Tue, 07 Jan 2003 20:57:48 +0000
[gaim-migrate @ 4474]
Is it ironic that I use Time Warner cable to help develope an
AIM client?
| 2086 | 1 | /* |
| 2 | The contents of this file are subject to the Mozilla Public License | |
| 3 | Version 1.1 (the "License"); you may not use this file except in | |
| 4 | compliance with the License. You may obtain a copy of the License at | |
| 5 | http://www.mozilla.org/MPL/ | |
| 6 | ||
| 7 | Software distributed under the License is distributed on an "AS IS" | |
| 8 | basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the | |
| 9 | License for the specific language governing rights and limitations | |
| 10 | under the License. | |
| 11 | ||
| 12 | The Original Code is expat. | |
| 13 | ||
| 14 | The Initial Developer of the Original Code is James Clark. | |
| 15 | Portions created by James Clark are Copyright (C) 1998, 1999 | |
| 16 | James Clark. All Rights Reserved. | |
| 17 | ||
| 18 | Contributor(s): | |
| 19 | ||
| 20 | */ | |
| 21 | ||
| 22 | #include "xmldef.h" | |
| 23 | #include "xmlparse.h" | |
| 24 | ||
| 25 | #ifdef XML_UNICODE | |
| 26 | #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX | |
| 27 | #define XmlConvert XmlUtf16Convert | |
| 28 | #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding | |
| 29 | #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS | |
| 30 | #define XmlEncode XmlUtf16Encode | |
| 31 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1)) | |
| 32 | typedef unsigned short ICHAR; | |
| 33 | #else | |
| 34 | #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX | |
| 35 | #define XmlConvert XmlUtf8Convert | |
| 36 | #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding | |
| 37 | #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS | |
| 38 | #define XmlEncode XmlUtf8Encode | |
| 39 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf8) | |
| 40 | typedef char ICHAR; | |
| 41 | #endif | |
| 42 | ||
| 43 | ||
| 44 | #ifndef XML_NS | |
| 45 | ||
| 46 | #define XmlInitEncodingNS XmlInitEncoding | |
| 47 | #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding | |
| 48 | #undef XmlGetInternalEncodingNS | |
| 49 | #define XmlGetInternalEncodingNS XmlGetInternalEncoding | |
| 50 | #define XmlParseXmlDeclNS XmlParseXmlDecl | |
| 51 | ||
| 52 | #endif | |
| 53 | ||
| 54 | ||
| 55 | #ifdef XML_UNICODE_WCHAR_T | |
| 56 | #define XML_T(x) L ## x | |
| 57 | #else | |
| 58 | #define XML_T(x) x | |
| 59 | #endif | |
| 60 | ||
| 61 | /* Round up n to be a multiple of sz, where sz is a power of 2. */ | |
| 62 | #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1)) | |
| 63 | ||
| 64 | #include "xmltok.h" | |
| 65 | #include "xmlrole.h" | |
| 66 | #include "hashtable.h" | |
| 67 | ||
| 68 | #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */ | |
| 69 | #define INIT_DATA_BUF_SIZE 1024 | |
| 70 | #define INIT_ATTS_SIZE 16 | |
| 71 | #define INIT_BLOCK_SIZE 1024 | |
| 72 | #define INIT_BUFFER_SIZE 1024 | |
| 73 | ||
| 74 | #define EXPAND_SPARE 24 | |
| 75 | ||
| 76 | typedef struct binding { | |
| 77 | struct prefix *prefix; | |
| 78 | struct binding *nextTagBinding; | |
| 79 | struct binding *prevPrefixBinding; | |
| 80 | const struct attribute_id *attId; | |
| 81 | XML_Char *uri; | |
| 82 | int uriLen; | |
| 83 | int uriAlloc; | |
| 84 | } BINDING; | |
| 85 | ||
| 86 | typedef struct prefix { | |
| 87 | const XML_Char *name; | |
| 88 | BINDING *binding; | |
| 89 | } PREFIX; | |
| 90 | ||
| 91 | typedef struct { | |
| 92 | const XML_Char *str; | |
| 93 | const XML_Char *localPart; | |
| 94 | int uriLen; | |
| 95 | } TAG_NAME; | |
| 96 | ||
| 97 | typedef struct tag { | |
| 98 | struct tag *parent; | |
| 99 | const char *rawName; | |
| 100 | int rawNameLength; | |
| 101 | TAG_NAME name; | |
| 102 | char *buf; | |
| 103 | char *bufEnd; | |
| 104 | BINDING *bindings; | |
| 105 | } TAG; | |
| 106 | ||
| 107 | typedef struct { | |
| 108 | const XML_Char *name; | |
| 109 | const XML_Char *textPtr; | |
| 110 | int textLen; | |
| 111 | const XML_Char *systemId; | |
| 112 | const XML_Char *base; | |
| 113 | const XML_Char *publicId; | |
| 114 | const XML_Char *notation; | |
| 115 | char open; | |
| 116 | } ENTITY; | |
| 117 | ||
| 118 | typedef struct block { | |
| 119 | struct block *next; | |
| 120 | int size; | |
| 121 | XML_Char s[1]; | |
| 122 | } BLOCK; | |
| 123 | ||
| 124 | typedef struct { | |
| 125 | BLOCK *blocks; | |
| 126 | BLOCK *freeBlocks; | |
| 127 | const XML_Char *end; | |
| 128 | XML_Char *ptr; | |
| 129 | XML_Char *start; | |
| 130 | } STRING_POOL; | |
| 131 | ||
| 132 | /* The XML_Char before the name is used to determine whether | |
| 133 | an attribute has been specified. */ | |
| 134 | typedef struct attribute_id { | |
| 135 | XML_Char *name; | |
| 136 | PREFIX *prefix; | |
| 137 | char maybeTokenized; | |
| 138 | char xmlns; | |
| 139 | } ATTRIBUTE_ID; | |
| 140 | ||
| 141 | typedef struct { | |
| 142 | const ATTRIBUTE_ID *id; | |
| 143 | char isCdata; | |
| 144 | const XML_Char *value; | |
| 145 | } DEFAULT_ATTRIBUTE; | |
| 146 | ||
| 147 | typedef struct { | |
| 148 | const XML_Char *name; | |
| 149 | PREFIX *prefix; | |
| 150 | int nDefaultAtts; | |
| 151 | int allocDefaultAtts; | |
| 152 | DEFAULT_ATTRIBUTE *defaultAtts; | |
| 153 | } ELEMENT_TYPE; | |
| 154 | ||
| 155 | typedef struct { | |
| 156 | HASH_TABLE generalEntities; | |
| 157 | HASH_TABLE elementTypes; | |
| 158 | HASH_TABLE attributeIds; | |
| 159 | HASH_TABLE prefixes; | |
| 160 | STRING_POOL pool; | |
| 161 | int complete; | |
| 162 | int standalone; | |
| 163 | const XML_Char *base; | |
| 164 | PREFIX defaultPrefix; | |
| 165 | } DTD; | |
| 166 | ||
| 167 | typedef struct open_internal_entity { | |
| 168 | const char *internalEventPtr; | |
| 169 | const char *internalEventEndPtr; | |
| 170 | struct open_internal_entity *next; | |
| 171 | ENTITY *entity; | |
| 172 | } OPEN_INTERNAL_ENTITY; | |
| 173 | ||
| 174 | typedef enum XML_Error Processor(XML_Parser parser, | |
| 175 | const char *start, | |
| 176 | const char *end, | |
| 177 | const char **endPtr); | |
| 178 | ||
| 179 | static Processor prologProcessor; | |
| 180 | static Processor prologInitProcessor; | |
| 181 | static Processor contentProcessor; | |
| 182 | static Processor cdataSectionProcessor; | |
| 183 | static Processor epilogProcessor; | |
| 184 | static Processor externalEntityInitProcessor; | |
| 185 | static Processor externalEntityInitProcessor2; | |
| 186 | static Processor externalEntityInitProcessor3; | |
| 187 | static Processor externalEntityContentProcessor; | |
| 188 | ||
| 189 | static enum XML_Error | |
| 190 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName); | |
| 191 | static enum XML_Error | |
| 192 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, const char *, const char *); | |
| 193 | static enum XML_Error | |
| 194 | initializeEncoding(XML_Parser parser); | |
| 195 | static enum XML_Error | |
| 196 | doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc, | |
| 197 | const char *start, const char *end, const char **endPtr); | |
| 198 | static enum XML_Error | |
| 199 | doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr); | |
| 200 | static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *, const char *s, | |
| 201 | TAG_NAME *tagNamePtr, BINDING **bindingsPtr); | |
| 202 | static | |
| 203 | int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr); | |
| 204 | static int | |
| 205 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *, int isCdata, const XML_Char *dfltValue); | |
| 206 | static enum XML_Error | |
| 207 | storeAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, | |
| 208 | STRING_POOL *); | |
| 209 | static enum XML_Error | |
| 210 | appendAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *, | |
| 211 | STRING_POOL *); | |
| 212 | static ATTRIBUTE_ID * | |
| 213 | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
| 214 | static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *); | |
| 215 | static enum XML_Error | |
| 216 | storeEntityValue(XML_Parser parser, const char *start, const char *end); | |
| 217 | static int | |
| 218 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
| 219 | static int | |
| 220 | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
| 221 | static void | |
| 222 | reportDefault(XML_Parser parser, const ENCODING *enc, const char *start, const char *end); | |
| 223 | ||
| 224 | static const XML_Char *getContext(XML_Parser parser); | |
| 225 | static int setContext(XML_Parser parser, const XML_Char *context); | |
| 226 | static void normalizePublicId(XML_Char *s); | |
| 227 | static int dtdInit(DTD *); | |
| 228 | static void dtdDestroy(DTD *); | |
| 229 | static int dtdCopy(DTD *newDtd, const DTD *oldDtd); | |
| 230 | static void poolInit(STRING_POOL *); | |
| 231 | static void poolClear(STRING_POOL *); | |
| 232 | static void poolDestroy(STRING_POOL *); | |
| 233 | static XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
| 234 | const char *ptr, const char *end); | |
| 235 | static XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
| 236 | const char *ptr, const char *end); | |
| 237 | static int poolGrow(STRING_POOL *pool); | |
| 238 | static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s); | |
| 239 | static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n); | |
| 240 | ||
| 241 | #define poolStart(pool) ((pool)->start) | |
| 242 | #define poolEnd(pool) ((pool)->ptr) | |
| 243 | #define poolLength(pool) ((pool)->ptr - (pool)->start) | |
| 244 | #define poolChop(pool) ((void)--(pool->ptr)) | |
| 245 | #define poolLastChar(pool) (((pool)->ptr)[-1]) | |
| 246 | #define poolDiscard(pool) ((pool)->ptr = (pool)->start) | |
| 247 | #define poolFinish(pool) ((pool)->start = (pool)->ptr) | |
| 248 | #define poolAppendChar(pool, c) \ | |
| 249 | (((pool)->ptr == (pool)->end && !poolGrow(pool)) \ | |
| 250 | ? 0 \ | |
| 251 | : ((*((pool)->ptr)++ = c), 1)) | |
| 252 | ||
| 253 | typedef struct { | |
| 254 | /* The first member must be userData so that the XML_GetUserData macro works. */ | |
| 255 | void *m_userData; | |
| 256 | void *m_handlerArg; | |
| 257 | char *m_buffer; | |
| 258 | /* first character to be parsed */ | |
| 259 | const char *m_bufferPtr; | |
| 260 | /* past last character to be parsed */ | |
| 261 | char *m_bufferEnd; | |
| 262 | /* allocated end of buffer */ | |
| 263 | const char *m_bufferLim; | |
| 264 | long m_parseEndByteIndex; | |
| 265 | const char *m_parseEndPtr; | |
| 266 | XML_Char *m_dataBuf; | |
| 267 | XML_Char *m_dataBufEnd; | |
| 268 | XML_StartElementHandler m_startElementHandler; | |
| 269 | XML_EndElementHandler m_endElementHandler; | |
| 270 | XML_CharacterDataHandler m_characterDataHandler; | |
| 271 | XML_ProcessingInstructionHandler m_processingInstructionHandler; | |
| 272 | XML_CommentHandler m_commentHandler; | |
| 273 | XML_StartCdataSectionHandler m_startCdataSectionHandler; | |
| 274 | XML_EndCdataSectionHandler m_endCdataSectionHandler; | |
| 275 | XML_DefaultHandler m_defaultHandler; | |
| 276 | XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler; | |
| 277 | XML_NotationDeclHandler m_notationDeclHandler; | |
| 278 | XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler; | |
| 279 | XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler; | |
| 280 | XML_NotStandaloneHandler m_notStandaloneHandler; | |
| 281 | XML_ExternalEntityRefHandler m_externalEntityRefHandler; | |
| 282 | void *m_externalEntityRefHandlerArg; | |
| 283 | XML_UnknownEncodingHandler m_unknownEncodingHandler; | |
| 284 | const ENCODING *m_encoding; | |
| 285 | INIT_ENCODING m_initEncoding; | |
| 286 | const XML_Char *m_protocolEncodingName; | |
| 287 | int m_ns; | |
| 288 | void *m_unknownEncodingMem; | |
| 289 | void *m_unknownEncodingData; | |
| 290 | void *m_unknownEncodingHandlerData; | |
| 291 | void (*m_unknownEncodingRelease)(void *); | |
| 292 | PROLOG_STATE m_prologState; | |
| 293 | Processor *m_processor; | |
| 294 | enum XML_Error m_errorCode; | |
| 295 | const char *m_eventPtr; | |
| 296 | const char *m_eventEndPtr; | |
| 297 | const char *m_positionPtr; | |
| 298 | OPEN_INTERNAL_ENTITY *m_openInternalEntities; | |
| 299 | int m_defaultExpandInternalEntities; | |
| 300 | int m_tagLevel; | |
| 301 | ENTITY *m_declEntity; | |
| 302 | const XML_Char *m_declNotationName; | |
| 303 | const XML_Char *m_declNotationPublicId; | |
| 304 | ELEMENT_TYPE *m_declElementType; | |
| 305 | ATTRIBUTE_ID *m_declAttributeId; | |
| 306 | char m_declAttributeIsCdata; | |
| 307 | DTD m_dtd; | |
| 308 | TAG *m_tagStack; | |
| 309 | TAG *m_freeTagList; | |
| 310 | BINDING *m_inheritedBindings; | |
| 311 | BINDING *m_freeBindingList; | |
| 312 | int m_attsSize; | |
| 313 | int m_nSpecifiedAtts; | |
| 314 | ATTRIBUTE *m_atts; | |
| 315 | POSITION m_position; | |
| 316 | STRING_POOL m_tempPool; | |
| 317 | STRING_POOL m_temp2Pool; | |
| 318 | char *m_groupConnector; | |
| 319 | unsigned m_groupSize; | |
| 320 | int m_hadExternalDoctype; | |
| 321 | XML_Char m_namespaceSeparator; | |
| 322 | } Parser; | |
| 323 | ||
| 324 | #define userData (((Parser *)parser)->m_userData) | |
| 325 | #define handlerArg (((Parser *)parser)->m_handlerArg) | |
| 326 | #define startElementHandler (((Parser *)parser)->m_startElementHandler) | |
| 327 | #define endElementHandler (((Parser *)parser)->m_endElementHandler) | |
| 328 | #define characterDataHandler (((Parser *)parser)->m_characterDataHandler) | |
| 329 | #define processingInstructionHandler (((Parser *)parser)->m_processingInstructionHandler) | |
| 330 | #define commentHandler (((Parser *)parser)->m_commentHandler) | |
| 331 | #define startCdataSectionHandler (((Parser *)parser)->m_startCdataSectionHandler) | |
| 332 | #define endCdataSectionHandler (((Parser *)parser)->m_endCdataSectionHandler) | |
| 333 | #define defaultHandler (((Parser *)parser)->m_defaultHandler) | |
| 334 | #define unparsedEntityDeclHandler (((Parser *)parser)->m_unparsedEntityDeclHandler) | |
| 335 | #define notationDeclHandler (((Parser *)parser)->m_notationDeclHandler) | |
| 336 | #define startNamespaceDeclHandler (((Parser *)parser)->m_startNamespaceDeclHandler) | |
| 337 | #define endNamespaceDeclHandler (((Parser *)parser)->m_endNamespaceDeclHandler) | |
| 338 | #define notStandaloneHandler (((Parser *)parser)->m_notStandaloneHandler) | |
| 339 | #define externalEntityRefHandler (((Parser *)parser)->m_externalEntityRefHandler) | |
| 340 | #define externalEntityRefHandlerArg (((Parser *)parser)->m_externalEntityRefHandlerArg) | |
| 341 | #define unknownEncodingHandler (((Parser *)parser)->m_unknownEncodingHandler) | |
| 342 | #define encoding (((Parser *)parser)->m_encoding) | |
| 343 | #define initEncoding (((Parser *)parser)->m_initEncoding) | |
| 344 | #define unknownEncodingMem (((Parser *)parser)->m_unknownEncodingMem) | |
| 345 | #define unknownEncodingData (((Parser *)parser)->m_unknownEncodingData) | |
| 346 | #define unknownEncodingHandlerData \ | |
| 347 | (((Parser *)parser)->m_unknownEncodingHandlerData) | |
| 348 | #define unknownEncodingRelease (((Parser *)parser)->m_unknownEncodingRelease) | |
| 349 | #define protocolEncodingName (((Parser *)parser)->m_protocolEncodingName) | |
| 350 | #define ns (((Parser *)parser)->m_ns) | |
| 351 | #define prologState (((Parser *)parser)->m_prologState) | |
| 352 | #define processor (((Parser *)parser)->m_processor) | |
| 353 | #define errorCode (((Parser *)parser)->m_errorCode) | |
| 354 | #define eventPtr (((Parser *)parser)->m_eventPtr) | |
| 355 | #define eventEndPtr (((Parser *)parser)->m_eventEndPtr) | |
| 356 | #define positionPtr (((Parser *)parser)->m_positionPtr) | |
| 357 | #define position (((Parser *)parser)->m_position) | |
| 358 | #define openInternalEntities (((Parser *)parser)->m_openInternalEntities) | |
| 359 | #define defaultExpandInternalEntities (((Parser *)parser)->m_defaultExpandInternalEntities) | |
| 360 | #define tagLevel (((Parser *)parser)->m_tagLevel) | |
| 361 | #define buffer (((Parser *)parser)->m_buffer) | |
| 362 | #define bufferPtr (((Parser *)parser)->m_bufferPtr) | |
| 363 | #define bufferEnd (((Parser *)parser)->m_bufferEnd) | |
| 364 | #define parseEndByteIndex (((Parser *)parser)->m_parseEndByteIndex) | |
| 365 | #define parseEndPtr (((Parser *)parser)->m_parseEndPtr) | |
| 366 | #define bufferLim (((Parser *)parser)->m_bufferLim) | |
| 367 | #define dataBuf (((Parser *)parser)->m_dataBuf) | |
| 368 | #define dataBufEnd (((Parser *)parser)->m_dataBufEnd) | |
| 369 | #define dtd (((Parser *)parser)->m_dtd) | |
| 370 | #define declEntity (((Parser *)parser)->m_declEntity) | |
| 371 | #define declNotationName (((Parser *)parser)->m_declNotationName) | |
| 372 | #define declNotationPublicId (((Parser *)parser)->m_declNotationPublicId) | |
| 373 | #define declElementType (((Parser *)parser)->m_declElementType) | |
| 374 | #define declAttributeId (((Parser *)parser)->m_declAttributeId) | |
| 375 | #define declAttributeIsCdata (((Parser *)parser)->m_declAttributeIsCdata) | |
| 376 | #define freeTagList (((Parser *)parser)->m_freeTagList) | |
| 377 | #define freeBindingList (((Parser *)parser)->m_freeBindingList) | |
| 378 | #define inheritedBindings (((Parser *)parser)->m_inheritedBindings) | |
| 379 | #define tagStack (((Parser *)parser)->m_tagStack) | |
| 380 | #define atts (((Parser *)parser)->m_atts) | |
| 381 | #define attsSize (((Parser *)parser)->m_attsSize) | |
| 382 | #define nSpecifiedAtts (((Parser *)parser)->m_nSpecifiedAtts) | |
| 383 | #define tempPool (((Parser *)parser)->m_tempPool) | |
| 384 | #define temp2Pool (((Parser *)parser)->m_temp2Pool) | |
| 385 | #define groupConnector (((Parser *)parser)->m_groupConnector) | |
| 386 | #define groupSize (((Parser *)parser)->m_groupSize) | |
| 387 | #define hadExternalDoctype (((Parser *)parser)->m_hadExternalDoctype) | |
| 388 | #define namespaceSeparator (((Parser *)parser)->m_namespaceSeparator) | |
| 389 | ||
| 390 | #ifdef _MSC_VER | |
| 391 | #ifdef _DEBUG | |
| 392 | Parser *asParser(XML_Parser parser) | |
| 393 | { | |
| 394 | return parser; | |
| 395 | } | |
| 396 | #endif | |
| 397 | #endif | |
| 398 | ||
| 399 | XML_Parser XML_ParserCreate(const XML_Char *encodingName) | |
| 400 | { | |
| 401 | XML_Parser parser = malloc(sizeof(Parser)); | |
| 402 | if (!parser) | |
| 403 | return parser; | |
| 404 | processor = prologInitProcessor; | |
| 405 | XmlPrologStateInit(&prologState); | |
| 406 | userData = 0; | |
| 407 | handlerArg = 0; | |
| 408 | startElementHandler = 0; | |
| 409 | endElementHandler = 0; | |
| 410 | characterDataHandler = 0; | |
| 411 | processingInstructionHandler = 0; | |
| 412 | commentHandler = 0; | |
| 413 | startCdataSectionHandler = 0; | |
| 414 | endCdataSectionHandler = 0; | |
| 415 | defaultHandler = 0; | |
| 416 | unparsedEntityDeclHandler = 0; | |
| 417 | notationDeclHandler = 0; | |
| 418 | startNamespaceDeclHandler = 0; | |
| 419 | endNamespaceDeclHandler = 0; | |
| 420 | notStandaloneHandler = 0; | |
| 421 | externalEntityRefHandler = 0; | |
| 422 | externalEntityRefHandlerArg = parser; | |
| 423 | unknownEncodingHandler = 0; | |
| 424 | buffer = 0; | |
| 425 | bufferPtr = 0; | |
| 426 | bufferEnd = 0; | |
| 427 | parseEndByteIndex = 0; | |
| 428 | parseEndPtr = 0; | |
| 429 | bufferLim = 0; | |
| 430 | declElementType = 0; | |
| 431 | declAttributeId = 0; | |
| 432 | declEntity = 0; | |
| 433 | declNotationName = 0; | |
| 434 | declNotationPublicId = 0; | |
| 435 | memset(&position, 0, sizeof(POSITION)); | |
| 436 | errorCode = XML_ERROR_NONE; | |
| 437 | eventPtr = 0; | |
| 438 | eventEndPtr = 0; | |
| 439 | positionPtr = 0; | |
| 440 | openInternalEntities = 0; | |
| 441 | tagLevel = 0; | |
| 442 | tagStack = 0; | |
| 443 | freeTagList = 0; | |
| 444 | freeBindingList = 0; | |
| 445 | inheritedBindings = 0; | |
| 446 | attsSize = INIT_ATTS_SIZE; | |
| 447 | atts = malloc(attsSize * sizeof(ATTRIBUTE)); | |
| 448 | nSpecifiedAtts = 0; | |
| 449 | dataBuf = malloc(INIT_DATA_BUF_SIZE * sizeof(XML_Char)); | |
| 450 | groupSize = 0; | |
| 451 | groupConnector = 0; | |
| 452 | hadExternalDoctype = 0; | |
| 453 | unknownEncodingMem = 0; | |
| 454 | unknownEncodingRelease = 0; | |
| 455 | unknownEncodingData = 0; | |
| 456 | unknownEncodingHandlerData = 0; | |
| 457 | namespaceSeparator = '!'; | |
| 458 | ns = 0; | |
| 459 | poolInit(&tempPool); | |
| 460 | poolInit(&temp2Pool); | |
| 461 | protocolEncodingName = encodingName ? poolCopyString(&tempPool, encodingName) : 0; | |
| 462 | if (!dtdInit(&dtd) || !atts || !dataBuf | |
| 463 | || (encodingName && !protocolEncodingName)) { | |
| 464 | XML_ParserFree(parser); | |
| 465 | return 0; | |
| 466 | } | |
| 467 | dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE; | |
| 468 | XmlInitEncoding(&initEncoding, &encoding, 0); | |
| 469 | return parser; | |
| 470 | } | |
| 471 | ||
| 472 | XML_Parser XML_ParserCreateNS(const XML_Char *encodingName, XML_Char nsSep) | |
| 473 | { | |
| 474 | static | |
| 475 | const XML_Char implicitContext[] = { | |
| 476 | XML_T('x'), XML_T('m'), XML_T('l'), XML_T('='), | |
| 477 | XML_T('h'), XML_T('t'), XML_T('t'), XML_T('p'), XML_T(':'), | |
| 478 | XML_T('/'), XML_T('/'), XML_T('w'), XML_T('w'), XML_T('w'), | |
| 479 | XML_T('.'), XML_T('w'), XML_T('3'), | |
| 480 | XML_T('.'), XML_T('o'), XML_T('r'), XML_T('g'), | |
| 481 | XML_T('/'), XML_T('X'), XML_T('M'), XML_T('L'), | |
| 482 | XML_T('/'), XML_T('1'), XML_T('9'), XML_T('9'), XML_T('8'), | |
| 483 | XML_T('/'), XML_T('n'), XML_T('a'), XML_T('m'), XML_T('e'), | |
| 484 | XML_T('s'), XML_T('p'), XML_T('a'), XML_T('c'), XML_T('e'), | |
| 485 | XML_T('\0') | |
| 486 | }; | |
| 487 | ||
| 488 | XML_Parser parser = XML_ParserCreate(encodingName); | |
| 489 | if (parser) { | |
| 490 | XmlInitEncodingNS(&initEncoding, &encoding, 0); | |
| 491 | ns = 1; | |
| 492 | namespaceSeparator = nsSep; | |
| 493 | } | |
| 494 | if (!setContext(parser, implicitContext)) { | |
| 495 | XML_ParserFree(parser); | |
| 496 | return 0; | |
| 497 | } | |
| 498 | return parser; | |
| 499 | } | |
| 500 | ||
| 501 | int XML_SetEncoding(XML_Parser parser, const XML_Char *encodingName) | |
| 502 | { | |
| 503 | if (!encodingName) | |
| 504 | protocolEncodingName = 0; | |
| 505 | else { | |
| 506 | protocolEncodingName = poolCopyString(&tempPool, encodingName); | |
| 507 | if (!protocolEncodingName) | |
| 508 | return 0; | |
| 509 | } | |
| 510 | return 1; | |
| 511 | } | |
| 512 | ||
| 513 | XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser, | |
| 514 | const XML_Char *context, | |
| 515 | const XML_Char *encodingName) | |
| 516 | { | |
| 517 | XML_Parser parser = oldParser; | |
| 518 | DTD *oldDtd = &dtd; | |
| 519 | XML_StartElementHandler oldStartElementHandler = startElementHandler; | |
| 520 | XML_EndElementHandler oldEndElementHandler = endElementHandler; | |
| 521 | XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler; | |
| 522 | XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler; | |
| 523 | XML_CommentHandler oldCommentHandler = commentHandler; | |
| 524 | XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler; | |
| 525 | XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler; | |
| 526 | XML_DefaultHandler oldDefaultHandler = defaultHandler; | |
| 527 | XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler = startNamespaceDeclHandler; | |
| 528 | XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler = endNamespaceDeclHandler; | |
| 529 | XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler; | |
| 530 | XML_ExternalEntityRefHandler oldExternalEntityRefHandler = externalEntityRefHandler; | |
| 531 | XML_UnknownEncodingHandler oldUnknownEncodingHandler = unknownEncodingHandler; | |
| 532 | void *oldUserData = userData; | |
| 533 | void *oldHandlerArg = handlerArg; | |
| 534 | int oldDefaultExpandInternalEntities = defaultExpandInternalEntities; | |
| 535 | void *oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg; | |
| 536 | ||
| 537 | parser = (ns | |
| 538 | ? XML_ParserCreateNS(encodingName, namespaceSeparator) | |
| 539 | : XML_ParserCreate(encodingName)); | |
| 540 | if (!parser) | |
| 541 | return 0; | |
| 542 | startElementHandler = oldStartElementHandler; | |
| 543 | endElementHandler = oldEndElementHandler; | |
| 544 | characterDataHandler = oldCharacterDataHandler; | |
| 545 | processingInstructionHandler = oldProcessingInstructionHandler; | |
| 546 | commentHandler = oldCommentHandler; | |
| 547 | startCdataSectionHandler = oldStartCdataSectionHandler; | |
| 548 | endCdataSectionHandler = oldEndCdataSectionHandler; | |
| 549 | defaultHandler = oldDefaultHandler; | |
| 550 | startNamespaceDeclHandler = oldStartNamespaceDeclHandler; | |
| 551 | endNamespaceDeclHandler = oldEndNamespaceDeclHandler; | |
| 552 | notStandaloneHandler = oldNotStandaloneHandler; | |
| 553 | externalEntityRefHandler = oldExternalEntityRefHandler; | |
| 554 | unknownEncodingHandler = oldUnknownEncodingHandler; | |
| 555 | userData = oldUserData; | |
| 556 | if (oldUserData == oldHandlerArg) | |
| 557 | handlerArg = userData; | |
| 558 | else | |
| 559 | handlerArg = parser; | |
| 560 | if (oldExternalEntityRefHandlerArg != oldParser) | |
| 561 | externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg; | |
| 562 | defaultExpandInternalEntities = oldDefaultExpandInternalEntities; | |
| 563 | if (!dtdCopy(&dtd, oldDtd) || !setContext(parser, context)) { | |
| 564 | XML_ParserFree(parser); | |
| 565 | return 0; | |
| 566 | } | |
| 567 | processor = externalEntityInitProcessor; | |
| 568 | return parser; | |
| 569 | } | |
| 570 | ||
| 571 | static | |
| 572 | void destroyBindings(BINDING *bindings) | |
| 573 | { | |
| 574 | for (;;) { | |
| 575 | BINDING *b = bindings; | |
| 576 | if (!b) | |
| 577 | break; | |
| 578 | bindings = b->nextTagBinding; | |
| 579 | free(b->uri); | |
| 580 | free(b); | |
| 581 | } | |
| 582 | } | |
| 583 | ||
| 584 | void XML_ParserFree(XML_Parser parser) | |
| 585 | { | |
| 586 | for (;;) { | |
| 587 | TAG *p; | |
| 588 | if (tagStack == 0) { | |
| 589 | if (freeTagList == 0) | |
| 590 | break; | |
| 591 | tagStack = freeTagList; | |
| 592 | freeTagList = 0; | |
| 593 | } | |
| 594 | p = tagStack; | |
| 595 | tagStack = tagStack->parent; | |
| 596 | free(p->buf); | |
| 597 | destroyBindings(p->bindings); | |
| 598 | free(p); | |
| 599 | } | |
| 600 | destroyBindings(freeBindingList); | |
| 601 | destroyBindings(inheritedBindings); | |
| 602 | poolDestroy(&tempPool); | |
| 603 | poolDestroy(&temp2Pool); | |
| 604 | dtdDestroy(&dtd); | |
| 605 | free((void *)atts); | |
| 606 | free(groupConnector); | |
| 607 | free(buffer); | |
| 608 | free(dataBuf); | |
| 609 | free(unknownEncodingMem); | |
| 610 | if (unknownEncodingRelease) | |
| 611 | unknownEncodingRelease(unknownEncodingData); | |
| 612 | free(parser); | |
| 613 | } | |
| 614 | ||
| 615 | void XML_UseParserAsHandlerArg(XML_Parser parser) | |
| 616 | { | |
| 617 | handlerArg = parser; | |
| 618 | } | |
| 619 | ||
| 620 | void XML_SetUserData(XML_Parser parser, void *p) | |
| 621 | { | |
| 622 | if (handlerArg == userData) | |
| 623 | handlerArg = userData = p; | |
| 624 | else | |
| 625 | userData = p; | |
| 626 | } | |
| 627 | ||
| 628 | int XML_SetBase(XML_Parser parser, const XML_Char *p) | |
| 629 | { | |
| 630 | if (p) { | |
| 631 | p = poolCopyString(&dtd.pool, p); | |
| 632 | if (!p) | |
| 633 | return 0; | |
| 634 | dtd.base = p; | |
| 635 | } | |
| 636 | else | |
| 637 | dtd.base = 0; | |
| 638 | return 1; | |
| 639 | } | |
| 640 | ||
| 641 | const XML_Char *XML_GetBase(XML_Parser parser) | |
| 642 | { | |
| 643 | return dtd.base; | |
| 644 | } | |
| 645 | ||
| 646 | int XML_GetSpecifiedAttributeCount(XML_Parser parser) | |
| 647 | { | |
| 648 | return nSpecifiedAtts; | |
| 649 | } | |
| 650 | ||
| 651 | void XML_SetElementHandler(XML_Parser parser, | |
| 652 | XML_StartElementHandler start, | |
| 653 | XML_EndElementHandler end) | |
| 654 | { | |
| 655 | startElementHandler = start; | |
| 656 | endElementHandler = end; | |
| 657 | } | |
| 658 | ||
| 659 | void XML_SetCharacterDataHandler(XML_Parser parser, | |
| 660 | XML_CharacterDataHandler handler) | |
| 661 | { | |
| 662 | characterDataHandler = handler; | |
| 663 | } | |
| 664 | ||
| 665 | void XML_SetProcessingInstructionHandler(XML_Parser parser, | |
| 666 | XML_ProcessingInstructionHandler handler) | |
| 667 | { | |
| 668 | processingInstructionHandler = handler; | |
| 669 | } | |
| 670 | ||
| 671 | void XML_SetCommentHandler(XML_Parser parser, | |
| 672 | XML_CommentHandler handler) | |
| 673 | { | |
| 674 | commentHandler = handler; | |
| 675 | } | |
| 676 | ||
| 677 | void XML_SetCdataSectionHandler(XML_Parser parser, | |
| 678 | XML_StartCdataSectionHandler start, | |
| 679 | XML_EndCdataSectionHandler end) | |
| 680 | { | |
| 681 | startCdataSectionHandler = start; | |
| 682 | endCdataSectionHandler = end; | |
| 683 | } | |
| 684 | ||
| 685 | void XML_SetDefaultHandler(XML_Parser parser, | |
| 686 | XML_DefaultHandler handler) | |
| 687 | { | |
| 688 | defaultHandler = handler; | |
| 689 | defaultExpandInternalEntities = 0; | |
| 690 | } | |
| 691 | ||
| 692 | void XML_SetDefaultHandlerExpand(XML_Parser parser, | |
| 693 | XML_DefaultHandler handler) | |
| 694 | { | |
| 695 | defaultHandler = handler; | |
| 696 | defaultExpandInternalEntities = 1; | |
| 697 | } | |
| 698 | ||
| 699 | void XML_SetUnparsedEntityDeclHandler(XML_Parser parser, | |
| 700 | XML_UnparsedEntityDeclHandler handler) | |
| 701 | { | |
| 702 | unparsedEntityDeclHandler = handler; | |
| 703 | } | |
| 704 | ||
| 705 | void XML_SetNotationDeclHandler(XML_Parser parser, | |
| 706 | XML_NotationDeclHandler handler) | |
| 707 | { | |
| 708 | notationDeclHandler = handler; | |
| 709 | } | |
| 710 | ||
| 711 | void XML_SetNamespaceDeclHandler(XML_Parser parser, | |
| 712 | XML_StartNamespaceDeclHandler start, | |
| 713 | XML_EndNamespaceDeclHandler end) | |
| 714 | { | |
| 715 | startNamespaceDeclHandler = start; | |
| 716 | endNamespaceDeclHandler = end; | |
| 717 | } | |
| 718 | ||
| 719 | void XML_SetNotStandaloneHandler(XML_Parser parser, | |
| 720 | XML_NotStandaloneHandler handler) | |
| 721 | { | |
| 722 | notStandaloneHandler = handler; | |
| 723 | } | |
| 724 | ||
| 725 | void XML_SetExternalEntityRefHandler(XML_Parser parser, | |
| 726 | XML_ExternalEntityRefHandler handler) | |
| 727 | { | |
| 728 | externalEntityRefHandler = handler; | |
| 729 | } | |
| 730 | ||
| 731 | void XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg) | |
| 732 | { | |
| 733 | if (arg) | |
| 734 | externalEntityRefHandlerArg = arg; | |
| 735 | else | |
| 736 | externalEntityRefHandlerArg = parser; | |
| 737 | } | |
| 738 | ||
| 739 | void XML_SetUnknownEncodingHandler(XML_Parser parser, | |
| 740 | XML_UnknownEncodingHandler handler, | |
| 741 | void *data) | |
| 742 | { | |
| 743 | unknownEncodingHandler = handler; | |
| 744 | unknownEncodingHandlerData = data; | |
| 745 | } | |
| 746 | ||
| 747 | int XML_Parse(XML_Parser parser, const char *s, int len, int isFinal) | |
| 748 | { | |
| 749 | if (len == 0) { | |
| 750 | if (!isFinal) | |
| 751 | return 1; | |
| 752 | positionPtr = bufferPtr; | |
| 753 | errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0); | |
| 754 | if (errorCode == XML_ERROR_NONE) | |
| 755 | return 1; | |
| 756 | eventEndPtr = eventPtr; | |
| 757 | return 0; | |
| 758 | } | |
| 759 | else if (bufferPtr == bufferEnd) { | |
| 760 | const char *end; | |
| 761 | int nLeftOver; | |
| 762 | parseEndByteIndex += len; | |
| 763 | positionPtr = s; | |
| 764 | if (isFinal) { | |
| 765 | errorCode = processor(parser, s, parseEndPtr = s + len, 0); | |
| 766 | if (errorCode == XML_ERROR_NONE) | |
| 767 | return 1; | |
| 768 | eventEndPtr = eventPtr; | |
| 769 | return 0; | |
| 770 | } | |
| 771 | errorCode = processor(parser, s, parseEndPtr = s + len, &end); | |
| 772 | if (errorCode != XML_ERROR_NONE) { | |
| 773 | eventEndPtr = eventPtr; | |
| 774 | return 0; | |
| 775 | } | |
| 776 | XmlUpdatePosition(encoding, positionPtr, end, &position); | |
| 777 | nLeftOver = s + len - end; | |
| 778 | if (nLeftOver) { | |
| 779 | if (buffer == 0 || nLeftOver > bufferLim - buffer) { | |
| 780 | /* FIXME avoid integer overflow */ | |
| 781 | buffer = buffer == 0 ? malloc(len * 2) : realloc(buffer, len * 2); | |
| 782 | if (!buffer) { | |
| 783 | errorCode = XML_ERROR_NO_MEMORY; | |
| 784 | eventPtr = eventEndPtr = 0; | |
| 785 | return 0; | |
| 786 | } | |
| 787 | bufferLim = buffer + len * 2; | |
| 788 | } | |
| 789 | memcpy(buffer, end, nLeftOver); | |
| 790 | bufferPtr = buffer; | |
| 791 | bufferEnd = buffer + nLeftOver; | |
| 792 | } | |
| 793 | return 1; | |
| 794 | } | |
| 795 | else { | |
| 796 | memcpy(XML_GetBuffer(parser, len), s, len); | |
| 797 | return XML_ParseBuffer(parser, len, isFinal); | |
| 798 | } | |
| 799 | } | |
| 800 | ||
| 801 | int XML_ParseBuffer(XML_Parser parser, int len, int isFinal) | |
| 802 | { | |
| 803 | const char *start = bufferPtr; | |
| 804 | positionPtr = start; | |
| 805 | bufferEnd += len; | |
| 806 | parseEndByteIndex += len; | |
| 807 | errorCode = processor(parser, start, parseEndPtr = bufferEnd, | |
| 808 | isFinal ? (const char **)0 : &bufferPtr); | |
| 809 | if (errorCode == XML_ERROR_NONE) { | |
| 810 | if (!isFinal) | |
| 811 | XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position); | |
| 812 | return 1; | |
| 813 | } | |
| 814 | else { | |
| 815 | eventEndPtr = eventPtr; | |
| 816 | return 0; | |
| 817 | } | |
| 818 | } | |
| 819 | ||
| 820 | void *XML_GetBuffer(XML_Parser parser, int len) | |
| 821 | { | |
| 822 | if (len > bufferLim - bufferEnd) { | |
| 823 | /* FIXME avoid integer overflow */ | |
| 824 | int neededSize = len + (bufferEnd - bufferPtr); | |
| 825 | if (neededSize <= bufferLim - buffer) { | |
| 826 | memmove(buffer, bufferPtr, bufferEnd - bufferPtr); | |
| 827 | bufferEnd = buffer + (bufferEnd - bufferPtr); | |
| 828 | bufferPtr = buffer; | |
| 829 | } | |
| 830 | else { | |
| 831 | char *newBuf; | |
| 832 | int bufferSize = bufferLim - bufferPtr; | |
| 833 | if (bufferSize == 0) | |
| 834 | bufferSize = INIT_BUFFER_SIZE; | |
| 835 | do { | |
| 836 | bufferSize *= 2; | |
| 837 | } while (bufferSize < neededSize); | |
| 838 | newBuf = malloc(bufferSize); | |
| 839 | if (newBuf == 0) { | |
| 840 | errorCode = XML_ERROR_NO_MEMORY; | |
| 841 | return 0; | |
| 842 | } | |
| 843 | bufferLim = newBuf + bufferSize; | |
| 844 | if (bufferPtr) { | |
| 845 | memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr); | |
| 846 | free(buffer); | |
| 847 | } | |
| 848 | bufferEnd = newBuf + (bufferEnd - bufferPtr); | |
| 849 | bufferPtr = buffer = newBuf; | |
| 850 | } | |
| 851 | } | |
| 852 | return bufferEnd; | |
| 853 | } | |
| 854 | ||
| 855 | enum XML_Error XML_GetErrorCode(XML_Parser parser) | |
| 856 | { | |
| 857 | return errorCode; | |
| 858 | } | |
| 859 | ||
| 860 | long XML_GetCurrentByteIndex(XML_Parser parser) | |
| 861 | { | |
| 862 | if (eventPtr) | |
| 863 | return parseEndByteIndex - (parseEndPtr - eventPtr); | |
| 864 | return -1; | |
| 865 | } | |
| 866 | ||
| 867 | int XML_GetCurrentByteCount(XML_Parser parser) | |
| 868 | { | |
| 869 | if (eventEndPtr && eventPtr) | |
| 870 | return eventEndPtr - eventPtr; | |
| 871 | return 0; | |
| 872 | } | |
| 873 | ||
| 874 | int XML_GetCurrentLineNumber(XML_Parser parser) | |
| 875 | { | |
| 876 | if (eventPtr) { | |
| 877 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
| 878 | positionPtr = eventPtr; | |
| 879 | } | |
| 880 | return position.lineNumber + 1; | |
| 881 | } | |
| 882 | ||
| 883 | int XML_GetCurrentColumnNumber(XML_Parser parser) | |
| 884 | { | |
| 885 | if (eventPtr) { | |
| 886 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position); | |
| 887 | positionPtr = eventPtr; | |
| 888 | } | |
| 889 | return position.columnNumber; | |
| 890 | } | |
| 891 | ||
| 892 | void XML_DefaultCurrent(XML_Parser parser) | |
| 893 | { | |
| 894 | if (defaultHandler) { | |
| 895 | if (openInternalEntities) | |
| 896 | reportDefault(parser, | |
| 897 | ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(), | |
| 898 | openInternalEntities->internalEventPtr, | |
| 899 | openInternalEntities->internalEventEndPtr); | |
| 900 | else | |
| 901 | reportDefault(parser, encoding, eventPtr, eventEndPtr); | |
| 902 | } | |
| 903 | } | |
| 904 | ||
| 905 | const XML_LChar *XML_ErrorString(int code) | |
| 906 | { | |
| 907 | static const XML_LChar *message[] = { | |
| 908 | 0, | |
| 909 | XML_T("out of memory"), | |
| 910 | XML_T("syntax error"), | |
| 911 | XML_T("no element found"), | |
| 912 | XML_T("not well-formed"), | |
| 913 | XML_T("unclosed token"), | |
| 914 | XML_T("unclosed token"), | |
| 915 | XML_T("mismatched tag"), | |
| 916 | XML_T("duplicate attribute"), | |
| 917 | XML_T("junk after document element"), | |
| 918 | XML_T("illegal parameter entity reference"), | |
| 919 | XML_T("undefined entity"), | |
| 920 | XML_T("recursive entity reference"), | |
| 921 | XML_T("asynchronous entity"), | |
| 922 | XML_T("reference to invalid character number"), | |
| 923 | XML_T("reference to binary entity"), | |
| 924 | XML_T("reference to external entity in attribute"), | |
| 925 | XML_T("xml processing instruction not at start of external entity"), | |
| 926 | XML_T("unknown encoding"), | |
| 927 | XML_T("encoding specified in XML declaration is incorrect"), | |
| 928 | XML_T("unclosed CDATA section"), | |
| 929 | XML_T("error in processing external entity reference"), | |
| 930 | XML_T("document is not standalone") | |
| 931 | }; | |
| 932 | if (code > 0 && code < sizeof(message)/sizeof(message[0])) | |
| 933 | return message[code]; | |
| 934 | return 0; | |
| 935 | } | |
| 936 | ||
| 937 | static | |
| 938 | enum XML_Error contentProcessor(XML_Parser parser, | |
| 939 | const char *start, | |
| 940 | const char *end, | |
| 941 | const char **endPtr) | |
| 942 | { | |
| 943 | return doContent(parser, 0, encoding, start, end, endPtr); | |
| 944 | } | |
| 945 | ||
| 946 | static | |
| 947 | enum XML_Error externalEntityInitProcessor(XML_Parser parser, | |
| 948 | const char *start, | |
| 949 | const char *end, | |
| 950 | const char **endPtr) | |
| 951 | { | |
| 952 | enum XML_Error result = initializeEncoding(parser); | |
| 953 | if (result != XML_ERROR_NONE) | |
| 954 | return result; | |
| 955 | processor = externalEntityInitProcessor2; | |
| 956 | return externalEntityInitProcessor2(parser, start, end, endPtr); | |
| 957 | } | |
| 958 | ||
| 959 | static | |
| 960 | enum XML_Error externalEntityInitProcessor2(XML_Parser parser, | |
| 961 | const char *start, | |
| 962 | const char *end, | |
| 963 | const char **endPtr) | |
| 964 | { | |
| 965 | const char *next; | |
| 966 | int tok = XmlContentTok(encoding, start, end, &next); | |
| 967 | switch (tok) { | |
| 968 | case XML_TOK_BOM: | |
| 969 | start = next; | |
| 970 | break; | |
| 971 | case XML_TOK_PARTIAL: | |
| 972 | if (endPtr) { | |
| 973 | *endPtr = start; | |
| 974 | return XML_ERROR_NONE; | |
| 975 | } | |
| 976 | eventPtr = start; | |
| 977 | return XML_ERROR_UNCLOSED_TOKEN; | |
| 978 | case XML_TOK_PARTIAL_CHAR: | |
| 979 | if (endPtr) { | |
| 980 | *endPtr = start; | |
| 981 | return XML_ERROR_NONE; | |
| 982 | } | |
| 983 | eventPtr = start; | |
| 984 | return XML_ERROR_PARTIAL_CHAR; | |
| 985 | } | |
| 986 | processor = externalEntityInitProcessor3; | |
| 987 | return externalEntityInitProcessor3(parser, start, end, endPtr); | |
| 988 | } | |
| 989 | ||
| 990 | static | |
| 991 | enum XML_Error externalEntityInitProcessor3(XML_Parser parser, | |
| 992 | const char *start, | |
| 993 | const char *end, | |
| 994 | const char **endPtr) | |
| 995 | { | |
| 996 | const char *next; | |
| 997 | int tok = XmlContentTok(encoding, start, end, &next); | |
| 998 | switch (tok) { | |
| 999 | case XML_TOK_XML_DECL: | |
| 1000 | { | |
| 1001 | enum XML_Error result = processXmlDecl(parser, 1, start, next); | |
| 1002 | if (result != XML_ERROR_NONE) | |
| 1003 | return result; | |
| 1004 | start = next; | |
| 1005 | } | |
| 1006 | break; | |
| 1007 | case XML_TOK_PARTIAL: | |
| 1008 | if (endPtr) { | |
| 1009 | *endPtr = start; | |
| 1010 | return XML_ERROR_NONE; | |
| 1011 | } | |
| 1012 | eventPtr = start; | |
| 1013 | return XML_ERROR_UNCLOSED_TOKEN; | |
| 1014 | case XML_TOK_PARTIAL_CHAR: | |
| 1015 | if (endPtr) { | |
| 1016 | *endPtr = start; | |
| 1017 | return XML_ERROR_NONE; | |
| 1018 | } | |
| 1019 | eventPtr = start; | |
| 1020 | return XML_ERROR_PARTIAL_CHAR; | |
| 1021 | } | |
| 1022 | processor = externalEntityContentProcessor; | |
| 1023 | tagLevel = 1; | |
| 1024 | return doContent(parser, 1, encoding, start, end, endPtr); | |
| 1025 | } | |
| 1026 | ||
| 1027 | static | |
| 1028 | enum XML_Error externalEntityContentProcessor(XML_Parser parser, | |
| 1029 | const char *start, | |
| 1030 | const char *end, | |
| 1031 | const char **endPtr) | |
| 1032 | { | |
| 1033 | return doContent(parser, 1, encoding, start, end, endPtr); | |
| 1034 | } | |
| 1035 | ||
| 1036 | static enum XML_Error | |
| 1037 | doContent(XML_Parser parser, | |
| 1038 | int startTagLevel, | |
| 1039 | const ENCODING *enc, | |
| 1040 | const char *s, | |
| 1041 | const char *end, | |
| 1042 | const char **nextPtr) | |
| 1043 | { | |
| 1044 | const ENCODING *internalEnc = ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(); | |
| 1045 | const char **eventPP; | |
| 1046 | const char **eventEndPP; | |
| 1047 | if (enc == encoding) { | |
| 1048 | eventPP = &eventPtr; | |
| 1049 | eventEndPP = &eventEndPtr; | |
| 1050 | } | |
| 1051 | else { | |
| 1052 | eventPP = &(openInternalEntities->internalEventPtr); | |
| 1053 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
| 1054 | } | |
| 1055 | *eventPP = s; | |
| 1056 | for (;;) { | |
| 1057 | const char *next = s; /* XmlContentTok doesn't always set the last arg */ | |
| 1058 | int tok = XmlContentTok(enc, s, end, &next); | |
| 1059 | *eventEndPP = next; | |
| 1060 | switch (tok) { | |
| 1061 | case XML_TOK_TRAILING_CR: | |
| 1062 | if (nextPtr) { | |
| 1063 | *nextPtr = s; | |
| 1064 | return XML_ERROR_NONE; | |
| 1065 | } | |
| 1066 | *eventEndPP = end; | |
| 1067 | if (characterDataHandler) { | |
| 1068 | XML_Char c = 0xA; | |
| 1069 | characterDataHandler(handlerArg, &c, 1); | |
| 1070 | } | |
| 1071 | else if (defaultHandler) | |
| 1072 | reportDefault(parser, enc, s, end); | |
| 1073 | if (startTagLevel == 0) | |
| 1074 | return XML_ERROR_NO_ELEMENTS; | |
| 1075 | if (tagLevel != startTagLevel) | |
| 1076 | return XML_ERROR_ASYNC_ENTITY; | |
| 1077 | return XML_ERROR_NONE; | |
| 1078 | case XML_TOK_NONE: | |
| 1079 | if (nextPtr) { | |
| 1080 | *nextPtr = s; | |
| 1081 | return XML_ERROR_NONE; | |
| 1082 | } | |
| 1083 | if (startTagLevel > 0) { | |
| 1084 | if (tagLevel != startTagLevel) | |
| 1085 | return XML_ERROR_ASYNC_ENTITY; | |
| 1086 | return XML_ERROR_NONE; | |
| 1087 | } | |
| 1088 | return XML_ERROR_NO_ELEMENTS; | |
| 1089 | case XML_TOK_INVALID: | |
| 1090 | *eventPP = next; | |
| 1091 | return XML_ERROR_INVALID_TOKEN; | |
| 1092 | case XML_TOK_PARTIAL: | |
| 1093 | if (nextPtr) { | |
| 1094 | *nextPtr = s; | |
| 1095 | return XML_ERROR_NONE; | |
| 1096 | } | |
| 1097 | return XML_ERROR_UNCLOSED_TOKEN; | |
| 1098 | case XML_TOK_PARTIAL_CHAR: | |
| 1099 | if (nextPtr) { | |
| 1100 | *nextPtr = s; | |
| 1101 | return XML_ERROR_NONE; | |
| 1102 | } | |
| 1103 | return XML_ERROR_PARTIAL_CHAR; | |
| 1104 | case XML_TOK_ENTITY_REF: | |
| 1105 | { | |
| 1106 | const XML_Char *name; | |
| 1107 | ENTITY *entity; | |
| 1108 | XML_Char ch = XmlPredefinedEntityName(enc, | |
| 1109 | s + enc->minBytesPerChar, | |
| 1110 | next - enc->minBytesPerChar); | |
| 1111 | if (ch) { | |
| 1112 | if (characterDataHandler) | |
| 1113 | characterDataHandler(handlerArg, &ch, 1); | |
| 1114 | else if (defaultHandler) | |
| 1115 | reportDefault(parser, enc, s, next); | |
| 1116 | break; | |
| 1117 | } | |
| 1118 | name = poolStoreString(&dtd.pool, enc, | |
| 1119 | s + enc->minBytesPerChar, | |
| 1120 | next - enc->minBytesPerChar); | |
| 1121 | if (!name) | |
| 1122 | return XML_ERROR_NO_MEMORY; | |
| 1123 | entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); | |
| 1124 | poolDiscard(&dtd.pool); | |
| 1125 | if (!entity) { | |
| 1126 | if (dtd.complete || dtd.standalone) | |
| 1127 | return XML_ERROR_UNDEFINED_ENTITY; | |
| 1128 | if (defaultHandler) | |
| 1129 | reportDefault(parser, enc, s, next); | |
| 1130 | break; | |
| 1131 | } | |
| 1132 | if (entity->open) | |
| 1133 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
| 1134 | if (entity->notation) | |
| 1135 | return XML_ERROR_BINARY_ENTITY_REF; | |
| 1136 | if (entity) { | |
| 1137 | if (entity->textPtr) { | |
| 1138 | enum XML_Error result; | |
| 1139 | OPEN_INTERNAL_ENTITY openEntity; | |
| 1140 | if (defaultHandler && !defaultExpandInternalEntities) { | |
| 1141 | reportDefault(parser, enc, s, next); | |
| 1142 | break; | |
| 1143 | } | |
| 1144 | entity->open = 1; | |
| 1145 | openEntity.next = openInternalEntities; | |
| 1146 | openInternalEntities = &openEntity; | |
| 1147 | openEntity.entity = entity; | |
| 1148 | openEntity.internalEventPtr = 0; | |
| 1149 | openEntity.internalEventEndPtr = 0; | |
| 1150 | result = doContent(parser, | |
| 1151 | tagLevel, | |
| 1152 | internalEnc, | |
| 1153 | (char *)entity->textPtr, | |
| 1154 | (char *)(entity->textPtr + entity->textLen), | |
| 1155 | 0); | |
| 1156 | entity->open = 0; | |
| 1157 | openInternalEntities = openEntity.next; | |
| 1158 | if (result) | |
| 1159 | return result; | |
| 1160 | } | |
| 1161 | else if (externalEntityRefHandler) { | |
| 1162 | const XML_Char *context; | |
| 1163 | entity->open = 1; | |
| 1164 | context = getContext(parser); | |
| 1165 | entity->open = 0; | |
| 1166 | if (!context) | |
| 1167 | return XML_ERROR_NO_MEMORY; | |
| 1168 | if (!externalEntityRefHandler(externalEntityRefHandlerArg, | |
| 1169 | context, | |
| 1170 | dtd.base, | |
| 1171 | entity->systemId, | |
| 1172 | entity->publicId)) | |
| 1173 | return XML_ERROR_EXTERNAL_ENTITY_HANDLING; | |
| 1174 | poolDiscard(&tempPool); | |
| 1175 | } | |
| 1176 | else if (defaultHandler) | |
| 1177 | reportDefault(parser, enc, s, next); | |
| 1178 | } | |
| 1179 | break; | |
| 1180 | } | |
| 1181 | case XML_TOK_START_TAG_WITH_ATTS: | |
| 1182 | if (!startElementHandler) { | |
| 1183 | enum XML_Error result = storeAtts(parser, enc, s, 0, 0); | |
| 1184 | if (result) | |
| 1185 | return result; | |
| 1186 | } | |
| 1187 | /* fall through */ | |
| 1188 | case XML_TOK_START_TAG_NO_ATTS: | |
| 1189 | { | |
| 1190 | TAG *tag; | |
| 1191 | if (freeTagList) { | |
| 1192 | tag = freeTagList; | |
| 1193 | freeTagList = freeTagList->parent; | |
| 1194 | } | |
| 1195 | else { | |
| 1196 | tag = malloc(sizeof(TAG)); | |
| 1197 | if (!tag) | |
| 1198 | return XML_ERROR_NO_MEMORY; | |
| 1199 | tag->buf = malloc(INIT_TAG_BUF_SIZE); | |
| 1200 | if (!tag->buf) | |
| 1201 | return XML_ERROR_NO_MEMORY; | |
| 1202 | tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE; | |
| 1203 | } | |
| 1204 | tag->bindings = 0; | |
| 1205 | tag->parent = tagStack; | |
| 1206 | tagStack = tag; | |
| 1207 | tag->name.localPart = 0; | |
| 1208 | tag->rawName = s + enc->minBytesPerChar; | |
| 1209 | tag->rawNameLength = XmlNameLength(enc, tag->rawName); | |
| 1210 | if (nextPtr) { | |
| 1211 | /* Need to guarantee that: | |
| 1212 | tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)) <= tag->bufEnd - sizeof(XML_Char) */ | |
| 1213 | if (tag->rawNameLength + (int)(sizeof(XML_Char) - 1) + (int)sizeof(XML_Char) > tag->bufEnd - tag->buf) { | |
| 1214 | int bufSize = tag->rawNameLength * 4; | |
| 1215 | bufSize = ROUND_UP(bufSize, sizeof(XML_Char)); | |
| 1216 | tag->buf = realloc(tag->buf, bufSize); | |
| 1217 | if (!tag->buf) | |
| 1218 | return XML_ERROR_NO_MEMORY; | |
| 1219 | tag->bufEnd = tag->buf + bufSize; | |
| 1220 | } | |
| 1221 | memcpy(tag->buf, tag->rawName, tag->rawNameLength); | |
| 1222 | tag->rawName = tag->buf; | |
| 1223 | } | |
| 1224 | ++tagLevel; | |
| 1225 | if (startElementHandler) { | |
| 1226 | enum XML_Error result; | |
| 1227 | XML_Char *toPtr; | |
| 1228 | for (;;) { | |
| 1229 | const char *rawNameEnd = tag->rawName + tag->rawNameLength; | |
| 1230 | const char *fromPtr = tag->rawName; | |
| 1231 | int bufSize; | |
| 1232 | if (nextPtr) | |
| 1233 | toPtr = (XML_Char *)(tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char))); | |
| 1234 | else | |
| 1235 | toPtr = (XML_Char *)tag->buf; | |
| 1236 | tag->name.str = toPtr; | |
| 1237 | XmlConvert(enc, | |
| 1238 | &fromPtr, rawNameEnd, | |
| 1239 | (ICHAR **)&toPtr, (ICHAR *)tag->bufEnd - 1); | |
| 1240 | if (fromPtr == rawNameEnd) | |
| 1241 | break; | |
| 1242 | bufSize = (tag->bufEnd - tag->buf) << 1; | |
| 1243 | tag->buf = realloc(tag->buf, bufSize); | |
| 1244 | if (!tag->buf) | |
| 1245 | return XML_ERROR_NO_MEMORY; | |
| 1246 | tag->bufEnd = tag->buf + bufSize; | |
| 1247 | if (nextPtr) | |
| 1248 | tag->rawName = tag->buf; | |
| 1249 | } | |
| 1250 | *toPtr = XML_T('\0'); | |
| 1251 | result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings)); | |
| 1252 | if (result) | |
| 1253 | return result; | |
| 1254 | startElementHandler(handlerArg, tag->name.str, (const XML_Char **)atts); | |
| 1255 | poolClear(&tempPool); | |
| 1256 | } | |
| 1257 | else { | |
| 1258 | tag->name.str = 0; | |
| 1259 | if (defaultHandler) | |
| 1260 | reportDefault(parser, enc, s, next); | |
| 1261 | } | |
| 1262 | break; | |
| 1263 | } | |
| 1264 | case XML_TOK_EMPTY_ELEMENT_WITH_ATTS: | |
| 1265 | if (!startElementHandler) { | |
| 1266 | enum XML_Error result = storeAtts(parser, enc, s, 0, 0); | |
| 1267 | if (result) | |
| 1268 | return result; | |
| 1269 | } | |
| 1270 | /* fall through */ | |
| 1271 | case XML_TOK_EMPTY_ELEMENT_NO_ATTS: | |
| 1272 | if (startElementHandler || endElementHandler) { | |
| 1273 | const char *rawName = s + enc->minBytesPerChar; | |
| 1274 | enum XML_Error result; | |
| 1275 | BINDING *bindings = 0; | |
| 1276 | TAG_NAME name; | |
| 1277 | name.str = poolStoreString(&tempPool, enc, rawName, | |
| 1278 | rawName + XmlNameLength(enc, rawName)); | |
| 1279 | if (!name.str) | |
| 1280 | return XML_ERROR_NO_MEMORY; | |
| 1281 | poolFinish(&tempPool); | |
| 1282 | result = storeAtts(parser, enc, s, &name, &bindings); | |
| 1283 | if (result) | |
| 1284 | return result; | |
| 1285 | poolFinish(&tempPool); | |
| 1286 | if (startElementHandler) | |
| 1287 | startElementHandler(handlerArg, name.str, (const XML_Char **)atts); | |
| 1288 | if (endElementHandler) { | |
| 1289 | if (startElementHandler) | |
| 1290 | *eventPP = *eventEndPP; | |
| 1291 | endElementHandler(handlerArg, name.str); | |
| 1292 | } | |
| 1293 | poolClear(&tempPool); | |
| 1294 | while (bindings) { | |
| 1295 | BINDING *b = bindings; | |
| 1296 | if (endNamespaceDeclHandler) | |
| 1297 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
| 1298 | bindings = bindings->nextTagBinding; | |
| 1299 | b->nextTagBinding = freeBindingList; | |
| 1300 | freeBindingList = b; | |
| 1301 | b->prefix->binding = b->prevPrefixBinding; | |
| 1302 | } | |
| 1303 | } | |
| 1304 | else if (defaultHandler) | |
| 1305 | reportDefault(parser, enc, s, next); | |
| 1306 | if (tagLevel == 0) | |
| 1307 | return epilogProcessor(parser, next, end, nextPtr); | |
| 1308 | break; | |
| 1309 | case XML_TOK_END_TAG: | |
| 1310 | if (tagLevel == startTagLevel) | |
| 1311 | return XML_ERROR_ASYNC_ENTITY; | |
| 1312 | else { | |
| 1313 | int len; | |
| 1314 | const char *rawName; | |
| 1315 | TAG *tag = tagStack; | |
| 1316 | tagStack = tag->parent; | |
| 1317 | tag->parent = freeTagList; | |
| 1318 | freeTagList = tag; | |
| 1319 | rawName = s + enc->minBytesPerChar*2; | |
| 1320 | len = XmlNameLength(enc, rawName); | |
| 1321 | if (len != tag->rawNameLength | |
| 1322 | || memcmp(tag->rawName, rawName, len) != 0) { | |
| 1323 | *eventPP = rawName; | |
| 1324 | return XML_ERROR_TAG_MISMATCH; | |
| 1325 | } | |
| 1326 | --tagLevel; | |
| 1327 | if (endElementHandler && tag->name.str) { | |
| 1328 | if (tag->name.localPart) { | |
| 1329 | XML_Char *to = (XML_Char *)tag->name.str + tag->name.uriLen; | |
| 1330 | const XML_Char *from = tag->name.localPart; | |
| 1331 | while ((*to++ = *from++) != 0) | |
| 1332 | ; | |
| 1333 | } | |
| 1334 | endElementHandler(handlerArg, tag->name.str); | |
| 1335 | } | |
| 1336 | else if (defaultHandler) | |
| 1337 | reportDefault(parser, enc, s, next); | |
| 1338 | while (tag->bindings) { | |
| 1339 | BINDING *b = tag->bindings; | |
| 1340 | if (endNamespaceDeclHandler) | |
| 1341 | endNamespaceDeclHandler(handlerArg, b->prefix->name); | |
| 1342 | tag->bindings = tag->bindings->nextTagBinding; | |
| 1343 | b->nextTagBinding = freeBindingList; | |
| 1344 | freeBindingList = b; | |
| 1345 | b->prefix->binding = b->prevPrefixBinding; | |
| 1346 | } | |
| 1347 | if (tagLevel == 0) | |
| 1348 | return epilogProcessor(parser, next, end, nextPtr); | |
| 1349 | } | |
| 1350 | break; | |
| 1351 | case XML_TOK_CHAR_REF: | |
| 1352 | { | |
| 1353 | int n = XmlCharRefNumber(enc, s); | |
| 1354 | if (n < 0) | |
| 1355 | return XML_ERROR_BAD_CHAR_REF; | |
| 1356 | if (characterDataHandler) { | |
| 1357 | XML_Char buf[XML_ENCODE_MAX]; | |
| 1358 | characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *)buf)); | |
| 1359 | } | |
| 1360 | else if (defaultHandler) | |
| 1361 | reportDefault(parser, enc, s, next); | |
| 1362 | } | |
| 1363 | break; | |
| 1364 | case XML_TOK_XML_DECL: | |
| 1365 | return XML_ERROR_MISPLACED_XML_PI; | |
| 1366 | case XML_TOK_DATA_NEWLINE: | |
| 1367 | if (characterDataHandler) { | |
| 1368 | XML_Char c = 0xA; | |
| 1369 | characterDataHandler(handlerArg, &c, 1); | |
| 1370 | } | |
| 1371 | else if (defaultHandler) | |
| 1372 | reportDefault(parser, enc, s, next); | |
| 1373 | break; | |
| 1374 | case XML_TOK_CDATA_SECT_OPEN: | |
| 1375 | { | |
| 1376 | enum XML_Error result; | |
| 1377 | if (startCdataSectionHandler) | |
| 1378 | startCdataSectionHandler(handlerArg); | |
| 1379 | #if 0 | |
| 1380 | /* Suppose you doing a transformation on a document that involves | |
| 1381 | changing only the character data. You set up a defaultHandler | |
| 1382 | and a characterDataHandler. The defaultHandler simply copies | |
| 1383 | characters through. The characterDataHandler does the transformation | |
| 1384 | and writes the characters out escaping them as necessary. This case | |
| 1385 | will fail to work if we leave out the following two lines (because & | |
| 1386 | and < inside CDATA sections will be incorrectly escaped). | |
| 1387 | ||
| 1388 | However, now we have a start/endCdataSectionHandler, so it seems | |
| 1389 | easier to let the user deal with this. */ | |
| 1390 | ||
| 1391 | else if (characterDataHandler) | |
| 1392 | characterDataHandler(handlerArg, dataBuf, 0); | |
| 1393 | #endif | |
| 1394 | else if (defaultHandler) | |
| 1395 | reportDefault(parser, enc, s, next); | |
| 1396 | result = doCdataSection(parser, enc, &next, end, nextPtr); | |
| 1397 | if (!next) { | |
| 1398 | processor = cdataSectionProcessor; | |
| 1399 | return result; | |
| 1400 | } | |
| 1401 | } | |
| 1402 | break; | |
| 1403 | case XML_TOK_TRAILING_RSQB: | |
| 1404 | if (nextPtr) { | |
| 1405 | *nextPtr = s; | |
| 1406 | return XML_ERROR_NONE; | |
| 1407 | } | |
| 1408 | if (characterDataHandler) { | |
| 1409 | if (MUST_CONVERT(enc, s)) { | |
| 1410 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
| 1411 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
| 1412 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
| 1413 | } | |
| 1414 | else | |
| 1415 | characterDataHandler(handlerArg, | |
| 1416 | (XML_Char *)s, | |
| 1417 | (XML_Char *)end - (XML_Char *)s); | |
| 1418 | } | |
| 1419 | else if (defaultHandler) | |
| 1420 | reportDefault(parser, enc, s, end); | |
| 1421 | if (startTagLevel == 0) { | |
| 1422 | *eventPP = end; | |
| 1423 | return XML_ERROR_NO_ELEMENTS; | |
| 1424 | } | |
| 1425 | if (tagLevel != startTagLevel) { | |
| 1426 | *eventPP = end; | |
| 1427 | return XML_ERROR_ASYNC_ENTITY; | |
| 1428 | } | |
| 1429 | return XML_ERROR_NONE; | |
| 1430 | case XML_TOK_DATA_CHARS: | |
| 1431 | if (characterDataHandler) { | |
| 1432 | if (MUST_CONVERT(enc, s)) { | |
| 1433 | for (;;) { | |
| 1434 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
| 1435 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
| 1436 | *eventEndPP = s; | |
| 1437 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
| 1438 | if (s == next) | |
| 1439 | break; | |
| 1440 | *eventPP = s; | |
| 1441 | } | |
| 1442 | } | |
| 1443 | else | |
| 1444 | characterDataHandler(handlerArg, | |
| 1445 | (XML_Char *)s, | |
| 1446 | (XML_Char *)next - (XML_Char *)s); | |
| 1447 | } | |
| 1448 | else if (defaultHandler) | |
| 1449 | reportDefault(parser, enc, s, next); | |
| 1450 | break; | |
| 1451 | case XML_TOK_PI: | |
| 1452 | if (!reportProcessingInstruction(parser, enc, s, next)) | |
| 1453 | return XML_ERROR_NO_MEMORY; | |
| 1454 | break; | |
| 1455 | case XML_TOK_COMMENT: | |
| 1456 | if (!reportComment(parser, enc, s, next)) | |
| 1457 | return XML_ERROR_NO_MEMORY; | |
| 1458 | break; | |
| 1459 | default: | |
| 1460 | if (defaultHandler) | |
| 1461 | reportDefault(parser, enc, s, next); | |
| 1462 | break; | |
| 1463 | } | |
| 1464 | *eventPP = s = next; | |
| 1465 | } | |
| 1466 | /* not reached */ | |
| 1467 | } | |
| 1468 | ||
| 1469 | /* If tagNamePtr is non-null, build a real list of attributes, | |
| 1470 | otherwise just check the attributes for well-formedness. */ | |
| 1471 | ||
| 1472 | static enum XML_Error storeAtts(XML_Parser parser, const ENCODING *enc, | |
| 1473 | const char *s, TAG_NAME *tagNamePtr, | |
| 1474 | BINDING **bindingsPtr) | |
| 1475 | { | |
| 1476 | ELEMENT_TYPE *elementType = 0; | |
| 1477 | int nDefaultAtts = 0; | |
| 1478 | const XML_Char **appAtts; | |
| 1479 | int attIndex = 0; | |
| 1480 | int i; | |
| 1481 | int n; | |
| 1482 | int nPrefixes = 0; | |
| 1483 | BINDING *binding; | |
| 1484 | const XML_Char *localPart; | |
| 1485 | ||
| 1486 | if (tagNamePtr) { | |
| 1487 | elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, 0); | |
| 1488 | if (!elementType) { | |
| 1489 | tagNamePtr->str = poolCopyString(&dtd.pool, tagNamePtr->str); | |
| 1490 | if (!tagNamePtr->str) | |
| 1491 | return XML_ERROR_NO_MEMORY; | |
| 1492 | elementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, tagNamePtr->str, sizeof(ELEMENT_TYPE)); | |
| 1493 | if (!elementType) | |
| 1494 | return XML_ERROR_NO_MEMORY; | |
| 1495 | if (ns && !setElementTypePrefix(parser, elementType)) | |
| 1496 | return XML_ERROR_NO_MEMORY; | |
| 1497 | } | |
| 1498 | nDefaultAtts = elementType->nDefaultAtts; | |
| 1499 | } | |
| 1500 | n = XmlGetAttributes(enc, s, attsSize, atts); | |
| 1501 | if (n + nDefaultAtts > attsSize) { | |
| 1502 | int oldAttsSize = attsSize; | |
| 1503 | attsSize = n + nDefaultAtts + INIT_ATTS_SIZE; | |
| 1504 | atts = realloc((void *)atts, attsSize * sizeof(ATTRIBUTE)); | |
| 1505 | if (!atts) | |
| 1506 | return XML_ERROR_NO_MEMORY; | |
| 1507 | if (n > oldAttsSize) | |
| 1508 | XmlGetAttributes(enc, s, n, atts); | |
| 1509 | } | |
| 1510 | appAtts = (const XML_Char **)atts; | |
| 1511 | for (i = 0; i < n; i++) { | |
| 1512 | ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name, | |
| 1513 | atts[i].name | |
| 1514 | + XmlNameLength(enc, atts[i].name)); | |
| 1515 | if (!attId) | |
| 1516 | return XML_ERROR_NO_MEMORY; | |
| 1517 | if ((attId->name)[-1]) { | |
| 1518 | if (enc == encoding) | |
| 1519 | eventPtr = atts[i].name; | |
| 1520 | return XML_ERROR_DUPLICATE_ATTRIBUTE; | |
| 1521 | } | |
| 1522 | (attId->name)[-1] = 1; | |
| 1523 | appAtts[attIndex++] = attId->name; | |
| 1524 | if (!atts[i].normalized) { | |
| 1525 | enum XML_Error result; | |
| 1526 | int isCdata = 1; | |
| 1527 | ||
| 1528 | if (attId->maybeTokenized) { | |
| 1529 | int j; | |
| 1530 | for (j = 0; j < nDefaultAtts; j++) { | |
| 1531 | if (attId == elementType->defaultAtts[j].id) { | |
| 1532 | isCdata = elementType->defaultAtts[j].isCdata; | |
| 1533 | break; | |
| 1534 | } | |
| 1535 | } | |
| 1536 | } | |
| 1537 | ||
| 1538 | result = storeAttributeValue(parser, enc, isCdata, | |
| 1539 | atts[i].valuePtr, atts[i].valueEnd, | |
| 1540 | &tempPool); | |
| 1541 | if (result) | |
| 1542 | return result; | |
| 1543 | if (tagNamePtr) { | |
| 1544 | appAtts[attIndex] = poolStart(&tempPool); | |
| 1545 | poolFinish(&tempPool); | |
| 1546 | } | |
| 1547 | else | |
| 1548 | poolDiscard(&tempPool); | |
| 1549 | } | |
| 1550 | else if (tagNamePtr) { | |
| 1551 | appAtts[attIndex] = poolStoreString(&tempPool, enc, atts[i].valuePtr, atts[i].valueEnd); | |
| 1552 | if (appAtts[attIndex] == 0) | |
| 1553 | return XML_ERROR_NO_MEMORY; | |
| 1554 | poolFinish(&tempPool); | |
| 1555 | } | |
| 1556 | if (attId->prefix && tagNamePtr) { | |
| 1557 | if (attId->xmlns) { | |
| 1558 | if (!addBinding(parser, attId->prefix, attId, appAtts[attIndex], bindingsPtr)) | |
| 1559 | return XML_ERROR_NO_MEMORY; | |
| 1560 | --attIndex; | |
| 1561 | } | |
| 1562 | else { | |
| 1563 | attIndex++; | |
| 1564 | nPrefixes++; | |
| 1565 | (attId->name)[-1] = 2; | |
| 1566 | } | |
| 1567 | } | |
| 1568 | else | |
| 1569 | attIndex++; | |
| 1570 | } | |
| 1571 | nSpecifiedAtts = attIndex; | |
| 1572 | if (tagNamePtr) { | |
| 1573 | int j; | |
| 1574 | for (j = 0; j < nDefaultAtts; j++) { | |
| 1575 | const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + j; | |
| 1576 | if (!(da->id->name)[-1] && da->value) { | |
| 1577 | if (da->id->prefix) { | |
| 1578 | if (da->id->xmlns) { | |
| 1579 | if (!addBinding(parser, da->id->prefix, da->id, da->value, bindingsPtr)) | |
| 1580 | return XML_ERROR_NO_MEMORY; | |
| 1581 | } | |
| 1582 | else { | |
| 1583 | (da->id->name)[-1] = 2; | |
| 1584 | nPrefixes++; | |
| 1585 | appAtts[attIndex++] = da->id->name; | |
| 1586 | appAtts[attIndex++] = da->value; | |
| 1587 | } | |
| 1588 | } | |
| 1589 | else { | |
| 1590 | (da->id->name)[-1] = 1; | |
| 1591 | appAtts[attIndex++] = da->id->name; | |
| 1592 | appAtts[attIndex++] = da->value; | |
| 1593 | } | |
| 1594 | } | |
| 1595 | } | |
| 1596 | appAtts[attIndex] = 0; | |
| 1597 | } | |
| 1598 | i = 0; | |
| 1599 | if (nPrefixes) { | |
| 1600 | for (; i < attIndex; i += 2) { | |
| 1601 | if (appAtts[i][-1] == 2) { | |
| 1602 | ATTRIBUTE_ID *id; | |
| 1603 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
| 1604 | id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, appAtts[i], 0); | |
| 1605 | if (id->prefix->binding) { | |
| 1606 | int j; | |
| 1607 | const BINDING *b = id->prefix->binding; | |
| 1608 | const XML_Char *s = appAtts[i]; | |
| 1609 | for (j = 0; j < b->uriLen; j++) { | |
| 1610 | if (!poolAppendChar(&tempPool, b->uri[j])) | |
| 1611 | return XML_ERROR_NO_MEMORY; | |
| 1612 | } | |
| 1613 | while (*s++ != ':') | |
| 1614 | ; | |
| 1615 | do { | |
| 1616 | if (!poolAppendChar(&tempPool, *s)) | |
| 1617 | return XML_ERROR_NO_MEMORY; | |
| 1618 | } while (*s++); | |
| 1619 | appAtts[i] = poolStart(&tempPool); | |
| 1620 | poolFinish(&tempPool); | |
| 1621 | } | |
| 1622 | if (!--nPrefixes) | |
| 1623 | break; | |
| 1624 | } | |
| 1625 | else | |
| 1626 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
| 1627 | } | |
| 1628 | } | |
| 1629 | for (; i < attIndex; i += 2) | |
| 1630 | ((XML_Char *)(appAtts[i]))[-1] = 0; | |
| 1631 | if (!tagNamePtr) | |
| 1632 | return XML_ERROR_NONE; | |
| 1633 | for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding) | |
| 1634 | binding->attId->name[-1] = 0; | |
| 1635 | if (elementType->prefix) { | |
| 1636 | binding = elementType->prefix->binding; | |
| 1637 | if (!binding) | |
| 1638 | return XML_ERROR_NONE; | |
| 1639 | localPart = tagNamePtr->str; | |
| 1640 | while (*localPart++ != XML_T(':')) | |
| 1641 | ; | |
| 1642 | } | |
| 1643 | else if (dtd.defaultPrefix.binding) { | |
| 1644 | binding = dtd.defaultPrefix.binding; | |
| 1645 | localPart = tagNamePtr->str; | |
| 1646 | } | |
| 1647 | else | |
| 1648 | return XML_ERROR_NONE; | |
| 1649 | tagNamePtr->localPart = localPart; | |
| 1650 | tagNamePtr->uriLen = binding->uriLen; | |
| 1651 | i = binding->uriLen; | |
| 1652 | do { | |
| 1653 | if (i == binding->uriAlloc) { | |
| 1654 | binding->uri = realloc(binding->uri, binding->uriAlloc *= 2); | |
| 1655 | if (!binding->uri) | |
| 1656 | return XML_ERROR_NO_MEMORY; | |
| 1657 | } | |
| 1658 | binding->uri[i++] = *localPart; | |
| 1659 | } while (*localPart++); | |
| 1660 | tagNamePtr->str = binding->uri; | |
| 1661 | return XML_ERROR_NONE; | |
| 1662 | } | |
| 1663 | ||
| 1664 | static | |
| 1665 | int addBinding(XML_Parser parser, PREFIX *prefix, const ATTRIBUTE_ID *attId, const XML_Char *uri, BINDING **bindingsPtr) | |
| 1666 | { | |
| 1667 | BINDING *b; | |
| 1668 | int len; | |
| 1669 | for (len = 0; uri[len]; len++) | |
| 1670 | ; | |
| 1671 | if (namespaceSeparator) | |
| 1672 | len++; | |
| 1673 | if (freeBindingList) { | |
| 1674 | b = freeBindingList; | |
| 1675 | if (len > b->uriAlloc) { | |
| 1676 | b->uri = realloc(b->uri, len + EXPAND_SPARE); | |
| 1677 | if (!b->uri) | |
| 1678 | return 0; | |
| 1679 | b->uriAlloc = len + EXPAND_SPARE; | |
| 1680 | } | |
| 1681 | freeBindingList = b->nextTagBinding; | |
| 1682 | } | |
| 1683 | else { | |
| 1684 | b = malloc(sizeof(BINDING)); | |
| 1685 | if (!b) | |
| 1686 | return 0; | |
| 1687 | b->uri = malloc(sizeof(XML_Char) * len + EXPAND_SPARE); | |
| 1688 | if (!b->uri) { | |
| 1689 | free(b); | |
| 1690 | return 0; | |
| 1691 | } | |
| 1692 | b->uriAlloc = len; | |
| 1693 | } | |
| 1694 | b->uriLen = len; | |
| 1695 | memcpy(b->uri, uri, len * sizeof(XML_Char)); | |
| 1696 | if (namespaceSeparator) | |
| 1697 | b->uri[len - 1] = namespaceSeparator; | |
| 1698 | b->prefix = prefix; | |
| 1699 | b->attId = attId; | |
| 1700 | b->prevPrefixBinding = prefix->binding; | |
| 1701 | if (*uri == XML_T('\0') && prefix == &dtd.defaultPrefix) | |
| 1702 | prefix->binding = 0; | |
| 1703 | else | |
| 1704 | prefix->binding = b; | |
| 1705 | b->nextTagBinding = *bindingsPtr; | |
| 1706 | *bindingsPtr = b; | |
| 1707 | if (startNamespaceDeclHandler) | |
| 1708 | startNamespaceDeclHandler(handlerArg, prefix->name, | |
| 1709 | prefix->binding ? uri : 0); | |
| 1710 | return 1; | |
| 1711 | } | |
| 1712 | ||
| 1713 | /* The idea here is to avoid using stack for each CDATA section when | |
| 1714 | the whole file is parsed with one call. */ | |
| 1715 | ||
| 1716 | static | |
| 1717 | enum XML_Error cdataSectionProcessor(XML_Parser parser, | |
| 1718 | const char *start, | |
| 1719 | const char *end, | |
| 1720 | const char **endPtr) | |
| 1721 | { | |
| 1722 | enum XML_Error result = doCdataSection(parser, encoding, &start, end, endPtr); | |
| 1723 | if (start) { | |
| 1724 | processor = contentProcessor; | |
| 1725 | return contentProcessor(parser, start, end, endPtr); | |
| 1726 | } | |
| 1727 | return result; | |
| 1728 | } | |
| 1729 | ||
| 1730 | /* startPtr gets set to non-null is the section is closed, and to null if | |
| 1731 | the section is not yet closed. */ | |
| 1732 | ||
| 1733 | static | |
| 1734 | enum XML_Error doCdataSection(XML_Parser parser, | |
| 1735 | const ENCODING *enc, | |
| 1736 | const char **startPtr, | |
| 1737 | const char *end, | |
| 1738 | const char **nextPtr) | |
| 1739 | { | |
| 1740 | const char *s = *startPtr; | |
| 1741 | const char **eventPP; | |
| 1742 | const char **eventEndPP; | |
| 1743 | if (enc == encoding) { | |
| 1744 | eventPP = &eventPtr; | |
| 1745 | *eventPP = s; | |
| 1746 | eventEndPP = &eventEndPtr; | |
| 1747 | } | |
| 1748 | else { | |
| 1749 | eventPP = &(openInternalEntities->internalEventPtr); | |
| 1750 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
| 1751 | } | |
| 1752 | *eventPP = s; | |
| 1753 | *startPtr = 0; | |
| 1754 | for (;;) { | |
| 1755 | const char *next; | |
| 1756 | int tok = XmlCdataSectionTok(enc, s, end, &next); | |
| 1757 | *eventEndPP = next; | |
| 1758 | switch (tok) { | |
| 1759 | case XML_TOK_CDATA_SECT_CLOSE: | |
| 1760 | if (endCdataSectionHandler) | |
| 1761 | endCdataSectionHandler(handlerArg); | |
| 1762 | #if 0 | |
| 1763 | /* see comment under XML_TOK_CDATA_SECT_OPEN */ | |
| 1764 | else if (characterDataHandler) | |
| 1765 | characterDataHandler(handlerArg, dataBuf, 0); | |
| 1766 | #endif | |
| 1767 | else if (defaultHandler) | |
| 1768 | reportDefault(parser, enc, s, next); | |
| 1769 | *startPtr = next; | |
| 1770 | return XML_ERROR_NONE; | |
| 1771 | case XML_TOK_DATA_NEWLINE: | |
| 1772 | if (characterDataHandler) { | |
| 1773 | XML_Char c = 0xA; | |
| 1774 | characterDataHandler(handlerArg, &c, 1); | |
| 1775 | } | |
| 1776 | else if (defaultHandler) | |
| 1777 | reportDefault(parser, enc, s, next); | |
| 1778 | break; | |
| 1779 | case XML_TOK_DATA_CHARS: | |
| 1780 | if (characterDataHandler) { | |
| 1781 | if (MUST_CONVERT(enc, s)) { | |
| 1782 | for (;;) { | |
| 1783 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
| 1784 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *)dataBufEnd); | |
| 1785 | *eventEndPP = next; | |
| 1786 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
| 1787 | if (s == next) | |
| 1788 | break; | |
| 1789 | *eventPP = s; | |
| 1790 | } | |
| 1791 | } | |
| 1792 | else | |
| 1793 | characterDataHandler(handlerArg, | |
| 1794 | (XML_Char *)s, | |
| 1795 | (XML_Char *)next - (XML_Char *)s); | |
| 1796 | } | |
| 1797 | else if (defaultHandler) | |
| 1798 | reportDefault(parser, enc, s, next); | |
| 1799 | break; | |
| 1800 | case XML_TOK_INVALID: | |
| 1801 | *eventPP = next; | |
| 1802 | return XML_ERROR_INVALID_TOKEN; | |
| 1803 | case XML_TOK_PARTIAL_CHAR: | |
| 1804 | if (nextPtr) { | |
| 1805 | *nextPtr = s; | |
| 1806 | return XML_ERROR_NONE; | |
| 1807 | } | |
| 1808 | return XML_ERROR_PARTIAL_CHAR; | |
| 1809 | case XML_TOK_PARTIAL: | |
| 1810 | case XML_TOK_NONE: | |
| 1811 | if (nextPtr) { | |
| 1812 | *nextPtr = s; | |
| 1813 | return XML_ERROR_NONE; | |
| 1814 | } | |
| 1815 | return XML_ERROR_UNCLOSED_CDATA_SECTION; | |
| 1816 | default: | |
| 1817 | abort(); | |
| 1818 | } | |
| 1819 | *eventPP = s = next; | |
| 1820 | } | |
| 1821 | /* not reached */ | |
| 1822 | } | |
| 1823 | ||
| 1824 | static enum XML_Error | |
| 1825 | initializeEncoding(XML_Parser parser) | |
| 1826 | { | |
| 1827 | const char *s; | |
| 1828 | #ifdef XML_UNICODE | |
| 1829 | char encodingBuf[128]; | |
| 1830 | if (!protocolEncodingName) | |
| 1831 | s = 0; | |
| 1832 | else { | |
| 1833 | int i; | |
| 1834 | for (i = 0; protocolEncodingName[i]; i++) { | |
| 1835 | if (i == sizeof(encodingBuf) - 1 | |
| 1836 | || protocolEncodingName[i] >= 0x80 | |
| 1837 | || protocolEncodingName[i] < 0) { | |
| 1838 | encodingBuf[0] = '\0'; | |
| 1839 | break; | |
| 1840 | } | |
| 1841 | encodingBuf[i] = (char)protocolEncodingName[i]; | |
| 1842 | } | |
| 1843 | encodingBuf[i] = '\0'; | |
| 1844 | s = encodingBuf; | |
| 1845 | } | |
| 1846 | #else | |
| 1847 | s = protocolEncodingName; | |
| 1848 | #endif | |
| 1849 | if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s)) | |
| 1850 | return XML_ERROR_NONE; | |
| 1851 | return handleUnknownEncoding(parser, protocolEncodingName); | |
| 1852 | } | |
| 1853 | ||
| 1854 | static enum XML_Error | |
| 1855 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, | |
| 1856 | const char *s, const char *next) | |
| 1857 | { | |
| 1858 | const char *encodingName = 0; | |
| 1859 | const ENCODING *newEncoding = 0; | |
| 1860 | const char *version; | |
| 1861 | int standalone = -1; | |
| 1862 | if (!(ns | |
| 1863 | ? XmlParseXmlDeclNS | |
| 1864 | : XmlParseXmlDecl)(isGeneralTextEntity, | |
| 1865 | encoding, | |
| 1866 | s, | |
| 1867 | next, | |
| 1868 | &eventPtr, | |
| 1869 | &version, | |
| 1870 | &encodingName, | |
| 1871 | &newEncoding, | |
| 1872 | &standalone)) | |
| 1873 | return XML_ERROR_SYNTAX; | |
| 1874 | if (!isGeneralTextEntity && standalone == 1) | |
| 1875 | dtd.standalone = 1; | |
| 1876 | if (defaultHandler) | |
| 1877 | reportDefault(parser, encoding, s, next); | |
| 1878 | if (!protocolEncodingName) { | |
| 1879 | if (newEncoding) { | |
| 1880 | if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) { | |
| 1881 | eventPtr = encodingName; | |
| 1882 | return XML_ERROR_INCORRECT_ENCODING; | |
| 1883 | } | |
| 1884 | encoding = newEncoding; | |
| 1885 | } | |
| 1886 | else if (encodingName) { | |
| 1887 | enum XML_Error result; | |
| 1888 | const XML_Char *s = poolStoreString(&tempPool, | |
| 1889 | encoding, | |
| 1890 | encodingName, | |
| 1891 | encodingName | |
| 1892 | + XmlNameLength(encoding, encodingName)); | |
| 1893 | if (!s) | |
| 1894 | return XML_ERROR_NO_MEMORY; | |
| 1895 | result = handleUnknownEncoding(parser, s); | |
| 1896 | poolDiscard(&tempPool); | |
| 1897 | if (result == XML_ERROR_UNKNOWN_ENCODING) | |
| 1898 | eventPtr = encodingName; | |
| 1899 | return result; | |
| 1900 | } | |
| 1901 | } | |
| 1902 | return XML_ERROR_NONE; | |
| 1903 | } | |
| 1904 | ||
| 1905 | static enum XML_Error | |
| 1906 | handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName) | |
| 1907 | { | |
| 1908 | if (unknownEncodingHandler) { | |
| 1909 | XML_Encoding info; | |
| 1910 | int i; | |
| 1911 | for (i = 0; i < 256; i++) | |
| 1912 | info.map[i] = -1; | |
| 1913 | info.convert = 0; | |
| 1914 | info.data = 0; | |
| 1915 | info.release = 0; | |
| 1916 | if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, &info)) { | |
| 1917 | ENCODING *enc; | |
| 1918 | unknownEncodingMem = malloc(XmlSizeOfUnknownEncoding()); | |
| 1919 | if (!unknownEncodingMem) { | |
| 1920 | if (info.release) | |
| 1921 | info.release(info.data); | |
| 1922 | return XML_ERROR_NO_MEMORY; | |
| 1923 | } | |
| 1924 | enc = (ns | |
| 1925 | ? XmlInitUnknownEncodingNS | |
| 1926 | : XmlInitUnknownEncoding)(unknownEncodingMem, | |
| 1927 | info.map, | |
| 1928 | info.convert, | |
| 1929 | info.data); | |
| 1930 | if (enc) { | |
| 1931 | unknownEncodingData = info.data; | |
| 1932 | unknownEncodingRelease = info.release; | |
| 1933 | encoding = enc; | |
| 1934 | return XML_ERROR_NONE; | |
| 1935 | } | |
| 1936 | } | |
| 1937 | if (info.release) | |
| 1938 | info.release(info.data); | |
| 1939 | } | |
| 1940 | return XML_ERROR_UNKNOWN_ENCODING; | |
| 1941 | } | |
| 1942 | ||
| 1943 | static enum XML_Error | |
| 1944 | prologInitProcessor(XML_Parser parser, | |
| 1945 | const char *s, | |
| 1946 | const char *end, | |
| 1947 | const char **nextPtr) | |
| 1948 | { | |
| 1949 | enum XML_Error result = initializeEncoding(parser); | |
| 1950 | if (result != XML_ERROR_NONE) | |
| 1951 | return result; | |
| 1952 | processor = prologProcessor; | |
| 1953 | return prologProcessor(parser, s, end, nextPtr); | |
| 1954 | } | |
| 1955 | ||
| 1956 | static enum XML_Error | |
| 1957 | prologProcessor(XML_Parser parser, | |
| 1958 | const char *s, | |
| 1959 | const char *end, | |
| 1960 | const char **nextPtr) | |
| 1961 | { | |
| 1962 | for (;;) { | |
| 1963 | const char *next; | |
| 1964 | int tok = XmlPrologTok(encoding, s, end, &next); | |
| 1965 | if (tok <= 0) { | |
| 1966 | if (nextPtr != 0 && tok != XML_TOK_INVALID) { | |
| 1967 | *nextPtr = s; | |
| 1968 | return XML_ERROR_NONE; | |
| 1969 | } | |
| 1970 | switch (tok) { | |
| 1971 | case XML_TOK_INVALID: | |
| 1972 | eventPtr = next; | |
| 1973 | return XML_ERROR_INVALID_TOKEN; | |
| 1974 | case XML_TOK_NONE: | |
| 1975 | return XML_ERROR_NO_ELEMENTS; | |
| 1976 | case XML_TOK_PARTIAL: | |
| 1977 | return XML_ERROR_UNCLOSED_TOKEN; | |
| 1978 | case XML_TOK_PARTIAL_CHAR: | |
| 1979 | return XML_ERROR_PARTIAL_CHAR; | |
| 1980 | case XML_TOK_TRAILING_CR: | |
| 1981 | eventPtr = s + encoding->minBytesPerChar; | |
| 1982 | return XML_ERROR_NO_ELEMENTS; | |
| 1983 | default: | |
| 1984 | abort(); | |
| 1985 | } | |
| 1986 | } | |
| 1987 | switch (XmlTokenRole(&prologState, tok, s, next, encoding)) { | |
| 1988 | case XML_ROLE_XML_DECL: | |
| 1989 | { | |
| 1990 | enum XML_Error result = processXmlDecl(parser, 0, s, next); | |
| 1991 | if (result != XML_ERROR_NONE) | |
| 1992 | return result; | |
| 1993 | } | |
| 1994 | break; | |
| 1995 | case XML_ROLE_DOCTYPE_SYSTEM_ID: | |
| 1996 | if (!dtd.standalone | |
| 1997 | && notStandaloneHandler | |
| 1998 | && !notStandaloneHandler(handlerArg)) | |
| 1999 | return XML_ERROR_NOT_STANDALONE; | |
| 2000 | hadExternalDoctype = 1; | |
| 2001 | break; | |
| 2002 | case XML_ROLE_DOCTYPE_PUBLIC_ID: | |
| 2003 | case XML_ROLE_ENTITY_PUBLIC_ID: | |
| 2004 | if (!XmlIsPublicId(encoding, s, next, &eventPtr)) | |
| 2005 | return XML_ERROR_SYNTAX; | |
| 2006 | if (declEntity) { | |
| 2007 | XML_Char *tem = poolStoreString(&dtd.pool, | |
| 2008 | encoding, | |
| 2009 | s + encoding->minBytesPerChar, | |
| 2010 | next - encoding->minBytesPerChar); | |
| 2011 | if (!tem) | |
| 2012 | return XML_ERROR_NO_MEMORY; | |
| 2013 | normalizePublicId(tem); | |
| 2014 | declEntity->publicId = tem; | |
| 2015 | poolFinish(&dtd.pool); | |
| 2016 | } | |
| 2017 | break; | |
| 2018 | case XML_ROLE_INSTANCE_START: | |
| 2019 | processor = contentProcessor; | |
| 2020 | if (hadExternalDoctype) | |
| 2021 | dtd.complete = 0; | |
| 2022 | return contentProcessor(parser, s, end, nextPtr); | |
| 2023 | case XML_ROLE_ATTLIST_ELEMENT_NAME: | |
| 2024 | { | |
| 2025 | const XML_Char *name = poolStoreString(&dtd.pool, encoding, s, next); | |
| 2026 | if (!name) | |
| 2027 | return XML_ERROR_NO_MEMORY; | |
| 2028 | declElementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, name, sizeof(ELEMENT_TYPE)); | |
| 2029 | if (!declElementType) | |
| 2030 | return XML_ERROR_NO_MEMORY; | |
| 2031 | if (declElementType->name != name) | |
| 2032 | poolDiscard(&dtd.pool); | |
| 2033 | else { | |
| 2034 | poolFinish(&dtd.pool); | |
| 2035 | if (!setElementTypePrefix(parser, declElementType)) | |
| 2036 | return XML_ERROR_NO_MEMORY; | |
| 2037 | } | |
| 2038 | break; | |
| 2039 | } | |
| 2040 | case XML_ROLE_ATTRIBUTE_NAME: | |
| 2041 | declAttributeId = getAttributeId(parser, encoding, s, next); | |
| 2042 | if (!declAttributeId) | |
| 2043 | return XML_ERROR_NO_MEMORY; | |
| 2044 | declAttributeIsCdata = 0; | |
| 2045 | break; | |
| 2046 | case XML_ROLE_ATTRIBUTE_TYPE_CDATA: | |
| 2047 | declAttributeIsCdata = 1; | |
| 2048 | break; | |
| 2049 | case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE: | |
| 2050 | case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE: | |
| 2051 | if (dtd.complete | |
| 2052 | && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, 0)) | |
| 2053 | return XML_ERROR_NO_MEMORY; | |
| 2054 | break; | |
| 2055 | case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE: | |
| 2056 | case XML_ROLE_FIXED_ATTRIBUTE_VALUE: | |
| 2057 | { | |
| 2058 | const XML_Char *attVal; | |
| 2059 | enum XML_Error result | |
| 2060 | = storeAttributeValue(parser, encoding, declAttributeIsCdata, | |
| 2061 | s + encoding->minBytesPerChar, | |
| 2062 | next - encoding->minBytesPerChar, | |
| 2063 | &dtd.pool); | |
| 2064 | if (result) | |
| 2065 | return result; | |
| 2066 | attVal = poolStart(&dtd.pool); | |
| 2067 | poolFinish(&dtd.pool); | |
| 2068 | if (dtd.complete | |
| 2069 | && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, attVal)) | |
| 2070 | return XML_ERROR_NO_MEMORY; | |
| 2071 | break; | |
| 2072 | } | |
| 2073 | case XML_ROLE_ENTITY_VALUE: | |
| 2074 | { | |
| 2075 | enum XML_Error result = storeEntityValue(parser, s, next); | |
| 2076 | if (result != XML_ERROR_NONE) | |
| 2077 | return result; | |
| 2078 | } | |
| 2079 | break; | |
| 2080 | case XML_ROLE_ENTITY_SYSTEM_ID: | |
| 2081 | if (declEntity) { | |
| 2082 | declEntity->systemId = poolStoreString(&dtd.pool, encoding, | |
| 2083 | s + encoding->minBytesPerChar, | |
| 2084 | next - encoding->minBytesPerChar); | |
| 2085 | if (!declEntity->systemId) | |
| 2086 | return XML_ERROR_NO_MEMORY; | |
| 2087 | declEntity->base = dtd.base; | |
| 2088 | poolFinish(&dtd.pool); | |
| 2089 | } | |
| 2090 | break; | |
| 2091 | case XML_ROLE_ENTITY_NOTATION_NAME: | |
| 2092 | if (declEntity) { | |
| 2093 | declEntity->notation = poolStoreString(&dtd.pool, encoding, s, next); | |
| 2094 | if (!declEntity->notation) | |
| 2095 | return XML_ERROR_NO_MEMORY; | |
| 2096 | poolFinish(&dtd.pool); | |
| 2097 | if (unparsedEntityDeclHandler) { | |
| 2098 | eventPtr = eventEndPtr = s; | |
| 2099 | unparsedEntityDeclHandler(handlerArg, | |
| 2100 | declEntity->name, | |
| 2101 | declEntity->base, | |
| 2102 | declEntity->systemId, | |
| 2103 | declEntity->publicId, | |
| 2104 | declEntity->notation); | |
| 2105 | } | |
| 2106 | ||
| 2107 | } | |
| 2108 | break; | |
| 2109 | case XML_ROLE_GENERAL_ENTITY_NAME: | |
| 2110 | { | |
| 2111 | const XML_Char *name; | |
| 2112 | if (XmlPredefinedEntityName(encoding, s, next)) { | |
| 2113 | declEntity = 0; | |
| 2114 | break; | |
| 2115 | } | |
| 2116 | name = poolStoreString(&dtd.pool, encoding, s, next); | |
| 2117 | if (!name) | |
| 2118 | return XML_ERROR_NO_MEMORY; | |
| 2119 | if (dtd.complete) { | |
| 2120 | declEntity = (ENTITY *)lookup(&dtd.generalEntities, name, sizeof(ENTITY)); | |
| 2121 | if (!declEntity) | |
| 2122 | return XML_ERROR_NO_MEMORY; | |
| 2123 | if (declEntity->name != name) { | |
| 2124 | poolDiscard(&dtd.pool); | |
| 2125 | declEntity = 0; | |
| 2126 | } | |
| 2127 | else | |
| 2128 | poolFinish(&dtd.pool); | |
| 2129 | } | |
| 2130 | else { | |
| 2131 | poolDiscard(&dtd.pool); | |
| 2132 | declEntity = 0; | |
| 2133 | } | |
| 2134 | } | |
| 2135 | break; | |
| 2136 | case XML_ROLE_PARAM_ENTITY_NAME: | |
| 2137 | declEntity = 0; | |
| 2138 | break; | |
| 2139 | case XML_ROLE_NOTATION_NAME: | |
| 2140 | declNotationPublicId = 0; | |
| 2141 | declNotationName = 0; | |
| 2142 | if (notationDeclHandler) { | |
| 2143 | declNotationName = poolStoreString(&tempPool, encoding, s, next); | |
| 2144 | if (!declNotationName) | |
| 2145 | return XML_ERROR_NO_MEMORY; | |
| 2146 | poolFinish(&tempPool); | |
| 2147 | } | |
| 2148 | break; | |
| 2149 | case XML_ROLE_NOTATION_PUBLIC_ID: | |
| 2150 | if (!XmlIsPublicId(encoding, s, next, &eventPtr)) | |
| 2151 | return XML_ERROR_SYNTAX; | |
| 2152 | if (declNotationName) { | |
| 2153 | XML_Char *tem = poolStoreString(&tempPool, | |
| 2154 | encoding, | |
| 2155 | s + encoding->minBytesPerChar, | |
| 2156 | next - encoding->minBytesPerChar); | |
| 2157 | if (!tem) | |
| 2158 | return XML_ERROR_NO_MEMORY; | |
| 2159 | normalizePublicId(tem); | |
| 2160 | declNotationPublicId = tem; | |
| 2161 | poolFinish(&tempPool); | |
| 2162 | } | |
| 2163 | break; | |
| 2164 | case XML_ROLE_NOTATION_SYSTEM_ID: | |
| 2165 | if (declNotationName && notationDeclHandler) { | |
| 2166 | const XML_Char *systemId | |
| 2167 | = poolStoreString(&tempPool, encoding, | |
| 2168 | s + encoding->minBytesPerChar, | |
| 2169 | next - encoding->minBytesPerChar); | |
| 2170 | if (!systemId) | |
| 2171 | return XML_ERROR_NO_MEMORY; | |
| 2172 | eventPtr = eventEndPtr = s; | |
| 2173 | notationDeclHandler(handlerArg, | |
| 2174 | declNotationName, | |
| 2175 | dtd.base, | |
| 2176 | systemId, | |
| 2177 | declNotationPublicId); | |
| 2178 | } | |
| 2179 | poolClear(&tempPool); | |
| 2180 | break; | |
| 2181 | case XML_ROLE_NOTATION_NO_SYSTEM_ID: | |
| 2182 | if (declNotationPublicId && notationDeclHandler) { | |
| 2183 | eventPtr = eventEndPtr = s; | |
| 2184 | notationDeclHandler(handlerArg, | |
| 2185 | declNotationName, | |
| 2186 | dtd.base, | |
| 2187 | 0, | |
| 2188 | declNotationPublicId); | |
| 2189 | } | |
| 2190 | poolClear(&tempPool); | |
| 2191 | break; | |
| 2192 | case XML_ROLE_ERROR: | |
| 2193 | eventPtr = s; | |
| 2194 | switch (tok) { | |
| 2195 | case XML_TOK_PARAM_ENTITY_REF: | |
| 2196 | return XML_ERROR_PARAM_ENTITY_REF; | |
| 2197 | case XML_TOK_XML_DECL: | |
| 2198 | return XML_ERROR_MISPLACED_XML_PI; | |
| 2199 | default: | |
| 2200 | return XML_ERROR_SYNTAX; | |
| 2201 | } | |
| 2202 | case XML_ROLE_GROUP_OPEN: | |
| 2203 | if (prologState.level >= groupSize) { | |
| 2204 | if (groupSize) | |
| 2205 | groupConnector = realloc(groupConnector, groupSize *= 2); | |
| 2206 | else | |
| 2207 | groupConnector = malloc(groupSize = 32); | |
| 2208 | if (!groupConnector) | |
| 2209 | return XML_ERROR_NO_MEMORY; | |
| 2210 | } | |
| 2211 | groupConnector[prologState.level] = 0; | |
| 2212 | break; | |
| 2213 | case XML_ROLE_GROUP_SEQUENCE: | |
| 2214 | if (groupConnector[prologState.level] == '|') { | |
| 2215 | eventPtr = s; | |
| 2216 | return XML_ERROR_SYNTAX; | |
| 2217 | } | |
| 2218 | groupConnector[prologState.level] = ','; | |
| 2219 | break; | |
| 2220 | case XML_ROLE_GROUP_CHOICE: | |
| 2221 | if (groupConnector[prologState.level] == ',') { | |
| 2222 | eventPtr = s; | |
| 2223 | return XML_ERROR_SYNTAX; | |
| 2224 | } | |
| 2225 | groupConnector[prologState.level] = '|'; | |
| 2226 | break; | |
| 2227 | case XML_ROLE_PARAM_ENTITY_REF: | |
| 2228 | if (!dtd.standalone | |
| 2229 | && notStandaloneHandler | |
| 2230 | && !notStandaloneHandler(handlerArg)) | |
| 2231 | return XML_ERROR_NOT_STANDALONE; | |
| 2232 | dtd.complete = 0; | |
| 2233 | break; | |
| 2234 | case XML_ROLE_NONE: | |
| 2235 | switch (tok) { | |
| 2236 | case XML_TOK_PI: | |
| 2237 | eventPtr = s; | |
| 2238 | eventEndPtr = next; | |
| 2239 | if (!reportProcessingInstruction(parser, encoding, s, next)) | |
| 2240 | return XML_ERROR_NO_MEMORY; | |
| 2241 | break; | |
| 2242 | case XML_TOK_COMMENT: | |
| 2243 | eventPtr = s; | |
| 2244 | eventEndPtr = next; | |
| 2245 | if (!reportComment(parser, encoding, s, next)) | |
| 2246 | return XML_ERROR_NO_MEMORY; | |
| 2247 | break; | |
| 2248 | } | |
| 2249 | break; | |
| 2250 | } | |
| 2251 | if (defaultHandler) { | |
| 2252 | switch (tok) { | |
| 2253 | case XML_TOK_PI: | |
| 2254 | case XML_TOK_COMMENT: | |
| 2255 | case XML_TOK_BOM: | |
| 2256 | case XML_TOK_XML_DECL: | |
| 2257 | break; | |
| 2258 | default: | |
| 2259 | eventPtr = s; | |
| 2260 | eventEndPtr = next; | |
| 2261 | reportDefault(parser, encoding, s, next); | |
| 2262 | } | |
| 2263 | } | |
| 2264 | s = next; | |
| 2265 | } | |
| 2266 | /* not reached */ | |
| 2267 | } | |
| 2268 | ||
| 2269 | static | |
| 2270 | enum XML_Error epilogProcessor(XML_Parser parser, | |
| 2271 | const char *s, | |
| 2272 | const char *end, | |
| 2273 | const char **nextPtr) | |
| 2274 | { | |
| 2275 | processor = epilogProcessor; | |
| 2276 | eventPtr = s; | |
| 2277 | for (;;) { | |
| 2278 | const char *next; | |
| 2279 | int tok = XmlPrologTok(encoding, s, end, &next); | |
| 2280 | eventEndPtr = next; | |
| 2281 | switch (tok) { | |
| 2282 | case XML_TOK_TRAILING_CR: | |
| 2283 | if (defaultHandler) { | |
| 2284 | eventEndPtr = end; | |
| 2285 | reportDefault(parser, encoding, s, end); | |
| 2286 | } | |
| 2287 | /* fall through */ | |
| 2288 | case XML_TOK_NONE: | |
| 2289 | if (nextPtr) | |
| 2290 | *nextPtr = end; | |
| 2291 | return XML_ERROR_NONE; | |
| 2292 | case XML_TOK_PROLOG_S: | |
| 2293 | if (defaultHandler) | |
| 2294 | reportDefault(parser, encoding, s, next); | |
| 2295 | break; | |
| 2296 | case XML_TOK_PI: | |
| 2297 | if (!reportProcessingInstruction(parser, encoding, s, next)) | |
| 2298 | return XML_ERROR_NO_MEMORY; | |
| 2299 | break; | |
| 2300 | case XML_TOK_COMMENT: | |
| 2301 | if (!reportComment(parser, encoding, s, next)) | |
| 2302 | return XML_ERROR_NO_MEMORY; | |
| 2303 | break; | |
| 2304 | case XML_TOK_INVALID: | |
| 2305 | eventPtr = next; | |
| 2306 | return XML_ERROR_INVALID_TOKEN; | |
| 2307 | case XML_TOK_PARTIAL: | |
| 2308 | if (nextPtr) { | |
| 2309 | *nextPtr = s; | |
| 2310 | return XML_ERROR_NONE; | |
| 2311 | } | |
| 2312 | return XML_ERROR_UNCLOSED_TOKEN; | |
| 2313 | case XML_TOK_PARTIAL_CHAR: | |
| 2314 | if (nextPtr) { | |
| 2315 | *nextPtr = s; | |
| 2316 | return XML_ERROR_NONE; | |
| 2317 | } | |
| 2318 | return XML_ERROR_PARTIAL_CHAR; | |
| 2319 | default: | |
| 2320 | return XML_ERROR_JUNK_AFTER_DOC_ELEMENT; | |
| 2321 | } | |
| 2322 | eventPtr = s = next; | |
| 2323 | } | |
| 2324 | } | |
| 2325 | ||
| 2326 | static enum XML_Error | |
| 2327 | storeAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, | |
| 2328 | const char *ptr, const char *end, | |
| 2329 | STRING_POOL *pool) | |
| 2330 | { | |
| 2331 | enum XML_Error result = appendAttributeValue(parser, enc, isCdata, ptr, end, pool); | |
| 2332 | if (result) | |
| 2333 | return result; | |
| 2334 | if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20) | |
| 2335 | poolChop(pool); | |
| 2336 | if (!poolAppendChar(pool, XML_T('\0'))) | |
| 2337 | return XML_ERROR_NO_MEMORY; | |
| 2338 | return XML_ERROR_NONE; | |
| 2339 | } | |
| 2340 | ||
| 2341 | static enum XML_Error | |
| 2342 | appendAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata, | |
| 2343 | const char *ptr, const char *end, | |
| 2344 | STRING_POOL *pool) | |
| 2345 | { | |
| 2346 | const ENCODING *internalEnc = ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(); | |
| 2347 | for (;;) { | |
| 2348 | const char *next; | |
| 2349 | int tok = XmlAttributeValueTok(enc, ptr, end, &next); | |
| 2350 | switch (tok) { | |
| 2351 | case XML_TOK_NONE: | |
| 2352 | return XML_ERROR_NONE; | |
| 2353 | case XML_TOK_INVALID: | |
| 2354 | if (enc == encoding) | |
| 2355 | eventPtr = next; | |
| 2356 | return XML_ERROR_INVALID_TOKEN; | |
| 2357 | case XML_TOK_PARTIAL: | |
| 2358 | if (enc == encoding) | |
| 2359 | eventPtr = ptr; | |
| 2360 | return XML_ERROR_INVALID_TOKEN; | |
| 2361 | case XML_TOK_CHAR_REF: | |
| 2362 | { | |
| 2363 | XML_Char buf[XML_ENCODE_MAX]; | |
| 2364 | int i; | |
| 2365 | int n = XmlCharRefNumber(enc, ptr); | |
| 2366 | if (n < 0) { | |
| 2367 | if (enc == encoding) | |
| 2368 | eventPtr = ptr; | |
| 2369 | return XML_ERROR_BAD_CHAR_REF; | |
| 2370 | } | |
| 2371 | if (!isCdata | |
| 2372 | && n == 0x20 /* space */ | |
| 2373 | && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
| 2374 | break; | |
| 2375 | n = XmlEncode(n, (ICHAR *)buf); | |
| 2376 | if (!n) { | |
| 2377 | if (enc == encoding) | |
| 2378 | eventPtr = ptr; | |
| 2379 | return XML_ERROR_BAD_CHAR_REF; | |
| 2380 | } | |
| 2381 | for (i = 0; i < n; i++) { | |
| 2382 | if (!poolAppendChar(pool, buf[i])) | |
| 2383 | return XML_ERROR_NO_MEMORY; | |
| 2384 | } | |
| 2385 | } | |
| 2386 | break; | |
| 2387 | case XML_TOK_DATA_CHARS: | |
| 2388 | if (!poolAppend(pool, enc, ptr, next)) | |
| 2389 | return XML_ERROR_NO_MEMORY; | |
| 2390 | break; | |
| 2391 | break; | |
| 2392 | case XML_TOK_TRAILING_CR: | |
| 2393 | next = ptr + enc->minBytesPerChar; | |
| 2394 | /* fall through */ | |
| 2395 | case XML_TOK_ATTRIBUTE_VALUE_S: | |
| 2396 | case XML_TOK_DATA_NEWLINE: | |
| 2397 | if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20)) | |
| 2398 | break; | |
| 2399 | if (!poolAppendChar(pool, 0x20)) | |
| 2400 | return XML_ERROR_NO_MEMORY; | |
| 2401 | break; | |
| 2402 | case XML_TOK_ENTITY_REF: | |
| 2403 | { | |
| 2404 | const XML_Char *name; | |
| 2405 | ENTITY *entity; | |
| 2406 | XML_Char ch = XmlPredefinedEntityName(enc, | |
| 2407 | ptr + enc->minBytesPerChar, | |
| 2408 | next - enc->minBytesPerChar); | |
| 2409 | if (ch) { | |
| 2410 | if (!poolAppendChar(pool, ch)) | |
| 2411 | return XML_ERROR_NO_MEMORY; | |
| 2412 | break; | |
| 2413 | } | |
| 2414 | name = poolStoreString(&temp2Pool, enc, | |
| 2415 | ptr + enc->minBytesPerChar, | |
| 2416 | next - enc->minBytesPerChar); | |
| 2417 | if (!name) | |
| 2418 | return XML_ERROR_NO_MEMORY; | |
| 2419 | entity = (ENTITY *)lookup(&dtd.generalEntities, name, 0); | |
| 2420 | poolDiscard(&temp2Pool); | |
| 2421 | if (!entity) { | |
| 2422 | if (dtd.complete) { | |
| 2423 | if (enc == encoding) | |
| 2424 | eventPtr = ptr; | |
| 2425 | return XML_ERROR_UNDEFINED_ENTITY; | |
| 2426 | } | |
| 2427 | } | |
| 2428 | else if (entity->open) { | |
| 2429 | if (enc == encoding) | |
| 2430 | eventPtr = ptr; | |
| 2431 | return XML_ERROR_RECURSIVE_ENTITY_REF; | |
| 2432 | } | |
| 2433 | else if (entity->notation) { | |
| 2434 | if (enc == encoding) | |
| 2435 | eventPtr = ptr; | |
| 2436 | return XML_ERROR_BINARY_ENTITY_REF; | |
| 2437 | } | |
| 2438 | else if (!entity->textPtr) { | |
| 2439 | if (enc == encoding) | |
| 2440 | eventPtr = ptr; | |
| 2441 | return XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF; | |
| 2442 | } | |
| 2443 | else { | |
| 2444 | enum XML_Error result; | |
| 2445 | const XML_Char *textEnd = entity->textPtr + entity->textLen; | |
| 2446 | entity->open = 1; | |
| 2447 | result = appendAttributeValue(parser, internalEnc, isCdata, (char *)entity->textPtr, (char *)textEnd, pool); | |
| 2448 | entity->open = 0; | |
| 2449 | if (result) | |
| 2450 | return result; | |
| 2451 | } | |
| 2452 | } | |
| 2453 | break; | |
| 2454 | default: | |
| 2455 | abort(); | |
| 2456 | } | |
| 2457 | ptr = next; | |
| 2458 | } | |
| 2459 | /* not reached */ | |
| 2460 | } | |
| 2461 | ||
| 2462 | static | |
| 2463 | enum XML_Error storeEntityValue(XML_Parser parser, | |
| 2464 | const char *entityTextPtr, | |
| 2465 | const char *entityTextEnd) | |
| 2466 | { | |
| 3127 | 2467 | const ENCODING *internalEnc; |
| 2086 | 2468 | STRING_POOL *pool = &(dtd.pool); |
| 2469 | entityTextPtr += encoding->minBytesPerChar; | |
| 2470 | entityTextEnd -= encoding->minBytesPerChar; | |
| 3127 | 2471 | internalEnc = ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(); |
| 2086 | 2472 | for (;;) { |
| 2473 | const char *next; | |
| 2474 | int tok = XmlEntityValueTok(encoding, entityTextPtr, entityTextEnd, &next); | |
| 2475 | switch (tok) { | |
| 2476 | case XML_TOK_PARAM_ENTITY_REF: | |
| 2477 | eventPtr = entityTextPtr; | |
| 2478 | return XML_ERROR_SYNTAX; | |
| 2479 | case XML_TOK_NONE: | |
| 2480 | if (declEntity) { | |
| 2481 | declEntity->textPtr = pool->start; | |
| 2482 | declEntity->textLen = pool->ptr - pool->start; | |
| 2483 | poolFinish(pool); | |
| 2484 | } | |
| 2485 | else | |
| 2486 | poolDiscard(pool); | |
| 2487 | return XML_ERROR_NONE; | |
| 2488 | case XML_TOK_ENTITY_REF: | |
| 2489 | case XML_TOK_DATA_CHARS: | |
| 2490 | if (!poolAppend(pool, encoding, entityTextPtr, next)) | |
| 2491 | return XML_ERROR_NO_MEMORY; | |
| 2492 | break; | |
| 2493 | case XML_TOK_TRAILING_CR: | |
| 2494 | next = entityTextPtr + encoding->minBytesPerChar; | |
| 2495 | /* fall through */ | |
| 2496 | case XML_TOK_DATA_NEWLINE: | |
| 2497 | if (pool->end == pool->ptr && !poolGrow(pool)) | |
| 2498 | return XML_ERROR_NO_MEMORY; | |
| 2499 | *(pool->ptr)++ = 0xA; | |
| 2500 | break; | |
| 2501 | case XML_TOK_CHAR_REF: | |
| 2502 | { | |
| 2503 | XML_Char buf[XML_ENCODE_MAX]; | |
| 2504 | int i; | |
| 2505 | int n = XmlCharRefNumber(encoding, entityTextPtr); | |
| 2506 | if (n < 0) { | |
| 2507 | eventPtr = entityTextPtr; | |
| 2508 | return XML_ERROR_BAD_CHAR_REF; | |
| 2509 | } | |
| 2510 | n = XmlEncode(n, (ICHAR *)buf); | |
| 2511 | if (!n) { | |
| 2512 | eventPtr = entityTextPtr; | |
| 2513 | return XML_ERROR_BAD_CHAR_REF; | |
| 2514 | } | |
| 2515 | for (i = 0; i < n; i++) { | |
| 2516 | if (pool->end == pool->ptr && !poolGrow(pool)) | |
| 2517 | return XML_ERROR_NO_MEMORY; | |
| 2518 | *(pool->ptr)++ = buf[i]; | |
| 2519 | } | |
| 2520 | } | |
| 2521 | break; | |
| 2522 | case XML_TOK_PARTIAL: | |
| 2523 | eventPtr = entityTextPtr; | |
| 2524 | return XML_ERROR_INVALID_TOKEN; | |
| 2525 | case XML_TOK_INVALID: | |
| 2526 | eventPtr = next; | |
| 2527 | return XML_ERROR_INVALID_TOKEN; | |
| 2528 | default: | |
| 2529 | abort(); | |
| 2530 | } | |
| 2531 | entityTextPtr = next; | |
| 2532 | } | |
| 2533 | /* not reached */ | |
| 2534 | } | |
| 2535 | ||
| 2536 | static void | |
| 2537 | normalizeLines(XML_Char *s) | |
| 2538 | { | |
| 2539 | XML_Char *p; | |
| 2540 | for (;; s++) { | |
| 2541 | if (*s == XML_T('\0')) | |
| 2542 | return; | |
| 2543 | if (*s == 0xD) | |
| 2544 | break; | |
| 2545 | } | |
| 2546 | p = s; | |
| 2547 | do { | |
| 2548 | if (*s == 0xD) { | |
| 2549 | *p++ = 0xA; | |
| 2550 | if (*++s == 0xA) | |
| 2551 | s++; | |
| 2552 | } | |
| 2553 | else | |
| 2554 | *p++ = *s++; | |
| 2555 | } while (*s); | |
| 2556 | *p = XML_T('\0'); | |
| 2557 | } | |
| 2558 | ||
| 2559 | static int | |
| 2560 | reportProcessingInstruction(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
| 2561 | { | |
| 2562 | const XML_Char *target; | |
| 2563 | XML_Char *data; | |
| 2564 | const char *tem; | |
| 2565 | if (!processingInstructionHandler) { | |
| 2566 | if (defaultHandler) | |
| 2567 | reportDefault(parser, enc, start, end); | |
| 2568 | return 1; | |
| 2569 | } | |
| 2570 | start += enc->minBytesPerChar * 2; | |
| 2571 | tem = start + XmlNameLength(enc, start); | |
| 2572 | target = poolStoreString(&tempPool, enc, start, tem); | |
| 2573 | if (!target) | |
| 2574 | return 0; | |
| 2575 | poolFinish(&tempPool); | |
| 2576 | data = poolStoreString(&tempPool, enc, | |
| 2577 | XmlSkipS(enc, tem), | |
| 2578 | end - enc->minBytesPerChar*2); | |
| 2579 | if (!data) | |
| 2580 | return 0; | |
| 2581 | normalizeLines(data); | |
| 2582 | processingInstructionHandler(handlerArg, target, data); | |
| 2583 | poolClear(&tempPool); | |
| 2584 | return 1; | |
| 2585 | } | |
| 2586 | ||
| 2587 | static int | |
| 2588 | reportComment(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
| 2589 | { | |
| 2590 | XML_Char *data; | |
| 2591 | if (!commentHandler) { | |
| 2592 | if (defaultHandler) | |
| 2593 | reportDefault(parser, enc, start, end); | |
| 2594 | return 1; | |
| 2595 | } | |
| 2596 | data = poolStoreString(&tempPool, | |
| 2597 | enc, | |
| 2598 | start + enc->minBytesPerChar * 4, | |
| 2599 | end - enc->minBytesPerChar * 3); | |
| 2600 | if (!data) | |
| 2601 | return 0; | |
| 2602 | normalizeLines(data); | |
| 2603 | commentHandler(handlerArg, data); | |
| 2604 | poolClear(&tempPool); | |
| 2605 | return 1; | |
| 2606 | } | |
| 2607 | ||
| 2608 | static void | |
| 2609 | reportDefault(XML_Parser parser, const ENCODING *enc, const char *s, const char *end) | |
| 2610 | { | |
| 2611 | if (MUST_CONVERT(enc, s)) { | |
| 2612 | const char **eventPP; | |
| 2613 | const char **eventEndPP; | |
| 2614 | if (enc == encoding) { | |
| 2615 | eventPP = &eventPtr; | |
| 2616 | eventEndPP = &eventEndPtr; | |
| 2617 | } | |
| 2618 | else { | |
| 2619 | eventPP = &(openInternalEntities->internalEventPtr); | |
| 2620 | eventEndPP = &(openInternalEntities->internalEventEndPtr); | |
| 2621 | } | |
| 2622 | do { | |
| 2623 | ICHAR *dataPtr = (ICHAR *)dataBuf; | |
| 2624 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *)dataBufEnd); | |
| 2625 | *eventEndPP = s; | |
| 2626 | defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *)dataBuf); | |
| 2627 | *eventPP = s; | |
| 2628 | } while (s != end); | |
| 2629 | } | |
| 2630 | else | |
| 2631 | defaultHandler(handlerArg, (XML_Char *)s, (XML_Char *)end - (XML_Char *)s); | |
| 2632 | } | |
| 2633 | ||
| 2634 | ||
| 2635 | static int | |
| 2636 | defineAttribute(ELEMENT_TYPE *type, ATTRIBUTE_ID *attId, int isCdata, const XML_Char *value) | |
| 2637 | { | |
| 2638 | DEFAULT_ATTRIBUTE *att; | |
| 2639 | if (type->nDefaultAtts == type->allocDefaultAtts) { | |
| 2640 | if (type->allocDefaultAtts == 0) { | |
| 2641 | type->allocDefaultAtts = 8; | |
| 2642 | type->defaultAtts = malloc(type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); | |
| 2643 | } | |
| 2644 | else { | |
| 2645 | type->allocDefaultAtts *= 2; | |
| 2646 | type->defaultAtts = realloc(type->defaultAtts, | |
| 2647 | type->allocDefaultAtts*sizeof(DEFAULT_ATTRIBUTE)); | |
| 2648 | } | |
| 2649 | if (!type->defaultAtts) | |
| 2650 | return 0; | |
| 2651 | } | |
| 2652 | att = type->defaultAtts + type->nDefaultAtts; | |
| 2653 | att->id = attId; | |
| 2654 | att->value = value; | |
| 2655 | att->isCdata = isCdata; | |
| 2656 | if (!isCdata) | |
| 2657 | attId->maybeTokenized = 1; | |
| 2658 | type->nDefaultAtts += 1; | |
| 2659 | return 1; | |
| 2660 | } | |
| 2661 | ||
| 2662 | static int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *elementType) | |
| 2663 | { | |
| 2664 | const XML_Char *name; | |
| 2665 | for (name = elementType->name; *name; name++) { | |
| 2666 | if (*name == XML_T(':')) { | |
| 2667 | PREFIX *prefix; | |
| 2668 | const XML_Char *s; | |
| 2669 | for (s = elementType->name; s != name; s++) { | |
| 2670 | if (!poolAppendChar(&dtd.pool, *s)) | |
| 2671 | return 0; | |
| 2672 | } | |
| 2673 | if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
| 2674 | return 0; | |
| 2675 | prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); | |
| 2676 | if (!prefix) | |
| 2677 | return 0; | |
| 2678 | if (prefix->name == poolStart(&dtd.pool)) | |
| 2679 | poolFinish(&dtd.pool); | |
| 2680 | else | |
| 2681 | poolDiscard(&dtd.pool); | |
| 2682 | elementType->prefix = prefix; | |
| 2683 | ||
| 2684 | } | |
| 2685 | } | |
| 2686 | return 1; | |
| 2687 | } | |
| 2688 | ||
| 2689 | static ATTRIBUTE_ID * | |
| 2690 | getAttributeId(XML_Parser parser, const ENCODING *enc, const char *start, const char *end) | |
| 2691 | { | |
| 2692 | ATTRIBUTE_ID *id; | |
| 2693 | const XML_Char *name; | |
| 2694 | if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
| 2695 | return 0; | |
| 2696 | name = poolStoreString(&dtd.pool, enc, start, end); | |
| 2697 | if (!name) | |
| 2698 | return 0; | |
| 2699 | ++name; | |
| 2700 | id = (ATTRIBUTE_ID *)lookup(&dtd.attributeIds, name, sizeof(ATTRIBUTE_ID)); | |
| 2701 | if (!id) | |
| 2702 | return 0; | |
| 2703 | if (id->name != name) | |
| 2704 | poolDiscard(&dtd.pool); | |
| 2705 | else { | |
| 2706 | poolFinish(&dtd.pool); | |
| 2707 | if (!ns) | |
| 2708 | ; | |
| 2709 | else if (name[0] == 'x' | |
| 2710 | && name[1] == 'm' | |
| 2711 | && name[2] == 'l' | |
| 2712 | && name[3] == 'n' | |
| 2713 | && name[4] == 's' | |
| 2714 | && (name[5] == XML_T('\0') || name[5] == XML_T(':'))) { | |
| 2715 | if (name[5] == '\0') | |
| 2716 | id->prefix = &dtd.defaultPrefix; | |
| 2717 | else | |
| 2718 | id->prefix = (PREFIX *)lookup(&dtd.prefixes, name + 6, sizeof(PREFIX)); | |
| 2719 | id->xmlns = 1; | |
| 2720 | } | |
| 2721 | else { | |
| 2722 | int i; | |
| 2723 | for (i = 0; name[i]; i++) { | |
| 2724 | if (name[i] == XML_T(':')) { | |
| 2725 | int j; | |
| 2726 | for (j = 0; j < i; j++) { | |
| 2727 | if (!poolAppendChar(&dtd.pool, name[j])) | |
| 2728 | return 0; | |
| 2729 | } | |
| 2730 | if (!poolAppendChar(&dtd.pool, XML_T('\0'))) | |
| 2731 | return 0; | |
| 2732 | id->prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX)); | |
| 2733 | if (id->prefix->name == poolStart(&dtd.pool)) | |
| 2734 | poolFinish(&dtd.pool); | |
| 2735 | else | |
| 2736 | poolDiscard(&dtd.pool); | |
| 2737 | break; | |
| 2738 | } | |
| 2739 | } | |
| 2740 | } | |
| 2741 | } | |
| 2742 | return id; | |
| 2743 | } | |
| 2744 | ||
| 2745 | #define CONTEXT_SEP XML_T('\f') | |
| 2746 | ||
| 2747 | static | |
| 2748 | const XML_Char *getContext(XML_Parser parser) | |
| 2749 | { | |
| 2750 | HASH_TABLE_ITER iter; | |
| 2751 | int needSep = 0; | |
| 2752 | ||
| 2753 | if (dtd.defaultPrefix.binding) { | |
| 2754 | int i; | |
| 2755 | int len; | |
| 2756 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
| 2757 | return 0; | |
| 2758 | len = dtd.defaultPrefix.binding->uriLen; | |
| 2759 | if (namespaceSeparator != XML_T('\0')) | |
| 2760 | len--; | |
| 2761 | for (i = 0; i < len; i++) | |
| 2762 | if (!poolAppendChar(&tempPool, dtd.defaultPrefix.binding->uri[i])) | |
| 2763 | return 0; | |
| 2764 | needSep = 1; | |
| 2765 | } | |
| 2766 | ||
| 2767 | hashTableIterInit(&iter, &(dtd.prefixes)); | |
| 2768 | for (;;) { | |
| 2769 | int i; | |
| 2770 | int len; | |
| 2771 | const XML_Char *s; | |
| 2772 | PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter); | |
| 2773 | if (!prefix) | |
| 2774 | break; | |
| 2775 | if (!prefix->binding) | |
| 2776 | continue; | |
| 2777 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
| 2778 | return 0; | |
| 2779 | for (s = prefix->name; *s; s++) | |
| 2780 | if (!poolAppendChar(&tempPool, *s)) | |
| 2781 | return 0; | |
| 2782 | if (!poolAppendChar(&tempPool, XML_T('='))) | |
| 2783 | return 0; | |
| 2784 | len = prefix->binding->uriLen; | |
| 2785 | if (namespaceSeparator != XML_T('\0')) | |
| 2786 | len--; | |
| 2787 | for (i = 0; i < len; i++) | |
| 2788 | if (!poolAppendChar(&tempPool, prefix->binding->uri[i])) | |
| 2789 | return 0; | |
| 2790 | needSep = 1; | |
| 2791 | } | |
| 2792 | ||
| 2793 | ||
| 2794 | hashTableIterInit(&iter, &(dtd.generalEntities)); | |
| 2795 | for (;;) { | |
| 2796 | const XML_Char *s; | |
| 2797 | ENTITY *e = (ENTITY *)hashTableIterNext(&iter); | |
| 2798 | if (!e) | |
| 2799 | break; | |
| 2800 | if (!e->open) | |
| 2801 | continue; | |
| 2802 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP)) | |
| 2803 | return 0; | |
| 2804 | for (s = e->name; *s; s++) | |
| 2805 | if (!poolAppendChar(&tempPool, *s)) | |
| 2806 | return 0; | |
| 2807 | needSep = 1; | |
| 2808 | } | |
| 2809 | ||
| 2810 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
| 2811 | return 0; | |
| 2812 | return tempPool.start; | |
| 2813 | } | |
| 2814 | ||
| 2815 | static | |
| 2816 | int setContext(XML_Parser parser, const XML_Char *context) | |
| 2817 | { | |
| 2818 | const XML_Char *s = context; | |
| 2819 | ||
| 2820 | while (*context != XML_T('\0')) { | |
| 2821 | if (*s == CONTEXT_SEP || *s == XML_T('\0')) { | |
| 2822 | ENTITY *e; | |
| 2823 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
| 2824 | return 0; | |
| 2825 | e = (ENTITY *)lookup(&dtd.generalEntities, poolStart(&tempPool), 0); | |
| 2826 | if (e) | |
| 2827 | e->open = 1; | |
| 2828 | if (*s != XML_T('\0')) | |
| 2829 | s++; | |
| 2830 | context = s; | |
| 2831 | poolDiscard(&tempPool); | |
| 2832 | } | |
| 2833 | else if (*s == '=') { | |
| 2834 | PREFIX *prefix; | |
| 2835 | if (poolLength(&tempPool) == 0) | |
| 2836 | prefix = &dtd.defaultPrefix; | |
| 2837 | else { | |
| 2838 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
| 2839 | return 0; | |
| 2840 | prefix = (PREFIX *)lookup(&dtd.prefixes, poolStart(&tempPool), sizeof(PREFIX)); | |
| 2841 | if (!prefix) | |
| 2842 | return 0; | |
| 2843 | if (prefix->name == poolStart(&tempPool)) | |
| 2844 | poolFinish(&tempPool); | |
| 2845 | else | |
| 2846 | poolDiscard(&tempPool); | |
| 2847 | } | |
| 2848 | for (context = s + 1; *context != CONTEXT_SEP && *context != XML_T('\0'); context++) | |
| 2849 | if (!poolAppendChar(&tempPool, *context)) | |
| 2850 | return 0; | |
| 2851 | if (!poolAppendChar(&tempPool, XML_T('\0'))) | |
| 2852 | return 0; | |
| 2853 | if (!addBinding(parser, prefix, 0, poolStart(&tempPool), &inheritedBindings)) | |
| 2854 | return 0; | |
| 2855 | poolDiscard(&tempPool); | |
| 2856 | if (*context != XML_T('\0')) | |
| 2857 | ++context; | |
| 2858 | s = context; | |
| 2859 | } | |
| 2860 | else { | |
| 2861 | if (!poolAppendChar(&tempPool, *s)) | |
| 2862 | return 0; | |
| 2863 | s++; | |
| 2864 | } | |
| 2865 | } | |
| 2866 | return 1; | |
| 2867 | } | |
| 2868 | ||
| 2869 | ||
| 2870 | static | |
| 2871 | void normalizePublicId(XML_Char *publicId) | |
| 2872 | { | |
| 2873 | XML_Char *p = publicId; | |
| 2874 | XML_Char *s; | |
| 2875 | for (s = publicId; *s; s++) { | |
| 2876 | switch (*s) { | |
| 2877 | case 0x20: | |
| 2878 | case 0xD: | |
| 2879 | case 0xA: | |
| 2880 | if (p != publicId && p[-1] != 0x20) | |
| 2881 | *p++ = 0x20; | |
| 2882 | break; | |
| 2883 | default: | |
| 2884 | *p++ = *s; | |
| 2885 | } | |
| 2886 | } | |
| 2887 | if (p != publicId && p[-1] == 0x20) | |
| 2888 | --p; | |
| 2889 | *p = XML_T('\0'); | |
| 2890 | } | |
| 2891 | ||
| 2892 | static int dtdInit(DTD *p) | |
| 2893 | { | |
| 2894 | poolInit(&(p->pool)); | |
| 2895 | hashTableInit(&(p->generalEntities)); | |
| 2896 | hashTableInit(&(p->elementTypes)); | |
| 2897 | hashTableInit(&(p->attributeIds)); | |
| 2898 | hashTableInit(&(p->prefixes)); | |
| 2899 | p->complete = 1; | |
| 2900 | p->standalone = 0; | |
| 2901 | p->base = 0; | |
| 2902 | p->defaultPrefix.name = 0; | |
| 2903 | p->defaultPrefix.binding = 0; | |
| 2904 | return 1; | |
| 2905 | } | |
| 2906 | ||
| 2907 | static void dtdDestroy(DTD *p) | |
| 2908 | { | |
| 2909 | HASH_TABLE_ITER iter; | |
| 2910 | hashTableIterInit(&iter, &(p->elementTypes)); | |
| 2911 | for (;;) { | |
| 2912 | ELEMENT_TYPE *e = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
| 2913 | if (!e) | |
| 2914 | break; | |
| 2915 | if (e->allocDefaultAtts != 0) | |
| 2916 | free(e->defaultAtts); | |
| 2917 | } | |
| 2918 | hashTableDestroy(&(p->generalEntities)); | |
| 2919 | hashTableDestroy(&(p->elementTypes)); | |
| 2920 | hashTableDestroy(&(p->attributeIds)); | |
| 2921 | hashTableDestroy(&(p->prefixes)); | |
| 2922 | poolDestroy(&(p->pool)); | |
| 2923 | } | |
| 2924 | ||
| 2925 | /* Do a deep copy of the DTD. Return 0 for out of memory; non-zero otherwise. | |
| 2926 | The new DTD has already been initialized. */ | |
| 2927 | ||
| 2928 | static int dtdCopy(DTD *newDtd, const DTD *oldDtd) | |
| 2929 | { | |
| 2930 | HASH_TABLE_ITER iter; | |
| 2931 | ||
| 2932 | if (oldDtd->base) { | |
| 2933 | const XML_Char *tem = poolCopyString(&(newDtd->pool), oldDtd->base); | |
| 2934 | if (!tem) | |
| 2935 | return 0; | |
| 2936 | newDtd->base = tem; | |
| 2937 | } | |
| 2938 | ||
| 2939 | /* Copy the prefix table. */ | |
| 2940 | ||
| 2941 | hashTableIterInit(&iter, &(oldDtd->prefixes)); | |
| 2942 | for (;;) { | |
| 2943 | const XML_Char *name; | |
| 2944 | const PREFIX *oldP = (PREFIX *)hashTableIterNext(&iter); | |
| 2945 | if (!oldP) | |
| 2946 | break; | |
| 2947 | name = poolCopyString(&(newDtd->pool), oldP->name); | |
| 2948 | if (!name) | |
| 2949 | return 0; | |
| 2950 | if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX))) | |
| 2951 | return 0; | |
| 2952 | } | |
| 2953 | ||
| 2954 | hashTableIterInit(&iter, &(oldDtd->attributeIds)); | |
| 2955 | ||
| 2956 | /* Copy the attribute id table. */ | |
| 2957 | ||
| 2958 | for (;;) { | |
| 2959 | ATTRIBUTE_ID *newA; | |
| 2960 | const XML_Char *name; | |
| 2961 | const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *)hashTableIterNext(&iter); | |
| 2962 | ||
| 2963 | if (!oldA) | |
| 2964 | break; | |
| 2965 | /* Remember to allocate the scratch byte before the name. */ | |
| 2966 | if (!poolAppendChar(&(newDtd->pool), XML_T('\0'))) | |
| 2967 | return 0; | |
| 2968 | name = poolCopyString(&(newDtd->pool), oldA->name); | |
| 2969 | if (!name) | |
| 2970 | return 0; | |
| 2971 | ++name; | |
| 2972 | newA = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), name, sizeof(ATTRIBUTE_ID)); | |
| 2973 | if (!newA) | |
| 2974 | return 0; | |
| 2975 | newA->maybeTokenized = oldA->maybeTokenized; | |
| 2976 | if (oldA->prefix) { | |
| 2977 | newA->xmlns = oldA->xmlns; | |
| 2978 | if (oldA->prefix == &oldDtd->defaultPrefix) | |
| 2979 | newA->prefix = &newDtd->defaultPrefix; | |
| 2980 | else | |
| 2981 | newA->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldA->prefix->name, 0); | |
| 2982 | } | |
| 2983 | } | |
| 2984 | ||
| 2985 | /* Copy the element type table. */ | |
| 2986 | ||
| 2987 | hashTableIterInit(&iter, &(oldDtd->elementTypes)); | |
| 2988 | ||
| 2989 | for (;;) { | |
| 2990 | int i; | |
| 2991 | ELEMENT_TYPE *newE; | |
| 2992 | const XML_Char *name; | |
| 2993 | const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *)hashTableIterNext(&iter); | |
| 2994 | if (!oldE) | |
| 2995 | break; | |
| 2996 | name = poolCopyString(&(newDtd->pool), oldE->name); | |
| 2997 | if (!name) | |
| 2998 | return 0; | |
| 2999 | newE = (ELEMENT_TYPE *)lookup(&(newDtd->elementTypes), name, sizeof(ELEMENT_TYPE)); | |
| 3000 | if (!newE) | |
| 3001 | return 0; | |
| 3002 | if (oldE->nDefaultAtts) { | |
| 3003 | newE->defaultAtts = (DEFAULT_ATTRIBUTE *)malloc(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE)); | |
| 3004 | if (!newE->defaultAtts) | |
| 3005 | return 0; | |
| 3006 | } | |
| 3007 | newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts; | |
| 3008 | if (oldE->prefix) | |
| 3009 | newE->prefix = (PREFIX *)lookup(&(newDtd->prefixes), oldE->prefix->name, 0); | |
| 3010 | for (i = 0; i < newE->nDefaultAtts; i++) { | |
| 3011 | newE->defaultAtts[i].id = (ATTRIBUTE_ID *)lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0); | |
| 3012 | newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata; | |
| 3013 | if (oldE->defaultAtts[i].value) { | |
| 3014 | newE->defaultAtts[i].value = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value); | |
| 3015 | if (!newE->defaultAtts[i].value) | |
| 3016 | return 0; | |
| 3017 | } | |
| 3018 | else | |
| 3019 | newE->defaultAtts[i].value = 0; | |
| 3020 | } | |
| 3021 | } | |
| 3022 | ||
| 3023 | /* Copy the entity table. */ | |
| 3024 | ||
| 3025 | hashTableIterInit(&iter, &(oldDtd->generalEntities)); | |
| 3026 | ||
| 3027 | for (;;) { | |
| 3028 | ENTITY *newE; | |
| 3029 | const XML_Char *name; | |
| 3030 | const ENTITY *oldE = (ENTITY *)hashTableIterNext(&iter); | |
| 3031 | if (!oldE) | |
| 3032 | break; | |
| 3033 | name = poolCopyString(&(newDtd->pool), oldE->name); | |
| 3034 | if (!name) | |
| 3035 | return 0; | |
| 3036 | newE = (ENTITY *)lookup(&(newDtd->generalEntities), name, sizeof(ENTITY)); | |
| 3037 | if (!newE) | |
| 3038 | return 0; | |
| 3039 | if (oldE->systemId) { | |
| 3040 | const XML_Char *tem = poolCopyString(&(newDtd->pool), oldE->systemId); | |
| 3041 | if (!tem) | |
| 3042 | return 0; | |
| 3043 | newE->systemId = tem; | |
| 3044 | if (oldE->base) { | |
| 3045 | if (oldE->base == oldDtd->base) | |
| 3046 | newE->base = newDtd->base; | |
| 3047 | tem = poolCopyString(&(newDtd->pool), oldE->base); | |
| 3048 | if (!tem) | |
| 3049 | return 0; | |
| 3050 | newE->base = tem; | |
| 3051 | } | |
| 3052 | } | |
| 3053 | else { | |
| 3054 | const XML_Char *tem = poolCopyStringN(&(newDtd->pool), oldE->textPtr, oldE->textLen); | |
| 3055 | if (!tem) | |
| 3056 | return 0; | |
| 3057 | newE->textPtr = tem; | |
| 3058 | newE->textLen = oldE->textLen; | |
| 3059 | } | |
| 3060 | if (oldE->notation) { | |
| 3061 | const XML_Char *tem = poolCopyString(&(newDtd->pool), oldE->notation); | |
| 3062 | if (!tem) | |
| 3063 | return 0; | |
| 3064 | newE->notation = tem; | |
| 3065 | } | |
| 3066 | } | |
| 3067 | ||
| 3068 | newDtd->complete = oldDtd->complete; | |
| 3069 | newDtd->standalone = oldDtd->standalone; | |
| 3070 | return 1; | |
| 3071 | } | |
| 3072 | ||
| 3073 | static | |
| 3074 | void poolInit(STRING_POOL *pool) | |
| 3075 | { | |
| 3076 | pool->blocks = 0; | |
| 3077 | pool->freeBlocks = 0; | |
| 3078 | pool->start = 0; | |
| 3079 | pool->ptr = 0; | |
| 3080 | pool->end = 0; | |
| 3081 | } | |
| 3082 | ||
| 3083 | static | |
| 3084 | void poolClear(STRING_POOL *pool) | |
| 3085 | { | |
| 3086 | if (!pool->freeBlocks) | |
| 3087 | pool->freeBlocks = pool->blocks; | |
| 3088 | else { | |
| 3089 | BLOCK *p = pool->blocks; | |
| 3090 | while (p) { | |
| 3091 | BLOCK *tem = p->next; | |
| 3092 | p->next = pool->freeBlocks; | |
| 3093 | pool->freeBlocks = p; | |
| 3094 | p = tem; | |
| 3095 | } | |
| 3096 | } | |
| 3097 | pool->blocks = 0; | |
| 3098 | pool->start = 0; | |
| 3099 | pool->ptr = 0; | |
| 3100 | pool->end = 0; | |
| 3101 | } | |
| 3102 | ||
| 3103 | static | |
| 3104 | void poolDestroy(STRING_POOL *pool) | |
| 3105 | { | |
| 3106 | BLOCK *p = pool->blocks; | |
| 3107 | while (p) { | |
| 3108 | BLOCK *tem = p->next; | |
| 3109 | free(p); | |
| 3110 | p = tem; | |
| 3111 | } | |
| 3112 | pool->blocks = 0; | |
| 3113 | p = pool->freeBlocks; | |
| 3114 | while (p) { | |
| 3115 | BLOCK *tem = p->next; | |
| 3116 | free(p); | |
| 3117 | p = tem; | |
| 3118 | } | |
| 3119 | pool->freeBlocks = 0; | |
| 3120 | pool->ptr = 0; | |
| 3121 | pool->start = 0; | |
| 3122 | pool->end = 0; | |
| 3123 | } | |
| 3124 | ||
| 3125 | static | |
| 3126 | XML_Char *poolAppend(STRING_POOL *pool, const ENCODING *enc, | |
| 3127 | const char *ptr, const char *end) | |
| 3128 | { | |
| 3129 | if (!pool->ptr && !poolGrow(pool)) | |
| 3130 | return 0; | |
| 3131 | for (;;) { | |
| 3132 | XmlConvert(enc, &ptr, end, (ICHAR **)&(pool->ptr), (ICHAR *)pool->end); | |
| 3133 | if (ptr == end) | |
| 3134 | break; | |
| 3135 | if (!poolGrow(pool)) | |
| 3136 | return 0; | |
| 3137 | } | |
| 3138 | return pool->start; | |
| 3139 | } | |
| 3140 | ||
| 3141 | static const XML_Char *poolCopyString(STRING_POOL *pool, const XML_Char *s) | |
| 3142 | { | |
| 3143 | do { | |
| 3144 | if (!poolAppendChar(pool, *s)) | |
| 3145 | return 0; | |
| 3146 | } while (*s++); | |
| 3147 | s = pool->start; | |
| 3148 | poolFinish(pool); | |
| 3149 | return s; | |
| 3150 | } | |
| 3151 | ||
| 3152 | static const XML_Char *poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n) | |
| 3153 | { | |
| 3154 | if (!pool->ptr && !poolGrow(pool)) | |
| 3155 | return 0; | |
| 3156 | for (; n > 0; --n, s++) { | |
| 3157 | if (!poolAppendChar(pool, *s)) | |
| 3158 | return 0; | |
| 3159 | ||
| 3160 | } | |
| 3161 | s = pool->start; | |
| 3162 | poolFinish(pool); | |
| 3163 | return s; | |
| 3164 | } | |
| 3165 | ||
| 3166 | static | |
| 3167 | XML_Char *poolStoreString(STRING_POOL *pool, const ENCODING *enc, | |
| 3168 | const char *ptr, const char *end) | |
| 3169 | { | |
| 3170 | if (!poolAppend(pool, enc, ptr, end)) | |
| 3171 | return 0; | |
| 3172 | if (pool->ptr == pool->end && !poolGrow(pool)) | |
| 3173 | return 0; | |
| 3174 | *(pool->ptr)++ = 0; | |
| 3175 | return pool->start; | |
| 3176 | } | |
| 3177 | ||
| 3178 | static | |
| 3179 | int poolGrow(STRING_POOL *pool) | |
| 3180 | { | |
| 3181 | if (pool->freeBlocks) { | |
| 3182 | if (pool->start == 0) { | |
| 3183 | pool->blocks = pool->freeBlocks; | |
| 3184 | pool->freeBlocks = pool->freeBlocks->next; | |
| 3185 | pool->blocks->next = 0; | |
| 3186 | pool->start = pool->blocks->s; | |
| 3187 | pool->end = pool->start + pool->blocks->size; | |
| 3188 | pool->ptr = pool->start; | |
| 3189 | return 1; | |
| 3190 | } | |
| 3191 | if (pool->end - pool->start < pool->freeBlocks->size) { | |
| 3192 | BLOCK *tem = pool->freeBlocks->next; | |
| 3193 | pool->freeBlocks->next = pool->blocks; | |
| 3194 | pool->blocks = pool->freeBlocks; | |
| 3195 | pool->freeBlocks = tem; | |
| 3196 | memcpy(pool->blocks->s, pool->start, (pool->end - pool->start) * sizeof(XML_Char)); | |
| 3197 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
| 3198 | pool->start = pool->blocks->s; | |
| 3199 | pool->end = pool->start + pool->blocks->size; | |
| 3200 | return 1; | |
| 3201 | } | |
| 3202 | } | |
| 3203 | if (pool->blocks && pool->start == pool->blocks->s) { | |
| 3204 | int blockSize = (pool->end - pool->start)*2; | |
| 3205 | pool->blocks = realloc(pool->blocks, offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); | |
| 3206 | if (!pool->blocks) | |
| 3207 | return 0; | |
| 3208 | pool->blocks->size = blockSize; | |
| 3209 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start); | |
| 3210 | pool->start = pool->blocks->s; | |
| 3211 | pool->end = pool->start + blockSize; | |
| 3212 | } | |
| 3213 | else { | |
| 3214 | BLOCK *tem; | |
| 3215 | int blockSize = pool->end - pool->start; | |
| 3216 | if (blockSize < INIT_BLOCK_SIZE) | |
| 3217 | blockSize = INIT_BLOCK_SIZE; | |
| 3218 | else | |
| 3219 | blockSize *= 2; | |
| 3220 | tem = malloc(offsetof(BLOCK, s) + blockSize * sizeof(XML_Char)); | |
| 3221 | if (!tem) | |
| 3222 | return 0; | |
| 3223 | tem->size = blockSize; | |
| 3224 | tem->next = pool->blocks; | |
| 3225 | pool->blocks = tem; | |
| 3226 | memcpy(tem->s, pool->start, (pool->ptr - pool->start) * sizeof(XML_Char)); | |
| 3227 | pool->ptr = tem->s + (pool->ptr - pool->start); | |
| 3228 | pool->start = tem->s; | |
| 3229 | pool->end = tem->s + blockSize; | |
| 3230 | } | |
| 3231 | return 1; | |
| 3232 | } |