]>
Commit | Line | Data |
---|---|---|
1 | /* Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd | |
2 | See the file COPYING for copying permission. | |
3 | */ | |
4 | ||
5 | #include <stddef.h> | |
6 | #include <string.h> /* memset(), memcpy() */ | |
7 | ||
8 | #ifdef COMPILED_FROM_DSP | |
9 | ||
10 | #include "winconfig.h" | |
11 | #include "expat.h" | |
12 | ||
13 | #elif defined(MACOS_CLASSIC) | |
14 | ||
15 | #include "macconfig.h" | |
16 | #include "expat.h" | |
17 | ||
18 | #else | |
19 | ||
20 | #include "expat_config.h" | |
21 | ||
22 | #ifdef __declspec | |
23 | #define XMLPARSEAPI(type) type __cdecl | |
24 | #endif | |
25 | ||
26 | #include "expat.h" | |
27 | ||
28 | #ifdef __declspec | |
29 | #undef XMLPARSEAPI | |
30 | #endif | |
31 | #endif /* ndef COMPILED_FROM_DSP */ | |
32 | ||
33 | #ifdef XML_UNICODE | |
34 | #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX | |
35 | #define XmlConvert XmlUtf16Convert | |
36 | #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding | |
37 | #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS | |
38 | #define XmlEncode XmlUtf16Encode | |
39 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1)) | |
40 | typedef unsigned short ICHAR; | |
41 | #else | |
42 | #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX | |
43 | #define XmlConvert XmlUtf8Convert | |
44 | #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding | |
45 | #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS | |
46 | #define XmlEncode XmlUtf8Encode | |
47 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf8) | |
48 | typedef char ICHAR; | |
49 | #endif | |
50 | ||
51 | ||
52 | #ifndef XML_NS | |
53 | ||
54 | #define XmlInitEncodingNS XmlInitEncoding | |
55 | #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding | |
56 | #undef XmlGetInternalEncodingNS | |
57 | #define XmlGetInternalEncodingNS XmlGetInternalEncoding | |
58 | #define XmlParseXmlDeclNS XmlParseXmlDecl | |
59 | ||
60 | #endif | |
61 | ||
62 | #ifdef XML_UNICODE | |
63 | ||
64 | #ifdef XML_UNICODE_WCHAR_T | |
65 | #define XML_T(x) (const wchar_t)x | |
66 | #define XML_L(x) L ## x | |
67 | #else | |
68 | #define XML_T(x) (const unsigned short)x | |
69 | #define XML_L(x) x | |
70 | #endif | |
71 | ||
72 | #else | |
73 | ||
74 | #define XML_T(x) x | |
75 | #define XML_L(x) x | |
76 | ||
77 | #endif | |
78 | ||
79 | /* Round up n to be a multiple of sz, where sz is a power of 2. */ | |
80 | #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1)) | |
81 | ||
82 | /* Handle the case where memmove() doesn't exist. */ | |
83 | #ifndef HAVE_MEMMOVE | |
84 | #ifdef HAVE_BCOPY | |
85 | #define memmove(d,s,l) bcopy((s),(d),(l)) | |
86 | #else | |
87 | #error memmove does not exist on this platform, nor is a substitute available | |
88 | #endif /* HAVE_BCOPY */ | |
89 | #endif /* HAVE_MEMMOVE */ | |
90 | ||
91 | #include "internal.h" | |
92 | #include "xmltok.h" | |
93 | #include "xmlrole.h" | |
94 | ||
95 | typedef const XML_Char *KEY; | |
96 | ||
97 | typedef struct { | |
98 | KEY name; | |
99 | } NAMED; | |
100 | ||
101 | typedef struct { | |
102 | NAMED **v; | |
103 | size_t size; | |
104 | size_t used; | |
105 | size_t usedLim; | |
106 | const XML_Memory_Handling_Suite *mem; | |
107 | } HASH_TABLE; | |
108 | ||
109 | typedef struct { | |
110 | NAMED **p; | |
111 | NAMED **end; | |
112 | } HASH_TABLE_ITER; | |
113 | ||
114 | #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ | |
115 | #define INIT_DATA_BUF_SIZE 1024 | |
116 | #define INIT_ATTS_SIZE 16 | |
117 | #define INIT_BLOCK_SIZE 1024 | |
118 | #define INIT_BUFFER_SIZE 1024 | |
119 | ||
120 | #define EXPAND_SPARE 24 | |
121 | ||
122 | typedef struct binding { | |
123 | struct prefix *prefix; | |
124 | struct binding *nextTagBinding; | |
125 | struct binding *prevPrefixBinding; | |
126 | const struct attribute_id *attId; | |
127 | XML_Char *uri; | |
128 | int uriLen; | |
129 | int uriAlloc; | |
130 | } BINDING; | |
131 | ||
132 | typedef struct prefix { | |
133 | const XML_Char *name; | |
134 | BINDING *binding; | |
135 | } PREFIX; | |
136 | ||
137 | typedef struct { | |
138 | const XML_Char *str; | |
139 | const XML_Char *localPart; | |
140 | const XML_Char *prefix; | |
141 | int strLen; | |
142 | int uriLen; | |
143 | int prefixLen; | |
144 | } TAG_NAME; | |
145 | ||
146 | /* TAG represents an open element. | |
147 | The name of the element is stored in both the document and API | |
148 | encodings. The memory buffer 'buf' is a separately-allocated | |
149 | memory area which stores the name. During the XML_Parse()/ | |
150 | XMLParseBuffer() when the element is open, the memory for the 'raw' | |
151 | version of the name (in the document encoding) is shared with the | |
152 | document buffer. If the element is open across calls to | |
153 | XML_Parse()/XML_ParseBuffer(), the buffer is re-allocated to | |
154 | contain the 'raw' name as well. | |
155 | ||
156 | A parser re-uses these structures, maintaining a list of allocated | |
157 | TAG objects in a free list. | |
158 | */ | |
159 | typedef struct tag { | |
160 | struct tag *parent; /* parent of this element */ | |
161 | const char *rawName; /* tagName in the original encoding */ | |
162 | int rawNameLength; | |
163 | TAG_NAME name; /* tagName in the API encoding */ | |
164 | char *buf; /* buffer for name components */ | |
165 | char *bufEnd; /* end of the buffer */ | |
166 | BINDING *bindings; | |
167 | } TAG; | |
168 | ||
169 | typedef struct { | |
170 | const XML_Char *name; | |
171 | const XML_Char *textPtr; | |
172 | int textLen; | |
173 | const XML_Char *systemId; | |
174 | const XML_Char *base; | |
175 | const XML_Char *publicId; | |
176 | const XML_Char *notation; | |
177 | XML_Bool open; | |
178 | XML_Bool is_param; | |
179 | XML_Bool is_internal; /* true if declared in internal subset outside PE */ | |
180 | } ENTITY; | |
181 | ||
182 | typedef struct { | |
183 | enum XML_Content_Type type; | |
184 | enum XML_Content_Quant quant; | |
185 | const XML_Char * name; | |
186 | int firstchild; | |
187 | int lastchild; | |
188 | int childcnt; | |
189 | int nextsib; | |
190 | } CONTENT_SCAFFOLD; | |
191 | ||
192 | #define INIT_SCAFFOLD_ELEMENTS 32 | |
193 | ||
194 | typedef struct block { | |
195 | struct block *next; | |
196 | int size; | |
197 | XML_Char s[1]; | |
198 | } BLOCK; | |
199 | ||
200 | typedef struct { | |
201 | BLOCK *blocks; | |
202 | BLOCK *freeBlocks; | |
203 | const XML_Char *end; | |
204 | XML_Char *ptr; | |
205 | XML_Char *start; | |
206 | const XML_Memory_Handling_Suite *mem; | |
207 | } STRING_POOL; | |
208 | ||
209 | /* The XML_Char before the name is used to determine whether | |
210 | an attribute has been specified. */ | |
211 | typedef struct attribute_id { | |
212 | XML_Char *name; | |
213 | PREFIX *prefix; | |
214 | XML_Bool maybeTokenized; | |
215 | XML_Bool xmlns; | |
216 | } ATTRIBUTE_ID; | |
217 | ||
218 | typedef struct { | |
219 | const ATTRIBUTE_ID *id; | |
220 | XML_Bool isCdata; | |
221 | const XML_Char *value; | |
222 | } DEFAULT_ATTRIBUTE; | |
223 | ||
224 | typedef struct { | |
225 | const XML_Char *name; | |
226 | PREFIX *prefix; | |
227 | const ATTRIBUTE_ID *idAtt; | |
228 | int nDefaultAtts; | |
229 | int allocDefaultAtts; | |
230 | DEFAULT_ATTRIBUTE *defaultAtts; | |
231 | } ELEMENT_TYPE; | |
232 | ||
233 | typedef struct { | |
234 | HASH_TABLE generalEntities; | |
235 | HASH_TABLE elementTypes; | |
236 | HASH_TABLE attributeIds; | |
237 | HASH_TABLE prefixes; | |
238 | STRING_POOL pool; | |
239 | STRING_POOL entityValuePool; | |
240 | /* false once a parameter entity reference has been skipped */ | |
241 | XML_Bool keepProcessing; | |
242 | /* true once an internal or external PE reference has been encountered; | |
243 | this includes the reference to an external subset */ | |
244 | XML_Bool hasParamEntityRefs; | |
245 | XML_Bool standalone; | |
246 | #ifdef XML_DTD | |
247 | /* indicates if external PE has been read */ | |
248 | XML_Bool paramEntityRead; | |
249 | HASH_TABLE paramEntities; | |
250 | #endif /* XML_DTD */ | |
251 | PREFIX defaultPrefix; | |
252 | /* === scaffolding for building content model === */ | |
253 | XML_Bool in_eldecl; | |
254 | CONTENT_SCAFFOLD *scaffold; | |
255 | unsigned contentStringLen; | |
256 | unsigned scaffSize; | |
257 | unsigned scaffCount; | |
258 | int scaffLevel; | |
259 | int *scaffIndex; | |
260 | } DTD; | |
261 | ||
262 | typedef struct open_internal_entity { | |
263 | const char *internalEventPtr; | |
264 | const char *internalEventEndPtr; | |
265 | struct open_internal_entity *next; | |
266 | ENTITY *entity; | |
267 | } OPEN_INTERNAL_ENTITY; | |
268 | ||
269 | typedef enum XML_Error PTRCALL Processor(XML_Parser parser, | |
270 | const char *start, | |
271 | const char *end, | |
272 | const char **endPtr); | |
273 | ||
274 | static Processor prologProcessor; | |
275 | static Processor prologInitProcessor; | |
276 | static Processor contentProcessor; | |
277 | static Processor cdataSectionProcessor; | |
278 | #ifdef XML_DTD | |
279 | static Processor ignoreSectionProcessor; | |
280 | static Processor externalParEntProcessor; | |
281 | static Processor externalParEntInitProcessor; | |
282 | static Processor entityValueProcessor; | |
283 | static Processor entityValueInitProcessor; | |
284 | #endif /* XML_DTD */ | |
285 | static Processor epilogProcessor; | |
286 | static Processor errorProcessor; | |
287 | static Processor externalEntityInitProcessor; | |
288 | static Processor externalEntityInitProcessor2; | |
289 | static Processor externalEntityInitProcessor3; | |
290 | static Processor externalEntityContentProcessor; | |
291 | ||
292 | static enum XML_Error | |
293 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName); | |
294 | static enum XML_Error | |
295 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
296 | const char *, const char *); | |
297 | static enum XML_Error | |
298 | initializeEncoding(XML_Parser parser); | |
299 | static enum XML_Error | |
300 | doProlog(XML_Parser parser, const ENCODING *enc, const char *s, | |
301 | const char *end, int tok, const char *next, const char **nextPtr); | |
302 | static enum XML_Error | |
303 | processInternalParamEntity(XML_Parser parser, ENTITY *entity); | |
304 | static enum XML_Error | |
305 | doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, | |
306 | const char *start, const char *end, const char **endPtr); | |
307 | static enum XML_Error | |
308 | doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, | |
309 | const char *end, const char **nextPtr); | |
310 | #ifdef XML_DTD | |
311 | static enum XML_Error | |
312 | doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr, | |
313 | const char *end, const char **nextPtr); | |
314 | #endif /* XML_DTD */ | |
315 | ||
316 | static enum XML_Error | |
317 | storeAtts(XML_Parser parser, const ENCODING *, const char *s, | |
318 | TAG_NAME *tagNamePtr, BINDING **bindingsPtr); | |
319 | static enum XML_Error | |
320 | addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, | |
321 | const XML_Char *uri, BINDING **bindingsPtr); | |
322 | static int | |
323 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, | |
324 | XML_Bool isCdata, XML_Bool isId, const XML_Char *dfltValue, | |
325 | XML_Parser parser); | |
326 | static enum XML_Error | |
327 | storeAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata, | |
328 | const char *, const char *, STRING_POOL *); | |
329 | static enum XML_Error | |
330 | appendAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata, | |
331 | const char *, const char *, STRING_POOL *); | |
332 | static ATTRIBUTE_ID * | |
333 | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, | |
334 | const char *end); | |
335 | static int | |
336 | setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); | |
337 | static enum XML_Error | |
338 | storeEntityValue(XML_Parser parser, const ENCODING *enc, const char *start, | |
339 | const char *end); | |
340 | static int | |
341 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, | |
342 | const char *start, const char *end); | |
343 | static int | |
344 | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, | |
345 | const char *end); | |
346 | static void | |
347 | reportDefault(XML_Parser parser, const ENCODING *enc, const char *start, | |
348 | const char *end); | |
349 | ||
350 | static const XML_Char * getContext(XML_Parser parser); | |
351 | static XML_Bool | |
352 | setContext(XML_Parser parser, const XML_Char *context); | |
353 | ||
354 | static void FASTCALL normalizePublicId(XML_Char *s); | |
355 | ||
356 | static DTD * dtdCreate(const XML_Memory_Handling_Suite *ms); | |
357 | /* do not call if parentParser != NULL */ | |
358 | static void dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms); | |
359 | static void | |
360 | dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms); | |
361 | static int | |
362 | dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms); | |
363 | static int | |
364 | copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *); | |
365 | ||
366 | static NAMED * | |
367 | lookup(HASH_TABLE *table, KEY name, size_t createSize); | |
368 | static void FASTCALL | |
369 | hashTableInit(HASH_TABLE *, const XML_Memory_Handling_Suite *ms); | |
370 | static void FASTCALL hashTableClear(HASH_TABLE *); | |
371 | static void FASTCALL hashTableDestroy(HASH_TABLE *); | |
372 | static void FASTCALL | |
373 | hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *); | |
374 | static NAMED * FASTCALL hashTableIterNext(HASH_TABLE_ITER *); | |
375 | ||
376 | static void FASTCALL | |
377 | poolInit(STRING_POOL *, const XML_Memory_Handling_Suite *ms); | |
378 | static void FASTCALL poolClear(STRING_POOL *); | |
379 | static void FASTCALL poolDestroy(STRING_POOL *); | |
380 | static XML_Char * | |
381 | poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
382 | const char *ptr, const char *end); | |
383 | static XML_Char * | |
384 | poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
385 | const char *ptr, const char *end); | |
386 | static XML_Bool FASTCALL poolGrow(STRING_POOL *pool); | |
387 | static const XML_Char * FASTCALL | |
388 | poolCopyString(STRING_POOL *pool, const XML_Char *s); | |
389 | static const XML_Char * | |
390 | poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n); | |
391 | static const XML_Char * FASTCALL | |
392 | poolAppendString(STRING_POOL *pool, const XML_Char *s); | |
393 | ||
394 | static int FASTCALL nextScaffoldPart(XML_Parser parser); | |
395 | static XML_Content * build_model(XML_Parser parser); | |
396 | static ELEMENT_TYPE * | |
397 | getElementType(XML_Parser parser, const ENCODING *enc, | |
398 | const char *ptr, const char *end); | |
399 | ||
400 | static XML_Parser | |
401 | parserCreate(const XML_Char *encodingName, | |
402 | const XML_Memory_Handling_Suite *memsuite, | |
403 | const XML_Char *nameSep, | |
404 | DTD *dtd); | |
405 | static void | |
406 | parserInit(XML_Parser parser, const XML_Char *encodingName); | |
407 | ||
408 | #define poolStart(pool) ((pool)->start) | |
409 | #define poolEnd(pool) ((pool)->ptr) | |
410 | #define poolLength(pool) ((pool)->ptr - (pool)->start) | |
411 | #define poolChop(pool) ((void)--(pool->ptr)) | |
412 | #define poolLastChar(pool) (((pool)->ptr)[-1]) | |
413 | #define poolDiscard(pool) ((pool)->ptr = (pool)->start) | |
414 | #define poolFinish(pool) ((pool)->start = (pool)->ptr) | |
415 | #define poolAppendChar(pool, c) \ | |
416 | (((pool)->ptr == (pool)->end && !poolGrow(pool)) \ | |
417 | ? 0 \ | |
418 | : ((*((pool)->ptr)++ = c), 1)) | |
419 | ||
420 | struct XML_ParserStruct { | |
421 | /* The first member must be userData so that the XML_GetUserData | |
422 | macro works. */ | |
423 | void *m_userData; | |
424 | void *m_handlerArg; | |
425 | char *m_buffer; | |
426 | const XML_Memory_Handling_Suite m_mem; | |
427 | /* first character to be parsed */ | |
428 | const char *m_bufferPtr; | |
429 | /* past last character to be parsed */ | |
430 | char *m_bufferEnd; | |
431 | /* allocated end of buffer */ | |
432 | const char *m_bufferLim; | |
433 | long m_parseEndByteIndex; | |
434 | const char *m_parseEndPtr; | |
435 | XML_Char *m_dataBuf; | |
436 | XML_Char *m_dataBufEnd; | |
437 | XML_StartElementHandler m_startElementHandler; | |
438 | XML_EndElementHandler m_endElementHandler; | |
439 | XML_CharacterDataHandler m_characterDataHandler; | |
440 | XML_ProcessingInstructionHandler m_processingInstructionHandler; | |
441 | XML_CommentHandler m_commentHandler; | |
442 | XML_StartCdataSectionHandler m_startCdataSectionHandler; | |
443 | XML_EndCdataSectionHandler m_endCdataSectionHandler; | |
444 | XML_DefaultHandler m_defaultHandler; | |
445 | XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler; | |
446 | XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler; | |
447 | XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; | |
448 | XML_NotationDeclHandler m_notationDeclHandler; | |
449 | XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; | |
450 | XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; | |
451 | XML_NotStandaloneHandler m_notStandaloneHandler; | |
452 | XML_ExternalEntityRefHandler m_externalEntityRefHandler; | |
453 | XML_Parser m_externalEntityRefHandlerArg; | |
454 | XML_SkippedEntityHandler m_skippedEntityHandler; | |
455 | XML_UnknownEncodingHandler m_unknownEncodingHandler; | |
456 | XML_ElementDeclHandler m_elementDeclHandler; | |
457 | XML_AttlistDeclHandler m_attlistDeclHandler; | |
458 | XML_EntityDeclHandler m_entityDeclHandler; | |
459 | XML_XmlDeclHandler m_xmlDeclHandler; | |
460 | const ENCODING *m_encoding; | |
461 | INIT_ENCODING m_initEncoding; | |
462 | const ENCODING *m_internalEncoding; | |
463 | const XML_Char *m_protocolEncodingName; | |
464 | XML_Bool m_ns; | |
465 | XML_Bool m_ns_triplets; | |
466 | void *m_unknownEncodingMem; | |
467 | void *m_unknownEncodingData; | |
468 | void *m_unknownEncodingHandlerData; | |
469 | void (*m_unknownEncodingRelease)(void *); | |
470 | PROLOG_STATE m_prologState; | |
471 | Processor *m_processor; | |
472 | enum XML_Error m_errorCode; | |
473 | const char *m_eventPtr; | |
474 | const char *m_eventEndPtr; | |
475 | const char *m_positionPtr; | |
476 | OPEN_INTERNAL_ENTITY *m_openInternalEntities; | |
477 | XML_Bool m_defaultExpandInternalEntities; | |
478 | int m_tagLevel; | |
479 | ENTITY *m_declEntity; | |
480 | const XML_Char *m_doctypeName; | |
481 | const XML_Char *m_doctypeSysid; | |
482 | const XML_Char *m_doctypePubid; | |
483 | const XML_Char *m_declAttributeType; | |
484 | const XML_Char *m_declNotationName; | |
485 | const XML_Char *m_declNotationPublicId; | |
486 | ELEMENT_TYPE *m_declElementType; | |
487 | ATTRIBUTE_ID *m_declAttributeId; | |
488 | XML_Bool m_declAttributeIsCdata; | |
489 | XML_Bool m_declAttributeIsId; | |
490 | DTD *m_dtd; | |
491 | const XML_Char *m_curBase; | |
492 | TAG *m_tagStack; | |
493 | TAG *m_freeTagList; | |
494 | BINDING *m_inheritedBindings; | |
495 | BINDING *m_freeBindingList; | |
496 | int m_attsSize; | |
497 | int m_nSpecifiedAtts; | |
498 | int m_idAttIndex; | |
499 | ATTRIBUTE *m_atts; | |
500 | POSITION m_position; | |
501 | STRING_POOL m_tempPool; | |
502 | STRING_POOL m_temp2Pool; | |
503 | char *m_groupConnector; | |
504 | unsigned m_groupSize; | |
505 | XML_Char m_namespaceSeparator; | |
506 | XML_Parser m_parentParser; | |
507 | #ifdef XML_DTD | |
508 | XML_Bool m_isParamEntity; | |
509 | XML_Bool m_useForeignDTD; | |
510 | enum XML_ParamEntityParsing m_paramEntityParsing; | |
511 | #endif | |
512 | }; | |
513 | ||
514 | #define MALLOC(s) (parser->m_mem.malloc_fcn((s))) | |
515 | #define REALLOC(p,s) (parser->m_mem.realloc_fcn((p),(s))) | |
516 | #define FREE(p) (parser->m_mem.free_fcn((p))) | |
517 | ||
518 | #define userData (parser->m_userData) | |
519 | #define handlerArg (parser->m_handlerArg) | |
520 | #define startElementHandler (parser->m_startElementHandler) | |
521 | #define endElementHandler (parser->m_endElementHandler) | |
522 | #define characterDataHandler (parser->m_characterDataHandler) | |
523 | #define processingInstructionHandler \ | |
524 | (parser->m_processingInstructionHandler) | |
525 | #define commentHandler (parser->m_commentHandler) | |
526 | #define startCdataSectionHandler \ | |
527 | (parser->m_startCdataSectionHandler) | |
528 | #define endCdataSectionHandler (parser->m_endCdataSectionHandler) | |
529 | #define defaultHandler (parser->m_defaultHandler) | |
530 | #define startDoctypeDeclHandler (parser->m_startDoctypeDeclHandler) | |
531 | #define endDoctypeDeclHandler (parser->m_endDoctypeDeclHandler) | |
532 | #define unparsedEntityDeclHandler \ | |
533 | (parser->m_unparsedEntityDeclHandler) | |
534 | #define notationDeclHandler (parser->m_notationDeclHandler) | |
535 | #define startNamespaceDeclHandler \ | |
536 | (parser->m_startNamespaceDeclHandler) | |
537 | #define endNamespaceDeclHandler (parser->m_endNamespaceDeclHandler) | |
538 | #define notStandaloneHandler (parser->m_notStandaloneHandler) | |
539 | #define externalEntityRefHandler \ | |
540 | (parser->m_externalEntityRefHandler) | |
541 | #define externalEntityRefHandlerArg \ | |
542 | (parser->m_externalEntityRefHandlerArg) | |
543 | #define internalEntityRefHandler \ | |
544 | (parser->m_internalEntityRefHandler) | |
545 | #define skippedEntityHandler (parser->m_skippedEntityHandler) | |
546 | #define unknownEncodingHandler (parser->m_unknownEncodingHandler) | |
547 | #define elementDeclHandler (parser->m_elementDeclHandler) | |
548 | #define attlistDeclHandler (parser->m_attlistDeclHandler) | |
549 | #define entityDeclHandler (parser->m_entityDeclHandler) | |
550 | #define xmlDeclHandler (parser->m_xmlDeclHandler) | |
551 | #define encoding (parser->m_encoding) | |
552 | #define initEncoding (parser->m_initEncoding) | |
553 | #define internalEncoding (parser->m_internalEncoding) | |
554 | #define unknownEncodingMem (parser->m_unknownEncodingMem) | |
555 | #define unknownEncodingData (parser->m_unknownEncodingData) | |
556 | #define unknownEncodingHandlerData \ | |
557 | (parser->m_unknownEncodingHandlerData) | |
558 | #define unknownEncodingRelease (parser->m_unknownEncodingRelease) | |
559 | #define protocolEncodingName (parser->m_protocolEncodingName) | |
560 | #define ns (parser->m_ns) | |
561 | #define ns_triplets (parser->m_ns_triplets) | |
562 | #define prologState (parser->m_prologState) | |
563 | #define processor (parser->m_processor) | |
564 | #define errorCode (parser->m_errorCode) | |
565 | #define eventPtr (parser->m_eventPtr) | |
566 | #define eventEndPtr (parser->m_eventEndPtr) | |
567 | #define positionPtr (parser->m_positionPtr) | |
568 | #define position (parser->m_position) | |
569 | #define openInternalEntities (parser->m_openInternalEntities) | |
570 | #define defaultExpandInternalEntities \ | |
571 | (parser->m_defaultExpandInternalEntities) | |
572 | #define tagLevel (parser->m_tagLevel) | |
573 | #define buffer (parser->m_buffer) | |
574 | #define bufferPtr (parser->m_bufferPtr) | |
575 | #define bufferEnd (parser->m_bufferEnd) | |
576 | #define parseEndByteIndex (parser->m_parseEndByteIndex) | |
577 | #define parseEndPtr (parser->m_parseEndPtr) | |
578 | #define bufferLim (parser->m_bufferLim) | |
579 | #define dataBuf (parser->m_dataBuf) | |
580 | #define dataBufEnd (parser->m_dataBufEnd) | |
581 | #define _dtd (parser->m_dtd) | |
582 | #define curBase (parser->m_curBase) | |
583 | #define declEntity (parser->m_declEntity) | |
584 | #define doctypeName (parser->m_doctypeName) | |
585 | #define doctypeSysid (parser->m_doctypeSysid) | |
586 | #define doctypePubid (parser->m_doctypePubid) | |
587 | #define declAttributeType (parser->m_declAttributeType) | |
588 | #define declNotationName (parser->m_declNotationName) | |
589 | #define declNotationPublicId (parser->m_declNotationPublicId) | |
590 | #define declElementType (parser->m_declElementType) | |
591 | #define declAttributeId (parser->m_declAttributeId) | |
592 | #define declAttributeIsCdata (parser->m_declAttributeIsCdata) | |
593 | #define declAttributeIsId (parser->m_declAttributeIsId) | |
594 | #define freeTagList (parser->m_freeTagList) | |
595 | #define freeBindingList (parser->m_freeBindingList) | |
596 | #define inheritedBindings (parser->m_inheritedBindings) | |
597 | #define tagStack (parser->m_tagStack) | |
598 | #define atts (parser->m_atts) | |
599 | #define attsSize (parser->m_attsSize) | |
600 | #define nSpecifiedAtts (parser->m_nSpecifiedAtts) | |
601 | #define idAttIndex (parser->m_idAttIndex) | |
602 | #define tempPool (parser->m_tempPool) | |
603 | #define temp2Pool (parser->m_temp2Pool) | |
604 | #define groupConnector (parser->m_groupConnector) | |
605 | #define groupSize (parser->m_groupSize) | |
606 | #define namespaceSeparator (parser->m_namespaceSeparator) | |
607 | #define parentParser (parser->m_parentParser) | |
608 | #ifdef XML_DTD | |
609 | #define isParamEntity (parser->m_isParamEntity) | |
610 | #define useForeignDTD (parser->m_useForeignDTD) | |
611 | #define paramEntityParsing (parser->m_paramEntityParsing) | |
612 | #endif /* XML_DTD */ | |
613 | ||
614 | #define parsing \ | |
615 | (parentParser \ | |
616 | ? \ | |
617 | (isParamEntity \ | |
618 | ? \ | |
619 | (processor != externalParEntInitProcessor) \ | |
620 | : \ | |
621 | (processor != externalEntityInitProcessor)) \ | |
622 | : \ | |
623 | (processor != prologInitProcessor)) | |
624 | ||
625 | XML_Parser | |
626 | XML_ParserCreate(const XML_Char *encodingName) | |
627 | { | |
628 | return XML_ParserCreate_MM(encodingName, NULL, NULL); | |
629 | } | |
630 | ||
631 | XML_Parser | |
632 | XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) | |
633 | { | |
634 | XML_Char tmp[2]; | |
635 | *tmp = nsSep; | |
636 | return XML_ParserCreate_MM(encodingName, NULL, tmp); | |
637 | } | |
638 | ||
639 | static const XML_Char implicitContext[] = { | |
640 | 'x', 'm', 'l', '=', 'h', 't', 't', 'p', ':', '/', '/', | |
641 | 'w', 'w', 'w', '.', 'w', '3', '.', 'o', 'r', 'g', '/', | |
642 | 'X', 'M', 'L', '/', '1', '9', '9', '8', '/', | |
643 | 'n', 'a', 'm', 'e', 's', 'p', 'a', 'c', 'e', '\0' | |
644 | }; | |
645 | ||
646 | XML_Parser | |
647 | XML_ParserCreate_MM(const XML_Char *encodingName, | |
648 | const XML_Memory_Handling_Suite *memsuite, | |
649 | const XML_Char *nameSep) | |
650 | { | |
651 | XML_Parser parser = parserCreate(encodingName, memsuite, nameSep, NULL); | |
652 | if (parser != NULL && ns) { | |
653 | /* implicit context only set for root parser, since child | |
654 | parsers (i.e. external entity parsers) will inherit it | |
655 | */ | |
656 | if (!setContext(parser, implicitContext)) { | |
657 | XML_ParserFree(parser); | |
658 | return NULL; | |
659 | } | |
660 | } | |
661 | return parser; | |
662 | } | |
663 | ||
664 | static XML_Parser | |
665 | parserCreate(const XML_Char *encodingName, | |
666 | const XML_Memory_Handling_Suite *memsuite, | |
667 | const XML_Char *nameSep, | |
668 | DTD *dtd) | |
669 | { | |
670 | XML_Parser parser; | |
671 | ||
672 | if (memsuite) { | |
673 | XML_Memory_Handling_Suite *mtemp; | |
674 | parser = (XML_Parser) | |
675 | memsuite->malloc_fcn(sizeof(struct XML_ParserStruct)); | |
676 | if (parser != NULL) { | |
677 | mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem); | |
678 | mtemp->malloc_fcn = memsuite->malloc_fcn; | |
679 | mtemp->realloc_fcn = memsuite->realloc_fcn; | |
680 | mtemp->free_fcn = memsuite->free_fcn; | |
681 | } | |
682 | } | |
683 | else { | |
684 | XML_Memory_Handling_Suite *mtemp; | |
685 | parser = (XML_Parser)malloc(sizeof(struct XML_ParserStruct)); | |
686 | if (parser != NULL) { | |
687 | mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem); | |
688 | mtemp->malloc_fcn = malloc; | |
689 | mtemp->realloc_fcn = realloc; | |
690 | mtemp->free_fcn = free; | |
691 | } | |
692 | } | |
693 | ||
694 | if (!parser) | |
695 | return parser; | |
696 | ||
697 | buffer = NULL; | |
698 | bufferLim = NULL; | |
699 | ||
700 | attsSize = INIT_ATTS_SIZE; | |
701 | atts = (ATTRIBUTE *)MALLOC(attsSize * sizeof(ATTRIBUTE)); | |
702 | if (atts == NULL) { | |
703 | FREE(parser); | |
704 | return NULL; | |
705 | } | |
706 | dataBuf = (XML_Char *)MALLOC(INIT_DATA_BUF_SIZE * sizeof(XML_Char)); | |
707 | if (dataBuf == NULL) { | |
708 | FREE(atts); | |
709 | FREE(parser); | |
710 | return NULL; | |
711 | } | |
712 | dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE; | |
713 | ||
714 | if (dtd) | |
715 | _dtd = dtd; | |
716 | else { | |
717 | _dtd = dtdCreate(&parser->m_mem); | |
718 | if (_dtd == NULL) { | |
719 | FREE(dataBuf); | |
720 | FREE(atts); | |
721 | FREE(parser); | |
722 | return NULL; | |
723 | } | |
724 | } | |
725 | ||
726 | freeBindingList = NULL; | |
727 | freeTagList = NULL; | |
728 | ||
729 | groupSize = 0; | |
730 | groupConnector = NULL; | |
731 | ||
732 | unknownEncodingHandler = NULL; | |
733 | unknownEncodingHandlerData = NULL; | |
734 | ||
735 | namespaceSeparator = '!'; | |
736 | ns = XML_FALSE; | |
737 | ns_triplets = XML_FALSE; | |
738 | ||
739 | poolInit(&tempPool, &(parser->m_mem)); | |
740 | poolInit(&temp2Pool, &(parser->m_mem)); | |
741 | parserInit(parser, encodingName); | |
742 | ||
743 | if (encodingName && !protocolEncodingName) { | |
744 | XML_ParserFree(parser); | |
745 | return NULL; | |
746 | } | |
747 | ||
748 | if (nameSep) { | |
749 | ns = XML_TRUE; | |
750 | internalEncoding = XmlGetInternalEncodingNS(); | |
751 | namespaceSeparator = *nameSep; | |
752 | } | |
753 | else { | |
754 | internalEncoding = XmlGetInternalEncoding(); | |
755 | } | |
756 | ||
757 | return parser; | |
758 | } | |
759 | ||
760 | static void | |
761 | parserInit(XML_Parser parser, const XML_Char *encodingName) | |
762 | { | |
763 | processor = prologInitProcessor; | |
764 | XmlPrologStateInit(&prologState); | |
765 | protocolEncodingName = (encodingName != NULL | |
766 | ? poolCopyString(&tempPool, encodingName) | |
767 | : NULL); | |
768 | curBase = NULL; | |
769 | XmlInitEncoding(&initEncoding, &encoding, 0); | |
770 | userData = NULL; | |
771 | handlerArg = NULL; | |
772 | startElementHandler = NULL; | |
773 | endElementHandler = NULL; | |
774 | characterDataHandler = NULL; | |
775 | processingInstructionHandler = NULL; | |
776 | commentHandler = NULL; | |
777 | startCdataSectionHandler = NULL; | |
778 | endCdataSectionHandler = NULL; | |
779 | defaultHandler = NULL; | |
780 | startDoctypeDeclHandler = NULL; | |
781 | endDoctypeDeclHandler = NULL; | |
782 | unparsedEntityDeclHandler = NULL; | |
783 | notationDeclHandler = NULL; | |
784 | startNamespaceDeclHandler = NULL; | |
785 | endNamespaceDeclHandler = NULL; | |
786 | notStandaloneHandler = NULL; | |
787 | externalEntityRefHandler = NULL; | |
788 | externalEntityRefHandlerArg = parser; | |
789 | skippedEntityHandler = NULL; | |
790 | elementDeclHandler = NULL; | |
791 | attlistDeclHandler = NULL; | |
792 | entityDeclHandler = NULL; | |
793 | xmlDeclHandler = NULL; | |
794 | bufferPtr = buffer; | |
795 | bufferEnd = buffer; | |
796 | parseEndByteIndex = 0; | |
797 | parseEndPtr = NULL; | |
798 | declElementType = NULL; | |
799 | declAttributeId = NULL; | |
800 | declEntity = NULL; | |
801 | doctypeName = NULL; | |
802 | doctypeSysid = NULL; | |
803 | doctypePubid = NULL; | |
804 | declAttributeType = NULL; | |
805 | declNotationName = NULL; | |
806 | declNotationPublicId = NULL; | |
807 | declAttributeIsCdata = XML_FALSE; | |
808 | declAttributeIsId = XML_FALSE; | |
809 | memset(&position, 0, sizeof(POSITION)); | |
810 | errorCode = XML_ERROR_NONE; | |
811 | eventPtr = NULL; | |
812 | eventEndPtr = NULL; | |
813 | positionPtr = NULL; | |
814 | openInternalEntities = 0; | |
815 | defaultExpandInternalEntities = XML_TRUE; | |
816 | tagLevel = 0; | |
817 | tagStack = NULL; | |
818 | inheritedBindings = NULL; | |
819 | nSpecifiedAtts = 0; | |
820 | unknownEncodingMem = NULL; | |
821 | unknownEncodingRelease = NULL; | |
822 | unknownEncodingData = NULL; | |
823 | parentParser = NULL; | |
824 | #ifdef XML_DTD | |
825 | isParamEntity = XML_FALSE; | |
826 | useForeignDTD = XML_FALSE; | |
827 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; | |
828 | #endif | |
829 | } | |
830 | ||
831 | /* moves list of bindings to freeBindingList */ | |
832 | static void FASTCALL | |
833 | moveToFreeBindingList(XML_Parser parser, BINDING *bindings) | |
834 | { | |
835 | while (bindings) { | |
836 | BINDING *b = bindings; | |
837 | bindings = bindings->nextTagBinding; | |
838 | b->nextTagBinding = freeBindingList; | |
839 | freeBindingList = b; | |
840 | } | |
841 | } | |
842 | ||
843 | XML_Bool | |
844 | XML_ParserReset(XML_Parser parser, const XML_Char *encodingName) | |
845 | { | |
846 | TAG *tStk; | |
847 | if (parentParser) | |
848 | return XML_FALSE; | |
849 | /* move tagStack to freeTagList */ | |
850 | tStk = tagStack; | |
851 | while (tStk) { | |
852 | TAG *tag = tStk; | |
853 | tStk = tStk->parent; | |
854 | tag->parent = freeTagList; | |
855 | moveToFreeBindingList(parser, tag->bindings); | |
856 | tag->bindings = NULL; | |
857 | freeTagList = tag; | |
858 | } | |
859 | moveToFreeBindingList(parser, inheritedBindings); | |
860 | if (unknownEncodingMem) | |
861 | FREE(unknownEncodingMem); | |
862 | if (unknownEncodingRelease) | |
863 | unknownEncodingRelease(unknownEncodingData); | |
864 | poolClear(&tempPool); | |
865 | poolClear(&temp2Pool); | |
866 | parserInit(parser, encodingName); | |
867 | dtdReset(_dtd, &parser->m_mem); | |
868 | return setContext(parser, implicitContext); | |
869 | } | |
870 | ||
871 | enum XML_Status | |
872 | XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) | |
873 | { | |
874 | /* Block after XML_Parse()/XML_ParseBuffer() has been called. | |
875 | XXX There's no way for the caller to determine which of the | |
876 | XXX possible error cases caused the XML_STATUS_ERROR return. | |
877 | */ | |
878 | if (parsing) | |
879 | return XML_STATUS_ERROR; | |
880 | if (encodingName == NULL) | |
881 | protocolEncodingName = NULL; | |
882 | else { | |
883 | protocolEncodingName = poolCopyString(&tempPool, encodingName); | |
884 | if (!protocolEncodingName) | |
885 | return XML_STATUS_ERROR; | |
886 | } | |
887 | return XML_STATUS_OK; | |
888 | } | |
889 | ||
890 | XML_Parser | |
891 | XML_ExternalEntityParserCreate(XML_Parser oldParser, | |
892 | const XML_Char *context, | |
893 | const XML_Char *encodingName) | |
894 | { | |
895 | XML_Parser parser = oldParser; | |
896 | DTD *newDtd = NULL; | |
897 | DTD *oldDtd = _dtd; | |
898 | XML_StartElementHandler oldStartElementHandler = startElementHandler; | |
899 | XML_EndElementHandler oldEndElementHandler = endElementHandler; | |
900 | XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler; | |
901 | XML_ProcessingInstructionHandler oldProcessingInstructionHandler | |
902 | = processingInstructionHandler; | |
903 | XML_CommentHandler oldCommentHandler = commentHandler; | |
904 | XML_StartCdataSectionHandler oldStartCdataSectionHandler | |
905 | = startCdataSectionHandler; | |
906 | XML_EndCdataSectionHandler oldEndCdataSectionHandler | |
907 | = endCdataSectionHandler; | |
908 | XML_DefaultHandler oldDefaultHandler = defaultHandler; | |
909 | XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler | |
910 | = unparsedEntityDeclHandler; | |
911 | XML_NotationDeclHandler oldNotationDeclHandler = notationDeclHandler; | |
912 | XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler | |
913 | = startNamespaceDeclHandler; | |
914 | XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler | |
915 | = endNamespaceDeclHandler; | |
916 | XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler; | |
917 | XML_ExternalEntityRefHandler oldExternalEntityRefHandler | |
918 | = externalEntityRefHandler; | |
919 | XML_SkippedEntityHandler oldSkippedEntityHandler = skippedEntityHandler; | |
920 | XML_UnknownEncodingHandler oldUnknownEncodingHandler | |
921 | = unknownEncodingHandler; | |
922 | XML_ElementDeclHandler oldElementDeclHandler = elementDeclHandler; | |
923 | XML_AttlistDeclHandler oldAttlistDeclHandler = attlistDeclHandler; | |
924 | XML_EntityDeclHandler oldEntityDeclHandler = entityDeclHandler; | |
925 | XML_XmlDeclHandler oldXmlDeclHandler = xmlDeclHandler; | |
926 | ELEMENT_TYPE * oldDeclElementType = declElementType; | |
927 | ||
928 | void *oldUserData = userData; | |
929 | void *oldHandlerArg = handlerArg; | |
930 | XML_Bool oldDefaultExpandInternalEntities = defaultExpandInternalEntities; | |
931 | XML_Parser oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg; | |
932 | #ifdef XML_DTD | |
933 | enum XML_ParamEntityParsing oldParamEntityParsing = paramEntityParsing; | |
934 | int oldInEntityValue = prologState.inEntityValue; | |
935 | #endif | |
936 | XML_Bool oldns_triplets = ns_triplets; | |
937 | ||
938 | #ifdef XML_DTD | |
939 | if (!context) | |
940 | newDtd = oldDtd; | |
941 | #endif /* XML_DTD */ | |
942 | ||
943 | /* Note that the magical uses of the pre-processor to make field | |
944 | access look more like C++ require that `parser' be overwritten | |
945 | here. This makes this function more painful to follow than it | |
946 | would be otherwise. | |
947 | */ | |
948 | if (ns) { | |
949 | XML_Char tmp[2]; | |
950 | *tmp = namespaceSeparator; | |
951 | parser = parserCreate(encodingName, &parser->m_mem, tmp, newDtd); | |
952 | } | |
953 | else { | |
954 | parser = parserCreate(encodingName, &parser->m_mem, NULL, newDtd); | |
955 | } | |
956 | ||
957 | if (!parser) | |
958 | return NULL; | |
959 | ||
960 | startElementHandler = oldStartElementHandler; | |
961 | endElementHandler = oldEndElementHandler; | |
962 | characterDataHandler = oldCharacterDataHandler; | |
963 | processingInstructionHandler = oldProcessingInstructionHandler; | |
964 | commentHandler = oldCommentHandler; | |
965 | startCdataSectionHandler = oldStartCdataSectionHandler; | |
966 | endCdataSectionHandler = oldEndCdataSectionHandler; | |
967 | defaultHandler = oldDefaultHandler; | |
968 | unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler; | |
969 | notationDeclHandler = oldNotationDeclHandler; | |
970 | startNamespaceDeclHandler = oldStartNamespaceDeclHandler; | |
971 | endNamespaceDeclHandler = oldEndNamespaceDeclHandler; | |
972 | notStandaloneHandler = oldNotStandaloneHandler; | |
973 | externalEntityRefHandler = oldExternalEntityRefHandler; | |
974 | skippedEntityHandler = oldSkippedEntityHandler; | |
975 | unknownEncodingHandler = oldUnknownEncodingHandler; | |
976 | elementDeclHandler = oldElementDeclHandler; | |
977 | attlistDeclHandler = oldAttlistDeclHandler; | |
978 | entityDeclHandler = oldEntityDeclHandler; | |
979 | xmlDeclHandler = oldXmlDeclHandler; | |
980 | declElementType = oldDeclElementType; | |
981 | userData = oldUserData; | |
982 | if (oldUserData == oldHandlerArg) | |
983 | handlerArg = userData; | |
984 | else | |
985 | handlerArg = parser; | |
986 | if (oldExternalEntityRefHandlerArg != oldParser) | |
987 | externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; | |
988 | defaultExpandInternalEntities = oldDefaultExpandInternalEntities; | |
989 | ns_triplets = oldns_triplets; | |
990 | parentParser = oldParser; | |
991 | #ifdef XML_DTD | |
992 | paramEntityParsing = oldParamEntityParsing; | |
993 | prologState.inEntityValue = oldInEntityValue; | |
994 | if (context) { | |
995 | #endif /* XML_DTD */ | |
996 | if (!dtdCopy(_dtd, oldDtd, &parser->m_mem) | |
997 | || !setContext(parser, context)) { | |
998 | XML_ParserFree(parser); | |
999 | return NULL; | |
1000 | } | |
1001 | processor = externalEntityInitProcessor; | |
1002 | #ifdef XML_DTD | |
1003 | } | |
1004 | else { | |
1005 | /* The DTD instance referenced by _dtd is shared between the document's | |
1006 | root parser and external PE parsers, therefore one does not need to | |
1007 | call setContext. In addition, one also *must* not call setContext, | |
1008 | because this would overwrite existing prefix->binding pointers in | |
1009 | _dtd with ones that get destroyed with the external PE parser. | |
1010 | This would leave those prefixes with dangling pointers. | |
1011 | */ | |
1012 | isParamEntity = XML_TRUE; | |
1013 | XmlPrologStateInitExternalEntity(&prologState); | |
1014 | processor = externalParEntInitProcessor; | |
1015 | } | |
1016 | #endif /* XML_DTD */ | |
1017 | return parser; | |
1018 | } | |
1019 | ||
1020 | static void FASTCALL | |
1021 | destroyBindings(BINDING *bindings, XML_Parser parser) | |
1022 | { | |
1023 | for (;;) { | |
1024 | BINDING *b = bindings; | |
1025 | if (!b) | |
1026 | break; | |
1027 | bindings = b->nextTagBinding; | |
1028 | FREE(b->uri); | |
1029 | FREE(b); | |
1030 | } | |
1031 | } | |
1032 | ||
1033 | void | |
1034 | XML_ParserFree(XML_Parser parser) | |
1035 | { | |
1036 | for (;;) { | |
1037 | TAG *p; | |
1038 | if (tagStack == NULL) { | |
1039 | if (freeTagList == NULL) | |
1040 | break; | |
1041 | tagStack = freeTagList; | |
1042 | freeTagList = NULL; | |
1043 | } | |
1044 | p = tagStack; | |
1045 | tagStack = tagStack->parent; | |
1046 | FREE(p->buf); | |
1047 | destroyBindings(p->bindings, parser); | |
1048 | FREE(p); | |
1049 | } | |
1050 | destroyBindings(freeBindingList, parser); | |
1051 | destroyBindings(inheritedBindings, parser); | |
1052 | poolDestroy(&tempPool); | |
1053 | poolDestroy(&temp2Pool); | |
1054 | #ifdef XML_DTD | |
1055 | /* external parameter entity parsers share the DTD structure | |
1056 | parser->m_dtd with the root parser, so we must not destroy it | |
1057 | */ | |
1058 | if (!isParamEntity && _dtd) | |
1059 | #else | |
1060 | if (_dtd) | |
1061 | #endif /* XML_DTD */ | |
1062 | dtdDestroy(_dtd, (XML_Bool)!parentParser, &parser->m_mem); | |
1063 | FREE((void *)atts); | |
1064 | if (groupConnector) | |
1065 | FREE(groupConnector); | |
1066 | if (buffer) | |
1067 | FREE(buffer); | |
1068 | FREE(dataBuf); | |
1069 | if (unknownEncodingMem) | |
1070 | FREE(unknownEncodingMem); | |
1071 | if (unknownEncodingRelease) | |
1072 | unknownEncodingRelease(unknownEncodingData); | |
1073 | FREE(parser); | |
1074 | } | |
1075 | ||
1076 | void | |
1077 | XML_UseParserAsHandlerArg(XML_Parser parser) | |
1078 | { | |
1079 | handlerArg = parser; | |
1080 | } | |
1081 | ||
1082 | enum XML_Error | |
1083 | XML_UseForeignDTD(XML_Parser parser, XML_Bool useDTD) | |
1084 | { | |
1085 | #ifdef XML_DTD | |
1086 | /* block after XML_Parse()/XML_ParseBuffer() has been called */ | |
1087 | if (parsing) | |
1088 | return XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING; | |
1089 | useForeignDTD = useDTD; | |
1090 | return XML_ERROR_NONE; | |
1091 | #else | |
1092 | return XML_ERROR_FEATURE_REQUIRES_XML_DTD; | |
1093 | #endif | |
1094 | } | |
1095 | ||
1096 | void | |
1097 | XML_SetReturnNSTriplet(XML_Parser parser, int do_nst) | |
1098 | { | |
1099 | /* block after XML_Parse()/XML_ParseBuffer() has been called */ | |
1100 | if (parsing) | |
1101 | return; | |
1102 | ns_triplets = do_nst ? XML_TRUE : XML_FALSE; | |
1103 | } | |
1104 | ||
1105 | void | |
1106 | XML_SetUserData(XML_Parser parser, void *p) | |
1107 | { | |
1108 | if (handlerArg == userData) | |
1109 | handlerArg = userData = p; | |
1110 | else | |
1111 | userData = p; | |
1112 | } | |
1113 | ||
1114 | enum XML_Status | |
1115 | XML_SetBase(XML_Parser parser, const XML_Char *p) | |
1116 | { | |
1117 | if (p) { | |
1118 | p = poolCopyString(&_dtd->pool, p); | |
1119 | if (!p) | |
1120 | return XML_STATUS_ERROR; | |
1121 | curBase = p; | |
1122 | } | |
1123 | else | |
1124 | curBase = NULL; | |
1125 | return XML_STATUS_OK; | |
1126 | } | |
1127 | ||
1128 | const XML_Char * | |
1129 | XML_GetBase(XML_Parser parser) | |
1130 | { | |
1131 | return curBase; | |
1132 | } | |
1133 | ||
1134 | int | |
1135 | XML_GetSpecifiedAttributeCount(XML_Parser parser) | |
1136 | { | |
1137 | return nSpecifiedAtts; | |
1138 | } | |
1139 | ||
1140 | int | |
1141 | XML_GetIdAttributeIndex(XML_Parser parser) | |
1142 | { | |
1143 | return idAttIndex; | |
1144 | } | |
1145 | ||
1146 | void | |
1147 | XML_SetElementHandler(XML_Parser parser, | |
1148 | XML_StartElementHandler start, | |
1149 | XML_EndElementHandler end) | |
1150 | { | |
1151 | startElementHandler = start; | |
1152 | endElementHandler = end; | |
1153 | } | |
1154 | ||
1155 | void | |
1156 | XML_SetStartElementHandler(XML_Parser parser, | |
1157 | XML_StartElementHandler start) { | |
1158 | startElementHandler = start; | |
1159 | } | |
1160 | ||
1161 | void | |
1162 | XML_SetEndElementHandler(XML_Parser parser, | |
1163 | XML_EndElementHandler end) { | |
1164 | endElementHandler = end; | |
1165 | } | |
1166 | ||
1167 | void | |
1168 | XML_SetCharacterDataHandler(XML_Parser parser, | |
1169 | XML_CharacterDataHandler handler) | |
1170 | { | |
1171 | characterDataHandler = handler; | |
1172 | } | |
1173 | ||
1174 | void | |
1175 | XML_SetProcessingInstructionHandler(XML_Parser parser, | |
1176 | XML_ProcessingInstructionHandler handler) | |
1177 | { | |
1178 | processingInstructionHandler = handler; | |
1179 | } | |
1180 | ||
1181 | void | |
1182 | XML_SetCommentHandler(XML_Parser parser, | |
1183 | XML_CommentHandler handler) | |
1184 | { | |
1185 | commentHandler = handler; | |
1186 | } | |
1187 | ||
1188 | void | |
1189 | XML_SetCdataSectionHandler(XML_Parser parser, | |
1190 | XML_StartCdataSectionHandler start, | |
1191 | XML_EndCdataSectionHandler end) | |
1192 | { | |
1193 | startCdataSectionHandler = start; | |
1194 | endCdataSectionHandler = end; | |
1195 | } | |
1196 | ||
1197 | void | |
1198 | XML_SetStartCdataSectionHandler(XML_Parser parser, | |
1199 | XML_StartCdataSectionHandler start) { | |
1200 | startCdataSectionHandler = start; | |
1201 | } | |
1202 | ||
1203 | void | |
1204 | XML_SetEndCdataSectionHandler(XML_Parser parser, | |
1205 | XML_EndCdataSectionHandler end) { | |
1206 | endCdataSectionHandler = end; | |
1207 | } | |
1208 | ||
1209 | void | |
1210 | XML_SetDefaultHandler(XML_Parser parser, | |
1211 | XML_DefaultHandler handler) | |
1212 | { | |
1213 | defaultHandler = handler; | |
1214 | defaultExpandInternalEntities = XML_FALSE; | |
1215 | } | |
1216 | ||
1217 | void | |
1218 | XML_SetDefaultHandlerExpand(XML_Parser parser, | |
1219 | XML_DefaultHandler handler) | |
1220 | { | |
1221 | defaultHandler = handler; | |
1222 | defaultExpandInternalEntities = XML_TRUE; | |
1223 | } | |
1224 | ||
1225 | void | |
1226 | XML_SetDoctypeDeclHandler(XML_Parser parser, | |
1227 | XML_StartDoctypeDeclHandler start, | |
1228 | XML_EndDoctypeDeclHandler end) | |
1229 | { | |
1230 | startDoctypeDeclHandler = start; | |
1231 | endDoctypeDeclHandler = end; | |
1232 | } | |
1233 | ||
1234 | void | |
1235 | XML_SetStartDoctypeDeclHandler(XML_Parser parser, | |
1236 | XML_StartDoctypeDeclHandler start) { | |
1237 | startDoctypeDeclHandler = start; | |
1238 | } | |
1239 | ||
1240 | void | |
1241 | XML_SetEndDoctypeDeclHandler(XML_Parser parser, | |
1242 | XML_EndDoctypeDeclHandler end) { | |
1243 | endDoctypeDeclHandler = end; | |
1244 | } | |
1245 | ||
1246 | void | |
1247 | XML_SetUnparsedEntityDeclHandler(XML_Parser parser, | |
1248 | XML_UnparsedEntityDeclHandler handler) | |
1249 | { | |
1250 | unparsedEntityDeclHandler = handler; | |
1251 | } | |
1252 | ||
1253 | void | |
1254 | XML_SetNotationDeclHandler(XML_Parser parser, | |
1255 | XML_NotationDeclHandler handler) | |
1256 | { | |
1257 | notationDeclHandler = handler; | |
1258 | } | |
1259 | ||
1260 | void | |
1261 | XML_SetNamespaceDeclHandler(XML_Parser parser, | |
1262 | XML_StartNamespaceDeclHandler start, | |
1263 | XML_EndNamespaceDeclHandler end) | |
1264 | { | |
1265 | startNamespaceDeclHandler = start; | |
1266 | endNamespaceDeclHandler = end; | |
1267 | } | |
1268 | ||
1269 | void | |
1270 | XML_SetStartNamespaceDeclHandler(XML_Parser parser, | |
1271 | XML_StartNamespaceDeclHandler start) { | |
1272 | startNamespaceDeclHandler = start; | |
1273 | } | |
1274 | ||
1275 | void | |
1276 | XML_SetEndNamespaceDeclHandler(XML_Parser parser, | |
1277 | XML_EndNamespaceDeclHandler end) { | |
1278 | endNamespaceDeclHandler = end; | |
1279 | } | |
1280 | ||
1281 | void | |
1282 | XML_SetNotStandaloneHandler(XML_Parser parser, | |
1283 | XML_NotStandaloneHandler handler) | |
1284 | { | |
1285 | notStandaloneHandler = handler; | |
1286 | } | |
1287 | ||
1288 | void | |
1289 | XML_SetExternalEntityRefHandler(XML_Parser parser, | |
1290 | XML_ExternalEntityRefHandler handler) | |
1291 | { | |
1292 | externalEntityRefHandler = handler; | |
1293 | } | |
1294 | ||
1295 | void | |
1296 | XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) | |
1297 | { | |
1298 | if (arg) | |
1299 | externalEntityRefHandlerArg = (XML_Parser)arg; | |
1300 | else | |
1301 | externalEntityRefHandlerArg = parser; | |
1302 | } | |
1303 | ||
1304 | void | |
1305 | XML_SetSkippedEntityHandler(XML_Parser parser, | |
1306 | XML_SkippedEntityHandler handler) | |
1307 | { | |
1308 | skippedEntityHandler = handler; | |
1309 | } | |
1310 | ||
1311 | void | |
1312 | XML_SetUnknownEncodingHandler(XML_Parser parser, | |
1313 | XML_UnknownEncodingHandler handler, | |
1314 | void *data) | |
1315 | { | |
1316 | unknownEncodingHandler = handler; | |
1317 | unknownEncodingHandlerData = data; | |
1318 | } | |
1319 | ||
1320 | void | |
1321 | XML_SetElementDeclHandler(XML_Parser parser, | |
1322 | XML_ElementDeclHandler eldecl) | |
1323 | { | |
1324 | elementDeclHandler = eldecl; | |
1325 | } | |
1326 | ||
1327 | void | |
1328 | XML_SetAttlistDeclHandler(XML_Parser parser, | |
1329 | XML_AttlistDeclHandler attdecl) | |
1330 | { | |
1331 | attlistDeclHandler = attdecl; | |
1332 | } | |
1333 | ||
1334 | void | |
1335 | XML_SetEntityDeclHandler(XML_Parser parser, | |
1336 | XML_EntityDeclHandler handler) | |
1337 | { | |
1338 | entityDeclHandler = handler; | |
1339 | } | |
1340 | ||
1341 | void | |
1342 | XML_SetXmlDeclHandler(XML_Parser parser, | |
1343 | XML_XmlDeclHandler handler) { | |
1344 | xmlDeclHandler = handler; | |
1345 | } | |
1346 | ||
1347 | int | |
1348 | XML_SetParamEntityParsing(XML_Parser parser, | |
1349 | enum XML_ParamEntityParsing peParsing) | |
1350 | { | |
1351 | /* block after XML_Parse()/XML_ParseBuffer() has been called */ | |
1352 | if (parsing) | |
1353 | return 0; | |
1354 | #ifdef XML_DTD | |
1355 | paramEntityParsing = peParsing; | |
1356 | return 1; | |
1357 | #else | |
1358 | return peParsing == XML_PARAM_ENTITY_PARSING_NEVER; | |
1359 | #endif | |
1360 | } | |
1361 | ||
1362 | enum XML_Status | |
1363 | XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) | |
1364 | { | |
1365 | if (len == 0) { | |
1366 | if (!isFinal) | |
1367 | return XML_STATUS_OK; | |
1368 | positionPtr = bufferPtr; | |
1369 | errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0); | |
1370 | if (errorCode == XML_ERROR_NONE) | |
1371 | return XML_STATUS_OK; | |
1372 | eventEndPtr = eventPtr; | |
1373 | processor = errorProcessor; | |
1374 | return XML_STATUS_ERROR; | |
1375 | } | |
1376 | #ifndef XML_CONTEXT_BYTES | |
1377 | else if (bufferPtr == bufferEnd) { | |
1378 | const char *end; | |
1379 | int nLeftOver; | |
1380 | parseEndByteIndex += len; | |
1381 | positionPtr = s; | |
1382 | if (isFinal) { | |
1383 | errorCode = processor(parser, s, parseEndPtr = s + len, 0); | |
1384 | if (errorCode == XML_ERROR_NONE) | |
1385 | return XML_STATUS_OK; | |
1386 | eventEndPtr = eventPtr; | |
1387 | processor = errorProcessor; | |
1388 | return XML_STATUS_ERROR; | |
1389 | } | |
1390 | errorCode = processor(parser, s, parseEndPtr = s + len, &end); | |
1391 | if (errorCode != XML_ERROR_NONE) { | |
1392 | eventEndPtr = eventPtr; | |
1393 | processor = errorProcessor; | |
1394 | return XML_STATUS_ERROR; | |
1395 | } | |
1396 | XmlUpdatePosition(encoding, positionPtr, end, &position); | |
1397 | positionPtr = end; | |
1398 | nLeftOver = s + len - end; | |
1399 | if (nLeftOver) { | |
1400 | if (buffer == NULL || nLeftOver > bufferLim - buffer) { | |
1401 | /* FIXME avoid integer overflow */ | |
1402 | char *temp; | |
1403 | temp = (buffer == NULL | |
1404 | ? (char *)MALLOC(len * 2) | |
1405 | : (char *)REALLOC(buffer, len * 2)); | |
1406 | if (temp == NULL) { | |
1407 | errorCode = XML_ERROR_NO_MEMORY; | |
1408 | return XML_STATUS_ERROR; | |
1409 | } | |
1410 | buffer = temp; | |
1411 | if (!buffer) { | |
1412 | errorCode = XML_ERROR_NO_MEMORY; | |
1413 | eventPtr = eventEndPtr = NULL; | |
1414 | processor = errorProcessor; | |
1415 | return XML_STATUS_ERROR; | |
1416 | } | |
1417 | bufferLim = buffer + len * 2; | |
1418 | } | |
1419 | memcpy(buffer, end, nLeftOver); | |
1420 | bufferPtr = buffer; | |
1421 | bufferEnd = buffer + nLeftOver; | |
1422 | } | |
1423 | return XML_STATUS_OK; | |
1424 | } | |
1425 | #endif /* not defined XML_CONTEXT_BYTES */ | |
1426 | else { | |
1427 | void *buff = XML_GetBuffer(parser, len); | |
1428 | if (buff == NULL) | |
1429 | return XML_STATUS_ERROR; | |
1430 | else { | |
1431 | memcpy(buff, s, len); | |
1432 | return XML_ParseBuffer(parser, len, isFinal); | |
1433 | } | |
1434 | } | |
1435 | } | |
1436 | ||
1437 | enum XML_Status | |
1438 | XML_ParseBuffer(XML_Parser parser, int len, int isFinal) | |
1439 | { | |
1440 | const char *start = bufferPtr; | |
1441 | positionPtr = start; | |
1442 | bufferEnd += len; | |
1443 | parseEndByteIndex += len; | |
1444 | errorCode = processor(parser, start, parseEndPtr = bufferEnd, | |
1445 | isFinal ? (const char **)NULL : &bufferPtr); | |
1446 | if (errorCode == XML_ERROR_NONE) { | |
1447 | if (!isFinal) { | |
1448 | XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position); | |
1449 | positionPtr = bufferPtr; | |
1450 | } | |
1451 | return XML_STATUS_OK; | |
1452 | } | |
1453 | else { | |
1454 | eventEndPtr = eventPtr; | |
1455 | processor = errorProcessor; | |
1456 | return XML_STATUS_ERROR; | |
1457 | } | |
1458 | } | |
1459 | ||
1460 | void * | |
1461 | XML_GetBuffer(XML_Parser parser, int len) | |
1462 | { | |
1463 | if (len > bufferLim - bufferEnd) { | |
1464 | /* FIXME avoid integer overflow */ | |
1465 | int neededSize = len + (bufferEnd - bufferPtr); | |
1466 | #ifdef XML_CONTEXT_BYTES | |
1467 | int keep = bufferPtr - buffer; | |
1468 | ||
1469 | if (keep > XML_CONTEXT_BYTES) | |
1470 | keep = XML_CONTEXT_BYTES; | |
1471 | neededSize += keep; | |
1472 | #endif /* defined XML_CONTEXT_BYTES */ | |
1473 | if (neededSize <= bufferLim - buffer) { | |
1474 | #ifdef XML_CONTEXT_BYTES | |
1475 | if (keep < bufferPtr - buffer) { | |
1476 | int offset = (bufferPtr - buffer) - keep; | |
1477 | memmove(buffer, &buffer[offset], bufferEnd - bufferPtr + keep); | |
1478 | bufferEnd -= offset; | |
1479 | bufferPtr -= offset; | |
1480 | } | |
1481 | #else | |
1482 | memmove(buffer, bufferPtr, bufferEnd - bufferPtr); | |
1483 | bufferEnd = buffer + (bufferEnd - bufferPtr); | |
1484 | bufferPtr = buffer; | |
1485 | #endif /* not defined XML_CONTEXT_BYTES */ | |
1486 | } | |
1487 | else { | |
1488 | char *newBuf; | |
1489 | int bufferSize = bufferLim - bufferPtr; | |
1490 | if (bufferSize == 0) | |
1491 | bufferSize = INIT_BUFFER_SIZE; | |
1492 | do { | |
1493 | bufferSize *= 2; | |
1494 | } while (bufferSize < neededSize); | |
1495 | newBuf = (char *)MALLOC(bufferSize); | |
1496 | if (newBuf == 0) { | |
1497 | errorCode = XML_ERROR_NO_MEMORY; | |
1498 | return NULL; | |
1499 | } | |
1500 | bufferLim = newBuf + bufferSize; | |
1501 | #ifdef XML_CONTEXT_BYTES | |
1502 | if (bufferPtr) { | |
1503 | int keep = bufferPtr - buffer; | |
1504 | if (keep > XML_CONTEXT_BYTES) | |
1505 | keep = XML_CONTEXT_BYTES; | |
1506 | memcpy(newBuf, &bufferPtr[-keep], bufferEnd - bufferPtr + keep); | |
1507 | FREE(buffer); | |
1508 | buffer = newBuf; | |
1509 | bufferEnd = buffer + (bufferEnd - bufferPtr) + keep; | |
1510 | bufferPtr = buffer + keep; | |
1511 | } | |
1512 | else { | |
1513 | bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
1514 | bufferPtr = buffer = newBuf; | |
1515 | } | |
1516 | #else | |
1517 | if (bufferPtr) { | |
1518 | memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr); | |
1519 | FREE(buffer); | |
1520 | } | |
1521 | bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
1522 | bufferPtr = buffer = newBuf; | |
1523 | #endif /* not defined XML_CONTEXT_BYTES */ | |
1524 | } | |
1525 | } | |
1526 | return bufferEnd; | |
1527 | } | |
1528 | ||
1529 | enum XML_Error | |
1530 | XML_GetErrorCode(XML_Parser parser) | |
1531 | { | |
1532 | return errorCode; | |
1533 | } | |
1534 | ||
1535 | long | |
1536 | XML_GetCurrentByteIndex(XML_Parser parser) | |
1537 | { | |
1538 | if (eventPtr) | |
1539 | return parseEndByteIndex - (parseEndPtr - eventPtr); | |
1540 | return -1; | |
1541 | } | |
1542 | ||
1543 | int | |
1544 | XML_GetCurrentByteCount(XML_Parser parser) | |
1545 | { | |
1546 | if (eventEndPtr && eventPtr) | |
1547 | return eventEndPtr - eventPtr; | |
1548 | return 0; | |
1549 | } | |
1550 | ||
1551 | const char * | |
1552 | XML_GetInputContext(XML_Parser parser, int *offset, int *size) | |
1553 | { | |
1554 | #ifdef XML_CONTEXT_BYTES | |
1555 | if (eventPtr && buffer) { | |
1556 | *offset = eventPtr - buffer; | |
1557 | *size = bufferEnd - buffer; | |
1558 | return buffer; | |
1559 | } | |
1560 | #endif /* defined XML_CONTEXT_BYTES */ | |
1561 | return (char *) 0; | |
1562 | } | |
1563 | ||
1564 | int | |
1565 | XML_GetCurrentLineNumber(XML_Parser parser) | |
1566 | { | |
1567 | if (eventPtr) { | |
1568 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
1569 | positionPtr = eventPtr; | |
1570 | } | |
1571 | return position.lineNumber + 1; | |
1572 | } | |
1573 | ||
1574 | int | |
1575 | XML_GetCurrentColumnNumber(XML_Parser parser) | |
1576 | { | |
1577 | if (eventPtr) { | |
1578 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
1579 | positionPtr = eventPtr; | |
1580 | } | |
1581 | return position.columnNumber; | |
1582 | } | |
1583 | ||
1584 | void | |
1585 | XML_FreeContentModel(XML_Parser parser, XML_Content *model) | |
1586 | { | |
1587 | FREE(model); | |
1588 | } | |
1589 | ||
1590 | void * | |
1591 | XML_MemMalloc(XML_Parser parser, size_t size) | |
1592 | { | |
1593 | return MALLOC(size); | |
1594 | } | |
1595 | ||
1596 | void * | |
1597 | XML_MemRealloc(XML_Parser parser, void *ptr, size_t size) | |
1598 | { | |
1599 | return REALLOC(ptr, size); | |
1600 | } | |
1601 | ||
1602 | void | |
1603 | XML_MemFree(XML_Parser parser, void *ptr) | |
1604 | { | |
1605 | FREE(ptr); | |
1606 | } | |
1607 | ||
1608 | void | |
1609 | XML_DefaultCurrent(XML_Parser parser) | |
1610 | { | |
1611 | if (defaultHandler) { | |
1612 | if (openInternalEntities) | |
1613 | reportDefault(parser, | |
1614 | internalEncoding, | |
1615 | openInternalEntities->internalEventPtr, | |
1616 | openInternalEntities->internalEventEndPtr); | |
1617 | else | |
1618 | reportDefault(parser, encoding, eventPtr, eventEndPtr); | |
1619 | } | |
1620 | } | |
1621 | ||
1622 | const XML_LChar * | |
1623 | XML_ErrorString(enum XML_Error code) | |
1624 | { | |
1625 | static const XML_LChar *message[] = { | |
1626 | 0, | |
1627 | XML_L("out of memory"), | |
1628 | XML_L("syntax error"), | |
1629 | XML_L("no element found"), | |
1630 | XML_L("not well-formed (invalid token)"), | |
1631 | XML_L("unclosed token"), | |
1632 | XML_L("partial character"), | |
1633 | XML_L("mismatched tag"), | |
1634 | XML_L("duplicate attribute"), | |
1635 | XML_L("junk after document element"), | |
1636 | XML_L("illegal parameter entity reference"), | |
1637 | XML_L("undefined entity"), | |
1638 | XML_L("recursive entity reference"), | |
1639 | XML_L("asynchronous entity"), | |
1640 | XML_L("reference to invalid character number"), | |
1641 | XML_L("reference to binary entity"), | |
1642 | XML_L("reference to external entity in attribute"), | |
1643 | XML_L("xml declaration not at start of external entity"), | |
1644 | XML_L("unknown encoding"), | |
1645 | XML_L("encoding specified in XML declaration is incorrect"), | |
1646 | XML_L("unclosed CDATA section"), | |
1647 | XML_L("error in processing external entity reference"), | |
1648 | XML_L("document is not standalone"), | |
1649 | XML_L("unexpected parser state - please send a bug report"), | |
1650 | XML_L("entity declared in parameter entity"), | |
1651 | XML_L("requested feature requires XML_DTD support in Expat"), | |
1652 | XML_L("cannot change setting once parsing has begun") | |
1653 | }; | |
1654 | if (code > 0 && code < sizeof(message)/sizeof(message[0])) | |
1655 | return message[code]; | |
1656 | return NULL; | |
1657 | } | |
1658 | ||
1659 | const XML_LChar * | |
1660 | XML_ExpatVersion(void) { | |
1661 | ||
1662 | /* V1 is used to string-ize the version number. However, it would | |
1663 | string-ize the actual version macro *names* unless we get them | |
1664 | substituted before being passed to V1. CPP is defined to expand | |
1665 | a macro, then rescan for more expansions. Thus, we use V2 to expand | |
1666 | the version macros, then CPP will expand the resulting V1() macro | |
1667 | with the correct numerals. */ | |
1668 | /* ### I'm assuming cpp is portable in this respect... */ | |
1669 | ||
1670 | #define V1(a,b,c) XML_L(#a)XML_L(".")XML_L(#b)XML_L(".")XML_L(#c) | |
1671 | #define V2(a,b,c) XML_L("expat_")V1(a,b,c) | |
1672 | ||
1673 | return V2(XML_MAJOR_VERSION, XML_MINOR_VERSION, XML_MICRO_VERSION); | |
1674 | ||
1675 | #undef V1 | |
1676 | #undef V2 | |
1677 | } | |
1678 | ||
1679 | XML_Expat_Version | |
1680 | XML_ExpatVersionInfo(void) | |
1681 | { | |
1682 | XML_Expat_Version version; | |
1683 | ||
1684 | version.major = XML_MAJOR_VERSION; | |
1685 | version.minor = XML_MINOR_VERSION; | |
1686 | version.micro = XML_MICRO_VERSION; | |
1687 | ||
1688 | return version; | |
1689 | } | |
1690 | ||
1691 | const XML_Feature * | |
1692 | XML_GetFeatureList(void) | |
1693 | { | |
1694 | static XML_Feature features[] = { | |
1695 | {XML_FEATURE_SIZEOF_XML_CHAR, XML_L("sizeof(XML_Char)")}, | |
1696 | {XML_FEATURE_SIZEOF_XML_LCHAR, XML_L("sizeof(XML_LChar)")}, | |
1697 | #ifdef XML_UNICODE | |
1698 | {XML_FEATURE_UNICODE, XML_L("XML_UNICODE")}, | |
1699 | #endif | |
1700 | #ifdef XML_UNICODE_WCHAR_T | |
1701 | {XML_FEATURE_UNICODE_WCHAR_T, XML_L("XML_UNICODE_WCHAR_T")}, | |
1702 | #endif | |
1703 | #ifdef XML_DTD | |
1704 | {XML_FEATURE_DTD, XML_L("XML_DTD")}, | |
1705 | #endif | |
1706 | #ifdef XML_CONTEXT_BYTES | |
1707 | {XML_FEATURE_CONTEXT_BYTES, XML_L("XML_CONTEXT_BYTES"), | |
1708 | XML_CONTEXT_BYTES}, | |
1709 | #endif | |
1710 | #ifdef XML_MIN_SIZE | |
1711 | {XML_FEATURE_MIN_SIZE, XML_L("XML_MIN_SIZE")}, | |
1712 | #endif | |
1713 | {XML_FEATURE_END, NULL} | |
1714 | }; | |
1715 | ||
1716 | features[0].value = sizeof(XML_Char); | |
1717 | features[1].value = sizeof(XML_LChar); | |
1718 | return features; | |
1719 | } | |
1720 | ||
1721 | /* Initially tag->rawName always points into the parse buffer; | |
1722 | for those TAG instances opened while the current parse buffer was | |
1723 | processed, and not yet closed, we need to store tag->rawName in a more | |
1724 | permanent location, since the parse buffer is about to be discarded. | |
1725 | */ | |
1726 | static XML_Bool | |
1727 | storeRawNames(XML_Parser parser) | |
1728 | { | |
1729 | TAG *tag = tagStack; | |
1730 | while (tag) { | |
1731 | int bufSize; | |
1732 | int nameLen = sizeof(XML_Char) * (tag->name.strLen + 1); | |
1733 | char *rawNameBuf = tag->buf + nameLen; | |
1734 | /* Stop if already stored. Since tagStack is a stack, we can stop | |
1735 | at the first entry that has already been copied; everything | |
1736 | below it in the stack is already been accounted for in a | |
1737 | previous call to this function. | |
1738 | */ | |
1739 | if (tag->rawName == rawNameBuf) | |
1740 | break; | |
1741 | /* For re-use purposes we need to ensure that the | |
1742 | size of tag->buf is a multiple of sizeof(XML_Char). | |
1743 | */ | |
1744 | bufSize = nameLen + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)); | |
1745 | if (bufSize > tag->bufEnd - tag->buf) { | |
1746 | char *temp = (char *)REALLOC(tag->buf, bufSize); | |
1747 | if (temp == NULL) | |
1748 | return XML_FALSE; | |
1749 | /* if tag->name.str points to tag->buf (only when namespace | |
1750 | processing is off) then we have to update it | |
1751 | */ | |
1752 | if (tag->name.str == (XML_Char *)tag->buf) | |
1753 | tag->name.str = (XML_Char *)temp; | |
1754 | /* if tag->name.localPart is set (when namespace processing is on) | |
1755 | then update it as well, since it will always point into tag->buf | |
1756 | */ | |
1757 | if (tag->name.localPart) | |
1758 | tag->name.localPart = (XML_Char *)temp + (tag->name.localPart - | |
1759 | (XML_Char *)tag->buf); | |
1760 | tag->buf = temp; | |
1761 | tag->bufEnd = temp + bufSize; | |
1762 | rawNameBuf = temp + nameLen; | |
1763 | } | |
1764 | memcpy(rawNameBuf, tag->rawName, tag->rawNameLength); | |
1765 | tag->rawName = rawNameBuf; | |
1766 | tag = tag->parent; | |
1767 | } | |
1768 | return XML_TRUE; | |
1769 | } | |
1770 | ||
1771 | static enum XML_Error PTRCALL | |
1772 | contentProcessor(XML_Parser parser, | |
1773 | const char *start, | |
1774 | const char *end, | |
1775 | const char **endPtr) | |
1776 | { | |
1777 | enum XML_Error result = | |
1778 | doContent(parser, 0, encoding, start, end, endPtr); | |
1779 | if (result != XML_ERROR_NONE) | |
1780 | return result; | |
1781 | if (!storeRawNames(parser)) | |
1782 | return XML_ERROR_NO_MEMORY; | |
1783 | return result; | |
1784 | } | |
1785 | ||
1786 | static enum XML_Error PTRCALL | |
1787 | externalEntityInitProcessor(XML_Parser parser, | |
1788 | const char *start, | |
1789 | const char *end, | |
1790 | const char **endPtr) | |
1791 | { | |
1792 | enum XML_Error result = initializeEncoding(parser); | |
1793 | if (result != XML_ERROR_NONE) | |
1794 | return result; | |
1795 | processor = externalEntityInitProcessor2; | |
1796 | return externalEntityInitProcessor2(parser, start, end, endPtr); | |
1797 | } | |
1798 | ||
1799 | static enum XML_Error PTRCALL | |
1800 | externalEntityInitProcessor2(XML_Parser parser, | |
1801 | const char *start, | |
1802 | const char *end, | |
1803 | const char **endPtr) | |
1804 | { | |
1805 | const char *next = start; /* XmlContentTok doesn't always set the last arg */ | |
1806 | int tok = XmlContentTok(encoding, start, end, &next); | |
1807 | switch (tok) { | |
1808 | case XML_TOK_BOM: | |
1809 | /* If we are at the end of the buffer, this would cause the next stage, | |
1810 | i.e. externalEntityInitProcessor3, to pass control directly to | |
1811 | doContent (by detecting XML_TOK_NONE) without processing any xml text | |
1812 | declaration - causing the error XML_ERROR_MISPLACED_XML_PI in doContent. | |
1813 | */ | |
1814 | if (next == end && endPtr) { | |
1815 | *endPtr = next; | |
1816 | return XML_ERROR_NONE; | |
1817 | } | |
1818 | start = next; | |
1819 | break; | |
1820 | case XML_TOK_PARTIAL: | |
1821 | if (endPtr) { | |
1822 | *endPtr = start; | |
1823 | return XML_ERROR_NONE; | |
1824 | } | |
1825 | eventPtr = start; | |
1826 | return XML_ERROR_UNCLOSED_TOKEN; | |
1827 | case XML_TOK_PARTIAL_CHAR: | |
1828 | if (endPtr) { | |
1829 | *endPtr = start; | |
1830 | return XML_ERROR_NONE; | |
1831 | } | |
1832 | eventPtr = start; | |
1833 | return XML_ERROR_PARTIAL_CHAR; | |
1834 | } | |
1835 | processor = externalEntityInitProcessor3; | |
1836 | return externalEntityInitProcessor3(parser, start, end, endPtr); | |
1837 | } | |
1838 | ||
1839 | static enum XML_Error PTRCALL | |
1840 | externalEntityInitProcessor3(XML_Parser parser, | |
1841 | const char *start, | |
1842 | const char *end, | |
1843 | const char **endPtr) | |
1844 | { | |
1845 | const char *next = start; /* XmlContentTok doesn't always set the last arg */ | |
1846 | int tok = XmlContentTok(encoding, start, end, &next); | |
1847 | switch (tok) { | |
1848 | case XML_TOK_XML_DECL: | |
1849 | { | |
1850 | enum XML_Error result = processXmlDecl(parser, 1, start, next); | |
1851 | if (result != XML_ERROR_NONE) | |
1852 | return result; | |
1853 | start = next; | |
1854 | } | |
1855 | break; | |
1856 | case XML_TOK_PARTIAL: | |
1857 | if (endPtr) { | |
1858 | *endPtr = start; | |
1859 | return XML_ERROR_NONE; | |
1860 | } | |
1861 | eventPtr = start; | |
1862 | return XML_ERROR_UNCLOSED_TOKEN; | |
1863 | case XML_TOK_PARTIAL_CHAR: | |
1864 | if (endPtr) { | |
1865 | *endPtr = start; | |
1866 | return XML_ERROR_NONE; | |
1867 | } | |
1868 | eventPtr = start; | |
1869 | return XML_ERROR_PARTIAL_CHAR; | |
1870 | } | |
1871 | processor = externalEntityContentProcessor; | |
1872 | tagLevel = 1; | |
1873 | return externalEntityContentProcessor(parser, start, end, endPtr); | |
1874 | } | |
1875 | ||
1876 | static enum XML_Error PTRCALL | |
1877 | externalEntityContentProcessor(XML_Parser parser, | |
1878 | const char *start, | |
1879 | const char *end, | |
1880 | const char **endPtr) | |
1881 | { | |
1882 | enum XML_Error result = | |
1883 | doContent(parser, 1, encoding, start, end, endPtr); | |
1884 | if (result != XML_ERROR_NONE) | |
1885 | return result; | |
1886 | if (!storeRawNames(parser)) | |
1887 | return XML_ERROR_NO_MEMORY; | |
1888 | return result; | |
1889 | } | |
1890 | ||
1891 | static enum XML_Error | |
1892 | doContent(XML_Parser parser, | |
1893 | int startTagLevel, | |
1894 | const ENCODING *enc, | |
1895 | const char *s, | |
1896 | const char *end, | |
1897 | const char **nextPtr) | |
1898 | { | |
1899 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
1900 | const char **eventPP; | |
1901 | const char **eventEndPP; | |
1902 | if (enc == encoding) { | |
1903 | eventPP = &eventPtr; | |
1904 | eventEndPP = &eventEndPtr; | |
1905 | } | |
1906 | else { | |
1907 | eventPP = &(openInternalEntities->internalEventPtr); | |
1908 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
1909 | } | |
1910 | *eventPP = s; | |
1911 | for (;;) { | |
1912 | const char *next = s; /* XmlContentTok doesn't always set the last arg */ | |
1913 | int tok = XmlContentTok(enc, s, end, &next); | |
1914 | *eventEndPP = next; | |
1915 | switch (tok) { | |
1916 | case XML_TOK_TRAILING_CR: | |
1917 | if (nextPtr) { | |
1918 | *nextPtr = s; | |
1919 | return XML_ERROR_NONE; | |
1920 | } | |
1921 | *eventEndPP = end; | |
1922 | if (characterDataHandler) { | |
1923 | XML_Char c = 0xA; | |
1924 | characterDataHandler(handlerArg, &c, 1); | |
1925 | } | |
1926 | else if (defaultHandler) | |
1927 | reportDefault(parser, enc, s, end); | |
1928 | if (startTagLevel == 0) | |
1929 | return XML_ERROR_NO_ELEMENTS; | |
1930 | if (tagLevel != startTagLevel) | |
1931 | return XML_ERROR_ASYNC_ENTITY; | |
1932 | return XML_ERROR_NONE; | |
1933 | case XML_TOK_NONE: | |
1934 | if (nextPtr) { | |
1935 | *nextPtr = s; | |
1936 | return XML_ERROR_NONE; | |
1937 | } | |
1938 | if (startTagLevel > 0) { | |
1939 | if (tagLevel != startTagLevel) | |
1940 | return XML_ERROR_ASYNC_ENTITY; | |
1941 | return XML_ERROR_NONE; | |
1942 | } | |
1943 | return XML_ERROR_NO_ELEMENTS; | |
1944 | case XML_TOK_INVALID: | |
1945 | *eventPP = next; | |
1946 | return XML_ERROR_INVALID_TOKEN; | |
1947 | case XML_TOK_PARTIAL: | |
1948 | if (nextPtr) { | |
1949 | *nextPtr = s; | |
1950 | return XML_ERROR_NONE; | |
1951 | } | |
1952 | return XML_ERROR_UNCLOSED_TOKEN; | |
1953 | case XML_TOK_PARTIAL_CHAR: | |
1954 | if (nextPtr) { | |
1955 | *nextPtr = s; | |
1956 | return XML_ERROR_NONE; | |
1957 | } | |
1958 | return XML_ERROR_PARTIAL_CHAR; | |
1959 | case XML_TOK_ENTITY_REF: | |
1960 | { | |
1961 | const XML_Char *name; | |
1962 | ENTITY *entity; | |
1963 | XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc, | |
1964 | s + enc->minBytesPerChar, | |
1965 | next - enc->minBytesPerChar); | |
1966 | if (ch) { | |
1967 | if (characterDataHandler) | |
1968 | characterDataHandler(handlerArg, &ch, 1); | |
1969 | else if (defaultHandler) | |
1970 | reportDefault(parser, enc, s, next); | |
1971 | break; | |
1972 | } | |
1973 | name = poolStoreString(&dtd->pool, enc, | |
1974 | s + enc->minBytesPerChar, | |
1975 | next - enc->minBytesPerChar); | |
1976 | if (!name) | |
1977 | return XML_ERROR_NO_MEMORY; | |
1978 | entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0); | |
1979 | poolDiscard(&dtd->pool); | |
1980 | /* First, determine if a check for an existing declaration is needed; | |
1981 | if yes, check that the entity exists, and that it is internal, | |
1982 | otherwise call the skipped entity or default handler. | |
1983 | */ | |
1984 | if (!dtd->hasParamEntityRefs || dtd->standalone) { | |
1985 | if (!entity) | |
1986 | return XML_ERROR_UNDEFINED_ENTITY; | |
1987 | else if (!entity->is_internal) | |
1988 | return XML_ERROR_ENTITY_DECLARED_IN_PE; | |
1989 | } | |
1990 | else if (!entity) { | |
1991 | if (skippedEntityHandler) | |
1992 | skippedEntityHandler(handlerArg, name, 0); | |
1993 | else if (defaultHandler) | |
1994 | reportDefault(parser, enc, s, next); | |
1995 | break; | |
1996 | } | |
1997 | if (entity->open) | |
1998 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
1999 | if (entity->notation) | |
2000 | return XML_ERROR_BINARY_ENTITY_REF; | |
2001 | if (entity->textPtr) { | |
2002 | enum XML_Error result; | |
2003 | OPEN_INTERNAL_ENTITY openEntity; | |
2004 | if (!defaultExpandInternalEntities) { | |
2005 | if (skippedEntityHandler) | |
2006 | skippedEntityHandler(handlerArg, entity->name, 0); | |
2007 | else if (defaultHandler) | |
2008 | reportDefault(parser, enc, s, next); | |
2009 | break; | |
2010 | } | |
2011 | entity->open = XML_TRUE; | |
2012 | openEntity.next = openInternalEntities; | |
2013 | openInternalEntities = &openEntity; | |
2014 | openEntity.entity = entity; | |
2015 | openEntity.internalEventPtr = NULL; | |
2016 | openEntity.internalEventEndPtr = NULL; | |
2017 | result = doContent(parser, | |
2018 | tagLevel, | |
2019 | internalEncoding, | |
2020 | (char *)entity->textPtr, | |
2021 | (char *)(entity->textPtr + entity->textLen), | |
2022 | 0); | |
2023 | entity->open = XML_FALSE; | |
2024 | openInternalEntities = openEntity.next; | |
2025 | if (result) | |
2026 | return result; | |
2027 | } | |
2028 | else if (externalEntityRefHandler) { | |
2029 | const XML_Char *context; | |
2030 | entity->open = XML_TRUE; | |
2031 | context = getContext(parser); | |
2032 | entity->open = XML_FALSE; | |
2033 | if (!context) | |
2034 | return XML_ERROR_NO_MEMORY; | |
2035 | if (!externalEntityRefHandler((XML_Parser)externalEntityRefHandlerArg, | |
2036 | context, | |
2037 | entity->base, | |
2038 | entity->systemId, | |
2039 | entity->publicId)) | |
2040 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
2041 | poolDiscard(&tempPool); | |
2042 | } | |
2043 | else if (defaultHandler) | |
2044 | reportDefault(parser, enc, s, next); | |
2045 | break; | |
2046 | } | |
2047 | case XML_TOK_START_TAG_NO_ATTS: | |
2048 | /* fall through */ | |
2049 | case XML_TOK_START_TAG_WITH_ATTS: | |
2050 | { | |
2051 | TAG *tag; | |
2052 | enum XML_Error result; | |
2053 | XML_Char *toPtr; | |
2054 | if (freeTagList) { | |
2055 | tag = freeTagList; | |
2056 | freeTagList = freeTagList->parent; | |
2057 | } | |
2058 | else { | |
2059 | tag = (TAG *)MALLOC(sizeof(TAG)); | |
2060 | if (!tag) | |
2061 | return XML_ERROR_NO_MEMORY; | |
2062 | tag->buf = (char *)MALLOC(INIT_TAG_BUF_SIZE); | |
2063 | if (!tag->buf) { | |
2064 | FREE(tag); | |
2065 | return XML_ERROR_NO_MEMORY; | |
2066 | } | |
2067 | tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; | |
2068 | } | |
2069 | tag->bindings = NULL; | |
2070 | tag->parent = tagStack; | |
2071 | tagStack = tag; | |
2072 | tag->name.localPart = NULL; | |
2073 | tag->name.prefix = NULL; | |
2074 | tag->rawName = s + enc->minBytesPerChar; | |
2075 | tag->rawNameLength = XmlNameLength(enc, tag->rawName); | |
2076 | ++tagLevel; | |
2077 | { | |
2078 | const char *rawNameEnd = tag->rawName + tag->rawNameLength; | |
2079 | const char *fromPtr = tag->rawName; | |
2080 | toPtr = (XML_Char *)tag->buf; | |
2081 | for (;;) { | |
2082 | int bufSize; | |
2083 | int convLen; | |
2084 | XmlConvert(enc, | |
2085 | &fromPtr, rawNameEnd, | |
2086 | (ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1); | |
2087 | convLen = toPtr - (XML_Char *)tag->buf; | |
2088 | if (fromPtr == rawNameEnd) { | |
2089 | tag->name.strLen = convLen; | |
2090 | break; | |
2091 | } | |
2092 | bufSize = (tag->bufEnd - tag->buf) << 1; | |
2093 | { | |
2094 | char *temp = (char *)REALLOC(tag->buf, bufSize); | |
2095 | if (temp == NULL) | |
2096 | return XML_ERROR_NO_MEMORY; | |
2097 | tag->buf = temp; | |
2098 | tag->bufEnd = temp + bufSize; | |
2099 | toPtr = (XML_Char *)temp + convLen; | |
2100 | } | |
2101 | } | |
2102 | } | |
2103 | tag->name.str = (XML_Char *)tag->buf; | |
2104 | *toPtr = XML_T('\0'); | |
2105 | result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings)); | |
2106 | if (result) | |
2107 | return result; | |
2108 | if (startElementHandler) | |
2109 | startElementHandler(handlerArg, tag->name.str, | |
2110 | (const XML_Char **)atts); | |
2111 | else if (defaultHandler) | |
2112 | reportDefault(parser, enc, s, next); | |
2113 | poolClear(&tempPool); | |
2114 | break; | |
2115 | } | |
2116 | case XML_TOK_EMPTY_ELEMENT_NO_ATTS: | |
2117 | /* fall through */ | |
2118 | case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: | |
2119 | { | |
2120 | const char *rawName = s + enc->minBytesPerChar; | |
2121 | enum XML_Error result; | |
2122 | BINDING *bindings = NULL; | |
2123 | XML_Bool noElmHandlers = XML_TRUE; | |
2124 | TAG_NAME name; | |
2125 | name.str = poolStoreString(&tempPool, enc, rawName, | |
2126 | rawName + XmlNameLength(enc, rawName)); | |
2127 | if (!name.str) | |
2128 | return XML_ERROR_NO_MEMORY; | |
2129 | poolFinish(&tempPool); | |
2130 | result = storeAtts(parser, enc, s, &name, &bindings); | |
2131 | if (result) | |
2132 | return result; | |
2133 | poolFinish(&tempPool); | |
2134 | if (startElementHandler) { | |
2135 | startElementHandler(handlerArg, name.str, (const XML_Char **)atts); | |
2136 | noElmHandlers = XML_FALSE; | |
2137 | } | |
2138 | if (endElementHandler) { | |
2139 | if (startElementHandler) | |
2140 | *eventPP = *eventEndPP; | |
2141 | endElementHandler(handlerArg, name.str); | |
2142 | noElmHandlers = XML_FALSE; | |
2143 | } | |
2144 | if (noElmHandlers && defaultHandler) | |
2145 | reportDefault(parser, enc, s, next); | |
2146 | poolClear(&tempPool); | |
2147 | while (bindings) { | |
2148 | BINDING *b = bindings; | |
2149 | if (endNamespaceDeclHandler) | |
2150 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
2151 | bindings = bindings->nextTagBinding; | |
2152 | b->nextTagBinding = freeBindingList; | |
2153 | freeBindingList = b; | |
2154 | b->prefix->binding = b->prevPrefixBinding; | |
2155 | } | |
2156 | } | |
2157 | if (tagLevel == 0) | |
2158 | return epilogProcessor(parser, next, end, nextPtr); | |
2159 | break; | |
2160 | case XML_TOK_END_TAG: | |
2161 | if (tagLevel == startTagLevel) | |
2162 | return XML_ERROR_ASYNC_ENTITY; | |
2163 | else { | |
2164 | int len; | |
2165 | const char *rawName; | |
2166 | TAG *tag = tagStack; | |
2167 | tagStack = tag->parent; | |
2168 | tag->parent = freeTagList; | |
2169 | freeTagList = tag; | |
2170 | rawName = s + enc->minBytesPerChar*2; | |
2171 | len = XmlNameLength(enc, rawName); | |
2172 | if (len != tag->rawNameLength | |
2173 | || memcmp(tag->rawName, rawName, len) != 0) { | |
2174 | *eventPP = rawName; | |
2175 | return XML_ERROR_TAG_MISMATCH; | |
2176 | } | |
2177 | --tagLevel; | |
2178 | if (endElementHandler) { | |
2179 | const XML_Char *localPart; | |
2180 | const XML_Char *prefix; | |
2181 | XML_Char *uri; | |
2182 | localPart = tag->name.localPart; | |
2183 | if (ns && localPart) { | |
2184 | /* localPart and prefix may have been overwritten in | |
2185 | tag->name.str, since this points to the binding->uri | |
2186 | buffer which gets re-used; so we have to add them again | |
2187 | */ | |
2188 | uri = (XML_Char *)tag->name.str + tag->name.uriLen; | |
2189 | /* don't need to check for space - already done in storeAtts() */ | |
2190 | while (*localPart) *uri++ = *localPart++; | |
2191 | prefix = (XML_Char *)tag->name.prefix; | |
2192 | if (ns_triplets && prefix) { | |
2193 | *uri++ = namespaceSeparator; | |
2194 | while (*prefix) *uri++ = *prefix++; | |
2195 | } | |
2196 | *uri = XML_T('\0'); | |
2197 | } | |
2198 | endElementHandler(handlerArg, tag->name.str); | |
2199 | } | |
2200 | else if (defaultHandler) | |
2201 | reportDefault(parser, enc, s, next); | |
2202 | while (tag->bindings) { | |
2203 | BINDING *b = tag->bindings; | |
2204 | if (endNamespaceDeclHandler) | |
2205 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
2206 | tag->bindings = tag->bindings->nextTagBinding; | |
2207 | b->nextTagBinding = freeBindingList; | |
2208 | freeBindingList = b; | |
2209 | b->prefix->binding = b->prevPrefixBinding; | |
2210 | } | |
2211 | if (tagLevel == 0) | |
2212 | return epilogProcessor(parser, next, end, nextPtr); | |
2213 | } | |
2214 | break; | |
2215 | case XML_TOK_CHAR_REF: | |
2216 | { | |
2217 | int n = XmlCharRefNumber(enc, s); | |
2218 | if (n < 0) | |
2219 | return XML_ERROR_BAD_CHAR_REF; | |
2220 | if (characterDataHandler) { | |
2221 | XML_Char buf[XML_ENCODE_MAX]; | |
2222 | characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf)); | |
2223 | } | |
2224 | else if (defaultHandler) | |
2225 | reportDefault(parser, enc, s, next); | |
2226 | } | |
2227 | break; | |
2228 | case XML_TOK_XML_DECL: | |
2229 | return XML_ERROR_MISPLACED_XML_PI; | |
2230 | case XML_TOK_DATA_NEWLINE: | |
2231 | if (characterDataHandler) { | |
2232 | XML_Char c = 0xA; | |
2233 | characterDataHandler(handlerArg, &c, 1); | |
2234 | } | |
2235 | else if (defaultHandler) | |
2236 | reportDefault(parser, enc, s, next); | |
2237 | break; | |
2238 | case XML_TOK_CDATA_SECT_OPEN: | |
2239 | { | |
2240 | enum XML_Error result; | |
2241 | if (startCdataSectionHandler) | |
2242 | startCdataSectionHandler(handlerArg); | |
2243 | #if 0 | |
2244 | /* Suppose you doing a transformation on a document that involves | |
2245 | changing only the character data. You set up a defaultHandler | |
2246 | and a characterDataHandler. The defaultHandler simply copies | |
2247 | characters through. The characterDataHandler does the | |
2248 | transformation and writes the characters out escaping them as | |
2249 | necessary. This case will fail to work if we leave out the | |
2250 | following two lines (because & and < inside CDATA sections will | |
2251 | be incorrectly escaped). | |
2252 | ||
2253 | However, now we have a start/endCdataSectionHandler, so it seems | |
2254 | easier to let the user deal with this. | |
2255 | */ | |
2256 | else if (characterDataHandler) | |
2257 | characterDataHandler(handlerArg, dataBuf, 0); | |
2258 | #endif | |
2259 | else if (defaultHandler) | |
2260 | reportDefault(parser, enc, s, next); | |
2261 | result = doCdataSection(parser, enc, &next, end, nextPtr); | |
2262 | if (!next) { | |
2263 | processor = cdataSectionProcessor; | |
2264 | return result; | |
2265 | } | |
2266 | } | |
2267 | break; | |
2268 | case XML_TOK_TRAILING_RSQB: | |
2269 | if (nextPtr) { | |
2270 | *nextPtr = s; | |
2271 | return XML_ERROR_NONE; | |
2272 | } | |
2273 | if (characterDataHandler) { | |
2274 | if (MUST_CONVERT(enc, s)) { | |
2275 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2276 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
2277 | characterDataHandler(handlerArg, dataBuf, | |
2278 | dataPtr - (ICHAR *)dataBuf); | |
2279 | } | |
2280 | else | |
2281 | characterDataHandler(handlerArg, | |
2282 | (XML_Char *)s, | |
2283 | (XML_Char *)end - (XML_Char *)s); | |
2284 | } | |
2285 | else if (defaultHandler) | |
2286 | reportDefault(parser, enc, s, end); | |
2287 | if (startTagLevel == 0) { | |
2288 | *eventPP = end; | |
2289 | return XML_ERROR_NO_ELEMENTS; | |
2290 | } | |
2291 | if (tagLevel != startTagLevel) { | |
2292 | *eventPP = end; | |
2293 | return XML_ERROR_ASYNC_ENTITY; | |
2294 | } | |
2295 | return XML_ERROR_NONE; | |
2296 | case XML_TOK_DATA_CHARS: | |
2297 | if (characterDataHandler) { | |
2298 | if (MUST_CONVERT(enc, s)) { | |
2299 | for (;;) { | |
2300 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2301 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
2302 | *eventEndPP = s; | |
2303 | characterDataHandler(handlerArg, dataBuf, | |
2304 | dataPtr - (ICHAR *)dataBuf); | |
2305 | if (s == next) | |
2306 | break; | |
2307 | *eventPP = s; | |
2308 | } | |
2309 | } | |
2310 | else | |
2311 | characterDataHandler(handlerArg, | |
2312 | (XML_Char *)s, | |
2313 | (XML_Char *)next - (XML_Char *)s); | |
2314 | } | |
2315 | else if (defaultHandler) | |
2316 | reportDefault(parser, enc, s, next); | |
2317 | break; | |
2318 | case XML_TOK_PI: | |
2319 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
2320 | return XML_ERROR_NO_MEMORY; | |
2321 | break; | |
2322 | case XML_TOK_COMMENT: | |
2323 | if (!reportComment(parser, enc, s, next)) | |
2324 | return XML_ERROR_NO_MEMORY; | |
2325 | break; | |
2326 | default: | |
2327 | if (defaultHandler) | |
2328 | reportDefault(parser, enc, s, next); | |
2329 | break; | |
2330 | } | |
2331 | *eventPP = s = next; | |
2332 | } | |
2333 | /* not reached */ | |
2334 | } | |
2335 | ||
2336 | /* Precondition: all arguments must be non-NULL; | |
2337 | Purpose: | |
2338 | - normalize attributes | |
2339 | - check attributes for well-formedness | |
2340 | - generate namespace aware attribute names (URI, prefix) | |
2341 | - build list of attributes for startElementHandler | |
2342 | - default attributes | |
2343 | - process namespace declarations (check and report them) | |
2344 | - generate namespace aware element name (URI, prefix) | |
2345 | */ | |
2346 | static enum XML_Error | |
2347 | storeAtts(XML_Parser parser, const ENCODING *enc, | |
2348 | const char *attStr, TAG_NAME *tagNamePtr, | |
2349 | BINDING **bindingsPtr) | |
2350 | { | |
2351 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
2352 | ELEMENT_TYPE *elementType = NULL; | |
2353 | int nDefaultAtts = 0; | |
2354 | const XML_Char **appAtts; /* the attribute list for the application */ | |
2355 | int attIndex = 0; | |
2356 | int prefixLen; | |
2357 | int i; | |
2358 | int n; | |
2359 | XML_Char *uri; | |
2360 | int nPrefixes = 0; | |
2361 | BINDING *binding; | |
2362 | const XML_Char *localPart; | |
2363 | ||
2364 | /* lookup the element type name */ | |
2365 | elementType = (ELEMENT_TYPE *)lookup(&dtd->elementTypes, tagNamePtr->str,0); | |
2366 | if (!elementType) { | |
2367 | const XML_Char *name = poolCopyString(&dtd->pool, tagNamePtr->str); | |
2368 | if (!name) | |
2369 | return XML_ERROR_NO_MEMORY; | |
2370 | elementType = (ELEMENT_TYPE *)lookup(&dtd->elementTypes, name, | |
2371 | sizeof(ELEMENT_TYPE)); | |
2372 | if (!elementType) | |
2373 | return XML_ERROR_NO_MEMORY; | |
2374 | if (ns && !setElementTypePrefix(parser, elementType)) | |
2375 | return XML_ERROR_NO_MEMORY; | |
2376 | } | |
2377 | nDefaultAtts = elementType->nDefaultAtts; | |
2378 | ||
2379 | /* get the attributes from the tokenizer */ | |
2380 | n = XmlGetAttributes(enc, attStr, attsSize, atts); | |
2381 | if (n + nDefaultAtts > attsSize) { | |
2382 | int oldAttsSize = attsSize; | |
2383 | ATTRIBUTE *temp; | |
2384 | attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; | |
2385 | temp = (ATTRIBUTE *)REALLOC((void *)atts, attsSize * sizeof(ATTRIBUTE)); | |
2386 | if (temp == NULL) | |
2387 | return XML_ERROR_NO_MEMORY; | |
2388 | atts = temp; | |
2389 | if (n > oldAttsSize) | |
2390 | XmlGetAttributes(enc, attStr, n, atts); | |
2391 | } | |
2392 | ||
2393 | appAtts = (const XML_Char **)atts; | |
2394 | for (i = 0; i < n; i++) { | |
2395 | /* add the name and value to the attribute list */ | |
2396 | ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name, | |
2397 | atts[i].name | |
2398 | + XmlNameLength(enc, atts[i].name)); | |
2399 | if (!attId) | |
2400 | return XML_ERROR_NO_MEMORY; | |
2401 | /* detect duplicate attributes */ | |
2402 | if ((attId->name)[-1]) { | |
2403 | if (enc == encoding) | |
2404 | eventPtr = atts[i].name; | |
2405 | return XML_ERROR_DUPLICATE_ATTRIBUTE; | |
2406 | } | |
2407 | (attId->name)[-1] = 1; | |
2408 | appAtts[attIndex++] = attId->name; | |
2409 | if (!atts[i].normalized) { | |
2410 | enum XML_Error result; | |
2411 | XML_Bool isCdata = XML_TRUE; | |
2412 | ||
2413 | /* figure out whether declared as other than CDATA */ | |
2414 | if (attId->maybeTokenized) { | |
2415 | int j; | |
2416 | for (j = 0; j < nDefaultAtts; j++) { | |
2417 | if (attId == elementType->defaultAtts[j].id) { | |
2418 | isCdata = elementType->defaultAtts[j].isCdata; | |
2419 | break; | |
2420 | } | |
2421 | } | |
2422 | } | |
2423 | ||
2424 | /* normalize the attribute value */ | |
2425 | result = storeAttributeValue(parser, enc, isCdata, | |
2426 | atts[i].valuePtr, atts[i].valueEnd, | |
2427 | &tempPool); | |
2428 | if (result) | |
2429 | return result; | |
2430 | appAtts[attIndex] = poolStart(&tempPool); | |
2431 | poolFinish(&tempPool); | |
2432 | } | |
2433 | else { | |
2434 | /* the value did not need normalizing */ | |
2435 | appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr, | |
2436 | atts[i].valueEnd); | |
2437 | if (appAtts[attIndex] == 0) | |
2438 | return XML_ERROR_NO_MEMORY; | |
2439 | poolFinish(&tempPool); | |
2440 | } | |
2441 | /* handle prefixed attribute names */ | |
2442 | if (attId->prefix) { | |
2443 | if (attId->xmlns) { | |
2444 | /* deal with namespace declarations here */ | |
2445 | enum XML_Error result = addBinding(parser, attId->prefix, attId, | |
2446 | appAtts[attIndex], bindingsPtr); | |
2447 | if (result) | |
2448 | return result; | |
2449 | --attIndex; | |
2450 | } | |
2451 | else { | |
2452 | /* deal with other prefixed names later */ | |
2453 | attIndex++; | |
2454 | nPrefixes++; | |
2455 | (attId->name)[-1] = 2; | |
2456 | } | |
2457 | } | |
2458 | else | |
2459 | attIndex++; | |
2460 | } | |
2461 | ||
2462 | /* set-up for XML_GetSpecifiedAttributeCount and XML_GetIdAttributeIndex */ | |
2463 | nSpecifiedAtts = attIndex; | |
2464 | if (elementType->idAtt && (elementType->idAtt->name)[-1]) { | |
2465 | for (i = 0; i < attIndex; i += 2) | |
2466 | if (appAtts[i] == elementType->idAtt->name) { | |
2467 | idAttIndex = i; | |
2468 | break; | |
2469 | } | |
2470 | } | |
2471 | else | |
2472 | idAttIndex = -1; | |
2473 | ||
2474 | /* do attribute defaulting */ | |
2475 | for (i = 0; i < nDefaultAtts; i++) { | |
2476 | const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + i; | |
2477 | if (!(da->id->name)[-1] && da->value) { | |
2478 | if (da->id->prefix) { | |
2479 | if (da->id->xmlns) { | |
2480 | enum XML_Error result = addBinding(parser, da->id->prefix, da->id, | |
2481 | da->value, bindingsPtr); | |
2482 | if (result) | |
2483 | return result; | |
2484 | } | |
2485 | else { | |
2486 | (da->id->name)[-1] = 2; | |
2487 | nPrefixes++; | |
2488 | appAtts[attIndex++] = da->id->name; | |
2489 | appAtts[attIndex++] = da->value; | |
2490 | } | |
2491 | } | |
2492 | else { | |
2493 | (da->id->name)[-1] = 1; | |
2494 | appAtts[attIndex++] = da->id->name; | |
2495 | appAtts[attIndex++] = da->value; | |
2496 | } | |
2497 | } | |
2498 | } | |
2499 | appAtts[attIndex] = 0; | |
2500 | ||
2501 | i = 0; | |
2502 | if (nPrefixes) { | |
2503 | /* expand prefixed attribute names */ | |
2504 | for (; i < attIndex; i += 2) { | |
2505 | if (appAtts[i][-1] == 2) { | |
2506 | ATTRIBUTE_ID *id; | |
2507 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
2508 | id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, appAtts[i], 0); | |
2509 | if (id->prefix->binding) { | |
2510 | int j; | |
2511 | const BINDING *b = id->prefix->binding; | |
2512 | const XML_Char *s = appAtts[i]; | |
2513 | for (j = 0; j < b->uriLen; j++) { | |
2514 | if (!poolAppendChar(&tempPool, b->uri[j])) | |
2515 | return XML_ERROR_NO_MEMORY; | |
2516 | } | |
2517 | while (*s++ != XML_T(':')) | |
2518 | ; | |
2519 | do { | |
2520 | if (!poolAppendChar(&tempPool, *s)) | |
2521 | return XML_ERROR_NO_MEMORY; | |
2522 | } while (*s++); | |
2523 | if (ns_triplets) { | |
2524 | tempPool.ptr[-1] = namespaceSeparator; | |
2525 | s = b->prefix->name; | |
2526 | do { | |
2527 | if (!poolAppendChar(&tempPool, *s)) | |
2528 | return XML_ERROR_NO_MEMORY; | |
2529 | } while (*s++); | |
2530 | } | |
2531 | ||
2532 | appAtts[i] = poolStart(&tempPool); | |
2533 | poolFinish(&tempPool); | |
2534 | } | |
2535 | if (!--nPrefixes) | |
2536 | break; | |
2537 | } | |
2538 | else | |
2539 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
2540 | } | |
2541 | } | |
2542 | /* clear the flags that say whether attributes were specified */ | |
2543 | for (; i < attIndex; i += 2) | |
2544 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
2545 | for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) | |
2546 | binding->attId->name[-1] = 0; | |
2547 | ||
2548 | /* expand the element type name */ | |
2549 | if (elementType->prefix) { | |
2550 | binding = elementType->prefix->binding; | |
2551 | if (!binding) | |
2552 | return XML_ERROR_NONE; | |
2553 | localPart = tagNamePtr->str; | |
2554 | while (*localPart++ != XML_T(':')) | |
2555 | ; | |
2556 | } | |
2557 | else if (dtd->defaultPrefix.binding) { | |
2558 | binding = dtd->defaultPrefix.binding; | |
2559 | localPart = tagNamePtr->str; | |
2560 | } | |
2561 | else | |
2562 | return XML_ERROR_NONE; | |
2563 | prefixLen = 0; | |
2564 | if (ns && ns_triplets && binding->prefix->name) { | |
2565 | for (; binding->prefix->name[prefixLen++];) | |
2566 | ; | |
2567 | } | |
2568 | tagNamePtr->localPart = localPart; | |
2569 | tagNamePtr->uriLen = binding->uriLen; | |
2570 | tagNamePtr->prefix = binding->prefix->name; | |
2571 | tagNamePtr->prefixLen = prefixLen; | |
2572 | for (i = 0; localPart[i++];) | |
2573 | ; | |
2574 | n = i + binding->uriLen + prefixLen; | |
2575 | if (n > binding->uriAlloc) { | |
2576 | TAG *p; | |
2577 | uri = (XML_Char *)MALLOC((n + EXPAND_SPARE) * sizeof(XML_Char)); | |
2578 | if (!uri) | |
2579 | return XML_ERROR_NO_MEMORY; | |
2580 | binding->uriAlloc = n + EXPAND_SPARE; | |
2581 | memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char)); | |
2582 | for (p = tagStack; p; p = p->parent) | |
2583 | if (p->name.str == binding->uri) | |
2584 | p->name.str = uri; | |
2585 | FREE(binding->uri); | |
2586 | binding->uri = uri; | |
2587 | } | |
2588 | uri = binding->uri + binding->uriLen; | |
2589 | memcpy(uri, localPart, i * sizeof(XML_Char)); | |
2590 | if (prefixLen) { | |
2591 | uri = uri + (i - 1); | |
2592 | if (namespaceSeparator) { *(uri) = namespaceSeparator; } | |
2593 | memcpy(uri + 1, binding->prefix->name, prefixLen * sizeof(XML_Char)); | |
2594 | } | |
2595 | tagNamePtr->str = binding->uri; | |
2596 | return XML_ERROR_NONE; | |
2597 | } | |
2598 | ||
2599 | /* addBinding() overwrites the value of prefix->binding without checking. | |
2600 | Therefore one must keep track of the old value outside of addBinding(). | |
2601 | */ | |
2602 | static enum XML_Error | |
2603 | addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, | |
2604 | const XML_Char *uri, BINDING **bindingsPtr) | |
2605 | { | |
2606 | BINDING *b; | |
2607 | int len; | |
2608 | ||
2609 | /* empty string is only valid when there is no prefix per XML NS 1.0 */ | |
2610 | if (*uri == XML_T('\0') && prefix->name) | |
2611 | return XML_ERROR_SYNTAX; | |
2612 | ||
2613 | for (len = 0; uri[len]; len++) | |
2614 | ; | |
2615 | if (namespaceSeparator) | |
2616 | len++; | |
2617 | if (freeBindingList) { | |
2618 | b = freeBindingList; | |
2619 | if (len > b->uriAlloc) { | |
2620 | XML_Char *temp = (XML_Char *)REALLOC(b->uri, | |
2621 | sizeof(XML_Char) * (len + EXPAND_SPARE)); | |
2622 | if (temp == NULL) | |
2623 | return XML_ERROR_NO_MEMORY; | |
2624 | b->uri = temp; | |
2625 | b->uriAlloc = len + EXPAND_SPARE; | |
2626 | } | |
2627 | freeBindingList = b->nextTagBinding; | |
2628 | } | |
2629 | else { | |
2630 | b = (BINDING *)MALLOC(sizeof(BINDING)); | |
2631 | if (!b) | |
2632 | return XML_ERROR_NO_MEMORY; | |
2633 | b->uri = (XML_Char *)MALLOC(sizeof(XML_Char) * (len + EXPAND_SPARE)); | |
2634 | if (!b->uri) { | |
2635 | FREE(b); | |
2636 | return XML_ERROR_NO_MEMORY; | |
2637 | } | |
2638 | b->uriAlloc = len + EXPAND_SPARE; | |
2639 | } | |
2640 | b->uriLen = len; | |
2641 | memcpy(b->uri, uri, len * sizeof(XML_Char)); | |
2642 | if (namespaceSeparator) | |
2643 | b->uri[len - 1] = namespaceSeparator; | |
2644 | b->prefix = prefix; | |
2645 | b->attId = attId; | |
2646 | b->prevPrefixBinding = prefix->binding; | |
2647 | if (*uri == XML_T('\0') && prefix == &_dtd->defaultPrefix) | |
2648 | prefix->binding = NULL; | |
2649 | else | |
2650 | prefix->binding = b; | |
2651 | b->nextTagBinding = *bindingsPtr; | |
2652 | *bindingsPtr = b; | |
2653 | if (startNamespaceDeclHandler) | |
2654 | startNamespaceDeclHandler(handlerArg, prefix->name, | |
2655 | prefix->binding ? uri : 0); | |
2656 | return XML_ERROR_NONE; | |
2657 | } | |
2658 | ||
2659 | /* The idea here is to avoid using stack for each CDATA section when | |
2660 | the whole file is parsed with one call. | |
2661 | */ | |
2662 | static enum XML_Error PTRCALL | |
2663 | cdataSectionProcessor(XML_Parser parser, | |
2664 | const char *start, | |
2665 | const char *end, | |
2666 | const char **endPtr) | |
2667 | { | |
2668 | enum XML_Error result = doCdataSection(parser, encoding, &start, | |
2669 | end, endPtr); | |
2670 | if (start) { | |
2671 | if (parentParser) { /* we are parsing an external entity */ | |
2672 | processor = externalEntityContentProcessor; | |
2673 | return externalEntityContentProcessor(parser, start, end, endPtr); | |
2674 | } | |
2675 | else { | |
2676 | processor = contentProcessor; | |
2677 | return contentProcessor(parser, start, end, endPtr); | |
2678 | } | |
2679 | } | |
2680 | return result; | |
2681 | } | |
2682 | ||
2683 | /* startPtr gets set to non-null is the section is closed, and to null if | |
2684 | the section is not yet closed. | |
2685 | */ | |
2686 | static enum XML_Error | |
2687 | doCdataSection(XML_Parser parser, | |
2688 | const ENCODING *enc, | |
2689 | const char **startPtr, | |
2690 | const char *end, | |
2691 | const char **nextPtr) | |
2692 | { | |
2693 | const char *s = *startPtr; | |
2694 | const char **eventPP; | |
2695 | const char **eventEndPP; | |
2696 | if (enc == encoding) { | |
2697 | eventPP = &eventPtr; | |
2698 | *eventPP = s; | |
2699 | eventEndPP = &eventEndPtr; | |
2700 | } | |
2701 | else { | |
2702 | eventPP = &(openInternalEntities->internalEventPtr); | |
2703 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2704 | } | |
2705 | *eventPP = s; | |
2706 | *startPtr = NULL; | |
2707 | for (;;) { | |
2708 | const char *next; | |
2709 | int tok = XmlCdataSectionTok(enc, s, end, &next); | |
2710 | *eventEndPP = next; | |
2711 | switch (tok) { | |
2712 | case XML_TOK_CDATA_SECT_CLOSE: | |
2713 | if (endCdataSectionHandler) | |
2714 | endCdataSectionHandler(handlerArg); | |
2715 | #if 0 | |
2716 | /* see comment under XML_TOK_CDATA_SECT_OPEN */ | |
2717 | else if (characterDataHandler) | |
2718 | characterDataHandler(handlerArg, dataBuf, 0); | |
2719 | #endif | |
2720 | else if (defaultHandler) | |
2721 | reportDefault(parser, enc, s, next); | |
2722 | *startPtr = next; | |
2723 | return XML_ERROR_NONE; | |
2724 | case XML_TOK_DATA_NEWLINE: | |
2725 | if (characterDataHandler) { | |
2726 | XML_Char c = 0xA; | |
2727 | characterDataHandler(handlerArg, &c, 1); | |
2728 | } | |
2729 | else if (defaultHandler) | |
2730 | reportDefault(parser, enc, s, next); | |
2731 | break; | |
2732 | case XML_TOK_DATA_CHARS: | |
2733 | if (characterDataHandler) { | |
2734 | if (MUST_CONVERT(enc, s)) { | |
2735 | for (;;) { | |
2736 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2737 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
2738 | *eventEndPP = next; | |
2739 | characterDataHandler(handlerArg, dataBuf, | |
2740 | dataPtr - (ICHAR *)dataBuf); | |
2741 | if (s == next) | |
2742 | break; | |
2743 | *eventPP = s; | |
2744 | } | |
2745 | } | |
2746 | else | |
2747 | characterDataHandler(handlerArg, | |
2748 | (XML_Char *)s, | |
2749 | (XML_Char *)next - (XML_Char *)s); | |
2750 | } | |
2751 | else if (defaultHandler) | |
2752 | reportDefault(parser, enc, s, next); | |
2753 | break; | |
2754 | case XML_TOK_INVALID: | |
2755 | *eventPP = next; | |
2756 | return XML_ERROR_INVALID_TOKEN; | |
2757 | case XML_TOK_PARTIAL_CHAR: | |
2758 | if (nextPtr) { | |
2759 | *nextPtr = s; | |
2760 | return XML_ERROR_NONE; | |
2761 | } | |
2762 | return XML_ERROR_PARTIAL_CHAR; | |
2763 | case XML_TOK_PARTIAL: | |
2764 | case XML_TOK_NONE: | |
2765 | if (nextPtr) { | |
2766 | *nextPtr = s; | |
2767 | return XML_ERROR_NONE; | |
2768 | } | |
2769 | return XML_ERROR_UNCLOSED_CDATA_SECTION; | |
2770 | default: | |
2771 | *eventPP = next; | |
2772 | return XML_ERROR_UNEXPECTED_STATE; | |
2773 | } | |
2774 | *eventPP = s = next; | |
2775 | } | |
2776 | /* not reached */ | |
2777 | } | |
2778 | ||
2779 | #ifdef XML_DTD | |
2780 | ||
2781 | /* The idea here is to avoid using stack for each IGNORE section when | |
2782 | the whole file is parsed with one call. | |
2783 | */ | |
2784 | static enum XML_Error PTRCALL | |
2785 | ignoreSectionProcessor(XML_Parser parser, | |
2786 | const char *start, | |
2787 | const char *end, | |
2788 | const char **endPtr) | |
2789 | { | |
2790 | enum XML_Error result = doIgnoreSection(parser, encoding, &start, | |
2791 | end, endPtr); | |
2792 | if (start) { | |
2793 | processor = prologProcessor; | |
2794 | return prologProcessor(parser, start, end, endPtr); | |
2795 | } | |
2796 | return result; | |
2797 | } | |
2798 | ||
2799 | /* startPtr gets set to non-null is the section is closed, and to null | |
2800 | if the section is not yet closed. | |
2801 | */ | |
2802 | static enum XML_Error | |
2803 | doIgnoreSection(XML_Parser parser, | |
2804 | const ENCODING *enc, | |
2805 | const char **startPtr, | |
2806 | const char *end, | |
2807 | const char **nextPtr) | |
2808 | { | |
2809 | const char *next; | |
2810 | int tok; | |
2811 | const char *s = *startPtr; | |
2812 | const char **eventPP; | |
2813 | const char **eventEndPP; | |
2814 | if (enc == encoding) { | |
2815 | eventPP = &eventPtr; | |
2816 | *eventPP = s; | |
2817 | eventEndPP = &eventEndPtr; | |
2818 | } | |
2819 | else { | |
2820 | eventPP = &(openInternalEntities->internalEventPtr); | |
2821 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2822 | } | |
2823 | *eventPP = s; | |
2824 | *startPtr = NULL; | |
2825 | tok = XmlIgnoreSectionTok(enc, s, end, &next); | |
2826 | *eventEndPP = next; | |
2827 | switch (tok) { | |
2828 | case XML_TOK_IGNORE_SECT: | |
2829 | if (defaultHandler) | |
2830 | reportDefault(parser, enc, s, next); | |
2831 | *startPtr = next; | |
2832 | return XML_ERROR_NONE; | |
2833 | case XML_TOK_INVALID: | |
2834 | *eventPP = next; | |
2835 | return XML_ERROR_INVALID_TOKEN; | |
2836 | case XML_TOK_PARTIAL_CHAR: | |
2837 | if (nextPtr) { | |
2838 | *nextPtr = s; | |
2839 | return XML_ERROR_NONE; | |
2840 | } | |
2841 | return XML_ERROR_PARTIAL_CHAR; | |
2842 | case XML_TOK_PARTIAL: | |
2843 | case XML_TOK_NONE: | |
2844 | if (nextPtr) { | |
2845 | *nextPtr = s; | |
2846 | return XML_ERROR_NONE; | |
2847 | } | |
2848 | return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */ | |
2849 | default: | |
2850 | *eventPP = next; | |
2851 | return XML_ERROR_UNEXPECTED_STATE; | |
2852 | } | |
2853 | /* not reached */ | |
2854 | } | |
2855 | ||
2856 | #endif /* XML_DTD */ | |
2857 | ||
2858 | static enum XML_Error | |
2859 | initializeEncoding(XML_Parser parser) | |
2860 | { | |
2861 | const char *s; | |
2862 | #ifdef XML_UNICODE | |
2863 | char encodingBuf[128]; | |
2864 | if (!protocolEncodingName) | |
2865 | s = NULL; | |
2866 | else { | |
2867 | int i; | |
2868 | for (i = 0; protocolEncodingName[i]; i++) { | |
2869 | if (i == sizeof(encodingBuf) - 1 | |
2870 | || (protocolEncodingName[i] & ~0x7f) != 0) { | |
2871 | encodingBuf[0] = '\0'; | |
2872 | break; | |
2873 | } | |
2874 | encodingBuf[i] = (char)protocolEncodingName[i]; | |
2875 | } | |
2876 | encodingBuf[i] = '\0'; | |
2877 | s = encodingBuf; | |
2878 | } | |
2879 | #else | |
2880 | s = protocolEncodingName; | |
2881 | #endif | |
2882 | if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s)) | |
2883 | return XML_ERROR_NONE; | |
2884 | return handleUnknownEncoding(parser, protocolEncodingName); | |
2885 | } | |
2886 | ||
2887 | static enum XML_Error | |
2888 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
2889 | const char *s, const char *next) | |
2890 | { | |
2891 | const char *encodingName = NULL; | |
2892 | const XML_Char *storedEncName = NULL; | |
2893 | const ENCODING *newEncoding = NULL; | |
2894 | const char *version = NULL; | |
2895 | const char *versionend; | |
2896 | const XML_Char *storedversion = NULL; | |
2897 | int standalone = -1; | |
2898 | if (!(ns | |
2899 | ? XmlParseXmlDeclNS | |
2900 | : XmlParseXmlDecl)(isGeneralTextEntity, | |
2901 | encoding, | |
2902 | s, | |
2903 | next, | |
2904 | &eventPtr, | |
2905 | &version, | |
2906 | &versionend, | |
2907 | &encodingName, | |
2908 | &newEncoding, | |
2909 | &standalone)) | |
2910 | return XML_ERROR_SYNTAX; | |
2911 | if (!isGeneralTextEntity && standalone == 1) { | |
2912 | _dtd->standalone = XML_TRUE; | |
2913 | #ifdef XML_DTD | |
2914 | if (paramEntityParsing == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE) | |
2915 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; | |
2916 | #endif /* XML_DTD */ | |
2917 | } | |
2918 | if (xmlDeclHandler) { | |
2919 | if (encodingName != NULL) { | |
2920 | storedEncName = poolStoreString(&temp2Pool, | |
2921 | encoding, | |
2922 | encodingName, | |
2923 | encodingName | |
2924 | + XmlNameLength(encoding, encodingName)); | |
2925 | if (!storedEncName) | |
2926 | return XML_ERROR_NO_MEMORY; | |
2927 | poolFinish(&temp2Pool); | |
2928 | } | |
2929 | if (version) { | |
2930 | storedversion = poolStoreString(&temp2Pool, | |
2931 | encoding, | |
2932 | version, | |
2933 | versionend - encoding->minBytesPerChar); | |
2934 | if (!storedversion) | |
2935 | return XML_ERROR_NO_MEMORY; | |
2936 | } | |
2937 | xmlDeclHandler(handlerArg, storedversion, storedEncName, standalone); | |
2938 | } | |
2939 | else if (defaultHandler) | |
2940 | reportDefault(parser, encoding, s, next); | |
2941 | if (protocolEncodingName == NULL) { | |
2942 | if (newEncoding) { | |
2943 | if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) { | |
2944 | eventPtr = encodingName; | |
2945 | return XML_ERROR_INCORRECT_ENCODING; | |
2946 | } | |
2947 | encoding = newEncoding; | |
2948 | } | |
2949 | else if (encodingName) { | |
2950 | enum XML_Error result; | |
2951 | if (!storedEncName) { | |
2952 | storedEncName = poolStoreString( | |
2953 | &temp2Pool, encoding, encodingName, | |
2954 | encodingName + XmlNameLength(encoding, encodingName)); | |
2955 | if (!storedEncName) | |
2956 | return XML_ERROR_NO_MEMORY; | |
2957 | } | |
2958 | result = handleUnknownEncoding(parser, storedEncName); | |
2959 | poolClear(&temp2Pool); | |
2960 | if (result == XML_ERROR_UNKNOWN_ENCODING) | |
2961 | eventPtr = encodingName; | |
2962 | return result; | |
2963 | } | |
2964 | } | |
2965 | ||
2966 | if (storedEncName || storedversion) | |
2967 | poolClear(&temp2Pool); | |
2968 | ||
2969 | return XML_ERROR_NONE; | |
2970 | } | |
2971 | ||
2972 | static enum XML_Error | |
2973 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) | |
2974 | { | |
2975 | if (unknownEncodingHandler) { | |
2976 | XML_Encoding info; | |
2977 | int i; | |
2978 | for (i = 0; i < 256; i++) | |
2979 | info.map[i] = -1; | |
2980 | info.convert = NULL; | |
2981 | info.data = NULL; | |
2982 | info.release = NULL; | |
2983 | if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, | |
2984 | &info)) { | |
2985 | ENCODING *enc; | |
2986 | unknownEncodingMem = MALLOC(XmlSizeOfUnknownEncoding()); | |
2987 | if (!unknownEncodingMem) { | |
2988 | if (info.release) | |
2989 | info.release(info.data); | |
2990 | return XML_ERROR_NO_MEMORY; | |
2991 | } | |
2992 | enc = (ns | |
2993 | ? XmlInitUnknownEncodingNS | |
2994 | : XmlInitUnknownEncoding)(unknownEncodingMem, | |
2995 | info.map, | |
2996 | info.convert, | |
2997 | info.data); | |
2998 | if (enc) { | |
2999 | unknownEncodingData = info.data; | |
3000 | unknownEncodingRelease = info.release; | |
3001 | encoding = enc; | |
3002 | return XML_ERROR_NONE; | |
3003 | } | |
3004 | } | |
3005 | if (info.release != NULL) | |
3006 | info.release(info.data); | |
3007 | } | |
3008 | return XML_ERROR_UNKNOWN_ENCODING; | |
3009 | } | |
3010 | ||
3011 | static enum XML_Error PTRCALL | |
3012 | prologInitProcessor(XML_Parser parser, | |
3013 | const char *s, | |
3014 | const char *end, | |
3015 | const char **nextPtr) | |
3016 | { | |
3017 | enum XML_Error result = initializeEncoding(parser); | |
3018 | if (result != XML_ERROR_NONE) | |
3019 | return result; | |
3020 | processor = prologProcessor; | |
3021 | return prologProcessor(parser, s, end, nextPtr); | |
3022 | } | |
3023 | ||
3024 | #ifdef XML_DTD | |
3025 | ||
3026 | static enum XML_Error PTRCALL | |
3027 | externalParEntInitProcessor(XML_Parser parser, | |
3028 | const char *s, | |
3029 | const char *end, | |
3030 | const char **nextPtr) | |
3031 | { | |
3032 | enum XML_Error result = initializeEncoding(parser); | |
3033 | if (result != XML_ERROR_NONE) | |
3034 | return result; | |
3035 | ||
3036 | /* we know now that XML_Parse(Buffer) has been called, | |
3037 | so we consider the external parameter entity read */ | |
3038 | _dtd->paramEntityRead = XML_TRUE; | |
3039 | ||
3040 | if (prologState.inEntityValue) { | |
3041 | processor = entityValueInitProcessor; | |
3042 | return entityValueInitProcessor(parser, s, end, nextPtr); | |
3043 | } | |
3044 | else { | |
3045 | processor = externalParEntProcessor; | |
3046 | return externalParEntProcessor(parser, s, end, nextPtr); | |
3047 | } | |
3048 | } | |
3049 | ||
3050 | static enum XML_Error PTRCALL | |
3051 | entityValueInitProcessor(XML_Parser parser, | |
3052 | const char *s, | |
3053 | const char *end, | |
3054 | const char **nextPtr) | |
3055 | { | |
3056 | const char *start = s; | |
3057 | const char *next = s; | |
3058 | int tok; | |
3059 | ||
3060 | for (;;) { | |
3061 | tok = XmlPrologTok(encoding, start, end, &next); | |
3062 | if (tok <= 0) { | |
3063 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3064 | *nextPtr = s; | |
3065 | return XML_ERROR_NONE; | |
3066 | } | |
3067 | switch (tok) { | |
3068 | case XML_TOK_INVALID: | |
3069 | return XML_ERROR_INVALID_TOKEN; | |
3070 | case XML_TOK_PARTIAL: | |
3071 | return XML_ERROR_UNCLOSED_TOKEN; | |
3072 | case XML_TOK_PARTIAL_CHAR: | |
3073 | return XML_ERROR_PARTIAL_CHAR; | |
3074 | case XML_TOK_NONE: /* start == end */ | |
3075 | default: | |
3076 | break; | |
3077 | } | |
3078 | return storeEntityValue(parser, encoding, s, end); | |
3079 | } | |
3080 | else if (tok == XML_TOK_XML_DECL) { | |
3081 | enum XML_Error result = processXmlDecl(parser, 0, start, next); | |
3082 | if (result != XML_ERROR_NONE) | |
3083 | return result; | |
3084 | if (nextPtr) *nextPtr = next; | |
3085 | /* stop scanning for text declaration - we found one */ | |
3086 | processor = entityValueProcessor; | |
3087 | return entityValueProcessor(parser, next, end, nextPtr); | |
3088 | } | |
3089 | /* If we are at the end of the buffer, this would cause XmlPrologTok to | |
3090 | return XML_TOK_NONE on the next call, which would then cause the | |
3091 | function to exit with *nextPtr set to s - that is what we want for other | |
3092 | tokens, but not for the BOM - we would rather like to skip it; | |
3093 | then, when this routine is entered the next time, XmlPrologTok will | |
3094 | return XML_TOK_INVALID, since the BOM is still in the buffer | |
3095 | */ | |
3096 | else if (tok == XML_TOK_BOM && next == end && nextPtr) { | |
3097 | *nextPtr = next; | |
3098 | return XML_ERROR_NONE; | |
3099 | } | |
3100 | start = next; | |
3101 | } | |
3102 | } | |
3103 | ||
3104 | static enum XML_Error PTRCALL | |
3105 | externalParEntProcessor(XML_Parser parser, | |
3106 | const char *s, | |
3107 | const char *end, | |
3108 | const char **nextPtr) | |
3109 | { | |
3110 | const char *start = s; | |
3111 | const char *next = s; | |
3112 | int tok; | |
3113 | ||
3114 | tok = XmlPrologTok(encoding, start, end, &next); | |
3115 | if (tok <= 0) { | |
3116 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3117 | *nextPtr = s; | |
3118 | return XML_ERROR_NONE; | |
3119 | } | |
3120 | switch (tok) { | |
3121 | case XML_TOK_INVALID: | |
3122 | return XML_ERROR_INVALID_TOKEN; | |
3123 | case XML_TOK_PARTIAL: | |
3124 | return XML_ERROR_UNCLOSED_TOKEN; | |
3125 | case XML_TOK_PARTIAL_CHAR: | |
3126 | return XML_ERROR_PARTIAL_CHAR; | |
3127 | case XML_TOK_NONE: /* start == end */ | |
3128 | default: | |
3129 | break; | |
3130 | } | |
3131 | } | |
3132 | /* This would cause the next stage, i.e. doProlog to be passed XML_TOK_BOM. | |
3133 | However, when parsing an external subset, doProlog will not accept a BOM | |
3134 | as valid, and report a syntax error, so we have to skip the BOM | |
3135 | */ | |
3136 | else if (tok == XML_TOK_BOM) { | |
3137 | s = next; | |
3138 | tok = XmlPrologTok(encoding, s, end, &next); | |
3139 | } | |
3140 | ||
3141 | processor = prologProcessor; | |
3142 | return doProlog(parser, encoding, s, end, tok, next, nextPtr); | |
3143 | } | |
3144 | ||
3145 | static enum XML_Error PTRCALL | |
3146 | entityValueProcessor(XML_Parser parser, | |
3147 | const char *s, | |
3148 | const char *end, | |
3149 | const char **nextPtr) | |
3150 | { | |
3151 | const char *start = s; | |
3152 | const char *next = s; | |
3153 | const ENCODING *enc = encoding; | |
3154 | int tok; | |
3155 | ||
3156 | for (;;) { | |
3157 | tok = XmlPrologTok(enc, start, end, &next); | |
3158 | if (tok <= 0) { | |
3159 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3160 | *nextPtr = s; | |
3161 | return XML_ERROR_NONE; | |
3162 | } | |
3163 | switch (tok) { | |
3164 | case XML_TOK_INVALID: | |
3165 | return XML_ERROR_INVALID_TOKEN; | |
3166 | case XML_TOK_PARTIAL: | |
3167 | return XML_ERROR_UNCLOSED_TOKEN; | |
3168 | case XML_TOK_PARTIAL_CHAR: | |
3169 | return XML_ERROR_PARTIAL_CHAR; | |
3170 | case XML_TOK_NONE: /* start == end */ | |
3171 | default: | |
3172 | break; | |
3173 | } | |
3174 | return storeEntityValue(parser, enc, s, end); | |
3175 | } | |
3176 | start = next; | |
3177 | } | |
3178 | } | |
3179 | ||
3180 | #endif /* XML_DTD */ | |
3181 | ||
3182 | static enum XML_Error PTRCALL | |
3183 | prologProcessor(XML_Parser parser, | |
3184 | const char *s, | |
3185 | const char *end, | |
3186 | const char **nextPtr) | |
3187 | { | |
3188 | const char *next = s; | |
3189 | int tok = XmlPrologTok(encoding, s, end, &next); | |
3190 | return doProlog(parser, encoding, s, end, tok, next, nextPtr); | |
3191 | } | |
3192 | ||
3193 | static enum XML_Error | |
3194 | doProlog(XML_Parser parser, | |
3195 | const ENCODING *enc, | |
3196 | const char *s, | |
3197 | const char *end, | |
3198 | int tok, | |
3199 | const char *next, | |
3200 | const char **nextPtr) | |
3201 | { | |
3202 | #ifdef XML_DTD | |
3203 | static const XML_Char externalSubsetName[] = { '#' , '\0' }; | |
3204 | #endif /* XML_DTD */ | |
3205 | static const XML_Char atypeCDATA[] = { 'C', 'D', 'A', 'T', 'A', '\0' }; | |
3206 | static const XML_Char atypeID[] = { 'I', 'D', '\0' }; | |
3207 | static const XML_Char atypeIDREF[] = { 'I', 'D', 'R', 'E', 'F', '\0' }; | |
3208 | static const XML_Char atypeIDREFS[] = { 'I', 'D', 'R', 'E', 'F', 'S', '\0' }; | |
3209 | static const XML_Char atypeENTITY[] = { 'E', 'N', 'T', 'I', 'T', 'Y', '\0' }; | |
3210 | static const XML_Char atypeENTITIES[] = | |
3211 | { 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S', '\0' }; | |
3212 | static const XML_Char atypeNMTOKEN[] = { | |
3213 | 'N', 'M', 'T', 'O', 'K', 'E', 'N', '\0' }; | |
3214 | static const XML_Char atypeNMTOKENS[] = { | |
3215 | 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S', '\0' }; | |
3216 | static const XML_Char notationPrefix[] = { | |
3217 | 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N', '(', '\0' }; | |
3218 | static const XML_Char enumValueSep[] = { '|', '\0' }; | |
3219 | static const XML_Char enumValueStart[] = { '(', '\0' }; | |
3220 | ||
3221 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
3222 | ||
3223 | const char **eventPP; | |
3224 | const char **eventEndPP; | |
3225 | enum XML_Content_Quant quant; | |
3226 | ||
3227 | if (enc == encoding) { | |
3228 | eventPP = &eventPtr; | |
3229 | eventEndPP = &eventEndPtr; | |
3230 | } | |
3231 | else { | |
3232 | eventPP = &(openInternalEntities->internalEventPtr); | |
3233 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
3234 | } | |
3235 | for (;;) { | |
3236 | int role; | |
3237 | XML_Bool handleDefault = XML_TRUE; | |
3238 | *eventPP = s; | |
3239 | *eventEndPP = next; | |
3240 | if (tok <= 0) { | |
3241 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3242 | *nextPtr = s; | |
3243 | return XML_ERROR_NONE; | |
3244 | } | |
3245 | switch (tok) { | |
3246 | case XML_TOK_INVALID: | |
3247 | *eventPP = next; | |
3248 | return XML_ERROR_INVALID_TOKEN; | |
3249 | case XML_TOK_PARTIAL: | |
3250 | return XML_ERROR_UNCLOSED_TOKEN; | |
3251 | case XML_TOK_PARTIAL_CHAR: | |
3252 | return XML_ERROR_PARTIAL_CHAR; | |
3253 | case XML_TOK_NONE: | |
3254 | #ifdef XML_DTD | |
3255 | if (enc != encoding) | |
3256 | return XML_ERROR_NONE; | |
3257 | if (isParamEntity) { | |
3258 | if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc) | |
3259 | == XML_ROLE_ERROR) | |
3260 | return XML_ERROR_SYNTAX; | |
3261 | return XML_ERROR_NONE; | |
3262 | } | |
3263 | #endif /* XML_DTD */ | |
3264 | return XML_ERROR_NO_ELEMENTS; | |
3265 | default: | |
3266 | tok = -tok; | |
3267 | next = end; | |
3268 | break; | |
3269 | } | |
3270 | } | |
3271 | role = XmlTokenRole(&prologState, tok, s, next, enc); | |
3272 | switch (role) { | |
3273 | case XML_ROLE_XML_DECL: | |
3274 | { | |
3275 | enum XML_Error result = processXmlDecl(parser, 0, s, next); | |
3276 | if (result != XML_ERROR_NONE) | |
3277 | return result; | |
3278 | enc = encoding; | |
3279 | handleDefault = XML_FALSE; | |
3280 | } | |
3281 | break; | |
3282 | case XML_ROLE_DOCTYPE_NAME: | |
3283 | if (startDoctypeDeclHandler) { | |
3284 | doctypeName = poolStoreString(&tempPool, enc, s, next); | |
3285 | if (!doctypeName) | |
3286 | return XML_ERROR_NO_MEMORY; | |
3287 | poolFinish(&tempPool); | |
3288 | doctypePubid = NULL; | |
3289 | handleDefault = XML_FALSE; | |
3290 | } | |
3291 | doctypeSysid = NULL; /* always initialize to NULL */ | |
3292 | break; | |
3293 | case XML_ROLE_DOCTYPE_INTERNAL_SUBSET: | |
3294 | if (startDoctypeDeclHandler) { | |
3295 | startDoctypeDeclHandler(handlerArg, doctypeName, doctypeSysid, | |
3296 | doctypePubid, 1); | |
3297 | doctypeName = NULL; | |
3298 | poolClear(&tempPool); | |
3299 | handleDefault = XML_FALSE; | |
3300 | } | |
3301 | break; | |
3302 | #ifdef XML_DTD | |
3303 | case XML_ROLE_TEXT_DECL: | |
3304 | { | |
3305 | enum XML_Error result = processXmlDecl(parser, 1, s, next); | |
3306 | if (result != XML_ERROR_NONE) | |
3307 | return result; | |
3308 | enc = encoding; | |
3309 | handleDefault = XML_FALSE; | |
3310 | } | |
3311 | break; | |
3312 | #endif /* XML_DTD */ | |
3313 | case XML_ROLE_DOCTYPE_PUBLIC_ID: | |
3314 | #ifdef XML_DTD | |
3315 | useForeignDTD = XML_FALSE; | |
3316 | #endif /* XML_DTD */ | |
3317 | dtd->hasParamEntityRefs = XML_TRUE; | |
3318 | if (startDoctypeDeclHandler) { | |
3319 | doctypePubid = poolStoreString(&tempPool, enc, | |
3320 | s + enc->minBytesPerChar, | |
3321 | next - enc->minBytesPerChar); | |
3322 | if (!doctypePubid) | |
3323 | return XML_ERROR_NO_MEMORY; | |
3324 | poolFinish(&tempPool); | |
3325 | handleDefault = XML_FALSE; | |
3326 | } | |
3327 | #ifdef XML_DTD | |
3328 | declEntity = (ENTITY *)lookup(&dtd->paramEntities, | |
3329 | externalSubsetName, | |
3330 | sizeof(ENTITY)); | |
3331 | if (!declEntity) | |
3332 | return XML_ERROR_NO_MEMORY; | |
3333 | #endif /* XML_DTD */ | |
3334 | /* fall through */ | |
3335 | case XML_ROLE_ENTITY_PUBLIC_ID: | |
3336 | if (!XmlIsPublicId(enc, s, next, eventPP)) | |
3337 | return XML_ERROR_SYNTAX; | |
3338 | if (dtd->keepProcessing && declEntity) { | |
3339 | XML_Char *tem = poolStoreString(&dtd->pool, | |
3340 | enc, | |
3341 | s + enc->minBytesPerChar, | |
3342 | next - enc->minBytesPerChar); | |
3343 | if (!tem) | |
3344 | return XML_ERROR_NO_MEMORY; | |
3345 | normalizePublicId(tem); | |
3346 | declEntity->publicId = tem; | |
3347 | poolFinish(&dtd->pool); | |
3348 | if (entityDeclHandler) | |
3349 | handleDefault = XML_FALSE; | |
3350 | } | |
3351 | break; | |
3352 | case XML_ROLE_DOCTYPE_CLOSE: | |
3353 | if (doctypeName) { | |
3354 | startDoctypeDeclHandler(handlerArg, doctypeName, | |
3355 | doctypeSysid, doctypePubid, 0); | |
3356 | poolClear(&tempPool); | |
3357 | handleDefault = XML_FALSE; | |
3358 | } | |
3359 | /* doctypeSysid will be non-NULL in the case of a previous | |
3360 | XML_ROLE_DOCTYPE_SYSTEM_ID, even if startDoctypeDeclHandler | |
3361 | was not set, indicating an external subset | |
3362 | */ | |
3363 | #ifdef XML_DTD | |
3364 | if (doctypeSysid || useForeignDTD) { | |
3365 | dtd->hasParamEntityRefs = XML_TRUE; /* when docTypeSysid == NULL */ | |
3366 | if (paramEntityParsing && externalEntityRefHandler) { | |
3367 | ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities, | |
3368 | externalSubsetName, | |
3369 | sizeof(ENTITY)); | |
3370 | if (!entity) | |
3371 | return XML_ERROR_NO_MEMORY; | |
3372 | if (useForeignDTD) | |
3373 | entity->base = curBase; | |
3374 | dtd->paramEntityRead = XML_FALSE; | |
3375 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
3376 | 0, | |
3377 | entity->base, | |
3378 | entity->systemId, | |
3379 | entity->publicId)) | |
3380 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
3381 | if (dtd->paramEntityRead && | |
3382 | !dtd->standalone && | |
3383 | notStandaloneHandler && | |
3384 | !notStandaloneHandler(handlerArg)) | |
3385 | return XML_ERROR_NOT_STANDALONE; | |
3386 | /* end of DTD - no need to update dtd->keepProcessing */ | |
3387 | } | |
3388 | useForeignDTD = XML_FALSE; | |
3389 | } | |
3390 | #endif /* XML_DTD */ | |
3391 | if (endDoctypeDeclHandler) { | |
3392 | endDoctypeDeclHandler(handlerArg); | |
3393 | handleDefault = XML_FALSE; | |
3394 | } | |
3395 | break; | |
3396 | case XML_ROLE_INSTANCE_START: | |
3397 | #ifdef XML_DTD | |
3398 | /* if there is no DOCTYPE declaration then now is the | |
3399 | last chance to read the foreign DTD | |
3400 | */ | |
3401 | if (useForeignDTD) { | |
3402 | dtd->hasParamEntityRefs = XML_TRUE; | |
3403 | if (paramEntityParsing && externalEntityRefHandler) { | |
3404 | ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities, | |
3405 | externalSubsetName, | |
3406 | sizeof(ENTITY)); | |
3407 | if (!entity) | |
3408 | return XML_ERROR_NO_MEMORY; | |
3409 | entity->base = curBase; | |
3410 | dtd->paramEntityRead = XML_FALSE; | |
3411 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
3412 | 0, | |
3413 | entity->base, | |
3414 | entity->systemId, | |
3415 | entity->publicId)) | |
3416 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
3417 | if (dtd->paramEntityRead && | |
3418 | !dtd->standalone && | |
3419 | notStandaloneHandler && | |
3420 | !notStandaloneHandler(handlerArg)) | |
3421 | return XML_ERROR_NOT_STANDALONE; | |
3422 | /* end of DTD - no need to update dtd->keepProcessing */ | |
3423 | } | |
3424 | } | |
3425 | #endif /* XML_DTD */ | |
3426 | processor = contentProcessor; | |
3427 | return contentProcessor(parser, s, end, nextPtr); | |
3428 | case XML_ROLE_ATTLIST_ELEMENT_NAME: | |
3429 | declElementType = getElementType(parser, enc, s, next); | |
3430 | if (!declElementType) | |
3431 | return XML_ERROR_NO_MEMORY; | |
3432 | goto checkAttListDeclHandler; | |
3433 | case XML_ROLE_ATTRIBUTE_NAME: | |
3434 | declAttributeId = getAttributeId(parser, enc, s, next); | |
3435 | if (!declAttributeId) | |
3436 | return XML_ERROR_NO_MEMORY; | |
3437 | declAttributeIsCdata = XML_FALSE; | |
3438 | declAttributeType = NULL; | |
3439 | declAttributeIsId = XML_FALSE; | |
3440 | goto checkAttListDeclHandler; | |
3441 | case XML_ROLE_ATTRIBUTE_TYPE_CDATA: | |
3442 | declAttributeIsCdata = XML_TRUE; | |
3443 | declAttributeType = atypeCDATA; | |
3444 | goto checkAttListDeclHandler; | |
3445 | case XML_ROLE_ATTRIBUTE_TYPE_ID: | |
3446 | declAttributeIsId = XML_TRUE; | |
3447 | declAttributeType = atypeID; | |
3448 | goto checkAttListDeclHandler; | |
3449 | case XML_ROLE_ATTRIBUTE_TYPE_IDREF: | |
3450 | declAttributeType = atypeIDREF; | |
3451 | goto checkAttListDeclHandler; | |
3452 | case XML_ROLE_ATTRIBUTE_TYPE_IDREFS: | |
3453 | declAttributeType = atypeIDREFS; | |
3454 | goto checkAttListDeclHandler; | |
3455 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITY: | |
3456 | declAttributeType = atypeENTITY; | |
3457 | goto checkAttListDeclHandler; | |
3458 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITIES: | |
3459 | declAttributeType = atypeENTITIES; | |
3460 | goto checkAttListDeclHandler; | |
3461 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKEN: | |
3462 | declAttributeType = atypeNMTOKEN; | |
3463 | goto checkAttListDeclHandler; | |
3464 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS: | |
3465 | declAttributeType = atypeNMTOKENS; | |
3466 | checkAttListDeclHandler: | |
3467 | if (dtd->keepProcessing && attlistDeclHandler) | |
3468 | handleDefault = XML_FALSE; | |
3469 | break; | |
3470 | case XML_ROLE_ATTRIBUTE_ENUM_VALUE: | |
3471 | case XML_ROLE_ATTRIBUTE_NOTATION_VALUE: | |
3472 | if (dtd->keepProcessing && attlistDeclHandler) { | |
3473 | const XML_Char *prefix; | |
3474 | if (declAttributeType) { | |
3475 | prefix = enumValueSep; | |
3476 | } | |
3477 | else { | |
3478 | prefix = (role == XML_ROLE_ATTRIBUTE_NOTATION_VALUE | |
3479 | ? notationPrefix | |
3480 | : enumValueStart); | |
3481 | } | |
3482 | if (!poolAppendString(&tempPool, prefix)) | |
3483 | return XML_ERROR_NO_MEMORY; | |
3484 | if (!poolAppend(&tempPool, enc, s, next)) | |
3485 | return XML_ERROR_NO_MEMORY; | |
3486 | declAttributeType = tempPool.start; | |
3487 | handleDefault = XML_FALSE; | |
3488 | } | |
3489 | break; | |
3490 | case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: | |
3491 | case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: | |
3492 | if (dtd->keepProcessing) { | |
3493 | if (!defineAttribute(declElementType, declAttributeId, | |
3494 | declAttributeIsCdata, declAttributeIsId, 0, | |
3495 | parser)) | |
3496 | return XML_ERROR_NO_MEMORY; | |
3497 | if (attlistDeclHandler && declAttributeType) { | |
3498 | if (*declAttributeType == XML_T('(') | |
3499 | || (*declAttributeType == XML_T('N') | |
3500 | && declAttributeType[1] == XML_T('O'))) { | |
3501 | /* Enumerated or Notation type */ | |
3502 | if (!poolAppendChar(&tempPool, XML_T(')')) | |
3503 | || !poolAppendChar(&tempPool, XML_T('\0'))) | |
3504 | return XML_ERROR_NO_MEMORY; | |
3505 | declAttributeType = tempPool.start; | |
3506 | poolFinish(&tempPool); | |
3507 | } | |
3508 | *eventEndPP = s; | |
3509 | attlistDeclHandler(handlerArg, declElementType->name, | |
3510 | declAttributeId->name, declAttributeType, | |
3511 | 0, role == XML_ROLE_REQUIRED_ATTRIBUTE_VALUE); | |
3512 | poolClear(&tempPool); | |
3513 | handleDefault = XML_FALSE; | |
3514 | } | |
3515 | } | |
3516 | break; | |
3517 | case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: | |
3518 | case XML_ROLE_FIXED_ATTRIBUTE_VALUE: | |
3519 | if (dtd->keepProcessing) { | |
3520 | const XML_Char *attVal; | |
3521 | enum XML_Error result | |
3522 | = storeAttributeValue(parser, enc, declAttributeIsCdata, | |
3523 | s + enc->minBytesPerChar, | |
3524 | next - enc->minBytesPerChar, | |
3525 | &dtd->pool); | |
3526 | if (result) | |
3527 | return result; | |
3528 | attVal = poolStart(&dtd->pool); | |
3529 | poolFinish(&dtd->pool); | |
3530 | /* ID attributes aren't allowed to have a default */ | |
3531 | if (!defineAttribute(declElementType, declAttributeId, | |
3532 | declAttributeIsCdata, XML_FALSE, attVal, parser)) | |
3533 | return XML_ERROR_NO_MEMORY; | |
3534 | if (attlistDeclHandler && declAttributeType) { | |
3535 | if (*declAttributeType == XML_T('(') | |
3536 | || (*declAttributeType == XML_T('N') | |
3537 | && declAttributeType[1] == XML_T('O'))) { | |
3538 | /* Enumerated or Notation type */ | |
3539 | if (!poolAppendChar(&tempPool, XML_T(')')) | |
3540 | || !poolAppendChar(&tempPool, XML_T('\0'))) | |
3541 | return XML_ERROR_NO_MEMORY; | |
3542 | declAttributeType = tempPool.start; | |
3543 | poolFinish(&tempPool); | |
3544 | } | |
3545 | *eventEndPP = s; | |
3546 | attlistDeclHandler(handlerArg, declElementType->name, | |
3547 | declAttributeId->name, declAttributeType, | |
3548 | attVal, | |
3549 | role == XML_ROLE_FIXED_ATTRIBUTE_VALUE); | |
3550 | poolClear(&tempPool); | |
3551 | handleDefault = XML_FALSE; | |
3552 | } | |
3553 | } | |
3554 | break; | |
3555 | case XML_ROLE_ENTITY_VALUE: | |
3556 | if (dtd->keepProcessing) { | |
3557 | enum XML_Error result = storeEntityValue(parser, enc, | |
3558 | s + enc->minBytesPerChar, | |
3559 | next - enc->minBytesPerChar); | |
3560 | if (declEntity) { | |
3561 | declEntity->textPtr = poolStart(&dtd->entityValuePool); | |
3562 | declEntity->textLen = poolLength(&dtd->entityValuePool); | |
3563 | poolFinish(&dtd->entityValuePool); | |
3564 | if (entityDeclHandler) { | |
3565 | *eventEndPP = s; | |
3566 | entityDeclHandler(handlerArg, | |
3567 | declEntity->name, | |
3568 | declEntity->is_param, | |
3569 | declEntity->textPtr, | |
3570 | declEntity->textLen, | |
3571 | curBase, 0, 0, 0); | |
3572 | handleDefault = XML_FALSE; | |
3573 | } | |
3574 | } | |
3575 | else | |
3576 | poolDiscard(&dtd->entityValuePool); | |
3577 | if (result != XML_ERROR_NONE) | |
3578 | return result; | |
3579 | } | |
3580 | break; | |
3581 | case XML_ROLE_DOCTYPE_SYSTEM_ID: | |
3582 | #ifdef XML_DTD | |
3583 | useForeignDTD = XML_FALSE; | |
3584 | #endif /* XML_DTD */ | |
3585 | dtd->hasParamEntityRefs = XML_TRUE; | |
3586 | if (startDoctypeDeclHandler) { | |
3587 | doctypeSysid = poolStoreString(&tempPool, enc, | |
3588 | s + enc->minBytesPerChar, | |
3589 | next - enc->minBytesPerChar); | |
3590 | if (doctypeSysid == NULL) | |
3591 | return XML_ERROR_NO_MEMORY; | |
3592 | poolFinish(&tempPool); | |
3593 | handleDefault = XML_FALSE; | |
3594 | } | |
3595 | #ifdef XML_DTD | |
3596 | else | |
3597 | /* use externalSubsetName to make doctypeSysid non-NULL | |
3598 | for the case where no startDoctypeDeclHandler is set */ | |
3599 | doctypeSysid = externalSubsetName; | |
3600 | #endif /* XML_DTD */ | |
3601 | if (!dtd->standalone | |
3602 | #ifdef XML_DTD | |
3603 | && !paramEntityParsing | |
3604 | #endif /* XML_DTD */ | |
3605 | && notStandaloneHandler | |
3606 | && !notStandaloneHandler(handlerArg)) | |
3607 | return XML_ERROR_NOT_STANDALONE; | |
3608 | #ifndef XML_DTD | |
3609 | break; | |
3610 | #else /* XML_DTD */ | |
3611 | if (!declEntity) { | |
3612 | declEntity = (ENTITY *)lookup(&dtd->paramEntities, | |
3613 | externalSubsetName, | |
3614 | sizeof(ENTITY)); | |
3615 | if (!declEntity) | |
3616 | return XML_ERROR_NO_MEMORY; | |
3617 | declEntity->publicId = NULL; | |
3618 | } | |
3619 | /* fall through */ | |
3620 | #endif /* XML_DTD */ | |
3621 | case XML_ROLE_ENTITY_SYSTEM_ID: | |
3622 | if (dtd->keepProcessing && declEntity) { | |
3623 | declEntity->systemId = poolStoreString(&dtd->pool, enc, | |
3624 | s + enc->minBytesPerChar, | |
3625 | next - enc->minBytesPerChar); | |
3626 | if (!declEntity->systemId) | |
3627 | return XML_ERROR_NO_MEMORY; | |
3628 | declEntity->base = curBase; | |
3629 | poolFinish(&dtd->pool); | |
3630 | if (entityDeclHandler) | |
3631 | handleDefault = XML_FALSE; | |
3632 | } | |
3633 | break; | |
3634 | case XML_ROLE_ENTITY_COMPLETE: | |
3635 | if (dtd->keepProcessing && declEntity && entityDeclHandler) { | |
3636 | *eventEndPP = s; | |
3637 | entityDeclHandler(handlerArg, | |
3638 | declEntity->name, | |
3639 | declEntity->is_param, | |
3640 | 0,0, | |
3641 | declEntity->base, | |
3642 | declEntity->systemId, | |
3643 | declEntity->publicId, | |
3644 | 0); | |
3645 | handleDefault = XML_FALSE; | |
3646 | } | |
3647 | break; | |
3648 | case XML_ROLE_ENTITY_NOTATION_NAME: | |
3649 | if (dtd->keepProcessing && declEntity) { | |
3650 | declEntity->notation = poolStoreString(&dtd->pool, enc, s, next); | |
3651 | if (!declEntity->notation) | |
3652 | return XML_ERROR_NO_MEMORY; | |
3653 | poolFinish(&dtd->pool); | |
3654 | if (unparsedEntityDeclHandler) { | |
3655 | *eventEndPP = s; | |
3656 | unparsedEntityDeclHandler(handlerArg, | |
3657 | declEntity->name, | |
3658 | declEntity->base, | |
3659 | declEntity->systemId, | |
3660 | declEntity->publicId, | |
3661 | declEntity->notation); | |
3662 | handleDefault = XML_FALSE; | |
3663 | } | |
3664 | else if (entityDeclHandler) { | |
3665 | *eventEndPP = s; | |
3666 | entityDeclHandler(handlerArg, | |
3667 | declEntity->name, | |
3668 | 0,0,0, | |
3669 | declEntity->base, | |
3670 | declEntity->systemId, | |
3671 | declEntity->publicId, | |
3672 | declEntity->notation); | |
3673 | handleDefault = XML_FALSE; | |
3674 | } | |
3675 | } | |
3676 | break; | |
3677 | case XML_ROLE_GENERAL_ENTITY_NAME: | |
3678 | { | |
3679 | if (XmlPredefinedEntityName(enc, s, next)) { | |
3680 | declEntity = NULL; | |
3681 | break; | |
3682 | } | |
3683 | if (dtd->keepProcessing) { | |
3684 | const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next); | |
3685 | if (!name) | |
3686 | return XML_ERROR_NO_MEMORY; | |
3687 | declEntity = (ENTITY *)lookup(&dtd->generalEntities, name, | |
3688 | sizeof(ENTITY)); | |
3689 | if (!declEntity) | |
3690 | return XML_ERROR_NO_MEMORY; | |
3691 | if (declEntity->name != name) { | |
3692 | poolDiscard(&dtd->pool); | |
3693 | declEntity = NULL; | |
3694 | } | |
3695 | else { | |
3696 | poolFinish(&dtd->pool); | |
3697 | declEntity->publicId = NULL; | |
3698 | declEntity->is_param = XML_FALSE; | |
3699 | /* if we have a parent parser or are reading an internal parameter | |
3700 | entity, then the entity declaration is not considered "internal" | |
3701 | */ | |
3702 | declEntity->is_internal = !(parentParser || openInternalEntities); | |
3703 | if (entityDeclHandler) | |
3704 | handleDefault = XML_FALSE; | |
3705 | } | |
3706 | } | |
3707 | else { | |
3708 | poolDiscard(&dtd->pool); | |
3709 | declEntity = NULL; | |
3710 | } | |
3711 | } | |
3712 | break; | |
3713 | case XML_ROLE_PARAM_ENTITY_NAME: | |
3714 | #ifdef XML_DTD | |
3715 | if (dtd->keepProcessing) { | |
3716 | const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next); | |
3717 | if (!name) | |
3718 | return XML_ERROR_NO_MEMORY; | |
3719 | declEntity = (ENTITY *)lookup(&dtd->paramEntities, | |
3720 | name, sizeof(ENTITY)); | |
3721 | if (!declEntity) | |
3722 | return XML_ERROR_NO_MEMORY; | |
3723 | if (declEntity->name != name) { | |
3724 | poolDiscard(&dtd->pool); | |
3725 | declEntity = NULL; | |
3726 | } | |
3727 | else { | |
3728 | poolFinish(&dtd->pool); | |
3729 | declEntity->publicId = NULL; | |
3730 | declEntity->is_param = XML_TRUE; | |
3731 | /* if we have a parent parser or are reading an internal parameter | |
3732 | entity, then the entity declaration is not considered "internal" | |
3733 | */ | |
3734 | declEntity->is_internal = !(parentParser || openInternalEntities); | |
3735 | if (entityDeclHandler) | |
3736 | handleDefault = XML_FALSE; | |
3737 | } | |
3738 | } | |
3739 | else { | |
3740 | poolDiscard(&dtd->pool); | |
3741 | declEntity = NULL; | |
3742 | } | |
3743 | #else /* not XML_DTD */ | |
3744 | declEntity = NULL; | |
3745 | #endif /* XML_DTD */ | |
3746 | break; | |
3747 | case XML_ROLE_NOTATION_NAME: | |
3748 | declNotationPublicId = NULL; | |
3749 | declNotationName = NULL; | |
3750 | if (notationDeclHandler) { | |
3751 | declNotationName = poolStoreString(&tempPool, enc, s, next); | |
3752 | if (!declNotationName) | |
3753 | return XML_ERROR_NO_MEMORY; | |
3754 | poolFinish(&tempPool); | |
3755 | handleDefault = XML_FALSE; | |
3756 | } | |
3757 | break; | |
3758 | case XML_ROLE_NOTATION_PUBLIC_ID: | |
3759 | if (!XmlIsPublicId(enc, s, next, eventPP)) | |
3760 | return XML_ERROR_SYNTAX; | |
3761 | if (declNotationName) { /* means notationDeclHandler != NULL */ | |
3762 | XML_Char *tem = poolStoreString(&tempPool, | |
3763 | enc, | |
3764 | s + enc->minBytesPerChar, | |
3765 | next - enc->minBytesPerChar); | |
3766 | if (!tem) | |
3767 | return XML_ERROR_NO_MEMORY; | |
3768 | normalizePublicId(tem); | |
3769 | declNotationPublicId = tem; | |
3770 | poolFinish(&tempPool); | |
3771 | handleDefault = XML_FALSE; | |
3772 | } | |
3773 | break; | |
3774 | case XML_ROLE_NOTATION_SYSTEM_ID: | |
3775 | if (declNotationName && notationDeclHandler) { | |
3776 | const XML_Char *systemId | |
3777 | = poolStoreString(&tempPool, enc, | |
3778 | s + enc->minBytesPerChar, | |
3779 | next - enc->minBytesPerChar); | |
3780 | if (!systemId) | |
3781 | return XML_ERROR_NO_MEMORY; | |
3782 | *eventEndPP = s; | |
3783 | notationDeclHandler(handlerArg, | |
3784 | declNotationName, | |
3785 | curBase, | |
3786 | systemId, | |
3787 | declNotationPublicId); | |
3788 | handleDefault = XML_FALSE; | |
3789 | } | |
3790 | poolClear(&tempPool); | |
3791 | break; | |
3792 | case XML_ROLE_NOTATION_NO_SYSTEM_ID: | |
3793 | if (declNotationPublicId && notationDeclHandler) { | |
3794 | *eventEndPP = s; | |
3795 | notationDeclHandler(handlerArg, | |
3796 | declNotationName, | |
3797 | curBase, | |
3798 | 0, | |
3799 | declNotationPublicId); | |
3800 | handleDefault = XML_FALSE; | |
3801 | } | |
3802 | poolClear(&tempPool); | |
3803 | break; | |
3804 | case XML_ROLE_ERROR: | |
3805 | switch (tok) { | |
3806 | case XML_TOK_PARAM_ENTITY_REF: | |
3807 | return XML_ERROR_PARAM_ENTITY_REF; | |
3808 | case XML_TOK_XML_DECL: | |
3809 | return XML_ERROR_MISPLACED_XML_PI; | |
3810 | default: | |
3811 | return XML_ERROR_SYNTAX; | |
3812 | } | |
3813 | #ifdef XML_DTD | |
3814 | case XML_ROLE_IGNORE_SECT: | |
3815 | { | |
3816 | enum XML_Error result; | |
3817 | if (defaultHandler) | |
3818 | reportDefault(parser, enc, s, next); | |
3819 | handleDefault = XML_FALSE; | |
3820 | result = doIgnoreSection(parser, enc, &next, end, nextPtr); | |
3821 | if (!next) { | |
3822 | processor = ignoreSectionProcessor; | |
3823 | return result; | |
3824 | } | |
3825 | } | |
3826 | break; | |
3827 | #endif /* XML_DTD */ | |
3828 | case XML_ROLE_GROUP_OPEN: | |
3829 | if (prologState.level >= groupSize) { | |
3830 | if (groupSize) { | |
3831 | char *temp = (char *)REALLOC(groupConnector, groupSize *= 2); | |
3832 | if (temp == NULL) | |
3833 | return XML_ERROR_NO_MEMORY; | |
3834 | groupConnector = temp; | |
3835 | if (dtd->scaffIndex) { | |
3836 | int *temp = (int *)REALLOC(dtd->scaffIndex, | |
3837 | groupSize * sizeof(int)); | |
3838 | if (temp == NULL) | |
3839 | return XML_ERROR_NO_MEMORY; | |
3840 | dtd->scaffIndex = temp; | |
3841 | } | |
3842 | } | |
3843 | else { | |
3844 | groupConnector = (char *)MALLOC(groupSize = 32); | |
3845 | if (!groupConnector) | |
3846 | return XML_ERROR_NO_MEMORY; | |
3847 | } | |
3848 | } | |
3849 | groupConnector[prologState.level] = 0; | |
3850 | if (dtd->in_eldecl) { | |
3851 | int myindex = nextScaffoldPart(parser); | |
3852 | if (myindex < 0) | |
3853 | return XML_ERROR_NO_MEMORY; | |
3854 | dtd->scaffIndex[dtd->scaffLevel] = myindex; | |
3855 | dtd->scaffLevel++; | |
3856 | dtd->scaffold[myindex].type = XML_CTYPE_SEQ; | |
3857 | if (elementDeclHandler) | |
3858 | handleDefault = XML_FALSE; | |
3859 | } | |
3860 | break; | |
3861 | case XML_ROLE_GROUP_SEQUENCE: | |
3862 | if (groupConnector[prologState.level] == '|') | |
3863 | return XML_ERROR_SYNTAX; | |
3864 | groupConnector[prologState.level] = ','; | |
3865 | if (dtd->in_eldecl && elementDeclHandler) | |
3866 | handleDefault = XML_FALSE; | |
3867 | break; | |
3868 | case XML_ROLE_GROUP_CHOICE: | |
3869 | if (groupConnector[prologState.level] == ',') | |
3870 | return XML_ERROR_SYNTAX; | |
3871 | if (dtd->in_eldecl | |
3872 | && !groupConnector[prologState.level] | |
3873 | && (dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type | |
3874 | != XML_CTYPE_MIXED) | |
3875 | ) { | |
3876 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type | |
3877 | = XML_CTYPE_CHOICE; | |
3878 | if (elementDeclHandler) | |
3879 | handleDefault = XML_FALSE; | |
3880 | } | |
3881 | groupConnector[prologState.level] = '|'; | |
3882 | break; | |
3883 | case XML_ROLE_PARAM_ENTITY_REF: | |
3884 | #ifdef XML_DTD | |
3885 | case XML_ROLE_INNER_PARAM_ENTITY_REF: | |
3886 | /* PE references in internal subset are | |
3887 | not allowed within declarations */ | |
3888 | if (prologState.documentEntity && | |
3889 | role == XML_ROLE_INNER_PARAM_ENTITY_REF) | |
3890 | return XML_ERROR_PARAM_ENTITY_REF; | |
3891 | dtd->hasParamEntityRefs = XML_TRUE; | |
3892 | if (!paramEntityParsing) | |
3893 | dtd->keepProcessing = dtd->standalone; | |
3894 | else { | |
3895 | const XML_Char *name; | |
3896 | ENTITY *entity; | |
3897 | name = poolStoreString(&dtd->pool, enc, | |
3898 | s + enc->minBytesPerChar, | |
3899 | next - enc->minBytesPerChar); | |
3900 | if (!name) | |
3901 | return XML_ERROR_NO_MEMORY; | |
3902 | entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0); | |
3903 | poolDiscard(&dtd->pool); | |
3904 | /* first, determine if a check for an existing declaration is needed; | |
3905 | if yes, check that the entity exists, and that it is internal, | |
3906 | otherwise call the skipped entity handler | |
3907 | */ | |
3908 | if (prologState.documentEntity && | |
3909 | (dtd->standalone | |
3910 | ? !openInternalEntities | |
3911 | : !dtd->hasParamEntityRefs)) { | |
3912 | if (!entity) | |
3913 | return XML_ERROR_UNDEFINED_ENTITY; | |
3914 | else if (!entity->is_internal) | |
3915 | return XML_ERROR_ENTITY_DECLARED_IN_PE; | |
3916 | } | |
3917 | else if (!entity) { | |
3918 | dtd->keepProcessing = dtd->standalone; | |
3919 | /* cannot report skipped entities in declarations */ | |
3920 | if ((role == XML_ROLE_PARAM_ENTITY_REF) && skippedEntityHandler) { | |
3921 | skippedEntityHandler(handlerArg, name, 1); | |
3922 | handleDefault = XML_FALSE; | |
3923 | } | |
3924 | break; | |
3925 | } | |
3926 | if (entity->open) | |
3927 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
3928 | if (entity->textPtr) { | |
3929 | enum XML_Error result; | |
3930 | result = processInternalParamEntity(parser, entity); | |
3931 | if (result != XML_ERROR_NONE) | |
3932 | return result; | |
3933 | handleDefault = XML_FALSE; | |
3934 | break; | |
3935 | } | |
3936 | if (externalEntityRefHandler) { | |
3937 | dtd->paramEntityRead = XML_FALSE; | |
3938 | entity->open = XML_TRUE; | |
3939 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
3940 | 0, | |
3941 | entity->base, | |
3942 | entity->systemId, | |
3943 | entity->publicId)) { | |
3944 | entity->open = XML_FALSE; | |
3945 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
3946 | } | |
3947 | entity->open = XML_FALSE; | |
3948 | handleDefault = XML_FALSE; | |
3949 | if (!dtd->paramEntityRead) { | |
3950 | dtd->keepProcessing = dtd->standalone; | |
3951 | break; | |
3952 | } | |
3953 | } | |
3954 | else { | |
3955 | dtd->keepProcessing = dtd->standalone; | |
3956 | break; | |
3957 | } | |
3958 | } | |
3959 | #endif /* XML_DTD */ | |
3960 | if (!dtd->standalone && | |
3961 | notStandaloneHandler && | |
3962 | !notStandaloneHandler(handlerArg)) | |
3963 | return XML_ERROR_NOT_STANDALONE; | |
3964 | break; | |
3965 | ||
3966 | /* Element declaration stuff */ | |
3967 | ||
3968 | case XML_ROLE_ELEMENT_NAME: | |
3969 | if (elementDeclHandler) { | |
3970 | declElementType = getElementType(parser, enc, s, next); | |
3971 | if (!declElementType) | |
3972 | return XML_ERROR_NO_MEMORY; | |
3973 | dtd->scaffLevel = 0; | |
3974 | dtd->scaffCount = 0; | |
3975 | dtd->in_eldecl = XML_TRUE; | |
3976 | handleDefault = XML_FALSE; | |
3977 | } | |
3978 | break; | |
3979 | ||
3980 | case XML_ROLE_CONTENT_ANY: | |
3981 | case XML_ROLE_CONTENT_EMPTY: | |
3982 | if (dtd->in_eldecl) { | |
3983 | if (elementDeclHandler) { | |
3984 | XML_Content * content = (XML_Content *) MALLOC(sizeof(XML_Content)); | |
3985 | if (!content) | |
3986 | return XML_ERROR_NO_MEMORY; | |
3987 | content->quant = XML_CQUANT_NONE; | |
3988 | content->name = NULL; | |
3989 | content->numchildren = 0; | |
3990 | content->children = NULL; | |
3991 | content->type = ((role == XML_ROLE_CONTENT_ANY) ? | |
3992 | XML_CTYPE_ANY : | |
3993 | XML_CTYPE_EMPTY); | |
3994 | *eventEndPP = s; | |
3995 | elementDeclHandler(handlerArg, declElementType->name, content); | |
3996 | handleDefault = XML_FALSE; | |
3997 | } | |
3998 | dtd->in_eldecl = XML_FALSE; | |
3999 | } | |
4000 | break; | |
4001 | ||
4002 | case XML_ROLE_CONTENT_PCDATA: | |
4003 | if (dtd->in_eldecl) { | |
4004 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type | |
4005 | = XML_CTYPE_MIXED; | |
4006 | if (elementDeclHandler) | |
4007 | handleDefault = XML_FALSE; | |
4008 | } | |
4009 | break; | |
4010 | ||
4011 | case XML_ROLE_CONTENT_ELEMENT: | |
4012 | quant = XML_CQUANT_NONE; | |
4013 | goto elementContent; | |
4014 | case XML_ROLE_CONTENT_ELEMENT_OPT: | |
4015 | quant = XML_CQUANT_OPT; | |
4016 | goto elementContent; | |
4017 | case XML_ROLE_CONTENT_ELEMENT_REP: | |
4018 | quant = XML_CQUANT_REP; | |
4019 | goto elementContent; | |
4020 | case XML_ROLE_CONTENT_ELEMENT_PLUS: | |
4021 | quant = XML_CQUANT_PLUS; | |
4022 | elementContent: | |
4023 | if (dtd->in_eldecl) { | |
4024 | ELEMENT_TYPE *el; | |
4025 | const XML_Char *name; | |
4026 | int nameLen; | |
4027 | const char *nxt = (quant == XML_CQUANT_NONE | |
4028 | ? next | |
4029 | : next - enc->minBytesPerChar); | |
4030 | int myindex = nextScaffoldPart(parser); | |
4031 | if (myindex < 0) | |
4032 | return XML_ERROR_NO_MEMORY; | |
4033 | dtd->scaffold[myindex].type = XML_CTYPE_NAME; | |
4034 | dtd->scaffold[myindex].quant = quant; | |
4035 | el = getElementType(parser, enc, s, nxt); | |
4036 | if (!el) | |
4037 | return XML_ERROR_NO_MEMORY; | |
4038 | name = el->name; | |
4039 | dtd->scaffold[myindex].name = name; | |
4040 | nameLen = 0; | |
4041 | for (; name[nameLen++]; ) | |
4042 | ; | |
4043 | dtd->contentStringLen += nameLen; | |
4044 | if (elementDeclHandler) | |
4045 | handleDefault = XML_FALSE; | |
4046 | } | |
4047 | break; | |
4048 | ||
4049 | case XML_ROLE_GROUP_CLOSE: | |
4050 | quant = XML_CQUANT_NONE; | |
4051 | goto closeGroup; | |
4052 | case XML_ROLE_GROUP_CLOSE_OPT: | |
4053 | quant = XML_CQUANT_OPT; | |
4054 | goto closeGroup; | |
4055 | case XML_ROLE_GROUP_CLOSE_REP: | |
4056 | quant = XML_CQUANT_REP; | |
4057 | goto closeGroup; | |
4058 | case XML_ROLE_GROUP_CLOSE_PLUS: | |
4059 | quant = XML_CQUANT_PLUS; | |
4060 | closeGroup: | |
4061 | if (dtd->in_eldecl) { | |
4062 | if (elementDeclHandler) | |
4063 | handleDefault = XML_FALSE; | |
4064 | dtd->scaffLevel--; | |
4065 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel]].quant = quant; | |
4066 | if (dtd->scaffLevel == 0) { | |
4067 | if (!handleDefault) { | |
4068 | XML_Content *model = build_model(parser); | |
4069 | if (!model) | |
4070 | return XML_ERROR_NO_MEMORY; | |
4071 | *eventEndPP = s; | |
4072 | elementDeclHandler(handlerArg, declElementType->name, model); | |
4073 | } | |
4074 | dtd->in_eldecl = XML_FALSE; | |
4075 | dtd->contentStringLen = 0; | |
4076 | } | |
4077 | } | |
4078 | break; | |
4079 | /* End element declaration stuff */ | |
4080 | ||
4081 | case XML_ROLE_PI: | |
4082 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
4083 | return XML_ERROR_NO_MEMORY; | |
4084 | handleDefault = XML_FALSE; | |
4085 | break; | |
4086 | case XML_ROLE_COMMENT: | |
4087 | if (!reportComment(parser, enc, s, next)) | |
4088 | return XML_ERROR_NO_MEMORY; | |
4089 | handleDefault = XML_FALSE; | |
4090 | break; | |
4091 | case XML_ROLE_NONE: | |
4092 | switch (tok) { | |
4093 | case XML_TOK_BOM: | |
4094 | handleDefault = XML_FALSE; | |
4095 | break; | |
4096 | } | |
4097 | break; | |
4098 | case XML_ROLE_DOCTYPE_NONE: | |
4099 | if (startDoctypeDeclHandler) | |
4100 | handleDefault = XML_FALSE; | |
4101 | break; | |
4102 | case XML_ROLE_ENTITY_NONE: | |
4103 | if (dtd->keepProcessing && entityDeclHandler) | |
4104 | handleDefault = XML_FALSE; | |
4105 | break; | |
4106 | case XML_ROLE_NOTATION_NONE: | |
4107 | if (notationDeclHandler) | |
4108 | handleDefault = XML_FALSE; | |
4109 | break; | |
4110 | case XML_ROLE_ATTLIST_NONE: | |
4111 | if (dtd->keepProcessing && attlistDeclHandler) | |
4112 | handleDefault = XML_FALSE; | |
4113 | break; | |
4114 | case XML_ROLE_ELEMENT_NONE: | |
4115 | if (elementDeclHandler) | |
4116 | handleDefault = XML_FALSE; | |
4117 | break; | |
4118 | } /* end of big switch */ | |
4119 | ||
4120 | if (handleDefault && defaultHandler) | |
4121 | reportDefault(parser, enc, s, next); | |
4122 | ||
4123 | s = next; | |
4124 | tok = XmlPrologTok(enc, s, end, &next); | |
4125 | } | |
4126 | /* not reached */ | |
4127 | } | |
4128 | ||
4129 | static enum XML_Error PTRCALL | |
4130 | epilogProcessor(XML_Parser parser, | |
4131 | const char *s, | |
4132 | const char *end, | |
4133 | const char **nextPtr) | |
4134 | { | |
4135 | processor = epilogProcessor; | |
4136 | eventPtr = s; | |
4137 | for (;;) { | |
4138 | const char *next = NULL; | |
4139 | int tok = XmlPrologTok(encoding, s, end, &next); | |
4140 | eventEndPtr = next; | |
4141 | switch (tok) { | |
4142 | /* report partial linebreak - it might be the last token */ | |
4143 | case -XML_TOK_PROLOG_S: | |
4144 | if (defaultHandler) { | |
4145 | eventEndPtr = next; | |
4146 | reportDefault(parser, encoding, s, next); | |
4147 | } | |
4148 | if (nextPtr) | |
4149 | *nextPtr = next; | |
4150 | return XML_ERROR_NONE; | |
4151 | case XML_TOK_NONE: | |
4152 | if (nextPtr) | |
4153 | *nextPtr = s; | |
4154 | return XML_ERROR_NONE; | |
4155 | case XML_TOK_PROLOG_S: | |
4156 | if (defaultHandler) | |
4157 | reportDefault(parser, encoding, s, next); | |
4158 | break; | |
4159 | case XML_TOK_PI: | |
4160 | if (!reportProcessingInstruction(parser, encoding, s, next)) | |
4161 | return XML_ERROR_NO_MEMORY; | |
4162 | break; | |
4163 | case XML_TOK_COMMENT: | |
4164 | if (!reportComment(parser, encoding, s, next)) | |
4165 | return XML_ERROR_NO_MEMORY; | |
4166 | break; | |
4167 | case XML_TOK_INVALID: | |
4168 | eventPtr = next; | |
4169 | return XML_ERROR_INVALID_TOKEN; | |
4170 | case XML_TOK_PARTIAL: | |
4171 | if (nextPtr) { | |
4172 | *nextPtr = s; | |
4173 | return XML_ERROR_NONE; | |
4174 | } | |
4175 | return XML_ERROR_UNCLOSED_TOKEN; | |
4176 | case XML_TOK_PARTIAL_CHAR: | |
4177 | if (nextPtr) { | |
4178 | *nextPtr = s; | |
4179 | return XML_ERROR_NONE; | |
4180 | } | |
4181 | return XML_ERROR_PARTIAL_CHAR; | |
4182 | default: | |
4183 | return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; | |
4184 | } | |
4185 | eventPtr = s = next; | |
4186 | } | |
4187 | } | |
4188 | ||
4189 | #ifdef XML_DTD | |
4190 | ||
4191 | static enum XML_Error | |
4192 | processInternalParamEntity(XML_Parser parser, ENTITY *entity) | |
4193 | { | |
4194 | const char *s, *end, *next; | |
4195 | int tok; | |
4196 | enum XML_Error result; | |
4197 | OPEN_INTERNAL_ENTITY openEntity; | |
4198 | entity->open = XML_TRUE; | |
4199 | openEntity.next = openInternalEntities; | |
4200 | openInternalEntities = &openEntity; | |
4201 | openEntity.entity = entity; | |
4202 | openEntity.internalEventPtr = NULL; | |
4203 | openEntity.internalEventEndPtr = NULL; | |
4204 | s = (char *)entity->textPtr; | |
4205 | end = (char *)(entity->textPtr + entity->textLen); | |
4206 | tok = XmlPrologTok(internalEncoding, s, end, &next); | |
4207 | result = doProlog(parser, internalEncoding, s, end, tok, next, 0); | |
4208 | entity->open = XML_FALSE; | |
4209 | openInternalEntities = openEntity.next; | |
4210 | return result; | |
4211 | } | |
4212 | ||
4213 | #endif /* XML_DTD */ | |
4214 | ||
4215 | static enum XML_Error PTRCALL | |
4216 | errorProcessor(XML_Parser parser, | |
4217 | const char *s, | |
4218 | const char *end, | |
4219 | const char **nextPtr) | |
4220 | { | |
4221 | return errorCode; | |
4222 | } | |
4223 | ||
4224 | static enum XML_Error | |
4225 | storeAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata, | |
4226 | const char *ptr, const char *end, | |
4227 | STRING_POOL *pool) | |
4228 | { | |
4229 | enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr, | |
4230 | end, pool); | |
4231 | if (result) | |
4232 | return result; | |
4233 | if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) | |
4234 | poolChop(pool); | |
4235 | if (!poolAppendChar(pool, XML_T('\0'))) | |
4236 | return XML_ERROR_NO_MEMORY; | |
4237 | return XML_ERROR_NONE; | |
4238 | } | |
4239 | ||
4240 | static enum XML_Error | |
4241 | appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata, | |
4242 | const char *ptr, const char *end, | |
4243 | STRING_POOL *pool) | |
4244 | { | |
4245 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4246 | for (;;) { | |
4247 | const char *next; | |
4248 | int tok = XmlAttributeValueTok(enc, ptr, end, &next); | |
4249 | switch (tok) { | |
4250 | case XML_TOK_NONE: | |
4251 | return XML_ERROR_NONE; | |
4252 | case XML_TOK_INVALID: | |
4253 | if (enc == encoding) | |
4254 | eventPtr = next; | |
4255 | return XML_ERROR_INVALID_TOKEN; | |
4256 | case XML_TOK_PARTIAL: | |
4257 | if (enc == encoding) | |
4258 | eventPtr = ptr; | |
4259 | return XML_ERROR_INVALID_TOKEN; | |
4260 | case XML_TOK_CHAR_REF: | |
4261 | { | |
4262 | XML_Char buf[XML_ENCODE_MAX]; | |
4263 | int i; | |
4264 | int n = XmlCharRefNumber(enc, ptr); | |
4265 | if (n < 0) { | |
4266 | if (enc == encoding) | |
4267 | eventPtr = ptr; | |
4268 | return XML_ERROR_BAD_CHAR_REF; | |
4269 | } | |
4270 | if (!isCdata | |
4271 | && n == 0x20 /* space */ | |
4272 | && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
4273 | break; | |
4274 | n = XmlEncode(n, (ICHAR *)buf); | |
4275 | if (!n) { | |
4276 | if (enc == encoding) | |
4277 | eventPtr = ptr; | |
4278 | return XML_ERROR_BAD_CHAR_REF; | |
4279 | } | |
4280 | for (i = 0; i < n; i++) { | |
4281 | if (!poolAppendChar(pool, buf[i])) | |
4282 | return XML_ERROR_NO_MEMORY; | |
4283 | } | |
4284 | } | |
4285 | break; | |
4286 | case XML_TOK_DATA_CHARS: | |
4287 | if (!poolAppend(pool, enc, ptr, next)) | |
4288 | return XML_ERROR_NO_MEMORY; | |
4289 | break; | |
4290 | case XML_TOK_TRAILING_CR: | |
4291 | next = ptr + enc->minBytesPerChar; | |
4292 | /* fall through */ | |
4293 | case XML_TOK_ATTRIBUTE_VALUE_S: | |
4294 | case XML_TOK_DATA_NEWLINE: | |
4295 | if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
4296 | break; | |
4297 | if (!poolAppendChar(pool, 0x20)) | |
4298 | return XML_ERROR_NO_MEMORY; | |
4299 | break; | |
4300 | case XML_TOK_ENTITY_REF: | |
4301 | { | |
4302 | const XML_Char *name; | |
4303 | ENTITY *entity; | |
4304 | char checkEntityDecl; | |
4305 | XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc, | |
4306 | ptr + enc->minBytesPerChar, | |
4307 | next - enc->minBytesPerChar); | |
4308 | if (ch) { | |
4309 | if (!poolAppendChar(pool, ch)) | |
4310 | return XML_ERROR_NO_MEMORY; | |
4311 | break; | |
4312 | } | |
4313 | name = poolStoreString(&temp2Pool, enc, | |
4314 | ptr + enc->minBytesPerChar, | |
4315 | next - enc->minBytesPerChar); | |
4316 | if (!name) | |
4317 | return XML_ERROR_NO_MEMORY; | |
4318 | entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0); | |
4319 | poolDiscard(&temp2Pool); | |
4320 | /* first, determine if a check for an existing declaration is needed; | |
4321 | if yes, check that the entity exists, and that it is internal, | |
4322 | otherwise call the default handler (if called from content) | |
4323 | */ | |
4324 | if (pool == &dtd->pool) /* are we called from prolog? */ | |
4325 | checkEntityDecl = | |
4326 | #ifdef XML_DTD | |
4327 | prologState.documentEntity && | |
4328 | #endif /* XML_DTD */ | |
4329 | (dtd->standalone | |
4330 | ? !openInternalEntities | |
4331 | : !dtd->hasParamEntityRefs); | |
4332 | else /* if (pool == &tempPool): we are called from content */ | |
4333 | checkEntityDecl = !dtd->hasParamEntityRefs || dtd->standalone; | |
4334 | if (checkEntityDecl) { | |
4335 | if (!entity) | |
4336 | return XML_ERROR_UNDEFINED_ENTITY; | |
4337 | else if (!entity->is_internal) | |
4338 | return XML_ERROR_ENTITY_DECLARED_IN_PE; | |
4339 | } | |
4340 | else if (!entity) { | |
4341 | /* cannot report skipped entity here - see comments on | |
4342 | skippedEntityHandler | |
4343 | if (skippedEntityHandler) | |
4344 | skippedEntityHandler(handlerArg, name, 0); | |
4345 | */ | |
4346 | if ((pool == &tempPool) && defaultHandler) | |
4347 | reportDefault(parser, enc, ptr, next); | |
4348 | break; | |
4349 | } | |
4350 | if (entity->open) { | |
4351 | if (enc == encoding) | |
4352 | eventPtr = ptr; | |
4353 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
4354 | } | |
4355 | if (entity->notation) { | |
4356 | if (enc == encoding) | |
4357 | eventPtr = ptr; | |
4358 | return XML_ERROR_BINARY_ENTITY_REF; | |
4359 | } | |
4360 | if (!entity->textPtr) { | |
4361 | if (enc == encoding) | |
4362 | eventPtr = ptr; | |
4363 | return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; | |
4364 | } | |
4365 | else { | |
4366 | enum XML_Error result; | |
4367 | const XML_Char *textEnd = entity->textPtr + entity->textLen; | |
4368 | entity->open = XML_TRUE; | |
4369 | result = appendAttributeValue(parser, internalEncoding, isCdata, | |
4370 | (char *)entity->textPtr, | |
4371 | (char *)textEnd, pool); | |
4372 | entity->open = XML_FALSE; | |
4373 | if (result) | |
4374 | return result; | |
4375 | } | |
4376 | } | |
4377 | break; | |
4378 | default: | |
4379 | if (enc == encoding) | |
4380 | eventPtr = ptr; | |
4381 | return XML_ERROR_UNEXPECTED_STATE; | |
4382 | } | |
4383 | ptr = next; | |
4384 | } | |
4385 | /* not reached */ | |
4386 | } | |
4387 | ||
4388 | static enum XML_Error | |
4389 | storeEntityValue(XML_Parser parser, | |
4390 | const ENCODING *enc, | |
4391 | const char *entityTextPtr, | |
4392 | const char *entityTextEnd) | |
4393 | { | |
4394 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4395 | STRING_POOL *pool = &(dtd->entityValuePool); | |
4396 | enum XML_Error result = XML_ERROR_NONE; | |
4397 | #ifdef XML_DTD | |
4398 | int oldInEntityValue = prologState.inEntityValue; | |
4399 | prologState.inEntityValue = 1; | |
4400 | #endif /* XML_DTD */ | |
4401 | /* never return Null for the value argument in EntityDeclHandler, | |
4402 | since this would indicate an external entity; therefore we | |
4403 | have to make sure that entityValuePool.start is not null */ | |
4404 | if (!pool->blocks) { | |
4405 | if (!poolGrow(pool)) | |
4406 | return XML_ERROR_NO_MEMORY; | |
4407 | } | |
4408 | ||
4409 | for (;;) { | |
4410 | const char *next; | |
4411 | int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next); | |
4412 | switch (tok) { | |
4413 | case XML_TOK_PARAM_ENTITY_REF: | |
4414 | #ifdef XML_DTD | |
4415 | if (isParamEntity || enc != encoding) { | |
4416 | const XML_Char *name; | |
4417 | ENTITY *entity; | |
4418 | name = poolStoreString(&tempPool, enc, | |
4419 | entityTextPtr + enc->minBytesPerChar, | |
4420 | next - enc->minBytesPerChar); | |
4421 | if (!name) { | |
4422 | result = XML_ERROR_NO_MEMORY; | |
4423 | goto endEntityValue; | |
4424 | } | |
4425 | entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0); | |
4426 | poolDiscard(&tempPool); | |
4427 | if (!entity) { | |
4428 | /* not a well-formedness error - see XML 1.0: WFC Entity Declared */ | |
4429 | /* cannot report skipped entity here - see comments on | |
4430 | skippedEntityHandler | |
4431 | if (skippedEntityHandler) | |
4432 | skippedEntityHandler(handlerArg, name, 0); | |
4433 | */ | |
4434 | dtd->keepProcessing = dtd->standalone; | |
4435 | goto endEntityValue; | |
4436 | } | |
4437 | if (entity->open) { | |
4438 | if (enc == encoding) | |
4439 | eventPtr = entityTextPtr; | |
4440 | result = XML_ERROR_RECURSIVE_ENTITY_REF; | |
4441 | goto endEntityValue; | |
4442 | } | |
4443 | if (entity->systemId) { | |
4444 | if (externalEntityRefHandler) { | |
4445 | dtd->paramEntityRead = XML_FALSE; | |
4446 | entity->open = XML_TRUE; | |
4447 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
4448 | 0, | |
4449 | entity->base, | |
4450 | entity->systemId, | |
4451 | entity->publicId)) { | |
4452 | entity->open = XML_FALSE; | |
4453 | result = XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
4454 | goto endEntityValue; | |
4455 | } | |
4456 | entity->open = XML_FALSE; | |
4457 | if (!dtd->paramEntityRead) | |
4458 | dtd->keepProcessing = dtd->standalone; | |
4459 | } | |
4460 | else | |
4461 | dtd->keepProcessing = dtd->standalone; | |
4462 | } | |
4463 | else { | |
4464 | entity->open = XML_TRUE; | |
4465 | result = storeEntityValue(parser, | |
4466 | internalEncoding, | |
4467 | (char *)entity->textPtr, | |
4468 | (char *)(entity->textPtr | |
4469 | + entity->textLen)); | |
4470 | entity->open = XML_FALSE; | |
4471 | if (result) | |
4472 | goto endEntityValue; | |
4473 | } | |
4474 | break; | |
4475 | } | |
4476 | #endif /* XML_DTD */ | |
4477 | /* in the internal subset, PE references are not legal | |
4478 | within markup declarations, e.g entity values in this case */ | |
4479 | eventPtr = entityTextPtr; | |
4480 | result = XML_ERROR_PARAM_ENTITY_REF; | |
4481 | goto endEntityValue; | |
4482 | case XML_TOK_NONE: | |
4483 | result = XML_ERROR_NONE; | |
4484 | goto endEntityValue; | |
4485 | case XML_TOK_ENTITY_REF: | |
4486 | case XML_TOK_DATA_CHARS: | |
4487 | if (!poolAppend(pool, enc, entityTextPtr, next)) { | |
4488 | result = XML_ERROR_NO_MEMORY; | |
4489 | goto endEntityValue; | |
4490 | } | |
4491 | break; | |
4492 | case XML_TOK_TRAILING_CR: | |
4493 | next = entityTextPtr + enc->minBytesPerChar; | |
4494 | /* fall through */ | |
4495 | case XML_TOK_DATA_NEWLINE: | |
4496 | if (pool->end == pool->ptr && !poolGrow(pool)) { | |
4497 | result = XML_ERROR_NO_MEMORY; | |
4498 | goto endEntityValue; | |
4499 | } | |
4500 | *(pool->ptr)++ = 0xA; | |
4501 | break; | |
4502 | case XML_TOK_CHAR_REF: | |
4503 | { | |
4504 | XML_Char buf[XML_ENCODE_MAX]; | |
4505 | int i; | |
4506 | int n = XmlCharRefNumber(enc, entityTextPtr); | |
4507 | if (n < 0) { | |
4508 | if (enc == encoding) | |
4509 | eventPtr = entityTextPtr; | |
4510 | result = XML_ERROR_BAD_CHAR_REF; | |
4511 | goto endEntityValue; | |
4512 | } | |
4513 | n = XmlEncode(n, (ICHAR *)buf); | |
4514 | if (!n) { | |
4515 | if (enc == encoding) | |
4516 | eventPtr = entityTextPtr; | |
4517 | result = XML_ERROR_BAD_CHAR_REF; | |
4518 | goto endEntityValue; | |
4519 | } | |
4520 | for (i = 0; i < n; i++) { | |
4521 | if (pool->end == pool->ptr && !poolGrow(pool)) { | |
4522 | result = XML_ERROR_NO_MEMORY; | |
4523 | goto endEntityValue; | |
4524 | } | |
4525 | *(pool->ptr)++ = buf[i]; | |
4526 | } | |
4527 | } | |
4528 | break; | |
4529 | case XML_TOK_PARTIAL: | |
4530 | if (enc == encoding) | |
4531 | eventPtr = entityTextPtr; | |
4532 | result = XML_ERROR_INVALID_TOKEN; | |
4533 | goto endEntityValue; | |
4534 | case XML_TOK_INVALID: | |
4535 | if (enc == encoding) | |
4536 | eventPtr = next; | |
4537 | result = XML_ERROR_INVALID_TOKEN; | |
4538 | goto endEntityValue; | |
4539 | default: | |
4540 | if (enc == encoding) | |
4541 | eventPtr = entityTextPtr; | |
4542 | result = XML_ERROR_UNEXPECTED_STATE; | |
4543 | goto endEntityValue; | |
4544 | } | |
4545 | entityTextPtr = next; | |
4546 | } | |
4547 | endEntityValue: | |
4548 | #ifdef XML_DTD | |
4549 | prologState.inEntityValue = oldInEntityValue; | |
4550 | #endif /* XML_DTD */ | |
4551 | return result; | |
4552 | } | |
4553 | ||
4554 | static void FASTCALL | |
4555 | normalizeLines(XML_Char *s) | |
4556 | { | |
4557 | XML_Char *p; | |
4558 | for (;; s++) { | |
4559 | if (*s == XML_T('\0')) | |
4560 | return; | |
4561 | if (*s == 0xD) | |
4562 | break; | |
4563 | } | |
4564 | p = s; | |
4565 | do { | |
4566 | if (*s == 0xD) { | |
4567 | *p++ = 0xA; | |
4568 | if (*++s == 0xA) | |
4569 | s++; | |
4570 | } | |
4571 | else | |
4572 | *p++ = *s++; | |
4573 | } while (*s); | |
4574 | *p = XML_T('\0'); | |
4575 | } | |
4576 | ||
4577 | static int | |
4578 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, | |
4579 | const char *start, const char *end) | |
4580 | { | |
4581 | const XML_Char *target; | |
4582 | XML_Char *data; | |
4583 | const char *tem; | |
4584 | if (!processingInstructionHandler) { | |
4585 | if (defaultHandler) | |
4586 | reportDefault(parser, enc, start, end); | |
4587 | return 1; | |
4588 | } | |
4589 | start += enc->minBytesPerChar * 2; | |
4590 | tem = start + XmlNameLength(enc, start); | |
4591 | target = poolStoreString(&tempPool, enc, start, tem); | |
4592 | if (!target) | |
4593 | return 0; | |
4594 | poolFinish(&tempPool); | |
4595 | data = poolStoreString(&tempPool, enc, | |
4596 | XmlSkipS(enc, tem), | |
4597 | end - enc->minBytesPerChar*2); | |
4598 | if (!data) | |
4599 | return 0; | |
4600 | normalizeLines(data); | |
4601 | processingInstructionHandler(handlerArg, target, data); | |
4602 | poolClear(&tempPool); | |
4603 | return 1; | |
4604 | } | |
4605 | ||
4606 | static int | |
4607 | reportComment(XML_Parser parser, const ENCODING *enc, | |
4608 | const char *start, const char *end) | |
4609 | { | |
4610 | XML_Char *data; | |
4611 | if (!commentHandler) { | |
4612 | if (defaultHandler) | |
4613 | reportDefault(parser, enc, start, end); | |
4614 | return 1; | |
4615 | } | |
4616 | data = poolStoreString(&tempPool, | |
4617 | enc, | |
4618 | start + enc->minBytesPerChar * 4, | |
4619 | end - enc->minBytesPerChar * 3); | |
4620 | if (!data) | |
4621 | return 0; | |
4622 | normalizeLines(data); | |
4623 | commentHandler(handlerArg, data); | |
4624 | poolClear(&tempPool); | |
4625 | return 1; | |
4626 | } | |
4627 | ||
4628 | static void | |
4629 | reportDefault(XML_Parser parser, const ENCODING *enc, | |
4630 | const char *s, const char *end) | |
4631 | { | |
4632 | if (MUST_CONVERT(enc, s)) { | |
4633 | const char **eventPP; | |
4634 | const char **eventEndPP; | |
4635 | if (enc == encoding) { | |
4636 | eventPP = &eventPtr; | |
4637 | eventEndPP = &eventEndPtr; | |
4638 | } | |
4639 | else { | |
4640 | eventPP = &(openInternalEntities->internalEventPtr); | |
4641 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
4642 | } | |
4643 | do { | |
4644 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
4645 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
4646 | *eventEndPP = s; | |
4647 | defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
4648 | *eventPP = s; | |
4649 | } while (s != end); | |
4650 | } | |
4651 | else | |
4652 | defaultHandler(handlerArg, (XML_Char *)s, (XML_Char *)end - (XML_Char *)s); | |
4653 | } | |
4654 | ||
4655 | ||
4656 | static int | |
4657 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, XML_Bool isCdata, | |
4658 | XML_Bool isId, const XML_Char *value, XML_Parser parser) | |
4659 | { | |
4660 | DEFAULT_ATTRIBUTE *att; | |
4661 | if (value || isId) { | |
4662 | /* The handling of default attributes gets messed up if we have | |
4663 | a default which duplicates a non-default. */ | |
4664 | int i; | |
4665 | for (i = 0; i < type->nDefaultAtts; i++) | |
4666 | if (attId == type->defaultAtts[i].id) | |
4667 | return 1; | |
4668 | if (isId && !type->idAtt && !attId->xmlns) | |
4669 | type->idAtt = attId; | |
4670 | } | |
4671 | if (type->nDefaultAtts == type->allocDefaultAtts) { | |
4672 | if (type->allocDefaultAtts == 0) { | |
4673 | type->allocDefaultAtts = 8; | |
4674 | type->defaultAtts = (DEFAULT_ATTRIBUTE *)MALLOC(type->allocDefaultAtts | |
4675 | * sizeof(DEFAULT_ATTRIBUTE)); | |
4676 | if (!type->defaultAtts) | |
4677 | return 0; | |
4678 | } | |
4679 | else { | |
4680 | DEFAULT_ATTRIBUTE *temp; | |
4681 | int count = type->allocDefaultAtts * 2; | |
4682 | temp = (DEFAULT_ATTRIBUTE *) | |
4683 | REALLOC(type->defaultAtts, (count * sizeof(DEFAULT_ATTRIBUTE))); | |
4684 | if (temp == NULL) | |
4685 | return 0; | |
4686 | type->allocDefaultAtts = count; | |
4687 | type->defaultAtts = temp; | |
4688 | } | |
4689 | } | |
4690 | att = type->defaultAtts + type->nDefaultAtts; | |
4691 | att->id = attId; | |
4692 | att->value = value; | |
4693 | att->isCdata = isCdata; | |
4694 | if (!isCdata) | |
4695 | attId->maybeTokenized = XML_TRUE; | |
4696 | type->nDefaultAtts += 1; | |
4697 | return 1; | |
4698 | } | |
4699 | ||
4700 | static int | |
4701 | setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) | |
4702 | { | |
4703 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4704 | const XML_Char *name; | |
4705 | for (name = elementType->name; *name; name++) { | |
4706 | if (*name == XML_T(':')) { | |
4707 | PREFIX *prefix; | |
4708 | const XML_Char *s; | |
4709 | for (s = elementType->name; s != name; s++) { | |
4710 | if (!poolAppendChar(&dtd->pool, *s)) | |
4711 | return 0; | |
4712 | } | |
4713 | if (!poolAppendChar(&dtd->pool, XML_T('\0'))) | |
4714 | return 0; | |
4715 | prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool), | |
4716 | sizeof(PREFIX)); | |
4717 | if (!prefix) | |
4718 | return 0; | |
4719 | if (prefix->name == poolStart(&dtd->pool)) | |
4720 | poolFinish(&dtd->pool); | |
4721 | else | |
4722 | poolDiscard(&dtd->pool); | |
4723 | elementType->prefix = prefix; | |
4724 | ||
4725 | } | |
4726 | } | |
4727 | return 1; | |
4728 | } | |
4729 | ||
4730 | static ATTRIBUTE_ID * | |
4731 | getAttributeId(XML_Parser parser, const ENCODING *enc, | |
4732 | const char *start, const char *end) | |
4733 | { | |
4734 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4735 | ATTRIBUTE_ID *id; | |
4736 | const XML_Char *name; | |
4737 | if (!poolAppendChar(&dtd->pool, XML_T('\0'))) | |
4738 | return NULL; | |
4739 | name = poolStoreString(&dtd->pool, enc, start, end); | |
4740 | if (!name) | |
4741 | return NULL; | |
4742 | ++name; | |
4743 | id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, name, sizeof(ATTRIBUTE_ID)); | |
4744 | if (!id) | |
4745 | return NULL; | |
4746 | if (id->name != name) | |
4747 | poolDiscard(&dtd->pool); | |
4748 | else { | |
4749 | poolFinish(&dtd->pool); | |
4750 | if (!ns) | |
4751 | ; | |
4752 | else if (name[0] == XML_T('x') | |
4753 | && name[1] == XML_T('m') | |
4754 | && name[2] == XML_T('l') | |
4755 | && name[3] == XML_T('n') | |
4756 | && name[4] == XML_T('s') | |
4757 | && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) { | |
4758 | if (name[5] == XML_T('\0')) | |
4759 | id->prefix = &dtd->defaultPrefix; | |
4760 | else | |
4761 | id->prefix = (PREFIX *)lookup(&dtd->prefixes, name + 6, sizeof(PREFIX)); | |
4762 | id->xmlns = XML_TRUE; | |
4763 | } | |
4764 | else { | |
4765 | int i; | |
4766 | for (i = 0; name[i]; i++) { | |
4767 | if (name[i] == XML_T(':')) { | |
4768 | int j; | |
4769 | for (j = 0; j < i; j++) { | |
4770 | if (!poolAppendChar(&dtd->pool, name[j])) | |
4771 | return NULL; | |
4772 | } | |
4773 | if (!poolAppendChar(&dtd->pool, XML_T('\0'))) | |
4774 | return NULL; | |
4775 | id->prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool), | |
4776 | sizeof(PREFIX)); | |
4777 | if (id->prefix->name == poolStart(&dtd->pool)) | |
4778 | poolFinish(&dtd->pool); | |
4779 | else | |
4780 | poolDiscard(&dtd->pool); | |
4781 | break; | |
4782 | } | |
4783 | } | |
4784 | } | |
4785 | } | |
4786 | return id; | |
4787 | } | |
4788 | ||
4789 | #define CONTEXT_SEP XML_T('\f') | |
4790 | ||
4791 | static const XML_Char * | |
4792 | getContext(XML_Parser parser) | |
4793 | { | |
4794 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4795 | HASH_TABLE_ITER iter; | |
4796 | XML_Bool needSep = XML_FALSE; | |
4797 | ||
4798 | if (dtd->defaultPrefix.binding) { | |
4799 | int i; | |
4800 | int len; | |
4801 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
4802 | return NULL; | |
4803 | len = dtd->defaultPrefix.binding->uriLen; | |
4804 | if (namespaceSeparator != XML_T('\0')) | |
4805 | len--; | |
4806 | for (i = 0; i < len; i++) | |
4807 | if (!poolAppendChar(&tempPool, dtd->defaultPrefix.binding->uri[i])) | |
4808 | return NULL; | |
4809 | needSep = XML_TRUE; | |
4810 | } | |
4811 | ||
4812 | hashTableIterInit(&iter, &(dtd->prefixes)); | |
4813 | for (;;) { | |
4814 | int i; | |
4815 | int len; | |
4816 | const XML_Char *s; | |
4817 | PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); | |
4818 | if (!prefix) | |
4819 | break; | |
4820 | if (!prefix->binding) | |
4821 | continue; | |
4822 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
4823 | return NULL; | |
4824 | for (s = prefix->name; *s; s++) | |
4825 | if (!poolAppendChar(&tempPool, *s)) | |
4826 | return NULL; | |
4827 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
4828 | return NULL; | |
4829 | len = prefix->binding->uriLen; | |
4830 | if (namespaceSeparator != XML_T('\0')) | |
4831 | len--; | |
4832 | for (i = 0; i < len; i++) | |
4833 | if (!poolAppendChar(&tempPool, prefix->binding->uri[i])) | |
4834 | return NULL; | |
4835 | needSep = XML_TRUE; | |
4836 | } | |
4837 | ||
4838 | ||
4839 | hashTableIterInit(&iter, &(dtd->generalEntities)); | |
4840 | for (;;) { | |
4841 | const XML_Char *s; | |
4842 | ENTITY *e = (ENTITY *)hashTableIterNext(&iter); | |
4843 | if (!e) | |
4844 | break; | |
4845 | if (!e->open) | |
4846 | continue; | |
4847 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
4848 | return NULL; | |
4849 | for (s = e->name; *s; s++) | |
4850 | if (!poolAppendChar(&tempPool, *s)) | |
4851 | return 0; | |
4852 | needSep = XML_TRUE; | |
4853 | } | |
4854 | ||
4855 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4856 | return NULL; | |
4857 | return tempPool.start; | |
4858 | } | |
4859 | ||
4860 | static XML_Bool | |
4861 | setContext(XML_Parser parser, const XML_Char *context) | |
4862 | { | |
4863 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4864 | const XML_Char *s = context; | |
4865 | ||
4866 | while (*context != XML_T('\0')) { | |
4867 | if (*s == CONTEXT_SEP || *s == XML_T('\0')) { | |
4868 | ENTITY *e; | |
4869 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4870 | return XML_FALSE; | |
4871 | e = (ENTITY *)lookup(&dtd->generalEntities, poolStart(&tempPool), 0); | |
4872 | if (e) | |
4873 | e->open = XML_TRUE; | |
4874 | if (*s != XML_T('\0')) | |
4875 | s++; | |
4876 | context = s; | |
4877 | poolDiscard(&tempPool); | |
4878 | } | |
4879 | else if (*s == XML_T('=')) { | |
4880 | PREFIX *prefix; | |
4881 | if (poolLength(&tempPool) == 0) | |
4882 | prefix = &dtd->defaultPrefix; | |
4883 | else { | |
4884 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4885 | return XML_FALSE; | |
4886 | prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&tempPool), | |
4887 | sizeof(PREFIX)); | |
4888 | if (!prefix) | |
4889 | return XML_FALSE; | |
4890 | if (prefix->name == poolStart(&tempPool)) { | |
4891 | prefix->name = poolCopyString(&dtd->pool, prefix->name); | |
4892 | if (!prefix->name) | |
4893 | return XML_FALSE; | |
4894 | } | |
4895 | poolDiscard(&tempPool); | |
4896 | } | |
4897 | for (context = s + 1; | |
4898 | *context != CONTEXT_SEP && *context != XML_T('\0'); | |
4899 | context++) | |
4900 | if (!poolAppendChar(&tempPool, *context)) | |
4901 | return XML_FALSE; | |
4902 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4903 | return XML_FALSE; | |
4904 | if (addBinding(parser, prefix, 0, poolStart(&tempPool), | |
4905 | &inheritedBindings) != XML_ERROR_NONE) | |
4906 | return XML_FALSE; | |
4907 | poolDiscard(&tempPool); | |
4908 | if (*context != XML_T('\0')) | |
4909 | ++context; | |
4910 | s = context; | |
4911 | } | |
4912 | else { | |
4913 | if (!poolAppendChar(&tempPool, *s)) | |
4914 | return XML_FALSE; | |
4915 | s++; | |
4916 | } | |
4917 | } | |
4918 | return XML_TRUE; | |
4919 | } | |
4920 | ||
4921 | static void FASTCALL | |
4922 | normalizePublicId(XML_Char *publicId) | |
4923 | { | |
4924 | XML_Char *p = publicId; | |
4925 | XML_Char *s; | |
4926 | for (s = publicId; *s; s++) { | |
4927 | switch (*s) { | |
4928 | case 0x20: | |
4929 | case 0xD: | |
4930 | case 0xA: | |
4931 | if (p != publicId && p[-1] != 0x20) | |
4932 | *p++ = 0x20; | |
4933 | break; | |
4934 | default: | |
4935 | *p++ = *s; | |
4936 | } | |
4937 | } | |
4938 | if (p != publicId && p[-1] == 0x20) | |
4939 | --p; | |
4940 | *p = XML_T('\0'); | |
4941 | } | |
4942 | ||
4943 | static DTD * | |
4944 | dtdCreate(const XML_Memory_Handling_Suite *ms) | |
4945 | { | |
4946 | DTD *p = (DTD *)ms->malloc_fcn(sizeof(DTD)); | |
4947 | if (p == NULL) | |
4948 | return p; | |
4949 | poolInit(&(p->pool), ms); | |
4950 | #ifdef XML_DTD | |
4951 | poolInit(&(p->entityValuePool), ms); | |
4952 | #endif /* XML_DTD */ | |
4953 | hashTableInit(&(p->generalEntities), ms); | |
4954 | hashTableInit(&(p->elementTypes), ms); | |
4955 | hashTableInit(&(p->attributeIds), ms); | |
4956 | hashTableInit(&(p->prefixes), ms); | |
4957 | #ifdef XML_DTD | |
4958 | p->paramEntityRead = XML_FALSE; | |
4959 | hashTableInit(&(p->paramEntities), ms); | |
4960 | #endif /* XML_DTD */ | |
4961 | p->defaultPrefix.name = NULL; | |
4962 | p->defaultPrefix.binding = NULL; | |
4963 | ||
4964 | p->in_eldecl = XML_FALSE; | |
4965 | p->scaffIndex = NULL; | |
4966 | p->scaffold = NULL; | |
4967 | p->scaffLevel = 0; | |
4968 | p->scaffSize = 0; | |
4969 | p->scaffCount = 0; | |
4970 | p->contentStringLen = 0; | |
4971 | ||
4972 | p->keepProcessing = XML_TRUE; | |
4973 | p->hasParamEntityRefs = XML_FALSE; | |
4974 | p->standalone = XML_FALSE; | |
4975 | return p; | |
4976 | } | |
4977 | ||
4978 | static void | |
4979 | dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms) | |
4980 | { | |
4981 | HASH_TABLE_ITER iter; | |
4982 | hashTableIterInit(&iter, &(p->elementTypes)); | |
4983 | for (;;) { | |
4984 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
4985 | if (!e) | |
4986 | break; | |
4987 | if (e->allocDefaultAtts != 0) | |
4988 | ms->free_fcn(e->defaultAtts); | |
4989 | } | |
4990 | hashTableClear(&(p->generalEntities)); | |
4991 | #ifdef XML_DTD | |
4992 | p->paramEntityRead = XML_FALSE; | |
4993 | hashTableClear(&(p->paramEntities)); | |
4994 | #endif /* XML_DTD */ | |
4995 | hashTableClear(&(p->elementTypes)); | |
4996 | hashTableClear(&(p->attributeIds)); | |
4997 | hashTableClear(&(p->prefixes)); | |
4998 | poolClear(&(p->pool)); | |
4999 | #ifdef XML_DTD | |
5000 | poolClear(&(p->entityValuePool)); | |
5001 | #endif /* XML_DTD */ | |
5002 | p->defaultPrefix.name = NULL; | |
5003 | p->defaultPrefix.binding = NULL; | |
5004 | ||
5005 | p->in_eldecl = XML_FALSE; | |
5006 | if (p->scaffIndex) { | |
5007 | ms->free_fcn(p->scaffIndex); | |
5008 | p->scaffIndex = NULL; | |
5009 | } | |
5010 | if (p->scaffold) { | |
5011 | ms->free_fcn(p->scaffold); | |
5012 | p->scaffold = NULL; | |
5013 | } | |
5014 | p->scaffLevel = 0; | |
5015 | p->scaffSize = 0; | |
5016 | p->scaffCount = 0; | |
5017 | p->contentStringLen = 0; | |
5018 | ||
5019 | p->keepProcessing = XML_TRUE; | |
5020 | p->hasParamEntityRefs = XML_FALSE; | |
5021 | p->standalone = XML_FALSE; | |
5022 | } | |
5023 | ||
5024 | static void | |
5025 | dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms) | |
5026 | { | |
5027 | HASH_TABLE_ITER iter; | |
5028 | hashTableIterInit(&iter, &(p->elementTypes)); | |
5029 | for (;;) { | |
5030 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
5031 | if (!e) | |
5032 | break; | |
5033 | if (e->allocDefaultAtts != 0) | |
5034 | ms->free_fcn(e->defaultAtts); | |
5035 | } | |
5036 | hashTableDestroy(&(p->generalEntities)); | |
5037 | #ifdef XML_DTD | |
5038 | hashTableDestroy(&(p->paramEntities)); | |
5039 | #endif /* XML_DTD */ | |
5040 | hashTableDestroy(&(p->elementTypes)); | |
5041 | hashTableDestroy(&(p->attributeIds)); | |
5042 | hashTableDestroy(&(p->prefixes)); | |
5043 | poolDestroy(&(p->pool)); | |
5044 | #ifdef XML_DTD | |
5045 | poolDestroy(&(p->entityValuePool)); | |
5046 | #endif /* XML_DTD */ | |
5047 | if (isDocEntity) { | |
5048 | if (p->scaffIndex) | |
5049 | ms->free_fcn(p->scaffIndex); | |
5050 | if (p->scaffold) | |
5051 | ms->free_fcn(p->scaffold); | |
5052 | } | |
5053 | ms->free_fcn(p); | |
5054 | } | |
5055 | ||
5056 | /* Do a deep copy of the DTD. Return 0 for out of memory, non-zero otherwise. | |
5057 | The new DTD has already been initialized. | |
5058 | */ | |
5059 | static int | |
5060 | dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms) | |
5061 | { | |
5062 | HASH_TABLE_ITER iter; | |
5063 | ||
5064 | /* Copy the prefix table. */ | |
5065 | ||
5066 | hashTableIterInit(&iter, &(oldDtd->prefixes)); | |
5067 | for (;;) { | |
5068 | const XML_Char *name; | |
5069 | const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); | |
5070 | if (!oldP) | |
5071 | break; | |
5072 | name = poolCopyString(&(newDtd->pool), oldP->name); | |
5073 | if (!name) | |
5074 | return 0; | |
5075 | if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX))) | |
5076 | return 0; | |
5077 | } | |
5078 | ||
5079 | hashTableIterInit(&iter, &(oldDtd->attributeIds)); | |
5080 | ||
5081 | /* Copy the attribute id table. */ | |
5082 | ||
5083 | for (;;) { | |
5084 | ATTRIBUTE_ID *newA; | |
5085 | const XML_Char *name; | |
5086 | const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); | |
5087 | ||
5088 | if (!oldA) | |
5089 | break; | |
5090 | /* Remember to allocate the scratch byte before the name. */ | |
5091 | if (!poolAppendChar(&(newDtd->pool), XML_T('\0'))) | |
5092 | return 0; | |
5093 | name = poolCopyString(&(newDtd->pool), oldA->name); | |
5094 | if (!name) | |
5095 | return 0; | |
5096 | ++name; | |
5097 | newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name, | |
5098 | sizeof(ATTRIBUTE_ID)); | |
5099 | if (!newA) | |
5100 | return 0; | |
5101 | newA->maybeTokenized = oldA->maybeTokenized; | |
5102 | if (oldA->prefix) { | |
5103 | newA->xmlns = oldA->xmlns; | |
5104 | if (oldA->prefix == &oldDtd->defaultPrefix) | |
5105 | newA->prefix = &newDtd->defaultPrefix; | |
5106 | else | |
5107 | newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes), | |
5108 | oldA->prefix->name, 0); | |
5109 | } | |
5110 | } | |
5111 | ||
5112 | /* Copy the element type table. */ | |
5113 | ||
5114 | hashTableIterInit(&iter, &(oldDtd->elementTypes)); | |
5115 | ||
5116 | for (;;) { | |
5117 | int i; | |
5118 | ELEMENT_TYPE *newE; | |
5119 | const XML_Char *name; | |
5120 | const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
5121 | if (!oldE) | |
5122 | break; | |
5123 | name = poolCopyString(&(newDtd->pool), oldE->name); | |
5124 | if (!name) | |
5125 | return 0; | |
5126 | newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name, | |
5127 | sizeof(ELEMENT_TYPE)); | |
5128 | if (!newE) | |
5129 | return 0; | |
5130 | if (oldE->nDefaultAtts) { | |
5131 | newE->defaultAtts = (DEFAULT_ATTRIBUTE *) | |
5132 | ms->malloc_fcn(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); | |
5133 | if (!newE->defaultAtts) { | |
5134 | ms->free_fcn(newE); | |
5135 | return 0; | |
5136 | } | |
5137 | } | |
5138 | if (oldE->idAtt) | |
5139 | newE->idAtt = (ATTRIBUTE_ID *) | |
5140 | lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0); | |
5141 | newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; | |
5142 | if (oldE->prefix) | |
5143 | newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes), | |
5144 | oldE->prefix->name, 0); | |
5145 | for (i = 0; i < newE->nDefaultAtts; i++) { | |
5146 | newE->defaultAtts[i].id = (ATTRIBUTE_ID *) | |
5147 | lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); | |
5148 | newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; | |
5149 | if (oldE->defaultAtts[i].value) { | |
5150 | newE->defaultAtts[i].value | |
5151 | = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); | |
5152 | if (!newE->defaultAtts[i].value) | |
5153 | return 0; | |
5154 | } | |
5155 | else | |
5156 | newE->defaultAtts[i].value = NULL; | |
5157 | } | |
5158 | } | |
5159 | ||
5160 | /* Copy the entity tables. */ | |
5161 | if (!copyEntityTable(&(newDtd->generalEntities), | |
5162 | &(newDtd->pool), | |
5163 | &(oldDtd->generalEntities))) | |
5164 | return 0; | |
5165 | ||
5166 | #ifdef XML_DTD | |
5167 | if (!copyEntityTable(&(newDtd->paramEntities), | |
5168 | &(newDtd->pool), | |
5169 | &(oldDtd->paramEntities))) | |
5170 | return 0; | |
5171 | newDtd->paramEntityRead = oldDtd->paramEntityRead; | |
5172 | #endif /* XML_DTD */ | |
5173 | ||
5174 | newDtd->keepProcessing = oldDtd->keepProcessing; | |
5175 | newDtd->hasParamEntityRefs = oldDtd->hasParamEntityRefs; | |
5176 | newDtd->standalone = oldDtd->standalone; | |
5177 | ||
5178 | /* Don't want deep copying for scaffolding */ | |
5179 | newDtd->in_eldecl = oldDtd->in_eldecl; | |
5180 | newDtd->scaffold = oldDtd->scaffold; | |
5181 | newDtd->contentStringLen = oldDtd->contentStringLen; | |
5182 | newDtd->scaffSize = oldDtd->scaffSize; | |
5183 | newDtd->scaffLevel = oldDtd->scaffLevel; | |
5184 | newDtd->scaffIndex = oldDtd->scaffIndex; | |
5185 | ||
5186 | return 1; | |
5187 | } /* End dtdCopy */ | |
5188 | ||
5189 | static int | |
5190 | copyEntityTable(HASH_TABLE *newTable, | |
5191 | STRING_POOL *newPool, | |
5192 | const HASH_TABLE *oldTable) | |
5193 | { | |
5194 | HASH_TABLE_ITER iter; | |
5195 | const XML_Char *cachedOldBase = NULL; | |
5196 | const XML_Char *cachedNewBase = NULL; | |
5197 | ||
5198 | hashTableIterInit(&iter, oldTable); | |
5199 | ||
5200 | for (;;) { | |
5201 | ENTITY *newE; | |
5202 | const XML_Char *name; | |
5203 | const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); | |
5204 | if (!oldE) | |
5205 | break; | |
5206 | name = poolCopyString(newPool, oldE->name); | |
5207 | if (!name) | |
5208 | return 0; | |
5209 | newE = (ENTITY *)lookup(newTable, name, sizeof(ENTITY)); | |
5210 | if (!newE) | |
5211 | return 0; | |
5212 | if (oldE->systemId) { | |
5213 | const XML_Char *tem = poolCopyString(newPool, oldE->systemId); | |
5214 | if (!tem) | |
5215 | return 0; | |
5216 | newE->systemId = tem; | |
5217 | if (oldE->base) { | |
5218 | if (oldE->base == cachedOldBase) | |
5219 | newE->base = cachedNewBase; | |
5220 | else { | |
5221 | cachedOldBase = oldE->base; | |
5222 | tem = poolCopyString(newPool, cachedOldBase); | |
5223 | if (!tem) | |
5224 | return 0; | |
5225 | cachedNewBase = newE->base = tem; | |
5226 | } | |
5227 | } | |
5228 | if (oldE->publicId) { | |
5229 | tem = poolCopyString(newPool, oldE->publicId); | |
5230 | if (!tem) | |
5231 | return 0; | |
5232 | newE->publicId = tem; | |
5233 | } | |
5234 | } | |
5235 | else { | |
5236 | const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr, | |
5237 | oldE->textLen); | |
5238 | if (!tem) | |
5239 | return 0; | |
5240 | newE->textPtr = tem; | |
5241 | newE->textLen = oldE->textLen; | |
5242 | } | |
5243 | if (oldE->notation) { | |
5244 | const XML_Char *tem = poolCopyString(newPool, oldE->notation); | |
5245 | if (!tem) | |
5246 | return 0; | |
5247 | newE->notation = tem; | |
5248 | } | |
5249 | newE->is_param = oldE->is_param; | |
5250 | newE->is_internal = oldE->is_internal; | |
5251 | } | |
5252 | return 1; | |
5253 | } | |
5254 | ||
5255 | #define INIT_SIZE 64 | |
5256 | ||
5257 | static int FASTCALL | |
5258 | keyeq(KEY s1, KEY s2) | |
5259 | { | |
5260 | for (; *s1 == *s2; s1++, s2++) | |
5261 | if (*s1 == 0) | |
5262 | return 1; | |
5263 | return 0; | |
5264 | } | |
5265 | ||
5266 | static unsigned long FASTCALL | |
5267 | hash(KEY s) | |
5268 | { | |
5269 | unsigned long h = 0; | |
5270 | while (*s) | |
5271 | h = (h << 5) + h + (unsigned char)*s++; | |
5272 | return h; | |
5273 | } | |
5274 | ||
5275 | static NAMED * | |
5276 | lookup(HASH_TABLE *table, KEY name, size_t createSize) | |
5277 | { | |
5278 | size_t i; | |
5279 | if (table->size == 0) { | |
5280 | size_t tsize; | |
5281 | ||
5282 | if (!createSize) | |
5283 | return NULL; | |
5284 | tsize = INIT_SIZE * sizeof(NAMED *); | |
5285 | table->v = (NAMED **)table->mem->malloc_fcn(tsize); | |
5286 | if (!table->v) | |
5287 | return NULL; | |
5288 | memset(table->v, 0, tsize); | |
5289 | table->size = INIT_SIZE; | |
5290 | table->usedLim = INIT_SIZE / 2; | |
5291 | i = hash(name) & (table->size - 1); | |
5292 | } | |
5293 | else { | |
5294 | unsigned long h = hash(name); | |
5295 | for (i = h & (table->size - 1); | |
5296 | table->v[i]; | |
5297 | i == 0 ? i = table->size - 1 : --i) { | |
5298 | if (keyeq(name, table->v[i]->name)) | |
5299 | return table->v[i]; | |
5300 | } | |
5301 | if (!createSize) | |
5302 | return NULL; | |
5303 | if (table->used == table->usedLim) { | |
5304 | /* check for overflow */ | |
5305 | size_t newSize = table->size * 2; | |
5306 | size_t tsize = newSize * sizeof(NAMED *); | |
5307 | NAMED **newV = (NAMED **)table->mem->malloc_fcn(tsize); | |
5308 | if (!newV) | |
5309 | return NULL; | |
5310 | memset(newV, 0, tsize); | |
5311 | for (i = 0; i < table->size; i++) | |
5312 | if (table->v[i]) { | |
5313 | size_t j; | |
5314 | for (j = hash(table->v[i]->name) & (newSize - 1); | |
5315 | newV[j]; | |
5316 | j == 0 ? j = newSize - 1 : --j) | |
5317 | ; | |
5318 | newV[j] = table->v[i]; | |
5319 | } | |
5320 | table->mem->free_fcn(table->v); | |
5321 | table->v = newV; | |
5322 | table->size = newSize; | |
5323 | table->usedLim = newSize/2; | |
5324 | for (i = h & (table->size - 1); | |
5325 | table->v[i]; | |
5326 | i == 0 ? i = table->size - 1 : --i) | |
5327 | ; | |
5328 | } | |
5329 | } | |
5330 | table->v[i] = (NAMED *)table->mem->malloc_fcn(createSize); | |
5331 | if (!table->v[i]) | |
5332 | return NULL; | |
5333 | memset(table->v[i], 0, createSize); | |
5334 | table->v[i]->name = name; | |
5335 | (table->used)++; | |
5336 | return table->v[i]; | |
5337 | } | |
5338 | ||
5339 | static void FASTCALL | |
5340 | hashTableClear(HASH_TABLE *table) | |
5341 | { | |
5342 | size_t i; | |
5343 | for (i = 0; i < table->size; i++) { | |
5344 | NAMED *p = table->v[i]; | |
5345 | if (p) { | |
5346 | table->mem->free_fcn(p); | |
5347 | table->v[i] = NULL; | |
5348 | } | |
5349 | } | |
5350 | table->usedLim = table->size / 2; | |
5351 | table->used = 0; | |
5352 | } | |
5353 | ||
5354 | static void FASTCALL | |
5355 | hashTableDestroy(HASH_TABLE *table) | |
5356 | { | |
5357 | size_t i; | |
5358 | for (i = 0; i < table->size; i++) { | |
5359 | NAMED *p = table->v[i]; | |
5360 | if (p) | |
5361 | table->mem->free_fcn(p); | |
5362 | } | |
5363 | if (table->v) | |
5364 | table->mem->free_fcn(table->v); | |
5365 | } | |
5366 | ||
5367 | static void FASTCALL | |
5368 | hashTableInit(HASH_TABLE *p, const XML_Memory_Handling_Suite *ms) | |
5369 | { | |
5370 | p->size = 0; | |
5371 | p->usedLim = 0; | |
5372 | p->used = 0; | |
5373 | p->v = NULL; | |
5374 | p->mem = ms; | |
5375 | } | |
5376 | ||
5377 | static void FASTCALL | |
5378 | hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table) | |
5379 | { | |
5380 | iter->p = table->v; | |
5381 | iter->end = iter->p + table->size; | |
5382 | } | |
5383 | ||
5384 | static NAMED * FASTCALL | |
5385 | hashTableIterNext(HASH_TABLE_ITER *iter) | |
5386 | { | |
5387 | while (iter->p != iter->end) { | |
5388 | NAMED *tem = *(iter->p)++; | |
5389 | if (tem) | |
5390 | return tem; | |
5391 | } | |
5392 | return NULL; | |
5393 | } | |
5394 | ||
5395 | static void FASTCALL | |
5396 | poolInit(STRING_POOL *pool, const XML_Memory_Handling_Suite *ms) | |
5397 | { | |
5398 | pool->blocks = NULL; | |
5399 | pool->freeBlocks = NULL; | |
5400 | pool->start = NULL; | |
5401 | pool->ptr = NULL; | |
5402 | pool->end = NULL; | |
5403 | pool->mem = ms; | |
5404 | } | |
5405 | ||
5406 | static void FASTCALL | |
5407 | poolClear(STRING_POOL *pool) | |
5408 | { | |
5409 | if (!pool->freeBlocks) | |
5410 | pool->freeBlocks = pool->blocks; | |
5411 | else { | |
5412 | BLOCK *p = pool->blocks; | |
5413 | while (p) { | |
5414 | BLOCK *tem = p->next; | |
5415 | p->next = pool->freeBlocks; | |
5416 | pool->freeBlocks = p; | |
5417 | p = tem; | |
5418 | } | |
5419 | } | |
5420 | pool->blocks = NULL; | |
5421 | pool->start = NULL; | |
5422 | pool->ptr = NULL; | |
5423 | pool->end = NULL; | |
5424 | } | |
5425 | ||
5426 | static void FASTCALL | |
5427 | poolDestroy(STRING_POOL *pool) | |
5428 | { | |
5429 | BLOCK *p = pool->blocks; | |
5430 | while (p) { | |
5431 | BLOCK *tem = p->next; | |
5432 | pool->mem->free_fcn(p); | |
5433 | p = tem; | |
5434 | } | |
5435 | p = pool->freeBlocks; | |
5436 | while (p) { | |
5437 | BLOCK *tem = p->next; | |
5438 | pool->mem->free_fcn(p); | |
5439 | p = tem; | |
5440 | } | |
5441 | } | |
5442 | ||
5443 | static XML_Char * | |
5444 | poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
5445 | const char *ptr, const char *end) | |
5446 | { | |
5447 | if (!pool->ptr && !poolGrow(pool)) | |
5448 | return NULL; | |
5449 | for (;;) { | |
5450 | XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); | |
5451 | if (ptr == end) | |
5452 | break; | |
5453 | if (!poolGrow(pool)) | |
5454 | return NULL; | |
5455 | } | |
5456 | return pool->start; | |
5457 | } | |
5458 | ||
5459 | static const XML_Char * FASTCALL | |
5460 | poolCopyString(STRING_POOL *pool, const XML_Char *s) | |
5461 | { | |
5462 | do { | |
5463 | if (!poolAppendChar(pool, *s)) | |
5464 | return NULL; | |
5465 | } while (*s++); | |
5466 | s = pool->start; | |
5467 | poolFinish(pool); | |
5468 | return s; | |
5469 | } | |
5470 | ||
5471 | static const XML_Char * | |
5472 | poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) | |
5473 | { | |
5474 | if (!pool->ptr && !poolGrow(pool)) | |
5475 | return NULL; | |
5476 | for (; n > 0; --n, s++) { | |
5477 | if (!poolAppendChar(pool, *s)) | |
5478 | return NULL; | |
5479 | } | |
5480 | s = pool->start; | |
5481 | poolFinish(pool); | |
5482 | return s; | |
5483 | } | |
5484 | ||
5485 | static const XML_Char * FASTCALL | |
5486 | poolAppendString(STRING_POOL *pool, const XML_Char *s) | |
5487 | { | |
5488 | while (*s) { | |
5489 | if (!poolAppendChar(pool, *s)) | |
5490 | return NULL; | |
5491 | s++; | |
5492 | } | |
5493 | return pool->start; | |
5494 | } | |
5495 | ||
5496 | static XML_Char * | |
5497 | poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
5498 | const char *ptr, const char *end) | |
5499 | { | |
5500 | if (!poolAppend(pool, enc, ptr, end)) | |
5501 | return NULL; | |
5502 | if (pool->ptr == pool->end && !poolGrow(pool)) | |
5503 | return NULL; | |
5504 | *(pool->ptr)++ = 0; | |
5505 | return pool->start; | |
5506 | } | |
5507 | ||
5508 | static XML_Bool FASTCALL | |
5509 | poolGrow(STRING_POOL *pool) | |
5510 | { | |
5511 | if (pool->freeBlocks) { | |
5512 | if (pool->start == 0) { | |
5513 | pool->blocks = pool->freeBlocks; | |
5514 | pool->freeBlocks = pool->freeBlocks->next; | |
5515 | pool->blocks->next = NULL; | |
5516 | pool->start = pool->blocks->s; | |
5517 | pool->end = pool->start + pool->blocks->size; | |
5518 | pool->ptr = pool->start; | |
5519 | return XML_TRUE; | |
5520 | } | |
5521 | if (pool->end - pool->start < pool->freeBlocks->size) { | |
5522 | BLOCK *tem = pool->freeBlocks->next; | |
5523 | pool->freeBlocks->next = pool->blocks; | |
5524 | pool->blocks = pool->freeBlocks; | |
5525 | pool->freeBlocks = tem; | |
5526 | memcpy(pool->blocks->s, pool->start, | |
5527 | (pool->end - pool->start) * sizeof(XML_Char)); | |
5528 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
5529 | pool->start = pool->blocks->s; | |
5530 | pool->end = pool->start + pool->blocks->size; | |
5531 | return XML_TRUE; | |
5532 | } | |
5533 | } | |
5534 | if (pool->blocks && pool->start == pool->blocks->s) { | |
5535 | int blockSize = (pool->end - pool->start)*2; | |
5536 | pool->blocks = (BLOCK *) | |
5537 | pool->mem->realloc_fcn(pool->blocks, | |
5538 | (offsetof(BLOCK, s) | |
5539 | + blockSize * sizeof(XML_Char))); | |
5540 | if (pool->blocks == NULL) | |
5541 | return XML_FALSE; | |
5542 | pool->blocks->size = blockSize; | |
5543 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
5544 | pool->start = pool->blocks->s; | |
5545 | pool->end = pool->start + blockSize; | |
5546 | } | |
5547 | else { | |
5548 | BLOCK *tem; | |
5549 | int blockSize = pool->end - pool->start; | |
5550 | if (blockSize < INIT_BLOCK_SIZE) | |
5551 | blockSize = INIT_BLOCK_SIZE; | |
5552 | else | |
5553 | blockSize *= 2; | |
5554 | tem = (BLOCK *)pool->mem->malloc_fcn(offsetof(BLOCK, s) | |
5555 | + blockSize * sizeof(XML_Char)); | |
5556 | if (!tem) | |
5557 | return XML_FALSE; | |
5558 | tem->size = blockSize; | |
5559 | tem->next = pool->blocks; | |
5560 | pool->blocks = tem; | |
5561 | if (pool->ptr != pool->start) | |
5562 | memcpy(tem->s, pool->start, | |
5563 | (pool->ptr - pool->start) * sizeof(XML_Char)); | |
5564 | pool->ptr = tem->s + (pool->ptr - pool->start); | |
5565 | pool->start = tem->s; | |
5566 | pool->end = tem->s + blockSize; | |
5567 | } | |
5568 | return XML_TRUE; | |
5569 | } | |
5570 | ||
5571 | static int FASTCALL | |
5572 | nextScaffoldPart(XML_Parser parser) | |
5573 | { | |
5574 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5575 | CONTENT_SCAFFOLD * me; | |
5576 | int next; | |
5577 | ||
5578 | if (!dtd->scaffIndex) { | |
5579 | dtd->scaffIndex = (int *)MALLOC(groupSize * sizeof(int)); | |
5580 | if (!dtd->scaffIndex) | |
5581 | return -1; | |
5582 | dtd->scaffIndex[0] = 0; | |
5583 | } | |
5584 | ||
5585 | if (dtd->scaffCount >= dtd->scaffSize) { | |
5586 | CONTENT_SCAFFOLD *temp; | |
5587 | if (dtd->scaffold) { | |
5588 | temp = (CONTENT_SCAFFOLD *) | |
5589 | REALLOC(dtd->scaffold, dtd->scaffSize * 2 * sizeof(CONTENT_SCAFFOLD)); | |
5590 | if (temp == NULL) | |
5591 | return -1; | |
5592 | dtd->scaffSize *= 2; | |
5593 | } | |
5594 | else { | |
5595 | temp = (CONTENT_SCAFFOLD *)MALLOC(INIT_SCAFFOLD_ELEMENTS | |
5596 | * sizeof(CONTENT_SCAFFOLD)); | |
5597 | if (temp == NULL) | |
5598 | return -1; | |
5599 | dtd->scaffSize = INIT_SCAFFOLD_ELEMENTS; | |
5600 | } | |
5601 | dtd->scaffold = temp; | |
5602 | } | |
5603 | next = dtd->scaffCount++; | |
5604 | me = &dtd->scaffold[next]; | |
5605 | if (dtd->scaffLevel) { | |
5606 | CONTENT_SCAFFOLD *parent = &dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel-1]]; | |
5607 | if (parent->lastchild) { | |
5608 | dtd->scaffold[parent->lastchild].nextsib = next; | |
5609 | } | |
5610 | if (!parent->childcnt) | |
5611 | parent->firstchild = next; | |
5612 | parent->lastchild = next; | |
5613 | parent->childcnt++; | |
5614 | } | |
5615 | me->firstchild = me->lastchild = me->childcnt = me->nextsib = 0; | |
5616 | return next; | |
5617 | } | |
5618 | ||
5619 | static void | |
5620 | build_node(XML_Parser parser, | |
5621 | int src_node, | |
5622 | XML_Content *dest, | |
5623 | XML_Content **contpos, | |
5624 | XML_Char **strpos) | |
5625 | { | |
5626 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5627 | dest->type = dtd->scaffold[src_node].type; | |
5628 | dest->quant = dtd->scaffold[src_node].quant; | |
5629 | if (dest->type == XML_CTYPE_NAME) { | |
5630 | const XML_Char *src; | |
5631 | dest->name = *strpos; | |
5632 | src = dtd->scaffold[src_node].name; | |
5633 | for (;;) { | |
5634 | *(*strpos)++ = *src; | |
5635 | if (!*src) | |
5636 | break; | |
5637 | src++; | |
5638 | } | |
5639 | dest->numchildren = 0; | |
5640 | dest->children = NULL; | |
5641 | } | |
5642 | else { | |
5643 | unsigned int i; | |
5644 | int cn; | |
5645 | dest->numchildren = dtd->scaffold[src_node].childcnt; | |
5646 | dest->children = *contpos; | |
5647 | *contpos += dest->numchildren; | |
5648 | for (i = 0, cn = dtd->scaffold[src_node].firstchild; | |
5649 | i < dest->numchildren; | |
5650 | i++, cn = dtd->scaffold[cn].nextsib) { | |
5651 | build_node(parser, cn, &(dest->children[i]), contpos, strpos); | |
5652 | } | |
5653 | dest->name = NULL; | |
5654 | } | |
5655 | } | |
5656 | ||
5657 | static XML_Content * | |
5658 | build_model (XML_Parser parser) | |
5659 | { | |
5660 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5661 | XML_Content *ret; | |
5662 | XML_Content *cpos; | |
5663 | XML_Char * str; | |
5664 | int allocsize = (dtd->scaffCount * sizeof(XML_Content) | |
5665 | + (dtd->contentStringLen * sizeof(XML_Char))); | |
5666 | ||
5667 | ret = (XML_Content *)MALLOC(allocsize); | |
5668 | if (!ret) | |
5669 | return NULL; | |
5670 | ||
5671 | str = (XML_Char *) (&ret[dtd->scaffCount]); | |
5672 | cpos = &ret[1]; | |
5673 | ||
5674 | build_node(parser, 0, ret, &cpos, &str); | |
5675 | return ret; | |
5676 | } | |
5677 | ||
5678 | static ELEMENT_TYPE * | |
5679 | getElementType(XML_Parser parser, | |
5680 | const ENCODING *enc, | |
5681 | const char *ptr, | |
5682 | const char *end) | |
5683 | { | |
5684 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5685 | const XML_Char *name = poolStoreString(&dtd->pool, enc, ptr, end); | |
5686 | ELEMENT_TYPE *ret; | |
5687 | ||
5688 | if (!name) | |
5689 | return NULL; | |
5690 | ret = (ELEMENT_TYPE *) lookup(&dtd->elementTypes, name, sizeof(ELEMENT_TYPE)); | |
5691 | if (!ret) | |
5692 | return NULL; | |
5693 | if (ret->name != name) | |
5694 | poolDiscard(&dtd->pool); | |
5695 | else { | |
5696 | poolFinish(&dtd->pool); | |
5697 | if (!setElementTypePrefix(parser, ret)) | |
5698 | return NULL; | |
5699 | } | |
5700 | return ret; | |
5701 | } |