1 |
adcroft |
1.1 |
/* |
2 |
|
|
Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd |
3 |
|
|
See the file copying.txt for copying permission. |
4 |
|
|
*/ |
5 |
|
|
|
6 |
|
|
#include "xmldef.h" |
7 |
|
|
#include "xmlparse.h" |
8 |
|
|
#include <stddef.h> |
9 |
|
|
|
10 |
|
|
#ifdef XML_UNICODE |
11 |
|
|
#define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX |
12 |
|
|
#define XmlConvert XmlUtf16Convert |
13 |
|
|
#define XmlGetInternalEncoding XmlGetUtf16InternalEncoding |
14 |
|
|
#define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS |
15 |
|
|
#define XmlEncode XmlUtf16Encode |
16 |
|
|
#define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1)) |
17 |
|
|
typedef unsigned short ICHAR; |
18 |
|
|
#else |
19 |
|
|
#define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX |
20 |
|
|
#define XmlConvert XmlUtf8Convert |
21 |
|
|
#define XmlGetInternalEncoding XmlGetUtf8InternalEncoding |
22 |
|
|
#define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS |
23 |
|
|
#define XmlEncode XmlUtf8Encode |
24 |
|
|
#define MUST_CONVERT(enc, s) (!(enc)->isUtf8) |
25 |
|
|
typedef char ICHAR; |
26 |
|
|
#endif |
27 |
|
|
|
28 |
|
|
|
29 |
|
|
#ifndef XML_NS |
30 |
|
|
|
31 |
|
|
#define XmlInitEncodingNS XmlInitEncoding |
32 |
|
|
#define XmlInitUnknownEncodingNS XmlInitUnknownEncoding |
33 |
|
|
#undef XmlGetInternalEncodingNS |
34 |
|
|
#define XmlGetInternalEncodingNS XmlGetInternalEncoding |
35 |
|
|
#define XmlParseXmlDeclNS XmlParseXmlDecl |
36 |
|
|
|
37 |
|
|
#endif |
38 |
|
|
|
39 |
|
|
#ifdef XML_UNICODE_WCHAR_T |
40 |
|
|
#define XML_T(x) L ## x |
41 |
|
|
#else |
42 |
|
|
#define XML_T(x) x |
43 |
|
|
#endif |
44 |
|
|
|
45 |
|
|
/* Round up n to be a multiple of sz, where sz is a power of 2. */ |
46 |
|
|
#define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1)) |
47 |
|
|
|
48 |
|
|
#include "xmltok.h" |
49 |
|
|
#include "xmlrole.h" |
50 |
|
|
|
51 |
|
|
typedef const XML_Char *KEY; |
52 |
|
|
|
53 |
|
|
typedef struct { |
54 |
|
|
KEY name; |
55 |
|
|
} NAMED; |
56 |
|
|
|
57 |
|
|
typedef struct { |
58 |
|
|
NAMED **v; |
59 |
|
|
size_t size; |
60 |
|
|
size_t used; |
61 |
|
|
size_t usedLim; |
62 |
|
|
} HASH_TABLE; |
63 |
|
|
|
64 |
|
|
typedef struct { |
65 |
|
|
NAMED **p; |
66 |
|
|
NAMED **end; |
67 |
|
|
} HASH_TABLE_ITER; |
68 |
|
|
|
69 |
|
|
#define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ |
70 |
|
|
#define INIT_DATA_BUF_SIZE 1024 |
71 |
|
|
#define INIT_ATTS_SIZE 16 |
72 |
|
|
#define INIT_BLOCK_SIZE 1024 |
73 |
|
|
#define INIT_BUFFER_SIZE 1024 |
74 |
|
|
|
75 |
|
|
#define EXPAND_SPARE 24 |
76 |
|
|
|
77 |
|
|
typedef struct binding { |
78 |
|
|
struct prefix *prefix; |
79 |
|
|
struct binding *nextTagBinding; |
80 |
|
|
struct binding *prevPrefixBinding; |
81 |
|
|
const struct attribute_id *attId; |
82 |
|
|
XML_Char *uri; |
83 |
|
|
int uriLen; |
84 |
|
|
int uriAlloc; |
85 |
|
|
} BINDING; |
86 |
|
|
|
87 |
|
|
typedef struct prefix { |
88 |
|
|
const XML_Char *name; |
89 |
|
|
BINDING *binding; |
90 |
|
|
} PREFIX; |
91 |
|
|
|
92 |
|
|
typedef struct { |
93 |
|
|
const XML_Char *str; |
94 |
|
|
const XML_Char *localPart; |
95 |
|
|
int uriLen; |
96 |
|
|
} TAG_NAME; |
97 |
|
|
|
98 |
|
|
typedef struct tag { |
99 |
|
|
struct tag *parent; |
100 |
|
|
const char *rawName; |
101 |
|
|
int rawNameLength; |
102 |
|
|
TAG_NAME name; |
103 |
|
|
char *buf; |
104 |
|
|
char *bufEnd; |
105 |
|
|
BINDING *bindings; |
106 |
|
|
} TAG; |
107 |
|
|
|
108 |
|
|
typedef struct { |
109 |
|
|
const XML_Char *name; |
110 |
|
|
const XML_Char *textPtr; |
111 |
|
|
int textLen; |
112 |
|
|
const XML_Char *systemId; |
113 |
|
|
const XML_Char *base; |
114 |
|
|
const XML_Char *publicId; |
115 |
|
|
const XML_Char *notation; |
116 |
|
|
char open; |
117 |
|
|
} ENTITY; |
118 |
|
|
|
119 |
|
|
typedef struct block { |
120 |
|
|
struct block *next; |
121 |
|
|
int size; |
122 |
|
|
XML_Char s[1]; |
123 |
|
|
} BLOCK; |
124 |
|
|
|
125 |
|
|
typedef struct { |
126 |
|
|
BLOCK *blocks; |
127 |
|
|
BLOCK *freeBlocks; |
128 |
|
|
const XML_Char *end; |
129 |
|
|
XML_Char *ptr; |
130 |
|
|
XML_Char *start; |
131 |
|
|
} STRING_POOL; |
132 |
|
|
|
133 |
|
|
/* The XML_Char before the name is used to determine whether |
134 |
|
|
an attribute has been specified. */ |
135 |
|
|
typedef struct attribute_id { |
136 |
|
|
XML_Char *name; |
137 |
|
|
PREFIX *prefix; |
138 |
|
|
char maybeTokenized; |
139 |
|
|
char xmlns; |
140 |
|
|
} ATTRIBUTE_ID; |
141 |
|
|
|
142 |
|
|
typedef struct { |
143 |
|
|
const ATTRIBUTE_ID *id; |
144 |
|
|
char isCdata; |
145 |
|
|
const XML_Char *value; |
146 |
|
|
} DEFAULT_ATTRIBUTE; |
147 |
|
|
|
148 |
|
|
typedef struct { |
149 |
|
|
const XML_Char *name; |
150 |
|
|
PREFIX *prefix; |
151 |
|
|
const ATTRIBUTE_ID *idAtt; |
152 |
|
|
int nDefaultAtts; |
153 |
|
|
int allocDefaultAtts; |
154 |
|
|
DEFAULT_ATTRIBUTE *defaultAtts; |
155 |
|
|
} ELEMENT_TYPE; |
156 |
|
|
|
157 |
|
|
typedef struct { |
158 |
|
|
HASH_TABLE generalEntities; |
159 |
|
|
HASH_TABLE elementTypes; |
160 |
|
|
HASH_TABLE attributeIds; |
161 |
|
|
HASH_TABLE prefixes; |
162 |
|
|
STRING_POOL pool; |
163 |
|
|
int complete; |
164 |
|
|
int standalone; |
165 |
|
|
#ifdef XML_DTD |
166 |
|
|
HASH_TABLE paramEntities; |
167 |
|
|
#endif /* XML_DTD */ |
168 |
|
|
PREFIX defaultPrefix; |
169 |
|
|
} DTD; |
170 |
|
|
|
171 |
|
|
typedef struct open_internal_entity { |
172 |
|
|
const char *internalEventPtr; |
173 |
|
|
const char *internalEventEndPtr; |
174 |
|
|
struct open_internal_entity *next; |
175 |
|
|
ENTITY *entity; |
176 |
|
|
} OPEN_INTERNAL_ENTITY; |
177 |
|
|
|
178 |
|
|
typedef enum XML_Error Processor(XML_Parser parser, |
179 |
|
|
const char *start, |
180 |
|
|
const char *end, |
181 |
|
|
const char **endPtr); |
182 |
|
|
|
183 |
|
|
static Processor prologProcessor; |
184 |
|
|
static Processor prologInitProcessor; |
185 |
|
|
static Processor contentProcessor; |
186 |
|
|
static Processor cdataSectionProcessor; |
187 |
|
|
#ifdef XML_DTD |
188 |
|
|
static Processor ignoreSectionProcessor; |
189 |
|
|
#endif /* XML_DTD */ |
190 |
|
|
static Processor epilogProcessor; |
191 |
|
|
static Processor errorProcessor; |
192 |
|
|
static Processor externalEntityInitProcessor; |
193 |
|
|
static Processor externalEntityInitProcessor2; |
194 |
|
|
static Processor externalEntityInitProcessor3; |
195 |
|
|
static Processor externalEntityContentProcessor; |
196 |
|
|
|
197 |
|
|
static enum XML_Error |
198 |
|
|
handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName); |
199 |
|
|
static enum XML_Error |
200 |
|
|
processXmlDecl(XML_Parser parser, int isGeneralTextEntity, const char *, const char *); |
201 |
|
|
static enum XML_Error |
202 |
|
|
initializeEncoding(XML_Parser parser); |
203 |
|
|
static enum XML_Error |
204 |
|
|
doProlog(XML_Parser parser, const ENCODING *enc, const char *s, |
205 |
|
|
const char *end, int tok, const char *next, const char **nextPtr); |
206 |
|
|
|
207 |
|
|
#ifdef XML_DTD |
208 |
|
|
static enum XML_Error |
209 |
|
|
processInternalParamEntity(XML_Parser parser, ENTITY *entity); |
210 |
|
|
#endif |
211 |
|
|
|
212 |
|
|
static enum XML_Error |
213 |
|
|
doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, |
214 |
|
|
const char *start, const char *end, const char **endPtr); |
215 |
|
|
static enum XML_Error |
216 |
|
|
doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr); |
217 |
|
|
#ifdef XML_DTD |
218 |
|
|
static enum XML_Error |
219 |
|
|
doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr); |
220 |
|
|
#endif /* XML_DTD */ |
221 |
|
|
static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *, const char *s, |
222 |
|
|
TAG_NAME *tagNamePtr, BINDING **bindingsPtr); |
223 |
|
|
static |
224 |
|
|
int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr); |
225 |
|
|
static int |
226 |
|
|
defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, int isCdata, int isId, const XML_Char *dfltValue); |
227 |
|
|
static enum XML_Error |
228 |
|
|
storeAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, |
229 |
|
|
STRING_POOL *); |
230 |
|
|
static enum XML_Error |
231 |
|
|
appendAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, |
232 |
|
|
STRING_POOL *); |
233 |
|
|
static ATTRIBUTE_ID * |
234 |
|
|
getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); |
235 |
|
|
static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); |
236 |
|
|
static enum XML_Error |
237 |
|
|
storeEntityValue(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); |
238 |
|
|
static int |
239 |
|
|
reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); |
240 |
|
|
static int |
241 |
|
|
reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); |
242 |
|
|
static void |
243 |
|
|
reportDefault(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); |
244 |
|
|
|
245 |
|
|
static const XML_Char *getContext(XML_Parser parser); |
246 |
|
|
static int setContext(XML_Parser parser, const XML_Char *context); |
247 |
|
|
static void normalizePublicId(XML_Char *s); |
248 |
|
|
static int dtdInit(DTD *); |
249 |
|
|
static void dtdDestroy(DTD *); |
250 |
|
|
static int dtdCopy(DTD *newDtd, const DTD *oldDtd); |
251 |
|
|
static int copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *); |
252 |
|
|
#ifdef XML_DTD |
253 |
|
|
static void dtdSwap(DTD *, DTD *); |
254 |
|
|
#endif /* XML_DTD */ |
255 |
|
|
static NAMED *lookup(HASH_TABLE *table, KEY name, size_t createSize); |
256 |
|
|
static void hashTableInit(HASH_TABLE *); |
257 |
|
|
static void hashTableDestroy(HASH_TABLE *); |
258 |
|
|
static void hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *); |
259 |
|
|
static NAMED *hashTableIterNext(HASH_TABLE_ITER *); |
260 |
|
|
static void poolInit(STRING_POOL *); |
261 |
|
|
static void poolClear(STRING_POOL *); |
262 |
|
|
static void poolDestroy(STRING_POOL *); |
263 |
|
|
static XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, |
264 |
|
|
const char *ptr, const char *end); |
265 |
|
|
static XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, |
266 |
|
|
const char *ptr, const char *end); |
267 |
|
|
static int poolGrow(STRING_POOL *pool); |
268 |
|
|
static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s); |
269 |
|
|
static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n); |
270 |
|
|
|
271 |
|
|
#define poolStart(pool) ((pool)->start) |
272 |
|
|
#define poolEnd(pool) ((pool)->ptr) |
273 |
|
|
#define poolLength(pool) ((pool)->ptr - (pool)->start) |
274 |
|
|
#define poolChop(pool) ((void)--(pool->ptr)) |
275 |
|
|
#define poolLastChar(pool) (((pool)->ptr)[-1]) |
276 |
|
|
#define poolDiscard(pool) ((pool)->ptr = (pool)->start) |
277 |
|
|
#define poolFinish(pool) ((pool)->start = (pool)->ptr) |
278 |
|
|
#define poolAppendChar(pool, c) \ |
279 |
|
|
(((pool)->ptr == (pool)->end && !poolGrow(pool)) \ |
280 |
|
|
? 0 \ |
281 |
|
|
: ((*((pool)->ptr)++ = c), 1)) |
282 |
|
|
|
283 |
|
|
typedef struct { |
284 |
|
|
/* The first member must be userData so that the XML_GetUserData macro works. */ |
285 |
|
|
void *m_userData; |
286 |
|
|
void *m_handlerArg; |
287 |
|
|
char *m_buffer; |
288 |
|
|
/* first character to be parsed */ |
289 |
|
|
const char *m_bufferPtr; |
290 |
|
|
/* past last character to be parsed */ |
291 |
|
|
char *m_bufferEnd; |
292 |
|
|
/* allocated end of buffer */ |
293 |
|
|
const char *m_bufferLim; |
294 |
|
|
long m_parseEndByteIndex; |
295 |
|
|
const char *m_parseEndPtr; |
296 |
|
|
XML_Char *m_dataBuf; |
297 |
|
|
XML_Char *m_dataBufEnd; |
298 |
|
|
XML_StartElementHandler m_startElementHandler; |
299 |
|
|
XML_EndElementHandler m_endElementHandler; |
300 |
|
|
XML_CharacterDataHandler m_characterDataHandler; |
301 |
|
|
XML_ProcessingInstructionHandler m_processingInstructionHandler; |
302 |
|
|
XML_CommentHandler m_commentHandler; |
303 |
|
|
XML_StartCdataSectionHandler m_startCdataSectionHandler; |
304 |
|
|
XML_EndCdataSectionHandler m_endCdataSectionHandler; |
305 |
|
|
XML_DefaultHandler m_defaultHandler; |
306 |
|
|
XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler; |
307 |
|
|
XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler; |
308 |
|
|
XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; |
309 |
|
|
XML_NotationDeclHandler m_notationDeclHandler; |
310 |
|
|
XML_ExternalParsedEntityDeclHandler m_externalParsedEntityDeclHandler; |
311 |
|
|
XML_InternalParsedEntityDeclHandler m_internalParsedEntityDeclHandler; |
312 |
|
|
XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; |
313 |
|
|
XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; |
314 |
|
|
XML_NotStandaloneHandler m_notStandaloneHandler; |
315 |
|
|
XML_ExternalEntityRefHandler m_externalEntityRefHandler; |
316 |
|
|
void *m_externalEntityRefHandlerArg; |
317 |
|
|
XML_UnknownEncodingHandler m_unknownEncodingHandler; |
318 |
|
|
const ENCODING *m_encoding; |
319 |
|
|
INIT_ENCODING m_initEncoding; |
320 |
|
|
const ENCODING *m_internalEncoding; |
321 |
|
|
const XML_Char *m_protocolEncodingName; |
322 |
|
|
int m_ns; |
323 |
|
|
void *m_unknownEncodingMem; |
324 |
|
|
void *m_unknownEncodingData; |
325 |
|
|
void *m_unknownEncodingHandlerData; |
326 |
|
|
void (*m_unknownEncodingRelease)(void *); |
327 |
|
|
PROLOG_STATE m_prologState; |
328 |
|
|
Processor *m_processor; |
329 |
|
|
enum XML_Error m_errorCode; |
330 |
|
|
const char *m_eventPtr; |
331 |
|
|
const char *m_eventEndPtr; |
332 |
|
|
const char *m_positionPtr; |
333 |
|
|
OPEN_INTERNAL_ENTITY *m_openInternalEntities; |
334 |
|
|
int m_defaultExpandInternalEntities; |
335 |
|
|
int m_tagLevel; |
336 |
|
|
ENTITY *m_declEntity; |
337 |
|
|
const XML_Char *m_declNotationName; |
338 |
|
|
const XML_Char *m_declNotationPublicId; |
339 |
|
|
ELEMENT_TYPE *m_declElementType; |
340 |
|
|
ATTRIBUTE_ID *m_declAttributeId; |
341 |
|
|
char m_declAttributeIsCdata; |
342 |
|
|
char m_declAttributeIsId; |
343 |
|
|
DTD m_dtd; |
344 |
|
|
const XML_Char *m_curBase; |
345 |
|
|
TAG *m_tagStack; |
346 |
|
|
TAG *m_freeTagList; |
347 |
|
|
BINDING *m_inheritedBindings; |
348 |
|
|
BINDING *m_freeBindingList; |
349 |
|
|
int m_attsSize; |
350 |
|
|
int m_nSpecifiedAtts; |
351 |
|
|
int m_idAttIndex; |
352 |
|
|
ATTRIBUTE *m_atts; |
353 |
|
|
POSITION m_position; |
354 |
|
|
STRING_POOL m_tempPool; |
355 |
|
|
STRING_POOL m_temp2Pool; |
356 |
|
|
char *m_groupConnector; |
357 |
|
|
unsigned m_groupSize; |
358 |
|
|
int m_hadExternalDoctype; |
359 |
|
|
XML_Char m_namespaceSeparator; |
360 |
|
|
#ifdef XML_DTD |
361 |
|
|
enum XML_ParamEntityParsing m_paramEntityParsing; |
362 |
|
|
XML_Parser m_parentParser; |
363 |
|
|
#endif |
364 |
|
|
} Parser; |
365 |
|
|
|
366 |
|
|
#define userData (((Parser *)parser)->m_userData) |
367 |
|
|
#define handlerArg (((Parser *)parser)->m_handlerArg) |
368 |
|
|
#define startElementHandler (((Parser *)parser)->m_startElementHandler) |
369 |
|
|
#define endElementHandler (((Parser *)parser)->m_endElementHandler) |
370 |
|
|
#define characterDataHandler (((Parser *)parser)->m_characterDataHandler) |
371 |
|
|
#define processingInstructionHandler (((Parser *)parser)->m_processingInstructionHandler) |
372 |
|
|
#define commentHandler (((Parser *)parser)->m_commentHandler) |
373 |
|
|
#define startCdataSectionHandler (((Parser *)parser)->m_startCdataSectionHandler) |
374 |
|
|
#define endCdataSectionHandler (((Parser *)parser)->m_endCdataSectionHandler) |
375 |
|
|
#define defaultHandler (((Parser *)parser)->m_defaultHandler) |
376 |
|
|
#define startDoctypeDeclHandler (((Parser *)parser)->m_startDoctypeDeclHandler) |
377 |
|
|
#define endDoctypeDeclHandler (((Parser *)parser)->m_endDoctypeDeclHandler) |
378 |
|
|
#define unparsedEntityDeclHandler (((Parser *)parser)->m_unparsedEntityDeclHandler) |
379 |
|
|
#define notationDeclHandler (((Parser *)parser)->m_notationDeclHandler) |
380 |
|
|
#define externalParsedEntityDeclHandler (((Parser *)parser)->m_externalParsedEntityDeclHandler) |
381 |
|
|
#define internalParsedEntityDeclHandler (((Parser *)parser)->m_internalParsedEntityDeclHandler) |
382 |
|
|
#define startNamespaceDeclHandler (((Parser *)parser)->m_startNamespaceDeclHandler) |
383 |
|
|
#define endNamespaceDeclHandler (((Parser *)parser)->m_endNamespaceDeclHandler) |
384 |
|
|
#define notStandaloneHandler (((Parser *)parser)->m_notStandaloneHandler) |
385 |
|
|
#define externalEntityRefHandler (((Parser *)parser)->m_externalEntityRefHandler) |
386 |
|
|
#define externalEntityRefHandlerArg (((Parser *)parser)->m_externalEntityRefHandlerArg) |
387 |
|
|
#define unknownEncodingHandler (((Parser *)parser)->m_unknownEncodingHandler) |
388 |
|
|
#define encoding (((Parser *)parser)->m_encoding) |
389 |
|
|
#define initEncoding (((Parser *)parser)->m_initEncoding) |
390 |
|
|
#define internalEncoding (((Parser *)parser)->m_internalEncoding) |
391 |
|
|
#define unknownEncodingMem (((Parser *)parser)->m_unknownEncodingMem) |
392 |
|
|
#define unknownEncodingData (((Parser *)parser)->m_unknownEncodingData) |
393 |
|
|
#define unknownEncodingHandlerData \ |
394 |
|
|
(((Parser *)parser)->m_unknownEncodingHandlerData) |
395 |
|
|
#define unknownEncodingRelease (((Parser *)parser)->m_unknownEncodingRelease) |
396 |
|
|
#define protocolEncodingName (((Parser *)parser)->m_protocolEncodingName) |
397 |
|
|
#define ns (((Parser *)parser)->m_ns) |
398 |
|
|
#define prologState (((Parser *)parser)->m_prologState) |
399 |
|
|
#define processor (((Parser *)parser)->m_processor) |
400 |
|
|
#define errorCode (((Parser *)parser)->m_errorCode) |
401 |
|
|
#define eventPtr (((Parser *)parser)->m_eventPtr) |
402 |
|
|
#define eventEndPtr (((Parser *)parser)->m_eventEndPtr) |
403 |
|
|
#define positionPtr (((Parser *)parser)->m_positionPtr) |
404 |
|
|
#define position (((Parser *)parser)->m_position) |
405 |
|
|
#define openInternalEntities (((Parser *)parser)->m_openInternalEntities) |
406 |
|
|
#define defaultExpandInternalEntities (((Parser *)parser)->m_defaultExpandInternalEntities) |
407 |
|
|
#define tagLevel (((Parser *)parser)->m_tagLevel) |
408 |
|
|
#define buffer (((Parser *)parser)->m_buffer) |
409 |
|
|
#define bufferPtr (((Parser *)parser)->m_bufferPtr) |
410 |
|
|
#define bufferEnd (((Parser *)parser)->m_bufferEnd) |
411 |
|
|
#define parseEndByteIndex (((Parser *)parser)->m_parseEndByteIndex) |
412 |
|
|
#define parseEndPtr (((Parser *)parser)->m_parseEndPtr) |
413 |
|
|
#define bufferLim (((Parser *)parser)->m_bufferLim) |
414 |
|
|
#define dataBuf (((Parser *)parser)->m_dataBuf) |
415 |
|
|
#define dataBufEnd (((Parser *)parser)->m_dataBufEnd) |
416 |
|
|
#define dtd (((Parser *)parser)->m_dtd) |
417 |
|
|
#define curBase (((Parser *)parser)->m_curBase) |
418 |
|
|
#define declEntity (((Parser *)parser)->m_declEntity) |
419 |
|
|
#define declNotationName (((Parser *)parser)->m_declNotationName) |
420 |
|
|
#define declNotationPublicId (((Parser *)parser)->m_declNotationPublicId) |
421 |
|
|
#define declElementType (((Parser *)parser)->m_declElementType) |
422 |
|
|
#define declAttributeId (((Parser *)parser)->m_declAttributeId) |
423 |
|
|
#define declAttributeIsCdata (((Parser *)parser)->m_declAttributeIsCdata) |
424 |
|
|
#define declAttributeIsId (((Parser *)parser)->m_declAttributeIsId) |
425 |
|
|
#define freeTagList (((Parser *)parser)->m_freeTagList) |
426 |
|
|
#define freeBindingList (((Parser *)parser)->m_freeBindingList) |
427 |
|
|
#define inheritedBindings (((Parser *)parser)->m_inheritedBindings) |
428 |
|
|
#define tagStack (((Parser *)parser)->m_tagStack) |
429 |
|
|
#define atts (((Parser *)parser)->m_atts) |
430 |
|
|
#define attsSize (((Parser *)parser)->m_attsSize) |
431 |
|
|
#define nSpecifiedAtts (((Parser *)parser)->m_nSpecifiedAtts) |
432 |
|
|
#define idAttIndex (((Parser *)parser)->m_idAttIndex) |
433 |
|
|
#define tempPool (((Parser *)parser)->m_tempPool) |
434 |
|
|
#define temp2Pool (((Parser *)parser)->m_temp2Pool) |
435 |
|
|
#define groupConnector (((Parser *)parser)->m_groupConnector) |
436 |
|
|
#define groupSize (((Parser *)parser)->m_groupSize) |
437 |
|
|
#define hadExternalDoctype (((Parser *)parser)->m_hadExternalDoctype) |
438 |
|
|
#define namespaceSeparator (((Parser *)parser)->m_namespaceSeparator) |
439 |
|
|
#ifdef XML_DTD |
440 |
|
|
#define parentParser (((Parser *)parser)->m_parentParser) |
441 |
|
|
#define paramEntityParsing (((Parser *)parser)->m_paramEntityParsing) |
442 |
|
|
#endif /* XML_DTD */ |
443 |
|
|
|
444 |
|
|
#ifdef _MSC_VER |
445 |
|
|
#ifdef _DEBUG |
446 |
|
|
Parser *asParser(XML_Parser parser) |
447 |
|
|
{ |
448 |
|
|
return parser; |
449 |
|
|
} |
450 |
|
|
#endif |
451 |
|
|
#endif |
452 |
|
|
|
453 |
|
|
XML_Parser XML_ParserCreate(const XML_Char *encodingName) |
454 |
|
|
{ |
455 |
|
|
XML_Parser parser = malloc(sizeof(Parser)); |
456 |
|
|
if (!parser) |
457 |
|
|
return parser; |
458 |
|
|
processor = prologInitProcessor; |
459 |
|
|
XmlPrologStateInit(&prologState); |
460 |
|
|
userData = 0; |
461 |
|
|
handlerArg = 0; |
462 |
|
|
startElementHandler = 0; |
463 |
|
|
endElementHandler = 0; |
464 |
|
|
characterDataHandler = 0; |
465 |
|
|
processingInstructionHandler = 0; |
466 |
|
|
commentHandler = 0; |
467 |
|
|
startCdataSectionHandler = 0; |
468 |
|
|
endCdataSectionHandler = 0; |
469 |
|
|
defaultHandler = 0; |
470 |
|
|
startDoctypeDeclHandler = 0; |
471 |
|
|
endDoctypeDeclHandler = 0; |
472 |
|
|
unparsedEntityDeclHandler = 0; |
473 |
|
|
notationDeclHandler = 0; |
474 |
|
|
externalParsedEntityDeclHandler = 0; |
475 |
|
|
internalParsedEntityDeclHandler = 0; |
476 |
|
|
startNamespaceDeclHandler = 0; |
477 |
|
|
endNamespaceDeclHandler = 0; |
478 |
|
|
notStandaloneHandler = 0; |
479 |
|
|
externalEntityRefHandler = 0; |
480 |
|
|
externalEntityRefHandlerArg = parser; |
481 |
|
|
unknownEncodingHandler = 0; |
482 |
|
|
buffer = 0; |
483 |
|
|
bufferPtr = 0; |
484 |
|
|
bufferEnd = 0; |
485 |
|
|
parseEndByteIndex = 0; |
486 |
|
|
parseEndPtr = 0; |
487 |
|
|
bufferLim = 0; |
488 |
|
|
declElementType = 0; |
489 |
|
|
declAttributeId = 0; |
490 |
|
|
declEntity = 0; |
491 |
|
|
declNotationName = 0; |
492 |
|
|
declNotationPublicId = 0; |
493 |
|
|
memset(&position, 0, sizeof(POSITION)); |
494 |
|
|
errorCode = XML_ERROR_NONE; |
495 |
|
|
eventPtr = 0; |
496 |
|
|
eventEndPtr = 0; |
497 |
|
|
positionPtr = 0; |
498 |
|
|
openInternalEntities = 0; |
499 |
|
|
tagLevel = 0; |
500 |
|
|
tagStack = 0; |
501 |
|
|
freeTagList = 0; |
502 |
|
|
freeBindingList = 0; |
503 |
|
|
inheritedBindings = 0; |
504 |
|
|
attsSize = INIT_ATTS_SIZE; |
505 |
|
|
atts = malloc(attsSize * sizeof(ATTRIBUTE)); |
506 |
|
|
nSpecifiedAtts = 0; |
507 |
|
|
dataBuf = malloc(INIT_DATA_BUF_SIZE * sizeof(XML_Char)); |
508 |
|
|
groupSize = 0; |
509 |
|
|
groupConnector = 0; |
510 |
|
|
hadExternalDoctype = 0; |
511 |
|
|
unknownEncodingMem = 0; |
512 |
|
|
unknownEncodingRelease = 0; |
513 |
|
|
unknownEncodingData = 0; |
514 |
|
|
unknownEncodingHandlerData = 0; |
515 |
|
|
namespaceSeparator = '!'; |
516 |
|
|
#ifdef XML_DTD |
517 |
|
|
parentParser = 0; |
518 |
|
|
paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; |
519 |
|
|
#endif |
520 |
|
|
ns = 0; |
521 |
|
|
poolInit(&tempPool); |
522 |
|
|
poolInit(&temp2Pool); |
523 |
|
|
protocolEncodingName = encodingName ? poolCopyString(&tempPool, encodingName) : 0; |
524 |
|
|
curBase = 0; |
525 |
|
|
if (!dtdInit(&dtd) || !atts || !dataBuf |
526 |
|
|
|| (encodingName && !protocolEncodingName)) { |
527 |
|
|
XML_ParserFree(parser); |
528 |
|
|
return 0; |
529 |
|
|
} |
530 |
|
|
dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE; |
531 |
|
|
XmlInitEncoding(&initEncoding, &encoding, 0); |
532 |
|
|
internalEncoding = XmlGetInternalEncoding(); |
533 |
|
|
return parser; |
534 |
|
|
} |
535 |
|
|
|
536 |
|
|
XML_Parser XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) |
537 |
|
|
{ |
538 |
|
|
static |
539 |
|
|
const XML_Char implicitContext[] = { |
540 |
|
|
XML_T('x'), XML_T('m'), XML_T('l'), XML_T('='), |
541 |
|
|
XML_T('h'), XML_T('t'), XML_T('t'), XML_T('p'), XML_T(':'), |
542 |
|
|
XML_T('/'), XML_T('/'), XML_T('w'), XML_T('w'), XML_T('w'), |
543 |
|
|
XML_T('.'), XML_T('w'), XML_T('3'), |
544 |
|
|
XML_T('.'), XML_T('o'), XML_T('r'), XML_T('g'), |
545 |
|
|
XML_T('/'), XML_T('X'), XML_T('M'), XML_T('L'), |
546 |
|
|
XML_T('/'), XML_T('1'), XML_T('9'), XML_T('9'), XML_T('8'), |
547 |
|
|
XML_T('/'), XML_T('n'), XML_T('a'), XML_T('m'), XML_T('e'), |
548 |
|
|
XML_T('s'), XML_T('p'), XML_T('a'), XML_T('c'), XML_T('e'), |
549 |
|
|
XML_T('\0') |
550 |
|
|
}; |
551 |
|
|
|
552 |
|
|
XML_Parser parser = XML_ParserCreate(encodingName); |
553 |
|
|
if (parser) { |
554 |
|
|
XmlInitEncodingNS(&initEncoding, &encoding, 0); |
555 |
|
|
ns = 1; |
556 |
|
|
internalEncoding = XmlGetInternalEncodingNS(); |
557 |
|
|
namespaceSeparator = nsSep; |
558 |
|
|
} |
559 |
|
|
if (!setContext(parser, implicitContext)) { |
560 |
|
|
XML_ParserFree(parser); |
561 |
|
|
return 0; |
562 |
|
|
} |
563 |
|
|
return parser; |
564 |
|
|
} |
565 |
|
|
|
566 |
|
|
int XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) |
567 |
|
|
{ |
568 |
|
|
if (!encodingName) |
569 |
|
|
protocolEncodingName = 0; |
570 |
|
|
else { |
571 |
|
|
protocolEncodingName = poolCopyString(&tempPool, encodingName); |
572 |
|
|
if (!protocolEncodingName) |
573 |
|
|
return 0; |
574 |
|
|
} |
575 |
|
|
return 1; |
576 |
|
|
} |
577 |
|
|
|
578 |
|
|
XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser, |
579 |
|
|
const XML_Char *context, |
580 |
|
|
const XML_Char *encodingName) |
581 |
|
|
{ |
582 |
|
|
XML_Parser parser = oldParser; |
583 |
|
|
DTD *oldDtd = &dtd; |
584 |
|
|
XML_StartElementHandler oldStartElementHandler = startElementHandler; |
585 |
|
|
XML_EndElementHandler oldEndElementHandler = endElementHandler; |
586 |
|
|
XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler; |
587 |
|
|
XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler; |
588 |
|
|
XML_CommentHandler oldCommentHandler = commentHandler; |
589 |
|
|
XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler; |
590 |
|
|
XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler; |
591 |
|
|
XML_DefaultHandler oldDefaultHandler = defaultHandler; |
592 |
|
|
XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler = unparsedEntityDeclHandler; |
593 |
|
|
XML_NotationDeclHandler oldNotationDeclHandler = notationDeclHandler; |
594 |
|
|
XML_ExternalParsedEntityDeclHandler oldExternalParsedEntityDeclHandler = externalParsedEntityDeclHandler; |
595 |
|
|
XML_InternalParsedEntityDeclHandler oldInternalParsedEntityDeclHandler = internalParsedEntityDeclHandler; |
596 |
|
|
XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler = startNamespaceDeclHandler; |
597 |
|
|
XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler = endNamespaceDeclHandler; |
598 |
|
|
XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler; |
599 |
|
|
XML_ExternalEntityRefHandler oldExternalEntityRefHandler = externalEntityRefHandler; |
600 |
|
|
XML_UnknownEncodingHandler oldUnknownEncodingHandler = unknownEncodingHandler; |
601 |
|
|
void *oldUserData = userData; |
602 |
|
|
void *oldHandlerArg = handlerArg; |
603 |
|
|
int oldDefaultExpandInternalEntities = defaultExpandInternalEntities; |
604 |
|
|
void *oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg; |
605 |
|
|
#ifdef XML_DTD |
606 |
|
|
int oldParamEntityParsing = paramEntityParsing; |
607 |
|
|
#endif |
608 |
|
|
parser = (ns |
609 |
|
|
? XML_ParserCreateNS(encodingName, namespaceSeparator) |
610 |
|
|
: XML_ParserCreate(encodingName)); |
611 |
|
|
if (!parser) |
612 |
|
|
return 0; |
613 |
|
|
startElementHandler = oldStartElementHandler; |
614 |
|
|
endElementHandler = oldEndElementHandler; |
615 |
|
|
characterDataHandler = oldCharacterDataHandler; |
616 |
|
|
processingInstructionHandler = oldProcessingInstructionHandler; |
617 |
|
|
commentHandler = oldCommentHandler; |
618 |
|
|
startCdataSectionHandler = oldStartCdataSectionHandler; |
619 |
|
|
endCdataSectionHandler = oldEndCdataSectionHandler; |
620 |
|
|
defaultHandler = oldDefaultHandler; |
621 |
|
|
unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler; |
622 |
|
|
notationDeclHandler = oldNotationDeclHandler; |
623 |
|
|
externalParsedEntityDeclHandler = oldExternalParsedEntityDeclHandler; |
624 |
|
|
internalParsedEntityDeclHandler = oldInternalParsedEntityDeclHandler; |
625 |
|
|
startNamespaceDeclHandler = oldStartNamespaceDeclHandler; |
626 |
|
|
endNamespaceDeclHandler = oldEndNamespaceDeclHandler; |
627 |
|
|
notStandaloneHandler = oldNotStandaloneHandler; |
628 |
|
|
externalEntityRefHandler = oldExternalEntityRefHandler; |
629 |
|
|
unknownEncodingHandler = oldUnknownEncodingHandler; |
630 |
|
|
userData = oldUserData; |
631 |
|
|
if (oldUserData == oldHandlerArg) |
632 |
|
|
handlerArg = userData; |
633 |
|
|
else |
634 |
|
|
handlerArg = parser; |
635 |
|
|
if (oldExternalEntityRefHandlerArg != oldParser) |
636 |
|
|
externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; |
637 |
|
|
defaultExpandInternalEntities = oldDefaultExpandInternalEntities; |
638 |
|
|
#ifdef XML_DTD |
639 |
|
|
paramEntityParsing = oldParamEntityParsing; |
640 |
|
|
if (context) { |
641 |
|
|
#endif /* XML_DTD */ |
642 |
|
|
if (!dtdCopy(&dtd, oldDtd) || !setContext(parser, context)) { |
643 |
|
|
XML_ParserFree(parser); |
644 |
|
|
return 0; |
645 |
|
|
} |
646 |
|
|
processor = externalEntityInitProcessor; |
647 |
|
|
#ifdef XML_DTD |
648 |
|
|
} |
649 |
|
|
else { |
650 |
|
|
dtdSwap(&dtd, oldDtd); |
651 |
|
|
parentParser = oldParser; |
652 |
|
|
XmlPrologStateInitExternalEntity(&prologState); |
653 |
|
|
dtd.complete = 1; |
654 |
|
|
hadExternalDoctype = 1; |
655 |
|
|
} |
656 |
|
|
#endif /* XML_DTD */ |
657 |
|
|
return parser; |
658 |
|
|
} |
659 |
|
|
|
660 |
|
|
static |
661 |
|
|
void destroyBindings(BINDING *bindings) |
662 |
|
|
{ |
663 |
|
|
for (;;) { |
664 |
|
|
BINDING *b = bindings; |
665 |
|
|
if (!b) |
666 |
|
|
break; |
667 |
|
|
bindings = b->nextTagBinding; |
668 |
|
|
free(b->uri); |
669 |
|
|
free(b); |
670 |
|
|
} |
671 |
|
|
} |
672 |
|
|
|
673 |
|
|
void XML_ParserFree(XML_Parser parser) |
674 |
|
|
{ |
675 |
|
|
for (;;) { |
676 |
|
|
TAG *p; |
677 |
|
|
if (tagStack == 0) { |
678 |
|
|
if (freeTagList == 0) |
679 |
|
|
break; |
680 |
|
|
tagStack = freeTagList; |
681 |
|
|
freeTagList = 0; |
682 |
|
|
} |
683 |
|
|
p = tagStack; |
684 |
|
|
tagStack = tagStack->parent; |
685 |
|
|
free(p->buf); |
686 |
|
|
destroyBindings(p->bindings); |
687 |
|
|
free(p); |
688 |
|
|
} |
689 |
|
|
destroyBindings(freeBindingList); |
690 |
|
|
destroyBindings(inheritedBindings); |
691 |
|
|
poolDestroy(&tempPool); |
692 |
|
|
poolDestroy(&temp2Pool); |
693 |
|
|
#ifdef XML_DTD |
694 |
|
|
if (parentParser) { |
695 |
|
|
if (hadExternalDoctype) |
696 |
|
|
dtd.complete = 0; |
697 |
|
|
dtdSwap(&dtd, &((Parser *)parentParser)->m_dtd); |
698 |
|
|
} |
699 |
|
|
#endif /* XML_DTD */ |
700 |
|
|
dtdDestroy(&dtd); |
701 |
|
|
free((void *)atts); |
702 |
|
|
free(groupConnector); |
703 |
|
|
free(buffer); |
704 |
|
|
free(dataBuf); |
705 |
|
|
free(unknownEncodingMem); |
706 |
|
|
if (unknownEncodingRelease) |
707 |
|
|
unknownEncodingRelease(unknownEncodingData); |
708 |
|
|
free(parser); |
709 |
|
|
} |
710 |
|
|
|
711 |
|
|
void XML_UseParserAsHandlerArg(XML_Parser parser) |
712 |
|
|
{ |
713 |
|
|
handlerArg = parser; |
714 |
|
|
} |
715 |
|
|
|
716 |
|
|
void XML_SetUserData(XML_Parser parser, void *p) |
717 |
|
|
{ |
718 |
|
|
if (handlerArg == userData) |
719 |
|
|
handlerArg = userData = p; |
720 |
|
|
else |
721 |
|
|
userData = p; |
722 |
|
|
} |
723 |
|
|
|
724 |
|
|
int XML_SetBase(XML_Parser parser, const XML_Char *p) |
725 |
|
|
{ |
726 |
|
|
if (p) { |
727 |
|
|
p = poolCopyString(&dtd.pool, p); |
728 |
|
|
if (!p) |
729 |
|
|
return 0; |
730 |
|
|
curBase = p; |
731 |
|
|
} |
732 |
|
|
else |
733 |
|
|
curBase = 0; |
734 |
|
|
return 1; |
735 |
|
|
} |
736 |
|
|
|
737 |
|
|
const XML_Char *XML_GetBase(XML_Parser parser) |
738 |
|
|
{ |
739 |
|
|
return curBase; |
740 |
|
|
} |
741 |
|
|
|
742 |
|
|
int XML_GetSpecifiedAttributeCount(XML_Parser parser) |
743 |
|
|
{ |
744 |
|
|
return nSpecifiedAtts; |
745 |
|
|
} |
746 |
|
|
|
747 |
|
|
int XML_GetIdAttributeIndex(XML_Parser parser) |
748 |
|
|
{ |
749 |
|
|
return idAttIndex; |
750 |
|
|
} |
751 |
|
|
|
752 |
|
|
void XML_SetElementHandler(XML_Parser parser, |
753 |
|
|
XML_StartElementHandler start, |
754 |
|
|
XML_EndElementHandler end) |
755 |
|
|
{ |
756 |
|
|
startElementHandler = start; |
757 |
|
|
endElementHandler = end; |
758 |
|
|
} |
759 |
|
|
|
760 |
|
|
void XML_SetCharacterDataHandler(XML_Parser parser, |
761 |
|
|
XML_CharacterDataHandler handler) |
762 |
|
|
{ |
763 |
|
|
characterDataHandler = handler; |
764 |
|
|
} |
765 |
|
|
|
766 |
|
|
void XML_SetProcessingInstructionHandler(XML_Parser parser, |
767 |
|
|
XML_ProcessingInstructionHandler handler) |
768 |
|
|
{ |
769 |
|
|
processingInstructionHandler = handler; |
770 |
|
|
} |
771 |
|
|
|
772 |
|
|
void XML_SetCommentHandler(XML_Parser parser, |
773 |
|
|
XML_CommentHandler handler) |
774 |
|
|
{ |
775 |
|
|
commentHandler = handler; |
776 |
|
|
} |
777 |
|
|
|
778 |
|
|
void XML_SetCdataSectionHandler(XML_Parser parser, |
779 |
|
|
XML_StartCdataSectionHandler start, |
780 |
|
|
XML_EndCdataSectionHandler end) |
781 |
|
|
{ |
782 |
|
|
startCdataSectionHandler = start; |
783 |
|
|
endCdataSectionHandler = end; |
784 |
|
|
} |
785 |
|
|
|
786 |
|
|
void XML_SetDefaultHandler(XML_Parser parser, |
787 |
|
|
XML_DefaultHandler handler) |
788 |
|
|
{ |
789 |
|
|
defaultHandler = handler; |
790 |
|
|
defaultExpandInternalEntities = 0; |
791 |
|
|
} |
792 |
|
|
|
793 |
|
|
void XML_SetDefaultHandlerExpand(XML_Parser parser, |
794 |
|
|
XML_DefaultHandler handler) |
795 |
|
|
{ |
796 |
|
|
defaultHandler = handler; |
797 |
|
|
defaultExpandInternalEntities = 1; |
798 |
|
|
} |
799 |
|
|
|
800 |
|
|
void XML_SetDoctypeDeclHandler(XML_Parser parser, |
801 |
|
|
XML_StartDoctypeDeclHandler start, |
802 |
|
|
XML_EndDoctypeDeclHandler end) |
803 |
|
|
{ |
804 |
|
|
startDoctypeDeclHandler = start; |
805 |
|
|
endDoctypeDeclHandler = end; |
806 |
|
|
} |
807 |
|
|
|
808 |
|
|
void XML_SetUnparsedEntityDeclHandler(XML_Parser parser, |
809 |
|
|
XML_UnparsedEntityDeclHandler handler) |
810 |
|
|
{ |
811 |
|
|
unparsedEntityDeclHandler = handler; |
812 |
|
|
} |
813 |
|
|
|
814 |
|
|
void XML_SetExternalParsedEntityDeclHandler(XML_Parser parser, |
815 |
|
|
XML_ExternalParsedEntityDeclHandler handler) |
816 |
|
|
{ |
817 |
|
|
externalParsedEntityDeclHandler = handler; |
818 |
|
|
} |
819 |
|
|
|
820 |
|
|
void XML_SetInternalParsedEntityDeclHandler(XML_Parser parser, |
821 |
|
|
XML_InternalParsedEntityDeclHandler handler) |
822 |
|
|
{ |
823 |
|
|
internalParsedEntityDeclHandler = handler; |
824 |
|
|
} |
825 |
|
|
|
826 |
|
|
void XML_SetNotationDeclHandler(XML_Parser parser, |
827 |
|
|
XML_NotationDeclHandler handler) |
828 |
|
|
{ |
829 |
|
|
notationDeclHandler = handler; |
830 |
|
|
} |
831 |
|
|
|
832 |
|
|
void XML_SetNamespaceDeclHandler(XML_Parser parser, |
833 |
|
|
XML_StartNamespaceDeclHandler start, |
834 |
|
|
XML_EndNamespaceDeclHandler end) |
835 |
|
|
{ |
836 |
|
|
startNamespaceDeclHandler = start; |
837 |
|
|
endNamespaceDeclHandler = end; |
838 |
|
|
} |
839 |
|
|
|
840 |
|
|
void XML_SetNotStandaloneHandler(XML_Parser parser, |
841 |
|
|
XML_NotStandaloneHandler handler) |
842 |
|
|
{ |
843 |
|
|
notStandaloneHandler = handler; |
844 |
|
|
} |
845 |
|
|
|
846 |
|
|
void XML_SetExternalEntityRefHandler(XML_Parser parser, |
847 |
|
|
XML_ExternalEntityRefHandler handler) |
848 |
|
|
{ |
849 |
|
|
externalEntityRefHandler = handler; |
850 |
|
|
} |
851 |
|
|
|
852 |
|
|
void XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) |
853 |
|
|
{ |
854 |
|
|
if (arg) |
855 |
|
|
externalEntityRefHandlerArg = arg; |
856 |
|
|
else |
857 |
|
|
externalEntityRefHandlerArg = parser; |
858 |
|
|
} |
859 |
|
|
|
860 |
|
|
void XML_SetUnknownEncodingHandler(XML_Parser parser, |
861 |
|
|
XML_UnknownEncodingHandler handler, |
862 |
|
|
void *data) |
863 |
|
|
{ |
864 |
|
|
unknownEncodingHandler = handler; |
865 |
|
|
unknownEncodingHandlerData = data; |
866 |
|
|
} |
867 |
|
|
|
868 |
|
|
int XML_SetParamEntityParsing(XML_Parser parser, |
869 |
|
|
enum XML_ParamEntityParsing parsing) |
870 |
|
|
{ |
871 |
|
|
#ifdef XML_DTD |
872 |
|
|
paramEntityParsing = parsing; |
873 |
|
|
return 1; |
874 |
|
|
#else |
875 |
|
|
return parsing == XML_PARAM_ENTITY_PARSING_NEVER; |
876 |
|
|
#endif |
877 |
|
|
} |
878 |
|
|
|
879 |
|
|
int XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) |
880 |
|
|
{ |
881 |
|
|
if (len == 0) { |
882 |
|
|
if (!isFinal) |
883 |
|
|
return 1; |
884 |
|
|
positionPtr = bufferPtr; |
885 |
|
|
errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0); |
886 |
|
|
if (errorCode == XML_ERROR_NONE) |
887 |
|
|
return 1; |
888 |
|
|
eventEndPtr = eventPtr; |
889 |
|
|
processor = errorProcessor; |
890 |
|
|
return 0; |
891 |
|
|
} |
892 |
|
|
else if (bufferPtr == bufferEnd) { |
893 |
|
|
const char *end; |
894 |
|
|
int nLeftOver; |
895 |
|
|
parseEndByteIndex += len; |
896 |
|
|
positionPtr = s; |
897 |
|
|
if (isFinal) { |
898 |
|
|
errorCode = processor(parser, s, parseEndPtr = s + len, 0); |
899 |
|
|
if (errorCode == XML_ERROR_NONE) |
900 |
|
|
return 1; |
901 |
|
|
eventEndPtr = eventPtr; |
902 |
|
|
processor = errorProcessor; |
903 |
|
|
return 0; |
904 |
|
|
} |
905 |
|
|
errorCode = processor(parser, s, parseEndPtr = s + len, &end); |
906 |
|
|
if (errorCode != XML_ERROR_NONE) { |
907 |
|
|
eventEndPtr = eventPtr; |
908 |
|
|
processor = errorProcessor; |
909 |
|
|
return 0; |
910 |
|
|
} |
911 |
|
|
XmlUpdatePosition(encoding, positionPtr, end, &position); |
912 |
|
|
nLeftOver = s + len - end; |
913 |
|
|
if (nLeftOver) { |
914 |
|
|
if (buffer == 0 || nLeftOver > bufferLim - buffer) { |
915 |
|
|
/* FIXME avoid integer overflow */ |
916 |
|
|
buffer = buffer == 0 ? malloc(len * 2) : realloc(buffer, len * 2); |
917 |
|
|
/* FIXME storage leak if realloc fails */ |
918 |
|
|
if (!buffer) { |
919 |
|
|
errorCode = XML_ERROR_NO_MEMORY; |
920 |
|
|
eventPtr = eventEndPtr = 0; |
921 |
|
|
processor = errorProcessor; |
922 |
|
|
return 0; |
923 |
|
|
} |
924 |
|
|
bufferLim = buffer + len * 2; |
925 |
|
|
} |
926 |
|
|
memcpy(buffer, end, nLeftOver); |
927 |
|
|
bufferPtr = buffer; |
928 |
|
|
bufferEnd = buffer + nLeftOver; |
929 |
|
|
} |
930 |
|
|
return 1; |
931 |
|
|
} |
932 |
|
|
else { |
933 |
|
|
memcpy(XML_GetBuffer(parser, len), s, len); |
934 |
|
|
return XML_ParseBuffer(parser, len, isFinal); |
935 |
|
|
} |
936 |
|
|
} |
937 |
|
|
|
938 |
|
|
int XML_ParseBuffer(XML_Parser parser, int len, int isFinal) |
939 |
|
|
{ |
940 |
|
|
const char *start = bufferPtr; |
941 |
|
|
positionPtr = start; |
942 |
|
|
bufferEnd += len; |
943 |
|
|
parseEndByteIndex += len; |
944 |
|
|
errorCode = processor(parser, start, parseEndPtr = bufferEnd, |
945 |
|
|
isFinal ? (const char **)0 : &bufferPtr); |
946 |
|
|
if (errorCode == XML_ERROR_NONE) { |
947 |
|
|
if (!isFinal) |
948 |
|
|
XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position); |
949 |
|
|
return 1; |
950 |
|
|
} |
951 |
|
|
else { |
952 |
|
|
eventEndPtr = eventPtr; |
953 |
|
|
processor = errorProcessor; |
954 |
|
|
return 0; |
955 |
|
|
} |
956 |
|
|
} |
957 |
|
|
|
958 |
|
|
void *XML_GetBuffer(XML_Parser parser, int len) |
959 |
|
|
{ |
960 |
|
|
if (len > bufferLim - bufferEnd) { |
961 |
|
|
/* FIXME avoid integer overflow */ |
962 |
|
|
int neededSize = len + (bufferEnd - bufferPtr); |
963 |
|
|
if (neededSize <= bufferLim - buffer) { |
964 |
|
|
memmove(buffer, bufferPtr, bufferEnd - bufferPtr); |
965 |
|
|
bufferEnd = buffer + (bufferEnd - bufferPtr); |
966 |
|
|
bufferPtr = buffer; |
967 |
|
|
} |
968 |
|
|
else { |
969 |
|
|
char *newBuf; |
970 |
|
|
int bufferSize = bufferLim - bufferPtr; |
971 |
|
|
if (bufferSize == 0) |
972 |
|
|
bufferSize = INIT_BUFFER_SIZE; |
973 |
|
|
do { |
974 |
|
|
bufferSize *= 2; |
975 |
|
|
} while (bufferSize < neededSize); |
976 |
|
|
newBuf = malloc(bufferSize); |
977 |
|
|
if (newBuf == 0) { |
978 |
|
|
errorCode = XML_ERROR_NO_MEMORY; |
979 |
|
|
return 0; |
980 |
|
|
} |
981 |
|
|
bufferLim = newBuf + bufferSize; |
982 |
|
|
if (bufferPtr) { |
983 |
|
|
memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr); |
984 |
|
|
free(buffer); |
985 |
|
|
} |
986 |
|
|
bufferEnd = newBuf + (bufferEnd - bufferPtr); |
987 |
|
|
bufferPtr = buffer = newBuf; |
988 |
|
|
} |
989 |
|
|
} |
990 |
|
|
return bufferEnd; |
991 |
|
|
} |
992 |
|
|
|
993 |
|
|
enum XML_Error XML_GetErrorCode(XML_Parser parser) |
994 |
|
|
{ |
995 |
|
|
return errorCode; |
996 |
|
|
} |
997 |
|
|
|
998 |
|
|
long XML_GetCurrentByteIndex(XML_Parser parser) |
999 |
|
|
{ |
1000 |
|
|
if (eventPtr) |
1001 |
|
|
return parseEndByteIndex - (parseEndPtr - eventPtr); |
1002 |
|
|
return -1; |
1003 |
|
|
} |
1004 |
|
|
|
1005 |
|
|
int XML_GetCurrentByteCount(XML_Parser parser) |
1006 |
|
|
{ |
1007 |
|
|
if (eventEndPtr && eventPtr) |
1008 |
|
|
return eventEndPtr - eventPtr; |
1009 |
|
|
return 0; |
1010 |
|
|
} |
1011 |
|
|
|
1012 |
|
|
int XML_GetCurrentLineNumber(XML_Parser parser) |
1013 |
|
|
{ |
1014 |
|
|
if (eventPtr) { |
1015 |
|
|
XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); |
1016 |
|
|
positionPtr = eventPtr; |
1017 |
|
|
} |
1018 |
|
|
return position.lineNumber + 1; |
1019 |
|
|
} |
1020 |
|
|
|
1021 |
|
|
int XML_GetCurrentColumnNumber(XML_Parser parser) |
1022 |
|
|
{ |
1023 |
|
|
if (eventPtr) { |
1024 |
|
|
XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); |
1025 |
|
|
positionPtr = eventPtr; |
1026 |
|
|
} |
1027 |
|
|
return position.columnNumber; |
1028 |
|
|
} |
1029 |
|
|
|
1030 |
|
|
void XML_DefaultCurrent(XML_Parser parser) |
1031 |
|
|
{ |
1032 |
|
|
if (defaultHandler) { |
1033 |
|
|
if (openInternalEntities) |
1034 |
|
|
reportDefault(parser, |
1035 |
|
|
internalEncoding, |
1036 |
|
|
openInternalEntities->internalEventPtr, |
1037 |
|
|
openInternalEntities->internalEventEndPtr); |
1038 |
|
|
else |
1039 |
|
|
reportDefault(parser, encoding, eventPtr, eventEndPtr); |
1040 |
|
|
} |
1041 |
|
|
} |
1042 |
|
|
|
1043 |
|
|
const XML_LChar *XML_ErrorString(int code) |
1044 |
|
|
{ |
1045 |
|
|
static const XML_LChar *message[] = { |
1046 |
|
|
0, |
1047 |
|
|
XML_T("out of memory"), |
1048 |
|
|
XML_T("syntax error"), |
1049 |
|
|
XML_T("no element found"), |
1050 |
|
|
XML_T("not well-formed"), |
1051 |
|
|
XML_T("unclosed token"), |
1052 |
|
|
XML_T("unclosed token"), |
1053 |
|
|
XML_T("mismatched tag"), |
1054 |
|
|
XML_T("duplicate attribute"), |
1055 |
|
|
XML_T("junk after document element"), |
1056 |
|
|
XML_T("illegal parameter entity reference"), |
1057 |
|
|
XML_T("undefined entity"), |
1058 |
|
|
XML_T("recursive entity reference"), |
1059 |
|
|
XML_T("asynchronous entity"), |
1060 |
|
|
XML_T("reference to invalid character number"), |
1061 |
|
|
XML_T("reference to binary entity"), |
1062 |
|
|
XML_T("reference to external entity in attribute"), |
1063 |
|
|
XML_T("xml processing instruction not at start of external entity"), |
1064 |
|
|
XML_T("unknown encoding"), |
1065 |
|
|
XML_T("encoding specified in XML declaration is incorrect"), |
1066 |
|
|
XML_T("unclosed CDATA section"), |
1067 |
|
|
XML_T("error in processing external entity reference"), |
1068 |
|
|
XML_T("document is not standalone") |
1069 |
|
|
}; |
1070 |
|
|
if (code > 0 && code < sizeof(message)/sizeof(message[0])) |
1071 |
|
|
return message[code]; |
1072 |
|
|
return 0; |
1073 |
|
|
} |
1074 |
|
|
|
1075 |
|
|
static |
1076 |
|
|
enum XML_Error contentProcessor(XML_Parser parser, |
1077 |
|
|
const char *start, |
1078 |
|
|
const char *end, |
1079 |
|
|
const char **endPtr) |
1080 |
|
|
{ |
1081 |
|
|
return doContent(parser, 0, encoding, start, end, endPtr); |
1082 |
|
|
} |
1083 |
|
|
|
1084 |
|
|
static |
1085 |
|
|
enum XML_Error externalEntityInitProcessor(XML_Parser parser, |
1086 |
|
|
const char *start, |
1087 |
|
|
const char *end, |
1088 |
|
|
const char **endPtr) |
1089 |
|
|
{ |
1090 |
|
|
enum XML_Error result = initializeEncoding(parser); |
1091 |
|
|
if (result != XML_ERROR_NONE) |
1092 |
|
|
return result; |
1093 |
|
|
processor = externalEntityInitProcessor2; |
1094 |
|
|
return externalEntityInitProcessor2(parser, start, end, endPtr); |
1095 |
|
|
} |
1096 |
|
|
|
1097 |
|
|
static |
1098 |
|
|
enum XML_Error externalEntityInitProcessor2(XML_Parser parser, |
1099 |
|
|
const char *start, |
1100 |
|
|
const char *end, |
1101 |
|
|
const char **endPtr) |
1102 |
|
|
{ |
1103 |
|
|
const char *next; |
1104 |
|
|
int tok = XmlContentTok(encoding, start, end, &next); |
1105 |
|
|
switch (tok) { |
1106 |
|
|
case XML_TOK_BOM: |
1107 |
|
|
start = next; |
1108 |
|
|
break; |
1109 |
|
|
case XML_TOK_PARTIAL: |
1110 |
|
|
if (endPtr) { |
1111 |
|
|
*endPtr = start; |
1112 |
|
|
return XML_ERROR_NONE; |
1113 |
|
|
} |
1114 |
|
|
eventPtr = start; |
1115 |
|
|
return XML_ERROR_UNCLOSED_TOKEN; |
1116 |
|
|
case XML_TOK_PARTIAL_CHAR: |
1117 |
|
|
if (endPtr) { |
1118 |
|
|
*endPtr = start; |
1119 |
|
|
return XML_ERROR_NONE; |
1120 |
|
|
} |
1121 |
|
|
eventPtr = start; |
1122 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
1123 |
|
|
} |
1124 |
|
|
processor = externalEntityInitProcessor3; |
1125 |
|
|
return externalEntityInitProcessor3(parser, start, end, endPtr); |
1126 |
|
|
} |
1127 |
|
|
|
1128 |
|
|
static |
1129 |
|
|
enum XML_Error externalEntityInitProcessor3(XML_Parser parser, |
1130 |
|
|
const char *start, |
1131 |
|
|
const char *end, |
1132 |
|
|
const char **endPtr) |
1133 |
|
|
{ |
1134 |
|
|
const char *next; |
1135 |
|
|
int tok = XmlContentTok(encoding, start, end, &next); |
1136 |
|
|
switch (tok) { |
1137 |
|
|
case XML_TOK_XML_DECL: |
1138 |
|
|
{ |
1139 |
|
|
enum XML_Error result = processXmlDecl(parser, 1, start, next); |
1140 |
|
|
if (result != XML_ERROR_NONE) |
1141 |
|
|
return result; |
1142 |
|
|
start = next; |
1143 |
|
|
} |
1144 |
|
|
break; |
1145 |
|
|
case XML_TOK_PARTIAL: |
1146 |
|
|
if (endPtr) { |
1147 |
|
|
*endPtr = start; |
1148 |
|
|
return XML_ERROR_NONE; |
1149 |
|
|
} |
1150 |
|
|
eventPtr = start; |
1151 |
|
|
return XML_ERROR_UNCLOSED_TOKEN; |
1152 |
|
|
case XML_TOK_PARTIAL_CHAR: |
1153 |
|
|
if (endPtr) { |
1154 |
|
|
*endPtr = start; |
1155 |
|
|
return XML_ERROR_NONE; |
1156 |
|
|
} |
1157 |
|
|
eventPtr = start; |
1158 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
1159 |
|
|
} |
1160 |
|
|
processor = externalEntityContentProcessor; |
1161 |
|
|
tagLevel = 1; |
1162 |
|
|
return doContent(parser, 1, encoding, start, end, endPtr); |
1163 |
|
|
} |
1164 |
|
|
|
1165 |
|
|
static |
1166 |
|
|
enum XML_Error externalEntityContentProcessor(XML_Parser parser, |
1167 |
|
|
const char *start, |
1168 |
|
|
const char *end, |
1169 |
|
|
const char **endPtr) |
1170 |
|
|
{ |
1171 |
|
|
return doContent(parser, 1, encoding, start, end, endPtr); |
1172 |
|
|
} |
1173 |
|
|
|
1174 |
|
|
static enum XML_Error |
1175 |
|
|
doContent(XML_Parser parser, |
1176 |
|
|
int startTagLevel, |
1177 |
|
|
const ENCODING *enc, |
1178 |
|
|
const char *s, |
1179 |
|
|
const char *end, |
1180 |
|
|
const char **nextPtr) |
1181 |
|
|
{ |
1182 |
|
|
const char **eventPP; |
1183 |
|
|
const char **eventEndPP; |
1184 |
|
|
if (enc == encoding) { |
1185 |
|
|
eventPP = &eventPtr; |
1186 |
|
|
eventEndPP = &eventEndPtr; |
1187 |
|
|
} |
1188 |
|
|
else { |
1189 |
|
|
eventPP = &(openInternalEntities->internalEventPtr); |
1190 |
|
|
eventEndPP = &(openInternalEntities->internalEventEndPtr); |
1191 |
|
|
} |
1192 |
|
|
*eventPP = s; |
1193 |
|
|
for (;;) { |
1194 |
|
|
const char *next = s; /* XmlContentTok doesn't always set the last arg */ |
1195 |
|
|
int tok = XmlContentTok(enc, s, end, &next); |
1196 |
|
|
*eventEndPP = next; |
1197 |
|
|
switch (tok) { |
1198 |
|
|
case XML_TOK_TRAILING_CR: |
1199 |
|
|
if (nextPtr) { |
1200 |
|
|
*nextPtr = s; |
1201 |
|
|
return XML_ERROR_NONE; |
1202 |
|
|
} |
1203 |
|
|
*eventEndPP = end; |
1204 |
|
|
if (characterDataHandler) { |
1205 |
|
|
XML_Char c = 0xA; |
1206 |
|
|
characterDataHandler(handlerArg, &c, 1); |
1207 |
|
|
} |
1208 |
|
|
else if (defaultHandler) |
1209 |
|
|
reportDefault(parser, enc, s, end); |
1210 |
|
|
if (startTagLevel == 0) |
1211 |
|
|
return XML_ERROR_NO_ELEMENTS; |
1212 |
|
|
if (tagLevel != startTagLevel) |
1213 |
|
|
return XML_ERROR_ASYNC_ENTITY; |
1214 |
|
|
return XML_ERROR_NONE; |
1215 |
|
|
case XML_TOK_NONE: |
1216 |
|
|
if (nextPtr) { |
1217 |
|
|
*nextPtr = s; |
1218 |
|
|
return XML_ERROR_NONE; |
1219 |
|
|
} |
1220 |
|
|
if (startTagLevel > 0) { |
1221 |
|
|
if (tagLevel != startTagLevel) |
1222 |
|
|
return XML_ERROR_ASYNC_ENTITY; |
1223 |
|
|
return XML_ERROR_NONE; |
1224 |
|
|
} |
1225 |
|
|
return XML_ERROR_NO_ELEMENTS; |
1226 |
|
|
case XML_TOK_INVALID: |
1227 |
|
|
*eventPP = next; |
1228 |
|
|
return XML_ERROR_INVALID_TOKEN; |
1229 |
|
|
case XML_TOK_PARTIAL: |
1230 |
|
|
if (nextPtr) { |
1231 |
|
|
*nextPtr = s; |
1232 |
|
|
return XML_ERROR_NONE; |
1233 |
|
|
} |
1234 |
|
|
return XML_ERROR_UNCLOSED_TOKEN; |
1235 |
|
|
case XML_TOK_PARTIAL_CHAR: |
1236 |
|
|
if (nextPtr) { |
1237 |
|
|
*nextPtr = s; |
1238 |
|
|
return XML_ERROR_NONE; |
1239 |
|
|
} |
1240 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
1241 |
|
|
case XML_TOK_ENTITY_REF: |
1242 |
|
|
{ |
1243 |
|
|
const XML_Char *name; |
1244 |
|
|
ENTITY *entity; |
1245 |
|
|
XML_Char ch = XmlPredefinedEntityName(enc, |
1246 |
|
|
s + enc->minBytesPerChar, |
1247 |
|
|
next - enc->minBytesPerChar); |
1248 |
|
|
if (ch) { |
1249 |
|
|
if (characterDataHandler) |
1250 |
|
|
characterDataHandler(handlerArg, &ch, 1); |
1251 |
|
|
else if (defaultHandler) |
1252 |
|
|
reportDefault(parser, enc, s, next); |
1253 |
|
|
break; |
1254 |
|
|
} |
1255 |
|
|
name = poolStoreString(&dtd.pool, enc, |
1256 |
|
|
s + enc->minBytesPerChar, |
1257 |
|
|
next - enc->minBytesPerChar); |
1258 |
|
|
if (!name) |
1259 |
|
|
return XML_ERROR_NO_MEMORY; |
1260 |
|
|
entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); |
1261 |
|
|
poolDiscard(&dtd.pool); |
1262 |
|
|
if (!entity) { |
1263 |
|
|
if (dtd.complete || dtd.standalone) |
1264 |
|
|
return XML_ERROR_UNDEFINED_ENTITY; |
1265 |
|
|
if (defaultHandler) |
1266 |
|
|
reportDefault(parser, enc, s, next); |
1267 |
|
|
break; |
1268 |
|
|
} |
1269 |
|
|
if (entity->open) |
1270 |
|
|
return XML_ERROR_RECURSIVE_ENTITY_REF; |
1271 |
|
|
if (entity->notation) |
1272 |
|
|
return XML_ERROR_BINARY_ENTITY_REF; |
1273 |
|
|
if (entity) { |
1274 |
|
|
if (entity->textPtr) { |
1275 |
|
|
enum XML_Error result; |
1276 |
|
|
OPEN_INTERNAL_ENTITY openEntity; |
1277 |
|
|
if (defaultHandler && !defaultExpandInternalEntities) { |
1278 |
|
|
reportDefault(parser, enc, s, next); |
1279 |
|
|
break; |
1280 |
|
|
} |
1281 |
|
|
entity->open = 1; |
1282 |
|
|
openEntity.next = openInternalEntities; |
1283 |
|
|
openInternalEntities = &openEntity; |
1284 |
|
|
openEntity.entity = entity; |
1285 |
|
|
openEntity.internalEventPtr = 0; |
1286 |
|
|
openEntity.internalEventEndPtr = 0; |
1287 |
|
|
result = doContent(parser, |
1288 |
|
|
tagLevel, |
1289 |
|
|
internalEncoding, |
1290 |
|
|
(char *)entity->textPtr, |
1291 |
|
|
(char *)(entity->textPtr + entity->textLen), |
1292 |
|
|
0); |
1293 |
|
|
entity->open = 0; |
1294 |
|
|
openInternalEntities = openEntity.next; |
1295 |
|
|
if (result) |
1296 |
|
|
return result; |
1297 |
|
|
} |
1298 |
|
|
else if (externalEntityRefHandler) { |
1299 |
|
|
const XML_Char *context; |
1300 |
|
|
entity->open = 1; |
1301 |
|
|
context = getContext(parser); |
1302 |
|
|
entity->open = 0; |
1303 |
|
|
if (!context) |
1304 |
|
|
return XML_ERROR_NO_MEMORY; |
1305 |
|
|
if (!externalEntityRefHandler(externalEntityRefHandlerArg, |
1306 |
|
|
context, |
1307 |
|
|
entity->base, |
1308 |
|
|
entity->systemId, |
1309 |
|
|
entity->publicId)) |
1310 |
|
|
return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
1311 |
|
|
poolDiscard(&tempPool); |
1312 |
|
|
} |
1313 |
|
|
else if (defaultHandler) |
1314 |
|
|
reportDefault(parser, enc, s, next); |
1315 |
|
|
} |
1316 |
|
|
break; |
1317 |
|
|
} |
1318 |
|
|
case XML_TOK_START_TAG_WITH_ATTS: |
1319 |
|
|
if (!startElementHandler) { |
1320 |
|
|
enum XML_Error result = storeAtts(parser, enc, s, 0, 0); |
1321 |
|
|
if (result) |
1322 |
|
|
return result; |
1323 |
|
|
} |
1324 |
|
|
/* fall through */ |
1325 |
|
|
case XML_TOK_START_TAG_NO_ATTS: |
1326 |
|
|
{ |
1327 |
|
|
TAG *tag; |
1328 |
|
|
if (freeTagList) { |
1329 |
|
|
tag = freeTagList; |
1330 |
|
|
freeTagList = freeTagList->parent; |
1331 |
|
|
} |
1332 |
|
|
else { |
1333 |
|
|
tag = malloc(sizeof(TAG)); |
1334 |
|
|
if (!tag) |
1335 |
|
|
return XML_ERROR_NO_MEMORY; |
1336 |
|
|
tag->buf = malloc(INIT_TAG_BUF_SIZE); |
1337 |
|
|
if (!tag->buf) |
1338 |
|
|
return XML_ERROR_NO_MEMORY; |
1339 |
|
|
tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; |
1340 |
|
|
} |
1341 |
|
|
tag->bindings = 0; |
1342 |
|
|
tag->parent = tagStack; |
1343 |
|
|
tagStack = tag; |
1344 |
|
|
tag->name.localPart = 0; |
1345 |
|
|
tag->rawName = s + enc->minBytesPerChar; |
1346 |
|
|
tag->rawNameLength = XmlNameLength(enc, tag->rawName); |
1347 |
|
|
if (nextPtr) { |
1348 |
|
|
/* Need to guarantee that: |
1349 |
|
|
tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)) <= tag->bufEnd - sizeof(XML_Char) */ |
1350 |
|
|
if (tag->rawNameLength + (int)(sizeof(XML_Char) - 1) + (int)sizeof(XML_Char) > tag->bufEnd - tag->buf) { |
1351 |
|
|
int bufSize = tag->rawNameLength * 4; |
1352 |
|
|
bufSize = ROUND_UP(bufSize, sizeof(XML_Char)); |
1353 |
|
|
tag->buf = realloc(tag->buf, bufSize); |
1354 |
|
|
if (!tag->buf) |
1355 |
|
|
return XML_ERROR_NO_MEMORY; |
1356 |
|
|
tag->bufEnd = tag->buf + bufSize; |
1357 |
|
|
} |
1358 |
|
|
memcpy(tag->buf, tag->rawName, tag->rawNameLength); |
1359 |
|
|
tag->rawName = tag->buf; |
1360 |
|
|
} |
1361 |
|
|
++tagLevel; |
1362 |
|
|
if (startElementHandler) { |
1363 |
|
|
enum XML_Error result; |
1364 |
|
|
XML_Char *toPtr; |
1365 |
|
|
for (;;) { |
1366 |
|
|
const char *rawNameEnd = tag->rawName + tag->rawNameLength; |
1367 |
|
|
const char *fromPtr = tag->rawName; |
1368 |
|
|
int bufSize; |
1369 |
|
|
if (nextPtr) |
1370 |
|
|
toPtr = (XML_Char *)(tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char))); |
1371 |
|
|
else |
1372 |
|
|
toPtr = (XML_Char *)tag->buf; |
1373 |
|
|
tag->name.str = toPtr; |
1374 |
|
|
XmlConvert(enc, |
1375 |
|
|
&fromPtr, rawNameEnd, |
1376 |
|
|
(ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1); |
1377 |
|
|
if (fromPtr == rawNameEnd) |
1378 |
|
|
break; |
1379 |
|
|
bufSize = (tag->bufEnd - tag->buf) << 1; |
1380 |
|
|
tag->buf = realloc(tag->buf, bufSize); |
1381 |
|
|
if (!tag->buf) |
1382 |
|
|
return XML_ERROR_NO_MEMORY; |
1383 |
|
|
tag->bufEnd = tag->buf + bufSize; |
1384 |
|
|
if (nextPtr) |
1385 |
|
|
tag->rawName = tag->buf; |
1386 |
|
|
} |
1387 |
|
|
*toPtr = XML_T('\0'); |
1388 |
|
|
result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings)); |
1389 |
|
|
if (result) |
1390 |
|
|
return result; |
1391 |
|
|
startElementHandler(handlerArg, tag->name.str, (const XML_Char **)atts); |
1392 |
|
|
poolClear(&tempPool); |
1393 |
|
|
} |
1394 |
|
|
else { |
1395 |
|
|
tag->name.str = 0; |
1396 |
|
|
if (defaultHandler) |
1397 |
|
|
reportDefault(parser, enc, s, next); |
1398 |
|
|
} |
1399 |
|
|
break; |
1400 |
|
|
} |
1401 |
|
|
case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: |
1402 |
|
|
if (!startElementHandler) { |
1403 |
|
|
enum XML_Error result = storeAtts(parser, enc, s, 0, 0); |
1404 |
|
|
if (result) |
1405 |
|
|
return result; |
1406 |
|
|
} |
1407 |
|
|
/* fall through */ |
1408 |
|
|
case XML_TOK_EMPTY_ELEMENT_NO_ATTS: |
1409 |
|
|
if (startElementHandler || endElementHandler) { |
1410 |
|
|
const char *rawName = s + enc->minBytesPerChar; |
1411 |
|
|
enum XML_Error result; |
1412 |
|
|
BINDING *bindings = 0; |
1413 |
|
|
TAG_NAME name; |
1414 |
|
|
name.str = poolStoreString(&tempPool, enc, rawName, |
1415 |
|
|
rawName + XmlNameLength(enc, rawName)); |
1416 |
|
|
if (!name.str) |
1417 |
|
|
return XML_ERROR_NO_MEMORY; |
1418 |
|
|
poolFinish(&tempPool); |
1419 |
|
|
result = storeAtts(parser, enc, s, &name, &bindings); |
1420 |
|
|
if (result) |
1421 |
|
|
return result; |
1422 |
|
|
poolFinish(&tempPool); |
1423 |
|
|
if (startElementHandler) |
1424 |
|
|
startElementHandler(handlerArg, name.str, (const XML_Char **)atts); |
1425 |
|
|
if (endElementHandler) { |
1426 |
|
|
if (startElementHandler) |
1427 |
|
|
*eventPP = *eventEndPP; |
1428 |
|
|
endElementHandler(handlerArg, name.str); |
1429 |
|
|
} |
1430 |
|
|
poolClear(&tempPool); |
1431 |
|
|
while (bindings) { |
1432 |
|
|
BINDING *b = bindings; |
1433 |
|
|
if (endNamespaceDeclHandler) |
1434 |
|
|
endNamespaceDeclHandler(handlerArg, b->prefix->name); |
1435 |
|
|
bindings = bindings->nextTagBinding; |
1436 |
|
|
b->nextTagBinding = freeBindingList; |
1437 |
|
|
freeBindingList = b; |
1438 |
|
|
b->prefix->binding = b->prevPrefixBinding; |
1439 |
|
|
} |
1440 |
|
|
} |
1441 |
|
|
else if (defaultHandler) |
1442 |
|
|
reportDefault(parser, enc, s, next); |
1443 |
|
|
if (tagLevel == 0) |
1444 |
|
|
return epilogProcessor(parser, next, end, nextPtr); |
1445 |
|
|
break; |
1446 |
|
|
case XML_TOK_END_TAG: |
1447 |
|
|
if (tagLevel == startTagLevel) |
1448 |
|
|
return XML_ERROR_ASYNC_ENTITY; |
1449 |
|
|
else { |
1450 |
|
|
int len; |
1451 |
|
|
const char *rawName; |
1452 |
|
|
TAG *tag = tagStack; |
1453 |
|
|
tagStack = tag->parent; |
1454 |
|
|
tag->parent = freeTagList; |
1455 |
|
|
freeTagList = tag; |
1456 |
|
|
rawName = s + enc->minBytesPerChar*2; |
1457 |
|
|
len = XmlNameLength(enc, rawName); |
1458 |
|
|
if (len != tag->rawNameLength |
1459 |
|
|
|| memcmp(tag->rawName, rawName, len) != 0) { |
1460 |
|
|
*eventPP = rawName; |
1461 |
|
|
return XML_ERROR_TAG_MISMATCH; |
1462 |
|
|
} |
1463 |
|
|
--tagLevel; |
1464 |
|
|
if (endElementHandler && tag->name.str) { |
1465 |
|
|
if (tag->name.localPart) { |
1466 |
|
|
XML_Char *to = (XML_Char *)tag->name.str + tag->name.uriLen; |
1467 |
|
|
const XML_Char *from = tag->name.localPart; |
1468 |
|
|
while ((*to++ = *from++) != 0) |
1469 |
|
|
; |
1470 |
|
|
} |
1471 |
|
|
endElementHandler(handlerArg, tag->name.str); |
1472 |
|
|
} |
1473 |
|
|
else if (defaultHandler) |
1474 |
|
|
reportDefault(parser, enc, s, next); |
1475 |
|
|
while (tag->bindings) { |
1476 |
|
|
BINDING *b = tag->bindings; |
1477 |
|
|
if (endNamespaceDeclHandler) |
1478 |
|
|
endNamespaceDeclHandler(handlerArg, b->prefix->name); |
1479 |
|
|
tag->bindings = tag->bindings->nextTagBinding; |
1480 |
|
|
b->nextTagBinding = freeBindingList; |
1481 |
|
|
freeBindingList = b; |
1482 |
|
|
b->prefix->binding = b->prevPrefixBinding; |
1483 |
|
|
} |
1484 |
|
|
if (tagLevel == 0) |
1485 |
|
|
return epilogProcessor(parser, next, end, nextPtr); |
1486 |
|
|
} |
1487 |
|
|
break; |
1488 |
|
|
case XML_TOK_CHAR_REF: |
1489 |
|
|
{ |
1490 |
|
|
int n = XmlCharRefNumber(enc, s); |
1491 |
|
|
if (n < 0) |
1492 |
|
|
return XML_ERROR_BAD_CHAR_REF; |
1493 |
|
|
if (characterDataHandler) { |
1494 |
|
|
XML_Char buf[XML_ENCODE_MAX]; |
1495 |
|
|
characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf)); |
1496 |
|
|
} |
1497 |
|
|
else if (defaultHandler) |
1498 |
|
|
reportDefault(parser, enc, s, next); |
1499 |
|
|
} |
1500 |
|
|
break; |
1501 |
|
|
case XML_TOK_XML_DECL: |
1502 |
|
|
return XML_ERROR_MISPLACED_XML_PI; |
1503 |
|
|
case XML_TOK_DATA_NEWLINE: |
1504 |
|
|
if (characterDataHandler) { |
1505 |
|
|
XML_Char c = 0xA; |
1506 |
|
|
characterDataHandler(handlerArg, &c, 1); |
1507 |
|
|
} |
1508 |
|
|
else if (defaultHandler) |
1509 |
|
|
reportDefault(parser, enc, s, next); |
1510 |
|
|
break; |
1511 |
|
|
case XML_TOK_CDATA_SECT_OPEN: |
1512 |
|
|
{ |
1513 |
|
|
enum XML_Error result; |
1514 |
|
|
if (startCdataSectionHandler) |
1515 |
|
|
startCdataSectionHandler(handlerArg); |
1516 |
|
|
#if 0 |
1517 |
|
|
/* Suppose you doing a transformation on a document that involves |
1518 |
|
|
changing only the character data. You set up a defaultHandler |
1519 |
|
|
and a characterDataHandler. The defaultHandler simply copies |
1520 |
|
|
characters through. The characterDataHandler does the transformation |
1521 |
|
|
and writes the characters out escaping them as necessary. This case |
1522 |
|
|
will fail to work if we leave out the following two lines (because & |
1523 |
|
|
and < inside CDATA sections will be incorrectly escaped). |
1524 |
|
|
|
1525 |
|
|
However, now we have a start/endCdataSectionHandler, so it seems |
1526 |
|
|
easier to let the user deal with this. */ |
1527 |
|
|
|
1528 |
|
|
else if (characterDataHandler) |
1529 |
|
|
characterDataHandler(handlerArg, dataBuf, 0); |
1530 |
|
|
#endif |
1531 |
|
|
else if (defaultHandler) |
1532 |
|
|
reportDefault(parser, enc, s, next); |
1533 |
|
|
result = doCdataSection(parser, enc, &next, end, nextPtr); |
1534 |
|
|
if (!next) { |
1535 |
|
|
processor = cdataSectionProcessor; |
1536 |
|
|
return result; |
1537 |
|
|
} |
1538 |
|
|
} |
1539 |
|
|
break; |
1540 |
|
|
case XML_TOK_TRAILING_RSQB: |
1541 |
|
|
if (nextPtr) { |
1542 |
|
|
*nextPtr = s; |
1543 |
|
|
return XML_ERROR_NONE; |
1544 |
|
|
} |
1545 |
|
|
if (characterDataHandler) { |
1546 |
|
|
if (MUST_CONVERT(enc, s)) { |
1547 |
|
|
ICHAR *dataPtr = (ICHAR *)dataBuf; |
1548 |
|
|
XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); |
1549 |
|
|
characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); |
1550 |
|
|
} |
1551 |
|
|
else |
1552 |
|
|
characterDataHandler(handlerArg, |
1553 |
|
|
(XML_Char *)s, |
1554 |
|
|
(XML_Char *)end - (XML_Char *)s); |
1555 |
|
|
} |
1556 |
|
|
else if (defaultHandler) |
1557 |
|
|
reportDefault(parser, enc, s, end); |
1558 |
|
|
if (startTagLevel == 0) { |
1559 |
|
|
*eventPP = end; |
1560 |
|
|
return XML_ERROR_NO_ELEMENTS; |
1561 |
|
|
} |
1562 |
|
|
if (tagLevel != startTagLevel) { |
1563 |
|
|
*eventPP = end; |
1564 |
|
|
return XML_ERROR_ASYNC_ENTITY; |
1565 |
|
|
} |
1566 |
|
|
return XML_ERROR_NONE; |
1567 |
|
|
case XML_TOK_DATA_CHARS: |
1568 |
|
|
if (characterDataHandler) { |
1569 |
|
|
if (MUST_CONVERT(enc, s)) { |
1570 |
|
|
for (;;) { |
1571 |
|
|
ICHAR *dataPtr = (ICHAR *)dataBuf; |
1572 |
|
|
XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); |
1573 |
|
|
*eventEndPP = s; |
1574 |
|
|
characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); |
1575 |
|
|
if (s == next) |
1576 |
|
|
break; |
1577 |
|
|
*eventPP = s; |
1578 |
|
|
} |
1579 |
|
|
} |
1580 |
|
|
else |
1581 |
|
|
characterDataHandler(handlerArg, |
1582 |
|
|
(XML_Char *)s, |
1583 |
|
|
(XML_Char *)next - (XML_Char *)s); |
1584 |
|
|
} |
1585 |
|
|
else if (defaultHandler) |
1586 |
|
|
reportDefault(parser, enc, s, next); |
1587 |
|
|
break; |
1588 |
|
|
case XML_TOK_PI: |
1589 |
|
|
if (!reportProcessingInstruction(parser, enc, s, next)) |
1590 |
|
|
return XML_ERROR_NO_MEMORY; |
1591 |
|
|
break; |
1592 |
|
|
case XML_TOK_COMMENT: |
1593 |
|
|
if (!reportComment(parser, enc, s, next)) |
1594 |
|
|
return XML_ERROR_NO_MEMORY; |
1595 |
|
|
break; |
1596 |
|
|
default: |
1597 |
|
|
if (defaultHandler) |
1598 |
|
|
reportDefault(parser, enc, s, next); |
1599 |
|
|
break; |
1600 |
|
|
} |
1601 |
|
|
*eventPP = s = next; |
1602 |
|
|
} |
1603 |
|
|
/* not reached */ |
1604 |
|
|
} |
1605 |
|
|
|
1606 |
|
|
/* If tagNamePtr is non-null, build a real list of attributes, |
1607 |
|
|
otherwise just check the attributes for well-formedness. */ |
1608 |
|
|
|
1609 |
|
|
static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *enc, |
1610 |
|
|
const char *attStr, TAG_NAME *tagNamePtr, |
1611 |
|
|
BINDING **bindingsPtr) |
1612 |
|
|
{ |
1613 |
|
|
ELEMENT_TYPE *elementType = 0; |
1614 |
|
|
int nDefaultAtts = 0; |
1615 |
|
|
const XML_Char **appAtts; /* the attribute list to pass to the application */ |
1616 |
|
|
int attIndex = 0; |
1617 |
|
|
int i; |
1618 |
|
|
int n; |
1619 |
|
|
int nPrefixes = 0; |
1620 |
|
|
BINDING *binding; |
1621 |
|
|
const XML_Char *localPart; |
1622 |
|
|
|
1623 |
|
|
/* lookup the element type name */ |
1624 |
|
|
if (tagNamePtr) { |
1625 |
|
|
elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, 0); |
1626 |
|
|
if (!elementType) { |
1627 |
|
|
tagNamePtr->str = poolCopyString(&dtd.pool, tagNamePtr->str); |
1628 |
|
|
if (!tagNamePtr->str) |
1629 |
|
|
return XML_ERROR_NO_MEMORY; |
1630 |
|
|
elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, sizeof(ELEMENT_TYPE)); |
1631 |
|
|
if (!elementType) |
1632 |
|
|
return XML_ERROR_NO_MEMORY; |
1633 |
|
|
if (ns && !setElementTypePrefix(parser, elementType)) |
1634 |
|
|
return XML_ERROR_NO_MEMORY; |
1635 |
|
|
} |
1636 |
|
|
nDefaultAtts = elementType->nDefaultAtts; |
1637 |
|
|
} |
1638 |
|
|
/* get the attributes from the tokenizer */ |
1639 |
|
|
n = XmlGetAttributes(enc, attStr, attsSize, atts); |
1640 |
|
|
if (n + nDefaultAtts > attsSize) { |
1641 |
|
|
int oldAttsSize = attsSize; |
1642 |
|
|
attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; |
1643 |
|
|
atts = realloc((void *)atts, attsSize * sizeof(ATTRIBUTE)); |
1644 |
|
|
if (!atts) |
1645 |
|
|
return XML_ERROR_NO_MEMORY; |
1646 |
|
|
if (n > oldAttsSize) |
1647 |
|
|
XmlGetAttributes(enc, attStr, n, atts); |
1648 |
|
|
} |
1649 |
|
|
appAtts = (const XML_Char **)atts; |
1650 |
|
|
for (i = 0; i < n; i++) { |
1651 |
|
|
/* add the name and value to the attribute list */ |
1652 |
|
|
ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name, |
1653 |
|
|
atts[i].name |
1654 |
|
|
+ XmlNameLength(enc, atts[i].name)); |
1655 |
|
|
if (!attId) |
1656 |
|
|
return XML_ERROR_NO_MEMORY; |
1657 |
|
|
/* detect duplicate attributes */ |
1658 |
|
|
if ((attId->name)[-1]) { |
1659 |
|
|
if (enc == encoding) |
1660 |
|
|
eventPtr = atts[i].name; |
1661 |
|
|
return XML_ERROR_DUPLICATE_ATTRIBUTE; |
1662 |
|
|
} |
1663 |
|
|
(attId->name)[-1] = 1; |
1664 |
|
|
appAtts[attIndex++] = attId->name; |
1665 |
|
|
if (!atts[i].normalized) { |
1666 |
|
|
enum XML_Error result; |
1667 |
|
|
int isCdata = 1; |
1668 |
|
|
|
1669 |
|
|
/* figure out whether declared as other than CDATA */ |
1670 |
|
|
if (attId->maybeTokenized) { |
1671 |
|
|
int j; |
1672 |
|
|
for (j = 0; j < nDefaultAtts; j++) { |
1673 |
|
|
if (attId == elementType->defaultAtts[j].id) { |
1674 |
|
|
isCdata = elementType->defaultAtts[j].isCdata; |
1675 |
|
|
break; |
1676 |
|
|
} |
1677 |
|
|
} |
1678 |
|
|
} |
1679 |
|
|
|
1680 |
|
|
/* normalize the attribute value */ |
1681 |
|
|
result = storeAttributeValue(parser, enc, isCdata, |
1682 |
|
|
atts[i].valuePtr, atts[i].valueEnd, |
1683 |
|
|
&tempPool); |
1684 |
|
|
if (result) |
1685 |
|
|
return result; |
1686 |
|
|
if (tagNamePtr) { |
1687 |
|
|
appAtts[attIndex] = poolStart(&tempPool); |
1688 |
|
|
poolFinish(&tempPool); |
1689 |
|
|
} |
1690 |
|
|
else |
1691 |
|
|
poolDiscard(&tempPool); |
1692 |
|
|
} |
1693 |
|
|
else if (tagNamePtr) { |
1694 |
|
|
/* the value did not need normalizing */ |
1695 |
|
|
appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr, atts[i].valueEnd); |
1696 |
|
|
if (appAtts[attIndex] == 0) |
1697 |
|
|
return XML_ERROR_NO_MEMORY; |
1698 |
|
|
poolFinish(&tempPool); |
1699 |
|
|
} |
1700 |
|
|
/* handle prefixed attribute names */ |
1701 |
|
|
if (attId->prefix && tagNamePtr) { |
1702 |
|
|
if (attId->xmlns) { |
1703 |
|
|
/* deal with namespace declarations here */ |
1704 |
|
|
if (!addBinding(parser, attId->prefix, attId, appAtts[attIndex], bindingsPtr)) |
1705 |
|
|
return XML_ERROR_NO_MEMORY; |
1706 |
|
|
--attIndex; |
1707 |
|
|
} |
1708 |
|
|
else { |
1709 |
|
|
/* deal with other prefixed names later */ |
1710 |
|
|
attIndex++; |
1711 |
|
|
nPrefixes++; |
1712 |
|
|
(attId->name)[-1] = 2; |
1713 |
|
|
} |
1714 |
|
|
} |
1715 |
|
|
else |
1716 |
|
|
attIndex++; |
1717 |
|
|
} |
1718 |
|
|
if (tagNamePtr) { |
1719 |
|
|
int j; |
1720 |
|
|
nSpecifiedAtts = attIndex; |
1721 |
|
|
if (elementType->idAtt && (elementType->idAtt->name)[-1]) { |
1722 |
|
|
for (i = 0; i < attIndex; i += 2) |
1723 |
|
|
if (appAtts[i] == elementType->idAtt->name) { |
1724 |
|
|
idAttIndex = i; |
1725 |
|
|
break; |
1726 |
|
|
} |
1727 |
|
|
} |
1728 |
|
|
else |
1729 |
|
|
idAttIndex = -1; |
1730 |
|
|
/* do attribute defaulting */ |
1731 |
|
|
for (j = 0; j < nDefaultAtts; j++) { |
1732 |
|
|
const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + j; |
1733 |
|
|
if (!(da->id->name)[-1] && da->value) { |
1734 |
|
|
if (da->id->prefix) { |
1735 |
|
|
if (da->id->xmlns) { |
1736 |
|
|
if (!addBinding(parser, da->id->prefix, da->id, da->value, bindingsPtr)) |
1737 |
|
|
return XML_ERROR_NO_MEMORY; |
1738 |
|
|
} |
1739 |
|
|
else { |
1740 |
|
|
(da->id->name)[-1] = 2; |
1741 |
|
|
nPrefixes++; |
1742 |
|
|
appAtts[attIndex++] = da->id->name; |
1743 |
|
|
appAtts[attIndex++] = da->value; |
1744 |
|
|
} |
1745 |
|
|
} |
1746 |
|
|
else { |
1747 |
|
|
(da->id->name)[-1] = 1; |
1748 |
|
|
appAtts[attIndex++] = da->id->name; |
1749 |
|
|
appAtts[attIndex++] = da->value; |
1750 |
|
|
} |
1751 |
|
|
} |
1752 |
|
|
} |
1753 |
|
|
appAtts[attIndex] = 0; |
1754 |
|
|
} |
1755 |
|
|
i = 0; |
1756 |
|
|
if (nPrefixes) { |
1757 |
|
|
/* expand prefixed attribute names */ |
1758 |
|
|
for (; i < attIndex; i += 2) { |
1759 |
|
|
if (appAtts[i][-1] == 2) { |
1760 |
|
|
ATTRIBUTE_ID *id; |
1761 |
|
|
((XML_Char *)(appAtts[i]))[-1] = 0; |
1762 |
|
|
id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, appAtts[i], 0); |
1763 |
|
|
if (id->prefix->binding) { |
1764 |
|
|
int j; |
1765 |
|
|
const BINDING *b = id->prefix->binding; |
1766 |
|
|
const XML_Char *s = appAtts[i]; |
1767 |
|
|
for (j = 0; j < b->uriLen; j++) { |
1768 |
|
|
if (!poolAppendChar(&tempPool, b->uri[j])) |
1769 |
|
|
return XML_ERROR_NO_MEMORY; |
1770 |
|
|
} |
1771 |
|
|
while (*s++ != ':') |
1772 |
|
|
; |
1773 |
|
|
do { |
1774 |
|
|
if (!poolAppendChar(&tempPool, *s)) |
1775 |
|
|
return XML_ERROR_NO_MEMORY; |
1776 |
|
|
} while (*s++); |
1777 |
|
|
appAtts[i] = poolStart(&tempPool); |
1778 |
|
|
poolFinish(&tempPool); |
1779 |
|
|
} |
1780 |
|
|
if (!--nPrefixes) |
1781 |
|
|
break; |
1782 |
|
|
} |
1783 |
|
|
else |
1784 |
|
|
((XML_Char *)(appAtts[i]))[-1] = 0; |
1785 |
|
|
} |
1786 |
|
|
} |
1787 |
|
|
/* clear the flags that say whether attributes were specified */ |
1788 |
|
|
for (; i < attIndex; i += 2) |
1789 |
|
|
((XML_Char *)(appAtts[i]))[-1] = 0; |
1790 |
|
|
if (!tagNamePtr) |
1791 |
|
|
return XML_ERROR_NONE; |
1792 |
|
|
for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) |
1793 |
|
|
binding->attId->name[-1] = 0; |
1794 |
|
|
/* expand the element type name */ |
1795 |
|
|
if (elementType->prefix) { |
1796 |
|
|
binding = elementType->prefix->binding; |
1797 |
|
|
if (!binding) |
1798 |
|
|
return XML_ERROR_NONE; |
1799 |
|
|
localPart = tagNamePtr->str; |
1800 |
|
|
while (*localPart++ != XML_T(':')) |
1801 |
|
|
; |
1802 |
|
|
} |
1803 |
|
|
else if (dtd.defaultPrefix.binding) { |
1804 |
|
|
binding = dtd.defaultPrefix.binding; |
1805 |
|
|
localPart = tagNamePtr->str; |
1806 |
|
|
} |
1807 |
|
|
else |
1808 |
|
|
return XML_ERROR_NONE; |
1809 |
|
|
tagNamePtr->localPart = localPart; |
1810 |
|
|
tagNamePtr->uriLen = binding->uriLen; |
1811 |
|
|
for (i = 0; localPart[i++];) |
1812 |
|
|
; |
1813 |
|
|
n = i + binding->uriLen; |
1814 |
|
|
if (n > binding->uriAlloc) { |
1815 |
|
|
TAG *p; |
1816 |
|
|
XML_Char *uri = malloc((n + EXPAND_SPARE) * sizeof(XML_Char)); |
1817 |
|
|
if (!uri) |
1818 |
|
|
return XML_ERROR_NO_MEMORY; |
1819 |
|
|
binding->uriAlloc = n + EXPAND_SPARE; |
1820 |
|
|
memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char)); |
1821 |
|
|
for (p = tagStack; p; p = p->parent) |
1822 |
|
|
if (p->name.str == binding->uri) |
1823 |
|
|
p->name.str = uri; |
1824 |
|
|
free(binding->uri); |
1825 |
|
|
binding->uri = uri; |
1826 |
|
|
} |
1827 |
|
|
memcpy(binding->uri + binding->uriLen, localPart, i * sizeof(XML_Char)); |
1828 |
|
|
tagNamePtr->str = binding->uri; |
1829 |
|
|
return XML_ERROR_NONE; |
1830 |
|
|
} |
1831 |
|
|
|
1832 |
|
|
static |
1833 |
|
|
int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr) |
1834 |
|
|
{ |
1835 |
|
|
BINDING *b; |
1836 |
|
|
int len; |
1837 |
|
|
for (len = 0; uri[len]; len++) |
1838 |
|
|
; |
1839 |
|
|
if (namespaceSeparator) |
1840 |
|
|
len++; |
1841 |
|
|
if (freeBindingList) { |
1842 |
|
|
b = freeBindingList; |
1843 |
|
|
if (len > b->uriAlloc) { |
1844 |
|
|
b->uri = realloc(b->uri, sizeof(XML_Char) * (len + EXPAND_SPARE)); |
1845 |
|
|
if (!b->uri) |
1846 |
|
|
return 0; |
1847 |
|
|
b->uriAlloc = len + EXPAND_SPARE; |
1848 |
|
|
} |
1849 |
|
|
freeBindingList = b->nextTagBinding; |
1850 |
|
|
} |
1851 |
|
|
else { |
1852 |
|
|
b = malloc(sizeof(BINDING)); |
1853 |
|
|
if (!b) |
1854 |
|
|
return 0; |
1855 |
|
|
b->uri = malloc(sizeof(XML_Char) * (len + EXPAND_SPARE)); |
1856 |
|
|
if (!b->uri) { |
1857 |
|
|
free(b); |
1858 |
|
|
return 0; |
1859 |
|
|
} |
1860 |
|
|
b->uriAlloc = len + EXPAND_SPARE; |
1861 |
|
|
} |
1862 |
|
|
b->uriLen = len; |
1863 |
|
|
memcpy(b->uri, uri, len * sizeof(XML_Char)); |
1864 |
|
|
if (namespaceSeparator) |
1865 |
|
|
b->uri[len - 1] = namespaceSeparator; |
1866 |
|
|
b->prefix = prefix; |
1867 |
|
|
b->attId = attId; |
1868 |
|
|
b->prevPrefixBinding = prefix->binding; |
1869 |
|
|
if (*uri == XML_T('\0') && prefix == &dtd.defaultPrefix) |
1870 |
|
|
prefix->binding = 0; |
1871 |
|
|
else |
1872 |
|
|
prefix->binding = b; |
1873 |
|
|
b->nextTagBinding = *bindingsPtr; |
1874 |
|
|
*bindingsPtr = b; |
1875 |
|
|
if (startNamespaceDeclHandler) |
1876 |
|
|
startNamespaceDeclHandler(handlerArg, prefix->name, |
1877 |
|
|
prefix->binding ? uri : 0); |
1878 |
|
|
return 1; |
1879 |
|
|
} |
1880 |
|
|
|
1881 |
|
|
/* The idea here is to avoid using stack for each CDATA section when |
1882 |
|
|
the whole file is parsed with one call. */ |
1883 |
|
|
|
1884 |
|
|
static |
1885 |
|
|
enum XML_Error cdataSectionProcessor(XML_Parser parser, |
1886 |
|
|
const char *start, |
1887 |
|
|
const char *end, |
1888 |
|
|
const char **endPtr) |
1889 |
|
|
{ |
1890 |
|
|
enum XML_Error result = doCdataSection(parser, encoding, &start, end, endPtr); |
1891 |
|
|
if (start) { |
1892 |
|
|
processor = contentProcessor; |
1893 |
|
|
return contentProcessor(parser, start, end, endPtr); |
1894 |
|
|
} |
1895 |
|
|
return result; |
1896 |
|
|
} |
1897 |
|
|
|
1898 |
|
|
/* startPtr gets set to non-null is the section is closed, and to null if |
1899 |
|
|
the section is not yet closed. */ |
1900 |
|
|
|
1901 |
|
|
static |
1902 |
|
|
enum XML_Error doCdataSection(XML_Parser parser, |
1903 |
|
|
const ENCODING *enc, |
1904 |
|
|
const char **startPtr, |
1905 |
|
|
const char *end, |
1906 |
|
|
const char **nextPtr) |
1907 |
|
|
{ |
1908 |
|
|
const char *s = *startPtr; |
1909 |
|
|
const char **eventPP; |
1910 |
|
|
const char **eventEndPP; |
1911 |
|
|
if (enc == encoding) { |
1912 |
|
|
eventPP = &eventPtr; |
1913 |
|
|
*eventPP = s; |
1914 |
|
|
eventEndPP = &eventEndPtr; |
1915 |
|
|
} |
1916 |
|
|
else { |
1917 |
|
|
eventPP = &(openInternalEntities->internalEventPtr); |
1918 |
|
|
eventEndPP = &(openInternalEntities->internalEventEndPtr); |
1919 |
|
|
} |
1920 |
|
|
*eventPP = s; |
1921 |
|
|
*startPtr = 0; |
1922 |
|
|
for (;;) { |
1923 |
|
|
const char *next; |
1924 |
|
|
int tok = XmlCdataSectionTok(enc, s, end, &next); |
1925 |
|
|
*eventEndPP = next; |
1926 |
|
|
switch (tok) { |
1927 |
|
|
case XML_TOK_CDATA_SECT_CLOSE: |
1928 |
|
|
if (endCdataSectionHandler) |
1929 |
|
|
endCdataSectionHandler(handlerArg); |
1930 |
|
|
#if 0 |
1931 |
|
|
/* see comment under XML_TOK_CDATA_SECT_OPEN */ |
1932 |
|
|
else if (characterDataHandler) |
1933 |
|
|
characterDataHandler(handlerArg, dataBuf, 0); |
1934 |
|
|
#endif |
1935 |
|
|
else if (defaultHandler) |
1936 |
|
|
reportDefault(parser, enc, s, next); |
1937 |
|
|
*startPtr = next; |
1938 |
|
|
return XML_ERROR_NONE; |
1939 |
|
|
case XML_TOK_DATA_NEWLINE: |
1940 |
|
|
if (characterDataHandler) { |
1941 |
|
|
XML_Char c = 0xA; |
1942 |
|
|
characterDataHandler(handlerArg, &c, 1); |
1943 |
|
|
} |
1944 |
|
|
else if (defaultHandler) |
1945 |
|
|
reportDefault(parser, enc, s, next); |
1946 |
|
|
break; |
1947 |
|
|
case XML_TOK_DATA_CHARS: |
1948 |
|
|
if (characterDataHandler) { |
1949 |
|
|
if (MUST_CONVERT(enc, s)) { |
1950 |
|
|
for (;;) { |
1951 |
|
|
ICHAR *dataPtr = (ICHAR *)dataBuf; |
1952 |
|
|
XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); |
1953 |
|
|
*eventEndPP = next; |
1954 |
|
|
characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); |
1955 |
|
|
if (s == next) |
1956 |
|
|
break; |
1957 |
|
|
*eventPP = s; |
1958 |
|
|
} |
1959 |
|
|
} |
1960 |
|
|
else |
1961 |
|
|
characterDataHandler(handlerArg, |
1962 |
|
|
(XML_Char *)s, |
1963 |
|
|
(XML_Char *)next - (XML_Char *)s); |
1964 |
|
|
} |
1965 |
|
|
else if (defaultHandler) |
1966 |
|
|
reportDefault(parser, enc, s, next); |
1967 |
|
|
break; |
1968 |
|
|
case XML_TOK_INVALID: |
1969 |
|
|
*eventPP = next; |
1970 |
|
|
return XML_ERROR_INVALID_TOKEN; |
1971 |
|
|
case XML_TOK_PARTIAL_CHAR: |
1972 |
|
|
if (nextPtr) { |
1973 |
|
|
*nextPtr = s; |
1974 |
|
|
return XML_ERROR_NONE; |
1975 |
|
|
} |
1976 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
1977 |
|
|
case XML_TOK_PARTIAL: |
1978 |
|
|
case XML_TOK_NONE: |
1979 |
|
|
if (nextPtr) { |
1980 |
|
|
*nextPtr = s; |
1981 |
|
|
return XML_ERROR_NONE; |
1982 |
|
|
} |
1983 |
|
|
return XML_ERROR_UNCLOSED_CDATA_SECTION; |
1984 |
|
|
default: |
1985 |
|
|
abort(); |
1986 |
|
|
} |
1987 |
|
|
*eventPP = s = next; |
1988 |
|
|
} |
1989 |
|
|
/* not reached */ |
1990 |
|
|
} |
1991 |
|
|
|
1992 |
|
|
#ifdef XML_DTD |
1993 |
|
|
|
1994 |
|
|
/* The idea here is to avoid using stack for each IGNORE section when |
1995 |
|
|
the whole file is parsed with one call. */ |
1996 |
|
|
|
1997 |
|
|
static |
1998 |
|
|
enum XML_Error ignoreSectionProcessor(XML_Parser parser, |
1999 |
|
|
const char *start, |
2000 |
|
|
const char *end, |
2001 |
|
|
const char **endPtr) |
2002 |
|
|
{ |
2003 |
|
|
enum XML_Error result = doIgnoreSection(parser, encoding, &start, end, endPtr); |
2004 |
|
|
if (start) { |
2005 |
|
|
processor = prologProcessor; |
2006 |
|
|
return prologProcessor(parser, start, end, endPtr); |
2007 |
|
|
} |
2008 |
|
|
return result; |
2009 |
|
|
} |
2010 |
|
|
|
2011 |
|
|
/* startPtr gets set to non-null is the section is closed, and to null if |
2012 |
|
|
the section is not yet closed. */ |
2013 |
|
|
|
2014 |
|
|
static |
2015 |
|
|
enum XML_Error doIgnoreSection(XML_Parser parser, |
2016 |
|
|
const ENCODING *enc, |
2017 |
|
|
const char **startPtr, |
2018 |
|
|
const char *end, |
2019 |
|
|
const char **nextPtr) |
2020 |
|
|
{ |
2021 |
|
|
const char *next; |
2022 |
|
|
int tok; |
2023 |
|
|
const char *s = *startPtr; |
2024 |
|
|
const char **eventPP; |
2025 |
|
|
const char **eventEndPP; |
2026 |
|
|
if (enc == encoding) { |
2027 |
|
|
eventPP = &eventPtr; |
2028 |
|
|
*eventPP = s; |
2029 |
|
|
eventEndPP = &eventEndPtr; |
2030 |
|
|
} |
2031 |
|
|
else { |
2032 |
|
|
eventPP = &(openInternalEntities->internalEventPtr); |
2033 |
|
|
eventEndPP = &(openInternalEntities->internalEventEndPtr); |
2034 |
|
|
} |
2035 |
|
|
*eventPP = s; |
2036 |
|
|
*startPtr = 0; |
2037 |
|
|
tok = XmlIgnoreSectionTok(enc, s, end, &next); |
2038 |
|
|
*eventEndPP = next; |
2039 |
|
|
switch (tok) { |
2040 |
|
|
case XML_TOK_IGNORE_SECT: |
2041 |
|
|
if (defaultHandler) |
2042 |
|
|
reportDefault(parser, enc, s, next); |
2043 |
|
|
*startPtr = next; |
2044 |
|
|
return XML_ERROR_NONE; |
2045 |
|
|
case XML_TOK_INVALID: |
2046 |
|
|
*eventPP = next; |
2047 |
|
|
return XML_ERROR_INVALID_TOKEN; |
2048 |
|
|
case XML_TOK_PARTIAL_CHAR: |
2049 |
|
|
if (nextPtr) { |
2050 |
|
|
*nextPtr = s; |
2051 |
|
|
return XML_ERROR_NONE; |
2052 |
|
|
} |
2053 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
2054 |
|
|
case XML_TOK_PARTIAL: |
2055 |
|
|
case XML_TOK_NONE: |
2056 |
|
|
if (nextPtr) { |
2057 |
|
|
*nextPtr = s; |
2058 |
|
|
return XML_ERROR_NONE; |
2059 |
|
|
} |
2060 |
|
|
return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */ |
2061 |
|
|
default: |
2062 |
|
|
abort(); |
2063 |
|
|
} |
2064 |
|
|
/* not reached */ |
2065 |
|
|
} |
2066 |
|
|
|
2067 |
|
|
#endif /* XML_DTD */ |
2068 |
|
|
|
2069 |
|
|
static enum XML_Error |
2070 |
|
|
initializeEncoding(XML_Parser parser) |
2071 |
|
|
{ |
2072 |
|
|
const char *s; |
2073 |
|
|
#ifdef XML_UNICODE |
2074 |
|
|
char encodingBuf[128]; |
2075 |
|
|
if (!protocolEncodingName) |
2076 |
|
|
s = 0; |
2077 |
|
|
else { |
2078 |
|
|
int i; |
2079 |
|
|
for (i = 0; protocolEncodingName[i]; i++) { |
2080 |
|
|
if (i == sizeof(encodingBuf) - 1 |
2081 |
|
|
|| (protocolEncodingName[i] & ~0x7f) != 0) { |
2082 |
|
|
encodingBuf[0] = '\0'; |
2083 |
|
|
break; |
2084 |
|
|
} |
2085 |
|
|
encodingBuf[i] = (char)protocolEncodingName[i]; |
2086 |
|
|
} |
2087 |
|
|
encodingBuf[i] = '\0'; |
2088 |
|
|
s = encodingBuf; |
2089 |
|
|
} |
2090 |
|
|
#else |
2091 |
|
|
s = protocolEncodingName; |
2092 |
|
|
#endif |
2093 |
|
|
if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s)) |
2094 |
|
|
return XML_ERROR_NONE; |
2095 |
|
|
return handleUnknownEncoding(parser, protocolEncodingName); |
2096 |
|
|
} |
2097 |
|
|
|
2098 |
|
|
static enum XML_Error |
2099 |
|
|
processXmlDecl(XML_Parser parser, int isGeneralTextEntity, |
2100 |
|
|
const char *s, const char *next) |
2101 |
|
|
{ |
2102 |
|
|
const char *encodingName = 0; |
2103 |
|
|
const ENCODING *newEncoding = 0; |
2104 |
|
|
const char *version; |
2105 |
|
|
int standalone = -1; |
2106 |
|
|
if (!(ns |
2107 |
|
|
? XmlParseXmlDeclNS |
2108 |
|
|
: XmlParseXmlDecl)(isGeneralTextEntity, |
2109 |
|
|
encoding, |
2110 |
|
|
s, |
2111 |
|
|
next, |
2112 |
|
|
&eventPtr, |
2113 |
|
|
&version, |
2114 |
|
|
&encodingName, |
2115 |
|
|
&newEncoding, |
2116 |
|
|
&standalone)) |
2117 |
|
|
return XML_ERROR_SYNTAX; |
2118 |
|
|
if (!isGeneralTextEntity && standalone == 1) { |
2119 |
|
|
dtd.standalone = 1; |
2120 |
|
|
#ifdef XML_DTD |
2121 |
|
|
if (paramEntityParsing == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE) |
2122 |
|
|
paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER; |
2123 |
|
|
#endif /* XML_DTD */ |
2124 |
|
|
} |
2125 |
|
|
if (defaultHandler) |
2126 |
|
|
reportDefault(parser, encoding, s, next); |
2127 |
|
|
if (!protocolEncodingName) { |
2128 |
|
|
if (newEncoding) { |
2129 |
|
|
if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) { |
2130 |
|
|
eventPtr = encodingName; |
2131 |
|
|
return XML_ERROR_INCORRECT_ENCODING; |
2132 |
|
|
} |
2133 |
|
|
encoding = newEncoding; |
2134 |
|
|
} |
2135 |
|
|
else if (encodingName) { |
2136 |
|
|
enum XML_Error result; |
2137 |
|
|
const XML_Char *s = poolStoreString(&tempPool, |
2138 |
|
|
encoding, |
2139 |
|
|
encodingName, |
2140 |
|
|
encodingName |
2141 |
|
|
+ XmlNameLength(encoding, encodingName)); |
2142 |
|
|
if (!s) |
2143 |
|
|
return XML_ERROR_NO_MEMORY; |
2144 |
|
|
result = handleUnknownEncoding(parser, s); |
2145 |
|
|
poolDiscard(&tempPool); |
2146 |
|
|
if (result == XML_ERROR_UNKNOWN_ENCODING) |
2147 |
|
|
eventPtr = encodingName; |
2148 |
|
|
return result; |
2149 |
|
|
} |
2150 |
|
|
} |
2151 |
|
|
return XML_ERROR_NONE; |
2152 |
|
|
} |
2153 |
|
|
|
2154 |
|
|
static enum XML_Error |
2155 |
|
|
handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) |
2156 |
|
|
{ |
2157 |
|
|
if (unknownEncodingHandler) { |
2158 |
|
|
XML_Encoding info; |
2159 |
|
|
int i; |
2160 |
|
|
for (i = 0; i < 256; i++) |
2161 |
|
|
info.map[i] = -1; |
2162 |
|
|
info.convert = 0; |
2163 |
|
|
info.data = 0; |
2164 |
|
|
info.release = 0; |
2165 |
|
|
if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, &info)) { |
2166 |
|
|
ENCODING *enc; |
2167 |
|
|
unknownEncodingMem = malloc(XmlSizeOfUnknownEncoding()); |
2168 |
|
|
if (!unknownEncodingMem) { |
2169 |
|
|
if (info.release) |
2170 |
|
|
info.release(info.data); |
2171 |
|
|
return XML_ERROR_NO_MEMORY; |
2172 |
|
|
} |
2173 |
|
|
enc = (ns |
2174 |
|
|
? XmlInitUnknownEncodingNS |
2175 |
|
|
: XmlInitUnknownEncoding)(unknownEncodingMem, |
2176 |
|
|
info.map, |
2177 |
|
|
info.convert, |
2178 |
|
|
info.data); |
2179 |
|
|
if (enc) { |
2180 |
|
|
unknownEncodingData = info.data; |
2181 |
|
|
unknownEncodingRelease = info.release; |
2182 |
|
|
encoding = enc; |
2183 |
|
|
return XML_ERROR_NONE; |
2184 |
|
|
} |
2185 |
|
|
} |
2186 |
|
|
if (info.release) |
2187 |
|
|
info.release(info.data); |
2188 |
|
|
} |
2189 |
|
|
return XML_ERROR_UNKNOWN_ENCODING; |
2190 |
|
|
} |
2191 |
|
|
|
2192 |
|
|
static enum XML_Error |
2193 |
|
|
prologInitProcessor(XML_Parser parser, |
2194 |
|
|
const char *s, |
2195 |
|
|
const char *end, |
2196 |
|
|
const char **nextPtr) |
2197 |
|
|
{ |
2198 |
|
|
enum XML_Error result = initializeEncoding(parser); |
2199 |
|
|
if (result != XML_ERROR_NONE) |
2200 |
|
|
return result; |
2201 |
|
|
processor = prologProcessor; |
2202 |
|
|
return prologProcessor(parser, s, end, nextPtr); |
2203 |
|
|
} |
2204 |
|
|
|
2205 |
|
|
static enum XML_Error |
2206 |
|
|
prologProcessor(XML_Parser parser, |
2207 |
|
|
const char *s, |
2208 |
|
|
const char *end, |
2209 |
|
|
const char **nextPtr) |
2210 |
|
|
{ |
2211 |
|
|
const char *next; |
2212 |
|
|
int tok = XmlPrologTok(encoding, s, end, &next); |
2213 |
|
|
return doProlog(parser, encoding, s, end, tok, next, nextPtr); |
2214 |
|
|
} |
2215 |
|
|
|
2216 |
|
|
static enum XML_Error |
2217 |
|
|
doProlog(XML_Parser parser, |
2218 |
|
|
const ENCODING *enc, |
2219 |
|
|
const char *s, |
2220 |
|
|
const char *end, |
2221 |
|
|
int tok, |
2222 |
|
|
const char *next, |
2223 |
|
|
const char **nextPtr) |
2224 |
|
|
{ |
2225 |
|
|
#ifdef XML_DTD |
2226 |
|
|
static const XML_Char externalSubsetName[] = { '#' , '\0' }; |
2227 |
|
|
#endif /* XML_DTD */ |
2228 |
|
|
|
2229 |
|
|
const char **eventPP; |
2230 |
|
|
const char **eventEndPP; |
2231 |
|
|
if (enc == encoding) { |
2232 |
|
|
eventPP = &eventPtr; |
2233 |
|
|
eventEndPP = &eventEndPtr; |
2234 |
|
|
} |
2235 |
|
|
else { |
2236 |
|
|
eventPP = &(openInternalEntities->internalEventPtr); |
2237 |
|
|
eventEndPP = &(openInternalEntities->internalEventEndPtr); |
2238 |
|
|
} |
2239 |
|
|
for (;;) { |
2240 |
|
|
int role; |
2241 |
|
|
*eventPP = s; |
2242 |
|
|
*eventEndPP = next; |
2243 |
|
|
if (tok <= 0) { |
2244 |
|
|
if (nextPtr != 0 && tok != XML_TOK_INVALID) { |
2245 |
|
|
*nextPtr = s; |
2246 |
|
|
return XML_ERROR_NONE; |
2247 |
|
|
} |
2248 |
|
|
switch (tok) { |
2249 |
|
|
case XML_TOK_INVALID: |
2250 |
|
|
*eventPP = next; |
2251 |
|
|
return XML_ERROR_INVALID_TOKEN; |
2252 |
|
|
case XML_TOK_PARTIAL: |
2253 |
|
|
return XML_ERROR_UNCLOSED_TOKEN; |
2254 |
|
|
case XML_TOK_PARTIAL_CHAR: |
2255 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
2256 |
|
|
case XML_TOK_NONE: |
2257 |
|
|
#ifdef XML_DTD |
2258 |
|
|
if (enc != encoding) |
2259 |
|
|
return XML_ERROR_NONE; |
2260 |
|
|
if (parentParser) { |
2261 |
|
|
if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc) |
2262 |
|
|
== XML_ROLE_ERROR) |
2263 |
|
|
return XML_ERROR_SYNTAX; |
2264 |
|
|
hadExternalDoctype = 0; |
2265 |
|
|
return XML_ERROR_NONE; |
2266 |
|
|
} |
2267 |
|
|
#endif /* XML_DTD */ |
2268 |
|
|
return XML_ERROR_NO_ELEMENTS; |
2269 |
|
|
default: |
2270 |
|
|
tok = -tok; |
2271 |
|
|
next = end; |
2272 |
|
|
break; |
2273 |
|
|
} |
2274 |
|
|
} |
2275 |
|
|
role = XmlTokenRole(&prologState, tok, s, next, enc); |
2276 |
|
|
switch (role) { |
2277 |
|
|
case XML_ROLE_XML_DECL: |
2278 |
|
|
{ |
2279 |
|
|
enum XML_Error result = processXmlDecl(parser, 0, s, next); |
2280 |
|
|
if (result != XML_ERROR_NONE) |
2281 |
|
|
return result; |
2282 |
|
|
enc = encoding; |
2283 |
|
|
} |
2284 |
|
|
break; |
2285 |
|
|
case XML_ROLE_DOCTYPE_NAME: |
2286 |
|
|
if (startDoctypeDeclHandler) { |
2287 |
|
|
const XML_Char *name = poolStoreString(&tempPool, enc, s, next); |
2288 |
|
|
if (!name) |
2289 |
|
|
return XML_ERROR_NO_MEMORY; |
2290 |
|
|
startDoctypeDeclHandler(handlerArg, name); |
2291 |
|
|
poolClear(&tempPool); |
2292 |
|
|
} |
2293 |
|
|
break; |
2294 |
|
|
#ifdef XML_DTD |
2295 |
|
|
case XML_ROLE_TEXT_DECL: |
2296 |
|
|
{ |
2297 |
|
|
enum XML_Error result = processXmlDecl(parser, 1, s, next); |
2298 |
|
|
if (result != XML_ERROR_NONE) |
2299 |
|
|
return result; |
2300 |
|
|
enc = encoding; |
2301 |
|
|
} |
2302 |
|
|
break; |
2303 |
|
|
#endif /* XML_DTD */ |
2304 |
|
|
case XML_ROLE_DOCTYPE_PUBLIC_ID: |
2305 |
|
|
#ifdef XML_DTD |
2306 |
|
|
declEntity = (ENTITY *)lookup(&dtd.paramEntities, |
2307 |
|
|
externalSubsetName, |
2308 |
|
|
sizeof(ENTITY)); |
2309 |
|
|
if (!declEntity) |
2310 |
|
|
return XML_ERROR_NO_MEMORY; |
2311 |
|
|
#endif /* XML_DTD */ |
2312 |
|
|
/* fall through */ |
2313 |
|
|
case XML_ROLE_ENTITY_PUBLIC_ID: |
2314 |
|
|
if (!XmlIsPublicId(enc, s, next, eventPP)) |
2315 |
|
|
return XML_ERROR_SYNTAX; |
2316 |
|
|
if (declEntity) { |
2317 |
|
|
XML_Char *tem = poolStoreString(&dtd.pool, |
2318 |
|
|
enc, |
2319 |
|
|
s + enc->minBytesPerChar, |
2320 |
|
|
next - enc->minBytesPerChar); |
2321 |
|
|
if (!tem) |
2322 |
|
|
return XML_ERROR_NO_MEMORY; |
2323 |
|
|
normalizePublicId(tem); |
2324 |
|
|
declEntity->publicId = tem; |
2325 |
|
|
poolFinish(&dtd.pool); |
2326 |
|
|
} |
2327 |
|
|
break; |
2328 |
|
|
case XML_ROLE_DOCTYPE_CLOSE: |
2329 |
|
|
if (dtd.complete && hadExternalDoctype) { |
2330 |
|
|
dtd.complete = 0; |
2331 |
|
|
#ifdef XML_DTD |
2332 |
|
|
if (paramEntityParsing && externalEntityRefHandler) { |
2333 |
|
|
ENTITY *entity = (ENTITY *)lookup(&dtd.paramEntities, |
2334 |
|
|
externalSubsetName, |
2335 |
|
|
0); |
2336 |
|
|
if (!externalEntityRefHandler(externalEntityRefHandlerArg, |
2337 |
|
|
0, |
2338 |
|
|
entity->base, |
2339 |
|
|
entity->systemId, |
2340 |
|
|
entity->publicId)) |
2341 |
|
|
return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
2342 |
|
|
} |
2343 |
|
|
#endif /* XML_DTD */ |
2344 |
|
|
if (!dtd.complete |
2345 |
|
|
&& !dtd.standalone |
2346 |
|
|
&& notStandaloneHandler |
2347 |
|
|
&& !notStandaloneHandler(handlerArg)) |
2348 |
|
|
return XML_ERROR_NOT_STANDALONE; |
2349 |
|
|
} |
2350 |
|
|
if (endDoctypeDeclHandler) |
2351 |
|
|
endDoctypeDeclHandler(handlerArg); |
2352 |
|
|
break; |
2353 |
|
|
case XML_ROLE_INSTANCE_START: |
2354 |
|
|
processor = contentProcessor; |
2355 |
|
|
return contentProcessor(parser, s, end, nextPtr); |
2356 |
|
|
case XML_ROLE_ATTLIST_ELEMENT_NAME: |
2357 |
|
|
{ |
2358 |
|
|
const XML_Char *name = poolStoreString(&dtd.pool, enc, s, next); |
2359 |
|
|
if (!name) |
2360 |
|
|
return XML_ERROR_NO_MEMORY; |
2361 |
|
|
declElementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, name, sizeof(ELEMENT_TYPE)); |
2362 |
|
|
if (!declElementType) |
2363 |
|
|
return XML_ERROR_NO_MEMORY; |
2364 |
|
|
if (declElementType->name != name) |
2365 |
|
|
poolDiscard(&dtd.pool); |
2366 |
|
|
else { |
2367 |
|
|
poolFinish(&dtd.pool); |
2368 |
|
|
if (!setElementTypePrefix(parser, declElementType)) |
2369 |
|
|
return XML_ERROR_NO_MEMORY; |
2370 |
|
|
} |
2371 |
|
|
break; |
2372 |
|
|
} |
2373 |
|
|
case XML_ROLE_ATTRIBUTE_NAME: |
2374 |
|
|
declAttributeId = getAttributeId(parser, enc, s, next); |
2375 |
|
|
if (!declAttributeId) |
2376 |
|
|
return XML_ERROR_NO_MEMORY; |
2377 |
|
|
declAttributeIsCdata = 0; |
2378 |
|
|
declAttributeIsId = 0; |
2379 |
|
|
break; |
2380 |
|
|
case XML_ROLE_ATTRIBUTE_TYPE_CDATA: |
2381 |
|
|
declAttributeIsCdata = 1; |
2382 |
|
|
break; |
2383 |
|
|
case XML_ROLE_ATTRIBUTE_TYPE_ID: |
2384 |
|
|
declAttributeIsId = 1; |
2385 |
|
|
break; |
2386 |
|
|
case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: |
2387 |
|
|
case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: |
2388 |
|
|
if (dtd.complete |
2389 |
|
|
&& !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, |
2390 |
|
|
declAttributeIsId, 0)) |
2391 |
|
|
return XML_ERROR_NO_MEMORY; |
2392 |
|
|
break; |
2393 |
|
|
case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: |
2394 |
|
|
case XML_ROLE_FIXED_ATTRIBUTE_VALUE: |
2395 |
|
|
{ |
2396 |
|
|
const XML_Char *attVal; |
2397 |
|
|
enum XML_Error result |
2398 |
|
|
= storeAttributeValue(parser, enc, declAttributeIsCdata, |
2399 |
|
|
s + enc->minBytesPerChar, |
2400 |
|
|
next - enc->minBytesPerChar, |
2401 |
|
|
&dtd.pool); |
2402 |
|
|
if (result) |
2403 |
|
|
return result; |
2404 |
|
|
attVal = poolStart(&dtd.pool); |
2405 |
|
|
poolFinish(&dtd.pool); |
2406 |
|
|
if (dtd.complete |
2407 |
|
|
// ID attributes aren't allowed to have a default |
2408 |
|
|
&& !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, 0, attVal)) |
2409 |
|
|
return XML_ERROR_NO_MEMORY; |
2410 |
|
|
break; |
2411 |
|
|
} |
2412 |
|
|
case XML_ROLE_ENTITY_VALUE: |
2413 |
|
|
{ |
2414 |
|
|
enum XML_Error result = storeEntityValue(parser, enc, |
2415 |
|
|
s + enc->minBytesPerChar, |
2416 |
|
|
next - enc->minBytesPerChar); |
2417 |
|
|
if (declEntity) { |
2418 |
|
|
declEntity->textPtr = poolStart(&dtd.pool); |
2419 |
|
|
declEntity->textLen = poolLength(&dtd.pool); |
2420 |
|
|
poolFinish(&dtd.pool); |
2421 |
|
|
if (internalParsedEntityDeclHandler |
2422 |
|
|
// Check it's not a parameter entity |
2423 |
|
|
&& ((ENTITY *)lookup(&dtd.generalEntities, declEntity->name, 0) |
2424 |
|
|
== declEntity)) { |
2425 |
|
|
*eventEndPP = s; |
2426 |
|
|
internalParsedEntityDeclHandler(handlerArg, |
2427 |
|
|
declEntity->name, |
2428 |
|
|
declEntity->textPtr, |
2429 |
|
|
declEntity->textLen); |
2430 |
|
|
} |
2431 |
|
|
} |
2432 |
|
|
else |
2433 |
|
|
poolDiscard(&dtd.pool); |
2434 |
|
|
if (result != XML_ERROR_NONE) |
2435 |
|
|
return result; |
2436 |
|
|
} |
2437 |
|
|
break; |
2438 |
|
|
case XML_ROLE_DOCTYPE_SYSTEM_ID: |
2439 |
|
|
if (!dtd.standalone |
2440 |
|
|
#ifdef XML_DTD |
2441 |
|
|
&& !paramEntityParsing |
2442 |
|
|
#endif /* XML_DTD */ |
2443 |
|
|
&& notStandaloneHandler |
2444 |
|
|
&& !notStandaloneHandler(handlerArg)) |
2445 |
|
|
return XML_ERROR_NOT_STANDALONE; |
2446 |
|
|
hadExternalDoctype = 1; |
2447 |
|
|
#ifndef XML_DTD |
2448 |
|
|
break; |
2449 |
|
|
#else /* XML_DTD */ |
2450 |
|
|
if (!declEntity) { |
2451 |
|
|
declEntity = (ENTITY *)lookup(&dtd.paramEntities, |
2452 |
|
|
externalSubsetName, |
2453 |
|
|
sizeof(ENTITY)); |
2454 |
|
|
if (!declEntity) |
2455 |
|
|
return XML_ERROR_NO_MEMORY; |
2456 |
|
|
} |
2457 |
|
|
/* fall through */ |
2458 |
|
|
#endif /* XML_DTD */ |
2459 |
|
|
case XML_ROLE_ENTITY_SYSTEM_ID: |
2460 |
|
|
if (declEntity) { |
2461 |
|
|
declEntity->systemId = poolStoreString(&dtd.pool, enc, |
2462 |
|
|
s + enc->minBytesPerChar, |
2463 |
|
|
next - enc->minBytesPerChar); |
2464 |
|
|
if (!declEntity->systemId) |
2465 |
|
|
return XML_ERROR_NO_MEMORY; |
2466 |
|
|
declEntity->base = curBase; |
2467 |
|
|
poolFinish(&dtd.pool); |
2468 |
|
|
} |
2469 |
|
|
break; |
2470 |
|
|
case XML_ROLE_ENTITY_NOTATION_NAME: |
2471 |
|
|
if (declEntity) { |
2472 |
|
|
declEntity->notation = poolStoreString(&dtd.pool, enc, s, next); |
2473 |
|
|
if (!declEntity->notation) |
2474 |
|
|
return XML_ERROR_NO_MEMORY; |
2475 |
|
|
poolFinish(&dtd.pool); |
2476 |
|
|
if (unparsedEntityDeclHandler) { |
2477 |
|
|
*eventEndPP = s; |
2478 |
|
|
unparsedEntityDeclHandler(handlerArg, |
2479 |
|
|
declEntity->name, |
2480 |
|
|
declEntity->base, |
2481 |
|
|
declEntity->systemId, |
2482 |
|
|
declEntity->publicId, |
2483 |
|
|
declEntity->notation); |
2484 |
|
|
} |
2485 |
|
|
|
2486 |
|
|
} |
2487 |
|
|
break; |
2488 |
|
|
case XML_ROLE_EXTERNAL_GENERAL_ENTITY_NO_NOTATION: |
2489 |
|
|
if (declEntity && externalParsedEntityDeclHandler) { |
2490 |
|
|
*eventEndPP = s; |
2491 |
|
|
externalParsedEntityDeclHandler(handlerArg, |
2492 |
|
|
declEntity->name, |
2493 |
|
|
declEntity->base, |
2494 |
|
|
declEntity->systemId, |
2495 |
|
|
declEntity->publicId); |
2496 |
|
|
} |
2497 |
|
|
break; |
2498 |
|
|
case XML_ROLE_GENERAL_ENTITY_NAME: |
2499 |
|
|
{ |
2500 |
|
|
const XML_Char *name; |
2501 |
|
|
if (XmlPredefinedEntityName(enc, s, next)) { |
2502 |
|
|
declEntity = 0; |
2503 |
|
|
break; |
2504 |
|
|
} |
2505 |
|
|
name = poolStoreString(&dtd.pool, enc, s, next); |
2506 |
|
|
if (!name) |
2507 |
|
|
return XML_ERROR_NO_MEMORY; |
2508 |
|
|
if (dtd.complete) { |
2509 |
|
|
declEntity = (ENTITY *)lookup(&dtd.generalEntities, name, sizeof(ENTITY)); |
2510 |
|
|
if (!declEntity) |
2511 |
|
|
return XML_ERROR_NO_MEMORY; |
2512 |
|
|
if (declEntity->name != name) { |
2513 |
|
|
poolDiscard(&dtd.pool); |
2514 |
|
|
declEntity = 0; |
2515 |
|
|
} |
2516 |
|
|
else |
2517 |
|
|
poolFinish(&dtd.pool); |
2518 |
|
|
} |
2519 |
|
|
else { |
2520 |
|
|
poolDiscard(&dtd.pool); |
2521 |
|
|
declEntity = 0; |
2522 |
|
|
} |
2523 |
|
|
} |
2524 |
|
|
break; |
2525 |
|
|
case XML_ROLE_PARAM_ENTITY_NAME: |
2526 |
|
|
#ifdef XML_DTD |
2527 |
|
|
if (dtd.complete) { |
2528 |
|
|
const XML_Char *name = poolStoreString(&dtd.pool, enc, s, next); |
2529 |
|
|
if (!name) |
2530 |
|
|
return XML_ERROR_NO_MEMORY; |
2531 |
|
|
declEntity = (ENTITY *)lookup(&dtd.paramEntities, name, sizeof(ENTITY)); |
2532 |
|
|
if (!declEntity) |
2533 |
|
|
return XML_ERROR_NO_MEMORY; |
2534 |
|
|
if (declEntity->name != name) { |
2535 |
|
|
poolDiscard(&dtd.pool); |
2536 |
|
|
declEntity = 0; |
2537 |
|
|
} |
2538 |
|
|
else |
2539 |
|
|
poolFinish(&dtd.pool); |
2540 |
|
|
} |
2541 |
|
|
#else /* not XML_DTD */ |
2542 |
|
|
declEntity = 0; |
2543 |
|
|
#endif /* not XML_DTD */ |
2544 |
|
|
break; |
2545 |
|
|
case XML_ROLE_NOTATION_NAME: |
2546 |
|
|
declNotationPublicId = 0; |
2547 |
|
|
declNotationName = 0; |
2548 |
|
|
if (notationDeclHandler) { |
2549 |
|
|
declNotationName = poolStoreString(&tempPool, enc, s, next); |
2550 |
|
|
if (!declNotationName) |
2551 |
|
|
return XML_ERROR_NO_MEMORY; |
2552 |
|
|
poolFinish(&tempPool); |
2553 |
|
|
} |
2554 |
|
|
break; |
2555 |
|
|
case XML_ROLE_NOTATION_PUBLIC_ID: |
2556 |
|
|
if (!XmlIsPublicId(enc, s, next, eventPP)) |
2557 |
|
|
return XML_ERROR_SYNTAX; |
2558 |
|
|
if (declNotationName) { |
2559 |
|
|
XML_Char *tem = poolStoreString(&tempPool, |
2560 |
|
|
enc, |
2561 |
|
|
s + enc->minBytesPerChar, |
2562 |
|
|
next - enc->minBytesPerChar); |
2563 |
|
|
if (!tem) |
2564 |
|
|
return XML_ERROR_NO_MEMORY; |
2565 |
|
|
normalizePublicId(tem); |
2566 |
|
|
declNotationPublicId = tem; |
2567 |
|
|
poolFinish(&tempPool); |
2568 |
|
|
} |
2569 |
|
|
break; |
2570 |
|
|
case XML_ROLE_NOTATION_SYSTEM_ID: |
2571 |
|
|
if (declNotationName && notationDeclHandler) { |
2572 |
|
|
const XML_Char *systemId |
2573 |
|
|
= poolStoreString(&tempPool, enc, |
2574 |
|
|
s + enc->minBytesPerChar, |
2575 |
|
|
next - enc->minBytesPerChar); |
2576 |
|
|
if (!systemId) |
2577 |
|
|
return XML_ERROR_NO_MEMORY; |
2578 |
|
|
*eventEndPP = s; |
2579 |
|
|
notationDeclHandler(handlerArg, |
2580 |
|
|
declNotationName, |
2581 |
|
|
curBase, |
2582 |
|
|
systemId, |
2583 |
|
|
declNotationPublicId); |
2584 |
|
|
} |
2585 |
|
|
poolClear(&tempPool); |
2586 |
|
|
break; |
2587 |
|
|
case XML_ROLE_NOTATION_NO_SYSTEM_ID: |
2588 |
|
|
if (declNotationPublicId && notationDeclHandler) { |
2589 |
|
|
*eventEndPP = s; |
2590 |
|
|
notationDeclHandler(handlerArg, |
2591 |
|
|
declNotationName, |
2592 |
|
|
curBase, |
2593 |
|
|
0, |
2594 |
|
|
declNotationPublicId); |
2595 |
|
|
} |
2596 |
|
|
poolClear(&tempPool); |
2597 |
|
|
break; |
2598 |
|
|
case XML_ROLE_ERROR: |
2599 |
|
|
switch (tok) { |
2600 |
|
|
case XML_TOK_PARAM_ENTITY_REF: |
2601 |
|
|
return XML_ERROR_PARAM_ENTITY_REF; |
2602 |
|
|
case XML_TOK_XML_DECL: |
2603 |
|
|
return XML_ERROR_MISPLACED_XML_PI; |
2604 |
|
|
default: |
2605 |
|
|
return XML_ERROR_SYNTAX; |
2606 |
|
|
} |
2607 |
|
|
#ifdef XML_DTD |
2608 |
|
|
case XML_ROLE_IGNORE_SECT: |
2609 |
|
|
{ |
2610 |
|
|
enum XML_Error result; |
2611 |
|
|
if (defaultHandler) |
2612 |
|
|
reportDefault(parser, enc, s, next); |
2613 |
|
|
result = doIgnoreSection(parser, enc, &next, end, nextPtr); |
2614 |
|
|
if (!next) { |
2615 |
|
|
processor = ignoreSectionProcessor; |
2616 |
|
|
return result; |
2617 |
|
|
} |
2618 |
|
|
} |
2619 |
|
|
break; |
2620 |
|
|
#endif /* XML_DTD */ |
2621 |
|
|
case XML_ROLE_GROUP_OPEN: |
2622 |
|
|
if (prologState.level >= groupSize) { |
2623 |
|
|
if (groupSize) |
2624 |
|
|
groupConnector = realloc(groupConnector, groupSize *= 2); |
2625 |
|
|
else |
2626 |
|
|
groupConnector = malloc(groupSize = 32); |
2627 |
|
|
if (!groupConnector) |
2628 |
|
|
return XML_ERROR_NO_MEMORY; |
2629 |
|
|
} |
2630 |
|
|
groupConnector[prologState.level] = 0; |
2631 |
|
|
break; |
2632 |
|
|
case XML_ROLE_GROUP_SEQUENCE: |
2633 |
|
|
if (groupConnector[prologState.level] == '|') |
2634 |
|
|
return XML_ERROR_SYNTAX; |
2635 |
|
|
groupConnector[prologState.level] = ','; |
2636 |
|
|
break; |
2637 |
|
|
case XML_ROLE_GROUP_CHOICE: |
2638 |
|
|
if (groupConnector[prologState.level] == ',') |
2639 |
|
|
return XML_ERROR_SYNTAX; |
2640 |
|
|
groupConnector[prologState.level] = '|'; |
2641 |
|
|
break; |
2642 |
|
|
case XML_ROLE_PARAM_ENTITY_REF: |
2643 |
|
|
#ifdef XML_DTD |
2644 |
|
|
case XML_ROLE_INNER_PARAM_ENTITY_REF: |
2645 |
|
|
if (paramEntityParsing |
2646 |
|
|
&& (dtd.complete || role == XML_ROLE_INNER_PARAM_ENTITY_REF)) { |
2647 |
|
|
const XML_Char *name; |
2648 |
|
|
ENTITY *entity; |
2649 |
|
|
name = poolStoreString(&dtd.pool, enc, |
2650 |
|
|
s + enc->minBytesPerChar, |
2651 |
|
|
next - enc->minBytesPerChar); |
2652 |
|
|
if (!name) |
2653 |
|
|
return XML_ERROR_NO_MEMORY; |
2654 |
|
|
entity = (ENTITY *)lookup(&dtd.paramEntities, name, 0); |
2655 |
|
|
poolDiscard(&dtd.pool); |
2656 |
|
|
if (!entity) { |
2657 |
|
|
/* FIXME what to do if !dtd.complete? */ |
2658 |
|
|
return XML_ERROR_UNDEFINED_ENTITY; |
2659 |
|
|
} |
2660 |
|
|
if (entity->open) |
2661 |
|
|
return XML_ERROR_RECURSIVE_ENTITY_REF; |
2662 |
|
|
if (entity->textPtr) { |
2663 |
|
|
enum XML_Error result; |
2664 |
|
|
result = processInternalParamEntity(parser, entity); |
2665 |
|
|
if (result != XML_ERROR_NONE) |
2666 |
|
|
return result; |
2667 |
|
|
break; |
2668 |
|
|
} |
2669 |
|
|
if (role == XML_ROLE_INNER_PARAM_ENTITY_REF) |
2670 |
|
|
return XML_ERROR_PARAM_ENTITY_REF; |
2671 |
|
|
if (externalEntityRefHandler) { |
2672 |
|
|
dtd.complete = 0; |
2673 |
|
|
entity->open = 1; |
2674 |
|
|
if (!externalEntityRefHandler(externalEntityRefHandlerArg, |
2675 |
|
|
0, |
2676 |
|
|
entity->base, |
2677 |
|
|
entity->systemId, |
2678 |
|
|
entity->publicId)) { |
2679 |
|
|
entity->open = 0; |
2680 |
|
|
return XML_ERROR_EXTERNAL_ENTITY_HANDLING; |
2681 |
|
|
} |
2682 |
|
|
entity->open = 0; |
2683 |
|
|
if (dtd.complete) |
2684 |
|
|
break; |
2685 |
|
|
} |
2686 |
|
|
} |
2687 |
|
|
#endif /* XML_DTD */ |
2688 |
|
|
if (!dtd.standalone |
2689 |
|
|
&& notStandaloneHandler |
2690 |
|
|
&& !notStandaloneHandler(handlerArg)) |
2691 |
|
|
return XML_ERROR_NOT_STANDALONE; |
2692 |
|
|
dtd.complete = 0; |
2693 |
|
|
if (defaultHandler) |
2694 |
|
|
reportDefault(parser, enc, s, next); |
2695 |
|
|
break; |
2696 |
|
|
case XML_ROLE_NONE: |
2697 |
|
|
switch (tok) { |
2698 |
|
|
case XML_TOK_PI: |
2699 |
|
|
if (!reportProcessingInstruction(parser, enc, s, next)) |
2700 |
|
|
return XML_ERROR_NO_MEMORY; |
2701 |
|
|
break; |
2702 |
|
|
case XML_TOK_COMMENT: |
2703 |
|
|
if (!reportComment(parser, enc, s, next)) |
2704 |
|
|
return XML_ERROR_NO_MEMORY; |
2705 |
|
|
break; |
2706 |
|
|
} |
2707 |
|
|
break; |
2708 |
|
|
} |
2709 |
|
|
if (defaultHandler) { |
2710 |
|
|
switch (tok) { |
2711 |
|
|
case XML_TOK_PI: |
2712 |
|
|
case XML_TOK_COMMENT: |
2713 |
|
|
case XML_TOK_BOM: |
2714 |
|
|
case XML_TOK_XML_DECL: |
2715 |
|
|
#ifdef XML_DTD |
2716 |
|
|
case XML_TOK_IGNORE_SECT: |
2717 |
|
|
#endif /* XML_DTD */ |
2718 |
|
|
case XML_TOK_PARAM_ENTITY_REF: |
2719 |
|
|
break; |
2720 |
|
|
default: |
2721 |
|
|
#ifdef XML_DTD |
2722 |
|
|
if (role != XML_ROLE_IGNORE_SECT) |
2723 |
|
|
#endif /* XML_DTD */ |
2724 |
|
|
reportDefault(parser, enc, s, next); |
2725 |
|
|
} |
2726 |
|
|
} |
2727 |
|
|
s = next; |
2728 |
|
|
tok = XmlPrologTok(enc, s, end, &next); |
2729 |
|
|
} |
2730 |
|
|
/* not reached */ |
2731 |
|
|
} |
2732 |
|
|
|
2733 |
|
|
static |
2734 |
|
|
enum XML_Error epilogProcessor(XML_Parser parser, |
2735 |
|
|
const char *s, |
2736 |
|
|
const char *end, |
2737 |
|
|
const char **nextPtr) |
2738 |
|
|
{ |
2739 |
|
|
processor = epilogProcessor; |
2740 |
|
|
eventPtr = s; |
2741 |
|
|
for (;;) { |
2742 |
|
|
const char *next; |
2743 |
|
|
int tok = XmlPrologTok(encoding, s, end, &next); |
2744 |
|
|
eventEndPtr = next; |
2745 |
|
|
switch (tok) { |
2746 |
|
|
case -XML_TOK_PROLOG_S: |
2747 |
|
|
if (defaultHandler) { |
2748 |
|
|
eventEndPtr = end; |
2749 |
|
|
reportDefault(parser, encoding, s, end); |
2750 |
|
|
} |
2751 |
|
|
/* fall through */ |
2752 |
|
|
case XML_TOK_NONE: |
2753 |
|
|
if (nextPtr) |
2754 |
|
|
*nextPtr = end; |
2755 |
|
|
return XML_ERROR_NONE; |
2756 |
|
|
case XML_TOK_PROLOG_S: |
2757 |
|
|
if (defaultHandler) |
2758 |
|
|
reportDefault(parser, encoding, s, next); |
2759 |
|
|
break; |
2760 |
|
|
case XML_TOK_PI: |
2761 |
|
|
if (!reportProcessingInstruction(parser, encoding, s, next)) |
2762 |
|
|
return XML_ERROR_NO_MEMORY; |
2763 |
|
|
break; |
2764 |
|
|
case XML_TOK_COMMENT: |
2765 |
|
|
if (!reportComment(parser, encoding, s, next)) |
2766 |
|
|
return XML_ERROR_NO_MEMORY; |
2767 |
|
|
break; |
2768 |
|
|
case XML_TOK_INVALID: |
2769 |
|
|
eventPtr = next; |
2770 |
|
|
return XML_ERROR_INVALID_TOKEN; |
2771 |
|
|
case XML_TOK_PARTIAL: |
2772 |
|
|
if (nextPtr) { |
2773 |
|
|
*nextPtr = s; |
2774 |
|
|
return XML_ERROR_NONE; |
2775 |
|
|
} |
2776 |
|
|
return XML_ERROR_UNCLOSED_TOKEN; |
2777 |
|
|
case XML_TOK_PARTIAL_CHAR: |
2778 |
|
|
if (nextPtr) { |
2779 |
|
|
*nextPtr = s; |
2780 |
|
|
return XML_ERROR_NONE; |
2781 |
|
|
} |
2782 |
|
|
return XML_ERROR_PARTIAL_CHAR; |
2783 |
|
|
default: |
2784 |
|
|
return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; |
2785 |
|
|
} |
2786 |
|
|
eventPtr = s = next; |
2787 |
|
|
} |
2788 |
|
|
} |
2789 |
|
|
|
2790 |
|
|
#ifdef XML_DTD |
2791 |
|
|
|
2792 |
|
|
static enum XML_Error |
2793 |
|
|
processInternalParamEntity(XML_Parser parser, ENTITY *entity) |
2794 |
|
|
{ |
2795 |
|
|
const char *s, *end, *next; |
2796 |
|
|
int tok; |
2797 |
|
|
enum XML_Error result; |
2798 |
|
|
OPEN_INTERNAL_ENTITY openEntity; |
2799 |
|
|
entity->open = 1; |
2800 |
|
|
openEntity.next = openInternalEntities; |
2801 |
|
|
openInternalEntities = &openEntity; |
2802 |
|
|
openEntity.entity = entity; |
2803 |
|
|
openEntity.internalEventPtr = 0; |
2804 |
|
|
openEntity.internalEventEndPtr = 0; |
2805 |
|
|
s = (char *)entity->textPtr; |
2806 |
|
|
end = (char *)(entity->textPtr + entity->textLen); |
2807 |
|
|
tok = XmlPrologTok(internalEncoding, s, end, &next); |
2808 |
|
|
result = doProlog(parser, internalEncoding, s, end, tok, next, 0); |
2809 |
|
|
entity->open = 0; |
2810 |
|
|
openInternalEntities = openEntity.next; |
2811 |
|
|
return result; |
2812 |
|
|
} |
2813 |
|
|
|
2814 |
|
|
#endif /* XML_DTD */ |
2815 |
|
|
|
2816 |
|
|
static |
2817 |
|
|
enum XML_Error errorProcessor(XML_Parser parser, |
2818 |
|
|
const char *s, |
2819 |
|
|
const char *end, |
2820 |
|
|
const char **nextPtr) |
2821 |
|
|
{ |
2822 |
|
|
return errorCode; |
2823 |
|
|
} |
2824 |
|
|
|
2825 |
|
|
static enum XML_Error |
2826 |
|
|
storeAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, |
2827 |
|
|
const char *ptr, const char *end, |
2828 |
|
|
STRING_POOL *pool) |
2829 |
|
|
{ |
2830 |
|
|
enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr, end, pool); |
2831 |
|
|
if (result) |
2832 |
|
|
return result; |
2833 |
|
|
if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) |
2834 |
|
|
poolChop(pool); |
2835 |
|
|
if (!poolAppendChar(pool, XML_T('\0'))) |
2836 |
|
|
return XML_ERROR_NO_MEMORY; |
2837 |
|
|
return XML_ERROR_NONE; |
2838 |
|
|
} |
2839 |
|
|
|
2840 |
|
|
static enum XML_Error |
2841 |
|
|
appendAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, |
2842 |
|
|
const char *ptr, const char *end, |
2843 |
|
|
STRING_POOL *pool) |
2844 |
|
|
{ |
2845 |
|
|
for (;;) { |
2846 |
|
|
const char *next; |
2847 |
|
|
int tok = XmlAttributeValueTok(enc, ptr, end, &next); |
2848 |
|
|
switch (tok) { |
2849 |
|
|
case XML_TOK_NONE: |
2850 |
|
|
return XML_ERROR_NONE; |
2851 |
|
|
case XML_TOK_INVALID: |
2852 |
|
|
if (enc == encoding) |
2853 |
|
|
eventPtr = next; |
2854 |
|
|
return XML_ERROR_INVALID_TOKEN; |
2855 |
|
|
case XML_TOK_PARTIAL: |
2856 |
|
|
if (enc == encoding) |
2857 |
|
|
eventPtr = ptr; |
2858 |
|
|
return XML_ERROR_INVALID_TOKEN; |
2859 |
|
|
case XML_TOK_CHAR_REF: |
2860 |
|
|
{ |
2861 |
|
|
XML_Char buf[XML_ENCODE_MAX]; |
2862 |
|
|
int i; |
2863 |
|
|
int n = XmlCharRefNumber(enc, ptr); |
2864 |
|
|
if (n < 0) { |
2865 |
|
|
if (enc == encoding) |
2866 |
|
|
eventPtr = ptr; |
2867 |
|
|
return XML_ERROR_BAD_CHAR_REF; |
2868 |
|
|
} |
2869 |
|
|
if (!isCdata |
2870 |
|
|
&& n == 0x20 /* space */ |
2871 |
|
|
&& (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) |
2872 |
|
|
break; |
2873 |
|
|
n = XmlEncode(n, (ICHAR *)buf); |
2874 |
|
|
if (!n) { |
2875 |
|
|
if (enc == encoding) |
2876 |
|
|
eventPtr = ptr; |
2877 |
|
|
return XML_ERROR_BAD_CHAR_REF; |
2878 |
|
|
} |
2879 |
|
|
for (i = 0; i < n; i++) { |
2880 |
|
|
if (!poolAppendChar(pool, buf[i])) |
2881 |
|
|
return XML_ERROR_NO_MEMORY; |
2882 |
|
|
} |
2883 |
|
|
} |
2884 |
|
|
break; |
2885 |
|
|
case XML_TOK_DATA_CHARS: |
2886 |
|
|
if (!poolAppend(pool, enc, ptr, next)) |
2887 |
|
|
return XML_ERROR_NO_MEMORY; |
2888 |
|
|
break; |
2889 |
|
|
break; |
2890 |
|
|
case XML_TOK_TRAILING_CR: |
2891 |
|
|
next = ptr + enc->minBytesPerChar; |
2892 |
|
|
/* fall through */ |
2893 |
|
|
case XML_TOK_ATTRIBUTE_VALUE_S: |
2894 |
|
|
case XML_TOK_DATA_NEWLINE: |
2895 |
|
|
if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) |
2896 |
|
|
break; |
2897 |
|
|
if (!poolAppendChar(pool, 0x20)) |
2898 |
|
|
return XML_ERROR_NO_MEMORY; |
2899 |
|
|
break; |
2900 |
|
|
case XML_TOK_ENTITY_REF: |
2901 |
|
|
{ |
2902 |
|
|
const XML_Char *name; |
2903 |
|
|
ENTITY *entity; |
2904 |
|
|
XML_Char ch = XmlPredefinedEntityName(enc, |
2905 |
|
|
ptr + enc->minBytesPerChar, |
2906 |
|
|
next - enc->minBytesPerChar); |
2907 |
|
|
if (ch) { |
2908 |
|
|
if (!poolAppendChar(pool, ch)) |
2909 |
|
|
return XML_ERROR_NO_MEMORY; |
2910 |
|
|
break; |
2911 |
|
|
} |
2912 |
|
|
name = poolStoreString(&temp2Pool, enc, |
2913 |
|
|
ptr + enc->minBytesPerChar, |
2914 |
|
|
next - enc->minBytesPerChar); |
2915 |
|
|
if (!name) |
2916 |
|
|
return XML_ERROR_NO_MEMORY; |
2917 |
|
|
entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); |
2918 |
|
|
poolDiscard(&temp2Pool); |
2919 |
|
|
if (!entity) { |
2920 |
|
|
if (dtd.complete) { |
2921 |
|
|
if (enc == encoding) |
2922 |
|
|
eventPtr = ptr; |
2923 |
|
|
return XML_ERROR_UNDEFINED_ENTITY; |
2924 |
|
|
} |
2925 |
|
|
} |
2926 |
|
|
else if (entity->open) { |
2927 |
|
|
if (enc == encoding) |
2928 |
|
|
eventPtr = ptr; |
2929 |
|
|
return XML_ERROR_RECURSIVE_ENTITY_REF; |
2930 |
|
|
} |
2931 |
|
|
else if (entity->notation) { |
2932 |
|
|
if (enc == encoding) |
2933 |
|
|
eventPtr = ptr; |
2934 |
|
|
return XML_ERROR_BINARY_ENTITY_REF; |
2935 |
|
|
} |
2936 |
|
|
else if (!entity->textPtr) { |
2937 |
|
|
if (enc == encoding) |
2938 |
|
|
eventPtr = ptr; |
2939 |
|
|
return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; |
2940 |
|
|
} |
2941 |
|
|
else { |
2942 |
|
|
enum XML_Error result; |
2943 |
|
|
const XML_Char *textEnd = entity->textPtr + entity->textLen; |
2944 |
|
|
entity->open = 1; |
2945 |
|
|
result = appendAttributeValue(parser, internalEncoding, isCdata, (char *)entity->textPtr, (char *)textEnd, pool); |
2946 |
|
|
entity->open = 0; |
2947 |
|
|
if (result) |
2948 |
|
|
return result; |
2949 |
|
|
} |
2950 |
|
|
} |
2951 |
|
|
break; |
2952 |
|
|
default: |
2953 |
|
|
abort(); |
2954 |
|
|
} |
2955 |
|
|
ptr = next; |
2956 |
|
|
} |
2957 |
|
|
/* not reached */ |
2958 |
|
|
} |
2959 |
|
|
|
2960 |
|
|
static |
2961 |
|
|
enum XML_Error storeEntityValue(XML_Parser parser, |
2962 |
|
|
const ENCODING *enc, |
2963 |
|
|
const char *entityTextPtr, |
2964 |
|
|
const char *entityTextEnd) |
2965 |
|
|
{ |
2966 |
|
|
STRING_POOL *pool = &(dtd.pool); |
2967 |
|
|
for (;;) { |
2968 |
|
|
const char *next; |
2969 |
|
|
int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next); |
2970 |
|
|
switch (tok) { |
2971 |
|
|
case XML_TOK_PARAM_ENTITY_REF: |
2972 |
|
|
#ifdef XML_DTD |
2973 |
|
|
if (parentParser || enc != encoding) { |
2974 |
|
|
enum XML_Error result; |
2975 |
|
|
const XML_Char *name; |
2976 |
|
|
ENTITY *entity; |
2977 |
|
|
name = poolStoreString(&tempPool, enc, |
2978 |
|
|
entityTextPtr + enc->minBytesPerChar, |
2979 |
|
|
next - enc->minBytesPerChar); |
2980 |
|
|
if (!name) |
2981 |
|
|
return XML_ERROR_NO_MEMORY; |
2982 |
|
|
entity = (ENTITY *)lookup(&dtd.paramEntities, name, 0); |
2983 |
|
|
poolDiscard(&tempPool); |
2984 |
|
|
if (!entity) { |
2985 |
|
|
if (enc == encoding) |
2986 |
|
|
eventPtr = entityTextPtr; |
2987 |
|
|
return XML_ERROR_UNDEFINED_ENTITY; |
2988 |
|
|
} |
2989 |
|
|
if (entity->open) { |
2990 |
|
|
if (enc == encoding) |
2991 |
|
|
eventPtr = entityTextPtr; |
2992 |
|
|
return XML_ERROR_RECURSIVE_ENTITY_REF; |
2993 |
|
|
} |
2994 |
|
|
if (entity->systemId) { |
2995 |
|
|
if (enc == encoding) |
2996 |
|
|
eventPtr = entityTextPtr; |
2997 |
|
|
return XML_ERROR_PARAM_ENTITY_REF; |
2998 |
|
|
} |
2999 |
|
|
entity->open = 1; |
3000 |
|
|
result = storeEntityValue(parser, |
3001 |
|
|
internalEncoding, |
3002 |
|
|
(char *)entity->textPtr, |
3003 |
|
|
(char *)(entity->textPtr + entity->textLen)); |
3004 |
|
|
entity->open = 0; |
3005 |
|
|
if (result) |
3006 |
|
|
return result; |
3007 |
|
|
break; |
3008 |
|
|
} |
3009 |
|
|
#endif /* XML_DTD */ |
3010 |
|
|
eventPtr = entityTextPtr; |
3011 |
|
|
return XML_ERROR_SYNTAX; |
3012 |
|
|
case XML_TOK_NONE: |
3013 |
|
|
return XML_ERROR_NONE; |
3014 |
|
|
case XML_TOK_ENTITY_REF: |
3015 |
|
|
case XML_TOK_DATA_CHARS: |
3016 |
|
|
if (!poolAppend(pool, enc, entityTextPtr, next)) |
3017 |
|
|
return XML_ERROR_NO_MEMORY; |
3018 |
|
|
break; |
3019 |
|
|
case XML_TOK_TRAILING_CR: |
3020 |
|
|
next = entityTextPtr + enc->minBytesPerChar; |
3021 |
|
|
/* fall through */ |
3022 |
|
|
case XML_TOK_DATA_NEWLINE: |
3023 |
|
|
if (pool->end == pool->ptr && !poolGrow(pool)) |
3024 |
|
|
return XML_ERROR_NO_MEMORY; |
3025 |
|
|
*(pool->ptr)++ = 0xA; |
3026 |
|
|
break; |
3027 |
|
|
case XML_TOK_CHAR_REF: |
3028 |
|
|
{ |
3029 |
|
|
XML_Char buf[XML_ENCODE_MAX]; |
3030 |
|
|
int i; |
3031 |
|
|
int n = XmlCharRefNumber(enc, entityTextPtr); |
3032 |
|
|
if (n < 0) { |
3033 |
|
|
if (enc == encoding) |
3034 |
|
|
eventPtr = entityTextPtr; |
3035 |
|
|
return XML_ERROR_BAD_CHAR_REF; |
3036 |
|
|
} |
3037 |
|
|
n = XmlEncode(n, (ICHAR *)buf); |
3038 |
|
|
if (!n) { |
3039 |
|
|
if (enc == encoding) |
3040 |
|
|
eventPtr = entityTextPtr; |
3041 |
|
|
return XML_ERROR_BAD_CHAR_REF; |
3042 |
|
|
} |
3043 |
|
|
for (i = 0; i < n; i++) { |
3044 |
|
|
if (pool->end == pool->ptr && !poolGrow(pool)) |
3045 |
|
|
return XML_ERROR_NO_MEMORY; |
3046 |
|
|
*(pool->ptr)++ = buf[i]; |
3047 |
|
|
} |
3048 |
|
|
} |
3049 |
|
|
break; |
3050 |
|
|
case XML_TOK_PARTIAL: |
3051 |
|
|
if (enc == encoding) |
3052 |
|
|
eventPtr = entityTextPtr; |
3053 |
|
|
return XML_ERROR_INVALID_TOKEN; |
3054 |
|
|
case XML_TOK_INVALID: |
3055 |
|
|
if (enc == encoding) |
3056 |
|
|
eventPtr = next; |
3057 |
|
|
return XML_ERROR_INVALID_TOKEN; |
3058 |
|
|
default: |
3059 |
|
|
abort(); |
3060 |
|
|
} |
3061 |
|
|
entityTextPtr = next; |
3062 |
|
|
} |
3063 |
|
|
/* not reached */ |
3064 |
|
|
} |
3065 |
|
|
|
3066 |
|
|
static void |
3067 |
|
|
normalizeLines(XML_Char *s) |
3068 |
|
|
{ |
3069 |
|
|
XML_Char *p; |
3070 |
|
|
for (;; s++) { |
3071 |
|
|
if (*s == XML_T('\0')) |
3072 |
|
|
return; |
3073 |
|
|
if (*s == 0xD) |
3074 |
|
|
break; |
3075 |
|
|
} |
3076 |
|
|
p = s; |
3077 |
|
|
do { |
3078 |
|
|
if (*s == 0xD) { |
3079 |
|
|
*p++ = 0xA; |
3080 |
|
|
if (*++s == 0xA) |
3081 |
|
|
s++; |
3082 |
|
|
} |
3083 |
|
|
else |
3084 |
|
|
*p++ = *s++; |
3085 |
|
|
} while (*s); |
3086 |
|
|
*p = XML_T('\0'); |
3087 |
|
|
} |
3088 |
|
|
|
3089 |
|
|
static int |
3090 |
|
|
reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) |
3091 |
|
|
{ |
3092 |
|
|
const XML_Char *target; |
3093 |
|
|
XML_Char *data; |
3094 |
|
|
const char *tem; |
3095 |
|
|
if (!processingInstructionHandler) { |
3096 |
|
|
if (defaultHandler) |
3097 |
|
|
reportDefault(parser, enc, start, end); |
3098 |
|
|
return 1; |
3099 |
|
|
} |
3100 |
|
|
start += enc->minBytesPerChar * 2; |
3101 |
|
|
tem = start + XmlNameLength(enc, start); |
3102 |
|
|
target = poolStoreString(&tempPool, enc, start, tem); |
3103 |
|
|
if (!target) |
3104 |
|
|
return 0; |
3105 |
|
|
poolFinish(&tempPool); |
3106 |
|
|
data = poolStoreString(&tempPool, enc, |
3107 |
|
|
XmlSkipS(enc, tem), |
3108 |
|
|
end - enc->minBytesPerChar*2); |
3109 |
|
|
if (!data) |
3110 |
|
|
return 0; |
3111 |
|
|
normalizeLines(data); |
3112 |
|
|
processingInstructionHandler(handlerArg, target, data); |
3113 |
|
|
poolClear(&tempPool); |
3114 |
|
|
return 1; |
3115 |
|
|
} |
3116 |
|
|
|
3117 |
|
|
static int |
3118 |
|
|
reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) |
3119 |
|
|
{ |
3120 |
|
|
XML_Char *data; |
3121 |
|
|
if (!commentHandler) { |
3122 |
|
|
if (defaultHandler) |
3123 |
|
|
reportDefault(parser, enc, start, end); |
3124 |
|
|
return 1; |
3125 |
|
|
} |
3126 |
|
|
data = poolStoreString(&tempPool, |
3127 |
|
|
enc, |
3128 |
|
|
start + enc->minBytesPerChar * 4, |
3129 |
|
|
end - enc->minBytesPerChar * 3); |
3130 |
|
|
if (!data) |
3131 |
|
|
return 0; |
3132 |
|
|
normalizeLines(data); |
3133 |
|
|
commentHandler(handlerArg, data); |
3134 |
|
|
poolClear(&tempPool); |
3135 |
|
|
return 1; |
3136 |
|
|
} |
3137 |
|
|
|
3138 |
|
|
static void |
3139 |
|
|
reportDefault(XML_Parser parser, const ENCODING *enc, const char *s, const char *end) |
3140 |
|
|
{ |
3141 |
|
|
if (MUST_CONVERT(enc, s)) { |
3142 |
|
|
const char **eventPP; |
3143 |
|
|
const char **eventEndPP; |
3144 |
|
|
if (enc == encoding) { |
3145 |
|
|
eventPP = &eventPtr; |
3146 |
|
|
eventEndPP = &eventEndPtr; |
3147 |
|
|
} |
3148 |
|
|
else { |
3149 |
|
|
eventPP = &(openInternalEntities->internalEventPtr); |
3150 |
|
|
eventEndPP = &(openInternalEntities->internalEventEndPtr); |
3151 |
|
|
} |
3152 |
|
|
do { |
3153 |
|
|
ICHAR *dataPtr = (ICHAR *)dataBuf; |
3154 |
|
|
XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); |
3155 |
|
|
*eventEndPP = s; |
3156 |
|
|
defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); |
3157 |
|
|
*eventPP = s; |
3158 |
|
|
} while (s != end); |
3159 |
|
|
} |
3160 |
|
|
else |
3161 |
|
|
defaultHandler(handlerArg, (XML_Char *)s, (XML_Char *)end - (XML_Char *)s); |
3162 |
|
|
} |
3163 |
|
|
|
3164 |
|
|
|
3165 |
|
|
static int |
3166 |
|
|
defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, int isCdata, int isId, const XML_Char *value) |
3167 |
|
|
{ |
3168 |
|
|
DEFAULT_ATTRIBUTE *att; |
3169 |
|
|
if (value || isId) { |
3170 |
|
|
/* The handling of default attributes gets messed up if we have |
3171 |
|
|
a default which duplicates a non-default. */ |
3172 |
|
|
int i; |
3173 |
|
|
for (i = 0; i < type->nDefaultAtts; i++) |
3174 |
|
|
if (attId == type->defaultAtts[i].id) |
3175 |
|
|
return 1; |
3176 |
|
|
if (isId && !type->idAtt && !attId->xmlns) |
3177 |
|
|
type->idAtt = attId; |
3178 |
|
|
} |
3179 |
|
|
if (type->nDefaultAtts == type->allocDefaultAtts) { |
3180 |
|
|
if (type->allocDefaultAtts == 0) { |
3181 |
|
|
type->allocDefaultAtts = 8; |
3182 |
|
|
type->defaultAtts = malloc(type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); |
3183 |
|
|
} |
3184 |
|
|
else { |
3185 |
|
|
type->allocDefaultAtts *= 2; |
3186 |
|
|
type->defaultAtts = realloc(type->defaultAtts, |
3187 |
|
|
type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); |
3188 |
|
|
} |
3189 |
|
|
if (!type->defaultAtts) |
3190 |
|
|
return 0; |
3191 |
|
|
} |
3192 |
|
|
att = type->defaultAtts + type->nDefaultAtts; |
3193 |
|
|
att->id = attId; |
3194 |
|
|
att->value = value; |
3195 |
|
|
att->isCdata = isCdata; |
3196 |
|
|
if (!isCdata) |
3197 |
|
|
attId->maybeTokenized = 1; |
3198 |
|
|
type->nDefaultAtts += 1; |
3199 |
|
|
return 1; |
3200 |
|
|
} |
3201 |
|
|
|
3202 |
|
|
static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) |
3203 |
|
|
{ |
3204 |
|
|
const XML_Char *name; |
3205 |
|
|
for (name = elementType->name; *name; name++) { |
3206 |
|
|
if (*name == XML_T(':')) { |
3207 |
|
|
PREFIX *prefix; |
3208 |
|
|
const XML_Char *s; |
3209 |
|
|
for (s = elementType->name; s != name; s++) { |
3210 |
|
|
if (!poolAppendChar(&dtd.pool, *s)) |
3211 |
|
|
return 0; |
3212 |
|
|
} |
3213 |
|
|
if (!poolAppendChar(&dtd.pool, XML_T('\0'))) |
3214 |
|
|
return 0; |
3215 |
|
|
prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); |
3216 |
|
|
if (!prefix) |
3217 |
|
|
return 0; |
3218 |
|
|
if (prefix->name == poolStart(&dtd.pool)) |
3219 |
|
|
poolFinish(&dtd.pool); |
3220 |
|
|
else |
3221 |
|
|
poolDiscard(&dtd.pool); |
3222 |
|
|
elementType->prefix = prefix; |
3223 |
|
|
|
3224 |
|
|
} |
3225 |
|
|
} |
3226 |
|
|
return 1; |
3227 |
|
|
} |
3228 |
|
|
|
3229 |
|
|
static ATTRIBUTE_ID * |
3230 |
|
|
getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) |
3231 |
|
|
{ |
3232 |
|
|
ATTRIBUTE_ID *id; |
3233 |
|
|
const XML_Char *name; |
3234 |
|
|
if (!poolAppendChar(&dtd.pool, XML_T('\0'))) |
3235 |
|
|
return 0; |
3236 |
|
|
name = poolStoreString(&dtd.pool, enc, start, end); |
3237 |
|
|
if (!name) |
3238 |
|
|
return 0; |
3239 |
|
|
++name; |
3240 |
|
|
id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, name, sizeof(ATTRIBUTE_ID)); |
3241 |
|
|
if (!id) |
3242 |
|
|
return 0; |
3243 |
|
|
if (id->name != name) |
3244 |
|
|
poolDiscard(&dtd.pool); |
3245 |
|
|
else { |
3246 |
|
|
poolFinish(&dtd.pool); |
3247 |
|
|
if (!ns) |
3248 |
|
|
; |
3249 |
|
|
else if (name[0] == 'x' |
3250 |
|
|
&& name[1] == 'm' |
3251 |
|
|
&& name[2] == 'l' |
3252 |
|
|
&& name[3] == 'n' |
3253 |
|
|
&& name[4] == 's' |
3254 |
|
|
&& (name[5] == XML_T('\0') || name[5] == XML_T(':'))) { |
3255 |
|
|
if (name[5] == '\0') |
3256 |
|
|
id->prefix = &dtd.defaultPrefix; |
3257 |
|
|
else |
3258 |
|
|
id->prefix = (PREFIX *)lookup(&dtd.prefixes, name + 6, sizeof(PREFIX)); |
3259 |
|
|
id->xmlns = 1; |
3260 |
|
|
} |
3261 |
|
|
else { |
3262 |
|
|
int i; |
3263 |
|
|
for (i = 0; name[i]; i++) { |
3264 |
|
|
if (name[i] == XML_T(':')) { |
3265 |
|
|
int j; |
3266 |
|
|
for (j = 0; j < i; j++) { |
3267 |
|
|
if (!poolAppendChar(&dtd.pool, name[j])) |
3268 |
|
|
return 0; |
3269 |
|
|
} |
3270 |
|
|
if (!poolAppendChar(&dtd.pool, XML_T('\0'))) |
3271 |
|
|
return 0; |
3272 |
|
|
id->prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); |
3273 |
|
|
if (id->prefix->name == poolStart(&dtd.pool)) |
3274 |
|
|
poolFinish(&dtd.pool); |
3275 |
|
|
else |
3276 |
|
|
poolDiscard(&dtd.pool); |
3277 |
|
|
break; |
3278 |
|
|
} |
3279 |
|
|
} |
3280 |
|
|
} |
3281 |
|
|
} |
3282 |
|
|
return id; |
3283 |
|
|
} |
3284 |
|
|
|
3285 |
|
|
#define CONTEXT_SEP XML_T('\f') |
3286 |
|
|
|
3287 |
|
|
static |
3288 |
|
|
const XML_Char *getContext(XML_Parser parser) |
3289 |
|
|
{ |
3290 |
|
|
HASH_TABLE_ITER iter; |
3291 |
|
|
int needSep = 0; |
3292 |
|
|
|
3293 |
|
|
if (dtd.defaultPrefix.binding) { |
3294 |
|
|
int i; |
3295 |
|
|
int len; |
3296 |
|
|
if (!poolAppendChar(&tempPool, XML_T('='))) |
3297 |
|
|
return 0; |
3298 |
|
|
len = dtd.defaultPrefix.binding->uriLen; |
3299 |
|
|
if (namespaceSeparator != XML_T('\0')) |
3300 |
|
|
len--; |
3301 |
|
|
for (i = 0; i < len; i++) |
3302 |
|
|
if (!poolAppendChar(&tempPool, dtd.defaultPrefix.binding->uri[i])) |
3303 |
|
|
return 0; |
3304 |
|
|
needSep = 1; |
3305 |
|
|
} |
3306 |
|
|
|
3307 |
|
|
hashTableIterInit(&iter, &(dtd.prefixes)); |
3308 |
|
|
for (;;) { |
3309 |
|
|
int i; |
3310 |
|
|
int len; |
3311 |
|
|
const XML_Char *s; |
3312 |
|
|
PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); |
3313 |
|
|
if (!prefix) |
3314 |
|
|
break; |
3315 |
|
|
if (!prefix->binding) |
3316 |
|
|
continue; |
3317 |
|
|
if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) |
3318 |
|
|
return 0; |
3319 |
|
|
for (s = prefix->name; *s; s++) |
3320 |
|
|
if (!poolAppendChar(&tempPool, *s)) |
3321 |
|
|
return 0; |
3322 |
|
|
if (!poolAppendChar(&tempPool, XML_T('='))) |
3323 |
|
|
return 0; |
3324 |
|
|
len = prefix->binding->uriLen; |
3325 |
|
|
if (namespaceSeparator != XML_T('\0')) |
3326 |
|
|
len--; |
3327 |
|
|
for (i = 0; i < len; i++) |
3328 |
|
|
if (!poolAppendChar(&tempPool, prefix->binding->uri[i])) |
3329 |
|
|
return 0; |
3330 |
|
|
needSep = 1; |
3331 |
|
|
} |
3332 |
|
|
|
3333 |
|
|
|
3334 |
|
|
hashTableIterInit(&iter, &(dtd.generalEntities)); |
3335 |
|
|
for (;;) { |
3336 |
|
|
const XML_Char *s; |
3337 |
|
|
ENTITY *e = (ENTITY *)hashTableIterNext(&iter); |
3338 |
|
|
if (!e) |
3339 |
|
|
break; |
3340 |
|
|
if (!e->open) |
3341 |
|
|
continue; |
3342 |
|
|
if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) |
3343 |
|
|
return 0; |
3344 |
|
|
for (s = e->name; *s; s++) |
3345 |
|
|
if (!poolAppendChar(&tempPool, *s)) |
3346 |
|
|
return 0; |
3347 |
|
|
needSep = 1; |
3348 |
|
|
} |
3349 |
|
|
|
3350 |
|
|
if (!poolAppendChar(&tempPool, XML_T('\0'))) |
3351 |
|
|
return 0; |
3352 |
|
|
return tempPool.start; |
3353 |
|
|
} |
3354 |
|
|
|
3355 |
|
|
static |
3356 |
|
|
int setContext(XML_Parser parser, const XML_Char *context) |
3357 |
|
|
{ |
3358 |
|
|
const XML_Char *s = context; |
3359 |
|
|
|
3360 |
|
|
while (*context != XML_T('\0')) { |
3361 |
|
|
if (*s == CONTEXT_SEP || *s == XML_T('\0')) { |
3362 |
|
|
ENTITY *e; |
3363 |
|
|
if (!poolAppendChar(&tempPool, XML_T('\0'))) |
3364 |
|
|
return 0; |
3365 |
|
|
e = (ENTITY *)lookup(&dtd.generalEntities, poolStart(&tempPool), 0); |
3366 |
|
|
if (e) |
3367 |
|
|
e->open = 1; |
3368 |
|
|
if (*s != XML_T('\0')) |
3369 |
|
|
s++; |
3370 |
|
|
context = s; |
3371 |
|
|
poolDiscard(&tempPool); |
3372 |
|
|
} |
3373 |
|
|
else if (*s == '=') { |
3374 |
|
|
PREFIX *prefix; |
3375 |
|
|
if (poolLength(&tempPool) == 0) |
3376 |
|
|
prefix = &dtd.defaultPrefix; |
3377 |
|
|
else { |
3378 |
|
|
if (!poolAppendChar(&tempPool, XML_T('\0'))) |
3379 |
|
|
return 0; |
3380 |
|
|
prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&tempPool), sizeof(PREFIX)); |
3381 |
|
|
if (!prefix) |
3382 |
|
|
return 0; |
3383 |
|
|
if (prefix->name == poolStart(&tempPool)) { |
3384 |
|
|
prefix->name = poolCopyString(&dtd.pool, prefix->name); |
3385 |
|
|
if (!prefix->name) |
3386 |
|
|
return 0; |
3387 |
|
|
} |
3388 |
|
|
poolDiscard(&tempPool); |
3389 |
|
|
} |
3390 |
|
|
for (context = s + 1; *context != CONTEXT_SEP && *context != XML_T('\0'); context++) |
3391 |
|
|
if (!poolAppendChar(&tempPool, *context)) |
3392 |
|
|
return 0; |
3393 |
|
|
if (!poolAppendChar(&tempPool, XML_T('\0'))) |
3394 |
|
|
return 0; |
3395 |
|
|
if (!addBinding(parser, prefix, 0, poolStart(&tempPool), &inheritedBindings)) |
3396 |
|
|
return 0; |
3397 |
|
|
poolDiscard(&tempPool); |
3398 |
|
|
if (*context != XML_T('\0')) |
3399 |
|
|
++context; |
3400 |
|
|
s = context; |
3401 |
|
|
} |
3402 |
|
|
else { |
3403 |
|
|
if (!poolAppendChar(&tempPool, *s)) |
3404 |
|
|
return 0; |
3405 |
|
|
s++; |
3406 |
|
|
} |
3407 |
|
|
} |
3408 |
|
|
return 1; |
3409 |
|
|
} |
3410 |
|
|
|
3411 |
|
|
|
3412 |
|
|
static |
3413 |
|
|
void normalizePublicId(XML_Char *publicId) |
3414 |
|
|
{ |
3415 |
|
|
XML_Char *p = publicId; |
3416 |
|
|
XML_Char *s; |
3417 |
|
|
for (s = publicId; *s; s++) { |
3418 |
|
|
switch (*s) { |
3419 |
|
|
case 0x20: |
3420 |
|
|
case 0xD: |
3421 |
|
|
case 0xA: |
3422 |
|
|
if (p != publicId && p[-1] != 0x20) |
3423 |
|
|
*p++ = 0x20; |
3424 |
|
|
break; |
3425 |
|
|
default: |
3426 |
|
|
*p++ = *s; |
3427 |
|
|
} |
3428 |
|
|
} |
3429 |
|
|
if (p != publicId && p[-1] == 0x20) |
3430 |
|
|
--p; |
3431 |
|
|
*p = XML_T('\0'); |
3432 |
|
|
} |
3433 |
|
|
|
3434 |
|
|
static int dtdInit(DTD *p) |
3435 |
|
|
{ |
3436 |
|
|
poolInit(&(p->pool)); |
3437 |
|
|
hashTableInit(&(p->generalEntities)); |
3438 |
|
|
hashTableInit(&(p->elementTypes)); |
3439 |
|
|
hashTableInit(&(p->attributeIds)); |
3440 |
|
|
hashTableInit(&(p->prefixes)); |
3441 |
|
|
p->complete = 1; |
3442 |
|
|
p->standalone = 0; |
3443 |
|
|
#ifdef XML_DTD |
3444 |
|
|
hashTableInit(&(p->paramEntities)); |
3445 |
|
|
#endif /* XML_DTD */ |
3446 |
|
|
p->defaultPrefix.name = 0; |
3447 |
|
|
p->defaultPrefix.binding = 0; |
3448 |
|
|
return 1; |
3449 |
|
|
} |
3450 |
|
|
|
3451 |
|
|
#ifdef XML_DTD |
3452 |
|
|
|
3453 |
|
|
static void dtdSwap(DTD *p1, DTD *p2) |
3454 |
|
|
{ |
3455 |
|
|
DTD tem; |
3456 |
|
|
memcpy(&tem, p1, sizeof(DTD)); |
3457 |
|
|
memcpy(p1, p2, sizeof(DTD)); |
3458 |
|
|
memcpy(p2, &tem, sizeof(DTD)); |
3459 |
|
|
} |
3460 |
|
|
|
3461 |
|
|
#endif /* XML_DTD */ |
3462 |
|
|
|
3463 |
|
|
static void dtdDestroy(DTD *p) |
3464 |
|
|
{ |
3465 |
|
|
HASH_TABLE_ITER iter; |
3466 |
|
|
hashTableIterInit(&iter, &(p->elementTypes)); |
3467 |
|
|
for (;;) { |
3468 |
|
|
ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); |
3469 |
|
|
if (!e) |
3470 |
|
|
break; |
3471 |
|
|
if (e->allocDefaultAtts != 0) |
3472 |
|
|
free(e->defaultAtts); |
3473 |
|
|
} |
3474 |
|
|
hashTableDestroy(&(p->generalEntities)); |
3475 |
|
|
#ifdef XML_DTD |
3476 |
|
|
hashTableDestroy(&(p->paramEntities)); |
3477 |
|
|
#endif /* XML_DTD */ |
3478 |
|
|
hashTableDestroy(&(p->elementTypes)); |
3479 |
|
|
hashTableDestroy(&(p->attributeIds)); |
3480 |
|
|
hashTableDestroy(&(p->prefixes)); |
3481 |
|
|
poolDestroy(&(p->pool)); |
3482 |
|
|
} |
3483 |
|
|
|
3484 |
|
|
/* Do a deep copy of the DTD. Return 0 for out of memory; non-zero otherwise. |
3485 |
|
|
The new DTD has already been initialized. */ |
3486 |
|
|
|
3487 |
|
|
static int dtdCopy(DTD *newDtd, const DTD *oldDtd) |
3488 |
|
|
{ |
3489 |
|
|
HASH_TABLE_ITER iter; |
3490 |
|
|
|
3491 |
|
|
/* Copy the prefix table. */ |
3492 |
|
|
|
3493 |
|
|
hashTableIterInit(&iter, &(oldDtd->prefixes)); |
3494 |
|
|
for (;;) { |
3495 |
|
|
const XML_Char *name; |
3496 |
|
|
const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); |
3497 |
|
|
if (!oldP) |
3498 |
|
|
break; |
3499 |
|
|
name = poolCopyString(&(newDtd->pool), oldP->name); |
3500 |
|
|
if (!name) |
3501 |
|
|
return 0; |
3502 |
|
|
if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX))) |
3503 |
|
|
return 0; |
3504 |
|
|
} |
3505 |
|
|
|
3506 |
|
|
hashTableIterInit(&iter, &(oldDtd->attributeIds)); |
3507 |
|
|
|
3508 |
|
|
/* Copy the attribute id table. */ |
3509 |
|
|
|
3510 |
|
|
for (;;) { |
3511 |
|
|
ATTRIBUTE_ID *newA; |
3512 |
|
|
const XML_Char *name; |
3513 |
|
|
const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); |
3514 |
|
|
|
3515 |
|
|
if (!oldA) |
3516 |
|
|
break; |
3517 |
|
|
/* Remember to allocate the scratch byte before the name. */ |
3518 |
|
|
if (!poolAppendChar(&(newDtd->pool), XML_T('\0'))) |
3519 |
|
|
return 0; |
3520 |
|
|
name = poolCopyString(&(newDtd->pool), oldA->name); |
3521 |
|
|
if (!name) |
3522 |
|
|
return 0; |
3523 |
|
|
++name; |
3524 |
|
|
newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name, sizeof(ATTRIBUTE_ID)); |
3525 |
|
|
if (!newA) |
3526 |
|
|
return 0; |
3527 |
|
|
newA->maybeTokenized = oldA->maybeTokenized; |
3528 |
|
|
if (oldA->prefix) { |
3529 |
|
|
newA->xmlns = oldA->xmlns; |
3530 |
|
|
if (oldA->prefix == &oldDtd->defaultPrefix) |
3531 |
|
|
newA->prefix = &newDtd->defaultPrefix; |
3532 |
|
|
else |
3533 |
|
|
newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldA->prefix->name, 0); |
3534 |
|
|
} |
3535 |
|
|
} |
3536 |
|
|
|
3537 |
|
|
/* Copy the element type table. */ |
3538 |
|
|
|
3539 |
|
|
hashTableIterInit(&iter, &(oldDtd->elementTypes)); |
3540 |
|
|
|
3541 |
|
|
for (;;) { |
3542 |
|
|
int i; |
3543 |
|
|
ELEMENT_TYPE *newE; |
3544 |
|
|
const XML_Char *name; |
3545 |
|
|
const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); |
3546 |
|
|
if (!oldE) |
3547 |
|
|
break; |
3548 |
|
|
name = poolCopyString(&(newDtd->pool), oldE->name); |
3549 |
|
|
if (!name) |
3550 |
|
|
return 0; |
3551 |
|
|
newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name, sizeof(ELEMENT_TYPE)); |
3552 |
|
|
if (!newE) |
3553 |
|
|
return 0; |
3554 |
|
|
if (oldE->nDefaultAtts) { |
3555 |
|
|
newE->defaultAtts = (DEFAULT_ATTRIBUTE *)malloc(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); |
3556 |
|
|
if (!newE->defaultAtts) |
3557 |
|
|
return 0; |
3558 |
|
|
} |
3559 |
|
|
if (oldE->idAtt) |
3560 |
|
|
newE->idAtt = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0); |
3561 |
|
|
newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; |
3562 |
|
|
if (oldE->prefix) |
3563 |
|
|
newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldE->prefix->name, 0); |
3564 |
|
|
for (i = 0; i < newE->nDefaultAtts; i++) { |
3565 |
|
|
newE->defaultAtts[i].id = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); |
3566 |
|
|
newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; |
3567 |
|
|
if (oldE->defaultAtts[i].value) { |
3568 |
|
|
newE->defaultAtts[i].value = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); |
3569 |
|
|
if (!newE->defaultAtts[i].value) |
3570 |
|
|
return 0; |
3571 |
|
|
} |
3572 |
|
|
else |
3573 |
|
|
newE->defaultAtts[i].value = 0; |
3574 |
|
|
} |
3575 |
|
|
} |
3576 |
|
|
|
3577 |
|
|
/* Copy the entity tables. */ |
3578 |
|
|
if (!copyEntityTable(&(newDtd->generalEntities), |
3579 |
|
|
&(newDtd->pool), |
3580 |
|
|
&(oldDtd->generalEntities))) |
3581 |
|
|
return 0; |
3582 |
|
|
|
3583 |
|
|
#ifdef XML_DTD |
3584 |
|
|
if (!copyEntityTable(&(newDtd->paramEntities), |
3585 |
|
|
&(newDtd->pool), |
3586 |
|
|
&(oldDtd->paramEntities))) |
3587 |
|
|
return 0; |
3588 |
|
|
#endif /* XML_DTD */ |
3589 |
|
|
|
3590 |
|
|
newDtd->complete = oldDtd->complete; |
3591 |
|
|
newDtd->standalone = oldDtd->standalone; |
3592 |
|
|
return 1; |
3593 |
|
|
} |
3594 |
|
|
|
3595 |
|
|
static int copyEntityTable(HASH_TABLE *newTable, |
3596 |
|
|
STRING_POOL *newPool, |
3597 |
|
|
const HASH_TABLE *oldTable) |
3598 |
|
|
{ |
3599 |
|
|
HASH_TABLE_ITER iter; |
3600 |
|
|
const XML_Char *cachedOldBase = 0; |
3601 |
|
|
const XML_Char *cachedNewBase = 0; |
3602 |
|
|
|
3603 |
|
|
hashTableIterInit(&iter, oldTable); |
3604 |
|
|
|
3605 |
|
|
for (;;) { |
3606 |
|
|
ENTITY *newE; |
3607 |
|
|
const XML_Char *name; |
3608 |
|
|
const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); |
3609 |
|
|
if (!oldE) |
3610 |
|
|
break; |
3611 |
|
|
name = poolCopyString(newPool, oldE->name); |
3612 |
|
|
if (!name) |
3613 |
|
|
return 0; |
3614 |
|
|
newE = (ENTITY *)lookup(newTable, name, sizeof(ENTITY)); |
3615 |
|
|
if (!newE) |
3616 |
|
|
return 0; |
3617 |
|
|
if (oldE->systemId) { |
3618 |
|
|
const XML_Char *tem = poolCopyString(newPool, oldE->systemId); |
3619 |
|
|
if (!tem) |
3620 |
|
|
return 0; |
3621 |
|
|
newE->systemId = tem; |
3622 |
|
|
if (oldE->base) { |
3623 |
|
|
if (oldE->base == cachedOldBase) |
3624 |
|
|
newE->base = cachedNewBase; |
3625 |
|
|
else { |
3626 |
|
|
cachedOldBase = oldE->base; |
3627 |
|
|
tem = poolCopyString(newPool, cachedOldBase); |
3628 |
|
|
if (!tem) |
3629 |
|
|
return 0; |
3630 |
|
|
cachedNewBase = newE->base = tem; |
3631 |
|
|
} |
3632 |
|
|
} |
3633 |
|
|
} |
3634 |
|
|
else { |
3635 |
|
|
const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr, oldE->textLen); |
3636 |
|
|
if (!tem) |
3637 |
|
|
return 0; |
3638 |
|
|
newE->textPtr = tem; |
3639 |
|
|
newE->textLen = oldE->textLen; |
3640 |
|
|
} |
3641 |
|
|
if (oldE->notation) { |
3642 |
|
|
const XML_Char *tem = poolCopyString(newPool, oldE->notation); |
3643 |
|
|
if (!tem) |
3644 |
|
|
return 0; |
3645 |
|
|
newE->notation = tem; |
3646 |
|
|
} |
3647 |
|
|
} |
3648 |
|
|
return 1; |
3649 |
|
|
} |
3650 |
|
|
|
3651 |
|
|
#define INIT_SIZE 64 |
3652 |
|
|
|
3653 |
|
|
static |
3654 |
|
|
int keyeq(KEY s1, KEY s2) |
3655 |
|
|
{ |
3656 |
|
|
for (; *s1 == *s2; s1++, s2++) |
3657 |
|
|
if (*s1 == 0) |
3658 |
|
|
return 1; |
3659 |
|
|
return 0; |
3660 |
|
|
} |
3661 |
|
|
|
3662 |
|
|
static |
3663 |
|
|
unsigned long hash(KEY s) |
3664 |
|
|
{ |
3665 |
|
|
unsigned long h = 0; |
3666 |
|
|
while (*s) |
3667 |
|
|
h = (h << 5) + h + (unsigned char)*s++; |
3668 |
|
|
return h; |
3669 |
|
|
} |
3670 |
|
|
|
3671 |
|
|
static |
3672 |
|
|
NAMED *lookup(HASH_TABLE *table, KEY name, size_t createSize) |
3673 |
|
|
{ |
3674 |
|
|
size_t i; |
3675 |
|
|
if (table->size == 0) { |
3676 |
|
|
if (!createSize) |
3677 |
|
|
return 0; |
3678 |
|
|
table->v = calloc(INIT_SIZE, sizeof(NAMED *)); |
3679 |
|
|
if (!table->v) |
3680 |
|
|
return 0; |
3681 |
|
|
table->size = INIT_SIZE; |
3682 |
|
|
table->usedLim = INIT_SIZE / 2; |
3683 |
|
|
i = hash(name) & (table->size - 1); |
3684 |
|
|
} |
3685 |
|
|
else { |
3686 |
|
|
unsigned long h = hash(name); |
3687 |
|
|
for (i = h & (table->size - 1); |
3688 |
|
|
table->v[i]; |
3689 |
|
|
i == 0 ? i = table->size - 1 : --i) { |
3690 |
|
|
if (keyeq(name, table->v[i]->name)) |
3691 |
|
|
return table->v[i]; |
3692 |
|
|
} |
3693 |
|
|
if (!createSize) |
3694 |
|
|
return 0; |
3695 |
|
|
if (table->used == table->usedLim) { |
3696 |
|
|
/* check for overflow */ |
3697 |
|
|
size_t newSize = table->size * 2; |
3698 |
|
|
NAMED **newV = calloc(newSize, sizeof(NAMED *)); |
3699 |
|
|
if (!newV) |
3700 |
|
|
return 0; |
3701 |
|
|
for (i = 0; i < table->size; i++) |
3702 |
|
|
if (table->v[i]) { |
3703 |
|
|
size_t j; |
3704 |
|
|
for (j = hash(table->v[i]->name) & (newSize - 1); |
3705 |
|
|
newV[j]; |
3706 |
|
|
j == 0 ? j = newSize - 1 : --j) |
3707 |
|
|
; |
3708 |
|
|
newV[j] = table->v[i]; |
3709 |
|
|
} |
3710 |
|
|
free(table->v); |
3711 |
|
|
table->v = newV; |
3712 |
|
|
table->size = newSize; |
3713 |
|
|
table->usedLim = newSize/2; |
3714 |
|
|
for (i = h & (table->size - 1); |
3715 |
|
|
table->v[i]; |
3716 |
|
|
i == 0 ? i = table->size - 1 : --i) |
3717 |
|
|
; |
3718 |
|
|
} |
3719 |
|
|
} |
3720 |
|
|
table->v[i] = calloc(1, createSize); |
3721 |
|
|
if (!table->v[i]) |
3722 |
|
|
return 0; |
3723 |
|
|
table->v[i]->name = name; |
3724 |
|
|
(table->used)++; |
3725 |
|
|
return table->v[i]; |
3726 |
|
|
} |
3727 |
|
|
|
3728 |
|
|
static |
3729 |
|
|
void hashTableDestroy(HASH_TABLE *table) |
3730 |
|
|
{ |
3731 |
|
|
size_t i; |
3732 |
|
|
for (i = 0; i < table->size; i++) { |
3733 |
|
|
NAMED *p = table->v[i]; |
3734 |
|
|
if (p) |
3735 |
|
|
free(p); |
3736 |
|
|
} |
3737 |
|
|
if (table->v) |
3738 |
|
|
free(table->v); |
3739 |
|
|
} |
3740 |
|
|
|
3741 |
|
|
static |
3742 |
|
|
void hashTableInit(HASH_TABLE *p) |
3743 |
|
|
{ |
3744 |
|
|
p->size = 0; |
3745 |
|
|
p->usedLim = 0; |
3746 |
|
|
p->used = 0; |
3747 |
|
|
p->v = 0; |
3748 |
|
|
} |
3749 |
|
|
|
3750 |
|
|
static |
3751 |
|
|
void hashTableIterInit(HASH_TABLE_ITER *iter, const HASH_TABLE *table) |
3752 |
|
|
{ |
3753 |
|
|
iter->p = table->v; |
3754 |
|
|
iter->end = iter->p + table->size; |
3755 |
|
|
} |
3756 |
|
|
|
3757 |
|
|
static |
3758 |
|
|
NAMED *hashTableIterNext(HASH_TABLE_ITER *iter) |
3759 |
|
|
{ |
3760 |
|
|
while (iter->p != iter->end) { |
3761 |
|
|
NAMED *tem = *(iter->p)++; |
3762 |
|
|
if (tem) |
3763 |
|
|
return tem; |
3764 |
|
|
} |
3765 |
|
|
return 0; |
3766 |
|
|
} |
3767 |
|
|
|
3768 |
|
|
|
3769 |
|
|
static |
3770 |
|
|
void poolInit(STRING_POOL *pool) |
3771 |
|
|
{ |
3772 |
|
|
pool->blocks = 0; |
3773 |
|
|
pool->freeBlocks = 0; |
3774 |
|
|
pool->start = 0; |
3775 |
|
|
pool->ptr = 0; |
3776 |
|
|
pool->end = 0; |
3777 |
|
|
} |
3778 |
|
|
|
3779 |
|
|
static |
3780 |
|
|
void poolClear(STRING_POOL *pool) |
3781 |
|
|
{ |
3782 |
|
|
if (!pool->freeBlocks) |
3783 |
|
|
pool->freeBlocks = pool->blocks; |
3784 |
|
|
else { |
3785 |
|
|
BLOCK *p = pool->blocks; |
3786 |
|
|
while (p) { |
3787 |
|
|
BLOCK *tem = p->next; |
3788 |
|
|
p->next = pool->freeBlocks; |
3789 |
|
|
pool->freeBlocks = p; |
3790 |
|
|
p = tem; |
3791 |
|
|
} |
3792 |
|
|
} |
3793 |
|
|
pool->blocks = 0; |
3794 |
|
|
pool->start = 0; |
3795 |
|
|
pool->ptr = 0; |
3796 |
|
|
pool->end = 0; |
3797 |
|
|
} |
3798 |
|
|
|
3799 |
|
|
static |
3800 |
|
|
void poolDestroy(STRING_POOL *pool) |
3801 |
|
|
{ |
3802 |
|
|
BLOCK *p = pool->blocks; |
3803 |
|
|
while (p) { |
3804 |
|
|
BLOCK *tem = p->next; |
3805 |
|
|
free(p); |
3806 |
|
|
p = tem; |
3807 |
|
|
} |
3808 |
|
|
pool->blocks = 0; |
3809 |
|
|
p = pool->freeBlocks; |
3810 |
|
|
while (p) { |
3811 |
|
|
BLOCK *tem = p->next; |
3812 |
|
|
free(p); |
3813 |
|
|
p = tem; |
3814 |
|
|
} |
3815 |
|
|
pool->freeBlocks = 0; |
3816 |
|
|
pool->ptr = 0; |
3817 |
|
|
pool->start = 0; |
3818 |
|
|
pool->end = 0; |
3819 |
|
|
} |
3820 |
|
|
|
3821 |
|
|
static |
3822 |
|
|
XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, |
3823 |
|
|
const char *ptr, const char *end) |
3824 |
|
|
{ |
3825 |
|
|
if (!pool->ptr && !poolGrow(pool)) |
3826 |
|
|
return 0; |
3827 |
|
|
for (;;) { |
3828 |
|
|
XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); |
3829 |
|
|
if (ptr == end) |
3830 |
|
|
break; |
3831 |
|
|
if (!poolGrow(pool)) |
3832 |
|
|
return 0; |
3833 |
|
|
} |
3834 |
|
|
return pool->start; |
3835 |
|
|
} |
3836 |
|
|
|
3837 |
|
|
static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s) |
3838 |
|
|
{ |
3839 |
|
|
do { |
3840 |
|
|
if (!poolAppendChar(pool, *s)) |
3841 |
|
|
return 0; |
3842 |
|
|
} while (*s++); |
3843 |
|
|
s = pool->start; |
3844 |
|
|
poolFinish(pool); |
3845 |
|
|
return s; |
3846 |
|
|
} |
3847 |
|
|
|
3848 |
|
|
static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) |
3849 |
|
|
{ |
3850 |
|
|
if (!pool->ptr && !poolGrow(pool)) |
3851 |
|
|
return 0; |
3852 |
|
|
for (; n > 0; --n, s++) { |
3853 |
|
|
if (!poolAppendChar(pool, *s)) |
3854 |
|
|
return 0; |
3855 |
|
|
|
3856 |
|
|
} |
3857 |
|
|
s = pool->start; |
3858 |
|
|
poolFinish(pool); |
3859 |
|
|
return s; |
3860 |
|
|
} |
3861 |
|
|
|
3862 |
|
|
static |
3863 |
|
|
XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, |
3864 |
|
|
const char *ptr, const char *end) |
3865 |
|
|
{ |
3866 |
|
|
if (!poolAppend(pool, enc, ptr, end)) |
3867 |
|
|
return 0; |
3868 |
|
|
if (pool->ptr == pool->end && !poolGrow(pool)) |
3869 |
|
|
return 0; |
3870 |
|
|
*(pool->ptr)++ = 0; |
3871 |
|
|
return pool->start; |
3872 |
|
|
} |
3873 |
|
|
|
3874 |
|
|
static |
3875 |
|
|
int poolGrow(STRING_POOL *pool) |
3876 |
|
|
{ |
3877 |
|
|
if (pool->freeBlocks) { |
3878 |
|
|
if (pool->start == 0) { |
3879 |
|
|
pool->blocks = pool->freeBlocks; |
3880 |
|
|
pool->freeBlocks = pool->freeBlocks->next; |
3881 |
|
|
pool->blocks->next = 0; |
3882 |
|
|
pool->start = pool->blocks->s; |
3883 |
|
|
pool->end = pool->start + pool->blocks->size; |
3884 |
|
|
pool->ptr = pool->start; |
3885 |
|
|
return 1; |
3886 |
|
|
} |
3887 |
|
|
if (pool->end - pool->start < pool->freeBlocks->size) { |
3888 |
|
|
BLOCK *tem = pool->freeBlocks->next; |
3889 |
|
|
pool->freeBlocks->next = pool->blocks; |
3890 |
|
|
pool->blocks = pool->freeBlocks; |
3891 |
|
|
pool->freeBlocks = tem; |
3892 |
|
|
memcpy(pool->blocks->s, pool->start, (pool->end - pool->start) * sizeof(XML_Char)); |
3893 |
|
|
pool->ptr = pool->blocks->s + (pool->ptr - pool->start); |
3894 |
|
|
pool->start = pool->blocks->s; |
3895 |
|
|
pool->end = pool->start + pool->blocks->size; |
3896 |
|
|
return 1; |
3897 |
|
|
} |
3898 |
|
|
} |
3899 |
|
|
if (pool->blocks && pool->start == pool->blocks->s) { |
3900 |
|
|
int blockSize = (pool->end - pool->start)*2; |
3901 |
|
|
pool->blocks = realloc(pool->blocks, offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); |
3902 |
|
|
if (!pool->blocks) |
3903 |
|
|
return 0; |
3904 |
|
|
pool->blocks->size = blockSize; |
3905 |
|
|
pool->ptr = pool->blocks->s + (pool->ptr - pool->start); |
3906 |
|
|
pool->start = pool->blocks->s; |
3907 |
|
|
pool->end = pool->start + blockSize; |
3908 |
|
|
} |
3909 |
|
|
else { |
3910 |
|
|
BLOCK *tem; |
3911 |
|
|
int blockSize = pool->end - pool->start; |
3912 |
|
|
if (blockSize < INIT_BLOCK_SIZE) |
3913 |
|
|
blockSize = INIT_BLOCK_SIZE; |
3914 |
|
|
else |
3915 |
|
|
blockSize *= 2; |
3916 |
|
|
tem = malloc(offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); |
3917 |
|
|
if (!tem) |
3918 |
|
|
return 0; |
3919 |
|
|
tem->size = blockSize; |
3920 |
|
|
tem->next = pool->blocks; |
3921 |
|
|
pool->blocks = tem; |
3922 |
|
|
if (pool->ptr != pool->start) |
3923 |
|
|
memcpy(tem->s, pool->start, (pool->ptr - pool->start) * sizeof(XML_Char)); |
3924 |
|
|
pool->ptr = tem->s + (pool->ptr - pool->start); |
3925 |
|
|
pool->start = tem->s; |
3926 |
|
|
pool->end = tem->s + blockSize; |
3927 |
|
|
} |
3928 |
|
|
return 1; |
3929 |
|
|
} |