haveMore);
329 static enum XML_Error
330 processInternalEntity(XML_Parser parser, ENTITY *entity,
331 XML_Bool betweenDecl);
332 static enum XML_Error
333 doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc,
334 const char *start, const char *end, const char **endPtr,
335 XML_Bool haveMore);
336 static enum XML_Error
337 doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr,
338 const char *end, const char **nextPtr, XML_Bool haveMore);
339 #ifdef XML_DTD
340 static enum XML_Error
341 doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr,
342 const char *end, const char **nextPtr, XML_Bool haveMore);
343 #endif /* XML_DTD */
345 static enum XML_Error
346 storeAtts(XML_Parser parser, const ENCODING *, const char *s,
347 TAG_NAME *tagNamePtr, BINDING **bindingsPtr);
348 static enum XML_Error
349 addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId,
350 const XML_Char *uri, BINDING **bindingsPtr);
351 static int
352 defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, XML_Bool isCdata,
353 XML_Bool isId, const XML_Char *dfltValue, XML_Parser parser);
354 static enum XML_Error
355 storeAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata,
356 const char *, const char *, STRING_POOL *);
357 static enum XML_Error
358 appendAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata,
359 const char *, const char *, STRING_POOL *);
360 static ATTRIBUTE_ID *
361 getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start,
362 const char *end);
363 static int
364 setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *);
365 static enum XML_Error
366 storeEntityValue(XML_Parser parser, const ENCODING *enc, const char *start,
367 const char *end);
368 static int
369 reportProcessingInstruction(XML_Parser parser, const ENCODING *enc,
370 const char *start, const char *end);
371 static int
372 reportComment(XML_Parser parser, const ENCODING *enc, const char *start,
373 const char *end);
374 static void
375 reportDefault(XML_Parser parser, const ENCODING *enc, const char *start,
376 const char *end);
378 static const XML_Char * getContext(XML_Parser parser);
379 static XML_Bool
380 setContext(XML_Parser parser, const XML_Char *context);
382 static void FASTCALL normalizePublicId(XML_Char *s);
384 static DTD * dtdCreate(const XML_Memory_Handling_Suite *ms);
385 /* do not call if parentParser != NULL */
386 static void dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms);
387 static void
388 dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms);
389 static int
390 dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms);
391 static int
392 copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *);
394 static NAMED *
395 lookup(HASH_TABLE *table, KEY name, size_t createSize);
396 static void FASTCALL
397 hashTableInit(HASH_TABLE *, const XML_Memory_Handling_Suite *ms);
398 static void FASTCALL hashTableClear(HASH_TABLE *);
399 static void FASTCALL hashTableDestroy(HASH_TABLE *);
400 static void FASTCALL
401 hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *);
402 static NAMED * FASTCALL hashTableIterNext(HASH_TABLE_ITER *);
404 static void FASTCALL
405 poolInit(STRING_POOL *, const XML_Memory_Handling_Suite *ms);
406 static void FASTCALL poolClear(STRING_POOL *);
407 static void FASTCALL poolDestroy(STRING_POOL *);
408 static XML_Char *
409 poolAppend(STRING_POOL *pool, const ENCODING *enc,
410 const char *ptr, const char *end);
411 static XML_Char *
412 poolStoreString(STRING_POOL *pool, const ENCODING *enc,
413 const char *ptr, const char *end);
414 static XML_Bool FASTCALL poolGrow(STRING_POOL *pool);
415 static const XML_Char * FASTCALL
416 poolCopyString(STRING_POOL *pool, const XML_Char *s);
417 static const XML_Char *
418 poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n);
419 static const XML_Char * FASTCALL
420 poolAppendString(STRING_POOL *pool, const XML_Char *s);
422 static int FASTCALL nextScaffoldPart(XML_Parser parser);
423 static XML_Content * build_model(XML_Parser parser);
424 static ELEMENT_TYPE *
425 getElementType(XML_Parser parser, const ENCODING *enc,
426 const char *ptr, const char *end);
428 static XML_Parser
429 parserCreate(const XML_Char *encodingName,
430 const XML_Memory_Handling_Suite *memsuite,
431 const XML_Char *nameSep,
432 DTD *dtd);
433 static void
434 parserInit(XML_Parser parser, const XML_Char *encodingName);
436 #define poolStart(pool) ((pool)->start)
437 #define poolEnd(pool) ((pool)->ptr)
438 #define poolLength(pool) ((pool)->ptr - (pool)->start)
439 #define poolChop(pool) ((void)--(pool->ptr))
440 #define poolLastChar(pool) (((pool)->ptr)[-1])
441 #define poolDiscard(pool) ((pool)->ptr = (pool)->start)
442 #define poolFinish(pool) ((pool)->start = (pool)->ptr)
443 #define poolAppendChar(pool, c) \
444 (((pool)->ptr == (pool)->end && !poolGrow(pool)) \
445 ? 0 \
446 : ((*((pool)->ptr)++ = c), 1))
448 struct XML_ParserStruct {
449 /* The first member must be userData so that the XML_GetUserData
450 macro works. */
451 void *m_userData;
452 void *m_handlerArg;
453 char *m_buffer;
454 const XML_Memory_Handling_Suite m_mem;
455 /* first character to be parsed */
456 const char *m_bufferPtr;
457 /* past last character to be parsed */
458 char *m_bufferEnd;
459 /* allocated end of buffer */
460 const char *m_bufferLim;
461 XML_Index m_parseEndByteIndex;
462 const char *m_parseEndPtr;
463 XML_Char *m_dataBuf;
464 XML_Char *m_dataBufEnd;
465 XML_StartElementHandler m_startElementHandler;
466 XML_EndElementHandler m_endElementHandler;
467 XML_CharacterDataHandler m_characterDataHandler;
468 XML_ProcessingInstructionHandler m_processingInstructionHandler;
469 XML_CommentHandler m_commentHandler;
470 XML_StartCdataSectionHandler m_startCdataSectionHandler;
471 XML_EndCdataSectionHandler m_endCdataSectionHandler;
472 XML_DefaultHandler m_defaultHandler;
473 XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler;
474 XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler;
475 XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler;
476 XML_NotationDeclHandler m_notationDeclHandler;
477 XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler;
478 XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler;
479 XML_NotStandaloneHandler m_notStandaloneHandler;
480 XML_ExternalEntityRefHandler m_externalEntityRefHandler;
481 XML_Parser m_externalEntityRefHandlerArg;
482 XML_SkippedEntityHandler m_skippedEntityHandler;
483 XML_UnknownEncodingHandler m_unknownEncodingHandler;
484 XML_ElementDeclHandler m_elementDeclHandler;
485 XML_AttlistDeclHandler m_attlistDeclHandler;
486 XML_EntityDeclHandler m_entityDeclHandler;
487 XML_XmlDeclHandler m_xmlDeclHandler;
488 const ENCODING *m_encoding;
489 INIT_ENCODING m_initEncoding;
490 const ENCODING *m_internalEncoding;
491 const XML_Char *m_protocolEncodingName;
492 XML_Bool m_ns;
493 XML_Bool m_ns_triplets;
494 void *m_unknownEncodingMem;
495 void *m_unknownEncodingData;
496 void *m_unknownEncodingHandlerData;
497 void (XMLCALL *m_unknownEncodingRelease)(void *);
498 PROLOG_STATE m_prologState;
499 Processor *m_processor;
500 enum XML_Error m_errorCode;
501 const char *m_eventPtr;
502 const char *m_eventEndPtr;
503 const char *m_positionPtr;
504 OPEN_INTERNAL_ENTITY *m_openInternalEntities;
505 OPEN_INTERNAL_ENTITY *m_freeInternalEntities;
506 XML_Bool m_defaultExpandInternalEntities;
507 int m_tagLevel;
508 ENTITY *m_declEntity;
509 const XML_Char *m_doctypeName;
510 const XML_Char *m_doctypeSysid;
511 const XML_Char *m_doctypePubid;
512 const XML_Char *m_declAttributeType;
513 const XML_Char *m_declNotationName;
514 const XML_Char *m_declNotationPublicId;
515 ELEMENT_TYPE *m_declElementType;
516 ATTRIBUTE_ID *m_declAttributeId;
517 XML_Bool m_declAttributeIsCdata;
518 XML_Bool m_declAttributeIsId;
519 DTD *m_dtd;
520 const XML_Char *m_curBase;
521 TAG *m_tagStack;
522 TAG *m_freeTagList;
523 BINDING *m_inheritedBindings;
524 BINDING *m_freeBindingList;
525 int m_attsSize;
526 int m_nSpecifiedAtts;
527 int m_idAttIndex;
528 ATTRIBUTE *m_atts;
529 NS_ATT *m_nsAtts;
530 unsigned long m_nsAttsVersion;
531 unsigned char m_nsAttsPower;
532 POSITION m_position;
533 STRING_POOL m_tempPool;
534 STRING_POOL m_temp2Pool;
535 char *m_groupConnector;
536 unsigned int m_groupSize;
537 XML_Char m_namespaceSeparator;
538 XML_Parser m_parentParser;
539 XML_ParsingStatus m_parsingStatus;
540 #ifdef XML_DTD
541 XML_Bool m_isParamEntity;
542 XML_Bool m_useForeignDTD;
543 enum XML_ParamEntityParsing m_paramEntityParsing;
544 #endif
547 #define MALLOC(s) (parser->m_mem.malloc_fcn((s)))
548 #define REALLOC(p,s) (parser->m_mem.realloc_fcn((p),(s)))
549 #define FREE(p) (parser->m_mem.free_fcn((p)))
551 #define userData (parser->m_userData)
552 #define handlerArg (parser->m_handlerArg)
553 #define startElementHandler (parser->m_startElementHandler)
554 #define endElementHandler (parser->m_endElementHandler)
555 #define characterDataHandler (parser->m_characterDataHandler)
556 #define processingInstructionHandler \
557 (parser->m_processingInstructionHandler)
558 #define commentHandler (parser->m_commentHandler)
559 #define startCdataSectionHandler \
560 (parser->m_startCdataSectionHandler)
561 #define endCdataSectionHandler (parser->m_endCdataSectionHandler)
562 #define defaultHandler (parser->m_defaultHandler)
563 #define startDoctypeDeclHandler (parser->m_startDoctypeDeclHandler)
564 #define endDoctypeDeclHandler (parser->m_endDoctypeDeclHandler)
565 #define unparsedEntityDeclHandler \
566 (parser->m_unparsedEntityDeclHandler)
567 #define notationDeclHandler (parser->m_notationDeclHandler)
568 #define startNamespaceDeclHandler \
569 (parser->m_startNamespaceDeclHandler)
570 #define endNamespaceDeclHandler (parser->m_endNamespaceDeclHandler)
571 #define notStandaloneHandler (parser->m_notStandaloneHandler)
572 #define externalEntityRefHandler \
573 (parser->m_externalEntityRefHandler)
574 #define externalEntityRefHandlerArg \
575 (parser->m_externalEntityRefHandlerArg)
576 #define internalEntityRefHandler \
577 (parser->m_internalEntityRefHandler)
578 #define skippedEntityHandler (parser->m_skippedEntityHandler)
579 #define unknownEncodingHandler (parser->m_unknownEncodingHandler)
580 #define elementDeclHandler (parser->m_elementDeclHandler)
581 #define attlistDeclHandler (parser->m_attlistDeclHandler)
582 #define entityDeclHandler (parser->m_entityDeclHandler)
583 #define xmlDeclHandler (parser->m_xmlDeclHandler)
584 #define encoding (parser->m_encoding)
585 #define initEncoding (parser->m_initEncoding)
586 #define internalEncoding (parser->m_internalEncoding)
587 #define unknownEncodingMem (parser->m_unknownEncodingMem)
588 #define unknownEncodingData (parser->m_unknownEncodingData)
589 #define unknownEncodingHandlerData \
590 (parser->m_unknownEncodingHandlerData)
591 #define unknownEncodingRelease (parser->m_unknownEncodingRelease)
592 #define protocolEncodingName (parser->m_protocolEncodingName)
593 #define ns (parser->m_ns)
594 #define ns_triplets (parser->m_ns_triplets)
595 #define prologState (parser->m_prologState)
596 #define processor (parser->m_processor)
597 #define errorCode (parser->m_errorCode)
598 #define eventPtr (parser->m_eventPtr)
599 #define eventEndPtr (parser->m_eventEndPtr)
600 #define positionPtr (parser->m_positionPtr)
601 #define position (parser->m_position)
602 #define openInternalEntities (parser->m_openInternalEntities)
603 #define freeInternalEntities (parser->m_freeInternalEntities)
604 #define defaultExpandInternalEntities \
605 (parser->m_defaultExpandInternalEntities)
606 #define tagLevel (parser->m_tagLevel)
607 #define buffer (parser->m_buffer)
608 #define bufferPtr (parser->m_bufferPtr)
609 #define bufferEnd (parser->m_bufferEnd)
610 #define parseEndByteIndex (parser->m_parseEndByteIndex)
611 #define parseEndPtr (parser->m_parseEndPtr)
612 #define bufferLim (parser->m_bufferLim)
613 #define dataBuf (parser->m_dataBuf)
614 #define dataBufEnd (parser->m_dataBufEnd)
615 #define _dtd (parser->m_dtd)
616 #define curBase (parser->m_curBase)
617 #define declEntity (parser->m_declEntity)
618 #define doctypeName (parser->m_doctypeName)
619 #define doctypeSysid (parser->m_doctypeSysid)
620 #define doctypePubid (parser->m_doctypePubid)
621 #define declAttributeType (parser->m_declAttributeType)
622 #define declNotationName (parser->m_declNotationName)
623 #define declNotationPublicId (parser->m_declNotationPublicId)
624 #define declElementType (parser->m_declElementType)
625 #define declAttributeId (parser->m_declAttributeId)
626 #define declAttributeIsCdata (parser->m_declAttributeIsCdata)
627 #define declAttributeIsId (parser->m_declAttributeIsId)
628 #define freeTagList (parser->m_freeTagList)
629 #define freeBindingList (parser->m_freeBindingList)
630 #define inheritedBindings (parser->m_inheritedBindings)
631 #define tagStack (parser->m_tagStack)
632 #define atts (parser->m_atts)
633 #define attsSize (parser->m_attsSize)
634 #define nSpecifiedAtts (parser->m_nSpecifiedAtts)
635 #define idAttIndex (parser->m_idAttIndex)
636 #define nsAtts (parser->m_nsAtts)
637 #define nsAttsVersion (parser->m_nsAttsVersion)
638 #define nsAttsPower (parser->m_nsAttsPower)
639 #define tempPool (parser->m_tempPool)
640 #define temp2Pool (parser->m_temp2Pool)
641 #define groupConnector (parser->m_groupConnector)
642 #define groupSize (parser->m_groupSize)
643 #define namespaceSeparator (parser->m_namespaceSeparator)
644 #define parentParser (parser->m_parentParser)
645 #define ps_parsing (parser->m_parsingStatus.parsing)
646 #define ps_finalBuffer (parser->m_parsingStatus.finalBuffer)
647 #ifdef XML_DTD
648 #define isParamEntity (parser->m_isParamEntity)
649 #define useForeignDTD (parser->m_useForeignDTD)
650 #define paramEntityParsing (parser->m_paramEntityParsing)
651 #endif /* XML_DTD */
653 XML_Parser XMLCALL
654 XML_ParserCreate(const XML_Char *encodingName)
656 return XML_ParserCreate_MM(encodingName, NULL, NULL);
659 XML_Parser XMLCALL
660 XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep)
662 XML_Char tmp[2];
663 *tmp = nsSep;
664 return XML_ParserCreate_MM(encodingName, NULL, tmp);
667 static const XML_Char implicitContext[] = {
668 'x', 'm', 'l', '=', 'h', 't', 't', 'p', ':', '/', '/',
669 'w', 'w', 'w', '.', 'w', '3', '.', 'o', 'r', 'g', '/',
670 'X', 'M', 'L', '/', '1', '9', '9', '8', '/',
671 'n', 'a', 'm', 'e', 's', 'p', 'a', 'c', 'e', '\0'
674 XML_Parser XMLCALL
675 XML_ParserCreate_MM(const XML_Char *encodingName,
676 const XML_Memory_Handling_Suite *memsuite,
677 const XML_Char *nameSep)
679 XML_Parser parser = parserCreate(encodingName, memsuite, nameSep, NULL);
680 if (parser != NULL && ns) {
681 /* implicit context only set for root parser, since child
682 parsers (i.e. external entity parsers) will inherit it
684 if (!setContext(parser, implicitContext)) {
685 XML_ParserFree(parser);
686 return NULL;
689 return parser;
692 static XML_Parser
693 parserCreate(const XML_Char *encodingName,
694 const XML_Memory_Handling_Suite *memsuite,
695 const XML_Char *nameSep,
696 DTD *dtd)
698 XML_Parser parser;
700 if (memsuite) {
701 XML_Memory_Handling_Suite *mtemp;
702 parser = (XML_Parser)
703 memsuite->malloc_fcn(sizeof(struct XML_ParserStruct));
704 if (parser != NULL) {
705 mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem);
706 mtemp->malloc_fcn = memsuite->malloc_fcn;
707 mtemp->realloc_fcn = memsuite->realloc_fcn;
708 mtemp->free_fcn = memsuite->free_fcn;
711 else {
712 XML_Memory_Handling_Suite *mtemp;
713 parser = (XML_Parser)malloc(sizeof(struct XML_ParserStruct));
714 if (parser != NULL) {
715 mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem);
716 mtemp->malloc_fcn = malloc;
717 mtemp->realloc_fcn = realloc;
718 mtemp->free_fcn = free;
722 if (!parser)
723 return parser;
725 buffer = NULL;
726 bufferLim = NULL;
728 attsSize = INIT_ATTS_SIZE;
729 atts = (ATTRIBUTE *)MALLOC(attsSize * sizeof(ATTRIBUTE));
730 if (atts == NULL) {
731 FREE(parser);
732 return NULL;
734 dataBuf = (XML_Char *)MALLOC(INIT_DATA_BUF_SIZE * sizeof(XML_Char));
735 if (dataBuf == NULL) {
736 FREE(atts);
737 FREE(parser);
738 return NULL;
740 dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE;
742 if (dtd)
743 _dtd = dtd;
744 else {
745 _dtd = dtdCreate(&parser->m_mem);
746 if (_dtd == NULL) {
747 FREE(dataBuf);
748 FREE(atts);
749 FREE(parser);
750 return NULL;
754 freeBindingList = NULL;
755 freeTagList = NULL;
756 freeInternalEntities = NULL;
758 groupSize = 0;
759 groupConnector = NULL;
761 unknownEncodingHandler = NULL;
762 unknownEncodingHandlerData = NULL;
764 namespaceSeparator = '!';
765 ns = XML_FALSE;
766 ns_triplets = XML_FALSE;
768 nsAtts = NULL;
769 nsAttsVersion = 0;
770 nsAttsPower = 0;
772 poolInit(&tempPool, &(parser->m_mem));
773 poolInit(&temp2Pool, &(parser->m_mem));
774 parserInit(parser, encodingName);
776 if (encodingName && !protocolEncodingName) {
777 XML_ParserFree(parser);
778 return NULL;
781 if (nameSep) {
782 ns = XML_TRUE;
783 internalEncoding = XmlGetInternalEncodingNS();
784 namespaceSeparator = *nameSep;
786 else {
787 internalEncoding = XmlGetInternalEncoding();
790 return parser;
793 static void
794 parserInit(XML_Parser parser, const XML_Char *encodingName)
796 processor = prologInitProcessor;
797 XmlPrologStateInit(&prologState);
798 protocolEncodingName = (encodingName != NULL
799 ? poolCopyString(&tempPool, encodingName)
800 : NULL);
801 curBase = NULL;
802 XmlInitEncoding(&initEncoding, &encoding, 0);
803 userData = NULL;
804 handlerArg = NULL;
805 startElementHandler = NULL;
806 endElementHandler = NULL;
807 characterDataHandler = NULL;
808 processingInstructionHandler = NULL;
809 commentHandler = NULL;
810 startCdataSectionHandler = NULL;
811 endCdataSectionHandler = NULL;
812 defaultHandler = NULL;
813 startDoctypeDeclHandler = NULL;
814 endDoctypeDeclHandler = NULL;
815 unparsedEntityDeclHandler = NULL;
816 notationDeclHandler = NULL;
817 startNamespaceDeclHandler = NULL;
818 endNamespaceDeclHandler = NULL;
819 notStandaloneHandler = NULL;
820 externalEntityRefHandler = NULL;
821 externalEntityRefHandlerArg = parser;
822 skippedEntityHandler = NULL;
823 elementDeclHandler = NULL;
824 attlistDeclHandler = NULL;
825 entityDeclHandler = NULL;
826 xmlDeclHandler = NULL;
827 bufferPtr = buffer;
828 bufferEnd = buffer;
829 parseEndByteIndex = 0;
830 parseEndPtr = NULL;
831 declElementType = NULL;
832 declAttributeId = NULL;
833 declEntity = NULL;
834 doctypeName = NULL;
835 doctypeSysid = NULL;
836 doctypePubid = NULL;
837 declAttributeType = NULL;
838 declNotationName = NULL;
839 declNotationPublicId = NULL;
840 declAttributeIsCdata = XML_FALSE;
841 declAttributeIsId = XML_FALSE;
842 memset(&position, 0, sizeof(POSITION));
843 errorCode = XML_ERROR_NONE;
844 eventPtr = NULL;
845 eventEndPtr = NULL;
846 positionPtr = NULL;
847 openInternalEntities = NULL;
848 defaultExpandInternalEntities = XML_TRUE;
849 tagLevel = 0;
850 tagStack = NULL;
851 inheritedBindings = NULL;
852 nSpecifiedAtts = 0;
853 unknownEncodingMem = NULL;
854 unknownEncodingRelease = NULL;
855 unknownEncodingData = NULL;
856 parentParser = NULL;
857 ps_parsing = XML_INITIALIZED;
858 #ifdef XML_DTD
859 isParamEntity = XML_FALSE;
860 useForeignDTD = XML_FALSE;
861 paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER;
862 #endif
865 /* moves list of bindings to freeBindingList */
866 static void FASTCALL
867 moveToFreeBindingList(XML_Parser parser, BINDING *bindings)
869 while (bindings) {
870 BINDING *b = bindings;
871 bindings = bindings->nextTagBinding;
872 b->nextTagBinding = freeBindingList;
873 freeBindingList = b;
877 XML_Bool XMLCALL
878 XML_ParserReset(XML_Parser parser, const XML_Char *encodingName)
880 TAG *tStk;
881 OPEN_INTERNAL_ENTITY *openEntityList;
882 if (parentParser)
883 return XML_FALSE;
884 /* move tagStack to freeTagList */
885 tStk = tagStack;
886 while (tStk) {
887 TAG *tag = tStk;
888 tStk = tStk->parent;
889 tag->parent = freeTagList;
890 moveToFreeBindingList(parser, tag->bindings);
891 tag->bindings = NULL;
892 freeTagList = tag;
894 /* move openInternalEntities to freeInternalEntities */
895 openEntityList = openInternalEntities;
896 while (openEntityList) {
897 OPEN_INTERNAL_ENTITY *openEntity = openEntityList;
898 openEntityList = openEntity->next;
899 openEntity->next = freeInternalEntities;
900 freeInternalEntities = openEntity;
902 moveToFreeBindingList(parser, inheritedBindings);
903 FREE(unknownEncodingMem);
904 if (unknownEncodingRelease)
905 unknownEncodingRelease(unknownEncodingData);
906 poolClear(&tempPool);
907 poolClear(&temp2Pool);
908 parserInit(parser, encodingName);
909 dtdReset(_dtd, &parser->m_mem);
910 return setContext(parser, implicitContext);
913 enum XML_Status XMLCALL
914 XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName)
916 /* Block after XML_Parse()/XML_ParseBuffer() has been called.
917 XXX There's no way for the caller to determine which of the
918 XXX possible error cases caused the XML_STATUS_ERROR return.
920 if (ps_parsing == XML_PARSING || ps_parsing == XML_SUSPENDED)
921 return XML_STATUS_ERROR;
922 if (encodingName == NULL)
923 protocolEncodingName = NULL;
924 else {
925 protocolEncodingName = poolCopyString(&tempPool, encodingName);
926 if (!protocolEncodingName)
927 return XML_STATUS_ERROR;
929 return XML_STATUS_OK;
932 XML_Parser XMLCALL
933 XML_ExternalEntityParserCreate(XML_Parser oldParser,
934 const XML_Char *context,
935 const XML_Char *encodingName)
937 XML_Parser parser = oldParser;
938 DTD *newDtd = NULL;
939 DTD *oldDtd = _dtd;
940 XML_StartElementHandler oldStartElementHandler = startElementHandler;
941 XML_EndElementHandler oldEndElementHandler = endElementHandler;
942 XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler;
943 XML_ProcessingInstructionHandler oldProcessingInstructionHandler
944 = processingInstructionHandler;
945 XML_CommentHandler oldCommentHandler = commentHandler;
946 XML_StartCdataSectionHandler oldStartCdataSectionHandler
947 = startCdataSectionHandler;
948 XML_EndCdataSectionHandler oldEndCdataSectionHandler
949 = endCdataSectionHandler;
950 XML_DefaultHandler oldDefaultHandler = defaultHandler;
951 XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler
952 = unparsedEntityDeclHandler;
953 XML_NotationDeclHandler oldNotationDeclHandler = notationDeclHandler;
954 XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler
955 = startNamespaceDeclHandler;
956 XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler
957 = endNamespaceDeclHandler;
958 XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler;
959 XML_ExternalEntityRefHandler oldExternalEntityRefHandler
960 = externalEntityRefHandler;
961 XML_SkippedEntityHandler oldSkippedEntityHandler = skippedEntityHandler;
962 XML_UnknownEncodingHandler oldUnknownEncodingHandler
963 = unknownEncodingHandler;
964 XML_ElementDeclHandler oldElementDeclHandler = elementDeclHandler;
965 XML_AttlistDeclHandler oldAttlistDeclHandler = attlistDeclHandler;
966 XML_EntityDeclHandler oldEntityDeclHandler = entityDeclHandler;
967 XML_XmlDeclHandler oldXmlDeclHandler = xmlDeclHandler;
968 ELEMENT_TYPE * oldDeclElementType = declElementType;
970 void *oldUserData = userData;
971 void *oldHandlerArg = handlerArg;
972 XML_Bool oldDefaultExpandInternalEntities = defaultExpandInternalEntities;
973 XML_Parser oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg;
974 #ifdef XML_DTD
975 enum XML_ParamEntityParsing oldParamEntityParsing = paramEntityParsing;
976 int oldInEntityValue = prologState.inEntityValue;
977 #endif
978 XML_Bool oldns_triplets = ns_triplets;
980 #ifdef XML_DTD
981 if (!context)
982 newDtd = oldDtd;
983 #endif /* XML_DTD */
985 /* Note that the magical uses of the pre-processor to make field
986 access look more like C++ require that `parser' be overwritten
987 here. This makes this function more painful to follow than it
988 would be otherwise.
990 if (ns) {
991 XML_Char tmp[2];
992 *tmp = namespaceSeparator;
993 parser = parserCreate(encodingName, &parser->m_mem, tmp, newDtd);
995 else {
996 parser = parserCreate(encodingName, &parser->m_mem, NULL, newDtd);
999 if (!parser)
1000 return NULL;
1002 startElementHandler = oldStartElementHandler;
1003 endElementHandler = oldEndElementHandler;
1004 characterDataHandler = oldCharacterDataHandler;
1005 processingInstructionHandler = oldProcessingInstructionHandler;
1006 commentHandler = oldCommentHandler;
1007 startCdataSectionHandler = oldStartCdataSectionHandler;
1008 endCdataSectionHandler = oldEndCdataSectionHandler;
1009 defaultHandler = oldDefaultHandler;
1010 unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler;
1011 notationDeclHandler = oldNotationDeclHandler;
1012 startNamespaceDeclHandler = oldStartNamespaceDeclHandler;
1013 endNamespaceDeclHandler = oldEndNamespaceDeclHandler;
1014 notStandaloneHandler = oldNotStandaloneHandler;
1015 externalEntityRefHandler = oldExternalEntityRefHandler;
1016 skippedEntityHandler = oldSkippedEntityHandler;
1017 unknownEncodingHandler = oldUnknownEncodingHandler;
1018 elementDeclHandler = oldElementDeclHandler;
1019 attlistDeclHandler = oldAttlistDeclHandler;
1020 entityDeclHandler = oldEntityDeclHandler;
1021 xmlDeclHandler = oldXmlDeclHandler;
1022 declElementType = oldDeclElementType;
1023 userData = oldUserData;
1024 if (oldUserData == oldHandlerArg)
1025 handlerArg = userData;
1026 else
1027 handlerArg = parser;
1028 if (oldExternalEntityRefHandlerArg != oldParser)
1029 externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg;
1030 defaultExpandInternalEntities = oldDefaultExpandInternalEntities;
1031 ns_triplets = oldns_triplets;
1032 parentParser = oldParser;
1033 #ifdef XML_DTD
1034 paramEntityParsing = oldParamEntityParsing;
1035 prologState.inEntityValue = oldInEntityValue;
1036 if (context) {
1037 #endif /* XML_DTD */
1038 if (!dtdCopy(_dtd, oldDtd, &parser->m_mem)
1039 || !setContext(parser, context)) {
1040 XML_ParserFree(parser);
1041 return NULL;
1043 processor = externalEntityInitProcessor;
1044 #ifdef XML_DTD
1046 else {
1047 /* The DTD instance referenced by _dtd is shared between the document's
1048 root parser and external PE parsers, therefore one does not need to
1049 call setContext. In addition, one also *must* not call setContext,
1050 because this would overwrite existing prefix->binding pointers in
1051 _dtd with ones that get destroyed with the external PE parser.
1052 This would leave those prefixes with dangling pointers.
1054 isParamEntity = XML_TRUE;
1055 XmlPrologStateInitExternalEntity(&prologState);
1056 processor = externalParEntInitProcessor;
1058 #endif /* XML_DTD */
1059 return parser;
1062 static void FASTCALL
1063 destroyBindings(BINDING *bindings, XML_Parser parser)
1065 for (;;) {
1066 BINDING *b = bindings;
1067 if (!b)
1068 break;
1069 bindings = b->nextTagBinding;
1070 FREE(b->uri);
1071 FREE(b);
1075 void XMLCALL
1076 XML_ParserFree(XML_Parser parser)
1078 TAG *tagList;
1079 OPEN_INTERNAL_ENTITY *entityList;
1080 if (parser == NULL)
1081 return;
1082 /* free tagStack and freeTagList */
1083 tagList = tagStack;
1084 for (;;) {
1085 TAG *p;
1086 if (tagList == NULL) {
1087 if (freeTagList == NULL)
1088 break;
1089 tagList = freeTagList;
1090 freeTagList = NULL;
1092 p = tagList;
1093 tagList = tagList->parent;
1094 FREE(p->buf);
1095 destroyBindings(p->bindings, parser);
1096 FREE(p);
1098 /* free openInternalEntities and freeInternalEntities */
1099 entityList = openInternalEntities;
1100 for (;;) {
1101 OPEN_INTERNAL_ENTITY *openEntity;
1102 if (entityList == NULL) {
1103 if (freeInternalEntities == NULL)
1104 break;
1105 entityList = freeInternalEntities;
1106 freeInternalEntities = NULL;
1108 openEntity = entityList;
1109 entityList = entityList->next;
1110 FREE(openEntity);
1113 destroyBindings(freeBindingList, parser);
1114 destroyBindings(inheritedBindings, parser);
1115 poolDestroy(&tempPool);
1116 poolDestroy(&temp2Pool);
1117 #ifdef XML_DTD
1118 /* external parameter entity parsers share the DTD structure
1119 parser->m_dtd with the root parser, so we must not destroy it
1121 if (!isParamEntity && _dtd)
1122 #else
1123 if (_dtd)
1124 #endif /* XML_DTD */
1125 dtdDestroy(_dtd, (XML_Bool)!parentParser, &parser->m_mem);
1126 FREE((void *)atts);
1127 FREE(groupConnector);
1128 FREE(buffer);
1129 FREE(dataBuf);
1130 FREE(nsAtts);
1131 FREE(unknownEncodingMem);
1132 if (unknownEncodingRelease)
1133 unknownEncodingRelease(unknownEncodingData);
1134 FREE(parser);
1137 void XMLCALL
1138 XML_UseParserAsHandlerArg(XML_Parser parser)
1140 handlerArg = parser;
1143 enum XML_Error XMLCALL
1144 XML_UseForeignDTD(XML_Parser parser, XML_Bool useDTD)
1146 #ifdef XML_DTD
1147 /* block after XML_Parse()/XML_ParseBuffer() has been called */
1148 if (ps_parsing == XML_PARSING || ps_parsing == XML_SUSPENDED)
1149 return XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING;
1150 useForeignDTD = useDTD;
1151 return XML_ERROR_NONE;
1152 #else
1153 return XML_ERROR_FEATURE_REQUIRES_XML_DTD;
1154 #endif
1157 void XMLCALL
1158 XML_SetReturnNSTriplet(XML_Parser parser, int do_nst)
1160 /* block after XML_Parse()/XML_ParseBuffer() has been called */
1161 if (ps_parsing == XML_PARSING || ps_parsing == XML_SUSPENDED)
1162 return;
1163 ns_triplets = do_nst ? XML_TRUE : XML_FALSE;
1166 void XMLCALL
1167 XML_SetUserData(XML_Parser parser, void *p)
1169 if (handlerArg == userData)
1170 handlerArg = userData = p;
1171 else
1172 userData = p;
1175 enum XML_Status XMLCALL
1176 XML_SetBase(XML_Parser parser, const XML_Char *p)
1178 if (p) {
1179 p = poolCopyString(&_dtd->pool, p);
1180 if (!p)
1181 return XML_STATUS_ERROR;
1182 curBase = p;
1184 else
1185 curBase = NULL;
1186 return XML_STATUS_OK;
1189 const XML_Char * XMLCALL
1190 XML_GetBase(XML_Parser parser)
1192 return curBase;
1195 int XMLCALL
1196 XML_GetSpecifiedAttributeCount(XML_Parser parser)
1198 return nSpecifiedAtts;
1201 int XMLCALL
1202 XML_GetIdAttributeIndex(XML_Parser parser)
1204 return idAttIndex;
1207 void XMLCALL
1208 XML_SetElementHandler(XML_Parser parser,
1209 XML_StartElementHandler start,
1210 XML_EndElementHandler end)
1212 startElementHandler = start;
1213 endElementHandler = end;
1216 void XMLCALL
1217 XML_SetStartElementHandler(XML_Parser parser,
1218 XML_StartElementHandler start) {
1219 startElementHandler = start;
1222 void XMLCALL
1223 XML_SetEndElementHandler(XML_Parser parser,
1224 XML_EndElementHandler end) {
1225 endElementHandler = end;
1228 void XMLCALL
1229 XML_SetCharacterDataHandler(XML_Parser parser,
1230 XML_CharacterDataHandler handler)
1232 characterDataHandler = handler;
1235 void XMLCALL
1236 XML_SetProcessingInstructionHandler(XML_Parser parser,
1237 XML_ProcessingInstructionHandler handler)
1239 processingInstructionHandler = handler;
1242 void XMLCALL
1243 XML_SetCommentHandler(XML_Parser parser,
1244 XML_CommentHandler handler)
1246 commentHandler = handler;
1249 void XMLCALL
1250 XML_SetCdataSectionHandler(XML_Parser parser,
1251 XML_StartCdataSectionHandler start,
1252 XML_EndCdataSectionHandler end)
1254 startCdataSectionHandler = start;
1255 endCdataSectionHandler = end;
1258 void XMLCALL
1259 XML_SetStartCdataSectionHandler(XML_Parser parser,
1260 XML_StartCdataSectionHandler start) {
1261 startCdataSectionHandler = start;
1264 void XMLCALL
1265 XML_SetEndCdataSectionHandler(XML_Parser parser,
1266 XML_EndCdataSectionHandler end) {
1267 endCdataSectionHandler = end;
1270 void XMLCALL
1271 XML_SetDefaultHandler(XML_Parser parser,
1272 XML_DefaultHandler handler)
1274 defaultHandler = handler;
1275 defaultExpandInternalEntities = XML_FALSE;
1278 void XMLCALL
1279 XML_SetDefaultHandlerExpand(XML_Parser parser,
1280 XML_DefaultHandler handler)
1282 defaultHandler = handler;
1283 defaultExpandInternalEntities = XML_TRUE;
1286 void XMLCALL
1287 XML_SetDoctypeDeclHandler(XML_Parser parser,
1288 XML_StartDoctypeDeclHandler start,
1289 XML_EndDoctypeDeclHandler end)
1291 startDoctypeDeclHandler = start;
1292 endDoctypeDeclHandler = end;
1295 void XMLCALL
1296 XML_SetStartDoctypeDeclHandler(XML_Parser parser,
1297 XML_StartDoctypeDeclHandler start) {
1298 startDoctypeDeclHandler = start;
1301 void XMLCALL
1302 XML_SetEndDoctypeDeclHandler(XML_Parser parser,
1303 XML_EndDoctypeDeclHandler end) {
1304 endDoctypeDeclHandler = end;
1307 void XMLCALL
1308 XML_SetUnparsedEntityDeclHandler(XML_Parser parser,
1309 XML_UnparsedEntityDeclHandler handler)
1311 unparsedEntityDeclHandler = handler;
1314 void XMLCALL
1315 XML_SetNotationDeclHandler(XML_Parser parser,
1316 XML_NotationDeclHandler handler)
1318 notationDeclHandler = handler;
1321 void XMLCALL
1322 XML_SetNamespaceDeclHandler(XML_Parser parser,
1323 XML_StartNamespaceDeclHandler start,
1324 XML_EndNamespaceDeclHandler end)
1326 startNamespaceDeclHandler = start;
1327 endNamespaceDeclHandler = end;
1330 void XMLCALL
1331 XML_SetStartNamespaceDeclHandler(XML_Parser parser,
1332 XML_StartNamespaceDeclHandler start) {
1333 startNamespaceDeclHandler = start;
1336 void XMLCALL
1337 XML_SetEndNamespaceDeclHandler(XML_Parser parser,
1338 XML_EndNamespaceDeclHandler end) {
1339 endNamespaceDeclHandler = end;
1342 void XMLCALL
1343 XML_SetNotStandaloneHandler(XML_Parser parser,
1344 XML_NotStandaloneHandler handler)
1346 notStandaloneHandler = handler;
1349 void XMLCALL
1350 XML_SetExternalEntityRefHandler(XML_Parser parser,
1351 XML_ExternalEntityRefHandler handler)
1353 externalEntityRefHandler = handler;
1356 void XMLCALL
1357 XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg)
1359 if (arg)
1360 externalEntityRefHandlerArg = (XML_Parser)arg;
1361 else
1362 externalEntityRefHandlerArg = parser;
1365 void XMLCALL
1366 XML_SetSkippedEntityHandler(XML_Parser parser,
1367 XML_SkippedEntityHandler handler)
1369 skippedEntityHandler = handler;
1372 void XMLCALL
1373 XML_SetUnknownEncodingHandler(XML_Parser parser,
1374 XML_UnknownEncodingHandler handler,
1375 void *data)
1377 unknownEncodingHandler = handler;
1378 unknownEncodingHandlerData = data;
1381 void XMLCALL
1382 XML_SetElementDeclHandler(XML_Parser parser,
1383 XML_ElementDeclHandler eldecl)
1385 elementDeclHandler = eldecl;
1388 void XMLCALL
1389 XML_SetAttlistDeclHandler(XML_Parser parser,
1390 XML_AttlistDeclHandler attdecl)
1392 attlistDeclHandler = attdecl;
1395 void XMLCALL
1396 XML_SetEntityDeclHandler(XML_Parser parser,
1397 XML_EntityDeclHandler handler)
1399 entityDeclHandler = handler;
1402 void XMLCALL
1403 XML_SetXmlDeclHandler(XML_Parser parser,
1404 XML_XmlDeclHandler handler) {
1405 xmlDeclHandler = handler;
1408 int XMLCALL
1409 XML_SetParamEntityParsing(XML_Parser parser,
1410 enum XML_ParamEntityParsing peParsing)
1412 /* block after XML_Parse()/XML_ParseBuffer() has been called */
1413 if (ps_parsing == XML_PARSING || ps_parsing == XML_SUSPENDED)
1414 return 0;
1415 #ifdef XML_DTD
1416 paramEntityParsing = peParsing;
1417 return 1;
1418 #else
1419 return peParsing == XML_PARAM_ENTITY_PARSING_NEVER;
1420 #endif
1423 enum XML_Status XMLCALL
1424 XML_Parse(XML_Parser parser, const char *s, int len, int isFinal)
1426 switch (ps_parsing) {
1427 case XML_SUSPENDED:
1428 errorCode = XML_ERROR_SUSPENDED;
1429 return XML_STATUS_ERROR;
1430 case XML_FINISHED:
1431 errorCode = XML_ERROR_FINISHED;
1432 return XML_STATUS_ERROR;
1433 default:
1434 ps_parsing = XML_PARSING;
1437 if (len == 0) {
1438 ps_finalBuffer = (XML_Bool)isFinal;
1439 if (!isFinal)
1440 return XML_STATUS_OK;
1441 positionPtr = bufferPtr;
1442 parseEndPtr = bufferEnd;
1444 /* If data are left over from last buffer, and we now know that these
1445 data are the final chunk of input, then we have to check them again
1446 to detect errors based on that fact.
1448 errorCode = processor(parser, bufferPtr, parseEndPtr, &bufferPtr);
1450 if (errorCode == XML_ERROR_NONE) {
1451 switch (ps_parsing) {
1452 case XML_SUSPENDED:
1453 XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
1454 positionPtr = bufferPtr;
1455 return XML_STATUS_SUSPENDED;
1456 case XML_INITIALIZED:
1457 case XML_PARSING:
1458 ps_parsing = XML_FINISHED;
1459 /* fall through */
1460 default:
1461 return XML_STATUS_OK;
1464 eventEndPtr = eventPtr;
1465 processor = errorProcessor;
1466 return XML_STATUS_ERROR;
1468 #ifndef XML_CONTEXT_BYTES
1469 else if (bufferPtr == bufferEnd) {
1470 const char *end;
1471 int nLeftOver;
1472 enum XML_Error result;
1473 parseEndByteIndex += len;
1474 positionPtr = s;
1475 ps_finalBuffer = (XML_Bool)isFinal;
1477 errorCode = processor(parser, s, parseEndPtr = s + len, &end);
1479 if (errorCode != XML_ERROR_NONE) {
1480 eventEndPtr = eventPtr;
1481 processor = errorProcessor;
1482 return XML_STATUS_ERROR;
1484 else {
1485 switch (ps_parsing) {
1486 case XML_SUSPENDED:
1487 result = XML_STATUS_SUSPENDED;
1488 break;
1489 case XML_INITIALIZED:
1490 case XML_PARSING:
1491 result = XML_STATUS_OK;
1492 if (isFinal) {
1493 ps_parsing = XML_FINISHED;
1494 return result;
1499 XmlUpdatePosition(encoding, positionPtr, end, &position);
1500 nLeftOver = s + len - end;
1501 if (nLeftOver) {
1502 if (buffer == NULL || nLeftOver > bufferLim - buffer) {
1503 /* FIXME avoid integer overflow */
1504 char *temp;
1505 temp = (buffer == NULL
1506 ? (char *)MALLOC(len * 2)
1507 : (char *)REALLOC(buffer, len * 2));
1508 if (temp == NULL) {
1509 errorCode = XML_ERROR_NO_MEMORY;
1510 return XML_STATUS_ERROR;
1512 buffer = temp;
1513 if (!buffer) {
1514 errorCode = XML_ERROR_NO_MEMORY;
1515 eventPtr = eventEndPtr = NULL;
1516 processor = errorProcessor;
1517 return XML_STATUS_ERROR;
1519 bufferLim = buffer + len * 2;
1521 memcpy(buffer, end, nLeftOver);
1523 bufferPtr = buffer;
1524 bufferEnd = buffer + nLeftOver;
1525 positionPtr = bufferPtr;
1526 parseEndPtr = bufferEnd;
1527 eventPtr = bufferPtr;
1528 eventEndPtr = bufferPtr;
1529 return result;
1531 #endif /* not defined XML_CONTEXT_BYTES */
1532 else {
1533 void *buff = XML_GetBuffer(parser, len);
1534 if (buff == NULL)
1535 return XML_STATUS_ERROR;
1536 else {
1537 memcpy(buff, s, len);
1538 return XML_ParseBuffer(parser, len, isFinal);
1543 enum XML_Status XMLCALL
1544 XML_ParseBuffer(XML_Parser parser, int len, int isFinal)
1546 const char *start;
1547 enum XML_Status result = XML_STATUS_OK;
1549 switch (ps_parsing) {
1550 case XML_SUSPENDED:
1551 errorCode = XML_ERROR_SUSPENDED;
1552 return XML_STATUS_ERROR;
1553 case XML_FINISHED:
1554 errorCode = XML_ERROR_FINISHED;
1555 return XML_STATUS_ERROR;
1556 default:
1557 ps_parsing = XML_PARSING;
1560 start = bufferPtr;
1561 positionPtr = start;
1562 bufferEnd += len;
1563 parseEndPtr = bufferEnd;
1564 parseEndByteIndex += len;
1565 ps_finalBuffer = (XML_Bool)isFinal;
1567 errorCode = processor(parser, start, parseEndPtr, &bufferPtr);
1569 if (errorCode != XML_ERROR_NONE) {
1570 eventEndPtr = eventPtr;
1571 processor = errorProcessor;
1572 return XML_STATUS_ERROR;
1574 else {
1575 switch (ps_parsing) {
1576 case XML_SUSPENDED:
1577 result = XML_STATUS_SUSPENDED;
1578 break;
1579 case XML_INITIALIZED:
1580 case XML_PARSING:
1581 if (isFinal) {
1582 ps_parsing = XML_FINISHED;
1583 return result;
1585 default: ; /* should not happen */
1589 XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
1590 positionPtr = bufferPtr;
1591 return result;
1594 void * XMLCALL
1595 XML_GetBuffer(XML_Parser parser, int len)
1597 switch (ps_parsing) {
1598 case XML_SUSPENDED:
1599 errorCode = XML_ERROR_SUSPENDED;
1600 return NULL;
1601 case XML_FINISHED:
1602 errorCode = XML_ERROR_FINISHED;
1603 return NULL;
1604 default: ;
1607 if (len > bufferLim - bufferEnd) {
1608 /* FIXME avoid integer overflow */
1609 int neededSize = len + (int)(bufferEnd - bufferPtr);
1610 #ifdef XML_CONTEXT_BYTES
1611 int keep = (int)(bufferPtr - buffer);
1613 if (keep > XML_CONTEXT_BYTES)
1614 keep = XML_CONTEXT_BYTES;
1615 neededSize += keep;
1616 #endif /* defined XML_CONTEXT_BYTES */
1617 if (neededSize <= bufferLim - buffer) {
1618 #ifdef XML_CONTEXT_BYTES
1619 if (keep < bufferPtr - buffer) {
1620 int offset = (int)(bufferPtr - buffer) - keep;
1621 memmove(buffer, &buffer[offset], bufferEnd - bufferPtr + keep);
1622 bufferEnd -= offset;
1623 bufferPtr -= offset;
1625 #else
1626 memmove(buffer, bufferPtr, bufferEnd - bufferPtr);
1627 bufferEnd = buffer + (bufferEnd - bufferPtr);
1628 bufferPtr = buffer;
1629 #endif /* not defined XML_CONTEXT_BYTES */
1631 else {
1632 char *newBuf;
1633 int bufferSize = (int)(bufferLim - bufferPtr);
1634 if (bufferSize == 0)
1635 bufferSize = INIT_BUFFER_SIZE;
1636 do {
1637 bufferSize *= 2;
1638 } while (bufferSize < neededSize);
1639 newBuf = (char *)MALLOC(bufferSize);
1640 if (newBuf == 0) {
1641 errorCode = XML_ERROR_NO_MEMORY;
1642 return NULL;
1644 bufferLim = newBuf + bufferSize;
1645 #ifdef XML_CONTEXT_BYTES
1646 if (bufferPtr) {
1647 int keep = (int)(bufferPtr - buffer);
1648 if (keep > XML_CONTEXT_BYTES)
1649 keep = XML_CONTEXT_BYTES;
1650 memcpy(newBuf, &bufferPtr[-keep], bufferEnd - bufferPtr + keep);
1651 FREE(buffer);
1652 buffer = newBuf;
1653 bufferEnd = buffer + (bufferEnd - bufferPtr) + keep;
1654 bufferPtr = buffer + keep;
1656 else {
1657 bufferEnd = newBuf + (bufferEnd - bufferPtr);
1658 bufferPtr = buffer = newBuf;
1660 #else
1661 if (bufferPtr) {
1662 memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr);
1663 FREE(buffer);
1665 bufferEnd = newBuf + (bufferEnd - bufferPtr);
1666 bufferPtr = buffer = newBuf;
1667 #endif /* not defined XML_CONTEXT_BYTES */
1670 return bufferEnd;
1673 enum XML_Status XMLCALL
1674 XML_StopParser(XML_Parser parser, XML_Bool resumable)
1676 switch (ps_parsing) {
1677 case XML_SUSPENDED:
1678 if (resumable) {
1679 errorCode = XML_ERROR_SUSPENDED;
1680 return XML_STATUS_ERROR;
1682 ps_parsing = XML_FINISHED;
1683 break;
1684 case XML_FINISHED:
1685 errorCode = XML_ERROR_FINISHED;
1686 return XML_STATUS_ERROR;
1687 default:
1688 if (resumable) {
1689 #ifdef XML_DTD
1690 if (isParamEntity) {
1691 errorCode = XML_ERROR_SUSPEND_PE;
1692 return XML_STATUS_ERROR;
1694 #endif
1695 ps_parsing = XML_SUSPENDED;
1697 else
1698 ps_parsing = XML_FINISHED;
1700 return XML_STATUS_OK;
1703 enum XML_Status XMLCALL
1704 XML_ResumeParser(XML_Parser parser)
1706 enum XML_Status result = XML_STATUS_OK;
1708 if (ps_parsing != XML_SUSPENDED) {
1709 errorCode = XML_ERROR_NOT_SUSPENDED;
1710 return XML_STATUS_ERROR;
1712 ps_parsing = XML_PARSING;
1714 errorCode = processor(parser, bufferPtr, parseEndPtr, &bufferPtr);
1716 if (errorCode != XML_ERROR_NONE) {
1717 eventEndPtr = eventPtr;
1718 processor = errorProcessor;
1719 return XML_STATUS_ERROR;
1721 else {
1722 switch (ps_parsing) {
1723 case XML_SUSPENDED:
1724 result = XML_STATUS_SUSPENDED;
1725 break;
1726 case XML_INITIALIZED:
1727 case XML_PARSING:
1728 if (ps_finalBuffer) {
1729 ps_parsing = XML_FINISHED;
1730 return result;
1732 default: ;
1736 XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
1737 positionPtr = bufferPtr;
1738 return result;
1741 void XMLCALL
1742 XML_GetParsingStatus(XML_Parser parser, XML_ParsingStatus *status)
1744 assert(status != NULL);
1745 *status = parser->m_parsingStatus;
1748 enum XML_Error XMLCALL
1749 XML_GetErrorCode(XML_Parser parser)
1751 return errorCode;
1754 XML_Index XMLCALL
1755 XML_GetCurrentByteIndex(XML_Parser parser)
1757 if (eventPtr)
1758 return parseEndByteIndex - (parseEndPtr - eventPtr);
1759 return -1;
1762 int XMLCALL
1763 XML_GetCurrentByteCount(XML_Parser parser)
1765 if (eventEndPtr && eventPtr)
1766 return (int)(eventEndPtr - eventPtr);
1767 return 0;
1770 const char * XMLCALL
1771 XML_GetInputContext(XML_Parser parser, int *offset, int *size)
1773 #ifdef XML_CONTEXT_BYTES
1774 if (eventPtr && buffer) {
1775 *offset = (int)(eventPtr - buffer);
1776 *size = (int)(bufferEnd - buffer);
1777 return buffer;
1779 #endif /* defined XML_CONTEXT_BYTES */
1780 return (char *) 0;
1783 XML_Size XMLCALL
1784 XML_GetCurrentLineNumber(XML_Parser parser)
1786 if (eventPtr && eventPtr >= positionPtr) {
1787 XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
1788 positionPtr = eventPtr;
1790 return position.lineNumber + 1;
1793 XML_Size XMLCALL
1794 XML_GetCurrentColumnNumber(XML_Parser parser)
1796 if (eventPtr && eventPtr >= positionPtr) {
1797 XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
1798 positionPtr = eventPtr;
1800 return position.columnNumber;
1803 void XMLCALL
1804 XML_FreeContentModel(XML_Parser parser, XML_Content *model)
1806 FREE(model);
1809 void * XMLCALL
1810 XML_MemMalloc(XML_Parser parser, size_t size)
1812 return MALLOC(size);
1815 void * XMLCALL
1816 XML_MemRealloc(XML_Parser parser, void *ptr, size_t size)
1818 return REALLOC(ptr, size);
1821 void XMLCALL
1822 XML_MemFree(XML_Parser parser, void *ptr)
1824 FREE(ptr);
1827 void XMLCALL
1828 XML_DefaultCurrent(XML_Parser parser)
1830 if (defaultHandler) {
1831 if (openInternalEntities)
1832 reportDefault(parser,
1833 internalEncoding,
1834 openInternalEntities->internalEventPtr,
1835 openInternalEntities->internalEventEndPtr);
1836 else
1837 reportDefault(parser, encoding, eventPtr, eventEndPtr);
1841 const XML_LChar * XMLCALL
1842 XML_ErrorString(enum XML_Error code)
1844 static const XML_LChar* const message[] = {
1846 XML_L("out of memory"),
1847 XML_L("syntax error"),
1848 XML_L("no element found"),
1849 XML_L("not well-formed (invalid token)"),
1850 XML_L("unclosed token"),
1851 XML_L("partial character"),
1852 XML_L("mismatched tag"),
1853 XML_L("duplicate attribute"),
1854 XML_L("junk after document element"),
1855 XML_L("illegal parameter entity reference"),
1856 XML_L("undefined entity"),
1857 XML_L("recursive entity reference"),
1858 XML_L("asynchronous entity"),
1859 XML_L("reference to invalid character number"),
1860 XML_L("reference to binary entity"),
1861 XML_L("reference to external entity in attribute"),
1862 XML_L("XML or text declaration not at start of entity"),
1863 XML_L("unknown encoding"),
1864 XML_L("encoding specified in XML declaration is incorrect"),
1865 XML_L("unclosed CDATA section"),
1866 XML_L("error in processing external entity reference"),
1867 XML_L("document is not standalone"),
1868 XML_L("unexpected parser state - please send a bug report"),
1869 XML_L("entity declared in parameter entity"),
1870 XML_L("requested feature requires XML_DTD support in Expat"),
1871 XML_L("cannot change setting once parsing has begun"),
1872 XML_L("unbound prefix"),
1873 XML_L("must not undeclare prefix"),
1874 XML_L("incomplete markup in parameter entity"),
1875 XML_L("XML declaration not well-formed"),
1876 XML_L("text declaration not well-formed"),
1877 XML_L("illegal character(s) in public id"),
1878 XML_L("parser suspended"),
1879 XML_L("parser not suspended"),
1880 XML_L("parsing aborted"),
1881 XML_L("parsing finished"),
1882 XML_L("cannot suspend in external parameter entity"),
1883 XML_L("reserved prefix (xml) must not be undeclared or bound to another namespace name"),
1884 XML_L("reserved prefix (xmlns) must not be declared or undeclared"),
1885 XML_L("prefix must not be bound to one of the reserved namespace names")
1887 if (code > 0 && code < sizeof(message)/sizeof(message[0]))
1888 return message[code];
1889 return NULL;
1892 const XML_LChar * XMLCALL
1893 XML_ExpatVersion(void) {
1895 /* V1 is used to string-ize the version number. However, it would
1896 string-ize the actual version macro *names* unless we get them
1897 substituted before being passed to V1. CPP is defined to expand
1898 a macro, then rescan for more expansions. Thus, we use V2 to expand
1899 the version macros, then CPP will expand the resulting V1() macro
1900 with the correct numerals. */
1901 /* ### I'm assuming cpp is portable in this respect... */
1903 #define V1(a,b,c) XML_L(#a)XML_L(".")XML_L(#b)XML_L(".")XML_L(#c)
1904 #define V2(a,b,c) XML_L("expat_")V1(a,b,c)
1906 return V2(XML_MAJOR_VERSION, XML_MINOR_VERSION, XML_MICRO_VERSION);
1908 #undef V1
1909 #undef V2
1912 XML_Expat_Version XMLCALL
1913 XML_ExpatVersionInfo(void)
1915 XML_Expat_Version version;
1917 version.major = XML_MAJOR_VERSION;
1918 version.minor = XML_MINOR_VERSION;
1919 version.micro = XML_MICRO_VERSION;
1921 return version;
1924 const XML_Feature * XMLCALL
1925 XML_GetFeatureList(void)
1927 static const XML_Feature features[] = {
1928 {XML_FEATURE_SIZEOF_XML_CHAR, XML_L("sizeof(XML_Char)"),
1929 sizeof(XML_Char)},
1930 {XML_FEATURE_SIZEOF_XML_LCHAR, XML_L("sizeof(XML_LChar)"),
1931 sizeof(XML_LChar)},
1932 #ifdef XML_UNICODE
1933 {XML_FEATURE_UNICODE, XML_L("XML_UNICODE"), 0},
1934 #endif
1935 #ifdef XML_UNICODE_WCHAR_T
1936 {XML_FEATURE_UNICODE_WCHAR_T, XML_L("XML_UNICODE_WCHAR_T"), 0},
1937 #endif
1938 #ifdef XML_DTD
1939 {XML_FEATURE_DTD, XML_L("XML_DTD"), 0},
1940 #endif
1941 #ifdef XML_CONTEXT_BYTES
1942 {XML_FEATURE_CONTEXT_BYTES, XML_L("XML_CONTEXT_BYTES"),
1943 XML_CONTEXT_BYTES},
1944 #endif
1945 #ifdef XML_MIN_SIZE
1946 {XML_FEATURE_MIN_SIZE, XML_L("XML_MIN_SIZE"), 0},
1947 #endif
1948 #ifdef XML_NS
1949 {XML_FEATURE_NS, XML_L("XML_NS"), 0},
1950 #endif
1951 {XML_FEATURE_END, NULL, 0}
1954 return features;
1957 /* Initially tag->rawName always points into the parse buffer;
1958 for those TAG instances opened while the current parse buffer was
1959 processed, and not yet closed, we need to store tag->rawName in a more
1960 permanent location, since the parse buffer is about to be discarded.
1962 static XML_Bool
1963 storeRawNames(XML_Parser parser)
1965 TAG *tag = tagStack;
1966 while (tag) {
1967 int bufSize;
1968 int nameLen = sizeof(XML_Char) * (tag->name.strLen + 1);
1969 char *rawNameBuf = tag->buf + nameLen;
1970 /* Stop if already stored. Since tagStack is a stack, we can stop
1971 at the first entry that has already been copied; everything
1972 below it in the stack is already been accounted for in a
1973 previous call to this function.
1975 if (tag->rawName == rawNameBuf)
1976 break;
1977 /* For re-use purposes we need to ensure that the
1978 size of tag->buf is a multiple of sizeof(XML_Char).
1980 bufSize = nameLen + ROUND_UP(tag->rawNameLength, sizeof(XML_Char));
1981 if (bufSize > tag->bufEnd - tag->buf) {
1982 char *temp = (char *)REALLOC(tag->buf, bufSize);
1983 if (temp == NULL)
1984 return XML_FALSE;
1985 /* if tag->name.str points to tag->buf (only when namespace
1986 processing is off) then we have to update it
1988 if (tag->name.str == (XML_Char *)tag->buf)
1989 tag->name.str = (XML_Char *)temp;
1990 /* if tag->name.localPart is set (when namespace processing is on)
1991 then update it as well, since it will always point into tag->buf
1993 if (tag->name.localPart)
1994 tag->name.localPart = (XML_Char *)temp + (tag->name.localPart -
1995 (XML_Char *)tag->buf);
1996 tag->buf = temp;
1997 tag->bufEnd = temp + bufSize;
1998 rawNameBuf = temp + nameLen;
2000 memcpy(rawNameBuf, tag->rawName, tag->rawNameLength);
2001 tag->rawName = rawNameBuf;
2002 tag = tag->parent;
2004 return XML_TRUE;
2007 static enum XML_Error PTRCALL
2008 contentProcessor(XML_Parser parser,
2009 const char *start,
2010 const char *end,
2011 const char **endPtr)
2013 enum XML_Error result = doContent(parser, 0, encoding, start, end,
2014 endPtr, (XML_Bool)!ps_finalBuffer);
2015 if (result == XML_ERROR_NONE) {
2016 if (!storeRawNames(parser))
2017 return XML_ERROR_NO_MEMORY;
2019 return result;
2022 static enum XML_Error PTRCALL
2023 externalEntityInitProcessor(XML_Parser parser,
2024 const char *start,
2025 const char *end,
2026 const char **endPtr)
2028 enum XML_Error result = initializeEncoding(parser);
2029 if (result != XML_ERROR_NONE)
2030 return result;
2031 processor = externalEntityInitProcessor2;
2032 return externalEntityInitProcessor2(parser, start, end, endPtr);
2035 static enum XML_Error PTRCALL
2036 externalEntityInitProcessor2(XML_Parser parser,
2037 const char *start,
2038 const char *end,
2039 const char **endPtr)
2041 const char *next = start; /* XmlContentTok doesn't always set the last arg */
2042 int tok = XmlContentTok(encoding, start, end, &next);
2043 switch (tok) {
2044 case XML_TOK_BOM:
2045 /* If we are at the end of the buffer, this would cause the next stage,
2046 i.e. externalEntityInitProcessor3, to pass control directly to
2047 doContent (by detecting XML_TOK_NONE) without processing any xml text
2048 declaration - causing the error XML_ERROR_MISPLACED_XML_PI in doContent.
2050 if (next == end && !ps_finalBuffer) {
2051 *endPtr = next;
2052 return XML_ERROR_NONE;
2054 start = next;
2055 break;
2056 case XML_TOK_PARTIAL:
2057 if (!ps_finalBuffer) {
2058 *endPtr = start;
2059 return XML_ERROR_NONE;
2061 eventPtr = start;
2062 return XML_ERROR_UNCLOSED_TOKEN;
2063 case XML_TOK_PARTIAL_CHAR:
2064 if (!ps_finalBuffer) {
2065 *endPtr = start;
2066 return XML_ERROR_NONE;
2068 eventPtr = start;
2069 return XML_ERROR_PARTIAL_CHAR;
2071 processor = externalEntityInitProcessor3;
2072 return externalEntityInitProcessor3(parser, start, end, endPtr);
2075 static enum XML_Error PTRCALL
2076 externalEntityInitProcessor3(XML_Parser parser,
2077 const char *start,
2078 const char *end,
2079 const char **endPtr)
2081 int tok;
2082 const char *next = start; /* XmlContentTok doesn't always set the last arg */
2083 eventPtr = start;
2084 tok = XmlContentTok(encoding, start, end, &next);
2085 eventEndPtr = next;
2087 switch (tok) {
2088 case XML_TOK_XML_DECL:
2090 enum XML_Error result;
2091 result = processXmlDecl(parser, 1, start, next);
2092 if (result != XML_ERROR_NONE)
2093 return result;
2094 switch (ps_parsing) {
2095 case XML_SUSPENDED:
2096 *endPtr = next;
2097 return XML_ERROR_NONE;
2098 case XML_FINISHED:
2099 return XML_ERROR_ABORTED;
2100 default:
2101 start = next;
2104 break;
2105 case XML_TOK_PARTIAL:
2106 if (!ps_finalBuffer) {
2107 *endPtr = start;
2108 return XML_ERROR_NONE;
2110 return XML_ERROR_UNCLOSED_TOKEN;
2111 case XML_TOK_PARTIAL_CHAR:
2112 if (!ps_finalBuffer) {
2113 *endPtr = start;
2114 return XML_ERROR_NONE;
2116 return XML_ERROR_PARTIAL_CHAR;
2118 processor = externalEntityContentProcessor;
2119 tagLevel = 1;
2120 return externalEntityContentProcessor(parser, start, end, endPtr);
2123 static enum XML_Error PTRCALL
2124 externalEntityContentProcessor(XML_Parser parser,
2125 const char *start,
2126 const char *end,
2127 const char **endPtr)
2129 enum XML_Error result = doContent(parser, 1, encoding, start, end,
2130 endPtr, (XML_Bool)!ps_finalBuffer);
2131 if (result == XML_ERROR_NONE) {
2132 if (!storeRawNames(parser))
2133 return XML_ERROR_NO_MEMORY;
2135 return result;
2138 static enum XML_Error
2139 doContent(XML_Parser parser,
2140 int startTagLevel,
2141 const ENCODING *enc,
2142 const char *s,
2143 const char *end,
2144 const char **nextPtr,
2145 XML_Bool haveMore)
2147 /* save one level of indirection */
2148 DTD * const dtd = _dtd;
2150 const char **eventPP;
2151 const char **eventEndPP;
2152 if (enc == encoding) {
2153 eventPP = &eventPtr;
2154 eventEndPP = &eventEndPtr;
2156 else {
2157 eventPP = &(openInternalEntities->internalEventPtr);
2158 eventEndPP = &(openInternalEntities->internalEventEndPtr);
2160 *eventPP = s;
2162 for (;;) {
2163 const char *next = s; /* XmlContentTok doesn't always set the last arg */
2164 int tok = XmlContentTok(enc, s, end, &next);
2165 *eventEndPP = next;
2166 switch (tok) {
2167 case XML_TOK_TRAILING_CR:
2168 if (haveMore) {
2169 *nextPtr = s;
2170 return XML_ERROR_NONE;
2172 *eventEndPP = end;
2173 if (characterDataHandler) {
2174 XML_Char c = 0xA;
2175 characterDataHandler(handlerArg, &c, 1);
2177 else if (defaultHandler)
2178 reportDefault(parser, enc, s, end);
2179 /* We are at the end of the final buffer, should we check for
2180 XML_SUSPENDED, XML_FINISHED?
2182 if (startTagLevel == 0)
2183 return XML_ERROR_NO_ELEMENTS;
2184 if (tagLevel != startTagLevel)
2185 return XML_ERROR_ASYNC_ENTITY;
2186 *nextPtr = end;
2187 return XML_ERROR_NONE;
2188 case XML_TOK_NONE:
2189 if (haveMore) {
2190 *nextPtr = s;
2191 return XML_ERROR_NONE;
2193 if (startTagLevel > 0) {
2194 if (tagLevel != startTagLevel)
2195 return XML_ERROR_ASYNC_ENTITY;
2196 *nextPtr = s;
2197 return XML_ERROR_NONE;
2199 return XML_ERROR_NO_ELEMENTS;
2200 case XML_TOK_INVALID:
2201 *eventPP = next;
2202 return XML_ERROR_INVALID_TOKEN;
2203 case XML_TOK_PARTIAL:
2204 if (haveMore) {
2205 *nextPtr = s;
2206 return XML_ERROR_NONE;
2208 return XML_ERROR_UNCLOSED_TOKEN;
2209 case XML_TOK_PARTIAL_CHAR:
2210 if (haveMore) {
2211 *nextPtr = s;
2212 return XML_ERROR_NONE;
2214 return XML_ERROR_PARTIAL_CHAR;
2215 case XML_TOK_ENTITY_REF:
2217 const XML_Char *name;
2218 ENTITY *entity;
2219 XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc,
2220 s + enc->minBytesPerChar,
2221 next - enc->minBytesPerChar);
2222 if (ch) {
2223 if (characterDataHandler)
2224 characterDataHandler(handlerArg, &ch, 1);
2225 else if (defaultHandler)
2226 reportDefault(parser, enc, s, next);
2227 break;
2229 name = poolStoreString(&dtd->pool, enc,
2230 s + enc->minBytesPerChar,
2231 next - enc->minBytesPerChar);
2232 if (!name)
2233 return XML_ERROR_NO_MEMORY;
2234 entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0);
2235 poolDiscard(&dtd->pool);
2236 /* First, determine if a check for an existing declaration is needed;
2237 if yes, check that the entity exists, and that it is internal,
2238 otherwise call the skipped entity or default handler.
2240 if (!dtd->hasParamEntityRefs || dtd->standalone) {
2241 if (!entity)
2242 return XML_ERROR_UNDEFINED_ENTITY;
2243 else if (!entity->is_internal)
2244 return XML_ERROR_ENTITY_DECLARED_IN_PE;
2246 else if (!entity) {
2247 if (skippedEntityHandler)
2248 skippedEntityHandler(handlerArg, name, 0);
2249 else if (defaultHandler)
2250 reportDefault(parser, enc, s, next);
2251 break;
2253 if (entity->open)
2254 return XML_ERROR_RECURSIVE_ENTITY_REF;
2255 if (entity->notation)
2256 return XML_ERROR_BINARY_ENTITY_REF;
2257 if (entity->textPtr) {
2258 enum XML_Error result;
2259 if (!defaultExpandInternalEntities) {
2260 if (skippedEntityHandler)
2261 skippedEntityHandler(handlerArg, entity->name, 0);
2262 else if (defaultHandler)
2263 reportDefault(parser, enc, s, next);
2264 break;
2266 result = processInternalEntity(parser, entity, XML_FALSE);
2267 if (result != XML_ERROR_NONE)
2268 return result;
2270 else if (externalEntityRefHandler) {
2271 const XML_Char *context;
2272 entity->open = XML_TRUE;
2273 context = getContext(parser);
2274 entity->open = XML_FALSE;
2275 if (!context)
2276 return XML_ERROR_NO_MEMORY;
2277 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
2278 context,
2279 entity->base,
2280 entity->systemId,
2281 entity->publicId))
2282 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
2283 poolDiscard(&tempPool);
2285 else if (defaultHandler)
2286 reportDefault(parser, enc, s, next);
2287 break;
2289 case XML_TOK_START_TAG_NO_ATTS:
2290 /* fall through */
2291 case XML_TOK_START_TAG_WITH_ATTS:
2293 TAG *tag;
2294 enum XML_Error result;
2295 XML_Char *toPtr;
2296 if (freeTagList) {
2297 tag = freeTagList;
2298 freeTagList = freeTagList->parent;
2300 else {
2301 tag = (TAG *)MALLOC(sizeof(TAG));
2302 if (!tag)
2303 return XML_ERROR_NO_MEMORY;
2304 tag->buf = (char *)MALLOC(INIT_TAG_BUF_SIZE);
2305 if (!tag->buf) {
2306 FREE(tag);
2307 return XML_ERROR_NO_MEMORY;
2309 tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE;
2311 tag->bindings = NULL;
2312 tag->parent = tagStack;
2313 tagStack = tag;
2314 tag->name.localPart = NULL;
2315 tag->name.prefix = NULL;
2316 tag->rawName = s + enc->minBytesPerChar;
2317 tag->rawNameLength = XmlNameLength(enc, tag->rawName);
2318 ++tagLevel;
2320 const char *rawNameEnd = tag->rawName + tag->rawNameLength;
2321 const char *fromPtr = tag->rawName;
2322 toPtr = (XML_Char *)tag->buf;
2323 for (;;) {
2324 int bufSize;
2325 int convLen;
2326 XmlConvert(enc,
2327 &fromPtr, rawNameEnd,
2328 (ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1);
2329 convLen = (int)(toPtr - (XML_Char *)tag->buf);
2330 if (fromPtr == rawNameEnd) {
2331 tag->name.strLen = convLen;
2332 break;
2334 bufSize = (int)(tag->bufEnd - tag->buf) << 1;
2336 char *temp = (char *)REALLOC(tag->buf, bufSize);
2337 if (temp == NULL)
2338 return XML_ERROR_NO_MEMORY;
2339 tag->buf = temp;
2340 tag->bufEnd = temp + bufSize;
2341 toPtr = (XML_Char *)temp + convLen;
2345 tag->name.str = (XML_Char *)tag->buf;
2346 *toPtr = XML_T('\0');
2347 result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings));
2348 if (result)
2349 return result;
2350 if (startElementHandler)
2351 startElementHandler(handlerArg, tag->name.str,
2352 (const XML_Char **)atts);
2353 else if (defaultHandler)
2354 reportDefault(parser, enc, s, next);
2355 poolClear(&tempPool);
2356 break;
2358 case XML_TOK_EMPTY_ELEMENT_NO_ATTS:
2359 /* fall through */
2360 case XML_TOK_EMPTY_ELEMENT_WITH_ATTS:
2362 const char *rawName = s + enc->minBytesPerChar;
2363 enum XML_Error result;
2364 BINDING *bindings = NULL;
2365 XML_Bool noElmHandlers = XML_TRUE;
2366 TAG_NAME name;
2367 name.str = poolStoreString(&tempPool, enc, rawName,
2368 rawName + XmlNameLength(enc, rawName));
2369 if (!name.str)
2370 return XML_ERROR_NO_MEMORY;
2371 poolFinish(&tempPool);
2372 result = storeAtts(parser, enc, s, &name, &bindings);
2373 if (result)
2374 return result;
2375 poolFinish(&tempPool);
2376 if (startElementHandler) {
2377 startElementHandler(handlerArg, name.str, (const XML_Char **)atts);
2378 noElmHandlers = XML_FALSE;
2380 if (endElementHandler) {
2381 if (startElementHandler)
2382 *eventPP = *eventEndPP;
2383 endElementHandler(handlerArg, name.str);
2384 noElmHandlers = XML_FALSE;
2386 if (noElmHandlers && defaultHandler)
2387 reportDefault(parser, enc, s, next);
2388 poolClear(&tempPool);
2389 while (bindings) {
2390 BINDING *b = bindings;
2391 if (endNamespaceDeclHandler)
2392 endNamespaceDeclHandler(handlerArg, b->prefix->name);
2393 bindings = bindings->nextTagBinding;
2394 b->nextTagBinding = freeBindingList;
2395 freeBindingList = b;
2396 b->prefix->binding = b->prevPrefixBinding;
2399 if (tagLevel == 0)
2400 return epilogProcessor(parser, next, end, nextPtr);
2401 break;
2402 case XML_TOK_END_TAG:
2403 if (tagLevel == startTagLevel)
2404 return XML_ERROR_ASYNC_ENTITY;
2405 else {
2406 int len;
2407 const char *rawName;
2408 TAG *tag = tagStack;
2409 tagStack = tag->parent;
2410 tag->parent = freeTagList;
2411 freeTagList = tag;
2412 rawName = s + enc->minBytesPerChar*2;
2413 len = XmlNameLength(enc, rawName);
2414 if (len != tag->rawNameLength
2415 || memcmp(tag->rawName, rawName, len) != 0) {
2416 *eventPP = rawName;
2417 return XML_ERROR_TAG_MISMATCH;
2419 --tagLevel;
2420 if (endElementHandler) {
2421 const XML_Char *localPart;
2422 const XML_Char *prefix;
2423 XML_Char *uri;
2424 localPart = tag->name.localPart;
2425 if (ns && localPart) {
2426 /* localPart and prefix may have been overwritten in
2427 tag->name.str, since this points to the binding->uri
2428 buffer which gets re-used; so we have to add them again
2430 uri = (XML_Char *)tag->name.str + tag->name.uriLen;
2431 /* don't need to check for space - already done in storeAtts() */
2432 while (*localPart) *uri++ = *localPart++;
2433 prefix = (XML_Char *)tag->name.prefix;
2434 if (ns_triplets && prefix) {
2435 *uri++ = namespaceSeparator;
2436 while (*prefix) *uri++ = *prefix++;
2438 *uri = XML_T('\0');
2440 endElementHandler(handlerArg, tag->name.str);
2442 else if (defaultHandler)
2443 reportDefault(parser, enc, s, next);
2444 while (tag->bindings) {
2445 BINDING *b = tag->bindings;
2446 if (endNamespaceDeclHandler)
2447 endNamespaceDeclHandler(handlerArg, b->prefix->name);
2448 tag->bindings = tag->bindings->nextTagBinding;
2449 b->nextTagBinding = freeBindingList;
2450 freeBindingList = b;
2451 b->prefix->binding = b->prevPrefixBinding;
2453 if (tagLevel == 0)
2454 return epilogProcessor(parser, next, end, nextPtr);
2456 break;
2457 case XML_TOK_CHAR_REF:
2459 int n = XmlCharRefNumber(enc, s);
2460 if (n < 0)
2461 return XML_ERROR_BAD_CHAR_REF;
2462 if (characterDataHandler) {
2463 XML_Char buf[XML_ENCODE_MAX];
2464 characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf));
2466 else if (defaultHandler)
2467 reportDefault(parser, enc, s, next);
2469 break;
2470 case XML_TOK_XML_DECL:
2471 return XML_ERROR_MISPLACED_XML_PI;
2472 case XML_TOK_DATA_NEWLINE:
2473 if (characterDataHandler) {
2474 XML_Char c = 0xA;
2475 characterDataHandler(handlerArg, &c, 1);
2477 else if (defaultHandler)
2478 reportDefault(parser, enc, s, next);
2479 break;
2480 case XML_TOK_CDATA_SECT_OPEN:
2482 enum XML_Error result;
2483 if (startCdataSectionHandler)
2484 startCdataSectionHandler(handlerArg);
2485 #if 0
2486 /* Suppose you doing a transformation on a document that involves
2487 changing only the character data. You set up a defaultHandler
2488 and a characterDataHandler. The defaultHandler simply copies
2489 characters through. The characterDataHandler does the
2490 transformation and writes the characters out escaping them as
2491 necessary. This case will fail to work if we leave out the
2492 following two lines (because & and < inside CDATA sections will
2493 be incorrectly escaped).
2495 However, now we have a start/endCdataSectionHandler, so it seems
2496 easier to let the user deal with this.
2498 else if (characterDataHandler)
2499 characterDataHandler(handlerArg, dataBuf, 0);
2500 #endif
2501 else if (defaultHandler)
2502 reportDefault(parser, enc, s, next);
2503 result = doCdataSection(parser, enc, &next, end, nextPtr, haveMore);
2504 if (result != XML_ERROR_NONE)
2505 return result;
2506 else if (!next) {
2507 processor = cdataSectionProcessor;
2508 return result;
2511 break;
2512 case XML_TOK_TRAILING_RSQB:
2513 if (haveMore) {
2514 *nextPtr = s;
2515 return XML_ERROR_NONE;
2517 if (characterDataHandler) {
2518 if (MUST_CONVERT(enc, s)) {
2519 ICHAR *dataPtr = (ICHAR *)dataBuf;
2520 XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd);
2521 characterDataHandler(handlerArg, dataBuf,
2522 (int)(dataPtr - (ICHAR *)dataBuf));
2524 else
2525 characterDataHandler(handlerArg,
2526 (XML_Char *)s,
2527 (int)((XML_Char *)end - (XML_Char *)s));
2529 else if (defaultHandler)
2530 reportDefault(parser, enc, s, end);
2531 /* We are at the end of the final buffer, should we check for
2532 XML_SUSPENDED, XML_FINISHED?
2534 if (startTagLevel == 0) {
2535 *eventPP = end;
2536 return XML_ERROR_NO_ELEMENTS;
2538 if (tagLevel != startTagLevel) {
2539 *eventPP = end;
2540 return XML_ERROR_ASYNC_ENTITY;
2542 *nextPtr = end;
2543 return XML_ERROR_NONE;
2544 case XML_TOK_DATA_CHARS:
2545 if (characterDataHandler) {
2546 if (MUST_CONVERT(enc, s)) {
2547 for (;;) {
2548 ICHAR *dataPtr = (ICHAR *)dataBuf;
2549 XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd);
2550 *eventEndPP = s;
2551 characterDataHandler(handlerArg, dataBuf,
2552 (int)(dataPtr - (ICHAR *)dataBuf));
2553 if (s == next)
2554 break;
2555 *eventPP = s;
2558 else
2559 characterDataHandler(handlerArg,
2560 (XML_Char *)s,
2561 (int)((XML_Char *)next - (XML_Char *)s));
2563 else if (defaultHandler)
2564 reportDefault(parser, enc, s, next);
2565 break;
2566 case XML_TOK_PI:
2567 if (!reportProcessingInstruction(parser, enc, s, next))
2568 return XML_ERROR_NO_MEMORY;
2569 break;
2570 case XML_TOK_COMMENT:
2571 if (!reportComment(parser, enc, s, next))
2572 return XML_ERROR_NO_MEMORY;
2573 break;
2574 default:
2575 if (defaultHandler)
2576 reportDefault(parser, enc, s, next);
2577 break;
2579 *eventPP = s = next;
2580 switch (ps_parsing) {
2581 case XML_SUSPENDED:
2582 *nextPtr = next;
2583 return XML_ERROR_NONE;
2584 case XML_FINISHED:
2585 return XML_ERROR_ABORTED;
2586 default: ;
2589 /* not reached */
2592 /* Precondition: all arguments must be non-NULL;
2593 Purpose:
2594 - normalize attributes
2595 - check attributes for well-formedness
2596 - generate namespace aware attribute names (URI, prefix)
2597 - build list of attributes for startElementHandler
2598 - default attributes
2599 - process namespace declarations (check and report them)
2600 - generate namespace aware element name (URI, prefix)
2602 static enum XML_Error
2603 storeAtts(XML_Parser parser, const ENCODING *enc,
2604 const char *attStr, TAG_NAME *tagNamePtr,
2605 BINDING **bindingsPtr)
2607 DTD * const dtd = _dtd; /* save one level of indirection */
2608 ELEMENT_TYPE *elementType;
2609 int nDefaultAtts;
2610 const XML_Char **appAtts; /* the attribute list for the application */
2611 int attIndex = 0;
2612 int prefixLen;
2613 int i;
2614 int n;
2615 XML_Char *uri;
2616 int nPrefixes = 0;
2617 BINDING *binding;
2618 const XML_Char *localPart;
2620 /* lookup the element type name */
2621 elementType = (ELEMENT_TYPE *)lookup(&dtd->elementTypes, tagNamePtr->str,0);
2622 if (!elementType) {
2623 const XML_Char *name = poolCopyString(&dtd->pool, tagNamePtr->str);
2624 if (!name)
2625 return XML_ERROR_NO_MEMORY;
2626 elementType = (ELEMENT_TYPE *)lookup(&dtd->elementTypes, name,
2627 sizeof(ELEMENT_TYPE));
2628 if (!elementType)
2629 return XML_ERROR_NO_MEMORY;
2630 if (ns && !setElementTypePrefix(parser, elementType))
2631 return XML_ERROR_NO_MEMORY;
2633 nDefaultAtts = elementType->nDefaultAtts;
2635 /* get the attributes from the tokenizer */
2636 n = XmlGetAttributes(enc, attStr, attsSize, atts);
2637 if (n + nDefaultAtts > attsSize) {
2638 int oldAttsSize = attsSize;
2639 ATTRIBUTE *temp;
2640 attsSize = n + nDefaultAtts + INIT_ATTS_SIZE;
2641 temp = (ATTRIBUTE *)REALLOC((void *)atts, attsSize * sizeof(ATTRIBUTE));
2642 if (temp == NULL)
2643 return XML_ERROR_NO_MEMORY;
2644 atts = temp;
2645 if (n > oldAttsSize)
2646 XmlGetAttributes(enc, attStr, n, atts);
2649 appAtts = (const XML_Char **)atts;
2650 for (i = 0; i < n; i++) {
2651 /* add the name and value to the attribute list */
2652 ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name,
2653 atts[i].name
2654 + XmlNameLength(enc, atts[i].name));
2655 if (!attId)
2656 return XML_ERROR_NO_MEMORY;
2657 /* Detect duplicate attributes by their QNames. This does not work when
2658 namespace processing is turned on and different prefixes for the same
2659 namespace are used. For this case we have a check further down.
2661 if ((attId->name)[-1]) {
2662 if (enc == encoding)
2663 eventPtr = atts[i].name;
2664 return XML_ERROR_DUPLICATE_ATTRIBUTE;
2666 (attId->name)[-1] = 1;
2667 appAtts[attIndex++] = attId->name;
2668 if (!atts[i].normalized) {
2669 enum XML_Error result;
2670 XML_Bool isCdata = XML_TRUE;
2672 /* figure out whether declared as other than CDATA */
2673 if (attId->maybeTokenized) {
2674 int j;
2675 for (j = 0; j < nDefaultAtts; j++) {
2676 if (attId == elementType->defaultAtts[j].id) {
2677 isCdata = elementType->defaultAtts[j].isCdata;
2678 break;
2683 /* normalize the attribute value */
2684 result = storeAttributeValue(parser, enc, isCdata,
2685 atts[i].valuePtr, atts[i].valueEnd,
2686 &tempPool);
2687 if (result)
2688 return result;
2689 appAtts[attIndex] = poolStart(&tempPool);
2690 poolFinish(&tempPool);
2692 else {
2693 /* the value did not need normalizing */
2694 appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr,
2695 atts[i].valueEnd);
2696 if (appAtts[attIndex] == 0)
2697 return XML_ERROR_NO_MEMORY;
2698 poolFinish(&tempPool);
2700 /* handle prefixed attribute names */
2701 if (attId->prefix) {
2702 if (attId->xmlns) {
2703 /* deal with namespace declarations here */
2704 enum XML_Error result = addBinding(parser, attId->prefix, attId,
2705 appAtts[attIndex], bindingsPtr);
2706 if (result)
2707 return result;
2708 --attIndex;
2710 else {
2711 /* deal with other prefixed names later */
2712 attIndex++;
2713 nPrefixes++;
2714 (attId->name)[-1] = 2;
2717 else
2718 attIndex++;
2721 /* set-up for XML_GetSpecifiedAttributeCount and XML_GetIdAttributeIndex */
2722 nSpecifiedAtts = attIndex;
2723 if (elementType->idAtt && (elementType->idAtt->name)[-1]) {
2724 for (i = 0; i < attIndex; i += 2)
2725 if (appAtts[i] == elementType->idAtt->name) {
2726 idAttIndex = i;
2727 break;
2730 else
2731 idAttIndex = -1;
2733 /* do attribute defaulting */
2734 for (i = 0; i < nDefaultAtts; i++) {
2735 const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + i;
2736 if (!(da->id->name)[-1] && da->value) {
2737 if (da->id->prefix) {
2738 if (da->id->xmlns) {
2739 enum XML_Error result = addBinding(parser, da->id->prefix, da->id,
2740 da->value, bindingsPtr);
2741 if (result)
2742 return result;
2744 else {
2745 (da->id->name)[-1] = 2;
2746 nPrefixes++;
2747 appAtts[attIndex++] = da->id->name;
2748 appAtts[attIndex++] = da->value;
2751 else {
2752 (da->id->name)[-1] = 1;
2753 appAtts[attIndex++] = da->id->name;
2754 appAtts[attIndex++] = da->value;
2758 appAtts[attIndex] = 0;
2760 /* expand prefixed attribute names, check for duplicates,
2761 and clear flags that say whether attributes were specified */
2762 i = 0;
2763 if (nPrefixes) {
2764 int j; /* hash table index */
2765 unsigned long version = nsAttsVersion;
2766 int nsAttsSize = (int)1 << nsAttsPower;
2767 /* size of hash table must be at least 2 * (# of prefixed attributes) */
2768 if ((nPrefixes << 1) >> nsAttsPower) { /* true for nsAttsPower = 0 */
2769 NS_ATT *temp;
2770 /* hash table size must also be a power of 2 and >= 8 */
2771 while (nPrefixes >> nsAttsPower++);
2772 if (nsAttsPower < 3)
2773 nsAttsPower = 3;
2774 nsAttsSize = (int)1 << nsAttsPower;
2775 temp = (NS_ATT *)REALLOC(nsAtts, nsAttsSize * sizeof(NS_ATT));
2776 if (!temp)
2777 return XML_ERROR_NO_MEMORY;
2778 nsAtts = temp;
2779 version = 0; /* force re-initialization of nsAtts hash table */
2781 /* using a version flag saves us from initializing nsAtts every time */
2782 if (!version) { /* initialize version flags when version wraps around */
2783 version = INIT_ATTS_VERSION;
2784 for (j = nsAttsSize; j != 0; )
2785 nsAtts[--j].version = version;
2787 nsAttsVersion = --version;
2789 /* expand prefixed names and check for duplicates */
2790 for (; i < attIndex; i += 2) {
2791 const XML_Char *s = appAtts[i];
2792 if (s[-1] == 2) { /* prefixed */
2793 ATTRIBUTE_ID *id;
2794 const BINDING *b;
2795 unsigned long uriHash = 0;
2796 ((XML_Char *)s)[-1] = 0; /* clear flag */
2797 id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, s, 0);
2798 if (!id)
2799 return XML_ERROR_NO_MEMORY;
2800 b = id->prefix->binding;
2801 if (!b)
2802 return XML_ERROR_UNBOUND_PREFIX;
2804 /* as we expand the name we also calculate its hash value */
2805 for (j = 0; j < b->uriLen; j++) {
2806 const XML_Char c = b->uri[j];
2807 if (!poolAppendChar(&tempPool, c))
2808 return XML_ERROR_NO_MEMORY;
2809 uriHash = CHAR_HASH(uriHash, c);
2811 while (*s++ != XML_T(':'))
2813 do { /* copies null terminator */
2814 const XML_Char c = *s;
2815 if (!poolAppendChar(&tempPool, *s))
2816 return XML_ERROR_NO_MEMORY;
2817 uriHash = CHAR_HASH(uriHash, c);
2818 } while (*s++);
2820 { /* Check hash table for duplicate of expanded name (uriName).
2821 Derived from code in lookup(HASH_TABLE *table, ...).
2823 unsigned char step = 0;
2824 unsigned long mask = nsAttsSize - 1;
2825 j = uriHash & mask; /* index into hash table */
2826 while (nsAtts[j].version == version) {
2827 /* for speed we compare stored hash values first */
2828 if (uriHash == nsAtts[j].hash) {
2829 const XML_Char *s1 = poolStart(&tempPool);
2830 const XML_Char *s2 = nsAtts[j].uriName;
2831 /* s1 is null terminated, but not s2 */
2832 for (; *s1 == *s2 && *s1 != 0; s1++, s2++);
2833 if (*s1 == 0)
2834 return XML_ERROR_DUPLICATE_ATTRIBUTE;
2836 if (!step)
2837 step = PROBE_STEP(uriHash, mask, nsAttsPower);
2838 j < step ? (j += nsAttsSize - step) : (j -= step);
2842 if (ns_triplets) { /* append namespace separator and prefix */
2843 tempPool.ptr[-1] = namespaceSeparator;
2844 s = b->prefix->name;
2845 do {
2846 if (!poolAppendChar(&tempPool, *s))
2847 return XML_ERROR_NO_MEMORY;
2848 } while (*s++);
2851 /* store expanded name in attribute list */
2852 s = poolStart(&tempPool);
2853 poolFinish(&tempPool);
2854 appAtts[i] = s;
2856 /* fill empty slot with new version, uriName and hash value */
2857 nsAtts[j].version = version;
2858 nsAtts[j].hash = uriHash;
2859 nsAtts[j].uriName = s;
2861 if (!--nPrefixes) {
2862 i += 2;
2863 break;
2866 else /* not prefixed */
2867 ((XML_Char *)s)[-1] = 0; /* clear flag */
2870 /* clear flags for the remaining attributes */
2871 for (; i < attIndex; i += 2)
2872 ((XML_Char *)(appAtts[i]))[-1] = 0;
2873 for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding)
2874 binding->attId->name[-1] = 0;
2876 if (!ns)
2877 return XML_ERROR_NONE;
2879 /* expand the element type name */
2880 if (elementType->prefix) {
2881 binding = elementType->prefix->binding;
2882 if (!binding)
2883 return XML_ERROR_UNBOUND_PREFIX;
2884 localPart = tagNamePtr->str;
2885 while (*localPart++ != XML_T(':'))
2888 else if (dtd->defaultPrefix.binding) {
2889 binding = dtd->defaultPrefix.binding;
2890 localPart = tagNamePtr->str;
2892 else
2893 return XML_ERROR_NONE;
2894 prefixLen = 0;
2895 if (ns_triplets && binding->prefix->name) {
2896 for (; binding->prefix->name[prefixLen++];)
2897 ; /* prefixLen includes null terminator */
2899 tagNamePtr->localPart = localPart;
2900 tagNamePtr->uriLen = binding->uriLen;
2901 tagNamePtr->prefix = binding->prefix->name;
2902 tagNamePtr->prefixLen = prefixLen;
2903 for (i = 0; localPart[i++];)
2904 ; /* i includes null terminator */
2905 n = i + binding->uriLen + prefixLen;
2906 if (n > binding->uriAlloc) {
2907 TAG *p;
2908 uri = (XML_Char *)MALLOC((n + EXPAND_SPARE) * sizeof(XML_Char));
2909 if (!uri)
2910 return XML_ERROR_NO_MEMORY;
2911 binding->uriAlloc = n + EXPAND_SPARE;
2912 memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char));
2913 for (p = tagStack; p; p = p->parent)
2914 if (p->name.str == binding->uri)
2915 p->name.str = uri;
2916 FREE(binding->uri);
2917 binding->uri = uri;
2919 /* if namespaceSeparator != '\0' then uri includes it already */
2920 uri = binding->uri + binding->uriLen;
2921 memcpy(uri, localPart, i * sizeof(XML_Char));
2922 /* we always have a namespace separator between localPart and prefix */
2923 if (prefixLen) {
2924 uri += i - 1;
2925 *uri = namespaceSeparator; /* replace null terminator */
2926 memcpy(uri + 1, binding->prefix->name, prefixLen * sizeof(XML_Char));
2928 tagNamePtr->str = binding->uri;
2929 return XML_ERROR_NONE;
2932 /* addBinding() overwrites the value of prefix->binding without checking.
2933 Therefore one must keep track of the old value outside of addBinding().
2935 static enum XML_Error
2936 addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId,
2937 const XML_Char *uri, BINDING **bindingsPtr)
2939 static const XML_Char xmlNamespace[] = {
2940 'h', 't', 't', 'p', ':', '/', '/',
2941 'w', 'w', 'w', '.', 'w', '3', '.', 'o', 'r', 'g', '/',
2942 'X', 'M', 'L', '/', '1', '9', '9', '8', '/',
2943 'n', 'a', 'm', 'e', 's', 'p', 'a', 'c', 'e', '\0'
2945 static const int xmlLen =
2946 (int)sizeof(xmlNamespace)/sizeof(XML_Char) - 1;
2947 static const XML_Char xmlnsNamespace[] = {
2948 'h', 't', 't', 'p', ':', '/', '/',
2949 'w', 'w', 'w', '.', 'w', '3', '.', 'o', 'r', 'g', '/',
2950 '2', '0', '0', '0', '/', 'x', 'm', 'l', 'n', 's', '/', '\0'
2952 static const int xmlnsLen =
2953 (int)sizeof(xmlnsNamespace)/sizeof(XML_Char) - 1;
2955 XML_Bool mustBeXML = XML_FALSE;
2956 XML_Bool isXML = XML_TRUE;
2957 XML_Bool isXMLNS = XML_TRUE;
2959 BINDING *b;
2960 int len;
2962 /* empty URI is only valid for default namespace per XML NS 1.0 (not 1.1) */
2963 if (*uri == XML_T('\0') && prefix->name)
2964 return XML_ERROR_UNDECLARING_PREFIX;
2966 if (prefix->name
2967 && prefix->name[0] == XML_T('x')
2968 && prefix->name[1] == XML_T('m')
2969 && prefix->name[2] == XML_T('l')) {
2971 /* Not allowed to bind xmlns */
2972 if (prefix->name[3] == XML_T('n')
2973 && prefix->name[4] == XML_T('s')
2974 && prefix->name[5] == XML_T('\0'))
2975 return XML_ERROR_RESERVED_PREFIX_XMLNS;
2977 if (prefix->name[3] == XML_T('\0'))
2978 mustBeXML = XML_TRUE;
2981 for (len = 0; uri[len]; len++) {
2982 if (isXML && (len > xmlLen || uri[len] != xmlNamespace[len]))
2983 isXML = XML_FALSE;
2985 if (!mustBeXML && isXMLNS
2986 && (len > xmlnsLen || uri[len] != xmlnsNamespace[len]))
2987 isXMLNS = XML_FALSE;
2989 isXML = isXML && len == xmlLen;
2990 isXMLNS = isXMLNS && len == xmlnsLen;
2992 if (mustBeXML != isXML)
2993 return mustBeXML ? XML_ERROR_RESERVED_PREFIX_XML
2994 : XML_ERROR_RESERVED_NAMESPACE_URI;
2996 if (isXMLNS)
2997 return XML_ERROR_RESERVED_NAMESPACE_URI;
2999 if (namespaceSeparator)
3000 len++;
3001 if (freeBindingList) {
3002 b = freeBindingList;
3003 if (len > b->uriAlloc) {
3004 XML_Char *temp = (XML_Char *)REALLOC(b->uri,
3005 sizeof(XML_Char) * (len + EXPAND_SPARE));
3006 if (temp == NULL)
3007 return XML_ERROR_NO_MEMORY;
3008 b->uri = temp;
3009 b->uriAlloc = len + EXPAND_SPARE;
3011 freeBindingList = b->nextTagBinding;
3013 else {
3014 b = (BINDING *)MALLOC(sizeof(BINDING));
3015 if (!b)
3016 return XML_ERROR_NO_MEMORY;
3017 b->uri = (XML_Char *)MALLOC(sizeof(XML_Char) * (len + EXPAND_SPARE));
3018 if (!b->uri) {
3019 FREE(b);
3020 return XML_ERROR_NO_MEMORY;
3022 b->uriAlloc = len + EXPAND_SPARE;
3024 b->uriLen = len;
3025 memcpy(b->uri, uri, len * sizeof(XML_Char));
3026 if (namespaceSeparator)
3027 b->uri[len - 1] = namespaceSeparator;
3028 b->prefix = prefix;
3029 b->attId = attId;
3030 b->prevPrefixBinding = prefix->binding;
3031 /* NULL binding when default namespace undeclared */
3032 if (*uri == XML_T('\0') && prefix == &_dtd->defaultPrefix)
3033 prefix->binding = NULL;
3034 else
3035 prefix->binding = b;
3036 b->nextTagBinding = *bindingsPtr;
3037 *bindingsPtr = b;
3038 /* if attId == NULL then we are not starting a namespace scope */
3039 if (attId && startNamespaceDeclHandler)
3040 startNamespaceDeclHandler(handlerArg, prefix->name,
3041 prefix->binding ? uri : 0);
3042 return XML_ERROR_NONE;
3045 /* The idea here is to avoid using stack for each CDATA section when
3046 the whole file is parsed with one call.
3048 static enum XML_Error PTRCALL
3049 cdataSectionProcessor(XML_Parser parser,
3050 const char *start,
3051 const char *end,
3052 const char **endPtr)
3054 enum XML_Error result = doCdataSection(parser, encoding, &start, end,
3055 endPtr, (XML_Bool)!ps_finalBuffer);
3056 if (result != XML_ERROR_NONE)
3057 return result;
3058 if (start) {
3059 if (parentParser) { /* we are parsing an external entity */
3060 processor = externalEntityContentProcessor;
3061 return externalEntityContentProcessor(parser, start, end, endPtr);
3063 else {
3064 processor = contentProcessor;
3065 return contentProcessor(parser, start, end, endPtr);
3068 return result;
3071 /* startPtr gets set to non-null if the section is closed, and to null if
3072 the section is not yet closed.
3074 static enum XML_Error
3075 doCdataSection(XML_Parser parser,
3076 const ENCODING *enc,
3077 const char **startPtr,
3078 const char *end,
3079 const char **nextPtr,
3080 XML_Bool haveMore)
3082 const char *s = *startPtr;
3083 const char **eventPP;
3084 const char **eventEndPP;
3085 if (enc == encoding) {
3086 eventPP = &eventPtr;
3087 *eventPP = s;
3088 eventEndPP = &eventEndPtr;
3090 else {
3091 eventPP = &(openInternalEntities->internalEventPtr);
3092 eventEndPP = &(openInternalEntities->internalEventEndPtr);
3094 *eventPP = s;
3095 *startPtr = NULL;
3097 for (;;) {
3098 const char *next;
3099 int tok = XmlCdataSectionTok(enc, s, end, &next);
3100 *eventEndPP = next;
3101 switch (tok) {
3102 case XML_TOK_CDATA_SECT_CLOSE:
3103 if (endCdataSectionHandler)
3104 endCdataSectionHandler(handlerArg);
3105 #if 0
3106 /* see comment under XML_TOK_CDATA_SECT_OPEN */
3107 else if (characterDataHandler)
3108 characterDataHandler(handlerArg, dataBuf, 0);
3109 #endif
3110 else if (defaultHandler)
3111 reportDefault(parser, enc, s, next);
3112 *startPtr = next;
3113 *nextPtr = next;
3114 if (ps_parsing == XML_FINISHED)
3115 return XML_ERROR_ABORTED;
3116 else
3117 return XML_ERROR_NONE;
3118 case XML_TOK_DATA_NEWLINE:
3119 if (characterDataHandler) {
3120 XML_Char c = 0xA;
3121 characterDataHandler(handlerArg, &c, 1);
3123 else if (defaultHandler)
3124 reportDefault(parser, enc, s, next);
3125 break;
3126 case XML_TOK_DATA_CHARS:
3127 if (characterDataHandler) {
3128 if (MUST_CONVERT(enc, s)) {
3129 for (;;) {
3130 ICHAR *dataPtr = (ICHAR *)dataBuf;
3131 XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd);
3132 *eventEndPP = next;
3133 characterDataHandler(handlerArg, dataBuf,
3134 (int)(dataPtr - (ICHAR *)dataBuf));
3135 if (s == next)
3136 break;
3137 *eventPP = s;
3140 else
3141 characterDataHandler(handlerArg,
3142 (XML_Char *)s,
3143 (int)((XML_Char *)next - (XML_Char *)s));
3145 else if (defaultHandler)
3146 reportDefault(parser, enc, s, next);
3147 break;
3148 case XML_TOK_INVALID:
3149 *eventPP = next;
3150 return XML_ERROR_INVALID_TOKEN;
3151 case XML_TOK_PARTIAL_CHAR:
3152 if (haveMore) {
3153 *nextPtr = s;
3154 return XML_ERROR_NONE;
3156 return XML_ERROR_PARTIAL_CHAR;
3157 case XML_TOK_PARTIAL:
3158 case XML_TOK_NONE:
3159 if (haveMore) {
3160 *nextPtr = s;
3161 return XML_ERROR_NONE;
3163 return XML_ERROR_UNCLOSED_CDATA_SECTION;
3164 default:
3165 *eventPP = next;
3166 return XML_ERROR_UNEXPECTED_STATE;
3169 *eventPP = s = next;
3170 switch (ps_parsing) {
3171 case XML_SUSPENDED:
3172 *nextPtr = next;
3173 return XML_ERROR_NONE;
3174 case XML_FINISHED:
3175 return XML_ERROR_ABORTED;
3176 default: ;
3179 /* not reached */
3182 #ifdef XML_DTD
3184 /* The idea here is to avoid using stack for each IGNORE section when
3185 the whole file is parsed with one call.
3187 static enum XML_Error PTRCALL
3188 ignoreSectionProcessor(XML_Parser parser,
3189 const char *start,
3190 const char *end,
3191 const char **endPtr)
3193 enum XML_Error result = doIgnoreSection(parser, encoding, &start, end,
3194 endPtr, (XML_Bool)!ps_finalBuffer);
3195 if (result != XML_ERROR_NONE)
3196 return result;
3197 if (start) {
3198 processor = prologProcessor;
3199 return prologProcessor(parser, start, end, endPtr);
3201 return result;
3204 /* startPtr gets set to non-null is the section is closed, and to null
3205 if the section is not yet closed.
3207 static enum XML_Error
3208 doIgnoreSection(XML_Parser parser,
3209 const ENCODING *enc,
3210 const char **startPtr,
3211 const char *end,
3212 const char **nextPtr,
3213 XML_Bool haveMore)
3215 const char *next;
3216 int tok;
3217 const char *s = *startPtr;
3218 const char **eventPP;
3219 const char **eventEndPP;
3220 if (enc == encoding) {
3221 eventPP = &eventPtr;
3222 *eventPP = s;
3223 eventEndPP = &eventEndPtr;
3225 else {
3226 eventPP = &(openInternalEntities->internalEventPtr);
3227 eventEndPP = &(openInternalEntities->internalEventEndPtr);
3229 *eventPP = s;
3230 *startPtr = NULL;
3231 tok = XmlIgnoreSectionTok(enc, s, end, &next);
3232 *eventEndPP = next;
3233 switch (tok) {
3234 case XML_TOK_IGNORE_SECT:
3235 if (defaultHandler)
3236 reportDefault(parser, enc, s, next);
3237 *startPtr = next;
3238 *nextPtr = next;
3239 if (ps_parsing == XML_FINISHED)
3240 return XML_ERROR_ABORTED;
3241 else
3242 return XML_ERROR_NONE;
3243 case XML_TOK_INVALID:
3244 *eventPP = next;
3245 return XML_ERROR_INVALID_TOKEN;
3246 case XML_TOK_PARTIAL_CHAR:
3247 if (haveMore) {
3248 *nextPtr = s;
3249 return XML_ERROR_NONE;
3251 return XML_ERROR_PARTIAL_CHAR;
3252 case XML_TOK_PARTIAL:
3253 case XML_TOK_NONE:
3254 if (haveMore) {
3255 *nextPtr = s;
3256 return XML_ERROR_NONE;
3258 return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */
3259 default:
3260 *eventPP = next;
3261 return XML_ERROR_UNEXPECTED_STATE;
3263 /* not reached */
3266 #endif /* XML_DTD */
3268 static enum XML_Error
3269 initializeEncoding(XML_Parser parser)
3271 const char *s;
3272 #ifdef XML_UNICODE
3273 char encodingBuf[128];
3274 if (!protocolEncodingName)
3275 s = NULL;
3276 else {
3277 int i;
3278 for (i = 0; protocolEncodingName[i]; i++) {
3279 if (i == sizeof(encodingBuf) - 1
3280 || (protocolEncodingName[i] & ~0x7f) != 0) {
3281 encodingBuf[0] = '\0';
3282 break;
3284 encodingBuf[i] = (char)protocolEncodingName[i];
3286 encodingBuf[i] = '\0';
3287 s = encodingBuf;
3289 #else
3290 s = protocolEncodingName;
3291 #endif
3292 if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s))
3293 return XML_ERROR_NONE;
3294 return handleUnknownEncoding(parser, protocolEncodingName);
3297 static enum XML_Error
3298 processXmlDecl(XML_Parser parser, int isGeneralTextEntity,
3299 const char *s, const char *next)
3301 const char *encodingName = NULL;
3302 const XML_Char *storedEncName = NULL;
3303 const ENCODING *newEncoding = NULL;
3304 const char *version = NULL;
3305 const char *versionend;
3306 const XML_Char *storedversion = NULL;
3307 int standalone = -1;
3308 if (!(ns
3309 ? XmlParseXmlDeclNS
3310 : XmlParseXmlDecl)(isGeneralTextEntity,
3311 encoding,
3313 next,
3314 &eventPtr,
3315 &version,
3316 &versionend,
3317 &encodingName,
3318 &newEncoding,
3319 &standalone)) {
3320 if (isGeneralTextEntity)
3321 return XML_ERROR_TEXT_DECL;
3322 else
3323 return XML_ERROR_XML_DECL;
3325 if (!isGeneralTextEntity && standalone == 1) {
3326 _dtd->standalone = XML_TRUE;
3327 #ifdef XML_DTD
3328 if (paramEntityParsing == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE)
3329 paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER;
3330 #endif /* XML_DTD */
3332 if (xmlDeclHandler) {
3333 if (encodingName != NULL) {
3334 storedEncName = poolStoreString(&temp2Pool,
3335 encoding,
3336 encodingName,
3337 encodingName
3338 + XmlNameLength(encoding, encodingName));
3339 if (!storedEncName)
3340 return XML_ERROR_NO_MEMORY;
3341 poolFinish(&temp2Pool);
3343 if (version) {
3344 storedversion = poolStoreString(&temp2Pool,
3345 encoding,
3346 version,
3347 versionend - encoding->minBytesPerChar);
3348 if (!storedversion)
3349 return XML_ERROR_NO_MEMORY;
3351 xmlDeclHandler(handlerArg, storedversion, storedEncName, standalone);
3353 else if (defaultHandler)
3354 reportDefault(parser, encoding, s, next);
3355 if (protocolEncodingName == NULL) {
3356 if (newEncoding) {
3357 if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) {
3358 eventPtr = encodingName;
3359 return XML_ERROR_INCORRECT_ENCODING;
3361 encoding = newEncoding;
3363 else if (encodingName) {
3364 enum XML_Error result;
3365 if (!storedEncName) {
3366 storedEncName = poolStoreString(
3367 &temp2Pool, encoding, encodingName,
3368 encodingName + XmlNameLength(encoding, encodingName));
3369 if (!storedEncName)
3370 return XML_ERROR_NO_MEMORY;
3372 result = handleUnknownEncoding(parser, storedEncName);
3373 poolClear(&temp2Pool);
3374 if (result == XML_ERROR_UNKNOWN_ENCODING)
3375 eventPtr = encodingName;
3376 return result;
3380 if (storedEncName || storedversion)
3381 poolClear(&temp2Pool);
3383 return XML_ERROR_NONE;
3386 static enum XML_Error
3387 handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName)
3389 if (unknownEncodingHandler) {
3390 XML_Encoding info;
3391 int i;
3392 for (i = 0; i < 256; i++)
3393 info.map[i] = -1;
3394 info.convert = NULL;
3395 info.data = NULL;
3396 info.release = NULL;
3397 if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName,
3398 &info)) {
3399 ENCODING *enc;
3400 unknownEncodingMem = MALLOC(XmlSizeOfUnknownEncoding());
3401 if (!unknownEncodingMem) {
3402 if (info.release)
3403 info.release(info.data);
3404 return XML_ERROR_NO_MEMORY;
3406 enc = (ns
3407 ? XmlInitUnknownEncodingNS
3408 : XmlInitUnknownEncoding)(unknownEncodingMem,
3409 info.map,
3410 info.convert,
3411 info.data);
3412 if (enc) {
3413 unknownEncodingData = info.data;
3414 unknownEncodingRelease = info.release;
3415 encoding = enc;
3416 return XML_ERROR_NONE;
3419 if (info.release != NULL)
3420 info.release(info.data);
3422 return XML_ERROR_UNKNOWN_ENCODING;
3425 static enum XML_Error PTRCALL
3426 prologInitProcessor(XML_Parser parser,
3427 const char *s,
3428 const char *end,
3429 const char **nextPtr)
3431 enum XML_Error result = initializeEncoding(parser);
3432 if (result != XML_ERROR_NONE)
3433 return result;
3434 processor = prologProcessor;
3435 return prologProcessor(parser, s, end, nextPtr);
3438 #ifdef XML_DTD
3440 static enum XML_Error PTRCALL
3441 externalParEntInitProcessor(XML_Parser parser,
3442 const char *s,
3443 const char *end,
3444 const char **nextPtr)
3446 enum XML_Error result = initializeEncoding(parser);
3447 if (result != XML_ERROR_NONE)
3448 return result;
3450 /* we know now that XML_Parse(Buffer) has been called,
3451 so we consider the external parameter entity read */
3452 _dtd->paramEntityRead = XML_TRUE;
3454 if (prologState.inEntityValue) {
3455 processor = entityValueInitProcessor;
3456 return entityValueInitProcessor(parser, s, end, nextPtr);
3458 else {
3459 processor = externalParEntProcessor;
3460 return externalParEntProcessor(parser, s, end, nextPtr);
3464 static enum XML_Error PTRCALL
3465 entityValueInitProcessor(XML_Parser parser,
3466 const char *s,
3467 const char *end,
3468 const char **nextPtr)
3470 int tok;
3471 const char *start = s;
3472 const char *next = start;
3473 eventPtr = start;
3475 for (;;) {
3476 tok = XmlPrologTok(encoding, start, end, &next);
3477 eventEndPtr = next;
3478 if (tok <= 0) {
3479 if (!ps_finalBuffer && tok != XML_TOK_INVALID) {
3480 *nextPtr = s;
3481 return XML_ERROR_NONE;
3483 switch (tok) {
3484 case XML_TOK_INVALID:
3485 return XML_ERROR_INVALID_TOKEN;
3486 case XML_TOK_PARTIAL:
3487 return XML_ERROR_UNCLOSED_TOKEN;
3488 case XML_TOK_PARTIAL_CHAR:
3489 return XML_ERROR_PARTIAL_CHAR;
3490 case XML_TOK_NONE: /* start == end */
3491 default:
3492 break;
3494 /* found end of entity value - can store it now */
3495 return storeEntityValue(parser, encoding, s, end);
3497 else if (tok == XML_TOK_XML_DECL) {
3498 enum XML_Error result;
3499 result = processXmlDecl(parser, 0, start, next);
3500 if (result != XML_ERROR_NONE)
3501 return result;
3502 switch (ps_parsing) {
3503 case XML_SUSPENDED:
3504 *nextPtr = next;
3505 return XML_ERROR_NONE;
3506 case XML_FINISHED:
3507 return XML_ERROR_ABORTED;
3508 default:
3509 *nextPtr = next;
3511 /* stop scanning for text declaration - we found one */
3512 processor = entityValueProcessor;
3513 return entityValueProcessor(parser, next, end, nextPtr);
3515 /* If we are at the end of the buffer, this would cause XmlPrologTok to
3516 return XML_TOK_NONE on the next call, which would then cause the
3517 function to exit with *nextPtr set to s - that is what we want for other
3518 tokens, but not for the BOM - we would rather like to skip it;
3519 then, when this routine is entered the next time, XmlPrologTok will
3520 return XML_TOK_INVALID, since the BOM is still in the buffer
3522 else if (tok == XML_TOK_BOM && next == end && !ps_finalBuffer) {
3523 *nextPtr = next;
3524 return XML_ERROR_NONE;
3526 start = next;
3527 eventPtr = start;
3531 static enum XML_Error PTRCALL
3532 externalParEntProcessor(XML_Parser parser,
3533 const char *s,
3534 const char *end,
3535 const char **nextPtr)
3537 const char *next = s;
3538 int tok;
3540 tok = XmlPrologTok(encoding, s, end, &next);
3541 if (tok <= 0) {
3542 if (!ps_finalBuffer && tok != XML_TOK_INVALID) {
3543 *nextPtr = s;
3544 return XML_ERROR_NONE;
3546 switch (tok) {
3547 case XML_TOK_INVALID:
3548 return XML_ERROR_INVALID_TOKEN;
3549 case XML_TOK_PARTIAL:
3550 return XML_ERROR_UNCLOSED_TOKEN;
3551 case XML_TOK_PARTIAL_CHAR:
3552 return XML_ERROR_PARTIAL_CHAR;
3553 case XML_TOK_NONE: /* start == end */
3554 default:
3555 break;
3558 /* This would cause the next stage, i.e. doProlog to be passed XML_TOK_BOM.
3559 However, when parsing an external subset, doProlog will not accept a BOM
3560 as valid, and report a syntax error, so we have to skip the BOM
3562 else if (tok == XML_TOK_BOM) {
3563 s = next;
3564 tok = XmlPrologTok(encoding, s, end, &next);
3567 processor = prologProcessor;
3568 return doProlog(parser, encoding, s, end, tok, next,
3569 nextPtr, (XML_Bool)!ps_finalBuffer);
3572 static enum XML_Error PTRCALL
3573 entityValueProcessor(XML_Parser parser,
3574 const char *s,
3575 const char *end,
3576 const char **nextPtr)
3578 const char *start = s;
3579 const char *next = s;
3580 const ENCODING *enc = encoding;
3581 int tok;
3583 for (;;) {
3584 tok = XmlPrologTok(enc, start, end, &next);
3585 if (tok <= 0) {
3586 if (!ps_finalBuffer && tok != XML_TOK_INVALID) {
3587 *nextPtr = s;
3588 return XML_ERROR_NONE;
3590 switch (tok) {
3591 case XML_TOK_INVALID:
3592 return XML_ERROR_INVALID_TOKEN;
3593 case XML_TOK_PARTIAL:
3594 return XML_ERROR_UNCLOSED_TOKEN;
3595 case XML_TOK_PARTIAL_CHAR:
3596 return XML_ERROR_PARTIAL_CHAR;
3597 case XML_TOK_NONE: /* start == end */
3598 default:
3599 break;
3601 /* found end of entity value - can store it now */
3602 return storeEntityValue(parser, enc, s, end);
3604 start = next;
3608 #endif /* XML_DTD */
3610 static enum XML_Error PTRCALL
3611 prologProcessor(XML_Parser parser,
3612 const char *s,
3613 const char *end,
3614 const char **nextPtr)
3616 const char *next = s;
3617 int tok = XmlPrologTok(encoding, s, end, &next);
3618 return doProlog(parser, encoding, s, end, tok, next,
3619 nextPtr, (XML_Bool)!ps_finalBuffer);
3622 static enum XML_Error
3623 doProlog(XML_Parser parser,
3624 const ENCODING *enc,
3625 const char *s,
3626 const char *end,
3627 int tok,
3628 const char *next,
3629 const char **nextPtr,
3630 XML_Bool haveMore)
3632 #ifdef XML_DTD
3633 static const XML_Char externalSubsetName[] = { '#' , '\0' };
3634 #endif /* XML_DTD */
3635 static const XML_Char atypeCDATA[] = { 'C', 'D', 'A', 'T', 'A', '\0' };
3636 static const XML_Char atypeID[] = { 'I', 'D', '\0' };
3637 static const XML_Char atypeIDREF[] = { 'I', 'D', 'R', 'E', 'F', '\0' };
3638 static const XML_Char atypeIDREFS[] = { 'I', 'D', 'R', 'E', 'F', 'S', '\0' };
3639 static const XML_Char atypeENTITY[] = { 'E', 'N', 'T', 'I', 'T', 'Y', '\0' };
3640 static const XML_Char atypeENTITIES[] =
3641 { 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S', '\0' };
3642 static const XML_Char atypeNMTOKEN[] = {
3643 'N', 'M', 'T', 'O', 'K', 'E', 'N', '\0' };
3644 static const XML_Char atypeNMTOKENS[] = {
3645 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S', '\0' };
3646 static const XML_Char notationPrefix[] = {
3647 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N', '(', '\0' };
3648 static const XML_Char enumValueSep[] = { '|', '\0' };
3649 static const XML_Char enumValueStart[] = { '(', '\0' };
3651 /* save one level of indirection */
3652 DTD * const dtd = _dtd;
3654 const char **eventPP;
3655 const char **eventEndPP;
3656 enum XML_Content_Quant quant;
3658 if (enc == encoding) {
3659 eventPP = &eventPtr;
3660 eventEndPP = &eventEndPtr;
3662 else {
3663 eventPP = &(openInternalEntities->internalEventPtr);
3664 eventEndPP = &(openInternalEntities->internalEventEndPtr);
3667 for (;;) {
3668 int role;
3669 XML_Bool handleDefault = XML_TRUE;
3670 *eventPP = s;
3671 *eventEndPP = next;
3672 if (tok <= 0) {
3673 if (haveMore && tok != XML_TOK_INVALID) {
3674 *nextPtr = s;
3675 return XML_ERROR_NONE;
3677 switch (tok) {
3678 case XML_TOK_INVALID:
3679 *eventPP = next;
3680 return XML_ERROR_INVALID_TOKEN;
3681 case XML_TOK_PARTIAL:
3682 return XML_ERROR_UNCLOSED_TOKEN;
3683 case XML_TOK_PARTIAL_CHAR:
3684 return XML_ERROR_PARTIAL_CHAR;
3685 case XML_TOK_NONE:
3686 #ifdef XML_DTD
3687 /* for internal PE NOT referenced between declarations */
3688 if (enc != encoding && !openInternalEntities->betweenDecl) {
3689 *nextPtr = s;
3690 return XML_ERROR_NONE;
3692 /* WFC: PE Between Declarations - must check that PE contains
3693 complete markup, not only for external PEs, but also for
3694 internal PEs if the reference occurs between declarations.
3696 if (isParamEntity || enc != encoding) {
3697 if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc)
3698 == XML_ROLE_ERROR)
3699 return XML_ERROR_INCOMPLETE_PE;
3700 *nextPtr = s;
3701 return XML_ERROR_NONE;
3703 #endif /* XML_DTD */
3704 return XML_ERROR_NO_ELEMENTS;
3705 default:
3706 tok = -tok;
3707 next = end;
3708 break;
3711 role = XmlTokenRole(&prologState, tok, s, next, enc);
3712 switch (role) {
3713 case XML_ROLE_XML_DECL:
3715 enum XML_Error result = processXmlDecl(parser, 0, s, next);
3716 if (result != XML_ERROR_NONE)
3717 return result;
3718 enc = encoding;
3719 handleDefault = XML_FALSE;
3721 break;
3722 case XML_ROLE_DOCTYPE_NAME:
3723 if (startDoctypeDeclHandler) {
3724 doctypeName = poolStoreString(&tempPool, enc, s, next);
3725 if (!doctypeName)
3726 return XML_ERROR_NO_MEMORY;
3727 poolFinish(&tempPool);
3728 doctypePubid = NULL;
3729 handleDefault = XML_FALSE;
3731 doctypeSysid = NULL; /* always initialize to NULL */
3732 break;
3733 case XML_ROLE_DOCTYPE_INTERNAL_SUBSET:
3734 if (startDoctypeDeclHandler) {
3735 startDoctypeDeclHandler(handlerArg, doctypeName, doctypeSysid,
3736 doctypePubid, 1);
3737 doctypeName = NULL;
3738 poolClear(&tempPool);
3739 handleDefault = XML_FALSE;
3741 break;
3742 #ifdef XML_DTD
3743 case XML_ROLE_TEXT_DECL:
3745 enum XML_Error result = processXmlDecl(parser, 1, s, next);
3746 if (result != XML_ERROR_NONE)
3747 return result;
3748 enc = encoding;
3749 handleDefault = XML_FALSE;
3751 break;
3752 #endif /* XML_DTD */
3753 case XML_ROLE_DOCTYPE_PUBLIC_ID:
3754 #ifdef XML_DTD
3755 useForeignDTD = XML_FALSE;
3756 declEntity = (ENTITY *)lookup(&dtd->paramEntities,
3757 externalSubsetName,
3758 sizeof(ENTITY));
3759 if (!declEntity)
3760 return XML_ERROR_NO_MEMORY;
3761 #endif /* XML_DTD */
3762 dtd->hasParamEntityRefs = XML_TRUE;
3763 if (startDoctypeDeclHandler) {
3764 if (!XmlIsPublicId(enc, s, next, eventPP))
3765 return XML_ERROR_PUBLICID;
3766 doctypePubid = poolStoreString(&tempPool, enc,
3767 s + enc->minBytesPerChar,
3768 next - enc->minBytesPerChar);
3769 if (!doctypePubid)
3770 return XML_ERROR_NO_MEMORY;
3771 normalizePublicId((XML_Char *)doctypePubid);
3772 poolFinish(&tempPool);
3773 handleDefault = XML_FALSE;
3774 goto alreadyChecked;
3776 /* fall through */
3777 case XML_ROLE_ENTITY_PUBLIC_ID:
3778 if (!XmlIsPublicId(enc, s, next, eventPP))
3779 return XML_ERROR_PUBLICID;
3780 alreadyChecked:
3781 if (dtd->keepProcessing && declEntity) {
3782 XML_Char *tem = poolStoreString(&dtd->pool,
3783 enc,
3784 s + enc->minBytesPerChar,
3785 next - enc->minBytesPerChar);
3786 if (!tem)
3787 return XML_ERROR_NO_MEMORY;
3788 normalizePublicId(tem);
3789 declEntity->publicId = tem;
3790 poolFinish(&dtd->pool);
3791 if (entityDeclHandler)
3792 handleDefault = XML_FALSE;
3794 break;
3795 case XML_ROLE_DOCTYPE_CLOSE:
3796 if (doctypeName) {
3797 startDoctypeDeclHandler(handlerArg, doctypeName,
3798 doctypeSysid, doctypePubid, 0);
3799 poolClear(&tempPool);
3800 handleDefault = XML_FALSE;
3802 /* doctypeSysid will be non-NULL in the case of a previous
3803 XML_ROLE_DOCTYPE_SYSTEM_ID, even if startDoctypeDeclHandler
3804 was not set, indicating an external subset
3806 #ifdef XML_DTD
3807 if (doctypeSysid || useForeignDTD) {
3808 XML_Bool hadParamEntityRefs = dtd->hasParamEntityRefs;
3809 dtd->hasParamEntityRefs = XML_TRUE;
3810 if (paramEntityParsing && externalEntityRefHandler) {
3811 ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities,
3812 externalSubsetName,
3813 sizeof(ENTITY));
3814 if (!entity)
3815 return XML_ERROR_NO_MEMORY;
3816 if (useForeignDTD)
3817 entity->base = curBase;
3818 dtd->paramEntityRead = XML_FALSE;
3819 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
3821 entity->base,
3822 entity->systemId,
3823 entity->publicId))
3824 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
3825 if (dtd->paramEntityRead) {
3826 if (!dtd->standalone &&
3827 notStandaloneHandler &&
3828 !notStandaloneHandler(handlerArg))
3829 return XML_ERROR_NOT_STANDALONE;
3831 /* if we didn't read the foreign DTD then this means that there
3832 is no external subset and we must reset dtd->hasParamEntityRefs
3834 else if (!doctypeSysid)
3835 dtd->hasParamEntityRefs = hadParamEntityRefs;
3836 /* end of DTD - no need to update dtd->keepProcessing */
3838 useForeignDTD = XML_FALSE;
3840 #endif /* XML_DTD */
3841 if (endDoctypeDeclHandler) {
3842 endDoctypeDeclHandler(handlerArg);
3843 handleDefault = XML_FALSE;
3845 break;
3846 case XML_ROLE_INSTANCE_START:
3847 #ifdef XML_DTD
3848 /* if there is no DOCTYPE declaration then now is the
3849 last chance to read the foreign DTD
3851 if (useForeignDTD) {
3852 XML_Bool hadParamEntityRefs = dtd->hasParamEntityRefs;
3853 dtd->hasParamEntityRefs = XML_TRUE;
3854 if (paramEntityParsing && externalEntityRefHandler) {
3855 ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities,
3856 externalSubsetName,
3857 sizeof(ENTITY));
3858 if (!entity)
3859 return XML_ERROR_NO_MEMORY;
3860 entity->base = curBase;
3861 dtd->paramEntityRead = XML_FALSE;
3862 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
3864 entity->base,
3865 entity->systemId,
3866 entity->publicId))
3867 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
3868 if (dtd->paramEntityRead) {
3869 if (!dtd->standalone &&
3870 notStandaloneHandler &&
3871 !notStandaloneHandler(handlerArg))
3872 return XML_ERROR_NOT_STANDALONE;
3874 /* if we didn't read the foreign DTD then this means that there
3875 is no external subset and we must reset dtd->hasParamEntityRefs
3877 else
3878 dtd->hasParamEntityRefs = hadParamEntityRefs;
3879 /* end of DTD - no need to update dtd->keepProcessing */
3882 #endif /* XML_DTD */
3883 processor = contentProcessor;
3884 return contentProcessor(parser, s, end, nextPtr);
3885 case XML_ROLE_ATTLIST_ELEMENT_NAME:
3886 declElementType = getElementType(parser, enc, s, next);
3887 if (!declElementType)
3888 return XML_ERROR_NO_MEMORY;
3889 goto checkAttListDeclHandler;
3890 case XML_ROLE_ATTRIBUTE_NAME:
3891 declAttributeId = getAttributeId(parser, enc, s, next);
3892 if (!declAttributeId)
3893 return XML_ERROR_NO_MEMORY;
3894 declAttributeIsCdata = XML_FALSE;
3895 declAttributeType = NULL;
3896 declAttributeIsId = XML_FALSE;
3897 goto checkAttListDeclHandler;
3898 case XML_ROLE_ATTRIBUTE_TYPE_CDATA:
3899 declAttributeIsCdata = XML_TRUE;
3900 declAttributeType = atypeCDATA;
3901 goto checkAttListDeclHandler;
3902 case XML_ROLE_ATTRIBUTE_TYPE_ID:
3903 declAttributeIsId = XML_TRUE;
3904 declAttributeType = atypeID;
3905 goto checkAttListDeclHandler;
3906 case XML_ROLE_ATTRIBUTE_TYPE_IDREF:
3907 declAttributeType = atypeIDREF;
3908 goto checkAttListDeclHandler;
3909 case XML_ROLE_ATTRIBUTE_TYPE_IDREFS:
3910 declAttributeType = atypeIDREFS;
3911 goto checkAttListDeclHandler;
3912 case XML_ROLE_ATTRIBUTE_TYPE_ENTITY:
3913 declAttributeType = atypeENTITY;
3914 goto checkAttListDeclHandler;
3915 case XML_ROLE_ATTRIBUTE_TYPE_ENTITIES:
3916 declAttributeType = atypeENTITIES;
3917 goto checkAttListDeclHandler;
3918 case XML_ROLE_ATTRIBUTE_TYPE_NMTOKEN:
3919 declAttributeType = atypeNMTOKEN;
3920 goto checkAttListDeclHandler;
3921 case XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS:
3922 declAttributeType = atypeNMTOKENS;
3923 checkAttListDeclHandler:
3924 if (dtd->keepProcessing && attlistDeclHandler)
3925 handleDefault = XML_FALSE;
3926 break;
3927 case XML_ROLE_ATTRIBUTE_ENUM_VALUE:
3928 case XML_ROLE_ATTRIBUTE_NOTATION_VALUE:
3929 if (dtd->keepProcessing && attlistDeclHandler) {
3930 const XML_Char *prefix;
3931 if (declAttributeType) {
3932 prefix = enumValueSep;
3934 else {
3935 prefix = (role == XML_ROLE_ATTRIBUTE_NOTATION_VALUE
3936 ? notationPrefix
3937 : enumValueStart);
3939 if (!poolAppendString(&tempPool, prefix))
3940 return XML_ERROR_NO_MEMORY;
3941 if (!poolAppend(&tempPool, enc, s, next))
3942 return XML_ERROR_NO_MEMORY;
3943 declAttributeType = tempPool.start;
3944 handleDefault = XML_FALSE;
3946 break;
3947 case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE:
3948 case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE:
3949 if (dtd->keepProcessing) {
3950 if (!defineAttribute(declElementType, declAttributeId,
3951 declAttributeIsCdata, declAttributeIsId,
3952 0, parser))
3953 return XML_ERROR_NO_MEMORY;
3954 if (attlistDeclHandler && declAttributeType) {
3955 if (*declAttributeType == XML_T('(')
3956 || (*declAttributeType == XML_T('N')
3957 && declAttributeType[1] == XML_T('O'))) {
3958 /* Enumerated or Notation type */
3959 if (!poolAppendChar(&tempPool, XML_T(')'))
3960 || !poolAppendChar(&tempPool, XML_T('\0')))
3961 return XML_ERROR_NO_MEMORY;
3962 declAttributeType = tempPool.start;
3963 poolFinish(&tempPool);
3965 *eventEndPP = s;
3966 attlistDeclHandler(handlerArg, declElementType->name,
3967 declAttributeId->name, declAttributeType,
3968 0, role == XML_ROLE_REQUIRED_ATTRIBUTE_VALUE);
3969 poolClear(&tempPool);
3970 handleDefault = XML_FALSE;
3973 break;
3974 case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE:
3975 case XML_ROLE_FIXED_ATTRIBUTE_VALUE:
3976 if (dtd->keepProcessing) {
3977 const XML_Char *attVal;
3978 enum XML_Error result =
3979 storeAttributeValue(parser, enc, declAttributeIsCdata,
3980 s + enc->minBytesPerChar,
3981 next - enc->minBytesPerChar,
3982 &dtd->pool);
3983 if (result)
3984 return result;
3985 attVal = poolStart(&dtd->pool);
3986 poolFinish(&dtd->pool);
3987 /* ID attributes aren't allowed to have a default */
3988 if (!defineAttribute(declElementType, declAttributeId,
3989 declAttributeIsCdata, XML_FALSE, attVal, parser))
3990 return XML_ERROR_NO_MEMORY;
3991 if (attlistDeclHandler && declAttributeType) {
3992 if (*declAttributeType == XML_T('(')
3993 || (*declAttributeType == XML_T('N')
3994 && declAttributeType[1] == XML_T('O'))) {
3995 /* Enumerated or Notation type */
3996 if (!poolAppendChar(&tempPool, XML_T(')'))
3997 || !poolAppendChar(&tempPool, XML_T('\0')))
3998 return XML_ERROR_NO_MEMORY;
3999 declAttributeType = tempPool.start;
4000 poolFinish(&tempPool);
4002 *eventEndPP = s;
4003 attlistDeclHandler(handlerArg, declElementType->name,
4004 declAttributeId->name, declAttributeType,
4005 attVal,
4006 role == XML_ROLE_FIXED_ATTRIBUTE_VALUE);
4007 poolClear(&tempPool);
4008 handleDefault = XML_FALSE;
4011 break;
4012 case XML_ROLE_ENTITY_VALUE:
4013 if (dtd->keepProcessing) {
4014 enum XML_Error result = storeEntityValue(parser, enc,
4015 s + enc->minBytesPerChar,
4016 next - enc->minBytesPerChar);
4017 if (declEntity) {
4018 declEntity->textPtr = poolStart(&dtd->entityValuePool);
4019 declEntity->textLen = (int)(poolLength(&dtd->entityValuePool));
4020 poolFinish(&dtd->entityValuePool);
4021 if (entityDeclHandler) {
4022 *eventEndPP = s;
4023 entityDeclHandler(handlerArg,
4024 declEntity->name,
4025 declEntity->is_param,
4026 declEntity->textPtr,
4027 declEntity->textLen,
4028 curBase, 0, 0, 0);
4029 handleDefault = XML_FALSE;
4032 else
4033 poolDiscard(&dtd->entityValuePool);
4034 if (result != XML_ERROR_NONE)
4035 return result;
4037 break;
4038 case XML_ROLE_DOCTYPE_SYSTEM_ID:
4039 #ifdef XML_DTD
4040 useForeignDTD = XML_FALSE;
4041 #endif /* XML_DTD */
4042 dtd->hasParamEntityRefs = XML_TRUE;
4043 if (startDoctypeDeclHandler) {
4044 doctypeSysid = poolStoreString(&tempPool, enc,
4045 s + enc->minBytesPerChar,
4046 next - enc->minBytesPerChar);
4047 if (doctypeSysid == NULL)
4048 return XML_ERROR_NO_MEMORY;
4049 poolFinish(&tempPool);
4050 handleDefault = XML_FALSE;
4052 #ifdef XML_DTD
4053 else
4054 /* use externalSubsetName to make doctypeSysid non-NULL
4055 for the case where no startDoctypeDeclHandler is set */
4056 doctypeSysid = externalSubsetName;
4057 #endif /* XML_DTD */
4058 if (!dtd->standalone
4059 #ifdef XML_DTD
4060 && !paramEntityParsing
4061 #endif /* XML_DTD */
4062 && notStandaloneHandler
4063 && !notStandaloneHandler(handlerArg))
4064 return XML_ERROR_NOT_STANDALONE;
4065 #ifndef XML_DTD
4066 break;
4067 #else /* XML_DTD */
4068 if (!declEntity) {
4069 declEntity = (ENTITY *)lookup(&dtd->paramEntities,
4070 externalSubsetName,
4071 sizeof(ENTITY));
4072 if (!declEntity)
4073 return XML_ERROR_NO_MEMORY;
4074 declEntity->publicId = NULL;
4076 /* fall through */
4077 #endif /* XML_DTD */
4078 case XML_ROLE_ENTITY_SYSTEM_ID:
4079 if (dtd->keepProcessing && declEntity) {
4080 declEntity->systemId = poolStoreString(&dtd->pool, enc,
4081 s + enc->minBytesPerChar,
4082 next - enc->minBytesPerChar);
4083 if (!declEntity->systemId)
4084 return XML_ERROR_NO_MEMORY;
4085 declEntity->base = curBase;
4086 poolFinish(&dtd->pool);
4087 if (entityDeclHandler)
4088 handleDefault = XML_FALSE;
4090 break;
4091 case XML_ROLE_ENTITY_COMPLETE:
4092 if (dtd->keepProcessing && declEntity && entityDeclHandler) {
4093 *eventEndPP = s;
4094 entityDeclHandler(handlerArg,
4095 declEntity->name,
4096 declEntity->is_param,
4097 0,0,
4098 declEntity->base,
4099 declEntity->systemId,
4100 declEntity->publicId,
4102 handleDefault = XML_FALSE;
4104 break;
4105 case XML_ROLE_ENTITY_NOTATION_NAME:
4106 if (dtd->keepProcessing && declEntity) {
4107 declEntity->notation = poolStoreString(&dtd->pool, enc, s, next);
4108 if (!declEntity->notation)
4109 return XML_ERROR_NO_MEMORY;
4110 poolFinish(&dtd->pool);
4111 if (unparsedEntityDeclHandler) {
4112 *eventEndPP = s;
4113 unparsedEntityDeclHandler(handlerArg,
4114 declEntity->name,
4115 declEntity->base,
4116 declEntity->systemId,
4117 declEntity->publicId,
4118 declEntity->notation);
4119 handleDefault = XML_FALSE;
4121 else if (entityDeclHandler) {
4122 *eventEndPP = s;
4123 entityDeclHandler(handlerArg,
4124 declEntity->name,
4125 0,0,0,
4126 declEntity->base,
4127 declEntity->systemId,
4128 declEntity->publicId,
4129 declEntity->notation);
4130 handleDefault = XML_FALSE;
4133 break;
4134 case XML_ROLE_GENERAL_ENTITY_NAME:
4136 if (XmlPredefinedEntityName(enc, s, next)) {
4137 declEntity = NULL;
4138 break;
4140 if (dtd->keepProcessing) {
4141 const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next);
4142 if (!name)
4143 return XML_ERROR_NO_MEMORY;
4144 declEntity = (ENTITY *)lookup(&dtd->generalEntities, name,
4145 sizeof(ENTITY));
4146 if (!declEntity)
4147 return XML_ERROR_NO_MEMORY;
4148 if (declEntity->name != name) {
4149 poolDiscard(&dtd->pool);
4150 declEntity = NULL;
4152 else {
4153 poolFinish(&dtd->pool);
4154 declEntity->publicId = NULL;
4155 declEntity->is_param = XML_FALSE;
4156 /* if we have a parent parser or are reading an internal parameter
4157 entity, then the entity declaration is not considered "internal"
4159 declEntity->is_internal = !(parentParser || openInternalEntities);
4160 if (entityDeclHandler)
4161 handleDefault = XML_FALSE;
4164 else {
4165 poolDiscard(&dtd->pool);
4166 declEntity = NULL;
4169 break;
4170 case XML_ROLE_PARAM_ENTITY_NAME:
4171 #ifdef XML_DTD
4172 if (dtd->keepProcessing) {
4173 const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next);
4174 if (!name)
4175 return XML_ERROR_NO_MEMORY;
4176 declEntity = (ENTITY *)lookup(&dtd->paramEntities,
4177 name, sizeof(ENTITY));
4178 if (!declEntity)
4179 return XML_ERROR_NO_MEMORY;
4180 if (declEntity->name != name) {
4181 poolDiscard(&dtd->pool);
4182 declEntity = NULL;
4184 else {
4185 poolFinish(&dtd->pool);
4186 declEntity->publicId = NULL;
4187 declEntity->is_param = XML_TRUE;
4188 /* if we have a parent parser or are reading an internal parameter
4189 entity, then the entity declaration is not considered "internal"
4191 declEntity->is_internal = !(parentParser || openInternalEntities);
4192 if (entityDeclHandler)
4193 handleDefault = XML_FALSE;
4196 else {
4197 poolDiscard(&dtd->pool);
4198 declEntity = NULL;
4200 #else /* not XML_DTD */
4201 declEntity = NULL;
4202 #endif /* XML_DTD */
4203 break;
4204 case XML_ROLE_NOTATION_NAME:
4205 declNotationPublicId = NULL;
4206 declNotationName = NULL;
4207 if (notationDeclHandler) {
4208 declNotationName = poolStoreString(&tempPool, enc, s, next);
4209 if (!declNotationName)
4210 return XML_ERROR_NO_MEMORY;
4211 poolFinish(&tempPool);
4212 handleDefault = XML_FALSE;
4214 break;
4215 case XML_ROLE_NOTATION_PUBLIC_ID:
4216 if (!XmlIsPublicId(enc, s, next, eventPP))
4217 return XML_ERROR_PUBLICID;
4218 if (declNotationName) { /* means notationDeclHandler != NULL */
4219 XML_Char *tem = poolStoreString(&tempPool,
4220 enc,
4221 s + enc->minBytesPerChar,
4222 next - enc->minBytesPerChar);
4223 if (!tem)
4224 return XML_ERROR_NO_MEMORY;
4225 normalizePublicId(tem);
4226 declNotationPublicId = tem;
4227 poolFinish(&tempPool);
4228 handleDefault = XML_FALSE;
4230 break;
4231 case XML_ROLE_NOTATION_SYSTEM_ID:
4232 if (declNotationName && notationDeclHandler) {
4233 const XML_Char *systemId
4234 = poolStoreString(&tempPool, enc,
4235 s + enc->minBytesPerChar,
4236 next - enc->minBytesPerChar);
4237 if (!systemId)
4238 return XML_ERROR_NO_MEMORY;
4239 *eventEndPP = s;
4240 notationDeclHandler(handlerArg,
4241 declNotationName,
4242 curBase,
4243 systemId,
4244 declNotationPublicId);
4245 handleDefault = XML_FALSE;
4247 poolClear(&tempPool);
4248 break;
4249 case XML_ROLE_NOTATION_NO_SYSTEM_ID:
4250 if (declNotationPublicId && notationDeclHandler) {
4251 *eventEndPP = s;
4252 notationDeclHandler(handlerArg,
4253 declNotationName,
4254 curBase,
4256 declNotationPublicId);
4257 handleDefault = XML_FALSE;
4259 poolClear(&tempPool);
4260 break;
4261 case XML_ROLE_ERROR:
4262 switch (tok) {
4263 case XML_TOK_PARAM_ENTITY_REF:
4264 /* PE references in internal subset are
4265 not allowed within declarations. */
4266 return XML_ERROR_PARAM_ENTITY_REF;
4267 case XML_TOK_XML_DECL:
4268 return XML_ERROR_MISPLACED_XML_PI;
4269 default:
4270 return XML_ERROR_SYNTAX;
4272 #ifdef XML_DTD
4273 case XML_ROLE_IGNORE_SECT:
4275 enum XML_Error result;
4276 if (defaultHandler)
4277 reportDefault(parser, enc, s, next);
4278 handleDefault = XML_FALSE;
4279 result = doIgnoreSection(parser, enc, &next, end, nextPtr, haveMore);
4280 if (result != XML_ERROR_NONE)
4281 return result;
4282 else if (!next) {
4283 processor = ignoreSectionProcessor;
4284 return result;
4287 break;
4288 #endif /* XML_DTD */
4289 case XML_ROLE_GROUP_OPEN:
4290 if (prologState.level >= groupSize) {
4291 if (groupSize) {
4292 char *temp = (char *)REALLOC(groupConnector, groupSize *= 2);
4293 if (temp == NULL)
4294 return XML_ERROR_NO_MEMORY;
4295 groupConnector = temp;
4296 if (dtd->scaffIndex) {
4297 int *temp = (int *)REALLOC(dtd->scaffIndex,
4298 groupSize * sizeof(int));
4299 if (temp == NULL)
4300 return XML_ERROR_NO_MEMORY;
4301 dtd->scaffIndex = temp;
4304 else {
4305 groupConnector = (char *)MALLOC(groupSize = 32);
4306 if (!groupConnector)
4307 return XML_ERROR_NO_MEMORY;
4310 groupConnector[prologState.level] = 0;
4311 if (dtd->in_eldecl) {
4312 int myindex = nextScaffoldPart(parser);
4313 if (myindex < 0)
4314 return XML_ERROR_NO_MEMORY;
4315 dtd->scaffIndex[dtd->scaffLevel] = myindex;
4316 dtd->scaffLevel++;
4317 dtd->scaffold[myindex].type = XML_CTYPE_SEQ;
4318 if (elementDeclHandler)
4319 handleDefault = XML_FALSE;
4321 break;
4322 case XML_ROLE_GROUP_SEQUENCE:
4323 if (groupConnector[prologState.level] == '|')
4324 return XML_ERROR_SYNTAX;
4325 groupConnector[prologState.level] = ',';
4326 if (dtd->in_eldecl && elementDeclHandler)
4327 handleDefault = XML_FALSE;
4328 break;
4329 case XML_ROLE_GROUP_CHOICE:
4330 if (groupConnector[prologState.level] == ',')
4331 return XML_ERROR_SYNTAX;
4332 if (dtd->in_eldecl
4333 && !groupConnector[prologState.level]
4334 && (dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type
4335 != XML_CTYPE_MIXED)
4337 dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type
4338 = XML_CTYPE_CHOICE;
4339 if (elementDeclHandler)
4340 handleDefault = XML_FALSE;
4342 groupConnector[prologState.level] = '|';
4343 break;
4344 case XML_ROLE_PARAM_ENTITY_REF:
4345 #ifdef XML_DTD
4346 case XML_ROLE_INNER_PARAM_ENTITY_REF:
4347 dtd->hasParamEntityRefs = XML_TRUE;
4348 if (!paramEntityParsing)
4349 dtd->keepProcessing = dtd->standalone;
4350 else {
4351 const XML_Char *name;
4352 ENTITY *entity;
4353 name = poolStoreString(&dtd->pool, enc,
4354 s + enc->minBytesPerChar,
4355 next - enc->minBytesPerChar);
4356 if (!name)
4357 return XML_ERROR_NO_MEMORY;
4358 entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0);
4359 poolDiscard(&dtd->pool);
4360 /* first, determine if a check for an existing declaration is needed;
4361 if yes, check that the entity exists, and that it is internal,
4362 otherwise call the skipped entity handler
4364 if (prologState.documentEntity &&
4365 (dtd->standalone
4366 ? !openInternalEntities
4367 : !dtd->hasParamEntityRefs)) {
4368 if (!entity)
4369 return XML_ERROR_UNDEFINED_ENTITY;
4370 else if (!entity->is_internal)
4371 return XML_ERROR_ENTITY_DECLARED_IN_PE;
4373 else if (!entity) {
4374 dtd->keepProcessing = dtd->standalone;
4375 /* cannot report skipped entities in declarations */
4376 if ((role == XML_ROLE_PARAM_ENTITY_REF) && skippedEntityHandler) {
4377 skippedEntityHandler(handlerArg, name, 1);
4378 handleDefault = XML_FALSE;
4380 break;
4382 if (entity->open)
4383 return XML_ERROR_RECURSIVE_ENTITY_REF;
4384 if (entity->textPtr) {
4385 enum XML_Error result;
4386 XML_Bool betweenDecl =
4387 (role == XML_ROLE_PARAM_ENTITY_REF ? XML_TRUE : XML_FALSE);
4388 result = processInternalEntity(parser, entity, betweenDecl);
4389 if (result != XML_ERROR_NONE)
4390 return result;
4391 handleDefault = XML_FALSE;
4392 break;
4394 if (externalEntityRefHandler) {
4395 dtd->paramEntityRead = XML_FALSE;
4396 entity->open = XML_TRUE;
4397 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
4399 entity->base,
4400 entity->systemId,
4401 entity->publicId)) {
4402 entity->open = XML_FALSE;
4403 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
4405 entity->open = XML_FALSE;
4406 handleDefault = XML_FALSE;
4407 if (!dtd->paramEntityRead) {
4408 dtd->keepProcessing = dtd->standalone;
4409 break;
4412 else {
4413 dtd->keepProcessing = dtd->standalone;
4414 break;
4417 #endif /* XML_DTD */
4418 if (!dtd->standalone &&
4419 notStandaloneHandler &&
4420 !notStandaloneHandler(handlerArg))
4421 return XML_ERROR_NOT_STANDALONE;
4422 break;
4424 /* Element declaration stuff */
4426 case XML_ROLE_ELEMENT_NAME:
4427 if (elementDeclHandler) {
4428 declElementType = getElementType(parser, enc, s, next);
4429 if (!declElementType)
4430 return XML_ERROR_NO_MEMORY;
4431 dtd->scaffLevel = 0;
4432 dtd->scaffCount = 0;
4433 dtd->in_eldecl = XML_TRUE;
4434 handleDefault = XML_FALSE;
4436 break;
4438 case XML_ROLE_CONTENT_ANY:
4439 case XML_ROLE_CONTENT_EMPTY:
4440 if (dtd->in_eldecl) {
4441 if (elementDeclHandler) {
4442 XML_Content * content = (XML_Content *) MALLOC(sizeof(XML_Content));
4443 if (!content)
4444 return XML_ERROR_NO_MEMORY;
4445 content->quant = XML_CQUANT_NONE;
4446 content->name = NULL;
4447 content->numchildren = 0;
4448 content->children = NULL;
4449 content->type = ((role == XML_ROLE_CONTENT_ANY) ?
4450 XML_CTYPE_ANY :
4451 XML_CTYPE_EMPTY);
4452 *eventEndPP = s;
4453 elementDeclHandler(handlerArg, declElementType->name, content);
4454 handleDefault = XML_FALSE;
4456 dtd->in_eldecl = XML_FALSE;
4458 break;
4460 case XML_ROLE_CONTENT_PCDATA:
4461 if (dtd->in_eldecl) {
4462 dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type
4463 = XML_CTYPE_MIXED;
4464 if (elementDeclHandler)
4465 handleDefault = XML_FALSE;
4467 break;
4469 case XML_ROLE_CONTENT_ELEMENT:
4470 quant = XML_CQUANT_NONE;
4471 goto elementContent;
4472 case XML_ROLE_CONTENT_ELEMENT_OPT:
4473 quant = XML_CQUANT_OPT;
4474 goto elementContent;
4475 case XML_ROLE_CONTENT_ELEMENT_REP:
4476 quant = XML_CQUANT_REP;
4477 goto elementContent;
4478 case XML_ROLE_CONTENT_ELEMENT_PLUS:
4479 quant = XML_CQUANT_PLUS;
4480 elementContent:
4481 if (dtd->in_eldecl) {
4482 ELEMENT_TYPE *el;
4483 const XML_Char *name;
4484 int nameLen;
4485 const char *nxt = (quant == XML_CQUANT_NONE
4486 ? next
4487 : next - enc->minBytesPerChar);
4488 int myindex = nextScaffoldPart(parser);
4489 if (myindex < 0)
4490 return XML_ERROR_NO_MEMORY;
4491 dtd->scaffold[myindex].type = XML_CTYPE_NAME;
4492 dtd->scaffold[myindex].quant = quant;
4493 el = getElementType(parser, enc, s, nxt);
4494 if (!el)
4495 return XML_ERROR_NO_MEMORY;
4496 name = el->name;
4497 dtd->scaffold[myindex].name = name;
4498 nameLen = 0;
4499 for (; name[nameLen++]; );
4500 dtd->contentStringLen += nameLen;
4501 if (elementDeclHandler)
4502 handleDefault = XML_FALSE;
4504 break;
4506 case XML_ROLE_GROUP_CLOSE:
4507 quant = XML_CQUANT_NONE;
4508 goto closeGroup;
4509 case XML_ROLE_GROUP_CLOSE_OPT:
4510 quant = XML_CQUANT_OPT;
4511 goto closeGroup;
4512 case XML_ROLE_GROUP_CLOSE_REP:
4513 quant = XML_CQUANT_REP;
4514 goto closeGroup;
4515 case XML_ROLE_GROUP_CLOSE_PLUS:
4516 quant = XML_CQUANT_PLUS;
4517 closeGroup:
4518 if (dtd->in_eldecl) {
4519 if (elementDeclHandler)
4520 handleDefault = XML_FALSE;
4521 dtd->scaffLevel--;
4522 dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel]].quant = quant;
4523 if (dtd->scaffLevel == 0) {
4524 if (!handleDefault) {
4525 XML_Content *model = build_model(parser);
4526 if (!model)
4527 return XML_ERROR_NO_MEMORY;
4528 *eventEndPP = s;
4529 elementDeclHandler(handlerArg, declElementType->name, model);
4531 dtd->in_eldecl = XML_FALSE;
4532 dtd->contentStringLen = 0;
4535 break;
4536 /* End element declaration stuff */
4538 case XML_ROLE_PI:
4539 if (!reportProcessingInstruction(parser, enc, s, next))
4540 return XML_ERROR_NO_MEMORY;
4541 handleDefault = XML_FALSE;
4542 break;
4543 case XML_ROLE_COMMENT:
4544 if (!reportComment(parser, enc, s, next))
4545 return XML_ERROR_NO_MEMORY;
4546 handleDefault = XML_FALSE;
4547 break;
4548 case XML_ROLE_NONE:
4549 switch (tok) {
4550 case XML_TOK_BOM:
4551 handleDefault = XML_FALSE;
4552 break;
4554 break;
4555 case XML_ROLE_DOCTYPE_NONE:
4556 if (startDoctypeDeclHandler)
4557 handleDefault = XML_FALSE;
4558 break;
4559 case XML_ROLE_ENTITY_NONE:
4560 if (dtd->keepProcessing && entityDeclHandler)
4561 handleDefault = XML_FALSE;
4562 break;
4563 case XML_ROLE_NOTATION_NONE:
4564 if (notationDeclHandler)
4565 handleDefault = XML_FALSE;
4566 break;
4567 case XML_ROLE_ATTLIST_NONE:
4568 if (dtd->keepProcessing && attlistDeclHandler)
4569 handleDefault = XML_FALSE;
4570 break;
4571 case XML_ROLE_ELEMENT_NONE:
4572 if (elementDeclHandler)
4573 handleDefault = XML_FALSE;
4574 break;
4575 } /* end of big switch */
4577 if (handleDefault && defaultHandler)
4578 reportDefault(parser, enc, s, next);
4580 switch (ps_parsing) {
4581 case XML_SUSPENDED:
4582 *nextPtr = next;
4583 return XML_ERROR_NONE;
4584 case XML_FINISHED:
4585 return XML_ERROR_ABORTED;
4586 default:
4587 s = next;
4588 tok = XmlPrologTok(enc, s, end, &next);
4591 /* not reached */
4594 static enum XML_Error PTRCALL
4595 epilogProcessor(XML_Parser parser,
4596 const char *s,
4597 const char *end,
4598 const char **nextPtr)
4600 processor = epilogProcessor;
4601 eventPtr = s;
4602 for (;;) {
4603 const char *next = NULL;
4604 int tok = XmlPrologTok(encoding, s, end, &next);
4605 eventEndPtr = next;
4606 switch (tok) {
4607 /* report partial linebreak - it might be the last token */
4608 case -XML_TOK_PROLOG_S:
4609 if (defaultHandler) {
4610 reportDefault(parser, encoding, s, next);
4611 if (ps_parsing == XML_FINISHED)
4612 return XML_ERROR_ABORTED;
4614 *nextPtr = next;
4615 return XML_ERROR_NONE;
4616 case XML_TOK_NONE:
4617 *nextPtr = s;
4618 return XML_ERROR_NONE;
4619 case XML_TOK_PROLOG_S:
4620 if (defaultHandler)
4621 reportDefault(parser, encoding, s, next);
4622 break;
4623 case XML_TOK_PI:
4624 if (!reportProcessingInstruction(parser, encoding, s, next))
4625 return XML_ERROR_NO_MEMORY;
4626 break;
4627 case XML_TOK_COMMENT:
4628 if (!reportComment(parser, encoding, s, next))
4629 return XML_ERROR_NO_MEMORY;
4630 break;
4631 case XML_TOK_INVALID:
4632 eventPtr = next;
4633 return XML_ERROR_INVALID_TOKEN;
4634 case XML_TOK_PARTIAL:
4635 if (!ps_finalBuffer) {
4636 *nextPtr = s;
4637 return XML_ERROR_NONE;
4639 return XML_ERROR_UNCLOSED_TOKEN;
4640 case XML_TOK_PARTIAL_CHAR:
4641 if (!ps_finalBuffer) {
4642 *nextPtr = s;
4643 return XML_ERROR_NONE;
4645 return XML_ERROR_PARTIAL_CHAR;
4646 default:
4647 return XML_ERROR_JUNK_AFTER_DOC_ELEMENT;
4649 eventPtr = s = next;
4650 switch (ps_parsing) {
4651 case XML_SUSPENDED:
4652 *nextPtr = next;
4653 return XML_ERROR_NONE;
4654 case XML_FINISHED:
4655 return XML_ERROR_ABORTED;
4656 default: ;
4661 static enum XML_Error
4662 processInternalEntity(XML_Parser parser, ENTITY *entity,
4663 XML_Bool betweenDecl)
4665 const char *textStart, *textEnd;
4666 const char *next;
4667 enum XML_Error result;
4668 OPEN_INTERNAL_ENTITY *openEntity;
4670 if (freeInternalEntities) {
4671 openEntity = freeInternalEntities;
4672 freeInternalEntities = openEntity->next;
4674 else {
4675 openEntity = (OPEN_INTERNAL_ENTITY *)MALLOC(sizeof(OPEN_INTERNAL_ENTITY));
4676 if (!openEntity)
4677 return XML_ERROR_NO_MEMORY;
4679 entity->open = XML_TRUE;
4680 entity->processed = 0;
4681 openEntity->next = openInternalEntities;
4682 openInternalEntities = openEntity;
4683 openEntity->entity = entity;
4684 openEntity->startTagLevel = tagLevel;
4685 openEntity->betweenDecl = betweenDecl;
4686 openEntity->internalEventPtr = NULL;
4687 openEntity->internalEventEndPtr = NULL;
4688 textStart = (char *)entity->textPtr;
4689 textEnd = (char *)(entity->textPtr + entity->textLen);
4691 #ifdef XML_DTD
4692 if (entity->is_param) {
4693 int tok = XmlPrologTok(internalEncoding, textStart, textEnd, &next);
4694 result = doProlog(parser, internalEncoding, textStart, textEnd, tok,
4695 next, &next, XML_FALSE);
4697 else
4698 #endif /* XML_DTD */
4699 result = doContent(parser, tagLevel, internalEncoding, textStart,
4700 textEnd, &next, XML_FALSE);
4702 if (result == XML_ERROR_NONE) {
4703 if (textEnd != next && ps_parsing == XML_SUSPENDED) {
4704 entity->processed = (int)(next - textStart);
4705 processor = internalEntityProcessor;
4707 else {
4708 entity->open = XML_FALSE;
4709 openInternalEntities = openEntity->next;
4710 /* put openEntity back in list of free instances */
4711 openEntity->next = freeInternalEntities;
4712 freeInternalEntities = openEntity;
4715 return result;
4718 static enum XML_Error PTRCALL
4719 internalEntityProcessor(XML_Parser parser,
4720 const char *s,
4721 const char *end,
4722 const char **nextPtr)
4724 ENTITY *entity;
4725 const char *textStart, *textEnd;
4726 const char *next;
4727 enum XML_Error result;
4728 OPEN_INTERNAL_ENTITY *openEntity = openInternalEntities;
4729 if (!openEntity)
4730 return XML_ERROR_UNEXPECTED_STATE;
4732 entity = openEntity->entity;
4733 textStart = ((char *)entity->textPtr) + entity->processed;
4734 textEnd = (char *)(entity->textPtr + entity->textLen);
4736 #ifdef XML_DTD
4737 if (entity->is_param) {
4738 int tok = XmlPrologTok(internalEncoding, textStart, textEnd, &next);
4739 result = doProlog(parser, internalEncoding, textStart, textEnd, tok,
4740 next, &next, XML_FALSE);
4742 else
4743 #endif /* XML_DTD */
4744 result = doContent(parser, openEntity->startTagLevel, internalEncoding,
4745 textStart, textEnd, &next, XML_FALSE);
4747 if (result != XML_ERROR_NONE)
4748 return result;
4749 else if (textEnd != next && ps_parsing == XML_SUSPENDED) {
4750 entity->processed = (int)(next - (char *)entity->textPtr);
4751 return result;
4753 else {
4754 entity->open = XML_FALSE;
4755 openInternalEntities = openEntity->next;
4756 /* put openEntity back in list of free instances */
4757 openEntity->next = freeInternalEntities;
4758 freeInternalEntities = openEntity;
4761 #ifdef XML_DTD
4762 if (entity->is_param) {
4763 int tok;
4764 processor = prologProcessor;
4765 tok = XmlPrologTok(encoding, s, end, &next);
4766 return doProlog(parser, encoding, s, end, tok, next, nextPtr,
4767 (XML_Bool)!ps_finalBuffer);
4769 else
4770 #endif /* XML_DTD */
4772 processor = contentProcessor;
4773 /* see externalEntityContentProcessor vs contentProcessor */
4774 return doContent(parser, parentParser ? 1 : 0, encoding, s, end,
4775 nextPtr, (XML_Bool)!ps_finalBuffer);
4779 static enum XML_Error PTRCALL
4780 errorProcessor(XML_Parser parser,
4781 const char *s,
4782 const char *end,
4783 const char **nextPtr)
4785 return errorCode;
4788 static enum XML_Error
4789 storeAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
4790 const char *ptr, const char *end,
4791 STRING_POOL *pool)
4793 enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr,
4794 end, pool);
4795 if (result)
4796 return result;
4797 if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20)
4798 poolChop(pool);
4799 if (!poolAppendChar(pool, XML_T('\0')))
4800 return XML_ERROR_NO_MEMORY;
4801 return XML_ERROR_NONE;
4804 static enum XML_Error
4805 appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
4806 const char *ptr, const char *end,
4807 STRING_POOL *pool)
4809 DTD * const dtd = _dtd; /* save one level of indirection */
4810 for (;;) {
4811 const char *next;
4812 int tok = XmlAttributeValueTok(enc, ptr, end, &next);
4813 switch (tok) {
4814 case XML_TOK_NONE:
4815 return XML_ERROR_NONE;
4816 case XML_TOK_INVALID:
4817 if (enc == encoding)
4818 eventPtr = next;
4819 return XML_ERROR_INVALID_TOKEN;
4820 case XML_TOK_PARTIAL:
4821 if (enc == encoding)
4822 eventPtr = ptr;
4823 return XML_ERROR_INVALID_TOKEN;
4824 case XML_TOK_CHAR_REF:
4826 XML_Char buf[XML_ENCODE_MAX];
4827 int i;
4828 int n = XmlCharRefNumber(enc, ptr);
4829 if (n < 0) {
4830 if (enc == encoding)
4831 eventPtr = ptr;
4832 return XML_ERROR_BAD_CHAR_REF;
4834 if (!isCdata
4835 && n == 0x20 /* space */
4836 && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
4837 break;
4838 n = XmlEncode(n, (ICHAR *)buf);
4839 if (!n) {
4840 if (enc == encoding)
4841 eventPtr = ptr;
4842 return XML_ERROR_BAD_CHAR_REF;
4844 for (i = 0; i < n; i++) {
4845 if (!poolAppendChar(pool, buf[i]))
4846 return XML_ERROR_NO_MEMORY;
4849 break;
4850 case XML_TOK_DATA_CHARS:
4851 if (!poolAppend(pool, enc, ptr, next))
4852 return XML_ERROR_NO_MEMORY;
4853 break;
4854 case XML_TOK_TRAILING_CR:
4855 next = ptr + enc->minBytesPerChar;
4856 /* fall through */
4857 case XML_TOK_ATTRIBUTE_VALUE_S:
4858 case XML_TOK_DATA_NEWLINE:
4859 if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
4860 break;
4861 if (!poolAppendChar(pool, 0x20))
4862 return XML_ERROR_NO_MEMORY;
4863 break;
4864 case XML_TOK_ENTITY_REF:
4866 const XML_Char *name;
4867 ENTITY *entity;
4868 char checkEntityDecl;
4869 XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc,
4870 ptr + enc->minBytesPerChar,
4871 next - enc->minBytesPerChar);
4872 if (ch) {
4873 if (!poolAppendChar(pool, ch))
4874 return XML_ERROR_NO_MEMORY;
4875 break;
4877 name = poolStoreString(&temp2Pool, enc,
4878 ptr + enc->minBytesPerChar,
4879 next - enc->minBytesPerChar);
4880 if (!name)
4881 return XML_ERROR_NO_MEMORY;
4882 entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0);
4883 poolDiscard(&temp2Pool);
4884 /* First, determine if a check for an existing declaration is needed;
4885 if yes, check that the entity exists, and that it is internal.
4887 if (pool == &dtd->pool) /* are we called from prolog? */
4888 checkEntityDecl =
4889 #ifdef XML_DTD
4890 prologState.documentEntity &&
4891 #endif /* XML_DTD */
4892 (dtd->standalone
4893 ? !openInternalEntities
4894 : !dtd->hasParamEntityRefs);
4895 else /* if (pool == &tempPool): we are called from content */
4896 checkEntityDecl = !dtd->hasParamEntityRefs || dtd->standalone;
4897 if (checkEntityDecl) {
4898 if (!entity)
4899 return XML_ERROR_UNDEFINED_ENTITY;
4900 else if (!entity->is_internal)
4901 return XML_ERROR_ENTITY_DECLARED_IN_PE;
4903 else if (!entity) {
4904 /* Cannot report skipped entity here - see comments on
4905 skippedEntityHandler.
4906 if (skippedEntityHandler)
4907 skippedEntityHandler(handlerArg, name, 0);
4909 /* Cannot call the default handler because this would be
4910 out of sync with the call to the startElementHandler.
4911 if ((pool == &tempPool) && defaultHandler)
4912 reportDefault(parser, enc, ptr, next);
4914 break;
4916 if (entity->open) {
4917 if (enc == encoding)
4918 eventPtr = ptr;
4919 return XML_ERROR_RECURSIVE_ENTITY_REF;
4921 if (entity->notation) {
4922 if (enc == encoding)
4923 eventPtr = ptr;
4924 return XML_ERROR_BINARY_ENTITY_REF;
4926 if (!entity->textPtr) {
4927 if (enc == encoding)
4928 eventPtr = ptr;
4929 return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF;
4931 else {
4932 enum XML_Error result;
4933 const XML_Char *textEnd = entity->textPtr + entity->textLen;
4934 entity->open = XML_TRUE;
4935 result = appendAttributeValue(parser, internalEncoding, isCdata,
4936 (char *)entity->textPtr,
4937 (char *)textEnd, pool);
4938 entity->open = XML_FALSE;
4939 if (result)
4940 return result;
4943 break;
4944 default:
4945 if (enc == encoding)
4946 eventPtr = ptr;
4947 return XML_ERROR_UNEXPECTED_STATE;
4949 ptr = next;
4951 /* not reached */
4954 static enum XML_Error
4955 storeEntityValue(XML_Parser parser,
4956 const ENCODING *enc,
4957 const char *entityTextPtr,
4958 const char *entityTextEnd)
4960 DTD * const dtd = _dtd; /* save one level of indirection */
4961 STRING_POOL *pool = &(dtd->entityValuePool);
4962 enum XML_Error result = XML_ERROR_NONE;
4963 #ifdef XML_DTD
4964 int oldInEntityValue = prologState.inEntityValue;
4965 prologState.inEntityValue = 1;
4966 #endif /* XML_DTD */
4967 /* never return Null for the value argument in EntityDeclHandler,
4968 since this would indicate an external entity; therefore we
4969 have to make sure that entityValuePool.start is not null */
4970 if (!pool->blocks) {
4971 if (!poolGrow(pool))
4972 return XML_ERROR_NO_MEMORY;
4975 for (;;) {
4976 const char *next;
4977 int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next);
4978 switch (tok) {
4979 case XML_TOK_PARAM_ENTITY_REF:
4980 #ifdef XML_DTD
4981 if (isParamEntity || enc != encoding) {
4982 const XML_Char *name;
4983 ENTITY *entity;
4984 name = poolStoreString(&tempPool, enc,
4985 entityTextPtr + enc->minBytesPerChar,
4986 next - enc->minBytesPerChar);
4987 if (!name) {
4988 result = XML_ERROR_NO_MEMORY;
4989 goto endEntityValue;
4991 entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0);
4992 poolDiscard(&tempPool);
4993 if (!entity) {
4994 /* not a well-formedness error - see XML 1.0: WFC Entity Declared */
4995 /* cannot report skipped entity here - see comments on
4996 skippedEntityHandler
4997 if (skippedEntityHandler)
4998 skippedEntityHandler(handlerArg, name, 0);
5000 dtd->keepProcessing = dtd->standalone;
5001 goto endEntityValue;
5003 if (entity->open) {
5004 if (enc == encoding)
5005 eventPtr = entityTextPtr;
5006 result = XML_ERROR_RECURSIVE_ENTITY_REF;
5007 goto endEntityValue;
5009 if (entity->systemId) {
5010 if (externalEntityRefHandler) {
5011 dtd->paramEntityRead = XML_FALSE;
5012 entity->open = XML_TRUE;
5013 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
5015 entity->base,
5016 entity->systemId,
5017 entity->publicId)) {
5018 entity->open = XML_FALSE;
5019 result = XML_ERROR_EXTERNAL_ENTITY_HANDLING;
5020 goto endEntityValue;
5022 entity->open = XML_FALSE;
5023 if (!dtd->paramEntityRead)
5024 dtd->keepProcessing = dtd->standalone;
5026 else
5027 dtd->keepProcessing = dtd->standalone;
5029 else {
5030 entity->open = XML_TRUE;
5031 result = storeEntityValue(parser,
5032 internalEncoding,
5033 (char *)entity->textPtr,
5034 (char *)(entity->textPtr
5035 + entity->textLen));
5036 entity->open = XML_FALSE;
5037 if (result)
5038 goto endEntityValue;
5040 break;
5042 #endif /* XML_DTD */
5043 /* In the internal subset, PE references are not legal
5044 within markup declarations, e.g entity values in this case. */
5045 eventPtr = entityTextPtr;
5046 result = XML_ERROR_PARAM_ENTITY_REF;
5047 goto endEntityValue;
5048 case XML_TOK_NONE:
5049 result = XML_ERROR_NONE;
5050 goto endEntityValue;
5051 case XML_TOK_ENTITY_REF:
5052 case XML_TOK_DATA_CHARS:
5053 if (!poolAppend(pool, enc, entityTextPtr, next)) {
5054 result = XML_ERROR_NO_MEMORY;
5055 goto endEntityValue;
5057 break;
5058 case XML_TOK_TRAILING_CR:
5059 next = entityTextPtr + enc->minBytesPerChar;
5060 /* fall through */
5061 case XML_TOK_DATA_NEWLINE:
5062 if (pool->end == pool->ptr && !poolGrow(pool)) {
5063 result = XML_ERROR_NO_MEMORY;
5064 goto endEntityValue;
5066 *(pool->ptr)++ = 0xA;
5067 break;
5068 case XML_TOK_CHAR_REF:
5070 XML_Char buf[XML_ENCODE_MAX];
5071 int i;
5072 int n = XmlCharRefNumber(enc, entityTextPtr);
5073 if (n < 0) {
5074 if (enc == encoding)
5075 eventPtr = entityTextPtr;
5076 result = XML_ERROR_BAD_CHAR_REF;
5077 goto endEntityValue;
5079 n = XmlEncode(n, (ICHAR *)buf);
5080 if (!n) {
5081 if (enc == encoding)
5082 eventPtr = entityTextPtr;
5083 result = XML_ERROR_BAD_CHAR_REF;
5084 goto endEntityValue;
5086 for (i = 0; i < n; i++) {
5087 if (pool->end == pool->ptr && !poolGrow(pool)) {
5088 result = XML_ERROR_NO_MEMORY;
5089 goto endEntityValue;
5091 *(pool->ptr)++ = buf[i];
5094 break;
5095 case XML_TOK_PARTIAL:
5096 if (enc == encoding)
5097 eventPtr = entityTextPtr;
5098 result = XML_ERROR_INVALID_TOKEN;
5099 goto endEntityValue;
5100 case XML_TOK_INVALID:
5101 if (enc == encoding)
5102 eventPtr = next;
5103 result = XML_ERROR_INVALID_TOKEN;
5104 goto endEntityValue;
5105 default:
5106 if (enc == encoding)
5107 eventPtr = entityTextPtr;
5108 result = XML_ERROR_UNEXPECTED_STATE;
5109 goto endEntityValue;
5111 entityTextPtr = next;
5113 endEntityValue:
5114 #ifdef XML_DTD
5115 prologState.inEntityValue = oldInEntityValue;
5116 #endif /* XML_DTD */
5117 return result;
5120 static void FASTCALL
5121 normalizeLines(XML_Char *s)
5123 XML_Char *p;
5124 for (;; s++) {
5125 if (*s == XML_T('\0'))
5126 return;
5127 if (*s == 0xD)
5128 break;
5130 p = s;
5131 do {
5132 if (*s == 0xD) {
5133 *p++ = 0xA;
5134 if (*++s == 0xA)
5135 s++;
5137 else
5138 *p++ = *s++;
5139 } while (*s);
5140 *p = XML_T('\0');
5143 static int
5144 reportProcessingInstruction(XML_Parser parser, const ENCODING *enc,
5145 const char *start, const char *end)
5147 const XML_Char *target;
5148 XML_Char *data;
5149 const char *tem;
5150 if (!processingInstructionHandler) {
5151 if (defaultHandler)
5152 reportDefault(parser, enc, start, end);
5153 return 1;
5155 start += enc->minBytesPerChar * 2;
5156 tem = start + XmlNameLength(enc, start);
5157 target = poolStoreString(&tempPool, enc, start, tem);
5158 if (!target)
5159 return 0;
5160 poolFinish(&tempPool);
5161 data = poolStoreString(&tempPool, enc,
5162 XmlSkipS(enc, tem),
5163 end - enc->minBytesPerChar*2);
5164 if (!data)
5165 return 0;
5166 normalizeLines(data);
5167 processingInstructionHandler(handlerArg, target, data);
5168 poolClear(&tempPool);
5169 return 1;
5172 static int
5173 reportComment(XML_Parser parser, const ENCODING *enc,
5174 const char *start, const char *end)
5176 XML_Char *data;
5177 if (!commentHandler) {
5178 if (defaultHandler)
5179 reportDefault(parser, enc, start, end);
5180 return 1;
5182 data = poolStoreString(&tempPool,
5183 enc,
5184 start + enc->minBytesPerChar * 4,
5185 end - enc->minBytesPerChar * 3);
5186 if (!data)
5187 return 0;
5188 normalizeLines(data);
5189 commentHandler(handlerArg, data);
5190 poolClear(&tempPool);
5191 return 1;
5194 static void
5195 reportDefault(XML_Parser parser, const ENCODING *enc,
5196 const char *s, const char *end)
5198 if (MUST_CONVERT(enc, s)) {
5199 const char **eventPP;
5200 const char **eventEndPP;
5201 if (enc == encoding) {
5202 eventPP = &eventPtr;
5203 eventEndPP = &eventEndPtr;
5205 else {
5206 eventPP = &(openInternalEntities->internalEventPtr);
5207 eventEndPP = &(openInternalEntities->internalEventEndPtr);
5209 do {
5210 ICHAR *dataPtr = (ICHAR *)dataBuf;
5211 XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd);
5212 *eventEndPP = s;
5213 defaultHandler(handlerArg, dataBuf, (int)(dataPtr - (ICHAR *)dataBuf));
5214 *eventPP = s;
5215 } while (s != end);
5217 else
5218 defaultHandler(handlerArg, (XML_Char *)s, (int)((XML_Char *)end - (XML_Char *)s));
5222 static int
5223 defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, XML_Bool isCdata,
5224 XML_Bool isId, const XML_Char *value, XML_Parser parser)
5226 DEFAULT_ATTRIBUTE *att;
5227 if (value || isId) {
5228 /* The handling of default attributes gets messed up if we have
5229 a default which duplicates a non-default. */
5230 int i;
5231 for (i = 0; i < type->nDefaultAtts; i++)
5232 if (attId == type->defaultAtts[i].id)
5233 return 1;
5234 if (isId && !type->idAtt && !attId->xmlns)
5235 type->idAtt = attId;
5237 if (type->nDefaultAtts == type->allocDefaultAtts) {
5238 if (type->allocDefaultAtts == 0) {
5239 type->allocDefaultAtts = 8;
5240 type->defaultAtts = (DEFAULT_ATTRIBUTE *)MALLOC(type->allocDefaultAtts
5241 * sizeof(DEFAULT_ATTRIBUTE));
5242 if (!type->defaultAtts)
5243 return 0;
5245 else {
5246 DEFAULT_ATTRIBUTE *temp;
5247 int count = type->allocDefaultAtts * 2;
5248 temp = (DEFAULT_ATTRIBUTE *)
5249 REALLOC(type->defaultAtts, (count * sizeof(DEFAULT_ATTRIBUTE)));
5250 if (temp == NULL)
5251 return 0;
5252 type->allocDefaultAtts = count;
5253 type->defaultAtts = temp;
5256 att = type->defaultAtts + type->nDefaultAtts;
5257 att->id = attId;
5258 att->value = value;
5259 att->isCdata = isCdata;
5260 if (!isCdata)
5261 attId->maybeTokenized = XML_TRUE;
5262 type->nDefaultAtts += 1;
5263 return 1;
5266 static int
5267 setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType)
5269 DTD * const dtd = _dtd; /* save one level of indirection */
5270 const XML_Char *name;
5271 for (name = elementType->name; *name; name++) {
5272 if (*name == XML_T(':')) {
5273 PREFIX *prefix;
5274 const XML_Char *s;
5275 for (s = elementType->name; s != name; s++) {
5276 if (!poolAppendChar(&dtd->pool, *s))
5277 return 0;
5279 if (!poolAppendChar(&dtd->pool, XML_T('\0')))
5280 return 0;
5281 prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool),
5282 sizeof(PREFIX));
5283 if (!prefix)
5284 return 0;
5285 if (prefix->name == poolStart(&dtd->pool))
5286 poolFinish(&dtd->pool);
5287 else
5288 poolDiscard(&dtd->pool);
5289 elementType->prefix = prefix;
5293 return 1;
5296 static ATTRIBUTE_ID *
5297 getAttributeId(XML_Parser parser, const ENCODING *enc,
5298 const char *start, const char *end)
5300 DTD * const dtd = _dtd; /* save one level of indirection */
5301 ATTRIBUTE_ID *id;
5302 const XML_Char *name;
5303 if (!poolAppendChar(&dtd->pool, XML_T('\0')))
5304 return NULL;
5305 name = poolStoreString(&dtd->pool, enc, start, end);
5306 if (!name)
5307 return NULL;
5308 /* skip quotation mark - its storage will be re-used (like in name[-1]) */
5309 ++name;
5310 id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, name, sizeof(ATTRIBUTE_ID));
5311 if (!id)
5312 return NULL;
5313 if (id->name != name)
5314 poolDiscard(&dtd->pool);
5315 else {
5316 poolFinish(&dtd->pool);
5317 if (!ns)
5319 else if (name[0] == XML_T('x')
5320 && name[1] == XML_T('m')
5321 && name[2] == XML_T('l')
5322 && name[3] == XML_T('n')
5323 && name[4] == XML_T('s')
5324 && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) {
5325 if (name[5] == XML_T('\0'))
5326 id->prefix = &dtd->defaultPrefix;
5327 else
5328 id->prefix = (PREFIX *)lookup(&dtd->prefixes, name + 6, sizeof(PREFIX));
5329 id->xmlns = XML_TRUE;
5331 else {
5332 int i;
5333 for (i = 0; name[i]; i++) {
5334 /* attributes without prefix are *not* in the default namespace */
5335 if (name[i] == XML_T(':')) {
5336 int j;
5337 for (j = 0; j < i; j++) {
5338 if (!poolAppendChar(&dtd->pool, name[j]))
5339 return NULL;
5341 if (!poolAppendChar(&dtd->pool, XML_T('\0')))
5342 return NULL;
5343 id->prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool),
5344 sizeof(PREFIX));
5345 if (!id->prefix)
5346 return NULL;
5347 if (id->prefix->name == poolStart(&dtd->pool))
5348 poolFinish(&dtd->pool);
5349 else
5350 poolDiscard(&dtd->pool);
5351 break;
5356 return id;
5359 #define CONTEXT_SEP XML_T('\f')
5361 static const XML_Char *
5362 getContext(XML_Parser parser)
5364 DTD * const dtd = _dtd; /* save one level of indirection */
5365 HASH_TABLE_ITER iter;
5366 XML_Bool needSep = XML_FALSE;
5368 if (dtd->defaultPrefix.binding) {
5369 int i;
5370 int len;
5371 if (!poolAppendChar(&tempPool, XML_T('=')))
5372 return NULL;
5373 len = dtd->defaultPrefix.binding->uriLen;
5374 if (namespaceSeparator)
5375 len--;
5376 for (i = 0; i < len; i++)
5377 if (!poolAppendChar(&tempPool, dtd->defaultPrefix.binding->uri[i]))
5378 return NULL;
5379 needSep = XML_TRUE;
5382 hashTableIterInit(&iter, &(dtd->prefixes));
5383 for (;;) {
5384 int i;
5385 int len;
5386 const XML_Char *s;
5387 PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter);
5388 if (!prefix)
5389 break;
5390 if (!prefix->binding)
5391 continue;
5392 if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
5393 return NULL;
5394 for (s = prefix->name; *s; s++)
5395 if (!poolAppendChar(&tempPool, *s))
5396 return NULL;
5397 if (!poolAppendChar(&tempPool, XML_T('=')))
5398 return NULL;
5399 len = prefix->binding->uriLen;
5400 if (namespaceSeparator)
5401 len--;
5402 for (i = 0; i < len; i++)
5403 if (!poolAppendChar(&tempPool, prefix->binding->uri[i]))
5404 return NULL;
5405 needSep = XML_TRUE;
5409 hashTableIterInit(&iter, &(dtd->generalEntities));
5410 for (;;) {
5411 const XML_Char *s;
5412 ENTITY *e = (ENTITY *)hashTableIterNext(&iter);
5413 if (!e)
5414 break;
5415 if (!e->open)
5416 continue;
5417 if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
5418 return NULL;
5419 for (s = e->name; *s; s++)
5420 if (!poolAppendChar(&tempPool, *s))
5421 return 0;
5422 needSep = XML_TRUE;
5425 if (!poolAppendChar(&tempPool, XML_T('\0')))
5426 return NULL;
5427 return tempPool.start;
5430 static XML_Bool
5431 setContext(XML_Parser parser, const XML_Char *context)
5433 DTD * const dtd = _dtd; /* save one level of indirection */
5434 const XML_Char *s = context;
5436 while (*context != XML_T('\0')) {
5437 if (*s == CONTEXT_SEP || *s == XML_T('\0')) {
5438 ENTITY *e;
5439 if (!poolAppendChar(&tempPool, XML_T('\0')))
5440 return XML_FALSE;
5441 e = (ENTITY *)lookup(&dtd->generalEntities, poolStart(&tempPool), 0);
5442 if (e)
5443 e->open = XML_TRUE;
5444 if (*s != XML_T('\0'))
5445 s++;
5446 context = s;
5447 poolDiscard(&tempPool);
5449 else if (*s == XML_T('=')) {
5450 PREFIX *prefix;
5451 if (poolLength(&tempPool) == 0)
5452 prefix = &dtd->defaultPrefix;
5453 else {
5454 if (!poolAppendChar(&tempPool, XML_T('\0')))
5455 return XML_FALSE;
5456 prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&tempPool),
5457 sizeof(PREFIX));
5458 if (!prefix)
5459 return XML_FALSE;
5460 if (prefix->name == poolStart(&tempPool)) {
5461 prefix->name = poolCopyString(&dtd->pool, prefix->name);
5462 if (!prefix->name)
5463 return XML_FALSE;
5465 poolDiscard(&tempPool);
5467 for (context = s + 1;
5468 *context != CONTEXT_SEP && *context != XML_T('\0');
5469 context++)
5470 if (!poolAppendChar(&tempPool, *context))
5471 return XML_FALSE;
5472 if (!poolAppendChar(&tempPool, XML_T('\0')))
5473 return XML_FALSE;
5474 if (addBinding(parser, prefix, NULL, poolStart(&tempPool),
5475 &inheritedBindings) != XML_ERROR_NONE)
5476 return XML_FALSE;
5477 poolDiscard(&tempPool);
5478 if (*context != XML_T('\0'))
5479 ++context;
5480 s = context;
5482 else {
5483 if (!poolAppendChar(&tempPool, *s))
5484 return XML_FALSE;
5485 s++;
5488 return XML_TRUE;
5491 static void FASTCALL
5492 normalizePublicId(XML_Char *publicId)
5494 XML_Char *p = publicId;
5495 XML_Char *s;
5496 for (s = publicId; *s; s++) {
5497 switch (*s) {
5498 case 0x20:
5499 case 0xD:
5500 case 0xA:
5501 if (p != publicId && p[-1] != 0x20)
5502 *p++ = 0x20;
5503 break;
5504 default:
5505 *p++ = *s;
5508 if (p != publicId && p[-1] == 0x20)
5509 --p;
5510 *p = XML_T('\0');
5513 static DTD *
5514 dtdCreate(const XML_Memory_Handling_Suite *ms)
5516 DTD *p = (DTD *)ms->malloc_fcn(sizeof(DTD));
5517 if (p == NULL)
5518 return p;
5519 poolInit(&(p->pool), ms);
5520 poolInit(&(p->entityValuePool), ms);
5521 hashTableInit(&(p->generalEntities), ms);
5522 hashTableInit(&(p->elementTypes), ms);
5523 hashTableInit(&(p->attributeIds), ms);
5524 hashTableInit(&(p->prefixes), ms);
5525 #ifdef XML_DTD
5526 p->paramEntityRead = XML_FALSE;
5527 hashTableInit(&(p->paramEntities), ms);
5528 #endif /* XML_DTD */
5529 p->defaultPrefix.name = NULL;
5530 p->defaultPrefix.binding = NULL;
5532 p->in_eldecl = XML_FALSE;
5533 p->scaffIndex = NULL;
5534 p->scaffold = NULL;
5535 p->scaffLevel = 0;
5536 p->scaffSize = 0;
5537 p->scaffCount = 0;
5538 p->contentStringLen = 0;
5540 p->keepProcessing = XML_TRUE;
5541 p->hasParamEntityRefs = XML_FALSE;
5542 p->standalone = XML_FALSE;
5543 return p;
5546 static void
5547 dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms)
5549 HASH_TABLE_ITER iter;
5550 hashTableIterInit(&iter, &(p->elementTypes));
5551 for (;;) {
5552 ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter);
5553 if (!e)
5554 break;
5555 if (e->allocDefaultAtts != 0)
5556 ms->free_fcn(e->defaultAtts);
5558 hashTableClear(&(p->generalEntities));
5559 #ifdef XML_DTD
5560 p->paramEntityRead = XML_FALSE;
5561 hashTableClear(&(p->paramEntities));
5562 #endif /* XML_DTD */
5563 hashTableClear(&(p->elementTypes));
5564 hashTableClear(&(p->attributeIds));
5565 hashTableClear(&(p->prefixes));
5566 poolClear(&(p->pool));
5567 poolClear(&(p->entityValuePool));
5568 p->defaultPrefix.name = NULL;
5569 p->defaultPrefix.binding = NULL;
5571 p->in_eldecl = XML_FALSE;
5573 ms->free_fcn(p->scaffIndex);
5574 p->scaffIndex = NULL;
5575 ms->free_fcn(p->scaffold);
5576 p->scaffold = NULL;
5578 p->scaffLevel = 0;
5579 p->scaffSize = 0;
5580 p->scaffCount = 0;
5581 p->contentStringLen = 0;
5583 p->keepProcessing = XML_TRUE;
5584 p->hasParamEntityRefs = XML_FALSE;
5585 p->standalone = XML_FALSE;
5588 static void
5589 dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms)
5591 HASH_TABLE_ITER iter;
5592 hashTableIterInit(&iter, &(p->elementTypes));
5593 for (;;) {
5594 ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter);
5595 if (!e)
5596 break;
5597 if (e->allocDefaultAtts != 0)
5598 ms->free_fcn(e->defaultAtts);
5600 hashTableDestroy(&(p->generalEntities));
5601 #ifdef XML_DTD
5602 hashTableDestroy(&(p->paramEntities));
5603 #endif /* XML_DTD */
5604 hashTableDestroy(&(p->elementTypes));
5605 hashTableDestroy(&(p->attributeIds));
5606 hashTableDestroy(&(p->prefixes));
5607 poolDestroy(&(p->pool));
5608 poolDestroy(&(p->entityValuePool));
5609 if (isDocEntity) {
5610 ms->free_fcn(p->scaffIndex);
5611 ms->free_fcn(p->scaffold);
5613 ms->free_fcn(p);
5616 /* Do a deep copy of the DTD. Return 0 for out of memory, non-zero otherwise.
5617 The new DTD has already been initialized.
5619 static int
5620 dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms)
5622 HASH_TABLE_ITER iter;
5624 /* Copy the prefix table. */
5626 hashTableIterInit(&iter, &(oldDtd->prefixes));
5627 for (;;) {
5628 const XML_Char *name;
5629 const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter);
5630 if (!oldP)
5631 break;
5632 name = poolCopyString(&(newDtd->pool), oldP->name);
5633 if (!name)
5634 return 0;
5635 if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX)))
5636 return 0;
5639 hashTableIterInit(&iter, &(oldDtd->attributeIds));
5641 /* Copy the attribute id table. */
5643 for (;;) {
5644 ATTRIBUTE_ID *newA;
5645 const XML_Char *name;
5646 const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter);
5648 if (!oldA)
5649 break;
5650 /* Remember to allocate the scratch byte before the name. */
5651 if (!poolAppendChar(&(newDtd->pool), XML_T('\0')))
5652 return 0;
5653 name = poolCopyString(&(newDtd->pool), oldA->name);
5654 if (!name)
5655 return 0;
5656 ++name;
5657 newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name,
5658 sizeof(ATTRIBUTE_ID));
5659 if (!newA)
5660 return 0;
5661 newA->maybeTokenized = oldA->maybeTokenized;
5662 if (oldA->prefix) {
5663 newA->xmlns = oldA->xmlns;
5664 if (oldA->prefix == &oldDtd->defaultPrefix)
5665 newA->prefix = &newDtd->defaultPrefix;
5666 else
5667 newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes),
5668 oldA->prefix->name, 0);
5672 /* Copy the element type table. */
5674 hashTableIterInit(&iter, &(oldDtd->elementTypes));
5676 for (;;) {
5677 int i;
5678 ELEMENT_TYPE *newE;
5679 const XML_Char *name;
5680 const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter);
5681 if (!oldE)
5682 break;
5683 name = poolCopyString(&(newDtd->pool), oldE->name);
5684 if (!name)
5685 return 0;
5686 newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name,
5687 sizeof(ELEMENT_TYPE));
5688 if (!newE)
5689 return 0;
5690 if (oldE->nDefaultAtts) {
5691 newE->defaultAtts = (DEFAULT_ATTRIBUTE *)
5692 ms->malloc_fcn(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE));
5693 if (!newE->defaultAtts) {
5694 ms->free_fcn(newE);
5695 return 0;
5698 if (oldE->idAtt)
5699 newE->idAtt = (ATTRIBUTE_ID *)
5700 lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0);
5701 newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts;
5702 if (oldE->prefix)
5703 newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes),
5704 oldE->prefix->name, 0);
5705 for (i = 0; i < newE->nDefaultAtts; i++) {
5706 newE->defaultAtts[i].id = (ATTRIBUTE_ID *)
5707 lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0);
5708 newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata;
5709 if (oldE->defaultAtts[i].value) {
5710 newE->defaultAtts[i].value
5711 = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value);
5712 if (!newE->defaultAtts[i].value)
5713 return 0;
5715 else
5716 newE->defaultAtts[i].value = NULL;
5720 /* Copy the entity tables. */
5721 if (!copyEntityTable(&(newDtd->generalEntities),
5722 &(newDtd->pool),
5723 &(oldDtd->generalEntities)))
5724 return 0;
5726 #ifdef XML_DTD
5727 if (!copyEntityTable(&(newDtd->paramEntities),
5728 &(newDtd->pool),
5729 &(oldDtd->paramEntities)))
5730 return 0;
5731 newDtd->paramEntityRead = oldDtd->paramEntityRead;
5732 #endif /* XML_DTD */
5734 newDtd->keepProcessing = oldDtd->keepProcessing;
5735 newDtd->hasParamEntityRefs = oldDtd->hasParamEntityRefs;
5736 newDtd->standalone = oldDtd->standalone;
5738 /* Don't want deep copying for scaffolding */
5739 newDtd->in_eldecl = oldDtd->in_eldecl;
5740 newDtd->scaffold = oldDtd->scaffold;
5741 newDtd->contentStringLen = oldDtd->contentStringLen;
5742 newDtd->scaffSize = oldDtd->scaffSize;
5743 newDtd->scaffLevel = oldDtd->scaffLevel;
5744 newDtd->scaffIndex = oldDtd->scaffIndex;
5746 return 1;
5747 } /* End dtdCopy */
5749 static int
5750 copyEntityTable(HASH_TABLE *newTable,
5751 STRING_POOL *newPool,
5752 const HASH_TABLE *oldTable)
5754 HASH_TABLE_ITER iter;
5755 const XML_Char *cachedOldBase = NULL;
5756 const XML_Char *cachedNewBase = NULL;
5758 hashTableIterInit(&iter, oldTable);
5760 for (;;) {
5761 ENTITY *newE;
5762 const XML_Char *name;
5763 const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter);
5764 if (!oldE)
5765 break;
5766 name = poolCopyString(newPool, oldE->name);
5767 if (!name)
5768 return 0;
5769 newE = (ENTITY *)lookup(newTable, name, sizeof(ENTITY));
5770 if (!newE)
5771 return 0;
5772 if (oldE->systemId) {
5773 const XML_Char *tem = poolCopyString(newPool, oldE->systemId);
5774 if (!tem)
5775 return 0;
5776 newE->systemId = tem;
5777 if (oldE->base) {
5778 if (oldE->base == cachedOldBase)
5779 newE->base = cachedNewBase;
5780 else {
5781 cachedOldBase = oldE->base;
5782 tem = poolCopyString(newPool, cachedOldBase);
5783 if (!tem)
5784 return 0;
5785 cachedNewBase = newE->base = tem;
5788 if (oldE->publicId) {
5789 tem = poolCopyString(newPool, oldE->publicId);
5790 if (!tem)
5791 return 0;
5792 newE->publicId = tem;
5795 else {
5796 const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr,
5797 oldE->textLen);
5798 if (!tem)
5799 return 0;
5800 newE->textPtr = tem;
5801 newE->textLen = oldE->textLen;
5803 if (oldE->notation) {
5804 const XML_Char *tem = poolCopyString(newPool, oldE->notation);
5805 if (!tem)
5806 return 0;
5807 newE->notation = tem;
5809 newE->is_param = oldE->is_param;
5810 newE->is_internal = oldE->is_internal;
5812 return 1;
5815 #define INIT_POWER 6
5817 static XML_Bool FASTCALL
5818 keyeq(KEY s1, KEY s2)
5820 for (; *s1 == *s2; s1++, s2++)
5821 if (*s1 == 0)
5822 return XML_TRUE;
5823 return XML_FALSE;
5826 static unsigned long FASTCALL
5827 hash(KEY s)
5829 unsigned long h = 0;
5830 while (*s)
5831 h = CHAR_HASH(h, *s++);
5832 return h;
5835 static NAMED *
5836 lookup(HASH_TABLE *table, KEY name, size_t createSize)
5838 size_t i;
5839 if (table->size == 0) {
5840 size_t tsize;
5841 if (!createSize)
5842 return NULL;
5843 table->power = INIT_POWER;
5844 /* table->size is a power of 2 */
5845 table->size = (size_t)1 << INIT_POWER;
5846 tsize = table->size * sizeof(NAMED *);
5847 table->v = (NAMED **)table->mem->malloc_fcn(tsize);
5848 if (!table->v) {
5849 table->size = 0;
5850 return NULL;
5852 memset(table->v, 0, tsize);
5853 i = hash(name) & ((unsigned long)table->size - 1);
5855 else {
5856 unsigned long h = hash(name);
5857 unsigned long mask = (unsigned long)table->size - 1;
5858 unsigned char step = 0;
5859 i = h & mask;
5860 while (table->v[i]) {
5861 if (keyeq(name, table->v[i]->name))
5862 return table->v[i];
5863 if (!step)
5864 step = PROBE_STEP(h, mask, table->power);
5865 i < step ? (i += table->size - step) : (i -= step);
5867 if (!createSize)
5868 return NULL;
5870 /* check for overflow (table is half full) */
5871 if (table->used >> (table->power - 1)) {
5872 unsigned char newPower = table->power + 1;
5873 size_t newSize = (size_t)1 << newPower;
5874 unsigned long newMask = (unsigned long)newSize - 1;
5875 size_t tsize = newSize * sizeof(NAMED *);
5876 NAMED **newV = (NAMED **)table->mem->malloc_fcn(tsize);
5877 if (!newV)
5878 return NULL;
5879 memset(newV, 0, tsize);
5880 for (i = 0; i < table->size; i++)
5881 if (table->v[i]) {
5882 unsigned long newHash = hash(table->v[i]->name);
5883 size_t j = newHash & newMask;
5884 step = 0;
5885 while (newV[j]) {
5886 if (!step)
5887 step = PROBE_STEP(newHash, newMask, newPower);
5888 j < step ? (j += newSize - step) : (j -= step);
5890 newV[j] = table->v[i];
5892 table->mem->free_fcn(table->v);
5893 table->v = newV;
5894 table->power = newPower;
5895 table->size = newSize;
5896 i = h & newMask;
5897 step = 0;
5898 while (table->v[i]) {
5899 if (!step)
5900 step = PROBE_STEP(h, newMask, newPower);
5901 i < step ? (i += newSize - step) : (i -= step);
5905 table->v[i] = (NAMED *)table->mem->malloc_fcn(createSize);
5906 if (!table->v[i])
5907 return NULL;
5908 memset(table->v[i], 0, createSize);
5909 table->v[i]->name = name;
5910 (table->used)++;
5911 return table->v[i];
5914 static void FASTCALL
5915 hashTableClear(HASH_TABLE *table)
5917 size_t i;
5918 for (i = 0; i < table->size; i++) {
5919 table->mem->free_fcn(table->v[i]);
5920 table->v[i] = NULL;
5922 table->used = 0;
5925 static void FASTCALL
5926 hashTableDestroy(HASH_TABLE *table)
5928 size_t i;
5929 for (i = 0; i < table->size; i++)
5930 table->mem->free_fcn(table->v[i]);
5931 table->mem->free_fcn(table->v);
5934 static void FASTCALL
5935 hashTableInit(HASH_TABLE *p, const XML_Memory_Handling_Suite *ms)
5937 p->power = 0;
5938 p->size = 0;
5939 p->used = 0;
5940 p->v = NULL;
5941 p->mem = ms;
5944 static void FASTCALL
5945 hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table)
5947 iter->p = table->v;
5948 iter->end = iter->p + table->size;
5951 static NAMED * FASTCALL
5952 hashTableIterNext(HASH_TABLE_ITER *iter)
5954 while (iter->p != iter->end) {
5955 NAMED *tem = *(iter->p)++;
5956 if (tem)
5957 return tem;
5959 return NULL;
5962 static void FASTCALL
5963 poolInit(STRING_POOL *pool, const XML_Memory_Handling_Suite *ms)
5965 pool->blocks = NULL;
5966 pool->freeBlocks = NULL;
5967 pool->start = NULL;
5968 pool->ptr = NULL;
5969 pool->end = NULL;
5970 pool->mem = ms;
5973 static void FASTCALL
5974 poolClear(STRING_POOL *pool)
5976 if (!pool->freeBlocks)
5977 pool->freeBlocks = pool->blocks;
5978 else {
5979 BLOCK *p = pool->blocks;
5980 while (p) {
5981 BLOCK *tem = p->next;
5982 p->next = pool->freeBlocks;
5983 pool->freeBlocks = p;
5984 p = tem;
5987 pool->blocks = NULL;
5988 pool->start = NULL;
5989 pool->ptr = NULL;
5990 pool->end = NULL;
5993 static void FASTCALL
5994 poolDestroy(STRING_POOL *pool)
5996 BLOCK *p = pool->blocks;
5997 while (p) {
5998 BLOCK *tem = p->next;
5999 pool->mem->free_fcn(p);
6000 p = tem;
6002 p = pool->freeBlocks;
6003 while (p) {
6004 BLOCK *tem = p->next;
6005 pool->mem->free_fcn(p);
6006 p = tem;
6010 static XML_Char *
6011 poolAppend(STRING_POOL *pool, const ENCODING *enc,
6012 const char *ptr, const char *end)
6014 if (!pool->ptr && !poolGrow(pool))
6015 return NULL;
6016 for (;;) {
6017 XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end);
6018 if (ptr == end)
6019 break;
6020 if (!poolGrow(pool))
6021 return NULL;
6023 return pool->start;
6026 static const XML_Char * FASTCALL
6027 poolCopyString(STRING_POOL *pool, const XML_Char *s)
6029 do {
6030 if (!poolAppendChar(pool, *s))
6031 return NULL;
6032 } while (*s++);
6033 s = pool->start;
6034 poolFinish(pool);
6035 return s;
6038 static const XML_Char *
6039 poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n)
6041 if (!pool->ptr && !poolGrow(pool))
6042 return NULL;
6043 for (; n > 0; --n, s++) {
6044 if (!poolAppendChar(pool, *s))
6045 return NULL;
6047 s = pool->start;
6048 poolFinish(pool);
6049 return s;
6052 static const XML_Char * FASTCALL
6053 poolAppendString(STRING_POOL *pool, const XML_Char *s)
6055 while (*s) {
6056 if (!poolAppendChar(pool, *s))
6057 return NULL;
6058 s++;
6060 return pool->start;
6063 static XML_Char *
6064 poolStoreString(STRING_POOL *pool, const ENCODING *enc,
6065 const char *ptr, const char *end)
6067 if (!poolAppend(pool, enc, ptr, end))
6068 return NULL;
6069 if (pool->ptr == pool->end && !poolGrow(pool))
6070 return NULL;
6071 *(pool->ptr)++ = 0;
6072 return pool->start;
6075 static XML_Bool FASTCALL
6076 poolGrow(STRING_POOL *pool)
6078 if (pool->freeBlocks) {
6079 if (pool->start == 0) {
6080 pool->blocks = pool->freeBlocks;
6081 pool->freeBlocks = pool->freeBlocks->next;
6082 pool->blocks->next = NULL;
6083 pool->start = pool->blocks->s;
6084 pool->end = pool->start + pool->blocks->size;
6085 pool->ptr = pool->start;
6086 return XML_TRUE;
6088 if (pool->end - pool->start < pool->freeBlocks->size) {
6089 BLOCK *tem = pool->freeBlocks->next;
6090 pool->freeBlocks->next = pool->blocks;
6091 pool->blocks = pool->freeBlocks;
6092 pool->freeBlocks = tem;
6093 memcpy(pool->blocks->s, pool->start,
6094 (pool->end - pool->start) * sizeof(XML_Char));
6095 pool->ptr = pool->blocks->s + (pool->ptr - pool->start);
6096 pool->start = pool->blocks->s;
6097 pool->end = pool->start + pool->blocks->size;
6098 return XML_TRUE;
6101 if (pool->blocks && pool->start == pool->blocks->s) {
6102 int blockSize = (int)(pool->end - pool->start)*2;
6103 pool->blocks = (BLOCK *)
6104 pool->mem->realloc_fcn(pool->blocks,
6105 (offsetof(BLOCK, s)
6106 + blockSize * sizeof(XML_Char)));
6107 if (pool->blocks == NULL)
6108 return XML_FALSE;
6109 pool->blocks->size = blockSize;
6110 pool->ptr = pool->blocks->s + (pool->ptr - pool->start);
6111 pool->start = pool->blocks->s;
6112 pool->end = pool->start + blockSize;
6114 else {
6115 BLOCK *tem;
6116 int blockSize = (int)(pool->end - pool->start);
6117 if (blockSize < INIT_BLOCK_SIZE)
6118 blockSize = INIT_BLOCK_SIZE;
6119 else
6120 blockSize *= 2;
6121 tem = (BLOCK *)pool->mem->malloc_fcn(offsetof(BLOCK, s)
6122 + blockSize * sizeof(XML_Char));
6123 if (!tem)
6124 return XML_FALSE;
6125 tem->size = blockSize;
6126 tem->next = pool->blocks;
6127 pool->blocks = tem;
6128 if (pool->ptr != pool->start)
6129 memcpy(tem->s, pool->start,
6130 (pool->ptr - pool->start) * sizeof(XML_Char));
6131 pool->ptr = tem->s + (pool->ptr - pool->start);
6132 pool->start = tem->s;
6133 pool->end = tem->s + blockSize;
6135 return XML_TRUE;
6138 static int FASTCALL
6139 nextScaffoldPart(XML_Parser parser)
6141 DTD * const dtd = _dtd; /* save one level of indirection */
6142 CONTENT_SCAFFOLD * me;
6143 int next;
6145 if (!dtd->scaffIndex) {
6146 dtd->scaffIndex = (int *)MALLOC(groupSize * sizeof(int));
6147 if (!dtd->scaffIndex)
6148 return -1;
6149 dtd->scaffIndex[0] = 0;
6152 if (dtd->scaffCount >= dtd->scaffSize) {
6153 CONTENT_SCAFFOLD *temp;
6154 if (dtd->scaffold) {
6155 temp = (CONTENT_SCAFFOLD *)
6156 REALLOC(dtd->scaffold, dtd->scaffSize * 2 * sizeof(CONTENT_SCAFFOLD));
6157 if (temp == NULL)
6158 return -1;
6159 dtd->scaffSize *= 2;
6161 else {
6162 temp = (CONTENT_SCAFFOLD *)MALLOC(INIT_SCAFFOLD_ELEMENTS
6163 * sizeof(CONTENT_SCAFFOLD));
6164 if (temp == NULL)
6165 return -1;
6166 dtd->scaffSize = INIT_SCAFFOLD_ELEMENTS;
6168 dtd->scaffold = temp;
6170 next = dtd->scaffCount++;
6171 me = &dtd->scaffold[next];
6172 if (dtd->scaffLevel) {
6173 CONTENT_SCAFFOLD *parent = &dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel-1]];
6174 if (parent->lastchild) {
6175 dtd->scaffold[parent->lastchild].nextsib = next;
6177 if (!parent->childcnt)
6178 parent->firstchild = next;
6179 parent->lastchild = next;
6180 parent->childcnt++;
6182 me->firstchild = me->lastchild = me->childcnt = me->nextsib = 0;
6183 return next;
6186 static void
6187 build_node(XML_Parser parser,
6188 int src_node,
6189 XML_Content *dest,
6190 XML_Content **contpos,
6191 XML_Char **strpos)
6193 DTD * const dtd = _dtd; /* save one level of indirection */
6194 dest->type = dtd->scaffold[src_node].type;
6195 dest->quant = dtd->scaffold[src_node].quant;
6196 if (dest->type == XML_CTYPE_NAME) {
6197 const XML_Char *src;
6198 dest->name = *strpos;
6199 src = dtd->scaffold[src_node].name;
6200 for (;;) {
6201 *(*strpos)++ = *src;
6202 if (!*src)
6203 break;
6204 src++;
6206 dest->numchildren = 0;
6207 dest->children = NULL;
6209 else {
6210 unsigned int i;
6211 int cn;
6212 dest->numchildren = dtd->scaffold[src_node].childcnt;
6213 dest->children = *contpos;
6214 *contpos += dest->numchildren;
6215 for (i = 0, cn = dtd->scaffold[src_node].firstchild;
6216 i < dest->numchildren;
6217 i++, cn = dtd->scaffold[cn].nextsib) {
6218 build_node(parser, cn, &(dest->children[i]), contpos, strpos);
6220 dest->name = NULL;
6224 static XML_Content *
6225 build_model (XML_Parser parser)
6227 DTD * const dtd = _dtd; /* save one level of indirection */
6228 XML_Content *ret;
6229 XML_Content *cpos;
6230 XML_Char * str;
6231 int allocsize = (dtd->scaffCount * sizeof(XML_Content)
6232 + (dtd->contentStringLen * sizeof(XML_Char)));
6234 ret = (XML_Content *)MALLOC(allocsize);
6235 if (!ret)
6236 return NULL;
6238 str = (XML_Char *) (&ret[dtd->scaffCount]);
6239 cpos = &ret[1];
6241 build_node(parser, 0, ret, &cpos, &str);
6242 return ret;
6245 static ELEMENT_TYPE *
6246 getElementType(XML_Parser parser,
6247 const ENCODING *enc,
6248 const char *ptr,
6249 const char *end)
6251 DTD * const dtd = _dtd; /* save one level of indirection */
6252 const XML_Char *name = poolStoreString(&dtd->pool, enc, ptr, end);
6253 ELEMENT_TYPE *ret;
6255 if (!name)
6256 return NULL;
6257 ret = (ELEMENT_TYPE *) lookup(&dtd->elementTypes, name, sizeof(ELEMENT_TYPE));
6258 if (!ret)
6259 return NULL;
6260 if (ret->name != name)
6261 poolDiscard(&dtd->pool);
6262 else {
6263 poolFinish(&dtd->pool);
6264 if (!setElementTypePrefix(parser, ret))
6265 return NULL;
6267 return ret;