]>
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(OS2_32) | |
14 | ||
15 | #include "os2config.h" | |
16 | #include "expat.h" | |
17 | ||
18 | #elif defined(MACOS_CLASSIC) | |
19 | ||
20 | #include "macconfig.h" | |
21 | #include "expat.h" | |
22 | ||
23 | #else | |
24 | ||
25 | #include "expat_config.h" | |
26 | ||
27 | #ifdef __declspec | |
28 | #define XMLPARSEAPI(type) type __cdecl | |
29 | #endif | |
30 | ||
31 | #include "expat.h" | |
32 | ||
33 | #ifdef __declspec | |
34 | #undef XMLPARSEAPI | |
35 | #endif | |
36 | #endif /* ndef COMPILED_FROM_DSP */ | |
37 | ||
38 | #ifdef XML_UNICODE | |
39 | #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX | |
40 | #define XmlConvert XmlUtf16Convert | |
41 | #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding | |
42 | #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS | |
43 | #define XmlEncode XmlUtf16Encode | |
44 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1)) | |
45 | typedef unsigned short ICHAR; | |
46 | #else | |
47 | #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX | |
48 | #define XmlConvert XmlUtf8Convert | |
49 | #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding | |
50 | #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS | |
51 | #define XmlEncode XmlUtf8Encode | |
52 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf8) | |
53 | typedef char ICHAR; | |
54 | #endif | |
55 | ||
56 | ||
57 | #ifndef XML_NS | |
58 | ||
59 | #define XmlInitEncodingNS XmlInitEncoding | |
60 | #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding | |
61 | #undef XmlGetInternalEncodingNS | |
62 | #define XmlGetInternalEncodingNS XmlGetInternalEncoding | |
63 | #define XmlParseXmlDeclNS XmlParseXmlDecl | |
64 | ||
65 | #endif | |
66 | ||
67 | #ifdef XML_UNICODE | |
68 | ||
69 | #ifdef XML_UNICODE_WCHAR_T | |
70 | #define XML_T(x) (const wchar_t)x | |
71 | #define XML_L(x) L ## x | |
72 | #else | |
73 | #define XML_T(x) (const unsigned short)x | |
74 | #define XML_L(x) x | |
75 | #endif | |
76 | ||
77 | #else | |
78 | ||
79 | #define XML_T(x) x | |
80 | #define XML_L(x) x | |
81 | ||
82 | #endif | |
83 | ||
84 | /* Round up n to be a multiple of sz, where sz is a power of 2. */ | |
85 | #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1)) | |
86 | ||
87 | /* Handle the case where memmove() doesn't exist. */ | |
88 | #ifndef HAVE_MEMMOVE | |
89 | #ifdef HAVE_BCOPY | |
90 | #define memmove(d,s,l) bcopy((s),(d),(l)) | |
91 | #else | |
92 | #error memmove does not exist on this platform, nor is a substitute available | |
93 | #endif /* HAVE_BCOPY */ | |
94 | #endif /* HAVE_MEMMOVE */ | |
95 | ||
96 | #include "internal.h" | |
97 | #include "xmltok.h" | |
98 | #include "xmlrole.h" | |
99 | ||
100 | typedef const XML_Char *KEY; | |
101 | ||
102 | typedef struct { | |
103 | KEY name; | |
104 | } NAMED; | |
105 | ||
106 | typedef struct { | |
107 | NAMED **v; | |
108 | size_t size; | |
109 | size_t used; | |
110 | size_t usedLim; | |
111 | const XML_Memory_Handling_Suite *mem; | |
112 | } HASH_TABLE; | |
113 | ||
114 | typedef struct { | |
115 | NAMED **p; | |
116 | NAMED **end; | |
117 | } HASH_TABLE_ITER; | |
118 | ||
119 | #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ | |
120 | #define INIT_DATA_BUF_SIZE 1024 | |
121 | #define INIT_ATTS_SIZE 16 | |
122 | #define INIT_BLOCK_SIZE 1024 | |
123 | #define INIT_BUFFER_SIZE 1024 | |
124 | ||
125 | #define EXPAND_SPARE 24 | |
126 | ||
127 | typedef struct binding { | |
128 | struct prefix *prefix; | |
129 | struct binding *nextTagBinding; | |
130 | struct binding *prevPrefixBinding; | |
131 | const struct attribute_id *attId; | |
132 | XML_Char *uri; | |
133 | int uriLen; | |
134 | int uriAlloc; | |
135 | } BINDING; | |
136 | ||
137 | typedef struct prefix { | |
138 | const XML_Char *name; | |
139 | BINDING *binding; | |
140 | } PREFIX; | |
141 | ||
142 | typedef struct { | |
143 | const XML_Char *str; | |
144 | const XML_Char *localPart; | |
145 | const XML_Char *prefix; | |
146 | int strLen; | |
147 | int uriLen; | |
148 | int prefixLen; | |
149 | } TAG_NAME; | |
150 | ||
151 | /* TAG represents an open element. | |
152 | The name of the element is stored in both the document and API | |
153 | encodings. The memory buffer 'buf' is a separately-allocated | |
154 | memory area which stores the name. During the XML_Parse()/ | |
155 | XMLParseBuffer() when the element is open, the memory for the 'raw' | |
156 | version of the name (in the document encoding) is shared with the | |
157 | document buffer. If the element is open across calls to | |
158 | XML_Parse()/XML_ParseBuffer(), the buffer is re-allocated to | |
159 | contain the 'raw' name as well. | |
160 | ||
161 | A parser re-uses these structures, maintaining a list of allocated | |
162 | TAG objects in a free list. | |
163 | */ | |
164 | typedef struct tag { | |
165 | struct tag *parent; /* parent of this element */ | |
166 | const char *rawName; /* tagName in the original encoding */ | |
167 | int rawNameLength; | |
168 | TAG_NAME name; /* tagName in the API encoding */ | |
169 | char *buf; /* buffer for name components */ | |
170 | char *bufEnd; /* end of the buffer */ | |
171 | BINDING *bindings; | |
172 | } TAG; | |
173 | ||
174 | typedef struct { | |
175 | const XML_Char *name; | |
176 | const XML_Char *textPtr; | |
177 | int textLen; | |
178 | const XML_Char *systemId; | |
179 | const XML_Char *base; | |
180 | const XML_Char *publicId; | |
181 | const XML_Char *notation; | |
182 | XML_Bool open; | |
183 | XML_Bool is_param; | |
184 | XML_Bool is_internal; /* true if declared in internal subset outside PE */ | |
185 | } ENTITY; | |
186 | ||
187 | typedef struct { | |
188 | enum XML_Content_Type type; | |
189 | enum XML_Content_Quant quant; | |
190 | const XML_Char * name; | |
191 | int firstchild; | |
192 | int lastchild; | |
193 | int childcnt; | |
194 | int nextsib; | |
195 | } CONTENT_SCAFFOLD; | |
196 | ||
197 | #define INIT_SCAFFOLD_ELEMENTS 32 | |
198 | ||
199 | typedef struct block { | |
200 | struct block *next; | |
201 | int size; | |
202 | XML_Char s[1]; | |
203 | } BLOCK; | |
204 | ||
205 | typedef struct { | |
206 | BLOCK *blocks; | |
207 | BLOCK *freeBlocks; | |
208 | const XML_Char *end; | |
209 | XML_Char *ptr; | |
210 | XML_Char *start; | |
211 | const XML_Memory_Handling_Suite *mem; | |
212 | } STRING_POOL; | |
213 | ||
214 | /* The XML_Char before the name is used to determine whether | |
215 | an attribute has been specified. */ | |
216 | typedef struct attribute_id { | |
217 | XML_Char *name; | |
218 | PREFIX *prefix; | |
219 | XML_Bool maybeTokenized; | |
220 | XML_Bool xmlns; | |
221 | } ATTRIBUTE_ID; | |
222 | ||
223 | typedef struct { | |
224 | const ATTRIBUTE_ID *id; | |
225 | XML_Bool isCdata; | |
226 | const XML_Char *value; | |
227 | } DEFAULT_ATTRIBUTE; | |
228 | ||
229 | typedef struct { | |
230 | const XML_Char *name; | |
231 | PREFIX *prefix; | |
232 | const ATTRIBUTE_ID *idAtt; | |
233 | int nDefaultAtts; | |
234 | int allocDefaultAtts; | |
235 | DEFAULT_ATTRIBUTE *defaultAtts; | |
236 | } ELEMENT_TYPE; | |
237 | ||
238 | typedef struct { | |
239 | HASH_TABLE generalEntities; | |
240 | HASH_TABLE elementTypes; | |
241 | HASH_TABLE attributeIds; | |
242 | HASH_TABLE prefixes; | |
243 | STRING_POOL pool; | |
244 | STRING_POOL entityValuePool; | |
245 | /* false once a parameter entity reference has been skipped */ | |
246 | XML_Bool keepProcessing; | |
247 | /* true once an internal or external PE reference has been encountered; | |
248 | this includes the reference to an external subset */ | |
249 | XML_Bool hasParamEntityRefs; | |
250 | XML_Bool standalone; | |
251 | #ifdef XML_DTD | |
252 | /* indicates if external PE has been read */ | |
253 | XML_Bool paramEntityRead; | |
254 | HASH_TABLE paramEntities; | |
255 | #endif /* XML_DTD */ | |
256 | PREFIX defaultPrefix; | |
257 | /* === scaffolding for building content model === */ | |
258 | XML_Bool in_eldecl; | |
259 | CONTENT_SCAFFOLD *scaffold; | |
260 | unsigned contentStringLen; | |
261 | unsigned scaffSize; | |
262 | unsigned scaffCount; | |
263 | int scaffLevel; | |
264 | int *scaffIndex; | |
265 | } DTD; | |
266 | ||
267 | typedef struct open_internal_entity { | |
268 | const char *internalEventPtr; | |
269 | const char *internalEventEndPtr; | |
270 | struct open_internal_entity *next; | |
271 | ENTITY *entity; | |
272 | } OPEN_INTERNAL_ENTITY; | |
273 | ||
274 | typedef enum XML_Error PTRCALL Processor(XML_Parser parser, | |
275 | const char *start, | |
276 | const char *end, | |
277 | const char **endPtr); | |
278 | ||
279 | static Processor prologProcessor; | |
280 | static Processor prologInitProcessor; | |
281 | static Processor contentProcessor; | |
282 | static Processor cdataSectionProcessor; | |
283 | #ifdef XML_DTD | |
284 | static Processor ignoreSectionProcessor; | |
285 | static Processor externalParEntProcessor; | |
286 | static Processor externalParEntInitProcessor; | |
287 | static Processor entityValueProcessor; | |
288 | static Processor entityValueInitProcessor; | |
289 | #endif /* XML_DTD */ | |
290 | static Processor epilogProcessor; | |
291 | static Processor errorProcessor; | |
292 | static Processor externalEntityInitProcessor; | |
293 | static Processor externalEntityInitProcessor2; | |
294 | static Processor externalEntityInitProcessor3; | |
295 | static Processor externalEntityContentProcessor; | |
296 | ||
297 | static enum XML_Error | |
298 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName); | |
299 | static enum XML_Error | |
300 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
301 | const char *, const char *); | |
302 | static enum XML_Error | |
303 | initializeEncoding(XML_Parser parser); | |
304 | static enum XML_Error | |
305 | doProlog(XML_Parser parser, const ENCODING *enc, const char *s, | |
306 | const char *end, int tok, const char *next, const char **nextPtr); | |
307 | static enum XML_Error | |
308 | processInternalParamEntity(XML_Parser parser, ENTITY *entity); | |
309 | static enum XML_Error | |
310 | doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, | |
311 | const char *start, const char *end, const char **endPtr); | |
312 | static enum XML_Error | |
313 | doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, | |
314 | const char *end, const char **nextPtr); | |
315 | #ifdef XML_DTD | |
316 | static enum XML_Error | |
317 | doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr, | |
318 | const char *end, const char **nextPtr); | |
319 | #endif /* XML_DTD */ | |
320 | ||
321 | static enum XML_Error | |
322 | storeAtts(XML_Parser parser, const ENCODING *, const char *s, | |
323 | TAG_NAME *tagNamePtr, BINDING **bindingsPtr); | |
324 | static enum XML_Error | |
325 | addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, | |
326 | const XML_Char *uri, BINDING **bindingsPtr); | |
327 | static int | |
328 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, | |
329 | XML_Bool isCdata, XML_Bool isId, const XML_Char *dfltValue, | |
330 | XML_Parser parser); | |
331 | static enum XML_Error | |
332 | storeAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata, | |
333 | const char *, const char *, STRING_POOL *); | |
334 | static enum XML_Error | |
335 | appendAttributeValue(XML_Parser parser, const ENCODING *, XML_Bool isCdata, | |
336 | const char *, const char *, STRING_POOL *); | |
337 | static ATTRIBUTE_ID * | |
338 | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, | |
339 | const char *end); | |
340 | static int | |
341 | setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); | |
342 | static enum XML_Error | |
343 | storeEntityValue(XML_Parser parser, const ENCODING *enc, const char *start, | |
344 | const char *end); | |
345 | static int | |
346 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, | |
347 | const char *start, const char *end); | |
348 | static int | |
349 | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, | |
350 | const char *end); | |
351 | static void | |
352 | reportDefault(XML_Parser parser, const ENCODING *enc, const char *start, | |
353 | const char *end); | |
354 | ||
355 | static const XML_Char * getContext(XML_Parser parser); | |
356 | static XML_Bool | |
357 | setContext(XML_Parser parser, const XML_Char *context); | |
358 | ||
359 | static void FASTCALL normalizePublicId(XML_Char *s); | |
360 | ||
361 | static DTD * dtdCreate(const XML_Memory_Handling_Suite *ms); | |
362 | /* do not call if parentParser != NULL */ | |
363 | static void dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms); | |
364 | static void | |
365 | dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms); | |
366 | static int | |
367 | dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms); | |
368 | static int | |
369 | copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *); | |
370 | ||
371 | static NAMED * | |
372 | lookup(HASH_TABLE *table, KEY name, size_t createSize); | |
373 | static void FASTCALL | |
374 | hashTableInit(HASH_TABLE *, const XML_Memory_Handling_Suite *ms); | |
375 | static void FASTCALL hashTableClear(HASH_TABLE *); | |
376 | static void FASTCALL hashTableDestroy(HASH_TABLE *); | |
377 | static void FASTCALL | |
378 | hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *); | |
379 | static NAMED * FASTCALL hashTableIterNext(HASH_TABLE_ITER *); | |
380 | ||
381 | static void FASTCALL | |
382 | poolInit(STRING_POOL *, const XML_Memory_Handling_Suite *ms); | |
383 | static void FASTCALL poolClear(STRING_POOL *); | |
384 | static void FASTCALL poolDestroy(STRING_POOL *); | |
385 | static XML_Char * | |
386 | poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
387 | const char *ptr, const char *end); | |
388 | static XML_Char * | |
389 | poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
390 | const char *ptr, const char *end); | |
391 | static XML_Bool FASTCALL poolGrow(STRING_POOL *pool); | |
392 | static const XML_Char * FASTCALL | |
393 | poolCopyString(STRING_POOL *pool, const XML_Char *s); | |
394 | static const XML_Char * | |
395 | poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n); | |
396 | static const XML_Char * FASTCALL | |
397 | poolAppendString(STRING_POOL *pool, const XML_Char *s); | |
398 | ||
399 | static int FASTCALL nextScaffoldPart(XML_Parser parser); | |
400 | static XML_Content * build_model(XML_Parser parser); | |
401 | static ELEMENT_TYPE * | |
402 | getElementType(XML_Parser parser, const ENCODING *enc, | |
403 | const char *ptr, const char *end); | |
404 | ||
405 | static XML_Parser | |
406 | parserCreate(const XML_Char *encodingName, | |
407 | const XML_Memory_Handling_Suite *memsuite, | |
408 | const XML_Char *nameSep, | |
409 | DTD *dtd); | |
410 | static void | |
411 | parserInit(XML_Parser parser, const XML_Char *encodingName); | |
412 | ||
413 | #define poolStart(pool) ((pool)->start) | |
414 | #define poolEnd(pool) ((pool)->ptr) | |
415 | #define poolLength(pool) ((pool)->ptr - (pool)->start) | |
416 | #define poolChop(pool) ((void)--(pool->ptr)) | |
417 | #define poolLastChar(pool) (((pool)->ptr)[-1]) | |
418 | #define poolDiscard(pool) ((pool)->ptr = (pool)->start) | |
419 | #define poolFinish(pool) ((pool)->start = (pool)->ptr) | |
420 | #define poolAppendChar(pool, c) \ | |
421 | (((pool)->ptr == (pool)->end && !poolGrow(pool)) \ | |
422 | ? 0 \ | |
423 | : ((*((pool)->ptr)++ = c), 1)) | |
424 | ||
425 | struct XML_ParserStruct { | |
426 | /* The first member must be userData so that the XML_GetUserData | |
427 | macro works. */ | |
428 | void *m_userData; | |
429 | void *m_handlerArg; | |
430 | char *m_buffer; | |
431 | const XML_Memory_Handling_Suite m_mem; | |
432 | /* first character to be parsed */ | |
433 | const char *m_bufferPtr; | |
434 | /* past last character to be parsed */ | |
435 | char *m_bufferEnd; | |
436 | /* allocated end of buffer */ | |
437 | const char *m_bufferLim; | |
438 | long m_parseEndByteIndex; | |
439 | const char *m_parseEndPtr; | |
440 | XML_Char *m_dataBuf; | |
441 | XML_Char *m_dataBufEnd; | |
442 | XML_StartElementHandler m_startElementHandler; | |
443 | XML_EndElementHandler m_endElementHandler; | |
444 | XML_CharacterDataHandler m_characterDataHandler; | |
445 | XML_ProcessingInstructionHandler m_processingInstructionHandler; | |
446 | XML_CommentHandler m_commentHandler; | |
447 | XML_StartCdataSectionHandler m_startCdataSectionHandler; | |
448 | XML_EndCdataSectionHandler m_endCdataSectionHandler; | |
449 | XML_DefaultHandler m_defaultHandler; | |
450 | XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler; | |
451 | XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler; | |
452 | XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; | |
453 | XML_NotationDeclHandler m_notationDeclHandler; | |
454 | XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; | |
455 | XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; | |
456 | XML_NotStandaloneHandler m_notStandaloneHandler; | |
457 | XML_ExternalEntityRefHandler m_externalEntityRefHandler; | |
458 | XML_Parser m_externalEntityRefHandlerArg; | |
459 | XML_SkippedEntityHandler m_skippedEntityHandler; | |
460 | XML_UnknownEncodingHandler m_unknownEncodingHandler; | |
461 | XML_ElementDeclHandler m_elementDeclHandler; | |
462 | XML_AttlistDeclHandler m_attlistDeclHandler; | |
463 | XML_EntityDeclHandler m_entityDeclHandler; | |
464 | XML_XmlDeclHandler m_xmlDeclHandler; | |
465 | const ENCODING *m_encoding; | |
466 | INIT_ENCODING m_initEncoding; | |
467 | const ENCODING *m_internalEncoding; | |
468 | const XML_Char *m_protocolEncodingName; | |
469 | XML_Bool m_ns; | |
470 | XML_Bool m_ns_triplets; | |
471 | void *m_unknownEncodingMem; | |
472 | void *m_unknownEncodingData; | |
473 | void *m_unknownEncodingHandlerData; | |
474 | void (*m_unknownEncodingRelease)(void *); | |
475 | PROLOG_STATE m_prologState; | |
476 | Processor *m_processor; | |
477 | enum XML_Error m_errorCode; | |
478 | const char *m_eventPtr; | |
479 | const char *m_eventEndPtr; | |
480 | const char *m_positionPtr; | |
481 | OPEN_INTERNAL_ENTITY *m_openInternalEntities; | |
482 | XML_Bool m_defaultExpandInternalEntities; | |
483 | int m_tagLevel; | |
484 | ENTITY *m_declEntity; | |
485 | const XML_Char *m_doctypeName; | |
486 | const XML_Char *m_doctypeSysid; | |
487 | const XML_Char *m_doctypePubid; | |
488 | const XML_Char *m_declAttributeType; | |
489 | const XML_Char *m_declNotationName; | |
490 | const XML_Char *m_declNotationPublicId; | |
491 | ELEMENT_TYPE *m_declElementType; | |
492 | ATTRIBUTE_ID *m_declAttributeId; | |
493 | XML_Bool m_declAttributeIsCdata; | |
494 | XML_Bool m_declAttributeIsId; | |
495 | DTD *m_dtd; | |
496 | const XML_Char *m_curBase; | |
497 | TAG *m_tagStack; | |
498 | TAG *m_freeTagList; | |
499 | BINDING *m_inheritedBindings; | |
500 | BINDING *m_freeBindingList; | |
501 | int m_attsSize; | |
502 | int m_nSpecifiedAtts; | |
503 | int m_idAttIndex; | |
504 | ATTRIBUTE *m_atts; | |
505 | POSITION m_position; | |
506 | STRING_POOL m_tempPool; | |
507 | STRING_POOL m_temp2Pool; | |
508 | char *m_groupConnector; | |
509 | unsigned m_groupSize; | |
510 | XML_Char m_namespaceSeparator; | |
511 | XML_Parser m_parentParser; | |
512 | #ifdef XML_DTD | |
513 | XML_Bool m_isParamEntity; | |
514 | XML_Bool m_useForeignDTD; | |
515 | enum XML_ParamEntityParsing m_paramEntityParsing; | |
516 | #endif | |
517 | }; | |
518 | ||
519 | #define MALLOC(s) (parser->m_mem.malloc_fcn((s))) | |
520 | #define REALLOC(p,s) (parser->m_mem.realloc_fcn((p),(s))) | |
521 | #define FREE(p) (parser->m_mem.free_fcn((p))) | |
522 | ||
523 | #define userData (parser->m_userData) | |
524 | #define handlerArg (parser->m_handlerArg) | |
525 | #define startElementHandler (parser->m_startElementHandler) | |
526 | #define endElementHandler (parser->m_endElementHandler) | |
527 | #define characterDataHandler (parser->m_characterDataHandler) | |
528 | #define processingInstructionHandler \ | |
529 | (parser->m_processingInstructionHandler) | |
530 | #define commentHandler (parser->m_commentHandler) | |
531 | #define startCdataSectionHandler \ | |
532 | (parser->m_startCdataSectionHandler) | |
533 | #define endCdataSectionHandler (parser->m_endCdataSectionHandler) | |
534 | #define defaultHandler (parser->m_defaultHandler) | |
535 | #define startDoctypeDeclHandler (parser->m_startDoctypeDeclHandler) | |
536 | #define endDoctypeDeclHandler (parser->m_endDoctypeDeclHandler) | |
537 | #define unparsedEntityDeclHandler \ | |
538 | (parser->m_unparsedEntityDeclHandler) | |
539 | #define notationDeclHandler (parser->m_notationDeclHandler) | |
540 | #define startNamespaceDeclHandler \ | |
541 | (parser->m_startNamespaceDeclHandler) | |
542 | #define endNamespaceDeclHandler (parser->m_endNamespaceDeclHandler) | |
543 | #define notStandaloneHandler (parser->m_notStandaloneHandler) | |
544 | #define externalEntityRefHandler \ | |
545 | (parser->m_externalEntityRefHandler) | |
546 | #define externalEntityRefHandlerArg \ | |
547 | (parser->m_externalEntityRefHandlerArg) | |
548 | #define internalEntityRefHandler \ | |
549 | (parser->m_internalEntityRefHandler) | |
550 | #define skippedEntityHandler (parser->m_skippedEntityHandler) | |
551 | #define unknownEncodingHandler (parser->m_unknownEncodingHandler) | |
552 | #define elementDeclHandler (parser->m_elementDeclHandler) | |
553 | #define attlistDeclHandler (parser->m_attlistDeclHandler) | |
554 | #define entityDeclHandler (parser->m_entityDeclHandler) | |
555 | #define xmlDeclHandler (parser->m_xmlDeclHandler) | |
556 | #define encoding (parser->m_encoding) | |
557 | #define initEncoding (parser->m_initEncoding) | |
558 | #define internalEncoding (parser->m_internalEncoding) | |
559 | #define unknownEncodingMem (parser->m_unknownEncodingMem) | |
560 | #define unknownEncodingData (parser->m_unknownEncodingData) | |
561 | #define unknownEncodingHandlerData \ | |
562 | (parser->m_unknownEncodingHandlerData) | |
563 | #define unknownEncodingRelease (parser->m_unknownEncodingRelease) | |
564 | #define protocolEncodingName (parser->m_protocolEncodingName) | |
565 | #define ns (parser->m_ns) | |
566 | #define ns_triplets (parser->m_ns_triplets) | |
567 | #define prologState (parser->m_prologState) | |
568 | #define processor (parser->m_processor) | |
569 | #define errorCode (parser->m_errorCode) | |
570 | #define eventPtr (parser->m_eventPtr) | |
571 | #define eventEndPtr (parser->m_eventEndPtr) | |
572 | #define positionPtr (parser->m_positionPtr) | |
573 | #define position (parser->m_position) | |
574 | #define openInternalEntities (parser->m_openInternalEntities) | |
575 | #define defaultExpandInternalEntities \ | |
576 | (parser->m_defaultExpandInternalEntities) | |
577 | #define tagLevel (parser->m_tagLevel) | |
578 | #define buffer (parser->m_buffer) | |
579 | #define bufferPtr (parser->m_bufferPtr) | |
580 | #define bufferEnd (parser->m_bufferEnd) | |
581 | #define parseEndByteIndex (parser->m_parseEndByteIndex) | |
582 | #define parseEndPtr (parser->m_parseEndPtr) | |
583 | #define bufferLim (parser->m_bufferLim) | |
584 | #define dataBuf (parser->m_dataBuf) | |
585 | #define dataBufEnd (parser->m_dataBufEnd) | |
586 | #define _dtd (parser->m_dtd) | |
587 | #define curBase (parser->m_curBase) | |
588 | #define declEntity (parser->m_declEntity) | |
589 | #define doctypeName (parser->m_doctypeName) | |
590 | #define doctypeSysid (parser->m_doctypeSysid) | |
591 | #define doctypePubid (parser->m_doctypePubid) | |
592 | #define declAttributeType (parser->m_declAttributeType) | |
593 | #define declNotationName (parser->m_declNotationName) | |
594 | #define declNotationPublicId (parser->m_declNotationPublicId) | |
595 | #define declElementType (parser->m_declElementType) | |
596 | #define declAttributeId (parser->m_declAttributeId) | |
597 | #define declAttributeIsCdata (parser->m_declAttributeIsCdata) | |
598 | #define declAttributeIsId (parser->m_declAttributeIsId) | |
599 | #define freeTagList (parser->m_freeTagList) | |
600 | #define freeBindingList (parser->m_freeBindingList) | |
601 | #define inheritedBindings (parser->m_inheritedBindings) | |
602 | #define tagStack (parser->m_tagStack) | |
603 | #define atts (parser->m_atts) | |
604 | #define attsSize (parser->m_attsSize) | |
605 | #define nSpecifiedAtts (parser->m_nSpecifiedAtts) | |
606 | #define idAttIndex (parser->m_idAttIndex) | |
607 | #define tempPool (parser->m_tempPool) | |
608 | #define temp2Pool (parser->m_temp2Pool) | |
609 | #define groupConnector (parser->m_groupConnector) | |
610 | #define groupSize (parser->m_groupSize) | |
611 | #define namespaceSeparator (parser->m_namespaceSeparator) | |
612 | #define parentParser (parser->m_parentParser) | |
613 | #ifdef XML_DTD | |
614 | #define isParamEntity (parser->m_isParamEntity) | |
615 | #define useForeignDTD (parser->m_useForeignDTD) | |
616 | #define paramEntityParsing (parser->m_paramEntityParsing) | |
617 | #endif /* XML_DTD */ | |
618 | ||
619 | #define parsing \ | |
620 | (parentParser \ | |
621 | ? \ | |
622 | (isParamEntity \ | |
623 | ? \ | |
624 | (processor != externalParEntInitProcessor) \ | |
625 | : \ | |
626 | (processor != externalEntityInitProcessor)) \ | |
627 | : \ | |
628 | (processor != prologInitProcessor)) | |
629 | ||
630 | XML_Parser | |
631 | XML_ParserCreate(const XML_Char *encodingName) | |
632 | { | |
633 | return XML_ParserCreate_MM(encodingName, NULL, NULL); | |
634 | } | |
635 | ||
636 | XML_Parser | |
637 | XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) | |
638 | { | |
639 | XML_Char tmp[2]; | |
640 | *tmp = nsSep; | |
641 | return XML_ParserCreate_MM(encodingName, NULL, tmp); | |
642 | } | |
643 | ||
644 | static const XML_Char implicitContext[] = { | |
645 | 'x', 'm', 'l', '=', 'h', 't', 't', 'p', ':', '/', '/', | |
646 | 'w', 'w', 'w', '.', 'w', '3', '.', 'o', 'r', 'g', '/', | |
647 | 'X', 'M', 'L', '/', '1', '9', '9', '8', '/', | |
648 | 'n', 'a', 'm', 'e', 's', 'p', 'a', 'c', 'e', '\0' | |
649 | }; | |
650 | ||
651 | XML_Parser | |
652 | XML_ParserCreate_MM(const XML_Char *encodingName, | |
653 | const XML_Memory_Handling_Suite *memsuite, | |
654 | const XML_Char *nameSep) | |
655 | { | |
656 | XML_Parser parser = parserCreate(encodingName, memsuite, nameSep, NULL); | |
657 | if (parser != NULL && ns) { | |
658 | /* implicit context only set for root parser, since child | |
659 | parsers (i.e. external entity parsers) will inherit it | |
660 | */ | |
661 | if (!setContext(parser, implicitContext)) { | |
662 | XML_ParserFree(parser); | |
663 | return NULL; | |
664 | } | |
665 | } | |
666 | return parser; | |
667 | } | |
668 | ||
669 | static XML_Parser | |
670 | parserCreate(const XML_Char *encodingName, | |
671 | const XML_Memory_Handling_Suite *memsuite, | |
672 | const XML_Char *nameSep, | |
673 | DTD *dtd) | |
674 | { | |
675 | XML_Parser parser; | |
676 | ||
677 | if (memsuite) { | |
678 | XML_Memory_Handling_Suite *mtemp; | |
679 | parser = (XML_Parser) | |
680 | memsuite->malloc_fcn(sizeof(struct XML_ParserStruct)); | |
681 | if (parser != NULL) { | |
682 | mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem); | |
683 | mtemp->malloc_fcn = memsuite->malloc_fcn; | |
684 | mtemp->realloc_fcn = memsuite->realloc_fcn; | |
685 | mtemp->free_fcn = memsuite->free_fcn; | |
686 | } | |
687 | } | |
688 | else { | |
689 | XML_Memory_Handling_Suite *mtemp; | |
690 | parser = (XML_Parser)malloc(sizeof(struct XML_ParserStruct)); | |
691 | if (parser != NULL) { | |
692 | mtemp = (XML_Memory_Handling_Suite *)&(parser->m_mem); | |
693 | mtemp->malloc_fcn = malloc; | |
694 | mtemp->realloc_fcn = realloc; | |
695 | mtemp->free_fcn = free; | |
696 | } | |
697 | } | |
698 | ||
699 | if (!parser) | |
700 | return parser; | |
701 | ||
702 | buffer = NULL; | |
703 | bufferLim = NULL; | |
704 | ||
705 | attsSize = INIT_ATTS_SIZE; | |
706 | atts = (ATTRIBUTE *)MALLOC(attsSize * sizeof(ATTRIBUTE)); | |
707 | if (atts == NULL) { | |
708 | FREE(parser); | |
709 | return NULL; | |
710 | } | |
711 | dataBuf = (XML_Char *)MALLOC(INIT_DATA_BUF_SIZE * sizeof(XML_Char)); | |
712 | if (dataBuf == NULL) { | |
713 | FREE(atts); | |
714 | FREE(parser); | |
715 | return NULL; | |
716 | } | |
717 | dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE; | |
718 | ||
719 | if (dtd) | |
720 | _dtd = dtd; | |
721 | else { | |
722 | _dtd = dtdCreate(&parser->m_mem); | |
723 | if (_dtd == NULL) { | |
724 | FREE(dataBuf); | |
725 | FREE(atts); | |
726 | FREE(parser); | |
727 | return NULL; | |
728 | } | |
729 | } | |
730 | ||
731 | freeBindingList = NULL; | |
732 | freeTagList = NULL; | |
733 | ||
734 | groupSize = 0; | |
735 | groupConnector = NULL; | |
736 | ||
737 | unknownEncodingHandler = NULL; | |
738 | unknownEncodingHandlerData = NULL; | |
739 | ||
740 | namespaceSeparator = '!'; | |
741 | ns = XML_FALSE; | |
742 | ns_triplets = XML_FALSE; | |
743 | ||
744 | poolInit(&tempPool, &(parser->m_mem)); | |
745 | poolInit(&temp2Pool, &(parser->m_mem)); | |
746 | parserInit(parser, encodingName); | |
747 | ||
748 | if (encodingName && !protocolEncodingName) { | |
749 | XML_ParserFree(parser); | |
750 | return NULL; | |
751 | } | |
752 | ||
753 | if (nameSep) { | |
754 | ns = XML_TRUE; | |
755 | internalEncoding = XmlGetInternalEncodingNS(); | |
756 | namespaceSeparator = *nameSep; | |
757 | } | |
758 | else { | |
759 | internalEncoding = XmlGetInternalEncoding(); | |
760 | } | |
761 | ||
762 | return parser; | |
763 | } | |
764 | ||
765 | static void | |
766 | parserInit(XML_Parser parser, const XML_Char *encodingName) | |
767 | { | |
768 | processor = prologInitProcessor; | |
769 | XmlPrologStateInit(&prologState); | |
770 | protocolEncodingName = (encodingName != NULL | |
771 | ? poolCopyString(&tempPool, encodingName) | |
772 | : NULL); | |
773 | curBase = NULL; | |
774 | XmlInitEncoding(&initEncoding, &encoding, 0); | |
775 | userData = NULL; | |
776 | handlerArg = NULL; | |
777 | startElementHandler = NULL; | |
778 | endElementHandler = NULL; | |
779 | characterDataHandler = NULL; | |
780 | processingInstructionHandler = NULL; | |
781 | commentHandler = NULL; | |
782 | startCdataSectionHandler = NULL; | |
783 | endCdataSectionHandler = NULL; | |
784 | defaultHandler = NULL; | |
785 | startDoctypeDeclHandler = NULL; | |
786 | endDoctypeDeclHandler = NULL; | |
787 | unparsedEntityDeclHandler = NULL; | |
788 | notationDeclHandler = NULL; | |
789 | startNamespaceDeclHandler = NULL; | |
790 | endNamespaceDeclHandler = NULL; | |
791 | notStandaloneHandler = NULL; | |
792 | externalEntityRefHandler = NULL; | |
793 | externalEntityRefHandlerArg = parser; | |
794 | skippedEntityHandler = NULL; | |
795 | elementDeclHandler = NULL; | |
796 | attlistDeclHandler = NULL; | |
797 | entityDeclHandler = NULL; | |
798 | xmlDeclHandler = NULL; | |
799 | bufferPtr = buffer; | |
800 | bufferEnd = buffer; | |
801 | parseEndByteIndex = 0; | |
802 | parseEndPtr = NULL; | |
803 | declElementType = NULL; | |
804 | declAttributeId = NULL; | |
805 | declEntity = NULL; | |
806 | doctypeName = NULL; | |
807 | doctypeSysid = NULL; | |
808 | doctypePubid = NULL; | |
809 | declAttributeType = NULL; | |
810 | declNotationName = NULL; | |
811 | declNotationPublicId = NULL; | |
812 | declAttributeIsCdata = XML_FALSE; | |
813 | declAttributeIsId = XML_FALSE; | |
814 | memset(&position, 0, sizeof(POSITION)); | |
815 | errorCode = XML_ERROR_NONE; | |
816 | eventPtr = NULL; | |
817 | eventEndPtr = NULL; | |
818 | positionPtr = NULL; | |
819 | openInternalEntities = 0; | |
820 | defaultExpandInternalEntities = XML_TRUE; | |
821 | tagLevel = 0; | |
822 | tagStack = NULL; | |
823 | inheritedBindings = NULL; | |
824 | nSpecifiedAtts = 0; | |
825 | unknownEncodingMem = NULL; | |
826 | unknownEncodingRelease = NULL; | |
827 | unknownEncodingData = NULL; | |
828 | parentParser = NULL; | |
829 | #ifdef XML_DTD | |
830 | isParamEntity = XML_FALSE; | |
831 | useForeignDTD = XML_FALSE; | |
832 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; | |
833 | #endif | |
834 | } | |
835 | ||
836 | /* moves list of bindings to freeBindingList */ | |
837 | static void FASTCALL | |
838 | moveToFreeBindingList(XML_Parser parser, BINDING *bindings) | |
839 | { | |
840 | while (bindings) { | |
841 | BINDING *b = bindings; | |
842 | bindings = bindings->nextTagBinding; | |
843 | b->nextTagBinding = freeBindingList; | |
844 | freeBindingList = b; | |
845 | } | |
846 | } | |
847 | ||
848 | XML_Bool | |
849 | XML_ParserReset(XML_Parser parser, const XML_Char *encodingName) | |
850 | { | |
851 | TAG *tStk; | |
852 | if (parentParser) | |
853 | return XML_FALSE; | |
854 | /* move tagStack to freeTagList */ | |
855 | tStk = tagStack; | |
856 | while (tStk) { | |
857 | TAG *tag = tStk; | |
858 | tStk = tStk->parent; | |
859 | tag->parent = freeTagList; | |
860 | moveToFreeBindingList(parser, tag->bindings); | |
861 | tag->bindings = NULL; | |
862 | freeTagList = tag; | |
863 | } | |
864 | moveToFreeBindingList(parser, inheritedBindings); | |
865 | if (unknownEncodingMem) | |
866 | FREE(unknownEncodingMem); | |
867 | if (unknownEncodingRelease) | |
868 | unknownEncodingRelease(unknownEncodingData); | |
869 | poolClear(&tempPool); | |
870 | poolClear(&temp2Pool); | |
871 | parserInit(parser, encodingName); | |
872 | dtdReset(_dtd, &parser->m_mem); | |
873 | return setContext(parser, implicitContext); | |
874 | } | |
875 | ||
876 | enum XML_Status | |
877 | XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) | |
878 | { | |
879 | /* Block after XML_Parse()/XML_ParseBuffer() has been called. | |
880 | XXX There's no way for the caller to determine which of the | |
881 | XXX possible error cases caused the XML_STATUS_ERROR return. | |
882 | */ | |
883 | if (parsing) | |
884 | return XML_STATUS_ERROR; | |
885 | if (encodingName == NULL) | |
886 | protocolEncodingName = NULL; | |
887 | else { | |
888 | protocolEncodingName = poolCopyString(&tempPool, encodingName); | |
889 | if (!protocolEncodingName) | |
890 | return XML_STATUS_ERROR; | |
891 | } | |
892 | return XML_STATUS_OK; | |
893 | } | |
894 | ||
895 | XML_Parser | |
896 | XML_ExternalEntityParserCreate(XML_Parser oldParser, | |
897 | const XML_Char *context, | |
898 | const XML_Char *encodingName) | |
899 | { | |
900 | XML_Parser parser = oldParser; | |
901 | DTD *newDtd = NULL; | |
902 | DTD *oldDtd = _dtd; | |
903 | XML_StartElementHandler oldStartElementHandler = startElementHandler; | |
904 | XML_EndElementHandler oldEndElementHandler = endElementHandler; | |
905 | XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler; | |
906 | XML_ProcessingInstructionHandler oldProcessingInstructionHandler | |
907 | = processingInstructionHandler; | |
908 | XML_CommentHandler oldCommentHandler = commentHandler; | |
909 | XML_StartCdataSectionHandler oldStartCdataSectionHandler | |
910 | = startCdataSectionHandler; | |
911 | XML_EndCdataSectionHandler oldEndCdataSectionHandler | |
912 | = endCdataSectionHandler; | |
913 | XML_DefaultHandler oldDefaultHandler = defaultHandler; | |
914 | XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler | |
915 | = unparsedEntityDeclHandler; | |
916 | XML_NotationDeclHandler oldNotationDeclHandler = notationDeclHandler; | |
917 | XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler | |
918 | = startNamespaceDeclHandler; | |
919 | XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler | |
920 | = endNamespaceDeclHandler; | |
921 | XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler; | |
922 | XML_ExternalEntityRefHandler oldExternalEntityRefHandler | |
923 | = externalEntityRefHandler; | |
924 | XML_SkippedEntityHandler oldSkippedEntityHandler = skippedEntityHandler; | |
925 | XML_UnknownEncodingHandler oldUnknownEncodingHandler | |
926 | = unknownEncodingHandler; | |
927 | XML_ElementDeclHandler oldElementDeclHandler = elementDeclHandler; | |
928 | XML_AttlistDeclHandler oldAttlistDeclHandler = attlistDeclHandler; | |
929 | XML_EntityDeclHandler oldEntityDeclHandler = entityDeclHandler; | |
930 | XML_XmlDeclHandler oldXmlDeclHandler = xmlDeclHandler; | |
931 | ELEMENT_TYPE * oldDeclElementType = declElementType; | |
932 | ||
933 | void *oldUserData = userData; | |
934 | void *oldHandlerArg = handlerArg; | |
935 | XML_Bool oldDefaultExpandInternalEntities = defaultExpandInternalEntities; | |
936 | XML_Parser oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg; | |
937 | #ifdef XML_DTD | |
938 | enum XML_ParamEntityParsing oldParamEntityParsing = paramEntityParsing; | |
939 | int oldInEntityValue = prologState.inEntityValue; | |
940 | #endif | |
941 | XML_Bool oldns_triplets = ns_triplets; | |
942 | ||
943 | #ifdef XML_DTD | |
944 | if (!context) | |
945 | newDtd = oldDtd; | |
946 | #endif /* XML_DTD */ | |
947 | ||
948 | /* Note that the magical uses of the pre-processor to make field | |
949 | access look more like C++ require that `parser' be overwritten | |
950 | here. This makes this function more painful to follow than it | |
951 | would be otherwise. | |
952 | */ | |
953 | if (ns) { | |
954 | XML_Char tmp[2]; | |
955 | *tmp = namespaceSeparator; | |
956 | parser = parserCreate(encodingName, &parser->m_mem, tmp, newDtd); | |
957 | } | |
958 | else { | |
959 | parser = parserCreate(encodingName, &parser->m_mem, NULL, newDtd); | |
960 | } | |
961 | ||
962 | if (!parser) | |
963 | return NULL; | |
964 | ||
965 | startElementHandler = oldStartElementHandler; | |
966 | endElementHandler = oldEndElementHandler; | |
967 | characterDataHandler = oldCharacterDataHandler; | |
968 | processingInstructionHandler = oldProcessingInstructionHandler; | |
969 | commentHandler = oldCommentHandler; | |
970 | startCdataSectionHandler = oldStartCdataSectionHandler; | |
971 | endCdataSectionHandler = oldEndCdataSectionHandler; | |
972 | defaultHandler = oldDefaultHandler; | |
973 | unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler; | |
974 | notationDeclHandler = oldNotationDeclHandler; | |
975 | startNamespaceDeclHandler = oldStartNamespaceDeclHandler; | |
976 | endNamespaceDeclHandler = oldEndNamespaceDeclHandler; | |
977 | notStandaloneHandler = oldNotStandaloneHandler; | |
978 | externalEntityRefHandler = oldExternalEntityRefHandler; | |
979 | skippedEntityHandler = oldSkippedEntityHandler; | |
980 | unknownEncodingHandler = oldUnknownEncodingHandler; | |
981 | elementDeclHandler = oldElementDeclHandler; | |
982 | attlistDeclHandler = oldAttlistDeclHandler; | |
983 | entityDeclHandler = oldEntityDeclHandler; | |
984 | xmlDeclHandler = oldXmlDeclHandler; | |
985 | declElementType = oldDeclElementType; | |
986 | userData = oldUserData; | |
987 | if (oldUserData == oldHandlerArg) | |
988 | handlerArg = userData; | |
989 | else | |
990 | handlerArg = parser; | |
991 | if (oldExternalEntityRefHandlerArg != oldParser) | |
992 | externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; | |
993 | defaultExpandInternalEntities = oldDefaultExpandInternalEntities; | |
994 | ns_triplets = oldns_triplets; | |
995 | parentParser = oldParser; | |
996 | #ifdef XML_DTD | |
997 | paramEntityParsing = oldParamEntityParsing; | |
998 | prologState.inEntityValue = oldInEntityValue; | |
999 | if (context) { | |
1000 | #endif /* XML_DTD */ | |
1001 | if (!dtdCopy(_dtd, oldDtd, &parser->m_mem) | |
1002 | || !setContext(parser, context)) { | |
1003 | XML_ParserFree(parser); | |
1004 | return NULL; | |
1005 | } | |
1006 | processor = externalEntityInitProcessor; | |
1007 | #ifdef XML_DTD | |
1008 | } | |
1009 | else { | |
1010 | /* The DTD instance referenced by _dtd is shared between the document's | |
1011 | root parser and external PE parsers, therefore one does not need to | |
1012 | call setContext. In addition, one also *must* not call setContext, | |
1013 | because this would overwrite existing prefix->binding pointers in | |
1014 | _dtd with ones that get destroyed with the external PE parser. | |
1015 | This would leave those prefixes with dangling pointers. | |
1016 | */ | |
1017 | isParamEntity = XML_TRUE; | |
1018 | XmlPrologStateInitExternalEntity(&prologState); | |
1019 | processor = externalParEntInitProcessor; | |
1020 | } | |
1021 | #endif /* XML_DTD */ | |
1022 | return parser; | |
1023 | } | |
1024 | ||
1025 | static void FASTCALL | |
1026 | destroyBindings(BINDING *bindings, XML_Parser parser) | |
1027 | { | |
1028 | for (;;) { | |
1029 | BINDING *b = bindings; | |
1030 | if (!b) | |
1031 | break; | |
1032 | bindings = b->nextTagBinding; | |
1033 | FREE(b->uri); | |
1034 | FREE(b); | |
1035 | } | |
1036 | } | |
1037 | ||
1038 | void | |
1039 | XML_ParserFree(XML_Parser parser) | |
1040 | { | |
1041 | for (;;) { | |
1042 | TAG *p; | |
1043 | if (tagStack == NULL) { | |
1044 | if (freeTagList == NULL) | |
1045 | break; | |
1046 | tagStack = freeTagList; | |
1047 | freeTagList = NULL; | |
1048 | } | |
1049 | p = tagStack; | |
1050 | tagStack = tagStack->parent; | |
1051 | FREE(p->buf); | |
1052 | destroyBindings(p->bindings, parser); | |
1053 | FREE(p); | |
1054 | } | |
1055 | destroyBindings(freeBindingList, parser); | |
1056 | destroyBindings(inheritedBindings, parser); | |
1057 | poolDestroy(&tempPool); | |
1058 | poolDestroy(&temp2Pool); | |
1059 | #ifdef XML_DTD | |
1060 | /* external parameter entity parsers share the DTD structure | |
1061 | parser->m_dtd with the root parser, so we must not destroy it | |
1062 | */ | |
1063 | if (!isParamEntity && _dtd) | |
1064 | #else | |
1065 | if (_dtd) | |
1066 | #endif /* XML_DTD */ | |
1067 | dtdDestroy(_dtd, (XML_Bool)!parentParser, &parser->m_mem); | |
1068 | FREE((void *)atts); | |
1069 | if (groupConnector) | |
1070 | FREE(groupConnector); | |
1071 | if (buffer) | |
1072 | FREE(buffer); | |
1073 | FREE(dataBuf); | |
1074 | if (unknownEncodingMem) | |
1075 | FREE(unknownEncodingMem); | |
1076 | if (unknownEncodingRelease) | |
1077 | unknownEncodingRelease(unknownEncodingData); | |
1078 | FREE(parser); | |
1079 | } | |
1080 | ||
1081 | void | |
1082 | XML_UseParserAsHandlerArg(XML_Parser parser) | |
1083 | { | |
1084 | handlerArg = parser; | |
1085 | } | |
1086 | ||
1087 | enum XML_Error | |
1088 | XML_UseForeignDTD(XML_Parser parser, XML_Bool useDTD) | |
1089 | { | |
1090 | #ifdef XML_DTD | |
1091 | /* block after XML_Parse()/XML_ParseBuffer() has been called */ | |
1092 | if (parsing) | |
1093 | return XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING; | |
1094 | useForeignDTD = useDTD; | |
1095 | return XML_ERROR_NONE; | |
1096 | #else | |
1097 | return XML_ERROR_FEATURE_REQUIRES_XML_DTD; | |
1098 | #endif | |
1099 | } | |
1100 | ||
1101 | void | |
1102 | XML_SetReturnNSTriplet(XML_Parser parser, int do_nst) | |
1103 | { | |
1104 | /* block after XML_Parse()/XML_ParseBuffer() has been called */ | |
1105 | if (parsing) | |
1106 | return; | |
1107 | ns_triplets = do_nst ? XML_TRUE : XML_FALSE; | |
1108 | } | |
1109 | ||
1110 | void | |
1111 | XML_SetUserData(XML_Parser parser, void *p) | |
1112 | { | |
1113 | if (handlerArg == userData) | |
1114 | handlerArg = userData = p; | |
1115 | else | |
1116 | userData = p; | |
1117 | } | |
1118 | ||
1119 | enum XML_Status | |
1120 | XML_SetBase(XML_Parser parser, const XML_Char *p) | |
1121 | { | |
1122 | if (p) { | |
1123 | p = poolCopyString(&_dtd->pool, p); | |
1124 | if (!p) | |
1125 | return XML_STATUS_ERROR; | |
1126 | curBase = p; | |
1127 | } | |
1128 | else | |
1129 | curBase = NULL; | |
1130 | return XML_STATUS_OK; | |
1131 | } | |
1132 | ||
1133 | const XML_Char * | |
1134 | XML_GetBase(XML_Parser parser) | |
1135 | { | |
1136 | return curBase; | |
1137 | } | |
1138 | ||
1139 | int | |
1140 | XML_GetSpecifiedAttributeCount(XML_Parser parser) | |
1141 | { | |
1142 | return nSpecifiedAtts; | |
1143 | } | |
1144 | ||
1145 | int | |
1146 | XML_GetIdAttributeIndex(XML_Parser parser) | |
1147 | { | |
1148 | return idAttIndex; | |
1149 | } | |
1150 | ||
1151 | void | |
1152 | XML_SetElementHandler(XML_Parser parser, | |
1153 | XML_StartElementHandler start, | |
1154 | XML_EndElementHandler end) | |
1155 | { | |
1156 | startElementHandler = start; | |
1157 | endElementHandler = end; | |
1158 | } | |
1159 | ||
1160 | void | |
1161 | XML_SetStartElementHandler(XML_Parser parser, | |
1162 | XML_StartElementHandler start) { | |
1163 | startElementHandler = start; | |
1164 | } | |
1165 | ||
1166 | void | |
1167 | XML_SetEndElementHandler(XML_Parser parser, | |
1168 | XML_EndElementHandler end) { | |
1169 | endElementHandler = end; | |
1170 | } | |
1171 | ||
1172 | void | |
1173 | XML_SetCharacterDataHandler(XML_Parser parser, | |
1174 | XML_CharacterDataHandler handler) | |
1175 | { | |
1176 | characterDataHandler = handler; | |
1177 | } | |
1178 | ||
1179 | void | |
1180 | XML_SetProcessingInstructionHandler(XML_Parser parser, | |
1181 | XML_ProcessingInstructionHandler handler) | |
1182 | { | |
1183 | processingInstructionHandler = handler; | |
1184 | } | |
1185 | ||
1186 | void | |
1187 | XML_SetCommentHandler(XML_Parser parser, | |
1188 | XML_CommentHandler handler) | |
1189 | { | |
1190 | commentHandler = handler; | |
1191 | } | |
1192 | ||
1193 | void | |
1194 | XML_SetCdataSectionHandler(XML_Parser parser, | |
1195 | XML_StartCdataSectionHandler start, | |
1196 | XML_EndCdataSectionHandler end) | |
1197 | { | |
1198 | startCdataSectionHandler = start; | |
1199 | endCdataSectionHandler = end; | |
1200 | } | |
1201 | ||
1202 | void | |
1203 | XML_SetStartCdataSectionHandler(XML_Parser parser, | |
1204 | XML_StartCdataSectionHandler start) { | |
1205 | startCdataSectionHandler = start; | |
1206 | } | |
1207 | ||
1208 | void | |
1209 | XML_SetEndCdataSectionHandler(XML_Parser parser, | |
1210 | XML_EndCdataSectionHandler end) { | |
1211 | endCdataSectionHandler = end; | |
1212 | } | |
1213 | ||
1214 | void | |
1215 | XML_SetDefaultHandler(XML_Parser parser, | |
1216 | XML_DefaultHandler handler) | |
1217 | { | |
1218 | defaultHandler = handler; | |
1219 | defaultExpandInternalEntities = XML_FALSE; | |
1220 | } | |
1221 | ||
1222 | void | |
1223 | XML_SetDefaultHandlerExpand(XML_Parser parser, | |
1224 | XML_DefaultHandler handler) | |
1225 | { | |
1226 | defaultHandler = handler; | |
1227 | defaultExpandInternalEntities = XML_TRUE; | |
1228 | } | |
1229 | ||
1230 | void | |
1231 | XML_SetDoctypeDeclHandler(XML_Parser parser, | |
1232 | XML_StartDoctypeDeclHandler start, | |
1233 | XML_EndDoctypeDeclHandler end) | |
1234 | { | |
1235 | startDoctypeDeclHandler = start; | |
1236 | endDoctypeDeclHandler = end; | |
1237 | } | |
1238 | ||
1239 | void | |
1240 | XML_SetStartDoctypeDeclHandler(XML_Parser parser, | |
1241 | XML_StartDoctypeDeclHandler start) { | |
1242 | startDoctypeDeclHandler = start; | |
1243 | } | |
1244 | ||
1245 | void | |
1246 | XML_SetEndDoctypeDeclHandler(XML_Parser parser, | |
1247 | XML_EndDoctypeDeclHandler end) { | |
1248 | endDoctypeDeclHandler = end; | |
1249 | } | |
1250 | ||
1251 | void | |
1252 | XML_SetUnparsedEntityDeclHandler(XML_Parser parser, | |
1253 | XML_UnparsedEntityDeclHandler handler) | |
1254 | { | |
1255 | unparsedEntityDeclHandler = handler; | |
1256 | } | |
1257 | ||
1258 | void | |
1259 | XML_SetNotationDeclHandler(XML_Parser parser, | |
1260 | XML_NotationDeclHandler handler) | |
1261 | { | |
1262 | notationDeclHandler = handler; | |
1263 | } | |
1264 | ||
1265 | void | |
1266 | XML_SetNamespaceDeclHandler(XML_Parser parser, | |
1267 | XML_StartNamespaceDeclHandler start, | |
1268 | XML_EndNamespaceDeclHandler end) | |
1269 | { | |
1270 | startNamespaceDeclHandler = start; | |
1271 | endNamespaceDeclHandler = end; | |
1272 | } | |
1273 | ||
1274 | void | |
1275 | XML_SetStartNamespaceDeclHandler(XML_Parser parser, | |
1276 | XML_StartNamespaceDeclHandler start) { | |
1277 | startNamespaceDeclHandler = start; | |
1278 | } | |
1279 | ||
1280 | void | |
1281 | XML_SetEndNamespaceDeclHandler(XML_Parser parser, | |
1282 | XML_EndNamespaceDeclHandler end) { | |
1283 | endNamespaceDeclHandler = end; | |
1284 | } | |
1285 | ||
1286 | void | |
1287 | XML_SetNotStandaloneHandler(XML_Parser parser, | |
1288 | XML_NotStandaloneHandler handler) | |
1289 | { | |
1290 | notStandaloneHandler = handler; | |
1291 | } | |
1292 | ||
1293 | void | |
1294 | XML_SetExternalEntityRefHandler(XML_Parser parser, | |
1295 | XML_ExternalEntityRefHandler handler) | |
1296 | { | |
1297 | externalEntityRefHandler = handler; | |
1298 | } | |
1299 | ||
1300 | void | |
1301 | XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) | |
1302 | { | |
1303 | if (arg) | |
1304 | externalEntityRefHandlerArg = (XML_Parser)arg; | |
1305 | else | |
1306 | externalEntityRefHandlerArg = parser; | |
1307 | } | |
1308 | ||
1309 | void | |
1310 | XML_SetSkippedEntityHandler(XML_Parser parser, | |
1311 | XML_SkippedEntityHandler handler) | |
1312 | { | |
1313 | skippedEntityHandler = handler; | |
1314 | } | |
1315 | ||
1316 | void | |
1317 | XML_SetUnknownEncodingHandler(XML_Parser parser, | |
1318 | XML_UnknownEncodingHandler handler, | |
1319 | void *data) | |
1320 | { | |
1321 | unknownEncodingHandler = handler; | |
1322 | unknownEncodingHandlerData = data; | |
1323 | } | |
1324 | ||
1325 | void | |
1326 | XML_SetElementDeclHandler(XML_Parser parser, | |
1327 | XML_ElementDeclHandler eldecl) | |
1328 | { | |
1329 | elementDeclHandler = eldecl; | |
1330 | } | |
1331 | ||
1332 | void | |
1333 | XML_SetAttlistDeclHandler(XML_Parser parser, | |
1334 | XML_AttlistDeclHandler attdecl) | |
1335 | { | |
1336 | attlistDeclHandler = attdecl; | |
1337 | } | |
1338 | ||
1339 | void | |
1340 | XML_SetEntityDeclHandler(XML_Parser parser, | |
1341 | XML_EntityDeclHandler handler) | |
1342 | { | |
1343 | entityDeclHandler = handler; | |
1344 | } | |
1345 | ||
1346 | void | |
1347 | XML_SetXmlDeclHandler(XML_Parser parser, | |
1348 | XML_XmlDeclHandler handler) { | |
1349 | xmlDeclHandler = handler; | |
1350 | } | |
1351 | ||
1352 | int | |
1353 | XML_SetParamEntityParsing(XML_Parser parser, | |
1354 | enum XML_ParamEntityParsing peParsing) | |
1355 | { | |
1356 | /* block after XML_Parse()/XML_ParseBuffer() has been called */ | |
1357 | if (parsing) | |
1358 | return 0; | |
1359 | #ifdef XML_DTD | |
1360 | paramEntityParsing = peParsing; | |
1361 | return 1; | |
1362 | #else | |
1363 | return peParsing == XML_PARAM_ENTITY_PARSING_NEVER; | |
1364 | #endif | |
1365 | } | |
1366 | ||
1367 | enum XML_Status | |
1368 | XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) | |
1369 | { | |
1370 | if (len == 0) { | |
1371 | if (!isFinal) | |
1372 | return XML_STATUS_OK; | |
1373 | positionPtr = bufferPtr; | |
1374 | errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0); | |
1375 | if (errorCode == XML_ERROR_NONE) | |
1376 | return XML_STATUS_OK; | |
1377 | eventEndPtr = eventPtr; | |
1378 | processor = errorProcessor; | |
1379 | return XML_STATUS_ERROR; | |
1380 | } | |
1381 | #ifndef XML_CONTEXT_BYTES | |
1382 | else if (bufferPtr == bufferEnd) { | |
1383 | const char *end; | |
1384 | int nLeftOver; | |
1385 | parseEndByteIndex += len; | |
1386 | positionPtr = s; | |
1387 | if (isFinal) { | |
1388 | errorCode = processor(parser, s, parseEndPtr = s + len, 0); | |
1389 | if (errorCode == XML_ERROR_NONE) | |
1390 | return XML_STATUS_OK; | |
1391 | eventEndPtr = eventPtr; | |
1392 | processor = errorProcessor; | |
1393 | return XML_STATUS_ERROR; | |
1394 | } | |
1395 | errorCode = processor(parser, s, parseEndPtr = s + len, &end); | |
1396 | if (errorCode != XML_ERROR_NONE) { | |
1397 | eventEndPtr = eventPtr; | |
1398 | processor = errorProcessor; | |
1399 | return XML_STATUS_ERROR; | |
1400 | } | |
1401 | XmlUpdatePosition(encoding, positionPtr, end, &position); | |
1402 | positionPtr = end; | |
1403 | nLeftOver = s + len - end; | |
1404 | if (nLeftOver) { | |
1405 | if (buffer == NULL || nLeftOver > bufferLim - buffer) { | |
1406 | /* FIXME avoid integer overflow */ | |
1407 | char *temp; | |
1408 | temp = (buffer == NULL | |
1409 | ? (char *)MALLOC(len * 2) | |
1410 | : (char *)REALLOC(buffer, len * 2)); | |
1411 | if (temp == NULL) { | |
1412 | errorCode = XML_ERROR_NO_MEMORY; | |
1413 | return XML_STATUS_ERROR; | |
1414 | } | |
1415 | buffer = temp; | |
1416 | if (!buffer) { | |
1417 | errorCode = XML_ERROR_NO_MEMORY; | |
1418 | eventPtr = eventEndPtr = NULL; | |
1419 | processor = errorProcessor; | |
1420 | return XML_STATUS_ERROR; | |
1421 | } | |
1422 | bufferLim = buffer + len * 2; | |
1423 | } | |
1424 | memcpy(buffer, end, nLeftOver); | |
1425 | bufferPtr = buffer; | |
1426 | bufferEnd = buffer + nLeftOver; | |
1427 | } | |
1428 | return XML_STATUS_OK; | |
1429 | } | |
1430 | #endif /* not defined XML_CONTEXT_BYTES */ | |
1431 | else { | |
1432 | void *buff = XML_GetBuffer(parser, len); | |
1433 | if (buff == NULL) | |
1434 | return XML_STATUS_ERROR; | |
1435 | else { | |
1436 | memcpy(buff, s, len); | |
1437 | return XML_ParseBuffer(parser, len, isFinal); | |
1438 | } | |
1439 | } | |
1440 | } | |
1441 | ||
1442 | enum XML_Status | |
1443 | XML_ParseBuffer(XML_Parser parser, int len, int isFinal) | |
1444 | { | |
1445 | const char *start = bufferPtr; | |
1446 | positionPtr = start; | |
1447 | bufferEnd += len; | |
1448 | parseEndByteIndex += len; | |
1449 | errorCode = processor(parser, start, parseEndPtr = bufferEnd, | |
1450 | isFinal ? (const char **)NULL : &bufferPtr); | |
1451 | if (errorCode == XML_ERROR_NONE) { | |
1452 | if (!isFinal) { | |
1453 | XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position); | |
1454 | positionPtr = bufferPtr; | |
1455 | } | |
1456 | return XML_STATUS_OK; | |
1457 | } | |
1458 | else { | |
1459 | eventEndPtr = eventPtr; | |
1460 | processor = errorProcessor; | |
1461 | return XML_STATUS_ERROR; | |
1462 | } | |
1463 | } | |
1464 | ||
1465 | void * | |
1466 | XML_GetBuffer(XML_Parser parser, int len) | |
1467 | { | |
1468 | if (len > bufferLim - bufferEnd) { | |
1469 | /* FIXME avoid integer overflow */ | |
1470 | int neededSize = len + (bufferEnd - bufferPtr); | |
1471 | #ifdef XML_CONTEXT_BYTES | |
1472 | int keep = bufferPtr - buffer; | |
1473 | ||
1474 | if (keep > XML_CONTEXT_BYTES) | |
1475 | keep = XML_CONTEXT_BYTES; | |
1476 | neededSize += keep; | |
1477 | #endif /* defined XML_CONTEXT_BYTES */ | |
1478 | if (neededSize <= bufferLim - buffer) { | |
1479 | #ifdef XML_CONTEXT_BYTES | |
1480 | if (keep < bufferPtr - buffer) { | |
1481 | int offset = (bufferPtr - buffer) - keep; | |
1482 | memmove(buffer, &buffer[offset], bufferEnd - bufferPtr + keep); | |
1483 | bufferEnd -= offset; | |
1484 | bufferPtr -= offset; | |
1485 | } | |
1486 | #else | |
1487 | memmove(buffer, bufferPtr, bufferEnd - bufferPtr); | |
1488 | bufferEnd = buffer + (bufferEnd - bufferPtr); | |
1489 | bufferPtr = buffer; | |
1490 | #endif /* not defined XML_CONTEXT_BYTES */ | |
1491 | } | |
1492 | else { | |
1493 | char *newBuf; | |
1494 | int bufferSize = bufferLim - bufferPtr; | |
1495 | if (bufferSize == 0) | |
1496 | bufferSize = INIT_BUFFER_SIZE; | |
1497 | do { | |
1498 | bufferSize *= 2; | |
1499 | } while (bufferSize < neededSize); | |
1500 | newBuf = (char *)MALLOC(bufferSize); | |
1501 | if (newBuf == 0) { | |
1502 | errorCode = XML_ERROR_NO_MEMORY; | |
1503 | return NULL; | |
1504 | } | |
1505 | bufferLim = newBuf + bufferSize; | |
1506 | #ifdef XML_CONTEXT_BYTES | |
1507 | if (bufferPtr) { | |
1508 | int keep = bufferPtr - buffer; | |
1509 | if (keep > XML_CONTEXT_BYTES) | |
1510 | keep = XML_CONTEXT_BYTES; | |
1511 | memcpy(newBuf, &bufferPtr[-keep], bufferEnd - bufferPtr + keep); | |
1512 | FREE(buffer); | |
1513 | buffer = newBuf; | |
1514 | bufferEnd = buffer + (bufferEnd - bufferPtr) + keep; | |
1515 | bufferPtr = buffer + keep; | |
1516 | } | |
1517 | else { | |
1518 | bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
1519 | bufferPtr = buffer = newBuf; | |
1520 | } | |
1521 | #else | |
1522 | if (bufferPtr) { | |
1523 | memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr); | |
1524 | FREE(buffer); | |
1525 | } | |
1526 | bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
1527 | bufferPtr = buffer = newBuf; | |
1528 | #endif /* not defined XML_CONTEXT_BYTES */ | |
1529 | } | |
1530 | } | |
1531 | return bufferEnd; | |
1532 | } | |
1533 | ||
1534 | enum XML_Error | |
1535 | XML_GetErrorCode(XML_Parser parser) | |
1536 | { | |
1537 | return errorCode; | |
1538 | } | |
1539 | ||
1540 | long | |
1541 | XML_GetCurrentByteIndex(XML_Parser parser) | |
1542 | { | |
1543 | if (eventPtr) | |
1544 | return parseEndByteIndex - (parseEndPtr - eventPtr); | |
1545 | return -1; | |
1546 | } | |
1547 | ||
1548 | int | |
1549 | XML_GetCurrentByteCount(XML_Parser parser) | |
1550 | { | |
1551 | if (eventEndPtr && eventPtr) | |
1552 | return eventEndPtr - eventPtr; | |
1553 | return 0; | |
1554 | } | |
1555 | ||
1556 | const char * | |
1557 | XML_GetInputContext(XML_Parser parser, int *offset, int *size) | |
1558 | { | |
1559 | #ifdef XML_CONTEXT_BYTES | |
1560 | if (eventPtr && buffer) { | |
1561 | *offset = eventPtr - buffer; | |
1562 | *size = bufferEnd - buffer; | |
1563 | return buffer; | |
1564 | } | |
1565 | #endif /* defined XML_CONTEXT_BYTES */ | |
1566 | return (char *) 0; | |
1567 | } | |
1568 | ||
1569 | int | |
1570 | XML_GetCurrentLineNumber(XML_Parser parser) | |
1571 | { | |
1572 | if (eventPtr) { | |
1573 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
1574 | positionPtr = eventPtr; | |
1575 | } | |
1576 | return position.lineNumber + 1; | |
1577 | } | |
1578 | ||
1579 | int | |
1580 | XML_GetCurrentColumnNumber(XML_Parser parser) | |
1581 | { | |
1582 | if (eventPtr) { | |
1583 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
1584 | positionPtr = eventPtr; | |
1585 | } | |
1586 | return position.columnNumber; | |
1587 | } | |
1588 | ||
1589 | void | |
1590 | XML_FreeContentModel(XML_Parser parser, XML_Content *model) | |
1591 | { | |
1592 | FREE(model); | |
1593 | } | |
1594 | ||
1595 | void * | |
1596 | XML_MemMalloc(XML_Parser parser, size_t size) | |
1597 | { | |
1598 | return MALLOC(size); | |
1599 | } | |
1600 | ||
1601 | void * | |
1602 | XML_MemRealloc(XML_Parser parser, void *ptr, size_t size) | |
1603 | { | |
1604 | return REALLOC(ptr, size); | |
1605 | } | |
1606 | ||
1607 | void | |
1608 | XML_MemFree(XML_Parser parser, void *ptr) | |
1609 | { | |
1610 | FREE(ptr); | |
1611 | } | |
1612 | ||
1613 | void | |
1614 | XML_DefaultCurrent(XML_Parser parser) | |
1615 | { | |
1616 | if (defaultHandler) { | |
1617 | if (openInternalEntities) | |
1618 | reportDefault(parser, | |
1619 | internalEncoding, | |
1620 | openInternalEntities->internalEventPtr, | |
1621 | openInternalEntities->internalEventEndPtr); | |
1622 | else | |
1623 | reportDefault(parser, encoding, eventPtr, eventEndPtr); | |
1624 | } | |
1625 | } | |
1626 | ||
1627 | const XML_LChar * | |
1628 | XML_ErrorString(enum XML_Error code) | |
1629 | { | |
1630 | static const XML_LChar *message[] = { | |
1631 | 0, | |
1632 | XML_L("out of memory"), | |
1633 | XML_L("syntax error"), | |
1634 | XML_L("no element found"), | |
1635 | XML_L("not well-formed (invalid token)"), | |
1636 | XML_L("unclosed token"), | |
1637 | XML_L("partial character"), | |
1638 | XML_L("mismatched tag"), | |
1639 | XML_L("duplicate attribute"), | |
1640 | XML_L("junk after document element"), | |
1641 | XML_L("illegal parameter entity reference"), | |
1642 | XML_L("undefined entity"), | |
1643 | XML_L("recursive entity reference"), | |
1644 | XML_L("asynchronous entity"), | |
1645 | XML_L("reference to invalid character number"), | |
1646 | XML_L("reference to binary entity"), | |
1647 | XML_L("reference to external entity in attribute"), | |
1648 | XML_L("xml declaration not at start of external entity"), | |
1649 | XML_L("unknown encoding"), | |
1650 | XML_L("encoding specified in XML declaration is incorrect"), | |
1651 | XML_L("unclosed CDATA section"), | |
1652 | XML_L("error in processing external entity reference"), | |
1653 | XML_L("document is not standalone"), | |
1654 | XML_L("unexpected parser state - please send a bug report"), | |
1655 | XML_L("entity declared in parameter entity"), | |
1656 | XML_L("requested feature requires XML_DTD support in Expat"), | |
1657 | XML_L("cannot change setting once parsing has begun") | |
1658 | }; | |
1659 | if (code > 0 && code < sizeof(message)/sizeof(message[0])) | |
1660 | return message[code]; | |
1661 | return NULL; | |
1662 | } | |
1663 | ||
1664 | const XML_LChar * | |
1665 | XML_ExpatVersion(void) { | |
1666 | ||
1667 | /* V1 is used to string-ize the version number. However, it would | |
1668 | string-ize the actual version macro *names* unless we get them | |
1669 | substituted before being passed to V1. CPP is defined to expand | |
1670 | a macro, then rescan for more expansions. Thus, we use V2 to expand | |
1671 | the version macros, then CPP will expand the resulting V1() macro | |
1672 | with the correct numerals. */ | |
1673 | /* ### I'm assuming cpp is portable in this respect... */ | |
1674 | ||
1675 | #define V1(a,b,c) XML_L(#a)XML_L(".")XML_L(#b)XML_L(".")XML_L(#c) | |
1676 | #define V2(a,b,c) XML_L("expat_")V1(a,b,c) | |
1677 | ||
1678 | return V2(XML_MAJOR_VERSION, XML_MINOR_VERSION, XML_MICRO_VERSION); | |
1679 | ||
1680 | #undef V1 | |
1681 | #undef V2 | |
1682 | } | |
1683 | ||
1684 | XML_Expat_Version | |
1685 | XML_ExpatVersionInfo(void) | |
1686 | { | |
1687 | XML_Expat_Version version; | |
1688 | ||
1689 | version.major = XML_MAJOR_VERSION; | |
1690 | version.minor = XML_MINOR_VERSION; | |
1691 | version.micro = XML_MICRO_VERSION; | |
1692 | ||
1693 | return version; | |
1694 | } | |
1695 | ||
1696 | const XML_Feature * | |
1697 | XML_GetFeatureList(void) | |
1698 | { | |
1699 | static XML_Feature features[] = { | |
1700 | {XML_FEATURE_SIZEOF_XML_CHAR, XML_L("sizeof(XML_Char)")}, | |
1701 | {XML_FEATURE_SIZEOF_XML_LCHAR, XML_L("sizeof(XML_LChar)")}, | |
1702 | #ifdef XML_UNICODE | |
1703 | {XML_FEATURE_UNICODE, XML_L("XML_UNICODE")}, | |
1704 | #endif | |
1705 | #ifdef XML_UNICODE_WCHAR_T | |
1706 | {XML_FEATURE_UNICODE_WCHAR_T, XML_L("XML_UNICODE_WCHAR_T")}, | |
1707 | #endif | |
1708 | #ifdef XML_DTD | |
1709 | {XML_FEATURE_DTD, XML_L("XML_DTD")}, | |
1710 | #endif | |
1711 | #ifdef XML_CONTEXT_BYTES | |
1712 | {XML_FEATURE_CONTEXT_BYTES, XML_L("XML_CONTEXT_BYTES"), | |
1713 | XML_CONTEXT_BYTES}, | |
1714 | #endif | |
1715 | #ifdef XML_MIN_SIZE | |
1716 | {XML_FEATURE_MIN_SIZE, XML_L("XML_MIN_SIZE")}, | |
1717 | #endif | |
1718 | {XML_FEATURE_END, NULL} | |
1719 | }; | |
1720 | ||
1721 | features[0].value = sizeof(XML_Char); | |
1722 | features[1].value = sizeof(XML_LChar); | |
1723 | return features; | |
1724 | } | |
1725 | ||
1726 | /* Initially tag->rawName always points into the parse buffer; | |
1727 | for those TAG instances opened while the current parse buffer was | |
1728 | processed, and not yet closed, we need to store tag->rawName in a more | |
1729 | permanent location, since the parse buffer is about to be discarded. | |
1730 | */ | |
1731 | static XML_Bool | |
1732 | storeRawNames(XML_Parser parser) | |
1733 | { | |
1734 | TAG *tag = tagStack; | |
1735 | while (tag) { | |
1736 | int bufSize; | |
1737 | int nameLen = sizeof(XML_Char) * (tag->name.strLen + 1); | |
1738 | char *rawNameBuf = tag->buf + nameLen; | |
1739 | /* Stop if already stored. Since tagStack is a stack, we can stop | |
1740 | at the first entry that has already been copied; everything | |
1741 | below it in the stack is already been accounted for in a | |
1742 | previous call to this function. | |
1743 | */ | |
1744 | if (tag->rawName == rawNameBuf) | |
1745 | break; | |
1746 | /* For re-use purposes we need to ensure that the | |
1747 | size of tag->buf is a multiple of sizeof(XML_Char). | |
1748 | */ | |
1749 | bufSize = nameLen + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)); | |
1750 | if (bufSize > tag->bufEnd - tag->buf) { | |
1751 | char *temp = (char *)REALLOC(tag->buf, bufSize); | |
1752 | if (temp == NULL) | |
1753 | return XML_FALSE; | |
1754 | /* if tag->name.str points to tag->buf (only when namespace | |
1755 | processing is off) then we have to update it | |
1756 | */ | |
1757 | if (tag->name.str == (XML_Char *)tag->buf) | |
1758 | tag->name.str = (XML_Char *)temp; | |
1759 | /* if tag->name.localPart is set (when namespace processing is on) | |
1760 | then update it as well, since it will always point into tag->buf | |
1761 | */ | |
1762 | if (tag->name.localPart) | |
1763 | tag->name.localPart = (XML_Char *)temp + (tag->name.localPart - | |
1764 | (XML_Char *)tag->buf); | |
1765 | tag->buf = temp; | |
1766 | tag->bufEnd = temp + bufSize; | |
1767 | rawNameBuf = temp + nameLen; | |
1768 | } | |
1769 | memcpy(rawNameBuf, tag->rawName, tag->rawNameLength); | |
1770 | tag->rawName = rawNameBuf; | |
1771 | tag = tag->parent; | |
1772 | } | |
1773 | return XML_TRUE; | |
1774 | } | |
1775 | ||
1776 | static enum XML_Error PTRCALL | |
1777 | contentProcessor(XML_Parser parser, | |
1778 | const char *start, | |
1779 | const char *end, | |
1780 | const char **endPtr) | |
1781 | { | |
1782 | enum XML_Error result = | |
1783 | doContent(parser, 0, encoding, start, end, endPtr); | |
1784 | if (result != XML_ERROR_NONE) | |
1785 | return result; | |
1786 | if (!storeRawNames(parser)) | |
1787 | return XML_ERROR_NO_MEMORY; | |
1788 | return result; | |
1789 | } | |
1790 | ||
1791 | static enum XML_Error PTRCALL | |
1792 | externalEntityInitProcessor(XML_Parser parser, | |
1793 | const char *start, | |
1794 | const char *end, | |
1795 | const char **endPtr) | |
1796 | { | |
1797 | enum XML_Error result = initializeEncoding(parser); | |
1798 | if (result != XML_ERROR_NONE) | |
1799 | return result; | |
1800 | processor = externalEntityInitProcessor2; | |
1801 | return externalEntityInitProcessor2(parser, start, end, endPtr); | |
1802 | } | |
1803 | ||
1804 | static enum XML_Error PTRCALL | |
1805 | externalEntityInitProcessor2(XML_Parser parser, | |
1806 | const char *start, | |
1807 | const char *end, | |
1808 | const char **endPtr) | |
1809 | { | |
1810 | const char *next = start; /* XmlContentTok doesn't always set the last arg */ | |
1811 | int tok = XmlContentTok(encoding, start, end, &next); | |
1812 | switch (tok) { | |
1813 | case XML_TOK_BOM: | |
1814 | /* If we are at the end of the buffer, this would cause the next stage, | |
1815 | i.e. externalEntityInitProcessor3, to pass control directly to | |
1816 | doContent (by detecting XML_TOK_NONE) without processing any xml text | |
1817 | declaration - causing the error XML_ERROR_MISPLACED_XML_PI in doContent. | |
1818 | */ | |
1819 | if (next == end && endPtr) { | |
1820 | *endPtr = next; | |
1821 | return XML_ERROR_NONE; | |
1822 | } | |
1823 | start = next; | |
1824 | break; | |
1825 | case XML_TOK_PARTIAL: | |
1826 | if (endPtr) { | |
1827 | *endPtr = start; | |
1828 | return XML_ERROR_NONE; | |
1829 | } | |
1830 | eventPtr = start; | |
1831 | return XML_ERROR_UNCLOSED_TOKEN; | |
1832 | case XML_TOK_PARTIAL_CHAR: | |
1833 | if (endPtr) { | |
1834 | *endPtr = start; | |
1835 | return XML_ERROR_NONE; | |
1836 | } | |
1837 | eventPtr = start; | |
1838 | return XML_ERROR_PARTIAL_CHAR; | |
1839 | } | |
1840 | processor = externalEntityInitProcessor3; | |
1841 | return externalEntityInitProcessor3(parser, start, end, endPtr); | |
1842 | } | |
1843 | ||
1844 | static enum XML_Error PTRCALL | |
1845 | externalEntityInitProcessor3(XML_Parser parser, | |
1846 | const char *start, | |
1847 | const char *end, | |
1848 | const char **endPtr) | |
1849 | { | |
1850 | const char *next = start; /* XmlContentTok doesn't always set the last arg */ | |
1851 | int tok = XmlContentTok(encoding, start, end, &next); | |
1852 | switch (tok) { | |
1853 | case XML_TOK_XML_DECL: | |
1854 | { | |
1855 | enum XML_Error result = processXmlDecl(parser, 1, start, next); | |
1856 | if (result != XML_ERROR_NONE) | |
1857 | return result; | |
1858 | start = next; | |
1859 | } | |
1860 | break; | |
1861 | case XML_TOK_PARTIAL: | |
1862 | if (endPtr) { | |
1863 | *endPtr = start; | |
1864 | return XML_ERROR_NONE; | |
1865 | } | |
1866 | eventPtr = start; | |
1867 | return XML_ERROR_UNCLOSED_TOKEN; | |
1868 | case XML_TOK_PARTIAL_CHAR: | |
1869 | if (endPtr) { | |
1870 | *endPtr = start; | |
1871 | return XML_ERROR_NONE; | |
1872 | } | |
1873 | eventPtr = start; | |
1874 | return XML_ERROR_PARTIAL_CHAR; | |
1875 | } | |
1876 | processor = externalEntityContentProcessor; | |
1877 | tagLevel = 1; | |
1878 | return externalEntityContentProcessor(parser, start, end, endPtr); | |
1879 | } | |
1880 | ||
1881 | static enum XML_Error PTRCALL | |
1882 | externalEntityContentProcessor(XML_Parser parser, | |
1883 | const char *start, | |
1884 | const char *end, | |
1885 | const char **endPtr) | |
1886 | { | |
1887 | enum XML_Error result = | |
1888 | doContent(parser, 1, encoding, start, end, endPtr); | |
1889 | if (result != XML_ERROR_NONE) | |
1890 | return result; | |
1891 | if (!storeRawNames(parser)) | |
1892 | return XML_ERROR_NO_MEMORY; | |
1893 | return result; | |
1894 | } | |
1895 | ||
1896 | static enum XML_Error | |
1897 | doContent(XML_Parser parser, | |
1898 | int startTagLevel, | |
1899 | const ENCODING *enc, | |
1900 | const char *s, | |
1901 | const char *end, | |
1902 | const char **nextPtr) | |
1903 | { | |
1904 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
1905 | const char **eventPP; | |
1906 | const char **eventEndPP; | |
1907 | if (enc == encoding) { | |
1908 | eventPP = &eventPtr; | |
1909 | eventEndPP = &eventEndPtr; | |
1910 | } | |
1911 | else { | |
1912 | eventPP = &(openInternalEntities->internalEventPtr); | |
1913 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
1914 | } | |
1915 | *eventPP = s; | |
1916 | for (;;) { | |
1917 | const char *next = s; /* XmlContentTok doesn't always set the last arg */ | |
1918 | int tok = XmlContentTok(enc, s, end, &next); | |
1919 | *eventEndPP = next; | |
1920 | switch (tok) { | |
1921 | case XML_TOK_TRAILING_CR: | |
1922 | if (nextPtr) { | |
1923 | *nextPtr = s; | |
1924 | return XML_ERROR_NONE; | |
1925 | } | |
1926 | *eventEndPP = end; | |
1927 | if (characterDataHandler) { | |
1928 | XML_Char c = 0xA; | |
1929 | characterDataHandler(handlerArg, &c, 1); | |
1930 | } | |
1931 | else if (defaultHandler) | |
1932 | reportDefault(parser, enc, s, end); | |
1933 | if (startTagLevel == 0) | |
1934 | return XML_ERROR_NO_ELEMENTS; | |
1935 | if (tagLevel != startTagLevel) | |
1936 | return XML_ERROR_ASYNC_ENTITY; | |
1937 | return XML_ERROR_NONE; | |
1938 | case XML_TOK_NONE: | |
1939 | if (nextPtr) { | |
1940 | *nextPtr = s; | |
1941 | return XML_ERROR_NONE; | |
1942 | } | |
1943 | if (startTagLevel > 0) { | |
1944 | if (tagLevel != startTagLevel) | |
1945 | return XML_ERROR_ASYNC_ENTITY; | |
1946 | return XML_ERROR_NONE; | |
1947 | } | |
1948 | return XML_ERROR_NO_ELEMENTS; | |
1949 | case XML_TOK_INVALID: | |
1950 | *eventPP = next; | |
1951 | return XML_ERROR_INVALID_TOKEN; | |
1952 | case XML_TOK_PARTIAL: | |
1953 | if (nextPtr) { | |
1954 | *nextPtr = s; | |
1955 | return XML_ERROR_NONE; | |
1956 | } | |
1957 | return XML_ERROR_UNCLOSED_TOKEN; | |
1958 | case XML_TOK_PARTIAL_CHAR: | |
1959 | if (nextPtr) { | |
1960 | *nextPtr = s; | |
1961 | return XML_ERROR_NONE; | |
1962 | } | |
1963 | return XML_ERROR_PARTIAL_CHAR; | |
1964 | case XML_TOK_ENTITY_REF: | |
1965 | { | |
1966 | const XML_Char *name; | |
1967 | ENTITY *entity; | |
1968 | XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc, | |
1969 | s + enc->minBytesPerChar, | |
1970 | next - enc->minBytesPerChar); | |
1971 | if (ch) { | |
1972 | if (characterDataHandler) | |
1973 | characterDataHandler(handlerArg, &ch, 1); | |
1974 | else if (defaultHandler) | |
1975 | reportDefault(parser, enc, s, next); | |
1976 | break; | |
1977 | } | |
1978 | name = poolStoreString(&dtd->pool, enc, | |
1979 | s + enc->minBytesPerChar, | |
1980 | next - enc->minBytesPerChar); | |
1981 | if (!name) | |
1982 | return XML_ERROR_NO_MEMORY; | |
1983 | entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0); | |
1984 | poolDiscard(&dtd->pool); | |
1985 | /* First, determine if a check for an existing declaration is needed; | |
1986 | if yes, check that the entity exists, and that it is internal, | |
1987 | otherwise call the skipped entity or default handler. | |
1988 | */ | |
1989 | if (!dtd->hasParamEntityRefs || dtd->standalone) { | |
1990 | if (!entity) | |
1991 | return XML_ERROR_UNDEFINED_ENTITY; | |
1992 | else if (!entity->is_internal) | |
1993 | return XML_ERROR_ENTITY_DECLARED_IN_PE; | |
1994 | } | |
1995 | else if (!entity) { | |
1996 | if (skippedEntityHandler) | |
1997 | skippedEntityHandler(handlerArg, name, 0); | |
1998 | else if (defaultHandler) | |
1999 | reportDefault(parser, enc, s, next); | |
2000 | break; | |
2001 | } | |
2002 | if (entity->open) | |
2003 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
2004 | if (entity->notation) | |
2005 | return XML_ERROR_BINARY_ENTITY_REF; | |
2006 | if (entity->textPtr) { | |
2007 | enum XML_Error result; | |
2008 | OPEN_INTERNAL_ENTITY openEntity; | |
2009 | if (!defaultExpandInternalEntities) { | |
2010 | if (skippedEntityHandler) | |
2011 | skippedEntityHandler(handlerArg, entity->name, 0); | |
2012 | else if (defaultHandler) | |
2013 | reportDefault(parser, enc, s, next); | |
2014 | break; | |
2015 | } | |
2016 | entity->open = XML_TRUE; | |
2017 | openEntity.next = openInternalEntities; | |
2018 | openInternalEntities = &openEntity; | |
2019 | openEntity.entity = entity; | |
2020 | openEntity.internalEventPtr = NULL; | |
2021 | openEntity.internalEventEndPtr = NULL; | |
2022 | result = doContent(parser, | |
2023 | tagLevel, | |
2024 | internalEncoding, | |
2025 | (char *)entity->textPtr, | |
2026 | (char *)(entity->textPtr + entity->textLen), | |
2027 | 0); | |
2028 | entity->open = XML_FALSE; | |
2029 | openInternalEntities = openEntity.next; | |
2030 | if (result) | |
2031 | return result; | |
2032 | } | |
2033 | else if (externalEntityRefHandler) { | |
2034 | const XML_Char *context; | |
2035 | entity->open = XML_TRUE; | |
2036 | context = getContext(parser); | |
2037 | entity->open = XML_FALSE; | |
2038 | if (!context) | |
2039 | return XML_ERROR_NO_MEMORY; | |
2040 | if (!externalEntityRefHandler((XML_Parser)externalEntityRefHandlerArg, | |
2041 | context, | |
2042 | entity->base, | |
2043 | entity->systemId, | |
2044 | entity->publicId)) | |
2045 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
2046 | poolDiscard(&tempPool); | |
2047 | } | |
2048 | else if (defaultHandler) | |
2049 | reportDefault(parser, enc, s, next); | |
2050 | break; | |
2051 | } | |
2052 | case XML_TOK_START_TAG_NO_ATTS: | |
2053 | /* fall through */ | |
2054 | case XML_TOK_START_TAG_WITH_ATTS: | |
2055 | { | |
2056 | TAG *tag; | |
2057 | enum XML_Error result; | |
2058 | XML_Char *toPtr; | |
2059 | if (freeTagList) { | |
2060 | tag = freeTagList; | |
2061 | freeTagList = freeTagList->parent; | |
2062 | } | |
2063 | else { | |
2064 | tag = (TAG *)MALLOC(sizeof(TAG)); | |
2065 | if (!tag) | |
2066 | return XML_ERROR_NO_MEMORY; | |
2067 | tag->buf = (char *)MALLOC(INIT_TAG_BUF_SIZE); | |
2068 | if (!tag->buf) { | |
2069 | FREE(tag); | |
2070 | return XML_ERROR_NO_MEMORY; | |
2071 | } | |
2072 | tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; | |
2073 | } | |
2074 | tag->bindings = NULL; | |
2075 | tag->parent = tagStack; | |
2076 | tagStack = tag; | |
2077 | tag->name.localPart = NULL; | |
2078 | tag->name.prefix = NULL; | |
2079 | tag->rawName = s + enc->minBytesPerChar; | |
2080 | tag->rawNameLength = XmlNameLength(enc, tag->rawName); | |
2081 | ++tagLevel; | |
2082 | { | |
2083 | const char *rawNameEnd = tag->rawName + tag->rawNameLength; | |
2084 | const char *fromPtr = tag->rawName; | |
2085 | toPtr = (XML_Char *)tag->buf; | |
2086 | for (;;) { | |
2087 | int bufSize; | |
2088 | int convLen; | |
2089 | XmlConvert(enc, | |
2090 | &fromPtr, rawNameEnd, | |
2091 | (ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1); | |
2092 | convLen = toPtr - (XML_Char *)tag->buf; | |
2093 | if (fromPtr == rawNameEnd) { | |
2094 | tag->name.strLen = convLen; | |
2095 | break; | |
2096 | } | |
2097 | bufSize = (tag->bufEnd - tag->buf) << 1; | |
2098 | { | |
2099 | char *temp = (char *)REALLOC(tag->buf, bufSize); | |
2100 | if (temp == NULL) | |
2101 | return XML_ERROR_NO_MEMORY; | |
2102 | tag->buf = temp; | |
2103 | tag->bufEnd = temp + bufSize; | |
2104 | toPtr = (XML_Char *)temp + convLen; | |
2105 | } | |
2106 | } | |
2107 | } | |
2108 | tag->name.str = (XML_Char *)tag->buf; | |
2109 | *toPtr = XML_T('\0'); | |
2110 | result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings)); | |
2111 | if (result) | |
2112 | return result; | |
2113 | if (startElementHandler) | |
2114 | startElementHandler(handlerArg, tag->name.str, | |
2115 | (const XML_Char **)atts); | |
2116 | else if (defaultHandler) | |
2117 | reportDefault(parser, enc, s, next); | |
2118 | poolClear(&tempPool); | |
2119 | break; | |
2120 | } | |
2121 | case XML_TOK_EMPTY_ELEMENT_NO_ATTS: | |
2122 | /* fall through */ | |
2123 | case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: | |
2124 | { | |
2125 | const char *rawName = s + enc->minBytesPerChar; | |
2126 | enum XML_Error result; | |
2127 | BINDING *bindings = NULL; | |
2128 | XML_Bool noElmHandlers = XML_TRUE; | |
2129 | TAG_NAME name; | |
2130 | name.str = poolStoreString(&tempPool, enc, rawName, | |
2131 | rawName + XmlNameLength(enc, rawName)); | |
2132 | if (!name.str) | |
2133 | return XML_ERROR_NO_MEMORY; | |
2134 | poolFinish(&tempPool); | |
2135 | result = storeAtts(parser, enc, s, &name, &bindings); | |
2136 | if (result) | |
2137 | return result; | |
2138 | poolFinish(&tempPool); | |
2139 | if (startElementHandler) { | |
2140 | startElementHandler(handlerArg, name.str, (const XML_Char **)atts); | |
2141 | noElmHandlers = XML_FALSE; | |
2142 | } | |
2143 | if (endElementHandler) { | |
2144 | if (startElementHandler) | |
2145 | *eventPP = *eventEndPP; | |
2146 | endElementHandler(handlerArg, name.str); | |
2147 | noElmHandlers = XML_FALSE; | |
2148 | } | |
2149 | if (noElmHandlers && defaultHandler) | |
2150 | reportDefault(parser, enc, s, next); | |
2151 | poolClear(&tempPool); | |
2152 | while (bindings) { | |
2153 | BINDING *b = bindings; | |
2154 | if (endNamespaceDeclHandler) | |
2155 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
2156 | bindings = bindings->nextTagBinding; | |
2157 | b->nextTagBinding = freeBindingList; | |
2158 | freeBindingList = b; | |
2159 | b->prefix->binding = b->prevPrefixBinding; | |
2160 | } | |
2161 | } | |
2162 | if (tagLevel == 0) | |
2163 | return epilogProcessor(parser, next, end, nextPtr); | |
2164 | break; | |
2165 | case XML_TOK_END_TAG: | |
2166 | if (tagLevel == startTagLevel) | |
2167 | return XML_ERROR_ASYNC_ENTITY; | |
2168 | else { | |
2169 | int len; | |
2170 | const char *rawName; | |
2171 | TAG *tag = tagStack; | |
2172 | tagStack = tag->parent; | |
2173 | tag->parent = freeTagList; | |
2174 | freeTagList = tag; | |
2175 | rawName = s + enc->minBytesPerChar*2; | |
2176 | len = XmlNameLength(enc, rawName); | |
2177 | if (len != tag->rawNameLength | |
2178 | || memcmp(tag->rawName, rawName, len) != 0) { | |
2179 | *eventPP = rawName; | |
2180 | return XML_ERROR_TAG_MISMATCH; | |
2181 | } | |
2182 | --tagLevel; | |
2183 | if (endElementHandler) { | |
2184 | const XML_Char *localPart; | |
2185 | const XML_Char *prefix; | |
2186 | XML_Char *uri; | |
2187 | localPart = tag->name.localPart; | |
2188 | if (ns && localPart) { | |
2189 | /* localPart and prefix may have been overwritten in | |
2190 | tag->name.str, since this points to the binding->uri | |
2191 | buffer which gets re-used; so we have to add them again | |
2192 | */ | |
2193 | uri = (XML_Char *)tag->name.str + tag->name.uriLen; | |
2194 | /* don't need to check for space - already done in storeAtts() */ | |
2195 | while (*localPart) *uri++ = *localPart++; | |
2196 | prefix = (XML_Char *)tag->name.prefix; | |
2197 | if (ns_triplets && prefix) { | |
2198 | *uri++ = namespaceSeparator; | |
2199 | while (*prefix) *uri++ = *prefix++; | |
2200 | } | |
2201 | *uri = XML_T('\0'); | |
2202 | } | |
2203 | endElementHandler(handlerArg, tag->name.str); | |
2204 | } | |
2205 | else if (defaultHandler) | |
2206 | reportDefault(parser, enc, s, next); | |
2207 | while (tag->bindings) { | |
2208 | BINDING *b = tag->bindings; | |
2209 | if (endNamespaceDeclHandler) | |
2210 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
2211 | tag->bindings = tag->bindings->nextTagBinding; | |
2212 | b->nextTagBinding = freeBindingList; | |
2213 | freeBindingList = b; | |
2214 | b->prefix->binding = b->prevPrefixBinding; | |
2215 | } | |
2216 | if (tagLevel == 0) | |
2217 | return epilogProcessor(parser, next, end, nextPtr); | |
2218 | } | |
2219 | break; | |
2220 | case XML_TOK_CHAR_REF: | |
2221 | { | |
2222 | int n = XmlCharRefNumber(enc, s); | |
2223 | if (n < 0) | |
2224 | return XML_ERROR_BAD_CHAR_REF; | |
2225 | if (characterDataHandler) { | |
2226 | XML_Char buf[XML_ENCODE_MAX]; | |
2227 | characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf)); | |
2228 | } | |
2229 | else if (defaultHandler) | |
2230 | reportDefault(parser, enc, s, next); | |
2231 | } | |
2232 | break; | |
2233 | case XML_TOK_XML_DECL: | |
2234 | return XML_ERROR_MISPLACED_XML_PI; | |
2235 | case XML_TOK_DATA_NEWLINE: | |
2236 | if (characterDataHandler) { | |
2237 | XML_Char c = 0xA; | |
2238 | characterDataHandler(handlerArg, &c, 1); | |
2239 | } | |
2240 | else if (defaultHandler) | |
2241 | reportDefault(parser, enc, s, next); | |
2242 | break; | |
2243 | case XML_TOK_CDATA_SECT_OPEN: | |
2244 | { | |
2245 | enum XML_Error result; | |
2246 | if (startCdataSectionHandler) | |
2247 | startCdataSectionHandler(handlerArg); | |
2248 | #if 0 | |
2249 | /* Suppose you doing a transformation on a document that involves | |
2250 | changing only the character data. You set up a defaultHandler | |
2251 | and a characterDataHandler. The defaultHandler simply copies | |
2252 | characters through. The characterDataHandler does the | |
2253 | transformation and writes the characters out escaping them as | |
2254 | necessary. This case will fail to work if we leave out the | |
2255 | following two lines (because & and < inside CDATA sections will | |
2256 | be incorrectly escaped). | |
2257 | ||
2258 | However, now we have a start/endCdataSectionHandler, so it seems | |
2259 | easier to let the user deal with this. | |
2260 | */ | |
2261 | else if (characterDataHandler) | |
2262 | characterDataHandler(handlerArg, dataBuf, 0); | |
2263 | #endif | |
2264 | else if (defaultHandler) | |
2265 | reportDefault(parser, enc, s, next); | |
2266 | result = doCdataSection(parser, enc, &next, end, nextPtr); | |
2267 | if (!next) { | |
2268 | processor = cdataSectionProcessor; | |
2269 | return result; | |
2270 | } | |
2271 | } | |
2272 | break; | |
2273 | case XML_TOK_TRAILING_RSQB: | |
2274 | if (nextPtr) { | |
2275 | *nextPtr = s; | |
2276 | return XML_ERROR_NONE; | |
2277 | } | |
2278 | if (characterDataHandler) { | |
2279 | if (MUST_CONVERT(enc, s)) { | |
2280 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2281 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
2282 | characterDataHandler(handlerArg, dataBuf, | |
2283 | dataPtr - (ICHAR *)dataBuf); | |
2284 | } | |
2285 | else | |
2286 | characterDataHandler(handlerArg, | |
2287 | (XML_Char *)s, | |
2288 | (XML_Char *)end - (XML_Char *)s); | |
2289 | } | |
2290 | else if (defaultHandler) | |
2291 | reportDefault(parser, enc, s, end); | |
2292 | if (startTagLevel == 0) { | |
2293 | *eventPP = end; | |
2294 | return XML_ERROR_NO_ELEMENTS; | |
2295 | } | |
2296 | if (tagLevel != startTagLevel) { | |
2297 | *eventPP = end; | |
2298 | return XML_ERROR_ASYNC_ENTITY; | |
2299 | } | |
2300 | return XML_ERROR_NONE; | |
2301 | case XML_TOK_DATA_CHARS: | |
2302 | if (characterDataHandler) { | |
2303 | if (MUST_CONVERT(enc, s)) { | |
2304 | for (;;) { | |
2305 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2306 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
2307 | *eventEndPP = s; | |
2308 | characterDataHandler(handlerArg, dataBuf, | |
2309 | dataPtr - (ICHAR *)dataBuf); | |
2310 | if (s == next) | |
2311 | break; | |
2312 | *eventPP = s; | |
2313 | } | |
2314 | } | |
2315 | else | |
2316 | characterDataHandler(handlerArg, | |
2317 | (XML_Char *)s, | |
2318 | (XML_Char *)next - (XML_Char *)s); | |
2319 | } | |
2320 | else if (defaultHandler) | |
2321 | reportDefault(parser, enc, s, next); | |
2322 | break; | |
2323 | case XML_TOK_PI: | |
2324 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
2325 | return XML_ERROR_NO_MEMORY; | |
2326 | break; | |
2327 | case XML_TOK_COMMENT: | |
2328 | if (!reportComment(parser, enc, s, next)) | |
2329 | return XML_ERROR_NO_MEMORY; | |
2330 | break; | |
2331 | default: | |
2332 | if (defaultHandler) | |
2333 | reportDefault(parser, enc, s, next); | |
2334 | break; | |
2335 | } | |
2336 | *eventPP = s = next; | |
2337 | } | |
2338 | /* not reached */ | |
2339 | } | |
2340 | ||
2341 | /* Precondition: all arguments must be non-NULL; | |
2342 | Purpose: | |
2343 | - normalize attributes | |
2344 | - check attributes for well-formedness | |
2345 | - generate namespace aware attribute names (URI, prefix) | |
2346 | - build list of attributes for startElementHandler | |
2347 | - default attributes | |
2348 | - process namespace declarations (check and report them) | |
2349 | - generate namespace aware element name (URI, prefix) | |
2350 | */ | |
2351 | static enum XML_Error | |
2352 | storeAtts(XML_Parser parser, const ENCODING *enc, | |
2353 | const char *attStr, TAG_NAME *tagNamePtr, | |
2354 | BINDING **bindingsPtr) | |
2355 | { | |
2356 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
2357 | ELEMENT_TYPE *elementType = NULL; | |
2358 | int nDefaultAtts = 0; | |
2359 | const XML_Char **appAtts; /* the attribute list for the application */ | |
2360 | int attIndex = 0; | |
2361 | int prefixLen; | |
2362 | int i; | |
2363 | int n; | |
2364 | XML_Char *uri; | |
2365 | int nPrefixes = 0; | |
2366 | BINDING *binding; | |
2367 | const XML_Char *localPart; | |
2368 | ||
2369 | /* lookup the element type name */ | |
2370 | elementType = (ELEMENT_TYPE *)lookup(&dtd->elementTypes, tagNamePtr->str,0); | |
2371 | if (!elementType) { | |
2372 | const XML_Char *name = poolCopyString(&dtd->pool, tagNamePtr->str); | |
2373 | if (!name) | |
2374 | return XML_ERROR_NO_MEMORY; | |
2375 | elementType = (ELEMENT_TYPE *)lookup(&dtd->elementTypes, name, | |
2376 | sizeof(ELEMENT_TYPE)); | |
2377 | if (!elementType) | |
2378 | return XML_ERROR_NO_MEMORY; | |
2379 | if (ns && !setElementTypePrefix(parser, elementType)) | |
2380 | return XML_ERROR_NO_MEMORY; | |
2381 | } | |
2382 | nDefaultAtts = elementType->nDefaultAtts; | |
2383 | ||
2384 | /* get the attributes from the tokenizer */ | |
2385 | n = XmlGetAttributes(enc, attStr, attsSize, atts); | |
2386 | if (n + nDefaultAtts > attsSize) { | |
2387 | int oldAttsSize = attsSize; | |
2388 | ATTRIBUTE *temp; | |
2389 | attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; | |
2390 | temp = (ATTRIBUTE *)REALLOC((void *)atts, attsSize * sizeof(ATTRIBUTE)); | |
2391 | if (temp == NULL) | |
2392 | return XML_ERROR_NO_MEMORY; | |
2393 | atts = temp; | |
2394 | if (n > oldAttsSize) | |
2395 | XmlGetAttributes(enc, attStr, n, atts); | |
2396 | } | |
2397 | ||
2398 | appAtts = (const XML_Char **)atts; | |
2399 | for (i = 0; i < n; i++) { | |
2400 | /* add the name and value to the attribute list */ | |
2401 | ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name, | |
2402 | atts[i].name | |
2403 | + XmlNameLength(enc, atts[i].name)); | |
2404 | if (!attId) | |
2405 | return XML_ERROR_NO_MEMORY; | |
2406 | /* detect duplicate attributes */ | |
2407 | if ((attId->name)[-1]) { | |
2408 | if (enc == encoding) | |
2409 | eventPtr = atts[i].name; | |
2410 | return XML_ERROR_DUPLICATE_ATTRIBUTE; | |
2411 | } | |
2412 | (attId->name)[-1] = 1; | |
2413 | appAtts[attIndex++] = attId->name; | |
2414 | if (!atts[i].normalized) { | |
2415 | enum XML_Error result; | |
2416 | XML_Bool isCdata = XML_TRUE; | |
2417 | ||
2418 | /* figure out whether declared as other than CDATA */ | |
2419 | if (attId->maybeTokenized) { | |
2420 | int j; | |
2421 | for (j = 0; j < nDefaultAtts; j++) { | |
2422 | if (attId == elementType->defaultAtts[j].id) { | |
2423 | isCdata = elementType->defaultAtts[j].isCdata; | |
2424 | break; | |
2425 | } | |
2426 | } | |
2427 | } | |
2428 | ||
2429 | /* normalize the attribute value */ | |
2430 | result = storeAttributeValue(parser, enc, isCdata, | |
2431 | atts[i].valuePtr, atts[i].valueEnd, | |
2432 | &tempPool); | |
2433 | if (result) | |
2434 | return result; | |
2435 | appAtts[attIndex] = poolStart(&tempPool); | |
2436 | poolFinish(&tempPool); | |
2437 | } | |
2438 | else { | |
2439 | /* the value did not need normalizing */ | |
2440 | appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr, | |
2441 | atts[i].valueEnd); | |
2442 | if (appAtts[attIndex] == 0) | |
2443 | return XML_ERROR_NO_MEMORY; | |
2444 | poolFinish(&tempPool); | |
2445 | } | |
2446 | /* handle prefixed attribute names */ | |
2447 | if (attId->prefix) { | |
2448 | if (attId->xmlns) { | |
2449 | /* deal with namespace declarations here */ | |
2450 | enum XML_Error result = addBinding(parser, attId->prefix, attId, | |
2451 | appAtts[attIndex], bindingsPtr); | |
2452 | if (result) | |
2453 | return result; | |
2454 | --attIndex; | |
2455 | } | |
2456 | else { | |
2457 | /* deal with other prefixed names later */ | |
2458 | attIndex++; | |
2459 | nPrefixes++; | |
2460 | (attId->name)[-1] = 2; | |
2461 | } | |
2462 | } | |
2463 | else | |
2464 | attIndex++; | |
2465 | } | |
2466 | ||
2467 | /* set-up for XML_GetSpecifiedAttributeCount and XML_GetIdAttributeIndex */ | |
2468 | nSpecifiedAtts = attIndex; | |
2469 | if (elementType->idAtt && (elementType->idAtt->name)[-1]) { | |
2470 | for (i = 0; i < attIndex; i += 2) | |
2471 | if (appAtts[i] == elementType->idAtt->name) { | |
2472 | idAttIndex = i; | |
2473 | break; | |
2474 | } | |
2475 | } | |
2476 | else | |
2477 | idAttIndex = -1; | |
2478 | ||
2479 | /* do attribute defaulting */ | |
2480 | for (i = 0; i < nDefaultAtts; i++) { | |
2481 | const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + i; | |
2482 | if (!(da->id->name)[-1] && da->value) { | |
2483 | if (da->id->prefix) { | |
2484 | if (da->id->xmlns) { | |
2485 | enum XML_Error result = addBinding(parser, da->id->prefix, da->id, | |
2486 | da->value, bindingsPtr); | |
2487 | if (result) | |
2488 | return result; | |
2489 | } | |
2490 | else { | |
2491 | (da->id->name)[-1] = 2; | |
2492 | nPrefixes++; | |
2493 | appAtts[attIndex++] = da->id->name; | |
2494 | appAtts[attIndex++] = da->value; | |
2495 | } | |
2496 | } | |
2497 | else { | |
2498 | (da->id->name)[-1] = 1; | |
2499 | appAtts[attIndex++] = da->id->name; | |
2500 | appAtts[attIndex++] = da->value; | |
2501 | } | |
2502 | } | |
2503 | } | |
2504 | appAtts[attIndex] = 0; | |
2505 | ||
2506 | i = 0; | |
2507 | if (nPrefixes) { | |
2508 | /* expand prefixed attribute names */ | |
2509 | for (; i < attIndex; i += 2) { | |
2510 | if (appAtts[i][-1] == 2) { | |
2511 | ATTRIBUTE_ID *id; | |
2512 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
2513 | id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, appAtts[i], 0); | |
2514 | if (id->prefix->binding) { | |
2515 | int j; | |
2516 | const BINDING *b = id->prefix->binding; | |
2517 | const XML_Char *s = appAtts[i]; | |
2518 | for (j = 0; j < b->uriLen; j++) { | |
2519 | if (!poolAppendChar(&tempPool, b->uri[j])) | |
2520 | return XML_ERROR_NO_MEMORY; | |
2521 | } | |
2522 | while (*s++ != XML_T(':')) | |
2523 | ; | |
2524 | do { | |
2525 | if (!poolAppendChar(&tempPool, *s)) | |
2526 | return XML_ERROR_NO_MEMORY; | |
2527 | } while (*s++); | |
2528 | if (ns_triplets) { | |
2529 | tempPool.ptr[-1] = namespaceSeparator; | |
2530 | s = b->prefix->name; | |
2531 | do { | |
2532 | if (!poolAppendChar(&tempPool, *s)) | |
2533 | return XML_ERROR_NO_MEMORY; | |
2534 | } while (*s++); | |
2535 | } | |
2536 | ||
2537 | appAtts[i] = poolStart(&tempPool); | |
2538 | poolFinish(&tempPool); | |
2539 | } | |
2540 | if (!--nPrefixes) | |
2541 | break; | |
2542 | } | |
2543 | else | |
2544 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
2545 | } | |
2546 | } | |
2547 | /* clear the flags that say whether attributes were specified */ | |
2548 | for (; i < attIndex; i += 2) | |
2549 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
2550 | for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) | |
2551 | binding->attId->name[-1] = 0; | |
2552 | ||
2553 | /* expand the element type name */ | |
2554 | if (elementType->prefix) { | |
2555 | binding = elementType->prefix->binding; | |
2556 | if (!binding) | |
2557 | return XML_ERROR_NONE; | |
2558 | localPart = tagNamePtr->str; | |
2559 | while (*localPart++ != XML_T(':')) | |
2560 | ; | |
2561 | } | |
2562 | else if (dtd->defaultPrefix.binding) { | |
2563 | binding = dtd->defaultPrefix.binding; | |
2564 | localPart = tagNamePtr->str; | |
2565 | } | |
2566 | else | |
2567 | return XML_ERROR_NONE; | |
2568 | prefixLen = 0; | |
2569 | if (ns && ns_triplets && binding->prefix->name) { | |
2570 | for (; binding->prefix->name[prefixLen++];) | |
2571 | ; | |
2572 | } | |
2573 | tagNamePtr->localPart = localPart; | |
2574 | tagNamePtr->uriLen = binding->uriLen; | |
2575 | tagNamePtr->prefix = binding->prefix->name; | |
2576 | tagNamePtr->prefixLen = prefixLen; | |
2577 | for (i = 0; localPart[i++];) | |
2578 | ; | |
2579 | n = i + binding->uriLen + prefixLen; | |
2580 | if (n > binding->uriAlloc) { | |
2581 | TAG *p; | |
2582 | uri = (XML_Char *)MALLOC((n + EXPAND_SPARE) * sizeof(XML_Char)); | |
2583 | if (!uri) | |
2584 | return XML_ERROR_NO_MEMORY; | |
2585 | binding->uriAlloc = n + EXPAND_SPARE; | |
2586 | memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char)); | |
2587 | for (p = tagStack; p; p = p->parent) | |
2588 | if (p->name.str == binding->uri) | |
2589 | p->name.str = uri; | |
2590 | FREE(binding->uri); | |
2591 | binding->uri = uri; | |
2592 | } | |
2593 | uri = binding->uri + binding->uriLen; | |
2594 | memcpy(uri, localPart, i * sizeof(XML_Char)); | |
2595 | if (prefixLen) { | |
2596 | uri = uri + (i - 1); | |
2597 | if (namespaceSeparator) { *(uri) = namespaceSeparator; } | |
2598 | memcpy(uri + 1, binding->prefix->name, prefixLen * sizeof(XML_Char)); | |
2599 | } | |
2600 | tagNamePtr->str = binding->uri; | |
2601 | return XML_ERROR_NONE; | |
2602 | } | |
2603 | ||
2604 | /* addBinding() overwrites the value of prefix->binding without checking. | |
2605 | Therefore one must keep track of the old value outside of addBinding(). | |
2606 | */ | |
2607 | static enum XML_Error | |
2608 | addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, | |
2609 | const XML_Char *uri, BINDING **bindingsPtr) | |
2610 | { | |
2611 | BINDING *b; | |
2612 | int len; | |
2613 | ||
2614 | /* empty string is only valid when there is no prefix per XML NS 1.0 */ | |
2615 | if (*uri == XML_T('\0') && prefix->name) | |
2616 | return XML_ERROR_SYNTAX; | |
2617 | ||
2618 | for (len = 0; uri[len]; len++) | |
2619 | ; | |
2620 | if (namespaceSeparator) | |
2621 | len++; | |
2622 | if (freeBindingList) { | |
2623 | b = freeBindingList; | |
2624 | if (len > b->uriAlloc) { | |
2625 | XML_Char *temp = (XML_Char *)REALLOC(b->uri, | |
2626 | sizeof(XML_Char) * (len + EXPAND_SPARE)); | |
2627 | if (temp == NULL) | |
2628 | return XML_ERROR_NO_MEMORY; | |
2629 | b->uri = temp; | |
2630 | b->uriAlloc = len + EXPAND_SPARE; | |
2631 | } | |
2632 | freeBindingList = b->nextTagBinding; | |
2633 | } | |
2634 | else { | |
2635 | b = (BINDING *)MALLOC(sizeof(BINDING)); | |
2636 | if (!b) | |
2637 | return XML_ERROR_NO_MEMORY; | |
2638 | b->uri = (XML_Char *)MALLOC(sizeof(XML_Char) * (len + EXPAND_SPARE)); | |
2639 | if (!b->uri) { | |
2640 | FREE(b); | |
2641 | return XML_ERROR_NO_MEMORY; | |
2642 | } | |
2643 | b->uriAlloc = len + EXPAND_SPARE; | |
2644 | } | |
2645 | b->uriLen = len; | |
2646 | memcpy(b->uri, uri, len * sizeof(XML_Char)); | |
2647 | if (namespaceSeparator) | |
2648 | b->uri[len - 1] = namespaceSeparator; | |
2649 | b->prefix = prefix; | |
2650 | b->attId = attId; | |
2651 | b->prevPrefixBinding = prefix->binding; | |
2652 | if (*uri == XML_T('\0') && prefix == &_dtd->defaultPrefix) | |
2653 | prefix->binding = NULL; | |
2654 | else | |
2655 | prefix->binding = b; | |
2656 | b->nextTagBinding = *bindingsPtr; | |
2657 | *bindingsPtr = b; | |
2658 | if (startNamespaceDeclHandler) | |
2659 | startNamespaceDeclHandler(handlerArg, prefix->name, | |
2660 | prefix->binding ? uri : 0); | |
2661 | return XML_ERROR_NONE; | |
2662 | } | |
2663 | ||
2664 | /* The idea here is to avoid using stack for each CDATA section when | |
2665 | the whole file is parsed with one call. | |
2666 | */ | |
2667 | static enum XML_Error PTRCALL | |
2668 | cdataSectionProcessor(XML_Parser parser, | |
2669 | const char *start, | |
2670 | const char *end, | |
2671 | const char **endPtr) | |
2672 | { | |
2673 | enum XML_Error result = doCdataSection(parser, encoding, &start, | |
2674 | end, endPtr); | |
2675 | if (start) { | |
2676 | if (parentParser) { /* we are parsing an external entity */ | |
2677 | processor = externalEntityContentProcessor; | |
2678 | return externalEntityContentProcessor(parser, start, end, endPtr); | |
2679 | } | |
2680 | else { | |
2681 | processor = contentProcessor; | |
2682 | return contentProcessor(parser, start, end, endPtr); | |
2683 | } | |
2684 | } | |
2685 | return result; | |
2686 | } | |
2687 | ||
2688 | /* startPtr gets set to non-null is the section is closed, and to null if | |
2689 | the section is not yet closed. | |
2690 | */ | |
2691 | static enum XML_Error | |
2692 | doCdataSection(XML_Parser parser, | |
2693 | const ENCODING *enc, | |
2694 | const char **startPtr, | |
2695 | const char *end, | |
2696 | const char **nextPtr) | |
2697 | { | |
2698 | const char *s = *startPtr; | |
2699 | const char **eventPP; | |
2700 | const char **eventEndPP; | |
2701 | if (enc == encoding) { | |
2702 | eventPP = &eventPtr; | |
2703 | *eventPP = s; | |
2704 | eventEndPP = &eventEndPtr; | |
2705 | } | |
2706 | else { | |
2707 | eventPP = &(openInternalEntities->internalEventPtr); | |
2708 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2709 | } | |
2710 | *eventPP = s; | |
2711 | *startPtr = NULL; | |
2712 | for (;;) { | |
2713 | const char *next; | |
2714 | int tok = XmlCdataSectionTok(enc, s, end, &next); | |
2715 | *eventEndPP = next; | |
2716 | switch (tok) { | |
2717 | case XML_TOK_CDATA_SECT_CLOSE: | |
2718 | if (endCdataSectionHandler) | |
2719 | endCdataSectionHandler(handlerArg); | |
2720 | #if 0 | |
2721 | /* see comment under XML_TOK_CDATA_SECT_OPEN */ | |
2722 | else if (characterDataHandler) | |
2723 | characterDataHandler(handlerArg, dataBuf, 0); | |
2724 | #endif | |
2725 | else if (defaultHandler) | |
2726 | reportDefault(parser, enc, s, next); | |
2727 | *startPtr = next; | |
2728 | return XML_ERROR_NONE; | |
2729 | case XML_TOK_DATA_NEWLINE: | |
2730 | if (characterDataHandler) { | |
2731 | XML_Char c = 0xA; | |
2732 | characterDataHandler(handlerArg, &c, 1); | |
2733 | } | |
2734 | else if (defaultHandler) | |
2735 | reportDefault(parser, enc, s, next); | |
2736 | break; | |
2737 | case XML_TOK_DATA_CHARS: | |
2738 | if (characterDataHandler) { | |
2739 | if (MUST_CONVERT(enc, s)) { | |
2740 | for (;;) { | |
2741 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2742 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
2743 | *eventEndPP = next; | |
2744 | characterDataHandler(handlerArg, dataBuf, | |
2745 | dataPtr - (ICHAR *)dataBuf); | |
2746 | if (s == next) | |
2747 | break; | |
2748 | *eventPP = s; | |
2749 | } | |
2750 | } | |
2751 | else | |
2752 | characterDataHandler(handlerArg, | |
2753 | (XML_Char *)s, | |
2754 | (XML_Char *)next - (XML_Char *)s); | |
2755 | } | |
2756 | else if (defaultHandler) | |
2757 | reportDefault(parser, enc, s, next); | |
2758 | break; | |
2759 | case XML_TOK_INVALID: | |
2760 | *eventPP = next; | |
2761 | return XML_ERROR_INVALID_TOKEN; | |
2762 | case XML_TOK_PARTIAL_CHAR: | |
2763 | if (nextPtr) { | |
2764 | *nextPtr = s; | |
2765 | return XML_ERROR_NONE; | |
2766 | } | |
2767 | return XML_ERROR_PARTIAL_CHAR; | |
2768 | case XML_TOK_PARTIAL: | |
2769 | case XML_TOK_NONE: | |
2770 | if (nextPtr) { | |
2771 | *nextPtr = s; | |
2772 | return XML_ERROR_NONE; | |
2773 | } | |
2774 | return XML_ERROR_UNCLOSED_CDATA_SECTION; | |
2775 | default: | |
2776 | *eventPP = next; | |
2777 | return XML_ERROR_UNEXPECTED_STATE; | |
2778 | } | |
2779 | *eventPP = s = next; | |
2780 | } | |
2781 | /* not reached */ | |
2782 | } | |
2783 | ||
2784 | #ifdef XML_DTD | |
2785 | ||
2786 | /* The idea here is to avoid using stack for each IGNORE section when | |
2787 | the whole file is parsed with one call. | |
2788 | */ | |
2789 | static enum XML_Error PTRCALL | |
2790 | ignoreSectionProcessor(XML_Parser parser, | |
2791 | const char *start, | |
2792 | const char *end, | |
2793 | const char **endPtr) | |
2794 | { | |
2795 | enum XML_Error result = doIgnoreSection(parser, encoding, &start, | |
2796 | end, endPtr); | |
2797 | if (start) { | |
2798 | processor = prologProcessor; | |
2799 | return prologProcessor(parser, start, end, endPtr); | |
2800 | } | |
2801 | return result; | |
2802 | } | |
2803 | ||
2804 | /* startPtr gets set to non-null is the section is closed, and to null | |
2805 | if the section is not yet closed. | |
2806 | */ | |
2807 | static enum XML_Error | |
2808 | doIgnoreSection(XML_Parser parser, | |
2809 | const ENCODING *enc, | |
2810 | const char **startPtr, | |
2811 | const char *end, | |
2812 | const char **nextPtr) | |
2813 | { | |
2814 | const char *next; | |
2815 | int tok; | |
2816 | const char *s = *startPtr; | |
2817 | const char **eventPP; | |
2818 | const char **eventEndPP; | |
2819 | if (enc == encoding) { | |
2820 | eventPP = &eventPtr; | |
2821 | *eventPP = s; | |
2822 | eventEndPP = &eventEndPtr; | |
2823 | } | |
2824 | else { | |
2825 | eventPP = &(openInternalEntities->internalEventPtr); | |
2826 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2827 | } | |
2828 | *eventPP = s; | |
2829 | *startPtr = NULL; | |
2830 | tok = XmlIgnoreSectionTok(enc, s, end, &next); | |
2831 | *eventEndPP = next; | |
2832 | switch (tok) { | |
2833 | case XML_TOK_IGNORE_SECT: | |
2834 | if (defaultHandler) | |
2835 | reportDefault(parser, enc, s, next); | |
2836 | *startPtr = next; | |
2837 | return XML_ERROR_NONE; | |
2838 | case XML_TOK_INVALID: | |
2839 | *eventPP = next; | |
2840 | return XML_ERROR_INVALID_TOKEN; | |
2841 | case XML_TOK_PARTIAL_CHAR: | |
2842 | if (nextPtr) { | |
2843 | *nextPtr = s; | |
2844 | return XML_ERROR_NONE; | |
2845 | } | |
2846 | return XML_ERROR_PARTIAL_CHAR; | |
2847 | case XML_TOK_PARTIAL: | |
2848 | case XML_TOK_NONE: | |
2849 | if (nextPtr) { | |
2850 | *nextPtr = s; | |
2851 | return XML_ERROR_NONE; | |
2852 | } | |
2853 | return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */ | |
2854 | default: | |
2855 | *eventPP = next; | |
2856 | return XML_ERROR_UNEXPECTED_STATE; | |
2857 | } | |
2858 | /* not reached */ | |
2859 | } | |
2860 | ||
2861 | #endif /* XML_DTD */ | |
2862 | ||
2863 | static enum XML_Error | |
2864 | initializeEncoding(XML_Parser parser) | |
2865 | { | |
2866 | const char *s; | |
2867 | #ifdef XML_UNICODE | |
2868 | char encodingBuf[128]; | |
2869 | if (!protocolEncodingName) | |
2870 | s = NULL; | |
2871 | else { | |
2872 | int i; | |
2873 | for (i = 0; protocolEncodingName[i]; i++) { | |
2874 | if (i == sizeof(encodingBuf) - 1 | |
2875 | || (protocolEncodingName[i] & ~0x7f) != 0) { | |
2876 | encodingBuf[0] = '\0'; | |
2877 | break; | |
2878 | } | |
2879 | encodingBuf[i] = (char)protocolEncodingName[i]; | |
2880 | } | |
2881 | encodingBuf[i] = '\0'; | |
2882 | s = encodingBuf; | |
2883 | } | |
2884 | #else | |
2885 | s = protocolEncodingName; | |
2886 | #endif | |
2887 | if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s)) | |
2888 | return XML_ERROR_NONE; | |
2889 | return handleUnknownEncoding(parser, protocolEncodingName); | |
2890 | } | |
2891 | ||
2892 | static enum XML_Error | |
2893 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
2894 | const char *s, const char *next) | |
2895 | { | |
2896 | const char *encodingName = NULL; | |
2897 | const XML_Char *storedEncName = NULL; | |
2898 | const ENCODING *newEncoding = NULL; | |
2899 | const char *version = NULL; | |
2900 | const char *versionend; | |
2901 | const XML_Char *storedversion = NULL; | |
2902 | int standalone = -1; | |
2903 | if (!(ns | |
2904 | ? XmlParseXmlDeclNS | |
2905 | : XmlParseXmlDecl)(isGeneralTextEntity, | |
2906 | encoding, | |
2907 | s, | |
2908 | next, | |
2909 | &eventPtr, | |
2910 | &version, | |
2911 | &versionend, | |
2912 | &encodingName, | |
2913 | &newEncoding, | |
2914 | &standalone)) | |
2915 | return XML_ERROR_SYNTAX; | |
2916 | if (!isGeneralTextEntity && standalone == 1) { | |
2917 | _dtd->standalone = XML_TRUE; | |
2918 | #ifdef XML_DTD | |
2919 | if (paramEntityParsing == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE) | |
2920 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; | |
2921 | #endif /* XML_DTD */ | |
2922 | } | |
2923 | if (xmlDeclHandler) { | |
2924 | if (encodingName != NULL) { | |
2925 | storedEncName = poolStoreString(&temp2Pool, | |
2926 | encoding, | |
2927 | encodingName, | |
2928 | encodingName | |
2929 | + XmlNameLength(encoding, encodingName)); | |
2930 | if (!storedEncName) | |
2931 | return XML_ERROR_NO_MEMORY; | |
2932 | poolFinish(&temp2Pool); | |
2933 | } | |
2934 | if (version) { | |
2935 | storedversion = poolStoreString(&temp2Pool, | |
2936 | encoding, | |
2937 | version, | |
2938 | versionend - encoding->minBytesPerChar); | |
2939 | if (!storedversion) | |
2940 | return XML_ERROR_NO_MEMORY; | |
2941 | } | |
2942 | xmlDeclHandler(handlerArg, storedversion, storedEncName, standalone); | |
2943 | } | |
2944 | else if (defaultHandler) | |
2945 | reportDefault(parser, encoding, s, next); | |
2946 | if (protocolEncodingName == NULL) { | |
2947 | if (newEncoding) { | |
2948 | if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) { | |
2949 | eventPtr = encodingName; | |
2950 | return XML_ERROR_INCORRECT_ENCODING; | |
2951 | } | |
2952 | encoding = newEncoding; | |
2953 | } | |
2954 | else if (encodingName) { | |
2955 | enum XML_Error result; | |
2956 | if (!storedEncName) { | |
2957 | storedEncName = poolStoreString( | |
2958 | &temp2Pool, encoding, encodingName, | |
2959 | encodingName + XmlNameLength(encoding, encodingName)); | |
2960 | if (!storedEncName) | |
2961 | return XML_ERROR_NO_MEMORY; | |
2962 | } | |
2963 | result = handleUnknownEncoding(parser, storedEncName); | |
2964 | poolClear(&temp2Pool); | |
2965 | if (result == XML_ERROR_UNKNOWN_ENCODING) | |
2966 | eventPtr = encodingName; | |
2967 | return result; | |
2968 | } | |
2969 | } | |
2970 | ||
2971 | if (storedEncName || storedversion) | |
2972 | poolClear(&temp2Pool); | |
2973 | ||
2974 | return XML_ERROR_NONE; | |
2975 | } | |
2976 | ||
2977 | static enum XML_Error | |
2978 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) | |
2979 | { | |
2980 | if (unknownEncodingHandler) { | |
2981 | XML_Encoding info; | |
2982 | int i; | |
2983 | for (i = 0; i < 256; i++) | |
2984 | info.map[i] = -1; | |
2985 | info.convert = NULL; | |
2986 | info.data = NULL; | |
2987 | info.release = NULL; | |
2988 | if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, | |
2989 | &info)) { | |
2990 | ENCODING *enc; | |
2991 | unknownEncodingMem = MALLOC(XmlSizeOfUnknownEncoding()); | |
2992 | if (!unknownEncodingMem) { | |
2993 | if (info.release) | |
2994 | info.release(info.data); | |
2995 | return XML_ERROR_NO_MEMORY; | |
2996 | } | |
2997 | enc = (ns | |
2998 | ? XmlInitUnknownEncodingNS | |
2999 | : XmlInitUnknownEncoding)(unknownEncodingMem, | |
3000 | info.map, | |
3001 | info.convert, | |
3002 | info.data); | |
3003 | if (enc) { | |
3004 | unknownEncodingData = info.data; | |
3005 | unknownEncodingRelease = info.release; | |
3006 | encoding = enc; | |
3007 | return XML_ERROR_NONE; | |
3008 | } | |
3009 | } | |
3010 | if (info.release != NULL) | |
3011 | info.release(info.data); | |
3012 | } | |
3013 | return XML_ERROR_UNKNOWN_ENCODING; | |
3014 | } | |
3015 | ||
3016 | static enum XML_Error PTRCALL | |
3017 | prologInitProcessor(XML_Parser parser, | |
3018 | const char *s, | |
3019 | const char *end, | |
3020 | const char **nextPtr) | |
3021 | { | |
3022 | enum XML_Error result = initializeEncoding(parser); | |
3023 | if (result != XML_ERROR_NONE) | |
3024 | return result; | |
3025 | processor = prologProcessor; | |
3026 | return prologProcessor(parser, s, end, nextPtr); | |
3027 | } | |
3028 | ||
3029 | #ifdef XML_DTD | |
3030 | ||
3031 | static enum XML_Error PTRCALL | |
3032 | externalParEntInitProcessor(XML_Parser parser, | |
3033 | const char *s, | |
3034 | const char *end, | |
3035 | const char **nextPtr) | |
3036 | { | |
3037 | enum XML_Error result = initializeEncoding(parser); | |
3038 | if (result != XML_ERROR_NONE) | |
3039 | return result; | |
3040 | ||
3041 | /* we know now that XML_Parse(Buffer) has been called, | |
3042 | so we consider the external parameter entity read */ | |
3043 | _dtd->paramEntityRead = XML_TRUE; | |
3044 | ||
3045 | if (prologState.inEntityValue) { | |
3046 | processor = entityValueInitProcessor; | |
3047 | return entityValueInitProcessor(parser, s, end, nextPtr); | |
3048 | } | |
3049 | else { | |
3050 | processor = externalParEntProcessor; | |
3051 | return externalParEntProcessor(parser, s, end, nextPtr); | |
3052 | } | |
3053 | } | |
3054 | ||
3055 | static enum XML_Error PTRCALL | |
3056 | entityValueInitProcessor(XML_Parser parser, | |
3057 | const char *s, | |
3058 | const char *end, | |
3059 | const char **nextPtr) | |
3060 | { | |
3061 | const char *start = s; | |
3062 | const char *next = s; | |
3063 | int tok; | |
3064 | ||
3065 | for (;;) { | |
3066 | tok = XmlPrologTok(encoding, start, end, &next); | |
3067 | if (tok <= 0) { | |
3068 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3069 | *nextPtr = s; | |
3070 | return XML_ERROR_NONE; | |
3071 | } | |
3072 | switch (tok) { | |
3073 | case XML_TOK_INVALID: | |
3074 | return XML_ERROR_INVALID_TOKEN; | |
3075 | case XML_TOK_PARTIAL: | |
3076 | return XML_ERROR_UNCLOSED_TOKEN; | |
3077 | case XML_TOK_PARTIAL_CHAR: | |
3078 | return XML_ERROR_PARTIAL_CHAR; | |
3079 | case XML_TOK_NONE: /* start == end */ | |
3080 | default: | |
3081 | break; | |
3082 | } | |
3083 | return storeEntityValue(parser, encoding, s, end); | |
3084 | } | |
3085 | else if (tok == XML_TOK_XML_DECL) { | |
3086 | enum XML_Error result = processXmlDecl(parser, 0, start, next); | |
3087 | if (result != XML_ERROR_NONE) | |
3088 | return result; | |
3089 | if (nextPtr) *nextPtr = next; | |
3090 | /* stop scanning for text declaration - we found one */ | |
3091 | processor = entityValueProcessor; | |
3092 | return entityValueProcessor(parser, next, end, nextPtr); | |
3093 | } | |
3094 | /* If we are at the end of the buffer, this would cause XmlPrologTok to | |
3095 | return XML_TOK_NONE on the next call, which would then cause the | |
3096 | function to exit with *nextPtr set to s - that is what we want for other | |
3097 | tokens, but not for the BOM - we would rather like to skip it; | |
3098 | then, when this routine is entered the next time, XmlPrologTok will | |
3099 | return XML_TOK_INVALID, since the BOM is still in the buffer | |
3100 | */ | |
3101 | else if (tok == XML_TOK_BOM && next == end && nextPtr) { | |
3102 | *nextPtr = next; | |
3103 | return XML_ERROR_NONE; | |
3104 | } | |
3105 | start = next; | |
3106 | } | |
3107 | } | |
3108 | ||
3109 | static enum XML_Error PTRCALL | |
3110 | externalParEntProcessor(XML_Parser parser, | |
3111 | const char *s, | |
3112 | const char *end, | |
3113 | const char **nextPtr) | |
3114 | { | |
3115 | const char *start = s; | |
3116 | const char *next = s; | |
3117 | int tok; | |
3118 | ||
3119 | tok = XmlPrologTok(encoding, start, end, &next); | |
3120 | if (tok <= 0) { | |
3121 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3122 | *nextPtr = s; | |
3123 | return XML_ERROR_NONE; | |
3124 | } | |
3125 | switch (tok) { | |
3126 | case XML_TOK_INVALID: | |
3127 | return XML_ERROR_INVALID_TOKEN; | |
3128 | case XML_TOK_PARTIAL: | |
3129 | return XML_ERROR_UNCLOSED_TOKEN; | |
3130 | case XML_TOK_PARTIAL_CHAR: | |
3131 | return XML_ERROR_PARTIAL_CHAR; | |
3132 | case XML_TOK_NONE: /* start == end */ | |
3133 | default: | |
3134 | break; | |
3135 | } | |
3136 | } | |
3137 | /* This would cause the next stage, i.e. doProlog to be passed XML_TOK_BOM. | |
3138 | However, when parsing an external subset, doProlog will not accept a BOM | |
3139 | as valid, and report a syntax error, so we have to skip the BOM | |
3140 | */ | |
3141 | else if (tok == XML_TOK_BOM) { | |
3142 | s = next; | |
3143 | tok = XmlPrologTok(encoding, s, end, &next); | |
3144 | } | |
3145 | ||
3146 | processor = prologProcessor; | |
3147 | return doProlog(parser, encoding, s, end, tok, next, nextPtr); | |
3148 | } | |
3149 | ||
3150 | static enum XML_Error PTRCALL | |
3151 | entityValueProcessor(XML_Parser parser, | |
3152 | const char *s, | |
3153 | const char *end, | |
3154 | const char **nextPtr) | |
3155 | { | |
3156 | const char *start = s; | |
3157 | const char *next = s; | |
3158 | const ENCODING *enc = encoding; | |
3159 | int tok; | |
3160 | ||
3161 | for (;;) { | |
3162 | tok = XmlPrologTok(enc, start, end, &next); | |
3163 | if (tok <= 0) { | |
3164 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3165 | *nextPtr = s; | |
3166 | return XML_ERROR_NONE; | |
3167 | } | |
3168 | switch (tok) { | |
3169 | case XML_TOK_INVALID: | |
3170 | return XML_ERROR_INVALID_TOKEN; | |
3171 | case XML_TOK_PARTIAL: | |
3172 | return XML_ERROR_UNCLOSED_TOKEN; | |
3173 | case XML_TOK_PARTIAL_CHAR: | |
3174 | return XML_ERROR_PARTIAL_CHAR; | |
3175 | case XML_TOK_NONE: /* start == end */ | |
3176 | default: | |
3177 | break; | |
3178 | } | |
3179 | return storeEntityValue(parser, enc, s, end); | |
3180 | } | |
3181 | start = next; | |
3182 | } | |
3183 | } | |
3184 | ||
3185 | #endif /* XML_DTD */ | |
3186 | ||
3187 | static enum XML_Error PTRCALL | |
3188 | prologProcessor(XML_Parser parser, | |
3189 | const char *s, | |
3190 | const char *end, | |
3191 | const char **nextPtr) | |
3192 | { | |
3193 | const char *next = s; | |
3194 | int tok = XmlPrologTok(encoding, s, end, &next); | |
3195 | return doProlog(parser, encoding, s, end, tok, next, nextPtr); | |
3196 | } | |
3197 | ||
3198 | static enum XML_Error | |
3199 | doProlog(XML_Parser parser, | |
3200 | const ENCODING *enc, | |
3201 | const char *s, | |
3202 | const char *end, | |
3203 | int tok, | |
3204 | const char *next, | |
3205 | const char **nextPtr) | |
3206 | { | |
3207 | #ifdef XML_DTD | |
3208 | static const XML_Char externalSubsetName[] = { '#' , '\0' }; | |
3209 | #endif /* XML_DTD */ | |
3210 | static const XML_Char atypeCDATA[] = { 'C', 'D', 'A', 'T', 'A', '\0' }; | |
3211 | static const XML_Char atypeID[] = { 'I', 'D', '\0' }; | |
3212 | static const XML_Char atypeIDREF[] = { 'I', 'D', 'R', 'E', 'F', '\0' }; | |
3213 | static const XML_Char atypeIDREFS[] = { 'I', 'D', 'R', 'E', 'F', 'S', '\0' }; | |
3214 | static const XML_Char atypeENTITY[] = { 'E', 'N', 'T', 'I', 'T', 'Y', '\0' }; | |
3215 | static const XML_Char atypeENTITIES[] = | |
3216 | { 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S', '\0' }; | |
3217 | static const XML_Char atypeNMTOKEN[] = { | |
3218 | 'N', 'M', 'T', 'O', 'K', 'E', 'N', '\0' }; | |
3219 | static const XML_Char atypeNMTOKENS[] = { | |
3220 | 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S', '\0' }; | |
3221 | static const XML_Char notationPrefix[] = { | |
3222 | 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N', '(', '\0' }; | |
3223 | static const XML_Char enumValueSep[] = { '|', '\0' }; | |
3224 | static const XML_Char enumValueStart[] = { '(', '\0' }; | |
3225 | ||
3226 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
3227 | ||
3228 | const char **eventPP; | |
3229 | const char **eventEndPP; | |
3230 | enum XML_Content_Quant quant; | |
3231 | ||
3232 | if (enc == encoding) { | |
3233 | eventPP = &eventPtr; | |
3234 | eventEndPP = &eventEndPtr; | |
3235 | } | |
3236 | else { | |
3237 | eventPP = &(openInternalEntities->internalEventPtr); | |
3238 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
3239 | } | |
3240 | for (;;) { | |
3241 | int role; | |
3242 | XML_Bool handleDefault = XML_TRUE; | |
3243 | *eventPP = s; | |
3244 | *eventEndPP = next; | |
3245 | if (tok <= 0) { | |
3246 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
3247 | *nextPtr = s; | |
3248 | return XML_ERROR_NONE; | |
3249 | } | |
3250 | switch (tok) { | |
3251 | case XML_TOK_INVALID: | |
3252 | *eventPP = next; | |
3253 | return XML_ERROR_INVALID_TOKEN; | |
3254 | case XML_TOK_PARTIAL: | |
3255 | return XML_ERROR_UNCLOSED_TOKEN; | |
3256 | case XML_TOK_PARTIAL_CHAR: | |
3257 | return XML_ERROR_PARTIAL_CHAR; | |
3258 | case XML_TOK_NONE: | |
3259 | #ifdef XML_DTD | |
3260 | if (enc != encoding) | |
3261 | return XML_ERROR_NONE; | |
3262 | if (isParamEntity) { | |
3263 | if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc) | |
3264 | == XML_ROLE_ERROR) | |
3265 | return XML_ERROR_SYNTAX; | |
3266 | return XML_ERROR_NONE; | |
3267 | } | |
3268 | #endif /* XML_DTD */ | |
3269 | return XML_ERROR_NO_ELEMENTS; | |
3270 | default: | |
3271 | tok = -tok; | |
3272 | next = end; | |
3273 | break; | |
3274 | } | |
3275 | } | |
3276 | role = XmlTokenRole(&prologState, tok, s, next, enc); | |
3277 | switch (role) { | |
3278 | case XML_ROLE_XML_DECL: | |
3279 | { | |
3280 | enum XML_Error result = processXmlDecl(parser, 0, s, next); | |
3281 | if (result != XML_ERROR_NONE) | |
3282 | return result; | |
3283 | enc = encoding; | |
3284 | handleDefault = XML_FALSE; | |
3285 | } | |
3286 | break; | |
3287 | case XML_ROLE_DOCTYPE_NAME: | |
3288 | if (startDoctypeDeclHandler) { | |
3289 | doctypeName = poolStoreString(&tempPool, enc, s, next); | |
3290 | if (!doctypeName) | |
3291 | return XML_ERROR_NO_MEMORY; | |
3292 | poolFinish(&tempPool); | |
3293 | doctypePubid = NULL; | |
3294 | handleDefault = XML_FALSE; | |
3295 | } | |
3296 | doctypeSysid = NULL; /* always initialize to NULL */ | |
3297 | break; | |
3298 | case XML_ROLE_DOCTYPE_INTERNAL_SUBSET: | |
3299 | if (startDoctypeDeclHandler) { | |
3300 | startDoctypeDeclHandler(handlerArg, doctypeName, doctypeSysid, | |
3301 | doctypePubid, 1); | |
3302 | doctypeName = NULL; | |
3303 | poolClear(&tempPool); | |
3304 | handleDefault = XML_FALSE; | |
3305 | } | |
3306 | break; | |
3307 | #ifdef XML_DTD | |
3308 | case XML_ROLE_TEXT_DECL: | |
3309 | { | |
3310 | enum XML_Error result = processXmlDecl(parser, 1, s, next); | |
3311 | if (result != XML_ERROR_NONE) | |
3312 | return result; | |
3313 | enc = encoding; | |
3314 | handleDefault = XML_FALSE; | |
3315 | } | |
3316 | break; | |
3317 | #endif /* XML_DTD */ | |
3318 | case XML_ROLE_DOCTYPE_PUBLIC_ID: | |
3319 | #ifdef XML_DTD | |
3320 | useForeignDTD = XML_FALSE; | |
3321 | #endif /* XML_DTD */ | |
3322 | dtd->hasParamEntityRefs = XML_TRUE; | |
3323 | if (startDoctypeDeclHandler) { | |
3324 | doctypePubid = poolStoreString(&tempPool, enc, | |
3325 | s + enc->minBytesPerChar, | |
3326 | next - enc->minBytesPerChar); | |
3327 | if (!doctypePubid) | |
3328 | return XML_ERROR_NO_MEMORY; | |
3329 | poolFinish(&tempPool); | |
3330 | handleDefault = XML_FALSE; | |
3331 | } | |
3332 | #ifdef XML_DTD | |
3333 | declEntity = (ENTITY *)lookup(&dtd->paramEntities, | |
3334 | externalSubsetName, | |
3335 | sizeof(ENTITY)); | |
3336 | if (!declEntity) | |
3337 | return XML_ERROR_NO_MEMORY; | |
3338 | #endif /* XML_DTD */ | |
3339 | /* fall through */ | |
3340 | case XML_ROLE_ENTITY_PUBLIC_ID: | |
3341 | if (!XmlIsPublicId(enc, s, next, eventPP)) | |
3342 | return XML_ERROR_SYNTAX; | |
3343 | if (dtd->keepProcessing && declEntity) { | |
3344 | XML_Char *tem = poolStoreString(&dtd->pool, | |
3345 | enc, | |
3346 | s + enc->minBytesPerChar, | |
3347 | next - enc->minBytesPerChar); | |
3348 | if (!tem) | |
3349 | return XML_ERROR_NO_MEMORY; | |
3350 | normalizePublicId(tem); | |
3351 | declEntity->publicId = tem; | |
3352 | poolFinish(&dtd->pool); | |
3353 | if (entityDeclHandler) | |
3354 | handleDefault = XML_FALSE; | |
3355 | } | |
3356 | break; | |
3357 | case XML_ROLE_DOCTYPE_CLOSE: | |
3358 | if (doctypeName) { | |
3359 | startDoctypeDeclHandler(handlerArg, doctypeName, | |
3360 | doctypeSysid, doctypePubid, 0); | |
3361 | poolClear(&tempPool); | |
3362 | handleDefault = XML_FALSE; | |
3363 | } | |
3364 | /* doctypeSysid will be non-NULL in the case of a previous | |
3365 | XML_ROLE_DOCTYPE_SYSTEM_ID, even if startDoctypeDeclHandler | |
3366 | was not set, indicating an external subset | |
3367 | */ | |
3368 | #ifdef XML_DTD | |
3369 | if (doctypeSysid || useForeignDTD) { | |
3370 | dtd->hasParamEntityRefs = XML_TRUE; /* when docTypeSysid == NULL */ | |
3371 | if (paramEntityParsing && externalEntityRefHandler) { | |
3372 | ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities, | |
3373 | externalSubsetName, | |
3374 | sizeof(ENTITY)); | |
3375 | if (!entity) | |
3376 | return XML_ERROR_NO_MEMORY; | |
3377 | if (useForeignDTD) | |
3378 | entity->base = curBase; | |
3379 | dtd->paramEntityRead = XML_FALSE; | |
3380 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
3381 | 0, | |
3382 | entity->base, | |
3383 | entity->systemId, | |
3384 | entity->publicId)) | |
3385 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
3386 | if (dtd->paramEntityRead && | |
3387 | !dtd->standalone && | |
3388 | notStandaloneHandler && | |
3389 | !notStandaloneHandler(handlerArg)) | |
3390 | return XML_ERROR_NOT_STANDALONE; | |
3391 | /* end of DTD - no need to update dtd->keepProcessing */ | |
3392 | } | |
3393 | useForeignDTD = XML_FALSE; | |
3394 | } | |
3395 | #endif /* XML_DTD */ | |
3396 | if (endDoctypeDeclHandler) { | |
3397 | endDoctypeDeclHandler(handlerArg); | |
3398 | handleDefault = XML_FALSE; | |
3399 | } | |
3400 | break; | |
3401 | case XML_ROLE_INSTANCE_START: | |
3402 | #ifdef XML_DTD | |
3403 | /* if there is no DOCTYPE declaration then now is the | |
3404 | last chance to read the foreign DTD | |
3405 | */ | |
3406 | if (useForeignDTD) { | |
3407 | dtd->hasParamEntityRefs = XML_TRUE; | |
3408 | if (paramEntityParsing && externalEntityRefHandler) { | |
3409 | ENTITY *entity = (ENTITY *)lookup(&dtd->paramEntities, | |
3410 | externalSubsetName, | |
3411 | sizeof(ENTITY)); | |
3412 | if (!entity) | |
3413 | return XML_ERROR_NO_MEMORY; | |
3414 | entity->base = curBase; | |
3415 | dtd->paramEntityRead = XML_FALSE; | |
3416 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
3417 | 0, | |
3418 | entity->base, | |
3419 | entity->systemId, | |
3420 | entity->publicId)) | |
3421 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
3422 | if (dtd->paramEntityRead && | |
3423 | !dtd->standalone && | |
3424 | notStandaloneHandler && | |
3425 | !notStandaloneHandler(handlerArg)) | |
3426 | return XML_ERROR_NOT_STANDALONE; | |
3427 | /* end of DTD - no need to update dtd->keepProcessing */ | |
3428 | } | |
3429 | } | |
3430 | #endif /* XML_DTD */ | |
3431 | processor = contentProcessor; | |
3432 | return contentProcessor(parser, s, end, nextPtr); | |
3433 | case XML_ROLE_ATTLIST_ELEMENT_NAME: | |
3434 | declElementType = getElementType(parser, enc, s, next); | |
3435 | if (!declElementType) | |
3436 | return XML_ERROR_NO_MEMORY; | |
3437 | goto checkAttListDeclHandler; | |
3438 | case XML_ROLE_ATTRIBUTE_NAME: | |
3439 | declAttributeId = getAttributeId(parser, enc, s, next); | |
3440 | if (!declAttributeId) | |
3441 | return XML_ERROR_NO_MEMORY; | |
3442 | declAttributeIsCdata = XML_FALSE; | |
3443 | declAttributeType = NULL; | |
3444 | declAttributeIsId = XML_FALSE; | |
3445 | goto checkAttListDeclHandler; | |
3446 | case XML_ROLE_ATTRIBUTE_TYPE_CDATA: | |
3447 | declAttributeIsCdata = XML_TRUE; | |
3448 | declAttributeType = atypeCDATA; | |
3449 | goto checkAttListDeclHandler; | |
3450 | case XML_ROLE_ATTRIBUTE_TYPE_ID: | |
3451 | declAttributeIsId = XML_TRUE; | |
3452 | declAttributeType = atypeID; | |
3453 | goto checkAttListDeclHandler; | |
3454 | case XML_ROLE_ATTRIBUTE_TYPE_IDREF: | |
3455 | declAttributeType = atypeIDREF; | |
3456 | goto checkAttListDeclHandler; | |
3457 | case XML_ROLE_ATTRIBUTE_TYPE_IDREFS: | |
3458 | declAttributeType = atypeIDREFS; | |
3459 | goto checkAttListDeclHandler; | |
3460 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITY: | |
3461 | declAttributeType = atypeENTITY; | |
3462 | goto checkAttListDeclHandler; | |
3463 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITIES: | |
3464 | declAttributeType = atypeENTITIES; | |
3465 | goto checkAttListDeclHandler; | |
3466 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKEN: | |
3467 | declAttributeType = atypeNMTOKEN; | |
3468 | goto checkAttListDeclHandler; | |
3469 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS: | |
3470 | declAttributeType = atypeNMTOKENS; | |
3471 | checkAttListDeclHandler: | |
3472 | if (dtd->keepProcessing && attlistDeclHandler) | |
3473 | handleDefault = XML_FALSE; | |
3474 | break; | |
3475 | case XML_ROLE_ATTRIBUTE_ENUM_VALUE: | |
3476 | case XML_ROLE_ATTRIBUTE_NOTATION_VALUE: | |
3477 | if (dtd->keepProcessing && attlistDeclHandler) { | |
3478 | const XML_Char *prefix; | |
3479 | if (declAttributeType) { | |
3480 | prefix = enumValueSep; | |
3481 | } | |
3482 | else { | |
3483 | prefix = (role == XML_ROLE_ATTRIBUTE_NOTATION_VALUE | |
3484 | ? notationPrefix | |
3485 | : enumValueStart); | |
3486 | } | |
3487 | if (!poolAppendString(&tempPool, prefix)) | |
3488 | return XML_ERROR_NO_MEMORY; | |
3489 | if (!poolAppend(&tempPool, enc, s, next)) | |
3490 | return XML_ERROR_NO_MEMORY; | |
3491 | declAttributeType = tempPool.start; | |
3492 | handleDefault = XML_FALSE; | |
3493 | } | |
3494 | break; | |
3495 | case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: | |
3496 | case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: | |
3497 | if (dtd->keepProcessing) { | |
3498 | if (!defineAttribute(declElementType, declAttributeId, | |
3499 | declAttributeIsCdata, declAttributeIsId, 0, | |
3500 | parser)) | |
3501 | return XML_ERROR_NO_MEMORY; | |
3502 | if (attlistDeclHandler && declAttributeType) { | |
3503 | if (*declAttributeType == XML_T('(') | |
3504 | || (*declAttributeType == XML_T('N') | |
3505 | && declAttributeType[1] == XML_T('O'))) { | |
3506 | /* Enumerated or Notation type */ | |
3507 | if (!poolAppendChar(&tempPool, XML_T(')')) | |
3508 | || !poolAppendChar(&tempPool, XML_T('\0'))) | |
3509 | return XML_ERROR_NO_MEMORY; | |
3510 | declAttributeType = tempPool.start; | |
3511 | poolFinish(&tempPool); | |
3512 | } | |
3513 | *eventEndPP = s; | |
3514 | attlistDeclHandler(handlerArg, declElementType->name, | |
3515 | declAttributeId->name, declAttributeType, | |
3516 | 0, role == XML_ROLE_REQUIRED_ATTRIBUTE_VALUE); | |
3517 | poolClear(&tempPool); | |
3518 | handleDefault = XML_FALSE; | |
3519 | } | |
3520 | } | |
3521 | break; | |
3522 | case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: | |
3523 | case XML_ROLE_FIXED_ATTRIBUTE_VALUE: | |
3524 | if (dtd->keepProcessing) { | |
3525 | const XML_Char *attVal; | |
3526 | enum XML_Error result | |
3527 | = storeAttributeValue(parser, enc, declAttributeIsCdata, | |
3528 | s + enc->minBytesPerChar, | |
3529 | next - enc->minBytesPerChar, | |
3530 | &dtd->pool); | |
3531 | if (result) | |
3532 | return result; | |
3533 | attVal = poolStart(&dtd->pool); | |
3534 | poolFinish(&dtd->pool); | |
3535 | /* ID attributes aren't allowed to have a default */ | |
3536 | if (!defineAttribute(declElementType, declAttributeId, | |
3537 | declAttributeIsCdata, XML_FALSE, attVal, parser)) | |
3538 | return XML_ERROR_NO_MEMORY; | |
3539 | if (attlistDeclHandler && declAttributeType) { | |
3540 | if (*declAttributeType == XML_T('(') | |
3541 | || (*declAttributeType == XML_T('N') | |
3542 | && declAttributeType[1] == XML_T('O'))) { | |
3543 | /* Enumerated or Notation type */ | |
3544 | if (!poolAppendChar(&tempPool, XML_T(')')) | |
3545 | || !poolAppendChar(&tempPool, XML_T('\0'))) | |
3546 | return XML_ERROR_NO_MEMORY; | |
3547 | declAttributeType = tempPool.start; | |
3548 | poolFinish(&tempPool); | |
3549 | } | |
3550 | *eventEndPP = s; | |
3551 | attlistDeclHandler(handlerArg, declElementType->name, | |
3552 | declAttributeId->name, declAttributeType, | |
3553 | attVal, | |
3554 | role == XML_ROLE_FIXED_ATTRIBUTE_VALUE); | |
3555 | poolClear(&tempPool); | |
3556 | handleDefault = XML_FALSE; | |
3557 | } | |
3558 | } | |
3559 | break; | |
3560 | case XML_ROLE_ENTITY_VALUE: | |
3561 | if (dtd->keepProcessing) { | |
3562 | enum XML_Error result = storeEntityValue(parser, enc, | |
3563 | s + enc->minBytesPerChar, | |
3564 | next - enc->minBytesPerChar); | |
3565 | if (declEntity) { | |
3566 | declEntity->textPtr = poolStart(&dtd->entityValuePool); | |
3567 | declEntity->textLen = poolLength(&dtd->entityValuePool); | |
3568 | poolFinish(&dtd->entityValuePool); | |
3569 | if (entityDeclHandler) { | |
3570 | *eventEndPP = s; | |
3571 | entityDeclHandler(handlerArg, | |
3572 | declEntity->name, | |
3573 | declEntity->is_param, | |
3574 | declEntity->textPtr, | |
3575 | declEntity->textLen, | |
3576 | curBase, 0, 0, 0); | |
3577 | handleDefault = XML_FALSE; | |
3578 | } | |
3579 | } | |
3580 | else | |
3581 | poolDiscard(&dtd->entityValuePool); | |
3582 | if (result != XML_ERROR_NONE) | |
3583 | return result; | |
3584 | } | |
3585 | break; | |
3586 | case XML_ROLE_DOCTYPE_SYSTEM_ID: | |
3587 | #ifdef XML_DTD | |
3588 | useForeignDTD = XML_FALSE; | |
3589 | #endif /* XML_DTD */ | |
3590 | dtd->hasParamEntityRefs = XML_TRUE; | |
3591 | if (startDoctypeDeclHandler) { | |
3592 | doctypeSysid = poolStoreString(&tempPool, enc, | |
3593 | s + enc->minBytesPerChar, | |
3594 | next - enc->minBytesPerChar); | |
3595 | if (doctypeSysid == NULL) | |
3596 | return XML_ERROR_NO_MEMORY; | |
3597 | poolFinish(&tempPool); | |
3598 | handleDefault = XML_FALSE; | |
3599 | } | |
3600 | #ifdef XML_DTD | |
3601 | else | |
3602 | /* use externalSubsetName to make doctypeSysid non-NULL | |
3603 | for the case where no startDoctypeDeclHandler is set */ | |
3604 | doctypeSysid = externalSubsetName; | |
3605 | #endif /* XML_DTD */ | |
3606 | if (!dtd->standalone | |
3607 | #ifdef XML_DTD | |
3608 | && !paramEntityParsing | |
3609 | #endif /* XML_DTD */ | |
3610 | && notStandaloneHandler | |
3611 | && !notStandaloneHandler(handlerArg)) | |
3612 | return XML_ERROR_NOT_STANDALONE; | |
3613 | #ifndef XML_DTD | |
3614 | break; | |
3615 | #else /* XML_DTD */ | |
3616 | if (!declEntity) { | |
3617 | declEntity = (ENTITY *)lookup(&dtd->paramEntities, | |
3618 | externalSubsetName, | |
3619 | sizeof(ENTITY)); | |
3620 | if (!declEntity) | |
3621 | return XML_ERROR_NO_MEMORY; | |
3622 | declEntity->publicId = NULL; | |
3623 | } | |
3624 | /* fall through */ | |
3625 | #endif /* XML_DTD */ | |
3626 | case XML_ROLE_ENTITY_SYSTEM_ID: | |
3627 | if (dtd->keepProcessing && declEntity) { | |
3628 | declEntity->systemId = poolStoreString(&dtd->pool, enc, | |
3629 | s + enc->minBytesPerChar, | |
3630 | next - enc->minBytesPerChar); | |
3631 | if (!declEntity->systemId) | |
3632 | return XML_ERROR_NO_MEMORY; | |
3633 | declEntity->base = curBase; | |
3634 | poolFinish(&dtd->pool); | |
3635 | if (entityDeclHandler) | |
3636 | handleDefault = XML_FALSE; | |
3637 | } | |
3638 | break; | |
3639 | case XML_ROLE_ENTITY_COMPLETE: | |
3640 | if (dtd->keepProcessing && declEntity && entityDeclHandler) { | |
3641 | *eventEndPP = s; | |
3642 | entityDeclHandler(handlerArg, | |
3643 | declEntity->name, | |
3644 | declEntity->is_param, | |
3645 | 0,0, | |
3646 | declEntity->base, | |
3647 | declEntity->systemId, | |
3648 | declEntity->publicId, | |
3649 | 0); | |
3650 | handleDefault = XML_FALSE; | |
3651 | } | |
3652 | break; | |
3653 | case XML_ROLE_ENTITY_NOTATION_NAME: | |
3654 | if (dtd->keepProcessing && declEntity) { | |
3655 | declEntity->notation = poolStoreString(&dtd->pool, enc, s, next); | |
3656 | if (!declEntity->notation) | |
3657 | return XML_ERROR_NO_MEMORY; | |
3658 | poolFinish(&dtd->pool); | |
3659 | if (unparsedEntityDeclHandler) { | |
3660 | *eventEndPP = s; | |
3661 | unparsedEntityDeclHandler(handlerArg, | |
3662 | declEntity->name, | |
3663 | declEntity->base, | |
3664 | declEntity->systemId, | |
3665 | declEntity->publicId, | |
3666 | declEntity->notation); | |
3667 | handleDefault = XML_FALSE; | |
3668 | } | |
3669 | else if (entityDeclHandler) { | |
3670 | *eventEndPP = s; | |
3671 | entityDeclHandler(handlerArg, | |
3672 | declEntity->name, | |
3673 | 0,0,0, | |
3674 | declEntity->base, | |
3675 | declEntity->systemId, | |
3676 | declEntity->publicId, | |
3677 | declEntity->notation); | |
3678 | handleDefault = XML_FALSE; | |
3679 | } | |
3680 | } | |
3681 | break; | |
3682 | case XML_ROLE_GENERAL_ENTITY_NAME: | |
3683 | { | |
3684 | if (XmlPredefinedEntityName(enc, s, next)) { | |
3685 | declEntity = NULL; | |
3686 | break; | |
3687 | } | |
3688 | if (dtd->keepProcessing) { | |
3689 | const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next); | |
3690 | if (!name) | |
3691 | return XML_ERROR_NO_MEMORY; | |
3692 | declEntity = (ENTITY *)lookup(&dtd->generalEntities, name, | |
3693 | sizeof(ENTITY)); | |
3694 | if (!declEntity) | |
3695 | return XML_ERROR_NO_MEMORY; | |
3696 | if (declEntity->name != name) { | |
3697 | poolDiscard(&dtd->pool); | |
3698 | declEntity = NULL; | |
3699 | } | |
3700 | else { | |
3701 | poolFinish(&dtd->pool); | |
3702 | declEntity->publicId = NULL; | |
3703 | declEntity->is_param = XML_FALSE; | |
3704 | /* if we have a parent parser or are reading an internal parameter | |
3705 | entity, then the entity declaration is not considered "internal" | |
3706 | */ | |
3707 | declEntity->is_internal = !(parentParser || openInternalEntities); | |
3708 | if (entityDeclHandler) | |
3709 | handleDefault = XML_FALSE; | |
3710 | } | |
3711 | } | |
3712 | else { | |
3713 | poolDiscard(&dtd->pool); | |
3714 | declEntity = NULL; | |
3715 | } | |
3716 | } | |
3717 | break; | |
3718 | case XML_ROLE_PARAM_ENTITY_NAME: | |
3719 | #ifdef XML_DTD | |
3720 | if (dtd->keepProcessing) { | |
3721 | const XML_Char *name = poolStoreString(&dtd->pool, enc, s, next); | |
3722 | if (!name) | |
3723 | return XML_ERROR_NO_MEMORY; | |
3724 | declEntity = (ENTITY *)lookup(&dtd->paramEntities, | |
3725 | name, sizeof(ENTITY)); | |
3726 | if (!declEntity) | |
3727 | return XML_ERROR_NO_MEMORY; | |
3728 | if (declEntity->name != name) { | |
3729 | poolDiscard(&dtd->pool); | |
3730 | declEntity = NULL; | |
3731 | } | |
3732 | else { | |
3733 | poolFinish(&dtd->pool); | |
3734 | declEntity->publicId = NULL; | |
3735 | declEntity->is_param = XML_TRUE; | |
3736 | /* if we have a parent parser or are reading an internal parameter | |
3737 | entity, then the entity declaration is not considered "internal" | |
3738 | */ | |
3739 | declEntity->is_internal = !(parentParser || openInternalEntities); | |
3740 | if (entityDeclHandler) | |
3741 | handleDefault = XML_FALSE; | |
3742 | } | |
3743 | } | |
3744 | else { | |
3745 | poolDiscard(&dtd->pool); | |
3746 | declEntity = NULL; | |
3747 | } | |
3748 | #else /* not XML_DTD */ | |
3749 | declEntity = NULL; | |
3750 | #endif /* XML_DTD */ | |
3751 | break; | |
3752 | case XML_ROLE_NOTATION_NAME: | |
3753 | declNotationPublicId = NULL; | |
3754 | declNotationName = NULL; | |
3755 | if (notationDeclHandler) { | |
3756 | declNotationName = poolStoreString(&tempPool, enc, s, next); | |
3757 | if (!declNotationName) | |
3758 | return XML_ERROR_NO_MEMORY; | |
3759 | poolFinish(&tempPool); | |
3760 | handleDefault = XML_FALSE; | |
3761 | } | |
3762 | break; | |
3763 | case XML_ROLE_NOTATION_PUBLIC_ID: | |
3764 | if (!XmlIsPublicId(enc, s, next, eventPP)) | |
3765 | return XML_ERROR_SYNTAX; | |
3766 | if (declNotationName) { /* means notationDeclHandler != NULL */ | |
3767 | XML_Char *tem = poolStoreString(&tempPool, | |
3768 | enc, | |
3769 | s + enc->minBytesPerChar, | |
3770 | next - enc->minBytesPerChar); | |
3771 | if (!tem) | |
3772 | return XML_ERROR_NO_MEMORY; | |
3773 | normalizePublicId(tem); | |
3774 | declNotationPublicId = tem; | |
3775 | poolFinish(&tempPool); | |
3776 | handleDefault = XML_FALSE; | |
3777 | } | |
3778 | break; | |
3779 | case XML_ROLE_NOTATION_SYSTEM_ID: | |
3780 | if (declNotationName && notationDeclHandler) { | |
3781 | const XML_Char *systemId | |
3782 | = poolStoreString(&tempPool, enc, | |
3783 | s + enc->minBytesPerChar, | |
3784 | next - enc->minBytesPerChar); | |
3785 | if (!systemId) | |
3786 | return XML_ERROR_NO_MEMORY; | |
3787 | *eventEndPP = s; | |
3788 | notationDeclHandler(handlerArg, | |
3789 | declNotationName, | |
3790 | curBase, | |
3791 | systemId, | |
3792 | declNotationPublicId); | |
3793 | handleDefault = XML_FALSE; | |
3794 | } | |
3795 | poolClear(&tempPool); | |
3796 | break; | |
3797 | case XML_ROLE_NOTATION_NO_SYSTEM_ID: | |
3798 | if (declNotationPublicId && notationDeclHandler) { | |
3799 | *eventEndPP = s; | |
3800 | notationDeclHandler(handlerArg, | |
3801 | declNotationName, | |
3802 | curBase, | |
3803 | 0, | |
3804 | declNotationPublicId); | |
3805 | handleDefault = XML_FALSE; | |
3806 | } | |
3807 | poolClear(&tempPool); | |
3808 | break; | |
3809 | case XML_ROLE_ERROR: | |
3810 | switch (tok) { | |
3811 | case XML_TOK_PARAM_ENTITY_REF: | |
3812 | return XML_ERROR_PARAM_ENTITY_REF; | |
3813 | case XML_TOK_XML_DECL: | |
3814 | return XML_ERROR_MISPLACED_XML_PI; | |
3815 | default: | |
3816 | return XML_ERROR_SYNTAX; | |
3817 | } | |
3818 | #ifdef XML_DTD | |
3819 | case XML_ROLE_IGNORE_SECT: | |
3820 | { | |
3821 | enum XML_Error result; | |
3822 | if (defaultHandler) | |
3823 | reportDefault(parser, enc, s, next); | |
3824 | handleDefault = XML_FALSE; | |
3825 | result = doIgnoreSection(parser, enc, &next, end, nextPtr); | |
3826 | if (!next) { | |
3827 | processor = ignoreSectionProcessor; | |
3828 | return result; | |
3829 | } | |
3830 | } | |
3831 | break; | |
3832 | #endif /* XML_DTD */ | |
3833 | case XML_ROLE_GROUP_OPEN: | |
3834 | if (prologState.level >= groupSize) { | |
3835 | if (groupSize) { | |
3836 | char *temp = (char *)REALLOC(groupConnector, groupSize *= 2); | |
3837 | if (temp == NULL) | |
3838 | return XML_ERROR_NO_MEMORY; | |
3839 | groupConnector = temp; | |
3840 | if (dtd->scaffIndex) { | |
3841 | int *temp = (int *)REALLOC(dtd->scaffIndex, | |
3842 | groupSize * sizeof(int)); | |
3843 | if (temp == NULL) | |
3844 | return XML_ERROR_NO_MEMORY; | |
3845 | dtd->scaffIndex = temp; | |
3846 | } | |
3847 | } | |
3848 | else { | |
3849 | groupConnector = (char *)MALLOC(groupSize = 32); | |
3850 | if (!groupConnector) | |
3851 | return XML_ERROR_NO_MEMORY; | |
3852 | } | |
3853 | } | |
3854 | groupConnector[prologState.level] = 0; | |
3855 | if (dtd->in_eldecl) { | |
3856 | int myindex = nextScaffoldPart(parser); | |
3857 | if (myindex < 0) | |
3858 | return XML_ERROR_NO_MEMORY; | |
3859 | dtd->scaffIndex[dtd->scaffLevel] = myindex; | |
3860 | dtd->scaffLevel++; | |
3861 | dtd->scaffold[myindex].type = XML_CTYPE_SEQ; | |
3862 | if (elementDeclHandler) | |
3863 | handleDefault = XML_FALSE; | |
3864 | } | |
3865 | break; | |
3866 | case XML_ROLE_GROUP_SEQUENCE: | |
3867 | if (groupConnector[prologState.level] == '|') | |
3868 | return XML_ERROR_SYNTAX; | |
3869 | groupConnector[prologState.level] = ','; | |
3870 | if (dtd->in_eldecl && elementDeclHandler) | |
3871 | handleDefault = XML_FALSE; | |
3872 | break; | |
3873 | case XML_ROLE_GROUP_CHOICE: | |
3874 | if (groupConnector[prologState.level] == ',') | |
3875 | return XML_ERROR_SYNTAX; | |
3876 | if (dtd->in_eldecl | |
3877 | && !groupConnector[prologState.level] | |
3878 | && (dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type | |
3879 | != XML_CTYPE_MIXED) | |
3880 | ) { | |
3881 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type | |
3882 | = XML_CTYPE_CHOICE; | |
3883 | if (elementDeclHandler) | |
3884 | handleDefault = XML_FALSE; | |
3885 | } | |
3886 | groupConnector[prologState.level] = '|'; | |
3887 | break; | |
3888 | case XML_ROLE_PARAM_ENTITY_REF: | |
3889 | #ifdef XML_DTD | |
3890 | case XML_ROLE_INNER_PARAM_ENTITY_REF: | |
3891 | /* PE references in internal subset are | |
3892 | not allowed within declarations */ | |
3893 | if (prologState.documentEntity && | |
3894 | role == XML_ROLE_INNER_PARAM_ENTITY_REF) | |
3895 | return XML_ERROR_PARAM_ENTITY_REF; | |
3896 | dtd->hasParamEntityRefs = XML_TRUE; | |
3897 | if (!paramEntityParsing) | |
3898 | dtd->keepProcessing = dtd->standalone; | |
3899 | else { | |
3900 | const XML_Char *name; | |
3901 | ENTITY *entity; | |
3902 | name = poolStoreString(&dtd->pool, enc, | |
3903 | s + enc->minBytesPerChar, | |
3904 | next - enc->minBytesPerChar); | |
3905 | if (!name) | |
3906 | return XML_ERROR_NO_MEMORY; | |
3907 | entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0); | |
3908 | poolDiscard(&dtd->pool); | |
3909 | /* first, determine if a check for an existing declaration is needed; | |
3910 | if yes, check that the entity exists, and that it is internal, | |
3911 | otherwise call the skipped entity handler | |
3912 | */ | |
3913 | if (prologState.documentEntity && | |
3914 | (dtd->standalone | |
3915 | ? !openInternalEntities | |
3916 | : !dtd->hasParamEntityRefs)) { | |
3917 | if (!entity) | |
3918 | return XML_ERROR_UNDEFINED_ENTITY; | |
3919 | else if (!entity->is_internal) | |
3920 | return XML_ERROR_ENTITY_DECLARED_IN_PE; | |
3921 | } | |
3922 | else if (!entity) { | |
3923 | dtd->keepProcessing = dtd->standalone; | |
3924 | /* cannot report skipped entities in declarations */ | |
3925 | if ((role == XML_ROLE_PARAM_ENTITY_REF) && skippedEntityHandler) { | |
3926 | skippedEntityHandler(handlerArg, name, 1); | |
3927 | handleDefault = XML_FALSE; | |
3928 | } | |
3929 | break; | |
3930 | } | |
3931 | if (entity->open) | |
3932 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
3933 | if (entity->textPtr) { | |
3934 | enum XML_Error result; | |
3935 | result = processInternalParamEntity(parser, entity); | |
3936 | if (result != XML_ERROR_NONE) | |
3937 | return result; | |
3938 | handleDefault = XML_FALSE; | |
3939 | break; | |
3940 | } | |
3941 | if (externalEntityRefHandler) { | |
3942 | dtd->paramEntityRead = XML_FALSE; | |
3943 | entity->open = XML_TRUE; | |
3944 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
3945 | 0, | |
3946 | entity->base, | |
3947 | entity->systemId, | |
3948 | entity->publicId)) { | |
3949 | entity->open = XML_FALSE; | |
3950 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
3951 | } | |
3952 | entity->open = XML_FALSE; | |
3953 | handleDefault = XML_FALSE; | |
3954 | if (!dtd->paramEntityRead) { | |
3955 | dtd->keepProcessing = dtd->standalone; | |
3956 | break; | |
3957 | } | |
3958 | } | |
3959 | else { | |
3960 | dtd->keepProcessing = dtd->standalone; | |
3961 | break; | |
3962 | } | |
3963 | } | |
3964 | #endif /* XML_DTD */ | |
3965 | if (!dtd->standalone && | |
3966 | notStandaloneHandler && | |
3967 | !notStandaloneHandler(handlerArg)) | |
3968 | return XML_ERROR_NOT_STANDALONE; | |
3969 | break; | |
3970 | ||
3971 | /* Element declaration stuff */ | |
3972 | ||
3973 | case XML_ROLE_ELEMENT_NAME: | |
3974 | if (elementDeclHandler) { | |
3975 | declElementType = getElementType(parser, enc, s, next); | |
3976 | if (!declElementType) | |
3977 | return XML_ERROR_NO_MEMORY; | |
3978 | dtd->scaffLevel = 0; | |
3979 | dtd->scaffCount = 0; | |
3980 | dtd->in_eldecl = XML_TRUE; | |
3981 | handleDefault = XML_FALSE; | |
3982 | } | |
3983 | break; | |
3984 | ||
3985 | case XML_ROLE_CONTENT_ANY: | |
3986 | case XML_ROLE_CONTENT_EMPTY: | |
3987 | if (dtd->in_eldecl) { | |
3988 | if (elementDeclHandler) { | |
3989 | XML_Content * content = (XML_Content *) MALLOC(sizeof(XML_Content)); | |
3990 | if (!content) | |
3991 | return XML_ERROR_NO_MEMORY; | |
3992 | content->quant = XML_CQUANT_NONE; | |
3993 | content->name = NULL; | |
3994 | content->numchildren = 0; | |
3995 | content->children = NULL; | |
3996 | content->type = ((role == XML_ROLE_CONTENT_ANY) ? | |
3997 | XML_CTYPE_ANY : | |
3998 | XML_CTYPE_EMPTY); | |
3999 | *eventEndPP = s; | |
4000 | elementDeclHandler(handlerArg, declElementType->name, content); | |
4001 | handleDefault = XML_FALSE; | |
4002 | } | |
4003 | dtd->in_eldecl = XML_FALSE; | |
4004 | } | |
4005 | break; | |
4006 | ||
4007 | case XML_ROLE_CONTENT_PCDATA: | |
4008 | if (dtd->in_eldecl) { | |
4009 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel - 1]].type | |
4010 | = XML_CTYPE_MIXED; | |
4011 | if (elementDeclHandler) | |
4012 | handleDefault = XML_FALSE; | |
4013 | } | |
4014 | break; | |
4015 | ||
4016 | case XML_ROLE_CONTENT_ELEMENT: | |
4017 | quant = XML_CQUANT_NONE; | |
4018 | goto elementContent; | |
4019 | case XML_ROLE_CONTENT_ELEMENT_OPT: | |
4020 | quant = XML_CQUANT_OPT; | |
4021 | goto elementContent; | |
4022 | case XML_ROLE_CONTENT_ELEMENT_REP: | |
4023 | quant = XML_CQUANT_REP; | |
4024 | goto elementContent; | |
4025 | case XML_ROLE_CONTENT_ELEMENT_PLUS: | |
4026 | quant = XML_CQUANT_PLUS; | |
4027 | elementContent: | |
4028 | if (dtd->in_eldecl) { | |
4029 | ELEMENT_TYPE *el; | |
4030 | const XML_Char *name; | |
4031 | int nameLen; | |
4032 | const char *nxt = (quant == XML_CQUANT_NONE | |
4033 | ? next | |
4034 | : next - enc->minBytesPerChar); | |
4035 | int myindex = nextScaffoldPart(parser); | |
4036 | if (myindex < 0) | |
4037 | return XML_ERROR_NO_MEMORY; | |
4038 | dtd->scaffold[myindex].type = XML_CTYPE_NAME; | |
4039 | dtd->scaffold[myindex].quant = quant; | |
4040 | el = getElementType(parser, enc, s, nxt); | |
4041 | if (!el) | |
4042 | return XML_ERROR_NO_MEMORY; | |
4043 | name = el->name; | |
4044 | dtd->scaffold[myindex].name = name; | |
4045 | nameLen = 0; | |
4046 | for (; name[nameLen++]; ) | |
4047 | ; | |
4048 | dtd->contentStringLen += nameLen; | |
4049 | if (elementDeclHandler) | |
4050 | handleDefault = XML_FALSE; | |
4051 | } | |
4052 | break; | |
4053 | ||
4054 | case XML_ROLE_GROUP_CLOSE: | |
4055 | quant = XML_CQUANT_NONE; | |
4056 | goto closeGroup; | |
4057 | case XML_ROLE_GROUP_CLOSE_OPT: | |
4058 | quant = XML_CQUANT_OPT; | |
4059 | goto closeGroup; | |
4060 | case XML_ROLE_GROUP_CLOSE_REP: | |
4061 | quant = XML_CQUANT_REP; | |
4062 | goto closeGroup; | |
4063 | case XML_ROLE_GROUP_CLOSE_PLUS: | |
4064 | quant = XML_CQUANT_PLUS; | |
4065 | closeGroup: | |
4066 | if (dtd->in_eldecl) { | |
4067 | if (elementDeclHandler) | |
4068 | handleDefault = XML_FALSE; | |
4069 | dtd->scaffLevel--; | |
4070 | dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel]].quant = quant; | |
4071 | if (dtd->scaffLevel == 0) { | |
4072 | if (!handleDefault) { | |
4073 | XML_Content *model = build_model(parser); | |
4074 | if (!model) | |
4075 | return XML_ERROR_NO_MEMORY; | |
4076 | *eventEndPP = s; | |
4077 | elementDeclHandler(handlerArg, declElementType->name, model); | |
4078 | } | |
4079 | dtd->in_eldecl = XML_FALSE; | |
4080 | dtd->contentStringLen = 0; | |
4081 | } | |
4082 | } | |
4083 | break; | |
4084 | /* End element declaration stuff */ | |
4085 | ||
4086 | case XML_ROLE_PI: | |
4087 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
4088 | return XML_ERROR_NO_MEMORY; | |
4089 | handleDefault = XML_FALSE; | |
4090 | break; | |
4091 | case XML_ROLE_COMMENT: | |
4092 | if (!reportComment(parser, enc, s, next)) | |
4093 | return XML_ERROR_NO_MEMORY; | |
4094 | handleDefault = XML_FALSE; | |
4095 | break; | |
4096 | case XML_ROLE_NONE: | |
4097 | switch (tok) { | |
4098 | case XML_TOK_BOM: | |
4099 | handleDefault = XML_FALSE; | |
4100 | break; | |
4101 | } | |
4102 | break; | |
4103 | case XML_ROLE_DOCTYPE_NONE: | |
4104 | if (startDoctypeDeclHandler) | |
4105 | handleDefault = XML_FALSE; | |
4106 | break; | |
4107 | case XML_ROLE_ENTITY_NONE: | |
4108 | if (dtd->keepProcessing && entityDeclHandler) | |
4109 | handleDefault = XML_FALSE; | |
4110 | break; | |
4111 | case XML_ROLE_NOTATION_NONE: | |
4112 | if (notationDeclHandler) | |
4113 | handleDefault = XML_FALSE; | |
4114 | break; | |
4115 | case XML_ROLE_ATTLIST_NONE: | |
4116 | if (dtd->keepProcessing && attlistDeclHandler) | |
4117 | handleDefault = XML_FALSE; | |
4118 | break; | |
4119 | case XML_ROLE_ELEMENT_NONE: | |
4120 | if (elementDeclHandler) | |
4121 | handleDefault = XML_FALSE; | |
4122 | break; | |
4123 | } /* end of big switch */ | |
4124 | ||
4125 | if (handleDefault && defaultHandler) | |
4126 | reportDefault(parser, enc, s, next); | |
4127 | ||
4128 | s = next; | |
4129 | tok = XmlPrologTok(enc, s, end, &next); | |
4130 | } | |
4131 | /* not reached */ | |
4132 | } | |
4133 | ||
4134 | static enum XML_Error PTRCALL | |
4135 | epilogProcessor(XML_Parser parser, | |
4136 | const char *s, | |
4137 | const char *end, | |
4138 | const char **nextPtr) | |
4139 | { | |
4140 | processor = epilogProcessor; | |
4141 | eventPtr = s; | |
4142 | for (;;) { | |
4143 | const char *next = NULL; | |
4144 | int tok = XmlPrologTok(encoding, s, end, &next); | |
4145 | eventEndPtr = next; | |
4146 | switch (tok) { | |
4147 | /* report partial linebreak - it might be the last token */ | |
4148 | case -XML_TOK_PROLOG_S: | |
4149 | if (defaultHandler) { | |
4150 | eventEndPtr = next; | |
4151 | reportDefault(parser, encoding, s, next); | |
4152 | } | |
4153 | if (nextPtr) | |
4154 | *nextPtr = next; | |
4155 | return XML_ERROR_NONE; | |
4156 | case XML_TOK_NONE: | |
4157 | if (nextPtr) | |
4158 | *nextPtr = s; | |
4159 | return XML_ERROR_NONE; | |
4160 | case XML_TOK_PROLOG_S: | |
4161 | if (defaultHandler) | |
4162 | reportDefault(parser, encoding, s, next); | |
4163 | break; | |
4164 | case XML_TOK_PI: | |
4165 | if (!reportProcessingInstruction(parser, encoding, s, next)) | |
4166 | return XML_ERROR_NO_MEMORY; | |
4167 | break; | |
4168 | case XML_TOK_COMMENT: | |
4169 | if (!reportComment(parser, encoding, s, next)) | |
4170 | return XML_ERROR_NO_MEMORY; | |
4171 | break; | |
4172 | case XML_TOK_INVALID: | |
4173 | eventPtr = next; | |
4174 | return XML_ERROR_INVALID_TOKEN; | |
4175 | case XML_TOK_PARTIAL: | |
4176 | if (nextPtr) { | |
4177 | *nextPtr = s; | |
4178 | return XML_ERROR_NONE; | |
4179 | } | |
4180 | return XML_ERROR_UNCLOSED_TOKEN; | |
4181 | case XML_TOK_PARTIAL_CHAR: | |
4182 | if (nextPtr) { | |
4183 | *nextPtr = s; | |
4184 | return XML_ERROR_NONE; | |
4185 | } | |
4186 | return XML_ERROR_PARTIAL_CHAR; | |
4187 | default: | |
4188 | return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; | |
4189 | } | |
4190 | eventPtr = s = next; | |
4191 | } | |
4192 | } | |
4193 | ||
4194 | #ifdef XML_DTD | |
4195 | ||
4196 | static enum XML_Error | |
4197 | processInternalParamEntity(XML_Parser parser, ENTITY *entity) | |
4198 | { | |
4199 | const char *s, *end, *next; | |
4200 | int tok; | |
4201 | enum XML_Error result; | |
4202 | OPEN_INTERNAL_ENTITY openEntity; | |
4203 | entity->open = XML_TRUE; | |
4204 | openEntity.next = openInternalEntities; | |
4205 | openInternalEntities = &openEntity; | |
4206 | openEntity.entity = entity; | |
4207 | openEntity.internalEventPtr = NULL; | |
4208 | openEntity.internalEventEndPtr = NULL; | |
4209 | s = (char *)entity->textPtr; | |
4210 | end = (char *)(entity->textPtr + entity->textLen); | |
4211 | tok = XmlPrologTok(internalEncoding, s, end, &next); | |
4212 | result = doProlog(parser, internalEncoding, s, end, tok, next, 0); | |
4213 | entity->open = XML_FALSE; | |
4214 | openInternalEntities = openEntity.next; | |
4215 | return result; | |
4216 | } | |
4217 | ||
4218 | #endif /* XML_DTD */ | |
4219 | ||
4220 | static enum XML_Error PTRCALL | |
4221 | errorProcessor(XML_Parser parser, | |
4222 | const char *s, | |
4223 | const char *end, | |
4224 | const char **nextPtr) | |
4225 | { | |
4226 | return errorCode; | |
4227 | } | |
4228 | ||
4229 | static enum XML_Error | |
4230 | storeAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata, | |
4231 | const char *ptr, const char *end, | |
4232 | STRING_POOL *pool) | |
4233 | { | |
4234 | enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr, | |
4235 | end, pool); | |
4236 | if (result) | |
4237 | return result; | |
4238 | if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) | |
4239 | poolChop(pool); | |
4240 | if (!poolAppendChar(pool, XML_T('\0'))) | |
4241 | return XML_ERROR_NO_MEMORY; | |
4242 | return XML_ERROR_NONE; | |
4243 | } | |
4244 | ||
4245 | static enum XML_Error | |
4246 | appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata, | |
4247 | const char *ptr, const char *end, | |
4248 | STRING_POOL *pool) | |
4249 | { | |
4250 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4251 | for (;;) { | |
4252 | const char *next; | |
4253 | int tok = XmlAttributeValueTok(enc, ptr, end, &next); | |
4254 | switch (tok) { | |
4255 | case XML_TOK_NONE: | |
4256 | return XML_ERROR_NONE; | |
4257 | case XML_TOK_INVALID: | |
4258 | if (enc == encoding) | |
4259 | eventPtr = next; | |
4260 | return XML_ERROR_INVALID_TOKEN; | |
4261 | case XML_TOK_PARTIAL: | |
4262 | if (enc == encoding) | |
4263 | eventPtr = ptr; | |
4264 | return XML_ERROR_INVALID_TOKEN; | |
4265 | case XML_TOK_CHAR_REF: | |
4266 | { | |
4267 | XML_Char buf[XML_ENCODE_MAX]; | |
4268 | int i; | |
4269 | int n = XmlCharRefNumber(enc, ptr); | |
4270 | if (n < 0) { | |
4271 | if (enc == encoding) | |
4272 | eventPtr = ptr; | |
4273 | return XML_ERROR_BAD_CHAR_REF; | |
4274 | } | |
4275 | if (!isCdata | |
4276 | && n == 0x20 /* space */ | |
4277 | && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
4278 | break; | |
4279 | n = XmlEncode(n, (ICHAR *)buf); | |
4280 | if (!n) { | |
4281 | if (enc == encoding) | |
4282 | eventPtr = ptr; | |
4283 | return XML_ERROR_BAD_CHAR_REF; | |
4284 | } | |
4285 | for (i = 0; i < n; i++) { | |
4286 | if (!poolAppendChar(pool, buf[i])) | |
4287 | return XML_ERROR_NO_MEMORY; | |
4288 | } | |
4289 | } | |
4290 | break; | |
4291 | case XML_TOK_DATA_CHARS: | |
4292 | if (!poolAppend(pool, enc, ptr, next)) | |
4293 | return XML_ERROR_NO_MEMORY; | |
4294 | break; | |
4295 | case XML_TOK_TRAILING_CR: | |
4296 | next = ptr + enc->minBytesPerChar; | |
4297 | /* fall through */ | |
4298 | case XML_TOK_ATTRIBUTE_VALUE_S: | |
4299 | case XML_TOK_DATA_NEWLINE: | |
4300 | if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
4301 | break; | |
4302 | if (!poolAppendChar(pool, 0x20)) | |
4303 | return XML_ERROR_NO_MEMORY; | |
4304 | break; | |
4305 | case XML_TOK_ENTITY_REF: | |
4306 | { | |
4307 | const XML_Char *name; | |
4308 | ENTITY *entity; | |
4309 | char checkEntityDecl; | |
4310 | XML_Char ch = (XML_Char) XmlPredefinedEntityName(enc, | |
4311 | ptr + enc->minBytesPerChar, | |
4312 | next - enc->minBytesPerChar); | |
4313 | if (ch) { | |
4314 | if (!poolAppendChar(pool, ch)) | |
4315 | return XML_ERROR_NO_MEMORY; | |
4316 | break; | |
4317 | } | |
4318 | name = poolStoreString(&temp2Pool, enc, | |
4319 | ptr + enc->minBytesPerChar, | |
4320 | next - enc->minBytesPerChar); | |
4321 | if (!name) | |
4322 | return XML_ERROR_NO_MEMORY; | |
4323 | entity = (ENTITY *)lookup(&dtd->generalEntities, name, 0); | |
4324 | poolDiscard(&temp2Pool); | |
4325 | /* first, determine if a check for an existing declaration is needed; | |
4326 | if yes, check that the entity exists, and that it is internal, | |
4327 | otherwise call the default handler (if called from content) | |
4328 | */ | |
4329 | if (pool == &dtd->pool) /* are we called from prolog? */ | |
4330 | checkEntityDecl = | |
4331 | #ifdef XML_DTD | |
4332 | prologState.documentEntity && | |
4333 | #endif /* XML_DTD */ | |
4334 | (dtd->standalone | |
4335 | ? !openInternalEntities | |
4336 | : !dtd->hasParamEntityRefs); | |
4337 | else /* if (pool == &tempPool): we are called from content */ | |
4338 | checkEntityDecl = !dtd->hasParamEntityRefs || dtd->standalone; | |
4339 | if (checkEntityDecl) { | |
4340 | if (!entity) | |
4341 | return XML_ERROR_UNDEFINED_ENTITY; | |
4342 | else if (!entity->is_internal) | |
4343 | return XML_ERROR_ENTITY_DECLARED_IN_PE; | |
4344 | } | |
4345 | else if (!entity) { | |
4346 | /* cannot report skipped entity here - see comments on | |
4347 | skippedEntityHandler | |
4348 | if (skippedEntityHandler) | |
4349 | skippedEntityHandler(handlerArg, name, 0); | |
4350 | */ | |
4351 | if ((pool == &tempPool) && defaultHandler) | |
4352 | reportDefault(parser, enc, ptr, next); | |
4353 | break; | |
4354 | } | |
4355 | if (entity->open) { | |
4356 | if (enc == encoding) | |
4357 | eventPtr = ptr; | |
4358 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
4359 | } | |
4360 | if (entity->notation) { | |
4361 | if (enc == encoding) | |
4362 | eventPtr = ptr; | |
4363 | return XML_ERROR_BINARY_ENTITY_REF; | |
4364 | } | |
4365 | if (!entity->textPtr) { | |
4366 | if (enc == encoding) | |
4367 | eventPtr = ptr; | |
4368 | return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; | |
4369 | } | |
4370 | else { | |
4371 | enum XML_Error result; | |
4372 | const XML_Char *textEnd = entity->textPtr + entity->textLen; | |
4373 | entity->open = XML_TRUE; | |
4374 | result = appendAttributeValue(parser, internalEncoding, isCdata, | |
4375 | (char *)entity->textPtr, | |
4376 | (char *)textEnd, pool); | |
4377 | entity->open = XML_FALSE; | |
4378 | if (result) | |
4379 | return result; | |
4380 | } | |
4381 | } | |
4382 | break; | |
4383 | default: | |
4384 | if (enc == encoding) | |
4385 | eventPtr = ptr; | |
4386 | return XML_ERROR_UNEXPECTED_STATE; | |
4387 | } | |
4388 | ptr = next; | |
4389 | } | |
4390 | /* not reached */ | |
4391 | } | |
4392 | ||
4393 | static enum XML_Error | |
4394 | storeEntityValue(XML_Parser parser, | |
4395 | const ENCODING *enc, | |
4396 | const char *entityTextPtr, | |
4397 | const char *entityTextEnd) | |
4398 | { | |
4399 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4400 | STRING_POOL *pool = &(dtd->entityValuePool); | |
4401 | enum XML_Error result = XML_ERROR_NONE; | |
4402 | #ifdef XML_DTD | |
4403 | int oldInEntityValue = prologState.inEntityValue; | |
4404 | prologState.inEntityValue = 1; | |
4405 | #endif /* XML_DTD */ | |
4406 | /* never return Null for the value argument in EntityDeclHandler, | |
4407 | since this would indicate an external entity; therefore we | |
4408 | have to make sure that entityValuePool.start is not null */ | |
4409 | if (!pool->blocks) { | |
4410 | if (!poolGrow(pool)) | |
4411 | return XML_ERROR_NO_MEMORY; | |
4412 | } | |
4413 | ||
4414 | for (;;) { | |
4415 | const char *next; | |
4416 | int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next); | |
4417 | switch (tok) { | |
4418 | case XML_TOK_PARAM_ENTITY_REF: | |
4419 | #ifdef XML_DTD | |
4420 | if (isParamEntity || enc != encoding) { | |
4421 | const XML_Char *name; | |
4422 | ENTITY *entity; | |
4423 | name = poolStoreString(&tempPool, enc, | |
4424 | entityTextPtr + enc->minBytesPerChar, | |
4425 | next - enc->minBytesPerChar); | |
4426 | if (!name) { | |
4427 | result = XML_ERROR_NO_MEMORY; | |
4428 | goto endEntityValue; | |
4429 | } | |
4430 | entity = (ENTITY *)lookup(&dtd->paramEntities, name, 0); | |
4431 | poolDiscard(&tempPool); | |
4432 | if (!entity) { | |
4433 | /* not a well-formedness error - see XML 1.0: WFC Entity Declared */ | |
4434 | /* cannot report skipped entity here - see comments on | |
4435 | skippedEntityHandler | |
4436 | if (skippedEntityHandler) | |
4437 | skippedEntityHandler(handlerArg, name, 0); | |
4438 | */ | |
4439 | dtd->keepProcessing = dtd->standalone; | |
4440 | goto endEntityValue; | |
4441 | } | |
4442 | if (entity->open) { | |
4443 | if (enc == encoding) | |
4444 | eventPtr = entityTextPtr; | |
4445 | result = XML_ERROR_RECURSIVE_ENTITY_REF; | |
4446 | goto endEntityValue; | |
4447 | } | |
4448 | if (entity->systemId) { | |
4449 | if (externalEntityRefHandler) { | |
4450 | dtd->paramEntityRead = XML_FALSE; | |
4451 | entity->open = XML_TRUE; | |
4452 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
4453 | 0, | |
4454 | entity->base, | |
4455 | entity->systemId, | |
4456 | entity->publicId)) { | |
4457 | entity->open = XML_FALSE; | |
4458 | result = XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
4459 | goto endEntityValue; | |
4460 | } | |
4461 | entity->open = XML_FALSE; | |
4462 | if (!dtd->paramEntityRead) | |
4463 | dtd->keepProcessing = dtd->standalone; | |
4464 | } | |
4465 | else | |
4466 | dtd->keepProcessing = dtd->standalone; | |
4467 | } | |
4468 | else { | |
4469 | entity->open = XML_TRUE; | |
4470 | result = storeEntityValue(parser, | |
4471 | internalEncoding, | |
4472 | (char *)entity->textPtr, | |
4473 | (char *)(entity->textPtr | |
4474 | + entity->textLen)); | |
4475 | entity->open = XML_FALSE; | |
4476 | if (result) | |
4477 | goto endEntityValue; | |
4478 | } | |
4479 | break; | |
4480 | } | |
4481 | #endif /* XML_DTD */ | |
4482 | /* in the internal subset, PE references are not legal | |
4483 | within markup declarations, e.g entity values in this case */ | |
4484 | eventPtr = entityTextPtr; | |
4485 | result = XML_ERROR_PARAM_ENTITY_REF; | |
4486 | goto endEntityValue; | |
4487 | case XML_TOK_NONE: | |
4488 | result = XML_ERROR_NONE; | |
4489 | goto endEntityValue; | |
4490 | case XML_TOK_ENTITY_REF: | |
4491 | case XML_TOK_DATA_CHARS: | |
4492 | if (!poolAppend(pool, enc, entityTextPtr, next)) { | |
4493 | result = XML_ERROR_NO_MEMORY; | |
4494 | goto endEntityValue; | |
4495 | } | |
4496 | break; | |
4497 | case XML_TOK_TRAILING_CR: | |
4498 | next = entityTextPtr + enc->minBytesPerChar; | |
4499 | /* fall through */ | |
4500 | case XML_TOK_DATA_NEWLINE: | |
4501 | if (pool->end == pool->ptr && !poolGrow(pool)) { | |
4502 | result = XML_ERROR_NO_MEMORY; | |
4503 | goto endEntityValue; | |
4504 | } | |
4505 | *(pool->ptr)++ = 0xA; | |
4506 | break; | |
4507 | case XML_TOK_CHAR_REF: | |
4508 | { | |
4509 | XML_Char buf[XML_ENCODE_MAX]; | |
4510 | int i; | |
4511 | int n = XmlCharRefNumber(enc, entityTextPtr); | |
4512 | if (n < 0) { | |
4513 | if (enc == encoding) | |
4514 | eventPtr = entityTextPtr; | |
4515 | result = XML_ERROR_BAD_CHAR_REF; | |
4516 | goto endEntityValue; | |
4517 | } | |
4518 | n = XmlEncode(n, (ICHAR *)buf); | |
4519 | if (!n) { | |
4520 | if (enc == encoding) | |
4521 | eventPtr = entityTextPtr; | |
4522 | result = XML_ERROR_BAD_CHAR_REF; | |
4523 | goto endEntityValue; | |
4524 | } | |
4525 | for (i = 0; i < n; i++) { | |
4526 | if (pool->end == pool->ptr && !poolGrow(pool)) { | |
4527 | result = XML_ERROR_NO_MEMORY; | |
4528 | goto endEntityValue; | |
4529 | } | |
4530 | *(pool->ptr)++ = buf[i]; | |
4531 | } | |
4532 | } | |
4533 | break; | |
4534 | case XML_TOK_PARTIAL: | |
4535 | if (enc == encoding) | |
4536 | eventPtr = entityTextPtr; | |
4537 | result = XML_ERROR_INVALID_TOKEN; | |
4538 | goto endEntityValue; | |
4539 | case XML_TOK_INVALID: | |
4540 | if (enc == encoding) | |
4541 | eventPtr = next; | |
4542 | result = XML_ERROR_INVALID_TOKEN; | |
4543 | goto endEntityValue; | |
4544 | default: | |
4545 | if (enc == encoding) | |
4546 | eventPtr = entityTextPtr; | |
4547 | result = XML_ERROR_UNEXPECTED_STATE; | |
4548 | goto endEntityValue; | |
4549 | } | |
4550 | entityTextPtr = next; | |
4551 | } | |
4552 | endEntityValue: | |
4553 | #ifdef XML_DTD | |
4554 | prologState.inEntityValue = oldInEntityValue; | |
4555 | #endif /* XML_DTD */ | |
4556 | return result; | |
4557 | } | |
4558 | ||
4559 | static void FASTCALL | |
4560 | normalizeLines(XML_Char *s) | |
4561 | { | |
4562 | XML_Char *p; | |
4563 | for (;; s++) { | |
4564 | if (*s == XML_T('\0')) | |
4565 | return; | |
4566 | if (*s == 0xD) | |
4567 | break; | |
4568 | } | |
4569 | p = s; | |
4570 | do { | |
4571 | if (*s == 0xD) { | |
4572 | *p++ = 0xA; | |
4573 | if (*++s == 0xA) | |
4574 | s++; | |
4575 | } | |
4576 | else | |
4577 | *p++ = *s++; | |
4578 | } while (*s); | |
4579 | *p = XML_T('\0'); | |
4580 | } | |
4581 | ||
4582 | static int | |
4583 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, | |
4584 | const char *start, const char *end) | |
4585 | { | |
4586 | const XML_Char *target; | |
4587 | XML_Char *data; | |
4588 | const char *tem; | |
4589 | if (!processingInstructionHandler) { | |
4590 | if (defaultHandler) | |
4591 | reportDefault(parser, enc, start, end); | |
4592 | return 1; | |
4593 | } | |
4594 | start += enc->minBytesPerChar * 2; | |
4595 | tem = start + XmlNameLength(enc, start); | |
4596 | target = poolStoreString(&tempPool, enc, start, tem); | |
4597 | if (!target) | |
4598 | return 0; | |
4599 | poolFinish(&tempPool); | |
4600 | data = poolStoreString(&tempPool, enc, | |
4601 | XmlSkipS(enc, tem), | |
4602 | end - enc->minBytesPerChar*2); | |
4603 | if (!data) | |
4604 | return 0; | |
4605 | normalizeLines(data); | |
4606 | processingInstructionHandler(handlerArg, target, data); | |
4607 | poolClear(&tempPool); | |
4608 | return 1; | |
4609 | } | |
4610 | ||
4611 | static int | |
4612 | reportComment(XML_Parser parser, const ENCODING *enc, | |
4613 | const char *start, const char *end) | |
4614 | { | |
4615 | XML_Char *data; | |
4616 | if (!commentHandler) { | |
4617 | if (defaultHandler) | |
4618 | reportDefault(parser, enc, start, end); | |
4619 | return 1; | |
4620 | } | |
4621 | data = poolStoreString(&tempPool, | |
4622 | enc, | |
4623 | start + enc->minBytesPerChar * 4, | |
4624 | end - enc->minBytesPerChar * 3); | |
4625 | if (!data) | |
4626 | return 0; | |
4627 | normalizeLines(data); | |
4628 | commentHandler(handlerArg, data); | |
4629 | poolClear(&tempPool); | |
4630 | return 1; | |
4631 | } | |
4632 | ||
4633 | static void | |
4634 | reportDefault(XML_Parser parser, const ENCODING *enc, | |
4635 | const char *s, const char *end) | |
4636 | { | |
4637 | if (MUST_CONVERT(enc, s)) { | |
4638 | const char **eventPP; | |
4639 | const char **eventEndPP; | |
4640 | if (enc == encoding) { | |
4641 | eventPP = &eventPtr; | |
4642 | eventEndPP = &eventEndPtr; | |
4643 | } | |
4644 | else { | |
4645 | eventPP = &(openInternalEntities->internalEventPtr); | |
4646 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
4647 | } | |
4648 | do { | |
4649 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
4650 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
4651 | *eventEndPP = s; | |
4652 | defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
4653 | *eventPP = s; | |
4654 | } while (s != end); | |
4655 | } | |
4656 | else | |
4657 | defaultHandler(handlerArg, (XML_Char *)s, (XML_Char *)end - (XML_Char *)s); | |
4658 | } | |
4659 | ||
4660 | ||
4661 | static int | |
4662 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, XML_Bool isCdata, | |
4663 | XML_Bool isId, const XML_Char *value, XML_Parser parser) | |
4664 | { | |
4665 | DEFAULT_ATTRIBUTE *att; | |
4666 | if (value || isId) { | |
4667 | /* The handling of default attributes gets messed up if we have | |
4668 | a default which duplicates a non-default. */ | |
4669 | int i; | |
4670 | for (i = 0; i < type->nDefaultAtts; i++) | |
4671 | if (attId == type->defaultAtts[i].id) | |
4672 | return 1; | |
4673 | if (isId && !type->idAtt && !attId->xmlns) | |
4674 | type->idAtt = attId; | |
4675 | } | |
4676 | if (type->nDefaultAtts == type->allocDefaultAtts) { | |
4677 | if (type->allocDefaultAtts == 0) { | |
4678 | type->allocDefaultAtts = 8; | |
4679 | type->defaultAtts = (DEFAULT_ATTRIBUTE *)MALLOC(type->allocDefaultAtts | |
4680 | * sizeof(DEFAULT_ATTRIBUTE)); | |
4681 | if (!type->defaultAtts) | |
4682 | return 0; | |
4683 | } | |
4684 | else { | |
4685 | DEFAULT_ATTRIBUTE *temp; | |
4686 | int count = type->allocDefaultAtts * 2; | |
4687 | temp = (DEFAULT_ATTRIBUTE *) | |
4688 | REALLOC(type->defaultAtts, (count * sizeof(DEFAULT_ATTRIBUTE))); | |
4689 | if (temp == NULL) | |
4690 | return 0; | |
4691 | type->allocDefaultAtts = count; | |
4692 | type->defaultAtts = temp; | |
4693 | } | |
4694 | } | |
4695 | att = type->defaultAtts + type->nDefaultAtts; | |
4696 | att->id = attId; | |
4697 | att->value = value; | |
4698 | att->isCdata = isCdata; | |
4699 | if (!isCdata) | |
4700 | attId->maybeTokenized = XML_TRUE; | |
4701 | type->nDefaultAtts += 1; | |
4702 | return 1; | |
4703 | } | |
4704 | ||
4705 | static int | |
4706 | setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) | |
4707 | { | |
4708 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4709 | const XML_Char *name; | |
4710 | for (name = elementType->name; *name; name++) { | |
4711 | if (*name == XML_T(':')) { | |
4712 | PREFIX *prefix; | |
4713 | const XML_Char *s; | |
4714 | for (s = elementType->name; s != name; s++) { | |
4715 | if (!poolAppendChar(&dtd->pool, *s)) | |
4716 | return 0; | |
4717 | } | |
4718 | if (!poolAppendChar(&dtd->pool, XML_T('\0'))) | |
4719 | return 0; | |
4720 | prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool), | |
4721 | sizeof(PREFIX)); | |
4722 | if (!prefix) | |
4723 | return 0; | |
4724 | if (prefix->name == poolStart(&dtd->pool)) | |
4725 | poolFinish(&dtd->pool); | |
4726 | else | |
4727 | poolDiscard(&dtd->pool); | |
4728 | elementType->prefix = prefix; | |
4729 | ||
4730 | } | |
4731 | } | |
4732 | return 1; | |
4733 | } | |
4734 | ||
4735 | static ATTRIBUTE_ID * | |
4736 | getAttributeId(XML_Parser parser, const ENCODING *enc, | |
4737 | const char *start, const char *end) | |
4738 | { | |
4739 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4740 | ATTRIBUTE_ID *id; | |
4741 | const XML_Char *name; | |
4742 | if (!poolAppendChar(&dtd->pool, XML_T('\0'))) | |
4743 | return NULL; | |
4744 | name = poolStoreString(&dtd->pool, enc, start, end); | |
4745 | if (!name) | |
4746 | return NULL; | |
4747 | ++name; | |
4748 | id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, name, sizeof(ATTRIBUTE_ID)); | |
4749 | if (!id) | |
4750 | return NULL; | |
4751 | if (id->name != name) | |
4752 | poolDiscard(&dtd->pool); | |
4753 | else { | |
4754 | poolFinish(&dtd->pool); | |
4755 | if (!ns) | |
4756 | ; | |
4757 | else if (name[0] == XML_T('x') | |
4758 | && name[1] == XML_T('m') | |
4759 | && name[2] == XML_T('l') | |
4760 | && name[3] == XML_T('n') | |
4761 | && name[4] == XML_T('s') | |
4762 | && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) { | |
4763 | if (name[5] == XML_T('\0')) | |
4764 | id->prefix = &dtd->defaultPrefix; | |
4765 | else | |
4766 | id->prefix = (PREFIX *)lookup(&dtd->prefixes, name + 6, sizeof(PREFIX)); | |
4767 | id->xmlns = XML_TRUE; | |
4768 | } | |
4769 | else { | |
4770 | int i; | |
4771 | for (i = 0; name[i]; i++) { | |
4772 | if (name[i] == XML_T(':')) { | |
4773 | int j; | |
4774 | for (j = 0; j < i; j++) { | |
4775 | if (!poolAppendChar(&dtd->pool, name[j])) | |
4776 | return NULL; | |
4777 | } | |
4778 | if (!poolAppendChar(&dtd->pool, XML_T('\0'))) | |
4779 | return NULL; | |
4780 | id->prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&dtd->pool), | |
4781 | sizeof(PREFIX)); | |
4782 | if (id->prefix->name == poolStart(&dtd->pool)) | |
4783 | poolFinish(&dtd->pool); | |
4784 | else | |
4785 | poolDiscard(&dtd->pool); | |
4786 | break; | |
4787 | } | |
4788 | } | |
4789 | } | |
4790 | } | |
4791 | return id; | |
4792 | } | |
4793 | ||
4794 | #define CONTEXT_SEP XML_T('\f') | |
4795 | ||
4796 | static const XML_Char * | |
4797 | getContext(XML_Parser parser) | |
4798 | { | |
4799 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4800 | HASH_TABLE_ITER iter; | |
4801 | XML_Bool needSep = XML_FALSE; | |
4802 | ||
4803 | if (dtd->defaultPrefix.binding) { | |
4804 | int i; | |
4805 | int len; | |
4806 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
4807 | return NULL; | |
4808 | len = dtd->defaultPrefix.binding->uriLen; | |
4809 | if (namespaceSeparator != XML_T('\0')) | |
4810 | len--; | |
4811 | for (i = 0; i < len; i++) | |
4812 | if (!poolAppendChar(&tempPool, dtd->defaultPrefix.binding->uri[i])) | |
4813 | return NULL; | |
4814 | needSep = XML_TRUE; | |
4815 | } | |
4816 | ||
4817 | hashTableIterInit(&iter, &(dtd->prefixes)); | |
4818 | for (;;) { | |
4819 | int i; | |
4820 | int len; | |
4821 | const XML_Char *s; | |
4822 | PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); | |
4823 | if (!prefix) | |
4824 | break; | |
4825 | if (!prefix->binding) | |
4826 | continue; | |
4827 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
4828 | return NULL; | |
4829 | for (s = prefix->name; *s; s++) | |
4830 | if (!poolAppendChar(&tempPool, *s)) | |
4831 | return NULL; | |
4832 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
4833 | return NULL; | |
4834 | len = prefix->binding->uriLen; | |
4835 | if (namespaceSeparator != XML_T('\0')) | |
4836 | len--; | |
4837 | for (i = 0; i < len; i++) | |
4838 | if (!poolAppendChar(&tempPool, prefix->binding->uri[i])) | |
4839 | return NULL; | |
4840 | needSep = XML_TRUE; | |
4841 | } | |
4842 | ||
4843 | ||
4844 | hashTableIterInit(&iter, &(dtd->generalEntities)); | |
4845 | for (;;) { | |
4846 | const XML_Char *s; | |
4847 | ENTITY *e = (ENTITY *)hashTableIterNext(&iter); | |
4848 | if (!e) | |
4849 | break; | |
4850 | if (!e->open) | |
4851 | continue; | |
4852 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
4853 | return NULL; | |
4854 | for (s = e->name; *s; s++) | |
4855 | if (!poolAppendChar(&tempPool, *s)) | |
4856 | return 0; | |
4857 | needSep = XML_TRUE; | |
4858 | } | |
4859 | ||
4860 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4861 | return NULL; | |
4862 | return tempPool.start; | |
4863 | } | |
4864 | ||
4865 | static XML_Bool | |
4866 | setContext(XML_Parser parser, const XML_Char *context) | |
4867 | { | |
4868 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
4869 | const XML_Char *s = context; | |
4870 | ||
4871 | while (*context != XML_T('\0')) { | |
4872 | if (*s == CONTEXT_SEP || *s == XML_T('\0')) { | |
4873 | ENTITY *e; | |
4874 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4875 | return XML_FALSE; | |
4876 | e = (ENTITY *)lookup(&dtd->generalEntities, poolStart(&tempPool), 0); | |
4877 | if (e) | |
4878 | e->open = XML_TRUE; | |
4879 | if (*s != XML_T('\0')) | |
4880 | s++; | |
4881 | context = s; | |
4882 | poolDiscard(&tempPool); | |
4883 | } | |
4884 | else if (*s == XML_T('=')) { | |
4885 | PREFIX *prefix; | |
4886 | if (poolLength(&tempPool) == 0) | |
4887 | prefix = &dtd->defaultPrefix; | |
4888 | else { | |
4889 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4890 | return XML_FALSE; | |
4891 | prefix = (PREFIX *)lookup(&dtd->prefixes, poolStart(&tempPool), | |
4892 | sizeof(PREFIX)); | |
4893 | if (!prefix) | |
4894 | return XML_FALSE; | |
4895 | if (prefix->name == poolStart(&tempPool)) { | |
4896 | prefix->name = poolCopyString(&dtd->pool, prefix->name); | |
4897 | if (!prefix->name) | |
4898 | return XML_FALSE; | |
4899 | } | |
4900 | poolDiscard(&tempPool); | |
4901 | } | |
4902 | for (context = s + 1; | |
4903 | *context != CONTEXT_SEP && *context != XML_T('\0'); | |
4904 | context++) | |
4905 | if (!poolAppendChar(&tempPool, *context)) | |
4906 | return XML_FALSE; | |
4907 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
4908 | return XML_FALSE; | |
4909 | if (addBinding(parser, prefix, 0, poolStart(&tempPool), | |
4910 | &inheritedBindings) != XML_ERROR_NONE) | |
4911 | return XML_FALSE; | |
4912 | poolDiscard(&tempPool); | |
4913 | if (*context != XML_T('\0')) | |
4914 | ++context; | |
4915 | s = context; | |
4916 | } | |
4917 | else { | |
4918 | if (!poolAppendChar(&tempPool, *s)) | |
4919 | return XML_FALSE; | |
4920 | s++; | |
4921 | } | |
4922 | } | |
4923 | return XML_TRUE; | |
4924 | } | |
4925 | ||
4926 | static void FASTCALL | |
4927 | normalizePublicId(XML_Char *publicId) | |
4928 | { | |
4929 | XML_Char *p = publicId; | |
4930 | XML_Char *s; | |
4931 | for (s = publicId; *s; s++) { | |
4932 | switch (*s) { | |
4933 | case 0x20: | |
4934 | case 0xD: | |
4935 | case 0xA: | |
4936 | if (p != publicId && p[-1] != 0x20) | |
4937 | *p++ = 0x20; | |
4938 | break; | |
4939 | default: | |
4940 | *p++ = *s; | |
4941 | } | |
4942 | } | |
4943 | if (p != publicId && p[-1] == 0x20) | |
4944 | --p; | |
4945 | *p = XML_T('\0'); | |
4946 | } | |
4947 | ||
4948 | static DTD * | |
4949 | dtdCreate(const XML_Memory_Handling_Suite *ms) | |
4950 | { | |
4951 | DTD *p = (DTD *)ms->malloc_fcn(sizeof(DTD)); | |
4952 | if (p == NULL) | |
4953 | return p; | |
4954 | poolInit(&(p->pool), ms); | |
4955 | #ifdef XML_DTD | |
4956 | poolInit(&(p->entityValuePool), ms); | |
4957 | #endif /* XML_DTD */ | |
4958 | hashTableInit(&(p->generalEntities), ms); | |
4959 | hashTableInit(&(p->elementTypes), ms); | |
4960 | hashTableInit(&(p->attributeIds), ms); | |
4961 | hashTableInit(&(p->prefixes), ms); | |
4962 | #ifdef XML_DTD | |
4963 | p->paramEntityRead = XML_FALSE; | |
4964 | hashTableInit(&(p->paramEntities), ms); | |
4965 | #endif /* XML_DTD */ | |
4966 | p->defaultPrefix.name = NULL; | |
4967 | p->defaultPrefix.binding = NULL; | |
4968 | ||
4969 | p->in_eldecl = XML_FALSE; | |
4970 | p->scaffIndex = NULL; | |
4971 | p->scaffold = NULL; | |
4972 | p->scaffLevel = 0; | |
4973 | p->scaffSize = 0; | |
4974 | p->scaffCount = 0; | |
4975 | p->contentStringLen = 0; | |
4976 | ||
4977 | p->keepProcessing = XML_TRUE; | |
4978 | p->hasParamEntityRefs = XML_FALSE; | |
4979 | p->standalone = XML_FALSE; | |
4980 | return p; | |
4981 | } | |
4982 | ||
4983 | static void | |
4984 | dtdReset(DTD *p, const XML_Memory_Handling_Suite *ms) | |
4985 | { | |
4986 | HASH_TABLE_ITER iter; | |
4987 | hashTableIterInit(&iter, &(p->elementTypes)); | |
4988 | for (;;) { | |
4989 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
4990 | if (!e) | |
4991 | break; | |
4992 | if (e->allocDefaultAtts != 0) | |
4993 | ms->free_fcn(e->defaultAtts); | |
4994 | } | |
4995 | hashTableClear(&(p->generalEntities)); | |
4996 | #ifdef XML_DTD | |
4997 | p->paramEntityRead = XML_FALSE; | |
4998 | hashTableClear(&(p->paramEntities)); | |
4999 | #endif /* XML_DTD */ | |
5000 | hashTableClear(&(p->elementTypes)); | |
5001 | hashTableClear(&(p->attributeIds)); | |
5002 | hashTableClear(&(p->prefixes)); | |
5003 | poolClear(&(p->pool)); | |
5004 | #ifdef XML_DTD | |
5005 | poolClear(&(p->entityValuePool)); | |
5006 | #endif /* XML_DTD */ | |
5007 | p->defaultPrefix.name = NULL; | |
5008 | p->defaultPrefix.binding = NULL; | |
5009 | ||
5010 | p->in_eldecl = XML_FALSE; | |
5011 | if (p->scaffIndex) { | |
5012 | ms->free_fcn(p->scaffIndex); | |
5013 | p->scaffIndex = NULL; | |
5014 | } | |
5015 | if (p->scaffold) { | |
5016 | ms->free_fcn(p->scaffold); | |
5017 | p->scaffold = NULL; | |
5018 | } | |
5019 | p->scaffLevel = 0; | |
5020 | p->scaffSize = 0; | |
5021 | p->scaffCount = 0; | |
5022 | p->contentStringLen = 0; | |
5023 | ||
5024 | p->keepProcessing = XML_TRUE; | |
5025 | p->hasParamEntityRefs = XML_FALSE; | |
5026 | p->standalone = XML_FALSE; | |
5027 | } | |
5028 | ||
5029 | static void | |
5030 | dtdDestroy(DTD *p, XML_Bool isDocEntity, const XML_Memory_Handling_Suite *ms) | |
5031 | { | |
5032 | HASH_TABLE_ITER iter; | |
5033 | hashTableIterInit(&iter, &(p->elementTypes)); | |
5034 | for (;;) { | |
5035 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
5036 | if (!e) | |
5037 | break; | |
5038 | if (e->allocDefaultAtts != 0) | |
5039 | ms->free_fcn(e->defaultAtts); | |
5040 | } | |
5041 | hashTableDestroy(&(p->generalEntities)); | |
5042 | #ifdef XML_DTD | |
5043 | hashTableDestroy(&(p->paramEntities)); | |
5044 | #endif /* XML_DTD */ | |
5045 | hashTableDestroy(&(p->elementTypes)); | |
5046 | hashTableDestroy(&(p->attributeIds)); | |
5047 | hashTableDestroy(&(p->prefixes)); | |
5048 | poolDestroy(&(p->pool)); | |
5049 | #ifdef XML_DTD | |
5050 | poolDestroy(&(p->entityValuePool)); | |
5051 | #endif /* XML_DTD */ | |
5052 | if (isDocEntity) { | |
5053 | if (p->scaffIndex) | |
5054 | ms->free_fcn(p->scaffIndex); | |
5055 | if (p->scaffold) | |
5056 | ms->free_fcn(p->scaffold); | |
5057 | } | |
5058 | ms->free_fcn(p); | |
5059 | } | |
5060 | ||
5061 | /* Do a deep copy of the DTD. Return 0 for out of memory, non-zero otherwise. | |
5062 | The new DTD has already been initialized. | |
5063 | */ | |
5064 | static int | |
5065 | dtdCopy(DTD *newDtd, const DTD *oldDtd, const XML_Memory_Handling_Suite *ms) | |
5066 | { | |
5067 | HASH_TABLE_ITER iter; | |
5068 | ||
5069 | /* Copy the prefix table. */ | |
5070 | ||
5071 | hashTableIterInit(&iter, &(oldDtd->prefixes)); | |
5072 | for (;;) { | |
5073 | const XML_Char *name; | |
5074 | const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); | |
5075 | if (!oldP) | |
5076 | break; | |
5077 | name = poolCopyString(&(newDtd->pool), oldP->name); | |
5078 | if (!name) | |
5079 | return 0; | |
5080 | if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX))) | |
5081 | return 0; | |
5082 | } | |
5083 | ||
5084 | hashTableIterInit(&iter, &(oldDtd->attributeIds)); | |
5085 | ||
5086 | /* Copy the attribute id table. */ | |
5087 | ||
5088 | for (;;) { | |
5089 | ATTRIBUTE_ID *newA; | |
5090 | const XML_Char *name; | |
5091 | const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); | |
5092 | ||
5093 | if (!oldA) | |
5094 | break; | |
5095 | /* Remember to allocate the scratch byte before the name. */ | |
5096 | if (!poolAppendChar(&(newDtd->pool), XML_T('\0'))) | |
5097 | return 0; | |
5098 | name = poolCopyString(&(newDtd->pool), oldA->name); | |
5099 | if (!name) | |
5100 | return 0; | |
5101 | ++name; | |
5102 | newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name, | |
5103 | sizeof(ATTRIBUTE_ID)); | |
5104 | if (!newA) | |
5105 | return 0; | |
5106 | newA->maybeTokenized = oldA->maybeTokenized; | |
5107 | if (oldA->prefix) { | |
5108 | newA->xmlns = oldA->xmlns; | |
5109 | if (oldA->prefix == &oldDtd->defaultPrefix) | |
5110 | newA->prefix = &newDtd->defaultPrefix; | |
5111 | else | |
5112 | newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes), | |
5113 | oldA->prefix->name, 0); | |
5114 | } | |
5115 | } | |
5116 | ||
5117 | /* Copy the element type table. */ | |
5118 | ||
5119 | hashTableIterInit(&iter, &(oldDtd->elementTypes)); | |
5120 | ||
5121 | for (;;) { | |
5122 | int i; | |
5123 | ELEMENT_TYPE *newE; | |
5124 | const XML_Char *name; | |
5125 | const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
5126 | if (!oldE) | |
5127 | break; | |
5128 | name = poolCopyString(&(newDtd->pool), oldE->name); | |
5129 | if (!name) | |
5130 | return 0; | |
5131 | newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name, | |
5132 | sizeof(ELEMENT_TYPE)); | |
5133 | if (!newE) | |
5134 | return 0; | |
5135 | if (oldE->nDefaultAtts) { | |
5136 | newE->defaultAtts = (DEFAULT_ATTRIBUTE *) | |
5137 | ms->malloc_fcn(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); | |
5138 | if (!newE->defaultAtts) { | |
5139 | ms->free_fcn(newE); | |
5140 | return 0; | |
5141 | } | |
5142 | } | |
5143 | if (oldE->idAtt) | |
5144 | newE->idAtt = (ATTRIBUTE_ID *) | |
5145 | lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0); | |
5146 | newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; | |
5147 | if (oldE->prefix) | |
5148 | newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes), | |
5149 | oldE->prefix->name, 0); | |
5150 | for (i = 0; i < newE->nDefaultAtts; i++) { | |
5151 | newE->defaultAtts[i].id = (ATTRIBUTE_ID *) | |
5152 | lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); | |
5153 | newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; | |
5154 | if (oldE->defaultAtts[i].value) { | |
5155 | newE->defaultAtts[i].value | |
5156 | = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); | |
5157 | if (!newE->defaultAtts[i].value) | |
5158 | return 0; | |
5159 | } | |
5160 | else | |
5161 | newE->defaultAtts[i].value = NULL; | |
5162 | } | |
5163 | } | |
5164 | ||
5165 | /* Copy the entity tables. */ | |
5166 | if (!copyEntityTable(&(newDtd->generalEntities), | |
5167 | &(newDtd->pool), | |
5168 | &(oldDtd->generalEntities))) | |
5169 | return 0; | |
5170 | ||
5171 | #ifdef XML_DTD | |
5172 | if (!copyEntityTable(&(newDtd->paramEntities), | |
5173 | &(newDtd->pool), | |
5174 | &(oldDtd->paramEntities))) | |
5175 | return 0; | |
5176 | newDtd->paramEntityRead = oldDtd->paramEntityRead; | |
5177 | #endif /* XML_DTD */ | |
5178 | ||
5179 | newDtd->keepProcessing = oldDtd->keepProcessing; | |
5180 | newDtd->hasParamEntityRefs = oldDtd->hasParamEntityRefs; | |
5181 | newDtd->standalone = oldDtd->standalone; | |
5182 | ||
5183 | /* Don't want deep copying for scaffolding */ | |
5184 | newDtd->in_eldecl = oldDtd->in_eldecl; | |
5185 | newDtd->scaffold = oldDtd->scaffold; | |
5186 | newDtd->contentStringLen = oldDtd->contentStringLen; | |
5187 | newDtd->scaffSize = oldDtd->scaffSize; | |
5188 | newDtd->scaffLevel = oldDtd->scaffLevel; | |
5189 | newDtd->scaffIndex = oldDtd->scaffIndex; | |
5190 | ||
5191 | return 1; | |
5192 | } /* End dtdCopy */ | |
5193 | ||
5194 | static int | |
5195 | copyEntityTable(HASH_TABLE *newTable, | |
5196 | STRING_POOL *newPool, | |
5197 | const HASH_TABLE *oldTable) | |
5198 | { | |
5199 | HASH_TABLE_ITER iter; | |
5200 | const XML_Char *cachedOldBase = NULL; | |
5201 | const XML_Char *cachedNewBase = NULL; | |
5202 | ||
5203 | hashTableIterInit(&iter, oldTable); | |
5204 | ||
5205 | for (;;) { | |
5206 | ENTITY *newE; | |
5207 | const XML_Char *name; | |
5208 | const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); | |
5209 | if (!oldE) | |
5210 | break; | |
5211 | name = poolCopyString(newPool, oldE->name); | |
5212 | if (!name) | |
5213 | return 0; | |
5214 | newE = (ENTITY *)lookup(newTable, name, sizeof(ENTITY)); | |
5215 | if (!newE) | |
5216 | return 0; | |
5217 | if (oldE->systemId) { | |
5218 | const XML_Char *tem = poolCopyString(newPool, oldE->systemId); | |
5219 | if (!tem) | |
5220 | return 0; | |
5221 | newE->systemId = tem; | |
5222 | if (oldE->base) { | |
5223 | if (oldE->base == cachedOldBase) | |
5224 | newE->base = cachedNewBase; | |
5225 | else { | |
5226 | cachedOldBase = oldE->base; | |
5227 | tem = poolCopyString(newPool, cachedOldBase); | |
5228 | if (!tem) | |
5229 | return 0; | |
5230 | cachedNewBase = newE->base = tem; | |
5231 | } | |
5232 | } | |
5233 | if (oldE->publicId) { | |
5234 | tem = poolCopyString(newPool, oldE->publicId); | |
5235 | if (!tem) | |
5236 | return 0; | |
5237 | newE->publicId = tem; | |
5238 | } | |
5239 | } | |
5240 | else { | |
5241 | const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr, | |
5242 | oldE->textLen); | |
5243 | if (!tem) | |
5244 | return 0; | |
5245 | newE->textPtr = tem; | |
5246 | newE->textLen = oldE->textLen; | |
5247 | } | |
5248 | if (oldE->notation) { | |
5249 | const XML_Char *tem = poolCopyString(newPool, oldE->notation); | |
5250 | if (!tem) | |
5251 | return 0; | |
5252 | newE->notation = tem; | |
5253 | } | |
5254 | newE->is_param = oldE->is_param; | |
5255 | newE->is_internal = oldE->is_internal; | |
5256 | } | |
5257 | return 1; | |
5258 | } | |
5259 | ||
5260 | #define INIT_SIZE 64 | |
5261 | ||
5262 | static int FASTCALL | |
5263 | keyeq(KEY s1, KEY s2) | |
5264 | { | |
5265 | for (; *s1 == *s2; s1++, s2++) | |
5266 | if (*s1 == 0) | |
5267 | return 1; | |
5268 | return 0; | |
5269 | } | |
5270 | ||
5271 | static unsigned long FASTCALL | |
5272 | hash(KEY s) | |
5273 | { | |
5274 | unsigned long h = 0; | |
5275 | while (*s) | |
5276 | h = (h << 5) + h + (unsigned char)*s++; | |
5277 | return h; | |
5278 | } | |
5279 | ||
5280 | static NAMED * | |
5281 | lookup(HASH_TABLE *table, KEY name, size_t createSize) | |
5282 | { | |
5283 | size_t i; | |
5284 | if (table->size == 0) { | |
5285 | size_t tsize; | |
5286 | ||
5287 | if (!createSize) | |
5288 | return NULL; | |
5289 | tsize = INIT_SIZE * sizeof(NAMED *); | |
5290 | table->v = (NAMED **)table->mem->malloc_fcn(tsize); | |
5291 | if (!table->v) | |
5292 | return NULL; | |
5293 | memset(table->v, 0, tsize); | |
5294 | table->size = INIT_SIZE; | |
5295 | table->usedLim = INIT_SIZE / 2; | |
5296 | i = hash(name) & (table->size - 1); | |
5297 | } | |
5298 | else { | |
5299 | unsigned long h = hash(name); | |
5300 | for (i = h & (table->size - 1); | |
5301 | table->v[i]; | |
5302 | i == 0 ? i = table->size - 1 : --i) { | |
5303 | if (keyeq(name, table->v[i]->name)) | |
5304 | return table->v[i]; | |
5305 | } | |
5306 | if (!createSize) | |
5307 | return NULL; | |
5308 | if (table->used == table->usedLim) { | |
5309 | /* check for overflow */ | |
5310 | size_t newSize = table->size * 2; | |
5311 | size_t tsize = newSize * sizeof(NAMED *); | |
5312 | NAMED **newV = (NAMED **)table->mem->malloc_fcn(tsize); | |
5313 | if (!newV) | |
5314 | return NULL; | |
5315 | memset(newV, 0, tsize); | |
5316 | for (i = 0; i < table->size; i++) | |
5317 | if (table->v[i]) { | |
5318 | size_t j; | |
5319 | for (j = hash(table->v[i]->name) & (newSize - 1); | |
5320 | newV[j]; | |
5321 | j == 0 ? j = newSize - 1 : --j) | |
5322 | ; | |
5323 | newV[j] = table->v[i]; | |
5324 | } | |
5325 | table->mem->free_fcn(table->v); | |
5326 | table->v = newV; | |
5327 | table->size = newSize; | |
5328 | table->usedLim = newSize/2; | |
5329 | for (i = h & (table->size - 1); | |
5330 | table->v[i]; | |
5331 | i == 0 ? i = table->size - 1 : --i) | |
5332 | ; | |
5333 | } | |
5334 | } | |
5335 | table->v[i] = (NAMED *)table->mem->malloc_fcn(createSize); | |
5336 | if (!table->v[i]) | |
5337 | return NULL; | |
5338 | memset(table->v[i], 0, createSize); | |
5339 | table->v[i]->name = name; | |
5340 | (table->used)++; | |
5341 | return table->v[i]; | |
5342 | } | |
5343 | ||
5344 | static void FASTCALL | |
5345 | hashTableClear(HASH_TABLE *table) | |
5346 | { | |
5347 | size_t i; | |
5348 | for (i = 0; i < table->size; i++) { | |
5349 | NAMED *p = table->v[i]; | |
5350 | if (p) { | |
5351 | table->mem->free_fcn(p); | |
5352 | table->v[i] = NULL; | |
5353 | } | |
5354 | } | |
5355 | table->usedLim = table->size / 2; | |
5356 | table->used = 0; | |
5357 | } | |
5358 | ||
5359 | static void FASTCALL | |
5360 | hashTableDestroy(HASH_TABLE *table) | |
5361 | { | |
5362 | size_t i; | |
5363 | for (i = 0; i < table->size; i++) { | |
5364 | NAMED *p = table->v[i]; | |
5365 | if (p) | |
5366 | table->mem->free_fcn(p); | |
5367 | } | |
5368 | if (table->v) | |
5369 | table->mem->free_fcn(table->v); | |
5370 | } | |
5371 | ||
5372 | static void FASTCALL | |
5373 | hashTableInit(HASH_TABLE *p, const XML_Memory_Handling_Suite *ms) | |
5374 | { | |
5375 | p->size = 0; | |
5376 | p->usedLim = 0; | |
5377 | p->used = 0; | |
5378 | p->v = NULL; | |
5379 | p->mem = ms; | |
5380 | } | |
5381 | ||
5382 | static void FASTCALL | |
5383 | hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table) | |
5384 | { | |
5385 | iter->p = table->v; | |
5386 | iter->end = iter->p + table->size; | |
5387 | } | |
5388 | ||
5389 | static NAMED * FASTCALL | |
5390 | hashTableIterNext(HASH_TABLE_ITER *iter) | |
5391 | { | |
5392 | while (iter->p != iter->end) { | |
5393 | NAMED *tem = *(iter->p)++; | |
5394 | if (tem) | |
5395 | return tem; | |
5396 | } | |
5397 | return NULL; | |
5398 | } | |
5399 | ||
5400 | static void FASTCALL | |
5401 | poolInit(STRING_POOL *pool, const XML_Memory_Handling_Suite *ms) | |
5402 | { | |
5403 | pool->blocks = NULL; | |
5404 | pool->freeBlocks = NULL; | |
5405 | pool->start = NULL; | |
5406 | pool->ptr = NULL; | |
5407 | pool->end = NULL; | |
5408 | pool->mem = ms; | |
5409 | } | |
5410 | ||
5411 | static void FASTCALL | |
5412 | poolClear(STRING_POOL *pool) | |
5413 | { | |
5414 | if (!pool->freeBlocks) | |
5415 | pool->freeBlocks = pool->blocks; | |
5416 | else { | |
5417 | BLOCK *p = pool->blocks; | |
5418 | while (p) { | |
5419 | BLOCK *tem = p->next; | |
5420 | p->next = pool->freeBlocks; | |
5421 | pool->freeBlocks = p; | |
5422 | p = tem; | |
5423 | } | |
5424 | } | |
5425 | pool->blocks = NULL; | |
5426 | pool->start = NULL; | |
5427 | pool->ptr = NULL; | |
5428 | pool->end = NULL; | |
5429 | } | |
5430 | ||
5431 | static void FASTCALL | |
5432 | poolDestroy(STRING_POOL *pool) | |
5433 | { | |
5434 | BLOCK *p = pool->blocks; | |
5435 | while (p) { | |
5436 | BLOCK *tem = p->next; | |
5437 | pool->mem->free_fcn(p); | |
5438 | p = tem; | |
5439 | } | |
5440 | p = pool->freeBlocks; | |
5441 | while (p) { | |
5442 | BLOCK *tem = p->next; | |
5443 | pool->mem->free_fcn(p); | |
5444 | p = tem; | |
5445 | } | |
5446 | } | |
5447 | ||
5448 | static XML_Char * | |
5449 | poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
5450 | const char *ptr, const char *end) | |
5451 | { | |
5452 | if (!pool->ptr && !poolGrow(pool)) | |
5453 | return NULL; | |
5454 | for (;;) { | |
5455 | XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); | |
5456 | if (ptr == end) | |
5457 | break; | |
5458 | if (!poolGrow(pool)) | |
5459 | return NULL; | |
5460 | } | |
5461 | return pool->start; | |
5462 | } | |
5463 | ||
5464 | static const XML_Char * FASTCALL | |
5465 | poolCopyString(STRING_POOL *pool, const XML_Char *s) | |
5466 | { | |
5467 | do { | |
5468 | if (!poolAppendChar(pool, *s)) | |
5469 | return NULL; | |
5470 | } while (*s++); | |
5471 | s = pool->start; | |
5472 | poolFinish(pool); | |
5473 | return s; | |
5474 | } | |
5475 | ||
5476 | static const XML_Char * | |
5477 | poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) | |
5478 | { | |
5479 | if (!pool->ptr && !poolGrow(pool)) | |
5480 | return NULL; | |
5481 | for (; n > 0; --n, s++) { | |
5482 | if (!poolAppendChar(pool, *s)) | |
5483 | return NULL; | |
5484 | } | |
5485 | s = pool->start; | |
5486 | poolFinish(pool); | |
5487 | return s; | |
5488 | } | |
5489 | ||
5490 | static const XML_Char * FASTCALL | |
5491 | poolAppendString(STRING_POOL *pool, const XML_Char *s) | |
5492 | { | |
5493 | while (*s) { | |
5494 | if (!poolAppendChar(pool, *s)) | |
5495 | return NULL; | |
5496 | s++; | |
5497 | } | |
5498 | return pool->start; | |
5499 | } | |
5500 | ||
5501 | static XML_Char * | |
5502 | poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
5503 | const char *ptr, const char *end) | |
5504 | { | |
5505 | if (!poolAppend(pool, enc, ptr, end)) | |
5506 | return NULL; | |
5507 | if (pool->ptr == pool->end && !poolGrow(pool)) | |
5508 | return NULL; | |
5509 | *(pool->ptr)++ = 0; | |
5510 | return pool->start; | |
5511 | } | |
5512 | ||
5513 | static XML_Bool FASTCALL | |
5514 | poolGrow(STRING_POOL *pool) | |
5515 | { | |
5516 | if (pool->freeBlocks) { | |
5517 | if (pool->start == 0) { | |
5518 | pool->blocks = pool->freeBlocks; | |
5519 | pool->freeBlocks = pool->freeBlocks->next; | |
5520 | pool->blocks->next = NULL; | |
5521 | pool->start = pool->blocks->s; | |
5522 | pool->end = pool->start + pool->blocks->size; | |
5523 | pool->ptr = pool->start; | |
5524 | return XML_TRUE; | |
5525 | } | |
5526 | if (pool->end - pool->start < pool->freeBlocks->size) { | |
5527 | BLOCK *tem = pool->freeBlocks->next; | |
5528 | pool->freeBlocks->next = pool->blocks; | |
5529 | pool->blocks = pool->freeBlocks; | |
5530 | pool->freeBlocks = tem; | |
5531 | memcpy(pool->blocks->s, pool->start, | |
5532 | (pool->end - pool->start) * sizeof(XML_Char)); | |
5533 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
5534 | pool->start = pool->blocks->s; | |
5535 | pool->end = pool->start + pool->blocks->size; | |
5536 | return XML_TRUE; | |
5537 | } | |
5538 | } | |
5539 | if (pool->blocks && pool->start == pool->blocks->s) { | |
5540 | int blockSize = (pool->end - pool->start)*2; | |
5541 | pool->blocks = (BLOCK *) | |
5542 | pool->mem->realloc_fcn(pool->blocks, | |
5543 | (offsetof(BLOCK, s) | |
5544 | + blockSize * sizeof(XML_Char))); | |
5545 | if (pool->blocks == NULL) | |
5546 | return XML_FALSE; | |
5547 | pool->blocks->size = blockSize; | |
5548 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
5549 | pool->start = pool->blocks->s; | |
5550 | pool->end = pool->start + blockSize; | |
5551 | } | |
5552 | else { | |
5553 | BLOCK *tem; | |
5554 | int blockSize = pool->end - pool->start; | |
5555 | if (blockSize < INIT_BLOCK_SIZE) | |
5556 | blockSize = INIT_BLOCK_SIZE; | |
5557 | else | |
5558 | blockSize *= 2; | |
5559 | tem = (BLOCK *)pool->mem->malloc_fcn(offsetof(BLOCK, s) | |
5560 | + blockSize * sizeof(XML_Char)); | |
5561 | if (!tem) | |
5562 | return XML_FALSE; | |
5563 | tem->size = blockSize; | |
5564 | tem->next = pool->blocks; | |
5565 | pool->blocks = tem; | |
5566 | if (pool->ptr != pool->start) | |
5567 | memcpy(tem->s, pool->start, | |
5568 | (pool->ptr - pool->start) * sizeof(XML_Char)); | |
5569 | pool->ptr = tem->s + (pool->ptr - pool->start); | |
5570 | pool->start = tem->s; | |
5571 | pool->end = tem->s + blockSize; | |
5572 | } | |
5573 | return XML_TRUE; | |
5574 | } | |
5575 | ||
5576 | static int FASTCALL | |
5577 | nextScaffoldPart(XML_Parser parser) | |
5578 | { | |
5579 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5580 | CONTENT_SCAFFOLD * me; | |
5581 | int next; | |
5582 | ||
5583 | if (!dtd->scaffIndex) { | |
5584 | dtd->scaffIndex = (int *)MALLOC(groupSize * sizeof(int)); | |
5585 | if (!dtd->scaffIndex) | |
5586 | return -1; | |
5587 | dtd->scaffIndex[0] = 0; | |
5588 | } | |
5589 | ||
5590 | if (dtd->scaffCount >= dtd->scaffSize) { | |
5591 | CONTENT_SCAFFOLD *temp; | |
5592 | if (dtd->scaffold) { | |
5593 | temp = (CONTENT_SCAFFOLD *) | |
5594 | REALLOC(dtd->scaffold, dtd->scaffSize * 2 * sizeof(CONTENT_SCAFFOLD)); | |
5595 | if (temp == NULL) | |
5596 | return -1; | |
5597 | dtd->scaffSize *= 2; | |
5598 | } | |
5599 | else { | |
5600 | temp = (CONTENT_SCAFFOLD *)MALLOC(INIT_SCAFFOLD_ELEMENTS | |
5601 | * sizeof(CONTENT_SCAFFOLD)); | |
5602 | if (temp == NULL) | |
5603 | return -1; | |
5604 | dtd->scaffSize = INIT_SCAFFOLD_ELEMENTS; | |
5605 | } | |
5606 | dtd->scaffold = temp; | |
5607 | } | |
5608 | next = dtd->scaffCount++; | |
5609 | me = &dtd->scaffold[next]; | |
5610 | if (dtd->scaffLevel) { | |
5611 | CONTENT_SCAFFOLD *parent = &dtd->scaffold[dtd->scaffIndex[dtd->scaffLevel-1]]; | |
5612 | if (parent->lastchild) { | |
5613 | dtd->scaffold[parent->lastchild].nextsib = next; | |
5614 | } | |
5615 | if (!parent->childcnt) | |
5616 | parent->firstchild = next; | |
5617 | parent->lastchild = next; | |
5618 | parent->childcnt++; | |
5619 | } | |
5620 | me->firstchild = me->lastchild = me->childcnt = me->nextsib = 0; | |
5621 | return next; | |
5622 | } | |
5623 | ||
5624 | static void | |
5625 | build_node(XML_Parser parser, | |
5626 | int src_node, | |
5627 | XML_Content *dest, | |
5628 | XML_Content **contpos, | |
5629 | XML_Char **strpos) | |
5630 | { | |
5631 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5632 | dest->type = dtd->scaffold[src_node].type; | |
5633 | dest->quant = dtd->scaffold[src_node].quant; | |
5634 | if (dest->type == XML_CTYPE_NAME) { | |
5635 | const XML_Char *src; | |
5636 | dest->name = *strpos; | |
5637 | src = dtd->scaffold[src_node].name; | |
5638 | for (;;) { | |
5639 | *(*strpos)++ = *src; | |
5640 | if (!*src) | |
5641 | break; | |
5642 | src++; | |
5643 | } | |
5644 | dest->numchildren = 0; | |
5645 | dest->children = NULL; | |
5646 | } | |
5647 | else { | |
5648 | unsigned int i; | |
5649 | int cn; | |
5650 | dest->numchildren = dtd->scaffold[src_node].childcnt; | |
5651 | dest->children = *contpos; | |
5652 | *contpos += dest->numchildren; | |
5653 | for (i = 0, cn = dtd->scaffold[src_node].firstchild; | |
5654 | i < dest->numchildren; | |
5655 | i++, cn = dtd->scaffold[cn].nextsib) { | |
5656 | build_node(parser, cn, &(dest->children[i]), contpos, strpos); | |
5657 | } | |
5658 | dest->name = NULL; | |
5659 | } | |
5660 | } | |
5661 | ||
5662 | static XML_Content * | |
5663 | build_model (XML_Parser parser) | |
5664 | { | |
5665 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5666 | XML_Content *ret; | |
5667 | XML_Content *cpos; | |
5668 | XML_Char * str; | |
5669 | int allocsize = (dtd->scaffCount * sizeof(XML_Content) | |
5670 | + (dtd->contentStringLen * sizeof(XML_Char))); | |
5671 | ||
5672 | ret = (XML_Content *)MALLOC(allocsize); | |
5673 | if (!ret) | |
5674 | return NULL; | |
5675 | ||
5676 | str = (XML_Char *) (&ret[dtd->scaffCount]); | |
5677 | cpos = &ret[1]; | |
5678 | ||
5679 | build_node(parser, 0, ret, &cpos, &str); | |
5680 | return ret; | |
5681 | } | |
5682 | ||
5683 | static ELEMENT_TYPE * | |
5684 | getElementType(XML_Parser parser, | |
5685 | const ENCODING *enc, | |
5686 | const char *ptr, | |
5687 | const char *end) | |
5688 | { | |
5689 | DTD * const dtd = _dtd; /* save one level of indirection */ | |
5690 | const XML_Char *name = poolStoreString(&dtd->pool, enc, ptr, end); | |
5691 | ELEMENT_TYPE *ret; | |
5692 | ||
5693 | if (!name) | |
5694 | return NULL; | |
5695 | ret = (ELEMENT_TYPE *) lookup(&dtd->elementTypes, name, sizeof(ELEMENT_TYPE)); | |
5696 | if (!ret) | |
5697 | return NULL; | |
5698 | if (ret->name != name) | |
5699 | poolDiscard(&dtd->pool); | |
5700 | else { | |
5701 | poolFinish(&dtd->pool); | |
5702 | if (!setElementTypePrefix(parser, ret)) | |
5703 | return NULL; | |
5704 | } | |
5705 | return ret; | |
5706 | } |