]>
Commit | Line | Data |
---|---|---|
75e01429 VS |
1 | /* |
2 | Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd | |
3 | See the file copying.txt for copying permission. | |
4 | */ | |
5 | ||
6 | #include "xmldef.h" | |
7 | #include "xmlparse.h" | |
8 | #include <stddef.h> | |
9 | ||
10 | #ifdef XML_UNICODE | |
11 | #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX | |
12 | #define XmlConvert XmlUtf16Convert | |
13 | #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding | |
14 | #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS | |
15 | #define XmlEncode XmlUtf16Encode | |
16 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1)) | |
17 | typedef unsigned short ICHAR; | |
18 | #else | |
19 | #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX | |
20 | #define XmlConvert XmlUtf8Convert | |
21 | #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding | |
22 | #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS | |
23 | #define XmlEncode XmlUtf8Encode | |
24 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf8) | |
25 | typedef char ICHAR; | |
26 | #endif | |
27 | ||
28 | ||
29 | #ifndef XML_NS | |
30 | ||
31 | #define XmlInitEncodingNS XmlInitEncoding | |
32 | #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding | |
33 | #undef XmlGetInternalEncodingNS | |
34 | #define XmlGetInternalEncodingNS XmlGetInternalEncoding | |
35 | #define XmlParseXmlDeclNS XmlParseXmlDecl | |
36 | ||
37 | #endif | |
38 | ||
39 | #ifdef XML_UNICODE_WCHAR_T | |
40 | #define XML_T(x) L ## x | |
41 | #else | |
42 | #define XML_T(x) x | |
43 | #endif | |
44 | ||
45 | /* Round up n to be a multiple of sz, where sz is a power of 2. */ | |
46 | #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1)) | |
47 | ||
48 | #include "xmltok.h" | |
49 | #include "xmlrole.h" | |
50 | ||
51 | typedef const XML_Char *KEY; | |
52 | ||
53 | typedef struct { | |
54 | KEY name; | |
55 | } NAMED; | |
56 | ||
57 | typedef struct { | |
58 | NAMED **v; | |
59 | size_t size; | |
60 | size_t used; | |
61 | size_t usedLim; | |
62 | } HASH_TABLE; | |
63 | ||
64 | typedef struct { | |
65 | NAMED **p; | |
66 | NAMED **end; | |
67 | } HASH_TABLE_ITER; | |
68 | ||
69 | #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ | |
70 | #define INIT_DATA_BUF_SIZE 1024 | |
71 | #define INIT_ATTS_SIZE 16 | |
72 | #define INIT_BLOCK_SIZE 1024 | |
73 | #define INIT_BUFFER_SIZE 1024 | |
74 | ||
75 | #define EXPAND_SPARE 24 | |
76 | ||
77 | typedef struct binding { | |
78 | struct prefix *prefix; | |
79 | struct binding *nextTagBinding; | |
80 | struct binding *prevPrefixBinding; | |
81 | const struct attribute_id *attId; | |
82 | XML_Char *uri; | |
83 | int uriLen; | |
84 | int uriAlloc; | |
85 | } BINDING; | |
86 | ||
87 | typedef struct prefix { | |
88 | const XML_Char *name; | |
89 | BINDING *binding; | |
90 | } PREFIX; | |
91 | ||
92 | typedef struct { | |
93 | const XML_Char *str; | |
94 | const XML_Char *localPart; | |
95 | int uriLen; | |
96 | } TAG_NAME; | |
97 | ||
98 | typedef struct tag { | |
99 | struct tag *parent; | |
100 | const char *rawName; | |
101 | int rawNameLength; | |
102 | TAG_NAME name; | |
103 | char *buf; | |
104 | char *bufEnd; | |
105 | BINDING *bindings; | |
106 | } TAG; | |
107 | ||
108 | typedef struct { | |
109 | const XML_Char *name; | |
110 | const XML_Char *textPtr; | |
111 | int textLen; | |
112 | const XML_Char *systemId; | |
113 | const XML_Char *base; | |
114 | const XML_Char *publicId; | |
115 | const XML_Char *notation; | |
116 | char open; | |
117 | } ENTITY; | |
118 | ||
119 | typedef struct block { | |
120 | struct block *next; | |
121 | int size; | |
122 | XML_Char s[1]; | |
123 | } BLOCK; | |
124 | ||
125 | typedef struct { | |
126 | BLOCK *blocks; | |
127 | BLOCK *freeBlocks; | |
128 | const XML_Char *end; | |
129 | XML_Char *ptr; | |
130 | XML_Char *start; | |
131 | } STRING_POOL; | |
132 | ||
133 | /* The XML_Char before the name is used to determine whether | |
134 | an attribute has been specified. */ | |
135 | typedef struct attribute_id { | |
136 | XML_Char *name; | |
137 | PREFIX *prefix; | |
138 | char maybeTokenized; | |
139 | char xmlns; | |
140 | } ATTRIBUTE_ID; | |
141 | ||
142 | typedef struct { | |
143 | const ATTRIBUTE_ID *id; | |
144 | char isCdata; | |
145 | const XML_Char *value; | |
146 | } DEFAULT_ATTRIBUTE; | |
147 | ||
148 | typedef struct { | |
149 | const XML_Char *name; | |
150 | PREFIX *prefix; | |
151 | const ATTRIBUTE_ID *idAtt; | |
152 | int nDefaultAtts; | |
153 | int allocDefaultAtts; | |
154 | DEFAULT_ATTRIBUTE *defaultAtts; | |
155 | } ELEMENT_TYPE; | |
156 | ||
157 | typedef struct { | |
158 | HASH_TABLE generalEntities; | |
159 | HASH_TABLE elementTypes; | |
160 | HASH_TABLE attributeIds; | |
161 | HASH_TABLE prefixes; | |
162 | STRING_POOL pool; | |
163 | int complete; | |
164 | int standalone; | |
165 | #ifdef XML_DTD | |
166 | HASH_TABLE paramEntities; | |
167 | #endif /* XML_DTD */ | |
168 | PREFIX defaultPrefix; | |
169 | } DTD; | |
170 | ||
171 | typedef struct open_internal_entity { | |
172 | const char *internalEventPtr; | |
173 | const char *internalEventEndPtr; | |
174 | struct open_internal_entity *next; | |
175 | ENTITY *entity; | |
176 | } OPEN_INTERNAL_ENTITY; | |
177 | ||
178 | typedef enum XML_Error Processor(XML_Parser parser, | |
179 | const char *start, | |
180 | const char *end, | |
181 | const char **endPtr); | |
182 | ||
183 | static Processor prologProcessor; | |
184 | static Processor prologInitProcessor; | |
185 | static Processor contentProcessor; | |
186 | static Processor cdataSectionProcessor; | |
187 | #ifdef XML_DTD | |
188 | static Processor ignoreSectionProcessor; | |
189 | #endif /* XML_DTD */ | |
190 | static Processor epilogProcessor; | |
191 | static Processor errorProcessor; | |
192 | static Processor externalEntityInitProcessor; | |
193 | static Processor externalEntityInitProcessor2; | |
194 | static Processor externalEntityInitProcessor3; | |
195 | static Processor externalEntityContentProcessor; | |
196 | ||
197 | static enum XML_Error | |
198 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName); | |
199 | static enum XML_Error | |
200 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, const char *, const char *); | |
201 | static enum XML_Error | |
202 | initializeEncoding(XML_Parser parser); | |
203 | static enum XML_Error | |
204 | doProlog(XML_Parser parser, const ENCODING *enc, const char *s, | |
205 | const char *end, int tok, const char *next, const char **nextPtr); | |
206 | static enum XML_Error | |
207 | processInternalParamEntity(XML_Parser parser, ENTITY *entity); | |
208 | static enum XML_Error | |
209 | doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, | |
210 | const char *start, const char *end, const char **endPtr); | |
211 | static enum XML_Error | |
212 | doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr); | |
213 | #ifdef XML_DTD | |
214 | static enum XML_Error | |
215 | doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr); | |
216 | #endif /* XML_DTD */ | |
217 | static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *, const char *s, | |
218 | TAG_NAME *tagNamePtr, BINDING **bindingsPtr); | |
219 | static | |
220 | int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr); | |
221 | static int | |
222 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, int isCdata, int isId, const XML_Char *dfltValue); | |
223 | static enum XML_Error | |
224 | storeAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, | |
225 | STRING_POOL *); | |
226 | static enum XML_Error | |
227 | appendAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, | |
228 | STRING_POOL *); | |
229 | static ATTRIBUTE_ID * | |
230 | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
231 | static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); | |
232 | static enum XML_Error | |
233 | storeEntityValue(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
234 | static int | |
235 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
236 | static int | |
237 | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
238 | static void | |
239 | reportDefault(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
240 | ||
241 | static const XML_Char *getContext(XML_Parser parser); | |
242 | static int setContext(XML_Parser parser, const XML_Char *context); | |
243 | static void normalizePublicId(XML_Char *s); | |
244 | static int dtdInit(DTD *); | |
245 | static void dtdDestroy(DTD *); | |
246 | static int dtdCopy(DTD *newDtd, const DTD *oldDtd); | |
247 | static int copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *); | |
248 | #ifdef XML_DTD | |
249 | static void dtdSwap(DTD *, DTD *); | |
250 | #endif /* XML_DTD */ | |
251 | static NAMED *lookup(HASH_TABLE *table, KEY name, size_t createSize); | |
252 | static void hashTableInit(HASH_TABLE *); | |
253 | static void hashTableDestroy(HASH_TABLE *); | |
254 | static void hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *); | |
255 | static NAMED *hashTableIterNext(HASH_TABLE_ITER *); | |
256 | static void poolInit(STRING_POOL *); | |
257 | static void poolClear(STRING_POOL *); | |
258 | static void poolDestroy(STRING_POOL *); | |
259 | static XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
260 | const char *ptr, const char *end); | |
261 | static XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
262 | const char *ptr, const char *end); | |
263 | static int poolGrow(STRING_POOL *pool); | |
264 | static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s); | |
265 | static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n); | |
266 | ||
267 | #define poolStart(pool) ((pool)->start) | |
268 | #define poolEnd(pool) ((pool)->ptr) | |
269 | #define poolLength(pool) ((pool)->ptr - (pool)->start) | |
270 | #define poolChop(pool) ((void)--(pool->ptr)) | |
271 | #define poolLastChar(pool) (((pool)->ptr)[-1]) | |
272 | #define poolDiscard(pool) ((pool)->ptr = (pool)->start) | |
273 | #define poolFinish(pool) ((pool)->start = (pool)->ptr) | |
274 | #define poolAppendChar(pool, c) \ | |
275 | (((pool)->ptr == (pool)->end && !poolGrow(pool)) \ | |
276 | ? 0 \ | |
277 | : ((*((pool)->ptr)++ = c), 1)) | |
278 | ||
279 | typedef struct { | |
280 | /* The first member must be userData so that the XML_GetUserData macro works. */ | |
281 | void *m_userData; | |
282 | void *m_handlerArg; | |
283 | char *m_buffer; | |
284 | /* first character to be parsed */ | |
285 | const char *m_bufferPtr; | |
286 | /* past last character to be parsed */ | |
287 | char *m_bufferEnd; | |
288 | /* allocated end of buffer */ | |
289 | const char *m_bufferLim; | |
290 | long m_parseEndByteIndex; | |
291 | const char *m_parseEndPtr; | |
292 | XML_Char *m_dataBuf; | |
293 | XML_Char *m_dataBufEnd; | |
294 | XML_StartElementHandler m_startElementHandler; | |
295 | XML_EndElementHandler m_endElementHandler; | |
296 | XML_CharacterDataHandler m_characterDataHandler; | |
297 | XML_ProcessingInstructionHandler m_processingInstructionHandler; | |
298 | XML_CommentHandler m_commentHandler; | |
299 | XML_StartCdataSectionHandler m_startCdataSectionHandler; | |
300 | XML_EndCdataSectionHandler m_endCdataSectionHandler; | |
301 | XML_DefaultHandler m_defaultHandler; | |
302 | XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler; | |
303 | XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler; | |
304 | XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; | |
305 | XML_NotationDeclHandler m_notationDeclHandler; | |
306 | XML_ExternalParsedEntityDeclHandler m_externalParsedEntityDeclHandler; | |
307 | XML_InternalParsedEntityDeclHandler m_internalParsedEntityDeclHandler; | |
308 | XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; | |
309 | XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; | |
310 | XML_NotStandaloneHandler m_notStandaloneHandler; | |
311 | XML_ExternalEntityRefHandler m_externalEntityRefHandler; | |
312 | void *m_externalEntityRefHandlerArg; | |
313 | XML_UnknownEncodingHandler m_unknownEncodingHandler; | |
314 | const ENCODING *m_encoding; | |
315 | INIT_ENCODING m_initEncoding; | |
316 | const ENCODING *m_internalEncoding; | |
317 | const XML_Char *m_protocolEncodingName; | |
318 | int m_ns; | |
319 | void *m_unknownEncodingMem; | |
320 | void *m_unknownEncodingData; | |
321 | void *m_unknownEncodingHandlerData; | |
322 | void (*m_unknownEncodingRelease)(void *); | |
323 | PROLOG_STATE m_prologState; | |
324 | Processor *m_processor; | |
325 | enum XML_Error m_errorCode; | |
326 | const char *m_eventPtr; | |
327 | const char *m_eventEndPtr; | |
328 | const char *m_positionPtr; | |
329 | OPEN_INTERNAL_ENTITY *m_openInternalEntities; | |
330 | int m_defaultExpandInternalEntities; | |
331 | int m_tagLevel; | |
332 | ENTITY *m_declEntity; | |
333 | const XML_Char *m_declNotationName; | |
334 | const XML_Char *m_declNotationPublicId; | |
335 | ELEMENT_TYPE *m_declElementType; | |
336 | ATTRIBUTE_ID *m_declAttributeId; | |
337 | char m_declAttributeIsCdata; | |
338 | char m_declAttributeIsId; | |
339 | DTD m_dtd; | |
340 | const XML_Char *m_curBase; | |
341 | TAG *m_tagStack; | |
342 | TAG *m_freeTagList; | |
343 | BINDING *m_inheritedBindings; | |
344 | BINDING *m_freeBindingList; | |
345 | int m_attsSize; | |
346 | int m_nSpecifiedAtts; | |
347 | int m_idAttIndex; | |
348 | ATTRIBUTE *m_atts; | |
349 | POSITION m_position; | |
350 | STRING_POOL m_tempPool; | |
351 | STRING_POOL m_temp2Pool; | |
352 | char *m_groupConnector; | |
353 | unsigned m_groupSize; | |
354 | int m_hadExternalDoctype; | |
355 | XML_Char m_namespaceSeparator; | |
356 | #ifdef XML_DTD | |
357 | enum XML_ParamEntityParsing m_paramEntityParsing; | |
358 | XML_Parser m_parentParser; | |
359 | #endif | |
360 | } Parser; | |
361 | ||
362 | #define userData (((Parser *)parser)->m_userData) | |
363 | #define handlerArg (((Parser *)parser)->m_handlerArg) | |
364 | #define startElementHandler (((Parser *)parser)->m_startElementHandler) | |
365 | #define endElementHandler (((Parser *)parser)->m_endElementHandler) | |
366 | #define characterDataHandler (((Parser *)parser)->m_characterDataHandler) | |
367 | #define processingInstructionHandler (((Parser *)parser)->m_processingInstructionHandler) | |
368 | #define commentHandler (((Parser *)parser)->m_commentHandler) | |
369 | #define startCdataSectionHandler (((Parser *)parser)->m_startCdataSectionHandler) | |
370 | #define endCdataSectionHandler (((Parser *)parser)->m_endCdataSectionHandler) | |
371 | #define defaultHandler (((Parser *)parser)->m_defaultHandler) | |
372 | #define startDoctypeDeclHandler (((Parser *)parser)->m_startDoctypeDeclHandler) | |
373 | #define endDoctypeDeclHandler (((Parser *)parser)->m_endDoctypeDeclHandler) | |
374 | #define unparsedEntityDeclHandler (((Parser *)parser)->m_unparsedEntityDeclHandler) | |
375 | #define notationDeclHandler (((Parser *)parser)->m_notationDeclHandler) | |
376 | #define externalParsedEntityDeclHandler (((Parser *)parser)->m_externalParsedEntityDeclHandler) | |
377 | #define internalParsedEntityDeclHandler (((Parser *)parser)->m_internalParsedEntityDeclHandler) | |
378 | #define startNamespaceDeclHandler (((Parser *)parser)->m_startNamespaceDeclHandler) | |
379 | #define endNamespaceDeclHandler (((Parser *)parser)->m_endNamespaceDeclHandler) | |
380 | #define notStandaloneHandler (((Parser *)parser)->m_notStandaloneHandler) | |
381 | #define externalEntityRefHandler (((Parser *)parser)->m_externalEntityRefHandler) | |
382 | #define externalEntityRefHandlerArg (((Parser *)parser)->m_externalEntityRefHandlerArg) | |
383 | #define unknownEncodingHandler (((Parser *)parser)->m_unknownEncodingHandler) | |
384 | #define encoding (((Parser *)parser)->m_encoding) | |
385 | #define initEncoding (((Parser *)parser)->m_initEncoding) | |
386 | #define internalEncoding (((Parser *)parser)->m_internalEncoding) | |
387 | #define unknownEncodingMem (((Parser *)parser)->m_unknownEncodingMem) | |
388 | #define unknownEncodingData (((Parser *)parser)->m_unknownEncodingData) | |
389 | #define unknownEncodingHandlerData \ | |
390 | (((Parser *)parser)->m_unknownEncodingHandlerData) | |
391 | #define unknownEncodingRelease (((Parser *)parser)->m_unknownEncodingRelease) | |
392 | #define protocolEncodingName (((Parser *)parser)->m_protocolEncodingName) | |
393 | #define ns (((Parser *)parser)->m_ns) | |
394 | #define prologState (((Parser *)parser)->m_prologState) | |
395 | #define processor (((Parser *)parser)->m_processor) | |
396 | #define errorCode (((Parser *)parser)->m_errorCode) | |
397 | #define eventPtr (((Parser *)parser)->m_eventPtr) | |
398 | #define eventEndPtr (((Parser *)parser)->m_eventEndPtr) | |
399 | #define positionPtr (((Parser *)parser)->m_positionPtr) | |
400 | #define position (((Parser *)parser)->m_position) | |
401 | #define openInternalEntities (((Parser *)parser)->m_openInternalEntities) | |
402 | #define defaultExpandInternalEntities (((Parser *)parser)->m_defaultExpandInternalEntities) | |
403 | #define tagLevel (((Parser *)parser)->m_tagLevel) | |
404 | #define buffer (((Parser *)parser)->m_buffer) | |
405 | #define bufferPtr (((Parser *)parser)->m_bufferPtr) | |
406 | #define bufferEnd (((Parser *)parser)->m_bufferEnd) | |
407 | #define parseEndByteIndex (((Parser *)parser)->m_parseEndByteIndex) | |
408 | #define parseEndPtr (((Parser *)parser)->m_parseEndPtr) | |
409 | #define bufferLim (((Parser *)parser)->m_bufferLim) | |
410 | #define dataBuf (((Parser *)parser)->m_dataBuf) | |
411 | #define dataBufEnd (((Parser *)parser)->m_dataBufEnd) | |
412 | #define dtd (((Parser *)parser)->m_dtd) | |
413 | #define curBase (((Parser *)parser)->m_curBase) | |
414 | #define declEntity (((Parser *)parser)->m_declEntity) | |
415 | #define declNotationName (((Parser *)parser)->m_declNotationName) | |
416 | #define declNotationPublicId (((Parser *)parser)->m_declNotationPublicId) | |
417 | #define declElementType (((Parser *)parser)->m_declElementType) | |
418 | #define declAttributeId (((Parser *)parser)->m_declAttributeId) | |
419 | #define declAttributeIsCdata (((Parser *)parser)->m_declAttributeIsCdata) | |
420 | #define declAttributeIsId (((Parser *)parser)->m_declAttributeIsId) | |
421 | #define freeTagList (((Parser *)parser)->m_freeTagList) | |
422 | #define freeBindingList (((Parser *)parser)->m_freeBindingList) | |
423 | #define inheritedBindings (((Parser *)parser)->m_inheritedBindings) | |
424 | #define tagStack (((Parser *)parser)->m_tagStack) | |
425 | #define atts (((Parser *)parser)->m_atts) | |
426 | #define attsSize (((Parser *)parser)->m_attsSize) | |
427 | #define nSpecifiedAtts (((Parser *)parser)->m_nSpecifiedAtts) | |
428 | #define idAttIndex (((Parser *)parser)->m_idAttIndex) | |
429 | #define tempPool (((Parser *)parser)->m_tempPool) | |
430 | #define temp2Pool (((Parser *)parser)->m_temp2Pool) | |
431 | #define groupConnector (((Parser *)parser)->m_groupConnector) | |
432 | #define groupSize (((Parser *)parser)->m_groupSize) | |
433 | #define hadExternalDoctype (((Parser *)parser)->m_hadExternalDoctype) | |
434 | #define namespaceSeparator (((Parser *)parser)->m_namespaceSeparator) | |
435 | #ifdef XML_DTD | |
436 | #define parentParser (((Parser *)parser)->m_parentParser) | |
437 | #define paramEntityParsing (((Parser *)parser)->m_paramEntityParsing) | |
438 | #endif /* XML_DTD */ | |
439 | ||
440 | #ifdef _MSC_VER | |
441 | #ifdef _DEBUG | |
442 | Parser *asParser(XML_Parser parser) | |
443 | { | |
444 | return parser; | |
445 | } | |
446 | #endif | |
447 | #endif | |
448 | ||
449 | XML_Parser XML_ParserCreate(const XML_Char *encodingName) | |
450 | { | |
451 | XML_Parser parser = malloc(sizeof(Parser)); | |
452 | if (!parser) | |
453 | return parser; | |
454 | processor = prologInitProcessor; | |
455 | XmlPrologStateInit(&prologState); | |
456 | userData = 0; | |
457 | handlerArg = 0; | |
458 | startElementHandler = 0; | |
459 | endElementHandler = 0; | |
460 | characterDataHandler = 0; | |
461 | processingInstructionHandler = 0; | |
462 | commentHandler = 0; | |
463 | startCdataSectionHandler = 0; | |
464 | endCdataSectionHandler = 0; | |
465 | defaultHandler = 0; | |
466 | startDoctypeDeclHandler = 0; | |
467 | endDoctypeDeclHandler = 0; | |
468 | unparsedEntityDeclHandler = 0; | |
469 | notationDeclHandler = 0; | |
470 | externalParsedEntityDeclHandler = 0; | |
471 | internalParsedEntityDeclHandler = 0; | |
472 | startNamespaceDeclHandler = 0; | |
473 | endNamespaceDeclHandler = 0; | |
474 | notStandaloneHandler = 0; | |
475 | externalEntityRefHandler = 0; | |
476 | externalEntityRefHandlerArg = parser; | |
477 | unknownEncodingHandler = 0; | |
478 | buffer = 0; | |
479 | bufferPtr = 0; | |
480 | bufferEnd = 0; | |
481 | parseEndByteIndex = 0; | |
482 | parseEndPtr = 0; | |
483 | bufferLim = 0; | |
484 | declElementType = 0; | |
485 | declAttributeId = 0; | |
486 | declEntity = 0; | |
487 | declNotationName = 0; | |
488 | declNotationPublicId = 0; | |
489 | memset(&position, 0, sizeof(POSITION)); | |
490 | errorCode = XML_ERROR_NONE; | |
491 | eventPtr = 0; | |
492 | eventEndPtr = 0; | |
493 | positionPtr = 0; | |
494 | openInternalEntities = 0; | |
495 | tagLevel = 0; | |
496 | tagStack = 0; | |
497 | freeTagList = 0; | |
498 | freeBindingList = 0; | |
499 | inheritedBindings = 0; | |
500 | attsSize = INIT_ATTS_SIZE; | |
501 | atts = malloc(attsSize * sizeof(ATTRIBUTE)); | |
502 | nSpecifiedAtts = 0; | |
503 | dataBuf = malloc(INIT_DATA_BUF_SIZE * sizeof(XML_Char)); | |
504 | groupSize = 0; | |
505 | groupConnector = 0; | |
506 | hadExternalDoctype = 0; | |
507 | unknownEncodingMem = 0; | |
508 | unknownEncodingRelease = 0; | |
509 | unknownEncodingData = 0; | |
510 | unknownEncodingHandlerData = 0; | |
511 | namespaceSeparator = '!'; | |
512 | #ifdef XML_DTD | |
513 | parentParser = 0; | |
514 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; | |
515 | #endif | |
516 | ns = 0; | |
517 | poolInit(&tempPool); | |
518 | poolInit(&temp2Pool); | |
519 | protocolEncodingName = encodingName ? poolCopyString(&tempPool, encodingName) : 0; | |
520 | curBase = 0; | |
521 | if (!dtdInit(&dtd) || !atts || !dataBuf | |
522 | || (encodingName && !protocolEncodingName)) { | |
523 | XML_ParserFree(parser); | |
524 | return 0; | |
525 | } | |
526 | dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE; | |
527 | XmlInitEncoding(&initEncoding, &encoding, 0); | |
528 | internalEncoding = XmlGetInternalEncoding(); | |
529 | return parser; | |
530 | } | |
531 | ||
532 | XML_Parser XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) | |
533 | { | |
534 | static | |
535 | const XML_Char implicitContext[] = { | |
536 | XML_T('x'), XML_T('m'), XML_T('l'), XML_T('='), | |
537 | XML_T('h'), XML_T('t'), XML_T('t'), XML_T('p'), XML_T(':'), | |
538 | XML_T('/'), XML_T('/'), XML_T('w'), XML_T('w'), XML_T('w'), | |
539 | XML_T('.'), XML_T('w'), XML_T('3'), | |
540 | XML_T('.'), XML_T('o'), XML_T('r'), XML_T('g'), | |
541 | XML_T('/'), XML_T('X'), XML_T('M'), XML_T('L'), | |
542 | XML_T('/'), XML_T('1'), XML_T('9'), XML_T('9'), XML_T('8'), | |
543 | XML_T('/'), XML_T('n'), XML_T('a'), XML_T('m'), XML_T('e'), | |
544 | XML_T('s'), XML_T('p'), XML_T('a'), XML_T('c'), XML_T('e'), | |
545 | XML_T('\0') | |
546 | }; | |
547 | ||
548 | XML_Parser parser = XML_ParserCreate(encodingName); | |
549 | if (parser) { | |
550 | XmlInitEncodingNS(&initEncoding, &encoding, 0); | |
551 | ns = 1; | |
552 | internalEncoding = XmlGetInternalEncodingNS(); | |
553 | namespaceSeparator = nsSep; | |
554 | } | |
555 | if (!setContext(parser, implicitContext)) { | |
556 | XML_ParserFree(parser); | |
557 | return 0; | |
558 | } | |
559 | return parser; | |
560 | } | |
561 | ||
562 | int XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) | |
563 | { | |
564 | if (!encodingName) | |
565 | protocolEncodingName = 0; | |
566 | else { | |
567 | protocolEncodingName = poolCopyString(&tempPool, encodingName); | |
568 | if (!protocolEncodingName) | |
569 | return 0; | |
570 | } | |
571 | return 1; | |
572 | } | |
573 | ||
574 | XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser, | |
575 | const XML_Char *context, | |
576 | const XML_Char *encodingName) | |
577 | { | |
578 | XML_Parser parser = oldParser; | |
579 | DTD *oldDtd = &dtd; | |
580 | XML_StartElementHandler oldStartElementHandler = startElementHandler; | |
581 | XML_EndElementHandler oldEndElementHandler = endElementHandler; | |
582 | XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler; | |
583 | XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler; | |
584 | XML_CommentHandler oldCommentHandler = commentHandler; | |
585 | XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler; | |
586 | XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler; | |
587 | XML_DefaultHandler oldDefaultHandler = defaultHandler; | |
588 | XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler = unparsedEntityDeclHandler; | |
589 | XML_NotationDeclHandler oldNotationDeclHandler = notationDeclHandler; | |
590 | XML_ExternalParsedEntityDeclHandler oldExternalParsedEntityDeclHandler = externalParsedEntityDeclHandler; | |
591 | XML_InternalParsedEntityDeclHandler oldInternalParsedEntityDeclHandler = internalParsedEntityDeclHandler; | |
592 | XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler = startNamespaceDeclHandler; | |
593 | XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler = endNamespaceDeclHandler; | |
594 | XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler; | |
595 | XML_ExternalEntityRefHandler oldExternalEntityRefHandler = externalEntityRefHandler; | |
596 | XML_UnknownEncodingHandler oldUnknownEncodingHandler = unknownEncodingHandler; | |
597 | void *oldUserData = userData; | |
598 | void *oldHandlerArg = handlerArg; | |
599 | int oldDefaultExpandInternalEntities = defaultExpandInternalEntities; | |
600 | void *oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg; | |
601 | #ifdef XML_DTD | |
602 | int oldParamEntityParsing = paramEntityParsing; | |
603 | #endif | |
604 | parser = (ns | |
605 | ? XML_ParserCreateNS(encodingName, namespaceSeparator) | |
606 | : XML_ParserCreate(encodingName)); | |
607 | if (!parser) | |
608 | return 0; | |
609 | startElementHandler = oldStartElementHandler; | |
610 | endElementHandler = oldEndElementHandler; | |
611 | characterDataHandler = oldCharacterDataHandler; | |
612 | processingInstructionHandler = oldProcessingInstructionHandler; | |
613 | commentHandler = oldCommentHandler; | |
614 | startCdataSectionHandler = oldStartCdataSectionHandler; | |
615 | endCdataSectionHandler = oldEndCdataSectionHandler; | |
616 | defaultHandler = oldDefaultHandler; | |
617 | unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler; | |
618 | notationDeclHandler = oldNotationDeclHandler; | |
619 | externalParsedEntityDeclHandler = oldExternalParsedEntityDeclHandler; | |
620 | internalParsedEntityDeclHandler = oldInternalParsedEntityDeclHandler; | |
621 | startNamespaceDeclHandler = oldStartNamespaceDeclHandler; | |
622 | endNamespaceDeclHandler = oldEndNamespaceDeclHandler; | |
623 | notStandaloneHandler = oldNotStandaloneHandler; | |
624 | externalEntityRefHandler = oldExternalEntityRefHandler; | |
625 | unknownEncodingHandler = oldUnknownEncodingHandler; | |
626 | userData = oldUserData; | |
627 | if (oldUserData == oldHandlerArg) | |
628 | handlerArg = userData; | |
629 | else | |
630 | handlerArg = parser; | |
631 | if (oldExternalEntityRefHandlerArg != oldParser) | |
632 | externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; | |
633 | defaultExpandInternalEntities = oldDefaultExpandInternalEntities; | |
634 | #ifdef XML_DTD | |
635 | paramEntityParsing = oldParamEntityParsing; | |
636 | if (context) { | |
637 | #endif /* XML_DTD */ | |
638 | if (!dtdCopy(&dtd, oldDtd) || !setContext(parser, context)) { | |
639 | XML_ParserFree(parser); | |
640 | return 0; | |
641 | } | |
642 | processor = externalEntityInitProcessor; | |
643 | #ifdef XML_DTD | |
644 | } | |
645 | else { | |
646 | dtdSwap(&dtd, oldDtd); | |
647 | parentParser = oldParser; | |
648 | XmlPrologStateInitExternalEntity(&prologState); | |
649 | dtd.complete = 1; | |
650 | hadExternalDoctype = 1; | |
651 | } | |
652 | #endif /* XML_DTD */ | |
653 | return parser; | |
654 | } | |
655 | ||
656 | static | |
657 | void destroyBindings(BINDING *bindings) | |
658 | { | |
659 | for (;;) { | |
660 | BINDING *b = bindings; | |
661 | if (!b) | |
662 | break; | |
663 | bindings = b->nextTagBinding; | |
664 | free(b->uri); | |
665 | free(b); | |
666 | } | |
667 | } | |
668 | ||
669 | void XML_ParserFree(XML_Parser parser) | |
670 | { | |
671 | for (;;) { | |
672 | TAG *p; | |
673 | if (tagStack == 0) { | |
674 | if (freeTagList == 0) | |
675 | break; | |
676 | tagStack = freeTagList; | |
677 | freeTagList = 0; | |
678 | } | |
679 | p = tagStack; | |
680 | tagStack = tagStack->parent; | |
681 | free(p->buf); | |
682 | destroyBindings(p->bindings); | |
683 | free(p); | |
684 | } | |
685 | destroyBindings(freeBindingList); | |
686 | destroyBindings(inheritedBindings); | |
687 | poolDestroy(&tempPool); | |
688 | poolDestroy(&temp2Pool); | |
689 | #ifdef XML_DTD | |
690 | if (parentParser) { | |
691 | if (hadExternalDoctype) | |
692 | dtd.complete = 0; | |
693 | dtdSwap(&dtd, &((Parser *)parentParser)->m_dtd); | |
694 | } | |
695 | #endif /* XML_DTD */ | |
696 | dtdDestroy(&dtd); | |
697 | free((void *)atts); | |
698 | free(groupConnector); | |
699 | free(buffer); | |
700 | free(dataBuf); | |
701 | free(unknownEncodingMem); | |
702 | if (unknownEncodingRelease) | |
703 | unknownEncodingRelease(unknownEncodingData); | |
704 | free(parser); | |
705 | } | |
706 | ||
707 | void XML_UseParserAsHandlerArg(XML_Parser parser) | |
708 | { | |
709 | handlerArg = parser; | |
710 | } | |
711 | ||
712 | void XML_SetUserData(XML_Parser parser, void *p) | |
713 | { | |
714 | if (handlerArg == userData) | |
715 | handlerArg = userData = p; | |
716 | else | |
717 | userData = p; | |
718 | } | |
719 | ||
720 | int XML_SetBase(XML_Parser parser, const XML_Char *p) | |
721 | { | |
722 | if (p) { | |
723 | p = poolCopyString(&dtd.pool, p); | |
724 | if (!p) | |
725 | return 0; | |
726 | curBase = p; | |
727 | } | |
728 | else | |
729 | curBase = 0; | |
730 | return 1; | |
731 | } | |
732 | ||
733 | const XML_Char *XML_GetBase(XML_Parser parser) | |
734 | { | |
735 | return curBase; | |
736 | } | |
737 | ||
738 | int XML_GetSpecifiedAttributeCount(XML_Parser parser) | |
739 | { | |
740 | return nSpecifiedAtts; | |
741 | } | |
742 | ||
743 | int XML_GetIdAttributeIndex(XML_Parser parser) | |
744 | { | |
745 | return idAttIndex; | |
746 | } | |
747 | ||
748 | void XML_SetElementHandler(XML_Parser parser, | |
749 | XML_StartElementHandler start, | |
750 | XML_EndElementHandler end) | |
751 | { | |
752 | startElementHandler = start; | |
753 | endElementHandler = end; | |
754 | } | |
755 | ||
756 | void XML_SetCharacterDataHandler(XML_Parser parser, | |
757 | XML_CharacterDataHandler handler) | |
758 | { | |
759 | characterDataHandler = handler; | |
760 | } | |
761 | ||
762 | void XML_SetProcessingInstructionHandler(XML_Parser parser, | |
763 | XML_ProcessingInstructionHandler handler) | |
764 | { | |
765 | processingInstructionHandler = handler; | |
766 | } | |
767 | ||
768 | void XML_SetCommentHandler(XML_Parser parser, | |
769 | XML_CommentHandler handler) | |
770 | { | |
771 | commentHandler = handler; | |
772 | } | |
773 | ||
774 | void XML_SetCdataSectionHandler(XML_Parser parser, | |
775 | XML_StartCdataSectionHandler start, | |
776 | XML_EndCdataSectionHandler end) | |
777 | { | |
778 | startCdataSectionHandler = start; | |
779 | endCdataSectionHandler = end; | |
780 | } | |
781 | ||
782 | void XML_SetDefaultHandler(XML_Parser parser, | |
783 | XML_DefaultHandler handler) | |
784 | { | |
785 | defaultHandler = handler; | |
786 | defaultExpandInternalEntities = 0; | |
787 | } | |
788 | ||
789 | void XML_SetDefaultHandlerExpand(XML_Parser parser, | |
790 | XML_DefaultHandler handler) | |
791 | { | |
792 | defaultHandler = handler; | |
793 | defaultExpandInternalEntities = 1; | |
794 | } | |
795 | ||
796 | void XML_SetDoctypeDeclHandler(XML_Parser parser, | |
797 | XML_StartDoctypeDeclHandler start, | |
798 | XML_EndDoctypeDeclHandler end) | |
799 | { | |
800 | startDoctypeDeclHandler = start; | |
801 | endDoctypeDeclHandler = end; | |
802 | } | |
803 | ||
804 | void XML_SetUnparsedEntityDeclHandler(XML_Parser parser, | |
805 | XML_UnparsedEntityDeclHandler handler) | |
806 | { | |
807 | unparsedEntityDeclHandler = handler; | |
808 | } | |
809 | ||
810 | void XML_SetExternalParsedEntityDeclHandler(XML_Parser parser, | |
811 | XML_ExternalParsedEntityDeclHandler handler) | |
812 | { | |
813 | externalParsedEntityDeclHandler = handler; | |
814 | } | |
815 | ||
816 | void XML_SetInternalParsedEntityDeclHandler(XML_Parser parser, | |
817 | XML_InternalParsedEntityDeclHandler handler) | |
818 | { | |
819 | internalParsedEntityDeclHandler = handler; | |
820 | } | |
821 | ||
822 | void XML_SetNotationDeclHandler(XML_Parser parser, | |
823 | XML_NotationDeclHandler handler) | |
824 | { | |
825 | notationDeclHandler = handler; | |
826 | } | |
827 | ||
828 | void XML_SetNamespaceDeclHandler(XML_Parser parser, | |
829 | XML_StartNamespaceDeclHandler start, | |
830 | XML_EndNamespaceDeclHandler end) | |
831 | { | |
832 | startNamespaceDeclHandler = start; | |
833 | endNamespaceDeclHandler = end; | |
834 | } | |
835 | ||
836 | void XML_SetNotStandaloneHandler(XML_Parser parser, | |
837 | XML_NotStandaloneHandler handler) | |
838 | { | |
839 | notStandaloneHandler = handler; | |
840 | } | |
841 | ||
842 | void XML_SetExternalEntityRefHandler(XML_Parser parser, | |
843 | XML_ExternalEntityRefHandler handler) | |
844 | { | |
845 | externalEntityRefHandler = handler; | |
846 | } | |
847 | ||
848 | void XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) | |
849 | { | |
850 | if (arg) | |
851 | externalEntityRefHandlerArg = arg; | |
852 | else | |
853 | externalEntityRefHandlerArg = parser; | |
854 | } | |
855 | ||
856 | void XML_SetUnknownEncodingHandler(XML_Parser parser, | |
857 | XML_UnknownEncodingHandler handler, | |
858 | void *data) | |
859 | { | |
860 | unknownEncodingHandler = handler; | |
861 | unknownEncodingHandlerData = data; | |
862 | } | |
863 | ||
864 | int XML_SetParamEntityParsing(XML_Parser parser, | |
865 | enum XML_ParamEntityParsing parsing) | |
866 | { | |
867 | #ifdef XML_DTD | |
868 | paramEntityParsing = parsing; | |
869 | return 1; | |
870 | #else | |
871 | return parsing == XML_PARAM_ENTITY_PARSING_NEVER; | |
872 | #endif | |
873 | } | |
874 | ||
875 | int XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) | |
876 | { | |
877 | if (len == 0) { | |
878 | if (!isFinal) | |
879 | return 1; | |
880 | positionPtr = bufferPtr; | |
881 | errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0); | |
882 | if (errorCode == XML_ERROR_NONE) | |
883 | return 1; | |
884 | eventEndPtr = eventPtr; | |
885 | processor = errorProcessor; | |
886 | return 0; | |
887 | } | |
888 | else if (bufferPtr == bufferEnd) { | |
889 | const char *end; | |
890 | int nLeftOver; | |
891 | parseEndByteIndex += len; | |
892 | positionPtr = s; | |
893 | if (isFinal) { | |
894 | errorCode = processor(parser, s, parseEndPtr = s + len, 0); | |
895 | if (errorCode == XML_ERROR_NONE) | |
896 | return 1; | |
897 | eventEndPtr = eventPtr; | |
898 | processor = errorProcessor; | |
899 | return 0; | |
900 | } | |
901 | errorCode = processor(parser, s, parseEndPtr = s + len, &end); | |
902 | if (errorCode != XML_ERROR_NONE) { | |
903 | eventEndPtr = eventPtr; | |
904 | processor = errorProcessor; | |
905 | return 0; | |
906 | } | |
907 | XmlUpdatePosition(encoding, positionPtr, end, &position); | |
908 | nLeftOver = s + len - end; | |
909 | if (nLeftOver) { | |
910 | if (buffer == 0 || nLeftOver > bufferLim - buffer) { | |
911 | /* FIXME avoid integer overflow */ | |
912 | buffer = buffer == 0 ? malloc(len * 2) : realloc(buffer, len * 2); | |
913 | /* FIXME storage leak if realloc fails */ | |
914 | if (!buffer) { | |
915 | errorCode = XML_ERROR_NO_MEMORY; | |
916 | eventPtr = eventEndPtr = 0; | |
917 | processor = errorProcessor; | |
918 | return 0; | |
919 | } | |
920 | bufferLim = buffer + len * 2; | |
921 | } | |
922 | memcpy(buffer, end, nLeftOver); | |
923 | bufferPtr = buffer; | |
924 | bufferEnd = buffer + nLeftOver; | |
925 | } | |
926 | return 1; | |
927 | } | |
928 | else { | |
929 | memcpy(XML_GetBuffer(parser, len), s, len); | |
930 | return XML_ParseBuffer(parser, len, isFinal); | |
931 | } | |
932 | } | |
933 | ||
934 | int XML_ParseBuffer(XML_Parser parser, int len, int isFinal) | |
935 | { | |
936 | const char *start = bufferPtr; | |
937 | positionPtr = start; | |
938 | bufferEnd += len; | |
939 | parseEndByteIndex += len; | |
940 | errorCode = processor(parser, start, parseEndPtr = bufferEnd, | |
941 | isFinal ? (const char **)0 : &bufferPtr); | |
942 | if (errorCode == XML_ERROR_NONE) { | |
943 | if (!isFinal) | |
944 | XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position); | |
945 | return 1; | |
946 | } | |
947 | else { | |
948 | eventEndPtr = eventPtr; | |
949 | processor = errorProcessor; | |
950 | return 0; | |
951 | } | |
952 | } | |
953 | ||
954 | void *XML_GetBuffer(XML_Parser parser, int len) | |
955 | { | |
956 | if (len > bufferLim - bufferEnd) { | |
957 | /* FIXME avoid integer overflow */ | |
958 | int neededSize = len + (bufferEnd - bufferPtr); | |
959 | if (neededSize <= bufferLim - buffer) { | |
960 | memmove(buffer, bufferPtr, bufferEnd - bufferPtr); | |
961 | bufferEnd = buffer + (bufferEnd - bufferPtr); | |
962 | bufferPtr = buffer; | |
963 | } | |
964 | else { | |
965 | char *newBuf; | |
966 | int bufferSize = bufferLim - bufferPtr; | |
967 | if (bufferSize == 0) | |
968 | bufferSize = INIT_BUFFER_SIZE; | |
969 | do { | |
970 | bufferSize *= 2; | |
971 | } while (bufferSize < neededSize); | |
972 | newBuf = malloc(bufferSize); | |
973 | if (newBuf == 0) { | |
974 | errorCode = XML_ERROR_NO_MEMORY; | |
975 | return 0; | |
976 | } | |
977 | bufferLim = newBuf + bufferSize; | |
978 | if (bufferPtr) { | |
979 | memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr); | |
980 | free(buffer); | |
981 | } | |
982 | bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
983 | bufferPtr = buffer = newBuf; | |
984 | } | |
985 | } | |
986 | return bufferEnd; | |
987 | } | |
988 | ||
989 | enum XML_Error XML_GetErrorCode(XML_Parser parser) | |
990 | { | |
991 | return errorCode; | |
992 | } | |
993 | ||
994 | long XML_GetCurrentByteIndex(XML_Parser parser) | |
995 | { | |
996 | if (eventPtr) | |
997 | return parseEndByteIndex - (parseEndPtr - eventPtr); | |
998 | return -1; | |
999 | } | |
1000 | ||
1001 | int XML_GetCurrentByteCount(XML_Parser parser) | |
1002 | { | |
1003 | if (eventEndPtr && eventPtr) | |
1004 | return eventEndPtr - eventPtr; | |
1005 | return 0; | |
1006 | } | |
1007 | ||
1008 | int XML_GetCurrentLineNumber(XML_Parser parser) | |
1009 | { | |
1010 | if (eventPtr) { | |
1011 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
1012 | positionPtr = eventPtr; | |
1013 | } | |
1014 | return position.lineNumber + 1; | |
1015 | } | |
1016 | ||
1017 | int XML_GetCurrentColumnNumber(XML_Parser parser) | |
1018 | { | |
1019 | if (eventPtr) { | |
1020 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
1021 | positionPtr = eventPtr; | |
1022 | } | |
1023 | return position.columnNumber; | |
1024 | } | |
1025 | ||
1026 | void XML_DefaultCurrent(XML_Parser parser) | |
1027 | { | |
1028 | if (defaultHandler) { | |
1029 | if (openInternalEntities) | |
1030 | reportDefault(parser, | |
1031 | internalEncoding, | |
1032 | openInternalEntities->internalEventPtr, | |
1033 | openInternalEntities->internalEventEndPtr); | |
1034 | else | |
1035 | reportDefault(parser, encoding, eventPtr, eventEndPtr); | |
1036 | } | |
1037 | } | |
1038 | ||
1039 | const XML_LChar *XML_ErrorString(int code) | |
1040 | { | |
1041 | static const XML_LChar *message[] = { | |
1042 | 0, | |
1043 | XML_T("out of memory"), | |
1044 | XML_T("syntax error"), | |
1045 | XML_T("no element found"), | |
1046 | XML_T("not well-formed"), | |
1047 | XML_T("unclosed token"), | |
1048 | XML_T("unclosed token"), | |
1049 | XML_T("mismatched tag"), | |
1050 | XML_T("duplicate attribute"), | |
1051 | XML_T("junk after document element"), | |
1052 | XML_T("illegal parameter entity reference"), | |
1053 | XML_T("undefined entity"), | |
1054 | XML_T("recursive entity reference"), | |
1055 | XML_T("asynchronous entity"), | |
1056 | XML_T("reference to invalid character number"), | |
1057 | XML_T("reference to binary entity"), | |
1058 | XML_T("reference to external entity in attribute"), | |
1059 | XML_T("xml processing instruction not at start of external entity"), | |
1060 | XML_T("unknown encoding"), | |
1061 | XML_T("encoding specified in XML declaration is incorrect"), | |
1062 | XML_T("unclosed CDATA section"), | |
1063 | XML_T("error in processing external entity reference"), | |
1064 | XML_T("document is not standalone") | |
1065 | }; | |
1066 | if (code > 0 && code < sizeof(message)/sizeof(message[0])) | |
1067 | return message[code]; | |
1068 | return 0; | |
1069 | } | |
1070 | ||
1071 | static | |
1072 | enum XML_Error contentProcessor(XML_Parser parser, | |
1073 | const char *start, | |
1074 | const char *end, | |
1075 | const char **endPtr) | |
1076 | { | |
1077 | return doContent(parser, 0, encoding, start, end, endPtr); | |
1078 | } | |
1079 | ||
1080 | static | |
1081 | enum XML_Error externalEntityInitProcessor(XML_Parser parser, | |
1082 | const char *start, | |
1083 | const char *end, | |
1084 | const char **endPtr) | |
1085 | { | |
1086 | enum XML_Error result = initializeEncoding(parser); | |
1087 | if (result != XML_ERROR_NONE) | |
1088 | return result; | |
1089 | processor = externalEntityInitProcessor2; | |
1090 | return externalEntityInitProcessor2(parser, start, end, endPtr); | |
1091 | } | |
1092 | ||
1093 | static | |
1094 | enum XML_Error externalEntityInitProcessor2(XML_Parser parser, | |
1095 | const char *start, | |
1096 | const char *end, | |
1097 | const char **endPtr) | |
1098 | { | |
1099 | const char *next; | |
1100 | int tok = XmlContentTok(encoding, start, end, &next); | |
1101 | switch (tok) { | |
1102 | case XML_TOK_BOM: | |
1103 | start = next; | |
1104 | break; | |
1105 | case XML_TOK_PARTIAL: | |
1106 | if (endPtr) { | |
1107 | *endPtr = start; | |
1108 | return XML_ERROR_NONE; | |
1109 | } | |
1110 | eventPtr = start; | |
1111 | return XML_ERROR_UNCLOSED_TOKEN; | |
1112 | case XML_TOK_PARTIAL_CHAR: | |
1113 | if (endPtr) { | |
1114 | *endPtr = start; | |
1115 | return XML_ERROR_NONE; | |
1116 | } | |
1117 | eventPtr = start; | |
1118 | return XML_ERROR_PARTIAL_CHAR; | |
1119 | } | |
1120 | processor = externalEntityInitProcessor3; | |
1121 | return externalEntityInitProcessor3(parser, start, end, endPtr); | |
1122 | } | |
1123 | ||
1124 | static | |
1125 | enum XML_Error externalEntityInitProcessor3(XML_Parser parser, | |
1126 | const char *start, | |
1127 | const char *end, | |
1128 | const char **endPtr) | |
1129 | { | |
1130 | const char *next; | |
1131 | int tok = XmlContentTok(encoding, start, end, &next); | |
1132 | switch (tok) { | |
1133 | case XML_TOK_XML_DECL: | |
1134 | { | |
1135 | enum XML_Error result = processXmlDecl(parser, 1, start, next); | |
1136 | if (result != XML_ERROR_NONE) | |
1137 | return result; | |
1138 | start = next; | |
1139 | } | |
1140 | break; | |
1141 | case XML_TOK_PARTIAL: | |
1142 | if (endPtr) { | |
1143 | *endPtr = start; | |
1144 | return XML_ERROR_NONE; | |
1145 | } | |
1146 | eventPtr = start; | |
1147 | return XML_ERROR_UNCLOSED_TOKEN; | |
1148 | case XML_TOK_PARTIAL_CHAR: | |
1149 | if (endPtr) { | |
1150 | *endPtr = start; | |
1151 | return XML_ERROR_NONE; | |
1152 | } | |
1153 | eventPtr = start; | |
1154 | return XML_ERROR_PARTIAL_CHAR; | |
1155 | } | |
1156 | processor = externalEntityContentProcessor; | |
1157 | tagLevel = 1; | |
1158 | return doContent(parser, 1, encoding, start, end, endPtr); | |
1159 | } | |
1160 | ||
1161 | static | |
1162 | enum XML_Error externalEntityContentProcessor(XML_Parser parser, | |
1163 | const char *start, | |
1164 | const char *end, | |
1165 | const char **endPtr) | |
1166 | { | |
1167 | return doContent(parser, 1, encoding, start, end, endPtr); | |
1168 | } | |
1169 | ||
1170 | static enum XML_Error | |
1171 | doContent(XML_Parser parser, | |
1172 | int startTagLevel, | |
1173 | const ENCODING *enc, | |
1174 | const char *s, | |
1175 | const char *end, | |
1176 | const char **nextPtr) | |
1177 | { | |
1178 | const char **eventPP; | |
1179 | const char **eventEndPP; | |
1180 | if (enc == encoding) { | |
1181 | eventPP = &eventPtr; | |
1182 | eventEndPP = &eventEndPtr; | |
1183 | } | |
1184 | else { | |
1185 | eventPP = &(openInternalEntities->internalEventPtr); | |
1186 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
1187 | } | |
1188 | *eventPP = s; | |
1189 | for (;;) { | |
1190 | const char *next = s; /* XmlContentTok doesn't always set the last arg */ | |
1191 | int tok = XmlContentTok(enc, s, end, &next); | |
1192 | *eventEndPP = next; | |
1193 | switch (tok) { | |
1194 | case XML_TOK_TRAILING_CR: | |
1195 | if (nextPtr) { | |
1196 | *nextPtr = s; | |
1197 | return XML_ERROR_NONE; | |
1198 | } | |
1199 | *eventEndPP = end; | |
1200 | if (characterDataHandler) { | |
1201 | XML_Char c = 0xA; | |
1202 | characterDataHandler(handlerArg, &c, 1); | |
1203 | } | |
1204 | else if (defaultHandler) | |
1205 | reportDefault(parser, enc, s, end); | |
1206 | if (startTagLevel == 0) | |
1207 | return XML_ERROR_NO_ELEMENTS; | |
1208 | if (tagLevel != startTagLevel) | |
1209 | return XML_ERROR_ASYNC_ENTITY; | |
1210 | return XML_ERROR_NONE; | |
1211 | case XML_TOK_NONE: | |
1212 | if (nextPtr) { | |
1213 | *nextPtr = s; | |
1214 | return XML_ERROR_NONE; | |
1215 | } | |
1216 | if (startTagLevel > 0) { | |
1217 | if (tagLevel != startTagLevel) | |
1218 | return XML_ERROR_ASYNC_ENTITY; | |
1219 | return XML_ERROR_NONE; | |
1220 | } | |
1221 | return XML_ERROR_NO_ELEMENTS; | |
1222 | case XML_TOK_INVALID: | |
1223 | *eventPP = next; | |
1224 | return XML_ERROR_INVALID_TOKEN; | |
1225 | case XML_TOK_PARTIAL: | |
1226 | if (nextPtr) { | |
1227 | *nextPtr = s; | |
1228 | return XML_ERROR_NONE; | |
1229 | } | |
1230 | return XML_ERROR_UNCLOSED_TOKEN; | |
1231 | case XML_TOK_PARTIAL_CHAR: | |
1232 | if (nextPtr) { | |
1233 | *nextPtr = s; | |
1234 | return XML_ERROR_NONE; | |
1235 | } | |
1236 | return XML_ERROR_PARTIAL_CHAR; | |
1237 | case XML_TOK_ENTITY_REF: | |
1238 | { | |
1239 | const XML_Char *name; | |
1240 | ENTITY *entity; | |
1241 | XML_Char ch = XmlPredefinedEntityName(enc, | |
1242 | s + enc->minBytesPerChar, | |
1243 | next - enc->minBytesPerChar); | |
1244 | if (ch) { | |
1245 | if (characterDataHandler) | |
1246 | characterDataHandler(handlerArg, &ch, 1); | |
1247 | else if (defaultHandler) | |
1248 | reportDefault(parser, enc, s, next); | |
1249 | break; | |
1250 | } | |
1251 | name = poolStoreString(&dtd.pool, enc, | |
1252 | s + enc->minBytesPerChar, | |
1253 | next - enc->minBytesPerChar); | |
1254 | if (!name) | |
1255 | return XML_ERROR_NO_MEMORY; | |
1256 | entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); | |
1257 | poolDiscard(&dtd.pool); | |
1258 | if (!entity) { | |
1259 | if (dtd.complete || dtd.standalone) | |
1260 | return XML_ERROR_UNDEFINED_ENTITY; | |
1261 | if (defaultHandler) | |
1262 | reportDefault(parser, enc, s, next); | |
1263 | break; | |
1264 | } | |
1265 | if (entity->open) | |
1266 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
1267 | if (entity->notation) | |
1268 | return XML_ERROR_BINARY_ENTITY_REF; | |
1269 | if (entity) { | |
1270 | if (entity->textPtr) { | |
1271 | enum XML_Error result; | |
1272 | OPEN_INTERNAL_ENTITY openEntity; | |
1273 | if (defaultHandler && !defaultExpandInternalEntities) { | |
1274 | reportDefault(parser, enc, s, next); | |
1275 | break; | |
1276 | } | |
1277 | entity->open = 1; | |
1278 | openEntity.next = openInternalEntities; | |
1279 | openInternalEntities = &openEntity; | |
1280 | openEntity.entity = entity; | |
1281 | openEntity.internalEventPtr = 0; | |
1282 | openEntity.internalEventEndPtr = 0; | |
1283 | result = doContent(parser, | |
1284 | tagLevel, | |
1285 | internalEncoding, | |
1286 | (char *)entity->textPtr, | |
1287 | (char *)(entity->textPtr + entity->textLen), | |
1288 | 0); | |
1289 | entity->open = 0; | |
1290 | openInternalEntities = openEntity.next; | |
1291 | if (result) | |
1292 | return result; | |
1293 | } | |
1294 | else if (externalEntityRefHandler) { | |
1295 | const XML_Char *context; | |
1296 | entity->open = 1; | |
1297 | context = getContext(parser); | |
1298 | entity->open = 0; | |
1299 | if (!context) | |
1300 | return XML_ERROR_NO_MEMORY; | |
1301 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
1302 | context, | |
1303 | entity->base, | |
1304 | entity->systemId, | |
1305 | entity->publicId)) | |
1306 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
1307 | poolDiscard(&tempPool); | |
1308 | } | |
1309 | else if (defaultHandler) | |
1310 | reportDefault(parser, enc, s, next); | |
1311 | } | |
1312 | break; | |
1313 | } | |
1314 | case XML_TOK_START_TAG_WITH_ATTS: | |
1315 | if (!startElementHandler) { | |
1316 | enum XML_Error result = storeAtts(parser, enc, s, 0, 0); | |
1317 | if (result) | |
1318 | return result; | |
1319 | } | |
1320 | /* fall through */ | |
1321 | case XML_TOK_START_TAG_NO_ATTS: | |
1322 | { | |
1323 | TAG *tag; | |
1324 | if (freeTagList) { | |
1325 | tag = freeTagList; | |
1326 | freeTagList = freeTagList->parent; | |
1327 | } | |
1328 | else { | |
1329 | tag = malloc(sizeof(TAG)); | |
1330 | if (!tag) | |
1331 | return XML_ERROR_NO_MEMORY; | |
1332 | tag->buf = malloc(INIT_TAG_BUF_SIZE); | |
1333 | if (!tag->buf) | |
1334 | return XML_ERROR_NO_MEMORY; | |
1335 | tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; | |
1336 | } | |
1337 | tag->bindings = 0; | |
1338 | tag->parent = tagStack; | |
1339 | tagStack = tag; | |
1340 | tag->name.localPart = 0; | |
1341 | tag->rawName = s + enc->minBytesPerChar; | |
1342 | tag->rawNameLength = XmlNameLength(enc, tag->rawName); | |
1343 | if (nextPtr) { | |
1344 | /* Need to guarantee that: | |
1345 | tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)) <= tag->bufEnd - sizeof(XML_Char) */ | |
1346 | if (tag->rawNameLength + (int)(sizeof(XML_Char) - 1) + (int)sizeof(XML_Char) > tag->bufEnd - tag->buf) { | |
1347 | int bufSize = tag->rawNameLength * 4; | |
1348 | bufSize = ROUND_UP(bufSize, sizeof(XML_Char)); | |
1349 | tag->buf = realloc(tag->buf, bufSize); | |
1350 | if (!tag->buf) | |
1351 | return XML_ERROR_NO_MEMORY; | |
1352 | tag->bufEnd = tag->buf + bufSize; | |
1353 | } | |
1354 | memcpy(tag->buf, tag->rawName, tag->rawNameLength); | |
1355 | tag->rawName = tag->buf; | |
1356 | } | |
1357 | ++tagLevel; | |
1358 | if (startElementHandler) { | |
1359 | enum XML_Error result; | |
1360 | XML_Char *toPtr; | |
1361 | for (;;) { | |
1362 | const char *rawNameEnd = tag->rawName + tag->rawNameLength; | |
1363 | const char *fromPtr = tag->rawName; | |
1364 | int bufSize; | |
1365 | if (nextPtr) | |
1366 | toPtr = (XML_Char *)(tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char))); | |
1367 | else | |
1368 | toPtr = (XML_Char *)tag->buf; | |
1369 | tag->name.str = toPtr; | |
1370 | XmlConvert(enc, | |
1371 | &fromPtr, rawNameEnd, | |
1372 | (ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1); | |
1373 | if (fromPtr == rawNameEnd) | |
1374 | break; | |
1375 | bufSize = (tag->bufEnd - tag->buf) << 1; | |
1376 | tag->buf = realloc(tag->buf, bufSize); | |
1377 | if (!tag->buf) | |
1378 | return XML_ERROR_NO_MEMORY; | |
1379 | tag->bufEnd = tag->buf + bufSize; | |
1380 | if (nextPtr) | |
1381 | tag->rawName = tag->buf; | |
1382 | } | |
1383 | *toPtr = XML_T('\0'); | |
1384 | result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings)); | |
1385 | if (result) | |
1386 | return result; | |
1387 | startElementHandler(handlerArg, tag->name.str, (const XML_Char **)atts); | |
1388 | poolClear(&tempPool); | |
1389 | } | |
1390 | else { | |
1391 | tag->name.str = 0; | |
1392 | if (defaultHandler) | |
1393 | reportDefault(parser, enc, s, next); | |
1394 | } | |
1395 | break; | |
1396 | } | |
1397 | case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: | |
1398 | if (!startElementHandler) { | |
1399 | enum XML_Error result = storeAtts(parser, enc, s, 0, 0); | |
1400 | if (result) | |
1401 | return result; | |
1402 | } | |
1403 | /* fall through */ | |
1404 | case XML_TOK_EMPTY_ELEMENT_NO_ATTS: | |
1405 | if (startElementHandler || endElementHandler) { | |
1406 | const char *rawName = s + enc->minBytesPerChar; | |
1407 | enum XML_Error result; | |
1408 | BINDING *bindings = 0; | |
1409 | TAG_NAME name; | |
1410 | name.str = poolStoreString(&tempPool, enc, rawName, | |
1411 | rawName + XmlNameLength(enc, rawName)); | |
1412 | if (!name.str) | |
1413 | return XML_ERROR_NO_MEMORY; | |
1414 | poolFinish(&tempPool); | |
1415 | result = storeAtts(parser, enc, s, &name, &bindings); | |
1416 | if (result) | |
1417 | return result; | |
1418 | poolFinish(&tempPool); | |
1419 | if (startElementHandler) | |
1420 | startElementHandler(handlerArg, name.str, (const XML_Char **)atts); | |
1421 | if (endElementHandler) { | |
1422 | if (startElementHandler) | |
1423 | *eventPP = *eventEndPP; | |
1424 | endElementHandler(handlerArg, name.str); | |
1425 | } | |
1426 | poolClear(&tempPool); | |
1427 | while (bindings) { | |
1428 | BINDING *b = bindings; | |
1429 | if (endNamespaceDeclHandler) | |
1430 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
1431 | bindings = bindings->nextTagBinding; | |
1432 | b->nextTagBinding = freeBindingList; | |
1433 | freeBindingList = b; | |
1434 | b->prefix->binding = b->prevPrefixBinding; | |
1435 | } | |
1436 | } | |
1437 | else if (defaultHandler) | |
1438 | reportDefault(parser, enc, s, next); | |
1439 | if (tagLevel == 0) | |
1440 | return epilogProcessor(parser, next, end, nextPtr); | |
1441 | break; | |
1442 | case XML_TOK_END_TAG: | |
1443 | if (tagLevel == startTagLevel) | |
1444 | return XML_ERROR_ASYNC_ENTITY; | |
1445 | else { | |
1446 | int len; | |
1447 | const char *rawName; | |
1448 | TAG *tag = tagStack; | |
1449 | tagStack = tag->parent; | |
1450 | tag->parent = freeTagList; | |
1451 | freeTagList = tag; | |
1452 | rawName = s + enc->minBytesPerChar*2; | |
1453 | len = XmlNameLength(enc, rawName); | |
1454 | if (len != tag->rawNameLength | |
1455 | || memcmp(tag->rawName, rawName, len) != 0) { | |
1456 | *eventPP = rawName; | |
1457 | return XML_ERROR_TAG_MISMATCH; | |
1458 | } | |
1459 | --tagLevel; | |
1460 | if (endElementHandler && tag->name.str) { | |
1461 | if (tag->name.localPart) { | |
1462 | XML_Char *to = (XML_Char *)tag->name.str + tag->name.uriLen; | |
1463 | const XML_Char *from = tag->name.localPart; | |
1464 | while ((*to++ = *from++) != 0) | |
1465 | ; | |
1466 | } | |
1467 | endElementHandler(handlerArg, tag->name.str); | |
1468 | } | |
1469 | else if (defaultHandler) | |
1470 | reportDefault(parser, enc, s, next); | |
1471 | while (tag->bindings) { | |
1472 | BINDING *b = tag->bindings; | |
1473 | if (endNamespaceDeclHandler) | |
1474 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
1475 | tag->bindings = tag->bindings->nextTagBinding; | |
1476 | b->nextTagBinding = freeBindingList; | |
1477 | freeBindingList = b; | |
1478 | b->prefix->binding = b->prevPrefixBinding; | |
1479 | } | |
1480 | if (tagLevel == 0) | |
1481 | return epilogProcessor(parser, next, end, nextPtr); | |
1482 | } | |
1483 | break; | |
1484 | case XML_TOK_CHAR_REF: | |
1485 | { | |
1486 | int n = XmlCharRefNumber(enc, s); | |
1487 | if (n < 0) | |
1488 | return XML_ERROR_BAD_CHAR_REF; | |
1489 | if (characterDataHandler) { | |
1490 | XML_Char buf[XML_ENCODE_MAX]; | |
1491 | characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf)); | |
1492 | } | |
1493 | else if (defaultHandler) | |
1494 | reportDefault(parser, enc, s, next); | |
1495 | } | |
1496 | break; | |
1497 | case XML_TOK_XML_DECL: | |
1498 | return XML_ERROR_MISPLACED_XML_PI; | |
1499 | case XML_TOK_DATA_NEWLINE: | |
1500 | if (characterDataHandler) { | |
1501 | XML_Char c = 0xA; | |
1502 | characterDataHandler(handlerArg, &c, 1); | |
1503 | } | |
1504 | else if (defaultHandler) | |
1505 | reportDefault(parser, enc, s, next); | |
1506 | break; | |
1507 | case XML_TOK_CDATA_SECT_OPEN: | |
1508 | { | |
1509 | enum XML_Error result; | |
1510 | if (startCdataSectionHandler) | |
1511 | startCdataSectionHandler(handlerArg); | |
1512 | #if 0 | |
1513 | /* Suppose you doing a transformation on a document that involves | |
1514 | changing only the character data. You set up a defaultHandler | |
1515 | and a characterDataHandler. The defaultHandler simply copies | |
1516 | characters through. The characterDataHandler does the transformation | |
1517 | and writes the characters out escaping them as necessary. This case | |
1518 | will fail to work if we leave out the following two lines (because & | |
1519 | and < inside CDATA sections will be incorrectly escaped). | |
1520 | ||
1521 | However, now we have a start/endCdataSectionHandler, so it seems | |
1522 | easier to let the user deal with this. */ | |
1523 | ||
1524 | else if (characterDataHandler) | |
1525 | characterDataHandler(handlerArg, dataBuf, 0); | |
1526 | #endif | |
1527 | else if (defaultHandler) | |
1528 | reportDefault(parser, enc, s, next); | |
1529 | result = doCdataSection(parser, enc, &next, end, nextPtr); | |
1530 | if (!next) { | |
1531 | processor = cdataSectionProcessor; | |
1532 | return result; | |
1533 | } | |
1534 | } | |
1535 | break; | |
1536 | case XML_TOK_TRAILING_RSQB: | |
1537 | if (nextPtr) { | |
1538 | *nextPtr = s; | |
1539 | return XML_ERROR_NONE; | |
1540 | } | |
1541 | if (characterDataHandler) { | |
1542 | if (MUST_CONVERT(enc, s)) { | |
1543 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
1544 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
1545 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
1546 | } | |
1547 | else | |
1548 | characterDataHandler(handlerArg, | |
1549 | (XML_Char *)s, | |
1550 | (XML_Char *)end - (XML_Char *)s); | |
1551 | } | |
1552 | else if (defaultHandler) | |
1553 | reportDefault(parser, enc, s, end); | |
1554 | if (startTagLevel == 0) { | |
1555 | *eventPP = end; | |
1556 | return XML_ERROR_NO_ELEMENTS; | |
1557 | } | |
1558 | if (tagLevel != startTagLevel) { | |
1559 | *eventPP = end; | |
1560 | return XML_ERROR_ASYNC_ENTITY; | |
1561 | } | |
1562 | return XML_ERROR_NONE; | |
1563 | case XML_TOK_DATA_CHARS: | |
1564 | if (characterDataHandler) { | |
1565 | if (MUST_CONVERT(enc, s)) { | |
1566 | for (;;) { | |
1567 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
1568 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
1569 | *eventEndPP = s; | |
1570 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
1571 | if (s == next) | |
1572 | break; | |
1573 | *eventPP = s; | |
1574 | } | |
1575 | } | |
1576 | else | |
1577 | characterDataHandler(handlerArg, | |
1578 | (XML_Char *)s, | |
1579 | (XML_Char *)next - (XML_Char *)s); | |
1580 | } | |
1581 | else if (defaultHandler) | |
1582 | reportDefault(parser, enc, s, next); | |
1583 | break; | |
1584 | case XML_TOK_PI: | |
1585 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
1586 | return XML_ERROR_NO_MEMORY; | |
1587 | break; | |
1588 | case XML_TOK_COMMENT: | |
1589 | if (!reportComment(parser, enc, s, next)) | |
1590 | return XML_ERROR_NO_MEMORY; | |
1591 | break; | |
1592 | default: | |
1593 | if (defaultHandler) | |
1594 | reportDefault(parser, enc, s, next); | |
1595 | break; | |
1596 | } | |
1597 | *eventPP = s = next; | |
1598 | } | |
1599 | /* not reached */ | |
1600 | } | |
1601 | ||
1602 | /* If tagNamePtr is non-null, build a real list of attributes, | |
1603 | otherwise just check the attributes for well-formedness. */ | |
1604 | ||
1605 | static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *enc, | |
1606 | const char *attStr, TAG_NAME *tagNamePtr, | |
1607 | BINDING **bindingsPtr) | |
1608 | { | |
1609 | ELEMENT_TYPE *elementType = 0; | |
1610 | int nDefaultAtts = 0; | |
1611 | const XML_Char **appAtts; /* the attribute list to pass to the application */ | |
1612 | int attIndex = 0; | |
1613 | int i; | |
1614 | int n; | |
1615 | int nPrefixes = 0; | |
1616 | BINDING *binding; | |
1617 | const XML_Char *localPart; | |
1618 | ||
1619 | /* lookup the element type name */ | |
1620 | if (tagNamePtr) { | |
1621 | elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, 0); | |
1622 | if (!elementType) { | |
1623 | tagNamePtr->str = poolCopyString(&dtd.pool, tagNamePtr->str); | |
1624 | if (!tagNamePtr->str) | |
1625 | return XML_ERROR_NO_MEMORY; | |
1626 | elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, sizeof(ELEMENT_TYPE)); | |
1627 | if (!elementType) | |
1628 | return XML_ERROR_NO_MEMORY; | |
1629 | if (ns && !setElementTypePrefix(parser, elementType)) | |
1630 | return XML_ERROR_NO_MEMORY; | |
1631 | } | |
1632 | nDefaultAtts = elementType->nDefaultAtts; | |
1633 | } | |
1634 | /* get the attributes from the tokenizer */ | |
1635 | n = XmlGetAttributes(enc, attStr, attsSize, atts); | |
1636 | if (n + nDefaultAtts > attsSize) { | |
1637 | int oldAttsSize = attsSize; | |
1638 | attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; | |
1639 | atts = realloc((void *)atts, attsSize * sizeof(ATTRIBUTE)); | |
1640 | if (!atts) | |
1641 | return XML_ERROR_NO_MEMORY; | |
1642 | if (n > oldAttsSize) | |
1643 | XmlGetAttributes(enc, attStr, n, atts); | |
1644 | } | |
1645 | appAtts = (const XML_Char **)atts; | |
1646 | for (i = 0; i < n; i++) { | |
1647 | /* add the name and value to the attribute list */ | |
1648 | ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name, | |
1649 | atts[i].name | |
1650 | + XmlNameLength(enc, atts[i].name)); | |
1651 | if (!attId) | |
1652 | return XML_ERROR_NO_MEMORY; | |
1653 | /* detect duplicate attributes */ | |
1654 | if ((attId->name)[-1]) { | |
1655 | if (enc == encoding) | |
1656 | eventPtr = atts[i].name; | |
1657 | return XML_ERROR_DUPLICATE_ATTRIBUTE; | |
1658 | } | |
1659 | (attId->name)[-1] = 1; | |
1660 | appAtts[attIndex++] = attId->name; | |
1661 | if (!atts[i].normalized) { | |
1662 | enum XML_Error result; | |
1663 | int isCdata = 1; | |
1664 | ||
1665 | /* figure out whether declared as other than CDATA */ | |
1666 | if (attId->maybeTokenized) { | |
1667 | int j; | |
1668 | for (j = 0; j < nDefaultAtts; j++) { | |
1669 | if (attId == elementType->defaultAtts[j].id) { | |
1670 | isCdata = elementType->defaultAtts[j].isCdata; | |
1671 | break; | |
1672 | } | |
1673 | } | |
1674 | } | |
1675 | ||
1676 | /* normalize the attribute value */ | |
1677 | result = storeAttributeValue(parser, enc, isCdata, | |
1678 | atts[i].valuePtr, atts[i].valueEnd, | |
1679 | &tempPool); | |
1680 | if (result) | |
1681 | return result; | |
1682 | if (tagNamePtr) { | |
1683 | appAtts[attIndex] = poolStart(&tempPool); | |
1684 | poolFinish(&tempPool); | |
1685 | } | |
1686 | else | |
1687 | poolDiscard(&tempPool); | |
1688 | } | |
1689 | else if (tagNamePtr) { | |
1690 | /* the value did not need normalizing */ | |
1691 | appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr, atts[i].valueEnd); | |
1692 | if (appAtts[attIndex] == 0) | |
1693 | return XML_ERROR_NO_MEMORY; | |
1694 | poolFinish(&tempPool); | |
1695 | } | |
1696 | /* handle prefixed attribute names */ | |
1697 | if (attId->prefix && tagNamePtr) { | |
1698 | if (attId->xmlns) { | |
1699 | /* deal with namespace declarations here */ | |
1700 | if (!addBinding(parser, attId->prefix, attId, appAtts[attIndex], bindingsPtr)) | |
1701 | return XML_ERROR_NO_MEMORY; | |
1702 | --attIndex; | |
1703 | } | |
1704 | else { | |
1705 | /* deal with other prefixed names later */ | |
1706 | attIndex++; | |
1707 | nPrefixes++; | |
1708 | (attId->name)[-1] = 2; | |
1709 | } | |
1710 | } | |
1711 | else | |
1712 | attIndex++; | |
1713 | } | |
1714 | if (tagNamePtr) { | |
1715 | int j; | |
1716 | nSpecifiedAtts = attIndex; | |
1717 | if (elementType->idAtt && (elementType->idAtt->name)[-1]) { | |
1718 | for (i = 0; i < attIndex; i += 2) | |
1719 | if (appAtts[i] == elementType->idAtt->name) { | |
1720 | idAttIndex = i; | |
1721 | break; | |
1722 | } | |
1723 | } | |
1724 | else | |
1725 | idAttIndex = -1; | |
1726 | /* do attribute defaulting */ | |
1727 | for (j = 0; j < nDefaultAtts; j++) { | |
1728 | const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + j; | |
1729 | if (!(da->id->name)[-1] && da->value) { | |
1730 | if (da->id->prefix) { | |
1731 | if (da->id->xmlns) { | |
1732 | if (!addBinding(parser, da->id->prefix, da->id, da->value, bindingsPtr)) | |
1733 | return XML_ERROR_NO_MEMORY; | |
1734 | } | |
1735 | else { | |
1736 | (da->id->name)[-1] = 2; | |
1737 | nPrefixes++; | |
1738 | appAtts[attIndex++] = da->id->name; | |
1739 | appAtts[attIndex++] = da->value; | |
1740 | } | |
1741 | } | |
1742 | else { | |
1743 | (da->id->name)[-1] = 1; | |
1744 | appAtts[attIndex++] = da->id->name; | |
1745 | appAtts[attIndex++] = da->value; | |
1746 | } | |
1747 | } | |
1748 | } | |
1749 | appAtts[attIndex] = 0; | |
1750 | } | |
1751 | i = 0; | |
1752 | if (nPrefixes) { | |
1753 | /* expand prefixed attribute names */ | |
1754 | for (; i < attIndex; i += 2) { | |
1755 | if (appAtts[i][-1] == 2) { | |
1756 | ATTRIBUTE_ID *id; | |
1757 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
1758 | id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, appAtts[i], 0); | |
1759 | if (id->prefix->binding) { | |
1760 | int j; | |
1761 | const BINDING *b = id->prefix->binding; | |
1762 | const XML_Char *s = appAtts[i]; | |
1763 | for (j = 0; j < b->uriLen; j++) { | |
1764 | if (!poolAppendChar(&tempPool, b->uri[j])) | |
1765 | return XML_ERROR_NO_MEMORY; | |
1766 | } | |
1767 | while (*s++ != ':') | |
1768 | ; | |
1769 | do { | |
1770 | if (!poolAppendChar(&tempPool, *s)) | |
1771 | return XML_ERROR_NO_MEMORY; | |
1772 | } while (*s++); | |
1773 | appAtts[i] = poolStart(&tempPool); | |
1774 | poolFinish(&tempPool); | |
1775 | } | |
1776 | if (!--nPrefixes) | |
1777 | break; | |
1778 | } | |
1779 | else | |
1780 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
1781 | } | |
1782 | } | |
1783 | /* clear the flags that say whether attributes were specified */ | |
1784 | for (; i < attIndex; i += 2) | |
1785 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
1786 | if (!tagNamePtr) | |
1787 | return XML_ERROR_NONE; | |
1788 | for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) | |
1789 | binding->attId->name[-1] = 0; | |
1790 | /* expand the element type name */ | |
1791 | if (elementType->prefix) { | |
1792 | binding = elementType->prefix->binding; | |
1793 | if (!binding) | |
1794 | return XML_ERROR_NONE; | |
1795 | localPart = tagNamePtr->str; | |
1796 | while (*localPart++ != XML_T(':')) | |
1797 | ; | |
1798 | } | |
1799 | else if (dtd.defaultPrefix.binding) { | |
1800 | binding = dtd.defaultPrefix.binding; | |
1801 | localPart = tagNamePtr->str; | |
1802 | } | |
1803 | else | |
1804 | return XML_ERROR_NONE; | |
1805 | tagNamePtr->localPart = localPart; | |
1806 | tagNamePtr->uriLen = binding->uriLen; | |
1807 | for (i = 0; localPart[i++];) | |
1808 | ; | |
1809 | n = i + binding->uriLen; | |
1810 | if (n > binding->uriAlloc) { | |
1811 | TAG *p; | |
1812 | XML_Char *uri = malloc((n + EXPAND_SPARE) * sizeof(XML_Char)); | |
1813 | if (!uri) | |
1814 | return XML_ERROR_NO_MEMORY; | |
1815 | binding->uriAlloc = n + EXPAND_SPARE; | |
1816 | memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char)); | |
1817 | for (p = tagStack; p; p = p->parent) | |
1818 | if (p->name.str == binding->uri) | |
1819 | p->name.str = uri; | |
1820 | free(binding->uri); | |
1821 | binding->uri = uri; | |
1822 | } | |
1823 | memcpy(binding->uri + binding->uriLen, localPart, i * sizeof(XML_Char)); | |
1824 | tagNamePtr->str = binding->uri; | |
1825 | return XML_ERROR_NONE; | |
1826 | } | |
1827 | ||
1828 | static | |
1829 | int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr) | |
1830 | { | |
1831 | BINDING *b; | |
1832 | int len; | |
1833 | for (len = 0; uri[len]; len++) | |
1834 | ; | |
1835 | if (namespaceSeparator) | |
1836 | len++; | |
1837 | if (freeBindingList) { | |
1838 | b = freeBindingList; | |
1839 | if (len > b->uriAlloc) { | |
1840 | b->uri = realloc(b->uri, sizeof(XML_Char) * (len + EXPAND_SPARE)); | |
1841 | if (!b->uri) | |
1842 | return 0; | |
1843 | b->uriAlloc = len + EXPAND_SPARE; | |
1844 | } | |
1845 | freeBindingList = b->nextTagBinding; | |
1846 | } | |
1847 | else { | |
1848 | b = malloc(sizeof(BINDING)); | |
1849 | if (!b) | |
1850 | return 0; | |
1851 | b->uri = malloc(sizeof(XML_Char) * (len + EXPAND_SPARE)); | |
1852 | if (!b->uri) { | |
1853 | free(b); | |
1854 | return 0; | |
1855 | } | |
1856 | b->uriAlloc = len + EXPAND_SPARE; | |
1857 | } | |
1858 | b->uriLen = len; | |
1859 | memcpy(b->uri, uri, len * sizeof(XML_Char)); | |
1860 | if (namespaceSeparator) | |
1861 | b->uri[len - 1] = namespaceSeparator; | |
1862 | b->prefix = prefix; | |
1863 | b->attId = attId; | |
1864 | b->prevPrefixBinding = prefix->binding; | |
1865 | if (*uri == XML_T('\0') && prefix == &dtd.defaultPrefix) | |
1866 | prefix->binding = 0; | |
1867 | else | |
1868 | prefix->binding = b; | |
1869 | b->nextTagBinding = *bindingsPtr; | |
1870 | *bindingsPtr = b; | |
1871 | if (startNamespaceDeclHandler) | |
1872 | startNamespaceDeclHandler(handlerArg, prefix->name, | |
1873 | prefix->binding ? uri : 0); | |
1874 | return 1; | |
1875 | } | |
1876 | ||
1877 | /* The idea here is to avoid using stack for each CDATA section when | |
1878 | the whole file is parsed with one call. */ | |
1879 | ||
1880 | static | |
1881 | enum XML_Error cdataSectionProcessor(XML_Parser parser, | |
1882 | const char *start, | |
1883 | const char *end, | |
1884 | const char **endPtr) | |
1885 | { | |
1886 | enum XML_Error result = doCdataSection(parser, encoding, &start, end, endPtr); | |
1887 | if (start) { | |
1888 | processor = contentProcessor; | |
1889 | return contentProcessor(parser, start, end, endPtr); | |
1890 | } | |
1891 | return result; | |
1892 | } | |
1893 | ||
1894 | /* startPtr gets set to non-null is the section is closed, and to null if | |
1895 | the section is not yet closed. */ | |
1896 | ||
1897 | static | |
1898 | enum XML_Error doCdataSection(XML_Parser parser, | |
1899 | const ENCODING *enc, | |
1900 | const char **startPtr, | |
1901 | const char *end, | |
1902 | const char **nextPtr) | |
1903 | { | |
1904 | const char *s = *startPtr; | |
1905 | const char **eventPP; | |
1906 | const char **eventEndPP; | |
1907 | if (enc == encoding) { | |
1908 | eventPP = &eventPtr; | |
1909 | *eventPP = s; | |
1910 | eventEndPP = &eventEndPtr; | |
1911 | } | |
1912 | else { | |
1913 | eventPP = &(openInternalEntities->internalEventPtr); | |
1914 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
1915 | } | |
1916 | *eventPP = s; | |
1917 | *startPtr = 0; | |
1918 | for (;;) { | |
1919 | const char *next; | |
1920 | int tok = XmlCdataSectionTok(enc, s, end, &next); | |
1921 | *eventEndPP = next; | |
1922 | switch (tok) { | |
1923 | case XML_TOK_CDATA_SECT_CLOSE: | |
1924 | if (endCdataSectionHandler) | |
1925 | endCdataSectionHandler(handlerArg); | |
1926 | #if 0 | |
1927 | /* see comment under XML_TOK_CDATA_SECT_OPEN */ | |
1928 | else if (characterDataHandler) | |
1929 | characterDataHandler(handlerArg, dataBuf, 0); | |
1930 | #endif | |
1931 | else if (defaultHandler) | |
1932 | reportDefault(parser, enc, s, next); | |
1933 | *startPtr = next; | |
1934 | return XML_ERROR_NONE; | |
1935 | case XML_TOK_DATA_NEWLINE: | |
1936 | if (characterDataHandler) { | |
1937 | XML_Char c = 0xA; | |
1938 | characterDataHandler(handlerArg, &c, 1); | |
1939 | } | |
1940 | else if (defaultHandler) | |
1941 | reportDefault(parser, enc, s, next); | |
1942 | break; | |
1943 | case XML_TOK_DATA_CHARS: | |
1944 | if (characterDataHandler) { | |
1945 | if (MUST_CONVERT(enc, s)) { | |
1946 | for (;;) { | |
1947 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
1948 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
1949 | *eventEndPP = next; | |
1950 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
1951 | if (s == next) | |
1952 | break; | |
1953 | *eventPP = s; | |
1954 | } | |
1955 | } | |
1956 | else | |
1957 | characterDataHandler(handlerArg, | |
1958 | (XML_Char *)s, | |
1959 | (XML_Char *)next - (XML_Char *)s); | |
1960 | } | |
1961 | else if (defaultHandler) | |
1962 | reportDefault(parser, enc, s, next); | |
1963 | break; | |
1964 | case XML_TOK_INVALID: | |
1965 | *eventPP = next; | |
1966 | return XML_ERROR_INVALID_TOKEN; | |
1967 | case XML_TOK_PARTIAL_CHAR: | |
1968 | if (nextPtr) { | |
1969 | *nextPtr = s; | |
1970 | return XML_ERROR_NONE; | |
1971 | } | |
1972 | return XML_ERROR_PARTIAL_CHAR; | |
1973 | case XML_TOK_PARTIAL: | |
1974 | case XML_TOK_NONE: | |
1975 | if (nextPtr) { | |
1976 | *nextPtr = s; | |
1977 | return XML_ERROR_NONE; | |
1978 | } | |
1979 | return XML_ERROR_UNCLOSED_CDATA_SECTION; | |
1980 | default: | |
1981 | abort(); | |
1982 | } | |
1983 | *eventPP = s = next; | |
1984 | } | |
1985 | /* not reached */ | |
1986 | } | |
1987 | ||
1988 | #ifdef XML_DTD | |
1989 | ||
1990 | /* The idea here is to avoid using stack for each IGNORE section when | |
1991 | the whole file is parsed with one call. */ | |
1992 | ||
1993 | static | |
1994 | enum XML_Error ignoreSectionProcessor(XML_Parser parser, | |
1995 | const char *start, | |
1996 | const char *end, | |
1997 | const char **endPtr) | |
1998 | { | |
1999 | enum XML_Error result = doIgnoreSection(parser, encoding, &start, end, endPtr); | |
2000 | if (start) { | |
2001 | processor = prologProcessor; | |
2002 | return prologProcessor(parser, start, end, endPtr); | |
2003 | } | |
2004 | return result; | |
2005 | } | |
2006 | ||
2007 | /* startPtr gets set to non-null is the section is closed, and to null if | |
2008 | the section is not yet closed. */ | |
2009 | ||
2010 | static | |
2011 | enum XML_Error doIgnoreSection(XML_Parser parser, | |
2012 | const ENCODING *enc, | |
2013 | const char **startPtr, | |
2014 | const char *end, | |
2015 | const char **nextPtr) | |
2016 | { | |
2017 | const char *next; | |
2018 | int tok; | |
2019 | const char *s = *startPtr; | |
2020 | const char **eventPP; | |
2021 | const char **eventEndPP; | |
2022 | if (enc == encoding) { | |
2023 | eventPP = &eventPtr; | |
2024 | *eventPP = s; | |
2025 | eventEndPP = &eventEndPtr; | |
2026 | } | |
2027 | else { | |
2028 | eventPP = &(openInternalEntities->internalEventPtr); | |
2029 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2030 | } | |
2031 | *eventPP = s; | |
2032 | *startPtr = 0; | |
2033 | tok = XmlIgnoreSectionTok(enc, s, end, &next); | |
2034 | *eventEndPP = next; | |
2035 | switch (tok) { | |
2036 | case XML_TOK_IGNORE_SECT: | |
2037 | if (defaultHandler) | |
2038 | reportDefault(parser, enc, s, next); | |
2039 | *startPtr = next; | |
2040 | return XML_ERROR_NONE; | |
2041 | case XML_TOK_INVALID: | |
2042 | *eventPP = next; | |
2043 | return XML_ERROR_INVALID_TOKEN; | |
2044 | case XML_TOK_PARTIAL_CHAR: | |
2045 | if (nextPtr) { | |
2046 | *nextPtr = s; | |
2047 | return XML_ERROR_NONE; | |
2048 | } | |
2049 | return XML_ERROR_PARTIAL_CHAR; | |
2050 | case XML_TOK_PARTIAL: | |
2051 | case XML_TOK_NONE: | |
2052 | if (nextPtr) { | |
2053 | *nextPtr = s; | |
2054 | return XML_ERROR_NONE; | |
2055 | } | |
2056 | return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */ | |
2057 | default: | |
2058 | abort(); | |
2059 | } | |
2060 | /* not reached */ | |
2061 | } | |
2062 | ||
2063 | #endif /* XML_DTD */ | |
2064 | ||
2065 | static enum XML_Error | |
2066 | initializeEncoding(XML_Parser parser) | |
2067 | { | |
2068 | const char *s; | |
2069 | #ifdef XML_UNICODE | |
2070 | char encodingBuf[128]; | |
2071 | if (!protocolEncodingName) | |
2072 | s = 0; | |
2073 | else { | |
2074 | int i; | |
2075 | for (i = 0; protocolEncodingName[i]; i++) { | |
2076 | if (i == sizeof(encodingBuf) - 1 | |
2077 | || (protocolEncodingName[i] & ~0x7f) != 0) { | |
2078 | encodingBuf[0] = '\0'; | |
2079 | break; | |
2080 | } | |
2081 | encodingBuf[i] = (char)protocolEncodingName[i]; | |
2082 | } | |
2083 | encodingBuf[i] = '\0'; | |
2084 | s = encodingBuf; | |
2085 | } | |
2086 | #else | |
2087 | s = protocolEncodingName; | |
2088 | #endif | |
2089 | if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s)) | |
2090 | return XML_ERROR_NONE; | |
2091 | return handleUnknownEncoding(parser, protocolEncodingName); | |
2092 | } | |
2093 | ||
2094 | static enum XML_Error | |
2095 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
2096 | const char *s, const char *next) | |
2097 | { | |
2098 | const char *encodingName = 0; | |
2099 | const ENCODING *newEncoding = 0; | |
2100 | const char *version; | |
2101 | int standalone = -1; | |
2102 | if (!(ns | |
2103 | ? XmlParseXmlDeclNS | |
2104 | : XmlParseXmlDecl)(isGeneralTextEntity, | |
2105 | encoding, | |
2106 | s, | |
2107 | next, | |
2108 | &eventPtr, | |
2109 | &version, | |
2110 | &encodingName, | |
2111 | &newEncoding, | |
2112 | &standalone)) | |
2113 | return XML_ERROR_SYNTAX; | |
2114 | if (!isGeneralTextEntity && standalone == 1) { | |
2115 | dtd.standalone = 1; | |
2116 | #ifdef XML_DTD | |
2117 | if (paramEntityParsing == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE) | |
2118 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; | |
2119 | #endif /* XML_DTD */ | |
2120 | } | |
2121 | if (defaultHandler) | |
2122 | reportDefault(parser, encoding, s, next); | |
2123 | if (!protocolEncodingName) { | |
2124 | if (newEncoding) { | |
2125 | if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) { | |
2126 | eventPtr = encodingName; | |
2127 | return XML_ERROR_INCORRECT_ENCODING; | |
2128 | } | |
2129 | encoding = newEncoding; | |
2130 | } | |
2131 | else if (encodingName) { | |
2132 | enum XML_Error result; | |
2133 | const XML_Char *s = poolStoreString(&tempPool, | |
2134 | encoding, | |
2135 | encodingName, | |
2136 | encodingName | |
2137 | + XmlNameLength(encoding, encodingName)); | |
2138 | if (!s) | |
2139 | return XML_ERROR_NO_MEMORY; | |
2140 | result = handleUnknownEncoding(parser, s); | |
2141 | poolDiscard(&tempPool); | |
2142 | if (result == XML_ERROR_UNKNOWN_ENCODING) | |
2143 | eventPtr = encodingName; | |
2144 | return result; | |
2145 | } | |
2146 | } | |
2147 | return XML_ERROR_NONE; | |
2148 | } | |
2149 | ||
2150 | static enum XML_Error | |
2151 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) | |
2152 | { | |
2153 | if (unknownEncodingHandler) { | |
2154 | XML_Encoding info; | |
2155 | int i; | |
2156 | for (i = 0; i < 256; i++) | |
2157 | info.map[i] = -1; | |
2158 | info.convert = 0; | |
2159 | info.data = 0; | |
2160 | info.release = 0; | |
2161 | if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, &info)) { | |
2162 | ENCODING *enc; | |
2163 | unknownEncodingMem = malloc(XmlSizeOfUnknownEncoding()); | |
2164 | if (!unknownEncodingMem) { | |
2165 | if (info.release) | |
2166 | info.release(info.data); | |
2167 | return XML_ERROR_NO_MEMORY; | |
2168 | } | |
2169 | enc = (ns | |
2170 | ? XmlInitUnknownEncodingNS | |
2171 | : XmlInitUnknownEncoding)(unknownEncodingMem, | |
2172 | info.map, | |
2173 | info.convert, | |
2174 | info.data); | |
2175 | if (enc) { | |
2176 | unknownEncodingData = info.data; | |
2177 | unknownEncodingRelease = info.release; | |
2178 | encoding = enc; | |
2179 | return XML_ERROR_NONE; | |
2180 | } | |
2181 | } | |
2182 | if (info.release) | |
2183 | info.release(info.data); | |
2184 | } | |
2185 | return XML_ERROR_UNKNOWN_ENCODING; | |
2186 | } | |
2187 | ||
2188 | static enum XML_Error | |
2189 | prologInitProcessor(XML_Parser parser, | |
2190 | const char *s, | |
2191 | const char *end, | |
2192 | const char **nextPtr) | |
2193 | { | |
2194 | enum XML_Error result = initializeEncoding(parser); | |
2195 | if (result != XML_ERROR_NONE) | |
2196 | return result; | |
2197 | processor = prologProcessor; | |
2198 | return prologProcessor(parser, s, end, nextPtr); | |
2199 | } | |
2200 | ||
2201 | static enum XML_Error | |
2202 | prologProcessor(XML_Parser parser, | |
2203 | const char *s, | |
2204 | const char *end, | |
2205 | const char **nextPtr) | |
2206 | { | |
2207 | const char *next; | |
2208 | int tok = XmlPrologTok(encoding, s, end, &next); | |
2209 | return doProlog(parser, encoding, s, end, tok, next, nextPtr); | |
2210 | } | |
2211 | ||
2212 | static enum XML_Error | |
2213 | doProlog(XML_Parser parser, | |
2214 | const ENCODING *enc, | |
2215 | const char *s, | |
2216 | const char *end, | |
2217 | int tok, | |
2218 | const char *next, | |
2219 | const char **nextPtr) | |
2220 | { | |
2221 | #ifdef XML_DTD | |
2222 | static const XML_Char externalSubsetName[] = { '#' , '\0' }; | |
2223 | #endif /* XML_DTD */ | |
2224 | ||
2225 | const char **eventPP; | |
2226 | const char **eventEndPP; | |
2227 | if (enc == encoding) { | |
2228 | eventPP = &eventPtr; | |
2229 | eventEndPP = &eventEndPtr; | |
2230 | } | |
2231 | else { | |
2232 | eventPP = &(openInternalEntities->internalEventPtr); | |
2233 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2234 | } | |
2235 | for (;;) { | |
2236 | int role; | |
2237 | *eventPP = s; | |
2238 | *eventEndPP = next; | |
2239 | if (tok <= 0) { | |
2240 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
2241 | *nextPtr = s; | |
2242 | return XML_ERROR_NONE; | |
2243 | } | |
2244 | switch (tok) { | |
2245 | case XML_TOK_INVALID: | |
2246 | *eventPP = next; | |
2247 | return XML_ERROR_INVALID_TOKEN; | |
2248 | case XML_TOK_PARTIAL: | |
2249 | return XML_ERROR_UNCLOSED_TOKEN; | |
2250 | case XML_TOK_PARTIAL_CHAR: | |
2251 | return XML_ERROR_PARTIAL_CHAR; | |
2252 | case XML_TOK_NONE: | |
2253 | #ifdef XML_DTD | |
2254 | if (enc != encoding) | |
2255 | return XML_ERROR_NONE; | |
2256 | if (parentParser) { | |
2257 | if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc) | |
2258 | == XML_ROLE_ERROR) | |
2259 | return XML_ERROR_SYNTAX; | |
2260 | hadExternalDoctype = 0; | |
2261 | return XML_ERROR_NONE; | |
2262 | } | |
2263 | #endif /* XML_DTD */ | |
2264 | return XML_ERROR_NO_ELEMENTS; | |
2265 | default: | |
2266 | tok = -tok; | |
2267 | next = end; | |
2268 | break; | |
2269 | } | |
2270 | } | |
2271 | role = XmlTokenRole(&prologState, tok, s, next, enc); | |
2272 | switch (role) { | |
2273 | case XML_ROLE_XML_DECL: | |
2274 | { | |
2275 | enum XML_Error result = processXmlDecl(parser, 0, s, next); | |
2276 | if (result != XML_ERROR_NONE) | |
2277 | return result; | |
2278 | enc = encoding; | |
2279 | } | |
2280 | break; | |
2281 | case XML_ROLE_DOCTYPE_NAME: | |
2282 | if (startDoctypeDeclHandler) { | |
2283 | const XML_Char *name = poolStoreString(&tempPool, enc, s, next); | |
2284 | if (!name) | |
2285 | return XML_ERROR_NO_MEMORY; | |
2286 | startDoctypeDeclHandler(handlerArg, name); | |
2287 | poolClear(&tempPool); | |
2288 | } | |
2289 | break; | |
2290 | #ifdef XML_DTD | |
2291 | case XML_ROLE_TEXT_DECL: | |
2292 | { | |
2293 | enum XML_Error result = processXmlDecl(parser, 1, s, next); | |
2294 | if (result != XML_ERROR_NONE) | |
2295 | return result; | |
2296 | enc = encoding; | |
2297 | } | |
2298 | break; | |
2299 | #endif /* XML_DTD */ | |
2300 | case XML_ROLE_DOCTYPE_PUBLIC_ID: | |
2301 | #ifdef XML_DTD | |
2302 | declEntity = (ENTITY *)lookup(&dtd.paramEntities, | |
2303 | externalSubsetName, | |
2304 | sizeof(ENTITY)); | |
2305 | if (!declEntity) | |
2306 | return XML_ERROR_NO_MEMORY; | |
2307 | #endif /* XML_DTD */ | |
2308 | /* fall through */ | |
2309 | case XML_ROLE_ENTITY_PUBLIC_ID: | |
2310 | if (!XmlIsPublicId(enc, s, next, eventPP)) | |
2311 | return XML_ERROR_SYNTAX; | |
2312 | if (declEntity) { | |
2313 | XML_Char *tem = poolStoreString(&dtd.pool, | |
2314 | enc, | |
2315 | s + enc->minBytesPerChar, | |
2316 | next - enc->minBytesPerChar); | |
2317 | if (!tem) | |
2318 | return XML_ERROR_NO_MEMORY; | |
2319 | normalizePublicId(tem); | |
2320 | declEntity->publicId = tem; | |
2321 | poolFinish(&dtd.pool); | |
2322 | } | |
2323 | break; | |
2324 | case XML_ROLE_DOCTYPE_CLOSE: | |
2325 | if (dtd.complete && hadExternalDoctype) { | |
2326 | dtd.complete = 0; | |
2327 | #ifdef XML_DTD | |
2328 | if (paramEntityParsing && externalEntityRefHandler) { | |
2329 | ENTITY *entity = (ENTITY *)lookup(&dtd.paramEntities, | |
2330 | externalSubsetName, | |
2331 | 0); | |
2332 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
2333 | 0, | |
2334 | entity->base, | |
2335 | entity->systemId, | |
2336 | entity->publicId)) | |
2337 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
2338 | } | |
2339 | #endif /* XML_DTD */ | |
2340 | if (!dtd.complete | |
2341 | && !dtd.standalone | |
2342 | && notStandaloneHandler | |
2343 | && !notStandaloneHandler(handlerArg)) | |
2344 | return XML_ERROR_NOT_STANDALONE; | |
2345 | } | |
2346 | if (endDoctypeDeclHandler) | |
2347 | endDoctypeDeclHandler(handlerArg); | |
2348 | break; | |
2349 | case XML_ROLE_INSTANCE_START: | |
2350 | processor = contentProcessor; | |
2351 | return contentProcessor(parser, s, end, nextPtr); | |
2352 | case XML_ROLE_ATTLIST_ELEMENT_NAME: | |
2353 | { | |
2354 | const XML_Char *name = poolStoreString(&dtd.pool, enc, s, next); | |
2355 | if (!name) | |
2356 | return XML_ERROR_NO_MEMORY; | |
2357 | declElementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, name, sizeof(ELEMENT_TYPE)); | |
2358 | if (!declElementType) | |
2359 | return XML_ERROR_NO_MEMORY; | |
2360 | if (declElementType->name != name) | |
2361 | poolDiscard(&dtd.pool); | |
2362 | else { | |
2363 | poolFinish(&dtd.pool); | |
2364 | if (!setElementTypePrefix(parser, declElementType)) | |
2365 | return XML_ERROR_NO_MEMORY; | |
2366 | } | |
2367 | break; | |
2368 | } | |
2369 | case XML_ROLE_ATTRIBUTE_NAME: | |
2370 | declAttributeId = getAttributeId(parser, enc, s, next); | |
2371 | if (!declAttributeId) | |
2372 | return XML_ERROR_NO_MEMORY; | |
2373 | declAttributeIsCdata = 0; | |
2374 | declAttributeIsId = 0; | |
2375 | break; | |
2376 | case XML_ROLE_ATTRIBUTE_TYPE_CDATA: | |
2377 | declAttributeIsCdata = 1; | |
2378 | break; | |
2379 | case XML_ROLE_ATTRIBUTE_TYPE_ID: | |
2380 | declAttributeIsId = 1; | |
2381 | break; | |
2382 | case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: | |
2383 | case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: | |
2384 | if (dtd.complete | |
2385 | && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, | |
2386 | declAttributeIsId, 0)) | |
2387 | return XML_ERROR_NO_MEMORY; | |
2388 | break; | |
2389 | case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: | |
2390 | case XML_ROLE_FIXED_ATTRIBUTE_VALUE: | |
2391 | { | |
2392 | const XML_Char *attVal; | |
2393 | enum XML_Error result | |
2394 | = storeAttributeValue(parser, enc, declAttributeIsCdata, | |
2395 | s + enc->minBytesPerChar, | |
2396 | next - enc->minBytesPerChar, | |
2397 | &dtd.pool); | |
2398 | if (result) | |
2399 | return result; | |
2400 | attVal = poolStart(&dtd.pool); | |
2401 | poolFinish(&dtd.pool); | |
2402 | if (dtd.complete | |
2403 | // ID attributes aren't allowed to have a default | |
2404 | && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, 0, attVal)) | |
2405 | return XML_ERROR_NO_MEMORY; | |
2406 | break; | |
2407 | } | |
2408 | case XML_ROLE_ENTITY_VALUE: | |
2409 | { | |
2410 | enum XML_Error result = storeEntityValue(parser, enc, | |
2411 | s + enc->minBytesPerChar, | |
2412 | next - enc->minBytesPerChar); | |
2413 | if (declEntity) { | |
2414 | declEntity->textPtr = poolStart(&dtd.pool); | |
2415 | declEntity->textLen = poolLength(&dtd.pool); | |
2416 | poolFinish(&dtd.pool); | |
2417 | if (internalParsedEntityDeclHandler | |
2418 | // Check it's not a parameter entity | |
2419 | && ((ENTITY *)lookup(&dtd.generalEntities, declEntity->name, 0) | |
2420 | == declEntity)) { | |
2421 | *eventEndPP = s; | |
2422 | internalParsedEntityDeclHandler(handlerArg, | |
2423 | declEntity->name, | |
2424 | declEntity->textPtr, | |
2425 | declEntity->textLen); | |
2426 | } | |
2427 | } | |
2428 | else | |
2429 | poolDiscard(&dtd.pool); | |
2430 | if (result != XML_ERROR_NONE) | |
2431 | return result; | |
2432 | } | |
2433 | break; | |
2434 | case XML_ROLE_DOCTYPE_SYSTEM_ID: | |
2435 | if (!dtd.standalone | |
2436 | #ifdef XML_DTD | |
2437 | && !paramEntityParsing | |
2438 | #endif /* XML_DTD */ | |
2439 | && notStandaloneHandler | |
2440 | && !notStandaloneHandler(handlerArg)) | |
2441 | return XML_ERROR_NOT_STANDALONE; | |
2442 | hadExternalDoctype = 1; | |
2443 | #ifndef XML_DTD | |
2444 | break; | |
2445 | #else /* XML_DTD */ | |
2446 | if (!declEntity) { | |
2447 | declEntity = (ENTITY *)lookup(&dtd.paramEntities, | |
2448 | externalSubsetName, | |
2449 | sizeof(ENTITY)); | |
2450 | if (!declEntity) | |
2451 | return XML_ERROR_NO_MEMORY; | |
2452 | } | |
2453 | /* fall through */ | |
2454 | #endif /* XML_DTD */ | |
2455 | case XML_ROLE_ENTITY_SYSTEM_ID: | |
2456 | if (declEntity) { | |
2457 | declEntity->systemId = poolStoreString(&dtd.pool, enc, | |
2458 | s + enc->minBytesPerChar, | |
2459 | next - enc->minBytesPerChar); | |
2460 | if (!declEntity->systemId) | |
2461 | return XML_ERROR_NO_MEMORY; | |
2462 | declEntity->base = curBase; | |
2463 | poolFinish(&dtd.pool); | |
2464 | } | |
2465 | break; | |
2466 | case XML_ROLE_ENTITY_NOTATION_NAME: | |
2467 | if (declEntity) { | |
2468 | declEntity->notation = poolStoreString(&dtd.pool, enc, s, next); | |
2469 | if (!declEntity->notation) | |
2470 | return XML_ERROR_NO_MEMORY; | |
2471 | poolFinish(&dtd.pool); | |
2472 | if (unparsedEntityDeclHandler) { | |
2473 | *eventEndPP = s; | |
2474 | unparsedEntityDeclHandler(handlerArg, | |
2475 | declEntity->name, | |
2476 | declEntity->base, | |
2477 | declEntity->systemId, | |
2478 | declEntity->publicId, | |
2479 | declEntity->notation); | |
2480 | } | |
2481 | ||
2482 | } | |
2483 | break; | |
2484 | case XML_ROLE_EXTERNAL_GENERAL_ENTITY_NO_NOTATION: | |
2485 | if (declEntity && externalParsedEntityDeclHandler) { | |
2486 | *eventEndPP = s; | |
2487 | externalParsedEntityDeclHandler(handlerArg, | |
2488 | declEntity->name, | |
2489 | declEntity->base, | |
2490 | declEntity->systemId, | |
2491 | declEntity->publicId); | |
2492 | } | |
2493 | break; | |
2494 | case XML_ROLE_GENERAL_ENTITY_NAME: | |
2495 | { | |
2496 | const XML_Char *name; | |
2497 | if (XmlPredefinedEntityName(enc, s, next)) { | |
2498 | declEntity = 0; | |
2499 | break; | |
2500 | } | |
2501 | name = poolStoreString(&dtd.pool, enc, s, next); | |
2502 | if (!name) | |
2503 | return XML_ERROR_NO_MEMORY; | |
2504 | if (dtd.complete) { | |
2505 | declEntity = (ENTITY *)lookup(&dtd.generalEntities, name, sizeof(ENTITY)); | |
2506 | if (!declEntity) | |
2507 | return XML_ERROR_NO_MEMORY; | |
2508 | if (declEntity->name != name) { | |
2509 | poolDiscard(&dtd.pool); | |
2510 | declEntity = 0; | |
2511 | } | |
2512 | else | |
2513 | poolFinish(&dtd.pool); | |
2514 | } | |
2515 | else { | |
2516 | poolDiscard(&dtd.pool); | |
2517 | declEntity = 0; | |
2518 | } | |
2519 | } | |
2520 | break; | |
2521 | case XML_ROLE_PARAM_ENTITY_NAME: | |
2522 | #ifdef XML_DTD | |
2523 | if (dtd.complete) { | |
2524 | const XML_Char *name = poolStoreString(&dtd.pool, enc, s, next); | |
2525 | if (!name) | |
2526 | return XML_ERROR_NO_MEMORY; | |
2527 | declEntity = (ENTITY *)lookup(&dtd.paramEntities, name, sizeof(ENTITY)); | |
2528 | if (!declEntity) | |
2529 | return XML_ERROR_NO_MEMORY; | |
2530 | if (declEntity->name != name) { | |
2531 | poolDiscard(&dtd.pool); | |
2532 | declEntity = 0; | |
2533 | } | |
2534 | else | |
2535 | poolFinish(&dtd.pool); | |
2536 | } | |
2537 | #else /* not XML_DTD */ | |
2538 | declEntity = 0; | |
2539 | #endif /* not XML_DTD */ | |
2540 | break; | |
2541 | case XML_ROLE_NOTATION_NAME: | |
2542 | declNotationPublicId = 0; | |
2543 | declNotationName = 0; | |
2544 | if (notationDeclHandler) { | |
2545 | declNotationName = poolStoreString(&tempPool, enc, s, next); | |
2546 | if (!declNotationName) | |
2547 | return XML_ERROR_NO_MEMORY; | |
2548 | poolFinish(&tempPool); | |
2549 | } | |
2550 | break; | |
2551 | case XML_ROLE_NOTATION_PUBLIC_ID: | |
2552 | if (!XmlIsPublicId(enc, s, next, eventPP)) | |
2553 | return XML_ERROR_SYNTAX; | |
2554 | if (declNotationName) { | |
2555 | XML_Char *tem = poolStoreString(&tempPool, | |
2556 | enc, | |
2557 | s + enc->minBytesPerChar, | |
2558 | next - enc->minBytesPerChar); | |
2559 | if (!tem) | |
2560 | return XML_ERROR_NO_MEMORY; | |
2561 | normalizePublicId(tem); | |
2562 | declNotationPublicId = tem; | |
2563 | poolFinish(&tempPool); | |
2564 | } | |
2565 | break; | |
2566 | case XML_ROLE_NOTATION_SYSTEM_ID: | |
2567 | if (declNotationName && notationDeclHandler) { | |
2568 | const XML_Char *systemId | |
2569 | = poolStoreString(&tempPool, enc, | |
2570 | s + enc->minBytesPerChar, | |
2571 | next - enc->minBytesPerChar); | |
2572 | if (!systemId) | |
2573 | return XML_ERROR_NO_MEMORY; | |
2574 | *eventEndPP = s; | |
2575 | notationDeclHandler(handlerArg, | |
2576 | declNotationName, | |
2577 | curBase, | |
2578 | systemId, | |
2579 | declNotationPublicId); | |
2580 | } | |
2581 | poolClear(&tempPool); | |
2582 | break; | |
2583 | case XML_ROLE_NOTATION_NO_SYSTEM_ID: | |
2584 | if (declNotationPublicId && notationDeclHandler) { | |
2585 | *eventEndPP = s; | |
2586 | notationDeclHandler(handlerArg, | |
2587 | declNotationName, | |
2588 | curBase, | |
2589 | 0, | |
2590 | declNotationPublicId); | |
2591 | } | |
2592 | poolClear(&tempPool); | |
2593 | break; | |
2594 | case XML_ROLE_ERROR: | |
2595 | switch (tok) { | |
2596 | case XML_TOK_PARAM_ENTITY_REF: | |
2597 | return XML_ERROR_PARAM_ENTITY_REF; | |
2598 | case XML_TOK_XML_DECL: | |
2599 | return XML_ERROR_MISPLACED_XML_PI; | |
2600 | default: | |
2601 | return XML_ERROR_SYNTAX; | |
2602 | } | |
2603 | #ifdef XML_DTD | |
2604 | case XML_ROLE_IGNORE_SECT: | |
2605 | { | |
2606 | enum XML_Error result; | |
2607 | if (defaultHandler) | |
2608 | reportDefault(parser, enc, s, next); | |
2609 | result = doIgnoreSection(parser, enc, &next, end, nextPtr); | |
2610 | if (!next) { | |
2611 | processor = ignoreSectionProcessor; | |
2612 | return result; | |
2613 | } | |
2614 | } | |
2615 | break; | |
2616 | #endif /* XML_DTD */ | |
2617 | case XML_ROLE_GROUP_OPEN: | |
2618 | if (prologState.level >= groupSize) { | |
2619 | if (groupSize) | |
2620 | groupConnector = realloc(groupConnector, groupSize *= 2); | |
2621 | else | |
2622 | groupConnector = malloc(groupSize = 32); | |
2623 | if (!groupConnector) | |
2624 | return XML_ERROR_NO_MEMORY; | |
2625 | } | |
2626 | groupConnector[prologState.level] = 0; | |
2627 | break; | |
2628 | case XML_ROLE_GROUP_SEQUENCE: | |
2629 | if (groupConnector[prologState.level] == '|') | |
2630 | return XML_ERROR_SYNTAX; | |
2631 | groupConnector[prologState.level] = ','; | |
2632 | break; | |
2633 | case XML_ROLE_GROUP_CHOICE: | |
2634 | if (groupConnector[prologState.level] == ',') | |
2635 | return XML_ERROR_SYNTAX; | |
2636 | groupConnector[prologState.level] = '|'; | |
2637 | break; | |
2638 | case XML_ROLE_PARAM_ENTITY_REF: | |
2639 | #ifdef XML_DTD | |
2640 | case XML_ROLE_INNER_PARAM_ENTITY_REF: | |
2641 | if (paramEntityParsing | |
2642 | && (dtd.complete || role == XML_ROLE_INNER_PARAM_ENTITY_REF)) { | |
2643 | const XML_Char *name; | |
2644 | ENTITY *entity; | |
2645 | name = poolStoreString(&dtd.pool, enc, | |
2646 | s + enc->minBytesPerChar, | |
2647 | next - enc->minBytesPerChar); | |
2648 | if (!name) | |
2649 | return XML_ERROR_NO_MEMORY; | |
2650 | entity = (ENTITY *)lookup(&dtd.paramEntities, name, 0); | |
2651 | poolDiscard(&dtd.pool); | |
2652 | if (!entity) { | |
2653 | /* FIXME what to do if !dtd.complete? */ | |
2654 | return XML_ERROR_UNDEFINED_ENTITY; | |
2655 | } | |
2656 | if (entity->open) | |
2657 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
2658 | if (entity->textPtr) { | |
2659 | enum XML_Error result; | |
2660 | result = processInternalParamEntity(parser, entity); | |
2661 | if (result != XML_ERROR_NONE) | |
2662 | return result; | |
2663 | break; | |
2664 | } | |
2665 | if (role == XML_ROLE_INNER_PARAM_ENTITY_REF) | |
2666 | return XML_ERROR_PARAM_ENTITY_REF; | |
2667 | if (externalEntityRefHandler) { | |
2668 | dtd.complete = 0; | |
2669 | entity->open = 1; | |
2670 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
2671 | 0, | |
2672 | entity->base, | |
2673 | entity->systemId, | |
2674 | entity->publicId)) { | |
2675 | entity->open = 0; | |
2676 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
2677 | } | |
2678 | entity->open = 0; | |
2679 | if (dtd.complete) | |
2680 | break; | |
2681 | } | |
2682 | } | |
2683 | #endif /* XML_DTD */ | |
2684 | if (!dtd.standalone | |
2685 | && notStandaloneHandler | |
2686 | && !notStandaloneHandler(handlerArg)) | |
2687 | return XML_ERROR_NOT_STANDALONE; | |
2688 | dtd.complete = 0; | |
2689 | if (defaultHandler) | |
2690 | reportDefault(parser, enc, s, next); | |
2691 | break; | |
2692 | case XML_ROLE_NONE: | |
2693 | switch (tok) { | |
2694 | case XML_TOK_PI: | |
2695 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
2696 | return XML_ERROR_NO_MEMORY; | |
2697 | break; | |
2698 | case XML_TOK_COMMENT: | |
2699 | if (!reportComment(parser, enc, s, next)) | |
2700 | return XML_ERROR_NO_MEMORY; | |
2701 | break; | |
2702 | } | |
2703 | break; | |
2704 | } | |
2705 | if (defaultHandler) { | |
2706 | switch (tok) { | |
2707 | case XML_TOK_PI: | |
2708 | case XML_TOK_COMMENT: | |
2709 | case XML_TOK_BOM: | |
2710 | case XML_TOK_XML_DECL: | |
2711 | #ifdef XML_DTD | |
2712 | case XML_TOK_IGNORE_SECT: | |
2713 | #endif /* XML_DTD */ | |
2714 | case XML_TOK_PARAM_ENTITY_REF: | |
2715 | break; | |
2716 | default: | |
2717 | #ifdef XML_DTD | |
2718 | if (role != XML_ROLE_IGNORE_SECT) | |
2719 | #endif /* XML_DTD */ | |
2720 | reportDefault(parser, enc, s, next); | |
2721 | } | |
2722 | } | |
2723 | s = next; | |
2724 | tok = XmlPrologTok(enc, s, end, &next); | |
2725 | } | |
2726 | /* not reached */ | |
2727 | } | |
2728 | ||
2729 | static | |
2730 | enum XML_Error epilogProcessor(XML_Parser parser, | |
2731 | const char *s, | |
2732 | const char *end, | |
2733 | const char **nextPtr) | |
2734 | { | |
2735 | processor = epilogProcessor; | |
2736 | eventPtr = s; | |
2737 | for (;;) { | |
2738 | const char *next; | |
2739 | int tok = XmlPrologTok(encoding, s, end, &next); | |
2740 | eventEndPtr = next; | |
2741 | switch (tok) { | |
2742 | case -XML_TOK_PROLOG_S: | |
2743 | if (defaultHandler) { | |
2744 | eventEndPtr = end; | |
2745 | reportDefault(parser, encoding, s, end); | |
2746 | } | |
2747 | /* fall through */ | |
2748 | case XML_TOK_NONE: | |
2749 | if (nextPtr) | |
2750 | *nextPtr = end; | |
2751 | return XML_ERROR_NONE; | |
2752 | case XML_TOK_PROLOG_S: | |
2753 | if (defaultHandler) | |
2754 | reportDefault(parser, encoding, s, next); | |
2755 | break; | |
2756 | case XML_TOK_PI: | |
2757 | if (!reportProcessingInstruction(parser, encoding, s, next)) | |
2758 | return XML_ERROR_NO_MEMORY; | |
2759 | break; | |
2760 | case XML_TOK_COMMENT: | |
2761 | if (!reportComment(parser, encoding, s, next)) | |
2762 | return XML_ERROR_NO_MEMORY; | |
2763 | break; | |
2764 | case XML_TOK_INVALID: | |
2765 | eventPtr = next; | |
2766 | return XML_ERROR_INVALID_TOKEN; | |
2767 | case XML_TOK_PARTIAL: | |
2768 | if (nextPtr) { | |
2769 | *nextPtr = s; | |
2770 | return XML_ERROR_NONE; | |
2771 | } | |
2772 | return XML_ERROR_UNCLOSED_TOKEN; | |
2773 | case XML_TOK_PARTIAL_CHAR: | |
2774 | if (nextPtr) { | |
2775 | *nextPtr = s; | |
2776 | return XML_ERROR_NONE; | |
2777 | } | |
2778 | return XML_ERROR_PARTIAL_CHAR; | |
2779 | default: | |
2780 | return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; | |
2781 | } | |
2782 | eventPtr = s = next; | |
2783 | } | |
2784 | } | |
2785 | ||
2786 | #ifdef XML_DTD | |
2787 | ||
2788 | static enum XML_Error | |
2789 | processInternalParamEntity(XML_Parser parser, ENTITY *entity) | |
2790 | { | |
2791 | const char *s, *end, *next; | |
2792 | int tok; | |
2793 | enum XML_Error result; | |
2794 | OPEN_INTERNAL_ENTITY openEntity; | |
2795 | entity->open = 1; | |
2796 | openEntity.next = openInternalEntities; | |
2797 | openInternalEntities = &openEntity; | |
2798 | openEntity.entity = entity; | |
2799 | openEntity.internalEventPtr = 0; | |
2800 | openEntity.internalEventEndPtr = 0; | |
2801 | s = (char *)entity->textPtr; | |
2802 | end = (char *)(entity->textPtr + entity->textLen); | |
2803 | tok = XmlPrologTok(internalEncoding, s, end, &next); | |
2804 | result = doProlog(parser, internalEncoding, s, end, tok, next, 0); | |
2805 | entity->open = 0; | |
2806 | openInternalEntities = openEntity.next; | |
2807 | return result; | |
2808 | } | |
2809 | ||
2810 | #endif /* XML_DTD */ | |
2811 | ||
2812 | static | |
2813 | enum XML_Error errorProcessor(XML_Parser parser, | |
2814 | const char *s, | |
2815 | const char *end, | |
2816 | const char **nextPtr) | |
2817 | { | |
2818 | return errorCode; | |
2819 | } | |
2820 | ||
2821 | static enum XML_Error | |
2822 | storeAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, | |
2823 | const char *ptr, const char *end, | |
2824 | STRING_POOL *pool) | |
2825 | { | |
2826 | enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr, end, pool); | |
2827 | if (result) | |
2828 | return result; | |
2829 | if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) | |
2830 | poolChop(pool); | |
2831 | if (!poolAppendChar(pool, XML_T('\0'))) | |
2832 | return XML_ERROR_NO_MEMORY; | |
2833 | return XML_ERROR_NONE; | |
2834 | } | |
2835 | ||
2836 | static enum XML_Error | |
2837 | appendAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, | |
2838 | const char *ptr, const char *end, | |
2839 | STRING_POOL *pool) | |
2840 | { | |
2841 | for (;;) { | |
2842 | const char *next; | |
2843 | int tok = XmlAttributeValueTok(enc, ptr, end, &next); | |
2844 | switch (tok) { | |
2845 | case XML_TOK_NONE: | |
2846 | return XML_ERROR_NONE; | |
2847 | case XML_TOK_INVALID: | |
2848 | if (enc == encoding) | |
2849 | eventPtr = next; | |
2850 | return XML_ERROR_INVALID_TOKEN; | |
2851 | case XML_TOK_PARTIAL: | |
2852 | if (enc == encoding) | |
2853 | eventPtr = ptr; | |
2854 | return XML_ERROR_INVALID_TOKEN; | |
2855 | case XML_TOK_CHAR_REF: | |
2856 | { | |
2857 | XML_Char buf[XML_ENCODE_MAX]; | |
2858 | int i; | |
2859 | int n = XmlCharRefNumber(enc, ptr); | |
2860 | if (n < 0) { | |
2861 | if (enc == encoding) | |
2862 | eventPtr = ptr; | |
2863 | return XML_ERROR_BAD_CHAR_REF; | |
2864 | } | |
2865 | if (!isCdata | |
2866 | && n == 0x20 /* space */ | |
2867 | && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
2868 | break; | |
2869 | n = XmlEncode(n, (ICHAR *)buf); | |
2870 | if (!n) { | |
2871 | if (enc == encoding) | |
2872 | eventPtr = ptr; | |
2873 | return XML_ERROR_BAD_CHAR_REF; | |
2874 | } | |
2875 | for (i = 0; i < n; i++) { | |
2876 | if (!poolAppendChar(pool, buf[i])) | |
2877 | return XML_ERROR_NO_MEMORY; | |
2878 | } | |
2879 | } | |
2880 | break; | |
2881 | case XML_TOK_DATA_CHARS: | |
2882 | if (!poolAppend(pool, enc, ptr, next)) | |
2883 | return XML_ERROR_NO_MEMORY; | |
2884 | break; | |
2885 | break; | |
2886 | case XML_TOK_TRAILING_CR: | |
2887 | next = ptr + enc->minBytesPerChar; | |
2888 | /* fall through */ | |
2889 | case XML_TOK_ATTRIBUTE_VALUE_S: | |
2890 | case XML_TOK_DATA_NEWLINE: | |
2891 | if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
2892 | break; | |
2893 | if (!poolAppendChar(pool, 0x20)) | |
2894 | return XML_ERROR_NO_MEMORY; | |
2895 | break; | |
2896 | case XML_TOK_ENTITY_REF: | |
2897 | { | |
2898 | const XML_Char *name; | |
2899 | ENTITY *entity; | |
2900 | XML_Char ch = XmlPredefinedEntityName(enc, | |
2901 | ptr + enc->minBytesPerChar, | |
2902 | next - enc->minBytesPerChar); | |
2903 | if (ch) { | |
2904 | if (!poolAppendChar(pool, ch)) | |
2905 | return XML_ERROR_NO_MEMORY; | |
2906 | break; | |
2907 | } | |
2908 | name = poolStoreString(&temp2Pool, enc, | |
2909 | ptr + enc->minBytesPerChar, | |
2910 | next - enc->minBytesPerChar); | |
2911 | if (!name) | |
2912 | return XML_ERROR_NO_MEMORY; | |
2913 | entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); | |
2914 | poolDiscard(&temp2Pool); | |
2915 | if (!entity) { | |
2916 | if (dtd.complete) { | |
2917 | if (enc == encoding) | |
2918 | eventPtr = ptr; | |
2919 | return XML_ERROR_UNDEFINED_ENTITY; | |
2920 | } | |
2921 | } | |
2922 | else if (entity->open) { | |
2923 | if (enc == encoding) | |
2924 | eventPtr = ptr; | |
2925 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
2926 | } | |
2927 | else if (entity->notation) { | |
2928 | if (enc == encoding) | |
2929 | eventPtr = ptr; | |
2930 | return XML_ERROR_BINARY_ENTITY_REF; | |
2931 | } | |
2932 | else if (!entity->textPtr) { | |
2933 | if (enc == encoding) | |
2934 | eventPtr = ptr; | |
2935 | return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; | |
2936 | } | |
2937 | else { | |
2938 | enum XML_Error result; | |
2939 | const XML_Char *textEnd = entity->textPtr + entity->textLen; | |
2940 | entity->open = 1; | |
2941 | result = appendAttributeValue(parser, internalEncoding, isCdata, (char *)entity->textPtr, (char *)textEnd, pool); | |
2942 | entity->open = 0; | |
2943 | if (result) | |
2944 | return result; | |
2945 | } | |
2946 | } | |
2947 | break; | |
2948 | default: | |
2949 | abort(); | |
2950 | } | |
2951 | ptr = next; | |
2952 | } | |
2953 | /* not reached */ | |
2954 | } | |
2955 | ||
2956 | static | |
2957 | enum XML_Error storeEntityValue(XML_Parser parser, | |
2958 | const ENCODING *enc, | |
2959 | const char *entityTextPtr, | |
2960 | const char *entityTextEnd) | |
2961 | { | |
2962 | STRING_POOL *pool = &(dtd.pool); | |
2963 | for (;;) { | |
2964 | const char *next; | |
2965 | int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next); | |
2966 | switch (tok) { | |
2967 | case XML_TOK_PARAM_ENTITY_REF: | |
2968 | #ifdef XML_DTD | |
2969 | if (parentParser || enc != encoding) { | |
2970 | enum XML_Error result; | |
2971 | const XML_Char *name; | |
2972 | ENTITY *entity; | |
2973 | name = poolStoreString(&tempPool, enc, | |
2974 | entityTextPtr + enc->minBytesPerChar, | |
2975 | next - enc->minBytesPerChar); | |
2976 | if (!name) | |
2977 | return XML_ERROR_NO_MEMORY; | |
2978 | entity = (ENTITY *)lookup(&dtd.paramEntities, name, 0); | |
2979 | poolDiscard(&tempPool); | |
2980 | if (!entity) { | |
2981 | if (enc == encoding) | |
2982 | eventPtr = entityTextPtr; | |
2983 | return XML_ERROR_UNDEFINED_ENTITY; | |
2984 | } | |
2985 | if (entity->open) { | |
2986 | if (enc == encoding) | |
2987 | eventPtr = entityTextPtr; | |
2988 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
2989 | } | |
2990 | if (entity->systemId) { | |
2991 | if (enc == encoding) | |
2992 | eventPtr = entityTextPtr; | |
2993 | return XML_ERROR_PARAM_ENTITY_REF; | |
2994 | } | |
2995 | entity->open = 1; | |
2996 | result = storeEntityValue(parser, | |
2997 | internalEncoding, | |
2998 | (char *)entity->textPtr, | |
2999 | (char *)(entity->textPtr + entity->textLen)); | |
3000 | entity->open = 0; | |
3001 | if (result) | |
3002 | return result; | |
3003 | break; | |
3004 | } | |
3005 | #endif /* XML_DTD */ | |
3006 | eventPtr = entityTextPtr; | |
3007 | return XML_ERROR_SYNTAX; | |
3008 | case XML_TOK_NONE: | |
3009 | return XML_ERROR_NONE; | |
3010 | case XML_TOK_ENTITY_REF: | |
3011 | case XML_TOK_DATA_CHARS: | |
3012 | if (!poolAppend(pool, enc, entityTextPtr, next)) | |
3013 | return XML_ERROR_NO_MEMORY; | |
3014 | break; | |
3015 | case XML_TOK_TRAILING_CR: | |
3016 | next = entityTextPtr + enc->minBytesPerChar; | |
3017 | /* fall through */ | |
3018 | case XML_TOK_DATA_NEWLINE: | |
3019 | if (pool->end == pool->ptr && !poolGrow(pool)) | |
3020 | return XML_ERROR_NO_MEMORY; | |
3021 | *(pool->ptr)++ = 0xA; | |
3022 | break; | |
3023 | case XML_TOK_CHAR_REF: | |
3024 | { | |
3025 | XML_Char buf[XML_ENCODE_MAX]; | |
3026 | int i; | |
3027 | int n = XmlCharRefNumber(enc, entityTextPtr); | |
3028 | if (n < 0) { | |
3029 | if (enc == encoding) | |
3030 | eventPtr = entityTextPtr; | |
3031 | return XML_ERROR_BAD_CHAR_REF; | |
3032 | } | |
3033 | n = XmlEncode(n, (ICHAR *)buf); | |
3034 | if (!n) { | |
3035 | if (enc == encoding) | |
3036 | eventPtr = entityTextPtr; | |
3037 | return XML_ERROR_BAD_CHAR_REF; | |
3038 | } | |
3039 | for (i = 0; i < n; i++) { | |
3040 | if (pool->end == pool->ptr && !poolGrow(pool)) | |
3041 | return XML_ERROR_NO_MEMORY; | |
3042 | *(pool->ptr)++ = buf[i]; | |
3043 | } | |
3044 | } | |
3045 | break; | |
3046 | case XML_TOK_PARTIAL: | |
3047 | if (enc == encoding) | |
3048 | eventPtr = entityTextPtr; | |
3049 | return XML_ERROR_INVALID_TOKEN; | |
3050 | case XML_TOK_INVALID: | |
3051 | if (enc == encoding) | |
3052 | eventPtr = next; | |
3053 | return XML_ERROR_INVALID_TOKEN; | |
3054 | default: | |
3055 | abort(); | |
3056 | } | |
3057 | entityTextPtr = next; | |
3058 | } | |
3059 | /* not reached */ | |
3060 | } | |
3061 | ||
3062 | static void | |
3063 | normalizeLines(XML_Char *s) | |
3064 | { | |
3065 | XML_Char *p; | |
3066 | for (;; s++) { | |
3067 | if (*s == XML_T('\0')) | |
3068 | return; | |
3069 | if (*s == 0xD) | |
3070 | break; | |
3071 | } | |
3072 | p = s; | |
3073 | do { | |
3074 | if (*s == 0xD) { | |
3075 | *p++ = 0xA; | |
3076 | if (*++s == 0xA) | |
3077 | s++; | |
3078 | } | |
3079 | else | |
3080 | *p++ = *s++; | |
3081 | } while (*s); | |
3082 | *p = XML_T('\0'); | |
3083 | } | |
3084 | ||
3085 | static int | |
3086 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
3087 | { | |
3088 | const XML_Char *target; | |
3089 | XML_Char *data; | |
3090 | const char *tem; | |
3091 | if (!processingInstructionHandler) { | |
3092 | if (defaultHandler) | |
3093 | reportDefault(parser, enc, start, end); | |
3094 | return 1; | |
3095 | } | |
3096 | start += enc->minBytesPerChar * 2; | |
3097 | tem = start + XmlNameLength(enc, start); | |
3098 | target = poolStoreString(&tempPool, enc, start, tem); | |
3099 | if (!target) | |
3100 | return 0; | |
3101 | poolFinish(&tempPool); | |
3102 | data = poolStoreString(&tempPool, enc, | |
3103 | XmlSkipS(enc, tem), | |
3104 | end - enc->minBytesPerChar*2); | |
3105 | if (!data) | |
3106 | return 0; | |
3107 | normalizeLines(data); | |
3108 | processingInstructionHandler(handlerArg, target, data); | |
3109 | poolClear(&tempPool); | |
3110 | return 1; | |
3111 | } | |
3112 | ||
3113 | static int | |
3114 | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
3115 | { | |
3116 | XML_Char *data; | |
3117 | if (!commentHandler) { | |
3118 | if (defaultHandler) | |
3119 | reportDefault(parser, enc, start, end); | |
3120 | return 1; | |
3121 | } | |
3122 | data = poolStoreString(&tempPool, | |
3123 | enc, | |
3124 | start + enc->minBytesPerChar * 4, | |
3125 | end - enc->minBytesPerChar * 3); | |
3126 | if (!data) | |
3127 | return 0; | |
3128 | normalizeLines(data); | |
3129 | commentHandler(handlerArg, data); | |
3130 | poolClear(&tempPool); | |
3131 | return 1; | |
3132 | } | |
3133 | ||
3134 | static void | |
3135 | reportDefault(XML_Parser parser, const ENCODING *enc, const char *s, const char *end) | |
3136 | { | |
3137 | if (MUST_CONVERT(enc, s)) { | |
3138 | const char **eventPP; | |
3139 | const char **eventEndPP; | |
3140 | if (enc == encoding) { | |
3141 | eventPP = &eventPtr; | |
3142 | eventEndPP = &eventEndPtr; | |
3143 | } | |
3144 | else { | |
3145 | eventPP = &(openInternalEntities->internalEventPtr); | |
3146 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
3147 | } | |
3148 | do { | |
3149 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
3150 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
3151 | *eventEndPP = s; | |
3152 | defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
3153 | *eventPP = s; | |
3154 | } while (s != end); | |
3155 | } | |
3156 | else | |
3157 | defaultHandler(handlerArg, (XML_Char *)s, (XML_Char *)end - (XML_Char *)s); | |
3158 | } | |
3159 | ||
3160 | ||
3161 | static int | |
3162 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, int isCdata, int isId, const XML_Char *value) | |
3163 | { | |
3164 | DEFAULT_ATTRIBUTE *att; | |
3165 | if (value || isId) { | |
3166 | /* The handling of default attributes gets messed up if we have | |
3167 | a default which duplicates a non-default. */ | |
3168 | int i; | |
3169 | for (i = 0; i < type->nDefaultAtts; i++) | |
3170 | if (attId == type->defaultAtts[i].id) | |
3171 | return 1; | |
3172 | if (isId && !type->idAtt && !attId->xmlns) | |
3173 | type->idAtt = attId; | |
3174 | } | |
3175 | if (type->nDefaultAtts == type->allocDefaultAtts) { | |
3176 | if (type->allocDefaultAtts == 0) { | |
3177 | type->allocDefaultAtts = 8; | |
3178 | type->defaultAtts = malloc(type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); | |
3179 | } | |
3180 | else { | |
3181 | type->allocDefaultAtts *= 2; | |
3182 | type->defaultAtts = realloc(type->defaultAtts, | |
3183 | type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); | |
3184 | } | |
3185 | if (!type->defaultAtts) | |
3186 | return 0; | |
3187 | } | |
3188 | att = type->defaultAtts + type->nDefaultAtts; | |
3189 | att->id = attId; | |
3190 | att->value = value; | |
3191 | att->isCdata = isCdata; | |
3192 | if (!isCdata) | |
3193 | attId->maybeTokenized = 1; | |
3194 | type->nDefaultAtts += 1; | |
3195 | return 1; | |
3196 | } | |
3197 | ||
3198 | static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) | |
3199 | { | |
3200 | const XML_Char *name; | |
3201 | for (name = elementType->name; *name; name++) { | |
3202 | if (*name == XML_T(':')) { | |
3203 | PREFIX *prefix; | |
3204 | const XML_Char *s; | |
3205 | for (s = elementType->name; s != name; s++) { | |
3206 | if (!poolAppendChar(&dtd.pool, *s)) | |
3207 | return 0; | |
3208 | } | |
3209 | if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
3210 | return 0; | |
3211 | prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); | |
3212 | if (!prefix) | |
3213 | return 0; | |
3214 | if (prefix->name == poolStart(&dtd.pool)) | |
3215 | poolFinish(&dtd.pool); | |
3216 | else | |
3217 | poolDiscard(&dtd.pool); | |
3218 | elementType->prefix = prefix; | |
3219 | ||
3220 | } | |
3221 | } | |
3222 | return 1; | |
3223 | } | |
3224 | ||
3225 | static ATTRIBUTE_ID * | |
3226 | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
3227 | { | |
3228 | ATTRIBUTE_ID *id; | |
3229 | const XML_Char *name; | |
3230 | if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
3231 | return 0; | |
3232 | name = poolStoreString(&dtd.pool, enc, start, end); | |
3233 | if (!name) | |
3234 | return 0; | |
3235 | ++name; | |
3236 | id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, name, sizeof(ATTRIBUTE_ID)); | |
3237 | if (!id) | |
3238 | return 0; | |
3239 | if (id->name != name) | |
3240 | poolDiscard(&dtd.pool); | |
3241 | else { | |
3242 | poolFinish(&dtd.pool); | |
3243 | if (!ns) | |
3244 | ; | |
3245 | else if (name[0] == 'x' | |
3246 | && name[1] == 'm' | |
3247 | && name[2] == 'l' | |
3248 | && name[3] == 'n' | |
3249 | && name[4] == 's' | |
3250 | && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) { | |
3251 | if (name[5] == '\0') | |
3252 | id->prefix = &dtd.defaultPrefix; | |
3253 | else | |
3254 | id->prefix = (PREFIX *)lookup(&dtd.prefixes, name + 6, sizeof(PREFIX)); | |
3255 | id->xmlns = 1; | |
3256 | } | |
3257 | else { | |
3258 | int i; | |
3259 | for (i = 0; name[i]; i++) { | |
3260 | if (name[i] == XML_T(':')) { | |
3261 | int j; | |
3262 | for (j = 0; j < i; j++) { | |
3263 | if (!poolAppendChar(&dtd.pool, name[j])) | |
3264 | return 0; | |
3265 | } | |
3266 | if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
3267 | return 0; | |
3268 | id->prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); | |
3269 | if (id->prefix->name == poolStart(&dtd.pool)) | |
3270 | poolFinish(&dtd.pool); | |
3271 | else | |
3272 | poolDiscard(&dtd.pool); | |
3273 | break; | |
3274 | } | |
3275 | } | |
3276 | } | |
3277 | } | |
3278 | return id; | |
3279 | } | |
3280 | ||
3281 | #define CONTEXT_SEP XML_T('\f') | |
3282 | ||
3283 | static | |
3284 | const XML_Char *getContext(XML_Parser parser) | |
3285 | { | |
3286 | HASH_TABLE_ITER iter; | |
3287 | int needSep = 0; | |
3288 | ||
3289 | if (dtd.defaultPrefix.binding) { | |
3290 | int i; | |
3291 | int len; | |
3292 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
3293 | return 0; | |
3294 | len = dtd.defaultPrefix.binding->uriLen; | |
3295 | if (namespaceSeparator != XML_T('\0')) | |
3296 | len--; | |
3297 | for (i = 0; i < len; i++) | |
3298 | if (!poolAppendChar(&tempPool, dtd.defaultPrefix.binding->uri[i])) | |
3299 | return 0; | |
3300 | needSep = 1; | |
3301 | } | |
3302 | ||
3303 | hashTableIterInit(&iter, &(dtd.prefixes)); | |
3304 | for (;;) { | |
3305 | int i; | |
3306 | int len; | |
3307 | const XML_Char *s; | |
3308 | PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); | |
3309 | if (!prefix) | |
3310 | break; | |
3311 | if (!prefix->binding) | |
3312 | continue; | |
3313 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
3314 | return 0; | |
3315 | for (s = prefix->name; *s; s++) | |
3316 | if (!poolAppendChar(&tempPool, *s)) | |
3317 | return 0; | |
3318 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
3319 | return 0; | |
3320 | len = prefix->binding->uriLen; | |
3321 | if (namespaceSeparator != XML_T('\0')) | |
3322 | len--; | |
3323 | for (i = 0; i < len; i++) | |
3324 | if (!poolAppendChar(&tempPool, prefix->binding->uri[i])) | |
3325 | return 0; | |
3326 | needSep = 1; | |
3327 | } | |
3328 | ||
3329 | ||
3330 | hashTableIterInit(&iter, &(dtd.generalEntities)); | |
3331 | for (;;) { | |
3332 | const XML_Char *s; | |
3333 | ENTITY *e = (ENTITY *)hashTableIterNext(&iter); | |
3334 | if (!e) | |
3335 | break; | |
3336 | if (!e->open) | |
3337 | continue; | |
3338 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
3339 | return 0; | |
3340 | for (s = e->name; *s; s++) | |
3341 | if (!poolAppendChar(&tempPool, *s)) | |
3342 | return 0; | |
3343 | needSep = 1; | |
3344 | } | |
3345 | ||
3346 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
3347 | return 0; | |
3348 | return tempPool.start; | |
3349 | } | |
3350 | ||
3351 | static | |
3352 | int setContext(XML_Parser parser, const XML_Char *context) | |
3353 | { | |
3354 | const XML_Char *s = context; | |
3355 | ||
3356 | while (*context != XML_T('\0')) { | |
3357 | if (*s == CONTEXT_SEP || *s == XML_T('\0')) { | |
3358 | ENTITY *e; | |
3359 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
3360 | return 0; | |
3361 | e = (ENTITY *)lookup(&dtd.generalEntities, poolStart(&tempPool), 0); | |
3362 | if (e) | |
3363 | e->open = 1; | |
3364 | if (*s != XML_T('\0')) | |
3365 | s++; | |
3366 | context = s; | |
3367 | poolDiscard(&tempPool); | |
3368 | } | |
3369 | else if (*s == '=') { | |
3370 | PREFIX *prefix; | |
3371 | if (poolLength(&tempPool) == 0) | |
3372 | prefix = &dtd.defaultPrefix; | |
3373 | else { | |
3374 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
3375 | return 0; | |
3376 | prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&tempPool), sizeof(PREFIX)); | |
3377 | if (!prefix) | |
3378 | return 0; | |
3379 | if (prefix->name == poolStart(&tempPool)) { | |
3380 | prefix->name = poolCopyString(&dtd.pool, prefix->name); | |
3381 | if (!prefix->name) | |
3382 | return 0; | |
3383 | } | |
3384 | poolDiscard(&tempPool); | |
3385 | } | |
3386 | for (context = s + 1; *context != CONTEXT_SEP && *context != XML_T('\0'); context++) | |
3387 | if (!poolAppendChar(&tempPool, *context)) | |
3388 | return 0; | |
3389 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
3390 | return 0; | |
3391 | if (!addBinding(parser, prefix, 0, poolStart(&tempPool), &inheritedBindings)) | |
3392 | return 0; | |
3393 | poolDiscard(&tempPool); | |
3394 | if (*context != XML_T('\0')) | |
3395 | ++context; | |
3396 | s = context; | |
3397 | } | |
3398 | else { | |
3399 | if (!poolAppendChar(&tempPool, *s)) | |
3400 | return 0; | |
3401 | s++; | |
3402 | } | |
3403 | } | |
3404 | return 1; | |
3405 | } | |
3406 | ||
3407 | ||
3408 | static | |
3409 | void normalizePublicId(XML_Char *publicId) | |
3410 | { | |
3411 | XML_Char *p = publicId; | |
3412 | XML_Char *s; | |
3413 | for (s = publicId; *s; s++) { | |
3414 | switch (*s) { | |
3415 | case 0x20: | |
3416 | case 0xD: | |
3417 | case 0xA: | |
3418 | if (p != publicId && p[-1] != 0x20) | |
3419 | *p++ = 0x20; | |
3420 | break; | |
3421 | default: | |
3422 | *p++ = *s; | |
3423 | } | |
3424 | } | |
3425 | if (p != publicId && p[-1] == 0x20) | |
3426 | --p; | |
3427 | *p = XML_T('\0'); | |
3428 | } | |
3429 | ||
3430 | static int dtdInit(DTD *p) | |
3431 | { | |
3432 | poolInit(&(p->pool)); | |
3433 | hashTableInit(&(p->generalEntities)); | |
3434 | hashTableInit(&(p->elementTypes)); | |
3435 | hashTableInit(&(p->attributeIds)); | |
3436 | hashTableInit(&(p->prefixes)); | |
3437 | p->complete = 1; | |
3438 | p->standalone = 0; | |
3439 | #ifdef XML_DTD | |
3440 | hashTableInit(&(p->paramEntities)); | |
3441 | #endif /* XML_DTD */ | |
3442 | p->defaultPrefix.name = 0; | |
3443 | p->defaultPrefix.binding = 0; | |
3444 | return 1; | |
3445 | } | |
3446 | ||
3447 | #ifdef XML_DTD | |
3448 | ||
3449 | static void dtdSwap(DTD *p1, DTD *p2) | |
3450 | { | |
3451 | DTD tem; | |
3452 | memcpy(&tem, p1, sizeof(DTD)); | |
3453 | memcpy(p1, p2, sizeof(DTD)); | |
3454 | memcpy(p2, &tem, sizeof(DTD)); | |
3455 | } | |
3456 | ||
3457 | #endif /* XML_DTD */ | |
3458 | ||
3459 | static void dtdDestroy(DTD *p) | |
3460 | { | |
3461 | HASH_TABLE_ITER iter; | |
3462 | hashTableIterInit(&iter, &(p->elementTypes)); | |
3463 | for (;;) { | |
3464 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
3465 | if (!e) | |
3466 | break; | |
3467 | if (e->allocDefaultAtts != 0) | |
3468 | free(e->defaultAtts); | |
3469 | } | |
3470 | hashTableDestroy(&(p->generalEntities)); | |
3471 | #ifdef XML_DTD | |
3472 | hashTableDestroy(&(p->paramEntities)); | |
3473 | #endif /* XML_DTD */ | |
3474 | hashTableDestroy(&(p->elementTypes)); | |
3475 | hashTableDestroy(&(p->attributeIds)); | |
3476 | hashTableDestroy(&(p->prefixes)); | |
3477 | poolDestroy(&(p->pool)); | |
3478 | } | |
3479 | ||
3480 | /* Do a deep copy of the DTD. Return 0 for out of memory; non-zero otherwise. | |
3481 | The new DTD has already been initialized. */ | |
3482 | ||
3483 | static int dtdCopy(DTD *newDtd, const DTD *oldDtd) | |
3484 | { | |
3485 | HASH_TABLE_ITER iter; | |
3486 | ||
3487 | /* Copy the prefix table. */ | |
3488 | ||
3489 | hashTableIterInit(&iter, &(oldDtd->prefixes)); | |
3490 | for (;;) { | |
3491 | const XML_Char *name; | |
3492 | const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); | |
3493 | if (!oldP) | |
3494 | break; | |
3495 | name = poolCopyString(&(newDtd->pool), oldP->name); | |
3496 | if (!name) | |
3497 | return 0; | |
3498 | if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX))) | |
3499 | return 0; | |
3500 | } | |
3501 | ||
3502 | hashTableIterInit(&iter, &(oldDtd->attributeIds)); | |
3503 | ||
3504 | /* Copy the attribute id table. */ | |
3505 | ||
3506 | for (;;) { | |
3507 | ATTRIBUTE_ID *newA; | |
3508 | const XML_Char *name; | |
3509 | const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); | |
3510 | ||
3511 | if (!oldA) | |
3512 | break; | |
3513 | /* Remember to allocate the scratch byte before the name. */ | |
3514 | if (!poolAppendChar(&(newDtd->pool), XML_T('\0'))) | |
3515 | return 0; | |
3516 | name = poolCopyString(&(newDtd->pool), oldA->name); | |
3517 | if (!name) | |
3518 | return 0; | |
3519 | ++name; | |
3520 | newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name, sizeof(ATTRIBUTE_ID)); | |
3521 | if (!newA) | |
3522 | return 0; | |
3523 | newA->maybeTokenized = oldA->maybeTokenized; | |
3524 | if (oldA->prefix) { | |
3525 | newA->xmlns = oldA->xmlns; | |
3526 | if (oldA->prefix == &oldDtd->defaultPrefix) | |
3527 | newA->prefix = &newDtd->defaultPrefix; | |
3528 | else | |
3529 | newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldA->prefix->name, 0); | |
3530 | } | |
3531 | } | |
3532 | ||
3533 | /* Copy the element type table. */ | |
3534 | ||
3535 | hashTableIterInit(&iter, &(oldDtd->elementTypes)); | |
3536 | ||
3537 | for (;;) { | |
3538 | int i; | |
3539 | ELEMENT_TYPE *newE; | |
3540 | const XML_Char *name; | |
3541 | const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
3542 | if (!oldE) | |
3543 | break; | |
3544 | name = poolCopyString(&(newDtd->pool), oldE->name); | |
3545 | if (!name) | |
3546 | return 0; | |
3547 | newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name, sizeof(ELEMENT_TYPE)); | |
3548 | if (!newE) | |
3549 | return 0; | |
3550 | if (oldE->nDefaultAtts) { | |
3551 | newE->defaultAtts = (DEFAULT_ATTRIBUTE *)malloc(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); | |
3552 | if (!newE->defaultAtts) | |
3553 | return 0; | |
3554 | } | |
3555 | if (oldE->idAtt) | |
3556 | newE->idAtt = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0); | |
3557 | newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; | |
3558 | if (oldE->prefix) | |
3559 | newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldE->prefix->name, 0); | |
3560 | for (i = 0; i < newE->nDefaultAtts; i++) { | |
3561 | newE->defaultAtts[i].id = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); | |
3562 | newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; | |
3563 | if (oldE->defaultAtts[i].value) { | |
3564 | newE->defaultAtts[i].value = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); | |
3565 | if (!newE->defaultAtts[i].value) | |
3566 | return 0; | |
3567 | } | |
3568 | else | |
3569 | newE->defaultAtts[i].value = 0; | |
3570 | } | |
3571 | } | |
3572 | ||
3573 | /* Copy the entity tables. */ | |
3574 | if (!copyEntityTable(&(newDtd->generalEntities), | |
3575 | &(newDtd->pool), | |
3576 | &(oldDtd->generalEntities))) | |
3577 | return 0; | |
3578 | ||
3579 | #ifdef XML_DTD | |
3580 | if (!copyEntityTable(&(newDtd->paramEntities), | |
3581 | &(newDtd->pool), | |
3582 | &(oldDtd->paramEntities))) | |
3583 | return 0; | |
3584 | #endif /* XML_DTD */ | |
3585 | ||
3586 | newDtd->complete = oldDtd->complete; | |
3587 | newDtd->standalone = oldDtd->standalone; | |
3588 | return 1; | |
3589 | } | |
3590 | ||
3591 | static int copyEntityTable(HASH_TABLE *newTable, | |
3592 | STRING_POOL *newPool, | |
3593 | const HASH_TABLE *oldTable) | |
3594 | { | |
3595 | HASH_TABLE_ITER iter; | |
3596 | const XML_Char *cachedOldBase = 0; | |
3597 | const XML_Char *cachedNewBase = 0; | |
3598 | ||
3599 | hashTableIterInit(&iter, oldTable); | |
3600 | ||
3601 | for (;;) { | |
3602 | ENTITY *newE; | |
3603 | const XML_Char *name; | |
3604 | const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); | |
3605 | if (!oldE) | |
3606 | break; | |
3607 | name = poolCopyString(newPool, oldE->name); | |
3608 | if (!name) | |
3609 | return 0; | |
3610 | newE = (ENTITY *)lookup(newTable, name, sizeof(ENTITY)); | |
3611 | if (!newE) | |
3612 | return 0; | |
3613 | if (oldE->systemId) { | |
3614 | const XML_Char *tem = poolCopyString(newPool, oldE->systemId); | |
3615 | if (!tem) | |
3616 | return 0; | |
3617 | newE->systemId = tem; | |
3618 | if (oldE->base) { | |
3619 | if (oldE->base == cachedOldBase) | |
3620 | newE->base = cachedNewBase; | |
3621 | else { | |
3622 | cachedOldBase = oldE->base; | |
3623 | tem = poolCopyString(newPool, cachedOldBase); | |
3624 | if (!tem) | |
3625 | return 0; | |
3626 | cachedNewBase = newE->base = tem; | |
3627 | } | |
3628 | } | |
3629 | } | |
3630 | else { | |
3631 | const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr, oldE->textLen); | |
3632 | if (!tem) | |
3633 | return 0; | |
3634 | newE->textPtr = tem; | |
3635 | newE->textLen = oldE->textLen; | |
3636 | } | |
3637 | if (oldE->notation) { | |
3638 | const XML_Char *tem = poolCopyString(newPool, oldE->notation); | |
3639 | if (!tem) | |
3640 | return 0; | |
3641 | newE->notation = tem; | |
3642 | } | |
3643 | } | |
3644 | return 1; | |
3645 | } | |
3646 | ||
3647 | #define INIT_SIZE 64 | |
3648 | ||
3649 | static | |
3650 | int keyeq(KEY s1, KEY s2) | |
3651 | { | |
3652 | for (; *s1 == *s2; s1++, s2++) | |
3653 | if (*s1 == 0) | |
3654 | return 1; | |
3655 | return 0; | |
3656 | } | |
3657 | ||
3658 | static | |
3659 | unsigned long hash(KEY s) | |
3660 | { | |
3661 | unsigned long h = 0; | |
3662 | while (*s) | |
3663 | h = (h << 5) + h + (unsigned char)*s++; | |
3664 | return h; | |
3665 | } | |
3666 | ||
3667 | static | |
3668 | NAMED *lookup(HASH_TABLE *table, KEY name, size_t createSize) | |
3669 | { | |
3670 | size_t i; | |
3671 | if (table->size == 0) { | |
3672 | if (!createSize) | |
3673 | return 0; | |
3674 | table->v = calloc(INIT_SIZE, sizeof(NAMED *)); | |
3675 | if (!table->v) | |
3676 | return 0; | |
3677 | table->size = INIT_SIZE; | |
3678 | table->usedLim = INIT_SIZE / 2; | |
3679 | i = hash(name) & (table->size - 1); | |
3680 | } | |
3681 | else { | |
3682 | unsigned long h = hash(name); | |
3683 | for (i = h & (table->size - 1); | |
3684 | table->v[i]; | |
3685 | i == 0 ? i = table->size - 1 : --i) { | |
3686 | if (keyeq(name, table->v[i]->name)) | |
3687 | return table->v[i]; | |
3688 | } | |
3689 | if (!createSize) | |
3690 | return 0; | |
3691 | if (table->used == table->usedLim) { | |
3692 | /* check for overflow */ | |
3693 | size_t newSize = table->size * 2; | |
3694 | NAMED **newV = calloc(newSize, sizeof(NAMED *)); | |
3695 | if (!newV) | |
3696 | return 0; | |
3697 | for (i = 0; i < table->size; i++) | |
3698 | if (table->v[i]) { | |
3699 | size_t j; | |
3700 | for (j = hash(table->v[i]->name) & (newSize - 1); | |
3701 | newV[j]; | |
3702 | j == 0 ? j = newSize - 1 : --j) | |
3703 | ; | |
3704 | newV[j] = table->v[i]; | |
3705 | } | |
3706 | free(table->v); | |
3707 | table->v = newV; | |
3708 | table->size = newSize; | |
3709 | table->usedLim = newSize/2; | |
3710 | for (i = h & (table->size - 1); | |
3711 | table->v[i]; | |
3712 | i == 0 ? i = table->size - 1 : --i) | |
3713 | ; | |
3714 | } | |
3715 | } | |
3716 | table->v[i] = calloc(1, createSize); | |
3717 | if (!table->v[i]) | |
3718 | return 0; | |
3719 | table->v[i]->name = name; | |
3720 | (table->used)++; | |
3721 | return table->v[i]; | |
3722 | } | |
3723 | ||
3724 | static | |
3725 | void hashTableDestroy(HASH_TABLE *table) | |
3726 | { | |
3727 | size_t i; | |
3728 | for (i = 0; i < table->size; i++) { | |
3729 | NAMED *p = table->v[i]; | |
3730 | if (p) | |
3731 | free(p); | |
3732 | } | |
3733 | if (table->v) | |
3734 | free(table->v); | |
3735 | } | |
3736 | ||
3737 | static | |
3738 | void hashTableInit(HASH_TABLE *p) | |
3739 | { | |
3740 | p->size = 0; | |
3741 | p->usedLim = 0; | |
3742 | p->used = 0; | |
3743 | p->v = 0; | |
3744 | } | |
3745 | ||
3746 | static | |
3747 | void hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table) | |
3748 | { | |
3749 | iter->p = table->v; | |
3750 | iter->end = iter->p + table->size; | |
3751 | } | |
3752 | ||
3753 | static | |
3754 | NAMED *hashTableIterNext(HASH_TABLE_ITER *iter) | |
3755 | { | |
3756 | while (iter->p != iter->end) { | |
3757 | NAMED *tem = *(iter->p)++; | |
3758 | if (tem) | |
3759 | return tem; | |
3760 | } | |
3761 | return 0; | |
3762 | } | |
3763 | ||
3764 | ||
3765 | static | |
3766 | void poolInit(STRING_POOL *pool) | |
3767 | { | |
3768 | pool->blocks = 0; | |
3769 | pool->freeBlocks = 0; | |
3770 | pool->start = 0; | |
3771 | pool->ptr = 0; | |
3772 | pool->end = 0; | |
3773 | } | |
3774 | ||
3775 | static | |
3776 | void poolClear(STRING_POOL *pool) | |
3777 | { | |
3778 | if (!pool->freeBlocks) | |
3779 | pool->freeBlocks = pool->blocks; | |
3780 | else { | |
3781 | BLOCK *p = pool->blocks; | |
3782 | while (p) { | |
3783 | BLOCK *tem = p->next; | |
3784 | p->next = pool->freeBlocks; | |
3785 | pool->freeBlocks = p; | |
3786 | p = tem; | |
3787 | } | |
3788 | } | |
3789 | pool->blocks = 0; | |
3790 | pool->start = 0; | |
3791 | pool->ptr = 0; | |
3792 | pool->end = 0; | |
3793 | } | |
3794 | ||
3795 | static | |
3796 | void poolDestroy(STRING_POOL *pool) | |
3797 | { | |
3798 | BLOCK *p = pool->blocks; | |
3799 | while (p) { | |
3800 | BLOCK *tem = p->next; | |
3801 | free(p); | |
3802 | p = tem; | |
3803 | } | |
3804 | pool->blocks = 0; | |
3805 | p = pool->freeBlocks; | |
3806 | while (p) { | |
3807 | BLOCK *tem = p->next; | |
3808 | free(p); | |
3809 | p = tem; | |
3810 | } | |
3811 | pool->freeBlocks = 0; | |
3812 | pool->ptr = 0; | |
3813 | pool->start = 0; | |
3814 | pool->end = 0; | |
3815 | } | |
3816 | ||
3817 | static | |
3818 | XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
3819 | const char *ptr, const char *end) | |
3820 | { | |
3821 | if (!pool->ptr && !poolGrow(pool)) | |
3822 | return 0; | |
3823 | for (;;) { | |
3824 | XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); | |
3825 | if (ptr == end) | |
3826 | break; | |
3827 | if (!poolGrow(pool)) | |
3828 | return 0; | |
3829 | } | |
3830 | return pool->start; | |
3831 | } | |
3832 | ||
3833 | static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s) | |
3834 | { | |
3835 | do { | |
3836 | if (!poolAppendChar(pool, *s)) | |
3837 | return 0; | |
3838 | } while (*s++); | |
3839 | s = pool->start; | |
3840 | poolFinish(pool); | |
3841 | return s; | |
3842 | } | |
3843 | ||
3844 | static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) | |
3845 | { | |
3846 | if (!pool->ptr && !poolGrow(pool)) | |
3847 | return 0; | |
3848 | for (; n > 0; --n, s++) { | |
3849 | if (!poolAppendChar(pool, *s)) | |
3850 | return 0; | |
3851 | ||
3852 | } | |
3853 | s = pool->start; | |
3854 | poolFinish(pool); | |
3855 | return s; | |
3856 | } | |
3857 | ||
3858 | static | |
3859 | XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
3860 | const char *ptr, const char *end) | |
3861 | { | |
3862 | if (!poolAppend(pool, enc, ptr, end)) | |
3863 | return 0; | |
3864 | if (pool->ptr == pool->end && !poolGrow(pool)) | |
3865 | return 0; | |
3866 | *(pool->ptr)++ = 0; | |
3867 | return pool->start; | |
3868 | } | |
3869 | ||
3870 | static | |
3871 | int poolGrow(STRING_POOL *pool) | |
3872 | { | |
3873 | if (pool->freeBlocks) { | |
3874 | if (pool->start == 0) { | |
3875 | pool->blocks = pool->freeBlocks; | |
3876 | pool->freeBlocks = pool->freeBlocks->next; | |
3877 | pool->blocks->next = 0; | |
3878 | pool->start = pool->blocks->s; | |
3879 | pool->end = pool->start + pool->blocks->size; | |
3880 | pool->ptr = pool->start; | |
3881 | return 1; | |
3882 | } | |
3883 | if (pool->end - pool->start < pool->freeBlocks->size) { | |
3884 | BLOCK *tem = pool->freeBlocks->next; | |
3885 | pool->freeBlocks->next = pool->blocks; | |
3886 | pool->blocks = pool->freeBlocks; | |
3887 | pool->freeBlocks = tem; | |
3888 | memcpy(pool->blocks->s, pool->start, (pool->end - pool->start) * sizeof(XML_Char)); | |
3889 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
3890 | pool->start = pool->blocks->s; | |
3891 | pool->end = pool->start + pool->blocks->size; | |
3892 | return 1; | |
3893 | } | |
3894 | } | |
3895 | if (pool->blocks && pool->start == pool->blocks->s) { | |
3896 | int blockSize = (pool->end - pool->start)*2; | |
3897 | pool->blocks = realloc(pool->blocks, offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); | |
3898 | if (!pool->blocks) | |
3899 | return 0; | |
3900 | pool->blocks->size = blockSize; | |
3901 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
3902 | pool->start = pool->blocks->s; | |
3903 | pool->end = pool->start + blockSize; | |
3904 | } | |
3905 | else { | |
3906 | BLOCK *tem; | |
3907 | int blockSize = pool->end - pool->start; | |
3908 | if (blockSize < INIT_BLOCK_SIZE) | |
3909 | blockSize = INIT_BLOCK_SIZE; | |
3910 | else | |
3911 | blockSize *= 2; | |
3912 | tem = malloc(offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); | |
3913 | if (!tem) | |
3914 | return 0; | |
3915 | tem->size = blockSize; | |
3916 | tem->next = pool->blocks; | |
3917 | pool->blocks = tem; | |
3918 | if (pool->ptr != pool->start) | |
3919 | memcpy(tem->s, pool->start, (pool->ptr - pool->start) * sizeof(XML_Char)); | |
3920 | pool->ptr = tem->s + (pool->ptr - pool->start); | |
3921 | pool->start = tem->s; | |
3922 | pool->end = tem->s + blockSize; | |
3923 | } | |
3924 | return 1; | |
3925 | } |