Mercurial > hg
annotate mcabber/libjabber/xmlparse.c @ 1030:923cf5a92b44
[BP-dd9e7eb5f8a8] Remove old UTF-8 conversions
author | Mikael Berthe <mikael@lilotux.net> |
---|---|
date | Thu, 16 Nov 2006 19:30:12 +0100 |
parents | 0aa9015f06df |
children |
rev | line source |
---|---|
25 | 1 /* |
2 The contents of this file are subject to the Mozilla Public License | |
3 Version 1.1 (the "License"); you may not use this file except in | |
4 compliance with the License. You may obtain a copy of the License at | |
5 http://www.mozilla.org/MPL/ | |
6 | |
7 Software distributed under the License is distributed on an "AS IS" | |
8 basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the | |
9 License for the specific language governing rights and limitations | |
10 under the License. | |
11 | |
12 The Original Code is expat. | |
13 | |
14 The Initial Developer of the Original Code is James Clark. | |
15 Portions created by James Clark are Copyright (C) 1998, 1999 | |
16 James Clark. All Rights Reserved. | |
17 | |
18 Contributor(s): | |
19 | |
20 Alternatively, the contents of this file may be used under the terms | |
21 of the GNU General Public License (the "GPL"), in which case the | |
22 provisions of the GPL are applicable instead of those above. If you | |
23 wish to allow use of your version of this file only under the terms of | |
24 the GPL and not to allow others to use your version of this file under | |
25 the MPL, indicate your decision by deleting the provisions above and | |
26 replace them with the notice and other provisions required by the | |
27 GPL. If you do not delete the provisions above, a recipient may use | |
28 your version of this file under either the MPL or the GPL. | |
29 */ | |
30 | |
31 #include "xmldef.h" | |
32 #include "xmlparse.h" | |
33 | |
34 #ifdef XML_UNICODE | |
35 #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX | |
36 #define XmlConvert XmlUtf16Convert | |
37 #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding | |
38 #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS | |
39 #define XmlEncode XmlUtf16Encode | |
40 #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1)) | |
41 typedef unsigned short ICHAR; | |
42 #else | |
43 #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX | |
44 #define XmlConvert XmlUtf8Convert | |
45 #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding | |
46 #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS | |
47 #define XmlEncode XmlUtf8Encode | |
48 #define MUST_CONVERT(enc, s) (!(enc)->isUtf8) | |
49 typedef char ICHAR; | |
50 #endif | |
51 | |
52 | |
53 #ifndef XML_NS | |
54 | |
55 #define XmlInitEncodingNS XmlInitEncoding | |
56 #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding | |
57 #undef XmlGetInternalEncodingNS | |
58 #define XmlGetInternalEncodingNS XmlGetInternalEncoding | |
59 #define XmlParseXmlDeclNS XmlParseXmlDecl | |
60 | |
61 #endif | |
62 | |
63 | |
64 #ifdef XML_UNICODE_WCHAR_T | |
65 #define XML_T(x) L ## x | |
66 #else | |
67 #define XML_T(x) x | |
68 #endif | |
69 | |
70 /* Round up n to be a multiple of sz, where sz is a power of 2. */ | |
71 #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1)) | |
72 | |
73 #include "xmltok.h" | |
74 #include "xmlrole.h" | |
75 #include "hashtable.h" | |
76 | |
77 #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ | |
78 #define INIT_DATA_BUF_SIZE 1024 | |
79 #define INIT_ATTS_SIZE 16 | |
80 #define INIT_BLOCK_SIZE 1024 | |
81 #define INIT_BUFFER_SIZE 1024 | |
82 | |
83 #define EXPAND_SPARE 24 | |
84 | |
85 typedef struct binding { | |
86 struct prefix *prefix; | |
87 struct binding *nextTagBinding; | |
88 struct binding *prevPrefixBinding; | |
89 const struct attribute_id *attId; | |
90 XML_Char *uri; | |
91 int uriLen; | |
92 int uriAlloc; | |
93 } BINDING; | |
94 | |
95 typedef struct prefix { | |
96 const XML_Char *name; | |
97 BINDING *binding; | |
98 } PREFIX; | |
99 | |
100 typedef struct { | |
101 const XML_Char *str; | |
102 const XML_Char *localPart; | |
103 int uriLen; | |
104 } TAG_NAME; | |
105 | |
106 typedef struct tag { | |
107 struct tag *parent; | |
108 const char *rawName; | |
109 int rawNameLength; | |
110 TAG_NAME name; | |
111 char *buf; | |
112 char *bufEnd; | |
113 BINDING *bindings; | |
114 } TAG; | |
115 | |
116 typedef struct { | |
117 const XML_Char *name; | |
118 const XML_Char *textPtr; | |
119 int textLen; | |
120 const XML_Char *systemId; | |
121 const XML_Char *base; | |
122 const XML_Char *publicId; | |
123 const XML_Char *notation; | |
124 char open; | |
125 } ENTITY; | |
126 | |
127 typedef struct block { | |
128 struct block *next; | |
129 int size; | |
130 XML_Char s[1]; | |
131 } BLOCK; | |
132 | |
133 typedef struct { | |
134 BLOCK *blocks; | |
135 BLOCK *freeBlocks; | |
136 const XML_Char *end; | |
137 XML_Char *ptr; | |
138 XML_Char *start; | |
139 } STRING_POOL; | |
140 | |
141 /* The XML_Char before the name is used to determine whether | |
142 an attribute has been specified. */ | |
143 typedef struct attribute_id { | |
144 XML_Char *name; | |
145 PREFIX *prefix; | |
146 char maybeTokenized; | |
147 char xmlns; | |
148 } ATTRIBUTE_ID; | |
149 | |
150 typedef struct { | |
151 const ATTRIBUTE_ID *id; | |
152 char isCdata; | |
153 const XML_Char *value; | |
154 } DEFAULT_ATTRIBUTE; | |
155 | |
156 typedef struct { | |
157 const XML_Char *name; | |
158 PREFIX *prefix; | |
159 int nDefaultAtts; | |
160 int allocDefaultAtts; | |
161 DEFAULT_ATTRIBUTE *defaultAtts; | |
162 } ELEMENT_TYPE; | |
163 | |
164 typedef struct { | |
165 HASH_TABLE generalEntities; | |
166 HASH_TABLE elementTypes; | |
167 HASH_TABLE attributeIds; | |
168 HASH_TABLE prefixes; | |
169 STRING_POOL pool; | |
170 int complete; | |
171 int standalone; | |
172 const XML_Char *base; | |
173 PREFIX defaultPrefix; | |
174 } DTD; | |
175 | |
176 typedef struct open_internal_entity { | |
177 const char *internalEventPtr; | |
178 const char *internalEventEndPtr; | |
179 struct open_internal_entity *next; | |
180 ENTITY *entity; | |
181 } OPEN_INTERNAL_ENTITY; | |
182 | |
183 typedef enum XML_Error Processor(XML_Parser parser, | |
184 const char *start, | |
185 const char *end, | |
186 const char **endPtr); | |
187 | |
188 static Processor prologProcessor; | |
189 static Processor prologInitProcessor; | |
190 static Processor contentProcessor; | |
191 static Processor cdataSectionProcessor; | |
192 static Processor epilogProcessor; | |
235 | 193 /* static Processor errorProcessor; */ |
25 | 194 static Processor externalEntityInitProcessor; |
195 static Processor externalEntityInitProcessor2; | |
196 static Processor externalEntityInitProcessor3; | |
197 static Processor externalEntityContentProcessor; | |
198 | |
199 static enum XML_Error | |
200 handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName); | |
201 static enum XML_Error | |
202 processXmlDecl(XML_Parser parser, int isGeneralTextEntity, const char *, const char *); | |
203 static enum XML_Error | |
204 initializeEncoding(XML_Parser parser); | |
205 static enum XML_Error | |
206 doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, | |
207 const char *start, const char *end, const char **endPtr); | |
208 static enum XML_Error | |
209 doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr); | |
210 static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *, const char *s, | |
211 TAG_NAME *tagNamePtr, BINDING **bindingsPtr); | |
212 static | |
213 int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr); | |
214 static int | |
215 defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, int isCdata, const XML_Char *dfltValue); | |
216 static enum XML_Error | |
217 storeAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, | |
218 STRING_POOL *); | |
219 static enum XML_Error | |
220 appendAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, | |
221 STRING_POOL *); | |
222 static ATTRIBUTE_ID * | |
223 getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
224 static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); | |
225 static enum XML_Error | |
226 storeEntityValue(XML_Parser parser, const char *start, const char *end); | |
227 static int | |
228 reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
229 static int | |
230 reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
231 static void | |
232 reportDefault(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
233 | |
234 static const XML_Char *getContext(XML_Parser parser); | |
235 static int setContext(XML_Parser parser, const XML_Char *context); | |
236 static void normalizePublicId(XML_Char *s); | |
237 static int dtdInit(DTD *); | |
238 static void dtdDestroy(DTD *); | |
239 static int dtdCopy(DTD *newDtd, const DTD *oldDtd); | |
240 static void poolInit(STRING_POOL *); | |
241 static void poolClear(STRING_POOL *); | |
242 static void poolDestroy(STRING_POOL *); | |
243 static XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
244 const char *ptr, const char *end); | |
245 static XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
246 const char *ptr, const char *end); | |
247 static int poolGrow(STRING_POOL *pool); | |
248 static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s); | |
249 static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n); | |
250 | |
251 #define poolStart(pool) ((pool)->start) | |
252 #define poolEnd(pool) ((pool)->ptr) | |
253 #define poolLength(pool) ((pool)->ptr - (pool)->start) | |
254 #define poolChop(pool) ((void)--(pool->ptr)) | |
255 #define poolLastChar(pool) (((pool)->ptr)[-1]) | |
256 #define poolDiscard(pool) ((pool)->ptr = (pool)->start) | |
257 #define poolFinish(pool) ((pool)->start = (pool)->ptr) | |
258 #define poolAppendChar(pool, c) \ | |
259 (((pool)->ptr == (pool)->end && !poolGrow(pool)) \ | |
260 ? 0 \ | |
261 : ((*((pool)->ptr)++ = c), 1)) | |
262 | |
263 typedef struct { | |
264 /* The first member must be userData so that the XML_GetUserData macro works. */ | |
265 void *m_userData; | |
266 void *m_handlerArg; | |
267 char *m_buffer; | |
268 /* first character to be parsed */ | |
269 const char *m_bufferPtr; | |
270 /* past last character to be parsed */ | |
271 char *m_bufferEnd; | |
272 /* allocated end of buffer */ | |
273 const char *m_bufferLim; | |
274 long m_parseEndByteIndex; | |
275 const char *m_parseEndPtr; | |
276 XML_Char *m_dataBuf; | |
277 XML_Char *m_dataBufEnd; | |
278 XML_StartElementHandler m_startElementHandler; | |
279 XML_EndElementHandler m_endElementHandler; | |
280 XML_CharacterDataHandler m_characterDataHandler; | |
281 XML_ProcessingInstructionHandler m_processingInstructionHandler; | |
282 XML_CommentHandler m_commentHandler; | |
283 XML_StartCdataSectionHandler m_startCdataSectionHandler; | |
284 XML_EndCdataSectionHandler m_endCdataSectionHandler; | |
285 XML_DefaultHandler m_defaultHandler; | |
286 XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; | |
287 XML_NotationDeclHandler m_notationDeclHandler; | |
288 XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; | |
289 XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; | |
290 XML_NotStandaloneHandler m_notStandaloneHandler; | |
291 XML_ExternalEntityRefHandler m_externalEntityRefHandler; | |
292 void *m_externalEntityRefHandlerArg; | |
293 XML_UnknownEncodingHandler m_unknownEncodingHandler; | |
294 const ENCODING *m_encoding; | |
295 INIT_ENCODING m_initEncoding; | |
296 const XML_Char *m_protocolEncodingName; | |
297 int m_ns; | |
298 void *m_unknownEncodingMem; | |
299 void *m_unknownEncodingData; | |
300 void *m_unknownEncodingHandlerData; | |
301 void (*m_unknownEncodingRelease)(void *); | |
302 PROLOG_STATE m_prologState; | |
303 Processor *m_processor; | |
304 enum XML_Error m_errorCode; | |
305 const char *m_eventPtr; | |
306 const char *m_eventEndPtr; | |
307 const char *m_positionPtr; | |
308 OPEN_INTERNAL_ENTITY *m_openInternalEntities; | |
309 int m_defaultExpandInternalEntities; | |
310 int m_tagLevel; | |
311 ENTITY *m_declEntity; | |
312 const XML_Char *m_declNotationName; | |
313 const XML_Char *m_declNotationPublicId; | |
314 ELEMENT_TYPE *m_declElementType; | |
315 ATTRIBUTE_ID *m_declAttributeId; | |
316 char m_declAttributeIsCdata; | |
317 DTD m_dtd; | |
318 TAG *m_tagStack; | |
319 TAG *m_freeTagList; | |
320 BINDING *m_inheritedBindings; | |
321 BINDING *m_freeBindingList; | |
322 int m_attsSize; | |
323 int m_nSpecifiedAtts; | |
324 ATTRIBUTE *m_atts; | |
325 POSITION m_position; | |
326 STRING_POOL m_tempPool; | |
327 STRING_POOL m_temp2Pool; | |
328 char *m_groupConnector; | |
329 unsigned m_groupSize; | |
330 int m_hadExternalDoctype; | |
331 XML_Char m_namespaceSeparator; | |
332 } Parser; | |
333 | |
334 #define userData (((Parser *)parser)->m_userData) | |
335 #define handlerArg (((Parser *)parser)->m_handlerArg) | |
336 #define startElementHandler (((Parser *)parser)->m_startElementHandler) | |
337 #define endElementHandler (((Parser *)parser)->m_endElementHandler) | |
338 #define characterDataHandler (((Parser *)parser)->m_characterDataHandler) | |
339 #define processingInstructionHandler (((Parser *)parser)->m_processingInstructionHandler) | |
340 #define commentHandler (((Parser *)parser)->m_commentHandler) | |
341 #define startCdataSectionHandler (((Parser *)parser)->m_startCdataSectionHandler) | |
342 #define endCdataSectionHandler (((Parser *)parser)->m_endCdataSectionHandler) | |
343 #define defaultHandler (((Parser *)parser)->m_defaultHandler) | |
344 #define unparsedEntityDeclHandler (((Parser *)parser)->m_unparsedEntityDeclHandler) | |
345 #define notationDeclHandler (((Parser *)parser)->m_notationDeclHandler) | |
346 #define startNamespaceDeclHandler (((Parser *)parser)->m_startNamespaceDeclHandler) | |
347 #define endNamespaceDeclHandler (((Parser *)parser)->m_endNamespaceDeclHandler) | |
348 #define notStandaloneHandler (((Parser *)parser)->m_notStandaloneHandler) | |
349 #define externalEntityRefHandler (((Parser *)parser)->m_externalEntityRefHandler) | |
350 #define externalEntityRefHandlerArg (((Parser *)parser)->m_externalEntityRefHandlerArg) | |
351 #define unknownEncodingHandler (((Parser *)parser)->m_unknownEncodingHandler) | |
352 #define encoding (((Parser *)parser)->m_encoding) | |
353 #define initEncoding (((Parser *)parser)->m_initEncoding) | |
354 #define unknownEncodingMem (((Parser *)parser)->m_unknownEncodingMem) | |
355 #define unknownEncodingData (((Parser *)parser)->m_unknownEncodingData) | |
356 #define unknownEncodingHandlerData \ | |
357 (((Parser *)parser)->m_unknownEncodingHandlerData) | |
358 #define unknownEncodingRelease (((Parser *)parser)->m_unknownEncodingRelease) | |
359 #define protocolEncodingName (((Parser *)parser)->m_protocolEncodingName) | |
360 #define ns (((Parser *)parser)->m_ns) | |
361 #define prologState (((Parser *)parser)->m_prologState) | |
362 #define processor (((Parser *)parser)->m_processor) | |
363 #define errorCode (((Parser *)parser)->m_errorCode) | |
364 #define eventPtr (((Parser *)parser)->m_eventPtr) | |
365 #define eventEndPtr (((Parser *)parser)->m_eventEndPtr) | |
366 #define positionPtr (((Parser *)parser)->m_positionPtr) | |
367 #define position (((Parser *)parser)->m_position) | |
368 #define openInternalEntities (((Parser *)parser)->m_openInternalEntities) | |
369 #define defaultExpandInternalEntities (((Parser *)parser)->m_defaultExpandInternalEntities) | |
370 #define tagLevel (((Parser *)parser)->m_tagLevel) | |
371 #define buffer (((Parser *)parser)->m_buffer) | |
372 #define bufferPtr (((Parser *)parser)->m_bufferPtr) | |
373 #define bufferEnd (((Parser *)parser)->m_bufferEnd) | |
374 #define parseEndByteIndex (((Parser *)parser)->m_parseEndByteIndex) | |
375 #define parseEndPtr (((Parser *)parser)->m_parseEndPtr) | |
376 #define bufferLim (((Parser *)parser)->m_bufferLim) | |
377 #define dataBuf (((Parser *)parser)->m_dataBuf) | |
378 #define dataBufEnd (((Parser *)parser)->m_dataBufEnd) | |
379 #define dtd (((Parser *)parser)->m_dtd) | |
380 #define declEntity (((Parser *)parser)->m_declEntity) | |
381 #define declNotationName (((Parser *)parser)->m_declNotationName) | |
382 #define declNotationPublicId (((Parser *)parser)->m_declNotationPublicId) | |
383 #define declElementType (((Parser *)parser)->m_declElementType) | |
384 #define declAttributeId (((Parser *)parser)->m_declAttributeId) | |
385 #define declAttributeIsCdata (((Parser *)parser)->m_declAttributeIsCdata) | |
386 #define freeTagList (((Parser *)parser)->m_freeTagList) | |
387 #define freeBindingList (((Parser *)parser)->m_freeBindingList) | |
388 #define inheritedBindings (((Parser *)parser)->m_inheritedBindings) | |
389 #define tagStack (((Parser *)parser)->m_tagStack) | |
390 #define atts (((Parser *)parser)->m_atts) | |
391 #define attsSize (((Parser *)parser)->m_attsSize) | |
392 #define nSpecifiedAtts (((Parser *)parser)->m_nSpecifiedAtts) | |
393 #define tempPool (((Parser *)parser)->m_tempPool) | |
394 #define temp2Pool (((Parser *)parser)->m_temp2Pool) | |
395 #define groupConnector (((Parser *)parser)->m_groupConnector) | |
396 #define groupSize (((Parser *)parser)->m_groupSize) | |
397 #define hadExternalDoctype (((Parser *)parser)->m_hadExternalDoctype) | |
398 #define namespaceSeparator (((Parser *)parser)->m_namespaceSeparator) | |
399 | |
400 #ifdef _MSC_VER | |
401 #ifdef _DEBUG | |
402 Parser *asParser(XML_Parser parser) | |
403 { | |
404 return parser; | |
405 } | |
406 #endif | |
407 #endif | |
408 | |
409 XML_Parser XML_ParserCreate(const XML_Char *encodingName) | |
410 { | |
411 XML_Parser parser = malloc(sizeof(Parser)); | |
412 if (!parser) | |
413 return parser; | |
414 processor = prologInitProcessor; | |
415 XmlPrologStateInit(&prologState); | |
416 userData = 0; | |
417 handlerArg = 0; | |
418 startElementHandler = 0; | |
419 endElementHandler = 0; | |
420 characterDataHandler = 0; | |
421 processingInstructionHandler = 0; | |
422 commentHandler = 0; | |
423 startCdataSectionHandler = 0; | |
424 endCdataSectionHandler = 0; | |
425 defaultHandler = 0; | |
426 unparsedEntityDeclHandler = 0; | |
427 notationDeclHandler = 0; | |
428 startNamespaceDeclHandler = 0; | |
429 endNamespaceDeclHandler = 0; | |
430 notStandaloneHandler = 0; | |
431 externalEntityRefHandler = 0; | |
432 externalEntityRefHandlerArg = parser; | |
433 unknownEncodingHandler = 0; | |
434 buffer = 0; | |
435 bufferPtr = 0; | |
436 bufferEnd = 0; | |
437 parseEndByteIndex = 0; | |
438 parseEndPtr = 0; | |
439 bufferLim = 0; | |
440 declElementType = 0; | |
441 declAttributeId = 0; | |
442 declEntity = 0; | |
443 declNotationName = 0; | |
444 declNotationPublicId = 0; | |
445 memset(&position, 0, sizeof(POSITION)); | |
446 errorCode = XML_ERROR_NONE; | |
447 eventPtr = 0; | |
448 eventEndPtr = 0; | |
449 positionPtr = 0; | |
450 openInternalEntities = 0; | |
451 tagLevel = 0; | |
452 tagStack = 0; | |
453 freeTagList = 0; | |
454 freeBindingList = 0; | |
455 inheritedBindings = 0; | |
456 attsSize = INIT_ATTS_SIZE; | |
457 atts = malloc(attsSize * sizeof(ATTRIBUTE)); | |
458 nSpecifiedAtts = 0; | |
459 dataBuf = malloc(INIT_DATA_BUF_SIZE * sizeof(XML_Char)); | |
460 groupSize = 0; | |
461 groupConnector = 0; | |
462 hadExternalDoctype = 0; | |
463 unknownEncodingMem = 0; | |
464 unknownEncodingRelease = 0; | |
465 unknownEncodingData = 0; | |
466 unknownEncodingHandlerData = 0; | |
467 namespaceSeparator = '!'; | |
468 ns = 0; | |
469 poolInit(&tempPool); | |
470 poolInit(&temp2Pool); | |
471 protocolEncodingName = encodingName ? poolCopyString(&tempPool, encodingName) : 0; | |
472 if (!dtdInit(&dtd) || !atts || !dataBuf | |
473 || (encodingName && !protocolEncodingName)) { | |
474 XML_ParserFree(parser); | |
475 return 0; | |
476 } | |
477 dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE; | |
478 XmlInitEncoding(&initEncoding, &encoding, 0); | |
479 return parser; | |
480 } | |
481 | |
482 XML_Parser XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) | |
483 { | |
484 static | |
485 const XML_Char implicitContext[] = { | |
486 XML_T('x'), XML_T('m'), XML_T('l'), XML_T('='), | |
487 XML_T('h'), XML_T('t'), XML_T('t'), XML_T('p'), XML_T(':'), | |
488 XML_T('/'), XML_T('/'), XML_T('w'), XML_T('w'), XML_T('w'), | |
489 XML_T('.'), XML_T('w'), XML_T('3'), | |
490 XML_T('.'), XML_T('o'), XML_T('r'), XML_T('g'), | |
491 XML_T('/'), XML_T('X'), XML_T('M'), XML_T('L'), | |
492 XML_T('/'), XML_T('1'), XML_T('9'), XML_T('9'), XML_T('8'), | |
493 XML_T('/'), XML_T('n'), XML_T('a'), XML_T('m'), XML_T('e'), | |
494 XML_T('s'), XML_T('p'), XML_T('a'), XML_T('c'), XML_T('e'), | |
495 XML_T('\0') | |
496 }; | |
497 | |
498 XML_Parser parser = XML_ParserCreate(encodingName); | |
499 if (parser) { | |
500 XmlInitEncodingNS(&initEncoding, &encoding, 0); | |
501 ns = 1; | |
502 namespaceSeparator = nsSep; | |
503 } | |
504 if (!setContext(parser, implicitContext)) { | |
505 XML_ParserFree(parser); | |
506 return 0; | |
507 } | |
508 return parser; | |
509 } | |
510 | |
511 int XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) | |
512 { | |
513 if (!encodingName) | |
514 protocolEncodingName = 0; | |
515 else { | |
516 protocolEncodingName = poolCopyString(&tempPool, encodingName); | |
517 if (!protocolEncodingName) | |
518 return 0; | |
519 } | |
520 return 1; | |
521 } | |
522 | |
523 XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser, | |
524 const XML_Char *context, | |
525 const XML_Char *encodingName) | |
526 { | |
527 XML_Parser parser = oldParser; | |
528 DTD *oldDtd = &dtd; | |
529 XML_StartElementHandler oldStartElementHandler = startElementHandler; | |
530 XML_EndElementHandler oldEndElementHandler = endElementHandler; | |
531 XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler; | |
532 XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler; | |
533 XML_CommentHandler oldCommentHandler = commentHandler; | |
534 XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler; | |
535 XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler; | |
536 XML_DefaultHandler oldDefaultHandler = defaultHandler; | |
537 XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler = startNamespaceDeclHandler; | |
538 XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler = endNamespaceDeclHandler; | |
539 XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler; | |
540 XML_ExternalEntityRefHandler oldExternalEntityRefHandler = externalEntityRefHandler; | |
541 XML_UnknownEncodingHandler oldUnknownEncodingHandler = unknownEncodingHandler; | |
542 void *oldUserData = userData; | |
543 void *oldHandlerArg = handlerArg; | |
544 int oldDefaultExpandInternalEntities = defaultExpandInternalEntities; | |
545 void *oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg; | |
546 | |
547 parser = (ns | |
548 ? XML_ParserCreateNS(encodingName, namespaceSeparator) | |
549 : XML_ParserCreate(encodingName)); | |
550 if (!parser) | |
551 return 0; | |
552 startElementHandler = oldStartElementHandler; | |
553 endElementHandler = oldEndElementHandler; | |
554 characterDataHandler = oldCharacterDataHandler; | |
555 processingInstructionHandler = oldProcessingInstructionHandler; | |
556 commentHandler = oldCommentHandler; | |
557 startCdataSectionHandler = oldStartCdataSectionHandler; | |
558 endCdataSectionHandler = oldEndCdataSectionHandler; | |
559 defaultHandler = oldDefaultHandler; | |
560 startNamespaceDeclHandler = oldStartNamespaceDeclHandler; | |
561 endNamespaceDeclHandler = oldEndNamespaceDeclHandler; | |
562 notStandaloneHandler = oldNotStandaloneHandler; | |
563 externalEntityRefHandler = oldExternalEntityRefHandler; | |
564 unknownEncodingHandler = oldUnknownEncodingHandler; | |
565 userData = oldUserData; | |
566 if (oldUserData == oldHandlerArg) | |
567 handlerArg = userData; | |
568 else | |
569 handlerArg = parser; | |
570 if (oldExternalEntityRefHandlerArg != oldParser) | |
571 externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; | |
572 defaultExpandInternalEntities = oldDefaultExpandInternalEntities; | |
573 if (!dtdCopy(&dtd, oldDtd) || !setContext(parser, context)) { | |
574 XML_ParserFree(parser); | |
575 return 0; | |
576 } | |
577 processor = externalEntityInitProcessor; | |
578 return parser; | |
579 } | |
580 | |
581 static | |
582 void destroyBindings(BINDING *bindings) | |
583 { | |
584 for (;;) { | |
585 BINDING *b = bindings; | |
586 if (!b) | |
587 break; | |
588 bindings = b->nextTagBinding; | |
589 free(b->uri); | |
590 free(b); | |
591 } | |
592 } | |
593 | |
594 void XML_ParserFree(XML_Parser parser) | |
595 { | |
596 for (;;) { | |
597 TAG *p; | |
598 if (tagStack == 0) { | |
599 if (freeTagList == 0) | |
600 break; | |
601 tagStack = freeTagList; | |
602 freeTagList = 0; | |
603 } | |
604 p = tagStack; | |
605 tagStack = tagStack->parent; | |
606 free(p->buf); | |
607 destroyBindings(p->bindings); | |
608 free(p); | |
609 } | |
610 destroyBindings(freeBindingList); | |
611 destroyBindings(inheritedBindings); | |
612 poolDestroy(&tempPool); | |
613 poolDestroy(&temp2Pool); | |
614 dtdDestroy(&dtd); | |
615 free((void *)atts); | |
616 free(groupConnector); | |
617 free(buffer); | |
618 free(dataBuf); | |
619 free(unknownEncodingMem); | |
620 if (unknownEncodingRelease) | |
621 unknownEncodingRelease(unknownEncodingData); | |
622 free(parser); | |
623 } | |
624 | |
625 void XML_UseParserAsHandlerArg(XML_Parser parser) | |
626 { | |
627 handlerArg = parser; | |
628 } | |
629 | |
630 void XML_SetUserData(XML_Parser parser, void *p) | |
631 { | |
632 if (handlerArg == userData) | |
633 handlerArg = userData = p; | |
634 else | |
635 userData = p; | |
636 } | |
637 | |
638 int XML_SetBase(XML_Parser parser, const XML_Char *p) | |
639 { | |
640 if (p) { | |
641 p = poolCopyString(&dtd.pool, p); | |
642 if (!p) | |
643 return 0; | |
644 dtd.base = p; | |
645 } | |
646 else | |
647 dtd.base = 0; | |
648 return 1; | |
649 } | |
650 | |
651 const XML_Char *XML_GetBase(XML_Parser parser) | |
652 { | |
653 return dtd.base; | |
654 } | |
655 | |
656 int XML_GetSpecifiedAttributeCount(XML_Parser parser) | |
657 { | |
658 return nSpecifiedAtts; | |
659 } | |
660 | |
661 void XML_SetElementHandler(XML_Parser parser, | |
662 XML_StartElementHandler start, | |
663 XML_EndElementHandler end) | |
664 { | |
665 startElementHandler = start; | |
666 endElementHandler = end; | |
667 } | |
668 | |
669 void XML_SetCharacterDataHandler(XML_Parser parser, | |
670 XML_CharacterDataHandler handler) | |
671 { | |
672 characterDataHandler = handler; | |
673 } | |
674 | |
675 void XML_SetProcessingInstructionHandler(XML_Parser parser, | |
676 XML_ProcessingInstructionHandler handler) | |
677 { | |
678 processingInstructionHandler = handler; | |
679 } | |
680 | |
681 void XML_SetCommentHandler(XML_Parser parser, | |
682 XML_CommentHandler handler) | |
683 { | |
684 commentHandler = handler; | |
685 } | |
686 | |
687 void XML_SetCdataSectionHandler(XML_Parser parser, | |
688 XML_StartCdataSectionHandler start, | |
689 XML_EndCdataSectionHandler end) | |
690 { | |
691 startCdataSectionHandler = start; | |
692 endCdataSectionHandler = end; | |
693 } | |
694 | |
695 void XML_SetDefaultHandler(XML_Parser parser, | |
696 XML_DefaultHandler handler) | |
697 { | |
698 defaultHandler = handler; | |
699 defaultExpandInternalEntities = 0; | |
700 } | |
701 | |
702 void XML_SetDefaultHandlerExpand(XML_Parser parser, | |
703 XML_DefaultHandler handler) | |
704 { | |
705 defaultHandler = handler; | |
706 defaultExpandInternalEntities = 1; | |
707 } | |
708 | |
709 void XML_SetUnparsedEntityDeclHandler(XML_Parser parser, | |
710 XML_UnparsedEntityDeclHandler handler) | |
711 { | |
712 unparsedEntityDeclHandler = handler; | |
713 } | |
714 | |
715 void XML_SetNotationDeclHandler(XML_Parser parser, | |
716 XML_NotationDeclHandler handler) | |
717 { | |
718 notationDeclHandler = handler; | |
719 } | |
720 | |
721 void XML_SetNamespaceDeclHandler(XML_Parser parser, | |
722 XML_StartNamespaceDeclHandler start, | |
723 XML_EndNamespaceDeclHandler end) | |
724 { | |
725 startNamespaceDeclHandler = start; | |
726 endNamespaceDeclHandler = end; | |
727 } | |
728 | |
729 void XML_SetNotStandaloneHandler(XML_Parser parser, | |
730 XML_NotStandaloneHandler handler) | |
731 { | |
732 notStandaloneHandler = handler; | |
733 } | |
734 | |
735 void XML_SetExternalEntityRefHandler(XML_Parser parser, | |
736 XML_ExternalEntityRefHandler handler) | |
737 { | |
738 externalEntityRefHandler = handler; | |
739 } | |
740 | |
741 void XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) | |
742 { | |
743 if (arg) | |
744 externalEntityRefHandlerArg = arg; | |
745 else | |
746 externalEntityRefHandlerArg = parser; | |
747 } | |
748 | |
749 void XML_SetUnknownEncodingHandler(XML_Parser parser, | |
750 XML_UnknownEncodingHandler handler, | |
751 void *data) | |
752 { | |
753 unknownEncodingHandler = handler; | |
754 unknownEncodingHandlerData = data; | |
755 } | |
756 | |
757 int XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) | |
758 { | |
759 if (len == 0) { | |
760 if (!isFinal) | |
761 return 1; | |
762 positionPtr = bufferPtr; | |
763 errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0); | |
764 if (errorCode == XML_ERROR_NONE) | |
765 return 1; | |
766 eventEndPtr = eventPtr; | |
767 return 0; | |
768 } | |
769 else if (bufferPtr == bufferEnd) { | |
770 const char *end; | |
771 int nLeftOver; | |
772 parseEndByteIndex += len; | |
773 positionPtr = s; | |
774 if (isFinal) { | |
775 errorCode = processor(parser, s, parseEndPtr = s + len, 0); | |
776 if (errorCode == XML_ERROR_NONE) | |
777 return 1; | |
778 eventEndPtr = eventPtr; | |
779 return 0; | |
780 } | |
781 errorCode = processor(parser, s, parseEndPtr = s + len, &end); | |
782 if (errorCode != XML_ERROR_NONE) { | |
783 eventEndPtr = eventPtr; | |
784 return 0; | |
785 } | |
786 XmlUpdatePosition(encoding, positionPtr, end, &position); | |
787 nLeftOver = s + len - end; | |
788 if (nLeftOver) { | |
789 if (buffer == 0 || nLeftOver > bufferLim - buffer) { | |
790 /* FIXME avoid integer overflow */ | |
791 buffer = buffer == 0 ? malloc(len * 2) : realloc(buffer, len * 2); | |
792 if (!buffer) { | |
793 errorCode = XML_ERROR_NO_MEMORY; | |
794 eventPtr = eventEndPtr = 0; | |
795 return 0; | |
796 } | |
797 bufferLim = buffer + len * 2; | |
798 } | |
799 memcpy(buffer, end, nLeftOver); | |
800 bufferPtr = buffer; | |
801 bufferEnd = buffer + nLeftOver; | |
802 } | |
803 return 1; | |
804 } | |
805 else { | |
806 memcpy(XML_GetBuffer(parser, len), s, len); | |
807 return XML_ParseBuffer(parser, len, isFinal); | |
808 } | |
809 } | |
810 | |
811 int XML_ParseBuffer(XML_Parser parser, int len, int isFinal) | |
812 { | |
813 const char *start = bufferPtr; | |
814 positionPtr = start; | |
815 bufferEnd += len; | |
816 parseEndByteIndex += len; | |
817 errorCode = processor(parser, start, parseEndPtr = bufferEnd, | |
818 isFinal ? (const char **)0 : &bufferPtr); | |
819 if (errorCode == XML_ERROR_NONE) { | |
820 if (!isFinal) | |
821 XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position); | |
822 return 1; | |
823 } | |
824 else { | |
825 eventEndPtr = eventPtr; | |
826 return 0; | |
827 } | |
828 } | |
829 | |
830 void *XML_GetBuffer(XML_Parser parser, int len) | |
831 { | |
832 if (len > bufferLim - bufferEnd) { | |
833 /* FIXME avoid integer overflow */ | |
834 int neededSize = len + (bufferEnd - bufferPtr); | |
835 if (neededSize <= bufferLim - buffer) { | |
836 memmove(buffer, bufferPtr, bufferEnd - bufferPtr); | |
837 bufferEnd = buffer + (bufferEnd - bufferPtr); | |
838 bufferPtr = buffer; | |
839 } | |
840 else { | |
841 char *newBuf; | |
842 int bufferSize = bufferLim - bufferPtr; | |
843 if (bufferSize == 0) | |
844 bufferSize = INIT_BUFFER_SIZE; | |
845 do { | |
846 bufferSize *= 2; | |
847 } while (bufferSize < neededSize); | |
848 newBuf = malloc(bufferSize); | |
849 if (newBuf == 0) { | |
850 errorCode = XML_ERROR_NO_MEMORY; | |
851 return 0; | |
852 } | |
853 bufferLim = newBuf + bufferSize; | |
854 if (bufferPtr) { | |
855 memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr); | |
856 free(buffer); | |
857 } | |
858 bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
859 bufferPtr = buffer = newBuf; | |
860 } | |
861 } | |
862 return bufferEnd; | |
863 } | |
864 | |
865 enum XML_Error XML_GetErrorCode(XML_Parser parser) | |
866 { | |
867 return errorCode; | |
868 } | |
869 | |
870 long XML_GetCurrentByteIndex(XML_Parser parser) | |
871 { | |
872 if (eventPtr) | |
873 return parseEndByteIndex - (parseEndPtr - eventPtr); | |
874 return -1; | |
875 } | |
876 | |
877 int XML_GetCurrentByteCount(XML_Parser parser) | |
878 { | |
879 if (eventEndPtr && eventPtr) | |
880 return eventEndPtr - eventPtr; | |
881 return 0; | |
882 } | |
883 | |
884 int XML_GetCurrentLineNumber(XML_Parser parser) | |
885 { | |
886 if (eventPtr) { | |
887 XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
888 positionPtr = eventPtr; | |
889 } | |
890 return position.lineNumber + 1; | |
891 } | |
892 | |
893 int XML_GetCurrentColumnNumber(XML_Parser parser) | |
894 { | |
895 if (eventPtr) { | |
896 XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
897 positionPtr = eventPtr; | |
898 } | |
899 return position.columnNumber; | |
900 } | |
901 | |
902 void XML_DefaultCurrent(XML_Parser parser) | |
903 { | |
904 if (defaultHandler) { | |
905 if (openInternalEntities) | |
906 reportDefault(parser, | |
907 ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(), | |
908 openInternalEntities->internalEventPtr, | |
909 openInternalEntities->internalEventEndPtr); | |
910 else | |
911 reportDefault(parser, encoding, eventPtr, eventEndPtr); | |
912 } | |
913 } | |
914 | |
915 const XML_LChar *XML_ErrorString(int code) | |
916 { | |
917 static const XML_LChar *message[] = { | |
918 0, | |
919 XML_T("out of memory"), | |
920 XML_T("syntax error"), | |
921 XML_T("no element found"), | |
922 XML_T("not well-formed"), | |
923 XML_T("unclosed token"), | |
924 XML_T("unclosed token"), | |
925 XML_T("mismatched tag"), | |
926 XML_T("duplicate attribute"), | |
927 XML_T("junk after document element"), | |
928 XML_T("illegal parameter entity reference"), | |
929 XML_T("undefined entity"), | |
930 XML_T("recursive entity reference"), | |
931 XML_T("asynchronous entity"), | |
932 XML_T("reference to invalid character number"), | |
933 XML_T("reference to binary entity"), | |
934 XML_T("reference to external entity in attribute"), | |
935 XML_T("xml processing instruction not at start of external entity"), | |
936 XML_T("unknown encoding"), | |
937 XML_T("encoding specified in XML declaration is incorrect"), | |
938 XML_T("unclosed CDATA section"), | |
939 XML_T("error in processing external entity reference"), | |
940 XML_T("document is not standalone") | |
941 }; | |
883
0aa9015f06df
Remove some more libjabber warnings
Mikael Berthe <mikael@lilotux.net>
parents:
235
diff
changeset
|
942 if (code > 0 && code < (int)(sizeof(message)/sizeof(message[0]))) |
25 | 943 return message[code]; |
944 return 0; | |
945 } | |
946 | |
947 static | |
948 enum XML_Error contentProcessor(XML_Parser parser, | |
949 const char *start, | |
950 const char *end, | |
951 const char **endPtr) | |
952 { | |
953 return doContent(parser, 0, encoding, start, end, endPtr); | |
954 } | |
955 | |
956 static | |
957 enum XML_Error externalEntityInitProcessor(XML_Parser parser, | |
958 const char *start, | |
959 const char *end, | |
960 const char **endPtr) | |
961 { | |
962 enum XML_Error result = initializeEncoding(parser); | |
963 if (result != XML_ERROR_NONE) | |
964 return result; | |
965 processor = externalEntityInitProcessor2; | |
966 return externalEntityInitProcessor2(parser, start, end, endPtr); | |
967 } | |
968 | |
969 static | |
970 enum XML_Error externalEntityInitProcessor2(XML_Parser parser, | |
971 const char *start, | |
972 const char *end, | |
973 const char **endPtr) | |
974 { | |
975 const char *next; | |
976 int tok = XmlContentTok(encoding, start, end, &next); | |
977 switch (tok) { | |
978 case XML_TOK_BOM: | |
979 start = next; | |
980 break; | |
981 case XML_TOK_PARTIAL: | |
982 if (endPtr) { | |
983 *endPtr = start; | |
984 return XML_ERROR_NONE; | |
985 } | |
986 eventPtr = start; | |
987 return XML_ERROR_UNCLOSED_TOKEN; | |
988 case XML_TOK_PARTIAL_CHAR: | |
989 if (endPtr) { | |
990 *endPtr = start; | |
991 return XML_ERROR_NONE; | |
992 } | |
993 eventPtr = start; | |
994 return XML_ERROR_PARTIAL_CHAR; | |
995 } | |
996 processor = externalEntityInitProcessor3; | |
997 return externalEntityInitProcessor3(parser, start, end, endPtr); | |
998 } | |
999 | |
1000 static | |
1001 enum XML_Error externalEntityInitProcessor3(XML_Parser parser, | |
1002 const char *start, | |
1003 const char *end, | |
1004 const char **endPtr) | |
1005 { | |
1006 const char *next; | |
1007 int tok = XmlContentTok(encoding, start, end, &next); | |
1008 switch (tok) { | |
1009 case XML_TOK_XML_DECL: | |
1010 { | |
1011 enum XML_Error result = processXmlDecl(parser, 1, start, next); | |
1012 if (result != XML_ERROR_NONE) | |
1013 return result; | |
1014 start = next; | |
1015 } | |
1016 break; | |
1017 case XML_TOK_PARTIAL: | |
1018 if (endPtr) { | |
1019 *endPtr = start; | |
1020 return XML_ERROR_NONE; | |
1021 } | |
1022 eventPtr = start; | |
1023 return XML_ERROR_UNCLOSED_TOKEN; | |
1024 case XML_TOK_PARTIAL_CHAR: | |
1025 if (endPtr) { | |
1026 *endPtr = start; | |
1027 return XML_ERROR_NONE; | |
1028 } | |
1029 eventPtr = start; | |
1030 return XML_ERROR_PARTIAL_CHAR; | |
1031 } | |
1032 processor = externalEntityContentProcessor; | |
1033 tagLevel = 1; | |
1034 return doContent(parser, 1, encoding, start, end, endPtr); | |
1035 } | |
1036 | |
1037 static | |
1038 enum XML_Error externalEntityContentProcessor(XML_Parser parser, | |
1039 const char *start, | |
1040 const char *end, | |
1041 const char **endPtr) | |
1042 { | |
1043 return doContent(parser, 1, encoding, start, end, endPtr); | |
1044 } | |
1045 | |
1046 static enum XML_Error | |
1047 doContent(XML_Parser parser, | |
1048 int startTagLevel, | |
1049 const ENCODING *enc, | |
1050 const char *s, | |
1051 const char *end, | |
1052 const char **nextPtr) | |
1053 { | |
1054 const ENCODING *internalEnc = ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(); | |
1055 const char **eventPP; | |
1056 const char **eventEndPP; | |
1057 if (enc == encoding) { | |
1058 eventPP = &eventPtr; | |
1059 eventEndPP = &eventEndPtr; | |
1060 } | |
1061 else { | |
1062 eventPP = &(openInternalEntities->internalEventPtr); | |
1063 eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
1064 } | |
1065 *eventPP = s; | |
1066 for (;;) { | |
1067 const char *next = s; /* XmlContentTok doesn't always set the last arg */ | |
1068 int tok = XmlContentTok(enc, s, end, &next); | |
1069 *eventEndPP = next; | |
1070 switch (tok) { | |
1071 case XML_TOK_TRAILING_CR: | |
1072 if (nextPtr) { | |
1073 *nextPtr = s; | |
1074 return XML_ERROR_NONE; | |
1075 } | |
1076 *eventEndPP = end; | |
1077 if (characterDataHandler) { | |
1078 XML_Char c = 0xA; | |
1079 characterDataHandler(handlerArg, &c, 1); | |
1080 } | |
1081 else if (defaultHandler) | |
1082 reportDefault(parser, enc, s, end); | |
1083 if (startTagLevel == 0) | |
1084 return XML_ERROR_NO_ELEMENTS; | |
1085 if (tagLevel != startTagLevel) | |
1086 return XML_ERROR_ASYNC_ENTITY; | |
1087 return XML_ERROR_NONE; | |
1088 case XML_TOK_NONE: | |
1089 if (nextPtr) { | |
1090 *nextPtr = s; | |
1091 return XML_ERROR_NONE; | |
1092 } | |
1093 if (startTagLevel > 0) { | |
1094 if (tagLevel != startTagLevel) | |
1095 return XML_ERROR_ASYNC_ENTITY; | |
1096 return XML_ERROR_NONE; | |
1097 } | |
1098 return XML_ERROR_NO_ELEMENTS; | |
1099 case XML_TOK_INVALID: | |
1100 *eventPP = next; | |
1101 return XML_ERROR_INVALID_TOKEN; | |
1102 case XML_TOK_PARTIAL: | |
1103 if (nextPtr) { | |
1104 *nextPtr = s; | |
1105 return XML_ERROR_NONE; | |
1106 } | |
1107 return XML_ERROR_UNCLOSED_TOKEN; | |
1108 case XML_TOK_PARTIAL_CHAR: | |
1109 if (nextPtr) { | |
1110 *nextPtr = s; | |
1111 return XML_ERROR_NONE; | |
1112 } | |
1113 return XML_ERROR_PARTIAL_CHAR; | |
1114 case XML_TOK_ENTITY_REF: | |
1115 { | |
1116 const XML_Char *name; | |
1117 ENTITY *entity; | |
1118 XML_Char ch = XmlPredefinedEntityName(enc, | |
1119 s + enc->minBytesPerChar, | |
1120 next - enc->minBytesPerChar); | |
1121 if (ch) { | |
1122 if (characterDataHandler) | |
1123 characterDataHandler(handlerArg, &ch, 1); | |
1124 else if (defaultHandler) | |
1125 reportDefault(parser, enc, s, next); | |
1126 break; | |
1127 } | |
1128 name = poolStoreString(&dtd.pool, enc, | |
1129 s + enc->minBytesPerChar, | |
1130 next - enc->minBytesPerChar); | |
1131 if (!name) | |
1132 return XML_ERROR_NO_MEMORY; | |
1133 entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); | |
1134 poolDiscard(&dtd.pool); | |
1135 if (!entity) { | |
1136 if (dtd.complete || dtd.standalone) | |
1137 return XML_ERROR_UNDEFINED_ENTITY; | |
1138 if (defaultHandler) | |
1139 reportDefault(parser, enc, s, next); | |
1140 break; | |
1141 } | |
1142 if (entity->open) | |
1143 return XML_ERROR_RECURSIVE_ENTITY_REF; | |
1144 if (entity->notation) | |
1145 return XML_ERROR_BINARY_ENTITY_REF; | |
1146 if (entity) { | |
1147 if (entity->textPtr) { | |
1148 enum XML_Error result; | |
1149 OPEN_INTERNAL_ENTITY openEntity; | |
1150 if (defaultHandler && !defaultExpandInternalEntities) { | |
1151 reportDefault(parser, enc, s, next); | |
1152 break; | |
1153 } | |
1154 entity->open = 1; | |
1155 openEntity.next = openInternalEntities; | |
1156 openInternalEntities = &openEntity; | |
1157 openEntity.entity = entity; | |
1158 openEntity.internalEventPtr = 0; | |
1159 openEntity.internalEventEndPtr = 0; | |
1160 result = doContent(parser, | |
1161 tagLevel, | |
1162 internalEnc, | |
1163 (char *)entity->textPtr, | |
1164 (char *)(entity->textPtr + entity->textLen), | |
1165 0); | |
1166 entity->open = 0; | |
1167 openInternalEntities = openEntity.next; | |
1168 if (result) | |
1169 return result; | |
1170 } | |
1171 else if (externalEntityRefHandler) { | |
1172 const XML_Char *context; | |
1173 entity->open = 1; | |
1174 context = getContext(parser); | |
1175 entity->open = 0; | |
1176 if (!context) | |
1177 return XML_ERROR_NO_MEMORY; | |
1178 if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
1179 context, | |
1180 dtd.base, | |
1181 entity->systemId, | |
1182 entity->publicId)) | |
1183 return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
1184 poolDiscard(&tempPool); | |
1185 } | |
1186 else if (defaultHandler) | |
1187 reportDefault(parser, enc, s, next); | |
1188 } | |
1189 break; | |
1190 } | |
1191 case XML_TOK_START_TAG_WITH_ATTS: | |
1192 if (!startElementHandler) { | |
1193 enum XML_Error result = storeAtts(parser, enc, s, 0, 0); | |
1194 if (result) | |
1195 return result; | |
1196 } | |
1197 /* fall through */ | |
1198 case XML_TOK_START_TAG_NO_ATTS: | |
1199 { | |
1200 TAG *tag; | |
1201 if (freeTagList) { | |
1202 tag = freeTagList; | |
1203 freeTagList = freeTagList->parent; | |
1204 } | |
1205 else { | |
1206 tag = malloc(sizeof(TAG)); | |
1207 if (!tag) | |
1208 return XML_ERROR_NO_MEMORY; | |
1209 tag->buf = malloc(INIT_TAG_BUF_SIZE); | |
1210 if (!tag->buf) | |
1211 return XML_ERROR_NO_MEMORY; | |
1212 tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; | |
1213 } | |
1214 tag->bindings = 0; | |
1215 tag->parent = tagStack; | |
1216 tagStack = tag; | |
1217 tag->name.localPart = 0; | |
1218 tag->rawName = s + enc->minBytesPerChar; | |
1219 tag->rawNameLength = XmlNameLength(enc, tag->rawName); | |
1220 if (nextPtr) { | |
1221 /* Need to guarantee that: | |
1222 tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)) <= tag->bufEnd - sizeof(XML_Char) */ | |
1223 if (tag->rawNameLength + (int)(sizeof(XML_Char) - 1) + (int)sizeof(XML_Char) > tag->bufEnd - tag->buf) { | |
1224 int bufSize = tag->rawNameLength * 4; | |
1225 bufSize = ROUND_UP(bufSize, sizeof(XML_Char)); | |
1226 tag->buf = realloc(tag->buf, bufSize); | |
1227 if (!tag->buf) | |
1228 return XML_ERROR_NO_MEMORY; | |
1229 tag->bufEnd = tag->buf + bufSize; | |
1230 } | |
1231 memcpy(tag->buf, tag->rawName, tag->rawNameLength); | |
1232 tag->rawName = tag->buf; | |
1233 } | |
1234 ++tagLevel; | |
1235 if (startElementHandler) { | |
1236 enum XML_Error result; | |
1237 XML_Char *toPtr; | |
1238 for (;;) { | |
1239 const char *rawNameEnd = tag->rawName + tag->rawNameLength; | |
1240 const char *fromPtr = tag->rawName; | |
1241 int bufSize; | |
1242 if (nextPtr) | |
1243 toPtr = (XML_Char *)(tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char))); | |
1244 else | |
1245 toPtr = (XML_Char *)tag->buf; | |
1246 tag->name.str = toPtr; | |
1247 XmlConvert(enc, | |
1248 &fromPtr, rawNameEnd, | |
1249 (ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1); | |
1250 if (fromPtr == rawNameEnd) | |
1251 break; | |
1252 bufSize = (tag->bufEnd - tag->buf) << 1; | |
1253 tag->buf = realloc(tag->buf, bufSize); | |
1254 if (!tag->buf) | |
1255 return XML_ERROR_NO_MEMORY; | |
1256 tag->bufEnd = tag->buf + bufSize; | |
1257 if (nextPtr) | |
1258 tag->rawName = tag->buf; | |
1259 } | |
1260 *toPtr = XML_T('\0'); | |
1261 result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings)); | |
1262 if (result) | |
1263 return result; | |
1264 startElementHandler(handlerArg, tag->name.str, (const XML_Char **)atts); | |
1265 poolClear(&tempPool); | |
1266 } | |
1267 else { | |
1268 tag->name.str = 0; | |
1269 if (defaultHandler) | |
1270 reportDefault(parser, enc, s, next); | |
1271 } | |
1272 break; | |
1273 } | |
1274 case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: | |
1275 if (!startElementHandler) { | |
1276 enum XML_Error result = storeAtts(parser, enc, s, 0, 0); | |
1277 if (result) | |
1278 return result; | |
1279 } | |
1280 /* fall through */ | |
1281 case XML_TOK_EMPTY_ELEMENT_NO_ATTS: | |
1282 if (startElementHandler || endElementHandler) { | |
1283 const char *rawName = s + enc->minBytesPerChar; | |
1284 enum XML_Error result; | |
1285 BINDING *bindings = 0; | |
1286 TAG_NAME name; | |
1287 name.str = poolStoreString(&tempPool, enc, rawName, | |
1288 rawName + XmlNameLength(enc, rawName)); | |
1289 if (!name.str) | |
1290 return XML_ERROR_NO_MEMORY; | |
1291 poolFinish(&tempPool); | |
1292 result = storeAtts(parser, enc, s, &name, &bindings); | |
1293 if (result) | |
1294 return result; | |
1295 poolFinish(&tempPool); | |
1296 if (startElementHandler) | |
1297 startElementHandler(handlerArg, name.str, (const XML_Char **)atts); | |
1298 if (endElementHandler) { | |
1299 if (startElementHandler) | |
1300 *eventPP = *eventEndPP; | |
1301 endElementHandler(handlerArg, name.str); | |
1302 } | |
1303 poolClear(&tempPool); | |
1304 while (bindings) { | |
1305 BINDING *b = bindings; | |
1306 if (endNamespaceDeclHandler) | |
1307 endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
1308 bindings = bindings->nextTagBinding; | |
1309 b->nextTagBinding = freeBindingList; | |
1310 freeBindingList = b; | |
1311 b->prefix->binding = b->prevPrefixBinding; | |
1312 } | |
1313 } | |
1314 else if (defaultHandler) | |
1315 reportDefault(parser, enc, s, next); | |
1316 if (tagLevel == 0) | |
1317 return epilogProcessor(parser, next, end, nextPtr); | |
1318 break; | |
1319 case XML_TOK_END_TAG: | |
1320 if (tagLevel == startTagLevel) | |
1321 return XML_ERROR_ASYNC_ENTITY; | |
1322 else { | |
1323 int len; | |
1324 const char *rawName; | |
1325 TAG *tag = tagStack; | |
1326 tagStack = tag->parent; | |
1327 tag->parent = freeTagList; | |
1328 freeTagList = tag; | |
1329 rawName = s + enc->minBytesPerChar*2; | |
1330 len = XmlNameLength(enc, rawName); | |
1331 if (len != tag->rawNameLength | |
1332 || memcmp(tag->rawName, rawName, len) != 0) { | |
1333 *eventPP = rawName; | |
1334 return XML_ERROR_TAG_MISMATCH; | |
1335 } | |
1336 --tagLevel; | |
1337 if (endElementHandler && tag->name.str) { | |
1338 if (tag->name.localPart) { | |
1339 XML_Char *to = (XML_Char *)tag->name.str + tag->name.uriLen; | |
1340 const XML_Char *from = tag->name.localPart; | |
1341 while ((*to++ = *from++) != 0) | |
1342 ; | |
1343 } | |
1344 endElementHandler(handlerArg, tag->name.str); | |
1345 } | |
1346 else if (defaultHandler) | |
1347 reportDefault(parser, enc, s, next); | |
1348 while (tag->bindings) { | |
1349 BINDING *b = tag->bindings; | |
1350 if (endNamespaceDeclHandler) | |
1351 endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
1352 tag->bindings = tag->bindings->nextTagBinding; | |
1353 b->nextTagBinding = freeBindingList; | |
1354 freeBindingList = b; | |
1355 b->prefix->binding = b->prevPrefixBinding; | |
1356 } | |
1357 if (tagLevel == 0) | |
1358 return epilogProcessor(parser, next, end, nextPtr); | |
1359 } | |
1360 break; | |
1361 case XML_TOK_CHAR_REF: | |
1362 { | |
1363 int n = XmlCharRefNumber(enc, s); | |
1364 if (n < 0) | |
1365 return XML_ERROR_BAD_CHAR_REF; | |
1366 if (characterDataHandler) { | |
1367 XML_Char buf[XML_ENCODE_MAX]; | |
1368 characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf)); | |
1369 } | |
1370 else if (defaultHandler) | |
1371 reportDefault(parser, enc, s, next); | |
1372 } | |
1373 break; | |
1374 case XML_TOK_XML_DECL: | |
1375 return XML_ERROR_MISPLACED_XML_PI; | |
1376 case XML_TOK_DATA_NEWLINE: | |
1377 if (characterDataHandler) { | |
1378 XML_Char c = 0xA; | |
1379 characterDataHandler(handlerArg, &c, 1); | |
1380 } | |
1381 else if (defaultHandler) | |
1382 reportDefault(parser, enc, s, next); | |
1383 break; | |
1384 case XML_TOK_CDATA_SECT_OPEN: | |
1385 { | |
1386 enum XML_Error result; | |
1387 if (startCdataSectionHandler) | |
1388 startCdataSectionHandler(handlerArg); | |
1389 #if 0 | |
1390 /* Suppose you doing a transformation on a document that involves | |
1391 changing only the character data. You set up a defaultHandler | |
1392 and a characterDataHandler. The defaultHandler simply copies | |
1393 characters through. The characterDataHandler does the transformation | |
1394 and writes the characters out escaping them as necessary. This case | |
1395 will fail to work if we leave out the following two lines (because & | |
1396 and < inside CDATA sections will be incorrectly escaped). | |
1397 | |
1398 However, now we have a start/endCdataSectionHandler, so it seems | |
1399 easier to let the user deal with this. */ | |
1400 | |
1401 else if (characterDataHandler) | |
1402 characterDataHandler(handlerArg, dataBuf, 0); | |
1403 #endif | |
1404 else if (defaultHandler) | |
1405 reportDefault(parser, enc, s, next); | |
1406 result = doCdataSection(parser, enc, &next, end, nextPtr); | |
1407 if (!next) { | |
1408 processor = cdataSectionProcessor; | |
1409 return result; | |
1410 } | |
1411 } | |
1412 break; | |
1413 case XML_TOK_TRAILING_RSQB: | |
1414 if (nextPtr) { | |
1415 *nextPtr = s; | |
1416 return XML_ERROR_NONE; | |
1417 } | |
1418 if (characterDataHandler) { | |
1419 if (MUST_CONVERT(enc, s)) { | |
1420 ICHAR *dataPtr = (ICHAR *)dataBuf; | |
1421 XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
1422 characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
1423 } | |
1424 else | |
1425 characterDataHandler(handlerArg, | |
1426 (XML_Char *)s, | |
1427 (XML_Char *)end - (XML_Char *)s); | |
1428 } | |
1429 else if (defaultHandler) | |
1430 reportDefault(parser, enc, s, end); | |
1431 if (startTagLevel == 0) { | |
1432 *eventPP = end; | |
1433 return XML_ERROR_NO_ELEMENTS; | |
1434 } | |
1435 if (tagLevel != startTagLevel) { | |
1436 *eventPP = end; | |
1437 return XML_ERROR_ASYNC_ENTITY; | |
1438 } | |
1439 return XML_ERROR_NONE; | |
1440 case XML_TOK_DATA_CHARS: | |
1441 if (characterDataHandler) { | |
1442 if (MUST_CONVERT(enc, s)) { | |
1443 for (;;) { | |
1444 ICHAR *dataPtr = (ICHAR *)dataBuf; | |
1445 XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
1446 *eventEndPP = s; | |
1447 characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
1448 if (s == next) | |
1449 break; | |
1450 *eventPP = s; | |
1451 } | |
1452 } | |
1453 else | |
1454 characterDataHandler(handlerArg, | |
1455 (XML_Char *)s, | |
1456 (XML_Char *)next - (XML_Char *)s); | |
1457 } | |
1458 else if (defaultHandler) | |
1459 reportDefault(parser, enc, s, next); | |
1460 break; | |
1461 case XML_TOK_PI: | |
1462 if (!reportProcessingInstruction(parser, enc, s, next)) | |
1463 return XML_ERROR_NO_MEMORY; | |
1464 break; | |
1465 case XML_TOK_COMMENT: | |
1466 if (!reportComment(parser, enc, s, next)) | |
1467 return XML_ERROR_NO_MEMORY; | |
1468 break; | |
1469 default: | |
1470 if (defaultHandler) | |
1471 reportDefault(parser, enc, s, next); | |
1472 break; | |
1473 } | |
1474 *eventPP = s = next; | |
1475 } | |
1476 /* not reached */ | |
1477 } | |
1478 | |
1479 /* If tagNamePtr is non-null, build a real list of attributes, | |
1480 otherwise just check the attributes for well-formedness. */ | |
1481 | |
1482 static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *enc, | |
1483 const char *s, TAG_NAME *tagNamePtr, | |
1484 BINDING **bindingsPtr) | |
1485 { | |
1486 ELEMENT_TYPE *elementType = 0; | |
1487 int nDefaultAtts = 0; | |
1488 const XML_Char **appAtts; | |
1489 int attIndex = 0; | |
1490 int i; | |
1491 int n; | |
1492 int nPrefixes = 0; | |
1493 BINDING *binding; | |
1494 const XML_Char *localPart; | |
1495 | |
1496 if (tagNamePtr) { | |
1497 elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, 0); | |
1498 if (!elementType) { | |
1499 tagNamePtr->str = poolCopyString(&dtd.pool, tagNamePtr->str); | |
1500 if (!tagNamePtr->str) | |
1501 return XML_ERROR_NO_MEMORY; | |
1502 elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, sizeof(ELEMENT_TYPE)); | |
1503 if (!elementType) | |
1504 return XML_ERROR_NO_MEMORY; | |
1505 if (ns && !setElementTypePrefix(parser, elementType)) | |
1506 return XML_ERROR_NO_MEMORY; | |
1507 } | |
1508 nDefaultAtts = elementType->nDefaultAtts; | |
1509 } | |
1510 n = XmlGetAttributes(enc, s, attsSize, atts); | |
1511 if (n + nDefaultAtts > attsSize) { | |
1512 int oldAttsSize = attsSize; | |
1513 attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; | |
1514 atts = realloc((void *)atts, attsSize * sizeof(ATTRIBUTE)); | |
1515 if (!atts) | |
1516 return XML_ERROR_NO_MEMORY; | |
1517 if (n > oldAttsSize) | |
1518 XmlGetAttributes(enc, s, n, atts); | |
1519 } | |
1520 appAtts = (const XML_Char **)atts; | |
1521 for (i = 0; i < n; i++) { | |
1522 ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name, | |
1523 atts[i].name | |
1524 + XmlNameLength(enc, atts[i].name)); | |
1525 if (!attId) | |
1526 return XML_ERROR_NO_MEMORY; | |
1527 if ((attId->name)[-1]) { | |
1528 if (enc == encoding) | |
1529 eventPtr = atts[i].name; | |
1530 return XML_ERROR_DUPLICATE_ATTRIBUTE; | |
1531 } | |
1532 (attId->name)[-1] = 1; | |
1533 appAtts[attIndex++] = attId->name; | |
1534 if (!atts[i].normalized) { | |
1535 enum XML_Error result; | |
1536 int isCdata = 1; | |
1537 | |
1538 if (attId->maybeTokenized) { | |
1539 int j; | |
1540 for (j = 0; j < nDefaultAtts; j++) { | |
1541 if (attId == elementType->defaultAtts[j].id) { | |
1542 isCdata = elementType->defaultAtts[j].isCdata; | |
1543 break; | |
1544 } | |
1545 } | |
1546 } | |
1547 | |
1548 result = storeAttributeValue(parser, enc, isCdata, | |
1549 atts[i].valuePtr, atts[i].valueEnd, | |
1550 &tempPool); | |
1551 if (result) | |
1552 return result; | |
1553 if (tagNamePtr) { | |
1554 appAtts[attIndex] = poolStart(&tempPool); | |
1555 poolFinish(&tempPool); | |
1556 } | |
1557 else | |
1558 poolDiscard(&tempPool); | |
1559 } | |
1560 else if (tagNamePtr) { | |
1561 appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr, atts[i].valueEnd); | |
1562 if (appAtts[attIndex] == 0) | |
1563 return XML_ERROR_NO_MEMORY; | |
1564 poolFinish(&tempPool); | |
1565 } | |
1566 if (attId->prefix && tagNamePtr) { | |
1567 if (attId->xmlns) { | |
1568 if (!addBinding(parser, attId->prefix, attId, appAtts[attIndex], bindingsPtr)) | |
1569 return XML_ERROR_NO_MEMORY; | |
1570 --attIndex; | |
1571 } | |
1572 else { | |
1573 attIndex++; | |
1574 nPrefixes++; | |
1575 (attId->name)[-1] = 2; | |
1576 } | |
1577 } | |
1578 else | |
1579 attIndex++; | |
1580 } | |
1581 nSpecifiedAtts = attIndex; | |
1582 if (tagNamePtr) { | |
1583 int j; | |
1584 for (j = 0; j < nDefaultAtts; j++) { | |
1585 const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + j; | |
1586 if (!(da->id->name)[-1] && da->value) { | |
1587 if (da->id->prefix) { | |
1588 if (da->id->xmlns) { | |
1589 if (!addBinding(parser, da->id->prefix, da->id, da->value, bindingsPtr)) | |
1590 return XML_ERROR_NO_MEMORY; | |
1591 } | |
1592 else { | |
1593 (da->id->name)[-1] = 2; | |
1594 nPrefixes++; | |
1595 appAtts[attIndex++] = da->id->name; | |
1596 appAtts[attIndex++] = da->value; | |
1597 } | |
1598 } | |
1599 else { | |
1600 (da->id->name)[-1] = 1; | |
1601 appAtts[attIndex++] = da->id->name; | |
1602 appAtts[attIndex++] = da->value; | |
1603 } | |
1604 } | |
1605 } | |
1606 appAtts[attIndex] = 0; | |
1607 } | |
1608 i = 0; | |
1609 if (nPrefixes) { | |
1610 for (; i < attIndex; i += 2) { | |
1611 if (appAtts[i][-1] == 2) { | |
1612 ATTRIBUTE_ID *id; | |
1613 ((XML_Char *)(appAtts[i]))[-1] = 0; | |
1614 id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, appAtts[i], 0); | |
1615 if (id->prefix->binding) { | |
1616 int j; | |
1617 const BINDING *b = id->prefix->binding; | |
1618 const XML_Char *s = appAtts[i]; | |
1619 for (j = 0; j < b->uriLen; j++) { | |
1620 if (!poolAppendChar(&tempPool, b->uri[j])) | |
1621 return XML_ERROR_NO_MEMORY; | |
1622 } | |
1623 while (*s++ != ':') | |
1624 ; | |
1625 do { | |
1626 if (!poolAppendChar(&tempPool, *s)) | |
1627 return XML_ERROR_NO_MEMORY; | |
1628 } while (*s++); | |
1629 appAtts[i] = poolStart(&tempPool); | |
1630 poolFinish(&tempPool); | |
1631 } | |
1632 if (!--nPrefixes) | |
1633 break; | |
1634 } | |
1635 else | |
1636 ((XML_Char *)(appAtts[i]))[-1] = 0; | |
1637 } | |
1638 } | |
1639 for (; i < attIndex; i += 2) | |
1640 ((XML_Char *)(appAtts[i]))[-1] = 0; | |
1641 if (!tagNamePtr) | |
1642 return XML_ERROR_NONE; | |
1643 for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) | |
1644 binding->attId->name[-1] = 0; | |
1645 if (elementType->prefix) { | |
1646 binding = elementType->prefix->binding; | |
1647 if (!binding) | |
1648 return XML_ERROR_NONE; | |
1649 localPart = tagNamePtr->str; | |
1650 while (*localPart++ != XML_T(':')) | |
1651 ; | |
1652 } | |
1653 else if (dtd.defaultPrefix.binding) { | |
1654 binding = dtd.defaultPrefix.binding; | |
1655 localPart = tagNamePtr->str; | |
1656 } | |
1657 else | |
1658 return XML_ERROR_NONE; | |
1659 tagNamePtr->localPart = localPart; | |
1660 tagNamePtr->uriLen = binding->uriLen; | |
1661 i = binding->uriLen; | |
1662 do { | |
1663 if (i == binding->uriAlloc) { | |
1664 binding->uri = realloc(binding->uri, binding->uriAlloc *= 2); | |
1665 if (!binding->uri) | |
1666 return XML_ERROR_NO_MEMORY; | |
1667 } | |
1668 binding->uri[i++] = *localPart; | |
1669 } while (*localPart++); | |
1670 tagNamePtr->str = binding->uri; | |
1671 return XML_ERROR_NONE; | |
1672 } | |
1673 | |
1674 static | |
1675 int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr) | |
1676 { | |
1677 BINDING *b; | |
1678 int len; | |
1679 for (len = 0; uri[len]; len++) | |
1680 ; | |
1681 if (namespaceSeparator) | |
1682 len++; | |
1683 if (freeBindingList) { | |
1684 b = freeBindingList; | |
1685 if (len > b->uriAlloc) { | |
1686 b->uri = realloc(b->uri, len + EXPAND_SPARE); | |
1687 if (!b->uri) | |
1688 return 0; | |
1689 b->uriAlloc = len + EXPAND_SPARE; | |
1690 } | |
1691 freeBindingList = b->nextTagBinding; | |
1692 } | |
1693 else { | |
1694 b = malloc(sizeof(BINDING)); | |
1695 if (!b) | |
1696 return 0; | |
1697 b->uri = malloc(sizeof(XML_Char) * len + EXPAND_SPARE); | |
1698 if (!b->uri) { | |
1699 free(b); | |
1700 return 0; | |
1701 } | |
1702 b->uriAlloc = len; | |
1703 } | |
1704 b->uriLen = len; | |
1705 memcpy(b->uri, uri, len * sizeof(XML_Char)); | |
1706 if (namespaceSeparator) | |
1707 b->uri[len - 1] = namespaceSeparator; | |
1708 b->prefix = prefix; | |
1709 b->attId = attId; | |
1710 b->prevPrefixBinding = prefix->binding; | |
1711 if (*uri == XML_T('\0') && prefix == &dtd.defaultPrefix) | |
1712 prefix->binding = 0; | |
1713 else | |
1714 prefix->binding = b; | |
1715 b->nextTagBinding = *bindingsPtr; | |
1716 *bindingsPtr = b; | |
1717 if (startNamespaceDeclHandler) | |
1718 startNamespaceDeclHandler(handlerArg, prefix->name, | |
1719 prefix->binding ? uri : 0); | |
1720 return 1; | |
1721 } | |
1722 | |
1723 /* The idea here is to avoid using stack for each CDATA section when | |
1724 the whole file is parsed with one call. */ | |
1725 | |
1726 static | |
1727 enum XML_Error cdataSectionProcessor(XML_Parser parser, | |
1728 const char *start, | |
1729 const char *end, | |
1730 const char **endPtr) | |
1731 { | |
1732 enum XML_Error result = doCdataSection(parser, encoding, &start, end, endPtr); | |
1733 if (start) { | |
1734 processor = contentProcessor; | |
1735 return contentProcessor(parser, start, end, endPtr); | |
1736 } | |
1737 return result; | |
1738 } | |
1739 | |
1740 /* startPtr gets set to non-null is the section is closed, and to null if | |
1741 the section is not yet closed. */ | |
1742 | |
1743 static | |
1744 enum XML_Error doCdataSection(XML_Parser parser, | |
1745 const ENCODING *enc, | |
1746 const char **startPtr, | |
1747 const char *end, | |
1748 const char **nextPtr) | |
1749 { | |
1750 const char *s = *startPtr; | |
1751 const char **eventPP; | |
1752 const char **eventEndPP; | |
1753 if (enc == encoding) { | |
1754 eventPP = &eventPtr; | |
1755 *eventPP = s; | |
1756 eventEndPP = &eventEndPtr; | |
1757 } | |
1758 else { | |
1759 eventPP = &(openInternalEntities->internalEventPtr); | |
1760 eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
1761 } | |
1762 *eventPP = s; | |
1763 *startPtr = 0; | |
1764 for (;;) { | |
1765 const char *next; | |
1766 int tok = XmlCdataSectionTok(enc, s, end, &next); | |
1767 *eventEndPP = next; | |
1768 switch (tok) { | |
1769 case XML_TOK_CDATA_SECT_CLOSE: | |
1770 if (endCdataSectionHandler) | |
1771 endCdataSectionHandler(handlerArg); | |
1772 #if 0 | |
1773 /* see comment under XML_TOK_CDATA_SECT_OPEN */ | |
1774 else if (characterDataHandler) | |
1775 characterDataHandler(handlerArg, dataBuf, 0); | |
1776 #endif | |
1777 else if (defaultHandler) | |
1778 reportDefault(parser, enc, s, next); | |
1779 *startPtr = next; | |
1780 return XML_ERROR_NONE; | |
1781 case XML_TOK_DATA_NEWLINE: | |
1782 if (characterDataHandler) { | |
1783 XML_Char c = 0xA; | |
1784 characterDataHandler(handlerArg, &c, 1); | |
1785 } | |
1786 else if (defaultHandler) | |
1787 reportDefault(parser, enc, s, next); | |
1788 break; | |
1789 case XML_TOK_DATA_CHARS: | |
1790 if (characterDataHandler) { | |
1791 if (MUST_CONVERT(enc, s)) { | |
1792 for (;;) { | |
1793 ICHAR *dataPtr = (ICHAR *)dataBuf; | |
1794 XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
1795 *eventEndPP = next; | |
1796 characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
1797 if (s == next) | |
1798 break; | |
1799 *eventPP = s; | |
1800 } | |
1801 } | |
1802 else | |
1803 characterDataHandler(handlerArg, | |
1804 (XML_Char *)s, | |
1805 (XML_Char *)next - (XML_Char *)s); | |
1806 } | |
1807 else if (defaultHandler) | |
1808 reportDefault(parser, enc, s, next); | |
1809 break; | |
1810 case XML_TOK_INVALID: | |
1811 *eventPP = next; | |
1812 return XML_ERROR_INVALID_TOKEN; | |
1813 case XML_TOK_PARTIAL_CHAR: | |
1814 if (nextPtr) { | |
1815 *nextPtr = s; | |
1816 return XML_ERROR_NONE; | |
1817 } | |
1818 return XML_ERROR_PARTIAL_CHAR; | |
1819 case XML_TOK_PARTIAL: | |
1820 case XML_TOK_NONE: | |
1821 if (nextPtr) { | |
1822 *nextPtr = s; | |
1823 return XML_ERROR_NONE; | |
1824 } | |
1825 return XML_ERROR_UNCLOSED_CDATA_SECTION; | |
1826 default: | |
1827 abort(); | |
1828 } | |
1829 *eventPP = s = next; | |
1830 } | |
1831 /* not reached */ | |
1832 } | |
1833 | |
1834 static enum XML_Error | |
1835 initializeEncoding(XML_Parser parser) | |
1836 { | |
1837 const char *s; | |
1838 #ifdef XML_UNICODE | |
1839 char encodingBuf[128]; | |
1840 if (!protocolEncodingName) | |
1841 s = 0; | |
1842 else { | |
1843 int i; | |
1844 for (i = 0; protocolEncodingName[i]; i++) { | |
1845 if (i == sizeof(encodingBuf) - 1 | |
1846 || protocolEncodingName[i] >= 0x80 | |
1847 || protocolEncodingName[i] < 0) { | |
1848 encodingBuf[0] = '\0'; | |
1849 break; | |
1850 } | |
1851 encodingBuf[i] = (char)protocolEncodingName[i]; | |
1852 } | |
1853 encodingBuf[i] = '\0'; | |
1854 s = encodingBuf; | |
1855 } | |
1856 #else | |
1857 s = protocolEncodingName; | |
1858 #endif | |
1859 if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s)) | |
1860 return XML_ERROR_NONE; | |
1861 return handleUnknownEncoding(parser, protocolEncodingName); | |
1862 } | |
1863 | |
1864 static enum XML_Error | |
1865 processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
1866 const char *s, const char *next) | |
1867 { | |
1868 const char *encodingName = 0; | |
1869 const ENCODING *newEncoding = 0; | |
1870 const char *version; | |
1871 int standalone = -1; | |
1872 if (!(ns | |
1873 ? XmlParseXmlDeclNS | |
1874 : XmlParseXmlDecl)(isGeneralTextEntity, | |
1875 encoding, | |
1876 s, | |
1877 next, | |
1878 &eventPtr, | |
1879 &version, | |
1880 &encodingName, | |
1881 &newEncoding, | |
1882 &standalone)) | |
1883 return XML_ERROR_SYNTAX; | |
1884 if (!isGeneralTextEntity && standalone == 1) | |
1885 dtd.standalone = 1; | |
1886 if (defaultHandler) | |
1887 reportDefault(parser, encoding, s, next); | |
1888 if (!protocolEncodingName) { | |
1889 if (newEncoding) { | |
1890 if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) { | |
1891 eventPtr = encodingName; | |
1892 return XML_ERROR_INCORRECT_ENCODING; | |
1893 } | |
1894 encoding = newEncoding; | |
1895 } | |
1896 else if (encodingName) { | |
1897 enum XML_Error result; | |
1898 const XML_Char *s = poolStoreString(&tempPool, | |
1899 encoding, | |
1900 encodingName, | |
1901 encodingName | |
1902 + XmlNameLength(encoding, encodingName)); | |
1903 if (!s) | |
1904 return XML_ERROR_NO_MEMORY; | |
1905 result = handleUnknownEncoding(parser, s); | |
1906 poolDiscard(&tempPool); | |
1907 if (result == XML_ERROR_UNKNOWN_ENCODING) | |
1908 eventPtr = encodingName; | |
1909 return result; | |
1910 } | |
1911 } | |
1912 return XML_ERROR_NONE; | |
1913 } | |
1914 | |
1915 static enum XML_Error | |
1916 handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) | |
1917 { | |
1918 if (unknownEncodingHandler) { | |
1919 XML_Encoding info; | |
1920 int i; | |
1921 for (i = 0; i < 256; i++) | |
1922 info.map[i] = -1; | |
1923 info.convert = 0; | |
1924 info.data = 0; | |
1925 info.release = 0; | |
1926 if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, &info)) { | |
1927 ENCODING *enc; | |
1928 unknownEncodingMem = malloc(XmlSizeOfUnknownEncoding()); | |
1929 if (!unknownEncodingMem) { | |
1930 if (info.release) | |
1931 info.release(info.data); | |
1932 return XML_ERROR_NO_MEMORY; | |
1933 } | |
1934 enc = (ns | |
1935 ? XmlInitUnknownEncodingNS | |
1936 : XmlInitUnknownEncoding)(unknownEncodingMem, | |
1937 info.map, | |
1938 info.convert, | |
1939 info.data); | |
1940 if (enc) { | |
1941 unknownEncodingData = info.data; | |
1942 unknownEncodingRelease = info.release; | |
1943 encoding = enc; | |
1944 return XML_ERROR_NONE; | |
1945 } | |
1946 } | |
1947 if (info.release) | |
1948 info.release(info.data); | |
1949 } | |
1950 return XML_ERROR_UNKNOWN_ENCODING; | |
1951 } | |
1952 | |
1953 static enum XML_Error | |
1954 prologInitProcessor(XML_Parser parser, | |
1955 const char *s, | |
1956 const char *end, | |
1957 const char **nextPtr) | |
1958 { | |
1959 enum XML_Error result = initializeEncoding(parser); | |
1960 if (result != XML_ERROR_NONE) | |
1961 return result; | |
1962 processor = prologProcessor; | |
1963 return prologProcessor(parser, s, end, nextPtr); | |
1964 } | |
1965 | |
1966 static enum XML_Error | |
1967 prologProcessor(XML_Parser parser, | |
1968 const char *s, | |
1969 const char *end, | |
1970 const char **nextPtr) | |
1971 { | |
1972 for (;;) { | |
1973 const char *next; | |
1974 int tok = XmlPrologTok(encoding, s, end, &next); | |
1975 if (tok <= 0) { | |
1976 if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
1977 *nextPtr = s; | |
1978 return XML_ERROR_NONE; | |
1979 } | |
1980 switch (tok) { | |
1981 case XML_TOK_INVALID: | |
1982 eventPtr = next; | |
1983 return XML_ERROR_INVALID_TOKEN; | |
1984 case XML_TOK_NONE: | |
1985 return XML_ERROR_NO_ELEMENTS; | |
1986 case XML_TOK_PARTIAL: | |
1987 return XML_ERROR_UNCLOSED_TOKEN; | |
1988 case XML_TOK_PARTIAL_CHAR: | |
1989 return XML_ERROR_PARTIAL_CHAR; | |
1990 case XML_TOK_TRAILING_CR: | |
1991 eventPtr = s + encoding->minBytesPerChar; | |
1992 return XML_ERROR_NO_ELEMENTS; | |
1993 default: | |
1994 abort(); | |
1995 } | |
1996 } | |
1997 switch (XmlTokenRole(&prologState, tok, s, next, encoding)) { | |
1998 case XML_ROLE_XML_DECL: | |
1999 { | |
2000 enum XML_Error result = processXmlDecl(parser, 0, s, next); | |
2001 if (result != XML_ERROR_NONE) | |
2002 return result; | |
2003 } | |
2004 break; | |
2005 case XML_ROLE_DOCTYPE_SYSTEM_ID: | |
2006 if (!dtd.standalone | |
2007 && notStandaloneHandler | |
2008 && !notStandaloneHandler(handlerArg)) | |
2009 return XML_ERROR_NOT_STANDALONE; | |
2010 hadExternalDoctype = 1; | |
2011 break; | |
2012 case XML_ROLE_DOCTYPE_PUBLIC_ID: | |
2013 case XML_ROLE_ENTITY_PUBLIC_ID: | |
2014 if (!XmlIsPublicId(encoding, s, next, &eventPtr)) | |
2015 return XML_ERROR_SYNTAX; | |
2016 if (declEntity) { | |
2017 XML_Char *tem = poolStoreString(&dtd.pool, | |
2018 encoding, | |
2019 s + encoding->minBytesPerChar, | |
2020 next - encoding->minBytesPerChar); | |
2021 if (!tem) | |
2022 return XML_ERROR_NO_MEMORY; | |
2023 normalizePublicId(tem); | |
2024 declEntity->publicId = tem; | |
2025 poolFinish(&dtd.pool); | |
2026 } | |
2027 break; | |
2028 case XML_ROLE_INSTANCE_START: | |
2029 processor = contentProcessor; | |
2030 if (hadExternalDoctype) | |
2031 dtd.complete = 0; | |
2032 return contentProcessor(parser, s, end, nextPtr); | |
2033 case XML_ROLE_ATTLIST_ELEMENT_NAME: | |
2034 { | |
2035 const XML_Char *name = poolStoreString(&dtd.pool, encoding, s, next); | |
2036 if (!name) | |
2037 return XML_ERROR_NO_MEMORY; | |
2038 declElementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, name, sizeof(ELEMENT_TYPE)); | |
2039 if (!declElementType) | |
2040 return XML_ERROR_NO_MEMORY; | |
2041 if (declElementType->name != name) | |
2042 poolDiscard(&dtd.pool); | |
2043 else { | |
2044 poolFinish(&dtd.pool); | |
2045 if (!setElementTypePrefix(parser, declElementType)) | |
2046 return XML_ERROR_NO_MEMORY; | |
2047 } | |
2048 break; | |
2049 } | |
2050 case XML_ROLE_ATTRIBUTE_NAME: | |
2051 declAttributeId = getAttributeId(parser, encoding, s, next); | |
2052 if (!declAttributeId) | |
2053 return XML_ERROR_NO_MEMORY; | |
2054 declAttributeIsCdata = 0; | |
2055 break; | |
2056 case XML_ROLE_ATTRIBUTE_TYPE_CDATA: | |
2057 declAttributeIsCdata = 1; | |
2058 break; | |
2059 case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: | |
2060 case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: | |
2061 if (dtd.complete | |
2062 && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, 0)) | |
2063 return XML_ERROR_NO_MEMORY; | |
2064 break; | |
2065 case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: | |
2066 case XML_ROLE_FIXED_ATTRIBUTE_VALUE: | |
2067 { | |
2068 const XML_Char *attVal; | |
2069 enum XML_Error result | |
2070 = storeAttributeValue(parser, encoding, declAttributeIsCdata, | |
2071 s + encoding->minBytesPerChar, | |
2072 next - encoding->minBytesPerChar, | |
2073 &dtd.pool); | |
2074 if (result) | |
2075 return result; | |
2076 attVal = poolStart(&dtd.pool); | |
2077 poolFinish(&dtd.pool); | |
2078 if (dtd.complete | |
2079 && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, attVal)) | |
2080 return XML_ERROR_NO_MEMORY; | |
2081 break; | |
2082 } | |
2083 case XML_ROLE_ENTITY_VALUE: | |
2084 { | |
2085 enum XML_Error result = storeEntityValue(parser, s, next); | |
2086 if (result != XML_ERROR_NONE) | |
2087 return result; | |
2088 } | |
2089 break; | |
2090 case XML_ROLE_ENTITY_SYSTEM_ID: | |
2091 if (declEntity) { | |
2092 declEntity->systemId = poolStoreString(&dtd.pool, encoding, | |
2093 s + encoding->minBytesPerChar, | |
2094 next - encoding->minBytesPerChar); | |
2095 if (!declEntity->systemId) | |
2096 return XML_ERROR_NO_MEMORY; | |
2097 declEntity->base = dtd.base; | |
2098 poolFinish(&dtd.pool); | |
2099 } | |
2100 break; | |
2101 case XML_ROLE_ENTITY_NOTATION_NAME: | |
2102 if (declEntity) { | |
2103 declEntity->notation = poolStoreString(&dtd.pool, encoding, s, next); | |
2104 if (!declEntity->notation) | |
2105 return XML_ERROR_NO_MEMORY; | |
2106 poolFinish(&dtd.pool); | |
2107 if (unparsedEntityDeclHandler) { | |
2108 eventPtr = eventEndPtr = s; | |
2109 unparsedEntityDeclHandler(handlerArg, | |
2110 declEntity->name, | |
2111 declEntity->base, | |
2112 declEntity->systemId, | |
2113 declEntity->publicId, | |
2114 declEntity->notation); | |
2115 } | |
2116 | |
2117 } | |
2118 break; | |
2119 case XML_ROLE_GENERAL_ENTITY_NAME: | |
2120 { | |
2121 const XML_Char *name; | |
2122 if (XmlPredefinedEntityName(encoding, s, next)) { | |
2123 declEntity = 0; | |
2124 break; | |
2125 } | |
2126 name = poolStoreString(&dtd.pool, encoding, s, next); | |
2127 if (!name) | |
2128 return XML_ERROR_NO_MEMORY; | |
2129 if (dtd.complete) { | |
2130 declEntity = (ENTITY *)lookup(&dtd.generalEntities, name, sizeof(ENTITY)); | |
2131 if (!declEntity) | |
2132 return XML_ERROR_NO_MEMORY; | |
2133 if (declEntity->name != name) { | |
2134 poolDiscard(&dtd.pool); | |
2135 declEntity = 0; | |
2136 } | |
2137 else | |
2138 poolFinish(&dtd.pool); | |
2139 } | |
2140 else { | |
2141 poolDiscard(&dtd.pool); | |
2142 declEntity = 0; | |
2143 } | |
2144 } | |
2145 break; | |
2146 case XML_ROLE_PARAM_ENTITY_NAME: | |
2147 declEntity = 0; | |
2148 break; | |
2149 case XML_ROLE_NOTATION_NAME: | |
2150 declNotationPublicId = 0; | |
2151 declNotationName = 0; | |
2152 if (notationDeclHandler) { | |
2153 declNotationName = poolStoreString(&tempPool, encoding, s, next); | |
2154 if (!declNotationName) | |
2155 return XML_ERROR_NO_MEMORY; | |
2156 poolFinish(&tempPool); | |
2157 } | |
2158 break; | |
2159 case XML_ROLE_NOTATION_PUBLIC_ID: | |
2160 if (!XmlIsPublicId(encoding, s, next, &eventPtr)) | |
2161 return XML_ERROR_SYNTAX; | |
2162 if (declNotationName) { | |
2163 XML_Char *tem = poolStoreString(&tempPool, | |
2164 encoding, | |
2165 s + encoding->minBytesPerChar, | |
2166 next - encoding->minBytesPerChar); | |
2167 if (!tem) | |
2168 return XML_ERROR_NO_MEMORY; | |
2169 normalizePublicId(tem); | |
2170 declNotationPublicId = tem; | |
2171 poolFinish(&tempPool); | |
2172 } | |
2173 break; | |
2174 case XML_ROLE_NOTATION_SYSTEM_ID: | |
2175 if (declNotationName && notationDeclHandler) { | |
2176 const XML_Char *systemId | |
2177 = poolStoreString(&tempPool, encoding, | |
2178 s + encoding->minBytesPerChar, | |
2179 next - encoding->minBytesPerChar); | |
2180 if (!systemId) | |
2181 return XML_ERROR_NO_MEMORY; | |
2182 eventPtr = eventEndPtr = s; | |
2183 notationDeclHandler(handlerArg, | |
2184 declNotationName, | |
2185 dtd.base, | |
2186 systemId, | |
2187 declNotationPublicId); | |
2188 } | |
2189 poolClear(&tempPool); | |
2190 break; | |
2191 case XML_ROLE_NOTATION_NO_SYSTEM_ID: | |
2192 if (declNotationPublicId && notationDeclHandler) { | |
2193 eventPtr = eventEndPtr = s; | |
2194 notationDeclHandler(handlerArg, | |
2195 declNotationName, | |
2196 dtd.base, | |
2197 0, | |
2198 declNotationPublicId); | |
2199 } | |
2200 poolClear(&tempPool); | |
2201 break; | |
2202 case XML_ROLE_ERROR: | |
2203 eventPtr = s; | |
2204 switch (tok) { | |
2205 case XML_TOK_PARAM_ENTITY_REF: | |
2206 return XML_ERROR_PARAM_ENTITY_REF; | |
2207 case XML_TOK_XML_DECL: | |
2208 return XML_ERROR_MISPLACED_XML_PI; | |
2209 default: | |
2210 return XML_ERROR_SYNTAX; | |
2211 } | |
2212 case XML_ROLE_GROUP_OPEN: | |
2213 if (prologState.level >= groupSize) { | |
2214 if (groupSize) | |
2215 groupConnector = realloc(groupConnector, groupSize *= 2); | |
2216 else | |
2217 groupConnector = malloc(groupSize = 32); | |
2218 if (!groupConnector) | |
2219 return XML_ERROR_NO_MEMORY; | |
2220 } | |
2221 groupConnector[prologState.level] = 0; | |
2222 break; | |
2223 case XML_ROLE_GROUP_SEQUENCE: | |
2224 if (groupConnector[prologState.level] == '|') { | |
2225 eventPtr = s; | |
2226 return XML_ERROR_SYNTAX; | |
2227 } | |
2228 groupConnector[prologState.level] = ','; | |
2229 break; | |
2230 case XML_ROLE_GROUP_CHOICE: | |
2231 if (groupConnector[prologState.level] == ',') { | |
2232 eventPtr = s; | |
2233 return XML_ERROR_SYNTAX; | |
2234 } | |
2235 groupConnector[prologState.level] = '|'; | |
2236 break; | |
2237 case XML_ROLE_PARAM_ENTITY_REF: | |
2238 if (!dtd.standalone | |
2239 && notStandaloneHandler | |
2240 && !notStandaloneHandler(handlerArg)) | |
2241 return XML_ERROR_NOT_STANDALONE; | |
2242 dtd.complete = 0; | |
2243 break; | |
2244 case XML_ROLE_NONE: | |
2245 switch (tok) { | |
2246 case XML_TOK_PI: | |
2247 eventPtr = s; | |
2248 eventEndPtr = next; | |
2249 if (!reportProcessingInstruction(parser, encoding, s, next)) | |
2250 return XML_ERROR_NO_MEMORY; | |
2251 break; | |
2252 case XML_TOK_COMMENT: | |
2253 eventPtr = s; | |
2254 eventEndPtr = next; | |
2255 if (!reportComment(parser, encoding, s, next)) | |
2256 return XML_ERROR_NO_MEMORY; | |
2257 break; | |
2258 } | |
2259 break; | |
2260 } | |
2261 if (defaultHandler) { | |
2262 switch (tok) { | |
2263 case XML_TOK_PI: | |
2264 case XML_TOK_COMMENT: | |
2265 case XML_TOK_BOM: | |
2266 case XML_TOK_XML_DECL: | |
2267 break; | |
2268 default: | |
2269 eventPtr = s; | |
2270 eventEndPtr = next; | |
2271 reportDefault(parser, encoding, s, next); | |
2272 } | |
2273 } | |
2274 s = next; | |
2275 } | |
2276 /* not reached */ | |
2277 } | |
2278 | |
2279 static | |
2280 enum XML_Error epilogProcessor(XML_Parser parser, | |
2281 const char *s, | |
2282 const char *end, | |
2283 const char **nextPtr) | |
2284 { | |
2285 processor = epilogProcessor; | |
2286 eventPtr = s; | |
2287 for (;;) { | |
2288 const char *next; | |
2289 int tok = XmlPrologTok(encoding, s, end, &next); | |
2290 eventEndPtr = next; | |
2291 switch (tok) { | |
2292 case XML_TOK_TRAILING_CR: | |
2293 if (defaultHandler) { | |
2294 eventEndPtr = end; | |
2295 reportDefault(parser, encoding, s, end); | |
2296 } | |
2297 /* fall through */ | |
2298 case XML_TOK_NONE: | |
2299 if (nextPtr) | |
2300 *nextPtr = end; | |
2301 return XML_ERROR_NONE; | |
2302 case XML_TOK_PROLOG_S: | |
2303 if (defaultHandler) | |
2304 reportDefault(parser, encoding, s, next); | |
2305 break; | |
2306 case XML_TOK_PI: | |
2307 if (!reportProcessingInstruction(parser, encoding, s, next)) | |
2308 return XML_ERROR_NO_MEMORY; | |
2309 break; | |
2310 case XML_TOK_COMMENT: | |
2311 if (!reportComment(parser, encoding, s, next)) | |
2312 return XML_ERROR_NO_MEMORY; | |
2313 break; | |
2314 case XML_TOK_INVALID: | |
2315 eventPtr = next; | |
2316 return XML_ERROR_INVALID_TOKEN; | |
2317 case XML_TOK_PARTIAL: | |
2318 if (nextPtr) { | |
2319 *nextPtr = s; | |
2320 return XML_ERROR_NONE; | |
2321 } | |
2322 return XML_ERROR_UNCLOSED_TOKEN; | |
2323 case XML_TOK_PARTIAL_CHAR: | |
2324 if (nextPtr) { | |
2325 *nextPtr = s; | |
2326 return XML_ERROR_NONE; | |
2327 } | |
2328 return XML_ERROR_PARTIAL_CHAR; | |
2329 default: | |
2330 return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; | |
2331 } | |
2332 eventPtr = s = next; | |
2333 } | |
2334 } | |
2335 | |
235 | 2336 /* |
25 | 2337 static |
2338 enum XML_Error errorProcessor(XML_Parser parser, | |
2339 const char *s, | |
2340 const char *end, | |
2341 const char **nextPtr) | |
2342 { | |
2343 return errorCode; | |
2344 } | |
235 | 2345 */ |
25 | 2346 |
2347 static enum XML_Error | |
2348 storeAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, | |
2349 const char *ptr, const char *end, | |
2350 STRING_POOL *pool) | |
2351 { | |
2352 enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr, end, pool); | |
2353 if (result) | |
2354 return result; | |
2355 if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) | |
2356 poolChop(pool); | |
2357 if (!poolAppendChar(pool, XML_T('\0'))) | |
2358 return XML_ERROR_NO_MEMORY; | |
2359 return XML_ERROR_NONE; | |
2360 } | |
2361 | |
2362 static enum XML_Error | |
2363 appendAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, | |
2364 const char *ptr, const char *end, | |
2365 STRING_POOL *pool) | |
2366 { | |
2367 const ENCODING *internalEnc = ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(); | |
2368 for (;;) { | |
2369 const char *next; | |
2370 int tok = XmlAttributeValueTok(enc, ptr, end, &next); | |
2371 switch (tok) { | |
2372 case XML_TOK_NONE: | |
2373 return XML_ERROR_NONE; | |
2374 case XML_TOK_INVALID: | |
2375 if (enc == encoding) | |
2376 eventPtr = next; | |
2377 return XML_ERROR_INVALID_TOKEN; | |
2378 case XML_TOK_PARTIAL: | |
2379 if (enc == encoding) | |
2380 eventPtr = ptr; | |
2381 return XML_ERROR_INVALID_TOKEN; | |
2382 case XML_TOK_CHAR_REF: | |
2383 { | |
2384 XML_Char buf[XML_ENCODE_MAX]; | |
2385 int i; | |
2386 int n = XmlCharRefNumber(enc, ptr); | |
2387 if (n < 0) { | |
2388 if (enc == encoding) | |
2389 eventPtr = ptr; | |
2390 return XML_ERROR_BAD_CHAR_REF; | |
2391 } | |
2392 if (!isCdata | |
2393 && n == 0x20 /* space */ | |
2394 && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
2395 break; | |
2396 n = XmlEncode(n, (ICHAR *)buf); | |
2397 if (!n) { | |
2398 if (enc == encoding) | |
2399 eventPtr = ptr; | |
2400 return XML_ERROR_BAD_CHAR_REF; | |
2401 } | |
2402 for (i = 0; i < n; i++) { | |
2403 if (!poolAppendChar(pool, buf[i])) | |
2404 return XML_ERROR_NO_MEMORY; | |
2405 } | |
2406 } | |
2407 break; | |
2408 case XML_TOK_DATA_CHARS: | |
2409 if (!poolAppend(pool, enc, ptr, next)) | |
2410 return XML_ERROR_NO_MEMORY; | |
2411 break; | |
2412 break; | |
2413 case XML_TOK_TRAILING_CR: | |
2414 next = ptr + enc->minBytesPerChar; | |
2415 /* fall through */ | |
2416 case XML_TOK_ATTRIBUTE_VALUE_S: | |
2417 case XML_TOK_DATA_NEWLINE: | |
2418 if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
2419 break; | |
2420 if (!poolAppendChar(pool, 0x20)) | |
2421 return XML_ERROR_NO_MEMORY; | |
2422 break; | |
2423 case XML_TOK_ENTITY_REF: | |
2424 { | |
2425 const XML_Char *name; | |
2426 ENTITY *entity; | |
2427 XML_Char ch = XmlPredefinedEntityName(enc, | |
2428 ptr + enc->minBytesPerChar, | |
2429 next - enc->minBytesPerChar); | |
2430 if (ch) { | |
2431 if (!poolAppendChar(pool, ch)) | |
2432 return XML_ERROR_NO_MEMORY; | |
2433 break; | |
2434 } | |
2435 name = poolStoreString(&temp2Pool, enc, | |
2436 ptr + enc->minBytesPerChar, | |
2437 next - enc->minBytesPerChar); | |
2438 if (!name) | |
2439 return XML_ERROR_NO_MEMORY; | |
2440 entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); | |
2441 poolDiscard(&temp2Pool); | |
2442 if (!entity) { | |
2443 if (dtd.complete) { | |
2444 if (enc == encoding) | |
2445 eventPtr = ptr; | |
2446 return XML_ERROR_UNDEFINED_ENTITY; | |
2447 } | |
2448 } | |
2449 else if (entity->open) { | |
2450 if (enc == encoding) | |
2451 eventPtr = ptr; | |
2452 return XML_ERROR_RECURSIVE_ENTITY_REF; | |
2453 } | |
2454 else if (entity->notation) { | |
2455 if (enc == encoding) | |
2456 eventPtr = ptr; | |
2457 return XML_ERROR_BINARY_ENTITY_REF; | |
2458 } | |
2459 else if (!entity->textPtr) { | |
2460 if (enc == encoding) | |
2461 eventPtr = ptr; | |
2462 return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; | |
2463 } | |
2464 else { | |
2465 enum XML_Error result; | |
2466 const XML_Char *textEnd = entity->textPtr + entity->textLen; | |
2467 entity->open = 1; | |
2468 result = appendAttributeValue(parser, internalEnc, isCdata, (char *)entity->textPtr, (char *)textEnd, pool); | |
2469 entity->open = 0; | |
2470 if (result) | |
2471 return result; | |
2472 } | |
2473 } | |
2474 break; | |
2475 default: | |
2476 abort(); | |
2477 } | |
2478 ptr = next; | |
2479 } | |
2480 /* not reached */ | |
2481 } | |
2482 | |
2483 static | |
2484 enum XML_Error storeEntityValue(XML_Parser parser, | |
2485 const char *entityTextPtr, | |
2486 const char *entityTextEnd) | |
2487 { | |
235 | 2488 /* const ENCODING *internalEnc = ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(); */ |
25 | 2489 STRING_POOL *pool = &(dtd.pool); |
2490 entityTextPtr += encoding->minBytesPerChar; | |
2491 entityTextEnd -= encoding->minBytesPerChar; | |
2492 for (;;) { | |
2493 const char *next; | |
2494 int tok = XmlEntityValueTok(encoding, entityTextPtr, entityTextEnd, &next); | |
2495 switch (tok) { | |
2496 case XML_TOK_PARAM_ENTITY_REF: | |
2497 eventPtr = entityTextPtr; | |
2498 return XML_ERROR_SYNTAX; | |
2499 case XML_TOK_NONE: | |
2500 if (declEntity) { | |
2501 declEntity->textPtr = pool->start; | |
2502 declEntity->textLen = pool->ptr - pool->start; | |
2503 poolFinish(pool); | |
2504 } | |
2505 else | |
2506 poolDiscard(pool); | |
2507 return XML_ERROR_NONE; | |
2508 case XML_TOK_ENTITY_REF: | |
2509 case XML_TOK_DATA_CHARS: | |
2510 if (!poolAppend(pool, encoding, entityTextPtr, next)) | |
2511 return XML_ERROR_NO_MEMORY; | |
2512 break; | |
2513 case XML_TOK_TRAILING_CR: | |
2514 next = entityTextPtr + encoding->minBytesPerChar; | |
2515 /* fall through */ | |
2516 case XML_TOK_DATA_NEWLINE: | |
2517 if (pool->end == pool->ptr && !poolGrow(pool)) | |
2518 return XML_ERROR_NO_MEMORY; | |
2519 *(pool->ptr)++ = 0xA; | |
2520 break; | |
2521 case XML_TOK_CHAR_REF: | |
2522 { | |
2523 XML_Char buf[XML_ENCODE_MAX]; | |
2524 int i; | |
2525 int n = XmlCharRefNumber(encoding, entityTextPtr); | |
2526 if (n < 0) { | |
2527 eventPtr = entityTextPtr; | |
2528 return XML_ERROR_BAD_CHAR_REF; | |
2529 } | |
2530 n = XmlEncode(n, (ICHAR *)buf); | |
2531 if (!n) { | |
2532 eventPtr = entityTextPtr; | |
2533 return XML_ERROR_BAD_CHAR_REF; | |
2534 } | |
2535 for (i = 0; i < n; i++) { | |
2536 if (pool->end == pool->ptr && !poolGrow(pool)) | |
2537 return XML_ERROR_NO_MEMORY; | |
2538 *(pool->ptr)++ = buf[i]; | |
2539 } | |
2540 } | |
2541 break; | |
2542 case XML_TOK_PARTIAL: | |
2543 eventPtr = entityTextPtr; | |
2544 return XML_ERROR_INVALID_TOKEN; | |
2545 case XML_TOK_INVALID: | |
2546 eventPtr = next; | |
2547 return XML_ERROR_INVALID_TOKEN; | |
2548 default: | |
2549 abort(); | |
2550 } | |
2551 entityTextPtr = next; | |
2552 } | |
2553 /* not reached */ | |
2554 } | |
2555 | |
2556 static void | |
2557 normalizeLines(XML_Char *s) | |
2558 { | |
2559 XML_Char *p; | |
2560 for (;; s++) { | |
2561 if (*s == XML_T('\0')) | |
2562 return; | |
2563 if (*s == 0xD) | |
2564 break; | |
2565 } | |
2566 p = s; | |
2567 do { | |
2568 if (*s == 0xD) { | |
2569 *p++ = 0xA; | |
2570 if (*++s == 0xA) | |
2571 s++; | |
2572 } | |
2573 else | |
2574 *p++ = *s++; | |
2575 } while (*s); | |
2576 *p = XML_T('\0'); | |
2577 } | |
2578 | |
2579 static int | |
2580 reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
2581 { | |
2582 const XML_Char *target; | |
2583 XML_Char *data; | |
2584 const char *tem; | |
2585 if (!processingInstructionHandler) { | |
2586 if (defaultHandler) | |
2587 reportDefault(parser, enc, start, end); | |
2588 return 1; | |
2589 } | |
2590 start += enc->minBytesPerChar * 2; | |
2591 tem = start + XmlNameLength(enc, start); | |
2592 target = poolStoreString(&tempPool, enc, start, tem); | |
2593 if (!target) | |
2594 return 0; | |
2595 poolFinish(&tempPool); | |
2596 data = poolStoreString(&tempPool, enc, | |
2597 XmlSkipS(enc, tem), | |
2598 end - enc->minBytesPerChar*2); | |
2599 if (!data) | |
2600 return 0; | |
2601 normalizeLines(data); | |
2602 processingInstructionHandler(handlerArg, target, data); | |
2603 poolClear(&tempPool); | |
2604 return 1; | |
2605 } | |
2606 | |
2607 static int | |
2608 reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
2609 { | |
2610 XML_Char *data; | |
2611 if (!commentHandler) { | |
2612 if (defaultHandler) | |
2613 reportDefault(parser, enc, start, end); | |
2614 return 1; | |
2615 } | |
2616 data = poolStoreString(&tempPool, | |
2617 enc, | |
2618 start + enc->minBytesPerChar * 4, | |
2619 end - enc->minBytesPerChar * 3); | |
2620 if (!data) | |
2621 return 0; | |
2622 normalizeLines(data); | |
2623 commentHandler(handlerArg, data); | |
2624 poolClear(&tempPool); | |
2625 return 1; | |
2626 } | |
2627 | |
2628 static void | |
2629 reportDefault(XML_Parser parser, const ENCODING *enc, const char *s, const char *end) | |
2630 { | |
2631 if (MUST_CONVERT(enc, s)) { | |
2632 const char **eventPP; | |
2633 const char **eventEndPP; | |
2634 if (enc == encoding) { | |
2635 eventPP = &eventPtr; | |
2636 eventEndPP = &eventEndPtr; | |
2637 } | |
2638 else { | |
2639 eventPP = &(openInternalEntities->internalEventPtr); | |
2640 eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
2641 } | |
2642 do { | |
2643 ICHAR *dataPtr = (ICHAR *)dataBuf; | |
2644 XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
2645 *eventEndPP = s; | |
2646 defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
2647 *eventPP = s; | |
2648 } while (s != end); | |
2649 } | |
2650 else | |
2651 defaultHandler(handlerArg, (XML_Char *)s, (XML_Char *)end - (XML_Char *)s); | |
2652 } | |
2653 | |
2654 | |
2655 static int | |
2656 defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, int isCdata, const XML_Char *value) | |
2657 { | |
2658 DEFAULT_ATTRIBUTE *att; | |
2659 if (type->nDefaultAtts == type->allocDefaultAtts) { | |
2660 if (type->allocDefaultAtts == 0) { | |
2661 type->allocDefaultAtts = 8; | |
2662 type->defaultAtts = malloc(type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); | |
2663 } | |
2664 else { | |
2665 type->allocDefaultAtts *= 2; | |
2666 type->defaultAtts = realloc(type->defaultAtts, | |
2667 type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); | |
2668 } | |
2669 if (!type->defaultAtts) | |
2670 return 0; | |
2671 } | |
2672 att = type->defaultAtts + type->nDefaultAtts; | |
2673 att->id = attId; | |
2674 att->value = value; | |
2675 att->isCdata = isCdata; | |
2676 if (!isCdata) | |
2677 attId->maybeTokenized = 1; | |
2678 type->nDefaultAtts += 1; | |
2679 return 1; | |
2680 } | |
2681 | |
2682 static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) | |
2683 { | |
2684 const XML_Char *name; | |
2685 for (name = elementType->name; *name; name++) { | |
2686 if (*name == XML_T(':')) { | |
2687 PREFIX *prefix; | |
2688 const XML_Char *s; | |
2689 for (s = elementType->name; s != name; s++) { | |
2690 if (!poolAppendChar(&dtd.pool, *s)) | |
2691 return 0; | |
2692 } | |
2693 if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
2694 return 0; | |
2695 prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); | |
2696 if (!prefix) | |
2697 return 0; | |
2698 if (prefix->name == poolStart(&dtd.pool)) | |
2699 poolFinish(&dtd.pool); | |
2700 else | |
2701 poolDiscard(&dtd.pool); | |
2702 elementType->prefix = prefix; | |
2703 | |
2704 } | |
2705 } | |
2706 return 1; | |
2707 } | |
2708 | |
2709 static ATTRIBUTE_ID * | |
2710 getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
2711 { | |
2712 ATTRIBUTE_ID *id; | |
2713 const XML_Char *name; | |
2714 if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
2715 return 0; | |
2716 name = poolStoreString(&dtd.pool, enc, start, end); | |
2717 if (!name) | |
2718 return 0; | |
2719 ++name; | |
2720 id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, name, sizeof(ATTRIBUTE_ID)); | |
2721 if (!id) | |
2722 return 0; | |
2723 if (id->name != name) | |
2724 poolDiscard(&dtd.pool); | |
2725 else { | |
2726 poolFinish(&dtd.pool); | |
2727 if (!ns) | |
2728 ; | |
2729 else if (name[0] == 'x' | |
2730 && name[1] == 'm' | |
2731 && name[2] == 'l' | |
2732 && name[3] == 'n' | |
2733 && name[4] == 's' | |
2734 && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) { | |
2735 if (name[5] == '\0') | |
2736 id->prefix = &dtd.defaultPrefix; | |
2737 else | |
2738 id->prefix = (PREFIX *)lookup(&dtd.prefixes, name + 6, sizeof(PREFIX)); | |
2739 id->xmlns = 1; | |
2740 } | |
2741 else { | |
2742 int i; | |
2743 for (i = 0; name[i]; i++) { | |
2744 if (name[i] == XML_T(':')) { | |
2745 int j; | |
2746 for (j = 0; j < i; j++) { | |
2747 if (!poolAppendChar(&dtd.pool, name[j])) | |
2748 return 0; | |
2749 } | |
2750 if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
2751 return 0; | |
2752 id->prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); | |
2753 if (id->prefix->name == poolStart(&dtd.pool)) | |
2754 poolFinish(&dtd.pool); | |
2755 else | |
2756 poolDiscard(&dtd.pool); | |
2757 break; | |
2758 } | |
2759 } | |
2760 } | |
2761 } | |
2762 return id; | |
2763 } | |
2764 | |
2765 #define CONTEXT_SEP XML_T('\f') | |
2766 | |
2767 static | |
2768 const XML_Char *getContext(XML_Parser parser) | |
2769 { | |
2770 HASH_TABLE_ITER iter; | |
2771 int needSep = 0; | |
2772 | |
2773 if (dtd.defaultPrefix.binding) { | |
2774 int i; | |
2775 int len; | |
2776 if (!poolAppendChar(&tempPool, XML_T('='))) | |
2777 return 0; | |
2778 len = dtd.defaultPrefix.binding->uriLen; | |
2779 if (namespaceSeparator != XML_T('\0')) | |
2780 len--; | |
2781 for (i = 0; i < len; i++) | |
2782 if (!poolAppendChar(&tempPool, dtd.defaultPrefix.binding->uri[i])) | |
2783 return 0; | |
2784 needSep = 1; | |
2785 } | |
2786 | |
2787 hashTableIterInit(&iter, &(dtd.prefixes)); | |
2788 for (;;) { | |
2789 int i; | |
2790 int len; | |
2791 const XML_Char *s; | |
2792 PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); | |
2793 if (!prefix) | |
2794 break; | |
2795 if (!prefix->binding) | |
2796 continue; | |
2797 if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
2798 return 0; | |
2799 for (s = prefix->name; *s; s++) | |
2800 if (!poolAppendChar(&tempPool, *s)) | |
2801 return 0; | |
2802 if (!poolAppendChar(&tempPool, XML_T('='))) | |
2803 return 0; | |
2804 len = prefix->binding->uriLen; | |
2805 if (namespaceSeparator != XML_T('\0')) | |
2806 len--; | |
2807 for (i = 0; i < len; i++) | |
2808 if (!poolAppendChar(&tempPool, prefix->binding->uri[i])) | |
2809 return 0; | |
2810 needSep = 1; | |
2811 } | |
2812 | |
2813 | |
2814 hashTableIterInit(&iter, &(dtd.generalEntities)); | |
2815 for (;;) { | |
2816 const XML_Char *s; | |
2817 ENTITY *e = (ENTITY *)hashTableIterNext(&iter); | |
2818 if (!e) | |
2819 break; | |
2820 if (!e->open) | |
2821 continue; | |
2822 if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
2823 return 0; | |
2824 for (s = e->name; *s; s++) | |
2825 if (!poolAppendChar(&tempPool, *s)) | |
2826 return 0; | |
2827 needSep = 1; | |
2828 } | |
2829 | |
2830 if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
2831 return 0; | |
2832 return tempPool.start; | |
2833 } | |
2834 | |
2835 static | |
2836 int setContext(XML_Parser parser, const XML_Char *context) | |
2837 { | |
2838 const XML_Char *s = context; | |
2839 | |
2840 while (*context != XML_T('\0')) { | |
2841 if (*s == CONTEXT_SEP || *s == XML_T('\0')) { | |
2842 ENTITY *e; | |
2843 if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
2844 return 0; | |
2845 e = (ENTITY *)lookup(&dtd.generalEntities, poolStart(&tempPool), 0); | |
2846 if (e) | |
2847 e->open = 1; | |
2848 if (*s != XML_T('\0')) | |
2849 s++; | |
2850 context = s; | |
2851 poolDiscard(&tempPool); | |
2852 } | |
2853 else if (*s == '=') { | |
2854 PREFIX *prefix; | |
2855 if (poolLength(&tempPool) == 0) | |
2856 prefix = &dtd.defaultPrefix; | |
2857 else { | |
2858 if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
2859 return 0; | |
2860 prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&tempPool), sizeof(PREFIX)); | |
2861 if (!prefix) | |
2862 return 0; | |
2863 if (prefix->name == poolStart(&tempPool)) | |
2864 poolFinish(&tempPool); | |
2865 else | |
2866 poolDiscard(&tempPool); | |
2867 } | |
2868 for (context = s + 1; *context != CONTEXT_SEP && *context != XML_T('\0'); context++) | |
2869 if (!poolAppendChar(&tempPool, *context)) | |
2870 return 0; | |
2871 if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
2872 return 0; | |
2873 if (!addBinding(parser, prefix, 0, poolStart(&tempPool), &inheritedBindings)) | |
2874 return 0; | |
2875 poolDiscard(&tempPool); | |
2876 if (*context != XML_T('\0')) | |
2877 ++context; | |
2878 s = context; | |
2879 } | |
2880 else { | |
2881 if (!poolAppendChar(&tempPool, *s)) | |
2882 return 0; | |
2883 s++; | |
2884 } | |
2885 } | |
2886 return 1; | |
2887 } | |
2888 | |
2889 | |
2890 static | |
2891 void normalizePublicId(XML_Char *publicId) | |
2892 { | |
2893 XML_Char *p = publicId; | |
2894 XML_Char *s; | |
2895 for (s = publicId; *s; s++) { | |
2896 switch (*s) { | |
2897 case 0x20: | |
2898 case 0xD: | |
2899 case 0xA: | |
2900 if (p != publicId && p[-1] != 0x20) | |
2901 *p++ = 0x20; | |
2902 break; | |
2903 default: | |
2904 *p++ = *s; | |
2905 } | |
2906 } | |
2907 if (p != publicId && p[-1] == 0x20) | |
2908 --p; | |
2909 *p = XML_T('\0'); | |
2910 } | |
2911 | |
2912 static int dtdInit(DTD *p) | |
2913 { | |
2914 poolInit(&(p->pool)); | |
2915 hashTableInit(&(p->generalEntities)); | |
2916 hashTableInit(&(p->elementTypes)); | |
2917 hashTableInit(&(p->attributeIds)); | |
2918 hashTableInit(&(p->prefixes)); | |
2919 p->complete = 1; | |
2920 p->standalone = 0; | |
2921 p->base = 0; | |
2922 p->defaultPrefix.name = 0; | |
2923 p->defaultPrefix.binding = 0; | |
2924 return 1; | |
2925 } | |
2926 | |
2927 static void dtdDestroy(DTD *p) | |
2928 { | |
2929 HASH_TABLE_ITER iter; | |
2930 hashTableIterInit(&iter, &(p->elementTypes)); | |
2931 for (;;) { | |
2932 ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
2933 if (!e) | |
2934 break; | |
2935 if (e->allocDefaultAtts != 0) | |
2936 free(e->defaultAtts); | |
2937 } | |
2938 hashTableDestroy(&(p->generalEntities)); | |
2939 hashTableDestroy(&(p->elementTypes)); | |
2940 hashTableDestroy(&(p->attributeIds)); | |
2941 hashTableDestroy(&(p->prefixes)); | |
2942 poolDestroy(&(p->pool)); | |
2943 } | |
2944 | |
2945 /* Do a deep copy of the DTD. Return 0 for out of memory; non-zero otherwise. | |
2946 The new DTD has already been initialized. */ | |
2947 | |
2948 static int dtdCopy(DTD *newDtd, const DTD *oldDtd) | |
2949 { | |
2950 HASH_TABLE_ITER iter; | |
2951 | |
2952 if (oldDtd->base) { | |
2953 const XML_Char *tem = poolCopyString(&(newDtd->pool), oldDtd->base); | |
2954 if (!tem) | |
2955 return 0; | |
2956 newDtd->base = tem; | |
2957 } | |
2958 | |
2959 /* Copy the prefix table. */ | |
2960 | |
2961 hashTableIterInit(&iter, &(oldDtd->prefixes)); | |
2962 for (;;) { | |
2963 const XML_Char *name; | |
2964 const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); | |
2965 if (!oldP) | |
2966 break; | |
2967 name = poolCopyString(&(newDtd->pool), oldP->name); | |
2968 if (!name) | |
2969 return 0; | |
2970 if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX))) | |
2971 return 0; | |
2972 } | |
2973 | |
2974 hashTableIterInit(&iter, &(oldDtd->attributeIds)); | |
2975 | |
2976 /* Copy the attribute id table. */ | |
2977 | |
2978 for (;;) { | |
2979 ATTRIBUTE_ID *newA; | |
2980 const XML_Char *name; | |
2981 const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); | |
2982 | |
2983 if (!oldA) | |
2984 break; | |
2985 /* Remember to allocate the scratch byte before the name. */ | |
2986 if (!poolAppendChar(&(newDtd->pool), XML_T('\0'))) | |
2987 return 0; | |
2988 name = poolCopyString(&(newDtd->pool), oldA->name); | |
2989 if (!name) | |
2990 return 0; | |
2991 ++name; | |
2992 newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name, sizeof(ATTRIBUTE_ID)); | |
2993 if (!newA) | |
2994 return 0; | |
2995 newA->maybeTokenized = oldA->maybeTokenized; | |
2996 if (oldA->prefix) { | |
2997 newA->xmlns = oldA->xmlns; | |
2998 if (oldA->prefix == &oldDtd->defaultPrefix) | |
2999 newA->prefix = &newDtd->defaultPrefix; | |
3000 else | |
3001 newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldA->prefix->name, 0); | |
3002 } | |
3003 } | |
3004 | |
3005 /* Copy the element type table. */ | |
3006 | |
3007 hashTableIterInit(&iter, &(oldDtd->elementTypes)); | |
3008 | |
3009 for (;;) { | |
3010 int i; | |
3011 ELEMENT_TYPE *newE; | |
3012 const XML_Char *name; | |
3013 const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
3014 if (!oldE) | |
3015 break; | |
3016 name = poolCopyString(&(newDtd->pool), oldE->name); | |
3017 if (!name) | |
3018 return 0; | |
3019 newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name, sizeof(ELEMENT_TYPE)); | |
3020 if (!newE) | |
3021 return 0; | |
3022 if (oldE->nDefaultAtts) { | |
3023 newE->defaultAtts = (DEFAULT_ATTRIBUTE *)malloc(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); | |
3024 if (!newE->defaultAtts) | |
3025 return 0; | |
3026 } | |
3027 newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; | |
3028 if (oldE->prefix) | |
3029 newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldE->prefix->name, 0); | |
3030 for (i = 0; i < newE->nDefaultAtts; i++) { | |
3031 newE->defaultAtts[i].id = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); | |
3032 newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; | |
3033 if (oldE->defaultAtts[i].value) { | |
3034 newE->defaultAtts[i].value = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); | |
3035 if (!newE->defaultAtts[i].value) | |
3036 return 0; | |
3037 } | |
3038 else | |
3039 newE->defaultAtts[i].value = 0; | |
3040 } | |
3041 } | |
3042 | |
3043 /* Copy the entity table. */ | |
3044 | |
3045 hashTableIterInit(&iter, &(oldDtd->generalEntities)); | |
3046 | |
3047 for (;;) { | |
3048 ENTITY *newE; | |
3049 const XML_Char *name; | |
3050 const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); | |
3051 if (!oldE) | |
3052 break; | |
3053 name = poolCopyString(&(newDtd->pool), oldE->name); | |
3054 if (!name) | |
3055 return 0; | |
3056 newE = (ENTITY *)lookup(&(newDtd->generalEntities), name, sizeof(ENTITY)); | |
3057 if (!newE) | |
3058 return 0; | |
3059 if (oldE->systemId) { | |
3060 const XML_Char *tem = poolCopyString(&(newDtd->pool), oldE->systemId); | |
3061 if (!tem) | |
3062 return 0; | |
3063 newE->systemId = tem; | |
3064 if (oldE->base) { | |
3065 if (oldE->base == oldDtd->base) | |
3066 newE->base = newDtd->base; | |
3067 tem = poolCopyString(&(newDtd->pool), oldE->base); | |
3068 if (!tem) | |
3069 return 0; | |
3070 newE->base = tem; | |
3071 } | |
3072 } | |
3073 else { | |
3074 const XML_Char *tem = poolCopyStringN(&(newDtd->pool), oldE->textPtr, oldE->textLen); | |
3075 if (!tem) | |
3076 return 0; | |
3077 newE->textPtr = tem; | |
3078 newE->textLen = oldE->textLen; | |
3079 } | |
3080 if (oldE->notation) { | |
3081 const XML_Char *tem = poolCopyString(&(newDtd->pool), oldE->notation); | |
3082 if (!tem) | |
3083 return 0; | |
3084 newE->notation = tem; | |
3085 } | |
3086 } | |
3087 | |
3088 newDtd->complete = oldDtd->complete; | |
3089 newDtd->standalone = oldDtd->standalone; | |
3090 return 1; | |
3091 } | |
3092 | |
3093 static | |
3094 void poolInit(STRING_POOL *pool) | |
3095 { | |
3096 pool->blocks = 0; | |
3097 pool->freeBlocks = 0; | |
3098 pool->start = 0; | |
3099 pool->ptr = 0; | |
3100 pool->end = 0; | |
3101 } | |
3102 | |
3103 static | |
3104 void poolClear(STRING_POOL *pool) | |
3105 { | |
3106 if (!pool->freeBlocks) | |
3107 pool->freeBlocks = pool->blocks; | |
3108 else { | |
3109 BLOCK *p = pool->blocks; | |
3110 while (p) { | |
3111 BLOCK *tem = p->next; | |
3112 p->next = pool->freeBlocks; | |
3113 pool->freeBlocks = p; | |
3114 p = tem; | |
3115 } | |
3116 } | |
3117 pool->blocks = 0; | |
3118 pool->start = 0; | |
3119 pool->ptr = 0; | |
3120 pool->end = 0; | |
3121 } | |
3122 | |
3123 static | |
3124 void poolDestroy(STRING_POOL *pool) | |
3125 { | |
3126 BLOCK *p = pool->blocks; | |
3127 while (p) { | |
3128 BLOCK *tem = p->next; | |
3129 free(p); | |
3130 p = tem; | |
3131 } | |
3132 pool->blocks = 0; | |
3133 p = pool->freeBlocks; | |
3134 while (p) { | |
3135 BLOCK *tem = p->next; | |
3136 free(p); | |
3137 p = tem; | |
3138 } | |
3139 pool->freeBlocks = 0; | |
3140 pool->ptr = 0; | |
3141 pool->start = 0; | |
3142 pool->end = 0; | |
3143 } | |
3144 | |
3145 static | |
3146 XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
3147 const char *ptr, const char *end) | |
3148 { | |
3149 if (!pool->ptr && !poolGrow(pool)) | |
3150 return 0; | |
3151 for (;;) { | |
3152 XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); | |
3153 if (ptr == end) | |
3154 break; | |
3155 if (!poolGrow(pool)) | |
3156 return 0; | |
3157 } | |
3158 return pool->start; | |
3159 } | |
3160 | |
3161 static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s) | |
3162 { | |
3163 do { | |
3164 if (!poolAppendChar(pool, *s)) | |
3165 return 0; | |
3166 } while (*s++); | |
3167 s = pool->start; | |
3168 poolFinish(pool); | |
3169 return s; | |
3170 } | |
3171 | |
3172 static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) | |
3173 { | |
3174 if (!pool->ptr && !poolGrow(pool)) | |
3175 return 0; | |
3176 for (; n > 0; --n, s++) { | |
3177 if (!poolAppendChar(pool, *s)) | |
3178 return 0; | |
3179 | |
3180 } | |
3181 s = pool->start; | |
3182 poolFinish(pool); | |
3183 return s; | |
3184 } | |
3185 | |
3186 static | |
3187 XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
3188 const char *ptr, const char *end) | |
3189 { | |
3190 if (!poolAppend(pool, enc, ptr, end)) | |
3191 return 0; | |
3192 if (pool->ptr == pool->end && !poolGrow(pool)) | |
3193 return 0; | |
3194 *(pool->ptr)++ = 0; | |
3195 return pool->start; | |
3196 } | |
3197 | |
3198 static | |
3199 int poolGrow(STRING_POOL *pool) | |
3200 { | |
3201 if (pool->freeBlocks) { | |
3202 if (pool->start == 0) { | |
3203 pool->blocks = pool->freeBlocks; | |
3204 pool->freeBlocks = pool->freeBlocks->next; | |
3205 pool->blocks->next = 0; | |
3206 pool->start = pool->blocks->s; | |
3207 pool->end = pool->start + pool->blocks->size; | |
3208 pool->ptr = pool->start; | |
3209 return 1; | |
3210 } | |
3211 if (pool->end - pool->start < pool->freeBlocks->size) { | |
3212 BLOCK *tem = pool->freeBlocks->next; | |
3213 pool->freeBlocks->next = pool->blocks; | |
3214 pool->blocks = pool->freeBlocks; | |
3215 pool->freeBlocks = tem; | |
3216 memcpy(pool->blocks->s, pool->start, (pool->end - pool->start) * sizeof(XML_Char)); | |
3217 pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
3218 pool->start = pool->blocks->s; | |
3219 pool->end = pool->start + pool->blocks->size; | |
3220 return 1; | |
3221 } | |
3222 } | |
3223 if (pool->blocks && pool->start == pool->blocks->s) { | |
3224 int blockSize = (pool->end - pool->start)*2; | |
3225 pool->blocks = realloc(pool->blocks, offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); | |
3226 if (!pool->blocks) | |
3227 return 0; | |
3228 pool->blocks->size = blockSize; | |
3229 pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
3230 pool->start = pool->blocks->s; | |
3231 pool->end = pool->start + blockSize; | |
3232 } | |
3233 else { | |
3234 BLOCK *tem; | |
3235 int blockSize = pool->end - pool->start; | |
3236 if (blockSize < INIT_BLOCK_SIZE) | |
3237 blockSize = INIT_BLOCK_SIZE; | |
3238 else | |
3239 blockSize *= 2; | |
3240 tem = malloc(offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); | |
3241 if (!tem) | |
3242 return 0; | |
3243 tem->size = blockSize; | |
3244 tem->next = pool->blocks; | |
3245 pool->blocks = tem; | |
3246 memcpy(tem->s, pool->start, (pool->ptr - pool->start) * sizeof(XML_Char)); | |
3247 pool->ptr = tem->s + (pool->ptr - pool->start); | |
3248 pool->start = tem->s; | |
3249 pool->end = tem->s + blockSize; | |
3250 } | |
3251 return 1; | |
3252 } |