| 1 |
|
|---|
| 2 | /*
|
|---|
| 3 | *@@sourcefile xmlparse.c:
|
|---|
| 4 | * contains the API of the expat XML parser 1.95.2, as released on
|
|---|
| 5 | * http://sourceforge.net/projects/expat/ on July 26, 2001.
|
|---|
| 6 | *
|
|---|
| 7 | * V1.95.1 was ported to and integrated with the xwphelpers
|
|---|
| 8 | * for V0.9.9 (2001-02-10) [umoeller]. V1.95.2 was re-ported
|
|---|
| 9 | * with V0.9.14 (2001-08-09) [umoeller].
|
|---|
| 10 | *
|
|---|
| 11 | * See xml.c for a general introduction to @XML support in the
|
|---|
| 12 | * xwphelpers.
|
|---|
| 13 | *
|
|---|
| 14 | * Expat is a library, written in C, for parsing XML @documents. It's
|
|---|
| 15 | * the underlying XML parser for the open source Mozilla project,
|
|---|
| 16 | * perl's XML::Parser, and other open-source XML parsers.
|
|---|
| 17 | *
|
|---|
| 18 | * Expat was written by James Clark, who also wrote groff (an nroff
|
|---|
| 19 | * look-alike), Jade (an implemention of ISO's DSSSL stylesheet
|
|---|
| 20 | * language for SGML), XP (a Java XML parser package), XT (a Java
|
|---|
| 21 | * XSL engine). He was also the technical lead on the XML Working
|
|---|
| 22 | * Group at W3 that produced the XML specification.
|
|---|
| 23 | *
|
|---|
| 24 | * Most of this documentation is the original documentation
|
|---|
| 25 | * that comes with expat as an HTML file. I (umoeller)
|
|---|
| 26 | * have integrated the docs into the sources and added a
|
|---|
| 27 | * couple of remarks while trying to implement parsers.
|
|---|
| 28 | *
|
|---|
| 29 | * See XML_ParserCreate for instructions how to parse XML.
|
|---|
| 30 | *
|
|---|
| 31 | * Note that expat is a non-validating XML processor. Validation
|
|---|
| 32 | * is supported by the top layer of xwphelpers XML support. See
|
|---|
| 33 | * xml.c.
|
|---|
| 34 | *
|
|---|
| 35 | *@@added V0.9.9 (2001-02-10) [umoeller]
|
|---|
| 36 | *@@header "expat\expat.h"
|
|---|
| 37 | */
|
|---|
| 38 |
|
|---|
| 39 | /*
|
|---|
| 40 | * Copyright (C) 2001 Ulrich Mller.
|
|---|
| 41 | * Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd.
|
|---|
| 42 | * and Clark Cooper.
|
|---|
| 43 | *
|
|---|
| 44 | * Permission is hereby granted, free of charge, to any person obtaining
|
|---|
| 45 | * a copy of this software and associated documentation files (the
|
|---|
| 46 | * "Software"), to deal in the Software without restriction, including
|
|---|
| 47 | * without limitation the rights to use, copy, modify, merge, publish,
|
|---|
| 48 | * distribute, sublicense, and/or sell copies of the Software, and to
|
|---|
| 49 | * permit persons to whom the Software is furnished to do so, subject to
|
|---|
| 50 | * the following conditions:
|
|---|
| 51 | *
|
|---|
| 52 | * The above copyright notice and this permission notice shall be included
|
|---|
| 53 | * in all copies or substantial portions of the Software.
|
|---|
| 54 | *
|
|---|
| 55 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|---|
| 56 | * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|---|
| 57 | * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
|
|---|
| 58 | * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
|
|---|
| 59 | * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
|
|---|
| 60 | * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
|
|---|
| 61 | * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
|---|
| 62 | */
|
|---|
| 63 |
|
|---|
| 64 | /*
|
|---|
| 65 | *@@category: Helpers\XML\expat
|
|---|
| 66 | * expat XML parser. See xmlparse.c.
|
|---|
| 67 | */
|
|---|
| 68 |
|
|---|
| 69 | #include "setup.h"
|
|---|
| 70 |
|
|---|
| 71 | #pragma info(norea, nogen)
|
|---|
| 72 | // disable "statement unreachable" and "missing break statement"
|
|---|
| 73 | // this code generates those options HEAVILY
|
|---|
| 74 |
|
|---|
| 75 |
|
|---|
| 76 | /* #ifdef COMPILED_FROM_DSP
|
|---|
| 77 | * # include "winconfig.h"
|
|---|
| 78 | * # define XMLPARSEAPI(type) __declspec(dllexport) type __cdecl
|
|---|
| 79 | * # include "expat.h"
|
|---|
| 80 | * # undef XMLPARSEAPI
|
|---|
| 81 | * #else
|
|---|
| 82 | * #include <config.h> */
|
|---|
| 83 |
|
|---|
| 84 | #ifdef __declspec
|
|---|
| 85 | #define XMLPARSEAPI(type) __declspec(dllexport) type __cdecl
|
|---|
| 86 | #endif
|
|---|
| 87 |
|
|---|
| 88 | #include "expat\expat.h"
|
|---|
| 89 |
|
|---|
| 90 | #ifdef __declspec
|
|---|
| 91 | #undef XMLPARSEAPI
|
|---|
| 92 | #endif
|
|---|
| 93 | // #endif /* ndef COMPILED_FROM_DSP */
|
|---|
| 94 |
|
|---|
| 95 | #include <stddef.h>
|
|---|
| 96 | #include <string.h>
|
|---|
| 97 |
|
|---|
| 98 | #ifdef XML_UNICODE
|
|---|
| 99 | #define XML_ENCODE_MAX XML_UTF16_ENCODE_MAX
|
|---|
| 100 | #define XmlConvert XmlUtf16Convert
|
|---|
| 101 | #define XmlGetInternalEncoding XmlGetUtf16InternalEncoding
|
|---|
| 102 | #define XmlGetInternalEncodingNS XmlGetUtf16InternalEncodingNS
|
|---|
| 103 | #define XmlEncode XmlUtf16Encode
|
|---|
| 104 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf16 || (((unsigned long)s) & 1))
|
|---|
| 105 | typedef unsigned short ICHAR;
|
|---|
| 106 |
|
|---|
| 107 | #else
|
|---|
| 108 | #define XML_ENCODE_MAX XML_UTF8_ENCODE_MAX
|
|---|
| 109 | #define XmlConvert XmlUtf8Convert
|
|---|
| 110 | #define XmlGetInternalEncoding XmlGetUtf8InternalEncoding
|
|---|
| 111 | #define XmlGetInternalEncodingNS XmlGetUtf8InternalEncodingNS
|
|---|
| 112 | #define XmlEncode XmlUtf8Encode
|
|---|
| 113 | #define MUST_CONVERT(enc, s) (!(enc)->isUtf8)
|
|---|
| 114 | typedef char ICHAR;
|
|---|
| 115 |
|
|---|
| 116 | #endif
|
|---|
| 117 |
|
|---|
| 118 |
|
|---|
| 119 | #ifndef XML_NS
|
|---|
| 120 |
|
|---|
| 121 | #define XmlInitEncodingNS XmlInitEncoding
|
|---|
| 122 | #define XmlInitUnknownEncodingNS XmlInitUnknownEncoding
|
|---|
| 123 | #undef XmlGetInternalEncodingNS
|
|---|
| 124 | #define XmlGetInternalEncodingNS XmlGetInternalEncoding
|
|---|
| 125 | #define XmlParseXmlDeclNS XmlParseXmlDecl
|
|---|
| 126 |
|
|---|
| 127 | #endif
|
|---|
| 128 |
|
|---|
| 129 | #ifdef XML_UNICODE_WCHAR_T
|
|---|
| 130 | #define XML_T(x) L ## x
|
|---|
| 131 | #else
|
|---|
| 132 | #define XML_T(x) x
|
|---|
| 133 | #endif
|
|---|
| 134 |
|
|---|
| 135 | /* Round up n to be a multiple of sz, where sz is a power of 2. */
|
|---|
| 136 | #define ROUND_UP(n, sz) (((n) + ((sz) - 1)) & ~((sz) - 1))
|
|---|
| 137 |
|
|---|
| 138 | #include "expat\xmltok.h"
|
|---|
| 139 | #include "expat\xmlrole.h"
|
|---|
| 140 |
|
|---|
| 141 | typedef const XML_Char *KEY;
|
|---|
| 142 |
|
|---|
| 143 | typedef struct
|
|---|
| 144 | {
|
|---|
| 145 | KEY name;
|
|---|
| 146 | }
|
|---|
| 147 | NAMED;
|
|---|
| 148 |
|
|---|
| 149 | typedef struct
|
|---|
| 150 | {
|
|---|
| 151 | NAMED **v;
|
|---|
| 152 | size_t size;
|
|---|
| 153 | size_t used;
|
|---|
| 154 | size_t usedLim;
|
|---|
| 155 | XML_Memory_Handling_Suite *mem;
|
|---|
| 156 | }
|
|---|
| 157 | HASH_TABLE;
|
|---|
| 158 |
|
|---|
| 159 | typedef struct
|
|---|
| 160 | {
|
|---|
| 161 | NAMED **p;
|
|---|
| 162 | NAMED **end;
|
|---|
| 163 | }
|
|---|
| 164 | HASH_TABLE_ITER;
|
|---|
| 165 |
|
|---|
| 166 | #define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */
|
|---|
| 167 | #define INIT_DATA_BUF_SIZE 1024
|
|---|
| 168 | #define INIT_ATTS_SIZE 16
|
|---|
| 169 | #define INIT_BLOCK_SIZE 1024
|
|---|
| 170 | #define INIT_BUFFER_SIZE 1024
|
|---|
| 171 |
|
|---|
| 172 | #define EXPAND_SPARE 24
|
|---|
| 173 |
|
|---|
| 174 | typedef struct binding
|
|---|
| 175 | {
|
|---|
| 176 | struct prefix *prefix;
|
|---|
| 177 | struct binding *nextTagBinding;
|
|---|
| 178 | struct binding *prevPrefixBinding;
|
|---|
| 179 | const struct attribute_id *attId;
|
|---|
| 180 | XML_Char *uri;
|
|---|
| 181 | int uriLen;
|
|---|
| 182 | int uriAlloc;
|
|---|
| 183 | }
|
|---|
| 184 | BINDING;
|
|---|
| 185 |
|
|---|
| 186 | typedef struct prefix
|
|---|
| 187 | {
|
|---|
| 188 | const XML_Char *name;
|
|---|
| 189 | BINDING *binding;
|
|---|
| 190 | }
|
|---|
| 191 | PREFIX;
|
|---|
| 192 |
|
|---|
| 193 | typedef struct
|
|---|
| 194 | {
|
|---|
| 195 | const XML_Char *str;
|
|---|
| 196 | const XML_Char *localPart;
|
|---|
| 197 | int uriLen;
|
|---|
| 198 | }
|
|---|
| 199 | TAG_NAME;
|
|---|
| 200 |
|
|---|
| 201 | typedef struct tag
|
|---|
| 202 | {
|
|---|
| 203 | struct tag *parent;
|
|---|
| 204 | const char *rawName;
|
|---|
| 205 | int rawNameLength;
|
|---|
| 206 | TAG_NAME name;
|
|---|
| 207 | char *buf;
|
|---|
| 208 | char *bufEnd;
|
|---|
| 209 | BINDING *bindings;
|
|---|
| 210 | }
|
|---|
| 211 | TAG;
|
|---|
| 212 |
|
|---|
| 213 | typedef struct
|
|---|
| 214 | {
|
|---|
| 215 | const XML_Char *name;
|
|---|
| 216 | const XML_Char *textPtr;
|
|---|
| 217 | int textLen;
|
|---|
| 218 | const XML_Char *systemId;
|
|---|
| 219 | const XML_Char *base;
|
|---|
| 220 | const XML_Char *publicId;
|
|---|
| 221 | const XML_Char *notation;
|
|---|
| 222 | char open;
|
|---|
| 223 | char is_param;
|
|---|
| 224 | }
|
|---|
| 225 | ENTITY;
|
|---|
| 226 |
|
|---|
| 227 | typedef struct
|
|---|
| 228 | {
|
|---|
| 229 | enum XML_Content_Type type;
|
|---|
| 230 | enum XML_Content_Quant quant;
|
|---|
| 231 | const XML_Char *name;
|
|---|
| 232 | int firstchild;
|
|---|
| 233 | int lastchild;
|
|---|
| 234 | int childcnt;
|
|---|
| 235 | int nextsib;
|
|---|
| 236 | }
|
|---|
| 237 | CONTENT_SCAFFOLD;
|
|---|
| 238 |
|
|---|
| 239 | typedef struct block
|
|---|
| 240 | {
|
|---|
| 241 | struct block *next;
|
|---|
| 242 | int size;
|
|---|
| 243 | XML_Char s[1];
|
|---|
| 244 | }
|
|---|
| 245 | BLOCK;
|
|---|
| 246 |
|
|---|
| 247 | typedef struct
|
|---|
| 248 | {
|
|---|
| 249 | BLOCK *blocks;
|
|---|
| 250 | BLOCK *freeBlocks;
|
|---|
| 251 | const XML_Char *end;
|
|---|
| 252 | XML_Char *ptr;
|
|---|
| 253 | XML_Char *start;
|
|---|
| 254 | XML_Memory_Handling_Suite *mem;
|
|---|
| 255 | }
|
|---|
| 256 | STRING_POOL;
|
|---|
| 257 |
|
|---|
| 258 | /* The XML_Char before the name is used to determine whether
|
|---|
| 259 | * an attribute has been specified. */
|
|---|
| 260 | typedef struct attribute_id
|
|---|
| 261 | {
|
|---|
| 262 | XML_Char *name;
|
|---|
| 263 | PREFIX *prefix;
|
|---|
| 264 | char maybeTokenized;
|
|---|
| 265 | char xmlns;
|
|---|
| 266 | }
|
|---|
| 267 | ATTRIBUTE_ID;
|
|---|
| 268 |
|
|---|
| 269 | typedef struct
|
|---|
| 270 | {
|
|---|
| 271 | const ATTRIBUTE_ID *id;
|
|---|
| 272 | char isCdata;
|
|---|
| 273 | const XML_Char *value;
|
|---|
| 274 | }
|
|---|
| 275 | DEFAULT_ATTRIBUTE;
|
|---|
| 276 |
|
|---|
| 277 | typedef struct
|
|---|
| 278 | {
|
|---|
| 279 | const XML_Char *name;
|
|---|
| 280 | PREFIX *prefix;
|
|---|
| 281 | const ATTRIBUTE_ID *idAtt;
|
|---|
| 282 | int nDefaultAtts;
|
|---|
| 283 | int allocDefaultAtts;
|
|---|
| 284 | DEFAULT_ATTRIBUTE *defaultAtts;
|
|---|
| 285 | }
|
|---|
| 286 | ELEMENT_TYPE;
|
|---|
| 287 |
|
|---|
| 288 | typedef struct
|
|---|
| 289 | {
|
|---|
| 290 | HASH_TABLE generalEntities;
|
|---|
| 291 | HASH_TABLE elementTypes;
|
|---|
| 292 | HASH_TABLE attributeIds;
|
|---|
| 293 | HASH_TABLE prefixes;
|
|---|
| 294 | STRING_POOL pool;
|
|---|
| 295 | int complete;
|
|---|
| 296 | int standalone;
|
|---|
| 297 | #ifdef XML_DTD
|
|---|
| 298 | HASH_TABLE paramEntities;
|
|---|
| 299 | #endif /* XML_DTD */
|
|---|
| 300 | PREFIX defaultPrefix;
|
|---|
| 301 | /* === scaffolding for building content model === */
|
|---|
| 302 | int in_eldecl;
|
|---|
| 303 | CONTENT_SCAFFOLD *scaffold;
|
|---|
| 304 | unsigned contentStringLen;
|
|---|
| 305 | unsigned scaffSize;
|
|---|
| 306 | unsigned scaffCount;
|
|---|
| 307 | int scaffLevel;
|
|---|
| 308 | int *scaffIndex;
|
|---|
| 309 | }
|
|---|
| 310 | DTD;
|
|---|
| 311 |
|
|---|
| 312 | typedef struct open_internal_entity
|
|---|
| 313 | {
|
|---|
| 314 | const char *internalEventPtr;
|
|---|
| 315 | const char *internalEventEndPtr;
|
|---|
| 316 | struct open_internal_entity *next;
|
|---|
| 317 | ENTITY *entity;
|
|---|
| 318 | }
|
|---|
| 319 | OPEN_INTERNAL_ENTITY;
|
|---|
| 320 |
|
|---|
| 321 | typedef XMLERROR Processor(XML_Parser parser,
|
|---|
| 322 | const char *start,
|
|---|
| 323 | const char *end,
|
|---|
| 324 | const char **endPtr);
|
|---|
| 325 |
|
|---|
| 326 | STATIC Processor prologProcessor;
|
|---|
| 327 | STATIC Processor prologInitProcessor;
|
|---|
| 328 | STATIC Processor contentProcessor;
|
|---|
| 329 | STATIC Processor cdataSectionProcessor;
|
|---|
| 330 |
|
|---|
| 331 | #ifdef XML_DTD
|
|---|
| 332 | STATIC Processor ignoreSectionProcessor;
|
|---|
| 333 |
|
|---|
| 334 | #endif /* XML_DTD */
|
|---|
| 335 | STATIC Processor epilogProcessor;
|
|---|
| 336 | STATIC Processor errorProcessor;
|
|---|
| 337 | STATIC Processor externalEntityInitProcessor;
|
|---|
| 338 | STATIC Processor externalEntityInitProcessor2;
|
|---|
| 339 | STATIC Processor externalEntityInitProcessor3;
|
|---|
| 340 | STATIC Processor externalEntityContentProcessor;
|
|---|
| 341 |
|
|---|
| 342 | STATIC XMLERROR
|
|---|
| 343 | handleUnknownEncoding(XML_Parser parser, const XML_Char * encodingName);
|
|---|
| 344 | STATIC XMLERROR
|
|---|
| 345 | processXmlDecl(XML_Parser parser, int isGeneralTextEntity, const char *, const char *);
|
|---|
| 346 | STATIC XMLERROR
|
|---|
| 347 | initializeEncoding(XML_Parser parser);
|
|---|
| 348 | STATIC XMLERROR
|
|---|
| 349 | doProlog(XML_Parser parser, const ENCODING * enc, const char *s,
|
|---|
| 350 | const char *end, int tok, const char *next, const char **nextPtr);
|
|---|
| 351 | STATIC XMLERROR
|
|---|
| 352 | processInternalParamEntity(XML_Parser parser, ENTITY * entity);
|
|---|
| 353 | STATIC XMLERROR
|
|---|
| 354 | doContent(XML_Parser parser, int startTagLevel, const ENCODING * enc,
|
|---|
| 355 | const char *start, const char *end, const char **endPtr);
|
|---|
| 356 | STATIC XMLERROR
|
|---|
| 357 | doCdataSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr);
|
|---|
| 358 |
|
|---|
| 359 | #ifdef XML_DTD
|
|---|
| 360 | STATIC XMLERROR
|
|---|
| 361 | doIgnoreSection(XML_Parser parser, const ENCODING *, const char **startPtr, const char *end, const char **nextPtr);
|
|---|
| 362 |
|
|---|
| 363 | #endif /* XML_DTD */
|
|---|
| 364 | STATIC XMLERROR storeAtts(XML_Parser parser, const ENCODING *, const char *s,
|
|---|
| 365 | TAG_NAME * tagNamePtr, BINDING ** bindingsPtr);
|
|---|
| 366 | STATIC
|
|---|
| 367 | int addBinding(XML_Parser parser, PREFIX * prefix, const ATTRIBUTE_ID * attId, const XML_Char * uri, BINDING ** bindingsPtr);
|
|---|
| 368 |
|
|---|
| 369 | STATIC int
|
|---|
| 370 | defineAttribute(ELEMENT_TYPE * type, ATTRIBUTE_ID *,
|
|---|
| 371 | int isCdata, int isId, const XML_Char * dfltValue,
|
|---|
| 372 | XML_Parser parser);
|
|---|
| 373 |
|
|---|
| 374 | STATIC XMLERROR
|
|---|
| 375 | storeAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *,
|
|---|
| 376 | STRING_POOL *);
|
|---|
| 377 | STATIC XMLERROR
|
|---|
| 378 | appendAttributeValue(XML_Parser parser, const ENCODING *, int isCdata, const char *, const char *,
|
|---|
| 379 | STRING_POOL *);
|
|---|
| 380 | STATIC ATTRIBUTE_ID *
|
|---|
| 381 | getAttributeId(XML_Parser parser, const ENCODING * enc, const char *start, const char *end);
|
|---|
| 382 | STATIC int setElementTypePrefix(XML_Parser parser, ELEMENT_TYPE *);
|
|---|
| 383 | STATIC XMLERROR
|
|---|
| 384 | storeEntityValue(XML_Parser parser, const ENCODING * enc, const char *start, const char *end);
|
|---|
| 385 | STATIC int
|
|---|
| 386 | reportProcessingInstruction(XML_Parser parser, const ENCODING * enc, const char *start, const char *end);
|
|---|
| 387 | STATIC int
|
|---|
| 388 | reportComment(XML_Parser parser, const ENCODING * enc, const char *start, const char *end);
|
|---|
| 389 | STATIC void
|
|---|
| 390 | reportDefault(XML_Parser parser, const ENCODING * enc, const char *start, const char *end);
|
|---|
| 391 |
|
|---|
| 392 | STATIC const XML_Char *getContext(XML_Parser parser);
|
|---|
| 393 | STATIC int setContext(XML_Parser parser, const XML_Char * context);
|
|---|
| 394 | STATIC void normalizePublicId(XML_Char * s);
|
|---|
| 395 | STATIC int dtdInit(DTD *, XML_Parser parser);
|
|---|
| 396 |
|
|---|
| 397 | STATIC void dtdDestroy(DTD *, XML_Parser parser);
|
|---|
| 398 |
|
|---|
| 399 | STATIC int dtdCopy(DTD * newDtd, const DTD * oldDtd, XML_Parser parser);
|
|---|
| 400 |
|
|---|
| 401 | STATIC int copyEntityTable(HASH_TABLE *, STRING_POOL *, const HASH_TABLE *,
|
|---|
| 402 | XML_Parser parser);
|
|---|
| 403 |
|
|---|
| 404 | #ifdef XML_DTD
|
|---|
| 405 | STATIC void dtdSwap(DTD *, DTD *);
|
|---|
| 406 |
|
|---|
| 407 | #endif /* XML_DTD */
|
|---|
| 408 |
|
|---|
| 409 | STATIC NAMED *lookup(HASH_TABLE * table, KEY name, size_t createSize);
|
|---|
| 410 |
|
|---|
| 411 | STATIC void hashTableInit(HASH_TABLE *, XML_Memory_Handling_Suite * ms);
|
|---|
| 412 |
|
|---|
| 413 | STATIC void hashTableDestroy(HASH_TABLE *);
|
|---|
| 414 | STATIC void hashTableIterInit(HASH_TABLE_ITER *, const HASH_TABLE *);
|
|---|
| 415 | STATIC NAMED *hashTableIterNext(HASH_TABLE_ITER *);
|
|---|
| 416 | STATIC void poolInit(STRING_POOL *, XML_Memory_Handling_Suite * ms);
|
|---|
| 417 | STATIC void poolClear(STRING_POOL *);
|
|---|
| 418 | STATIC void poolDestroy(STRING_POOL *);
|
|---|
| 419 | STATIC XML_Char *poolAppend(STRING_POOL *pool,
|
|---|
| 420 | const ENCODING *enc,
|
|---|
| 421 | const char *ptr,
|
|---|
| 422 | const char *end,
|
|---|
| 423 | unsigned long *pulOfs); // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 424 | STATIC XML_Char *poolStoreString(STRING_POOL *pool,
|
|---|
| 425 | const ENCODING *enc,
|
|---|
| 426 | const char *ptr,
|
|---|
| 427 | const char *end,
|
|---|
| 428 | unsigned long *pulOfs); // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 429 |
|
|---|
| 430 | STATIC int poolGrow(STRING_POOL * pool);
|
|---|
| 431 |
|
|---|
| 432 | STATIC int nextScaffoldPart(XML_Parser parser);
|
|---|
| 433 | STATIC XMLCONTENT *build_model(XML_Parser parser);
|
|---|
| 434 |
|
|---|
| 435 | STATIC const XML_Char *poolCopyString(STRING_POOL * pool, const XML_Char * s);
|
|---|
| 436 | STATIC const XML_Char *poolCopyStringN(STRING_POOL * pool, const XML_Char * s, int n);
|
|---|
| 437 | STATIC const XML_Char *poolAppendString(STRING_POOL * pool, const XML_Char * s);
|
|---|
| 438 | STATIC ELEMENT_TYPE *getElementType(XML_Parser Paraser,
|
|---|
| 439 | const ENCODING * enc,
|
|---|
| 440 | const char *ptr,
|
|---|
| 441 | const char *end);
|
|---|
| 442 |
|
|---|
| 443 | #define poolStart(pool) ((pool)->start)
|
|---|
| 444 | #define poolEnd(pool) ((pool)->ptr)
|
|---|
| 445 | #define poolLength(pool) ((pool)->ptr - (pool)->start)
|
|---|
| 446 | #define poolChop(pool) ((void)--(pool->ptr))
|
|---|
| 447 | #define poolLastChar(pool) (((pool)->ptr)[-1])
|
|---|
| 448 | #define poolDiscard(pool) ((pool)->ptr = (pool)->start)
|
|---|
| 449 | #define poolFinish(pool) ((pool)->start = (pool)->ptr)
|
|---|
| 450 | #define poolAppendChar(pool, c) \
|
|---|
| 451 | (((pool)->ptr == (pool)->end && !poolGrow(pool)) \
|
|---|
| 452 | ? 0 \
|
|---|
| 453 | : ((*((pool)->ptr)++ = c), 1))
|
|---|
| 454 |
|
|---|
| 455 | typedef struct
|
|---|
| 456 | {
|
|---|
| 457 | /* The first member must be userData so that the XML_GetUserData macro works. */
|
|---|
| 458 | void *m_userData;
|
|---|
| 459 | void *m_handlerArg;
|
|---|
| 460 | char *m_buffer;
|
|---|
| 461 | XML_Memory_Handling_Suite m_mem;
|
|---|
| 462 | /* first character to be parsed */
|
|---|
| 463 | const char *m_bufferPtr;
|
|---|
| 464 | /* past last character to be parsed */
|
|---|
| 465 | char *m_bufferEnd;
|
|---|
| 466 | /* allocated end of buffer */
|
|---|
| 467 | const char *m_bufferLim;
|
|---|
| 468 | long m_parseEndByteIndex;
|
|---|
| 469 | const char *m_parseEndPtr;
|
|---|
| 470 | XML_Char *m_dataBuf;
|
|---|
| 471 | XML_Char *m_dataBufEnd;
|
|---|
| 472 | XML_StartElementHandler m_startElementHandler;
|
|---|
| 473 | XML_EndElementHandler m_endElementHandler;
|
|---|
| 474 | XML_CharacterDataHandler m_characterDataHandler;
|
|---|
| 475 | XML_ProcessingInstructionHandler m_processingInstructionHandler;
|
|---|
| 476 | XML_CommentHandler m_commentHandler;
|
|---|
| 477 | XML_StartCdataSectionHandler m_startCdataSectionHandler;
|
|---|
| 478 | XML_EndCdataSectionHandler m_endCdataSectionHandler;
|
|---|
| 479 | XML_DefaultHandler m_defaultHandler;
|
|---|
| 480 | XML_StartDoctypeDeclHandler m_startDoctypeDeclHandler;
|
|---|
| 481 | XML_EndDoctypeDeclHandler m_endDoctypeDeclHandler;
|
|---|
| 482 | XML_UnparsedEntityDeclHandler m_unparsedEntityDeclHandler;
|
|---|
| 483 | XML_NotationDeclHandler m_notationDeclHandler;
|
|---|
| 484 | XML_StartNamespaceDeclHandler m_startNamespaceDeclHandler;
|
|---|
| 485 | XML_EndNamespaceDeclHandler m_endNamespaceDeclHandler;
|
|---|
| 486 | XML_NotStandaloneHandler m_notStandaloneHandler;
|
|---|
| 487 | XML_ExternalEntityRefHandler m_externalEntityRefHandler;
|
|---|
| 488 | void *m_externalEntityRefHandlerArg;
|
|---|
| 489 | XML_UnknownEncodingHandler m_unknownEncodingHandler;
|
|---|
| 490 | XML_ElementDeclHandler m_elementDeclHandler;
|
|---|
| 491 | XML_AttlistDeclHandler m_attlistDeclHandler;
|
|---|
| 492 | XML_EntityDeclHandler m_entityDeclHandler;
|
|---|
| 493 | XML_XmlDeclHandler m_xmlDeclHandler;
|
|---|
| 494 | const ENCODING *m_encoding;
|
|---|
| 495 | INIT_ENCODING m_initEncoding;
|
|---|
| 496 | const ENCODING *m_internalEncoding;
|
|---|
| 497 | const XML_Char *m_protocolEncodingName;
|
|---|
| 498 | int m_ns;
|
|---|
| 499 | int m_ns_triplets;
|
|---|
| 500 | void *m_unknownEncodingMem;
|
|---|
| 501 | void *m_unknownEncodingData;
|
|---|
| 502 | void *m_unknownEncodingHandlerData;
|
|---|
| 503 | void (*m_unknownEncodingRelease) (void *);
|
|---|
| 504 | PROLOG_STATE m_prologState;
|
|---|
| 505 | Processor *m_processor;
|
|---|
| 506 | XMLERROR m_errorCode;
|
|---|
| 507 | const char *m_eventPtr;
|
|---|
| 508 | const char *m_eventEndPtr;
|
|---|
| 509 | const char *m_positionPtr;
|
|---|
| 510 | OPEN_INTERNAL_ENTITY *m_openInternalEntities;
|
|---|
| 511 | int m_defaultExpandInternalEntities;
|
|---|
| 512 | int m_tagLevel;
|
|---|
| 513 | ENTITY *m_declEntity;
|
|---|
| 514 | const XML_Char *m_doctypeName;
|
|---|
| 515 | const XML_Char *m_doctypeSysid;
|
|---|
| 516 | const XML_Char *m_doctypePubid;
|
|---|
| 517 | const XML_Char *m_declAttributeType;
|
|---|
| 518 | const XML_Char *m_declNotationName;
|
|---|
| 519 | const XML_Char *m_declNotationPublicId;
|
|---|
| 520 | ELEMENT_TYPE *m_declElementType;
|
|---|
| 521 | ATTRIBUTE_ID *m_declAttributeId;
|
|---|
| 522 | char m_declAttributeIsCdata;
|
|---|
| 523 | char m_declAttributeIsId;
|
|---|
| 524 | DTD m_dtd;
|
|---|
| 525 | const XML_Char *m_curBase;
|
|---|
| 526 | TAG *m_tagStack;
|
|---|
| 527 | TAG *m_freeTagList;
|
|---|
| 528 | BINDING *m_inheritedBindings;
|
|---|
| 529 | BINDING *m_freeBindingList;
|
|---|
| 530 | int m_attsSize;
|
|---|
| 531 | int m_nSpecifiedAtts;
|
|---|
| 532 | int m_idAttIndex;
|
|---|
| 533 | ATTRIBUTE *m_atts;
|
|---|
| 534 | POSITION m_position;
|
|---|
| 535 | STRING_POOL m_tempPool;
|
|---|
| 536 | STRING_POOL m_temp2Pool;
|
|---|
| 537 | char *m_groupConnector;
|
|---|
| 538 | unsigned m_groupSize;
|
|---|
| 539 | int m_hadExternalDoctype;
|
|---|
| 540 | XML_Char m_namespaceSeparator;
|
|---|
| 541 | #ifdef XML_DTD
|
|---|
| 542 | enum XML_ParamEntityParsing m_paramEntityParsing;
|
|---|
| 543 | XML_Parser m_parentParser;
|
|---|
| 544 | #endif
|
|---|
| 545 | }
|
|---|
| 546 | Parser;
|
|---|
| 547 |
|
|---|
| 548 | #define MALLOC(s) (((Parser *)parser)->m_mem.malloc_fcn((s)))
|
|---|
| 549 | #define REALLOC(p,s) (((Parser *)parser)->m_mem.realloc_fcn((p),(s)))
|
|---|
| 550 | #define FREE(p) (((Parser *)parser)->m_mem.free_fcn((p)))
|
|---|
| 551 |
|
|---|
| 552 | #define userData (((Parser *)parser)->m_userData)
|
|---|
| 553 | #define handlerArg (((Parser *)parser)->m_handlerArg)
|
|---|
| 554 | #define startElementHandler (((Parser *)parser)->m_startElementHandler)
|
|---|
| 555 | #define endElementHandler (((Parser *)parser)->m_endElementHandler)
|
|---|
| 556 | #define characterDataHandler (((Parser *)parser)->m_characterDataHandler)
|
|---|
| 557 | #define processingInstructionHandler (((Parser *)parser)->m_processingInstructionHandler)
|
|---|
| 558 | #define commentHandler (((Parser *)parser)->m_commentHandler)
|
|---|
| 559 | #define startCdataSectionHandler (((Parser *)parser)->m_startCdataSectionHandler)
|
|---|
| 560 | #define endCdataSectionHandler (((Parser *)parser)->m_endCdataSectionHandler)
|
|---|
| 561 | #define defaultHandler (((Parser *)parser)->m_defaultHandler)
|
|---|
| 562 | #define startDoctypeDeclHandler (((Parser *)parser)->m_startDoctypeDeclHandler)
|
|---|
| 563 | #define endDoctypeDeclHandler (((Parser *)parser)->m_endDoctypeDeclHandler)
|
|---|
| 564 | #define unparsedEntityDeclHandler (((Parser *)parser)->m_unparsedEntityDeclHandler)
|
|---|
| 565 | #define notationDeclHandler (((Parser *)parser)->m_notationDeclHandler)
|
|---|
| 566 | #define startNamespaceDeclHandler (((Parser *)parser)->m_startNamespaceDeclHandler)
|
|---|
| 567 | #define endNamespaceDeclHandler (((Parser *)parser)->m_endNamespaceDeclHandler)
|
|---|
| 568 | #define notStandaloneHandler (((Parser *)parser)->m_notStandaloneHandler)
|
|---|
| 569 | #define externalEntityRefHandler (((Parser *)parser)->m_externalEntityRefHandler)
|
|---|
| 570 | #define externalEntityRefHandlerArg (((Parser *)parser)->m_externalEntityRefHandlerArg)
|
|---|
| 571 | #define internalEntityRefHandler (((Parser *)parser)->m_internalEntityRefHandler)
|
|---|
| 572 | #define unknownEncodingHandler (((Parser *)parser)->m_unknownEncodingHandler)
|
|---|
| 573 | #define elementDeclHandler (((Parser *)parser)->m_elementDeclHandler)
|
|---|
| 574 | #define attlistDeclHandler (((Parser *)parser)->m_attlistDeclHandler)
|
|---|
| 575 | #define entityDeclHandler (((Parser *)parser)->m_entityDeclHandler)
|
|---|
| 576 | #define xmlDeclHandler (((Parser *)parser)->m_xmlDeclHandler)
|
|---|
| 577 | #define encoding (((Parser *)parser)->m_encoding)
|
|---|
| 578 | #define initEncoding (((Parser *)parser)->m_initEncoding)
|
|---|
| 579 | #define internalEncoding (((Parser *)parser)->m_internalEncoding)
|
|---|
| 580 | #define unknownEncodingMem (((Parser *)parser)->m_unknownEncodingMem)
|
|---|
| 581 | #define unknownEncodingData (((Parser *)parser)->m_unknownEncodingData)
|
|---|
| 582 | #define unknownEncodingHandlerData \
|
|---|
| 583 | (((Parser *)parser)->m_unknownEncodingHandlerData)
|
|---|
| 584 | #define unknownEncodingRelease (((Parser *)parser)->m_unknownEncodingRelease)
|
|---|
| 585 | #define protocolEncodingName (((Parser *)parser)->m_protocolEncodingName)
|
|---|
| 586 | #define ns (((Parser *)parser)->m_ns)
|
|---|
| 587 | #define ns_triplets (((Parser *)parser)->m_ns_triplets)
|
|---|
| 588 | #define prologState (((Parser *)parser)->m_prologState)
|
|---|
| 589 | #define processor (((Parser *)parser)->m_processor)
|
|---|
| 590 | #define errorCode (((Parser *)parser)->m_errorCode)
|
|---|
| 591 | #define eventPtr (((Parser *)parser)->m_eventPtr)
|
|---|
| 592 | #define eventEndPtr (((Parser *)parser)->m_eventEndPtr)
|
|---|
| 593 | #define positionPtr (((Parser *)parser)->m_positionPtr)
|
|---|
| 594 | #define position (((Parser *)parser)->m_position)
|
|---|
| 595 | #define openInternalEntities (((Parser *)parser)->m_openInternalEntities)
|
|---|
| 596 | #define defaultExpandInternalEntities (((Parser *)parser)->m_defaultExpandInternalEntities)
|
|---|
| 597 | #define tagLevel (((Parser *)parser)->m_tagLevel)
|
|---|
| 598 | #define buffer (((Parser *)parser)->m_buffer)
|
|---|
| 599 | #define bufferPtr (((Parser *)parser)->m_bufferPtr)
|
|---|
| 600 | #define bufferEnd (((Parser *)parser)->m_bufferEnd)
|
|---|
| 601 | #define parseEndByteIndex (((Parser *)parser)->m_parseEndByteIndex)
|
|---|
| 602 | #define parseEndPtr (((Parser *)parser)->m_parseEndPtr)
|
|---|
| 603 | #define bufferLim (((Parser *)parser)->m_bufferLim)
|
|---|
| 604 | #define dataBuf (((Parser *)parser)->m_dataBuf)
|
|---|
| 605 | #define dataBufEnd (((Parser *)parser)->m_dataBufEnd)
|
|---|
| 606 | #define dtd (((Parser *)parser)->m_dtd)
|
|---|
| 607 | #define curBase (((Parser *)parser)->m_curBase)
|
|---|
| 608 | #define declEntity (((Parser *)parser)->m_declEntity)
|
|---|
| 609 | #define doctypeName (((Parser *)parser)->m_doctypeName)
|
|---|
| 610 | #define doctypeSysid (((Parser *)parser)->m_doctypeSysid)
|
|---|
| 611 | #define doctypePubid (((Parser *)parser)->m_doctypePubid)
|
|---|
| 612 | #define declAttributeType (((Parser *)parser)->m_declAttributeType)
|
|---|
| 613 | #define declNotationName (((Parser *)parser)->m_declNotationName)
|
|---|
| 614 | #define declNotationPublicId (((Parser *)parser)->m_declNotationPublicId)
|
|---|
| 615 | #define declElementType (((Parser *)parser)->m_declElementType)
|
|---|
| 616 | #define declAttributeId (((Parser *)parser)->m_declAttributeId)
|
|---|
| 617 | #define declAttributeIsCdata (((Parser *)parser)->m_declAttributeIsCdata)
|
|---|
| 618 | #define declAttributeIsId (((Parser *)parser)->m_declAttributeIsId)
|
|---|
| 619 | #define freeTagList (((Parser *)parser)->m_freeTagList)
|
|---|
| 620 | #define freeBindingList (((Parser *)parser)->m_freeBindingList)
|
|---|
| 621 | #define inheritedBindings (((Parser *)parser)->m_inheritedBindings)
|
|---|
| 622 | #define tagStack (((Parser *)parser)->m_tagStack)
|
|---|
| 623 | #define atts (((Parser *)parser)->m_atts)
|
|---|
| 624 | #define attsSize (((Parser *)parser)->m_attsSize)
|
|---|
| 625 | #define nSpecifiedAtts (((Parser *)parser)->m_nSpecifiedAtts)
|
|---|
| 626 | #define idAttIndex (((Parser *)parser)->m_idAttIndex)
|
|---|
| 627 | #define tempPool (((Parser *)parser)->m_tempPool)
|
|---|
| 628 | #define temp2Pool (((Parser *)parser)->m_temp2Pool)
|
|---|
| 629 | #define groupConnector (((Parser *)parser)->m_groupConnector)
|
|---|
| 630 | #define groupSize (((Parser *)parser)->m_groupSize)
|
|---|
| 631 | #define hadExternalDoctype (((Parser *)parser)->m_hadExternalDoctype)
|
|---|
| 632 | #define namespaceSeparator (((Parser *)parser)->m_namespaceSeparator)
|
|---|
| 633 | #ifdef XML_DTD
|
|---|
| 634 | #define parentParser (((Parser *)parser)->m_parentParser)
|
|---|
| 635 | #define paramEntityParsing (((Parser *)parser)->m_paramEntityParsing)
|
|---|
| 636 | #endif /* XML_DTD */
|
|---|
| 637 |
|
|---|
| 638 | #ifdef COMPILED_FROM_DSP
|
|---|
| 639 | BOOL WINAPI DllMain(HINSTANCE h, DWORD r, LPVOID p)
|
|---|
| 640 | {
|
|---|
| 641 | return TRUE;
|
|---|
| 642 | }
|
|---|
| 643 | #endif /* def COMPILED_FROM_DSP */
|
|---|
| 644 |
|
|---|
| 645 | #ifdef _MSC_VER
|
|---|
| 646 | #ifdef _DEBUG
|
|---|
| 647 | Parser *asParser(XML_Parser parser)
|
|---|
| 648 | {
|
|---|
| 649 | return parser;
|
|---|
| 650 | }
|
|---|
| 651 | #endif
|
|---|
| 652 | #endif
|
|---|
| 653 |
|
|---|
| 654 | /*
|
|---|
| 655 | *@@ XML_ParserCreate:
|
|---|
| 656 | * constructs a new parser. If encoding is non-null, it specifies
|
|---|
| 657 | * a character encoding to use for the document (see below).
|
|---|
| 658 | * This overrides the document encoding declaration.
|
|---|
| 659 | *
|
|---|
| 660 | * Expat is a stream-oriented parser. You register callback (or
|
|---|
| 661 | * handler) functions with the parser and then start feeding it
|
|---|
| 662 | * the document. As the parser recognizes parts of the document,
|
|---|
| 663 | * it will call the appropriate handler for that part (if you've
|
|---|
| 664 | * registered one). The document is fed to the parser in pieces,
|
|---|
| 665 | * so you can start parsing before you have all the document.
|
|---|
| 666 | * This also allows you to parse really huge documents that won't
|
|---|
| 667 | * fit into memory.
|
|---|
| 668 | *
|
|---|
| 669 | * Expat can be intimidating due to the many kinds of handlers
|
|---|
| 670 | * and options you can set. But you only need to learn four
|
|---|
| 671 | * functions in order to do 90% of what you'll want to do with it:
|
|---|
| 672 | *
|
|---|
| 673 | * -- XML_ParserCreate: Create a new parser object.
|
|---|
| 674 | *
|
|---|
| 675 | * -- XML_SetElementHandler: Set handlers for start and end tags.
|
|---|
| 676 | *
|
|---|
| 677 | * -- XML_SetCharacterDataHandler: Set handler for text.
|
|---|
| 678 | *
|
|---|
| 679 | * -- XML_Parse: Pass a buffer full of document to the parser.
|
|---|
| 680 | *
|
|---|
| 681 | * So the first step in parsing an XML document with expat is
|
|---|
| 682 | * to create a parser object. There are three functions in the
|
|---|
| 683 | * expat API for creating a parser object. However, only two of
|
|---|
| 684 | * these (XML_ParserCreate and XML_ParserCreateNS) can be used
|
|---|
| 685 | * for constructing a parser for a top-level document. The object
|
|---|
| 686 | * returned by these functions is an opaque pointer (i.e. expat.h
|
|---|
| 687 | * declares it as void*) to data with further internal structure.
|
|---|
| 688 | * In order to free the memory associated with this object you
|
|---|
| 689 | * must call XML_ParserFree. Note that if you have provided
|
|---|
| 690 | * any user data that gets stored in the parser, then your
|
|---|
| 691 | * application is responsible for freeing it prior to calling
|
|---|
| 692 | * XML_ParserFree.
|
|---|
| 693 | *
|
|---|
| 694 | * The objects returned by the parser creation functions are
|
|---|
| 695 | * good for parsing only one XML document or external parsed
|
|---|
| 696 | * entity. If your application needs to parse many XML documents,
|
|---|
| 697 | * then it needs to create a parser object for each one. The
|
|---|
| 698 | * best way to deal with this is to create a higher level object
|
|---|
| 699 | * that contains all the default initialization you want for
|
|---|
| 700 | * your parser objects.
|
|---|
| 701 | *
|
|---|
| 702 | * Walking through a document hierarchy with a stream oriented
|
|---|
| 703 | * parser will require a good stack mechanism in order to keep
|
|---|
| 704 | * track of current context. For instance, to answer the simple
|
|---|
| 705 | * question, "What element does this text belong to?" requires
|
|---|
| 706 | * a stack, since the parser may have descended into other
|
|---|
| 707 | * elements that are children of the current one and has
|
|---|
| 708 | * encountered this text on the way out.
|
|---|
| 709 | *
|
|---|
| 710 | * The things you're likely to want to keep on a stack are the
|
|---|
| 711 | * currently opened element and its attributes. You push this
|
|---|
| 712 | * information onto the stack in the start handler and you pop
|
|---|
| 713 | * it off in the end handler.
|
|---|
| 714 | *
|
|---|
| 715 | * For some tasks, it is sufficient to just keep information on
|
|---|
| 716 | * what the depth of the stack is (or would be if you had one).
|
|---|
| 717 | * The outline program shown above presents one example. Another
|
|---|
| 718 | * such task would be skipping over a complete element. When you
|
|---|
| 719 | * see the start tag for the element you want to skip, you set a
|
|---|
| 720 | * skip flag and record the depth at which the element started.
|
|---|
| 721 | * When the end tag handler encounters the same depth, the
|
|---|
| 722 | * skipped element has ended and the flag may be cleared. If you
|
|---|
| 723 | * follow the convention that the root element starts at 1, then
|
|---|
| 724 | * you can use the same variable for skip flag and skip depth.
|
|---|
| 725 | *
|
|---|
| 726 | * <B>Namespace Processing</B>
|
|---|
| 727 | *
|
|---|
| 728 | * When the parser is created using the XML_ParserCreateNS,
|
|---|
| 729 | * function, expat performs namespace processing. See that
|
|---|
| 730 | * function for details.
|
|---|
| 731 | *
|
|---|
| 732 | * <B>Character Encodings</B>
|
|---|
| 733 | *
|
|---|
| 734 | * While XML is based on Unicode, and every XML processor is
|
|---|
| 735 | * required to recognized UTF-8 and UTF-16 (1 and 2 byte
|
|---|
| 736 | * encodings of Unicode), other @encodings may be declared in
|
|---|
| 737 | * XML documents or entities. For the main document, an XML
|
|---|
| 738 | * declaration may contain an encoding declaration in its
|
|---|
| 739 | * @text_declaration. See @encodings for additional
|
|---|
| 740 | * information.
|
|---|
| 741 | *
|
|---|
| 742 | * With expat, you may also specify an encoding at the time
|
|---|
| 743 | * of creating a parser. This is useful when the encoding
|
|---|
| 744 | * information may come from a source outside the document
|
|---|
| 745 | * itself (like a higher level protocol).
|
|---|
| 746 | *
|
|---|
| 747 | * See XML_SetUnknownEncodingHandler for encodings directly
|
|---|
| 748 | * supported by expat and for how to handle other encodings.
|
|---|
| 749 | *
|
|---|
| 750 | * One pitfall that novice expat users are likely to fall into is
|
|---|
| 751 | * that although expat may accept input in various encodings, the
|
|---|
| 752 | * strings that it passes to the handlers are always encoded in
|
|---|
| 753 | * UTF-8. Your application is responsible for any translation of
|
|---|
| 754 | * these strings into other encodings.
|
|---|
| 755 | *
|
|---|
| 756 | * <B>Handling External Entity References</B>
|
|---|
| 757 | *
|
|---|
| 758 | * Expat does not read or parse @external_entities directly. Note that
|
|---|
| 759 | * any external @DTD is a special case of an external entity. For how to
|
|---|
| 760 | * handle external entities, see XML_SetExternalEntityRefHandler.
|
|---|
| 761 | *
|
|---|
| 762 | * <B>Parsing Parameter Entities</B>
|
|---|
| 763 | *
|
|---|
| 764 | * In order to parse @parameter_entities, before starting the parse,
|
|---|
| 765 | * you must call XML_SetParamEntityParsing.
|
|---|
| 766 | */
|
|---|
| 767 |
|
|---|
| 768 | XML_Parser XML_ParserCreate(const XML_Char * encodingName)
|
|---|
| 769 | {
|
|---|
| 770 | return XML_ParserCreate_MM(encodingName, NULL, NULL);
|
|---|
| 771 | }
|
|---|
| 772 |
|
|---|
| 773 | /*
|
|---|
| 774 | *@@ XML_ParserCreateNS:
|
|---|
| 775 | * constructs a new parser that has namespace processing
|
|---|
| 776 | * in effect. Namespace expanded element names and attribute
|
|---|
| 777 | * names are returned as a concatenation of the namespace URI,
|
|---|
| 778 | * sep, and the local part of the name. This means that you
|
|---|
| 779 | * should pick a character for sep that can't be part of a
|
|---|
| 780 | * legal URI.
|
|---|
| 781 | *
|
|---|
| 782 | * Under namespace processing, expat consumes xmlns and xmlns:...
|
|---|
| 783 | * attributes, which declare namespaces for the scope of the
|
|---|
| 784 | * element in which they occur. This means that your start
|
|---|
| 785 | * handler will not see these attributes. Your application can
|
|---|
| 786 | * still be informed of these declarations by setting namespace
|
|---|
| 787 | * declaration handlers with XML_SetNamespaceDeclHandler.
|
|---|
| 788 | *
|
|---|
| 789 | * Element type and attribute names that belong to a given
|
|---|
| 790 | * namespace are passed to the appropriate handler in expanded
|
|---|
| 791 | * form. By default this expanded form is a concatenation of the
|
|---|
| 792 | * namespace URI, the separator character (which is the 2nd
|
|---|
| 793 | * argument to XML_ParserCreateNS), and the local name (i.e.
|
|---|
| 794 | * the part after the colon). Names with undeclared prefixes
|
|---|
| 795 | * are passed through to the handlers unchanged, with the prefix
|
|---|
| 796 | * and colon still attached. Unprefixed attribute names are never
|
|---|
| 797 | * expanded, and unprefixed element names are only expanded when
|
|---|
| 798 | * they are in the scope of a default namespace.
|
|---|
| 799 | *
|
|---|
| 800 | * However if XML_SetReturnNSTriplet has been called with a
|
|---|
| 801 | * non-zero do_nst parameter, then the expanded form for names
|
|---|
| 802 | * with an explicit prefix is a concatenation of: URI, separator,
|
|---|
| 803 | * local name, separator, prefix.
|
|---|
| 804 | *
|
|---|
| 805 | * You can set handlers for the start of a namespace declaration
|
|---|
| 806 | * and for the end of a scope of a declaration with the
|
|---|
| 807 | * XML_SetNamespaceDeclHandler function. The
|
|---|
| 808 | * StartNamespaceDeclHandler is called prior to the start
|
|---|
| 809 | * tag handler and the EndNamespaceDeclHandler is called before
|
|---|
| 810 | * the corresponding end tag that ends the namespace's scope.
|
|---|
| 811 | * The namespace start handler gets passed the prefix and URI
|
|---|
| 812 | * for the namespace. For a default namespace declaration
|
|---|
| 813 | * (xmlns='...'), the prefix will be null. The URI will be null
|
|---|
| 814 | * for the case where the default namespace is being unset. The
|
|---|
| 815 | * namespace end handler just gets the prefix for the closing
|
|---|
| 816 | * scope.
|
|---|
| 817 | *
|
|---|
| 818 | * These handlers are called for each declaration. So if, for
|
|---|
| 819 | * instance, a start tag had three namespace declarations, then
|
|---|
| 820 | * the StartNamespaceDeclHandler would be called three times
|
|---|
| 821 | * before the start tag handler is called, once for each
|
|---|
| 822 | * declaration.
|
|---|
| 823 | *
|
|---|
| 824 | * The namespace.c example demonstrates the use of these
|
|---|
| 825 | * features. Like outline.c, it produces an outline, but in
|
|---|
| 826 | * addition it annotates when a namespace scope starts and
|
|---|
| 827 | * when it ends. This example also demonstrates use of
|
|---|
| 828 | * application user data.
|
|---|
| 829 | */
|
|---|
| 830 |
|
|---|
| 831 | XML_Parser XML_ParserCreateNS(const XML_Char * encodingName,
|
|---|
| 832 | XML_Char nsSep)
|
|---|
| 833 | {
|
|---|
| 834 | XML_Char tmp[2];
|
|---|
| 835 |
|
|---|
| 836 | *tmp = nsSep;
|
|---|
| 837 | return XML_ParserCreate_MM(encodingName, NULL, tmp);
|
|---|
| 838 | }
|
|---|
| 839 |
|
|---|
| 840 | /*
|
|---|
| 841 | *@@ XML_ParserCreate_MM:
|
|---|
| 842 | * constructs a new parser using the suite of memory handling
|
|---|
| 843 | * functions specified in ms. If ms is NULL, then use the
|
|---|
| 844 | * standard set of memory management functions. If sep is
|
|---|
| 845 | * non-NULL, then namespace processing is enabled in the
|
|---|
| 846 | * created parser and the character pointed at by sep is
|
|---|
| 847 | * used as the separator between the namespace URI and the
|
|---|
| 848 | * local part of the name.
|
|---|
| 849 | */
|
|---|
| 850 |
|
|---|
| 851 | XML_Parser XML_ParserCreate_MM(const XML_Char * encodingName,
|
|---|
| 852 | const XML_Memory_Handling_Suite * memsuite,
|
|---|
| 853 | const XML_Char * nameSep)
|
|---|
| 854 | {
|
|---|
| 855 |
|
|---|
| 856 | XML_Parser parser;
|
|---|
| 857 | static
|
|---|
| 858 | const XML_Char implicitContext[] =
|
|---|
| 859 | {
|
|---|
| 860 | XML_T('x'), XML_T('m'), XML_T('l'), XML_T('='),
|
|---|
| 861 | XML_T('h'), XML_T('t'), XML_T('t'), XML_T('p'), XML_T(':'),
|
|---|
| 862 | XML_T('/'), XML_T('/'), XML_T('w'), XML_T('w'), XML_T('w'),
|
|---|
| 863 | XML_T('.'), XML_T('w'), XML_T('3'),
|
|---|
| 864 | XML_T('.'), XML_T('o'), XML_T('r'), XML_T('g'),
|
|---|
| 865 | XML_T('/'), XML_T('X'), XML_T('M'), XML_T('L'),
|
|---|
| 866 | XML_T('/'), XML_T('1'), XML_T('9'), XML_T('9'), XML_T('8'),
|
|---|
| 867 | XML_T('/'), XML_T('n'), XML_T('a'), XML_T('m'), XML_T('e'),
|
|---|
| 868 | XML_T('s'), XML_T('p'), XML_T('a'), XML_T('c'), XML_T('e'),
|
|---|
| 869 | XML_T('\0')
|
|---|
| 870 | };
|
|---|
| 871 |
|
|---|
| 872 |
|
|---|
| 873 | if (memsuite)
|
|---|
| 874 | {
|
|---|
| 875 | XML_Memory_Handling_Suite *mtemp;
|
|---|
| 876 |
|
|---|
| 877 | parser = memsuite->malloc_fcn(sizeof(Parser));
|
|---|
| 878 | mtemp = &(((Parser *) parser)->m_mem);
|
|---|
| 879 | mtemp->malloc_fcn = memsuite->malloc_fcn;
|
|---|
| 880 | mtemp->realloc_fcn = memsuite->realloc_fcn;
|
|---|
| 881 | mtemp->free_fcn = memsuite->free_fcn;
|
|---|
| 882 | }
|
|---|
| 883 | else
|
|---|
| 884 | {
|
|---|
| 885 | XML_Memory_Handling_Suite *mtemp;
|
|---|
| 886 |
|
|---|
| 887 | parser = malloc(sizeof(Parser));
|
|---|
| 888 | mtemp = &(((Parser *) parser)->m_mem);
|
|---|
| 889 | mtemp->malloc_fcn = malloc;
|
|---|
| 890 | mtemp->realloc_fcn = realloc;
|
|---|
| 891 | mtemp->free_fcn = free;
|
|---|
| 892 | }
|
|---|
| 893 |
|
|---|
| 894 | if (!parser)
|
|---|
| 895 | return parser;
|
|---|
| 896 | processor = prologInitProcessor;
|
|---|
| 897 | XmlPrologStateInit(&prologState);
|
|---|
| 898 | userData = 0;
|
|---|
| 899 | handlerArg = 0;
|
|---|
| 900 | startElementHandler = 0;
|
|---|
| 901 | endElementHandler = 0;
|
|---|
| 902 | characterDataHandler = 0;
|
|---|
| 903 | processingInstructionHandler = 0;
|
|---|
| 904 | commentHandler = 0;
|
|---|
| 905 | startCdataSectionHandler = 0;
|
|---|
| 906 | endCdataSectionHandler = 0;
|
|---|
| 907 | defaultHandler = 0;
|
|---|
| 908 | startDoctypeDeclHandler = 0;
|
|---|
| 909 | endDoctypeDeclHandler = 0;
|
|---|
| 910 | unparsedEntityDeclHandler = 0;
|
|---|
| 911 | notationDeclHandler = 0;
|
|---|
| 912 | startNamespaceDeclHandler = 0;
|
|---|
| 913 | endNamespaceDeclHandler = 0;
|
|---|
| 914 | notStandaloneHandler = 0;
|
|---|
| 915 | externalEntityRefHandler = 0;
|
|---|
| 916 | externalEntityRefHandlerArg = parser;
|
|---|
| 917 | unknownEncodingHandler = 0;
|
|---|
| 918 | elementDeclHandler = 0;
|
|---|
| 919 | attlistDeclHandler = 0;
|
|---|
| 920 | entityDeclHandler = 0;
|
|---|
| 921 | xmlDeclHandler = 0;
|
|---|
| 922 | buffer = 0;
|
|---|
| 923 | bufferPtr = 0;
|
|---|
| 924 | bufferEnd = 0;
|
|---|
| 925 | parseEndByteIndex = 0;
|
|---|
| 926 | parseEndPtr = 0;
|
|---|
| 927 | bufferLim = 0;
|
|---|
| 928 | declElementType = 0;
|
|---|
| 929 | declAttributeId = 0;
|
|---|
| 930 | declEntity = 0;
|
|---|
| 931 | doctypeName = 0;
|
|---|
| 932 | doctypeSysid = 0;
|
|---|
| 933 | doctypePubid = 0;
|
|---|
| 934 | declAttributeType = 0;
|
|---|
| 935 | declNotationName = 0;
|
|---|
| 936 | declNotationPublicId = 0;
|
|---|
| 937 | memset(&position, 0, sizeof(POSITION));
|
|---|
| 938 | errorCode = ERROR_EXPAT_NONE;
|
|---|
| 939 | eventPtr = 0;
|
|---|
| 940 | eventEndPtr = 0;
|
|---|
| 941 | positionPtr = 0;
|
|---|
| 942 | openInternalEntities = 0;
|
|---|
| 943 | tagLevel = 0;
|
|---|
| 944 | tagStack = 0;
|
|---|
| 945 | freeTagList = 0;
|
|---|
| 946 | freeBindingList = 0;
|
|---|
| 947 | inheritedBindings = 0;
|
|---|
| 948 | attsSize = INIT_ATTS_SIZE;
|
|---|
| 949 | atts = (ATTRIBUTE*)MALLOC(attsSize * sizeof(ATTRIBUTE));
|
|---|
| 950 | nSpecifiedAtts = 0;
|
|---|
| 951 | dataBuf = (XML_Char*)MALLOC(INIT_DATA_BUF_SIZE * sizeof(XML_Char));
|
|---|
| 952 | groupSize = 0;
|
|---|
| 953 | groupConnector = 0;
|
|---|
| 954 | hadExternalDoctype = 0;
|
|---|
| 955 | unknownEncodingMem = 0;
|
|---|
| 956 | unknownEncodingRelease = 0;
|
|---|
| 957 | unknownEncodingData = 0;
|
|---|
| 958 | unknownEncodingHandlerData = 0;
|
|---|
| 959 | namespaceSeparator = '!';
|
|---|
| 960 | #ifdef XML_DTD
|
|---|
| 961 | parentParser = 0;
|
|---|
| 962 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER;
|
|---|
| 963 | #endif
|
|---|
| 964 | ns = 0;
|
|---|
| 965 | ns_triplets = 0;
|
|---|
| 966 | poolInit(&tempPool, &(((Parser *) parser)->m_mem));
|
|---|
| 967 | poolInit(&temp2Pool, &(((Parser *) parser)->m_mem));
|
|---|
| 968 | protocolEncodingName = encodingName ? poolCopyString(&tempPool, encodingName) : 0;
|
|---|
| 969 | curBase = 0;
|
|---|
| 970 | if (!dtdInit(&dtd, parser) || !atts || !dataBuf
|
|---|
| 971 | || (encodingName && !protocolEncodingName))
|
|---|
| 972 | {
|
|---|
| 973 | XML_ParserFree(parser);
|
|---|
| 974 | return 0;
|
|---|
| 975 | }
|
|---|
| 976 | dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE;
|
|---|
| 977 |
|
|---|
| 978 | if (nameSep)
|
|---|
| 979 | {
|
|---|
| 980 | XmlInitEncodingNS(&initEncoding, &encoding, 0);
|
|---|
| 981 | ns = 1;
|
|---|
| 982 | internalEncoding = XmlGetInternalEncodingNS();
|
|---|
| 983 | namespaceSeparator = *nameSep;
|
|---|
| 984 |
|
|---|
| 985 | if (!setContext(parser, implicitContext))
|
|---|
| 986 | {
|
|---|
| 987 | XML_ParserFree(parser);
|
|---|
| 988 | return 0;
|
|---|
| 989 | }
|
|---|
| 990 | }
|
|---|
| 991 | else
|
|---|
| 992 | {
|
|---|
| 993 | XmlInitEncoding(&initEncoding, &encoding, 0);
|
|---|
| 994 | internalEncoding = XmlGetInternalEncoding();
|
|---|
| 995 | }
|
|---|
| 996 |
|
|---|
| 997 | return parser;
|
|---|
| 998 | } /* End XML_ParserCreate_MM */
|
|---|
| 999 |
|
|---|
| 1000 | /*
|
|---|
| 1001 | *@@ XML_SetEncoding:
|
|---|
| 1002 | * sets the encoding to be used by the parser. It is
|
|---|
| 1003 | * equivalent to passing a non-null encoding argument
|
|---|
| 1004 | * to the parser creation functions. It must not be
|
|---|
| 1005 | * called after XML_Parser or XML_ParseBuffer have
|
|---|
| 1006 | * been called on the given parser.
|
|---|
| 1007 | */
|
|---|
| 1008 |
|
|---|
| 1009 | int XML_SetEncoding(XML_Parser parser,
|
|---|
| 1010 | const XML_Char * encodingName)
|
|---|
| 1011 | {
|
|---|
| 1012 | if (!encodingName)
|
|---|
| 1013 | protocolEncodingName = 0;
|
|---|
| 1014 | else
|
|---|
| 1015 | {
|
|---|
| 1016 | protocolEncodingName = poolCopyString(&tempPool, encodingName);
|
|---|
| 1017 | if (!protocolEncodingName)
|
|---|
| 1018 | return 0;
|
|---|
| 1019 | }
|
|---|
| 1020 | return 1;
|
|---|
| 1021 | }
|
|---|
| 1022 |
|
|---|
| 1023 | /*
|
|---|
| 1024 | *@@ XML_ExternalEntityParserCreate:
|
|---|
| 1025 | * constructs a new XML_Parser object for parsing an external
|
|---|
| 1026 | * general entity. Context is the context argument passed in a
|
|---|
| 1027 | * call to a ExternalEntityRefHandler. Other state information
|
|---|
| 1028 | * such as handlers, user data, namespace processing is
|
|---|
| 1029 | * inherited from the parser passed as the 1st argument. So you
|
|---|
| 1030 | * shouldn't need to call any of the behavior changing
|
|---|
| 1031 | * functions on this parser (unless you want it to act
|
|---|
| 1032 | * differently than the parent parser.)
|
|---|
| 1033 | */
|
|---|
| 1034 |
|
|---|
| 1035 | XML_Parser XML_ExternalEntityParserCreate(XML_Parser oldParser,
|
|---|
| 1036 | const XML_Char * context,
|
|---|
| 1037 | const XML_Char * encodingName)
|
|---|
| 1038 | {
|
|---|
| 1039 | XML_Parser parser = oldParser;
|
|---|
| 1040 | DTD *oldDtd = &dtd;
|
|---|
| 1041 | XML_StartElementHandler oldStartElementHandler = startElementHandler;
|
|---|
| 1042 | XML_EndElementHandler oldEndElementHandler = endElementHandler;
|
|---|
| 1043 | XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler;
|
|---|
| 1044 | XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler;
|
|---|
| 1045 | XML_CommentHandler oldCommentHandler = commentHandler;
|
|---|
| 1046 | XML_StartCdataSectionHandler oldStartCdataSectionHandler = startCdataSectionHandler;
|
|---|
| 1047 | XML_EndCdataSectionHandler oldEndCdataSectionHandler = endCdataSectionHandler;
|
|---|
| 1048 | XML_DefaultHandler oldDefaultHandler = defaultHandler;
|
|---|
| 1049 | XML_UnparsedEntityDeclHandler oldUnparsedEntityDeclHandler = unparsedEntityDeclHandler;
|
|---|
| 1050 | XML_NotationDeclHandler oldNotationDeclHandler = notationDeclHandler;
|
|---|
| 1051 | XML_StartNamespaceDeclHandler oldStartNamespaceDeclHandler = startNamespaceDeclHandler;
|
|---|
| 1052 | XML_EndNamespaceDeclHandler oldEndNamespaceDeclHandler = endNamespaceDeclHandler;
|
|---|
| 1053 | XML_NotStandaloneHandler oldNotStandaloneHandler = notStandaloneHandler;
|
|---|
| 1054 | XML_ExternalEntityRefHandler oldExternalEntityRefHandler = externalEntityRefHandler;
|
|---|
| 1055 | XML_UnknownEncodingHandler oldUnknownEncodingHandler = unknownEncodingHandler;
|
|---|
| 1056 | XML_ElementDeclHandler oldElementDeclHandler = elementDeclHandler;
|
|---|
| 1057 | XML_AttlistDeclHandler oldAttlistDeclHandler = attlistDeclHandler;
|
|---|
| 1058 | XML_EntityDeclHandler oldEntityDeclHandler = entityDeclHandler;
|
|---|
| 1059 | XML_XmlDeclHandler oldXmlDeclHandler = xmlDeclHandler;
|
|---|
| 1060 | ELEMENT_TYPE *oldDeclElementType = declElementType;
|
|---|
| 1061 |
|
|---|
| 1062 | void *oldUserData = userData;
|
|---|
| 1063 | void *oldHandlerArg = handlerArg;
|
|---|
| 1064 | int oldDefaultExpandInternalEntities = defaultExpandInternalEntities;
|
|---|
| 1065 | void *oldExternalEntityRefHandlerArg = externalEntityRefHandlerArg;
|
|---|
| 1066 |
|
|---|
| 1067 | #ifdef XML_DTD
|
|---|
| 1068 | enum XML_ParamEntityParsing oldParamEntityParsing = paramEntityParsing;
|
|---|
| 1069 |
|
|---|
| 1070 | #endif
|
|---|
| 1071 | int oldns_triplets = ns_triplets;
|
|---|
| 1072 |
|
|---|
| 1073 | if (ns)
|
|---|
| 1074 | {
|
|---|
| 1075 | XML_Char tmp[2];
|
|---|
| 1076 |
|
|---|
| 1077 | *tmp = namespaceSeparator;
|
|---|
| 1078 | parser = XML_ParserCreate_MM(encodingName, &((Parser *) parser)->m_mem,
|
|---|
| 1079 | tmp);
|
|---|
| 1080 | }
|
|---|
| 1081 | else
|
|---|
| 1082 | {
|
|---|
| 1083 | parser = XML_ParserCreate_MM(encodingName, &((Parser *) parser)->m_mem,
|
|---|
| 1084 | NULL);
|
|---|
| 1085 | }
|
|---|
| 1086 |
|
|---|
| 1087 | if (!parser)
|
|---|
| 1088 | return 0;
|
|---|
| 1089 |
|
|---|
| 1090 | startElementHandler = oldStartElementHandler;
|
|---|
| 1091 | endElementHandler = oldEndElementHandler;
|
|---|
| 1092 | characterDataHandler = oldCharacterDataHandler;
|
|---|
| 1093 | processingInstructionHandler = oldProcessingInstructionHandler;
|
|---|
| 1094 | commentHandler = oldCommentHandler;
|
|---|
| 1095 | startCdataSectionHandler = oldStartCdataSectionHandler;
|
|---|
| 1096 | endCdataSectionHandler = oldEndCdataSectionHandler;
|
|---|
| 1097 | defaultHandler = oldDefaultHandler;
|
|---|
| 1098 | unparsedEntityDeclHandler = oldUnparsedEntityDeclHandler;
|
|---|
| 1099 | notationDeclHandler = oldNotationDeclHandler;
|
|---|
| 1100 | startNamespaceDeclHandler = oldStartNamespaceDeclHandler;
|
|---|
| 1101 | endNamespaceDeclHandler = oldEndNamespaceDeclHandler;
|
|---|
| 1102 | notStandaloneHandler = oldNotStandaloneHandler;
|
|---|
| 1103 | externalEntityRefHandler = oldExternalEntityRefHandler;
|
|---|
| 1104 | unknownEncodingHandler = oldUnknownEncodingHandler;
|
|---|
| 1105 | elementDeclHandler = oldElementDeclHandler;
|
|---|
| 1106 | attlistDeclHandler = oldAttlistDeclHandler;
|
|---|
| 1107 | entityDeclHandler = oldEntityDeclHandler;
|
|---|
| 1108 | xmlDeclHandler = oldXmlDeclHandler;
|
|---|
| 1109 | declElementType = oldDeclElementType;
|
|---|
| 1110 | userData = oldUserData;
|
|---|
| 1111 | if (oldUserData == oldHandlerArg)
|
|---|
| 1112 | handlerArg = userData;
|
|---|
| 1113 | else
|
|---|
| 1114 | handlerArg = parser;
|
|---|
| 1115 | if (oldExternalEntityRefHandlerArg != oldParser)
|
|---|
| 1116 | externalEntityRefHandlerArg = oldExternalEntityRefHandlerArg;
|
|---|
| 1117 | defaultExpandInternalEntities = oldDefaultExpandInternalEntities;
|
|---|
| 1118 | ns_triplets = oldns_triplets;
|
|---|
| 1119 | #ifdef XML_DTD
|
|---|
| 1120 | paramEntityParsing = oldParamEntityParsing;
|
|---|
| 1121 | if (context)
|
|---|
| 1122 | {
|
|---|
| 1123 | #endif /* XML_DTD */
|
|---|
| 1124 | if (!dtdCopy(&dtd, oldDtd, parser) || !setContext(parser, context))
|
|---|
| 1125 | {
|
|---|
| 1126 | XML_ParserFree(parser);
|
|---|
| 1127 | return 0;
|
|---|
| 1128 | }
|
|---|
| 1129 | processor = externalEntityInitProcessor;
|
|---|
| 1130 | #ifdef XML_DTD
|
|---|
| 1131 | }
|
|---|
| 1132 | else
|
|---|
| 1133 | {
|
|---|
| 1134 | dtdSwap(&dtd, oldDtd);
|
|---|
| 1135 | parentParser = oldParser;
|
|---|
| 1136 | XmlPrologStateInitExternalEntity(&prologState);
|
|---|
| 1137 | dtd.complete = 1;
|
|---|
| 1138 | hadExternalDoctype = 1;
|
|---|
| 1139 | }
|
|---|
| 1140 | #endif /* XML_DTD */
|
|---|
| 1141 | return parser;
|
|---|
| 1142 | }
|
|---|
| 1143 |
|
|---|
| 1144 | STATIC void destroyBindings(BINDING * bindings,
|
|---|
| 1145 | XML_Parser parser)
|
|---|
| 1146 | {
|
|---|
| 1147 | for (;;)
|
|---|
| 1148 | {
|
|---|
| 1149 | BINDING *b = bindings;
|
|---|
| 1150 |
|
|---|
| 1151 | if (!b)
|
|---|
| 1152 | break;
|
|---|
| 1153 | bindings = b->nextTagBinding;
|
|---|
| 1154 | FREE(b->uri);
|
|---|
| 1155 | FREE(b);
|
|---|
| 1156 | }
|
|---|
| 1157 | }
|
|---|
| 1158 |
|
|---|
| 1159 | /*
|
|---|
| 1160 | *@@ XML_ParserFree:
|
|---|
| 1161 | * free memory used by the parser. Your application is
|
|---|
| 1162 | * responsible for freeing any memory associated
|
|---|
| 1163 | * with UserData.
|
|---|
| 1164 | */
|
|---|
| 1165 |
|
|---|
| 1166 | void XML_ParserFree(XML_Parser parser)
|
|---|
| 1167 | {
|
|---|
| 1168 | for (;;)
|
|---|
| 1169 | {
|
|---|
| 1170 | TAG *p;
|
|---|
| 1171 |
|
|---|
| 1172 | if (tagStack == 0)
|
|---|
| 1173 | {
|
|---|
| 1174 | if (freeTagList == 0)
|
|---|
| 1175 | break;
|
|---|
| 1176 | tagStack = freeTagList;
|
|---|
| 1177 | freeTagList = 0;
|
|---|
| 1178 | }
|
|---|
| 1179 | p = tagStack;
|
|---|
| 1180 | tagStack = tagStack->parent;
|
|---|
| 1181 | FREE(p->buf);
|
|---|
| 1182 | destroyBindings(p->bindings, parser);
|
|---|
| 1183 | FREE(p);
|
|---|
| 1184 | }
|
|---|
| 1185 | destroyBindings(freeBindingList, parser);
|
|---|
| 1186 | destroyBindings(inheritedBindings, parser);
|
|---|
| 1187 | poolDestroy(&tempPool);
|
|---|
| 1188 | poolDestroy(&temp2Pool);
|
|---|
| 1189 | #ifdef XML_DTD
|
|---|
| 1190 | if (parentParser)
|
|---|
| 1191 | {
|
|---|
| 1192 | if (hadExternalDoctype)
|
|---|
| 1193 | dtd.complete = 0;
|
|---|
| 1194 | dtdSwap(&dtd, &((Parser *) parentParser)->m_dtd);
|
|---|
| 1195 | }
|
|---|
| 1196 | #endif /* XML_DTD */
|
|---|
| 1197 | dtdDestroy(&dtd, parser);
|
|---|
| 1198 | FREE((void *)atts);
|
|---|
| 1199 | if (groupConnector)
|
|---|
| 1200 | FREE(groupConnector);
|
|---|
| 1201 | if (buffer)
|
|---|
| 1202 | FREE(buffer);
|
|---|
| 1203 | FREE(dataBuf);
|
|---|
| 1204 | if (unknownEncodingMem)
|
|---|
| 1205 | FREE(unknownEncodingMem);
|
|---|
| 1206 | if (unknownEncodingRelease)
|
|---|
| 1207 | unknownEncodingRelease(unknownEncodingData);
|
|---|
| 1208 | FREE(parser);
|
|---|
| 1209 | }
|
|---|
| 1210 |
|
|---|
| 1211 | /*
|
|---|
| 1212 | *@@ XML_UseParserAsHandlerArg:
|
|---|
| 1213 | * after this is called, handlers receive the parser in the
|
|---|
| 1214 | * userData argument (see XML_SetUserData). The userData
|
|---|
| 1215 | * information can still be obtained using the XML_GetUserData
|
|---|
| 1216 | * function.
|
|---|
| 1217 | */
|
|---|
| 1218 |
|
|---|
| 1219 | void XML_UseParserAsHandlerArg(XML_Parser parser)
|
|---|
| 1220 | {
|
|---|
| 1221 | handlerArg = parser;
|
|---|
| 1222 | }
|
|---|
| 1223 |
|
|---|
| 1224 | /*
|
|---|
| 1225 | *@@ XML_SetReturnNSTriplet:
|
|---|
| 1226 | * this function only has an effect when using a parser
|
|---|
| 1227 | * created with XML_ParserCreateNS, i.e. when namespace
|
|---|
| 1228 | * processing is in effect. The do_nst sets whether or
|
|---|
| 1229 | * not prefixes are returned with names qualified with
|
|---|
| 1230 | * a namespace prefix. If this function is called with
|
|---|
| 1231 | * do_nst non-zero, then afterwards namespace qualified
|
|---|
| 1232 | * names (that is qualified with a prefix as opposed to
|
|---|
| 1233 | * belonging to a default namespace) are returned as a
|
|---|
| 1234 | * triplet with the three parts separated by the namespace
|
|---|
| 1235 | * separator specified when the parser was created. The
|
|---|
| 1236 | * order of returned parts is URI, local name, and prefix.
|
|---|
| 1237 | *
|
|---|
| 1238 | * If do_nst is zero, then namespaces are reported in
|
|---|
| 1239 | * the default manner, URI then local_name separated by
|
|---|
| 1240 | * the namespace separator.
|
|---|
| 1241 | */
|
|---|
| 1242 |
|
|---|
| 1243 | void XML_SetReturnNSTriplet(XML_Parser parser,
|
|---|
| 1244 | int do_nst)
|
|---|
| 1245 | {
|
|---|
| 1246 | ns_triplets = do_nst;
|
|---|
| 1247 | }
|
|---|
| 1248 |
|
|---|
| 1249 | /*
|
|---|
| 1250 | *@@ XML_SetUserData:
|
|---|
| 1251 | * sets the user data pointer that gets passed to handlers.
|
|---|
| 1252 | *
|
|---|
| 1253 | * It overwrites any previous value for this pointer. Note that
|
|---|
| 1254 | * the application is responsible for freeing the memory
|
|---|
| 1255 | * associated with userData when it is finished with the parser.
|
|---|
| 1256 | * So if you call this when there's already a pointer there, and
|
|---|
| 1257 | * you haven't freed the memory associated with it, then you've
|
|---|
| 1258 | * probably just leaked memory.
|
|---|
| 1259 | *
|
|---|
| 1260 | * Also see XML_UseParserAsHandlerArg.
|
|---|
| 1261 | */
|
|---|
| 1262 |
|
|---|
| 1263 | void XML_SetUserData(XML_Parser parser,
|
|---|
| 1264 | void *p)
|
|---|
| 1265 | {
|
|---|
| 1266 | if (handlerArg == userData)
|
|---|
| 1267 | handlerArg = userData = p;
|
|---|
| 1268 | else
|
|---|
| 1269 | userData = p;
|
|---|
| 1270 | }
|
|---|
| 1271 |
|
|---|
| 1272 | /*
|
|---|
| 1273 | *@@ XML_SetBase:
|
|---|
| 1274 | * set the base to be used for resolving relative URIs in
|
|---|
| 1275 | * system identifiers. The return value is 0 if there's no
|
|---|
| 1276 | * memory to store base, otherwise it's non-zero.
|
|---|
| 1277 | */
|
|---|
| 1278 |
|
|---|
| 1279 | int XML_SetBase(XML_Parser parser,
|
|---|
| 1280 | const XML_Char * p)
|
|---|
| 1281 | {
|
|---|
| 1282 | if (p)
|
|---|
| 1283 | {
|
|---|
| 1284 | p = poolCopyString(&dtd.pool, p);
|
|---|
| 1285 | if (!p)
|
|---|
| 1286 | return 0;
|
|---|
| 1287 | curBase = p;
|
|---|
| 1288 | }
|
|---|
| 1289 | else
|
|---|
| 1290 | curBase = 0;
|
|---|
| 1291 | return 1;
|
|---|
| 1292 | }
|
|---|
| 1293 |
|
|---|
| 1294 | /*
|
|---|
| 1295 | *@@ XML_GetBase:
|
|---|
| 1296 | * returns the base for resolving relative URIs.
|
|---|
| 1297 | * See XML_SetBase.
|
|---|
| 1298 | */
|
|---|
| 1299 |
|
|---|
| 1300 | const XML_Char *XML_GetBase(XML_Parser parser)
|
|---|
| 1301 | {
|
|---|
| 1302 | return curBase;
|
|---|
| 1303 | }
|
|---|
| 1304 |
|
|---|
| 1305 | /*
|
|---|
| 1306 | *@@ XML_GetSpecifiedAttributeCount:
|
|---|
| 1307 | * when attributes are reported to the start handler in the
|
|---|
| 1308 | * atts vector, attributes that were explicitly set in the
|
|---|
| 1309 | * element occur before any attributes that receive their
|
|---|
| 1310 | * value from default information in an ATTLIST declaration.
|
|---|
| 1311 | * This function returns the number of attributes that were
|
|---|
| 1312 | * explicitly set times two, thus giving the offset in the
|
|---|
| 1313 | * atts array passed to the start tag handler of the first
|
|---|
| 1314 | * attribute set due to defaults. It supplies information
|
|---|
| 1315 | * for the last call to a start handler. If called inside a
|
|---|
| 1316 | * start handler, then that means the current call.
|
|---|
| 1317 | */
|
|---|
| 1318 |
|
|---|
| 1319 | int XML_GetSpecifiedAttributeCount(XML_Parser parser)
|
|---|
| 1320 | {
|
|---|
| 1321 | return nSpecifiedAtts;
|
|---|
| 1322 | }
|
|---|
| 1323 |
|
|---|
| 1324 | /*
|
|---|
| 1325 | *@@ XML_GetIdAttributeIndex:
|
|---|
| 1326 | * returns the index of the ID attribute passed in the atts
|
|---|
| 1327 | * array in the last call to XML_StartElementHandler, or -1
|
|---|
| 1328 | * if there is no ID attribute. If called inside a start
|
|---|
| 1329 | * handler, then that means the current call.
|
|---|
| 1330 | */
|
|---|
| 1331 |
|
|---|
| 1332 | int XML_GetIdAttributeIndex(XML_Parser parser)
|
|---|
| 1333 | {
|
|---|
| 1334 | return idAttIndex;
|
|---|
| 1335 | }
|
|---|
| 1336 |
|
|---|
| 1337 | /*
|
|---|
| 1338 | *@@ XML_SetElementHandler:
|
|---|
| 1339 | * sets handlers for start and end tags with one call.
|
|---|
| 1340 | *
|
|---|
| 1341 | * See XML_SetStartElementHandler and XML_SetEndElementHandler
|
|---|
| 1342 | * for details.
|
|---|
| 1343 | */
|
|---|
| 1344 |
|
|---|
| 1345 | void XML_SetElementHandler(XML_Parser parser,
|
|---|
| 1346 | XML_StartElementHandler start,
|
|---|
| 1347 | XML_EndElementHandler end)
|
|---|
| 1348 | {
|
|---|
| 1349 | startElementHandler = start;
|
|---|
| 1350 | endElementHandler = end;
|
|---|
| 1351 | }
|
|---|
| 1352 |
|
|---|
| 1353 | /*
|
|---|
| 1354 | *@@ XML_SetStartElementHandler:
|
|---|
| 1355 | * sets handler for start (and empty) tags.
|
|---|
| 1356 | *
|
|---|
| 1357 | * This handler must have the following prototype:
|
|---|
| 1358 | *
|
|---|
| 1359 | + void EXPATENTRY StartElementHandler(void *pUserData,
|
|---|
| 1360 | + const XML_Char *name,
|
|---|
| 1361 | + const XML_Char **atts);
|
|---|
| 1362 | +
|
|---|
| 1363 | * "data" is the user data pointer set with XML_SetUserData.
|
|---|
| 1364 | *
|
|---|
| 1365 | * "name" is the element name.
|
|---|
| 1366 | *
|
|---|
| 1367 | * Attributes are passed to the start handler as a pointer
|
|---|
| 1368 | * to a vector of char pointers. Each attribute seen in a
|
|---|
| 1369 | * start (or empty) tag occupies 2 consecutive places in
|
|---|
| 1370 | * this vector: the attribute name followed by the attribute
|
|---|
| 1371 | * value. These pairs are terminated by a null pointer.
|
|---|
| 1372 | *
|
|---|
| 1373 | * Note that an empty tag generates a call to both start
|
|---|
| 1374 | * and end handlers (in that order).
|
|---|
| 1375 | *
|
|---|
| 1376 | * Although handlers are typically set prior to parsing and left
|
|---|
| 1377 | * alone, an application may choose to set or change the handler
|
|---|
| 1378 | * for a parsing event while the parse is in progress. For
|
|---|
| 1379 | * instance, your application may choose to ignore all text not
|
|---|
| 1380 | * descended from a para element. One way it could do this is to
|
|---|
| 1381 | * set the character handler when a para start tag is seen, and
|
|---|
| 1382 | * unset it for the corresponding end tag.
|
|---|
| 1383 | *
|
|---|
| 1384 | * A handler may be unset by providing a NULL pointer to the
|
|---|
| 1385 | * appropriate handler setter. None of the handler setting
|
|---|
| 1386 | * functions have a return value.
|
|---|
| 1387 | *
|
|---|
| 1388 | * Your handlers will be receiving strings in arrays of type
|
|---|
| 1389 | * XML_Char. This type is defined in expat.h as char* and
|
|---|
| 1390 | * contains bytes encoding UTF-8. Note that you'll receive them
|
|---|
| 1391 | * in this form independent of the original encoding of the
|
|---|
| 1392 | * document.
|
|---|
| 1393 | */
|
|---|
| 1394 |
|
|---|
| 1395 | void XML_SetStartElementHandler(XML_Parser parser,
|
|---|
| 1396 | XML_StartElementHandler start)
|
|---|
| 1397 | {
|
|---|
| 1398 | startElementHandler = start;
|
|---|
| 1399 | }
|
|---|
| 1400 |
|
|---|
| 1401 | /*
|
|---|
| 1402 | *@@ XML_SetEndElementHandler:
|
|---|
| 1403 | * sets handler for end (and empty) tags. As noted above, an
|
|---|
| 1404 | * empty tag generates a call to both start and end handlers.
|
|---|
| 1405 | *
|
|---|
| 1406 | * Handler prototype:
|
|---|
| 1407 | +
|
|---|
| 1408 | + void EXPATENTRY EndElementHandler(void *pUserData,
|
|---|
| 1409 | + const XML_Char *name);
|
|---|
| 1410 | */
|
|---|
| 1411 |
|
|---|
| 1412 | void XML_SetEndElementHandler(XML_Parser parser,
|
|---|
| 1413 | XML_EndElementHandler end)
|
|---|
| 1414 | {
|
|---|
| 1415 | endElementHandler = end;
|
|---|
| 1416 | }
|
|---|
| 1417 |
|
|---|
| 1418 | /*
|
|---|
| 1419 | *@@ XML_SetCharacterDataHandler:
|
|---|
| 1420 | * sets a character data (text, @content) handler.
|
|---|
| 1421 | *
|
|---|
| 1422 | * Handler prototype:
|
|---|
| 1423 | *
|
|---|
| 1424 | + void EXPATENTRY CharacterDataHandler(void *pUserData,
|
|---|
| 1425 | + const XML_Char *s,
|
|---|
| 1426 | + int len);
|
|---|
| 1427 | +
|
|---|
| 1428 | * The string your handler receives is NOT zero terminated.
|
|---|
| 1429 | * You have to use the length argument to deal with the end
|
|---|
| 1430 | * of the string. A single block of contiguous text free of
|
|---|
| 1431 | * markup may still result in a sequence of calls to this
|
|---|
| 1432 | * handler. In other words, if you're searching for a pattern
|
|---|
| 1433 | * in the text, it may be split across calls to this handler.
|
|---|
| 1434 | */
|
|---|
| 1435 |
|
|---|
| 1436 | void XML_SetCharacterDataHandler(XML_Parser parser,
|
|---|
| 1437 | XML_CharacterDataHandler handler)
|
|---|
| 1438 | {
|
|---|
| 1439 | characterDataHandler = handler;
|
|---|
| 1440 | }
|
|---|
| 1441 |
|
|---|
| 1442 | /*
|
|---|
| 1443 | *@@ XML_SetProcessingInstructionHandler:
|
|---|
| 1444 | * sets a handler for processing instructions.
|
|---|
| 1445 | *
|
|---|
| 1446 | * Handler prototype:
|
|---|
| 1447 | *
|
|---|
| 1448 | + void EXPATENTRY ProcessingInstructionHandler(void *pUserData,
|
|---|
| 1449 | + const XML_Char *target,
|
|---|
| 1450 | + const XML_Char *data);
|
|---|
| 1451 | +
|
|---|
| 1452 | * The target is the first word in the processing instruction. The data
|
|---|
| 1453 | * is the rest of the characters in it after skipping all
|
|---|
| 1454 | * whitespace after the initial word.
|
|---|
| 1455 | */
|
|---|
| 1456 |
|
|---|
| 1457 | void XML_SetProcessingInstructionHandler(XML_Parser parser,
|
|---|
| 1458 | XML_ProcessingInstructionHandler handler)
|
|---|
| 1459 | {
|
|---|
| 1460 | processingInstructionHandler = handler;
|
|---|
| 1461 | }
|
|---|
| 1462 |
|
|---|
| 1463 | /*
|
|---|
| 1464 | *@@ XML_SetCommentHandler:
|
|---|
| 1465 | * sets a handler for comments. The data is all text inside
|
|---|
| 1466 | * the comment delimiters.
|
|---|
| 1467 | *
|
|---|
| 1468 | * Handler prototype:
|
|---|
| 1469 | *
|
|---|
| 1470 | + void EXPATENTRY CommentHandler(void *pUserData,
|
|---|
| 1471 | + const XML_Char *data);
|
|---|
| 1472 | *
|
|---|
| 1473 | */
|
|---|
| 1474 |
|
|---|
| 1475 | void XML_SetCommentHandler(XML_Parser parser,
|
|---|
| 1476 | XML_CommentHandler handler)
|
|---|
| 1477 | {
|
|---|
| 1478 | commentHandler = handler;
|
|---|
| 1479 | }
|
|---|
| 1480 |
|
|---|
| 1481 | /*
|
|---|
| 1482 | *@@ XML_SetCdataSectionHandler:
|
|---|
| 1483 | * combination of XML_SetStartCdataSectionHandler and
|
|---|
| 1484 | * XML_SetEndCdataSectionHandler in one call.
|
|---|
| 1485 | */
|
|---|
| 1486 |
|
|---|
| 1487 | void XML_SetCdataSectionHandler(XML_Parser parser,
|
|---|
| 1488 | XML_StartCdataSectionHandler start,
|
|---|
| 1489 | XML_EndCdataSectionHandler end)
|
|---|
| 1490 | {
|
|---|
| 1491 | startCdataSectionHandler = start;
|
|---|
| 1492 | endCdataSectionHandler = end;
|
|---|
| 1493 | }
|
|---|
| 1494 |
|
|---|
| 1495 | /*
|
|---|
| 1496 | *@@ XML_SetCdataSectionHandler:
|
|---|
| 1497 | * sets a handler that gets called at the beginning of a CDATA
|
|---|
| 1498 | * section.
|
|---|
| 1499 | *
|
|---|
| 1500 | * Handler prototype:
|
|---|
| 1501 | *
|
|---|
| 1502 | + void EXPATENTRY StartCdataSectionHandler(void *pUserData);
|
|---|
| 1503 | *
|
|---|
| 1504 | */
|
|---|
| 1505 |
|
|---|
| 1506 | void XML_SetStartCdataSectionHandler(XML_Parser parser,
|
|---|
| 1507 | XML_StartCdataSectionHandler start)
|
|---|
| 1508 | {
|
|---|
| 1509 | startCdataSectionHandler = start;
|
|---|
| 1510 | }
|
|---|
| 1511 |
|
|---|
| 1512 | /*
|
|---|
| 1513 | *@@ XML_SetCdataSectionHandler:
|
|---|
| 1514 | * sets a handler that gets called at the end of a CDATA
|
|---|
| 1515 | * section.
|
|---|
| 1516 | *
|
|---|
| 1517 | * Handler prototype:
|
|---|
| 1518 | *
|
|---|
| 1519 | * void EXPATENTRY EndCdataSectionHandler(void *pUserData);
|
|---|
| 1520 | */
|
|---|
| 1521 |
|
|---|
| 1522 | void XML_SetEndCdataSectionHandler(XML_Parser parser,
|
|---|
| 1523 | XML_EndCdataSectionHandler end)
|
|---|
| 1524 | {
|
|---|
| 1525 | endCdataSectionHandler = end;
|
|---|
| 1526 | }
|
|---|
| 1527 |
|
|---|
| 1528 | /*
|
|---|
| 1529 | *@@ XML_SetDefaultHandler:
|
|---|
| 1530 | * sets a handler for any characters in the document for
|
|---|
| 1531 | * which there is no applicable handler.
|
|---|
| 1532 | *
|
|---|
| 1533 | * This includes both data for which no handlers can be set
|
|---|
| 1534 | * (like some kinds of DTD declarations) and data which could
|
|---|
| 1535 | * be reported but for which there has no handler supplied.
|
|---|
| 1536 | *
|
|---|
| 1537 | * Handler prototype:
|
|---|
| 1538 | +
|
|---|
| 1539 | + void EXPATENTRY DefaultHandler(void *pUserData,
|
|---|
| 1540 | + const XML_Char *s,
|
|---|
| 1541 | + int len);
|
|---|
| 1542 | *
|
|---|
| 1543 | * The characters are passed exactly as they were in the XML
|
|---|
| 1544 | * document except that they will be encoded in UTF-8. Line
|
|---|
| 1545 | * boundaries are not normalized.
|
|---|
| 1546 | *
|
|---|
| 1547 | * Note that a byte order mark character is not passed to the
|
|---|
| 1548 | * default handler.
|
|---|
| 1549 | *
|
|---|
| 1550 | * Note that a contiguous piece of data that is destined to
|
|---|
| 1551 | * be reported to the default handler may actually be reported
|
|---|
| 1552 | * over several calls to the handler.
|
|---|
| 1553 | *
|
|---|
| 1554 | * Setting the handler with this call has the side effect of
|
|---|
| 1555 | * turning off expansion of references to internally defined
|
|---|
| 1556 | * general entities. Instead these references are passed to
|
|---|
| 1557 | * the default handler. To avoid that, use XML_SetDefaultHandlerExpand.
|
|---|
| 1558 | */
|
|---|
| 1559 |
|
|---|
| 1560 | void XML_SetDefaultHandler(XML_Parser parser,
|
|---|
| 1561 | XML_DefaultHandler handler)
|
|---|
| 1562 | {
|
|---|
| 1563 | defaultHandler = handler;
|
|---|
| 1564 | defaultExpandInternalEntities = 0;
|
|---|
| 1565 | }
|
|---|
| 1566 |
|
|---|
| 1567 | /*
|
|---|
| 1568 | *@@ XML_SetDefaultHandlerExpand:
|
|---|
| 1569 | * this sets a default handler, but doesn't affect expansion of
|
|---|
| 1570 | * internal entity references.
|
|---|
| 1571 | *
|
|---|
| 1572 | * See XML_SetDefaultHandler.
|
|---|
| 1573 | */
|
|---|
| 1574 |
|
|---|
| 1575 | void XML_SetDefaultHandlerExpand(XML_Parser parser,
|
|---|
| 1576 | XML_DefaultHandler handler)
|
|---|
| 1577 | {
|
|---|
| 1578 | defaultHandler = handler;
|
|---|
| 1579 | defaultExpandInternalEntities = 1;
|
|---|
| 1580 | }
|
|---|
| 1581 |
|
|---|
| 1582 | /*
|
|---|
| 1583 | *@@ XML_SetDoctypeDeclHandler:
|
|---|
| 1584 | * combination of XML_SetStartDoctypeDeclHandler and
|
|---|
| 1585 | * XML_SetEndDoctypeDeclHandler.
|
|---|
| 1586 | */
|
|---|
| 1587 |
|
|---|
| 1588 | void XML_SetDoctypeDeclHandler(XML_Parser parser,
|
|---|
| 1589 | XML_StartDoctypeDeclHandler start,
|
|---|
| 1590 | XML_EndDoctypeDeclHandler end)
|
|---|
| 1591 | {
|
|---|
| 1592 | startDoctypeDeclHandler = start;
|
|---|
| 1593 | endDoctypeDeclHandler = end;
|
|---|
| 1594 | }
|
|---|
| 1595 |
|
|---|
| 1596 | /*
|
|---|
| 1597 | *@@ XML_SetStartDoctypeDeclHandler:
|
|---|
| 1598 | * sets a handler that is called at the start of a DOCTYPE
|
|---|
| 1599 | * declaration, before any external or internal subset is
|
|---|
| 1600 | * parsed.
|
|---|
| 1601 | *
|
|---|
| 1602 | * Handler prototype:
|
|---|
| 1603 | *
|
|---|
| 1604 | + void EXPATENTRY StartDoctypeDeclHandler(void *pUserData,
|
|---|
| 1605 | + const XML_Char *pcszDoctypeName,
|
|---|
| 1606 | + const XML_Char *pcszSysid,
|
|---|
| 1607 | + const XML_Char *pcszPubid,
|
|---|
| 1608 | + int fHasInternalSubset);
|
|---|
| 1609 | *
|
|---|
| 1610 | * Both pcszSysid and pcszPubid may be NULL. "fHasInternalSubset"
|
|---|
| 1611 | * will be non-zero if the DOCTYPE declaration has an internal subset.
|
|---|
| 1612 | */
|
|---|
| 1613 |
|
|---|
| 1614 | void XML_SetStartDoctypeDeclHandler(XML_Parser parser,
|
|---|
| 1615 | XML_StartDoctypeDeclHandler start)
|
|---|
| 1616 | {
|
|---|
| 1617 | startDoctypeDeclHandler = start;
|
|---|
| 1618 | }
|
|---|
| 1619 |
|
|---|
| 1620 | /*
|
|---|
| 1621 | *@@ XML_SetEndDoctypeDeclHandler:
|
|---|
| 1622 | * sets a handler that is called at the end of a DOCTYPE
|
|---|
| 1623 | * declaration, after parsing any external subset.
|
|---|
| 1624 | *
|
|---|
| 1625 | * Handler prototype:
|
|---|
| 1626 | *
|
|---|
| 1627 | + void EXPATENTRY EndDoctypeDeclHandler(void *pUserData);
|
|---|
| 1628 | *
|
|---|
| 1629 | */
|
|---|
| 1630 |
|
|---|
| 1631 | void XML_SetEndDoctypeDeclHandler(XML_Parser parser,
|
|---|
| 1632 | XML_EndDoctypeDeclHandler end)
|
|---|
| 1633 | {
|
|---|
| 1634 | endDoctypeDeclHandler = end;
|
|---|
| 1635 | }
|
|---|
| 1636 |
|
|---|
| 1637 | /*
|
|---|
| 1638 | *@@ XML_SetUnparsedEntityDeclHandler:
|
|---|
| 1639 | * sets a handler that receives declarations of unparsed
|
|---|
| 1640 | * entities. These are entity declarations that have a
|
|---|
| 1641 | * notation (NDATA) field:
|
|---|
| 1642 | *
|
|---|
| 1643 | + <!ENTITY logo SYSTEM "images/logo.gif" NDATA gif>
|
|---|
| 1644 | *
|
|---|
| 1645 | * This handler is obsolete and is provided for backwards
|
|---|
| 1646 | * compatibility. Use instead XML_SetEntityDeclHandler.
|
|---|
| 1647 | */
|
|---|
| 1648 |
|
|---|
| 1649 | void XML_SetUnparsedEntityDeclHandler(XML_Parser parser,
|
|---|
| 1650 | XML_UnparsedEntityDeclHandler handler)
|
|---|
| 1651 | {
|
|---|
| 1652 | unparsedEntityDeclHandler = handler;
|
|---|
| 1653 | }
|
|---|
| 1654 |
|
|---|
| 1655 | /*
|
|---|
| 1656 | *@@ XML_SetNotationDeclHandler:
|
|---|
| 1657 | * sets a handler that receives notation declarations.
|
|---|
| 1658 | *
|
|---|
| 1659 | * Handler prototype:
|
|---|
| 1660 | +
|
|---|
| 1661 | + void EXPATENTRY NotationDeclHandler(void *pUserData,
|
|---|
| 1662 | + const XML_Char *pcszNotationName,
|
|---|
| 1663 | + const XML_Char *pcszBase,
|
|---|
| 1664 | + const XML_Char *pcszSystemId,
|
|---|
| 1665 | + const XML_Char *pcszPublicId);
|
|---|
| 1666 | +
|
|---|
| 1667 | */
|
|---|
| 1668 |
|
|---|
| 1669 | void XML_SetNotationDeclHandler(XML_Parser parser,
|
|---|
| 1670 | XML_NotationDeclHandler handler)
|
|---|
| 1671 | {
|
|---|
| 1672 | notationDeclHandler = handler;
|
|---|
| 1673 | }
|
|---|
| 1674 |
|
|---|
| 1675 | /*
|
|---|
| 1676 | *@@ XML_SetNamespaceDeclHandler:
|
|---|
| 1677 | * combination of XML_SetEndNamespaceDeclHandler and
|
|---|
| 1678 | * XML_SetEndNamespaceDeclHandler.
|
|---|
| 1679 | */
|
|---|
| 1680 |
|
|---|
| 1681 | void XML_SetNamespaceDeclHandler(XML_Parser parser,
|
|---|
| 1682 | XML_StartNamespaceDeclHandler start,
|
|---|
| 1683 | XML_EndNamespaceDeclHandler end)
|
|---|
| 1684 | {
|
|---|
| 1685 | startNamespaceDeclHandler = start;
|
|---|
| 1686 | endNamespaceDeclHandler = end;
|
|---|
| 1687 | }
|
|---|
| 1688 |
|
|---|
| 1689 | /*
|
|---|
| 1690 | *@@ XML_SetStartNamespaceDeclHandler:
|
|---|
| 1691 | * sets a handler to be called when a namespace is declared.
|
|---|
| 1692 | * Namespace declarations occur inside start tags. But the
|
|---|
| 1693 | * namespace declaration start handler is called before the
|
|---|
| 1694 | * start tag handler for each namespace declared in that start
|
|---|
| 1695 | * tag.
|
|---|
| 1696 | *
|
|---|
| 1697 | * Handler prototype:
|
|---|
| 1698 | *
|
|---|
| 1699 | + void EXPATENTRY StartNamespaceDeclHandler(void *pUserData,
|
|---|
| 1700 | + const XML_Char *prefix,
|
|---|
| 1701 | + const XML_Char *uri);
|
|---|
| 1702 | +
|
|---|
| 1703 | */
|
|---|
| 1704 |
|
|---|
| 1705 | void XML_SetStartNamespaceDeclHandler(XML_Parser parser,
|
|---|
| 1706 | XML_StartNamespaceDeclHandler start)
|
|---|
| 1707 | {
|
|---|
| 1708 | startNamespaceDeclHandler = start;
|
|---|
| 1709 | }
|
|---|
| 1710 |
|
|---|
| 1711 | /*
|
|---|
| 1712 | *@@ XML_SetEndNamespaceDeclHandler:
|
|---|
| 1713 | * sets a handler to be called when leaving the scope of a
|
|---|
| 1714 | * namespace declaration. This will be called, for each
|
|---|
| 1715 | * namespace declaration, after the handler for the end tag
|
|---|
| 1716 | * of the element in which the namespace was declared.
|
|---|
| 1717 | *
|
|---|
| 1718 | * Handler prototype:
|
|---|
| 1719 | +
|
|---|
| 1720 | + void EXPATENTRY EndNamespaceDeclHandler(void *pUserData,
|
|---|
| 1721 | + const XML_Char *prefix);
|
|---|
| 1722 | */
|
|---|
| 1723 |
|
|---|
| 1724 | void XML_SetEndNamespaceDeclHandler(XML_Parser parser,
|
|---|
| 1725 | XML_EndNamespaceDeclHandler end)
|
|---|
| 1726 | {
|
|---|
| 1727 | endNamespaceDeclHandler = end;
|
|---|
| 1728 | }
|
|---|
| 1729 |
|
|---|
| 1730 |
|
|---|
| 1731 | /*
|
|---|
| 1732 | *@@ XML_SetNotStandaloneHandler:
|
|---|
| 1733 | * sets a handler that is called if the document is not
|
|---|
| 1734 | * "standalone". This happens when there is an external
|
|---|
| 1735 | * subset or a reference to a parameter entity, but does
|
|---|
| 1736 | * not have standalone set to "yes" in an XML declaration.
|
|---|
| 1737 | * If this handler returns 0, then the parser will throw an
|
|---|
| 1738 | * ERROR_EXPAT_NOT_STANDALONE error.
|
|---|
| 1739 | *
|
|---|
| 1740 | * Handler prototype:
|
|---|
| 1741 | *
|
|---|
| 1742 | + int EXPATENTRY NotStandaloneHandler(void *pUserData);
|
|---|
| 1743 | */
|
|---|
| 1744 |
|
|---|
| 1745 | void XML_SetNotStandaloneHandler(XML_Parser parser,
|
|---|
| 1746 | XML_NotStandaloneHandler handler)
|
|---|
| 1747 | {
|
|---|
| 1748 | notStandaloneHandler = handler;
|
|---|
| 1749 | }
|
|---|
| 1750 |
|
|---|
| 1751 | /*
|
|---|
| 1752 | *@@ XML_SetExternalEntityRefHandler:
|
|---|
| 1753 | * sets a handler for references to @external_entities.
|
|---|
| 1754 | *
|
|---|
| 1755 | * This handler is also called for processing an external DTD
|
|---|
| 1756 | * subset if parameter entity parsing is in effect.
|
|---|
| 1757 | * (See XML_SetParamEntityParsing.)
|
|---|
| 1758 | *
|
|---|
| 1759 | * Warning: If you have set no ExternalEntityRefHandler, then
|
|---|
| 1760 | * external entity references are silently ignored. Otherwise,
|
|---|
| 1761 | * they trigger a call to your handler with the information
|
|---|
| 1762 | * needed to read and parse the external entity.
|
|---|
| 1763 | *
|
|---|
| 1764 | * Handler prototype:
|
|---|
| 1765 | +
|
|---|
| 1766 | + int EXPATENTRY ExternalEntityRefHandler(void *pUser,
|
|---|
| 1767 | + XML_Parser parser,
|
|---|
| 1768 | + const XML_Char *pcszContext,
|
|---|
| 1769 | + const XML_Char *pcszBase,
|
|---|
| 1770 | + const XML_Char *pcszSystemId,
|
|---|
| 1771 | + const XML_Char *pcszPublicId);
|
|---|
| 1772 | +
|
|---|
| 1773 | * The pcszContext argument specifies the parsing context in the
|
|---|
| 1774 | * format expected by the context argument to
|
|---|
| 1775 | * XML_ExternalEntityParserCreate; pcszContext is valid only until
|
|---|
| 1776 | * the handler returns, so if the referenced entity is to be
|
|---|
| 1777 | * parsed later, it must be copied.
|
|---|
| 1778 | *
|
|---|
| 1779 | * The pcszBase parameter is the base to use for relative system
|
|---|
| 1780 | * identifiers. It is set by XML_SetBase and may be null.
|
|---|
| 1781 | *
|
|---|
| 1782 | * The pcszPublicId parameter is the public id given in the entity
|
|---|
| 1783 | * declaration and may be null.
|
|---|
| 1784 | *
|
|---|
| 1785 | * The pcszSystemId is the system identifier specified in the
|
|---|
| 1786 | * entity declaration and is never null.
|
|---|
| 1787 | *
|
|---|
| 1788 | * There are a couple of ways in which this handler differs
|
|---|
| 1789 | * from others. First, this handler returns an integer. A
|
|---|
| 1790 | * non-zero value should be returned for successful handling
|
|---|
| 1791 | * of the external entity reference. Returning a zero indicates
|
|---|
| 1792 | * failure, and causes the calling parser to return an
|
|---|
| 1793 | * ERROR_EXPAT_EXTERNAL_ENTITY_HANDLING error.
|
|---|
| 1794 | *
|
|---|
| 1795 | * Second, instead of having pUserData as its first argument,
|
|---|
| 1796 | * it receives the parser that encountered the entity reference.
|
|---|
| 1797 | * This, along with the context parameter, may be used as
|
|---|
| 1798 | * arguments to a call to XML_ExternalEntityParserCreate.
|
|---|
| 1799 | * Using the returned parser, the body of the external entity
|
|---|
| 1800 | * can be recursively parsed.
|
|---|
| 1801 | *
|
|---|
| 1802 | * Since this handler may be called recursively, it should not
|
|---|
| 1803 | * be saving information into global or static variables.
|
|---|
| 1804 | *
|
|---|
| 1805 | * Your handler isn't actually responsible for parsing the entity,
|
|---|
| 1806 | * but it is responsible for creating a subsidiary parser with
|
|---|
| 1807 | * XML_ExternalEntityParserCreate that will do the job. That returns
|
|---|
| 1808 | * an instance of XML_Parser that has handlers and other data
|
|---|
| 1809 | * structures initialized from the parent parser. You may then use
|
|---|
| 1810 | * XML_Parse or XML_ParseBuffer calls against that parser. Since
|
|---|
| 1811 | * external entities may refer to other external entities, your
|
|---|
| 1812 | * handler should be prepared to be called recursively.
|
|---|
| 1813 | *
|
|---|
| 1814 | *@@changed V0.9.14 (2001-08-09) [umoeller]: changed prototype to contain user data
|
|---|
| 1815 | */
|
|---|
| 1816 |
|
|---|
| 1817 | void XML_SetExternalEntityRefHandler(XML_Parser parser,
|
|---|
| 1818 | XML_ExternalEntityRefHandler handler)
|
|---|
| 1819 | {
|
|---|
| 1820 | externalEntityRefHandler = handler;
|
|---|
| 1821 | }
|
|---|
| 1822 |
|
|---|
| 1823 | void XML_SetExternalEntityRefHandlerArg(XML_Parser parser, void *arg)
|
|---|
| 1824 | {
|
|---|
| 1825 | if (arg)
|
|---|
| 1826 | externalEntityRefHandlerArg = arg;
|
|---|
| 1827 | else
|
|---|
| 1828 | externalEntityRefHandlerArg = parser;
|
|---|
| 1829 | }
|
|---|
| 1830 |
|
|---|
| 1831 | /*
|
|---|
| 1832 | *@@ XML_SetUnknownEncodingHandler:
|
|---|
| 1833 | * sets a handler to deal with @encodings other than the
|
|---|
| 1834 | * built-in set.
|
|---|
| 1835 | *
|
|---|
| 1836 | * There are four built-in encodings in expat:
|
|---|
| 1837 | *
|
|---|
| 1838 | * -- UTF-8: 8-bit encoding of Unicode.
|
|---|
| 1839 | *
|
|---|
| 1840 | * -- UTF-16: 16-bit encoding of Unicode.
|
|---|
| 1841 | *
|
|---|
| 1842 | * -- ISO-8859-1: that's "latin 1".
|
|---|
| 1843 | *
|
|---|
| 1844 | * -- US-ASCII
|
|---|
| 1845 | *
|
|---|
| 1846 | * Anything else discovered in an encoding declaration or in
|
|---|
| 1847 | * the protocol encoding specified in the parser constructor
|
|---|
| 1848 | * triggers a call to the UnknownEncodingHandler.
|
|---|
| 1849 | *
|
|---|
| 1850 | * Handler prototype:
|
|---|
| 1851 | *
|
|---|
| 1852 | + int EXPATENTRY UnknownEncodingHandler(void *encodingHandlerData,
|
|---|
| 1853 | + const XML_Char *name,
|
|---|
| 1854 | + XML_Encoding *info);
|
|---|
| 1855 | +
|
|---|
| 1856 | * The encodingHandlerData argument is that which was passed as the
|
|---|
| 1857 | * second argument to XML_SetUnknownEncodingHandler.
|
|---|
| 1858 | *
|
|---|
| 1859 | * The name argument gives the name of the encoding as specified in
|
|---|
| 1860 | * the encoding declaration.
|
|---|
| 1861 | *
|
|---|
| 1862 | * If the callback can provide information about the encoding,
|
|---|
| 1863 | * it must fill in the XML_Encoding structure, and return 1.
|
|---|
| 1864 | * Otherwise it must return 0.
|
|---|
| 1865 | * If info does not describe a suitable encoding,
|
|---|
| 1866 | * then the parser will return an XML_UNKNOWN_ENCODING error.
|
|---|
| 1867 | *
|
|---|
| 1868 | * See _XML_Encoding for more details about that structure and
|
|---|
| 1869 | * what to do with it, and restrictions imposed by expat.
|
|---|
| 1870 | *
|
|---|
| 1871 | */
|
|---|
| 1872 |
|
|---|
| 1873 | void XML_SetUnknownEncodingHandler(XML_Parser parser,
|
|---|
| 1874 | XML_UnknownEncodingHandler handler,
|
|---|
| 1875 | void *data)
|
|---|
| 1876 | {
|
|---|
| 1877 | unknownEncodingHandler = handler;
|
|---|
| 1878 | unknownEncodingHandlerData = data;
|
|---|
| 1879 | }
|
|---|
| 1880 |
|
|---|
| 1881 | /*
|
|---|
| 1882 | *@@ XML_SetElementDeclHandler:
|
|---|
| 1883 | * sets a handler for an @element_declaration in a @DTD. The
|
|---|
| 1884 | * handler gets called with the name of the element in
|
|---|
| 1885 | * the declaration and a pointer to a structure that contains
|
|---|
| 1886 | * the element model. It is the application's responsibility
|
|---|
| 1887 | * to free this data structure.
|
|---|
| 1888 | *
|
|---|
| 1889 | * This handler must have the following prototype:
|
|---|
| 1890 | *
|
|---|
| 1891 | + void EXPATENTRY ElementDeclHandler(void *pUserData,
|
|---|
| 1892 | + const XML_Char *name,
|
|---|
| 1893 | + XMLCONTENT *model);
|
|---|
| 1894 | *
|
|---|
| 1895 | * See _XMLCONTENT for details.
|
|---|
| 1896 | */
|
|---|
| 1897 |
|
|---|
| 1898 | void XML_SetElementDeclHandler(XML_Parser parser,
|
|---|
| 1899 | XML_ElementDeclHandler eldecl)
|
|---|
| 1900 | {
|
|---|
| 1901 | elementDeclHandler = eldecl;
|
|---|
| 1902 | }
|
|---|
| 1903 |
|
|---|
| 1904 | /*
|
|---|
| 1905 | *@@ XML_SetAttlistDeclHandler:
|
|---|
| 1906 | * sets a handler for an @attribute_declaration in the @DTD.
|
|---|
| 1907 | *
|
|---|
| 1908 | * This handler must have the following prototype:
|
|---|
| 1909 | *
|
|---|
| 1910 | + void EXPATENTRY AttlistDeclHandler(void *pUserData,
|
|---|
| 1911 | + const XML_Char *pcszElementName,
|
|---|
| 1912 | + const XML_Char *pcszAttribName,
|
|---|
| 1913 | + const XML_Char *pcszAttribType,
|
|---|
| 1914 | + const XML_Char *pcszDefault,
|
|---|
| 1915 | + int fIsRequired);
|
|---|
| 1916 | *
|
|---|
| 1917 | * This handler is called for each attribute. So a single attlist
|
|---|
| 1918 | * declaration with multiple attributes declared will generate
|
|---|
| 1919 | * multiple calls to this handler.
|
|---|
| 1920 | *
|
|---|
| 1921 | * -- pcszElementName is the name of the element for which the
|
|---|
| 1922 | * attribute is being declared.
|
|---|
| 1923 | *
|
|---|
| 1924 | * -- pcszAttribName has the attribute name being declared.
|
|---|
| 1925 | *
|
|---|
| 1926 | * -- pcszAttribType is the attribute type.
|
|---|
| 1927 | * It is the string representing the type in the declaration
|
|---|
| 1928 | * with whitespace removed.
|
|---|
| 1929 | *
|
|---|
| 1930 | * -- pcszDefault holds the default value. It will be
|
|---|
| 1931 | * NULL in the case of "#IMPLIED" or "#REQUIRED" attributes.
|
|---|
| 1932 | * You can distinguish these two cases by checking the
|
|---|
| 1933 | * fIsRequired parameter, which will be true in the case of
|
|---|
| 1934 | * "#REQUIRED" attributes. Attributes which are "#FIXED"
|
|---|
| 1935 | * will have also have a TRUE fIsRequired, but they will have
|
|---|
| 1936 | * the non-NULL fixed value in the pcszDefault parameter.
|
|---|
| 1937 | */
|
|---|
| 1938 |
|
|---|
| 1939 | void XML_SetAttlistDeclHandler(XML_Parser parser,
|
|---|
| 1940 | XML_AttlistDeclHandler attdecl)
|
|---|
| 1941 | {
|
|---|
| 1942 | attlistDeclHandler = attdecl;
|
|---|
| 1943 | }
|
|---|
| 1944 |
|
|---|
| 1945 | /*
|
|---|
| 1946 | *@@ XML_SetEntityDeclHandler:
|
|---|
| 1947 | * sets a handler that will be called for all entity declarations.
|
|---|
| 1948 | *
|
|---|
| 1949 | * Handler prototype:
|
|---|
| 1950 | *
|
|---|
| 1951 | + void EXPATENTRY EntityDeclHandler(void *pUserData,
|
|---|
| 1952 | + const XML_Char *pcszEntityName,
|
|---|
| 1953 | + int fIsParameterEntity,
|
|---|
| 1954 | + const XML_Char *pcszValue,
|
|---|
| 1955 | + int iValueLength,
|
|---|
| 1956 | + const XML_Char *pcszBase,
|
|---|
| 1957 | + const XML_Char *pcszSystemId,
|
|---|
| 1958 | + const XML_Char *pcszPublicId,
|
|---|
| 1959 | + const XML_Char *pcszNotationName);
|
|---|
| 1960 | +
|
|---|
| 1961 | * The fIsParameterEntity argument will be non-zero in the case
|
|---|
| 1962 | * of parameter entities and zero otherwise.
|
|---|
| 1963 | *
|
|---|
| 1964 | * For internal entities (<!ENTITY foo "bar">), pcszValue will be
|
|---|
| 1965 | * non-NULL and pcszSystemId, pcszPublicId, and pcszNotationName
|
|---|
| 1966 | * will all be NULL. The value string is not NULL terminated; the
|
|---|
| 1967 | * lengthis provided in the iValueLength parameter. Do not use
|
|---|
| 1968 | * iValueLength to test for internal entities, since it is legal
|
|---|
| 1969 | * to have zero-length values. Instead check for whether or not
|
|---|
| 1970 | * pcszValue is NULL.
|
|---|
| 1971 | *
|
|---|
| 1972 | * The pcszNotationName argument will have a non-NULL value only
|
|---|
| 1973 | * for unparsed entity declarations.
|
|---|
| 1974 | */
|
|---|
| 1975 |
|
|---|
| 1976 | void XML_SetEntityDeclHandler(XML_Parser parser,
|
|---|
| 1977 | XML_EntityDeclHandler handler)
|
|---|
| 1978 | {
|
|---|
| 1979 | entityDeclHandler = handler;
|
|---|
| 1980 | }
|
|---|
| 1981 |
|
|---|
| 1982 | /*
|
|---|
| 1983 | *@@ XML_SetXmlDeclHandler:
|
|---|
| 1984 | * sets a handler that is called for XML declarations and also
|
|---|
| 1985 | * for text declarations discovered in external entities.
|
|---|
| 1986 | *
|
|---|
| 1987 | * Handler prototype:
|
|---|
| 1988 | *
|
|---|
| 1989 | + void EXPATENTRY XmlDeclHandler(void *pUserData,
|
|---|
| 1990 | + const XML_Char *pcszVersion,
|
|---|
| 1991 | + const XML_Char *pcszEncoding,
|
|---|
| 1992 | + int standalone);
|
|---|
| 1993 | *
|
|---|
| 1994 | * The way to distinguish is that the version parameter will
|
|---|
| 1995 | * be NULL for text declarations. The encoding parameter may
|
|---|
| 1996 | * be NULL for an XML declaration. The standalone argument will
|
|---|
| 1997 | * contain -1, 0, or 1 indicating respectively that there was no
|
|---|
| 1998 | * standalone parameter in the declaration, that it was given
|
|---|
| 1999 | * as no, or that it was given as yes.
|
|---|
| 2000 | */
|
|---|
| 2001 |
|
|---|
| 2002 | void XML_SetXmlDeclHandler(XML_Parser parser,
|
|---|
| 2003 | XML_XmlDeclHandler handler)
|
|---|
| 2004 | {
|
|---|
| 2005 | xmlDeclHandler = handler;
|
|---|
| 2006 | }
|
|---|
| 2007 |
|
|---|
| 2008 | /*
|
|---|
| 2009 | *@@ XML_SetParamEntityParsing:
|
|---|
| 2010 | * this enables parsing of @parameter_entities, including the
|
|---|
| 2011 | * external parameter entity that is the external @DTD subset,
|
|---|
| 2012 | * according to code.
|
|---|
| 2013 | *
|
|---|
| 2014 | * If parsing of parameter entities is enabled, then references
|
|---|
| 2015 | * to external parameter entities (including the external DTD
|
|---|
| 2016 | * subset) will be passed to the handler set with
|
|---|
| 2017 | * XML_SetExternalEntityRefHandler. The context passed will be 0.
|
|---|
| 2018 | *
|
|---|
| 2019 | * Unlike external general entities, external parameter entities
|
|---|
| 2020 | * can only be parsed synchronously. If the external parameter
|
|---|
| 2021 | * entity is to be parsed, it must be parsed during the call to
|
|---|
| 2022 | * the external entity ref handler: the complete sequence of
|
|---|
| 2023 | * XML_ExternalEntityParserCreate, XML_Parse/XML_ParseBuffer and
|
|---|
| 2024 | * XML_ParserFree calls must be made during this call.
|
|---|
| 2025 | *
|
|---|
| 2026 | * After XML_ExternalEntityParserCreate has been called to create
|
|---|
| 2027 | * the parser for the external parameter entity (context must be 0
|
|---|
| 2028 | * for this call), it is illegal to make any calls on the old parser
|
|---|
| 2029 | * until XML_ParserFree has been called on the newly created parser.
|
|---|
| 2030 | * If the library has been compiled without support for parameter
|
|---|
| 2031 | * entity parsing (ie without XML_DTD being defined), then
|
|---|
| 2032 | * XML_SetParamEntityParsing will return 0 if parsing of parameter
|
|---|
| 2033 | * entities is requested; otherwise it will return non-zero.
|
|---|
| 2034 | *
|
|---|
| 2035 | * NOTE: The xwphelpers implementation #defines XML_DTD in
|
|---|
| 2036 | * include\expat\expat_setup.h, which is included in all expat
|
|---|
| 2037 | * files.
|
|---|
| 2038 | *
|
|---|
| 2039 | * The choices for code are:
|
|---|
| 2040 | *
|
|---|
| 2041 | * -- XML_PARAM_ENTITY_PARSING_NEVER:
|
|---|
| 2042 | * Don't parse parameter entities or the external subset.
|
|---|
| 2043 | * This is the only choice if XML_DTD is not defined.
|
|---|
| 2044 | *
|
|---|
| 2045 | * -- XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE:
|
|---|
| 2046 | * Parse parameter entities and the external subset unless
|
|---|
| 2047 | * standalone was set to "yes" in the XML declaration.
|
|---|
| 2048 | *
|
|---|
| 2049 | * -- XML_PARAM_ENTITY_PARSING_ALWAYS
|
|---|
| 2050 | * Always parse parameter entities and the external subset.
|
|---|
| 2051 | *
|
|---|
| 2052 | * In order to read an external DTD, you also have to set an
|
|---|
| 2053 | * external entity reference handler (XML_SetExternalEntityRefHandler).
|
|---|
| 2054 | */
|
|---|
| 2055 |
|
|---|
| 2056 | int XML_SetParamEntityParsing(XML_Parser parser,
|
|---|
| 2057 | enum XML_ParamEntityParsing parsing)
|
|---|
| 2058 | {
|
|---|
| 2059 | #ifdef XML_DTD
|
|---|
| 2060 | paramEntityParsing = parsing;
|
|---|
| 2061 | return 1;
|
|---|
| 2062 | #else
|
|---|
| 2063 | return parsing == XML_PARAM_ENTITY_PARSING_NEVER;
|
|---|
| 2064 | #endif
|
|---|
| 2065 | }
|
|---|
| 2066 |
|
|---|
| 2067 | /*
|
|---|
| 2068 | *@@ XML_Parse:
|
|---|
| 2069 | * parses some more of the document.
|
|---|
| 2070 | *
|
|---|
| 2071 | * The string s is a buffer containing part (or perhaps all)
|
|---|
| 2072 | * of the document. The number of bytes of s that are part
|
|---|
| 2073 | * of the document is indicated by len. This means that s
|
|---|
| 2074 | * doesn't have to be null terminated. It also means that
|
|---|
| 2075 | * if len is larger than the number of bytes in the block
|
|---|
| 2076 | * of memory that s points at, then a memory fault is
|
|---|
| 2077 | * likely. The isFinal parameter informs the parser that
|
|---|
| 2078 | * this is the last piece of the document. Frequently, the
|
|---|
| 2079 | * last piece is empty (i.e. len is zero.)
|
|---|
| 2080 | *
|
|---|
| 2081 | * If a parse error occurred, it returns 0. Otherwise it
|
|---|
| 2082 | * returns a non-zero value.
|
|---|
| 2083 | */
|
|---|
| 2084 |
|
|---|
| 2085 | int XML_Parse(XML_Parser parser,
|
|---|
| 2086 | const char *s,
|
|---|
| 2087 | int len,
|
|---|
| 2088 | int isFinal)
|
|---|
| 2089 | {
|
|---|
| 2090 | if (len == 0)
|
|---|
| 2091 | {
|
|---|
| 2092 | if (!isFinal)
|
|---|
| 2093 | return 1;
|
|---|
| 2094 | positionPtr = bufferPtr;
|
|---|
| 2095 | errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0);
|
|---|
| 2096 | if (errorCode == ERROR_EXPAT_NONE)
|
|---|
| 2097 | return 1;
|
|---|
| 2098 | eventEndPtr = eventPtr;
|
|---|
| 2099 | processor = errorProcessor;
|
|---|
| 2100 | return 0;
|
|---|
| 2101 | }
|
|---|
| 2102 | #ifndef XML_CONTEXT_BYTES
|
|---|
| 2103 | else if (bufferPtr == bufferEnd)
|
|---|
| 2104 | {
|
|---|
| 2105 | const char *end;
|
|---|
| 2106 | int nLeftOver;
|
|---|
| 2107 |
|
|---|
| 2108 | parseEndByteIndex += len;
|
|---|
| 2109 | positionPtr = s;
|
|---|
| 2110 | if (isFinal)
|
|---|
| 2111 | {
|
|---|
| 2112 | errorCode = processor(parser, s, parseEndPtr = s + len, 0);
|
|---|
| 2113 | if (errorCode == ERROR_EXPAT_NONE)
|
|---|
| 2114 | return 1;
|
|---|
| 2115 | eventEndPtr = eventPtr;
|
|---|
| 2116 | processor = errorProcessor;
|
|---|
| 2117 | return 0;
|
|---|
| 2118 | }
|
|---|
| 2119 | errorCode = processor(parser, s, parseEndPtr = s + len, &end);
|
|---|
| 2120 | if (errorCode != ERROR_EXPAT_NONE)
|
|---|
| 2121 | {
|
|---|
| 2122 | eventEndPtr = eventPtr;
|
|---|
| 2123 | processor = errorProcessor;
|
|---|
| 2124 | return 0;
|
|---|
| 2125 | }
|
|---|
| 2126 | XmlUpdatePosition(encoding, positionPtr, end, &position);
|
|---|
| 2127 | nLeftOver = s + len - end;
|
|---|
| 2128 | if (nLeftOver)
|
|---|
| 2129 | {
|
|---|
| 2130 | if (buffer == 0 || nLeftOver > bufferLim - buffer)
|
|---|
| 2131 | {
|
|---|
| 2132 | /* FIXME avoid integer overflow */
|
|---|
| 2133 | buffer = buffer == 0 ? MALLOC(len * 2) : REALLOC(buffer, len * 2);
|
|---|
| 2134 | /* FIXME storage leak if realloc fails */
|
|---|
| 2135 | if (!buffer)
|
|---|
| 2136 | {
|
|---|
| 2137 | errorCode = ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2138 | eventPtr = eventEndPtr = 0;
|
|---|
| 2139 | processor = errorProcessor;
|
|---|
| 2140 | return 0;
|
|---|
| 2141 | }
|
|---|
| 2142 | bufferLim = buffer + len * 2;
|
|---|
| 2143 | }
|
|---|
| 2144 | memcpy(buffer, end, nLeftOver);
|
|---|
| 2145 | bufferPtr = buffer;
|
|---|
| 2146 | bufferEnd = buffer + nLeftOver;
|
|---|
| 2147 | }
|
|---|
| 2148 | return 1;
|
|---|
| 2149 | }
|
|---|
| 2150 | #endif /* not defined XML_CONTEXT_BYTES */
|
|---|
| 2151 | else
|
|---|
| 2152 | {
|
|---|
| 2153 | memcpy(XML_GetBuffer(parser, len), s, len);
|
|---|
| 2154 | return XML_ParseBuffer(parser, len, isFinal);
|
|---|
| 2155 | }
|
|---|
| 2156 | }
|
|---|
| 2157 |
|
|---|
| 2158 | /*
|
|---|
| 2159 | *@@ XML_ParseBuffer:
|
|---|
| 2160 | * this is just like XML_Parse, except in this case expat
|
|---|
| 2161 | * provides the buffer. By obtaining the buffer from expat
|
|---|
| 2162 | * with the XML_GetBuffer function, the application can
|
|---|
| 2163 | * avoid double copying of the input.
|
|---|
| 2164 | */
|
|---|
| 2165 |
|
|---|
| 2166 | int XML_ParseBuffer(XML_Parser parser,
|
|---|
| 2167 | int len,
|
|---|
| 2168 | int isFinal)
|
|---|
| 2169 | {
|
|---|
| 2170 | const char *start = bufferPtr;
|
|---|
| 2171 |
|
|---|
| 2172 | positionPtr = start;
|
|---|
| 2173 | bufferEnd += len;
|
|---|
| 2174 | parseEndByteIndex += len;
|
|---|
| 2175 | errorCode = processor(parser, start, parseEndPtr = bufferEnd,
|
|---|
| 2176 | isFinal ? (const char **)0 : &bufferPtr);
|
|---|
| 2177 | if (errorCode == ERROR_EXPAT_NONE)
|
|---|
| 2178 | {
|
|---|
| 2179 | if (!isFinal)
|
|---|
| 2180 | XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
|
|---|
| 2181 | return 1;
|
|---|
| 2182 | }
|
|---|
| 2183 | else
|
|---|
| 2184 | {
|
|---|
| 2185 | eventEndPtr = eventPtr;
|
|---|
| 2186 | processor = errorProcessor;
|
|---|
| 2187 | return 0;
|
|---|
| 2188 | }
|
|---|
| 2189 | }
|
|---|
| 2190 |
|
|---|
| 2191 | /*
|
|---|
| 2192 | *@@ XML_GetBuffer:
|
|---|
| 2193 | * obtains a buffer of size len to read a piece of the
|
|---|
| 2194 | * document into. A NULL value is returned if expat can't
|
|---|
| 2195 | * allocate enough memory for this buffer. This has to be
|
|---|
| 2196 | * called prior to every call to XML_ParseBuffer.
|
|---|
| 2197 | *
|
|---|
| 2198 | * A typical use would look like this:
|
|---|
| 2199 | *
|
|---|
| 2200 | + for (;;)
|
|---|
| 2201 | + {
|
|---|
| 2202 | + int bytes_read;
|
|---|
| 2203 | + void *buff = XML_GetBuffer(p, BUFF_SIZE);
|
|---|
| 2204 | + if (buff == NULL)
|
|---|
| 2205 | + // handle error
|
|---|
| 2206 | +
|
|---|
| 2207 | + bytes_read = read(docfd, buff, BUFF_SIZE);
|
|---|
| 2208 | + if (bytes_read < 0)
|
|---|
| 2209 | + // handle error
|
|---|
| 2210 | +
|
|---|
| 2211 | + if (!XML_ParseBuffer(p, bytes_read, bytes_read == 0))
|
|---|
| 2212 | + // handle parse error
|
|---|
| 2213 | +
|
|---|
| 2214 | + if (bytes_read == 0)
|
|---|
| 2215 | + break;
|
|---|
| 2216 | + }
|
|---|
| 2217 | *
|
|---|
| 2218 | */
|
|---|
| 2219 |
|
|---|
| 2220 | void *XML_GetBuffer(XML_Parser parser,
|
|---|
| 2221 | int len)
|
|---|
| 2222 | {
|
|---|
| 2223 | if (len > bufferLim - bufferEnd)
|
|---|
| 2224 | {
|
|---|
| 2225 | /* FIXME avoid integer overflow */
|
|---|
| 2226 | int neededSize = len + (bufferEnd - bufferPtr);
|
|---|
| 2227 |
|
|---|
| 2228 | #ifdef XML_CONTEXT_BYTES
|
|---|
| 2229 | int keep = bufferPtr - buffer;
|
|---|
| 2230 |
|
|---|
| 2231 | if (keep > XML_CONTEXT_BYTES)
|
|---|
| 2232 | keep = XML_CONTEXT_BYTES;
|
|---|
| 2233 | neededSize += keep;
|
|---|
| 2234 | #endif /* defined XML_CONTEXT_BYTES */
|
|---|
| 2235 | if (neededSize <= bufferLim - buffer)
|
|---|
| 2236 | {
|
|---|
| 2237 | #ifdef XML_CONTEXT_BYTES
|
|---|
| 2238 | if (keep < bufferPtr - buffer)
|
|---|
| 2239 | {
|
|---|
| 2240 | int offset = (bufferPtr - buffer) - keep;
|
|---|
| 2241 |
|
|---|
| 2242 | memmove(buffer, &buffer[offset], bufferEnd - bufferPtr + keep);
|
|---|
| 2243 | bufferEnd -= offset;
|
|---|
| 2244 | bufferPtr -= offset;
|
|---|
| 2245 | }
|
|---|
| 2246 | #else
|
|---|
| 2247 | memmove(buffer, bufferPtr, bufferEnd - bufferPtr);
|
|---|
| 2248 | bufferEnd = buffer + (bufferEnd - bufferPtr);
|
|---|
| 2249 | bufferPtr = buffer;
|
|---|
| 2250 | #endif /* not defined XML_CONTEXT_BYTES */
|
|---|
| 2251 | }
|
|---|
| 2252 | else
|
|---|
| 2253 | {
|
|---|
| 2254 | char *newBuf;
|
|---|
| 2255 | int bufferSize = bufferLim - bufferPtr;
|
|---|
| 2256 |
|
|---|
| 2257 | if (bufferSize == 0)
|
|---|
| 2258 | bufferSize = INIT_BUFFER_SIZE;
|
|---|
| 2259 | do
|
|---|
| 2260 | {
|
|---|
| 2261 | bufferSize *= 2;
|
|---|
| 2262 | }
|
|---|
| 2263 | while (bufferSize < neededSize);
|
|---|
| 2264 | newBuf = (char*)MALLOC(bufferSize);
|
|---|
| 2265 | if (newBuf == 0)
|
|---|
| 2266 | {
|
|---|
| 2267 | errorCode = ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2268 | return 0;
|
|---|
| 2269 | }
|
|---|
| 2270 | bufferLim = newBuf + bufferSize;
|
|---|
| 2271 | #ifdef XML_CONTEXT_BYTES
|
|---|
| 2272 | if (bufferPtr)
|
|---|
| 2273 | {
|
|---|
| 2274 | int keep = bufferPtr - buffer;
|
|---|
| 2275 |
|
|---|
| 2276 | if (keep > XML_CONTEXT_BYTES)
|
|---|
| 2277 | keep = XML_CONTEXT_BYTES;
|
|---|
| 2278 | memcpy(newBuf, &bufferPtr[-keep], bufferEnd - bufferPtr + keep);
|
|---|
| 2279 | FREE(buffer);
|
|---|
| 2280 | buffer = newBuf;
|
|---|
| 2281 | bufferEnd = buffer + (bufferEnd - bufferPtr) + keep;
|
|---|
| 2282 | bufferPtr = buffer + keep;
|
|---|
| 2283 | }
|
|---|
| 2284 | else
|
|---|
| 2285 | {
|
|---|
| 2286 | bufferEnd = newBuf + (bufferEnd - bufferPtr);
|
|---|
| 2287 | bufferPtr = buffer = newBuf;
|
|---|
| 2288 | }
|
|---|
| 2289 | #else
|
|---|
| 2290 | if (bufferPtr)
|
|---|
| 2291 | {
|
|---|
| 2292 | memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr);
|
|---|
| 2293 | FREE(buffer);
|
|---|
| 2294 | }
|
|---|
| 2295 | bufferEnd = newBuf + (bufferEnd - bufferPtr);
|
|---|
| 2296 | bufferPtr = buffer = newBuf;
|
|---|
| 2297 | #endif /* not defined XML_CONTEXT_BYTES */
|
|---|
| 2298 | }
|
|---|
| 2299 | }
|
|---|
| 2300 | return bufferEnd;
|
|---|
| 2301 | }
|
|---|
| 2302 |
|
|---|
| 2303 | /*
|
|---|
| 2304 | *@@ XML_GetErrorCode:
|
|---|
| 2305 | * returns what type of error has occurred.
|
|---|
| 2306 | *
|
|---|
| 2307 | * To be called when the parse functions return 0 (i.e. a
|
|---|
| 2308 | * parse error has ocurred), although the position reporting
|
|---|
| 2309 | * functions are useful outside of errors. The position reported
|
|---|
| 2310 | * is the byte position (in the original document or entity
|
|---|
| 2311 | * encoding) of the first of the sequence of characters that
|
|---|
| 2312 | * generated the current event (or the error that caused the
|
|---|
| 2313 | * parse functions to return 0.)
|
|---|
| 2314 | *
|
|---|
| 2315 | * The position reporting functions are accurate only outside
|
|---|
| 2316 | * of the DTD. In other words, they usually return bogus
|
|---|
| 2317 | * information when called from within a DTD declaration handler.
|
|---|
| 2318 | */
|
|---|
| 2319 |
|
|---|
| 2320 | XMLERROR XML_GetErrorCode(XML_Parser parser)
|
|---|
| 2321 | {
|
|---|
| 2322 | return errorCode;
|
|---|
| 2323 | }
|
|---|
| 2324 |
|
|---|
| 2325 | /*
|
|---|
| 2326 | *@@ XML_GetCurrentByteIndex:
|
|---|
| 2327 | * returns the byte offset of the position.
|
|---|
| 2328 | *
|
|---|
| 2329 | * To be called on parser errors. See XML_GetErrorCode.
|
|---|
| 2330 | */
|
|---|
| 2331 |
|
|---|
| 2332 | long XML_GetCurrentByteIndex(XML_Parser parser)
|
|---|
| 2333 | {
|
|---|
| 2334 | if (eventPtr)
|
|---|
| 2335 | return parseEndByteIndex - (parseEndPtr - eventPtr);
|
|---|
| 2336 | return -1;
|
|---|
| 2337 | }
|
|---|
| 2338 |
|
|---|
| 2339 | /*
|
|---|
| 2340 | *@@ XML_GetCurrentByteCount:
|
|---|
| 2341 | * return the number of bytes in the current event. Returns
|
|---|
| 2342 | * 0 if the event is inside a reference to an internal entity.
|
|---|
| 2343 | *
|
|---|
| 2344 | * To be called on parser errors. See XML_GetErrorCode.
|
|---|
| 2345 | */
|
|---|
| 2346 |
|
|---|
| 2347 | int XML_GetCurrentByteCount(XML_Parser parser)
|
|---|
| 2348 | {
|
|---|
| 2349 | if (eventEndPtr && eventPtr)
|
|---|
| 2350 | return eventEndPtr - eventPtr;
|
|---|
| 2351 | return 0;
|
|---|
| 2352 | }
|
|---|
| 2353 |
|
|---|
| 2354 | /*
|
|---|
| 2355 | *@@ XML_GetInputContext:
|
|---|
| 2356 | * returns the parser's input buffer, sets the integer pointed
|
|---|
| 2357 | * at by offset to the offset within this buffer of the current
|
|---|
| 2358 | * parse position, and set the integer pointed at by size to the
|
|---|
| 2359 | * size of the returned buffer.
|
|---|
| 2360 | *
|
|---|
| 2361 | * This should only be called from within a handler during an
|
|---|
| 2362 | * active parse and the returned buffer should only be referred
|
|---|
| 2363 | * to from within the handler that made the call. This input
|
|---|
| 2364 | * buffer contains the untranslated bytes of the input.
|
|---|
| 2365 | *
|
|---|
| 2366 | * Only a limited amount of context is kept, so if the event
|
|---|
| 2367 | * triggering a call spans over a very large amount of input,
|
|---|
| 2368 | * the actual parse position may be before the beginning of the buffer.
|
|---|
| 2369 | */
|
|---|
| 2370 |
|
|---|
| 2371 | const char *XML_GetInputContext(XML_Parser parser,
|
|---|
| 2372 | int *offset,
|
|---|
| 2373 | int *size)
|
|---|
| 2374 | {
|
|---|
| 2375 | #ifdef XML_CONTEXT_BYTES
|
|---|
| 2376 | if (eventPtr && buffer)
|
|---|
| 2377 | {
|
|---|
| 2378 | *offset = eventPtr - buffer;
|
|---|
| 2379 | *size = bufferEnd - buffer;
|
|---|
| 2380 | return buffer;
|
|---|
| 2381 | }
|
|---|
| 2382 | #endif /* defined XML_CONTEXT_BYTES */
|
|---|
| 2383 | return (char *)0;
|
|---|
| 2384 | }
|
|---|
| 2385 |
|
|---|
| 2386 | /*
|
|---|
| 2387 | *@@ XML_GetCurrentLineNumber:
|
|---|
| 2388 | * returns the line number of the position.
|
|---|
| 2389 | *
|
|---|
| 2390 | * Can be called on parser errors. See XML_GetErrorCode.
|
|---|
| 2391 | */
|
|---|
| 2392 |
|
|---|
| 2393 | int XML_GetCurrentLineNumber(XML_Parser parser)
|
|---|
| 2394 | {
|
|---|
| 2395 | if (eventPtr)
|
|---|
| 2396 | {
|
|---|
| 2397 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
|
|---|
| 2398 | positionPtr = eventPtr;
|
|---|
| 2399 | }
|
|---|
| 2400 | return position.lineNumber + 1;
|
|---|
| 2401 | }
|
|---|
| 2402 |
|
|---|
| 2403 | /*
|
|---|
| 2404 | *@@ XML_GetCurrentColumnNumber:
|
|---|
| 2405 | * returns the offset, from the beginning of the current
|
|---|
| 2406 | * line, of the position.
|
|---|
| 2407 | *
|
|---|
| 2408 | * Can be called on parser errors. See XML_GetErrorCode.
|
|---|
| 2409 | */
|
|---|
| 2410 |
|
|---|
| 2411 | int XML_GetCurrentColumnNumber(XML_Parser parser)
|
|---|
| 2412 | {
|
|---|
| 2413 | if (eventPtr)
|
|---|
| 2414 | {
|
|---|
| 2415 | XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
|
|---|
| 2416 | positionPtr = eventPtr;
|
|---|
| 2417 | }
|
|---|
| 2418 | return position.columnNumber;
|
|---|
| 2419 | }
|
|---|
| 2420 |
|
|---|
| 2421 | /*
|
|---|
| 2422 | *@@ XML_DefaultCurrent:
|
|---|
| 2423 | * not in the expat docs.
|
|---|
| 2424 | */
|
|---|
| 2425 |
|
|---|
| 2426 | void XML_DefaultCurrent(XML_Parser parser)
|
|---|
| 2427 | {
|
|---|
| 2428 | if (defaultHandler)
|
|---|
| 2429 | {
|
|---|
| 2430 | if (openInternalEntities)
|
|---|
| 2431 | reportDefault(parser,
|
|---|
| 2432 | internalEncoding,
|
|---|
| 2433 | openInternalEntities->internalEventPtr,
|
|---|
| 2434 | openInternalEntities->internalEventEndPtr);
|
|---|
| 2435 | else
|
|---|
| 2436 | reportDefault(parser, encoding, eventPtr, eventEndPtr);
|
|---|
| 2437 | }
|
|---|
| 2438 | }
|
|---|
| 2439 |
|
|---|
| 2440 | const XML_LChar *XML_ErrorString(int code)
|
|---|
| 2441 | {
|
|---|
| 2442 | static const XML_LChar *message[] =
|
|---|
| 2443 | {
|
|---|
| 2444 | 0,
|
|---|
| 2445 | XML_T("out of memory"),
|
|---|
| 2446 | XML_T("syntax error"),
|
|---|
| 2447 | XML_T("no element found"),
|
|---|
| 2448 | XML_T("not well-formed (invalid token)"),
|
|---|
| 2449 | XML_T("unclosed token"),
|
|---|
| 2450 | XML_T("unclosed token"),
|
|---|
| 2451 | XML_T("mismatched tag"),
|
|---|
| 2452 | XML_T("duplicate attribute"),
|
|---|
| 2453 | XML_T("junk after document element"),
|
|---|
| 2454 | XML_T("illegal parameter entity reference"),
|
|---|
| 2455 | XML_T("undefined entity"),
|
|---|
| 2456 | XML_T("recursive entity reference"),
|
|---|
| 2457 | XML_T("asynchronous entity"),
|
|---|
| 2458 | XML_T("reference to invalid character number"),
|
|---|
| 2459 | XML_T("reference to binary entity"),
|
|---|
| 2460 | XML_T("reference to external entity in attribute"),
|
|---|
| 2461 | XML_T("xml processing instruction not at start of external entity"),
|
|---|
| 2462 | XML_T("unknown encoding"),
|
|---|
| 2463 | XML_T("encoding specified in XML declaration is incorrect"),
|
|---|
| 2464 | XML_T("unclosed CDATA section"),
|
|---|
| 2465 | XML_T("error in processing external entity reference"),
|
|---|
| 2466 | XML_T("document is not standalone"),
|
|---|
| 2467 | XML_T("unexpected parser state - please send a bug report")
|
|---|
| 2468 | };
|
|---|
| 2469 |
|
|---|
| 2470 | if (code > 0 && code < sizeof(message) / sizeof(message[0]))
|
|---|
| 2471 | return message[code];
|
|---|
| 2472 | return 0;
|
|---|
| 2473 | }
|
|---|
| 2474 |
|
|---|
| 2475 | /* const XML_LChar *
|
|---|
| 2476 | * XML_ExpatVersion(void) {
|
|---|
| 2477 | * return VERSION;
|
|---|
| 2478 | * }
|
|---|
| 2479 | *
|
|---|
| 2480 | * XML_Expat_Version
|
|---|
| 2481 | * XML_ExpatVersionInfo(void) {
|
|---|
| 2482 | * XML_Expat_Version version;
|
|---|
| 2483 | *
|
|---|
| 2484 | * version.major = XML_MAJOR_VERSION;
|
|---|
| 2485 | * version.minor = XML_MINOR_VERSION;
|
|---|
| 2486 | * version.micro = XML_MICRO_VERSION;
|
|---|
| 2487 | *
|
|---|
| 2488 | * return version;
|
|---|
| 2489 | * } */
|
|---|
| 2490 |
|
|---|
| 2491 | STATIC XMLERROR contentProcessor(XML_Parser parser,
|
|---|
| 2492 | const char *start,
|
|---|
| 2493 | const char *end,
|
|---|
| 2494 | const char **endPtr)
|
|---|
| 2495 | {
|
|---|
| 2496 | return doContent(parser, 0, encoding, start, end, endPtr);
|
|---|
| 2497 | }
|
|---|
| 2498 |
|
|---|
| 2499 | STATIC XMLERROR externalEntityInitProcessor(XML_Parser parser,
|
|---|
| 2500 | const char *start,
|
|---|
| 2501 | const char *end,
|
|---|
| 2502 | const char **endPtr)
|
|---|
| 2503 | {
|
|---|
| 2504 | XMLERROR result = initializeEncoding(parser);
|
|---|
| 2505 |
|
|---|
| 2506 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 2507 | return result;
|
|---|
| 2508 | processor = externalEntityInitProcessor2;
|
|---|
| 2509 | return externalEntityInitProcessor2(parser, start, end, endPtr);
|
|---|
| 2510 | }
|
|---|
| 2511 |
|
|---|
| 2512 | STATIC XMLERROR externalEntityInitProcessor2(XML_Parser parser,
|
|---|
| 2513 | const char *start,
|
|---|
| 2514 | const char *end,
|
|---|
| 2515 | const char **endPtr)
|
|---|
| 2516 | {
|
|---|
| 2517 | const char *next;
|
|---|
| 2518 | int tok = XmlContentTok(encoding, start, end, &next);
|
|---|
| 2519 |
|
|---|
| 2520 | switch (tok)
|
|---|
| 2521 | {
|
|---|
| 2522 | case XML_TOK_BOM:
|
|---|
| 2523 | start = next;
|
|---|
| 2524 | break;
|
|---|
| 2525 | case XML_TOK_PARTIAL:
|
|---|
| 2526 | if (endPtr)
|
|---|
| 2527 | {
|
|---|
| 2528 | *endPtr = start;
|
|---|
| 2529 | return ERROR_EXPAT_NONE;
|
|---|
| 2530 | }
|
|---|
| 2531 | eventPtr = start;
|
|---|
| 2532 | return ERROR_EXPAT_UNCLOSED_TOKEN;
|
|---|
| 2533 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 2534 | if (endPtr)
|
|---|
| 2535 | {
|
|---|
| 2536 | *endPtr = start;
|
|---|
| 2537 | return ERROR_EXPAT_NONE;
|
|---|
| 2538 | }
|
|---|
| 2539 | eventPtr = start;
|
|---|
| 2540 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 2541 | }
|
|---|
| 2542 | processor = externalEntityInitProcessor3;
|
|---|
| 2543 | return externalEntityInitProcessor3(parser, start, end, endPtr);
|
|---|
| 2544 | }
|
|---|
| 2545 |
|
|---|
| 2546 | STATIC XMLERROR externalEntityInitProcessor3(XML_Parser parser,
|
|---|
| 2547 | const char *start,
|
|---|
| 2548 | const char *end,
|
|---|
| 2549 | const char **endPtr)
|
|---|
| 2550 | {
|
|---|
| 2551 | const char *next;
|
|---|
| 2552 | int tok = XmlContentTok(encoding, start, end, &next);
|
|---|
| 2553 |
|
|---|
| 2554 | switch (tok)
|
|---|
| 2555 | {
|
|---|
| 2556 | case XML_TOK_XML_DECL:
|
|---|
| 2557 | {
|
|---|
| 2558 | XMLERROR result = processXmlDecl(parser, 1, start, next);
|
|---|
| 2559 |
|
|---|
| 2560 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 2561 | return result;
|
|---|
| 2562 | start = next;
|
|---|
| 2563 | }
|
|---|
| 2564 | break;
|
|---|
| 2565 | case XML_TOK_PARTIAL:
|
|---|
| 2566 | if (endPtr)
|
|---|
| 2567 | {
|
|---|
| 2568 | *endPtr = start;
|
|---|
| 2569 | return ERROR_EXPAT_NONE;
|
|---|
| 2570 | }
|
|---|
| 2571 | eventPtr = start;
|
|---|
| 2572 | return ERROR_EXPAT_UNCLOSED_TOKEN;
|
|---|
| 2573 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 2574 | if (endPtr)
|
|---|
| 2575 | {
|
|---|
| 2576 | *endPtr = start;
|
|---|
| 2577 | return ERROR_EXPAT_NONE;
|
|---|
| 2578 | }
|
|---|
| 2579 | eventPtr = start;
|
|---|
| 2580 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 2581 | }
|
|---|
| 2582 | processor = externalEntityContentProcessor;
|
|---|
| 2583 | tagLevel = 1;
|
|---|
| 2584 | return doContent(parser, 1, encoding, start, end, endPtr);
|
|---|
| 2585 | }
|
|---|
| 2586 |
|
|---|
| 2587 | STATIC XMLERROR externalEntityContentProcessor(XML_Parser parser,
|
|---|
| 2588 | const char *start,
|
|---|
| 2589 | const char *end,
|
|---|
| 2590 | const char **endPtr)
|
|---|
| 2591 | {
|
|---|
| 2592 | return doContent(parser, 1, encoding, start, end, endPtr);
|
|---|
| 2593 | }
|
|---|
| 2594 |
|
|---|
| 2595 | /*
|
|---|
| 2596 | *@@ doContent:
|
|---|
| 2597 | *
|
|---|
| 2598 | *@@changed V0.9.14 (2001-08-09) [umoeller]: fixed ERROR_EXPAT_UNDEFINED_ENTITY with callback-defined encodings
|
|---|
| 2599 | */
|
|---|
| 2600 |
|
|---|
| 2601 | STATIC XMLERROR doContent(XML_Parser parser,
|
|---|
| 2602 | int startTagLevel,
|
|---|
| 2603 | const ENCODING * enc,
|
|---|
| 2604 | const char *s,
|
|---|
| 2605 | const char *end,
|
|---|
| 2606 | const char **nextPtr)
|
|---|
| 2607 | {
|
|---|
| 2608 | const char **eventPP;
|
|---|
| 2609 | const char **eventEndPP;
|
|---|
| 2610 |
|
|---|
| 2611 | if (enc == encoding)
|
|---|
| 2612 | {
|
|---|
| 2613 | eventPP = &eventPtr;
|
|---|
| 2614 | eventEndPP = &eventEndPtr;
|
|---|
| 2615 | }
|
|---|
| 2616 | else
|
|---|
| 2617 | {
|
|---|
| 2618 | eventPP = &(openInternalEntities->internalEventPtr);
|
|---|
| 2619 | eventEndPP = &(openInternalEntities->internalEventEndPtr);
|
|---|
| 2620 | }
|
|---|
| 2621 | *eventPP = s;
|
|---|
| 2622 | for (;;)
|
|---|
| 2623 | {
|
|---|
| 2624 | const char *next = s; /* XmlContentTok doesn't always set the last arg */
|
|---|
| 2625 | int tok = XmlContentTok(enc, s, end, &next);
|
|---|
| 2626 |
|
|---|
| 2627 | *eventEndPP = next;
|
|---|
| 2628 | switch (tok)
|
|---|
| 2629 | {
|
|---|
| 2630 | case XML_TOK_TRAILING_CR:
|
|---|
| 2631 | if (nextPtr)
|
|---|
| 2632 | {
|
|---|
| 2633 | *nextPtr = s;
|
|---|
| 2634 | return ERROR_EXPAT_NONE;
|
|---|
| 2635 | }
|
|---|
| 2636 | *eventEndPP = end;
|
|---|
| 2637 | if (characterDataHandler)
|
|---|
| 2638 | {
|
|---|
| 2639 | XML_Char c = 0xA;
|
|---|
| 2640 |
|
|---|
| 2641 | characterDataHandler(handlerArg, &c, 1);
|
|---|
| 2642 | }
|
|---|
| 2643 | else if (defaultHandler)
|
|---|
| 2644 | reportDefault(parser, enc, s, end);
|
|---|
| 2645 | if (startTagLevel == 0)
|
|---|
| 2646 | return ERROR_EXPAT_NO_ELEMENTS;
|
|---|
| 2647 | if (tagLevel != startTagLevel)
|
|---|
| 2648 | return ERROR_EXPAT_ASYNC_ENTITY;
|
|---|
| 2649 | return ERROR_EXPAT_NONE;
|
|---|
| 2650 | case XML_TOK_NONE:
|
|---|
| 2651 | if (nextPtr)
|
|---|
| 2652 | {
|
|---|
| 2653 | *nextPtr = s;
|
|---|
| 2654 | return ERROR_EXPAT_NONE;
|
|---|
| 2655 | }
|
|---|
| 2656 | if (startTagLevel > 0)
|
|---|
| 2657 | {
|
|---|
| 2658 | if (tagLevel != startTagLevel)
|
|---|
| 2659 | return ERROR_EXPAT_ASYNC_ENTITY;
|
|---|
| 2660 | return ERROR_EXPAT_NONE;
|
|---|
| 2661 | }
|
|---|
| 2662 | return ERROR_EXPAT_NO_ELEMENTS;
|
|---|
| 2663 | case XML_TOK_INVALID:
|
|---|
| 2664 | *eventPP = next;
|
|---|
| 2665 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 2666 | case XML_TOK_PARTIAL:
|
|---|
| 2667 | if (nextPtr)
|
|---|
| 2668 | {
|
|---|
| 2669 | *nextPtr = s;
|
|---|
| 2670 | return ERROR_EXPAT_NONE;
|
|---|
| 2671 | }
|
|---|
| 2672 | return ERROR_EXPAT_UNCLOSED_TOKEN;
|
|---|
| 2673 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 2674 | if (nextPtr)
|
|---|
| 2675 | {
|
|---|
| 2676 | *nextPtr = s;
|
|---|
| 2677 | return ERROR_EXPAT_NONE;
|
|---|
| 2678 | }
|
|---|
| 2679 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 2680 | case XML_TOK_ENTITY_REF:
|
|---|
| 2681 | {
|
|---|
| 2682 | const XML_Char *name;
|
|---|
| 2683 | ENTITY *entity;
|
|---|
| 2684 | unsigned long ulOfs; // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 2685 | XML_Char ch = XmlPredefinedEntityName(enc,
|
|---|
| 2686 | s + enc->minBytesPerChar,
|
|---|
| 2687 | next - enc->minBytesPerChar);
|
|---|
| 2688 |
|
|---|
| 2689 | if (ch)
|
|---|
| 2690 | {
|
|---|
| 2691 | if (characterDataHandler)
|
|---|
| 2692 | characterDataHandler(handlerArg, &ch, 1);
|
|---|
| 2693 | else if (defaultHandler)
|
|---|
| 2694 | reportDefault(parser, enc, s, next);
|
|---|
| 2695 | break;
|
|---|
| 2696 | }
|
|---|
| 2697 | name = poolStoreString(&dtd.pool,
|
|---|
| 2698 | enc,
|
|---|
| 2699 | s + enc->minBytesPerChar,
|
|---|
| 2700 | next - enc->minBytesPerChar,
|
|---|
| 2701 | &ulOfs);
|
|---|
| 2702 | if (!name)
|
|---|
| 2703 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2704 | entity = (ENTITY*)lookup(&dtd.generalEntities,
|
|---|
| 2705 | name + ulOfs, // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 2706 | 0);
|
|---|
| 2707 | poolDiscard(&dtd.pool);
|
|---|
| 2708 | if (!entity)
|
|---|
| 2709 | {
|
|---|
| 2710 | if (dtd.complete || dtd.standalone)
|
|---|
| 2711 | return ERROR_EXPAT_UNDEFINED_ENTITY;
|
|---|
| 2712 | if (defaultHandler)
|
|---|
| 2713 | reportDefault(parser, enc, s, next);
|
|---|
| 2714 | break;
|
|---|
| 2715 | }
|
|---|
| 2716 | if (entity->open)
|
|---|
| 2717 | return ERROR_EXPAT_RECURSIVE_ENTITY_REF;
|
|---|
| 2718 | if (entity->notation)
|
|---|
| 2719 | return ERROR_EXPAT_BINARY_ENTITY_REF;
|
|---|
| 2720 | if (entity)
|
|---|
| 2721 | {
|
|---|
| 2722 | if (entity->textPtr)
|
|---|
| 2723 | {
|
|---|
| 2724 | XMLERROR result;
|
|---|
| 2725 | OPEN_INTERNAL_ENTITY openEntity;
|
|---|
| 2726 |
|
|---|
| 2727 | if (defaultHandler && !defaultExpandInternalEntities)
|
|---|
| 2728 | {
|
|---|
| 2729 | reportDefault(parser, enc, s, next);
|
|---|
| 2730 | break;
|
|---|
| 2731 | }
|
|---|
| 2732 | entity->open = 1;
|
|---|
| 2733 | openEntity.next = openInternalEntities;
|
|---|
| 2734 | openInternalEntities = &openEntity;
|
|---|
| 2735 | openEntity.entity = entity;
|
|---|
| 2736 | openEntity.internalEventPtr = 0;
|
|---|
| 2737 | openEntity.internalEventEndPtr = 0;
|
|---|
| 2738 | result = doContent(parser,
|
|---|
| 2739 | tagLevel,
|
|---|
| 2740 | internalEncoding,
|
|---|
| 2741 | (char *)entity->textPtr,
|
|---|
| 2742 | (char *)(entity->textPtr + entity->textLen),
|
|---|
| 2743 | 0);
|
|---|
| 2744 | entity->open = 0;
|
|---|
| 2745 | openInternalEntities = openEntity.next;
|
|---|
| 2746 | if (result)
|
|---|
| 2747 | return result;
|
|---|
| 2748 | }
|
|---|
| 2749 | else if (externalEntityRefHandler)
|
|---|
| 2750 | {
|
|---|
| 2751 | const XML_Char *context;
|
|---|
| 2752 |
|
|---|
| 2753 | entity->open = 1;
|
|---|
| 2754 | context = getContext(parser);
|
|---|
| 2755 | entity->open = 0;
|
|---|
| 2756 | if (!context)
|
|---|
| 2757 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2758 | if (!externalEntityRefHandler(handlerArg, // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 2759 | externalEntityRefHandlerArg,
|
|---|
| 2760 | context,
|
|---|
| 2761 | entity->base,
|
|---|
| 2762 | entity->systemId,
|
|---|
| 2763 | entity->publicId))
|
|---|
| 2764 | return ERROR_EXPAT_EXTERNAL_ENTITY_HANDLING;
|
|---|
| 2765 | poolDiscard(&tempPool);
|
|---|
| 2766 | }
|
|---|
| 2767 | else if (defaultHandler)
|
|---|
| 2768 | reportDefault(parser, enc, s, next);
|
|---|
| 2769 | }
|
|---|
| 2770 | break;
|
|---|
| 2771 | }
|
|---|
| 2772 | case XML_TOK_START_TAG_WITH_ATTS:
|
|---|
| 2773 | if (!startElementHandler)
|
|---|
| 2774 | {
|
|---|
| 2775 | XMLERROR result = storeAtts(parser, enc, s, 0, 0);
|
|---|
| 2776 |
|
|---|
| 2777 | if (result)
|
|---|
| 2778 | return result;
|
|---|
| 2779 | }
|
|---|
| 2780 | /* fall through */
|
|---|
| 2781 | case XML_TOK_START_TAG_NO_ATTS:
|
|---|
| 2782 | {
|
|---|
| 2783 | TAG *tag;
|
|---|
| 2784 |
|
|---|
| 2785 | if (freeTagList)
|
|---|
| 2786 | {
|
|---|
| 2787 | tag = freeTagList;
|
|---|
| 2788 | freeTagList = freeTagList->parent;
|
|---|
| 2789 | }
|
|---|
| 2790 | else
|
|---|
| 2791 | {
|
|---|
| 2792 | tag = (TAG*)MALLOC(sizeof(TAG));
|
|---|
| 2793 | if (!tag)
|
|---|
| 2794 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2795 | tag->buf = (char*)MALLOC(INIT_TAG_BUF_SIZE);
|
|---|
| 2796 | if (!tag->buf)
|
|---|
| 2797 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2798 | tag->bufEnd = tag->buf + INIT_TAG_BUF_SIZE;
|
|---|
| 2799 | }
|
|---|
| 2800 | tag->bindings = 0;
|
|---|
| 2801 | tag->parent = tagStack;
|
|---|
| 2802 | tagStack = tag;
|
|---|
| 2803 | tag->name.localPart = 0;
|
|---|
| 2804 | tag->rawName = s + enc->minBytesPerChar;
|
|---|
| 2805 | tag->rawNameLength = XmlNameLength(enc, tag->rawName);
|
|---|
| 2806 | if (nextPtr)
|
|---|
| 2807 | {
|
|---|
| 2808 | /* Need to guarantee that:
|
|---|
| 2809 | * tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)) <= tag->bufEnd - sizeof(XML_Char) */
|
|---|
| 2810 | if (tag->rawNameLength + (int)(sizeof(XML_Char) - 1) + (int)sizeof(XML_Char) > tag->bufEnd - tag->buf)
|
|---|
| 2811 | {
|
|---|
| 2812 | int bufSize = tag->rawNameLength * 4;
|
|---|
| 2813 |
|
|---|
| 2814 | bufSize = ROUND_UP(bufSize, sizeof(XML_Char));
|
|---|
| 2815 | tag->buf = (char*)REALLOC(tag->buf, bufSize);
|
|---|
| 2816 | if (!tag->buf)
|
|---|
| 2817 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2818 | tag->bufEnd = tag->buf + bufSize;
|
|---|
| 2819 | }
|
|---|
| 2820 | memcpy(tag->buf, tag->rawName, tag->rawNameLength);
|
|---|
| 2821 | tag->rawName = tag->buf;
|
|---|
| 2822 | }
|
|---|
| 2823 | ++tagLevel;
|
|---|
| 2824 | if (startElementHandler)
|
|---|
| 2825 | {
|
|---|
| 2826 | XMLERROR result;
|
|---|
| 2827 | XML_Char *toPtr;
|
|---|
| 2828 |
|
|---|
| 2829 | for (;;)
|
|---|
| 2830 | {
|
|---|
| 2831 | const char *rawNameEnd = tag->rawName + tag->rawNameLength;
|
|---|
| 2832 | const char *fromPtr = tag->rawName;
|
|---|
| 2833 | int bufSize;
|
|---|
| 2834 |
|
|---|
| 2835 | if (nextPtr)
|
|---|
| 2836 | toPtr = (XML_Char *) (tag->buf + ROUND_UP(tag->rawNameLength, sizeof(XML_Char)));
|
|---|
| 2837 | else
|
|---|
| 2838 | toPtr = (XML_Char *) tag->buf;
|
|---|
| 2839 | tag->name.str = toPtr;
|
|---|
| 2840 | XmlConvert(enc,
|
|---|
| 2841 | &fromPtr, rawNameEnd,
|
|---|
| 2842 | (ICHAR **) & toPtr, (ICHAR *) tag->bufEnd - 1);
|
|---|
| 2843 | if (fromPtr == rawNameEnd)
|
|---|
| 2844 | break;
|
|---|
| 2845 | bufSize = (tag->bufEnd - tag->buf) << 1;
|
|---|
| 2846 | tag->buf = (char*)REALLOC(tag->buf, bufSize);
|
|---|
| 2847 | if (!tag->buf)
|
|---|
| 2848 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2849 | tag->bufEnd = tag->buf + bufSize;
|
|---|
| 2850 | if (nextPtr)
|
|---|
| 2851 | tag->rawName = tag->buf;
|
|---|
| 2852 | }
|
|---|
| 2853 | *toPtr = XML_T('\0');
|
|---|
| 2854 | result = storeAtts(parser, enc, s, &(tag->name), &(tag->bindings));
|
|---|
| 2855 | if (result)
|
|---|
| 2856 | return result;
|
|---|
| 2857 | startElementHandler(handlerArg, tag->name.str, (const XML_Char **)atts);
|
|---|
| 2858 | poolClear(&tempPool);
|
|---|
| 2859 | }
|
|---|
| 2860 | else
|
|---|
| 2861 | {
|
|---|
| 2862 | tag->name.str = 0;
|
|---|
| 2863 | if (defaultHandler)
|
|---|
| 2864 | reportDefault(parser, enc, s, next);
|
|---|
| 2865 | }
|
|---|
| 2866 | break;
|
|---|
| 2867 | }
|
|---|
| 2868 | case XML_TOK_EMPTY_ELEMENT_WITH_ATTS:
|
|---|
| 2869 | if (!startElementHandler)
|
|---|
| 2870 | {
|
|---|
| 2871 | XMLERROR result = storeAtts(parser, enc, s, 0, 0);
|
|---|
| 2872 |
|
|---|
| 2873 | if (result)
|
|---|
| 2874 | return result;
|
|---|
| 2875 | }
|
|---|
| 2876 | /* fall through */
|
|---|
| 2877 | case XML_TOK_EMPTY_ELEMENT_NO_ATTS:
|
|---|
| 2878 | if (startElementHandler || endElementHandler)
|
|---|
| 2879 | {
|
|---|
| 2880 | const char *rawName = s + enc->minBytesPerChar;
|
|---|
| 2881 | XMLERROR result;
|
|---|
| 2882 | BINDING *bindings = 0;
|
|---|
| 2883 | TAG_NAME name;
|
|---|
| 2884 |
|
|---|
| 2885 | name.str = poolStoreString(&tempPool,
|
|---|
| 2886 | enc,
|
|---|
| 2887 | rawName,
|
|---|
| 2888 | rawName + XmlNameLength(enc, rawName),
|
|---|
| 2889 | NULL);
|
|---|
| 2890 | if (!name.str)
|
|---|
| 2891 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 2892 | poolFinish(&tempPool);
|
|---|
| 2893 | result = storeAtts(parser, enc, s, &name, &bindings);
|
|---|
| 2894 | if (result)
|
|---|
| 2895 | return result;
|
|---|
| 2896 | poolFinish(&tempPool);
|
|---|
| 2897 | if (startElementHandler)
|
|---|
| 2898 | startElementHandler(handlerArg, name.str, (const XML_Char **)atts);
|
|---|
| 2899 | if (endElementHandler)
|
|---|
| 2900 | {
|
|---|
| 2901 | if (startElementHandler)
|
|---|
| 2902 | *eventPP = *eventEndPP;
|
|---|
| 2903 | endElementHandler(handlerArg, name.str);
|
|---|
| 2904 | }
|
|---|
| 2905 | poolClear(&tempPool);
|
|---|
| 2906 | while (bindings)
|
|---|
| 2907 | {
|
|---|
| 2908 | BINDING *b = bindings;
|
|---|
| 2909 |
|
|---|
| 2910 | if (endNamespaceDeclHandler)
|
|---|
| 2911 | endNamespaceDeclHandler(handlerArg, b->prefix->name);
|
|---|
| 2912 | bindings = bindings->nextTagBinding;
|
|---|
| 2913 | b->nextTagBinding = freeBindingList;
|
|---|
| 2914 | freeBindingList = b;
|
|---|
| 2915 | b->prefix->binding = b->prevPrefixBinding;
|
|---|
| 2916 | }
|
|---|
| 2917 | }
|
|---|
| 2918 | else if (defaultHandler)
|
|---|
| 2919 | reportDefault(parser, enc, s, next);
|
|---|
| 2920 | if (tagLevel == 0)
|
|---|
| 2921 | return epilogProcessor(parser, next, end, nextPtr);
|
|---|
| 2922 | break;
|
|---|
| 2923 | case XML_TOK_END_TAG:
|
|---|
| 2924 | if (tagLevel == startTagLevel)
|
|---|
| 2925 | return ERROR_EXPAT_ASYNC_ENTITY;
|
|---|
| 2926 | else
|
|---|
| 2927 | {
|
|---|
| 2928 | int len;
|
|---|
| 2929 | const char *rawName;
|
|---|
| 2930 | TAG *tag = tagStack;
|
|---|
| 2931 |
|
|---|
| 2932 | tagStack = tag->parent;
|
|---|
| 2933 | tag->parent = freeTagList;
|
|---|
| 2934 | freeTagList = tag;
|
|---|
| 2935 | rawName = s + enc->minBytesPerChar * 2;
|
|---|
| 2936 | len = XmlNameLength(enc, rawName);
|
|---|
| 2937 | if (len != tag->rawNameLength
|
|---|
| 2938 | || memcmp(tag->rawName, rawName, len) != 0)
|
|---|
| 2939 | {
|
|---|
| 2940 | *eventPP = rawName;
|
|---|
| 2941 | return ERROR_EXPAT_TAG_MISMATCH;
|
|---|
| 2942 | }
|
|---|
| 2943 | --tagLevel;
|
|---|
| 2944 | if (endElementHandler && tag->name.str)
|
|---|
| 2945 | {
|
|---|
| 2946 | if (tag->name.localPart)
|
|---|
| 2947 | {
|
|---|
| 2948 | XML_Char *to = (XML_Char *) tag->name.str + tag->name.uriLen;
|
|---|
| 2949 | const XML_Char *from = tag->name.localPart;
|
|---|
| 2950 |
|
|---|
| 2951 | while ((*to++ = *from++) != 0)
|
|---|
| 2952 | ;
|
|---|
| 2953 | }
|
|---|
| 2954 | endElementHandler(handlerArg, tag->name.str);
|
|---|
| 2955 | }
|
|---|
| 2956 | else if (defaultHandler)
|
|---|
| 2957 | reportDefault(parser, enc, s, next);
|
|---|
| 2958 | while (tag->bindings)
|
|---|
| 2959 | {
|
|---|
| 2960 | BINDING *b = tag->bindings;
|
|---|
| 2961 |
|
|---|
| 2962 | if (endNamespaceDeclHandler)
|
|---|
| 2963 | endNamespaceDeclHandler(handlerArg, b->prefix->name);
|
|---|
| 2964 | tag->bindings = tag->bindings->nextTagBinding;
|
|---|
| 2965 | b->nextTagBinding = freeBindingList;
|
|---|
| 2966 | freeBindingList = b;
|
|---|
| 2967 | b->prefix->binding = b->prevPrefixBinding;
|
|---|
| 2968 | }
|
|---|
| 2969 | if (tagLevel == 0)
|
|---|
| 2970 | return epilogProcessor(parser, next, end, nextPtr);
|
|---|
| 2971 | }
|
|---|
| 2972 | break;
|
|---|
| 2973 | case XML_TOK_CHAR_REF:
|
|---|
| 2974 | {
|
|---|
| 2975 | int n = XmlCharRefNumber(enc, s);
|
|---|
| 2976 |
|
|---|
| 2977 | if (n < 0)
|
|---|
| 2978 | return ERROR_EXPAT_BAD_CHAR_REF;
|
|---|
| 2979 | if (characterDataHandler)
|
|---|
| 2980 | {
|
|---|
| 2981 | XML_Char buf[XML_ENCODE_MAX];
|
|---|
| 2982 |
|
|---|
| 2983 | characterDataHandler(handlerArg, buf, XmlEncode(n, (ICHAR *) buf));
|
|---|
| 2984 | }
|
|---|
| 2985 | else if (defaultHandler)
|
|---|
| 2986 | reportDefault(parser, enc, s, next);
|
|---|
| 2987 | }
|
|---|
| 2988 | break;
|
|---|
| 2989 | case XML_TOK_XML_DECL:
|
|---|
| 2990 | return ERROR_EXPAT_MISPLACED_XML_PI;
|
|---|
| 2991 | case XML_TOK_DATA_NEWLINE:
|
|---|
| 2992 | if (characterDataHandler)
|
|---|
| 2993 | {
|
|---|
| 2994 | XML_Char c = 0xA;
|
|---|
| 2995 |
|
|---|
| 2996 | characterDataHandler(handlerArg, &c, 1);
|
|---|
| 2997 | }
|
|---|
| 2998 | else if (defaultHandler)
|
|---|
| 2999 | reportDefault(parser, enc, s, next);
|
|---|
| 3000 | break;
|
|---|
| 3001 | case XML_TOK_CDATA_SECT_OPEN:
|
|---|
| 3002 | {
|
|---|
| 3003 | XMLERROR result;
|
|---|
| 3004 |
|
|---|
| 3005 | if (startCdataSectionHandler)
|
|---|
| 3006 | startCdataSectionHandler(handlerArg);
|
|---|
| 3007 | #if 0
|
|---|
| 3008 | /* Suppose you doing a transformation on a document that involves
|
|---|
| 3009 | * changing only the character data. You set up a defaultHandler
|
|---|
| 3010 | * and a characterDataHandler. The defaultHandler simply copies
|
|---|
| 3011 | * characters through. The characterDataHandler does the transformation
|
|---|
| 3012 | * and writes the characters out escaping them as necessary. This case
|
|---|
| 3013 | * will fail to work if we leave out the following two lines (because &
|
|---|
| 3014 | * and < inside CDATA sections will be incorrectly escaped).
|
|---|
| 3015 | *
|
|---|
| 3016 | * However, now we have a start/endCdataSectionHandler, so it seems
|
|---|
| 3017 | * easier to let the user deal with this. */
|
|---|
| 3018 |
|
|---|
| 3019 | else if (characterDataHandler)
|
|---|
| 3020 | characterDataHandler(handlerArg, dataBuf, 0);
|
|---|
| 3021 | #endif
|
|---|
| 3022 | else if (defaultHandler)
|
|---|
| 3023 | reportDefault(parser, enc, s, next);
|
|---|
| 3024 | result = doCdataSection(parser, enc, &next, end, nextPtr);
|
|---|
| 3025 | if (!next)
|
|---|
| 3026 | {
|
|---|
| 3027 | processor = cdataSectionProcessor;
|
|---|
| 3028 | return result;
|
|---|
| 3029 | }
|
|---|
| 3030 | }
|
|---|
| 3031 | break;
|
|---|
| 3032 | case XML_TOK_TRAILING_RSQB:
|
|---|
| 3033 | if (nextPtr)
|
|---|
| 3034 | {
|
|---|
| 3035 | *nextPtr = s;
|
|---|
| 3036 | return ERROR_EXPAT_NONE;
|
|---|
| 3037 | }
|
|---|
| 3038 | if (characterDataHandler)
|
|---|
| 3039 | {
|
|---|
| 3040 | if (MUST_CONVERT(enc, s))
|
|---|
| 3041 | {
|
|---|
| 3042 | ICHAR *dataPtr = (ICHAR *) dataBuf;
|
|---|
| 3043 |
|
|---|
| 3044 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *) dataBufEnd);
|
|---|
| 3045 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *) dataBuf);
|
|---|
| 3046 | }
|
|---|
| 3047 | else
|
|---|
| 3048 | characterDataHandler(handlerArg,
|
|---|
| 3049 | (XML_Char *) s,
|
|---|
| 3050 | (XML_Char *) end - (XML_Char *) s);
|
|---|
| 3051 | }
|
|---|
| 3052 | else if (defaultHandler)
|
|---|
| 3053 | reportDefault(parser, enc, s, end);
|
|---|
| 3054 | if (startTagLevel == 0)
|
|---|
| 3055 | {
|
|---|
| 3056 | *eventPP = end;
|
|---|
| 3057 | return ERROR_EXPAT_NO_ELEMENTS;
|
|---|
| 3058 | }
|
|---|
| 3059 | if (tagLevel != startTagLevel)
|
|---|
| 3060 | {
|
|---|
| 3061 | *eventPP = end;
|
|---|
| 3062 | return ERROR_EXPAT_ASYNC_ENTITY;
|
|---|
| 3063 | }
|
|---|
| 3064 | return ERROR_EXPAT_NONE;
|
|---|
| 3065 | case XML_TOK_DATA_CHARS:
|
|---|
| 3066 | if (characterDataHandler)
|
|---|
| 3067 | {
|
|---|
| 3068 | if (MUST_CONVERT(enc, s))
|
|---|
| 3069 | {
|
|---|
| 3070 | for (;;)
|
|---|
| 3071 | {
|
|---|
| 3072 | ICHAR *dataPtr = (ICHAR *) dataBuf;
|
|---|
| 3073 |
|
|---|
| 3074 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *) dataBufEnd);
|
|---|
| 3075 | *eventEndPP = s;
|
|---|
| 3076 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *) dataBuf);
|
|---|
| 3077 | if (s == next)
|
|---|
| 3078 | break;
|
|---|
| 3079 | *eventPP = s;
|
|---|
| 3080 | }
|
|---|
| 3081 | }
|
|---|
| 3082 | else
|
|---|
| 3083 | characterDataHandler(handlerArg,
|
|---|
| 3084 | (XML_Char *) s,
|
|---|
| 3085 | (XML_Char *) next - (XML_Char *) s);
|
|---|
| 3086 | }
|
|---|
| 3087 | else if (defaultHandler)
|
|---|
| 3088 | reportDefault(parser, enc, s, next);
|
|---|
| 3089 | break;
|
|---|
| 3090 | case XML_TOK_PI:
|
|---|
| 3091 | if (!reportProcessingInstruction(parser, enc, s, next))
|
|---|
| 3092 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3093 | break;
|
|---|
| 3094 | case XML_TOK_COMMENT:
|
|---|
| 3095 | if (!reportComment(parser, enc, s, next))
|
|---|
| 3096 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3097 | break;
|
|---|
| 3098 | default:
|
|---|
| 3099 | if (defaultHandler)
|
|---|
| 3100 | reportDefault(parser, enc, s, next);
|
|---|
| 3101 | break;
|
|---|
| 3102 | }
|
|---|
| 3103 | *eventPP = s = next;
|
|---|
| 3104 | }
|
|---|
| 3105 | /* not reached */
|
|---|
| 3106 | }
|
|---|
| 3107 |
|
|---|
| 3108 | /* If tagNamePtr is non-null, build a real list of attributes,
|
|---|
| 3109 | * otherwise just check the attributes for well-formedness. */
|
|---|
| 3110 |
|
|---|
| 3111 | STATIC XMLERROR storeAtts(XML_Parser parser,
|
|---|
| 3112 | const ENCODING * enc,
|
|---|
| 3113 | const char *attStr,
|
|---|
| 3114 | TAG_NAME * tagNamePtr,
|
|---|
| 3115 | BINDING ** bindingsPtr)
|
|---|
| 3116 | {
|
|---|
| 3117 | ELEMENT_TYPE *elementType = 0;
|
|---|
| 3118 | int nDefaultAtts = 0;
|
|---|
| 3119 | const XML_Char **appAtts; /* the attribute list to pass to the application */
|
|---|
| 3120 | int attIndex = 0;
|
|---|
| 3121 | int i;
|
|---|
| 3122 | int n;
|
|---|
| 3123 | int nPrefixes = 0;
|
|---|
| 3124 | BINDING *binding;
|
|---|
| 3125 | const XML_Char *localPart;
|
|---|
| 3126 |
|
|---|
| 3127 | /* lookup the element type name */
|
|---|
| 3128 | if (tagNamePtr)
|
|---|
| 3129 | {
|
|---|
| 3130 | elementType = (ELEMENT_TYPE *) lookup(&dtd.elementTypes, tagNamePtr->str, 0);
|
|---|
| 3131 | if (!elementType)
|
|---|
| 3132 | {
|
|---|
| 3133 | tagNamePtr->str = poolCopyString(&dtd.pool, tagNamePtr->str);
|
|---|
| 3134 | if (!tagNamePtr->str)
|
|---|
| 3135 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3136 | elementType = (ELEMENT_TYPE *) lookup(&dtd.elementTypes, tagNamePtr->str, sizeof(ELEMENT_TYPE));
|
|---|
| 3137 | if (!elementType)
|
|---|
| 3138 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3139 | if (ns && !setElementTypePrefix(parser, elementType))
|
|---|
| 3140 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3141 | }
|
|---|
| 3142 | nDefaultAtts = elementType->nDefaultAtts;
|
|---|
| 3143 | }
|
|---|
| 3144 | /* get the attributes from the tokenizer */
|
|---|
| 3145 | n = XmlGetAttributes(enc, attStr, attsSize, atts);
|
|---|
| 3146 | if (n + nDefaultAtts > attsSize)
|
|---|
| 3147 | {
|
|---|
| 3148 | int oldAttsSize = attsSize;
|
|---|
| 3149 |
|
|---|
| 3150 | attsSize = n + nDefaultAtts + INIT_ATTS_SIZE;
|
|---|
| 3151 | atts = (ATTRIBUTE*)REALLOC((void *)atts, attsSize * sizeof(ATTRIBUTE));
|
|---|
| 3152 | if (!atts)
|
|---|
| 3153 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3154 | if (n > oldAttsSize)
|
|---|
| 3155 | XmlGetAttributes(enc, attStr, n, atts);
|
|---|
| 3156 | }
|
|---|
| 3157 | appAtts = (const XML_Char **)atts;
|
|---|
| 3158 | for (i = 0; i < n; i++)
|
|---|
| 3159 | {
|
|---|
| 3160 | /* add the name and value to the attribute list */
|
|---|
| 3161 | ATTRIBUTE_ID *attId = getAttributeId(parser, enc, atts[i].name,
|
|---|
| 3162 | atts[i].name
|
|---|
| 3163 | + XmlNameLength(enc, atts[i].name));
|
|---|
| 3164 |
|
|---|
| 3165 | if (!attId)
|
|---|
| 3166 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3167 | /* detect duplicate attributes */
|
|---|
| 3168 | if ((attId->name)[-1])
|
|---|
| 3169 | {
|
|---|
| 3170 | if (enc == encoding)
|
|---|
| 3171 | eventPtr = atts[i].name;
|
|---|
| 3172 | return ERROR_EXPAT_DUPLICATE_ATTRIBUTE;
|
|---|
| 3173 | }
|
|---|
| 3174 | (attId->name)[-1] = 1;
|
|---|
| 3175 | appAtts[attIndex++] = attId->name;
|
|---|
| 3176 | if (!atts[i].normalized)
|
|---|
| 3177 | {
|
|---|
| 3178 | XMLERROR result;
|
|---|
| 3179 | int isCdata = 1;
|
|---|
| 3180 |
|
|---|
| 3181 | /* figure out whether declared as other than CDATA */
|
|---|
| 3182 | if (attId->maybeTokenized)
|
|---|
| 3183 | {
|
|---|
| 3184 | int j;
|
|---|
| 3185 |
|
|---|
| 3186 | for (j = 0; j < nDefaultAtts; j++)
|
|---|
| 3187 | {
|
|---|
| 3188 | if (attId == elementType->defaultAtts[j].id)
|
|---|
| 3189 | {
|
|---|
| 3190 | isCdata = elementType->defaultAtts[j].isCdata;
|
|---|
| 3191 | break;
|
|---|
| 3192 | }
|
|---|
| 3193 | }
|
|---|
| 3194 | }
|
|---|
| 3195 |
|
|---|
| 3196 | /* normalize the attribute value */
|
|---|
| 3197 | result = storeAttributeValue(parser, enc, isCdata,
|
|---|
| 3198 | atts[i].valuePtr, atts[i].valueEnd,
|
|---|
| 3199 | &tempPool);
|
|---|
| 3200 | if (result)
|
|---|
| 3201 | return result;
|
|---|
| 3202 | if (tagNamePtr)
|
|---|
| 3203 | {
|
|---|
| 3204 | appAtts[attIndex] = poolStart(&tempPool);
|
|---|
| 3205 | poolFinish(&tempPool);
|
|---|
| 3206 | }
|
|---|
| 3207 | else
|
|---|
| 3208 | poolDiscard(&tempPool);
|
|---|
| 3209 | }
|
|---|
| 3210 | else if (tagNamePtr)
|
|---|
| 3211 | {
|
|---|
| 3212 | /* the value did not need normalizing */
|
|---|
| 3213 | appAtts[attIndex] = poolStoreString(&tempPool,
|
|---|
| 3214 | enc,
|
|---|
| 3215 | atts[i].valuePtr,
|
|---|
| 3216 | atts[i].valueEnd,
|
|---|
| 3217 | NULL);
|
|---|
| 3218 | if (appAtts[attIndex] == 0)
|
|---|
| 3219 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3220 | poolFinish(&tempPool);
|
|---|
| 3221 | }
|
|---|
| 3222 | /* handle prefixed attribute names */
|
|---|
| 3223 | if (attId->prefix && tagNamePtr)
|
|---|
| 3224 | {
|
|---|
| 3225 | if (attId->xmlns)
|
|---|
| 3226 | {
|
|---|
| 3227 | /* deal with namespace declarations here */
|
|---|
| 3228 | if (!addBinding(parser, attId->prefix, attId, appAtts[attIndex], bindingsPtr))
|
|---|
| 3229 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3230 | --attIndex;
|
|---|
| 3231 | }
|
|---|
| 3232 | else
|
|---|
| 3233 | {
|
|---|
| 3234 | /* deal with other prefixed names later */
|
|---|
| 3235 | attIndex++;
|
|---|
| 3236 | nPrefixes++;
|
|---|
| 3237 | (attId->name)[-1] = 2;
|
|---|
| 3238 | }
|
|---|
| 3239 | }
|
|---|
| 3240 | else
|
|---|
| 3241 | attIndex++;
|
|---|
| 3242 | }
|
|---|
| 3243 | if (tagNamePtr)
|
|---|
| 3244 | {
|
|---|
| 3245 | int j;
|
|---|
| 3246 |
|
|---|
| 3247 | nSpecifiedAtts = attIndex;
|
|---|
| 3248 | if (elementType->idAtt && (elementType->idAtt->name)[-1])
|
|---|
| 3249 | {
|
|---|
| 3250 | for (i = 0; i < attIndex; i += 2)
|
|---|
| 3251 | if (appAtts[i] == elementType->idAtt->name)
|
|---|
| 3252 | {
|
|---|
| 3253 | idAttIndex = i;
|
|---|
| 3254 | break;
|
|---|
| 3255 | }
|
|---|
| 3256 | }
|
|---|
| 3257 | else
|
|---|
| 3258 | idAttIndex = -1;
|
|---|
| 3259 | /* do attribute defaulting */
|
|---|
| 3260 | for (j = 0; j < nDefaultAtts; j++)
|
|---|
| 3261 | {
|
|---|
| 3262 | const DEFAULT_ATTRIBUTE *da = elementType->defaultAtts + j;
|
|---|
| 3263 |
|
|---|
| 3264 | if (!(da->id->name)[-1] && da->value)
|
|---|
| 3265 | {
|
|---|
| 3266 | if (da->id->prefix)
|
|---|
| 3267 | {
|
|---|
| 3268 | if (da->id->xmlns)
|
|---|
| 3269 | {
|
|---|
| 3270 | if (!addBinding(parser, da->id->prefix, da->id, da->value, bindingsPtr))
|
|---|
| 3271 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3272 | }
|
|---|
| 3273 | else
|
|---|
| 3274 | {
|
|---|
| 3275 | (da->id->name)[-1] = 2;
|
|---|
| 3276 | nPrefixes++;
|
|---|
| 3277 | appAtts[attIndex++] = da->id->name;
|
|---|
| 3278 | appAtts[attIndex++] = da->value;
|
|---|
| 3279 | }
|
|---|
| 3280 | }
|
|---|
| 3281 | else
|
|---|
| 3282 | {
|
|---|
| 3283 | (da->id->name)[-1] = 1;
|
|---|
| 3284 | appAtts[attIndex++] = da->id->name;
|
|---|
| 3285 | appAtts[attIndex++] = da->value;
|
|---|
| 3286 | }
|
|---|
| 3287 | }
|
|---|
| 3288 | }
|
|---|
| 3289 | appAtts[attIndex] = 0;
|
|---|
| 3290 | }
|
|---|
| 3291 | i = 0;
|
|---|
| 3292 | if (nPrefixes)
|
|---|
| 3293 | {
|
|---|
| 3294 | /* expand prefixed attribute names */
|
|---|
| 3295 | for (; i < attIndex; i += 2)
|
|---|
| 3296 | {
|
|---|
| 3297 | if (appAtts[i][-1] == 2)
|
|---|
| 3298 | {
|
|---|
| 3299 | ATTRIBUTE_ID *id;
|
|---|
| 3300 |
|
|---|
| 3301 | ((XML_Char *) (appAtts[i]))[-1] = 0;
|
|---|
| 3302 | id = (ATTRIBUTE_ID *) lookup(&dtd.attributeIds, appAtts[i], 0);
|
|---|
| 3303 | if (id->prefix->binding)
|
|---|
| 3304 | {
|
|---|
| 3305 | int j;
|
|---|
| 3306 | const BINDING *b = id->prefix->binding;
|
|---|
| 3307 | const XML_Char *s = appAtts[i];
|
|---|
| 3308 |
|
|---|
| 3309 | for (j = 0; j < b->uriLen; j++)
|
|---|
| 3310 | {
|
|---|
| 3311 | if (!poolAppendChar(&tempPool, b->uri[j]))
|
|---|
| 3312 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3313 | }
|
|---|
| 3314 | while (*s++ != ':')
|
|---|
| 3315 | ;
|
|---|
| 3316 | do
|
|---|
| 3317 | {
|
|---|
| 3318 | if (!poolAppendChar(&tempPool, *s))
|
|---|
| 3319 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3320 | }
|
|---|
| 3321 | while (*s++);
|
|---|
| 3322 | if (ns_triplets)
|
|---|
| 3323 | {
|
|---|
| 3324 | tempPool.ptr[-1] = namespaceSeparator;
|
|---|
| 3325 | s = b->prefix->name;
|
|---|
| 3326 | do
|
|---|
| 3327 | {
|
|---|
| 3328 | if (!poolAppendChar(&tempPool, *s))
|
|---|
| 3329 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3330 | }
|
|---|
| 3331 | while (*s++);
|
|---|
| 3332 | }
|
|---|
| 3333 |
|
|---|
| 3334 | appAtts[i] = poolStart(&tempPool);
|
|---|
| 3335 | poolFinish(&tempPool);
|
|---|
| 3336 | }
|
|---|
| 3337 | if (!--nPrefixes)
|
|---|
| 3338 | break;
|
|---|
| 3339 | }
|
|---|
| 3340 | else
|
|---|
| 3341 | ((XML_Char *) (appAtts[i]))[-1] = 0;
|
|---|
| 3342 | }
|
|---|
| 3343 | }
|
|---|
| 3344 | /* clear the flags that say whether attributes were specified */
|
|---|
| 3345 | for (; i < attIndex; i += 2)
|
|---|
| 3346 | ((XML_Char *) (appAtts[i]))[-1] = 0;
|
|---|
| 3347 | if (!tagNamePtr)
|
|---|
| 3348 | return ERROR_EXPAT_NONE;
|
|---|
| 3349 | for (binding = *bindingsPtr; binding; binding = binding->nextTagBinding)
|
|---|
| 3350 | binding->attId->name[-1] = 0;
|
|---|
| 3351 | /* expand the element type name */
|
|---|
| 3352 | if (elementType->prefix)
|
|---|
| 3353 | {
|
|---|
| 3354 | binding = elementType->prefix->binding;
|
|---|
| 3355 | if (!binding)
|
|---|
| 3356 | return ERROR_EXPAT_NONE;
|
|---|
| 3357 | localPart = tagNamePtr->str;
|
|---|
| 3358 | while (*localPart++ != XML_T(':'))
|
|---|
| 3359 | ;
|
|---|
| 3360 | }
|
|---|
| 3361 | else if (dtd.defaultPrefix.binding)
|
|---|
| 3362 | {
|
|---|
| 3363 | binding = dtd.defaultPrefix.binding;
|
|---|
| 3364 | localPart = tagNamePtr->str;
|
|---|
| 3365 | }
|
|---|
| 3366 | else
|
|---|
| 3367 | return ERROR_EXPAT_NONE;
|
|---|
| 3368 | tagNamePtr->localPart = localPart;
|
|---|
| 3369 | tagNamePtr->uriLen = binding->uriLen;
|
|---|
| 3370 | for (i = 0; localPart[i++];)
|
|---|
| 3371 | ;
|
|---|
| 3372 | n = i + binding->uriLen;
|
|---|
| 3373 | if (n > binding->uriAlloc)
|
|---|
| 3374 | {
|
|---|
| 3375 | TAG *p;
|
|---|
| 3376 | XML_Char *uri = (XML_Char*)MALLOC((n + EXPAND_SPARE) * sizeof(XML_Char));
|
|---|
| 3377 |
|
|---|
| 3378 | if (!uri)
|
|---|
| 3379 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3380 | binding->uriAlloc = n + EXPAND_SPARE;
|
|---|
| 3381 | memcpy(uri, binding->uri, binding->uriLen * sizeof(XML_Char));
|
|---|
| 3382 | for (p = tagStack; p; p = p->parent)
|
|---|
| 3383 | if (p->name.str == binding->uri)
|
|---|
| 3384 | p->name.str = uri;
|
|---|
| 3385 | FREE(binding->uri);
|
|---|
| 3386 | binding->uri = uri;
|
|---|
| 3387 | }
|
|---|
| 3388 | memcpy(binding->uri + binding->uriLen, localPart, i * sizeof(XML_Char));
|
|---|
| 3389 | tagNamePtr->str = binding->uri;
|
|---|
| 3390 | return ERROR_EXPAT_NONE;
|
|---|
| 3391 | }
|
|---|
| 3392 |
|
|---|
| 3393 | STATIC int addBinding(XML_Parser parser,
|
|---|
| 3394 | PREFIX * prefix,
|
|---|
| 3395 | const ATTRIBUTE_ID * attId,
|
|---|
| 3396 | const XML_Char * uri,
|
|---|
| 3397 | BINDING ** bindingsPtr)
|
|---|
| 3398 | {
|
|---|
| 3399 | BINDING *b;
|
|---|
| 3400 | int len;
|
|---|
| 3401 |
|
|---|
| 3402 | for (len = 0; uri[len]; len++)
|
|---|
| 3403 | ;
|
|---|
| 3404 | if (namespaceSeparator)
|
|---|
| 3405 | len++;
|
|---|
| 3406 | if (freeBindingList)
|
|---|
| 3407 | {
|
|---|
| 3408 | b = freeBindingList;
|
|---|
| 3409 | if (len > b->uriAlloc)
|
|---|
| 3410 | {
|
|---|
| 3411 | b->uri = (XML_Char*)REALLOC(b->uri, sizeof(XML_Char) * (len + EXPAND_SPARE));
|
|---|
| 3412 | if (!b->uri)
|
|---|
| 3413 | return 0;
|
|---|
| 3414 | b->uriAlloc = len + EXPAND_SPARE;
|
|---|
| 3415 | }
|
|---|
| 3416 | freeBindingList = b->nextTagBinding;
|
|---|
| 3417 | }
|
|---|
| 3418 | else
|
|---|
| 3419 | {
|
|---|
| 3420 | b = (BINDING*)MALLOC(sizeof(BINDING));
|
|---|
| 3421 | if (!b)
|
|---|
| 3422 | return 0;
|
|---|
| 3423 | b->uri = (XML_Char*)MALLOC(sizeof(XML_Char) * (len + EXPAND_SPARE));
|
|---|
| 3424 | if (!b->uri)
|
|---|
| 3425 | {
|
|---|
| 3426 | FREE(b);
|
|---|
| 3427 | return 0;
|
|---|
| 3428 | }
|
|---|
| 3429 | b->uriAlloc = len + EXPAND_SPARE;
|
|---|
| 3430 | }
|
|---|
| 3431 | b->uriLen = len;
|
|---|
| 3432 | memcpy(b->uri, uri, len * sizeof(XML_Char));
|
|---|
| 3433 | if (namespaceSeparator)
|
|---|
| 3434 | b->uri[len - 1] = namespaceSeparator;
|
|---|
| 3435 | b->prefix = prefix;
|
|---|
| 3436 | b->attId = attId;
|
|---|
| 3437 | b->prevPrefixBinding = prefix->binding;
|
|---|
| 3438 | if (*uri == XML_T('\0') && prefix == &dtd.defaultPrefix)
|
|---|
| 3439 | prefix->binding = 0;
|
|---|
| 3440 | else
|
|---|
| 3441 | prefix->binding = b;
|
|---|
| 3442 | b->nextTagBinding = *bindingsPtr;
|
|---|
| 3443 | *bindingsPtr = b;
|
|---|
| 3444 | if (startNamespaceDeclHandler)
|
|---|
| 3445 | startNamespaceDeclHandler(handlerArg, prefix->name,
|
|---|
| 3446 | prefix->binding ? uri : 0);
|
|---|
| 3447 | return 1;
|
|---|
| 3448 | }
|
|---|
| 3449 |
|
|---|
| 3450 | /* The idea here is to avoid using stack for each CDATA section when
|
|---|
| 3451 | * the whole file is parsed with one call. */
|
|---|
| 3452 |
|
|---|
| 3453 | STATIC XMLERROR cdataSectionProcessor(XML_Parser parser,
|
|---|
| 3454 | const char *start,
|
|---|
| 3455 | const char *end,
|
|---|
| 3456 | const char **endPtr)
|
|---|
| 3457 | {
|
|---|
| 3458 | XMLERROR result = doCdataSection(parser, encoding, &start, end, endPtr);
|
|---|
| 3459 |
|
|---|
| 3460 | if (start)
|
|---|
| 3461 | {
|
|---|
| 3462 | processor = contentProcessor;
|
|---|
| 3463 | return contentProcessor(parser, start, end, endPtr);
|
|---|
| 3464 | }
|
|---|
| 3465 | return result;
|
|---|
| 3466 | }
|
|---|
| 3467 |
|
|---|
| 3468 | /* startPtr gets set to non-null is the section is closed, and to null if
|
|---|
| 3469 | * the section is not yet closed. */
|
|---|
| 3470 |
|
|---|
| 3471 | STATIC XMLERROR doCdataSection(XML_Parser parser,
|
|---|
| 3472 | const ENCODING * enc,
|
|---|
| 3473 | const char **startPtr,
|
|---|
| 3474 | const char *end,
|
|---|
| 3475 | const char **nextPtr)
|
|---|
| 3476 | {
|
|---|
| 3477 | const char *s = *startPtr;
|
|---|
| 3478 | const char **eventPP;
|
|---|
| 3479 | const char **eventEndPP;
|
|---|
| 3480 |
|
|---|
| 3481 | if (enc == encoding)
|
|---|
| 3482 | {
|
|---|
| 3483 | eventPP = &eventPtr;
|
|---|
| 3484 | *eventPP = s;
|
|---|
| 3485 | eventEndPP = &eventEndPtr;
|
|---|
| 3486 | }
|
|---|
| 3487 | else
|
|---|
| 3488 | {
|
|---|
| 3489 | eventPP = &(openInternalEntities->internalEventPtr);
|
|---|
| 3490 | eventEndPP = &(openInternalEntities->internalEventEndPtr);
|
|---|
| 3491 | }
|
|---|
| 3492 | *eventPP = s;
|
|---|
| 3493 | *startPtr = 0;
|
|---|
| 3494 | for (;;)
|
|---|
| 3495 | {
|
|---|
| 3496 | const char *next;
|
|---|
| 3497 | int tok = XmlCdataSectionTok(enc, s, end, &next);
|
|---|
| 3498 |
|
|---|
| 3499 | *eventEndPP = next;
|
|---|
| 3500 | switch (tok)
|
|---|
| 3501 | {
|
|---|
| 3502 | case XML_TOK_CDATA_SECT_CLOSE:
|
|---|
| 3503 | if (endCdataSectionHandler)
|
|---|
| 3504 | endCdataSectionHandler(handlerArg);
|
|---|
| 3505 | #if 0
|
|---|
| 3506 | /* see comment under XML_TOK_CDATA_SECT_OPEN */
|
|---|
| 3507 | else if (characterDataHandler)
|
|---|
| 3508 | characterDataHandler(handlerArg, dataBuf, 0);
|
|---|
| 3509 | #endif
|
|---|
| 3510 | else if (defaultHandler)
|
|---|
| 3511 | reportDefault(parser, enc, s, next);
|
|---|
| 3512 | *startPtr = next;
|
|---|
| 3513 | return ERROR_EXPAT_NONE;
|
|---|
| 3514 | case XML_TOK_DATA_NEWLINE:
|
|---|
| 3515 | if (characterDataHandler)
|
|---|
| 3516 | {
|
|---|
| 3517 | XML_Char c = 0xA;
|
|---|
| 3518 |
|
|---|
| 3519 | characterDataHandler(handlerArg, &c, 1);
|
|---|
| 3520 | }
|
|---|
| 3521 | else if (defaultHandler)
|
|---|
| 3522 | reportDefault(parser, enc, s, next);
|
|---|
| 3523 | break;
|
|---|
| 3524 | case XML_TOK_DATA_CHARS:
|
|---|
| 3525 | if (characterDataHandler)
|
|---|
| 3526 | {
|
|---|
| 3527 | if (MUST_CONVERT(enc, s))
|
|---|
| 3528 | {
|
|---|
| 3529 | for (;;)
|
|---|
| 3530 | {
|
|---|
| 3531 | ICHAR *dataPtr = (ICHAR *) dataBuf;
|
|---|
| 3532 |
|
|---|
| 3533 | XmlConvert(enc, &s, next, &dataPtr, (ICHAR *) dataBufEnd);
|
|---|
| 3534 | *eventEndPP = next;
|
|---|
| 3535 | characterDataHandler(handlerArg, dataBuf, dataPtr - (ICHAR *) dataBuf);
|
|---|
| 3536 | if (s == next)
|
|---|
| 3537 | break;
|
|---|
| 3538 | *eventPP = s;
|
|---|
| 3539 | }
|
|---|
| 3540 | }
|
|---|
| 3541 | else
|
|---|
| 3542 | characterDataHandler(handlerArg,
|
|---|
| 3543 | (XML_Char *) s,
|
|---|
| 3544 | (XML_Char *) next - (XML_Char *) s);
|
|---|
| 3545 | }
|
|---|
| 3546 | else if (defaultHandler)
|
|---|
| 3547 | reportDefault(parser, enc, s, next);
|
|---|
| 3548 | break;
|
|---|
| 3549 | case XML_TOK_INVALID:
|
|---|
| 3550 | *eventPP = next;
|
|---|
| 3551 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 3552 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 3553 | if (nextPtr)
|
|---|
| 3554 | {
|
|---|
| 3555 | *nextPtr = s;
|
|---|
| 3556 | return ERROR_EXPAT_NONE;
|
|---|
| 3557 | }
|
|---|
| 3558 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 3559 | case XML_TOK_PARTIAL:
|
|---|
| 3560 | case XML_TOK_NONE:
|
|---|
| 3561 | if (nextPtr)
|
|---|
| 3562 | {
|
|---|
| 3563 | *nextPtr = s;
|
|---|
| 3564 | return ERROR_EXPAT_NONE;
|
|---|
| 3565 | }
|
|---|
| 3566 | return ERROR_EXPAT_UNCLOSED_CDATA_SECTION;
|
|---|
| 3567 | default:
|
|---|
| 3568 | *eventPP = next;
|
|---|
| 3569 | return ERROR_EXPAT_UNEXPECTED_STATE;
|
|---|
| 3570 | }
|
|---|
| 3571 | *eventPP = s = next;
|
|---|
| 3572 | }
|
|---|
| 3573 | /* not reached */
|
|---|
| 3574 | }
|
|---|
| 3575 |
|
|---|
| 3576 | #ifdef XML_DTD
|
|---|
| 3577 |
|
|---|
| 3578 | /* The idea here is to avoid using stack for each IGNORE section when
|
|---|
| 3579 | * the whole file is parsed with one call. */
|
|---|
| 3580 |
|
|---|
| 3581 | STATIC XMLERROR ignoreSectionProcessor(XML_Parser parser,
|
|---|
| 3582 | const char *start,
|
|---|
| 3583 | const char *end,
|
|---|
| 3584 | const char **endPtr)
|
|---|
| 3585 | {
|
|---|
| 3586 | XMLERROR result = doIgnoreSection(parser, encoding, &start, end, endPtr);
|
|---|
| 3587 |
|
|---|
| 3588 | if (start)
|
|---|
| 3589 | {
|
|---|
| 3590 | processor = prologProcessor;
|
|---|
| 3591 | return prologProcessor(parser, start, end, endPtr);
|
|---|
| 3592 | }
|
|---|
| 3593 | return result;
|
|---|
| 3594 | }
|
|---|
| 3595 |
|
|---|
| 3596 | /* startPtr gets set to non-null is the section is closed, and to null if
|
|---|
| 3597 | * the section is not yet closed. */
|
|---|
| 3598 |
|
|---|
| 3599 | STATIC XMLERROR doIgnoreSection(XML_Parser parser,
|
|---|
| 3600 | const ENCODING * enc,
|
|---|
| 3601 | const char **startPtr,
|
|---|
| 3602 | const char *end,
|
|---|
| 3603 | const char **nextPtr)
|
|---|
| 3604 | {
|
|---|
| 3605 | const char *next;
|
|---|
| 3606 | int tok;
|
|---|
| 3607 | const char *s = *startPtr;
|
|---|
| 3608 | const char **eventPP;
|
|---|
| 3609 | const char **eventEndPP;
|
|---|
| 3610 |
|
|---|
| 3611 | if (enc == encoding)
|
|---|
| 3612 | {
|
|---|
| 3613 | eventPP = &eventPtr;
|
|---|
| 3614 | *eventPP = s;
|
|---|
| 3615 | eventEndPP = &eventEndPtr;
|
|---|
| 3616 | }
|
|---|
| 3617 | else
|
|---|
| 3618 | {
|
|---|
| 3619 | eventPP = &(openInternalEntities->internalEventPtr);
|
|---|
| 3620 | eventEndPP = &(openInternalEntities->internalEventEndPtr);
|
|---|
| 3621 | }
|
|---|
| 3622 | *eventPP = s;
|
|---|
| 3623 | *startPtr = 0;
|
|---|
| 3624 | tok = XmlIgnoreSectionTok(enc, s, end, &next);
|
|---|
| 3625 | *eventEndPP = next;
|
|---|
| 3626 | switch (tok)
|
|---|
| 3627 | {
|
|---|
| 3628 | case XML_TOK_IGNORE_SECT:
|
|---|
| 3629 | if (defaultHandler)
|
|---|
| 3630 | reportDefault(parser, enc, s, next);
|
|---|
| 3631 | *startPtr = next;
|
|---|
| 3632 | return ERROR_EXPAT_NONE;
|
|---|
| 3633 | case XML_TOK_INVALID:
|
|---|
| 3634 | *eventPP = next;
|
|---|
| 3635 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 3636 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 3637 | if (nextPtr)
|
|---|
| 3638 | {
|
|---|
| 3639 | *nextPtr = s;
|
|---|
| 3640 | return ERROR_EXPAT_NONE;
|
|---|
| 3641 | }
|
|---|
| 3642 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 3643 | case XML_TOK_PARTIAL:
|
|---|
| 3644 | case XML_TOK_NONE:
|
|---|
| 3645 | if (nextPtr)
|
|---|
| 3646 | {
|
|---|
| 3647 | *nextPtr = s;
|
|---|
| 3648 | return ERROR_EXPAT_NONE;
|
|---|
| 3649 | }
|
|---|
| 3650 | return ERROR_EXPAT_SYNTAX; /* ERROR_EXPAT_UNCLOSED_IGNORE_SECTION */
|
|---|
| 3651 | default:
|
|---|
| 3652 | *eventPP = next;
|
|---|
| 3653 | return ERROR_EXPAT_UNEXPECTED_STATE;
|
|---|
| 3654 | }
|
|---|
| 3655 | /* not reached */
|
|---|
| 3656 | }
|
|---|
| 3657 |
|
|---|
| 3658 | #endif /* XML_DTD */
|
|---|
| 3659 |
|
|---|
| 3660 | STATIC XMLERROR initializeEncoding(XML_Parser parser)
|
|---|
| 3661 | {
|
|---|
| 3662 | const char *s;
|
|---|
| 3663 |
|
|---|
| 3664 | #ifdef XML_UNICODE
|
|---|
| 3665 | char encodingBuf[128];
|
|---|
| 3666 |
|
|---|
| 3667 | if (!protocolEncodingName)
|
|---|
| 3668 | s = 0;
|
|---|
| 3669 | else
|
|---|
| 3670 | {
|
|---|
| 3671 | int i;
|
|---|
| 3672 |
|
|---|
| 3673 | for (i = 0; protocolEncodingName[i]; i++)
|
|---|
| 3674 | {
|
|---|
| 3675 | if (i == sizeof(encodingBuf) - 1
|
|---|
| 3676 | || (protocolEncodingName[i] & ~0x7f) != 0)
|
|---|
| 3677 | {
|
|---|
| 3678 | encodingBuf[0] = '\0';
|
|---|
| 3679 | break;
|
|---|
| 3680 | }
|
|---|
| 3681 | encodingBuf[i] = (char)protocolEncodingName[i];
|
|---|
| 3682 | }
|
|---|
| 3683 | encodingBuf[i] = '\0';
|
|---|
| 3684 | s = encodingBuf;
|
|---|
| 3685 | }
|
|---|
| 3686 | #else
|
|---|
| 3687 | s = protocolEncodingName;
|
|---|
| 3688 | #endif
|
|---|
| 3689 | if ((ns ? XmlInitEncodingNS : XmlInitEncoding) (&initEncoding, &encoding, s))
|
|---|
| 3690 | return ERROR_EXPAT_NONE;
|
|---|
| 3691 | return handleUnknownEncoding(parser, protocolEncodingName);
|
|---|
| 3692 | }
|
|---|
| 3693 |
|
|---|
| 3694 | STATIC XMLERROR processXmlDecl(XML_Parser parser,
|
|---|
| 3695 | int isGeneralTextEntity,
|
|---|
| 3696 | const char *s,
|
|---|
| 3697 | const char *next)
|
|---|
| 3698 | {
|
|---|
| 3699 | const char *encodingName = 0;
|
|---|
| 3700 | const char *storedEncName = 0;
|
|---|
| 3701 | const ENCODING *newEncoding = 0;
|
|---|
| 3702 | const char *version = 0;
|
|---|
| 3703 | const char *versionend;
|
|---|
| 3704 | const char *storedversion = 0;
|
|---|
| 3705 | int standalone = -1;
|
|---|
| 3706 |
|
|---|
| 3707 | if (!(ns
|
|---|
| 3708 | ? XmlParseXmlDeclNS
|
|---|
| 3709 | : XmlParseXmlDecl) (isGeneralTextEntity,
|
|---|
| 3710 | encoding,
|
|---|
| 3711 | s,
|
|---|
| 3712 | next,
|
|---|
| 3713 | &eventPtr,
|
|---|
| 3714 | &version,
|
|---|
| 3715 | &versionend,
|
|---|
| 3716 | &encodingName,
|
|---|
| 3717 | &newEncoding,
|
|---|
| 3718 | &standalone))
|
|---|
| 3719 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 3720 | if (!isGeneralTextEntity && standalone == 1)
|
|---|
| 3721 | {
|
|---|
| 3722 | dtd.standalone = 1;
|
|---|
| 3723 | #ifdef XML_DTD
|
|---|
| 3724 | if (paramEntityParsing == XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE)
|
|---|
| 3725 | paramEntityParsing = XML_PARAM_ENTITY_PARSING_NEVER;
|
|---|
| 3726 | #endif /* XML_DTD */
|
|---|
| 3727 | }
|
|---|
| 3728 | if (xmlDeclHandler)
|
|---|
| 3729 | {
|
|---|
| 3730 | if (encodingName)
|
|---|
| 3731 | {
|
|---|
| 3732 | storedEncName = poolStoreString(&temp2Pool,
|
|---|
| 3733 | encoding,
|
|---|
| 3734 | encodingName,
|
|---|
| 3735 | encodingName
|
|---|
| 3736 | + XmlNameLength(encoding, encodingName),
|
|---|
| 3737 | NULL);
|
|---|
| 3738 | if (!storedEncName)
|
|---|
| 3739 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3740 | poolFinish(&temp2Pool);
|
|---|
| 3741 | }
|
|---|
| 3742 | if (version)
|
|---|
| 3743 | {
|
|---|
| 3744 | storedversion = poolStoreString(&temp2Pool,
|
|---|
| 3745 | encoding,
|
|---|
| 3746 | version,
|
|---|
| 3747 | versionend - encoding->minBytesPerChar,
|
|---|
| 3748 | NULL);
|
|---|
| 3749 | if (!storedversion)
|
|---|
| 3750 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3751 | }
|
|---|
| 3752 | xmlDeclHandler(handlerArg, storedversion, storedEncName, standalone);
|
|---|
| 3753 | }
|
|---|
| 3754 | else if (defaultHandler)
|
|---|
| 3755 | reportDefault(parser, encoding, s, next);
|
|---|
| 3756 | if (!protocolEncodingName)
|
|---|
| 3757 | {
|
|---|
| 3758 | if (newEncoding)
|
|---|
| 3759 | {
|
|---|
| 3760 | if (newEncoding->minBytesPerChar != encoding->minBytesPerChar)
|
|---|
| 3761 | {
|
|---|
| 3762 | eventPtr = encodingName;
|
|---|
| 3763 | return ERROR_EXPAT_INCORRECT_ENCODING;
|
|---|
| 3764 | }
|
|---|
| 3765 | encoding = newEncoding;
|
|---|
| 3766 | }
|
|---|
| 3767 | else if (encodingName)
|
|---|
| 3768 | {
|
|---|
| 3769 | XMLERROR result;
|
|---|
| 3770 |
|
|---|
| 3771 | if (!storedEncName)
|
|---|
| 3772 | {
|
|---|
| 3773 | storedEncName = poolStoreString(&temp2Pool,
|
|---|
| 3774 | encoding,
|
|---|
| 3775 | encodingName,
|
|---|
| 3776 | encodingName
|
|---|
| 3777 | + XmlNameLength(encoding, encodingName),
|
|---|
| 3778 | NULL);
|
|---|
| 3779 | if (!storedEncName)
|
|---|
| 3780 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3781 | }
|
|---|
| 3782 | result = handleUnknownEncoding(parser, storedEncName);
|
|---|
| 3783 | poolClear(&tempPool);
|
|---|
| 3784 | if (result == ERROR_EXPAT_UNKNOWN_ENCODING)
|
|---|
| 3785 | eventPtr = encodingName;
|
|---|
| 3786 | return result;
|
|---|
| 3787 | }
|
|---|
| 3788 | }
|
|---|
| 3789 |
|
|---|
| 3790 | if (storedEncName || storedversion)
|
|---|
| 3791 | poolClear(&temp2Pool);
|
|---|
| 3792 |
|
|---|
| 3793 | return ERROR_EXPAT_NONE;
|
|---|
| 3794 | }
|
|---|
| 3795 |
|
|---|
| 3796 | STATIC XMLERROR handleUnknownEncoding(XML_Parser parser,
|
|---|
| 3797 | const XML_Char * encodingName)
|
|---|
| 3798 | {
|
|---|
| 3799 | if (unknownEncodingHandler)
|
|---|
| 3800 | {
|
|---|
| 3801 | XML_Encoding info;
|
|---|
| 3802 | int i;
|
|---|
| 3803 |
|
|---|
| 3804 | for (i = 0; i < 256; i++)
|
|---|
| 3805 | info.map[i] = -1;
|
|---|
| 3806 | info.convert = 0;
|
|---|
| 3807 | info.data = 0;
|
|---|
| 3808 | info.release = 0;
|
|---|
| 3809 | if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, &info))
|
|---|
| 3810 | {
|
|---|
| 3811 | ENCODING *enc;
|
|---|
| 3812 |
|
|---|
| 3813 | unknownEncodingMem = MALLOC(XmlSizeOfUnknownEncoding());
|
|---|
| 3814 | if (!unknownEncodingMem)
|
|---|
| 3815 | {
|
|---|
| 3816 | if (info.release)
|
|---|
| 3817 | info.release(info.data);
|
|---|
| 3818 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3819 | }
|
|---|
| 3820 | enc = (ns
|
|---|
| 3821 | ? XmlInitUnknownEncodingNS
|
|---|
| 3822 | : XmlInitUnknownEncoding) (unknownEncodingMem,
|
|---|
| 3823 | info.map,
|
|---|
| 3824 | info.convert,
|
|---|
| 3825 | info.data);
|
|---|
| 3826 | if (enc)
|
|---|
| 3827 | {
|
|---|
| 3828 | unknownEncodingData = info.data;
|
|---|
| 3829 | unknownEncodingRelease = (void(*)(void*))info.release;
|
|---|
| 3830 | encoding = enc;
|
|---|
| 3831 | return ERROR_EXPAT_NONE;
|
|---|
| 3832 | }
|
|---|
| 3833 | }
|
|---|
| 3834 | if (info.release)
|
|---|
| 3835 | info.release(info.data);
|
|---|
| 3836 | }
|
|---|
| 3837 | return ERROR_EXPAT_UNKNOWN_ENCODING;
|
|---|
| 3838 | }
|
|---|
| 3839 |
|
|---|
| 3840 | STATIC XMLERROR prologInitProcessor(XML_Parser parser,
|
|---|
| 3841 | const char *s,
|
|---|
| 3842 | const char *end,
|
|---|
| 3843 | const char **nextPtr)
|
|---|
| 3844 | {
|
|---|
| 3845 | XMLERROR result = initializeEncoding(parser);
|
|---|
| 3846 |
|
|---|
| 3847 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 3848 | return result;
|
|---|
| 3849 | processor = prologProcessor;
|
|---|
| 3850 | return prologProcessor(parser, s, end, nextPtr);
|
|---|
| 3851 | }
|
|---|
| 3852 |
|
|---|
| 3853 | STATIC XMLERROR prologProcessor(XML_Parser parser,
|
|---|
| 3854 | const char *s,
|
|---|
| 3855 | const char *end,
|
|---|
| 3856 | const char **nextPtr)
|
|---|
| 3857 | {
|
|---|
| 3858 | const char *next;
|
|---|
| 3859 | int tok = XmlPrologTok(encoding, s, end, &next);
|
|---|
| 3860 |
|
|---|
| 3861 | return doProlog(parser, encoding, s, end, tok, next, nextPtr);
|
|---|
| 3862 | }
|
|---|
| 3863 |
|
|---|
| 3864 | STATIC XMLERROR doProlog(XML_Parser parser,
|
|---|
| 3865 | const ENCODING * enc,
|
|---|
| 3866 | const char *s,
|
|---|
| 3867 | const char *end,
|
|---|
| 3868 | int tok,
|
|---|
| 3869 | const char *next,
|
|---|
| 3870 | const char **nextPtr)
|
|---|
| 3871 | {
|
|---|
| 3872 | #ifdef XML_DTD
|
|---|
| 3873 | static const XML_Char externalSubsetName[] =
|
|---|
| 3874 | {'#', '\0'};
|
|---|
| 3875 |
|
|---|
| 3876 | #endif /* XML_DTD */
|
|---|
| 3877 |
|
|---|
| 3878 | const char **eventPP;
|
|---|
| 3879 | const char **eventEndPP;
|
|---|
| 3880 | enum XML_Content_Quant quant;
|
|---|
| 3881 |
|
|---|
| 3882 | if (enc == encoding)
|
|---|
| 3883 | {
|
|---|
| 3884 | eventPP = &eventPtr;
|
|---|
| 3885 | eventEndPP = &eventEndPtr;
|
|---|
| 3886 | }
|
|---|
| 3887 | else
|
|---|
| 3888 | {
|
|---|
| 3889 | eventPP = &(openInternalEntities->internalEventPtr);
|
|---|
| 3890 | eventEndPP = &(openInternalEntities->internalEventEndPtr);
|
|---|
| 3891 | }
|
|---|
| 3892 | for (;;)
|
|---|
| 3893 | {
|
|---|
| 3894 | int role;
|
|---|
| 3895 |
|
|---|
| 3896 | *eventPP = s;
|
|---|
| 3897 | *eventEndPP = next;
|
|---|
| 3898 | if (tok <= 0)
|
|---|
| 3899 | {
|
|---|
| 3900 | if (nextPtr != 0 && tok != XML_TOK_INVALID)
|
|---|
| 3901 | {
|
|---|
| 3902 | *nextPtr = s;
|
|---|
| 3903 | return ERROR_EXPAT_NONE;
|
|---|
| 3904 | }
|
|---|
| 3905 | switch (tok)
|
|---|
| 3906 | {
|
|---|
| 3907 | case XML_TOK_INVALID:
|
|---|
| 3908 | *eventPP = next;
|
|---|
| 3909 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 3910 | case XML_TOK_PARTIAL:
|
|---|
| 3911 | return ERROR_EXPAT_UNCLOSED_TOKEN;
|
|---|
| 3912 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 3913 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 3914 | case XML_TOK_NONE:
|
|---|
| 3915 | #ifdef XML_DTD
|
|---|
| 3916 | if (enc != encoding)
|
|---|
| 3917 | return ERROR_EXPAT_NONE;
|
|---|
| 3918 | if (parentParser)
|
|---|
| 3919 | {
|
|---|
| 3920 | if (XmlTokenRole(&prologState, XML_TOK_NONE, end, end, enc)
|
|---|
| 3921 | == XML_ROLE_ERROR)
|
|---|
| 3922 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 3923 | hadExternalDoctype = 0;
|
|---|
| 3924 | return ERROR_EXPAT_NONE;
|
|---|
| 3925 | }
|
|---|
| 3926 | #endif /* XML_DTD */
|
|---|
| 3927 | return ERROR_EXPAT_NO_ELEMENTS;
|
|---|
| 3928 | default:
|
|---|
| 3929 | tok = -tok;
|
|---|
| 3930 | next = end;
|
|---|
| 3931 | break;
|
|---|
| 3932 | }
|
|---|
| 3933 | }
|
|---|
| 3934 | role = XmlTokenRole(&prologState, tok, s, next, enc);
|
|---|
| 3935 | switch (role)
|
|---|
| 3936 | {
|
|---|
| 3937 | case XML_ROLE_XML_DECL:
|
|---|
| 3938 | {
|
|---|
| 3939 | XMLERROR result = processXmlDecl(parser, 0, s, next);
|
|---|
| 3940 |
|
|---|
| 3941 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 3942 | return result;
|
|---|
| 3943 | enc = encoding;
|
|---|
| 3944 | }
|
|---|
| 3945 | break;
|
|---|
| 3946 | case XML_ROLE_DOCTYPE_NAME:
|
|---|
| 3947 | if (startDoctypeDeclHandler)
|
|---|
| 3948 | {
|
|---|
| 3949 | doctypeName = poolStoreString(&tempPool, enc, s, next, NULL);
|
|---|
| 3950 | if (!doctypeName)
|
|---|
| 3951 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3952 | poolFinish(&tempPool);
|
|---|
| 3953 | doctypeSysid = 0;
|
|---|
| 3954 | doctypePubid = 0;
|
|---|
| 3955 | }
|
|---|
| 3956 | break;
|
|---|
| 3957 | case XML_ROLE_DOCTYPE_INTERNAL_SUBSET:
|
|---|
| 3958 | if (startDoctypeDeclHandler)
|
|---|
| 3959 | {
|
|---|
| 3960 | startDoctypeDeclHandler(handlerArg, doctypeName, doctypeSysid,
|
|---|
| 3961 | doctypePubid, 1);
|
|---|
| 3962 | doctypeName = 0;
|
|---|
| 3963 | poolClear(&tempPool);
|
|---|
| 3964 | }
|
|---|
| 3965 | break;
|
|---|
| 3966 | #ifdef XML_DTD
|
|---|
| 3967 | case XML_ROLE_TEXT_DECL:
|
|---|
| 3968 | {
|
|---|
| 3969 | XMLERROR result = processXmlDecl(parser, 1, s, next);
|
|---|
| 3970 |
|
|---|
| 3971 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 3972 | return result;
|
|---|
| 3973 | enc = encoding;
|
|---|
| 3974 | }
|
|---|
| 3975 | break;
|
|---|
| 3976 | #endif /* XML_DTD */
|
|---|
| 3977 | case XML_ROLE_DOCTYPE_PUBLIC_ID:
|
|---|
| 3978 | if (startDoctypeDeclHandler)
|
|---|
| 3979 | {
|
|---|
| 3980 | doctypePubid = poolStoreString(&tempPool, enc, s + 1, next - 1, NULL);
|
|---|
| 3981 | if (!doctypePubid)
|
|---|
| 3982 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3983 | poolFinish(&tempPool);
|
|---|
| 3984 | }
|
|---|
| 3985 | #ifdef XML_DTD
|
|---|
| 3986 | declEntity = (ENTITY *) lookup(&dtd.paramEntities,
|
|---|
| 3987 | externalSubsetName,
|
|---|
| 3988 | sizeof(ENTITY));
|
|---|
| 3989 | if (!declEntity)
|
|---|
| 3990 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 3991 | #endif /* XML_DTD */
|
|---|
| 3992 | /* fall through */
|
|---|
| 3993 | case XML_ROLE_ENTITY_PUBLIC_ID:
|
|---|
| 3994 | if (!XmlIsPublicId(enc, s, next, eventPP))
|
|---|
| 3995 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 3996 | if (declEntity)
|
|---|
| 3997 | {
|
|---|
| 3998 | XML_Char *tem = poolStoreString(&dtd.pool,
|
|---|
| 3999 | enc,
|
|---|
| 4000 | s + enc->minBytesPerChar,
|
|---|
| 4001 | next - enc->minBytesPerChar,
|
|---|
| 4002 | NULL);
|
|---|
| 4003 |
|
|---|
| 4004 | if (!tem)
|
|---|
| 4005 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4006 | normalizePublicId(tem);
|
|---|
| 4007 | declEntity->publicId = tem;
|
|---|
| 4008 | poolFinish(&dtd.pool);
|
|---|
| 4009 | }
|
|---|
| 4010 | break;
|
|---|
| 4011 | case XML_ROLE_DOCTYPE_CLOSE:
|
|---|
| 4012 | if (doctypeName)
|
|---|
| 4013 | {
|
|---|
| 4014 | startDoctypeDeclHandler(handlerArg, doctypeName,
|
|---|
| 4015 | doctypeSysid, doctypePubid, 0);
|
|---|
| 4016 | poolClear(&tempPool);
|
|---|
| 4017 | }
|
|---|
| 4018 | if (dtd.complete && hadExternalDoctype)
|
|---|
| 4019 | {
|
|---|
| 4020 | dtd.complete = 0;
|
|---|
| 4021 | #ifdef XML_DTD
|
|---|
| 4022 | if (paramEntityParsing && externalEntityRefHandler)
|
|---|
| 4023 | {
|
|---|
| 4024 | ENTITY *entity = (ENTITY *) lookup(&dtd.paramEntities,
|
|---|
| 4025 | externalSubsetName,
|
|---|
| 4026 | 0);
|
|---|
| 4027 |
|
|---|
| 4028 | if (!externalEntityRefHandler(handlerArg, // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 4029 | externalEntityRefHandlerArg,
|
|---|
| 4030 | 0,
|
|---|
| 4031 | entity->base,
|
|---|
| 4032 | entity->systemId,
|
|---|
| 4033 | entity->publicId))
|
|---|
| 4034 | return ERROR_EXPAT_EXTERNAL_ENTITY_HANDLING;
|
|---|
| 4035 | }
|
|---|
| 4036 | #endif /* XML_DTD */
|
|---|
| 4037 | if (!dtd.complete
|
|---|
| 4038 | && !dtd.standalone
|
|---|
| 4039 | && notStandaloneHandler
|
|---|
| 4040 | && !notStandaloneHandler(handlerArg))
|
|---|
| 4041 | return ERROR_EXPAT_NOT_STANDALONE;
|
|---|
| 4042 | }
|
|---|
| 4043 | if (endDoctypeDeclHandler)
|
|---|
| 4044 | endDoctypeDeclHandler(handlerArg);
|
|---|
| 4045 | break;
|
|---|
| 4046 | case XML_ROLE_INSTANCE_START:
|
|---|
| 4047 | processor = contentProcessor;
|
|---|
| 4048 | return contentProcessor(parser, s, end, nextPtr);
|
|---|
| 4049 | case XML_ROLE_ATTLIST_ELEMENT_NAME:
|
|---|
| 4050 | declElementType = getElementType(parser, enc, s, next);
|
|---|
| 4051 | if (!declElementType)
|
|---|
| 4052 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4053 | break;
|
|---|
| 4054 | case XML_ROLE_ATTRIBUTE_NAME:
|
|---|
| 4055 | declAttributeId = getAttributeId(parser, enc, s, next);
|
|---|
| 4056 | if (!declAttributeId)
|
|---|
| 4057 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4058 | declAttributeIsCdata = 0;
|
|---|
| 4059 | declAttributeType = 0;
|
|---|
| 4060 | declAttributeIsId = 0;
|
|---|
| 4061 | break;
|
|---|
| 4062 | case XML_ROLE_ATTRIBUTE_TYPE_CDATA:
|
|---|
| 4063 | declAttributeIsCdata = 1;
|
|---|
| 4064 | declAttributeType = "CDATA";
|
|---|
| 4065 | break;
|
|---|
| 4066 | case XML_ROLE_ATTRIBUTE_TYPE_ID:
|
|---|
| 4067 | declAttributeIsId = 1;
|
|---|
| 4068 | declAttributeType = "ID";
|
|---|
| 4069 | break;
|
|---|
| 4070 | case XML_ROLE_ATTRIBUTE_TYPE_IDREF:
|
|---|
| 4071 | declAttributeType = "IDREF";
|
|---|
| 4072 | break;
|
|---|
| 4073 | case XML_ROLE_ATTRIBUTE_TYPE_IDREFS:
|
|---|
| 4074 | declAttributeType = "IDREFS";
|
|---|
| 4075 | break;
|
|---|
| 4076 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITY:
|
|---|
| 4077 | declAttributeType = "ENTITY";
|
|---|
| 4078 | break;
|
|---|
| 4079 | case XML_ROLE_ATTRIBUTE_TYPE_ENTITIES:
|
|---|
| 4080 | declAttributeType = "ENTITIES";
|
|---|
| 4081 | break;
|
|---|
| 4082 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKEN:
|
|---|
| 4083 | declAttributeType = "NMTOKEN";
|
|---|
| 4084 | break;
|
|---|
| 4085 | case XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS:
|
|---|
| 4086 | declAttributeType = "NMTOKENS";
|
|---|
| 4087 | break;
|
|---|
| 4088 |
|
|---|
| 4089 | case XML_ROLE_ATTRIBUTE_ENUM_VALUE:
|
|---|
| 4090 | case XML_ROLE_ATTRIBUTE_NOTATION_VALUE:
|
|---|
| 4091 | if (attlistDeclHandler)
|
|---|
| 4092 | {
|
|---|
| 4093 | char *prefix;
|
|---|
| 4094 |
|
|---|
| 4095 | if (declAttributeType)
|
|---|
| 4096 | {
|
|---|
| 4097 | prefix = "|";
|
|---|
| 4098 | }
|
|---|
| 4099 | else
|
|---|
| 4100 | {
|
|---|
| 4101 | prefix = (role == XML_ROLE_ATTRIBUTE_NOTATION_VALUE
|
|---|
| 4102 | ? "NOTATION("
|
|---|
| 4103 | : "(");
|
|---|
| 4104 | }
|
|---|
| 4105 | if (!poolAppendString(&tempPool, prefix))
|
|---|
| 4106 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4107 | if (!poolAppend(&tempPool, enc, s, next, NULL))
|
|---|
| 4108 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4109 | declAttributeType = tempPool.start;
|
|---|
| 4110 | }
|
|---|
| 4111 | break;
|
|---|
| 4112 | case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE:
|
|---|
| 4113 | case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE:
|
|---|
| 4114 | if (dtd.complete
|
|---|
| 4115 | && !defineAttribute(declElementType, declAttributeId,
|
|---|
| 4116 | declAttributeIsCdata, declAttributeIsId, 0,
|
|---|
| 4117 | parser))
|
|---|
| 4118 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4119 | if (attlistDeclHandler && declAttributeType)
|
|---|
| 4120 | {
|
|---|
| 4121 | if (*declAttributeType == '('
|
|---|
| 4122 | || (*declAttributeType == 'N' && declAttributeType[1] == 'O'))
|
|---|
| 4123 | {
|
|---|
| 4124 | /* Enumerated or Notation type */
|
|---|
| 4125 | if (!poolAppendChar(&tempPool, ')')
|
|---|
| 4126 | || !poolAppendChar(&tempPool, '\0'))
|
|---|
| 4127 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4128 | declAttributeType = tempPool.start;
|
|---|
| 4129 | poolFinish(&tempPool);
|
|---|
| 4130 | }
|
|---|
| 4131 | *eventEndPP = s;
|
|---|
| 4132 | attlistDeclHandler(handlerArg, declElementType->name,
|
|---|
| 4133 | declAttributeId->name, declAttributeType,
|
|---|
| 4134 | 0, role == XML_ROLE_REQUIRED_ATTRIBUTE_VALUE);
|
|---|
| 4135 | poolClear(&tempPool);
|
|---|
| 4136 | }
|
|---|
| 4137 | break;
|
|---|
| 4138 | case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE:
|
|---|
| 4139 | case XML_ROLE_FIXED_ATTRIBUTE_VALUE:
|
|---|
| 4140 | {
|
|---|
| 4141 | const XML_Char *attVal;
|
|---|
| 4142 | XMLERROR result
|
|---|
| 4143 | = storeAttributeValue(parser, enc, declAttributeIsCdata,
|
|---|
| 4144 | s + enc->minBytesPerChar,
|
|---|
| 4145 | next - enc->minBytesPerChar,
|
|---|
| 4146 | &dtd.pool);
|
|---|
| 4147 |
|
|---|
| 4148 | if (result)
|
|---|
| 4149 | return result;
|
|---|
| 4150 | attVal = poolStart(&dtd.pool);
|
|---|
| 4151 | poolFinish(&dtd.pool);
|
|---|
| 4152 | if (dtd.complete
|
|---|
| 4153 | /* ID attributes aren't allowed to have a default */
|
|---|
| 4154 | && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, 0, attVal, parser))
|
|---|
| 4155 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4156 | if (attlistDeclHandler && declAttributeType)
|
|---|
| 4157 | {
|
|---|
| 4158 | if (*declAttributeType == '('
|
|---|
| 4159 | || (*declAttributeType == 'N' && declAttributeType[1] == 'O'))
|
|---|
| 4160 | {
|
|---|
| 4161 | /* Enumerated or Notation type */
|
|---|
| 4162 | if (!poolAppendChar(&tempPool, ')')
|
|---|
| 4163 | || !poolAppendChar(&tempPool, '\0'))
|
|---|
| 4164 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4165 | declAttributeType = tempPool.start;
|
|---|
| 4166 | poolFinish(&tempPool);
|
|---|
| 4167 | }
|
|---|
| 4168 | *eventEndPP = s;
|
|---|
| 4169 | attlistDeclHandler(handlerArg, declElementType->name,
|
|---|
| 4170 | declAttributeId->name, declAttributeType,
|
|---|
| 4171 | attVal,
|
|---|
| 4172 | role == XML_ROLE_FIXED_ATTRIBUTE_VALUE);
|
|---|
| 4173 | poolClear(&tempPool);
|
|---|
| 4174 | }
|
|---|
| 4175 | break;
|
|---|
| 4176 | }
|
|---|
| 4177 | case XML_ROLE_ENTITY_VALUE:
|
|---|
| 4178 | {
|
|---|
| 4179 | XMLERROR result = storeEntityValue(parser, enc,
|
|---|
| 4180 | s + enc->minBytesPerChar,
|
|---|
| 4181 | next - enc->minBytesPerChar);
|
|---|
| 4182 |
|
|---|
| 4183 | if (declEntity)
|
|---|
| 4184 | {
|
|---|
| 4185 | declEntity->textPtr = poolStart(&dtd.pool);
|
|---|
| 4186 | declEntity->textLen = poolLength(&dtd.pool);
|
|---|
| 4187 | poolFinish(&dtd.pool);
|
|---|
| 4188 | if (entityDeclHandler)
|
|---|
| 4189 | {
|
|---|
| 4190 | *eventEndPP = s;
|
|---|
| 4191 | entityDeclHandler(handlerArg,
|
|---|
| 4192 | declEntity->name,
|
|---|
| 4193 | declEntity->is_param,
|
|---|
| 4194 | declEntity->textPtr,
|
|---|
| 4195 | declEntity->textLen,
|
|---|
| 4196 | curBase, 0, 0, 0);
|
|---|
| 4197 | }
|
|---|
| 4198 | }
|
|---|
| 4199 | else
|
|---|
| 4200 | poolDiscard(&dtd.pool);
|
|---|
| 4201 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 4202 | return result;
|
|---|
| 4203 | }
|
|---|
| 4204 | break;
|
|---|
| 4205 | case XML_ROLE_DOCTYPE_SYSTEM_ID:
|
|---|
| 4206 | if (startDoctypeDeclHandler)
|
|---|
| 4207 | {
|
|---|
| 4208 | doctypeSysid = poolStoreString(&tempPool,
|
|---|
| 4209 | enc,
|
|---|
| 4210 | s + 1,
|
|---|
| 4211 | next - 1,
|
|---|
| 4212 | NULL);
|
|---|
| 4213 | if (!doctypeSysid)
|
|---|
| 4214 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4215 | poolFinish(&tempPool);
|
|---|
| 4216 | }
|
|---|
| 4217 | if (!dtd.standalone
|
|---|
| 4218 | #ifdef XML_DTD
|
|---|
| 4219 | && !paramEntityParsing
|
|---|
| 4220 | #endif /* XML_DTD */
|
|---|
| 4221 | && notStandaloneHandler
|
|---|
| 4222 | && !notStandaloneHandler(handlerArg))
|
|---|
| 4223 | return ERROR_EXPAT_NOT_STANDALONE;
|
|---|
| 4224 | hadExternalDoctype = 1;
|
|---|
| 4225 | #ifndef XML_DTD
|
|---|
| 4226 | break;
|
|---|
| 4227 | #else /* XML_DTD */
|
|---|
| 4228 | if (!declEntity)
|
|---|
| 4229 | {
|
|---|
| 4230 | declEntity = (ENTITY *) lookup(&dtd.paramEntities,
|
|---|
| 4231 | externalSubsetName,
|
|---|
| 4232 | sizeof(ENTITY));
|
|---|
| 4233 | declEntity->publicId = 0;
|
|---|
| 4234 | if (!declEntity)
|
|---|
| 4235 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4236 | }
|
|---|
| 4237 | /* fall through */
|
|---|
| 4238 | #endif /* XML_DTD */
|
|---|
| 4239 | case XML_ROLE_ENTITY_SYSTEM_ID:
|
|---|
| 4240 | if (declEntity)
|
|---|
| 4241 | {
|
|---|
| 4242 | declEntity->systemId = poolStoreString(&dtd.pool,
|
|---|
| 4243 | enc,
|
|---|
| 4244 | s + enc->minBytesPerChar,
|
|---|
| 4245 | next - enc->minBytesPerChar,
|
|---|
| 4246 | NULL);
|
|---|
| 4247 | if (!declEntity->systemId)
|
|---|
| 4248 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4249 | declEntity->base = curBase;
|
|---|
| 4250 | poolFinish(&dtd.pool);
|
|---|
| 4251 | }
|
|---|
| 4252 | break;
|
|---|
| 4253 | case XML_ROLE_ENTITY_COMPLETE:
|
|---|
| 4254 | if (declEntity && entityDeclHandler)
|
|---|
| 4255 | {
|
|---|
| 4256 | *eventEndPP = s;
|
|---|
| 4257 | entityDeclHandler(handlerArg,
|
|---|
| 4258 | declEntity->name,
|
|---|
| 4259 | 0, 0, 0,
|
|---|
| 4260 | declEntity->base,
|
|---|
| 4261 | declEntity->systemId,
|
|---|
| 4262 | declEntity->publicId,
|
|---|
| 4263 | 0);
|
|---|
| 4264 | }
|
|---|
| 4265 | break;
|
|---|
| 4266 | case XML_ROLE_ENTITY_NOTATION_NAME:
|
|---|
| 4267 | if (declEntity)
|
|---|
| 4268 | {
|
|---|
| 4269 | declEntity->notation = poolStoreString(&dtd.pool,
|
|---|
| 4270 | enc,
|
|---|
| 4271 | s,
|
|---|
| 4272 | next,
|
|---|
| 4273 | NULL);
|
|---|
| 4274 | if (!declEntity->notation)
|
|---|
| 4275 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4276 | poolFinish(&dtd.pool);
|
|---|
| 4277 | if (unparsedEntityDeclHandler)
|
|---|
| 4278 | {
|
|---|
| 4279 | *eventEndPP = s;
|
|---|
| 4280 | unparsedEntityDeclHandler(handlerArg,
|
|---|
| 4281 | declEntity->name,
|
|---|
| 4282 | declEntity->base,
|
|---|
| 4283 | declEntity->systemId,
|
|---|
| 4284 | declEntity->publicId,
|
|---|
| 4285 | declEntity->notation);
|
|---|
| 4286 | }
|
|---|
| 4287 | else if (entityDeclHandler)
|
|---|
| 4288 | {
|
|---|
| 4289 | *eventEndPP = s;
|
|---|
| 4290 | entityDeclHandler(handlerArg,
|
|---|
| 4291 | declEntity->name,
|
|---|
| 4292 | 0, 0, 0,
|
|---|
| 4293 | declEntity->base,
|
|---|
| 4294 | declEntity->systemId,
|
|---|
| 4295 | declEntity->publicId,
|
|---|
| 4296 | declEntity->notation);
|
|---|
| 4297 | }
|
|---|
| 4298 | }
|
|---|
| 4299 | break;
|
|---|
| 4300 | case XML_ROLE_GENERAL_ENTITY_NAME:
|
|---|
| 4301 | {
|
|---|
| 4302 | const XML_Char *name;
|
|---|
| 4303 |
|
|---|
| 4304 | if (XmlPredefinedEntityName(enc, s, next))
|
|---|
| 4305 | {
|
|---|
| 4306 | declEntity = 0;
|
|---|
| 4307 | break;
|
|---|
| 4308 | }
|
|---|
| 4309 | name = poolStoreString(&dtd.pool,
|
|---|
| 4310 | enc,
|
|---|
| 4311 | s,
|
|---|
| 4312 | next,
|
|---|
| 4313 | NULL);
|
|---|
| 4314 | if (!name)
|
|---|
| 4315 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4316 | if (dtd.complete)
|
|---|
| 4317 | {
|
|---|
| 4318 | declEntity = (ENTITY *) lookup(&dtd.generalEntities, name, sizeof(ENTITY));
|
|---|
| 4319 | if (!declEntity)
|
|---|
| 4320 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4321 | if (declEntity->name != name)
|
|---|
| 4322 | {
|
|---|
| 4323 | poolDiscard(&dtd.pool);
|
|---|
| 4324 | declEntity = 0;
|
|---|
| 4325 | }
|
|---|
| 4326 | else
|
|---|
| 4327 | {
|
|---|
| 4328 | poolFinish(&dtd.pool);
|
|---|
| 4329 | declEntity->publicId = 0;
|
|---|
| 4330 | declEntity->is_param = 0;
|
|---|
| 4331 | }
|
|---|
| 4332 | }
|
|---|
| 4333 | else
|
|---|
| 4334 | {
|
|---|
| 4335 | poolDiscard(&dtd.pool);
|
|---|
| 4336 | declEntity = 0;
|
|---|
| 4337 | }
|
|---|
| 4338 | }
|
|---|
| 4339 | break;
|
|---|
| 4340 | case XML_ROLE_PARAM_ENTITY_NAME:
|
|---|
| 4341 | #ifdef XML_DTD
|
|---|
| 4342 | if (dtd.complete)
|
|---|
| 4343 | {
|
|---|
| 4344 | const XML_Char *name = poolStoreString(&dtd.pool, enc, s, next, NULL);
|
|---|
| 4345 |
|
|---|
| 4346 | if (!name)
|
|---|
| 4347 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4348 | declEntity = (ENTITY *) lookup(&dtd.paramEntities,
|
|---|
| 4349 | name, sizeof(ENTITY));
|
|---|
| 4350 | if (!declEntity)
|
|---|
| 4351 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4352 | if (declEntity->name != name)
|
|---|
| 4353 | {
|
|---|
| 4354 | poolDiscard(&dtd.pool);
|
|---|
| 4355 | declEntity = 0;
|
|---|
| 4356 | }
|
|---|
| 4357 | else
|
|---|
| 4358 | {
|
|---|
| 4359 | poolFinish(&dtd.pool);
|
|---|
| 4360 | declEntity->publicId = 0;
|
|---|
| 4361 | declEntity->is_param = 1;
|
|---|
| 4362 | }
|
|---|
| 4363 | }
|
|---|
| 4364 | #else /* not XML_DTD */
|
|---|
| 4365 | declEntity = 0;
|
|---|
| 4366 | #endif /* not XML_DTD */
|
|---|
| 4367 | break;
|
|---|
| 4368 | case XML_ROLE_NOTATION_NAME:
|
|---|
| 4369 | declNotationPublicId = 0;
|
|---|
| 4370 | declNotationName = 0;
|
|---|
| 4371 | if (notationDeclHandler)
|
|---|
| 4372 | {
|
|---|
| 4373 | declNotationName = poolStoreString(&tempPool, enc, s, next, NULL);
|
|---|
| 4374 | if (!declNotationName)
|
|---|
| 4375 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4376 | poolFinish(&tempPool);
|
|---|
| 4377 | }
|
|---|
| 4378 | break;
|
|---|
| 4379 | case XML_ROLE_NOTATION_PUBLIC_ID:
|
|---|
| 4380 | if (!XmlIsPublicId(enc, s, next, eventPP))
|
|---|
| 4381 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 4382 | if (declNotationName)
|
|---|
| 4383 | {
|
|---|
| 4384 | XML_Char *tem = poolStoreString(&tempPool,
|
|---|
| 4385 | enc,
|
|---|
| 4386 | s + enc->minBytesPerChar,
|
|---|
| 4387 | next - enc->minBytesPerChar,
|
|---|
| 4388 | NULL);
|
|---|
| 4389 |
|
|---|
| 4390 | if (!tem)
|
|---|
| 4391 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4392 | normalizePublicId(tem);
|
|---|
| 4393 | declNotationPublicId = tem;
|
|---|
| 4394 | poolFinish(&tempPool);
|
|---|
| 4395 | }
|
|---|
| 4396 | break;
|
|---|
| 4397 | case XML_ROLE_NOTATION_SYSTEM_ID:
|
|---|
| 4398 | if (declNotationName && notationDeclHandler)
|
|---|
| 4399 | {
|
|---|
| 4400 | const XML_Char *systemId
|
|---|
| 4401 | = poolStoreString(&tempPool, enc,
|
|---|
| 4402 | s + enc->minBytesPerChar,
|
|---|
| 4403 | next - enc->minBytesPerChar,
|
|---|
| 4404 | NULL);
|
|---|
| 4405 |
|
|---|
| 4406 | if (!systemId)
|
|---|
| 4407 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4408 | *eventEndPP = s;
|
|---|
| 4409 | notationDeclHandler(handlerArg,
|
|---|
| 4410 | declNotationName,
|
|---|
| 4411 | curBase,
|
|---|
| 4412 | systemId,
|
|---|
| 4413 | declNotationPublicId);
|
|---|
| 4414 | }
|
|---|
| 4415 | poolClear(&tempPool);
|
|---|
| 4416 | break;
|
|---|
| 4417 | case XML_ROLE_NOTATION_NO_SYSTEM_ID:
|
|---|
| 4418 | if (declNotationPublicId && notationDeclHandler)
|
|---|
| 4419 | {
|
|---|
| 4420 | *eventEndPP = s;
|
|---|
| 4421 | notationDeclHandler(handlerArg,
|
|---|
| 4422 | declNotationName,
|
|---|
| 4423 | curBase,
|
|---|
| 4424 | 0,
|
|---|
| 4425 | declNotationPublicId);
|
|---|
| 4426 | }
|
|---|
| 4427 | poolClear(&tempPool);
|
|---|
| 4428 | break;
|
|---|
| 4429 | case XML_ROLE_ERROR:
|
|---|
| 4430 | switch (tok)
|
|---|
| 4431 | {
|
|---|
| 4432 | case XML_TOK_PARAM_ENTITY_REF:
|
|---|
| 4433 | return ERROR_EXPAT_PARAM_ENTITY_REF;
|
|---|
| 4434 | case XML_TOK_XML_DECL:
|
|---|
| 4435 | return ERROR_EXPAT_MISPLACED_XML_PI;
|
|---|
| 4436 | default:
|
|---|
| 4437 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 4438 | }
|
|---|
| 4439 | #ifdef XML_DTD
|
|---|
| 4440 | case XML_ROLE_IGNORE_SECT:
|
|---|
| 4441 | {
|
|---|
| 4442 | XMLERROR result;
|
|---|
| 4443 |
|
|---|
| 4444 | if (defaultHandler)
|
|---|
| 4445 | reportDefault(parser, enc, s, next);
|
|---|
| 4446 | result = doIgnoreSection(parser, enc, &next, end, nextPtr);
|
|---|
| 4447 | if (!next)
|
|---|
| 4448 | {
|
|---|
| 4449 | processor = ignoreSectionProcessor;
|
|---|
| 4450 | return result;
|
|---|
| 4451 | }
|
|---|
| 4452 | }
|
|---|
| 4453 | break;
|
|---|
| 4454 | #endif /* XML_DTD */
|
|---|
| 4455 | case XML_ROLE_GROUP_OPEN:
|
|---|
| 4456 | if (prologState.level >= groupSize)
|
|---|
| 4457 | {
|
|---|
| 4458 | if (groupSize)
|
|---|
| 4459 | {
|
|---|
| 4460 | groupConnector = (char*)REALLOC(groupConnector, groupSize *= 2);
|
|---|
| 4461 | if (dtd.scaffIndex)
|
|---|
| 4462 | dtd.scaffIndex = (int*)REALLOC(dtd.scaffIndex, groupSize * sizeof(int));
|
|---|
| 4463 | }
|
|---|
| 4464 | else
|
|---|
| 4465 | groupConnector = (char*)MALLOC(groupSize = 32);
|
|---|
| 4466 | if (!groupConnector)
|
|---|
| 4467 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4468 | }
|
|---|
| 4469 | groupConnector[prologState.level] = 0;
|
|---|
| 4470 | if (dtd.in_eldecl)
|
|---|
| 4471 | {
|
|---|
| 4472 | int myindex = nextScaffoldPart(parser);
|
|---|
| 4473 |
|
|---|
| 4474 | if (myindex < 0)
|
|---|
| 4475 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4476 | dtd.scaffIndex[dtd.scaffLevel] = myindex;
|
|---|
| 4477 | dtd.scaffLevel++;
|
|---|
| 4478 | dtd.scaffold[myindex].type = XML_CTYPE_SEQ;
|
|---|
| 4479 | }
|
|---|
| 4480 | break;
|
|---|
| 4481 | case XML_ROLE_GROUP_SEQUENCE:
|
|---|
| 4482 | if (groupConnector[prologState.level] == '|')
|
|---|
| 4483 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 4484 | groupConnector[prologState.level] = ',';
|
|---|
| 4485 | break;
|
|---|
| 4486 | case XML_ROLE_GROUP_CHOICE:
|
|---|
| 4487 | if (groupConnector[prologState.level] == ',')
|
|---|
| 4488 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 4489 | if (dtd.in_eldecl
|
|---|
| 4490 | && !groupConnector[prologState.level]
|
|---|
| 4491 | && dtd.scaffold[dtd.scaffIndex[dtd.scaffLevel - 1]].type != XML_CTYPE_MIXED
|
|---|
| 4492 | )
|
|---|
| 4493 | {
|
|---|
| 4494 | dtd.scaffold[dtd.scaffIndex[dtd.scaffLevel - 1]].type = XML_CTYPE_CHOICE;
|
|---|
| 4495 | }
|
|---|
| 4496 | groupConnector[prologState.level] = '|';
|
|---|
| 4497 | break;
|
|---|
| 4498 | case XML_ROLE_PARAM_ENTITY_REF:
|
|---|
| 4499 | #ifdef XML_DTD
|
|---|
| 4500 | case XML_ROLE_INNER_PARAM_ENTITY_REF:
|
|---|
| 4501 | if (paramEntityParsing
|
|---|
| 4502 | && (dtd.complete || role == XML_ROLE_INNER_PARAM_ENTITY_REF))
|
|---|
| 4503 | {
|
|---|
| 4504 | const XML_Char *name;
|
|---|
| 4505 | ENTITY *entity;
|
|---|
| 4506 |
|
|---|
| 4507 | name = poolStoreString(&dtd.pool, enc,
|
|---|
| 4508 | s + enc->minBytesPerChar,
|
|---|
| 4509 | next - enc->minBytesPerChar,
|
|---|
| 4510 | NULL);
|
|---|
| 4511 | if (!name)
|
|---|
| 4512 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4513 | entity = (ENTITY *) lookup(&dtd.paramEntities, name, 0);
|
|---|
| 4514 | poolDiscard(&dtd.pool);
|
|---|
| 4515 | if (!entity)
|
|---|
| 4516 | {
|
|---|
| 4517 | /* FIXME what to do if !dtd.complete? */
|
|---|
| 4518 | return ERROR_EXPAT_UNDEFINED_ENTITY;
|
|---|
| 4519 | }
|
|---|
| 4520 | if (entity->open)
|
|---|
| 4521 | return ERROR_EXPAT_RECURSIVE_ENTITY_REF;
|
|---|
| 4522 | if (entity->textPtr)
|
|---|
| 4523 | {
|
|---|
| 4524 | XMLERROR result;
|
|---|
| 4525 |
|
|---|
| 4526 | result = processInternalParamEntity(parser, entity);
|
|---|
| 4527 | if (result != ERROR_EXPAT_NONE)
|
|---|
| 4528 | return result;
|
|---|
| 4529 | break;
|
|---|
| 4530 | }
|
|---|
| 4531 | if (role == XML_ROLE_INNER_PARAM_ENTITY_REF)
|
|---|
| 4532 | return ERROR_EXPAT_PARAM_ENTITY_REF;
|
|---|
| 4533 | if (externalEntityRefHandler)
|
|---|
| 4534 | {
|
|---|
| 4535 | dtd.complete = 0;
|
|---|
| 4536 | entity->open = 1;
|
|---|
| 4537 | if (!externalEntityRefHandler(handlerArg, // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 4538 | externalEntityRefHandlerArg,
|
|---|
| 4539 | 0,
|
|---|
| 4540 | entity->base,
|
|---|
| 4541 | entity->systemId,
|
|---|
| 4542 | entity->publicId))
|
|---|
| 4543 | {
|
|---|
| 4544 | entity->open = 0;
|
|---|
| 4545 | return ERROR_EXPAT_EXTERNAL_ENTITY_HANDLING;
|
|---|
| 4546 | }
|
|---|
| 4547 | entity->open = 0;
|
|---|
| 4548 | if (dtd.complete)
|
|---|
| 4549 | break;
|
|---|
| 4550 | }
|
|---|
| 4551 | }
|
|---|
| 4552 | #endif /* XML_DTD */
|
|---|
| 4553 | if (!dtd.standalone
|
|---|
| 4554 | && notStandaloneHandler
|
|---|
| 4555 | && !notStandaloneHandler(handlerArg))
|
|---|
| 4556 | return ERROR_EXPAT_NOT_STANDALONE;
|
|---|
| 4557 | dtd.complete = 0;
|
|---|
| 4558 | if (defaultHandler)
|
|---|
| 4559 | reportDefault(parser, enc, s, next);
|
|---|
| 4560 | break;
|
|---|
| 4561 |
|
|---|
| 4562 | /* Element declaration stuff */
|
|---|
| 4563 |
|
|---|
| 4564 | case XML_ROLE_ELEMENT_NAME:
|
|---|
| 4565 | if (elementDeclHandler)
|
|---|
| 4566 | {
|
|---|
| 4567 | declElementType = getElementType(parser, enc, s, next);
|
|---|
| 4568 | if (!declElementType)
|
|---|
| 4569 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4570 | dtd.scaffLevel = 0;
|
|---|
| 4571 | dtd.scaffCount = 0;
|
|---|
| 4572 | dtd.in_eldecl = 1;
|
|---|
| 4573 | }
|
|---|
| 4574 | break;
|
|---|
| 4575 |
|
|---|
| 4576 | case XML_ROLE_CONTENT_ANY:
|
|---|
| 4577 | case XML_ROLE_CONTENT_EMPTY:
|
|---|
| 4578 | if (dtd.in_eldecl)
|
|---|
| 4579 | {
|
|---|
| 4580 | if (elementDeclHandler)
|
|---|
| 4581 | {
|
|---|
| 4582 | XMLCONTENT *content = (XMLCONTENT *) MALLOC(sizeof(XMLCONTENT));
|
|---|
| 4583 |
|
|---|
| 4584 | if (!content)
|
|---|
| 4585 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4586 | content->quant = XML_CQUANT_NONE;
|
|---|
| 4587 | content->name = 0;
|
|---|
| 4588 | content->numchildren = 0;
|
|---|
| 4589 | content->children = 0;
|
|---|
| 4590 | content->type = ((role == XML_ROLE_CONTENT_ANY) ?
|
|---|
| 4591 | XML_CTYPE_ANY :
|
|---|
| 4592 | XML_CTYPE_EMPTY);
|
|---|
| 4593 | *eventEndPP = s;
|
|---|
| 4594 | elementDeclHandler(handlerArg, declElementType->name, content);
|
|---|
| 4595 | }
|
|---|
| 4596 | dtd.in_eldecl = 0;
|
|---|
| 4597 | }
|
|---|
| 4598 | break;
|
|---|
| 4599 |
|
|---|
| 4600 | case XML_ROLE_CONTENT_PCDATA:
|
|---|
| 4601 | if (dtd.in_eldecl)
|
|---|
| 4602 | {
|
|---|
| 4603 | dtd.scaffold[dtd.scaffIndex[dtd.scaffLevel - 1]].type = XML_CTYPE_MIXED;
|
|---|
| 4604 | }
|
|---|
| 4605 | break;
|
|---|
| 4606 |
|
|---|
| 4607 | case XML_ROLE_CONTENT_ELEMENT:
|
|---|
| 4608 | quant = XML_CQUANT_NONE;
|
|---|
| 4609 | goto elementContent;
|
|---|
| 4610 | case XML_ROLE_CONTENT_ELEMENT_OPT:
|
|---|
| 4611 | quant = XML_CQUANT_OPT;
|
|---|
| 4612 | goto elementContent;
|
|---|
| 4613 | case XML_ROLE_CONTENT_ELEMENT_REP:
|
|---|
| 4614 | quant = XML_CQUANT_REP;
|
|---|
| 4615 | goto elementContent;
|
|---|
| 4616 | case XML_ROLE_CONTENT_ELEMENT_PLUS:
|
|---|
| 4617 | quant = XML_CQUANT_PLUS;
|
|---|
| 4618 | elementContent:
|
|---|
| 4619 | if (dtd.in_eldecl)
|
|---|
| 4620 | {
|
|---|
| 4621 | ELEMENT_TYPE *el;
|
|---|
| 4622 | const char *nxt = quant == XML_CQUANT_NONE ? next : next - 1;
|
|---|
| 4623 | int myindex = nextScaffoldPart(parser);
|
|---|
| 4624 |
|
|---|
| 4625 | if (myindex < 0)
|
|---|
| 4626 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4627 | dtd.scaffold[myindex].type = XML_CTYPE_NAME;
|
|---|
| 4628 | dtd.scaffold[myindex].quant = quant;
|
|---|
| 4629 | el = getElementType(parser, enc, s, nxt);
|
|---|
| 4630 | if (!el)
|
|---|
| 4631 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4632 | dtd.scaffold[myindex].name = el->name;
|
|---|
| 4633 | dtd.contentStringLen += nxt - s + 1;
|
|---|
| 4634 | }
|
|---|
| 4635 | break;
|
|---|
| 4636 |
|
|---|
| 4637 | case XML_ROLE_GROUP_CLOSE:
|
|---|
| 4638 | quant = XML_CQUANT_NONE;
|
|---|
| 4639 | goto closeGroup;
|
|---|
| 4640 | case XML_ROLE_GROUP_CLOSE_OPT:
|
|---|
| 4641 | quant = XML_CQUANT_OPT;
|
|---|
| 4642 | goto closeGroup;
|
|---|
| 4643 | case XML_ROLE_GROUP_CLOSE_REP:
|
|---|
| 4644 | quant = XML_CQUANT_REP;
|
|---|
| 4645 | goto closeGroup;
|
|---|
| 4646 | case XML_ROLE_GROUP_CLOSE_PLUS:
|
|---|
| 4647 | quant = XML_CQUANT_PLUS;
|
|---|
| 4648 | closeGroup:
|
|---|
| 4649 | if (dtd.in_eldecl)
|
|---|
| 4650 | {
|
|---|
| 4651 | dtd.scaffLevel--;
|
|---|
| 4652 | dtd.scaffold[dtd.scaffIndex[dtd.scaffLevel]].quant = quant;
|
|---|
| 4653 | if (dtd.scaffLevel == 0)
|
|---|
| 4654 | {
|
|---|
| 4655 | if (elementDeclHandler)
|
|---|
| 4656 | {
|
|---|
| 4657 | XMLCONTENT *model = build_model(parser);
|
|---|
| 4658 |
|
|---|
| 4659 | if (!model)
|
|---|
| 4660 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4661 | *eventEndPP = s;
|
|---|
| 4662 | elementDeclHandler(handlerArg, declElementType->name, model);
|
|---|
| 4663 | }
|
|---|
| 4664 | dtd.in_eldecl = 0;
|
|---|
| 4665 | dtd.contentStringLen = 0;
|
|---|
| 4666 | }
|
|---|
| 4667 | }
|
|---|
| 4668 | break;
|
|---|
| 4669 | /* End element declaration stuff */
|
|---|
| 4670 |
|
|---|
| 4671 | case XML_ROLE_NONE:
|
|---|
| 4672 | switch (tok)
|
|---|
| 4673 | {
|
|---|
| 4674 | case XML_TOK_PI:
|
|---|
| 4675 | if (!reportProcessingInstruction(parser, enc, s, next))
|
|---|
| 4676 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4677 | break;
|
|---|
| 4678 | case XML_TOK_COMMENT:
|
|---|
| 4679 | if (!reportComment(parser, enc, s, next))
|
|---|
| 4680 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4681 | break;
|
|---|
| 4682 | }
|
|---|
| 4683 | break;
|
|---|
| 4684 | }
|
|---|
| 4685 | if (defaultHandler)
|
|---|
| 4686 | {
|
|---|
| 4687 | switch (tok)
|
|---|
| 4688 | {
|
|---|
| 4689 | case XML_TOK_PI:
|
|---|
| 4690 | case XML_TOK_COMMENT:
|
|---|
| 4691 | case XML_TOK_BOM:
|
|---|
| 4692 | case XML_TOK_XML_DECL:
|
|---|
| 4693 | #ifdef XML_DTD
|
|---|
| 4694 | case XML_TOK_IGNORE_SECT:
|
|---|
| 4695 | #endif /* XML_DTD */
|
|---|
| 4696 | case XML_TOK_PARAM_ENTITY_REF:
|
|---|
| 4697 | break;
|
|---|
| 4698 | default:
|
|---|
| 4699 | #ifdef XML_DTD
|
|---|
| 4700 | if (role != XML_ROLE_IGNORE_SECT)
|
|---|
| 4701 | #endif /* XML_DTD */
|
|---|
| 4702 | reportDefault(parser, enc, s, next);
|
|---|
| 4703 | }
|
|---|
| 4704 | }
|
|---|
| 4705 | s = next;
|
|---|
| 4706 | tok = XmlPrologTok(enc, s, end, &next);
|
|---|
| 4707 | }
|
|---|
| 4708 | /* not reached */
|
|---|
| 4709 | }
|
|---|
| 4710 |
|
|---|
| 4711 | STATIC XMLERROR epilogProcessor(XML_Parser parser,
|
|---|
| 4712 | const char *s,
|
|---|
| 4713 | const char *end,
|
|---|
| 4714 | const char **nextPtr)
|
|---|
| 4715 | {
|
|---|
| 4716 | processor = epilogProcessor;
|
|---|
| 4717 | eventPtr = s;
|
|---|
| 4718 | for (;;)
|
|---|
| 4719 | {
|
|---|
| 4720 | const char *next;
|
|---|
| 4721 | int tok = XmlPrologTok(encoding, s, end, &next);
|
|---|
| 4722 |
|
|---|
| 4723 | eventEndPtr = next;
|
|---|
| 4724 | switch (tok)
|
|---|
| 4725 | {
|
|---|
| 4726 | case -XML_TOK_PROLOG_S:
|
|---|
| 4727 | if (defaultHandler)
|
|---|
| 4728 | {
|
|---|
| 4729 | eventEndPtr = end;
|
|---|
| 4730 | reportDefault(parser, encoding, s, end);
|
|---|
| 4731 | }
|
|---|
| 4732 | /* fall through */
|
|---|
| 4733 | case XML_TOK_NONE:
|
|---|
| 4734 | if (nextPtr)
|
|---|
| 4735 | *nextPtr = end;
|
|---|
| 4736 | return ERROR_EXPAT_NONE;
|
|---|
| 4737 | case XML_TOK_PROLOG_S:
|
|---|
| 4738 | if (defaultHandler)
|
|---|
| 4739 | reportDefault(parser, encoding, s, next);
|
|---|
| 4740 | break;
|
|---|
| 4741 | case XML_TOK_PI:
|
|---|
| 4742 | if (!reportProcessingInstruction(parser, encoding, s, next))
|
|---|
| 4743 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4744 | break;
|
|---|
| 4745 | case XML_TOK_COMMENT:
|
|---|
| 4746 | if (!reportComment(parser, encoding, s, next))
|
|---|
| 4747 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4748 | break;
|
|---|
| 4749 | case XML_TOK_INVALID:
|
|---|
| 4750 | eventPtr = next;
|
|---|
| 4751 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 4752 | case XML_TOK_PARTIAL:
|
|---|
| 4753 | if (nextPtr)
|
|---|
| 4754 | {
|
|---|
| 4755 | *nextPtr = s;
|
|---|
| 4756 | return ERROR_EXPAT_NONE;
|
|---|
| 4757 | }
|
|---|
| 4758 | return ERROR_EXPAT_UNCLOSED_TOKEN;
|
|---|
| 4759 | case XML_TOK_PARTIAL_CHAR:
|
|---|
| 4760 | if (nextPtr)
|
|---|
| 4761 | {
|
|---|
| 4762 | *nextPtr = s;
|
|---|
| 4763 | return ERROR_EXPAT_NONE;
|
|---|
| 4764 | }
|
|---|
| 4765 | return ERROR_EXPAT_PARTIAL_CHAR;
|
|---|
| 4766 | default:
|
|---|
| 4767 | return ERROR_EXPAT_JUNK_AFTER_DOC_ELEMENT;
|
|---|
| 4768 | }
|
|---|
| 4769 | eventPtr = s = next;
|
|---|
| 4770 | }
|
|---|
| 4771 | }
|
|---|
| 4772 |
|
|---|
| 4773 | #ifdef XML_DTD
|
|---|
| 4774 |
|
|---|
| 4775 | STATIC XMLERROR processInternalParamEntity(XML_Parser parser,
|
|---|
| 4776 | ENTITY * entity)
|
|---|
| 4777 | {
|
|---|
| 4778 | const char *s, *end, *next;
|
|---|
| 4779 | int tok;
|
|---|
| 4780 | XMLERROR result;
|
|---|
| 4781 | OPEN_INTERNAL_ENTITY openEntity;
|
|---|
| 4782 |
|
|---|
| 4783 | entity->open = 1;
|
|---|
| 4784 | openEntity.next = openInternalEntities;
|
|---|
| 4785 | openInternalEntities = &openEntity;
|
|---|
| 4786 | openEntity.entity = entity;
|
|---|
| 4787 | openEntity.internalEventPtr = 0;
|
|---|
| 4788 | openEntity.internalEventEndPtr = 0;
|
|---|
| 4789 | s = (char *)entity->textPtr;
|
|---|
| 4790 | end = (char *)(entity->textPtr + entity->textLen);
|
|---|
| 4791 | tok = XmlPrologTok(internalEncoding, s, end, &next);
|
|---|
| 4792 | result = doProlog(parser, internalEncoding, s, end, tok, next, 0);
|
|---|
| 4793 | entity->open = 0;
|
|---|
| 4794 | openInternalEntities = openEntity.next;
|
|---|
| 4795 | return result;
|
|---|
| 4796 | }
|
|---|
| 4797 |
|
|---|
| 4798 | #endif /* XML_DTD */
|
|---|
| 4799 |
|
|---|
| 4800 | STATIC XMLERROR errorProcessor(XML_Parser parser,
|
|---|
| 4801 | const char *s,
|
|---|
| 4802 | const char *end,
|
|---|
| 4803 | const char **nextPtr)
|
|---|
| 4804 | {
|
|---|
| 4805 | return errorCode;
|
|---|
| 4806 | }
|
|---|
| 4807 |
|
|---|
| 4808 | STATIC XMLERROR storeAttributeValue(XML_Parser parser,
|
|---|
| 4809 | const ENCODING * enc,
|
|---|
| 4810 | int isCdata,
|
|---|
| 4811 | const char *ptr,
|
|---|
| 4812 | const char *end,
|
|---|
| 4813 | STRING_POOL * pool)
|
|---|
| 4814 | {
|
|---|
| 4815 | XMLERROR result = appendAttributeValue(parser, enc, isCdata, ptr, end, pool);
|
|---|
| 4816 |
|
|---|
| 4817 | if (result)
|
|---|
| 4818 | return result;
|
|---|
| 4819 | if (!isCdata && poolLength(pool) && poolLastChar(pool) == 0x20)
|
|---|
| 4820 | poolChop(pool);
|
|---|
| 4821 | if (!poolAppendChar(pool, XML_T('\0')))
|
|---|
| 4822 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4823 | return ERROR_EXPAT_NONE;
|
|---|
| 4824 | }
|
|---|
| 4825 |
|
|---|
| 4826 |
|
|---|
| 4827 | /*
|
|---|
| 4828 | *@@ appendAttributeValue:
|
|---|
| 4829 | *
|
|---|
| 4830 | *@@changed V0.9.14 (2001-08-09) [umoeller]: fixed ERROR_EXPAT_UNDEFINED_ENTITY with callback-defined encodings
|
|---|
| 4831 | */
|
|---|
| 4832 |
|
|---|
| 4833 | STATIC XMLERROR appendAttributeValue(XML_Parser parser,
|
|---|
| 4834 | const ENCODING * enc,
|
|---|
| 4835 | int isCdata,
|
|---|
| 4836 | const char *ptr,
|
|---|
| 4837 | const char *end,
|
|---|
| 4838 | STRING_POOL * pool)
|
|---|
| 4839 | {
|
|---|
| 4840 | for (;;)
|
|---|
| 4841 | {
|
|---|
| 4842 | const char *next;
|
|---|
| 4843 | int tok = XmlAttributeValueTok(enc, ptr, end, &next);
|
|---|
| 4844 |
|
|---|
| 4845 | switch (tok)
|
|---|
| 4846 | {
|
|---|
| 4847 | case XML_TOK_NONE:
|
|---|
| 4848 | return ERROR_EXPAT_NONE;
|
|---|
| 4849 | case XML_TOK_INVALID:
|
|---|
| 4850 | if (enc == encoding)
|
|---|
| 4851 | eventPtr = next;
|
|---|
| 4852 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 4853 | case XML_TOK_PARTIAL:
|
|---|
| 4854 | if (enc == encoding)
|
|---|
| 4855 | eventPtr = ptr;
|
|---|
| 4856 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 4857 | case XML_TOK_CHAR_REF:
|
|---|
| 4858 | {
|
|---|
| 4859 | XML_Char buf[XML_ENCODE_MAX];
|
|---|
| 4860 | int i;
|
|---|
| 4861 | int n = XmlCharRefNumber(enc, ptr);
|
|---|
| 4862 |
|
|---|
| 4863 | if (n < 0)
|
|---|
| 4864 | {
|
|---|
| 4865 | if (enc == encoding)
|
|---|
| 4866 | eventPtr = ptr;
|
|---|
| 4867 | return ERROR_EXPAT_BAD_CHAR_REF;
|
|---|
| 4868 | }
|
|---|
| 4869 | if (!isCdata
|
|---|
| 4870 | && n == 0x20 /* space */
|
|---|
| 4871 | && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
|
|---|
| 4872 | break;
|
|---|
| 4873 | n = XmlEncode(n, (ICHAR *) buf);
|
|---|
| 4874 | if (!n)
|
|---|
| 4875 | {
|
|---|
| 4876 | if (enc == encoding)
|
|---|
| 4877 | eventPtr = ptr;
|
|---|
| 4878 | return ERROR_EXPAT_BAD_CHAR_REF;
|
|---|
| 4879 | }
|
|---|
| 4880 | for (i = 0; i < n; i++)
|
|---|
| 4881 | {
|
|---|
| 4882 | if (!poolAppendChar(pool, buf[i]))
|
|---|
| 4883 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4884 | }
|
|---|
| 4885 | }
|
|---|
| 4886 | break;
|
|---|
| 4887 | case XML_TOK_DATA_CHARS:
|
|---|
| 4888 | if (!poolAppend(pool, enc, ptr, next, NULL))
|
|---|
| 4889 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4890 | break;
|
|---|
| 4891 | break;
|
|---|
| 4892 | case XML_TOK_TRAILING_CR:
|
|---|
| 4893 | next = ptr + enc->minBytesPerChar;
|
|---|
| 4894 | /* fall through */
|
|---|
| 4895 | case XML_TOK_ATTRIBUTE_VALUE_S:
|
|---|
| 4896 | case XML_TOK_DATA_NEWLINE:
|
|---|
| 4897 | if (!isCdata && (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
|
|---|
| 4898 | break;
|
|---|
| 4899 | if (!poolAppendChar(pool, 0x20))
|
|---|
| 4900 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4901 | break;
|
|---|
| 4902 | case XML_TOK_ENTITY_REF:
|
|---|
| 4903 | {
|
|---|
| 4904 | const XML_Char *name;
|
|---|
| 4905 | ENTITY *entity;
|
|---|
| 4906 | unsigned long ulOfs; // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 4907 | XML_Char ch = XmlPredefinedEntityName(enc,
|
|---|
| 4908 | ptr + enc->minBytesPerChar,
|
|---|
| 4909 | next - enc->minBytesPerChar);
|
|---|
| 4910 |
|
|---|
| 4911 | if (ch)
|
|---|
| 4912 | {
|
|---|
| 4913 | if (!poolAppendChar(pool, ch))
|
|---|
| 4914 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4915 | break;
|
|---|
| 4916 | }
|
|---|
| 4917 | name = poolStoreString(&temp2Pool, enc,
|
|---|
| 4918 | ptr + enc->minBytesPerChar,
|
|---|
| 4919 | next - enc->minBytesPerChar,
|
|---|
| 4920 | &ulOfs); // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 4921 | if (!name)
|
|---|
| 4922 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 4923 | entity = (ENTITY *) lookup(&dtd.generalEntities,
|
|---|
| 4924 | name + ulOfs, // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 4925 | 0);
|
|---|
| 4926 | poolDiscard(&temp2Pool);
|
|---|
| 4927 | if (!entity)
|
|---|
| 4928 | {
|
|---|
| 4929 | if (dtd.complete)
|
|---|
| 4930 | {
|
|---|
| 4931 | if (enc == encoding)
|
|---|
| 4932 | eventPtr = ptr;
|
|---|
| 4933 | return ERROR_EXPAT_UNDEFINED_ENTITY;
|
|---|
| 4934 | }
|
|---|
| 4935 | }
|
|---|
| 4936 | else if (entity->open)
|
|---|
| 4937 | {
|
|---|
| 4938 | if (enc == encoding)
|
|---|
| 4939 | eventPtr = ptr;
|
|---|
| 4940 | return ERROR_EXPAT_RECURSIVE_ENTITY_REF;
|
|---|
| 4941 | }
|
|---|
| 4942 | else if (entity->notation)
|
|---|
| 4943 | {
|
|---|
| 4944 | if (enc == encoding)
|
|---|
| 4945 | eventPtr = ptr;
|
|---|
| 4946 | return ERROR_EXPAT_BINARY_ENTITY_REF;
|
|---|
| 4947 | }
|
|---|
| 4948 | else if (!entity->textPtr)
|
|---|
| 4949 | {
|
|---|
| 4950 | if (enc == encoding)
|
|---|
| 4951 | eventPtr = ptr;
|
|---|
| 4952 | return ERROR_EXPAT_ATTRIBUTE_EXTERNAL_ENTITY_REF;
|
|---|
| 4953 | }
|
|---|
| 4954 | else
|
|---|
| 4955 | {
|
|---|
| 4956 | XMLERROR result;
|
|---|
| 4957 | const XML_Char *textEnd = entity->textPtr + entity->textLen;
|
|---|
| 4958 |
|
|---|
| 4959 | entity->open = 1;
|
|---|
| 4960 | result = appendAttributeValue(parser, internalEncoding, isCdata, (char *)entity->textPtr, (char *)textEnd, pool);
|
|---|
| 4961 | entity->open = 0;
|
|---|
| 4962 | if (result)
|
|---|
| 4963 | return result;
|
|---|
| 4964 | }
|
|---|
| 4965 | }
|
|---|
| 4966 | break;
|
|---|
| 4967 | default:
|
|---|
| 4968 | if (enc == encoding)
|
|---|
| 4969 | eventPtr = ptr;
|
|---|
| 4970 | return ERROR_EXPAT_UNEXPECTED_STATE;
|
|---|
| 4971 | }
|
|---|
| 4972 | ptr = next;
|
|---|
| 4973 | }
|
|---|
| 4974 | /* not reached */
|
|---|
| 4975 | }
|
|---|
| 4976 |
|
|---|
| 4977 | STATIC XMLERROR storeEntityValue(XML_Parser parser,
|
|---|
| 4978 | const ENCODING * enc,
|
|---|
| 4979 | const char *entityTextPtr,
|
|---|
| 4980 | const char *entityTextEnd)
|
|---|
| 4981 | {
|
|---|
| 4982 | STRING_POOL *pool = &(dtd.pool);
|
|---|
| 4983 |
|
|---|
| 4984 | for (;;)
|
|---|
| 4985 | {
|
|---|
| 4986 | const char *next;
|
|---|
| 4987 | int tok = XmlEntityValueTok(enc, entityTextPtr, entityTextEnd, &next);
|
|---|
| 4988 |
|
|---|
| 4989 | switch (tok)
|
|---|
| 4990 | {
|
|---|
| 4991 | case XML_TOK_PARAM_ENTITY_REF:
|
|---|
| 4992 | #ifdef XML_DTD
|
|---|
| 4993 | if (parentParser || enc != encoding)
|
|---|
| 4994 | {
|
|---|
| 4995 | XMLERROR result;
|
|---|
| 4996 | const XML_Char *name;
|
|---|
| 4997 | ENTITY *entity;
|
|---|
| 4998 |
|
|---|
| 4999 | name = poolStoreString(&tempPool, enc,
|
|---|
| 5000 | entityTextPtr + enc->minBytesPerChar,
|
|---|
| 5001 | next - enc->minBytesPerChar,
|
|---|
| 5002 | NULL);
|
|---|
| 5003 | if (!name)
|
|---|
| 5004 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 5005 | entity = (ENTITY *) lookup(&dtd.paramEntities, name, 0);
|
|---|
| 5006 | poolDiscard(&tempPool);
|
|---|
| 5007 | if (!entity)
|
|---|
| 5008 | {
|
|---|
| 5009 | if (enc == encoding)
|
|---|
| 5010 | eventPtr = entityTextPtr;
|
|---|
| 5011 | return ERROR_EXPAT_UNDEFINED_ENTITY;
|
|---|
| 5012 | }
|
|---|
| 5013 | if (entity->open)
|
|---|
| 5014 | {
|
|---|
| 5015 | if (enc == encoding)
|
|---|
| 5016 | eventPtr = entityTextPtr;
|
|---|
| 5017 | return ERROR_EXPAT_RECURSIVE_ENTITY_REF;
|
|---|
| 5018 | }
|
|---|
| 5019 | if (entity->systemId)
|
|---|
| 5020 | {
|
|---|
| 5021 | if (enc == encoding)
|
|---|
| 5022 | eventPtr = entityTextPtr;
|
|---|
| 5023 | return ERROR_EXPAT_PARAM_ENTITY_REF;
|
|---|
| 5024 | }
|
|---|
| 5025 | entity->open = 1;
|
|---|
| 5026 | result = storeEntityValue(parser,
|
|---|
| 5027 | internalEncoding,
|
|---|
| 5028 | (char *)entity->textPtr,
|
|---|
| 5029 | (char *)(entity->textPtr + entity->textLen));
|
|---|
| 5030 | entity->open = 0;
|
|---|
| 5031 | if (result)
|
|---|
| 5032 | return result;
|
|---|
| 5033 | break;
|
|---|
| 5034 | }
|
|---|
| 5035 | #endif /* XML_DTD */
|
|---|
| 5036 | eventPtr = entityTextPtr;
|
|---|
| 5037 | return ERROR_EXPAT_SYNTAX;
|
|---|
| 5038 | case XML_TOK_NONE:
|
|---|
| 5039 | return ERROR_EXPAT_NONE;
|
|---|
| 5040 | case XML_TOK_ENTITY_REF:
|
|---|
| 5041 | case XML_TOK_DATA_CHARS:
|
|---|
| 5042 | if (!poolAppend(pool, enc, entityTextPtr, next, NULL))
|
|---|
| 5043 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 5044 | break;
|
|---|
| 5045 | case XML_TOK_TRAILING_CR:
|
|---|
| 5046 | next = entityTextPtr + enc->minBytesPerChar;
|
|---|
| 5047 | /* fall through */
|
|---|
| 5048 | case XML_TOK_DATA_NEWLINE:
|
|---|
| 5049 | if (pool->end == pool->ptr && !poolGrow(pool))
|
|---|
| 5050 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 5051 | *(pool->ptr)++ = 0xA;
|
|---|
| 5052 | break;
|
|---|
| 5053 | case XML_TOK_CHAR_REF:
|
|---|
| 5054 | {
|
|---|
| 5055 | XML_Char buf[XML_ENCODE_MAX];
|
|---|
| 5056 | int i;
|
|---|
| 5057 | int n = XmlCharRefNumber(enc, entityTextPtr);
|
|---|
| 5058 |
|
|---|
| 5059 | if (n < 0)
|
|---|
| 5060 | {
|
|---|
| 5061 | if (enc == encoding)
|
|---|
| 5062 | eventPtr = entityTextPtr;
|
|---|
| 5063 | return ERROR_EXPAT_BAD_CHAR_REF;
|
|---|
| 5064 | }
|
|---|
| 5065 | n = XmlEncode(n, (ICHAR *) buf);
|
|---|
| 5066 | if (!n)
|
|---|
| 5067 | {
|
|---|
| 5068 | if (enc == encoding)
|
|---|
| 5069 | eventPtr = entityTextPtr;
|
|---|
| 5070 | return ERROR_EXPAT_BAD_CHAR_REF;
|
|---|
| 5071 | }
|
|---|
| 5072 | for (i = 0; i < n; i++)
|
|---|
| 5073 | {
|
|---|
| 5074 | if (pool->end == pool->ptr && !poolGrow(pool))
|
|---|
| 5075 | return ERROR_EXPAT_NO_MEMORY;
|
|---|
| 5076 | *(pool->ptr)++ = buf[i];
|
|---|
| 5077 | }
|
|---|
| 5078 | }
|
|---|
| 5079 | break;
|
|---|
| 5080 | case XML_TOK_PARTIAL:
|
|---|
| 5081 | if (enc == encoding)
|
|---|
| 5082 | eventPtr = entityTextPtr;
|
|---|
| 5083 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 5084 | case XML_TOK_INVALID:
|
|---|
| 5085 | if (enc == encoding)
|
|---|
| 5086 | eventPtr = next;
|
|---|
| 5087 | return ERROR_EXPAT_INVALID_TOKEN;
|
|---|
| 5088 | default:
|
|---|
| 5089 | if (enc == encoding)
|
|---|
| 5090 | eventPtr = entityTextPtr;
|
|---|
| 5091 | return ERROR_EXPAT_UNEXPECTED_STATE;
|
|---|
| 5092 | }
|
|---|
| 5093 | entityTextPtr = next;
|
|---|
| 5094 | }
|
|---|
| 5095 | /* not reached */
|
|---|
| 5096 | }
|
|---|
| 5097 |
|
|---|
| 5098 | STATIC void normalizeLines(XML_Char * s)
|
|---|
| 5099 | {
|
|---|
| 5100 | XML_Char *p;
|
|---|
| 5101 |
|
|---|
| 5102 | for (;; s++)
|
|---|
| 5103 | {
|
|---|
| 5104 | if (*s == XML_T('\0'))
|
|---|
| 5105 | return;
|
|---|
| 5106 | if (*s == 0xD)
|
|---|
| 5107 | break;
|
|---|
| 5108 | }
|
|---|
| 5109 | p = s;
|
|---|
| 5110 | do
|
|---|
| 5111 | {
|
|---|
| 5112 | if (*s == 0xD)
|
|---|
| 5113 | {
|
|---|
| 5114 | *p++ = 0xA;
|
|---|
| 5115 | if (*++s == 0xA)
|
|---|
| 5116 | s++;
|
|---|
| 5117 | }
|
|---|
| 5118 | else
|
|---|
| 5119 | *p++ = *s++;
|
|---|
| 5120 | }
|
|---|
| 5121 | while (*s);
|
|---|
| 5122 | *p = XML_T('\0');
|
|---|
| 5123 | }
|
|---|
| 5124 |
|
|---|
| 5125 | STATIC int reportProcessingInstruction(XML_Parser parser,
|
|---|
| 5126 | const ENCODING * enc,
|
|---|
| 5127 | const char *start,
|
|---|
| 5128 | const char *end)
|
|---|
| 5129 | {
|
|---|
| 5130 | const XML_Char *target;
|
|---|
| 5131 | XML_Char *data;
|
|---|
| 5132 | const char *tem;
|
|---|
| 5133 |
|
|---|
| 5134 | if (!processingInstructionHandler)
|
|---|
| 5135 | {
|
|---|
| 5136 | if (defaultHandler)
|
|---|
| 5137 | reportDefault(parser, enc, start, end);
|
|---|
| 5138 | return 1;
|
|---|
| 5139 | }
|
|---|
| 5140 | start += enc->minBytesPerChar * 2;
|
|---|
| 5141 | tem = start + XmlNameLength(enc, start);
|
|---|
| 5142 | target = poolStoreString(&tempPool, enc, start, tem, NULL);
|
|---|
| 5143 | if (!target)
|
|---|
| 5144 | return 0;
|
|---|
| 5145 | poolFinish(&tempPool);
|
|---|
| 5146 | data = poolStoreString(&tempPool, enc,
|
|---|
| 5147 | XmlSkipS(enc, tem),
|
|---|
| 5148 | end - enc->minBytesPerChar * 2,
|
|---|
| 5149 | NULL);
|
|---|
| 5150 | if (!data)
|
|---|
| 5151 | return 0;
|
|---|
| 5152 | normalizeLines(data);
|
|---|
| 5153 | processingInstructionHandler(handlerArg, target, data);
|
|---|
| 5154 | poolClear(&tempPool);
|
|---|
| 5155 | return 1;
|
|---|
| 5156 | }
|
|---|
| 5157 |
|
|---|
| 5158 | STATIC int reportComment(XML_Parser parser,
|
|---|
| 5159 | const ENCODING * enc,
|
|---|
| 5160 | const char *start,
|
|---|
| 5161 | const char *end)
|
|---|
| 5162 | {
|
|---|
| 5163 | XML_Char *data;
|
|---|
| 5164 |
|
|---|
| 5165 | if (!commentHandler)
|
|---|
| 5166 | {
|
|---|
| 5167 | if (defaultHandler)
|
|---|
| 5168 | reportDefault(parser, enc, start, end);
|
|---|
| 5169 | return 1;
|
|---|
| 5170 | }
|
|---|
| 5171 | data = poolStoreString(&tempPool,
|
|---|
| 5172 | enc,
|
|---|
| 5173 | start + enc->minBytesPerChar * 4,
|
|---|
| 5174 | end - enc->minBytesPerChar * 3,
|
|---|
| 5175 | NULL);
|
|---|
| 5176 | if (!data)
|
|---|
| 5177 | return 0;
|
|---|
| 5178 | normalizeLines(data);
|
|---|
| 5179 | commentHandler(handlerArg, data);
|
|---|
| 5180 | poolClear(&tempPool);
|
|---|
| 5181 | return 1;
|
|---|
| 5182 | }
|
|---|
| 5183 |
|
|---|
| 5184 | STATIC void reportDefault(XML_Parser parser,
|
|---|
| 5185 | const ENCODING * enc,
|
|---|
| 5186 | const char *s,
|
|---|
| 5187 | const char *end)
|
|---|
| 5188 | {
|
|---|
| 5189 | if (MUST_CONVERT(enc, s))
|
|---|
| 5190 | {
|
|---|
| 5191 | const char **eventPP;
|
|---|
| 5192 | const char **eventEndPP;
|
|---|
| 5193 |
|
|---|
| 5194 | if (enc == encoding)
|
|---|
| 5195 | {
|
|---|
| 5196 | eventPP = &eventPtr;
|
|---|
| 5197 | eventEndPP = &eventEndPtr;
|
|---|
| 5198 | }
|
|---|
| 5199 | else
|
|---|
| 5200 | {
|
|---|
| 5201 | eventPP = &(openInternalEntities->internalEventPtr);
|
|---|
| 5202 | eventEndPP = &(openInternalEntities->internalEventEndPtr);
|
|---|
| 5203 | }
|
|---|
| 5204 | do
|
|---|
| 5205 | {
|
|---|
| 5206 | ICHAR *dataPtr = (ICHAR *) dataBuf;
|
|---|
| 5207 |
|
|---|
| 5208 | XmlConvert(enc, &s, end, &dataPtr, (ICHAR *) dataBufEnd);
|
|---|
| 5209 | *eventEndPP = s;
|
|---|
| 5210 | defaultHandler(handlerArg, dataBuf, dataPtr - (ICHAR *) dataBuf);
|
|---|
| 5211 | *eventPP = s;
|
|---|
| 5212 | }
|
|---|
| 5213 | while (s != end);
|
|---|
| 5214 | }
|
|---|
| 5215 | else
|
|---|
| 5216 | defaultHandler(handlerArg, (XML_Char *) s, (XML_Char *) end - (XML_Char *) s);
|
|---|
| 5217 | }
|
|---|
| 5218 |
|
|---|
| 5219 |
|
|---|
| 5220 | STATIC int defineAttribute(ELEMENT_TYPE * type,
|
|---|
| 5221 | ATTRIBUTE_ID * attId,
|
|---|
| 5222 | int isCdata,
|
|---|
| 5223 | int isId,
|
|---|
| 5224 | const XML_Char * value,
|
|---|
| 5225 | XML_Parser parser)
|
|---|
| 5226 | {
|
|---|
| 5227 | DEFAULT_ATTRIBUTE *att;
|
|---|
| 5228 |
|
|---|
| 5229 | if (value || isId)
|
|---|
| 5230 | {
|
|---|
| 5231 | /* The handling of default attributes gets messed up if we have
|
|---|
| 5232 | * a default which duplicates a non-default. */
|
|---|
| 5233 | int i;
|
|---|
| 5234 |
|
|---|
| 5235 | for (i = 0; i < type->nDefaultAtts; i++)
|
|---|
| 5236 | if (attId == type->defaultAtts[i].id)
|
|---|
| 5237 | return 1;
|
|---|
| 5238 | if (isId && !type->idAtt && !attId->xmlns)
|
|---|
| 5239 | type->idAtt = attId;
|
|---|
| 5240 | }
|
|---|
| 5241 | if (type->nDefaultAtts == type->allocDefaultAtts)
|
|---|
| 5242 | {
|
|---|
| 5243 | if (type->allocDefaultAtts == 0)
|
|---|
| 5244 | {
|
|---|
| 5245 | type->allocDefaultAtts = 8;
|
|---|
| 5246 | type->defaultAtts = (DEFAULT_ATTRIBUTE*)MALLOC(type->allocDefaultAtts * sizeof(DEFAULT_ATTRIBUTE));
|
|---|
| 5247 | }
|
|---|
| 5248 | else
|
|---|
| 5249 | {
|
|---|
| 5250 | type->allocDefaultAtts *= 2;
|
|---|
| 5251 | type->defaultAtts = (DEFAULT_ATTRIBUTE*)REALLOC(type->defaultAtts,
|
|---|
| 5252 | type->allocDefaultAtts * sizeof(DEFAULT_ATTRIBUTE));
|
|---|
| 5253 | }
|
|---|
| 5254 | if (!type->defaultAtts)
|
|---|
| 5255 | return 0;
|
|---|
| 5256 | }
|
|---|
| 5257 | att = type->defaultAtts + type->nDefaultAtts;
|
|---|
| 5258 | att->id = attId;
|
|---|
| 5259 | att->value = value;
|
|---|
| 5260 | att->isCdata = isCdata;
|
|---|
| 5261 | if (!isCdata)
|
|---|
| 5262 | attId->maybeTokenized = 1;
|
|---|
| 5263 | type->nDefaultAtts += 1;
|
|---|
| 5264 | return 1;
|
|---|
| 5265 | }
|
|---|
| 5266 |
|
|---|
| 5267 | STATIC int setElementTypePrefix(XML_Parser parser,
|
|---|
| 5268 | ELEMENT_TYPE * elementType)
|
|---|
| 5269 | {
|
|---|
| 5270 | const XML_Char *name;
|
|---|
| 5271 |
|
|---|
| 5272 | for (name = elementType->name; *name; name++)
|
|---|
| 5273 | {
|
|---|
| 5274 | if (*name == XML_T(':'))
|
|---|
| 5275 | {
|
|---|
| 5276 | PREFIX *prefix;
|
|---|
| 5277 | const XML_Char *s;
|
|---|
| 5278 |
|
|---|
| 5279 | for (s = elementType->name; s != name; s++)
|
|---|
| 5280 | {
|
|---|
| 5281 | if (!poolAppendChar(&dtd.pool, *s))
|
|---|
| 5282 | return 0;
|
|---|
| 5283 | }
|
|---|
| 5284 | if (!poolAppendChar(&dtd.pool, XML_T('\0')))
|
|---|
| 5285 | return 0;
|
|---|
| 5286 | prefix = (PREFIX *) lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX));
|
|---|
| 5287 | if (!prefix)
|
|---|
| 5288 | return 0;
|
|---|
| 5289 | if (prefix->name == poolStart(&dtd.pool))
|
|---|
| 5290 | poolFinish(&dtd.pool);
|
|---|
| 5291 | else
|
|---|
| 5292 | poolDiscard(&dtd.pool);
|
|---|
| 5293 | elementType->prefix = prefix;
|
|---|
| 5294 |
|
|---|
| 5295 | }
|
|---|
| 5296 | }
|
|---|
| 5297 | return 1;
|
|---|
| 5298 | }
|
|---|
| 5299 |
|
|---|
| 5300 | STATIC ATTRIBUTE_ID * getAttributeId(XML_Parser parser,
|
|---|
| 5301 | const ENCODING * enc,
|
|---|
| 5302 | const char *start,
|
|---|
| 5303 | const char *end)
|
|---|
| 5304 | {
|
|---|
| 5305 | ATTRIBUTE_ID *id;
|
|---|
| 5306 | const XML_Char *name;
|
|---|
| 5307 |
|
|---|
| 5308 | if (!poolAppendChar(&dtd.pool, XML_T('\0')))
|
|---|
| 5309 | return 0;
|
|---|
| 5310 | name = poolStoreString(&dtd.pool, enc, start, end, NULL);
|
|---|
| 5311 | if (!name)
|
|---|
| 5312 | return 0;
|
|---|
| 5313 | ++name;
|
|---|
| 5314 | id = (ATTRIBUTE_ID *) lookup(&dtd.attributeIds, name, sizeof(ATTRIBUTE_ID));
|
|---|
| 5315 | if (!id)
|
|---|
| 5316 | return 0;
|
|---|
| 5317 | if (id->name != name)
|
|---|
| 5318 | poolDiscard(&dtd.pool);
|
|---|
| 5319 | else
|
|---|
| 5320 | {
|
|---|
| 5321 | poolFinish(&dtd.pool);
|
|---|
| 5322 | if (!ns)
|
|---|
| 5323 | ;
|
|---|
| 5324 | else if (name[0] == 'x'
|
|---|
| 5325 | && name[1] == 'm'
|
|---|
| 5326 | && name[2] == 'l'
|
|---|
| 5327 | && name[3] == 'n'
|
|---|
| 5328 | && name[4] == 's'
|
|---|
| 5329 | && (name[5] == XML_T('\0') || name[5] == XML_T(':')))
|
|---|
| 5330 | {
|
|---|
| 5331 | if (name[5] == '\0')
|
|---|
| 5332 | id->prefix = &dtd.defaultPrefix;
|
|---|
| 5333 | else
|
|---|
| 5334 | id->prefix = (PREFIX *) lookup(&dtd.prefixes, name + 6, sizeof(PREFIX));
|
|---|
| 5335 | id->xmlns = 1;
|
|---|
| 5336 | }
|
|---|
| 5337 | else
|
|---|
| 5338 | {
|
|---|
| 5339 | int i;
|
|---|
| 5340 |
|
|---|
| 5341 | for (i = 0; name[i]; i++)
|
|---|
| 5342 | {
|
|---|
| 5343 | if (name[i] == XML_T(':'))
|
|---|
| 5344 | {
|
|---|
| 5345 | int j;
|
|---|
| 5346 |
|
|---|
| 5347 | for (j = 0; j < i; j++)
|
|---|
| 5348 | {
|
|---|
| 5349 | if (!poolAppendChar(&dtd.pool, name[j]))
|
|---|
| 5350 | return 0;
|
|---|
| 5351 | }
|
|---|
| 5352 | if (!poolAppendChar(&dtd.pool, XML_T('\0')))
|
|---|
| 5353 | return 0;
|
|---|
| 5354 | id->prefix = (PREFIX *) lookup(&dtd.prefixes, poolStart(&dtd.pool), sizeof(PREFIX));
|
|---|
| 5355 | if (id->prefix->name == poolStart(&dtd.pool))
|
|---|
| 5356 | poolFinish(&dtd.pool);
|
|---|
| 5357 | else
|
|---|
| 5358 | poolDiscard(&dtd.pool);
|
|---|
| 5359 | break;
|
|---|
| 5360 | }
|
|---|
| 5361 | }
|
|---|
| 5362 | }
|
|---|
| 5363 | }
|
|---|
| 5364 | return id;
|
|---|
| 5365 | }
|
|---|
| 5366 |
|
|---|
| 5367 | #define CONTEXT_SEP XML_T('\f')
|
|---|
| 5368 |
|
|---|
| 5369 | STATIC const XML_Char *getContext(XML_Parser parser)
|
|---|
| 5370 | {
|
|---|
| 5371 | HASH_TABLE_ITER iter;
|
|---|
| 5372 | int needSep = 0;
|
|---|
| 5373 |
|
|---|
| 5374 | if (dtd.defaultPrefix.binding)
|
|---|
| 5375 | {
|
|---|
| 5376 | int i;
|
|---|
| 5377 | int len;
|
|---|
| 5378 |
|
|---|
| 5379 | if (!poolAppendChar(&tempPool, XML_T('=')))
|
|---|
| 5380 | return 0;
|
|---|
| 5381 | len = dtd.defaultPrefix.binding->uriLen;
|
|---|
| 5382 | if (namespaceSeparator != XML_T('\0'))
|
|---|
| 5383 | len--;
|
|---|
| 5384 | for (i = 0; i < len; i++)
|
|---|
| 5385 | if (!poolAppendChar(&tempPool, dtd.defaultPrefix.binding->uri[i]))
|
|---|
| 5386 | return 0;
|
|---|
| 5387 | needSep = 1;
|
|---|
| 5388 | }
|
|---|
| 5389 |
|
|---|
| 5390 | hashTableIterInit(&iter, &(dtd.prefixes));
|
|---|
| 5391 | for (;;)
|
|---|
| 5392 | {
|
|---|
| 5393 | int i;
|
|---|
| 5394 | int len;
|
|---|
| 5395 | const XML_Char *s;
|
|---|
| 5396 | PREFIX *prefix = (PREFIX *) hashTableIterNext(&iter);
|
|---|
| 5397 |
|
|---|
| 5398 | if (!prefix)
|
|---|
| 5399 | break;
|
|---|
| 5400 | if (!prefix->binding)
|
|---|
| 5401 | continue;
|
|---|
| 5402 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
|
|---|
| 5403 | return 0;
|
|---|
| 5404 | for (s = prefix->name; *s; s++)
|
|---|
| 5405 | if (!poolAppendChar(&tempPool, *s))
|
|---|
| 5406 | return 0;
|
|---|
| 5407 | if (!poolAppendChar(&tempPool, XML_T('=')))
|
|---|
| 5408 | return 0;
|
|---|
| 5409 | len = prefix->binding->uriLen;
|
|---|
| 5410 | if (namespaceSeparator != XML_T('\0'))
|
|---|
| 5411 | len--;
|
|---|
| 5412 | for (i = 0; i < len; i++)
|
|---|
| 5413 | if (!poolAppendChar(&tempPool, prefix->binding->uri[i]))
|
|---|
| 5414 | return 0;
|
|---|
| 5415 | needSep = 1;
|
|---|
| 5416 | }
|
|---|
| 5417 |
|
|---|
| 5418 |
|
|---|
| 5419 | hashTableIterInit(&iter, &(dtd.generalEntities));
|
|---|
| 5420 | for (;;)
|
|---|
| 5421 | {
|
|---|
| 5422 | const XML_Char *s;
|
|---|
| 5423 | ENTITY *e = (ENTITY *) hashTableIterNext(&iter);
|
|---|
| 5424 |
|
|---|
| 5425 | if (!e)
|
|---|
| 5426 | break;
|
|---|
| 5427 | if (!e->open)
|
|---|
| 5428 | continue;
|
|---|
| 5429 | if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
|
|---|
| 5430 | return 0;
|
|---|
| 5431 | for (s = e->name; *s; s++)
|
|---|
| 5432 | if (!poolAppendChar(&tempPool, *s))
|
|---|
| 5433 | return 0;
|
|---|
| 5434 | needSep = 1;
|
|---|
| 5435 | }
|
|---|
| 5436 |
|
|---|
| 5437 | if (!poolAppendChar(&tempPool, XML_T('\0')))
|
|---|
| 5438 | return 0;
|
|---|
| 5439 | return tempPool.start;
|
|---|
| 5440 | }
|
|---|
| 5441 |
|
|---|
| 5442 | STATIC int setContext(XML_Parser parser,
|
|---|
| 5443 | const XML_Char * context)
|
|---|
| 5444 | {
|
|---|
| 5445 | const XML_Char *s = context;
|
|---|
| 5446 |
|
|---|
| 5447 | while (*context != XML_T('\0'))
|
|---|
| 5448 | {
|
|---|
| 5449 | if (*s == CONTEXT_SEP || *s == XML_T('\0'))
|
|---|
| 5450 | {
|
|---|
| 5451 | ENTITY *e;
|
|---|
| 5452 |
|
|---|
| 5453 | if (!poolAppendChar(&tempPool, XML_T('\0')))
|
|---|
| 5454 | return 0;
|
|---|
| 5455 | e = (ENTITY *) lookup(&dtd.generalEntities, poolStart(&tempPool), 0);
|
|---|
| 5456 | if (e)
|
|---|
| 5457 | e->open = 1;
|
|---|
| 5458 | if (*s != XML_T('\0'))
|
|---|
| 5459 | s++;
|
|---|
| 5460 | context = s;
|
|---|
| 5461 | poolDiscard(&tempPool);
|
|---|
| 5462 | }
|
|---|
| 5463 | else if (*s == '=')
|
|---|
| 5464 | {
|
|---|
| 5465 | PREFIX *prefix;
|
|---|
| 5466 |
|
|---|
| 5467 | if (poolLength(&tempPool) == 0)
|
|---|
| 5468 | prefix = &dtd.defaultPrefix;
|
|---|
| 5469 | else
|
|---|
| 5470 | {
|
|---|
| 5471 | if (!poolAppendChar(&tempPool, XML_T('\0')))
|
|---|
| 5472 | return 0;
|
|---|
| 5473 | prefix = (PREFIX *) lookup(&dtd.prefixes, poolStart(&tempPool), sizeof(PREFIX));
|
|---|
| 5474 | if (!prefix)
|
|---|
| 5475 | return 0;
|
|---|
| 5476 | if (prefix->name == poolStart(&tempPool))
|
|---|
| 5477 | {
|
|---|
| 5478 | prefix->name = poolCopyString(&dtd.pool, prefix->name);
|
|---|
| 5479 | if (!prefix->name)
|
|---|
| 5480 | return 0;
|
|---|
| 5481 | }
|
|---|
| 5482 | poolDiscard(&tempPool);
|
|---|
| 5483 | }
|
|---|
| 5484 | for (context = s + 1; *context != CONTEXT_SEP && *context != XML_T('\0'); context++)
|
|---|
| 5485 | if (!poolAppendChar(&tempPool, *context))
|
|---|
| 5486 | return 0;
|
|---|
| 5487 | if (!poolAppendChar(&tempPool, XML_T('\0')))
|
|---|
| 5488 | return 0;
|
|---|
| 5489 | if (!addBinding(parser, prefix, 0, poolStart(&tempPool), &inheritedBindings))
|
|---|
| 5490 | return 0;
|
|---|
| 5491 | poolDiscard(&tempPool);
|
|---|
| 5492 | if (*context != XML_T('\0'))
|
|---|
| 5493 | ++context;
|
|---|
| 5494 | s = context;
|
|---|
| 5495 | }
|
|---|
| 5496 | else
|
|---|
| 5497 | {
|
|---|
| 5498 | if (!poolAppendChar(&tempPool, *s))
|
|---|
| 5499 | return 0;
|
|---|
| 5500 | s++;
|
|---|
| 5501 | }
|
|---|
| 5502 | }
|
|---|
| 5503 | return 1;
|
|---|
| 5504 | }
|
|---|
| 5505 |
|
|---|
| 5506 |
|
|---|
| 5507 | STATIC void normalizePublicId(XML_Char * publicId)
|
|---|
| 5508 | {
|
|---|
| 5509 | XML_Char *p = publicId;
|
|---|
| 5510 | XML_Char *s;
|
|---|
| 5511 |
|
|---|
| 5512 | for (s = publicId; *s; s++)
|
|---|
| 5513 | {
|
|---|
| 5514 | switch (*s)
|
|---|
| 5515 | {
|
|---|
| 5516 | case 0x20:
|
|---|
| 5517 | case 0xD:
|
|---|
| 5518 | case 0xA:
|
|---|
| 5519 | if (p != publicId && p[-1] != 0x20)
|
|---|
| 5520 | *p++ = 0x20;
|
|---|
| 5521 | break;
|
|---|
| 5522 | default:
|
|---|
| 5523 | *p++ = *s;
|
|---|
| 5524 | }
|
|---|
| 5525 | }
|
|---|
| 5526 | if (p != publicId && p[-1] == 0x20)
|
|---|
| 5527 | --p;
|
|---|
| 5528 | *p = XML_T('\0');
|
|---|
| 5529 | }
|
|---|
| 5530 |
|
|---|
| 5531 | STATIC int dtdInit(DTD * p,
|
|---|
| 5532 | XML_Parser parser)
|
|---|
| 5533 | {
|
|---|
| 5534 | XML_Memory_Handling_Suite *ms = &((Parser *) parser)->m_mem;
|
|---|
| 5535 |
|
|---|
| 5536 | poolInit(&(p->pool), ms);
|
|---|
| 5537 | hashTableInit(&(p->generalEntities), ms);
|
|---|
| 5538 | hashTableInit(&(p->elementTypes), ms);
|
|---|
| 5539 | hashTableInit(&(p->attributeIds), ms);
|
|---|
| 5540 | hashTableInit(&(p->prefixes), ms);
|
|---|
| 5541 | p->complete = 1;
|
|---|
| 5542 | p->standalone = 0;
|
|---|
| 5543 | #ifdef XML_DTD
|
|---|
| 5544 | hashTableInit(&(p->paramEntities), ms);
|
|---|
| 5545 | #endif /* XML_DTD */
|
|---|
| 5546 | p->defaultPrefix.name = 0;
|
|---|
| 5547 | p->defaultPrefix.binding = 0;
|
|---|
| 5548 |
|
|---|
| 5549 | p->in_eldecl = 0;
|
|---|
| 5550 | p->scaffIndex = 0;
|
|---|
| 5551 | p->scaffLevel = 0;
|
|---|
| 5552 | p->scaffold = 0;
|
|---|
| 5553 | p->contentStringLen = 0;
|
|---|
| 5554 | p->scaffSize = 0;
|
|---|
| 5555 | p->scaffCount = 0;
|
|---|
| 5556 |
|
|---|
| 5557 | return 1;
|
|---|
| 5558 | }
|
|---|
| 5559 |
|
|---|
| 5560 | #ifdef XML_DTD
|
|---|
| 5561 |
|
|---|
| 5562 | STATIC void dtdSwap(DTD * p1, DTD * p2)
|
|---|
| 5563 | {
|
|---|
| 5564 | DTD tem;
|
|---|
| 5565 |
|
|---|
| 5566 | memcpy(&tem, p1, sizeof(DTD));
|
|---|
| 5567 | memcpy(p1, p2, sizeof(DTD));
|
|---|
| 5568 | memcpy(p2, &tem, sizeof(DTD));
|
|---|
| 5569 | }
|
|---|
| 5570 |
|
|---|
| 5571 | #endif /* XML_DTD */
|
|---|
| 5572 |
|
|---|
| 5573 | STATIC void dtdDestroy(DTD * p, XML_Parser parser)
|
|---|
| 5574 | {
|
|---|
| 5575 | HASH_TABLE_ITER iter;
|
|---|
| 5576 |
|
|---|
| 5577 | hashTableIterInit(&iter, &(p->elementTypes));
|
|---|
| 5578 | for (;;)
|
|---|
| 5579 | {
|
|---|
| 5580 | ELEMENT_TYPE *e = (ELEMENT_TYPE *) hashTableIterNext(&iter);
|
|---|
| 5581 |
|
|---|
| 5582 | if (!e)
|
|---|
| 5583 | break;
|
|---|
| 5584 | if (e->allocDefaultAtts != 0)
|
|---|
| 5585 | FREE(e->defaultAtts);
|
|---|
| 5586 | }
|
|---|
| 5587 | hashTableDestroy(&(p->generalEntities));
|
|---|
| 5588 | #ifdef XML_DTD
|
|---|
| 5589 | hashTableDestroy(&(p->paramEntities));
|
|---|
| 5590 | #endif /* XML_DTD */
|
|---|
| 5591 | hashTableDestroy(&(p->elementTypes));
|
|---|
| 5592 | hashTableDestroy(&(p->attributeIds));
|
|---|
| 5593 | hashTableDestroy(&(p->prefixes));
|
|---|
| 5594 | poolDestroy(&(p->pool));
|
|---|
| 5595 | if (p->scaffIndex)
|
|---|
| 5596 | FREE(p->scaffIndex);
|
|---|
| 5597 | if (p->scaffold)
|
|---|
| 5598 | FREE(p->scaffold);
|
|---|
| 5599 | }
|
|---|
| 5600 |
|
|---|
| 5601 | /* Do a deep copy of the DTD. Return 0 for out of memory; non-zero otherwise.
|
|---|
| 5602 | * The new DTD has already been initialized. */
|
|---|
| 5603 |
|
|---|
| 5604 | STATIC int dtdCopy(DTD * newDtd, const DTD * oldDtd, XML_Parser parser)
|
|---|
| 5605 | {
|
|---|
| 5606 | HASH_TABLE_ITER iter;
|
|---|
| 5607 |
|
|---|
| 5608 | /* Copy the prefix table. */
|
|---|
| 5609 |
|
|---|
| 5610 | hashTableIterInit(&iter, &(oldDtd->prefixes));
|
|---|
| 5611 | for (;;)
|
|---|
| 5612 | {
|
|---|
| 5613 | const XML_Char *name;
|
|---|
| 5614 | const PREFIX *oldP = (PREFIX *) hashTableIterNext(&iter);
|
|---|
| 5615 |
|
|---|
| 5616 | if (!oldP)
|
|---|
| 5617 | break;
|
|---|
| 5618 | name = poolCopyString(&(newDtd->pool), oldP->name);
|
|---|
| 5619 | if (!name)
|
|---|
| 5620 | return 0;
|
|---|
| 5621 | if (!lookup(&(newDtd->prefixes), name, sizeof(PREFIX)))
|
|---|
| 5622 | return 0;
|
|---|
| 5623 | }
|
|---|
| 5624 |
|
|---|
| 5625 | hashTableIterInit(&iter, &(oldDtd->attributeIds));
|
|---|
| 5626 |
|
|---|
| 5627 | /* Copy the attribute id table. */
|
|---|
| 5628 |
|
|---|
| 5629 | for (;;)
|
|---|
| 5630 | {
|
|---|
| 5631 | ATTRIBUTE_ID *newA;
|
|---|
| 5632 | const XML_Char *name;
|
|---|
| 5633 | const ATTRIBUTE_ID *oldA = (ATTRIBUTE_ID *) hashTableIterNext(&iter);
|
|---|
| 5634 |
|
|---|
| 5635 | if (!oldA)
|
|---|
| 5636 | break;
|
|---|
| 5637 | /* Remember to allocate the scratch byte before the name. */
|
|---|
| 5638 | if (!poolAppendChar(&(newDtd->pool), XML_T('\0')))
|
|---|
| 5639 | return 0;
|
|---|
| 5640 | name = poolCopyString(&(newDtd->pool), oldA->name);
|
|---|
| 5641 | if (!name)
|
|---|
| 5642 | return 0;
|
|---|
| 5643 | ++name;
|
|---|
| 5644 | newA = (ATTRIBUTE_ID *) lookup(&(newDtd->attributeIds), name, sizeof(ATTRIBUTE_ID));
|
|---|
| 5645 | if (!newA)
|
|---|
| 5646 | return 0;
|
|---|
| 5647 | newA->maybeTokenized = oldA->maybeTokenized;
|
|---|
| 5648 | if (oldA->prefix)
|
|---|
| 5649 | {
|
|---|
| 5650 | newA->xmlns = oldA->xmlns;
|
|---|
| 5651 | if (oldA->prefix == &oldDtd->defaultPrefix)
|
|---|
| 5652 | newA->prefix = &newDtd->defaultPrefix;
|
|---|
| 5653 | else
|
|---|
| 5654 | newA->prefix = (PREFIX *) lookup(&(newDtd->prefixes), oldA->prefix->name, 0);
|
|---|
| 5655 | }
|
|---|
| 5656 | }
|
|---|
| 5657 |
|
|---|
| 5658 | /* Copy the element type table. */
|
|---|
| 5659 |
|
|---|
| 5660 | hashTableIterInit(&iter, &(oldDtd->elementTypes));
|
|---|
| 5661 |
|
|---|
| 5662 | for (;;)
|
|---|
| 5663 | {
|
|---|
| 5664 | int i;
|
|---|
| 5665 | ELEMENT_TYPE *newE;
|
|---|
| 5666 | const XML_Char *name;
|
|---|
| 5667 | const ELEMENT_TYPE *oldE = (ELEMENT_TYPE *) hashTableIterNext(&iter);
|
|---|
| 5668 |
|
|---|
| 5669 | if (!oldE)
|
|---|
| 5670 | break;
|
|---|
| 5671 | name = poolCopyString(&(newDtd->pool), oldE->name);
|
|---|
| 5672 | if (!name)
|
|---|
| 5673 | return 0;
|
|---|
| 5674 | newE = (ELEMENT_TYPE *) lookup(&(newDtd->elementTypes), name, sizeof(ELEMENT_TYPE));
|
|---|
| 5675 | if (!newE)
|
|---|
| 5676 | return 0;
|
|---|
| 5677 | if (oldE->nDefaultAtts)
|
|---|
| 5678 | {
|
|---|
| 5679 | newE->defaultAtts = (DEFAULT_ATTRIBUTE *) MALLOC(oldE->nDefaultAtts * sizeof(DEFAULT_ATTRIBUTE));
|
|---|
| 5680 | if (!newE->defaultAtts)
|
|---|
| 5681 | return 0;
|
|---|
| 5682 | }
|
|---|
| 5683 | if (oldE->idAtt)
|
|---|
| 5684 | newE->idAtt = (ATTRIBUTE_ID *) lookup(&(newDtd->attributeIds), oldE->idAtt->name, 0);
|
|---|
| 5685 | newE->allocDefaultAtts = newE->nDefaultAtts = oldE->nDefaultAtts;
|
|---|
| 5686 | if (oldE->prefix)
|
|---|
| 5687 | newE->prefix = (PREFIX *) lookup(&(newDtd->prefixes), oldE->prefix->name, 0);
|
|---|
| 5688 | for (i = 0; i < newE->nDefaultAtts; i++)
|
|---|
| 5689 | {
|
|---|
| 5690 | newE->defaultAtts[i].id = (ATTRIBUTE_ID *) lookup(&(newDtd->attributeIds), oldE->defaultAtts[i].id->name, 0);
|
|---|
| 5691 | newE->defaultAtts[i].isCdata = oldE->defaultAtts[i].isCdata;
|
|---|
| 5692 | if (oldE->defaultAtts[i].value)
|
|---|
| 5693 | {
|
|---|
| 5694 | newE->defaultAtts[i].value = poolCopyString(&(newDtd->pool), oldE->defaultAtts[i].value);
|
|---|
| 5695 | if (!newE->defaultAtts[i].value)
|
|---|
| 5696 | return 0;
|
|---|
| 5697 | }
|
|---|
| 5698 | else
|
|---|
| 5699 | newE->defaultAtts[i].value = 0;
|
|---|
| 5700 | }
|
|---|
| 5701 | }
|
|---|
| 5702 |
|
|---|
| 5703 | /* Copy the entity tables. */
|
|---|
| 5704 | if (!copyEntityTable(&(newDtd->generalEntities),
|
|---|
| 5705 | &(newDtd->pool),
|
|---|
| 5706 | &(oldDtd->generalEntities), parser))
|
|---|
| 5707 | return 0;
|
|---|
| 5708 |
|
|---|
| 5709 | #ifdef XML_DTD
|
|---|
| 5710 | if (!copyEntityTable(&(newDtd->paramEntities),
|
|---|
| 5711 | &(newDtd->pool),
|
|---|
| 5712 | &(oldDtd->paramEntities), parser))
|
|---|
| 5713 | return 0;
|
|---|
| 5714 | #endif /* XML_DTD */
|
|---|
| 5715 |
|
|---|
| 5716 | newDtd->complete = oldDtd->complete;
|
|---|
| 5717 | newDtd->standalone = oldDtd->standalone;
|
|---|
| 5718 |
|
|---|
| 5719 | /* Don't want deep copying for scaffolding */
|
|---|
| 5720 | newDtd->in_eldecl = oldDtd->in_eldecl;
|
|---|
| 5721 | newDtd->scaffold = oldDtd->scaffold;
|
|---|
| 5722 | newDtd->contentStringLen = oldDtd->contentStringLen;
|
|---|
| 5723 | newDtd->scaffSize = oldDtd->scaffSize;
|
|---|
| 5724 | newDtd->scaffLevel = oldDtd->scaffLevel;
|
|---|
| 5725 | newDtd->scaffIndex = oldDtd->scaffIndex;
|
|---|
| 5726 |
|
|---|
| 5727 | return 1;
|
|---|
| 5728 | } /* End dtdCopy */
|
|---|
| 5729 |
|
|---|
| 5730 | STATIC int copyEntityTable(HASH_TABLE * newTable,
|
|---|
| 5731 | STRING_POOL * newPool,
|
|---|
| 5732 | const HASH_TABLE * oldTable,
|
|---|
| 5733 | XML_Parser parser)
|
|---|
| 5734 | {
|
|---|
| 5735 | HASH_TABLE_ITER iter;
|
|---|
| 5736 | const XML_Char *cachedOldBase = 0;
|
|---|
| 5737 | const XML_Char *cachedNewBase = 0;
|
|---|
| 5738 |
|
|---|
| 5739 | hashTableIterInit(&iter, oldTable);
|
|---|
| 5740 |
|
|---|
| 5741 | for (;;)
|
|---|
| 5742 | {
|
|---|
| 5743 | ENTITY *newE;
|
|---|
| 5744 | const XML_Char *name;
|
|---|
| 5745 | const ENTITY *oldE = (ENTITY *) hashTableIterNext(&iter);
|
|---|
| 5746 |
|
|---|
| 5747 | if (!oldE)
|
|---|
| 5748 | break;
|
|---|
| 5749 | name = poolCopyString(newPool, oldE->name);
|
|---|
| 5750 | if (!name)
|
|---|
| 5751 | return 0;
|
|---|
| 5752 | newE = (ENTITY *) lookup(newTable, name, sizeof(ENTITY));
|
|---|
| 5753 | if (!newE)
|
|---|
| 5754 | return 0;
|
|---|
| 5755 | if (oldE->systemId)
|
|---|
| 5756 | {
|
|---|
| 5757 | const XML_Char *tem = poolCopyString(newPool, oldE->systemId);
|
|---|
| 5758 |
|
|---|
| 5759 | if (!tem)
|
|---|
| 5760 | return 0;
|
|---|
| 5761 | newE->systemId = tem;
|
|---|
| 5762 | if (oldE->base)
|
|---|
| 5763 | {
|
|---|
| 5764 | if (oldE->base == cachedOldBase)
|
|---|
| 5765 | newE->base = cachedNewBase;
|
|---|
| 5766 | else
|
|---|
| 5767 | {
|
|---|
| 5768 | cachedOldBase = oldE->base;
|
|---|
| 5769 | tem = poolCopyString(newPool, cachedOldBase);
|
|---|
| 5770 | if (!tem)
|
|---|
| 5771 | return 0;
|
|---|
| 5772 | cachedNewBase = newE->base = tem;
|
|---|
| 5773 | }
|
|---|
| 5774 | }
|
|---|
| 5775 | }
|
|---|
| 5776 | else
|
|---|
| 5777 | {
|
|---|
| 5778 | const XML_Char *tem = poolCopyStringN(newPool, oldE->textPtr, oldE->textLen);
|
|---|
| 5779 |
|
|---|
| 5780 | if (!tem)
|
|---|
| 5781 | return 0;
|
|---|
| 5782 | newE->textPtr = tem;
|
|---|
| 5783 | newE->textLen = oldE->textLen;
|
|---|
| 5784 | }
|
|---|
| 5785 | if (oldE->notation)
|
|---|
| 5786 | {
|
|---|
| 5787 | const XML_Char *tem = poolCopyString(newPool, oldE->notation);
|
|---|
| 5788 |
|
|---|
| 5789 | if (!tem)
|
|---|
| 5790 | return 0;
|
|---|
| 5791 | newE->notation = tem;
|
|---|
| 5792 | }
|
|---|
| 5793 | }
|
|---|
| 5794 | return 1;
|
|---|
| 5795 | }
|
|---|
| 5796 |
|
|---|
| 5797 | #define INIT_SIZE 64
|
|---|
| 5798 |
|
|---|
| 5799 | STATIC int keyeq(KEY s1, KEY s2)
|
|---|
| 5800 | {
|
|---|
| 5801 | for (; *s1 == *s2; s1++, s2++)
|
|---|
| 5802 | if (*s1 == 0)
|
|---|
| 5803 | return 1;
|
|---|
| 5804 | return 0;
|
|---|
| 5805 | }
|
|---|
| 5806 |
|
|---|
| 5807 | STATIC unsigned long hash(KEY s)
|
|---|
| 5808 | {
|
|---|
| 5809 | unsigned long h = 0;
|
|---|
| 5810 |
|
|---|
| 5811 | while (*s)
|
|---|
| 5812 | h = (h << 5) + h + (unsigned char)*s++;
|
|---|
| 5813 | return h;
|
|---|
| 5814 | }
|
|---|
| 5815 |
|
|---|
| 5816 | STATIC NAMED *lookup(HASH_TABLE * table,
|
|---|
| 5817 | KEY name,
|
|---|
| 5818 | size_t createSize)
|
|---|
| 5819 | {
|
|---|
| 5820 | size_t i;
|
|---|
| 5821 |
|
|---|
| 5822 | if (table->size == 0)
|
|---|
| 5823 | {
|
|---|
| 5824 | size_t tsize;
|
|---|
| 5825 |
|
|---|
| 5826 | if (!createSize)
|
|---|
| 5827 | return 0;
|
|---|
| 5828 | tsize = INIT_SIZE * sizeof(NAMED *);
|
|---|
| 5829 | table->v = (NAMED**)table->mem->malloc_fcn(tsize);
|
|---|
| 5830 | if (!table->v)
|
|---|
| 5831 | return 0;
|
|---|
| 5832 | memset(table->v, 0, tsize);
|
|---|
| 5833 | table->size = INIT_SIZE;
|
|---|
| 5834 | table->usedLim = INIT_SIZE / 2;
|
|---|
| 5835 | i = hash(name) & (table->size - 1);
|
|---|
| 5836 | }
|
|---|
| 5837 | else
|
|---|
| 5838 | {
|
|---|
| 5839 | unsigned long h = hash(name);
|
|---|
| 5840 |
|
|---|
| 5841 | for (i = h & (table->size - 1);
|
|---|
| 5842 | table->v[i];
|
|---|
| 5843 | i == 0 ? i = table->size - 1 : --i)
|
|---|
| 5844 | {
|
|---|
| 5845 | if (keyeq(name, table->v[i]->name))
|
|---|
| 5846 | return table->v[i];
|
|---|
| 5847 | }
|
|---|
| 5848 | if (!createSize)
|
|---|
| 5849 | return 0;
|
|---|
| 5850 | if (table->used == table->usedLim)
|
|---|
| 5851 | {
|
|---|
| 5852 | /* check for overflow */
|
|---|
| 5853 | size_t newSize = table->size * 2;
|
|---|
| 5854 | size_t tsize = newSize * sizeof(NAMED *);
|
|---|
| 5855 | NAMED **newV = (NAMED**)table->mem->malloc_fcn(tsize);
|
|---|
| 5856 |
|
|---|
| 5857 | if (!newV)
|
|---|
| 5858 | return 0;
|
|---|
| 5859 | memset(newV, 0, tsize);
|
|---|
| 5860 | for (i = 0; i < table->size; i++)
|
|---|
| 5861 | if (table->v[i])
|
|---|
| 5862 | {
|
|---|
| 5863 | size_t j;
|
|---|
| 5864 |
|
|---|
| 5865 | for (j = hash(table->v[i]->name) & (newSize - 1);
|
|---|
| 5866 | newV[j];
|
|---|
| 5867 | j == 0 ? j = newSize - 1 : --j)
|
|---|
| 5868 | ;
|
|---|
| 5869 | newV[j] = table->v[i];
|
|---|
| 5870 | }
|
|---|
| 5871 | table->mem->free_fcn(table->v);
|
|---|
| 5872 | table->v = newV;
|
|---|
| 5873 | table->size = newSize;
|
|---|
| 5874 | table->usedLim = newSize / 2;
|
|---|
| 5875 | for (i = h & (table->size - 1);
|
|---|
| 5876 | table->v[i];
|
|---|
| 5877 | i == 0 ? i = table->size - 1 : --i)
|
|---|
| 5878 | ;
|
|---|
| 5879 | }
|
|---|
| 5880 | }
|
|---|
| 5881 | table->v[i] = (NAMED*)table->mem->malloc_fcn(createSize);
|
|---|
| 5882 | if (!table->v[i])
|
|---|
| 5883 | return 0;
|
|---|
| 5884 | memset(table->v[i], 0, createSize);
|
|---|
| 5885 | table->v[i]->name = name;
|
|---|
| 5886 | (table->used)++;
|
|---|
| 5887 | return table->v[i];
|
|---|
| 5888 | }
|
|---|
| 5889 |
|
|---|
| 5890 | STATIC void hashTableDestroy(HASH_TABLE * table)
|
|---|
| 5891 | {
|
|---|
| 5892 | size_t i;
|
|---|
| 5893 |
|
|---|
| 5894 | for (i = 0; i < table->size; i++)
|
|---|
| 5895 | {
|
|---|
| 5896 | NAMED *p = table->v[i];
|
|---|
| 5897 |
|
|---|
| 5898 | if (p)
|
|---|
| 5899 | table->mem->free_fcn(p);
|
|---|
| 5900 | }
|
|---|
| 5901 | if (table->v)
|
|---|
| 5902 | table->mem->free_fcn(table->v);
|
|---|
| 5903 | }
|
|---|
| 5904 |
|
|---|
| 5905 | STATIC void hashTableInit(HASH_TABLE * p, XML_Memory_Handling_Suite * ms)
|
|---|
| 5906 | {
|
|---|
| 5907 | p->size = 0;
|
|---|
| 5908 | p->usedLim = 0;
|
|---|
| 5909 | p->used = 0;
|
|---|
| 5910 | p->v = 0;
|
|---|
| 5911 | p->mem = ms;
|
|---|
| 5912 | }
|
|---|
| 5913 |
|
|---|
| 5914 | STATIC void hashTableIterInit(HASH_TABLE_ITER * iter, const HASH_TABLE * table)
|
|---|
| 5915 | {
|
|---|
| 5916 | iter->p = table->v;
|
|---|
| 5917 | iter->end = iter->p + table->size;
|
|---|
| 5918 | }
|
|---|
| 5919 |
|
|---|
| 5920 | STATIC NAMED *hashTableIterNext(HASH_TABLE_ITER * iter)
|
|---|
| 5921 | {
|
|---|
| 5922 | while (iter->p != iter->end)
|
|---|
| 5923 | {
|
|---|
| 5924 | NAMED *tem = *(iter->p)++;
|
|---|
| 5925 |
|
|---|
| 5926 | if (tem)
|
|---|
| 5927 | return tem;
|
|---|
| 5928 | }
|
|---|
| 5929 | return 0;
|
|---|
| 5930 | }
|
|---|
| 5931 |
|
|---|
| 5932 |
|
|---|
| 5933 | STATIC void poolInit(STRING_POOL * pool, XML_Memory_Handling_Suite * ms)
|
|---|
| 5934 | {
|
|---|
| 5935 | pool->blocks = 0;
|
|---|
| 5936 | pool->freeBlocks = 0;
|
|---|
| 5937 | pool->start = 0;
|
|---|
| 5938 | pool->ptr = 0;
|
|---|
| 5939 | pool->end = 0;
|
|---|
| 5940 | pool->mem = ms;
|
|---|
| 5941 | }
|
|---|
| 5942 |
|
|---|
| 5943 | STATIC void poolClear(STRING_POOL * pool)
|
|---|
| 5944 | {
|
|---|
| 5945 | if (!pool->freeBlocks)
|
|---|
| 5946 | pool->freeBlocks = pool->blocks;
|
|---|
| 5947 | else
|
|---|
| 5948 | {
|
|---|
| 5949 | BLOCK *p = pool->blocks;
|
|---|
| 5950 |
|
|---|
| 5951 | while (p)
|
|---|
| 5952 | {
|
|---|
| 5953 | BLOCK *tem = p->next;
|
|---|
| 5954 |
|
|---|
| 5955 | p->next = pool->freeBlocks;
|
|---|
| 5956 | pool->freeBlocks = p;
|
|---|
| 5957 | p = tem;
|
|---|
| 5958 | }
|
|---|
| 5959 | }
|
|---|
| 5960 | pool->blocks = 0;
|
|---|
| 5961 | pool->start = 0;
|
|---|
| 5962 | pool->ptr = 0;
|
|---|
| 5963 | pool->end = 0;
|
|---|
| 5964 | }
|
|---|
| 5965 |
|
|---|
| 5966 | STATIC void poolDestroy(STRING_POOL * pool)
|
|---|
| 5967 | {
|
|---|
| 5968 | BLOCK *p = pool->blocks;
|
|---|
| 5969 |
|
|---|
| 5970 | while (p)
|
|---|
| 5971 | {
|
|---|
| 5972 | BLOCK *tem = p->next;
|
|---|
| 5973 |
|
|---|
| 5974 | pool->mem->free_fcn(p);
|
|---|
| 5975 | p = tem;
|
|---|
| 5976 | }
|
|---|
| 5977 | pool->blocks = 0;
|
|---|
| 5978 | p = pool->freeBlocks;
|
|---|
| 5979 | while (p)
|
|---|
| 5980 | {
|
|---|
| 5981 | BLOCK *tem = p->next;
|
|---|
| 5982 |
|
|---|
| 5983 | pool->mem->free_fcn(p);
|
|---|
| 5984 | p = tem;
|
|---|
| 5985 | }
|
|---|
| 5986 | pool->freeBlocks = 0;
|
|---|
| 5987 | pool->ptr = 0;
|
|---|
| 5988 | pool->start = 0;
|
|---|
| 5989 | pool->end = 0;
|
|---|
| 5990 | }
|
|---|
| 5991 |
|
|---|
| 5992 | /*
|
|---|
| 5993 | *@@ poolAppend:
|
|---|
| 5994 | * appends a new string to a pool.
|
|---|
| 5995 | * Returns pool->start if the pool is valid,
|
|---|
| 5996 | * or NULL if storing failed.
|
|---|
| 5997 | *
|
|---|
| 5998 | * V0.9.14: I added the pulOfs parameter which, after
|
|---|
| 5999 | * the string was stored, receives the offset from
|
|---|
| 6000 | * pool->start at which the string was stored in the
|
|---|
| 6001 | * pool. I am not quite sure how all this works, but
|
|---|
| 6002 | * if a user-defined encoding is in place (via
|
|---|
| 6003 | * XML_SetUnknownEncodingHandler), the first entry
|
|---|
| 6004 | * in the pool is always the encoding name from the
|
|---|
| 6005 | * xml header, and all entity references failed here
|
|---|
| 6006 | * because the lookup would always be done for
|
|---|
| 6007 | * the encoding name instead of the entity name
|
|---|
| 6008 | * (which then is the second entry in the pool).
|
|---|
| 6009 | * Whatever.
|
|---|
| 6010 | *
|
|---|
| 6011 | *@@changed V0.9.14 (2001-08-09) [umoeller]: added pulOfs param
|
|---|
| 6012 | */
|
|---|
| 6013 |
|
|---|
| 6014 | STATIC XML_Char* poolAppend(STRING_POOL * pool,
|
|---|
| 6015 | const ENCODING * enc,
|
|---|
| 6016 | const char *ptr,
|
|---|
| 6017 | const char *end,
|
|---|
| 6018 | unsigned long *pulOfs) // out: offset of beginning of stored
|
|---|
| 6019 | // string from pool->start
|
|---|
| 6020 | // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 6021 | {
|
|---|
| 6022 | if (!pool->ptr && !poolGrow(pool))
|
|---|
| 6023 | return 0;
|
|---|
| 6024 | for (;;)
|
|---|
| 6025 | {
|
|---|
| 6026 | char *pOldStart = pool->ptr;
|
|---|
| 6027 | XmlConvert(enc,
|
|---|
| 6028 | &ptr, // fromP
|
|---|
| 6029 | end,
|
|---|
| 6030 | (ICHAR**)&(pool->ptr), // toP
|
|---|
| 6031 | (ICHAR *)pool->end);
|
|---|
| 6032 | // expands to:
|
|---|
| 6033 | // (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
|
|---|
| 6034 | // -- for cp850, we end up in
|
|---|
| 6035 | // -- for Latin1, we end up in
|
|---|
| 6036 | if (pulOfs)
|
|---|
| 6037 | *pulOfs = pOldStart - pool->start; // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 6038 |
|
|---|
| 6039 | if (ptr == end)
|
|---|
| 6040 | break;
|
|---|
| 6041 | if (!poolGrow(pool))
|
|---|
| 6042 | return 0;
|
|---|
| 6043 | }
|
|---|
| 6044 | return pool->start;
|
|---|
| 6045 | }
|
|---|
| 6046 |
|
|---|
| 6047 | STATIC const XML_Char* poolCopyString(STRING_POOL *pool,
|
|---|
| 6048 | const XML_Char *s)
|
|---|
| 6049 | {
|
|---|
| 6050 | do
|
|---|
| 6051 | {
|
|---|
| 6052 | if (!poolAppendChar(pool, *s))
|
|---|
| 6053 | return 0;
|
|---|
| 6054 | }
|
|---|
| 6055 | while (*s++);
|
|---|
| 6056 | s = pool->start;
|
|---|
| 6057 | poolFinish(pool);
|
|---|
| 6058 | return s;
|
|---|
| 6059 | }
|
|---|
| 6060 |
|
|---|
| 6061 | STATIC const XML_Char *poolCopyStringN(STRING_POOL * pool, const XML_Char * s, int n)
|
|---|
| 6062 | {
|
|---|
| 6063 | if (!pool->ptr && !poolGrow(pool))
|
|---|
| 6064 | return 0;
|
|---|
| 6065 | for (; n > 0; --n, s++)
|
|---|
| 6066 | {
|
|---|
| 6067 | if (!poolAppendChar(pool, *s))
|
|---|
| 6068 | return 0;
|
|---|
| 6069 |
|
|---|
| 6070 | }
|
|---|
| 6071 | s = pool->start;
|
|---|
| 6072 | poolFinish(pool);
|
|---|
| 6073 | return s;
|
|---|
| 6074 | }
|
|---|
| 6075 |
|
|---|
| 6076 | STATIC const XML_Char* poolAppendString(STRING_POOL *pool,
|
|---|
| 6077 | const XML_Char *s)
|
|---|
| 6078 | {
|
|---|
| 6079 | while (*s)
|
|---|
| 6080 | {
|
|---|
| 6081 | if (!poolAppendChar(pool, *s))
|
|---|
| 6082 | return 0;
|
|---|
| 6083 | s++;
|
|---|
| 6084 | }
|
|---|
| 6085 | return pool->start;
|
|---|
| 6086 | } /* End poolAppendString */
|
|---|
| 6087 |
|
|---|
| 6088 | /*
|
|---|
| 6089 | *@@ poolStoreString:
|
|---|
| 6090 | *
|
|---|
| 6091 | *@@changed V0.9.14 (2001-08-09) [umoeller]: added pulOfs param
|
|---|
| 6092 | */
|
|---|
| 6093 |
|
|---|
| 6094 | STATIC XML_Char* poolStoreString(STRING_POOL *pool,
|
|---|
| 6095 | const ENCODING *enc,
|
|---|
| 6096 | const char *ptr,
|
|---|
| 6097 | const char *end,
|
|---|
| 6098 | unsigned long *pulOfs) // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 6099 | {
|
|---|
| 6100 | if (!poolAppend(pool,
|
|---|
| 6101 | enc,
|
|---|
| 6102 | ptr,
|
|---|
| 6103 | end,
|
|---|
| 6104 | pulOfs)) // V0.9.14 (2001-08-09) [umoeller]
|
|---|
| 6105 | return 0;
|
|---|
| 6106 | if ( (pool->ptr == pool->end)
|
|---|
| 6107 | && (!poolGrow(pool))
|
|---|
| 6108 | )
|
|---|
| 6109 | return 0;
|
|---|
| 6110 | *(pool->ptr)++ = 0;
|
|---|
| 6111 | return pool->start;
|
|---|
| 6112 | }
|
|---|
| 6113 |
|
|---|
| 6114 | STATIC int poolGrow(STRING_POOL *pool)
|
|---|
| 6115 | {
|
|---|
| 6116 | if (pool->freeBlocks)
|
|---|
| 6117 | {
|
|---|
| 6118 | if (pool->start == 0)
|
|---|
| 6119 | {
|
|---|
| 6120 | pool->blocks = pool->freeBlocks;
|
|---|
| 6121 | pool->freeBlocks = pool->freeBlocks->next;
|
|---|
| 6122 | pool->blocks->next = 0;
|
|---|
| 6123 | pool->start = pool->blocks->s;
|
|---|
| 6124 | pool->end = pool->start + pool->blocks->size;
|
|---|
| 6125 | pool->ptr = pool->start;
|
|---|
| 6126 | return 1;
|
|---|
| 6127 | }
|
|---|
| 6128 | if (pool->end - pool->start < pool->freeBlocks->size)
|
|---|
| 6129 | {
|
|---|
| 6130 | BLOCK *tem = pool->freeBlocks->next;
|
|---|
| 6131 |
|
|---|
| 6132 | pool->freeBlocks->next = pool->blocks;
|
|---|
| 6133 | pool->blocks = pool->freeBlocks;
|
|---|
| 6134 | pool->freeBlocks = tem;
|
|---|
| 6135 | memcpy(pool->blocks->s,
|
|---|
| 6136 | pool->start,
|
|---|
| 6137 | (pool->end - pool->start) * sizeof(XML_Char));
|
|---|
| 6138 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start);
|
|---|
| 6139 | pool->start = pool->blocks->s;
|
|---|
| 6140 | pool->end = pool->start + pool->blocks->size;
|
|---|
| 6141 | return 1;
|
|---|
| 6142 | }
|
|---|
| 6143 | }
|
|---|
| 6144 | if (pool->blocks && pool->start == pool->blocks->s)
|
|---|
| 6145 | {
|
|---|
| 6146 | int blockSize = (pool->end - pool->start) * 2;
|
|---|
| 6147 |
|
|---|
| 6148 | pool->blocks = (BLOCK*)pool->mem->realloc_fcn(pool->blocks, offsetof(BLOCK, s) + blockSize * sizeof(XML_Char));
|
|---|
| 6149 | if (!pool->blocks)
|
|---|
| 6150 | return 0;
|
|---|
| 6151 | pool->blocks->size = blockSize;
|
|---|
| 6152 | pool->ptr = pool->blocks->s + (pool->ptr - pool->start);
|
|---|
| 6153 | pool->start = pool->blocks->s;
|
|---|
| 6154 | pool->end = pool->start + blockSize;
|
|---|
| 6155 | }
|
|---|
| 6156 | else
|
|---|
| 6157 | {
|
|---|
| 6158 | BLOCK *tem;
|
|---|
| 6159 | int blockSize = pool->end - pool->start;
|
|---|
| 6160 |
|
|---|
| 6161 | if (blockSize < INIT_BLOCK_SIZE)
|
|---|
| 6162 | blockSize = INIT_BLOCK_SIZE;
|
|---|
| 6163 | else
|
|---|
| 6164 | blockSize *= 2;
|
|---|
| 6165 | tem = (BLOCK*)pool->mem->malloc_fcn(offsetof(BLOCK, s) + blockSize * sizeof(XML_Char));
|
|---|
| 6166 | if (!tem)
|
|---|
| 6167 | return 0;
|
|---|
| 6168 | tem->size = blockSize;
|
|---|
| 6169 | tem->next = pool->blocks;
|
|---|
| 6170 | pool->blocks = tem;
|
|---|
| 6171 | if (pool->ptr != pool->start)
|
|---|
| 6172 | memcpy(tem->s, pool->start, (pool->ptr - pool->start) * sizeof(XML_Char));
|
|---|
| 6173 | pool->ptr = tem->s + (pool->ptr - pool->start);
|
|---|
| 6174 | pool->start = tem->s;
|
|---|
| 6175 | pool->end = tem->s + blockSize;
|
|---|
| 6176 | }
|
|---|
| 6177 | return 1;
|
|---|
| 6178 | }
|
|---|
| 6179 |
|
|---|
| 6180 | STATIC int nextScaffoldPart(XML_Parser parser)
|
|---|
| 6181 | {
|
|---|
| 6182 | CONTENT_SCAFFOLD *me;
|
|---|
| 6183 | int next;
|
|---|
| 6184 |
|
|---|
| 6185 | if (!dtd.scaffIndex)
|
|---|
| 6186 | {
|
|---|
| 6187 | dtd.scaffIndex = (int*)MALLOC(groupSize * sizeof(int));
|
|---|
| 6188 |
|
|---|
| 6189 | if (!dtd.scaffIndex)
|
|---|
| 6190 | return -1;
|
|---|
| 6191 | dtd.scaffIndex[0] = 0;
|
|---|
| 6192 | }
|
|---|
| 6193 |
|
|---|
| 6194 | if (dtd.scaffCount >= dtd.scaffSize)
|
|---|
| 6195 | {
|
|---|
| 6196 | if (dtd.scaffold)
|
|---|
| 6197 | {
|
|---|
| 6198 | dtd.scaffSize *= 2;
|
|---|
| 6199 | dtd.scaffold = (CONTENT_SCAFFOLD *) REALLOC(dtd.scaffold,
|
|---|
| 6200 | dtd.scaffSize * sizeof(CONTENT_SCAFFOLD));
|
|---|
| 6201 | }
|
|---|
| 6202 | else
|
|---|
| 6203 | {
|
|---|
| 6204 | dtd.scaffSize = 32;
|
|---|
| 6205 | dtd.scaffold = (CONTENT_SCAFFOLD *) MALLOC(dtd.scaffSize * sizeof(CONTENT_SCAFFOLD));
|
|---|
| 6206 | }
|
|---|
| 6207 | if (!dtd.scaffold)
|
|---|
| 6208 | return -1;
|
|---|
| 6209 | }
|
|---|
| 6210 | next = dtd.scaffCount++;
|
|---|
| 6211 | me = &dtd.scaffold[next];
|
|---|
| 6212 | if (dtd.scaffLevel)
|
|---|
| 6213 | {
|
|---|
| 6214 | CONTENT_SCAFFOLD *parent = &dtd.scaffold[dtd.scaffIndex[dtd.scaffLevel - 1]];
|
|---|
| 6215 |
|
|---|
| 6216 | if (parent->lastchild)
|
|---|
| 6217 | {
|
|---|
| 6218 | dtd.scaffold[parent->lastchild].nextsib = next;
|
|---|
| 6219 | }
|
|---|
| 6220 | if (!parent->childcnt)
|
|---|
| 6221 | parent->firstchild = next;
|
|---|
| 6222 | parent->lastchild = next;
|
|---|
| 6223 | parent->childcnt++;
|
|---|
| 6224 | }
|
|---|
| 6225 | me->firstchild = me->lastchild = me->childcnt = me->nextsib = 0;
|
|---|
| 6226 | return next;
|
|---|
| 6227 | } /* End nextScaffoldPart */
|
|---|
| 6228 |
|
|---|
| 6229 | STATIC void build_node(XML_Parser parser,
|
|---|
| 6230 | int src_node,
|
|---|
| 6231 | XMLCONTENT * dest,
|
|---|
| 6232 | XMLCONTENT ** contpos,
|
|---|
| 6233 | char **strpos)
|
|---|
| 6234 | {
|
|---|
| 6235 | dest->type = dtd.scaffold[src_node].type;
|
|---|
| 6236 | dest->quant = dtd.scaffold[src_node].quant;
|
|---|
| 6237 | if (dest->type == XML_CTYPE_NAME)
|
|---|
| 6238 | {
|
|---|
| 6239 | const char *src;
|
|---|
| 6240 |
|
|---|
| 6241 | dest->name = *strpos;
|
|---|
| 6242 | src = dtd.scaffold[src_node].name;
|
|---|
| 6243 | for (;;)
|
|---|
| 6244 | {
|
|---|
| 6245 | *(*strpos)++ = *src;
|
|---|
| 6246 | if (!*src)
|
|---|
| 6247 | break;
|
|---|
| 6248 | src++;
|
|---|
| 6249 | }
|
|---|
| 6250 | dest->numchildren = 0;
|
|---|
| 6251 | dest->children = 0;
|
|---|
| 6252 | }
|
|---|
| 6253 | else
|
|---|
| 6254 | {
|
|---|
| 6255 | unsigned int i;
|
|---|
| 6256 | int cn;
|
|---|
| 6257 |
|
|---|
| 6258 | dest->numchildren = dtd.scaffold[src_node].childcnt;
|
|---|
| 6259 | dest->children = *contpos;
|
|---|
| 6260 | *contpos += dest->numchildren;
|
|---|
| 6261 | for (i = 0, cn = dtd.scaffold[src_node].firstchild;
|
|---|
| 6262 | i < dest->numchildren;
|
|---|
| 6263 | i++, cn = dtd.scaffold[cn].nextsib)
|
|---|
| 6264 | {
|
|---|
| 6265 | build_node(parser, cn, &(dest->children[i]), contpos, strpos);
|
|---|
| 6266 | }
|
|---|
| 6267 | dest->name = 0;
|
|---|
| 6268 | }
|
|---|
| 6269 | } /* End build_node */
|
|---|
| 6270 |
|
|---|
| 6271 | STATIC XMLCONTENT * build_model(XML_Parser parser)
|
|---|
| 6272 | {
|
|---|
| 6273 | XMLCONTENT *ret;
|
|---|
| 6274 | XMLCONTENT *cpos;
|
|---|
| 6275 | char *str;
|
|---|
| 6276 | int allocsize = dtd.scaffCount * sizeof(XMLCONTENT) + dtd.contentStringLen;
|
|---|
| 6277 |
|
|---|
| 6278 | ret = (XMLCONTENT*)MALLOC(allocsize);
|
|---|
| 6279 | if (!ret)
|
|---|
| 6280 | return 0;
|
|---|
| 6281 |
|
|---|
| 6282 | str = (char *)(&ret[dtd.scaffCount]);
|
|---|
| 6283 | cpos = &ret[1];
|
|---|
| 6284 |
|
|---|
| 6285 | build_node(parser, 0, ret, &cpos, &str);
|
|---|
| 6286 | return ret;
|
|---|
| 6287 | } /* End build_model */
|
|---|
| 6288 |
|
|---|
| 6289 | STATIC ELEMENT_TYPE * getElementType(XML_Parser parser,
|
|---|
| 6290 | const ENCODING * enc,
|
|---|
| 6291 | const char *ptr,
|
|---|
| 6292 | const char *end)
|
|---|
| 6293 | {
|
|---|
| 6294 | const XML_Char *name = poolStoreString(&dtd.pool, enc, ptr, end, NULL);
|
|---|
| 6295 | ELEMENT_TYPE *ret;
|
|---|
| 6296 |
|
|---|
| 6297 | if (!name)
|
|---|
| 6298 | return 0;
|
|---|
| 6299 | ret = (ELEMENT_TYPE *) lookup(&dtd.elementTypes, name, sizeof(ELEMENT_TYPE));
|
|---|
| 6300 | if (!ret)
|
|---|
| 6301 | return 0;
|
|---|
| 6302 | if (ret->name != name)
|
|---|
| 6303 | poolDiscard(&dtd.pool);
|
|---|
| 6304 | else
|
|---|
| 6305 | {
|
|---|
| 6306 | poolFinish(&dtd.pool);
|
|---|
| 6307 | if (!setElementTypePrefix(parser, ret))
|
|---|
| 6308 | return 0;
|
|---|
| 6309 | }
|
|---|
| 6310 | return ret;
|
|---|
| 6311 | } /* End getElementType */
|
|---|