source: python/trunk/Modules/expat/xmlparse.c@ 10

Last change on this file since 10 was 2, checked in by Yuri Dario, 15 years ago

Initial import for vendor code.

  • Property svn:eol-style set to native
File size: 188.1 KB
Line 
1/* Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd
2 See the file COPYING for copying permission.
3*/
4
5#define XML_BUILDING_EXPAT 1
6
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 */
16
17#include <stddef.h>
18#include <string.h> /* memset(), memcpy() */
19#include <assert.h>
20
21#include "expat.h"
22
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))
30typedef 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)
38typedef char ICHAR;
39#endif
40
41
42#ifndef XML_NS
43
44#define XmlInitEncodingNS XmlInitEncoding
45#define XmlInitUnknownEncodingNS XmlInitUnknownEncoding
46#undef XmlGetInternalEncodingNS
47#define XmlGetInternalEncodingNS XmlGetInternalEncoding
48#define XmlParseXmlDeclNS XmlParseXmlDecl
49
50#endif
51
52#ifdef XML_UNICODE
53
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
61
62#else
63
64#define XML_T(x) x
65#define XML_L(x) x
66
67#endif
68
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))
71
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 */
80
81#include "internal.h"
82#include "xmltok.h"
83#include "xmlrole.h"
84
85typedef const XML_Char *KEY;
86
87typedef struct {
88 KEY name;
89} NAMED;
90
91typedef 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;
98
99/* Basic character hash algorithm, taken from Python's string hash:
100 h = h * 1000003 ^ character, the constant being a prime number.
101
102*/
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
110
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.
118*/
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))
123
124typedef struct {
125 NAMED **p;
126 NAMED **end;
127} HASH_TABLE_ITER;
128
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
135
136#define EXPAND_SPARE 24
137
138typedef 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;
147
148typedef struct prefix {
149 const XML_Char *name;
150 BINDING *binding;
151} PREFIX;
152
153typedef 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;
161
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.
171
172 A parser re-uses these structures, maintaining a list of allocated
173 TAG objects in a free list.
174*/
175typedef 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;
184
185typedef 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;
198
199typedef 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;
208
209#define INIT_SCAFFOLD_ELEMENTS 32
210
211typedef struct block {
212 struct block *next;
213 int size;
214 XML_Char s[1];
215} BLOCK;
216
217typedef 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;
225
226/* The XML_Char before the name is used to determine whether
227 an attribute has been specified. */
228typedef struct attribute_id {
229 XML_Char *name;
230 PREFIX *prefix;
231 XML_Bool maybeTokenized;
232 XML_Bool xmlns;
233} ATTRIBUTE_ID;
234
235typedef struct {
236 const ATTRIBUTE_ID *id;
237 XML_Bool isCdata;
238 const XML_Char *value;
239} DEFAULT_ATTRIBUTE;
240
241typedef struct {
242 unsigned long version;
243 unsigned long hash;
244 const XML_Char *uriName;
245} NS_ATT;
246
247typedef 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;
255
256typedef 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;
284
285typedef 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;
293
294typedef enum XML_Error PTRCALL Processor(XML_Parser parser,
295 const char *start,
296 const char *end,
297 const char **endPtr);
298
299static Processor prologProcessor;
300static Processor prologInitProcessor;
301static Processor contentProcessor;
302static Processor cdataSectionProcessor;
303#ifdef XML_DTD
304static Processor ignoreSectionProcessor;
305static Processor externalParEntProcessor;
306static Processor externalParEntInitProcessor;
307static Processor entityValueProcessor;
308static Processor entityValueInitProcessor;
309#endif /* XML_DTD */
310static Processor epilogProcessor;
311static Processor errorProcessor;
312static Processor externalEntityInitProcessor;
313static Processor externalEntityInitProcessor2;
314static Processor externalEntityInitProcessor3;
315static Processor externalEntityContentProcessor;
316static Processor internalEntityProcessor;
317
318static enum XML_Error
319handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName);
320static enum XML_Error
321processXmlDecl(XML_Parser parser, int isGeneralTextEntity,
322 const char *s, const char *next);
323static enum XML_Error
324initializeEncoding(XML_Parser parser);
325static enum XML_Error
326doProlog(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);
329static enum XML_Error
330processInternalEntity(XML_Parser parser, ENTITY *entity,
331 XML_Bool betweenDecl);
332static enum XML_Error
333doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc,
334 const char *start, const char *end, const char **endPtr,
335 XML_Bool haveMore);
336static enum XML_Error
337doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr,
338 const char *end, const char **nextPtr, XML_Bool haveMore);
339#ifdef XML_DTD
340static enum XML_Error
341doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr,
342 const char *end, const char **nextPtr, XML_Bool haveMore);
343#endif /* XML_DTD */
344
345static enum XML_Error
346storeAtts(XML_Parser parser, const ENCODING *, const char *s,
347 TAG_NAME *tagNamePtr, BINDING **bindingsPtr);
348static enum XML_Error
349addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId,
350 const XML_Char *uri, BINDING **bindingsPtr);
351static int
352defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, XML_Bool isCdata,
353 XML_Bool isId, const XML_Char *dfltValue, XML_Parser parser);
354static enum XML_Error
355storeAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata,
356 const char *, const char *, STRING_POOL *);
357static enum XML_Error
358appendAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata,
359 const char *, const char *, STRING_POOL *);
360static ATTRIBUTE_ID *
361getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start,
362 const char *end);
363static int
364setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *);
365static enum XML_Error
366storeEntityValue(XML_Parser parser, const ENCODING *enc, const char *start,
367 const char *end);
368static int
369reportProcessingInstruction(XML_Parser parser, const ENCODING *enc,
370 const char *start, const char *end);
371static int
372reportComment(XML_Parser parser, const ENCODING *enc, const char *start,
373 const char *end);
374static void
375reportDefault(XML_Parser parser, const ENCODING *enc, const char *start,
376 const char *end);
377
378static const XML_Char * getContext(XML_Parser parser);
379static XML_Bool
380setContext(XML_Parser parser, const XML_Char *context);
381
382static void FASTCALL normalizePublicId(XML_Char *s);
383
384static DTD * dtdCreate(const XML_Memory_Handling_Suite *ms);
385/* do not call if parentParser != NULL */
386static void dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms);
387static void
388dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms);
389static int
390dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms);
391static int
392copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *);
393
394static NAMED *
395lookup(HASH_TABLE *table, KEY name, size_t createSize);
396static void FASTCALL
397hashTableInit(HASH_TABLE *, const XML_Memory_Handling_Suite *ms);
398static void FASTCALL hashTableClear(HASH_TABLE *);
399static void FASTCALL hashTableDestroy(HASH_TABLE *);
400static void FASTCALL
401hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *);
402static NAMED * FASTCALL hashTableIterNext(HASH_TABLE_ITER *);
403
404static void FASTCALL
405poolInit(STRING_POOL *, const XML_Memory_Handling_Suite *ms);
406static void FASTCALL poolClear(STRING_POOL *);
407static void FASTCALL poolDestroy(STRING_POOL *);
408static XML_Char *
409poolAppend(STRING_POOL *pool, const ENCODING *enc,
410 const char *ptr, const char *end);
411static XML_Char *
412poolStoreString(STRING_POOL *pool, const ENCODING *enc,
413 const char *ptr, const char *end);
414static XML_Bool FASTCALL poolGrow(STRING_POOL *pool);
415static const XML_Char * FASTCALL
416poolCopyString(STRING_POOL *pool, const XML_Char *s);
417static const XML_Char *
418poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n);
419static const XML_Char * FASTCALL
420poolAppendString(STRING_POOL *pool, const XML_Char *s);
421
422static int FASTCALL nextScaffoldPart(XML_Parser parser);
423static XML_Content * build_model(XML_Parser parser);
424static ELEMENT_TYPE *
425getElementType(XML_Parser parser, const ENCODING *enc,
426 const char *ptr, const char *end);
427
428static XML_Parser
429parserCreate(const XML_Char *encodingName,
430 const XML_Memory_Handling_Suite *memsuite,
431 const XML_Char *nameSep,
432 DTD *dtd);
433static void
434parserInit(XML_Parser parser, const XML_Char *encodingName);
435
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))
447
448struct 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
545};
546
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)))
550
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 */
652
653XML_Parser XMLCALL
654XML_ParserCreate(const XML_Char *encodingName)
655{
656 return XML_ParserCreate_MM(encodingName, NULL, NULL);
657}
658
659XML_Parser XMLCALL
660XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep)
661{
662 XML_Char tmp[2];
663 *tmp = nsSep;
664 return XML_ParserCreate_MM(encodingName, NULL, tmp);
665}
666
667static 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'
672};
673
674XML_Parser XMLCALL
675XML_ParserCreate_MM(const XML_Char *encodingName,
676 const XML_Memory_Handling_Suite *memsuite,
677 const XML_Char *nameSep)
678{
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
683 */
684 if (!setContext(parser, implicitContext)) {
685 XML_ParserFree(parser);
686 return NULL;
687 }
688 }
689 return parser;
690}
691
692static XML_Parser
693parserCreate(const XML_Char *encodingName,
694 const XML_Memory_Handling_Suite *memsuite,
695 const XML_Char *nameSep,
696 DTD *dtd)
697{
698 XML_Parser parser;
699
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;
709 }
710 }
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;
719 }
720 }
721
722 if (!parser)
723 return parser;
724
725 buffer = NULL;
726 bufferLim = NULL;
727
728 attsSize = INIT_ATTS_SIZE;
729 atts = (ATTRIBUTE *)MALLOC(attsSize * sizeof(ATTRIBUTE));
730 if (atts == NULL) {
731 FREE(parser);
732 return NULL;
733 }
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;
739 }
740 dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE;
741
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;
751 }
752 }
753
754 freeBindingList = NULL;
755 freeTagList = NULL;
756 freeInternalEntities = NULL;
757
758 groupSize = 0;
759 groupConnector = NULL;
760
761 unknownEncodingHandler = NULL;
762 unknownEncodingHandlerData = NULL;
763
764 namespaceSeparator = '!';
765 ns = XML_FALSE;
766 ns_triplets = XML_FALSE;
767
768 nsAtts = NULL;
769 nsAttsVersion = 0;
770 nsAttsPower = 0;
771
772 poolInit(&tempPool, &(parser->m_mem));
773 poolInit(&temp2Pool, &(parser->m_mem));
774 parserInit(parser, encodingName);
775
776 if (encodingName && !protocolEncodingName) {
777 XML_ParserFree(parser);
778 return NULL;
779 }
780
781 if (nameSep) {
782 ns = XML_TRUE;
783 internalEncoding = XmlGetInternalEncodingNS();
784 namespaceSeparator = *nameSep;
785 }
786 else {
787 internalEncoding = XmlGetInternalEncoding();
788 }
789
790 return parser;
791}
792
793static void
794parserInit(XML_Parser parser, const XML_Char *encodingName)
795{
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
863}
864
865/* moves list of bindings to freeBindingList */
866static void FASTCALL
867moveToFreeBindingList(XML_Parser parser, BINDING *bindings)
868{
869 while (bindings) {
870 BINDING *b = bindings;
871 bindings = bindings->nextTagBinding;
872 b->nextTagBinding = freeBindingList;
873 freeBindingList = b;
874 }
875}
876
877XML_Bool XMLCALL
878XML_ParserReset(XML_Parser parser, const XML_Char *encodingName)
879{
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;
893 }
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;
901 }
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);
911}
912
913enum XML_Status XMLCALL
914XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName)
915{
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.
919 */
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;
928 }
929 return XML_STATUS_OK;
930}
931
932XML_Parser XMLCALL
933XML_ExternalEntityParserCreate(XML_Parser oldParser,
934 const XML_Char *context,
935 const XML_Char *encodingName)
936{
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;
969
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;
979
980#ifdef XML_DTD
981 if (!context)
982 newDtd = oldDtd;
983#endif /* XML_DTD */
984
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.
989 */
990 if (ns) {
991 XML_Char tmp[2];
992 *tmp = namespaceSeparator;
993 parser = parserCreate(encodingName, &parser->m_mem, tmp, newDtd);
994 }
995 else {
996 parser = parserCreate(encodingName, &parser->m_mem, NULL, newDtd);
997 }
998
999 if (!parser)
1000 return NULL;
1001
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;
1042 }
1043 processor = externalEntityInitProcessor;
1044#ifdef XML_DTD
1045 }
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.
1053 */
1054 isParamEntity = XML_TRUE;
1055 XmlPrologStateInitExternalEntity(&prologState);
1056 processor = externalParEntInitProcessor;
1057 }
1058#endif /* XML_DTD */
1059 return parser;
1060}
1061
1062static void FASTCALL
1063destroyBindings(BINDING *bindings, XML_Parser parser)
1064{
1065 for (;;) {
1066 BINDING *b = bindings;
1067 if (!b)
1068 break;
1069 bindings = b->nextTagBinding;
1070 FREE(b->uri);
1071 FREE(b);
1072 }
1073}
1074
1075void XMLCALL
1076XML_ParserFree(XML_Parser parser)
1077{
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;
1091 }
1092 p = tagList;
1093 tagList = tagList->parent;
1094 FREE(p->buf);
1095 destroyBindings(p->bindings, parser);
1096 FREE(p);
1097 }
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;
1107 }
1108 openEntity = entityList;
1109 entityList = entityList->next;
1110 FREE(openEntity);
1111 }
1112
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
1120 */
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);
1135}
1136
1137void XMLCALL
1138XML_UseParserAsHandlerArg(XML_Parser parser)
1139{
1140 handlerArg = parser;
1141}
1142
1143enum XML_Error XMLCALL
1144XML_UseForeignDTD(XML_Parser parser, XML_Bool useDTD)
1145{
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
1155}
1156
1157void XMLCALL
1158XML_SetReturnNSTriplet(XML_Parser parser, int do_nst)
1159{
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;
1164}
1165
1166void XMLCALL
1167XML_SetUserData(XML_Parser parser, void *p)
1168{
1169 if (handlerArg == userData)
1170 handlerArg = userData = p;
1171 else
1172 userData = p;
1173}
1174
1175enum XML_Status XMLCALL
1176XML_SetBase(XML_Parser parser, const XML_Char *p)
1177{
1178 if (p) {
1179 p = poolCopyString(&_dtd->pool, p);
1180 if (!p)
1181 return XML_STATUS_ERROR;
1182 curBase = p;
1183 }
1184 else
1185 curBase = NULL;
1186 return XML_STATUS_OK;
1187}
1188
1189const XML_Char * XMLCALL
1190XML_GetBase(XML_Parser parser)
1191{
1192 return curBase;
1193}
1194
1195int XMLCALL
1196XML_GetSpecifiedAttributeCount(XML_Parser parser)
1197{
1198 return nSpecifiedAtts;
1199}
1200
1201int XMLCALL
1202XML_GetIdAttributeIndex(XML_Parser parser)
1203{
1204 return idAttIndex;
1205}
1206
1207void XMLCALL
1208XML_SetElementHandler(XML_Parser parser,
1209 XML_StartElementHandler start,
1210 XML_EndElementHandler end)
1211{
1212 startElementHandler = start;
1213 endElementHandler = end;
1214}
1215
1216void XMLCALL
1217XML_SetStartElementHandler(XML_Parser parser,
1218 XML_StartElementHandler start) {
1219 startElementHandler = start;
1220}
1221
1222void XMLCALL
1223XML_SetEndElementHandler(XML_Parser parser,
1224 XML_EndElementHandler end) {
1225 endElementHandler = end;
1226}
1227
1228void XMLCALL
1229XML_SetCharacterDataHandler(XML_Parser parser,
1230 XML_CharacterDataHandler handler)
1231{
1232 characterDataHandler = handler;
1233}
1234
1235void XMLCALL
1236XML_SetProcessingInstructionHandler(XML_Parser parser,
1237 XML_ProcessingInstructionHandler handler)
1238{
1239 processingInstructionHandler = handler;
1240}
1241
1242void XMLCALL
1243XML_SetCommentHandler(XML_Parser parser,
1244 XML_CommentHandler handler)
1245{
1246 commentHandler = handler;
1247}
1248
1249void XMLCALL
1250XML_SetCdataSectionHandler(XML_Parser parser,
1251 XML_StartCdataSectionHandler start,
1252 XML_EndCdataSectionHandler end)
1253{
1254 startCdataSectionHandler = start;
1255 endCdataSectionHandler = end;
1256}
1257
1258void XMLCALL
1259XML_SetStartCdataSectionHandler(XML_Parser parser,
1260 XML_StartCdataSectionHandler start) {
1261 startCdataSectionHandler = start;
1262}
1263
1264void XMLCALL
1265XML_SetEndCdataSectionHandler(XML_Parser parser,
1266 XML_EndCdataSectionHandler end) {
1267 endCdataSectionHandler = end;
1268}
1269
1270void XMLCALL
1271XML_SetDefaultHandler(XML_Parser parser,
1272 XML_DefaultHandler handler)
1273{
1274 defaultHandler = handler;
1275 defaultExpandInternalEntities = XML_FALSE;
1276}
1277
1278void XMLCALL
1279XML_SetDefaultHandlerExpand(XML_Parser parser,
1280 XML_DefaultHandler handler)
1281{
1282 defaultHandler = handler;
1283 defaultExpandInternalEntities = XML_TRUE;
1284}
1285
1286void XMLCALL
1287XML_SetDoctypeDeclHandler(XML_Parser parser,
1288 XML_StartDoctypeDeclHandler start,
1289 XML_EndDoctypeDeclHandler end)
1290{
1291 startDoctypeDeclHandler = start;
1292 endDoctypeDeclHandler = end;
1293}
1294
1295void XMLCALL
1296XML_SetStartDoctypeDeclHandler(XML_Parser parser,
1297 XML_StartDoctypeDeclHandler start) {
1298 startDoctypeDeclHandler = start;
1299}
1300
1301void XMLCALL
1302XML_SetEndDoctypeDeclHandler(XML_Parser parser,
1303 XML_EndDoctypeDeclHandler end) {
1304 endDoctypeDeclHandler = end;
1305}
1306
1307void XMLCALL
1308XML_SetUnparsedEntityDeclHandler(XML_Parser parser,
1309 XML_UnparsedEntityDeclHandler handler)
1310{
1311 unparsedEntityDeclHandler = handler;
1312}
1313
1314void XMLCALL
1315XML_SetNotationDeclHandler(XML_Parser parser,
1316 XML_NotationDeclHandler handler)
1317{
1318 notationDeclHandler = handler;
1319}
1320
1321void XMLCALL
1322XML_SetNamespaceDeclHandler(XML_Parser parser,
1323 XML_StartNamespaceDeclHandler start,
1324 XML_EndNamespaceDeclHandler end)
1325{
1326 startNamespaceDeclHandler = start;
1327 endNamespaceDeclHandler = end;
1328}
1329
1330void XMLCALL
1331XML_SetStartNamespaceDeclHandler(XML_Parser parser,
1332 XML_StartNamespaceDeclHandler start) {
1333 startNamespaceDeclHandler = start;
1334}
1335
1336void XMLCALL
1337XML_SetEndNamespaceDeclHandler(XML_Parser parser,
1338 XML_EndNamespaceDeclHandler end) {
1339 endNamespaceDeclHandler = end;
1340}
1341
1342void XMLCALL
1343XML_SetNotStandaloneHandler(XML_Parser parser,
1344 XML_NotStandaloneHandler handler)
1345{
1346 notStandaloneHandler = handler;
1347}
1348
1349void XMLCALL
1350XML_SetExternalEntityRefHandler(XML_Parser parser,
1351 XML_ExternalEntityRefHandler handler)
1352{
1353 externalEntityRefHandler = handler;
1354}
1355
1356void XMLCALL
1357XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg)
1358{
1359 if (arg)
1360 externalEntityRefHandlerArg = (XML_Parser)arg;
1361 else
1362 externalEntityRefHandlerArg = parser;
1363}
1364
1365void XMLCALL
1366XML_SetSkippedEntityHandler(XML_Parser parser,
1367 XML_SkippedEntityHandler handler)
1368{
1369 skippedEntityHandler = handler;
1370}
1371
1372void XMLCALL
1373XML_SetUnknownEncodingHandler(XML_Parser parser,
1374 XML_UnknownEncodingHandler handler,
1375 void *data)
1376{
1377 unknownEncodingHandler = handler;
1378 unknownEncodingHandlerData = data;
1379}
1380
1381void XMLCALL
1382XML_SetElementDeclHandler(XML_Parser parser,
1383 XML_ElementDeclHandler eldecl)
1384{
1385 elementDeclHandler = eldecl;
1386}
1387
1388void XMLCALL
1389XML_SetAttlistDeclHandler(XML_Parser parser,
1390 XML_AttlistDeclHandler attdecl)
1391{
1392 attlistDeclHandler = attdecl;
1393}
1394
1395void XMLCALL
1396XML_SetEntityDeclHandler(XML_Parser parser,
1397 XML_EntityDeclHandler handler)
1398{
1399 entityDeclHandler = handler;
1400}
1401
1402void XMLCALL
1403XML_SetXmlDeclHandler(XML_Parser parser,
1404 XML_XmlDeclHandler handler) {
1405 xmlDeclHandler = handler;
1406}
1407
1408int XMLCALL
1409XML_SetParamEntityParsing(XML_Parser parser,
1410 enum XML_ParamEntityParsing peParsing)
1411{
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
1421}
1422
1423enum XML_Status XMLCALL
1424XML_Parse(XML_Parser parser, const char *s, int len, int isFinal)
1425{
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;
1435 }
1436
1437 if (len == 0) {
1438 ps_finalBuffer = (XML_Bool)isFinal;
1439 if (!isFinal)
1440 return XML_STATUS_OK;
1441 positionPtr = bufferPtr;
1442 parseEndPtr = bufferEnd;
1443
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.
1447 */
1448 errorCode = processor(parser, bufferPtr, parseEndPtr, &bufferPtr);
1449
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;
1462 }
1463 }
1464 eventEndPtr = eventPtr;
1465 processor = errorProcessor;
1466 return XML_STATUS_ERROR;
1467 }
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;
1476
1477 errorCode = processor(parser, s, parseEndPtr = s + len, &end);
1478
1479 if (errorCode != XML_ERROR_NONE) {
1480 eventEndPtr = eventPtr;
1481 processor = errorProcessor;
1482 return XML_STATUS_ERROR;
1483 }
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;
1495 }
1496 }
1497 }
1498
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;
1511 }
1512 buffer = temp;
1513 if (!buffer) {
1514 errorCode = XML_ERROR_NO_MEMORY;
1515 eventPtr = eventEndPtr = NULL;
1516 processor = errorProcessor;
1517 return XML_STATUS_ERROR;
1518 }
1519 bufferLim = buffer + len * 2;
1520 }
1521 memcpy(buffer, end, nLeftOver);
1522 }
1523 bufferPtr = buffer;
1524 bufferEnd = buffer + nLeftOver;
1525 positionPtr = bufferPtr;
1526 parseEndPtr = bufferEnd;
1527 eventPtr = bufferPtr;
1528 eventEndPtr = bufferPtr;
1529 return result;
1530 }
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);
1539 }
1540 }
1541}
1542
1543enum XML_Status XMLCALL
1544XML_ParseBuffer(XML_Parser parser, int len, int isFinal)
1545{
1546 const char *start;
1547 enum XML_Status result = XML_STATUS_OK;
1548
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;
1558 }
1559
1560 start = bufferPtr;
1561 positionPtr = start;
1562 bufferEnd += len;
1563 parseEndPtr = bufferEnd;
1564 parseEndByteIndex += len;
1565 ps_finalBuffer = (XML_Bool)isFinal;
1566
1567 errorCode = processor(parser, start, parseEndPtr, &bufferPtr);
1568
1569 if (errorCode != XML_ERROR_NONE) {
1570 eventEndPtr = eventPtr;
1571 processor = errorProcessor;
1572 return XML_STATUS_ERROR;
1573 }
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;
1584 }
1585 default: ; /* should not happen */
1586 }
1587 }
1588
1589 XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
1590 positionPtr = bufferPtr;
1591 return result;
1592}
1593
1594void * XMLCALL
1595XML_GetBuffer(XML_Parser parser, int len)
1596{
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: ;
1605 }
1606
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);
1612
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;
1624 }
1625#else
1626 memmove(buffer, bufferPtr, bufferEnd - bufferPtr);
1627 bufferEnd = buffer + (bufferEnd - bufferPtr);
1628 bufferPtr = buffer;
1629#endif /* not defined XML_CONTEXT_BYTES */
1630 }
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;
1643 }
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;
1655 }
1656 else {
1657 bufferEnd = newBuf + (bufferEnd - bufferPtr);
1658 bufferPtr = buffer = newBuf;
1659 }
1660#else
1661 if (bufferPtr) {
1662 memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr);
1663 FREE(buffer);
1664 }
1665 bufferEnd = newBuf + (bufferEnd - bufferPtr);
1666 bufferPtr = buffer = newBuf;
1667#endif /* not defined XML_CONTEXT_BYTES */
1668 }
1669 }
1670 return bufferEnd;
1671}
1672
1673enum XML_Status XMLCALL
1674XML_StopParser(XML_Parser parser, XML_Bool resumable)
1675{
1676 switch (ps_parsing) {
1677 case XML_SUSPENDED:
1678 if (resumable) {
1679 errorCode = XML_ERROR_SUSPENDED;
1680 return XML_STATUS_ERROR;
1681 }
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;
1693 }
1694#endif
1695 ps_parsing = XML_SUSPENDED;
1696 }
1697 else
1698 ps_parsing = XML_FINISHED;
1699 }
1700 return XML_STATUS_OK;
1701}
1702
1703enum XML_Status XMLCALL
1704XML_ResumeParser(XML_Parser parser)
1705{
1706 enum XML_Status result = XML_STATUS_OK;
1707
1708 if (ps_parsing != XML_SUSPENDED) {
1709 errorCode = XML_ERROR_NOT_SUSPENDED;
1710 return XML_STATUS_ERROR;
1711 }
1712 ps_parsing = XML_PARSING;
1713
1714 errorCode = processor(parser, bufferPtr, parseEndPtr, &bufferPtr);
1715
1716 if (errorCode != XML_ERROR_NONE) {
1717 eventEndPtr = eventPtr;
1718 processor = errorProcessor;
1719 return XML_STATUS_ERROR;
1720 }
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;
1731 }
1732 default: ;
1733 }
1734 }
1735
1736 XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
1737 positionPtr = bufferPtr;
1738 return result;
1739}
1740
1741void XMLCALL
1742XML_GetParsingStatus(XML_Parser parser, XML_ParsingStatus *status)
1743{
1744 assert(status != NULL);
1745 *status = parser->m_parsingStatus;
1746}
1747
1748enum XML_Error XMLCALL
1749XML_GetErrorCode(XML_Parser parser)
1750{
1751 return errorCode;
1752}
1753
1754XML_Index XMLCALL
1755XML_GetCurrentByteIndex(XML_Parser parser)
1756{
1757 if (eventPtr)
1758 return parseEndByteIndex - (parseEndPtr - eventPtr);
1759 return -1;
1760}
1761
1762int XMLCALL
1763XML_GetCurrentByteCount(XML_Parser parser)
1764{
1765 if (eventEndPtr && eventPtr)
1766 return (int)(eventEndPtr - eventPtr);
1767 return 0;
1768}
1769
1770const char * XMLCALL
1771XML_GetInputContext(XML_Parser parser, int *offset, int *size)
1772{
1773#ifdef XML_CONTEXT_BYTES
1774 if (eventPtr && buffer) {
1775 *offset = (int)(eventPtr - buffer);
1776 *size = (int)(bufferEnd - buffer);
1777 return buffer;
1778 }
1779#endif /* defined XML_CONTEXT_BYTES */
1780 return (char *) 0;
1781}
1782
1783XML_Size XMLCALL
1784XML_GetCurrentLineNumber(XML_Parser parser)
1785{
1786 if (eventPtr && eventPtr >= positionPtr) {
1787 XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
1788 positionPtr = eventPtr;
1789 }
1790 return position.lineNumber + 1;
1791}
1792
1793XML_Size XMLCALL
1794XML_GetCurrentColumnNumber(XML_Parser parser)
1795{
1796 if (eventPtr && eventPtr >= positionPtr) {
1797 XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
1798 positionPtr = eventPtr;
1799 }
1800 return position.columnNumber;
1801}
1802
1803void XMLCALL
1804XML_FreeContentModel(XML_Parser parser, XML_Content *model)
1805{
1806 FREE(model);
1807}
1808
1809void * XMLCALL
1810XML_MemMalloc(XML_Parser parser, size_t size)
1811{
1812 return MALLOC(size);
1813}
1814
1815void * XMLCALL
1816XML_MemRealloc(XML_Parser parser, void *ptr, size_t size)
1817{
1818 return REALLOC(ptr, size);
1819}
1820
1821void XMLCALL
1822XML_MemFree(XML_Parser parser, void *ptr)
1823{
1824 FREE(ptr);
1825}
1826
1827void XMLCALL
1828XML_DefaultCurrent(XML_Parser parser)
1829{
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);
1838 }
1839}
1840
1841const XML_LChar * XMLCALL
1842XML_ErrorString(enum XML_Error code)
1843{
1844 static const XML_LChar* const message[] = {
1845 0,
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")
1886 };
1887 if (code > 0 && code < sizeof(message)/sizeof(message[0]))
1888 return message[code];
1889 return NULL;
1890}
1891
1892const XML_LChar * XMLCALL
1893XML_ExpatVersion(void) {
1894
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... */
1902
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)
1905
1906 return V2(XML_MAJOR_VERSION, XML_MINOR_VERSION, XML_MICRO_VERSION);
1907
1908#undef V1
1909#undef V2
1910}
1911
1912XML_Expat_Version XMLCALL
1913XML_ExpatVersionInfo(void)
1914{
1915 XML_Expat_Version version;
1916
1917 version.major = XML_MAJOR_VERSION;
1918 version.minor = XML_MINOR_VERSION;
1919 version.micro = XML_MICRO_VERSION;
1920
1921 return version;
1922}
1923
1924const XML_Feature * XMLCALL
1925XML_GetFeatureList(void)
1926{
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}
1952 };
1953
1954 return features;
1955}
1956
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.
1961*/
1962static XML_Bool
1963storeRawNames(XML_Parser parser)
1964{
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.
1974 */
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).
1979 */
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
1987 */
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
1992 */
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;
1999 }
2000 memcpy(rawNameBuf, tag->rawName, tag->rawNameLength);
2001 tag->rawName = rawNameBuf;
2002 tag = tag->parent;
2003 }
2004 return XML_TRUE;
2005}
2006
2007static enum XML_Error PTRCALL
2008contentProcessor(XML_Parser parser,
2009 const char *start,
2010 const char *end,
2011 const char **endPtr)
2012{
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;
2018 }
2019 return result;
2020}
2021
2022static enum XML_Error PTRCALL
2023externalEntityInitProcessor(XML_Parser parser,
2024 const char *start,
2025 const char *end,
2026 const char **endPtr)
2027{
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);
2033}
2034
2035static enum XML_Error PTRCALL
2036externalEntityInitProcessor2(XML_Parser parser,
2037 const char *start,
2038 const char *end,
2039 const char **endPtr)
2040{
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.
2049 */
2050 if (next == end && !ps_finalBuffer) {
2051 *endPtr = next;
2052 return XML_ERROR_NONE;
2053 }
2054 start = next;
2055 break;
2056 case XML_TOK_PARTIAL:
2057 if (!ps_finalBuffer) {
2058 *endPtr = start;
2059 return XML_ERROR_NONE;
2060 }
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;
2067 }
2068 eventPtr = start;
2069 return XML_ERROR_PARTIAL_CHAR;
2070 }
2071 processor = externalEntityInitProcessor3;
2072 return externalEntityInitProcessor3(parser, start, end, endPtr);
2073}
2074
2075static enum XML_Error PTRCALL
2076externalEntityInitProcessor3(XML_Parser parser,
2077 const char *start,
2078 const char *end,
2079 const char **endPtr)
2080{
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;
2086
2087 switch (tok) {
2088 case XML_TOK_XML_DECL:
2089 {
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;
2102 }
2103 }
2104 break;
2105 case XML_TOK_PARTIAL:
2106 if (!ps_finalBuffer) {
2107 *endPtr = start;
2108 return XML_ERROR_NONE;
2109 }
2110 return XML_ERROR_UNCLOSED_TOKEN;
2111 case XML_TOK_PARTIAL_CHAR:
2112 if (!ps_finalBuffer) {
2113 *endPtr = start;
2114 return XML_ERROR_NONE;
2115 }
2116 return XML_ERROR_PARTIAL_CHAR;
2117 }
2118 processor = externalEntityContentProcessor;
2119 tagLevel = 1;
2120 return externalEntityContentProcessor(parser, start, end, endPtr);
2121}
2122
2123static enum XML_Error PTRCALL
2124externalEntityContentProcessor(XML_Parser parser,
2125 const char *start,
2126 const char *end,
2127 const char **endPtr)
2128{
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;
2134 }
2135 return result;
2136}
2137
2138static enum XML_Error
2139doContent(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)
2146{
2147 /* save one level of indirection */
2148 DTD * const dtd = _dtd;
2149
2150 const char **eventPP;
2151 const char **eventEndPP;
2152 if (enc == encoding) {
2153 eventPP = &eventPtr;
2154 eventEndPP = &eventEndPtr;
2155 }
2156 else {
2157 eventPP = &(openInternalEntities->internalEventPtr);
2158 eventEndPP = &(openInternalEntities->internalEventEndPtr);
2159 }
2160 *eventPP = s;
2161
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;
2171 }
2172 *eventEndPP = end;
2173 if (characterDataHandler) {
2174 XML_Char c = 0xA;
2175 characterDataHandler(handlerArg, &c, 1);
2176 }
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?
2181 */
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;
2192 }
2193 if (startTagLevel > 0) {
2194 if (tagLevel != startTagLevel)
2195 return XML_ERROR_ASYNC_ENTITY;
2196 *nextPtr = s;
2197 return XML_ERROR_NONE;
2198 }
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;
2207 }
2208 return XML_ERROR_UNCLOSED_TOKEN;
2209 case XML_TOK_PARTIAL_CHAR:
2210 if (haveMore) {
2211 *nextPtr = s;
2212 return XML_ERROR_NONE;
2213 }
2214 return XML_ERROR_PARTIAL_CHAR;
2215 case XML_TOK_ENTITY_REF:
2216 {
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;
2228 }
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.
2239 */
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;
2245 }
2246 else if (!entity) {
2247 if (skippedEntityHandler)
2248 skippedEntityHandler(handlerArg, name, 0);
2249 else if (defaultHandler)
2250 reportDefault(parser, enc, s, next);
2251 break;
2252 }
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;
2265 }
2266 result = processInternalEntity(parser, entity, XML_FALSE);
2267 if (result != XML_ERROR_NONE)
2268 return result;
2269 }
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);
2284 }
2285 else if (defaultHandler)
2286 reportDefault(parser, enc, s, next);
2287 break;
2288 }
2289 case XML_TOK_START_TAG_NO_ATTS:
2290 /* fall through */
2291 case XML_TOK_START_TAG_WITH_ATTS:
2292 {
2293 TAG *tag;
2294 enum XML_Error result;
2295 XML_Char *toPtr;
2296 if (freeTagList) {
2297 tag = freeTagList;
2298 freeTagList = freeTagList->parent;
2299 }
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;
2308 }
2309 tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE;
2310 }
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;
2319 {
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;
2333 }
2334 bufSize = (int)(tag->bufEnd - tag->buf) << 1;
2335 {
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;
2342 }
2343 }
2344 }
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;
2357 }
2358 case XML_TOK_EMPTY_ELEMENT_NO_ATTS:
2359 /* fall through */
2360 case XML_TOK_EMPTY_ELEMENT_WITH_ATTS:
2361 {
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;
2379 }
2380 if (endElementHandler) {
2381 if (startElementHandler)
2382 *eventPP = *eventEndPP;
2383 endElementHandler(handlerArg, name.str);
2384 noElmHandlers = XML_FALSE;
2385 }
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;
2397 }
2398 }
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;
2418 }
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
2429 */
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++;
2437 }
2438 *uri = XML_T('\0');
2439 }
2440 endElementHandler(handlerArg, tag->name.str);
2441 }
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;
2452 }
2453 if (tagLevel == 0)
2454 return epilogProcessor(parser, next, end, nextPtr);
2455 }
2456 break;
2457 case XML_TOK_CHAR_REF:
2458 {
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));
2465 }
2466 else if (defaultHandler)
2467 reportDefault(parser, enc, s, next);
2468 }
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);
2476 }
2477 else if (defaultHandler)
2478 reportDefault(parser, enc, s, next);
2479 break;
2480 case XML_TOK_CDATA_SECT_OPEN:
2481 {
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).
2494
2495 However, now we have a start/endCdataSectionHandler, so it seems
2496 easier to let the user deal with this.
2497 */
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;
2509 }
2510 }
2511 break;
2512 case XML_TOK_TRAILING_RSQB:
2513 if (haveMore) {
2514 *nextPtr = s;
2515 return XML_ERROR_NONE;
2516 }
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));
2523 }
2524 else
2525 characterDataHandler(handlerArg,
2526 (XML_Char *)s,
2527 (int)((XML_Char *)end - (XML_Char *)s));
2528 }
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?
2533 */
2534 if (startTagLevel == 0) {
2535 *eventPP = end;
2536 return XML_ERROR_NO_ELEMENTS;
2537 }
2538 if (tagLevel != startTagLevel) {
2539 *eventPP = end;
2540 return XML_ERROR_ASYNC_ENTITY;
2541 }
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;
2556 }
2557 }
2558 else
2559 characterDataHandler(handlerArg,
2560 (XML_Char *)s,
2561 (int)((XML_Char *)next - (XML_Char *)s));
2562 }
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;
2578 }
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: ;
2587 }
2588 }
2589 /* not reached */
2590}
2591
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)
2601*/
2602static enum XML_Error
2603storeAtts(XML_Parser parser, const ENCODING *enc,
2604 const char *attStr, TAG_NAME *tagNamePtr,
2605 BINDING **bindingsPtr)
2606{
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;
2619
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;
2632 }
2633 nDefaultAtts = elementType->nDefaultAtts;
2634
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);
2647 }
2648
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.
2660 */
2661 if ((attId->name)[-1]) {
2662 if (enc == encoding)
2663 eventPtr = atts[i].name;
2664 return XML_ERROR_DUPLICATE_ATTRIBUTE;
2665 }
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;
2671
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;
2679 }
2680 }
2681 }
2682
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);
2691 }
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);
2699 }
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;
2709 }
2710 else {
2711 /* deal with other prefixed names later */
2712 attIndex++;
2713 nPrefixes++;
2714 (attId->name)[-1] = 2;
2715 }
2716 }
2717 else
2718 attIndex++;
2719 }
2720
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;
2728 }
2729 }
2730 else
2731 idAttIndex = -1;
2732
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;
2743 }
2744 else {
2745 (da->id->name)[-1] = 2;
2746 nPrefixes++;
2747 appAtts[attIndex++] = da->id->name;
2748 appAtts[attIndex++] = da->value;
2749 }
2750 }
2751 else {
2752 (da->id->name)[-1] = 1;
2753 appAtts[attIndex++] = da->id->name;
2754 appAtts[attIndex++] = da->value;
2755 }
2756 }
2757 }
2758 appAtts[attIndex] = 0;
2759
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 */
2780 }
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;
2786 }
2787 nsAttsVersion = --version;
2788
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;
2803
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);
2810 }
2811 while (*s++ != XML_T(':'))
2812 ;
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++);
2819
2820 { /* Check hash table for duplicate of expanded name (uriName).
2821 Derived from code in lookup(HASH_TABLE *table, ...).
2822 */
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;
2835 }
2836 if (!step)
2837 step = PROBE_STEP(uriHash, mask, nsAttsPower);
2838 j < step ? (j += nsAttsSize - step) : (j -= step);
2839 }
2840 }
2841
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++);
2849 }
2850
2851 /* store expanded name in attribute list */
2852 s = poolStart(&tempPool);
2853 poolFinish(&tempPool);
2854 appAtts[i] = s;
2855
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;
2860
2861 if (!--nPrefixes) {
2862 i += 2;
2863 break;
2864 }
2865 }
2866 else /* not prefixed */
2867 ((XML_Char *)s)[-1] = 0; /* clear flag */
2868 }
2869 }
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;
2875
2876 if (!ns)
2877 return XML_ERROR_NONE;
2878
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(':'))
2886 ;
2887 }
2888 else if (dtd->defaultPrefix.binding) {
2889 binding = dtd->defaultPrefix.binding;
2890 localPart = tagNamePtr->str;
2891 }
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 */
2898 }
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;
2918 }
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));
2927 }
2928 tagNamePtr->str = binding->uri;
2929 return XML_ERROR_NONE;
2930}
2931
2932/* addBinding() overwrites the value of prefix->binding without checking.
2933 Therefore one must keep track of the old value outside of addBinding().
2934*/
2935static enum XML_Error
2936addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId,
2937 const XML_Char *uri, BINDING **bindingsPtr)
2938{
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'
2944 };
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'
2951 };
2952 static const int xmlnsLen =
2953 (int)sizeof(xmlnsNamespace)/sizeof(XML_Char) - 1;
2954
2955 XML_Bool mustBeXML = XML_FALSE;
2956 XML_Bool isXML = XML_TRUE;
2957 XML_Bool isXMLNS = XML_TRUE;
2958
2959 BINDING *b;
2960 int len;
2961
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;
2965
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')) {
2970
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;
2976
2977 if (prefix->name[3] == XML_T('\0'))
2978 mustBeXML = XML_TRUE;
2979 }
2980
2981 for (len = 0; uri[len]; len++) {
2982 if (isXML && (len > xmlLen || uri[len] != xmlNamespace[len]))
2983 isXML = XML_FALSE;
2984
2985 if (!mustBeXML && isXMLNS
2986 && (len > xmlnsLen || uri[len] != xmlnsNamespace[len]))
2987 isXMLNS = XML_FALSE;
2988 }
2989 isXML = isXML && len == xmlLen;
2990 isXMLNS = isXMLNS && len == xmlnsLen;
2991
2992 if (mustBeXML != isXML)
2993 return mustBeXML ? XML_ERROR_RESERVED_PREFIX_XML
2994 : XML_ERROR_RESERVED_NAMESPACE_URI;
2995
2996 if (isXMLNS)
2997 return XML_ERROR_RESERVED_NAMESPACE_URI;
2998
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;
3010 }
3011 freeBindingList = b->nextTagBinding;
3012 }
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;
3021 }
3022 b->uriAlloc = len + EXPAND_SPARE;
3023 }
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;
3043}
3044
3045/* The idea here is to avoid using stack for each CDATA section when
3046 the whole file is parsed with one call.
3047*/
3048static enum XML_Error PTRCALL
3049cdataSectionProcessor(XML_Parser parser,
3050 const char *start,
3051 const char *end,
3052 const char **endPtr)
3053{
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);
3062 }
3063 else {
3064 processor = contentProcessor;
3065 return contentProcessor(parser, start, end, endPtr);
3066 }
3067 }
3068 return result;
3069}
3070
3071/* startPtr gets set to non-null if the section is closed, and to null if
3072 the section is not yet closed.
3073*/
3074static enum XML_Error
3075doCdataSection(XML_Parser parser,
3076 const ENCODING *enc,
3077 const char **startPtr,
3078 const char *end,
3079 const char **nextPtr,
3080 XML_Bool haveMore)
3081{
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;
3089 }
3090 else {
3091 eventPP = &(openInternalEntities->internalEventPtr);
3092 eventEndPP = &(openInternalEntities->internalEventEndPtr);
3093 }
3094 *eventPP = s;
3095 *startPtr = NULL;
3096
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);
3122 }
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;
3138 }
3139 }
3140 else
3141 characterDataHandler(handlerArg,
3142 (XML_Char *)s,
3143 (int)((XML_Char *)next - (XML_Char *)s));
3144 }
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;
3155 }
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;
3162 }
3163 return XML_ERROR_UNCLOSED_CDATA_SECTION;
3164 default:
3165 *eventPP = next;
3166 return XML_ERROR_UNEXPECTED_STATE;
3167 }
3168
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: ;
3177 }
3178 }
3179 /* not reached */
3180}
3181
3182#ifdef XML_DTD
3183
3184/* The idea here is to avoid using stack for each IGNORE section when
3185 the whole file is parsed with one call.
3186*/
3187static enum XML_Error PTRCALL
3188ignoreSectionProcessor(XML_Parser parser,
3189 const char *start,
3190 const char *end,
3191 const char **endPtr)
3192{
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);
3200 }
3201 return result;
3202}
3203
3204/* startPtr gets set to non-null is the section is closed, and to null
3205 if the section is not yet closed.
3206*/
3207static enum XML_Error
3208doIgnoreSection(XML_Parser parser,
3209 const ENCODING *enc,
3210 const char **startPtr,
3211 const char *end,
3212 const char **nextPtr,
3213 XML_Bool haveMore)
3214{
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;
3224 }
3225 else {
3226 eventPP = &(openInternalEntities->internalEventPtr);
3227 eventEndPP = &(openInternalEntities->internalEventEndPtr);
3228 }
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;
3250 }
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;
3257 }
3258 return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */
3259 default:
3260 *eventPP = next;
3261 return XML_ERROR_UNEXPECTED_STATE;
3262 }
3263 /* not reached */
3264}
3265
3266#endif /* XML_DTD */
3267
3268static enum XML_Error
3269initializeEncoding(XML_Parser parser)
3270{
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;
3283 }
3284 encodingBuf[i] = (char)protocolEncodingName[i];
3285 }
3286 encodingBuf[i] = '\0';
3287 s = encodingBuf;
3288 }
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);
3295}
3296
3297static enum XML_Error
3298processXmlDecl(XML_Parser parser, int isGeneralTextEntity,
3299 const char *s, const char *next)
3300{
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,
3312 s,
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;
3324 }
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 */
3331 }
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);
3342 }
3343 if (version) {
3344 storedversion = poolStoreString(&temp2Pool,
3345 encoding,
3346 version,
3347 versionend - encoding->minBytesPerChar);
3348 if (!storedversion)
3349 return XML_ERROR_NO_MEMORY;
3350 }
3351 xmlDeclHandler(handlerArg, storedversion, storedEncName, standalone);
3352 }
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;
3360 }
3361 encoding = newEncoding;
3362 }
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;
3371 }
3372 result = handleUnknownEncoding(parser, storedEncName);
3373 poolClear(&temp2Pool);
3374 if (result == XML_ERROR_UNKNOWN_ENCODING)
3375 eventPtr = encodingName;
3376 return result;
3377 }
3378 }
3379
3380 if (storedEncName || storedversion)
3381 poolClear(&temp2Pool);
3382
3383 return XML_ERROR_NONE;
3384}
3385
3386static enum XML_Error
3387handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName)
3388{
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;
3405 }
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;
3417 }
3418 }
3419 if (info.release != NULL)
3420 info.release(info.data);
3421 }
3422 return XML_ERROR_UNKNOWN_ENCODING;
3423}
3424
3425static enum XML_Error PTRCALL
3426prologInitProcessor(XML_Parser parser,
3427 const char *s,
3428 const char *end,
3429 const char **nextPtr)
3430{
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);
3436}
3437
3438#ifdef XML_DTD
3439
3440static enum XML_Error PTRCALL
3441externalParEntInitProcessor(XML_Parser parser,
3442 const char *s,
3443 const char *end,
3444 const char **nextPtr)
3445{
3446 enum XML_Error result = initializeEncoding(parser);
3447 if (result != XML_ERROR_NONE)
3448 return result;
3449
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;
3453
3454 if (prologState.inEntityValue) {
3455 processor = entityValueInitProcessor;
3456 return entityValueInitProcessor(parser, s, end, nextPtr);
3457 }
3458 else {
3459 processor = externalParEntProcessor;
3460 return externalParEntProcessor(parser, s, end, nextPtr);
3461 }
3462}
3463
3464static enum XML_Error PTRCALL
3465entityValueInitProcessor(XML_Parser parser,
3466 const char *s,
3467 const char *end,
3468 const char **nextPtr)
3469{
3470 int tok;
3471 const char *start = s;
3472 const char *next = start;
3473 eventPtr = start;
3474
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;
3482 }
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;
3493 }
3494 /* found end of entity value - can store it now */
3495 return storeEntityValue(parser, encoding, s, end);
3496 }
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;
3510 }
3511 /* stop scanning for text declaration - we found one */
3512 processor = entityValueProcessor;
3513 return entityValueProcessor(parser, next, end, nextPtr);
3514 }
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
3521 */
3522 else if (tok == XML_TOK_BOM && next == end && !ps_finalBuffer) {
3523 *nextPtr = next;
3524 return XML_ERROR_NONE;
3525 }
3526 start = next;
3527 eventPtr = start;
3528 }
3529}
3530
3531static enum XML_Error PTRCALL
3532externalParEntProcessor(XML_Parser parser,
3533 const char *s,
3534 const char *end,
3535 const char **nextPtr)
3536{
3537 const char *next = s;
3538 int tok;
3539
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;
3545 }
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;
3556 }
3557 }
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
3561 */
3562 else if (tok == XML_TOK_BOM) {
3563 s = next;
3564 tok = XmlPrologTok(encoding, s, end, &next);
3565 }
3566
3567 processor = prologProcessor;
3568 return doProlog(parser, encoding, s, end, tok, next,
3569 nextPtr, (XML_Bool)!ps_finalBuffer);
3570}
3571
3572static enum XML_Error PTRCALL
3573entityValueProcessor(XML_Parser parser,
3574 const char *s,
3575 const char *end,
3576 const char **nextPtr)
3577{
3578 const char *start = s;
3579 const char *next = s;
3580 const ENCODING *enc = encoding;
3581 int tok;
3582
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;
3589 }
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;
3600 }
3601 /* found end of entity value - can store it now */
3602 return storeEntityValue(parser, enc, s, end);
3603 }
3604 start = next;
3605 }
3606}
3607
3608#endif /* XML_DTD */
3609
3610static enum XML_Error PTRCALL
3611prologProcessor(XML_Parser parser,
3612 const char *s,
3613 const char *end,
3614 const char **nextPtr)
3615{
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);
3620}
3621
3622static enum XML_Error
3623doProlog(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)
3631{
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' };
3650
3651 /* save one level of indirection */
3652 DTD * const dtd = _dtd;
3653
3654 const char **eventPP;
3655 const char **eventEndPP;
3656 enum XML_Content_Quant quant;
3657
3658 if (enc == encoding) {
3659 eventPP = &eventPtr;
3660 eventEndPP = &eventEndPtr;
3661 }
3662 else {
3663 eventPP = &(openInternalEntities->internalEventPtr);
3664 eventEndPP = &(openInternalEntities->internalEventEndPtr);
3665 }
3666
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;
3676 }
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_PROLOG_S:
3686 tok = -tok;
3687 break;
3688 case XML_TOK_NONE:
3689#ifdef XML_DTD
3690 /* for internal PE NOT referenced between declarations */
3691 if (enc != encoding && !openInternalEntities->betweenDecl) {
3692 *nextPtr = s;
3693 return XML_ERROR_NONE;
3694 }
3695 /* WFC: PE Between Declarations - must check that PE contains
3696 complete markup, not only for external PEs, but also for
3697 internal PEs if the reference occurs between declarations.
3698 */
3699 if (isParamEntity || enc != encoding) {
3700 if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc)
3701 == XML_ROLE_ERROR)
3702 return XML_ERROR_INCOMPLETE_PE;
3703 *nextPtr = s;
3704 return XML_ERROR_NONE;
3705 }
3706#endif /* XML_DTD */
3707 return XML_ERROR_NO_ELEMENTS;
3708 default:
3709 tok = -tok;
3710 next = end;
3711 break;
3712 }
3713 }
3714 role = XmlTokenRole(&prologState, tok, s, next, enc);
3715 switch (role) {
3716 case XML_ROLE_XML_DECL:
3717 {
3718 enum XML_Error result = processXmlDecl(parser, 0, s, next);
3719 if (result != XML_ERROR_NONE)
3720 return result;
3721 enc = encoding;
3722 handleDefault = XML_FALSE;
3723 }
3724 break;
3725 case XML_ROLE_DOCTYPE_NAME:
3726 if (startDoctypeDeclHandler) {
3727 doctypeName = poolStoreString(&tempPool, enc, s, next);
3728 if (!doctypeName)
3729 return XML_ERROR_NO_MEMORY;
3730 poolFinish(&tempPool);
3731 doctypePubid = NULL;
3732 handleDefault = XML_FALSE;
3733 }
3734 doctypeSysid = NULL; /* always initialize to NULL */
3735 break;
3736 case XML_ROLE_DOCTYPE_INTERNAL_SUBSET:
3737 if (startDoctypeDeclHandler) {
3738 startDoctypeDeclHandler(handlerArg, doctypeName, doctypeSysid,
3739 doctypePubid, 1);
3740 doctypeName = NULL;
3741 poolClear(&tempPool);
3742 handleDefault = XML_FALSE;
3743 }
3744 break;
3745#ifdef XML_DTD
3746 case XML_ROLE_TEXT_DECL:
3747 {
3748 enum XML_Error result = processXmlDecl(parser, 1, s, next);
3749 if (result != XML_ERROR_NONE)
3750 return result;
3751 enc = encoding;
3752 handleDefault = XML_FALSE;
3753 }
3754 break;
3755#endif /* XML_DTD */
3756 case XML_ROLE_DOCTYPE_PUBLIC_ID:
3757#ifdef XML_DTD
3758 useForeignDTD = XML_FALSE;
3759 declEntity = (ENTITY *)lookup(&dtd->paramEntities,
3760 externalSubsetName,
3761 sizeof(ENTITY));
3762 if (!declEntity)
3763 return XML_ERROR_NO_MEMORY;
3764#endif /* XML_DTD */
3765 dtd->hasParamEntityRefs = XML_TRUE;
3766 if (startDoctypeDeclHandler) {
3767 if (!XmlIsPublicId(enc, s, next, eventPP))
3768 return XML_ERROR_PUBLICID;
3769 doctypePubid = poolStoreString(&tempPool, enc,
3770 s + enc->minBytesPerChar,
3771 next - enc->minBytesPerChar);
3772 if (!doctypePubid)
3773 return XML_ERROR_NO_MEMORY;
3774 normalizePublicId((XML_Char *)doctypePubid);
3775 poolFinish(&tempPool);
3776 handleDefault = XML_FALSE;
3777 goto alreadyChecked;
3778 }
3779 /* fall through */
3780 case XML_ROLE_ENTITY_PUBLIC_ID:
3781 if (!XmlIsPublicId(enc, s, next, eventPP))
3782 return XML_ERROR_PUBLICID;
3783 alreadyChecked:
3784 if (dtd->keepProcessing && declEntity) {
3785 XML_Char *tem = poolStoreString(&dtd->pool,
3786 enc,
3787 s + enc->minBytesPerChar,
3788 next - enc->minBytesPerChar);
3789 if (!tem)
3790 return XML_ERROR_NO_MEMORY;
3791 normalizePublicId(tem);
3792 declEntity->publicId = tem;
3793 poolFinish(&dtd->pool);
3794 if (entityDeclHandler)
3795 handleDefault = XML_FALSE;
3796 }
3797 break;
3798 case XML_ROLE_DOCTYPE_CLOSE:
3799 if (doctypeName) {
3800 startDoctypeDeclHandler(handlerArg, doctypeName,
3801 doctypeSysid, doctypePubid, 0);
3802 poolClear(&tempPool);
3803 handleDefault = XML_FALSE;
3804 }
3805 /* doctypeSysid will be non-NULL in the case of a previous
3806 XML_ROLE_DOCTYPE_SYSTEM_ID, even if startDoctypeDeclHandler
3807 was not set, indicating an external subset
3808 */
3809#ifdef XML_DTD
3810 if (doctypeSysid || useForeignDTD) {
3811 XML_Bool hadParamEntityRefs = dtd->hasParamEntityRefs;
3812 dtd->hasParamEntityRefs = XML_TRUE;
3813 if (paramEntityParsing && externalEntityRefHandler) {
3814 ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities,
3815 externalSubsetName,
3816 sizeof(ENTITY));
3817 if (!entity)
3818 return XML_ERROR_NO_MEMORY;
3819 if (useForeignDTD)
3820 entity->base = curBase;
3821 dtd->paramEntityRead = XML_FALSE;
3822 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
3823 0,
3824 entity->base,
3825 entity->systemId,
3826 entity->publicId))
3827 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
3828 if (dtd->paramEntityRead) {
3829 if (!dtd->standalone &&
3830 notStandaloneHandler &&
3831 !notStandaloneHandler(handlerArg))
3832 return XML_ERROR_NOT_STANDALONE;
3833 }
3834 /* if we didn't read the foreign DTD then this means that there
3835 is no external subset and we must reset dtd->hasParamEntityRefs
3836 */
3837 else if (!doctypeSysid)
3838 dtd->hasParamEntityRefs = hadParamEntityRefs;
3839 /* end of DTD - no need to update dtd->keepProcessing */
3840 }
3841 useForeignDTD = XML_FALSE;
3842 }
3843#endif /* XML_DTD */
3844 if (endDoctypeDeclHandler) {
3845 endDoctypeDeclHandler(handlerArg);
3846 handleDefault = XML_FALSE;
3847 }
3848 break;
3849 case XML_ROLE_INSTANCE_START:
3850#ifdef XML_DTD
3851 /* if there is no DOCTYPE declaration then now is the
3852 last chance to read the foreign DTD
3853 */
3854 if (useForeignDTD) {
3855 XML_Bool hadParamEntityRefs = dtd->hasParamEntityRefs;
3856 dtd->hasParamEntityRefs = XML_TRUE;
3857 if (paramEntityParsing && externalEntityRefHandler) {
3858 ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities,
3859 externalSubsetName,
3860 sizeof(ENTITY));
3861 if (!entity)
3862 return XML_ERROR_NO_MEMORY;
3863 entity->base = curBase;
3864 dtd->paramEntityRead = XML_FALSE;
3865 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
3866 0,
3867 entity->base,
3868 entity->systemId,
3869 entity->publicId))
3870 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
3871 if (dtd->paramEntityRead) {
3872 if (!dtd->standalone &&
3873 notStandaloneHandler &&
3874 !notStandaloneHandler(handlerArg))
3875 return XML_ERROR_NOT_STANDALONE;
3876 }
3877 /* if we didn't read the foreign DTD then this means that there
3878 is no external subset and we must reset dtd->hasParamEntityRefs
3879 */
3880 else
3881 dtd->hasParamEntityRefs = hadParamEntityRefs;
3882 /* end of DTD - no need to update dtd->keepProcessing */
3883 }
3884 }
3885#endif /* XML_DTD */
3886 processor = contentProcessor;
3887 return contentProcessor(parser, s, end, nextPtr);
3888 case XML_ROLE_ATTLIST_ELEMENT_NAME:
3889 declElementType = getElementType(parser, enc, s, next);
3890 if (!declElementType)
3891 return XML_ERROR_NO_MEMORY;
3892 goto checkAttListDeclHandler;
3893 case XML_ROLE_ATTRIBUTE_NAME:
3894 declAttributeId = getAttributeId(parser, enc, s, next);
3895 if (!declAttributeId)
3896 return XML_ERROR_NO_MEMORY;
3897 declAttributeIsCdata = XML_FALSE;
3898 declAttributeType = NULL;
3899 declAttributeIsId = XML_FALSE;
3900 goto checkAttListDeclHandler;
3901 case XML_ROLE_ATTRIBUTE_TYPE_CDATA:
3902 declAttributeIsCdata = XML_TRUE;
3903 declAttributeType = atypeCDATA;
3904 goto checkAttListDeclHandler;
3905 case XML_ROLE_ATTRIBUTE_TYPE_ID:
3906 declAttributeIsId = XML_TRUE;
3907 declAttributeType = atypeID;
3908 goto checkAttListDeclHandler;
3909 case XML_ROLE_ATTRIBUTE_TYPE_IDREF:
3910 declAttributeType = atypeIDREF;
3911 goto checkAttListDeclHandler;
3912 case XML_ROLE_ATTRIBUTE_TYPE_IDREFS:
3913 declAttributeType = atypeIDREFS;
3914 goto checkAttListDeclHandler;
3915 case XML_ROLE_ATTRIBUTE_TYPE_ENTITY:
3916 declAttributeType = atypeENTITY;
3917 goto checkAttListDeclHandler;
3918 case XML_ROLE_ATTRIBUTE_TYPE_ENTITIES:
3919 declAttributeType = atypeENTITIES;
3920 goto checkAttListDeclHandler;
3921 case XML_ROLE_ATTRIBUTE_TYPE_NMTOKEN:
3922 declAttributeType = atypeNMTOKEN;
3923 goto checkAttListDeclHandler;
3924 case XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS:
3925 declAttributeType = atypeNMTOKENS;
3926 checkAttListDeclHandler:
3927 if (dtd->keepProcessing && attlistDeclHandler)
3928 handleDefault = XML_FALSE;
3929 break;
3930 case XML_ROLE_ATTRIBUTE_ENUM_VALUE:
3931 case XML_ROLE_ATTRIBUTE_NOTATION_VALUE:
3932 if (dtd->keepProcessing && attlistDeclHandler) {
3933 const XML_Char *prefix;
3934 if (declAttributeType) {
3935 prefix = enumValueSep;
3936 }
3937 else {
3938 prefix = (role == XML_ROLE_ATTRIBUTE_NOTATION_VALUE
3939 ? notationPrefix
3940 : enumValueStart);
3941 }
3942 if (!poolAppendString(&tempPool, prefix))
3943 return XML_ERROR_NO_MEMORY;
3944 if (!poolAppend(&tempPool, enc, s, next))
3945 return XML_ERROR_NO_MEMORY;
3946 declAttributeType = tempPool.start;
3947 handleDefault = XML_FALSE;
3948 }
3949 break;
3950 case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE:
3951 case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE:
3952 if (dtd->keepProcessing) {
3953 if (!defineAttribute(declElementType, declAttributeId,
3954 declAttributeIsCdata, declAttributeIsId,
3955 0, parser))
3956 return XML_ERROR_NO_MEMORY;
3957 if (attlistDeclHandler && declAttributeType) {
3958 if (*declAttributeType == XML_T('(')
3959 || (*declAttributeType == XML_T('N')
3960 && declAttributeType[1] == XML_T('O'))) {
3961 /* Enumerated or Notation type */
3962 if (!poolAppendChar(&tempPool, XML_T(')'))
3963 || !poolAppendChar(&tempPool, XML_T('\0')))
3964 return XML_ERROR_NO_MEMORY;
3965 declAttributeType = tempPool.start;
3966 poolFinish(&tempPool);
3967 }
3968 *eventEndPP = s;
3969 attlistDeclHandler(handlerArg, declElementType->name,
3970 declAttributeId->name, declAttributeType,
3971 0, role == XML_ROLE_REQUIRED_ATTRIBUTE_VALUE);
3972 poolClear(&tempPool);
3973 handleDefault = XML_FALSE;
3974 }
3975 }
3976 break;
3977 case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE:
3978 case XML_ROLE_FIXED_ATTRIBUTE_VALUE:
3979 if (dtd->keepProcessing) {
3980 const XML_Char *attVal;
3981 enum XML_Error result =
3982 storeAttributeValue(parser, enc, declAttributeIsCdata,
3983 s + enc->minBytesPerChar,
3984 next - enc->minBytesPerChar,
3985 &dtd->pool);
3986 if (result)
3987 return result;
3988 attVal = poolStart(&dtd->pool);
3989 poolFinish(&dtd->pool);
3990 /* ID attributes aren't allowed to have a default */
3991 if (!defineAttribute(declElementType, declAttributeId,
3992 declAttributeIsCdata, XML_FALSE, attVal, parser))
3993 return XML_ERROR_NO_MEMORY;
3994 if (attlistDeclHandler && declAttributeType) {
3995 if (*declAttributeType == XML_T('(')
3996 || (*declAttributeType == XML_T('N')
3997 && declAttributeType[1] == XML_T('O'))) {
3998 /* Enumerated or Notation type */
3999 if (!poolAppendChar(&tempPool, XML_T(')'))
4000 || !poolAppendChar(&tempPool, XML_T('\0')))
4001 return XML_ERROR_NO_MEMORY;
4002 declAttributeType = tempPool.start;
4003 poolFinish(&tempPool);
4004 }
4005 *eventEndPP = s;
4006 attlistDeclHandler(handlerArg, declElementType->name,
4007 declAttributeId->name, declAttributeType,
4008 attVal,
4009 role == XML_ROLE_FIXED_ATTRIBUTE_VALUE);
4010 poolClear(&tempPool);
4011 handleDefault = XML_FALSE;
4012 }
4013 }
4014 break;
4015 case XML_ROLE_ENTITY_VALUE:
4016 if (dtd->keepProcessing) {
4017 enum XML_Error result = storeEntityValue(parser, enc,
4018 s + enc->minBytesPerChar,
4019 next - enc->minBytesPerChar);
4020 if (declEntity) {
4021 declEntity->textPtr = poolStart(&dtd->entityValuePool);
4022 declEntity->textLen = (int)(poolLength(&dtd->entityValuePool));
4023 poolFinish(&dtd->entityValuePool);
4024 if (entityDeclHandler) {
4025 *eventEndPP = s;
4026 entityDeclHandler(handlerArg,
4027 declEntity->name,
4028 declEntity->is_param,
4029 declEntity->textPtr,
4030 declEntity->textLen,
4031 curBase, 0, 0, 0);
4032 handleDefault = XML_FALSE;
4033 }
4034 }
4035 else
4036 poolDiscard(&dtd->entityValuePool);
4037 if (result != XML_ERROR_NONE)
4038 return result;
4039 }
4040 break;
4041 case XML_ROLE_DOCTYPE_SYSTEM_ID:
4042#ifdef XML_DTD
4043 useForeignDTD = XML_FALSE;
4044#endif /* XML_DTD */
4045 dtd->hasParamEntityRefs = XML_TRUE;
4046 if (startDoctypeDeclHandler) {
4047 doctypeSysid = poolStoreString(&tempPool, enc,
4048 s + enc->minBytesPerChar,
4049 next - enc->minBytesPerChar);
4050 if (doctypeSysid == NULL)
4051 return XML_ERROR_NO_MEMORY;
4052 poolFinish(&tempPool);
4053 handleDefault = XML_FALSE;
4054 }
4055#ifdef XML_DTD
4056 else
4057 /* use externalSubsetName to make doctypeSysid non-NULL
4058 for the case where no startDoctypeDeclHandler is set */
4059 doctypeSysid = externalSubsetName;
4060#endif /* XML_DTD */
4061 if (!dtd->standalone
4062#ifdef XML_DTD
4063 && !paramEntityParsing
4064#endif /* XML_DTD */
4065 && notStandaloneHandler
4066 && !notStandaloneHandler(handlerArg))
4067 return XML_ERROR_NOT_STANDALONE;
4068#ifndef XML_DTD
4069 break;
4070#else /* XML_DTD */
4071 if (!declEntity) {
4072 declEntity = (ENTITY *)lookup(&dtd->paramEntities,
4073 externalSubsetName,
4074 sizeof(ENTITY));
4075 if (!declEntity)
4076 return XML_ERROR_NO_MEMORY;
4077 declEntity->publicId = NULL;
4078 }
4079 /* fall through */
4080#endif /* XML_DTD */
4081 case XML_ROLE_ENTITY_SYSTEM_ID:
4082 if (dtd->keepProcessing && declEntity) {
4083 declEntity->systemId = poolStoreString(&dtd->pool, enc,
4084 s + enc->minBytesPerChar,
4085 next - enc->minBytesPerChar);
4086 if (!declEntity->systemId)
4087 return XML_ERROR_NO_MEMORY;
4088 declEntity->base = curBase;
4089 poolFinish(&dtd->pool);
4090 if (entityDeclHandler)
4091 handleDefault = XML_FALSE;
4092 }
4093 break;
4094 case XML_ROLE_ENTITY_COMPLETE:
4095 if (dtd->keepProcessing && declEntity && entityDeclHandler) {
4096 *eventEndPP = s;
4097 entityDeclHandler(handlerArg,
4098 declEntity->name,
4099 declEntity->is_param,
4100 0,0,
4101 declEntity->base,
4102 declEntity->systemId,
4103 declEntity->publicId,
4104 0);
4105 handleDefault = XML_FALSE;
4106 }
4107 break;
4108 case XML_ROLE_ENTITY_NOTATION_NAME:
4109 if (dtd->keepProcessing && declEntity) {
4110 declEntity->notation = poolStoreString(&dtd->pool, enc, s, next);
4111 if (!declEntity->notation)
4112 return XML_ERROR_NO_MEMORY;
4113 poolFinish(&dtd->pool);
4114 if (unparsedEntityDeclHandler) {
4115 *eventEndPP = s;
4116 unparsedEntityDeclHandler(handlerArg,
4117 declEntity->name,
4118 declEntity->base,
4119 declEntity->systemId,
4120 declEntity->publicId,
4121 declEntity->notation);
4122 handleDefault = XML_FALSE;
4123 }
4124 else if (entityDeclHandler) {
4125 *eventEndPP = s;
4126 entityDeclHandler(handlerArg,
4127 declEntity->name,
4128 0,0,0,
4129 declEntity->base,
4130 declEntity->systemId,
4131 declEntity->publicId,
4132 declEntity->notation);
4133 handleDefault = XML_FALSE;
4134 }
4135 }
4136 break;
4137 case XML_ROLE_GENERAL_ENTITY_NAME:
4138 {
4139 if (XmlPredefinedEntityName(enc, s, next)) {
4140 declEntity = NULL;
4141 break;
4142 }
4143 if (dtd->keepProcessing) {
4144 const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next);
4145 if (!name)
4146 return XML_ERROR_NO_MEMORY;
4147 declEntity = (ENTITY *)lookup(&dtd->generalEntities, name,
4148 sizeof(ENTITY));
4149 if (!declEntity)
4150 return XML_ERROR_NO_MEMORY;
4151 if (declEntity->name != name) {
4152 poolDiscard(&dtd->pool);
4153 declEntity = NULL;
4154 }
4155 else {
4156 poolFinish(&dtd->pool);
4157 declEntity->publicId = NULL;
4158 declEntity->is_param = XML_FALSE;
4159 /* if we have a parent parser or are reading an internal parameter
4160 entity, then the entity declaration is not considered "internal"
4161 */
4162 declEntity->is_internal = !(parentParser || openInternalEntities);
4163 if (entityDeclHandler)
4164 handleDefault = XML_FALSE;
4165 }
4166 }
4167 else {
4168 poolDiscard(&dtd->pool);
4169 declEntity = NULL;
4170 }
4171 }
4172 break;
4173 case XML_ROLE_PARAM_ENTITY_NAME:
4174#ifdef XML_DTD
4175 if (dtd->keepProcessing) {
4176 const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next);
4177 if (!name)
4178 return XML_ERROR_NO_MEMORY;
4179 declEntity = (ENTITY *)lookup(&dtd->paramEntities,
4180 name, sizeof(ENTITY));
4181 if (!declEntity)
4182 return XML_ERROR_NO_MEMORY;
4183 if (declEntity->name != name) {
4184 poolDiscard(&dtd->pool);
4185 declEntity = NULL;
4186 }
4187 else {
4188 poolFinish(&dtd->pool);
4189 declEntity->publicId = NULL;
4190 declEntity->is_param = XML_TRUE;
4191 /* if we have a parent parser or are reading an internal parameter
4192 entity, then the entity declaration is not considered "internal"
4193 */
4194 declEntity->is_internal = !(parentParser || openInternalEntities);
4195 if (entityDeclHandler)
4196 handleDefault = XML_FALSE;
4197 }
4198 }
4199 else {
4200 poolDiscard(&dtd->pool);
4201 declEntity = NULL;
4202 }
4203#else /* not XML_DTD */
4204 declEntity = NULL;
4205#endif /* XML_DTD */
4206 break;
4207 case XML_ROLE_NOTATION_NAME:
4208 declNotationPublicId = NULL;
4209 declNotationName = NULL;
4210 if (notationDeclHandler) {
4211 declNotationName = poolStoreString(&tempPool, enc, s, next);
4212 if (!declNotationName)
4213 return XML_ERROR_NO_MEMORY;
4214 poolFinish(&tempPool);
4215 handleDefault = XML_FALSE;
4216 }
4217 break;
4218 case XML_ROLE_NOTATION_PUBLIC_ID:
4219 if (!XmlIsPublicId(enc, s, next, eventPP))
4220 return XML_ERROR_PUBLICID;
4221 if (declNotationName) { /* means notationDeclHandler != NULL */
4222 XML_Char *tem = poolStoreString(&tempPool,
4223 enc,
4224 s + enc->minBytesPerChar,
4225 next - enc->minBytesPerChar);
4226 if (!tem)
4227 return XML_ERROR_NO_MEMORY;
4228 normalizePublicId(tem);
4229 declNotationPublicId = tem;
4230 poolFinish(&tempPool);
4231 handleDefault = XML_FALSE;
4232 }
4233 break;
4234 case XML_ROLE_NOTATION_SYSTEM_ID:
4235 if (declNotationName && notationDeclHandler) {
4236 const XML_Char *systemId
4237 = poolStoreString(&tempPool, enc,
4238 s + enc->minBytesPerChar,
4239 next - enc->minBytesPerChar);
4240 if (!systemId)
4241 return XML_ERROR_NO_MEMORY;
4242 *eventEndPP = s;
4243 notationDeclHandler(handlerArg,
4244 declNotationName,
4245 curBase,
4246 systemId,
4247 declNotationPublicId);
4248 handleDefault = XML_FALSE;
4249 }
4250 poolClear(&tempPool);
4251 break;
4252 case XML_ROLE_NOTATION_NO_SYSTEM_ID:
4253 if (declNotationPublicId && notationDeclHandler) {
4254 *eventEndPP = s;
4255 notationDeclHandler(handlerArg,
4256 declNotationName,
4257 curBase,
4258 0,
4259 declNotationPublicId);
4260 handleDefault = XML_FALSE;
4261 }
4262 poolClear(&tempPool);
4263 break;
4264 case XML_ROLE_ERROR:
4265 switch (tok) {
4266 case XML_TOK_PARAM_ENTITY_REF:
4267 /* PE references in internal subset are
4268 not allowed within declarations. */
4269 return XML_ERROR_PARAM_ENTITY_REF;
4270 case XML_TOK_XML_DECL:
4271 return XML_ERROR_MISPLACED_XML_PI;
4272 default:
4273 return XML_ERROR_SYNTAX;
4274 }
4275#ifdef XML_DTD
4276 case XML_ROLE_IGNORE_SECT:
4277 {
4278 enum XML_Error result;
4279 if (defaultHandler)
4280 reportDefault(parser, enc, s, next);
4281 handleDefault = XML_FALSE;
4282 result = doIgnoreSection(parser, enc, &next, end, nextPtr, haveMore);
4283 if (result != XML_ERROR_NONE)
4284 return result;
4285 else if (!next) {
4286 processor = ignoreSectionProcessor;
4287 return result;
4288 }
4289 }
4290 break;
4291#endif /* XML_DTD */
4292 case XML_ROLE_GROUP_OPEN:
4293 if (prologState.level >= groupSize) {
4294 if (groupSize) {
4295 char *temp = (char *)REALLOC(groupConnector, groupSize *= 2);
4296 if (temp == NULL)
4297 return XML_ERROR_NO_MEMORY;
4298 groupConnector = temp;
4299 if (dtd->scaffIndex) {
4300 int *temp = (int *)REALLOC(dtd->scaffIndex,
4301 groupSize * sizeof(int));
4302 if (temp == NULL)
4303 return XML_ERROR_NO_MEMORY;
4304 dtd->scaffIndex = temp;
4305 }
4306 }
4307 else {
4308 groupConnector = (char *)MALLOC(groupSize = 32);
4309 if (!groupConnector)
4310 return XML_ERROR_NO_MEMORY;
4311 }
4312 }
4313 groupConnector[prologState.level] = 0;
4314 if (dtd->in_eldecl) {
4315 int myindex = nextScaffoldPart(parser);
4316 if (myindex < 0)
4317 return XML_ERROR_NO_MEMORY;
4318 dtd->scaffIndex[dtd->scaffLevel] = myindex;
4319 dtd->scaffLevel++;
4320 dtd->scaffold[myindex].type = XML_CTYPE_SEQ;
4321 if (elementDeclHandler)
4322 handleDefault = XML_FALSE;
4323 }
4324 break;
4325 case XML_ROLE_GROUP_SEQUENCE:
4326 if (groupConnector[prologState.level] == '|')
4327 return XML_ERROR_SYNTAX;
4328 groupConnector[prologState.level] = ',';
4329 if (dtd->in_eldecl && elementDeclHandler)
4330 handleDefault = XML_FALSE;
4331 break;
4332 case XML_ROLE_GROUP_CHOICE:
4333 if (groupConnector[prologState.level] == ',')
4334 return XML_ERROR_SYNTAX;
4335 if (dtd->in_eldecl
4336 && !groupConnector[prologState.level]
4337 && (dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type
4338 != XML_CTYPE_MIXED)
4339 ) {
4340 dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type
4341 = XML_CTYPE_CHOICE;
4342 if (elementDeclHandler)
4343 handleDefault = XML_FALSE;
4344 }
4345 groupConnector[prologState.level] = '|';
4346 break;
4347 case XML_ROLE_PARAM_ENTITY_REF:
4348#ifdef XML_DTD
4349 case XML_ROLE_INNER_PARAM_ENTITY_REF:
4350 dtd->hasParamEntityRefs = XML_TRUE;
4351 if (!paramEntityParsing)
4352 dtd->keepProcessing = dtd->standalone;
4353 else {
4354 const XML_Char *name;
4355 ENTITY *entity;
4356 name = poolStoreString(&dtd->pool, enc,
4357 s + enc->minBytesPerChar,
4358 next - enc->minBytesPerChar);
4359 if (!name)
4360 return XML_ERROR_NO_MEMORY;
4361 entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0);
4362 poolDiscard(&dtd->pool);
4363 /* first, determine if a check for an existing declaration is needed;
4364 if yes, check that the entity exists, and that it is internal,
4365 otherwise call the skipped entity handler
4366 */
4367 if (prologState.documentEntity &&
4368 (dtd->standalone
4369 ? !openInternalEntities
4370 : !dtd->hasParamEntityRefs)) {
4371 if (!entity)
4372 return XML_ERROR_UNDEFINED_ENTITY;
4373 else if (!entity->is_internal)
4374 return XML_ERROR_ENTITY_DECLARED_IN_PE;
4375 }
4376 else if (!entity) {
4377 dtd->keepProcessing = dtd->standalone;
4378 /* cannot report skipped entities in declarations */
4379 if ((role == XML_ROLE_PARAM_ENTITY_REF) && skippedEntityHandler) {
4380 skippedEntityHandler(handlerArg, name, 1);
4381 handleDefault = XML_FALSE;
4382 }
4383 break;
4384 }
4385 if (entity->open)
4386 return XML_ERROR_RECURSIVE_ENTITY_REF;
4387 if (entity->textPtr) {
4388 enum XML_Error result;
4389 XML_Bool betweenDecl =
4390 (role == XML_ROLE_PARAM_ENTITY_REF ? XML_TRUE : XML_FALSE);
4391 result = processInternalEntity(parser, entity, betweenDecl);
4392 if (result != XML_ERROR_NONE)
4393 return result;
4394 handleDefault = XML_FALSE;
4395 break;
4396 }
4397 if (externalEntityRefHandler) {
4398 dtd->paramEntityRead = XML_FALSE;
4399 entity->open = XML_TRUE;
4400 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
4401 0,
4402 entity->base,
4403 entity->systemId,
4404 entity->publicId)) {
4405 entity->open = XML_FALSE;
4406 return XML_ERROR_EXTERNAL_ENTITY_HANDLING;
4407 }
4408 entity->open = XML_FALSE;
4409 handleDefault = XML_FALSE;
4410 if (!dtd->paramEntityRead) {
4411 dtd->keepProcessing = dtd->standalone;
4412 break;
4413 }
4414 }
4415 else {
4416 dtd->keepProcessing = dtd->standalone;
4417 break;
4418 }
4419 }
4420#endif /* XML_DTD */
4421 if (!dtd->standalone &&
4422 notStandaloneHandler &&
4423 !notStandaloneHandler(handlerArg))
4424 return XML_ERROR_NOT_STANDALONE;
4425 break;
4426
4427 /* Element declaration stuff */
4428
4429 case XML_ROLE_ELEMENT_NAME:
4430 if (elementDeclHandler) {
4431 declElementType = getElementType(parser, enc, s, next);
4432 if (!declElementType)
4433 return XML_ERROR_NO_MEMORY;
4434 dtd->scaffLevel = 0;
4435 dtd->scaffCount = 0;
4436 dtd->in_eldecl = XML_TRUE;
4437 handleDefault = XML_FALSE;
4438 }
4439 break;
4440
4441 case XML_ROLE_CONTENT_ANY:
4442 case XML_ROLE_CONTENT_EMPTY:
4443 if (dtd->in_eldecl) {
4444 if (elementDeclHandler) {
4445 XML_Content * content = (XML_Content *) MALLOC(sizeof(XML_Content));
4446 if (!content)
4447 return XML_ERROR_NO_MEMORY;
4448 content->quant = XML_CQUANT_NONE;
4449 content->name = NULL;
4450 content->numchildren = 0;
4451 content->children = NULL;
4452 content->type = ((role == XML_ROLE_CONTENT_ANY) ?
4453 XML_CTYPE_ANY :
4454 XML_CTYPE_EMPTY);
4455 *eventEndPP = s;
4456 elementDeclHandler(handlerArg, declElementType->name, content);
4457 handleDefault = XML_FALSE;
4458 }
4459 dtd->in_eldecl = XML_FALSE;
4460 }
4461 break;
4462
4463 case XML_ROLE_CONTENT_PCDATA:
4464 if (dtd->in_eldecl) {
4465 dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type
4466 = XML_CTYPE_MIXED;
4467 if (elementDeclHandler)
4468 handleDefault = XML_FALSE;
4469 }
4470 break;
4471
4472 case XML_ROLE_CONTENT_ELEMENT:
4473 quant = XML_CQUANT_NONE;
4474 goto elementContent;
4475 case XML_ROLE_CONTENT_ELEMENT_OPT:
4476 quant = XML_CQUANT_OPT;
4477 goto elementContent;
4478 case XML_ROLE_CONTENT_ELEMENT_REP:
4479 quant = XML_CQUANT_REP;
4480 goto elementContent;
4481 case XML_ROLE_CONTENT_ELEMENT_PLUS:
4482 quant = XML_CQUANT_PLUS;
4483 elementContent:
4484 if (dtd->in_eldecl) {
4485 ELEMENT_TYPE *el;
4486 const XML_Char *name;
4487 int nameLen;
4488 const char *nxt = (quant == XML_CQUANT_NONE
4489 ? next
4490 : next - enc->minBytesPerChar);
4491 int myindex = nextScaffoldPart(parser);
4492 if (myindex < 0)
4493 return XML_ERROR_NO_MEMORY;
4494 dtd->scaffold[myindex].type = XML_CTYPE_NAME;
4495 dtd->scaffold[myindex].quant = quant;
4496 el = getElementType(parser, enc, s, nxt);
4497 if (!el)
4498 return XML_ERROR_NO_MEMORY;
4499 name = el->name;
4500 dtd->scaffold[myindex].name = name;
4501 nameLen = 0;
4502 for (; name[nameLen++]; );
4503 dtd->contentStringLen += nameLen;
4504 if (elementDeclHandler)
4505 handleDefault = XML_FALSE;
4506 }
4507 break;
4508
4509 case XML_ROLE_GROUP_CLOSE:
4510 quant = XML_CQUANT_NONE;
4511 goto closeGroup;
4512 case XML_ROLE_GROUP_CLOSE_OPT:
4513 quant = XML_CQUANT_OPT;
4514 goto closeGroup;
4515 case XML_ROLE_GROUP_CLOSE_REP:
4516 quant = XML_CQUANT_REP;
4517 goto closeGroup;
4518 case XML_ROLE_GROUP_CLOSE_PLUS:
4519 quant = XML_CQUANT_PLUS;
4520 closeGroup:
4521 if (dtd->in_eldecl) {
4522 if (elementDeclHandler)
4523 handleDefault = XML_FALSE;
4524 dtd->scaffLevel--;
4525 dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel]].quant = quant;
4526 if (dtd->scaffLevel == 0) {
4527 if (!handleDefault) {
4528 XML_Content *model = build_model(parser);
4529 if (!model)
4530 return XML_ERROR_NO_MEMORY;
4531 *eventEndPP = s;
4532 elementDeclHandler(handlerArg, declElementType->name, model);
4533 }
4534 dtd->in_eldecl = XML_FALSE;
4535 dtd->contentStringLen = 0;
4536 }
4537 }
4538 break;
4539 /* End element declaration stuff */
4540
4541 case XML_ROLE_PI:
4542 if (!reportProcessingInstruction(parser, enc, s, next))
4543 return XML_ERROR_NO_MEMORY;
4544 handleDefault = XML_FALSE;
4545 break;
4546 case XML_ROLE_COMMENT:
4547 if (!reportComment(parser, enc, s, next))
4548 return XML_ERROR_NO_MEMORY;
4549 handleDefault = XML_FALSE;
4550 break;
4551 case XML_ROLE_NONE:
4552 switch (tok) {
4553 case XML_TOK_BOM:
4554 handleDefault = XML_FALSE;
4555 break;
4556 }
4557 break;
4558 case XML_ROLE_DOCTYPE_NONE:
4559 if (startDoctypeDeclHandler)
4560 handleDefault = XML_FALSE;
4561 break;
4562 case XML_ROLE_ENTITY_NONE:
4563 if (dtd->keepProcessing && entityDeclHandler)
4564 handleDefault = XML_FALSE;
4565 break;
4566 case XML_ROLE_NOTATION_NONE:
4567 if (notationDeclHandler)
4568 handleDefault = XML_FALSE;
4569 break;
4570 case XML_ROLE_ATTLIST_NONE:
4571 if (dtd->keepProcessing && attlistDeclHandler)
4572 handleDefault = XML_FALSE;
4573 break;
4574 case XML_ROLE_ELEMENT_NONE:
4575 if (elementDeclHandler)
4576 handleDefault = XML_FALSE;
4577 break;
4578 } /* end of big switch */
4579
4580 if (handleDefault && defaultHandler)
4581 reportDefault(parser, enc, s, next);
4582
4583 switch (ps_parsing) {
4584 case XML_SUSPENDED:
4585 *nextPtr = next;
4586 return XML_ERROR_NONE;
4587 case XML_FINISHED:
4588 return XML_ERROR_ABORTED;
4589 default:
4590 s = next;
4591 tok = XmlPrologTok(enc, s, end, &next);
4592 }
4593 }
4594 /* not reached */
4595}
4596
4597static enum XML_Error PTRCALL
4598epilogProcessor(XML_Parser parser,
4599 const char *s,
4600 const char *end,
4601 const char **nextPtr)
4602{
4603 processor = epilogProcessor;
4604 eventPtr = s;
4605 for (;;) {
4606 const char *next = NULL;
4607 int tok = XmlPrologTok(encoding, s, end, &next);
4608 eventEndPtr = next;
4609 switch (tok) {
4610 /* report partial linebreak - it might be the last token */
4611 case -XML_TOK_PROLOG_S:
4612 if (defaultHandler) {
4613 reportDefault(parser, encoding, s, next);
4614 if (ps_parsing == XML_FINISHED)
4615 return XML_ERROR_ABORTED;
4616 }
4617 *nextPtr = next;
4618 return XML_ERROR_NONE;
4619 case XML_TOK_NONE:
4620 *nextPtr = s;
4621 return XML_ERROR_NONE;
4622 case XML_TOK_PROLOG_S:
4623 if (defaultHandler)
4624 reportDefault(parser, encoding, s, next);
4625 break;
4626 case XML_TOK_PI:
4627 if (!reportProcessingInstruction(parser, encoding, s, next))
4628 return XML_ERROR_NO_MEMORY;
4629 break;
4630 case XML_TOK_COMMENT:
4631 if (!reportComment(parser, encoding, s, next))
4632 return XML_ERROR_NO_MEMORY;
4633 break;
4634 case XML_TOK_INVALID:
4635 eventPtr = next;
4636 return XML_ERROR_INVALID_TOKEN;
4637 case XML_TOK_PARTIAL:
4638 if (!ps_finalBuffer) {
4639 *nextPtr = s;
4640 return XML_ERROR_NONE;
4641 }
4642 return XML_ERROR_UNCLOSED_TOKEN;
4643 case XML_TOK_PARTIAL_CHAR:
4644 if (!ps_finalBuffer) {
4645 *nextPtr = s;
4646 return XML_ERROR_NONE;
4647 }
4648 return XML_ERROR_PARTIAL_CHAR;
4649 default:
4650 return XML_ERROR_JUNK_AFTER_DOC_ELEMENT;
4651 }
4652 eventPtr = s = next;
4653 switch (ps_parsing) {
4654 case XML_SUSPENDED:
4655 *nextPtr = next;
4656 return XML_ERROR_NONE;
4657 case XML_FINISHED:
4658 return XML_ERROR_ABORTED;
4659 default: ;
4660 }
4661 }
4662}
4663
4664static enum XML_Error
4665processInternalEntity(XML_Parser parser, ENTITY *entity,
4666 XML_Bool betweenDecl)
4667{
4668 const char *textStart, *textEnd;
4669 const char *next;
4670 enum XML_Error result;
4671 OPEN_INTERNAL_ENTITY *openEntity;
4672
4673 if (freeInternalEntities) {
4674 openEntity = freeInternalEntities;
4675 freeInternalEntities = openEntity->next;
4676 }
4677 else {
4678 openEntity = (OPEN_INTERNAL_ENTITY *)MALLOC(sizeof(OPEN_INTERNAL_ENTITY));
4679 if (!openEntity)
4680 return XML_ERROR_NO_MEMORY;
4681 }
4682 entity->open = XML_TRUE;
4683 entity->processed = 0;
4684 openEntity->next = openInternalEntities;
4685 openInternalEntities = openEntity;
4686 openEntity->entity = entity;
4687 openEntity->startTagLevel = tagLevel;
4688 openEntity->betweenDecl = betweenDecl;
4689 openEntity->internalEventPtr = NULL;
4690 openEntity->internalEventEndPtr = NULL;
4691 textStart = (char *)entity->textPtr;
4692 textEnd = (char *)(entity->textPtr + entity->textLen);
4693
4694#ifdef XML_DTD
4695 if (entity->is_param) {
4696 int tok = XmlPrologTok(internalEncoding, textStart, textEnd, &next);
4697 result = doProlog(parser, internalEncoding, textStart, textEnd, tok,
4698 next, &next, XML_FALSE);
4699 }
4700 else
4701#endif /* XML_DTD */
4702 result = doContent(parser, tagLevel, internalEncoding, textStart,
4703 textEnd, &next, XML_FALSE);
4704
4705 if (result == XML_ERROR_NONE) {
4706 if (textEnd != next && ps_parsing == XML_SUSPENDED) {
4707 entity->processed = (int)(next - textStart);
4708 processor = internalEntityProcessor;
4709 }
4710 else {
4711 entity->open = XML_FALSE;
4712 openInternalEntities = openEntity->next;
4713 /* put openEntity back in list of free instances */
4714 openEntity->next = freeInternalEntities;
4715 freeInternalEntities = openEntity;
4716 }
4717 }
4718 return result;
4719}
4720
4721static enum XML_Error PTRCALL
4722internalEntityProcessor(XML_Parser parser,
4723 const char *s,
4724 const char *end,
4725 const char **nextPtr)
4726{
4727 ENTITY *entity;
4728 const char *textStart, *textEnd;
4729 const char *next;
4730 enum XML_Error result;
4731 OPEN_INTERNAL_ENTITY *openEntity = openInternalEntities;
4732 if (!openEntity)
4733 return XML_ERROR_UNEXPECTED_STATE;
4734
4735 entity = openEntity->entity;
4736 textStart = ((char *)entity->textPtr) + entity->processed;
4737 textEnd = (char *)(entity->textPtr + entity->textLen);
4738
4739#ifdef XML_DTD
4740 if (entity->is_param) {
4741 int tok = XmlPrologTok(internalEncoding, textStart, textEnd, &next);
4742 result = doProlog(parser, internalEncoding, textStart, textEnd, tok,
4743 next, &next, XML_FALSE);
4744 }
4745 else
4746#endif /* XML_DTD */
4747 result = doContent(parser, openEntity->startTagLevel, internalEncoding,
4748 textStart, textEnd, &next, XML_FALSE);
4749
4750 if (result != XML_ERROR_NONE)
4751 return result;
4752 else if (textEnd != next && ps_parsing == XML_SUSPENDED) {
4753 entity->processed = (int)(next - (char *)entity->textPtr);
4754 return result;
4755 }
4756 else {
4757 entity->open = XML_FALSE;
4758 openInternalEntities = openEntity->next;
4759 /* put openEntity back in list of free instances */
4760 openEntity->next = freeInternalEntities;
4761 freeInternalEntities = openEntity;
4762 }
4763
4764#ifdef XML_DTD
4765 if (entity->is_param) {
4766 int tok;
4767 processor = prologProcessor;
4768 tok = XmlPrologTok(encoding, s, end, &next);
4769 return doProlog(parser, encoding, s, end, tok, next, nextPtr,
4770 (XML_Bool)!ps_finalBuffer);
4771 }
4772 else
4773#endif /* XML_DTD */
4774 {
4775 processor = contentProcessor;
4776 /* see externalEntityContentProcessor vs contentProcessor */
4777 return doContent(parser, parentParser ? 1 : 0, encoding, s, end,
4778 nextPtr, (XML_Bool)!ps_finalBuffer);
4779 }
4780}
4781
4782static enum XML_Error PTRCALL
4783errorProcessor(XML_Parser parser,
4784 const char *s,
4785 const char *end,
4786 const char **nextPtr)
4787{
4788 return errorCode;
4789}
4790
4791static enum XML_Error
4792storeAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
4793 const char *ptr, const char *end,
4794 STRING_POOL *pool)
4795{
4796 enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr,
4797 end, pool);
4798 if (result)
4799 return result;
4800 if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20)
4801 poolChop(pool);
4802 if (!poolAppendChar(pool, XML_T('\0')))
4803 return XML_ERROR_NO_MEMORY;
4804 return XML_ERROR_NONE;
4805}
4806
4807static enum XML_Error
4808appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
4809 const char *ptr, const char *end,
4810 STRING_POOL *pool)
4811{
4812 DTD * const dtd = _dtd; /* save one level of indirection */
4813 for (;;) {
4814 const char *next;
4815 int tok = XmlAttributeValueTok(enc, ptr, end, &next);
4816 switch (tok) {
4817 case XML_TOK_NONE:
4818 return XML_ERROR_NONE;
4819 case XML_TOK_INVALID:
4820 if (enc == encoding)
4821 eventPtr = next;
4822 return XML_ERROR_INVALID_TOKEN;
4823 case XML_TOK_PARTIAL:
4824 if (enc == encoding)
4825 eventPtr = ptr;
4826 return XML_ERROR_INVALID_TOKEN;
4827 case XML_TOK_CHAR_REF:
4828 {
4829 XML_Char buf[XML_ENCODE_MAX];
4830 int i;
4831 int n = XmlCharRefNumber(enc, ptr);
4832 if (n < 0) {
4833 if (enc == encoding)
4834 eventPtr = ptr;
4835 return XML_ERROR_BAD_CHAR_REF;
4836 }
4837 if (!isCdata
4838 && n == 0x20 /* space */
4839 && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
4840 break;
4841 n = XmlEncode(n, (ICHAR *)buf);
4842 if (!n) {
4843 if (enc == encoding)
4844 eventPtr = ptr;
4845 return XML_ERROR_BAD_CHAR_REF;
4846 }
4847 for (i = 0; i < n; i++) {
4848 if (!poolAppendChar(pool, buf[i]))
4849 return XML_ERROR_NO_MEMORY;
4850 }
4851 }
4852 break;
4853 case XML_TOK_DATA_CHARS:
4854 if (!poolAppend(pool, enc, ptr, next))
4855 return XML_ERROR_NO_MEMORY;
4856 break;
4857 case XML_TOK_TRAILING_CR:
4858 next = ptr + enc->minBytesPerChar;
4859 /* fall through */
4860 case XML_TOK_ATTRIBUTE_VALUE_S:
4861 case XML_TOK_DATA_NEWLINE:
4862 if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
4863 break;
4864 if (!poolAppendChar(pool, 0x20))
4865 return XML_ERROR_NO_MEMORY;
4866 break;
4867 case XML_TOK_ENTITY_REF:
4868 {
4869 const XML_Char *name;
4870 ENTITY *entity;
4871 char checkEntityDecl;
4872 XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc,
4873 ptr + enc->minBytesPerChar,
4874 next - enc->minBytesPerChar);
4875 if (ch) {
4876 if (!poolAppendChar(pool, ch))
4877 return XML_ERROR_NO_MEMORY;
4878 break;
4879 }
4880 name = poolStoreString(&temp2Pool, enc,
4881 ptr + enc->minBytesPerChar,
4882 next - enc->minBytesPerChar);
4883 if (!name)
4884 return XML_ERROR_NO_MEMORY;
4885 entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0);
4886 poolDiscard(&temp2Pool);
4887 /* First, determine if a check for an existing declaration is needed;
4888 if yes, check that the entity exists, and that it is internal.
4889 */
4890 if (pool == &dtd->pool) /* are we called from prolog? */
4891 checkEntityDecl =
4892#ifdef XML_DTD
4893 prologState.documentEntity &&
4894#endif /* XML_DTD */
4895 (dtd->standalone
4896 ? !openInternalEntities
4897 : !dtd->hasParamEntityRefs);
4898 else /* if (pool == &tempPool): we are called from content */
4899 checkEntityDecl = !dtd->hasParamEntityRefs || dtd->standalone;
4900 if (checkEntityDecl) {
4901 if (!entity)
4902 return XML_ERROR_UNDEFINED_ENTITY;
4903 else if (!entity->is_internal)
4904 return XML_ERROR_ENTITY_DECLARED_IN_PE;
4905 }
4906 else if (!entity) {
4907 /* Cannot report skipped entity here - see comments on
4908 skippedEntityHandler.
4909 if (skippedEntityHandler)
4910 skippedEntityHandler(handlerArg, name, 0);
4911 */
4912 /* Cannot call the default handler because this would be
4913 out of sync with the call to the startElementHandler.
4914 if ((pool == &tempPool) && defaultHandler)
4915 reportDefault(parser, enc, ptr, next);
4916 */
4917 break;
4918 }
4919 if (entity->open) {
4920 if (enc == encoding)
4921 eventPtr = ptr;
4922 return XML_ERROR_RECURSIVE_ENTITY_REF;
4923 }
4924 if (entity->notation) {
4925 if (enc == encoding)
4926 eventPtr = ptr;
4927 return XML_ERROR_BINARY_ENTITY_REF;
4928 }
4929 if (!entity->textPtr) {
4930 if (enc == encoding)
4931 eventPtr = ptr;
4932 return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF;
4933 }
4934 else {
4935 enum XML_Error result;
4936 const XML_Char *textEnd = entity->textPtr + entity->textLen;
4937 entity->open = XML_TRUE;
4938 result = appendAttributeValue(parser, internalEncoding, isCdata,
4939 (char *)entity->textPtr,
4940 (char *)textEnd, pool);
4941 entity->open = XML_FALSE;
4942 if (result)
4943 return result;
4944 }
4945 }
4946 break;
4947 default:
4948 if (enc == encoding)
4949 eventPtr = ptr;
4950 return XML_ERROR_UNEXPECTED_STATE;
4951 }
4952 ptr = next;
4953 }
4954 /* not reached */
4955}
4956
4957static enum XML_Error
4958storeEntityValue(XML_Parser parser,
4959 const ENCODING *enc,
4960 const char *entityTextPtr,
4961 const char *entityTextEnd)
4962{
4963 DTD * const dtd = _dtd; /* save one level of indirection */
4964 STRING_POOL *pool = &(dtd->entityValuePool);
4965 enum XML_Error result = XML_ERROR_NONE;
4966#ifdef XML_DTD
4967 int oldInEntityValue = prologState.inEntityValue;
4968 prologState.inEntityValue = 1;
4969#endif /* XML_DTD */
4970 /* never return Null for the value argument in EntityDeclHandler,
4971 since this would indicate an external entity; therefore we
4972 have to make sure that entityValuePool.start is not null */
4973 if (!pool->blocks) {
4974 if (!poolGrow(pool))
4975 return XML_ERROR_NO_MEMORY;
4976 }
4977
4978 for (;;) {
4979 const char *next;
4980 int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next);
4981 switch (tok) {
4982 case XML_TOK_PARAM_ENTITY_REF:
4983#ifdef XML_DTD
4984 if (isParamEntity || enc != encoding) {
4985 const XML_Char *name;
4986 ENTITY *entity;
4987 name = poolStoreString(&tempPool, enc,
4988 entityTextPtr + enc->minBytesPerChar,
4989 next - enc->minBytesPerChar);
4990 if (!name) {
4991 result = XML_ERROR_NO_MEMORY;
4992 goto endEntityValue;
4993 }
4994 entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0);
4995 poolDiscard(&tempPool);
4996 if (!entity) {
4997 /* not a well-formedness error - see XML 1.0: WFC Entity Declared */
4998 /* cannot report skipped entity here - see comments on
4999 skippedEntityHandler
5000 if (skippedEntityHandler)
5001 skippedEntityHandler(handlerArg, name, 0);
5002 */
5003 dtd->keepProcessing = dtd->standalone;
5004 goto endEntityValue;
5005 }
5006 if (entity->open) {
5007 if (enc == encoding)
5008 eventPtr = entityTextPtr;
5009 result = XML_ERROR_RECURSIVE_ENTITY_REF;
5010 goto endEntityValue;
5011 }
5012 if (entity->systemId) {
5013 if (externalEntityRefHandler) {
5014 dtd->paramEntityRead = XML_FALSE;
5015 entity->open = XML_TRUE;
5016 if (!externalEntityRefHandler(externalEntityRefHandlerArg,
5017 0,
5018 entity->base,
5019 entity->systemId,
5020 entity->publicId)) {
5021 entity->open = XML_FALSE;
5022 result = XML_ERROR_EXTERNAL_ENTITY_HANDLING;
5023 goto endEntityValue;
5024 }
5025 entity->open = XML_FALSE;
5026 if (!dtd->paramEntityRead)
5027 dtd->keepProcessing = dtd->standalone;
5028 }
5029 else
5030 dtd->keepProcessing = dtd->standalone;
5031 }
5032 else {
5033 entity->open = XML_TRUE;
5034 result = storeEntityValue(parser,
5035 internalEncoding,
5036 (char *)entity->textPtr,
5037 (char *)(entity->textPtr
5038 + entity->textLen));
5039 entity->open = XML_FALSE;
5040 if (result)
5041 goto endEntityValue;
5042 }
5043 break;
5044 }
5045#endif /* XML_DTD */
5046 /* In the internal subset, PE references are not legal
5047 within markup declarations, e.g entity values in this case. */
5048 eventPtr = entityTextPtr;
5049 result = XML_ERROR_PARAM_ENTITY_REF;
5050 goto endEntityValue;
5051 case XML_TOK_NONE:
5052 result = XML_ERROR_NONE;
5053 goto endEntityValue;
5054 case XML_TOK_ENTITY_REF:
5055 case XML_TOK_DATA_CHARS:
5056 if (!poolAppend(pool, enc, entityTextPtr, next)) {
5057 result = XML_ERROR_NO_MEMORY;
5058 goto endEntityValue;
5059 }
5060 break;
5061 case XML_TOK_TRAILING_CR:
5062 next = entityTextPtr + enc->minBytesPerChar;
5063 /* fall through */
5064 case XML_TOK_DATA_NEWLINE:
5065 if (pool->end == pool->ptr && !poolGrow(pool)) {
5066 result = XML_ERROR_NO_MEMORY;
5067 goto endEntityValue;
5068 }
5069 *(pool->ptr)++ = 0xA;
5070 break;
5071 case XML_TOK_CHAR_REF:
5072 {
5073 XML_Char buf[XML_ENCODE_MAX];
5074 int i;
5075 int n = XmlCharRefNumber(enc, entityTextPtr);
5076 if (n < 0) {
5077 if (enc == encoding)
5078 eventPtr = entityTextPtr;
5079 result = XML_ERROR_BAD_CHAR_REF;
5080 goto endEntityValue;
5081 }
5082 n = XmlEncode(n, (ICHAR *)buf);
5083 if (!n) {
5084 if (enc == encoding)
5085 eventPtr = entityTextPtr;
5086 result = XML_ERROR_BAD_CHAR_REF;
5087 goto endEntityValue;
5088 }
5089 for (i = 0; i < n; i++) {
5090 if (pool->end == pool->ptr && !poolGrow(pool)) {
5091 result = XML_ERROR_NO_MEMORY;
5092 goto endEntityValue;
5093 }
5094 *(pool->ptr)++ = buf[i];
5095 }
5096 }
5097 break;
5098 case XML_TOK_PARTIAL:
5099 if (enc == encoding)
5100 eventPtr = entityTextPtr;
5101 result = XML_ERROR_INVALID_TOKEN;
5102 goto endEntityValue;
5103 case XML_TOK_INVALID:
5104 if (enc == encoding)
5105 eventPtr = next;
5106 result = XML_ERROR_INVALID_TOKEN;
5107 goto endEntityValue;
5108 default:
5109 if (enc == encoding)
5110 eventPtr = entityTextPtr;
5111 result = XML_ERROR_UNEXPECTED_STATE;
5112 goto endEntityValue;
5113 }
5114 entityTextPtr = next;
5115 }
5116endEntityValue:
5117#ifdef XML_DTD
5118 prologState.inEntityValue = oldInEntityValue;
5119#endif /* XML_DTD */
5120 return result;
5121}
5122
5123static void FASTCALL
5124normalizeLines(XML_Char *s)
5125{
5126 XML_Char *p;
5127 for (;; s++) {
5128 if (*s == XML_T('\0'))
5129 return;
5130 if (*s == 0xD)
5131 break;
5132 }
5133 p = s;
5134 do {
5135 if (*s == 0xD) {
5136 *p++ = 0xA;
5137 if (*++s == 0xA)
5138 s++;
5139 }
5140 else
5141 *p++ = *s++;
5142 } while (*s);
5143 *p = XML_T('\0');
5144}
5145
5146static int
5147reportProcessingInstruction(XML_Parser parser, const ENCODING *enc,
5148 const char *start, const char *end)
5149{
5150 const XML_Char *target;
5151 XML_Char *data;
5152 const char *tem;
5153 if (!processingInstructionHandler) {
5154 if (defaultHandler)
5155 reportDefault(parser, enc, start, end);
5156 return 1;
5157 }
5158 start += enc->minBytesPerChar * 2;
5159 tem = start + XmlNameLength(enc, start);
5160 target = poolStoreString(&tempPool, enc, start, tem);
5161 if (!target)
5162 return 0;
5163 poolFinish(&tempPool);
5164 data = poolStoreString(&tempPool, enc,
5165 XmlSkipS(enc, tem),
5166 end - enc->minBytesPerChar*2);
5167 if (!data)
5168 return 0;
5169 normalizeLines(data);
5170 processingInstructionHandler(handlerArg, target, data);
5171 poolClear(&tempPool);
5172 return 1;
5173}
5174
5175static int
5176reportComment(XML_Parser parser, const ENCODING *enc,
5177 const char *start, const char *end)
5178{
5179 XML_Char *data;
5180 if (!commentHandler) {
5181 if (defaultHandler)
5182 reportDefault(parser, enc, start, end);
5183 return 1;
5184 }
5185 data = poolStoreString(&tempPool,
5186 enc,
5187 start + enc->minBytesPerChar * 4,
5188 end - enc->minBytesPerChar * 3);
5189 if (!data)
5190 return 0;
5191 normalizeLines(data);
5192 commentHandler(handlerArg, data);
5193 poolClear(&tempPool);
5194 return 1;
5195}
5196
5197static void
5198reportDefault(XML_Parser parser, const ENCODING *enc,
5199 const char *s, const char *end)
5200{
5201 if (MUST_CONVERT(enc, s)) {
5202 const char **eventPP;
5203 const char **eventEndPP;
5204 if (enc == encoding) {
5205 eventPP = &eventPtr;
5206 eventEndPP = &eventEndPtr;
5207 }
5208 else {
5209 eventPP = &(openInternalEntities->internalEventPtr);
5210 eventEndPP = &(openInternalEntities->internalEventEndPtr);
5211 }
5212 do {
5213 ICHAR *dataPtr = (ICHAR *)dataBuf;
5214 XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd);
5215 *eventEndPP = s;
5216 defaultHandler(handlerArg, dataBuf, (int)(dataPtr - (ICHAR *)dataBuf));
5217 *eventPP = s;
5218 } while (s != end);
5219 }
5220 else
5221 defaultHandler(handlerArg, (XML_Char *)s, (int)((XML_Char *)end - (XML_Char *)s));
5222}
5223
5224
5225static int
5226defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, XML_Bool isCdata,
5227 XML_Bool isId, const XML_Char *value, XML_Parser parser)
5228{
5229 DEFAULT_ATTRIBUTE *att;
5230 if (value || isId) {
5231 /* The handling of default attributes gets messed up if we have
5232 a default which duplicates a non-default. */
5233 int i;
5234 for (i = 0; i < type->nDefaultAtts; i++)
5235 if (attId == type->defaultAtts[i].id)
5236 return 1;
5237 if (isId && !type->idAtt && !attId->xmlns)
5238 type->idAtt = attId;
5239 }
5240 if (type->nDefaultAtts == type->allocDefaultAtts) {
5241 if (type->allocDefaultAtts == 0) {
5242 type->allocDefaultAtts = 8;
5243 type->defaultAtts = (DEFAULT_ATTRIBUTE *)MALLOC(type->allocDefaultAtts
5244 * sizeof(DEFAULT_ATTRIBUTE));
5245 if (!type->defaultAtts)
5246 return 0;
5247 }
5248 else {
5249 DEFAULT_ATTRIBUTE *temp;
5250 int count = type->allocDefaultAtts * 2;
5251 temp = (DEFAULT_ATTRIBUTE *)
5252 REALLOC(type->defaultAtts, (count * sizeof(DEFAULT_ATTRIBUTE)));
5253 if (temp == NULL)
5254 return 0;
5255 type->allocDefaultAtts = count;
5256 type->defaultAtts = temp;
5257 }
5258 }
5259 att = type->defaultAtts + type->nDefaultAtts;
5260 att->id = attId;
5261 att->value = value;
5262 att->isCdata = isCdata;
5263 if (!isCdata)
5264 attId->maybeTokenized = XML_TRUE;
5265 type->nDefaultAtts += 1;
5266 return 1;
5267}
5268
5269static int
5270setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType)
5271{
5272 DTD * const dtd = _dtd; /* save one level of indirection */
5273 const XML_Char *name;
5274 for (name = elementType->name; *name; name++) {
5275 if (*name == XML_T(':')) {
5276 PREFIX *prefix;
5277 const XML_Char *s;
5278 for (s = elementType->name; s != name; s++) {
5279 if (!poolAppendChar(&dtd->pool, *s))
5280 return 0;
5281 }
5282 if (!poolAppendChar(&dtd->pool, XML_T('\0')))
5283 return 0;
5284 prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool),
5285 sizeof(PREFIX));
5286 if (!prefix)
5287 return 0;
5288 if (prefix->name == poolStart(&dtd->pool))
5289 poolFinish(&dtd->pool);
5290 else
5291 poolDiscard(&dtd->pool);
5292 elementType->prefix = prefix;
5293
5294 }
5295 }
5296 return 1;
5297}
5298
5299static ATTRIBUTE_ID *
5300getAttributeId(XML_Parser parser, const ENCODING *enc,
5301 const char *start, const char *end)
5302{
5303 DTD * const dtd = _dtd; /* save one level of indirection */
5304 ATTRIBUTE_ID *id;
5305 const XML_Char *name;
5306 if (!poolAppendChar(&dtd->pool, XML_T('\0')))
5307 return NULL;
5308 name = poolStoreString(&dtd->pool, enc, start, end);
5309 if (!name)
5310 return NULL;
5311 /* skip quotation mark - its storage will be re-used (like in name[-1]) */
5312 ++name;
5313 id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, name, sizeof(ATTRIBUTE_ID));
5314 if (!id)
5315 return NULL;
5316 if (id->name != name)
5317 poolDiscard(&dtd->pool);
5318 else {
5319 poolFinish(&dtd->pool);
5320 if (!ns)
5321 ;
5322 else if (name[0] == XML_T('x')
5323 && name[1] == XML_T('m')
5324 && name[2] == XML_T('l')
5325 && name[3] == XML_T('n')
5326 && name[4] == XML_T('s')
5327 && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) {
5328 if (name[5] == XML_T('\0'))
5329 id->prefix = &dtd->defaultPrefix;
5330 else
5331 id->prefix = (PREFIX *)lookup(&dtd->prefixes, name + 6, sizeof(PREFIX));
5332 id->xmlns = XML_TRUE;
5333 }
5334 else {
5335 int i;
5336 for (i = 0; name[i]; i++) {
5337 /* attributes without prefix are *not* in the default namespace */
5338 if (name[i] == XML_T(':')) {
5339 int j;
5340 for (j = 0; j < i; j++) {
5341 if (!poolAppendChar(&dtd->pool, name[j]))
5342 return NULL;
5343 }
5344 if (!poolAppendChar(&dtd->pool, XML_T('\0')))
5345 return NULL;
5346 id->prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool),
5347 sizeof(PREFIX));
5348 if (!id->prefix)
5349 return NULL;
5350 if (id->prefix->name == poolStart(&dtd->pool))
5351 poolFinish(&dtd->pool);
5352 else
5353 poolDiscard(&dtd->pool);
5354 break;
5355 }
5356 }
5357 }
5358 }
5359 return id;
5360}
5361
5362#define CONTEXT_SEP XML_T('\f')
5363
5364static const XML_Char *
5365getContext(XML_Parser parser)
5366{
5367 DTD * const dtd = _dtd; /* save one level of indirection */
5368 HASH_TABLE_ITER iter;
5369 XML_Bool needSep = XML_FALSE;
5370
5371 if (dtd->defaultPrefix.binding) {
5372 int i;
5373 int len;
5374 if (!poolAppendChar(&tempPool, XML_T('=')))
5375 return NULL;
5376 len = dtd->defaultPrefix.binding->uriLen;
5377 if (namespaceSeparator)
5378 len--;
5379 for (i = 0; i < len; i++)
5380 if (!poolAppendChar(&tempPool, dtd->defaultPrefix.binding->uri[i]))
5381 return NULL;
5382 needSep = XML_TRUE;
5383 }
5384
5385 hashTableIterInit(&iter, &(dtd->prefixes));
5386 for (;;) {
5387 int i;
5388 int len;
5389 const XML_Char *s;
5390 PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter);
5391 if (!prefix)
5392 break;
5393 if (!prefix->binding)
5394 continue;
5395 if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
5396 return NULL;
5397 for (s = prefix->name; *s; s++)
5398 if (!poolAppendChar(&tempPool, *s))
5399 return NULL;
5400 if (!poolAppendChar(&tempPool, XML_T('=')))
5401 return NULL;
5402 len = prefix->binding->uriLen;
5403 if (namespaceSeparator)
5404 len--;
5405 for (i = 0; i < len; i++)
5406 if (!poolAppendChar(&tempPool, prefix->binding->uri[i]))
5407 return NULL;
5408 needSep = XML_TRUE;
5409 }
5410
5411
5412 hashTableIterInit(&iter, &(dtd->generalEntities));
5413 for (;;) {
5414 const XML_Char *s;
5415 ENTITY *e = (ENTITY *)hashTableIterNext(&iter);
5416 if (!e)
5417 break;
5418 if (!e->open)
5419 continue;
5420 if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
5421 return NULL;
5422 for (s = e->name; *s; s++)
5423 if (!poolAppendChar(&tempPool, *s))
5424 return 0;
5425 needSep = XML_TRUE;
5426 }
5427
5428 if (!poolAppendChar(&tempPool, XML_T('\0')))
5429 return NULL;
5430 return tempPool.start;
5431}
5432
5433static XML_Bool
5434setContext(XML_Parser parser, const XML_Char *context)
5435{
5436 DTD * const dtd = _dtd; /* save one level of indirection */
5437 const XML_Char *s = context;
5438
5439 while (*context != XML_T('\0')) {
5440 if (*s == CONTEXT_SEP || *s == XML_T('\0')) {
5441 ENTITY *e;
5442 if (!poolAppendChar(&tempPool, XML_T('\0')))
5443 return XML_FALSE;
5444 e = (ENTITY *)lookup(&dtd->generalEntities, poolStart(&tempPool), 0);
5445 if (e)
5446 e->open = XML_TRUE;
5447 if (*s != XML_T('\0'))
5448 s++;
5449 context = s;
5450 poolDiscard(&tempPool);
5451 }
5452 else if (*s == XML_T('=')) {
5453 PREFIX *prefix;
5454 if (poolLength(&tempPool) == 0)
5455 prefix = &dtd->defaultPrefix;
5456 else {
5457 if (!poolAppendChar(&tempPool, XML_T('\0')))
5458 return XML_FALSE;
5459 prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&tempPool),
5460 sizeof(PREFIX));
5461 if (!prefix)
5462 return XML_FALSE;
5463 if (prefix->name == poolStart(&tempPool)) {
5464 prefix->name = poolCopyString(&dtd->pool, prefix->name);
5465 if (!prefix->name)
5466 return XML_FALSE;
5467 }
5468 poolDiscard(&tempPool);
5469 }
5470 for (context = s + 1;
5471 *context != CONTEXT_SEP && *context != XML_T('\0');
5472 context++)
5473 if (!poolAppendChar(&tempPool, *context))
5474 return XML_FALSE;
5475 if (!poolAppendChar(&tempPool, XML_T('\0')))
5476 return XML_FALSE;
5477 if (addBinding(parser, prefix, NULL, poolStart(&tempPool),
5478 &inheritedBindings) != XML_ERROR_NONE)
5479 return XML_FALSE;
5480 poolDiscard(&tempPool);
5481 if (*context != XML_T('\0'))
5482 ++context;
5483 s = context;
5484 }
5485 else {
5486 if (!poolAppendChar(&tempPool, *s))
5487 return XML_FALSE;
5488 s++;
5489 }
5490 }
5491 return XML_TRUE;
5492}
5493
5494static void FASTCALL
5495normalizePublicId(XML_Char *publicId)
5496{
5497 XML_Char *p = publicId;
5498 XML_Char *s;
5499 for (s = publicId; *s; s++) {
5500 switch (*s) {
5501 case 0x20:
5502 case 0xD:
5503 case 0xA:
5504 if (p != publicId && p[-1] != 0x20)
5505 *p++ = 0x20;
5506 break;
5507 default:
5508 *p++ = *s;
5509 }
5510 }
5511 if (p != publicId && p[-1] == 0x20)
5512 --p;
5513 *p = XML_T('\0');
5514}
5515
5516static DTD *
5517dtdCreate(const XML_Memory_Handling_Suite *ms)
5518{
5519 DTD *p = (DTD *)ms->malloc_fcn(sizeof(DTD));
5520 if (p == NULL)
5521 return p;
5522 poolInit(&(p->pool), ms);
5523 poolInit(&(p->entityValuePool), ms);
5524 hashTableInit(&(p->generalEntities), ms);
5525 hashTableInit(&(p->elementTypes), ms);
5526 hashTableInit(&(p->attributeIds), ms);
5527 hashTableInit(&(p->prefixes), ms);
5528#ifdef XML_DTD
5529 p->paramEntityRead = XML_FALSE;
5530 hashTableInit(&(p->paramEntities), ms);
5531#endif /* XML_DTD */
5532 p->defaultPrefix.name = NULL;
5533 p->defaultPrefix.binding = NULL;
5534
5535 p->in_eldecl = XML_FALSE;
5536 p->scaffIndex = NULL;
5537 p->scaffold = NULL;
5538 p->scaffLevel = 0;
5539 p->scaffSize = 0;
5540 p->scaffCount = 0;
5541 p->contentStringLen = 0;
5542
5543 p->keepProcessing = XML_TRUE;
5544 p->hasParamEntityRefs = XML_FALSE;
5545 p->standalone = XML_FALSE;
5546 return p;
5547}
5548
5549static void
5550dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms)
5551{
5552 HASH_TABLE_ITER iter;
5553 hashTableIterInit(&iter, &(p->elementTypes));
5554 for (;;) {
5555 ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter);
5556 if (!e)
5557 break;
5558 if (e->allocDefaultAtts != 0)
5559 ms->free_fcn(e->defaultAtts);
5560 }
5561 hashTableClear(&(p->generalEntities));
5562#ifdef XML_DTD
5563 p->paramEntityRead = XML_FALSE;
5564 hashTableClear(&(p->paramEntities));
5565#endif /* XML_DTD */
5566 hashTableClear(&(p->elementTypes));
5567 hashTableClear(&(p->attributeIds));
5568 hashTableClear(&(p->prefixes));
5569 poolClear(&(p->pool));
5570 poolClear(&(p->entityValuePool));
5571 p->defaultPrefix.name = NULL;
5572 p->defaultPrefix.binding = NULL;
5573
5574 p->in_eldecl = XML_FALSE;
5575
5576 ms->free_fcn(p->scaffIndex);
5577 p->scaffIndex = NULL;
5578 ms->free_fcn(p->scaffold);
5579 p->scaffold = NULL;
5580
5581 p->scaffLevel = 0;
5582 p->scaffSize = 0;
5583 p->scaffCount = 0;
5584 p->contentStringLen = 0;
5585
5586 p->keepProcessing = XML_TRUE;
5587 p->hasParamEntityRefs = XML_FALSE;
5588 p->standalone = XML_FALSE;
5589}
5590
5591static void
5592dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms)
5593{
5594 HASH_TABLE_ITER iter;
5595 hashTableIterInit(&iter, &(p->elementTypes));
5596 for (;;) {
5597 ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter);
5598 if (!e)
5599 break;
5600 if (e->allocDefaultAtts != 0)
5601 ms->free_fcn(e->defaultAtts);
5602 }
5603 hashTableDestroy(&(p->generalEntities));
5604#ifdef XML_DTD
5605 hashTableDestroy(&(p->paramEntities));
5606#endif /* XML_DTD */
5607 hashTableDestroy(&(p->elementTypes));
5608 hashTableDestroy(&(p->attributeIds));
5609 hashTableDestroy(&(p->prefixes));
5610 poolDestroy(&(p->pool));
5611 poolDestroy(&(p->entityValuePool));
5612 if (isDocEntity) {
5613 ms->free_fcn(p->scaffIndex);
5614 ms->free_fcn(p->scaffold);
5615 }
5616 ms->free_fcn(p);
5617}
5618
5619/* Do a deep copy of the DTD. Return 0 for out of memory, non-zero otherwise.
5620 The new DTD has already been initialized.
5621*/
5622static int
5623dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms)
5624{
5625 HASH_TABLE_ITER iter;
5626
5627 /* Copy the prefix table. */
5628
5629 hashTableIterInit(&iter, &(oldDtd->prefixes));
5630 for (;;) {
5631 const XML_Char *name;
5632 const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter);
5633 if (!oldP)
5634 break;
5635 name = poolCopyString(&(newDtd->pool), oldP->name);
5636 if (!name)
5637 return 0;
5638 if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX)))
5639 return 0;
5640 }
5641
5642 hashTableIterInit(&iter, &(oldDtd->attributeIds));
5643
5644 /* Copy the attribute id table. */
5645
5646 for (;;) {
5647 ATTRIBUTE_ID *newA;
5648 const XML_Char *name;
5649 const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter);
5650
5651 if (!oldA)
5652 break;
5653 /* Remember to allocate the scratch byte before the name. */
5654 if (!poolAppendChar(&(newDtd->pool), XML_T('\0')))
5655 return 0;
5656 name = poolCopyString(&(newDtd->pool), oldA->name);
5657 if (!name)
5658 return 0;
5659 ++name;
5660 newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name,
5661 sizeof(ATTRIBUTE_ID));
5662 if (!newA)
5663 return 0;
5664 newA->maybeTokenized = oldA->maybeTokenized;
5665 if (oldA->prefix) {
5666 newA->xmlns = oldA->xmlns;
5667 if (oldA->prefix == &oldDtd->defaultPrefix)
5668 newA->prefix = &newDtd->defaultPrefix;
5669 else
5670 newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes),
5671 oldA->prefix->name, 0);
5672 }
5673 }
5674
5675 /* Copy the element type table. */
5676
5677 hashTableIterInit(&iter, &(oldDtd->elementTypes));
5678
5679 for (;;) {
5680 int i;
5681 ELEMENT_TYPE *newE;
5682 const XML_Char *name;
5683 const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter);
5684 if (!oldE)
5685 break;
5686 name = poolCopyString(&(newDtd->pool), oldE->name);
5687 if (!name)
5688 return 0;
5689 newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name,
5690 sizeof(ELEMENT_TYPE));
5691 if (!newE)
5692 return 0;
5693 if (oldE->nDefaultAtts) {
5694 newE->defaultAtts = (DEFAULT_ATTRIBUTE *)
5695 ms->malloc_fcn(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE));
5696 if (!newE->defaultAtts) {
5697 ms->free_fcn(newE);
5698 return 0;
5699 }
5700 }
5701 if (oldE->idAtt)
5702 newE->idAtt = (ATTRIBUTE_ID *)
5703 lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0);
5704 newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts;
5705 if (oldE->prefix)
5706 newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes),
5707 oldE->prefix->name, 0);
5708 for (i = 0; i < newE->nDefaultAtts; i++) {
5709 newE->defaultAtts[i].id = (ATTRIBUTE_ID *)
5710 lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0);
5711 newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata;
5712 if (oldE->defaultAtts[i].value) {
5713 newE->defaultAtts[i].value
5714 = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value);
5715 if (!newE->defaultAtts[i].value)
5716 return 0;
5717 }
5718 else
5719 newE->defaultAtts[i].value = NULL;
5720 }
5721 }
5722
5723 /* Copy the entity tables. */
5724 if (!copyEntityTable(&(newDtd->generalEntities),
5725 &(newDtd->pool),
5726 &(oldDtd->generalEntities)))
5727 return 0;
5728
5729#ifdef XML_DTD
5730 if (!copyEntityTable(&(newDtd->paramEntities),
5731 &(newDtd->pool),
5732 &(oldDtd->paramEntities)))
5733 return 0;
5734 newDtd->paramEntityRead = oldDtd->paramEntityRead;
5735#endif /* XML_DTD */
5736
5737 newDtd->keepProcessing = oldDtd->keepProcessing;
5738 newDtd->hasParamEntityRefs = oldDtd->hasParamEntityRefs;
5739 newDtd->standalone = oldDtd->standalone;
5740
5741 /* Don't want deep copying for scaffolding */
5742 newDtd->in_eldecl = oldDtd->in_eldecl;
5743 newDtd->scaffold = oldDtd->scaffold;
5744 newDtd->contentStringLen = oldDtd->contentStringLen;
5745 newDtd->scaffSize = oldDtd->scaffSize;
5746 newDtd->scaffLevel = oldDtd->scaffLevel;
5747 newDtd->scaffIndex = oldDtd->scaffIndex;
5748
5749 return 1;
5750} /* End dtdCopy */
5751
5752static int
5753copyEntityTable(HASH_TABLE *newTable,
5754 STRING_POOL *newPool,
5755 const HASH_TABLE *oldTable)
5756{
5757 HASH_TABLE_ITER iter;
5758 const XML_Char *cachedOldBase = NULL;
5759 const XML_Char *cachedNewBase = NULL;
5760
5761 hashTableIterInit(&iter, oldTable);
5762
5763 for (;;) {
5764 ENTITY *newE;
5765 const XML_Char *name;
5766 const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter);
5767 if (!oldE)
5768 break;
5769 name = poolCopyString(newPool, oldE->name);
5770 if (!name)
5771 return 0;
5772 newE = (ENTITY *)lookup(newTable, name, sizeof(ENTITY));
5773 if (!newE)
5774 return 0;
5775 if (oldE->systemId) {
5776 const XML_Char *tem = poolCopyString(newPool, oldE->systemId);
5777 if (!tem)
5778 return 0;
5779 newE->systemId = tem;
5780 if (oldE->base) {
5781 if (oldE->base == cachedOldBase)
5782 newE->base = cachedNewBase;
5783 else {
5784 cachedOldBase = oldE->base;
5785 tem = poolCopyString(newPool, cachedOldBase);
5786 if (!tem)
5787 return 0;
5788 cachedNewBase = newE->base = tem;
5789 }
5790 }
5791 if (oldE->publicId) {
5792 tem = poolCopyString(newPool, oldE->publicId);
5793 if (!tem)
5794 return 0;
5795 newE->publicId = tem;
5796 }
5797 }
5798 else {
5799 const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr,
5800 oldE->textLen);
5801 if (!tem)
5802 return 0;
5803 newE->textPtr = tem;
5804 newE->textLen = oldE->textLen;
5805 }
5806 if (oldE->notation) {
5807 const XML_Char *tem = poolCopyString(newPool, oldE->notation);
5808 if (!tem)
5809 return 0;
5810 newE->notation = tem;
5811 }
5812 newE->is_param = oldE->is_param;
5813 newE->is_internal = oldE->is_internal;
5814 }
5815 return 1;
5816}
5817
5818#define INIT_POWER 6
5819
5820static XML_Bool FASTCALL
5821keyeq(KEY s1, KEY s2)
5822{
5823 for (; *s1 == *s2; s1++, s2++)
5824 if (*s1 == 0)
5825 return XML_TRUE;
5826 return XML_FALSE;
5827}
5828
5829static unsigned long FASTCALL
5830hash(KEY s)
5831{
5832 unsigned long h = 0;
5833 while (*s)
5834 h = CHAR_HASH(h, *s++);
5835 return h;
5836}
5837
5838static NAMED *
5839lookup(HASH_TABLE *table, KEY name, size_t createSize)
5840{
5841 size_t i;
5842 if (table->size == 0) {
5843 size_t tsize;
5844 if (!createSize)
5845 return NULL;
5846 table->power = INIT_POWER;
5847 /* table->size is a power of 2 */
5848 table->size = (size_t)1 << INIT_POWER;
5849 tsize = table->size * sizeof(NAMED *);
5850 table->v = (NAMED **)table->mem->malloc_fcn(tsize);
5851 if (!table->v) {
5852 table->size = 0;
5853 return NULL;
5854 }
5855 memset(table->v, 0, tsize);
5856 i = hash(name) & ((unsigned long)table->size - 1);
5857 }
5858 else {
5859 unsigned long h = hash(name);
5860 unsigned long mask = (unsigned long)table->size - 1;
5861 unsigned char step = 0;
5862 i = h & mask;
5863 while (table->v[i]) {
5864 if (keyeq(name, table->v[i]->name))
5865 return table->v[i];
5866 if (!step)
5867 step = PROBE_STEP(h, mask, table->power);
5868 i < step ? (i += table->size - step) : (i -= step);
5869 }
5870 if (!createSize)
5871 return NULL;
5872
5873 /* check for overflow (table is half full) */
5874 if (table->used >> (table->power - 1)) {
5875 unsigned char newPower = table->power + 1;
5876 size_t newSize = (size_t)1 << newPower;
5877 unsigned long newMask = (unsigned long)newSize - 1;
5878 size_t tsize = newSize * sizeof(NAMED *);
5879 NAMED **newV = (NAMED **)table->mem->malloc_fcn(tsize);
5880 if (!newV)
5881 return NULL;
5882 memset(newV, 0, tsize);
5883 for (i = 0; i < table->size; i++)
5884 if (table->v[i]) {
5885 unsigned long newHash = hash(table->v[i]->name);
5886 size_t j = newHash & newMask;
5887 step = 0;
5888 while (newV[j]) {
5889 if (!step)
5890 step = PROBE_STEP(newHash, newMask, newPower);
5891 j < step ? (j += newSize - step) : (j -= step);
5892 }
5893 newV[j] = table->v[i];
5894 }
5895 table->mem->free_fcn(table->v);
5896 table->v = newV;
5897 table->power = newPower;
5898 table->size = newSize;
5899 i = h & newMask;
5900 step = 0;
5901 while (table->v[i]) {
5902 if (!step)
5903 step = PROBE_STEP(h, newMask, newPower);
5904 i < step ? (i += newSize - step) : (i -= step);
5905 }
5906 }
5907 }
5908 table->v[i] = (NAMED *)table->mem->malloc_fcn(createSize);
5909 if (!table->v[i])
5910 return NULL;
5911 memset(table->v[i], 0, createSize);
5912 table->v[i]->name = name;
5913 (table->used)++;
5914 return table->v[i];
5915}
5916
5917static void FASTCALL
5918hashTableClear(HASH_TABLE *table)
5919{
5920 size_t i;
5921 for (i = 0; i < table->size; i++) {
5922 table->mem->free_fcn(table->v[i]);
5923 table->v[i] = NULL;
5924 }
5925 table->used = 0;
5926}
5927
5928static void FASTCALL
5929hashTableDestroy(HASH_TABLE *table)
5930{
5931 size_t i;
5932 for (i = 0; i < table->size; i++)
5933 table->mem->free_fcn(table->v[i]);
5934 table->mem->free_fcn(table->v);
5935}
5936
5937static void FASTCALL
5938hashTableInit(HASH_TABLE *p, const XML_Memory_Handling_Suite *ms)
5939{
5940 p->power = 0;
5941 p->size = 0;
5942 p->used = 0;
5943 p->v = NULL;
5944 p->mem = ms;
5945}
5946
5947static void FASTCALL
5948hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table)
5949{
5950 iter->p = table->v;
5951 iter->end = iter->p + table->size;
5952}
5953
5954static NAMED * FASTCALL
5955hashTableIterNext(HASH_TABLE_ITER *iter)
5956{
5957 while (iter->p != iter->end) {
5958 NAMED *tem = *(iter->p)++;
5959 if (tem)
5960 return tem;
5961 }
5962 return NULL;
5963}
5964
5965static void FASTCALL
5966poolInit(STRING_POOL *pool, const XML_Memory_Handling_Suite *ms)
5967{
5968 pool->blocks = NULL;
5969 pool->freeBlocks = NULL;
5970 pool->start = NULL;
5971 pool->ptr = NULL;
5972 pool->end = NULL;
5973 pool->mem = ms;
5974}
5975
5976static void FASTCALL
5977poolClear(STRING_POOL *pool)
5978{
5979 if (!pool->freeBlocks)
5980 pool->freeBlocks = pool->blocks;
5981 else {
5982 BLOCK *p = pool->blocks;
5983 while (p) {
5984 BLOCK *tem = p->next;
5985 p->next = pool->freeBlocks;
5986 pool->freeBlocks = p;
5987 p = tem;
5988 }
5989 }
5990 pool->blocks = NULL;
5991 pool->start = NULL;
5992 pool->ptr = NULL;
5993 pool->end = NULL;
5994}
5995
5996static void FASTCALL
5997poolDestroy(STRING_POOL *pool)
5998{
5999 BLOCK *p = pool->blocks;
6000 while (p) {
6001 BLOCK *tem = p->next;
6002 pool->mem->free_fcn(p);
6003 p = tem;
6004 }
6005 p = pool->freeBlocks;
6006 while (p) {
6007 BLOCK *tem = p->next;
6008 pool->mem->free_fcn(p);
6009 p = tem;
6010 }
6011}
6012
6013static XML_Char *
6014poolAppend(STRING_POOL *pool, const ENCODING *enc,
6015 const char *ptr, const char *end)
6016{
6017 if (!pool->ptr && !poolGrow(pool))
6018 return NULL;
6019 for (;;) {
6020 XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end);
6021 if (ptr == end)
6022 break;
6023 if (!poolGrow(pool))
6024 return NULL;
6025 }
6026 return pool->start;
6027}
6028
6029static const XML_Char * FASTCALL
6030poolCopyString(STRING_POOL *pool, const XML_Char *s)
6031{
6032 do {
6033 if (!poolAppendChar(pool, *s))
6034 return NULL;
6035 } while (*s++);
6036 s = pool->start;
6037 poolFinish(pool);
6038 return s;
6039}
6040
6041static const XML_Char *
6042poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n)
6043{
6044 if (!pool->ptr && !poolGrow(pool))
6045 return NULL;
6046 for (; n > 0; --n, s++) {
6047 if (!poolAppendChar(pool, *s))
6048 return NULL;
6049 }
6050 s = pool->start;
6051 poolFinish(pool);
6052 return s;
6053}
6054
6055static const XML_Char * FASTCALL
6056poolAppendString(STRING_POOL *pool, const XML_Char *s)
6057{
6058 while (*s) {
6059 if (!poolAppendChar(pool, *s))
6060 return NULL;
6061 s++;
6062 }
6063 return pool->start;
6064}
6065
6066static XML_Char *
6067poolStoreString(STRING_POOL *pool, const ENCODING *enc,
6068 const char *ptr, const char *end)
6069{
6070 if (!poolAppend(pool, enc, ptr, end))
6071 return NULL;
6072 if (pool->ptr == pool->end && !poolGrow(pool))
6073 return NULL;
6074 *(pool->ptr)++ = 0;
6075 return pool->start;
6076}
6077
6078static XML_Bool FASTCALL
6079poolGrow(STRING_POOL *pool)
6080{
6081 if (pool->freeBlocks) {
6082 if (pool->start == 0) {
6083 pool->blocks = pool->freeBlocks;
6084 pool->freeBlocks = pool->freeBlocks->next;
6085 pool->blocks->next = NULL;
6086 pool->start = pool->blocks->s;
6087 pool->end = pool->start + pool->blocks->size;
6088 pool->ptr = pool->start;
6089 return XML_TRUE;
6090 }
6091 if (pool->end - pool->start < pool->freeBlocks->size) {
6092 BLOCK *tem = pool->freeBlocks->next;
6093 pool->freeBlocks->next = pool->blocks;
6094 pool->blocks = pool->freeBlocks;
6095 pool->freeBlocks = tem;
6096 memcpy(pool->blocks->s, pool->start,
6097 (pool->end - pool->start) * sizeof(XML_Char));
6098 pool->ptr = pool->blocks->s + (pool->ptr - pool->start);
6099 pool->start = pool->blocks->s;
6100 pool->end = pool->start + pool->blocks->size;
6101 return XML_TRUE;
6102 }
6103 }
6104 if (pool->blocks && pool->start == pool->blocks->s) {
6105 int blockSize = (int)(pool->end - pool->start)*2;
6106 pool->blocks = (BLOCK *)
6107 pool->mem->realloc_fcn(pool->blocks,
6108 (offsetof(BLOCK, s)
6109 + blockSize * sizeof(XML_Char)));
6110 if (pool->blocks == NULL)
6111 return XML_FALSE;
6112 pool->blocks->size = blockSize;
6113 pool->ptr = pool->blocks->s + (pool->ptr - pool->start);
6114 pool->start = pool->blocks->s;
6115 pool->end = pool->start + blockSize;
6116 }
6117 else {
6118 BLOCK *tem;
6119 int blockSize = (int)(pool->end - pool->start);
6120 if (blockSize < INIT_BLOCK_SIZE)
6121 blockSize = INIT_BLOCK_SIZE;
6122 else
6123 blockSize *= 2;
6124 tem = (BLOCK *)pool->mem->malloc_fcn(offsetof(BLOCK, s)
6125 + blockSize * sizeof(XML_Char));
6126 if (!tem)
6127 return XML_FALSE;
6128 tem->size = blockSize;
6129 tem->next = pool->blocks;
6130 pool->blocks = tem;
6131 if (pool->ptr != pool->start)
6132 memcpy(tem->s, pool->start,
6133 (pool->ptr - pool->start) * sizeof(XML_Char));
6134 pool->ptr = tem->s + (pool->ptr - pool->start);
6135 pool->start = tem->s;
6136 pool->end = tem->s + blockSize;
6137 }
6138 return XML_TRUE;
6139}
6140
6141static int FASTCALL
6142nextScaffoldPart(XML_Parser parser)
6143{
6144 DTD * const dtd = _dtd; /* save one level of indirection */
6145 CONTENT_SCAFFOLD * me;
6146 int next;
6147
6148 if (!dtd->scaffIndex) {
6149 dtd->scaffIndex = (int *)MALLOC(groupSize * sizeof(int));
6150 if (!dtd->scaffIndex)
6151 return -1;
6152 dtd->scaffIndex[0] = 0;
6153 }
6154
6155 if (dtd->scaffCount >= dtd->scaffSize) {
6156 CONTENT_SCAFFOLD *temp;
6157 if (dtd->scaffold) {
6158 temp = (CONTENT_SCAFFOLD *)
6159 REALLOC(dtd->scaffold, dtd->scaffSize * 2 * sizeof(CONTENT_SCAFFOLD));
6160 if (temp == NULL)
6161 return -1;
6162 dtd->scaffSize *= 2;
6163 }
6164 else {
6165 temp = (CONTENT_SCAFFOLD *)MALLOC(INIT_SCAFFOLD_ELEMENTS
6166 * sizeof(CONTENT_SCAFFOLD));
6167 if (temp == NULL)
6168 return -1;
6169 dtd->scaffSize = INIT_SCAFFOLD_ELEMENTS;
6170 }
6171 dtd->scaffold = temp;
6172 }
6173 next = dtd->scaffCount++;
6174 me = &dtd->scaffold[next];
6175 if (dtd->scaffLevel) {
6176 CONTENT_SCAFFOLD *parent = &dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel-1]];
6177 if (parent->lastchild) {
6178 dtd->scaffold[parent->lastchild].nextsib = next;
6179 }
6180 if (!parent->childcnt)
6181 parent->firstchild = next;
6182 parent->lastchild = next;
6183 parent->childcnt++;
6184 }
6185 me->firstchild = me->lastchild = me->childcnt = me->nextsib = 0;
6186 return next;
6187}
6188
6189static void
6190build_node(XML_Parser parser,
6191 int src_node,
6192 XML_Content *dest,
6193 XML_Content **contpos,
6194 XML_Char **strpos)
6195{
6196 DTD * const dtd = _dtd; /* save one level of indirection */
6197 dest->type = dtd->scaffold[src_node].type;
6198 dest->quant = dtd->scaffold[src_node].quant;
6199 if (dest->type == XML_CTYPE_NAME) {
6200 const XML_Char *src;
6201 dest->name = *strpos;
6202 src = dtd->scaffold[src_node].name;
6203 for (;;) {
6204 *(*strpos)++ = *src;
6205 if (!*src)
6206 break;
6207 src++;
6208 }
6209 dest->numchildren = 0;
6210 dest->children = NULL;
6211 }
6212 else {
6213 unsigned int i;
6214 int cn;
6215 dest->numchildren = dtd->scaffold[src_node].childcnt;
6216 dest->children = *contpos;
6217 *contpos += dest->numchildren;
6218 for (i = 0, cn = dtd->scaffold[src_node].firstchild;
6219 i < dest->numchildren;
6220 i++, cn = dtd->scaffold[cn].nextsib) {
6221 build_node(parser, cn, &(dest->children[i]), contpos, strpos);
6222 }
6223 dest->name = NULL;
6224 }
6225}
6226
6227static XML_Content *
6228build_model (XML_Parser parser)
6229{
6230 DTD * const dtd = _dtd; /* save one level of indirection */
6231 XML_Content *ret;
6232 XML_Content *cpos;
6233 XML_Char * str;
6234 int allocsize = (dtd->scaffCount * sizeof(XML_Content)
6235 + (dtd->contentStringLen * sizeof(XML_Char)));
6236
6237 ret = (XML_Content *)MALLOC(allocsize);
6238 if (!ret)
6239 return NULL;
6240
6241 str = (XML_Char *) (&ret[dtd->scaffCount]);
6242 cpos = &ret[1];
6243
6244 build_node(parser, 0, ret, &cpos, &str);
6245 return ret;
6246}
6247
6248static ELEMENT_TYPE *
6249getElementType(XML_Parser parser,
6250 const ENCODING *enc,
6251 const char *ptr,
6252 const char *end)
6253{
6254 DTD * const dtd = _dtd; /* save one level of indirection */
6255 const XML_Char *name = poolStoreString(&dtd->pool, enc, ptr, end);
6256 ELEMENT_TYPE *ret;
6257
6258 if (!name)
6259 return NULL;
6260 ret = (ELEMENT_TYPE *) lookup(&dtd->elementTypes, name, sizeof(ELEMENT_TYPE));
6261 if (!ret)
6262 return NULL;
6263 if (ret->name != name)
6264 poolDiscard(&dtd->pool);
6265 else {
6266 poolFinish(&dtd->pool);
6267 if (!setElementTypePrefix(parser, ret))
6268 return NULL;
6269 }
6270 return ret;
6271}
Note: See TracBrowser for help on using the repository browser.