Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1 | /* |
| 2 | * parser.c : an XML 1.0 parser, namespaces and validity support are mostly |
| 3 | * implemented on top of the SAX interfaces |
| 4 | * |
| 5 | * References: |
| 6 | * The XML specification: |
| 7 | * http://www.w3.org/TR/REC-xml |
| 8 | * Original 1.0 version: |
| 9 | * http://www.w3.org/TR/1998/REC-xml-19980210 |
| 10 | * XML second edition working draft |
| 11 | * http://www.w3.org/TR/2000/WD-xml-2e-20000814 |
| 12 | * |
| 13 | * Okay this is a big file, the parser core is around 7000 lines, then it |
| 14 | * is followed by the progressive parser top routines, then the various |
| 15 | * high level APIs to call the parser and a few miscelaneous functions. |
| 16 | * A number of helper functions and deprecated ones have been moved to |
| 17 | * parserInternals.c to reduce this file size. |
| 18 | * As much as possible the functions are associated with their relative |
| 19 | * production in the XML specification. A few productions defining the |
| 20 | * different ranges of character are actually implanted either in |
| 21 | * parserInternals.h or parserInternals.c |
| 22 | * The DOM tree build is realized from the default SAX callbacks in |
| 23 | * the module SAX.c. |
| 24 | * The routines doing the validation checks are in valid.c and called either |
| 25 | * from the SAx callbacks or as standalones functions using a preparsed |
| 26 | * document. |
| 27 | * |
| 28 | * See Copyright for the status of this software. |
| 29 | * |
| 30 | * Daniel.Veillard@w3.org |
| 31 | * |
| 32 | * 14 Nov 2000 ht - truncated definitions of xmlSubstituteEntitiesDefaultValue |
| 33 | * and xmlDoValidityCheckingDefaultValue for VMS |
| 34 | */ |
| 35 | |
Bjorn Reese | 70a9da5 | 2001-04-21 16:57:29 +0000 | [diff] [blame] | 36 | #include "libxml.h" |
| 37 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 38 | #ifdef WIN32 |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 39 | #define XML_DIR_SEP '\\' |
| 40 | #else |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 41 | #define XML_DIR_SEP '/' |
| 42 | #endif |
| 43 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 44 | #include <stdlib.h> |
| 45 | #include <string.h> |
| 46 | #include <libxml/xmlmemory.h> |
| 47 | #include <libxml/tree.h> |
| 48 | #include <libxml/parser.h> |
| 49 | #include <libxml/parserInternals.h> |
| 50 | #include <libxml/valid.h> |
| 51 | #include <libxml/entities.h> |
| 52 | #include <libxml/xmlerror.h> |
| 53 | #include <libxml/encoding.h> |
| 54 | #include <libxml/xmlIO.h> |
| 55 | #include <libxml/uri.h> |
| 56 | |
| 57 | #ifdef HAVE_CTYPE_H |
| 58 | #include <ctype.h> |
| 59 | #endif |
| 60 | #ifdef HAVE_STDLIB_H |
| 61 | #include <stdlib.h> |
| 62 | #endif |
| 63 | #ifdef HAVE_SYS_STAT_H |
| 64 | #include <sys/stat.h> |
| 65 | #endif |
| 66 | #ifdef HAVE_FCNTL_H |
| 67 | #include <fcntl.h> |
| 68 | #endif |
| 69 | #ifdef HAVE_UNISTD_H |
| 70 | #include <unistd.h> |
| 71 | #endif |
| 72 | #ifdef HAVE_ZLIB_H |
| 73 | #include <zlib.h> |
| 74 | #endif |
| 75 | |
| 76 | |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 77 | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 78 | #define XML_PARSER_BUFFER_SIZE 100 |
| 79 | |
| 80 | /* |
| 81 | * Various global defaults for parsing |
| 82 | */ |
| 83 | int xmlGetWarningsDefaultValue = 1; |
| 84 | int xmlParserDebugEntities = 0; |
| 85 | #ifdef VMS |
| 86 | int xmlSubstituteEntitiesDefaultVal = 0; |
| 87 | #define xmlSubstituteEntitiesDefaultValue xmlSubstituteEntitiesDefaultVal |
| 88 | int xmlDoValidityCheckingDefaultVal = 0; |
| 89 | #define xmlDoValidityCheckingDefaultValue xmlDoValidityCheckingDefaultVal |
| 90 | #else |
| 91 | int xmlSubstituteEntitiesDefaultValue = 0; |
| 92 | int xmlDoValidityCheckingDefaultValue = 0; |
| 93 | #endif |
| 94 | int xmlLoadExtDtdDefaultValue = 0; |
| 95 | int xmlPedanticParserDefaultValue = 0; |
| 96 | int xmlKeepBlanksDefaultValue = 1; |
| 97 | |
| 98 | /* |
| 99 | * List of XML prefixed PI allowed by W3C specs |
| 100 | */ |
| 101 | |
| 102 | const char *xmlW3CPIs[] = { |
| 103 | "xml-stylesheet", |
| 104 | NULL |
| 105 | }; |
| 106 | |
| 107 | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
| 108 | void xmlParserHandlePEReference(xmlParserCtxtPtr ctxt); |
| 109 | xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
| 110 | const xmlChar **str); |
| 111 | |
| 112 | |
| 113 | /************************************************************************ |
| 114 | * * |
| 115 | * Parser stacks related functions and macros * |
| 116 | * * |
| 117 | ************************************************************************/ |
| 118 | |
| 119 | xmlEntityPtr xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, |
| 120 | const xmlChar ** str); |
| 121 | |
| 122 | /* |
| 123 | * Generic function for accessing stacks in the Parser Context |
| 124 | */ |
| 125 | |
| 126 | #define PUSH_AND_POP(scope, type, name) \ |
| 127 | scope int name##Push(xmlParserCtxtPtr ctxt, type value) { \ |
| 128 | if (ctxt->name##Nr >= ctxt->name##Max) { \ |
| 129 | ctxt->name##Max *= 2; \ |
| 130 | ctxt->name##Tab = (type *) xmlRealloc(ctxt->name##Tab, \ |
| 131 | ctxt->name##Max * sizeof(ctxt->name##Tab[0])); \ |
| 132 | if (ctxt->name##Tab == NULL) { \ |
| 133 | xmlGenericError(xmlGenericErrorContext, \ |
| 134 | "realloc failed !\n"); \ |
| 135 | return(0); \ |
| 136 | } \ |
| 137 | } \ |
| 138 | ctxt->name##Tab[ctxt->name##Nr] = value; \ |
| 139 | ctxt->name = value; \ |
| 140 | return(ctxt->name##Nr++); \ |
| 141 | } \ |
| 142 | scope type name##Pop(xmlParserCtxtPtr ctxt) { \ |
| 143 | type ret; \ |
| 144 | if (ctxt->name##Nr <= 0) return(0); \ |
| 145 | ctxt->name##Nr--; \ |
| 146 | if (ctxt->name##Nr > 0) \ |
| 147 | ctxt->name = ctxt->name##Tab[ctxt->name##Nr - 1]; \ |
| 148 | else \ |
| 149 | ctxt->name = NULL; \ |
| 150 | ret = ctxt->name##Tab[ctxt->name##Nr]; \ |
| 151 | ctxt->name##Tab[ctxt->name##Nr] = 0; \ |
| 152 | return(ret); \ |
| 153 | } \ |
| 154 | |
| 155 | /* |
| 156 | * Those macros actually generate the functions |
| 157 | */ |
| 158 | PUSH_AND_POP(extern, xmlParserInputPtr, input) |
| 159 | PUSH_AND_POP(extern, xmlNodePtr, node) |
| 160 | PUSH_AND_POP(extern, xmlChar*, name) |
| 161 | |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 162 | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 163 | if (ctxt->spaceNr >= ctxt->spaceMax) { |
| 164 | ctxt->spaceMax *= 2; |
| 165 | ctxt->spaceTab = (int *) xmlRealloc(ctxt->spaceTab, |
| 166 | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
| 167 | if (ctxt->spaceTab == NULL) { |
| 168 | xmlGenericError(xmlGenericErrorContext, |
| 169 | "realloc failed !\n"); |
| 170 | return(0); |
| 171 | } |
| 172 | } |
| 173 | ctxt->spaceTab[ctxt->spaceNr] = val; |
| 174 | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
| 175 | return(ctxt->spaceNr++); |
| 176 | } |
| 177 | |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 178 | static int spacePop(xmlParserCtxtPtr ctxt) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 179 | int ret; |
| 180 | if (ctxt->spaceNr <= 0) return(0); |
| 181 | ctxt->spaceNr--; |
| 182 | if (ctxt->spaceNr > 0) |
| 183 | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
| 184 | else |
| 185 | ctxt->space = NULL; |
| 186 | ret = ctxt->spaceTab[ctxt->spaceNr]; |
| 187 | ctxt->spaceTab[ctxt->spaceNr] = -1; |
| 188 | return(ret); |
| 189 | } |
| 190 | |
| 191 | /* |
| 192 | * Macros for accessing the content. Those should be used only by the parser, |
| 193 | * and not exported. |
| 194 | * |
| 195 | * Dirty macros, i.e. one often need to make assumption on the context to |
| 196 | * use them |
| 197 | * |
| 198 | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
| 199 | * To be used with extreme caution since operations consuming |
| 200 | * characters may move the input buffer to a different location ! |
| 201 | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
| 202 | * This should be used internally by the parser |
| 203 | * only to compare to ASCII values otherwise it would break when |
| 204 | * running with UTF-8 encoding. |
| 205 | * RAW same as CUR but in the input buffer, bypass any token |
| 206 | * extraction that may have been done |
| 207 | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
| 208 | * to compare on ASCII based substring. |
| 209 | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
| 210 | * strings within the parser. |
| 211 | * |
| 212 | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
| 213 | * |
| 214 | * NEXT Skip to the next character, this does the proper decoding |
| 215 | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
| 216 | * NEXTL(l) Skip l xmlChars in the input buffer |
| 217 | * CUR_CHAR(l) returns the current unicode character (int), set l |
| 218 | * to the number of xmlChars used for the encoding [0-5]. |
| 219 | * CUR_SCHAR same but operate on a string instead of the context |
| 220 | * COPY_BUF copy the current unicode char to the target buffer, increment |
| 221 | * the index |
| 222 | * GROW, SHRINK handling of input buffers |
| 223 | */ |
| 224 | |
| 225 | #define RAW (ctxt->token ? -1 : (*ctxt->input->cur)) |
| 226 | #define CUR (ctxt->token ? ctxt->token : (*ctxt->input->cur)) |
| 227 | #define NXT(val) ctxt->input->cur[(val)] |
| 228 | #define CUR_PTR ctxt->input->cur |
| 229 | |
| 230 | #define SKIP(val) do { \ |
| 231 | ctxt->nbChars += (val),ctxt->input->cur += (val); \ |
| 232 | if (*ctxt->input->cur == '%') xmlParserHandlePEReference(ctxt); \ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 233 | if ((*ctxt->input->cur == 0) && \ |
| 234 | (xmlParserInputGrow(ctxt->input, INPUT_CHUNK) <= 0)) \ |
| 235 | xmlPopInput(ctxt); \ |
| 236 | } while (0) |
| 237 | |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 238 | #define SHRINK if (ctxt->input->cur - ctxt->input->base > INPUT_CHUNK) {\ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 239 | xmlParserInputShrink(ctxt->input); \ |
| 240 | if ((*ctxt->input->cur == 0) && \ |
| 241 | (xmlParserInputGrow(ctxt->input, INPUT_CHUNK) <= 0)) \ |
| 242 | xmlPopInput(ctxt); \ |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 243 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 244 | |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 245 | #define GROW if (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK) { \ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 246 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
| 247 | if ((*ctxt->input->cur == 0) && \ |
| 248 | (xmlParserInputGrow(ctxt->input, INPUT_CHUNK) <= 0)) \ |
| 249 | xmlPopInput(ctxt); \ |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 250 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 251 | |
| 252 | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
| 253 | |
| 254 | #define NEXT xmlNextChar(ctxt) |
| 255 | |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 256 | #define NEXT1 { \ |
| 257 | ctxt->input->cur++; \ |
| 258 | ctxt->nbChars++; \ |
| 259 | if (*ctxt->input->cur == 0) \ |
| 260 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
| 261 | } |
| 262 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 263 | #define NEXTL(l) do { \ |
| 264 | if (*(ctxt->input->cur) == '\n') { \ |
| 265 | ctxt->input->line++; ctxt->input->col = 1; \ |
| 266 | } else ctxt->input->col++; \ |
| 267 | ctxt->token = 0; ctxt->input->cur += l; \ |
| 268 | if (*ctxt->input->cur == '%') xmlParserHandlePEReference(ctxt); \ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 269 | } while (0) |
| 270 | |
| 271 | #define CUR_CHAR(l) xmlCurrentChar(ctxt, &l) |
| 272 | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
| 273 | |
| 274 | #define COPY_BUF(l,b,i,v) \ |
| 275 | if (l == 1) b[i++] = (xmlChar) v; \ |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 276 | else i += xmlCopyCharMultiByte(&b[i],v) |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 277 | |
| 278 | /** |
| 279 | * xmlSkipBlankChars: |
| 280 | * @ctxt: the XML parser context |
| 281 | * |
| 282 | * skip all blanks character found at that point in the input streams. |
| 283 | * It pops up finished entities in the process if allowable at that point. |
| 284 | * |
| 285 | * Returns the number of space chars skipped |
| 286 | */ |
| 287 | |
| 288 | int |
| 289 | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
Daniel Veillard | 02141ea | 2001-04-30 11:46:40 +0000 | [diff] [blame^] | 290 | int res = 0; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 291 | |
Daniel Veillard | 02141ea | 2001-04-30 11:46:40 +0000 | [diff] [blame^] | 292 | if (ctxt->token != 0) { |
| 293 | if (!IS_BLANK(ctxt->token)) |
| 294 | return(0); |
| 295 | ctxt->token = 0; |
| 296 | res++; |
| 297 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 298 | /* |
| 299 | * It's Okay to use CUR/NEXT here since all the blanks are on |
| 300 | * the ASCII range. |
| 301 | */ |
Daniel Veillard | 02141ea | 2001-04-30 11:46:40 +0000 | [diff] [blame^] | 302 | if ((ctxt->inputNr == 1) && (ctxt->instate != XML_PARSER_DTD)) { |
| 303 | const xmlChar *cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 304 | /* |
Daniel Veillard | 02141ea | 2001-04-30 11:46:40 +0000 | [diff] [blame^] | 305 | * if we are in the document content, go really fast |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 306 | */ |
Daniel Veillard | 02141ea | 2001-04-30 11:46:40 +0000 | [diff] [blame^] | 307 | cur = ctxt->input->cur; |
| 308 | while (IS_BLANK(*cur)) { |
| 309 | if (*cur == '\n') { |
| 310 | ctxt->input->line++; ctxt->input->col = 1; |
| 311 | } |
| 312 | cur++; |
| 313 | res++; |
| 314 | if (*cur == 0) { |
| 315 | ctxt->input->cur = cur; |
| 316 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
| 317 | cur = ctxt->input->cur; |
| 318 | } |
| 319 | } |
| 320 | ctxt->input->cur = cur; |
| 321 | } else { |
| 322 | int cur; |
| 323 | do { |
| 324 | cur = CUR; |
| 325 | while (IS_BLANK(cur)) { /* CHECKED tstblanks.xml */ |
| 326 | NEXT; |
| 327 | cur = CUR; |
| 328 | res++; |
| 329 | } |
| 330 | while ((cur == 0) && (ctxt->inputNr > 1) && |
| 331 | (ctxt->instate != XML_PARSER_COMMENT)) { |
| 332 | xmlPopInput(ctxt); |
| 333 | cur = CUR; |
| 334 | } |
| 335 | /* |
| 336 | * Need to handle support of entities branching here |
| 337 | */ |
| 338 | if (*ctxt->input->cur == '%') xmlParserHandlePEReference(ctxt); |
| 339 | } while (IS_BLANK(cur)); /* CHECKED tstblanks.xml */ |
| 340 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 341 | return(res); |
| 342 | } |
| 343 | |
| 344 | /************************************************************************ |
| 345 | * * |
| 346 | * Commodity functions to handle entities * |
| 347 | * * |
| 348 | ************************************************************************/ |
| 349 | |
| 350 | /** |
| 351 | * xmlPopInput: |
| 352 | * @ctxt: an XML parser context |
| 353 | * |
| 354 | * xmlPopInput: the current input pointed by ctxt->input came to an end |
| 355 | * pop it and return the next char. |
| 356 | * |
| 357 | * Returns the current xmlChar in the parser context |
| 358 | */ |
| 359 | xmlChar |
| 360 | xmlPopInput(xmlParserCtxtPtr ctxt) { |
| 361 | if (ctxt->inputNr == 1) return(0); /* End of main Input */ |
| 362 | if (xmlParserDebugEntities) |
| 363 | xmlGenericError(xmlGenericErrorContext, |
| 364 | "Popping input %d\n", ctxt->inputNr); |
| 365 | xmlFreeInputStream(inputPop(ctxt)); |
| 366 | if ((*ctxt->input->cur == 0) && |
| 367 | (xmlParserInputGrow(ctxt->input, INPUT_CHUNK) <= 0)) |
| 368 | return(xmlPopInput(ctxt)); |
| 369 | return(CUR); |
| 370 | } |
| 371 | |
| 372 | /** |
| 373 | * xmlPushInput: |
| 374 | * @ctxt: an XML parser context |
| 375 | * @input: an XML parser input fragment (entity, XML fragment ...). |
| 376 | * |
| 377 | * xmlPushInput: switch to a new input stream which is stacked on top |
| 378 | * of the previous one(s). |
| 379 | */ |
| 380 | void |
| 381 | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
| 382 | if (input == NULL) return; |
| 383 | |
| 384 | if (xmlParserDebugEntities) { |
| 385 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
| 386 | xmlGenericError(xmlGenericErrorContext, |
| 387 | "%s(%d): ", ctxt->input->filename, |
| 388 | ctxt->input->line); |
| 389 | xmlGenericError(xmlGenericErrorContext, |
| 390 | "Pushing input %d : %.30s\n", ctxt->inputNr+1, input->cur); |
| 391 | } |
| 392 | inputPush(ctxt, input); |
| 393 | GROW; |
| 394 | } |
| 395 | |
| 396 | /** |
| 397 | * xmlParseCharRef: |
| 398 | * @ctxt: an XML parser context |
| 399 | * |
| 400 | * parse Reference declarations |
| 401 | * |
| 402 | * [66] CharRef ::= '&#' [0-9]+ ';' | |
| 403 | * '&#x' [0-9a-fA-F]+ ';' |
| 404 | * |
| 405 | * [ WFC: Legal Character ] |
| 406 | * Characters referred to using character references must match the |
| 407 | * production for Char. |
| 408 | * |
| 409 | * Returns the value parsed (as an int), 0 in case of error |
| 410 | */ |
| 411 | int |
| 412 | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
Daniel Veillard | 5058211 | 2001-03-26 22:52:16 +0000 | [diff] [blame] | 413 | unsigned int val = 0; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 414 | int count = 0; |
| 415 | |
| 416 | if (ctxt->token != 0) { |
| 417 | val = ctxt->token; |
| 418 | ctxt->token = 0; |
| 419 | return(val); |
| 420 | } |
| 421 | /* |
| 422 | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
| 423 | */ |
| 424 | if ((RAW == '&') && (NXT(1) == '#') && |
| 425 | (NXT(2) == 'x')) { |
| 426 | SKIP(3); |
| 427 | GROW; |
| 428 | while (RAW != ';') { /* loop blocked by count */ |
| 429 | if ((RAW >= '0') && (RAW <= '9') && (count < 20)) |
| 430 | val = val * 16 + (CUR - '0'); |
| 431 | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
| 432 | val = val * 16 + (CUR - 'a') + 10; |
| 433 | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
| 434 | val = val * 16 + (CUR - 'A') + 10; |
| 435 | else { |
| 436 | ctxt->errNo = XML_ERR_INVALID_HEX_CHARREF; |
| 437 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 438 | ctxt->sax->error(ctxt->userData, |
| 439 | "xmlParseCharRef: invalid hexadecimal value\n"); |
| 440 | ctxt->wellFormed = 0; |
| 441 | ctxt->disableSAX = 1; |
| 442 | val = 0; |
| 443 | break; |
| 444 | } |
| 445 | NEXT; |
| 446 | count++; |
| 447 | } |
| 448 | if (RAW == ';') { |
| 449 | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
| 450 | ctxt->nbChars ++; |
| 451 | ctxt->input->cur++; |
| 452 | } |
| 453 | } else if ((RAW == '&') && (NXT(1) == '#')) { |
| 454 | SKIP(2); |
| 455 | GROW; |
| 456 | while (RAW != ';') { /* loop blocked by count */ |
| 457 | if ((RAW >= '0') && (RAW <= '9') && (count < 20)) |
| 458 | val = val * 10 + (CUR - '0'); |
| 459 | else { |
| 460 | ctxt->errNo = XML_ERR_INVALID_DEC_CHARREF; |
| 461 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 462 | ctxt->sax->error(ctxt->userData, |
| 463 | "xmlParseCharRef: invalid decimal value\n"); |
| 464 | ctxt->wellFormed = 0; |
| 465 | ctxt->disableSAX = 1; |
| 466 | val = 0; |
| 467 | break; |
| 468 | } |
| 469 | NEXT; |
| 470 | count++; |
| 471 | } |
| 472 | if (RAW == ';') { |
| 473 | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
| 474 | ctxt->nbChars ++; |
| 475 | ctxt->input->cur++; |
| 476 | } |
| 477 | } else { |
| 478 | ctxt->errNo = XML_ERR_INVALID_CHARREF; |
| 479 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 480 | ctxt->sax->error(ctxt->userData, |
| 481 | "xmlParseCharRef: invalid value\n"); |
| 482 | ctxt->wellFormed = 0; |
| 483 | ctxt->disableSAX = 1; |
| 484 | } |
| 485 | |
| 486 | /* |
| 487 | * [ WFC: Legal Character ] |
| 488 | * Characters referred to using character references must match the |
| 489 | * production for Char. |
| 490 | */ |
| 491 | if (IS_CHAR(val)) { |
| 492 | return(val); |
| 493 | } else { |
| 494 | ctxt->errNo = XML_ERR_INVALID_CHAR; |
| 495 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 496 | ctxt->sax->error(ctxt->userData, "CharRef: invalid xmlChar value %d\n", |
| 497 | val); |
| 498 | ctxt->wellFormed = 0; |
| 499 | ctxt->disableSAX = 1; |
| 500 | } |
| 501 | return(0); |
| 502 | } |
| 503 | |
| 504 | /** |
| 505 | * xmlParseStringCharRef: |
| 506 | * @ctxt: an XML parser context |
| 507 | * @str: a pointer to an index in the string |
| 508 | * |
| 509 | * parse Reference declarations, variant parsing from a string rather |
| 510 | * than an an input flow. |
| 511 | * |
| 512 | * [66] CharRef ::= '&#' [0-9]+ ';' | |
| 513 | * '&#x' [0-9a-fA-F]+ ';' |
| 514 | * |
| 515 | * [ WFC: Legal Character ] |
| 516 | * Characters referred to using character references must match the |
| 517 | * production for Char. |
| 518 | * |
| 519 | * Returns the value parsed (as an int), 0 in case of error, str will be |
| 520 | * updated to the current value of the index |
| 521 | */ |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 522 | static int |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 523 | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
| 524 | const xmlChar *ptr; |
| 525 | xmlChar cur; |
| 526 | int val = 0; |
| 527 | |
| 528 | if ((str == NULL) || (*str == NULL)) return(0); |
| 529 | ptr = *str; |
| 530 | cur = *ptr; |
| 531 | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
| 532 | ptr += 3; |
| 533 | cur = *ptr; |
| 534 | while (cur != ';') { /* Non input consuming loop */ |
| 535 | if ((cur >= '0') && (cur <= '9')) |
| 536 | val = val * 16 + (cur - '0'); |
| 537 | else if ((cur >= 'a') && (cur <= 'f')) |
| 538 | val = val * 16 + (cur - 'a') + 10; |
| 539 | else if ((cur >= 'A') && (cur <= 'F')) |
| 540 | val = val * 16 + (cur - 'A') + 10; |
| 541 | else { |
| 542 | ctxt->errNo = XML_ERR_INVALID_HEX_CHARREF; |
| 543 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 544 | ctxt->sax->error(ctxt->userData, |
| 545 | "xmlParseStringCharRef: invalid hexadecimal value\n"); |
| 546 | ctxt->wellFormed = 0; |
| 547 | ctxt->disableSAX = 1; |
| 548 | val = 0; |
| 549 | break; |
| 550 | } |
| 551 | ptr++; |
| 552 | cur = *ptr; |
| 553 | } |
| 554 | if (cur == ';') |
| 555 | ptr++; |
| 556 | } else if ((cur == '&') && (ptr[1] == '#')){ |
| 557 | ptr += 2; |
| 558 | cur = *ptr; |
| 559 | while (cur != ';') { /* Non input consuming loops */ |
| 560 | if ((cur >= '0') && (cur <= '9')) |
| 561 | val = val * 10 + (cur - '0'); |
| 562 | else { |
| 563 | ctxt->errNo = XML_ERR_INVALID_DEC_CHARREF; |
| 564 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 565 | ctxt->sax->error(ctxt->userData, |
| 566 | "xmlParseStringCharRef: invalid decimal value\n"); |
| 567 | ctxt->wellFormed = 0; |
| 568 | ctxt->disableSAX = 1; |
| 569 | val = 0; |
| 570 | break; |
| 571 | } |
| 572 | ptr++; |
| 573 | cur = *ptr; |
| 574 | } |
| 575 | if (cur == ';') |
| 576 | ptr++; |
| 577 | } else { |
| 578 | ctxt->errNo = XML_ERR_INVALID_CHARREF; |
| 579 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 580 | ctxt->sax->error(ctxt->userData, |
| 581 | "xmlParseCharRef: invalid value\n"); |
| 582 | ctxt->wellFormed = 0; |
| 583 | ctxt->disableSAX = 1; |
| 584 | return(0); |
| 585 | } |
| 586 | *str = ptr; |
| 587 | |
| 588 | /* |
| 589 | * [ WFC: Legal Character ] |
| 590 | * Characters referred to using character references must match the |
| 591 | * production for Char. |
| 592 | */ |
| 593 | if (IS_CHAR(val)) { |
| 594 | return(val); |
| 595 | } else { |
| 596 | ctxt->errNo = XML_ERR_INVALID_CHAR; |
| 597 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 598 | ctxt->sax->error(ctxt->userData, |
| 599 | "CharRef: invalid xmlChar value %d\n", val); |
| 600 | ctxt->wellFormed = 0; |
| 601 | ctxt->disableSAX = 1; |
| 602 | } |
| 603 | return(0); |
| 604 | } |
| 605 | |
| 606 | /** |
| 607 | * xmlParserHandlePEReference: |
| 608 | * @ctxt: the parser context |
| 609 | * |
| 610 | * [69] PEReference ::= '%' Name ';' |
| 611 | * |
| 612 | * [ WFC: No Recursion ] |
| 613 | * A parsed entity must not contain a recursive |
| 614 | * reference to itself, either directly or indirectly. |
| 615 | * |
| 616 | * [ WFC: Entity Declared ] |
| 617 | * In a document without any DTD, a document with only an internal DTD |
| 618 | * subset which contains no parameter entity references, or a document |
| 619 | * with "standalone='yes'", ... ... The declaration of a parameter |
| 620 | * entity must precede any reference to it... |
| 621 | * |
| 622 | * [ VC: Entity Declared ] |
| 623 | * In a document with an external subset or external parameter entities |
| 624 | * with "standalone='no'", ... ... The declaration of a parameter entity |
| 625 | * must precede any reference to it... |
| 626 | * |
| 627 | * [ WFC: In DTD ] |
| 628 | * Parameter-entity references may only appear in the DTD. |
| 629 | * NOTE: misleading but this is handled. |
| 630 | * |
| 631 | * A PEReference may have been detected in the current input stream |
| 632 | * the handling is done accordingly to |
| 633 | * http://www.w3.org/TR/REC-xml#entproc |
| 634 | * i.e. |
| 635 | * - Included in literal in entity values |
| 636 | * - Included as Paraemeter Entity reference within DTDs |
| 637 | */ |
| 638 | void |
| 639 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
| 640 | xmlChar *name; |
| 641 | xmlEntityPtr entity = NULL; |
| 642 | xmlParserInputPtr input; |
| 643 | |
| 644 | if (ctxt->token != 0) { |
| 645 | return; |
| 646 | } |
| 647 | if (RAW != '%') return; |
| 648 | switch(ctxt->instate) { |
| 649 | case XML_PARSER_CDATA_SECTION: |
| 650 | return; |
| 651 | case XML_PARSER_COMMENT: |
| 652 | return; |
| 653 | case XML_PARSER_START_TAG: |
| 654 | return; |
| 655 | case XML_PARSER_END_TAG: |
| 656 | return; |
| 657 | case XML_PARSER_EOF: |
| 658 | ctxt->errNo = XML_ERR_PEREF_AT_EOF; |
| 659 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 660 | ctxt->sax->error(ctxt->userData, "PEReference at EOF\n"); |
| 661 | ctxt->wellFormed = 0; |
| 662 | ctxt->disableSAX = 1; |
| 663 | return; |
| 664 | case XML_PARSER_PROLOG: |
| 665 | case XML_PARSER_START: |
| 666 | case XML_PARSER_MISC: |
| 667 | ctxt->errNo = XML_ERR_PEREF_IN_PROLOG; |
| 668 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 669 | ctxt->sax->error(ctxt->userData, "PEReference in prolog!\n"); |
| 670 | ctxt->wellFormed = 0; |
| 671 | ctxt->disableSAX = 1; |
| 672 | return; |
| 673 | case XML_PARSER_ENTITY_DECL: |
| 674 | case XML_PARSER_CONTENT: |
| 675 | case XML_PARSER_ATTRIBUTE_VALUE: |
| 676 | case XML_PARSER_PI: |
| 677 | case XML_PARSER_SYSTEM_LITERAL: |
| 678 | /* we just ignore it there */ |
| 679 | return; |
| 680 | case XML_PARSER_EPILOG: |
| 681 | ctxt->errNo = XML_ERR_PEREF_IN_EPILOG; |
| 682 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 683 | ctxt->sax->error(ctxt->userData, "PEReference in epilog!\n"); |
| 684 | ctxt->wellFormed = 0; |
| 685 | ctxt->disableSAX = 1; |
| 686 | return; |
| 687 | case XML_PARSER_ENTITY_VALUE: |
| 688 | /* |
| 689 | * NOTE: in the case of entity values, we don't do the |
| 690 | * substitution here since we need the literal |
| 691 | * entity value to be able to save the internal |
| 692 | * subset of the document. |
| 693 | * This will be handled by xmlStringDecodeEntities |
| 694 | */ |
| 695 | return; |
| 696 | case XML_PARSER_DTD: |
| 697 | /* |
| 698 | * [WFC: Well-Formedness Constraint: PEs in Internal Subset] |
| 699 | * In the internal DTD subset, parameter-entity references |
| 700 | * can occur only where markup declarations can occur, not |
| 701 | * within markup declarations. |
| 702 | * In that case this is handled in xmlParseMarkupDecl |
| 703 | */ |
| 704 | if ((ctxt->external == 0) && (ctxt->inputNr == 1)) |
| 705 | return; |
| 706 | break; |
| 707 | case XML_PARSER_IGNORE: |
| 708 | return; |
| 709 | } |
| 710 | |
| 711 | NEXT; |
| 712 | name = xmlParseName(ctxt); |
| 713 | if (xmlParserDebugEntities) |
| 714 | xmlGenericError(xmlGenericErrorContext, |
| 715 | "PE Reference: %s\n", name); |
| 716 | if (name == NULL) { |
| 717 | ctxt->errNo = XML_ERR_PEREF_NO_NAME; |
| 718 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 719 | ctxt->sax->error(ctxt->userData, "xmlHandlePEReference: no name\n"); |
| 720 | ctxt->wellFormed = 0; |
| 721 | ctxt->disableSAX = 1; |
| 722 | } else { |
| 723 | if (RAW == ';') { |
| 724 | NEXT; |
| 725 | if ((ctxt->sax != NULL) && (ctxt->sax->getParameterEntity != NULL)) |
| 726 | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
| 727 | if (entity == NULL) { |
| 728 | |
| 729 | /* |
| 730 | * [ WFC: Entity Declared ] |
| 731 | * In a document without any DTD, a document with only an |
| 732 | * internal DTD subset which contains no parameter entity |
| 733 | * references, or a document with "standalone='yes'", ... |
| 734 | * ... The declaration of a parameter entity must precede |
| 735 | * any reference to it... |
| 736 | */ |
| 737 | if ((ctxt->standalone == 1) || |
| 738 | ((ctxt->hasExternalSubset == 0) && |
| 739 | (ctxt->hasPErefs == 0))) { |
| 740 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 741 | ctxt->sax->error(ctxt->userData, |
| 742 | "PEReference: %%%s; not found\n", name); |
| 743 | ctxt->wellFormed = 0; |
| 744 | ctxt->disableSAX = 1; |
| 745 | } else { |
| 746 | /* |
| 747 | * [ VC: Entity Declared ] |
| 748 | * In a document with an external subset or external |
| 749 | * parameter entities with "standalone='no'", ... |
| 750 | * ... The declaration of a parameter entity must precede |
| 751 | * any reference to it... |
| 752 | */ |
| 753 | if ((!ctxt->disableSAX) && |
| 754 | (ctxt->validate) && (ctxt->vctxt.error != NULL)) { |
| 755 | ctxt->vctxt.error(ctxt->vctxt.userData, |
| 756 | "PEReference: %%%s; not found\n", name); |
| 757 | } else if ((!ctxt->disableSAX) && |
| 758 | (ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 759 | ctxt->sax->warning(ctxt->userData, |
| 760 | "PEReference: %%%s; not found\n", name); |
| 761 | ctxt->valid = 0; |
| 762 | } |
| 763 | } else { |
| 764 | if ((entity->etype == XML_INTERNAL_PARAMETER_ENTITY) || |
| 765 | (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)) { |
| 766 | /* |
| 767 | * handle the extra spaces added before and after |
| 768 | * c.f. http://www.w3.org/TR/REC-xml#as-PE |
| 769 | * this is done independantly. |
| 770 | */ |
| 771 | input = xmlNewEntityInputStream(ctxt, entity); |
| 772 | xmlPushInput(ctxt, input); |
| 773 | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
| 774 | (RAW == '<') && (NXT(1) == '?') && |
| 775 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 776 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 777 | xmlParseTextDecl(ctxt); |
| 778 | } |
| 779 | if (ctxt->token == 0) |
| 780 | ctxt->token = ' '; |
| 781 | } else { |
| 782 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 783 | ctxt->sax->error(ctxt->userData, |
| 784 | "xmlHandlePEReference: %s is not a parameter entity\n", |
| 785 | name); |
| 786 | ctxt->wellFormed = 0; |
| 787 | ctxt->disableSAX = 1; |
| 788 | } |
| 789 | } |
| 790 | } else { |
| 791 | ctxt->errNo = XML_ERR_PEREF_SEMICOL_MISSING; |
| 792 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 793 | ctxt->sax->error(ctxt->userData, |
| 794 | "xmlHandlePEReference: expecting ';'\n"); |
| 795 | ctxt->wellFormed = 0; |
| 796 | ctxt->disableSAX = 1; |
| 797 | } |
| 798 | xmlFree(name); |
| 799 | } |
| 800 | } |
| 801 | |
| 802 | /* |
| 803 | * Macro used to grow the current buffer. |
| 804 | */ |
| 805 | #define growBuffer(buffer) { \ |
| 806 | buffer##_size *= 2; \ |
| 807 | buffer = (xmlChar *) \ |
| 808 | xmlRealloc(buffer, buffer##_size * sizeof(xmlChar)); \ |
| 809 | if (buffer == NULL) { \ |
| 810 | perror("realloc failed"); \ |
| 811 | return(NULL); \ |
| 812 | } \ |
| 813 | } |
| 814 | |
| 815 | /** |
| 816 | * xmlStringDecodeEntities: |
| 817 | * @ctxt: the parser context |
| 818 | * @str: the input string |
| 819 | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
| 820 | * @end: an end marker xmlChar, 0 if none |
| 821 | * @end2: an end marker xmlChar, 0 if none |
| 822 | * @end3: an end marker xmlChar, 0 if none |
| 823 | * |
| 824 | * Takes a entity string content and process to do the adequate subtitutions. |
| 825 | * |
| 826 | * [67] Reference ::= EntityRef | CharRef |
| 827 | * |
| 828 | * [69] PEReference ::= '%' Name ';' |
| 829 | * |
| 830 | * Returns A newly allocated string with the substitution done. The caller |
| 831 | * must deallocate it ! |
| 832 | */ |
| 833 | xmlChar * |
| 834 | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int what, |
| 835 | xmlChar end, xmlChar end2, xmlChar end3) { |
| 836 | xmlChar *buffer = NULL; |
| 837 | int buffer_size = 0; |
| 838 | |
| 839 | xmlChar *current = NULL; |
| 840 | xmlEntityPtr ent; |
| 841 | int c,l; |
| 842 | int nbchars = 0; |
| 843 | |
| 844 | if (str == NULL) |
| 845 | return(NULL); |
| 846 | |
| 847 | if (ctxt->depth > 40) { |
| 848 | ctxt->errNo = XML_ERR_ENTITY_LOOP; |
| 849 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 850 | ctxt->sax->error(ctxt->userData, |
| 851 | "Detected entity reference loop\n"); |
| 852 | ctxt->wellFormed = 0; |
| 853 | ctxt->disableSAX = 1; |
| 854 | return(NULL); |
| 855 | } |
| 856 | |
| 857 | /* |
| 858 | * allocate a translation buffer. |
| 859 | */ |
| 860 | buffer_size = XML_PARSER_BIG_BUFFER_SIZE; |
| 861 | buffer = (xmlChar *) xmlMalloc(buffer_size * sizeof(xmlChar)); |
| 862 | if (buffer == NULL) { |
| 863 | perror("xmlDecodeEntities: malloc failed"); |
| 864 | return(NULL); |
| 865 | } |
| 866 | |
| 867 | /* |
| 868 | * Ok loop until we reach one of the ending char or a size limit. |
| 869 | * we are operating on already parsed values. |
| 870 | */ |
| 871 | c = CUR_SCHAR(str, l); |
| 872 | while ((c != 0) && (c != end) && /* non input consuming loop */ |
| 873 | (c != end2) && (c != end3)) { |
| 874 | |
| 875 | if (c == 0) break; |
| 876 | if ((c == '&') && (str[1] == '#')) { |
| 877 | int val = xmlParseStringCharRef(ctxt, &str); |
| 878 | if (val != 0) { |
| 879 | COPY_BUF(0,buffer,nbchars,val); |
| 880 | } |
| 881 | } else if ((c == '&') && (what & XML_SUBSTITUTE_REF)) { |
| 882 | if (xmlParserDebugEntities) |
| 883 | xmlGenericError(xmlGenericErrorContext, |
| 884 | "String decoding Entity Reference: %.30s\n", |
| 885 | str); |
| 886 | ent = xmlParseStringEntityRef(ctxt, &str); |
| 887 | if ((ent != NULL) && |
| 888 | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
| 889 | if (ent->content != NULL) { |
| 890 | COPY_BUF(0,buffer,nbchars,ent->content[0]); |
| 891 | } else { |
| 892 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 893 | ctxt->sax->error(ctxt->userData, |
| 894 | "internal error entity has no content\n"); |
| 895 | } |
| 896 | } else if ((ent != NULL) && (ent->content != NULL)) { |
| 897 | xmlChar *rep; |
| 898 | |
| 899 | ctxt->depth++; |
| 900 | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
| 901 | 0, 0, 0); |
| 902 | ctxt->depth--; |
| 903 | if (rep != NULL) { |
| 904 | current = rep; |
| 905 | while (*current != 0) { /* non input consuming loop */ |
| 906 | buffer[nbchars++] = *current++; |
| 907 | if (nbchars > |
| 908 | buffer_size - XML_PARSER_BUFFER_SIZE) { |
| 909 | growBuffer(buffer); |
| 910 | } |
| 911 | } |
| 912 | xmlFree(rep); |
| 913 | } |
| 914 | } else if (ent != NULL) { |
| 915 | int i = xmlStrlen(ent->name); |
| 916 | const xmlChar *cur = ent->name; |
| 917 | |
| 918 | buffer[nbchars++] = '&'; |
| 919 | if (nbchars > buffer_size - i - XML_PARSER_BUFFER_SIZE) { |
| 920 | growBuffer(buffer); |
| 921 | } |
| 922 | for (;i > 0;i--) |
| 923 | buffer[nbchars++] = *cur++; |
| 924 | buffer[nbchars++] = ';'; |
| 925 | } |
| 926 | } else if (c == '%' && (what & XML_SUBSTITUTE_PEREF)) { |
| 927 | if (xmlParserDebugEntities) |
| 928 | xmlGenericError(xmlGenericErrorContext, |
| 929 | "String decoding PE Reference: %.30s\n", str); |
| 930 | ent = xmlParseStringPEReference(ctxt, &str); |
| 931 | if (ent != NULL) { |
| 932 | xmlChar *rep; |
| 933 | |
| 934 | ctxt->depth++; |
| 935 | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
| 936 | 0, 0, 0); |
| 937 | ctxt->depth--; |
| 938 | if (rep != NULL) { |
| 939 | current = rep; |
| 940 | while (*current != 0) { /* non input consuming loop */ |
| 941 | buffer[nbchars++] = *current++; |
| 942 | if (nbchars > |
| 943 | buffer_size - XML_PARSER_BUFFER_SIZE) { |
| 944 | growBuffer(buffer); |
| 945 | } |
| 946 | } |
| 947 | xmlFree(rep); |
| 948 | } |
| 949 | } |
| 950 | } else { |
| 951 | COPY_BUF(l,buffer,nbchars,c); |
| 952 | str += l; |
| 953 | if (nbchars > buffer_size - XML_PARSER_BUFFER_SIZE) { |
| 954 | growBuffer(buffer); |
| 955 | } |
| 956 | } |
| 957 | c = CUR_SCHAR(str, l); |
| 958 | } |
| 959 | buffer[nbchars++] = 0; |
| 960 | return(buffer); |
| 961 | } |
| 962 | |
| 963 | |
| 964 | /************************************************************************ |
| 965 | * * |
| 966 | * Commodity functions to handle xmlChars * |
| 967 | * * |
| 968 | ************************************************************************/ |
| 969 | |
| 970 | /** |
| 971 | * xmlStrndup: |
| 972 | * @cur: the input xmlChar * |
| 973 | * @len: the len of @cur |
| 974 | * |
| 975 | * a strndup for array of xmlChar's |
| 976 | * |
| 977 | * Returns a new xmlChar * or NULL |
| 978 | */ |
| 979 | xmlChar * |
| 980 | xmlStrndup(const xmlChar *cur, int len) { |
| 981 | xmlChar *ret; |
| 982 | |
| 983 | if ((cur == NULL) || (len < 0)) return(NULL); |
| 984 | ret = (xmlChar *) xmlMalloc((len + 1) * sizeof(xmlChar)); |
| 985 | if (ret == NULL) { |
| 986 | xmlGenericError(xmlGenericErrorContext, |
| 987 | "malloc of %ld byte failed\n", |
| 988 | (len + 1) * (long)sizeof(xmlChar)); |
| 989 | return(NULL); |
| 990 | } |
| 991 | memcpy(ret, cur, len * sizeof(xmlChar)); |
| 992 | ret[len] = 0; |
| 993 | return(ret); |
| 994 | } |
| 995 | |
| 996 | /** |
| 997 | * xmlStrdup: |
| 998 | * @cur: the input xmlChar * |
| 999 | * |
| 1000 | * a strdup for array of xmlChar's. Since they are supposed to be |
| 1001 | * encoded in UTF-8 or an encoding with 8bit based chars, we assume |
| 1002 | * a termination mark of '0'. |
| 1003 | * |
| 1004 | * Returns a new xmlChar * or NULL |
| 1005 | */ |
| 1006 | xmlChar * |
| 1007 | xmlStrdup(const xmlChar *cur) { |
| 1008 | const xmlChar *p = cur; |
| 1009 | |
| 1010 | if (cur == NULL) return(NULL); |
| 1011 | while (*p != 0) p++; /* non input consuming */ |
| 1012 | return(xmlStrndup(cur, p - cur)); |
| 1013 | } |
| 1014 | |
| 1015 | /** |
| 1016 | * xmlCharStrndup: |
| 1017 | * @cur: the input char * |
| 1018 | * @len: the len of @cur |
| 1019 | * |
| 1020 | * a strndup for char's to xmlChar's |
| 1021 | * |
| 1022 | * Returns a new xmlChar * or NULL |
| 1023 | */ |
| 1024 | |
| 1025 | xmlChar * |
| 1026 | xmlCharStrndup(const char *cur, int len) { |
| 1027 | int i; |
| 1028 | xmlChar *ret; |
| 1029 | |
| 1030 | if ((cur == NULL) || (len < 0)) return(NULL); |
| 1031 | ret = (xmlChar *) xmlMalloc((len + 1) * sizeof(xmlChar)); |
| 1032 | if (ret == NULL) { |
| 1033 | xmlGenericError(xmlGenericErrorContext, "malloc of %ld byte failed\n", |
| 1034 | (len + 1) * (long)sizeof(xmlChar)); |
| 1035 | return(NULL); |
| 1036 | } |
| 1037 | for (i = 0;i < len;i++) |
| 1038 | ret[i] = (xmlChar) cur[i]; |
| 1039 | ret[len] = 0; |
| 1040 | return(ret); |
| 1041 | } |
| 1042 | |
| 1043 | /** |
| 1044 | * xmlCharStrdup: |
| 1045 | * @cur: the input char * |
| 1046 | * @len: the len of @cur |
| 1047 | * |
| 1048 | * a strdup for char's to xmlChar's |
| 1049 | * |
| 1050 | * Returns a new xmlChar * or NULL |
| 1051 | */ |
| 1052 | |
| 1053 | xmlChar * |
| 1054 | xmlCharStrdup(const char *cur) { |
| 1055 | const char *p = cur; |
| 1056 | |
| 1057 | if (cur == NULL) return(NULL); |
| 1058 | while (*p != '\0') p++; /* non input consuming */ |
| 1059 | return(xmlCharStrndup(cur, p - cur)); |
| 1060 | } |
| 1061 | |
| 1062 | /** |
| 1063 | * xmlStrcmp: |
| 1064 | * @str1: the first xmlChar * |
| 1065 | * @str2: the second xmlChar * |
| 1066 | * |
| 1067 | * a strcmp for xmlChar's |
| 1068 | * |
| 1069 | * Returns the integer result of the comparison |
| 1070 | */ |
| 1071 | |
| 1072 | int |
| 1073 | xmlStrcmp(const xmlChar *str1, const xmlChar *str2) { |
| 1074 | register int tmp; |
| 1075 | |
| 1076 | if (str1 == str2) return(0); |
| 1077 | if (str1 == NULL) return(-1); |
| 1078 | if (str2 == NULL) return(1); |
| 1079 | do { |
| 1080 | tmp = *str1++ - *str2; |
| 1081 | if (tmp != 0) return(tmp); |
| 1082 | } while (*str2++ != 0); |
| 1083 | return 0; |
| 1084 | } |
| 1085 | |
| 1086 | /** |
| 1087 | * xmlStrEqual: |
| 1088 | * @str1: the first xmlChar * |
| 1089 | * @str2: the second xmlChar * |
| 1090 | * |
| 1091 | * Check if both string are equal of have same content |
| 1092 | * Should be a bit more readable and faster than xmlStrEqual() |
| 1093 | * |
| 1094 | * Returns 1 if they are equal, 0 if they are different |
| 1095 | */ |
| 1096 | |
| 1097 | int |
| 1098 | xmlStrEqual(const xmlChar *str1, const xmlChar *str2) { |
| 1099 | if (str1 == str2) return(1); |
| 1100 | if (str1 == NULL) return(0); |
| 1101 | if (str2 == NULL) return(0); |
| 1102 | do { |
| 1103 | if (*str1++ != *str2) return(0); |
| 1104 | } while (*str2++); |
| 1105 | return(1); |
| 1106 | } |
| 1107 | |
| 1108 | /** |
| 1109 | * xmlStrncmp: |
| 1110 | * @str1: the first xmlChar * |
| 1111 | * @str2: the second xmlChar * |
| 1112 | * @len: the max comparison length |
| 1113 | * |
| 1114 | * a strncmp for xmlChar's |
| 1115 | * |
| 1116 | * Returns the integer result of the comparison |
| 1117 | */ |
| 1118 | |
| 1119 | int |
| 1120 | xmlStrncmp(const xmlChar *str1, const xmlChar *str2, int len) { |
| 1121 | register int tmp; |
| 1122 | |
| 1123 | if (len <= 0) return(0); |
| 1124 | if (str1 == str2) return(0); |
| 1125 | if (str1 == NULL) return(-1); |
| 1126 | if (str2 == NULL) return(1); |
| 1127 | do { |
| 1128 | tmp = *str1++ - *str2; |
| 1129 | if (tmp != 0 || --len == 0) return(tmp); |
| 1130 | } while (*str2++ != 0); |
| 1131 | return 0; |
| 1132 | } |
| 1133 | |
| 1134 | static xmlChar casemap[256] = { |
| 1135 | 0x00,0x01,0x02,0x03,0x04,0x05,0x06,0x07, |
| 1136 | 0x08,0x09,0x0A,0x0B,0x0C,0x0D,0x0E,0x0F, |
| 1137 | 0x10,0x11,0x12,0x13,0x14,0x15,0x16,0x17, |
| 1138 | 0x18,0x19,0x1A,0x1B,0x1C,0x1D,0x1E,0x1F, |
| 1139 | 0x20,0x21,0x22,0x23,0x24,0x25,0x26,0x27, |
| 1140 | 0x28,0x29,0x2A,0x2B,0x2C,0x2D,0x2E,0x2F, |
| 1141 | 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37, |
| 1142 | 0x38,0x39,0x3A,0x3B,0x3C,0x3D,0x3E,0x3F, |
| 1143 | 0x40,0x61,0x62,0x63,0x64,0x65,0x66,0x67, |
| 1144 | 0x68,0x69,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F, |
| 1145 | 0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77, |
| 1146 | 0x78,0x79,0x7A,0x7B,0x5C,0x5D,0x5E,0x5F, |
| 1147 | 0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67, |
| 1148 | 0x68,0x69,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F, |
| 1149 | 0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77, |
| 1150 | 0x78,0x79,0x7A,0x7B,0x7C,0x7D,0x7E,0x7F, |
| 1151 | 0x80,0x81,0x82,0x83,0x84,0x85,0x86,0x87, |
| 1152 | 0x88,0x89,0x8A,0x8B,0x8C,0x8D,0x8E,0x8F, |
| 1153 | 0x90,0x91,0x92,0x93,0x94,0x95,0x96,0x97, |
| 1154 | 0x98,0x99,0x9A,0x9B,0x9C,0x9D,0x9E,0x9F, |
| 1155 | 0xA0,0xA1,0xA2,0xA3,0xA4,0xA5,0xA6,0xA7, |
| 1156 | 0xA8,0xA9,0xAA,0xAB,0xAC,0xAD,0xAE,0xAF, |
| 1157 | 0xB0,0xB1,0xB2,0xB3,0xB4,0xB5,0xB6,0xB7, |
| 1158 | 0xB8,0xB9,0xBA,0xBB,0xBC,0xBD,0xBE,0xBF, |
| 1159 | 0xC0,0xC1,0xC2,0xC3,0xC4,0xC5,0xC6,0xC7, |
| 1160 | 0xC8,0xC9,0xCA,0xCB,0xCC,0xCD,0xCE,0xCF, |
| 1161 | 0xD0,0xD1,0xD2,0xD3,0xD4,0xD5,0xD6,0xD7, |
| 1162 | 0xD8,0xD9,0xDA,0xDB,0xDC,0xDD,0xDE,0xDF, |
| 1163 | 0xE0,0xE1,0xE2,0xE3,0xE4,0xE5,0xE6,0xE7, |
| 1164 | 0xE8,0xE9,0xEA,0xEB,0xEC,0xED,0xEE,0xEF, |
| 1165 | 0xF0,0xF1,0xF2,0xF3,0xF4,0xF5,0xF6,0xF7, |
| 1166 | 0xF8,0xF9,0xFA,0xFB,0xFC,0xFD,0xFE,0xFF |
| 1167 | }; |
| 1168 | |
| 1169 | /** |
| 1170 | * xmlStrcasecmp: |
| 1171 | * @str1: the first xmlChar * |
| 1172 | * @str2: the second xmlChar * |
| 1173 | * |
| 1174 | * a strcasecmp for xmlChar's |
| 1175 | * |
| 1176 | * Returns the integer result of the comparison |
| 1177 | */ |
| 1178 | |
| 1179 | int |
| 1180 | xmlStrcasecmp(const xmlChar *str1, const xmlChar *str2) { |
| 1181 | register int tmp; |
| 1182 | |
| 1183 | if (str1 == str2) return(0); |
| 1184 | if (str1 == NULL) return(-1); |
| 1185 | if (str2 == NULL) return(1); |
| 1186 | do { |
| 1187 | tmp = casemap[*str1++] - casemap[*str2]; |
| 1188 | if (tmp != 0) return(tmp); |
| 1189 | } while (*str2++ != 0); |
| 1190 | return 0; |
| 1191 | } |
| 1192 | |
| 1193 | /** |
| 1194 | * xmlStrncasecmp: |
| 1195 | * @str1: the first xmlChar * |
| 1196 | * @str2: the second xmlChar * |
| 1197 | * @len: the max comparison length |
| 1198 | * |
| 1199 | * a strncasecmp for xmlChar's |
| 1200 | * |
| 1201 | * Returns the integer result of the comparison |
| 1202 | */ |
| 1203 | |
| 1204 | int |
| 1205 | xmlStrncasecmp(const xmlChar *str1, const xmlChar *str2, int len) { |
| 1206 | register int tmp; |
| 1207 | |
| 1208 | if (len <= 0) return(0); |
| 1209 | if (str1 == str2) return(0); |
| 1210 | if (str1 == NULL) return(-1); |
| 1211 | if (str2 == NULL) return(1); |
| 1212 | do { |
| 1213 | tmp = casemap[*str1++] - casemap[*str2]; |
| 1214 | if (tmp != 0 || --len == 0) return(tmp); |
| 1215 | } while (*str2++ != 0); |
| 1216 | return 0; |
| 1217 | } |
| 1218 | |
| 1219 | /** |
| 1220 | * xmlStrchr: |
| 1221 | * @str: the xmlChar * array |
| 1222 | * @val: the xmlChar to search |
| 1223 | * |
| 1224 | * a strchr for xmlChar's |
| 1225 | * |
| 1226 | * Returns the xmlChar * for the first occurence or NULL. |
| 1227 | */ |
| 1228 | |
| 1229 | const xmlChar * |
| 1230 | xmlStrchr(const xmlChar *str, xmlChar val) { |
| 1231 | if (str == NULL) return(NULL); |
| 1232 | while (*str != 0) { /* non input consuming */ |
| 1233 | if (*str == val) return((xmlChar *) str); |
| 1234 | str++; |
| 1235 | } |
| 1236 | return(NULL); |
| 1237 | } |
| 1238 | |
| 1239 | /** |
| 1240 | * xmlStrstr: |
| 1241 | * @str: the xmlChar * array (haystack) |
| 1242 | * @val: the xmlChar to search (needle) |
| 1243 | * |
| 1244 | * a strstr for xmlChar's |
| 1245 | * |
| 1246 | * Returns the xmlChar * for the first occurence or NULL. |
| 1247 | */ |
| 1248 | |
| 1249 | const xmlChar * |
| 1250 | xmlStrstr(const xmlChar *str, xmlChar *val) { |
| 1251 | int n; |
| 1252 | |
| 1253 | if (str == NULL) return(NULL); |
| 1254 | if (val == NULL) return(NULL); |
| 1255 | n = xmlStrlen(val); |
| 1256 | |
| 1257 | if (n == 0) return(str); |
| 1258 | while (*str != 0) { /* non input consuming */ |
| 1259 | if (*str == *val) { |
| 1260 | if (!xmlStrncmp(str, val, n)) return((const xmlChar *) str); |
| 1261 | } |
| 1262 | str++; |
| 1263 | } |
| 1264 | return(NULL); |
| 1265 | } |
| 1266 | |
| 1267 | /** |
| 1268 | * xmlStrcasestr: |
| 1269 | * @str: the xmlChar * array (haystack) |
| 1270 | * @val: the xmlChar to search (needle) |
| 1271 | * |
| 1272 | * a case-ignoring strstr for xmlChar's |
| 1273 | * |
| 1274 | * Returns the xmlChar * for the first occurence or NULL. |
| 1275 | */ |
| 1276 | |
| 1277 | const xmlChar * |
| 1278 | xmlStrcasestr(const xmlChar *str, xmlChar *val) { |
| 1279 | int n; |
| 1280 | |
| 1281 | if (str == NULL) return(NULL); |
| 1282 | if (val == NULL) return(NULL); |
| 1283 | n = xmlStrlen(val); |
| 1284 | |
| 1285 | if (n == 0) return(str); |
| 1286 | while (*str != 0) { /* non input consuming */ |
| 1287 | if (casemap[*str] == casemap[*val]) |
| 1288 | if (!xmlStrncasecmp(str, val, n)) return(str); |
| 1289 | str++; |
| 1290 | } |
| 1291 | return(NULL); |
| 1292 | } |
| 1293 | |
| 1294 | /** |
| 1295 | * xmlStrsub: |
| 1296 | * @str: the xmlChar * array (haystack) |
| 1297 | * @start: the index of the first char (zero based) |
| 1298 | * @len: the length of the substring |
| 1299 | * |
| 1300 | * Extract a substring of a given string |
| 1301 | * |
| 1302 | * Returns the xmlChar * for the first occurence or NULL. |
| 1303 | */ |
| 1304 | |
| 1305 | xmlChar * |
| 1306 | xmlStrsub(const xmlChar *str, int start, int len) { |
| 1307 | int i; |
| 1308 | |
| 1309 | if (str == NULL) return(NULL); |
| 1310 | if (start < 0) return(NULL); |
| 1311 | if (len < 0) return(NULL); |
| 1312 | |
| 1313 | for (i = 0;i < start;i++) { |
| 1314 | if (*str == 0) return(NULL); |
| 1315 | str++; |
| 1316 | } |
| 1317 | if (*str == 0) return(NULL); |
| 1318 | return(xmlStrndup(str, len)); |
| 1319 | } |
| 1320 | |
| 1321 | /** |
| 1322 | * xmlStrlen: |
| 1323 | * @str: the xmlChar * array |
| 1324 | * |
| 1325 | * length of a xmlChar's string |
| 1326 | * |
| 1327 | * Returns the number of xmlChar contained in the ARRAY. |
| 1328 | */ |
| 1329 | |
| 1330 | int |
| 1331 | xmlStrlen(const xmlChar *str) { |
| 1332 | int len = 0; |
| 1333 | |
| 1334 | if (str == NULL) return(0); |
| 1335 | while (*str != 0) { /* non input consuming */ |
| 1336 | str++; |
| 1337 | len++; |
| 1338 | } |
| 1339 | return(len); |
| 1340 | } |
| 1341 | |
| 1342 | /** |
| 1343 | * xmlStrncat: |
| 1344 | * @cur: the original xmlChar * array |
| 1345 | * @add: the xmlChar * array added |
| 1346 | * @len: the length of @add |
| 1347 | * |
| 1348 | * a strncat for array of xmlChar's, it will extend cur with the len |
| 1349 | * first bytes of @add. |
| 1350 | * |
| 1351 | * Returns a new xmlChar *, the original @cur is reallocated if needed |
| 1352 | * and should not be freed |
| 1353 | */ |
| 1354 | |
| 1355 | xmlChar * |
| 1356 | xmlStrncat(xmlChar *cur, const xmlChar *add, int len) { |
| 1357 | int size; |
| 1358 | xmlChar *ret; |
| 1359 | |
| 1360 | if ((add == NULL) || (len == 0)) |
| 1361 | return(cur); |
| 1362 | if (cur == NULL) |
| 1363 | return(xmlStrndup(add, len)); |
| 1364 | |
| 1365 | size = xmlStrlen(cur); |
| 1366 | ret = (xmlChar *) xmlRealloc(cur, (size + len + 1) * sizeof(xmlChar)); |
| 1367 | if (ret == NULL) { |
| 1368 | xmlGenericError(xmlGenericErrorContext, |
| 1369 | "xmlStrncat: realloc of %ld byte failed\n", |
| 1370 | (size + len + 1) * (long)sizeof(xmlChar)); |
| 1371 | return(cur); |
| 1372 | } |
| 1373 | memcpy(&ret[size], add, len * sizeof(xmlChar)); |
| 1374 | ret[size + len] = 0; |
| 1375 | return(ret); |
| 1376 | } |
| 1377 | |
| 1378 | /** |
| 1379 | * xmlStrcat: |
| 1380 | * @cur: the original xmlChar * array |
| 1381 | * @add: the xmlChar * array added |
| 1382 | * |
| 1383 | * a strcat for array of xmlChar's. Since they are supposed to be |
| 1384 | * encoded in UTF-8 or an encoding with 8bit based chars, we assume |
| 1385 | * a termination mark of '0'. |
| 1386 | * |
| 1387 | * Returns a new xmlChar * containing the concatenated string. |
| 1388 | */ |
| 1389 | xmlChar * |
| 1390 | xmlStrcat(xmlChar *cur, const xmlChar *add) { |
| 1391 | const xmlChar *p = add; |
| 1392 | |
| 1393 | if (add == NULL) return(cur); |
| 1394 | if (cur == NULL) |
| 1395 | return(xmlStrdup(add)); |
| 1396 | |
| 1397 | while (*p != 0) p++; /* non input consuming */ |
| 1398 | return(xmlStrncat(cur, add, p - add)); |
| 1399 | } |
| 1400 | |
| 1401 | /************************************************************************ |
| 1402 | * * |
| 1403 | * Commodity functions, cleanup needed ? * |
| 1404 | * * |
| 1405 | ************************************************************************/ |
| 1406 | |
| 1407 | /** |
| 1408 | * areBlanks: |
| 1409 | * @ctxt: an XML parser context |
| 1410 | * @str: a xmlChar * |
| 1411 | * @len: the size of @str |
| 1412 | * |
| 1413 | * Is this a sequence of blank chars that one can ignore ? |
| 1414 | * |
| 1415 | * Returns 1 if ignorable 0 otherwise. |
| 1416 | */ |
| 1417 | |
| 1418 | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) { |
| 1419 | int i, ret; |
| 1420 | xmlNodePtr lastChild; |
| 1421 | |
Daniel Veillard | 2f36224 | 2001-03-02 17:36:21 +0000 | [diff] [blame] | 1422 | if (ctxt->keepBlanks) |
| 1423 | return(0); |
| 1424 | |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1425 | /* |
| 1426 | * Check for xml:space value. |
| 1427 | */ |
| 1428 | if (*(ctxt->space) == 1) |
| 1429 | return(0); |
| 1430 | |
| 1431 | /* |
| 1432 | * Check that the string is made of blanks |
| 1433 | */ |
| 1434 | for (i = 0;i < len;i++) |
| 1435 | if (!(IS_BLANK(str[i]))) return(0); |
| 1436 | |
| 1437 | /* |
| 1438 | * Look if the element is mixed content in the Dtd if available |
| 1439 | */ |
| 1440 | if (ctxt->myDoc != NULL) { |
| 1441 | ret = xmlIsMixedElement(ctxt->myDoc, ctxt->node->name); |
| 1442 | if (ret == 0) return(1); |
| 1443 | if (ret == 1) return(0); |
| 1444 | } |
| 1445 | |
| 1446 | /* |
| 1447 | * Otherwise, heuristic :-\ |
| 1448 | */ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1449 | if (RAW != '<') return(0); |
| 1450 | if (ctxt->node == NULL) return(0); |
| 1451 | if ((ctxt->node->children == NULL) && |
| 1452 | (RAW == '<') && (NXT(1) == '/')) return(0); |
| 1453 | |
| 1454 | lastChild = xmlGetLastChild(ctxt->node); |
| 1455 | if (lastChild == NULL) { |
| 1456 | if (ctxt->node->content != NULL) return(0); |
| 1457 | } else if (xmlNodeIsText(lastChild)) |
| 1458 | return(0); |
| 1459 | else if ((ctxt->node->children != NULL) && |
| 1460 | (xmlNodeIsText(ctxt->node->children))) |
| 1461 | return(0); |
| 1462 | return(1); |
| 1463 | } |
| 1464 | |
| 1465 | /* |
| 1466 | * Forward definition for recusive behaviour. |
| 1467 | */ |
| 1468 | void xmlParsePEReference(xmlParserCtxtPtr ctxt); |
| 1469 | void xmlParseReference(xmlParserCtxtPtr ctxt); |
| 1470 | |
| 1471 | /************************************************************************ |
| 1472 | * * |
| 1473 | * Extra stuff for namespace support * |
| 1474 | * Relates to http://www.w3.org/TR/WD-xml-names * |
| 1475 | * * |
| 1476 | ************************************************************************/ |
| 1477 | |
| 1478 | /** |
| 1479 | * xmlSplitQName: |
| 1480 | * @ctxt: an XML parser context |
| 1481 | * @name: an XML parser context |
| 1482 | * @prefix: a xmlChar ** |
| 1483 | * |
| 1484 | * parse an UTF8 encoded XML qualified name string |
| 1485 | * |
| 1486 | * [NS 5] QName ::= (Prefix ':')? LocalPart |
| 1487 | * |
| 1488 | * [NS 6] Prefix ::= NCName |
| 1489 | * |
| 1490 | * [NS 7] LocalPart ::= NCName |
| 1491 | * |
| 1492 | * Returns the local part, and prefix is updated |
| 1493 | * to get the Prefix if any. |
| 1494 | */ |
| 1495 | |
| 1496 | xmlChar * |
| 1497 | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefix) { |
| 1498 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
| 1499 | xmlChar *buffer = NULL; |
| 1500 | int len = 0; |
| 1501 | int max = XML_MAX_NAMELEN; |
| 1502 | xmlChar *ret = NULL; |
| 1503 | const xmlChar *cur = name; |
| 1504 | int c; |
| 1505 | |
| 1506 | *prefix = NULL; |
| 1507 | |
| 1508 | #ifndef XML_XML_NAMESPACE |
| 1509 | /* xml: prefix is not really a namespace */ |
| 1510 | if ((cur[0] == 'x') && (cur[1] == 'm') && |
| 1511 | (cur[2] == 'l') && (cur[3] == ':')) |
| 1512 | return(xmlStrdup(name)); |
| 1513 | #endif |
| 1514 | |
| 1515 | /* nasty but valid */ |
| 1516 | if (cur[0] == ':') |
| 1517 | return(xmlStrdup(name)); |
| 1518 | |
| 1519 | c = *cur++; |
| 1520 | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
| 1521 | buf[len++] = c; |
| 1522 | c = *cur++; |
| 1523 | } |
| 1524 | if (len >= max) { |
| 1525 | /* |
| 1526 | * Okay someone managed to make a huge name, so he's ready to pay |
| 1527 | * for the processing speed. |
| 1528 | */ |
| 1529 | max = len * 2; |
| 1530 | |
| 1531 | buffer = (xmlChar *) xmlMalloc(max * sizeof(xmlChar)); |
| 1532 | if (buffer == NULL) { |
| 1533 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1534 | ctxt->sax->error(ctxt->userData, |
| 1535 | "xmlSplitQName: out of memory\n"); |
| 1536 | return(NULL); |
| 1537 | } |
| 1538 | memcpy(buffer, buf, len); |
| 1539 | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
| 1540 | if (len + 10 > max) { |
| 1541 | max *= 2; |
| 1542 | buffer = (xmlChar *) xmlRealloc(buffer, |
| 1543 | max * sizeof(xmlChar)); |
| 1544 | if (buffer == NULL) { |
| 1545 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1546 | ctxt->sax->error(ctxt->userData, |
| 1547 | "xmlSplitQName: out of memory\n"); |
| 1548 | return(NULL); |
| 1549 | } |
| 1550 | } |
| 1551 | buffer[len++] = c; |
| 1552 | c = *cur++; |
| 1553 | } |
| 1554 | buffer[len] = 0; |
| 1555 | } |
| 1556 | |
| 1557 | if (buffer == NULL) |
| 1558 | ret = xmlStrndup(buf, len); |
| 1559 | else { |
| 1560 | ret = buffer; |
| 1561 | buffer = NULL; |
| 1562 | max = XML_MAX_NAMELEN; |
| 1563 | } |
| 1564 | |
| 1565 | |
| 1566 | if (c == ':') { |
| 1567 | c = *cur++; |
| 1568 | if (c == 0) return(ret); |
| 1569 | *prefix = ret; |
| 1570 | len = 0; |
| 1571 | |
| 1572 | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
| 1573 | buf[len++] = c; |
| 1574 | c = *cur++; |
| 1575 | } |
| 1576 | if (len >= max) { |
| 1577 | /* |
| 1578 | * Okay someone managed to make a huge name, so he's ready to pay |
| 1579 | * for the processing speed. |
| 1580 | */ |
| 1581 | max = len * 2; |
| 1582 | |
| 1583 | buffer = (xmlChar *) xmlMalloc(max * sizeof(xmlChar)); |
| 1584 | if (buffer == NULL) { |
| 1585 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1586 | ctxt->sax->error(ctxt->userData, |
| 1587 | "xmlSplitQName: out of memory\n"); |
| 1588 | return(NULL); |
| 1589 | } |
| 1590 | memcpy(buffer, buf, len); |
| 1591 | while (c != 0) { /* tested bigname2.xml */ |
| 1592 | if (len + 10 > max) { |
| 1593 | max *= 2; |
| 1594 | buffer = (xmlChar *) xmlRealloc(buffer, |
| 1595 | max * sizeof(xmlChar)); |
| 1596 | if (buffer == NULL) { |
| 1597 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1598 | ctxt->sax->error(ctxt->userData, |
| 1599 | "xmlSplitQName: out of memory\n"); |
| 1600 | return(NULL); |
| 1601 | } |
| 1602 | } |
| 1603 | buffer[len++] = c; |
| 1604 | c = *cur++; |
| 1605 | } |
| 1606 | buffer[len] = 0; |
| 1607 | } |
| 1608 | |
| 1609 | if (buffer == NULL) |
| 1610 | ret = xmlStrndup(buf, len); |
| 1611 | else { |
| 1612 | ret = buffer; |
| 1613 | } |
| 1614 | } |
| 1615 | |
| 1616 | return(ret); |
| 1617 | } |
| 1618 | |
| 1619 | /************************************************************************ |
| 1620 | * * |
| 1621 | * The parser itself * |
| 1622 | * Relates to http://www.w3.org/TR/REC-xml * |
| 1623 | * * |
| 1624 | ************************************************************************/ |
| 1625 | |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 1626 | xmlChar *xmlParseNameComplex(xmlParserCtxtPtr ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1627 | /** |
| 1628 | * xmlParseName: |
| 1629 | * @ctxt: an XML parser context |
| 1630 | * |
| 1631 | * parse an XML name. |
| 1632 | * |
| 1633 | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
| 1634 | * CombiningChar | Extender |
| 1635 | * |
| 1636 | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
| 1637 | * |
| 1638 | * [6] Names ::= Name (S Name)* |
| 1639 | * |
| 1640 | * Returns the Name parsed or NULL |
| 1641 | */ |
| 1642 | |
| 1643 | xmlChar * |
| 1644 | xmlParseName(xmlParserCtxtPtr ctxt) { |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 1645 | const xmlChar *in; |
| 1646 | xmlChar *ret; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1647 | int count = 0; |
| 1648 | |
| 1649 | GROW; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 1650 | |
| 1651 | /* |
| 1652 | * Accelerator for simple ASCII names |
| 1653 | */ |
| 1654 | in = ctxt->input->cur; |
| 1655 | if (((*in >= 0x61) && (*in <= 0x7A)) || |
| 1656 | ((*in >= 0x41) && (*in <= 0x5A)) || |
| 1657 | (*in == '_') || (*in == ':')) { |
| 1658 | in++; |
| 1659 | while (((*in >= 0x61) && (*in <= 0x7A)) || |
| 1660 | ((*in >= 0x41) && (*in <= 0x5A)) || |
| 1661 | ((*in >= 0x30) && (*in <= 0x39)) || |
| 1662 | (*in == '_') || (*in == ':')) |
| 1663 | in++; |
| 1664 | if ((*in == ' ') || (*in == '>') || (*in == '/')) { |
| 1665 | count = in - ctxt->input->cur; |
| 1666 | ret = xmlStrndup(ctxt->input->cur, count); |
| 1667 | ctxt->input->cur = in; |
| 1668 | return(ret); |
| 1669 | } |
| 1670 | } |
Daniel Veillard | 2f36224 | 2001-03-02 17:36:21 +0000 | [diff] [blame] | 1671 | return(xmlParseNameComplex(ctxt)); |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 1672 | } |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 1673 | |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 1674 | xmlChar * |
| 1675 | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
| 1676 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
| 1677 | int len = 0, l; |
| 1678 | int c; |
| 1679 | int count = 0; |
| 1680 | |
| 1681 | /* |
| 1682 | * Handler for more complex cases |
| 1683 | */ |
| 1684 | GROW; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1685 | c = CUR_CHAR(l); |
| 1686 | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
| 1687 | (!IS_LETTER(c) && (c != '_') && |
| 1688 | (c != ':'))) { |
| 1689 | return(NULL); |
| 1690 | } |
| 1691 | |
| 1692 | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
| 1693 | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
| 1694 | (c == '.') || (c == '-') || |
| 1695 | (c == '_') || (c == ':') || |
| 1696 | (IS_COMBINING(c)) || |
| 1697 | (IS_EXTENDER(c)))) { |
| 1698 | if (count++ > 100) { |
| 1699 | count = 0; |
| 1700 | GROW; |
| 1701 | } |
| 1702 | COPY_BUF(l,buf,len,c); |
| 1703 | NEXTL(l); |
| 1704 | c = CUR_CHAR(l); |
| 1705 | if (len >= XML_MAX_NAMELEN) { |
| 1706 | /* |
| 1707 | * Okay someone managed to make a huge name, so he's ready to pay |
| 1708 | * for the processing speed. |
| 1709 | */ |
| 1710 | xmlChar *buffer; |
| 1711 | int max = len * 2; |
| 1712 | |
| 1713 | buffer = (xmlChar *) xmlMalloc(max * sizeof(xmlChar)); |
| 1714 | if (buffer == NULL) { |
| 1715 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1716 | ctxt->sax->error(ctxt->userData, |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 1717 | "xmlParseNameComplex: out of memory\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1718 | return(NULL); |
| 1719 | } |
| 1720 | memcpy(buffer, buf, len); |
| 1721 | while ((IS_LETTER(c)) || (IS_DIGIT(c)) || /* test bigname.xml */ |
| 1722 | (c == '.') || (c == '-') || |
| 1723 | (c == '_') || (c == ':') || |
| 1724 | (IS_COMBINING(c)) || |
| 1725 | (IS_EXTENDER(c))) { |
| 1726 | if (count++ > 100) { |
| 1727 | count = 0; |
| 1728 | GROW; |
| 1729 | } |
| 1730 | if (len + 10 > max) { |
| 1731 | max *= 2; |
| 1732 | buffer = (xmlChar *) xmlRealloc(buffer, |
| 1733 | max * sizeof(xmlChar)); |
| 1734 | if (buffer == NULL) { |
| 1735 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1736 | ctxt->sax->error(ctxt->userData, |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 1737 | "xmlParseNameComplex: out of memory\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1738 | return(NULL); |
| 1739 | } |
| 1740 | } |
| 1741 | COPY_BUF(l,buffer,len,c); |
| 1742 | NEXTL(l); |
| 1743 | c = CUR_CHAR(l); |
| 1744 | } |
| 1745 | buffer[len] = 0; |
| 1746 | return(buffer); |
| 1747 | } |
| 1748 | } |
| 1749 | return(xmlStrndup(buf, len)); |
| 1750 | } |
| 1751 | |
| 1752 | /** |
| 1753 | * xmlParseStringName: |
| 1754 | * @ctxt: an XML parser context |
| 1755 | * @str: a pointer to the string pointer (IN/OUT) |
| 1756 | * |
| 1757 | * parse an XML name. |
| 1758 | * |
| 1759 | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
| 1760 | * CombiningChar | Extender |
| 1761 | * |
| 1762 | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
| 1763 | * |
| 1764 | * [6] Names ::= Name (S Name)* |
| 1765 | * |
| 1766 | * Returns the Name parsed or NULL. The str pointer |
| 1767 | * is updated to the current location in the string. |
| 1768 | */ |
| 1769 | |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 1770 | static xmlChar * |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1771 | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
| 1772 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
| 1773 | const xmlChar *cur = *str; |
| 1774 | int len = 0, l; |
| 1775 | int c; |
| 1776 | |
| 1777 | c = CUR_SCHAR(cur, l); |
| 1778 | if (!IS_LETTER(c) && (c != '_') && |
| 1779 | (c != ':')) { |
| 1780 | return(NULL); |
| 1781 | } |
| 1782 | |
| 1783 | while ((IS_LETTER(c)) || (IS_DIGIT(c)) || /* test bigentname.xml */ |
| 1784 | (c == '.') || (c == '-') || |
| 1785 | (c == '_') || (c == ':') || |
| 1786 | (IS_COMBINING(c)) || |
| 1787 | (IS_EXTENDER(c))) { |
| 1788 | COPY_BUF(l,buf,len,c); |
| 1789 | cur += l; |
| 1790 | c = CUR_SCHAR(cur, l); |
| 1791 | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
| 1792 | /* |
| 1793 | * Okay someone managed to make a huge name, so he's ready to pay |
| 1794 | * for the processing speed. |
| 1795 | */ |
| 1796 | xmlChar *buffer; |
| 1797 | int max = len * 2; |
| 1798 | |
| 1799 | buffer = (xmlChar *) xmlMalloc(max * sizeof(xmlChar)); |
| 1800 | if (buffer == NULL) { |
| 1801 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1802 | ctxt->sax->error(ctxt->userData, |
| 1803 | "xmlParseStringName: out of memory\n"); |
| 1804 | return(NULL); |
| 1805 | } |
| 1806 | memcpy(buffer, buf, len); |
| 1807 | while ((IS_LETTER(c)) || (IS_DIGIT(c)) || /* test bigentname.xml */ |
| 1808 | (c == '.') || (c == '-') || |
| 1809 | (c == '_') || (c == ':') || |
| 1810 | (IS_COMBINING(c)) || |
| 1811 | (IS_EXTENDER(c))) { |
| 1812 | if (len + 10 > max) { |
| 1813 | max *= 2; |
| 1814 | buffer = (xmlChar *) xmlRealloc(buffer, |
| 1815 | max * sizeof(xmlChar)); |
| 1816 | if (buffer == NULL) { |
| 1817 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1818 | ctxt->sax->error(ctxt->userData, |
| 1819 | "xmlParseStringName: out of memory\n"); |
| 1820 | return(NULL); |
| 1821 | } |
| 1822 | } |
| 1823 | COPY_BUF(l,buffer,len,c); |
| 1824 | cur += l; |
| 1825 | c = CUR_SCHAR(cur, l); |
| 1826 | } |
| 1827 | buffer[len] = 0; |
| 1828 | *str = cur; |
| 1829 | return(buffer); |
| 1830 | } |
| 1831 | } |
| 1832 | *str = cur; |
| 1833 | return(xmlStrndup(buf, len)); |
| 1834 | } |
| 1835 | |
| 1836 | /** |
| 1837 | * xmlParseNmtoken: |
| 1838 | * @ctxt: an XML parser context |
| 1839 | * |
| 1840 | * parse an XML Nmtoken. |
| 1841 | * |
| 1842 | * [7] Nmtoken ::= (NameChar)+ |
| 1843 | * |
| 1844 | * [8] Nmtokens ::= Nmtoken (S Nmtoken)* |
| 1845 | * |
| 1846 | * Returns the Nmtoken parsed or NULL |
| 1847 | */ |
| 1848 | |
| 1849 | xmlChar * |
| 1850 | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
| 1851 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
| 1852 | int len = 0, l; |
| 1853 | int c; |
| 1854 | int count = 0; |
| 1855 | |
| 1856 | GROW; |
| 1857 | c = CUR_CHAR(l); |
| 1858 | |
| 1859 | while ((IS_LETTER(c)) || (IS_DIGIT(c)) || /* test bigtoken.xml */ |
| 1860 | (c == '.') || (c == '-') || |
| 1861 | (c == '_') || (c == ':') || |
| 1862 | (IS_COMBINING(c)) || |
| 1863 | (IS_EXTENDER(c))) { |
| 1864 | if (count++ > 100) { |
| 1865 | count = 0; |
| 1866 | GROW; |
| 1867 | } |
| 1868 | COPY_BUF(l,buf,len,c); |
| 1869 | NEXTL(l); |
| 1870 | c = CUR_CHAR(l); |
| 1871 | if (len >= XML_MAX_NAMELEN) { |
| 1872 | /* |
| 1873 | * Okay someone managed to make a huge token, so he's ready to pay |
| 1874 | * for the processing speed. |
| 1875 | */ |
| 1876 | xmlChar *buffer; |
| 1877 | int max = len * 2; |
| 1878 | |
| 1879 | buffer = (xmlChar *) xmlMalloc(max * sizeof(xmlChar)); |
| 1880 | if (buffer == NULL) { |
| 1881 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1882 | ctxt->sax->error(ctxt->userData, |
| 1883 | "xmlParseNmtoken: out of memory\n"); |
| 1884 | return(NULL); |
| 1885 | } |
| 1886 | memcpy(buffer, buf, len); |
| 1887 | while ((IS_LETTER(c)) || (IS_DIGIT(c)) || /* test bigtoken.xml */ |
| 1888 | (c == '.') || (c == '-') || |
| 1889 | (c == '_') || (c == ':') || |
| 1890 | (IS_COMBINING(c)) || |
| 1891 | (IS_EXTENDER(c))) { |
| 1892 | if (count++ > 100) { |
| 1893 | count = 0; |
| 1894 | GROW; |
| 1895 | } |
| 1896 | if (len + 10 > max) { |
| 1897 | max *= 2; |
| 1898 | buffer = (xmlChar *) xmlRealloc(buffer, |
| 1899 | max * sizeof(xmlChar)); |
| 1900 | if (buffer == NULL) { |
| 1901 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1902 | ctxt->sax->error(ctxt->userData, |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 1903 | "xmlParseNameComplex: out of memory\n"); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 1904 | return(NULL); |
| 1905 | } |
| 1906 | } |
| 1907 | COPY_BUF(l,buffer,len,c); |
| 1908 | NEXTL(l); |
| 1909 | c = CUR_CHAR(l); |
| 1910 | } |
| 1911 | buffer[len] = 0; |
| 1912 | return(buffer); |
| 1913 | } |
| 1914 | } |
| 1915 | if (len == 0) |
| 1916 | return(NULL); |
| 1917 | return(xmlStrndup(buf, len)); |
| 1918 | } |
| 1919 | |
| 1920 | /** |
| 1921 | * xmlParseEntityValue: |
| 1922 | * @ctxt: an XML parser context |
| 1923 | * @orig: if non-NULL store a copy of the original entity value |
| 1924 | * |
| 1925 | * parse a value for ENTITY declarations |
| 1926 | * |
| 1927 | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
| 1928 | * "'" ([^%&'] | PEReference | Reference)* "'" |
| 1929 | * |
| 1930 | * Returns the EntityValue parsed with reference substitued or NULL |
| 1931 | */ |
| 1932 | |
| 1933 | xmlChar * |
| 1934 | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
| 1935 | xmlChar *buf = NULL; |
| 1936 | int len = 0; |
| 1937 | int size = XML_PARSER_BUFFER_SIZE; |
| 1938 | int c, l; |
| 1939 | xmlChar stop; |
| 1940 | xmlChar *ret = NULL; |
| 1941 | const xmlChar *cur = NULL; |
| 1942 | xmlParserInputPtr input; |
| 1943 | |
| 1944 | if (RAW == '"') stop = '"'; |
| 1945 | else if (RAW == '\'') stop = '\''; |
| 1946 | else { |
| 1947 | ctxt->errNo = XML_ERR_ENTITY_NOT_STARTED; |
| 1948 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 1949 | ctxt->sax->error(ctxt->userData, "EntityValue: \" or ' expected\n"); |
| 1950 | ctxt->wellFormed = 0; |
| 1951 | ctxt->disableSAX = 1; |
| 1952 | return(NULL); |
| 1953 | } |
| 1954 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 1955 | if (buf == NULL) { |
| 1956 | xmlGenericError(xmlGenericErrorContext, |
| 1957 | "malloc of %d byte failed\n", size); |
| 1958 | return(NULL); |
| 1959 | } |
| 1960 | |
| 1961 | /* |
| 1962 | * The content of the entity definition is copied in a buffer. |
| 1963 | */ |
| 1964 | |
| 1965 | ctxt->instate = XML_PARSER_ENTITY_VALUE; |
| 1966 | input = ctxt->input; |
| 1967 | GROW; |
| 1968 | NEXT; |
| 1969 | c = CUR_CHAR(l); |
| 1970 | /* |
| 1971 | * NOTE: 4.4.5 Included in Literal |
| 1972 | * When a parameter entity reference appears in a literal entity |
| 1973 | * value, ... a single or double quote character in the replacement |
| 1974 | * text is always treated as a normal data character and will not |
| 1975 | * terminate the literal. |
| 1976 | * In practice it means we stop the loop only when back at parsing |
| 1977 | * the initial entity and the quote is found |
| 1978 | */ |
| 1979 | while ((IS_CHAR(c)) && ((c != stop) || /* checked */ |
| 1980 | (ctxt->input != input))) { |
| 1981 | if (len + 5 >= size) { |
| 1982 | size *= 2; |
| 1983 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 1984 | if (buf == NULL) { |
| 1985 | xmlGenericError(xmlGenericErrorContext, |
| 1986 | "realloc of %d byte failed\n", size); |
| 1987 | return(NULL); |
| 1988 | } |
| 1989 | } |
| 1990 | COPY_BUF(l,buf,len,c); |
| 1991 | NEXTL(l); |
| 1992 | /* |
| 1993 | * Pop-up of finished entities. |
| 1994 | */ |
| 1995 | while ((RAW == 0) && (ctxt->inputNr > 1)) /* non input consuming */ |
| 1996 | xmlPopInput(ctxt); |
| 1997 | |
| 1998 | GROW; |
| 1999 | c = CUR_CHAR(l); |
| 2000 | if (c == 0) { |
| 2001 | GROW; |
| 2002 | c = CUR_CHAR(l); |
| 2003 | } |
| 2004 | } |
| 2005 | buf[len] = 0; |
| 2006 | |
| 2007 | /* |
| 2008 | * Raise problem w.r.t. '&' and '%' being used in non-entities |
| 2009 | * reference constructs. Note Charref will be handled in |
| 2010 | * xmlStringDecodeEntities() |
| 2011 | */ |
| 2012 | cur = buf; |
| 2013 | while (*cur != 0) { /* non input consuming */ |
| 2014 | if ((*cur == '%') || ((*cur == '&') && (cur[1] != '#'))) { |
| 2015 | xmlChar *name; |
| 2016 | xmlChar tmp = *cur; |
| 2017 | |
| 2018 | cur++; |
| 2019 | name = xmlParseStringName(ctxt, &cur); |
| 2020 | if ((name == NULL) || (*cur != ';')) { |
| 2021 | ctxt->errNo = XML_ERR_ENTITY_CHAR_ERROR; |
| 2022 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2023 | ctxt->sax->error(ctxt->userData, |
| 2024 | "EntityValue: '%c' forbidden except for entities references\n", |
| 2025 | tmp); |
| 2026 | ctxt->wellFormed = 0; |
| 2027 | ctxt->disableSAX = 1; |
| 2028 | } |
| 2029 | if ((ctxt->inSubset == 1) && (tmp == '%')) { |
| 2030 | ctxt->errNo = XML_ERR_ENTITY_PE_INTERNAL; |
| 2031 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2032 | ctxt->sax->error(ctxt->userData, |
| 2033 | "EntityValue: PEReferences forbidden in internal subset\n", |
| 2034 | tmp); |
| 2035 | ctxt->wellFormed = 0; |
| 2036 | ctxt->disableSAX = 1; |
| 2037 | } |
| 2038 | if (name != NULL) |
| 2039 | xmlFree(name); |
| 2040 | } |
| 2041 | cur++; |
| 2042 | } |
| 2043 | |
| 2044 | /* |
| 2045 | * Then PEReference entities are substituted. |
| 2046 | */ |
| 2047 | if (c != stop) { |
| 2048 | ctxt->errNo = XML_ERR_ENTITY_NOT_FINISHED; |
| 2049 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2050 | ctxt->sax->error(ctxt->userData, "EntityValue: \" expected\n"); |
| 2051 | ctxt->wellFormed = 0; |
| 2052 | ctxt->disableSAX = 1; |
| 2053 | xmlFree(buf); |
| 2054 | } else { |
| 2055 | NEXT; |
| 2056 | /* |
| 2057 | * NOTE: 4.4.7 Bypassed |
| 2058 | * When a general entity reference appears in the EntityValue in |
| 2059 | * an entity declaration, it is bypassed and left as is. |
| 2060 | * so XML_SUBSTITUTE_REF is not set here. |
| 2061 | */ |
| 2062 | ret = xmlStringDecodeEntities(ctxt, buf, XML_SUBSTITUTE_PEREF, |
| 2063 | 0, 0, 0); |
| 2064 | if (orig != NULL) |
| 2065 | *orig = buf; |
| 2066 | else |
| 2067 | xmlFree(buf); |
| 2068 | } |
| 2069 | |
| 2070 | return(ret); |
| 2071 | } |
| 2072 | |
| 2073 | /** |
| 2074 | * xmlParseAttValue: |
| 2075 | * @ctxt: an XML parser context |
| 2076 | * |
| 2077 | * parse a value for an attribute |
| 2078 | * Note: the parser won't do substitution of entities here, this |
| 2079 | * will be handled later in xmlStringGetNodeList |
| 2080 | * |
| 2081 | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
| 2082 | * "'" ([^<&'] | Reference)* "'" |
| 2083 | * |
| 2084 | * 3.3.3 Attribute-Value Normalization: |
| 2085 | * Before the value of an attribute is passed to the application or |
| 2086 | * checked for validity, the XML processor must normalize it as follows: |
| 2087 | * - a character reference is processed by appending the referenced |
| 2088 | * character to the attribute value |
| 2089 | * - an entity reference is processed by recursively processing the |
| 2090 | * replacement text of the entity |
| 2091 | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
| 2092 | * appending #x20 to the normalized value, except that only a single |
| 2093 | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
| 2094 | * parsed entity or the literal entity value of an internal parsed entity |
| 2095 | * - other characters are processed by appending them to the normalized value |
| 2096 | * If the declared value is not CDATA, then the XML processor must further |
| 2097 | * process the normalized attribute value by discarding any leading and |
| 2098 | * trailing space (#x20) characters, and by replacing sequences of space |
| 2099 | * (#x20) characters by a single space (#x20) character. |
| 2100 | * All attributes for which no declaration has been read should be treated |
| 2101 | * by a non-validating parser as if declared CDATA. |
| 2102 | * |
| 2103 | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
| 2104 | */ |
| 2105 | |
| 2106 | xmlChar * |
| 2107 | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
| 2108 | xmlChar limit = 0; |
| 2109 | xmlChar *buf = NULL; |
| 2110 | int len = 0; |
| 2111 | int buf_size = 0; |
| 2112 | int c, l; |
| 2113 | xmlChar *current = NULL; |
| 2114 | xmlEntityPtr ent; |
| 2115 | |
| 2116 | |
| 2117 | SHRINK; |
| 2118 | if (NXT(0) == '"') { |
| 2119 | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
| 2120 | limit = '"'; |
| 2121 | NEXT; |
| 2122 | } else if (NXT(0) == '\'') { |
| 2123 | limit = '\''; |
| 2124 | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
| 2125 | NEXT; |
| 2126 | } else { |
| 2127 | ctxt->errNo = XML_ERR_ATTRIBUTE_NOT_STARTED; |
| 2128 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2129 | ctxt->sax->error(ctxt->userData, "AttValue: \" or ' expected\n"); |
| 2130 | ctxt->wellFormed = 0; |
| 2131 | ctxt->disableSAX = 1; |
| 2132 | return(NULL); |
| 2133 | } |
| 2134 | |
| 2135 | /* |
| 2136 | * allocate a translation buffer. |
| 2137 | */ |
| 2138 | buf_size = XML_PARSER_BUFFER_SIZE; |
| 2139 | buf = (xmlChar *) xmlMalloc(buf_size * sizeof(xmlChar)); |
| 2140 | if (buf == NULL) { |
| 2141 | perror("xmlParseAttValue: malloc failed"); |
| 2142 | return(NULL); |
| 2143 | } |
| 2144 | |
| 2145 | /* |
| 2146 | * Ok loop until we reach one of the ending char or a size limit. |
| 2147 | */ |
| 2148 | c = CUR_CHAR(l); |
| 2149 | while (((NXT(0) != limit) && /* checked */ |
| 2150 | (c != '<')) || (ctxt->token != 0)) { |
| 2151 | if (c == 0) break; |
| 2152 | if (ctxt->token == '&') { |
| 2153 | /* |
| 2154 | * The reparsing will be done in xmlStringGetNodeList() |
| 2155 | * called by the attribute() function in SAX.c |
| 2156 | */ |
| 2157 | static xmlChar buffer[6] = "&"; |
| 2158 | |
| 2159 | if (len > buf_size - 10) { |
| 2160 | growBuffer(buf); |
| 2161 | } |
| 2162 | current = &buffer[0]; |
| 2163 | while (*current != 0) { /* non input consuming */ |
| 2164 | buf[len++] = *current++; |
| 2165 | } |
| 2166 | ctxt->token = 0; |
| 2167 | } else if (c == '&') { |
| 2168 | if (NXT(1) == '#') { |
| 2169 | int val = xmlParseCharRef(ctxt); |
| 2170 | if (val == '&') { |
| 2171 | /* |
| 2172 | * The reparsing will be done in xmlStringGetNodeList() |
| 2173 | * called by the attribute() function in SAX.c |
| 2174 | */ |
| 2175 | static xmlChar buffer[6] = "&"; |
| 2176 | |
| 2177 | if (len > buf_size - 10) { |
| 2178 | growBuffer(buf); |
| 2179 | } |
| 2180 | current = &buffer[0]; |
| 2181 | while (*current != 0) { /* non input consuming */ |
| 2182 | buf[len++] = *current++; |
| 2183 | } |
| 2184 | } else { |
Daniel Veillard | 0b6b55b | 2001-03-20 11:27:34 +0000 | [diff] [blame] | 2185 | if (len > buf_size - 10) { |
| 2186 | growBuffer(buf); |
| 2187 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 2188 | len += xmlCopyChar(0, &buf[len], val); |
| 2189 | } |
| 2190 | } else { |
| 2191 | ent = xmlParseEntityRef(ctxt); |
| 2192 | if ((ent != NULL) && |
| 2193 | (ctxt->replaceEntities != 0)) { |
| 2194 | xmlChar *rep; |
| 2195 | |
| 2196 | if (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) { |
| 2197 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
| 2198 | XML_SUBSTITUTE_REF, 0, 0, 0); |
| 2199 | if (rep != NULL) { |
| 2200 | current = rep; |
| 2201 | while (*current != 0) { /* non input consuming */ |
| 2202 | buf[len++] = *current++; |
| 2203 | if (len > buf_size - 10) { |
| 2204 | growBuffer(buf); |
| 2205 | } |
| 2206 | } |
| 2207 | xmlFree(rep); |
| 2208 | } |
| 2209 | } else { |
Daniel Veillard | 0b6b55b | 2001-03-20 11:27:34 +0000 | [diff] [blame] | 2210 | if (len > buf_size - 10) { |
| 2211 | growBuffer(buf); |
| 2212 | } |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 2213 | if (ent->content != NULL) |
| 2214 | buf[len++] = ent->content[0]; |
| 2215 | } |
| 2216 | } else if (ent != NULL) { |
| 2217 | int i = xmlStrlen(ent->name); |
| 2218 | const xmlChar *cur = ent->name; |
| 2219 | |
| 2220 | /* |
| 2221 | * This may look absurd but is needed to detect |
| 2222 | * entities problems |
| 2223 | */ |
| 2224 | if ((ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
| 2225 | (ent->content != NULL)) { |
| 2226 | xmlChar *rep; |
| 2227 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
| 2228 | XML_SUBSTITUTE_REF, 0, 0, 0); |
| 2229 | if (rep != NULL) |
| 2230 | xmlFree(rep); |
| 2231 | } |
| 2232 | |
| 2233 | /* |
| 2234 | * Just output the reference |
| 2235 | */ |
| 2236 | buf[len++] = '&'; |
| 2237 | if (len > buf_size - i - 10) { |
| 2238 | growBuffer(buf); |
| 2239 | } |
| 2240 | for (;i > 0;i--) |
| 2241 | buf[len++] = *cur++; |
| 2242 | buf[len++] = ';'; |
| 2243 | } |
| 2244 | } |
| 2245 | } else { |
| 2246 | if ((c == 0x20) || (c == 0xD) || (c == 0xA) || (c == 0x9)) { |
| 2247 | COPY_BUF(l,buf,len,0x20); |
| 2248 | if (len > buf_size - 10) { |
| 2249 | growBuffer(buf); |
| 2250 | } |
| 2251 | } else { |
| 2252 | COPY_BUF(l,buf,len,c); |
| 2253 | if (len > buf_size - 10) { |
| 2254 | growBuffer(buf); |
| 2255 | } |
| 2256 | } |
| 2257 | NEXTL(l); |
| 2258 | } |
| 2259 | GROW; |
| 2260 | c = CUR_CHAR(l); |
| 2261 | } |
| 2262 | buf[len++] = 0; |
| 2263 | if (RAW == '<') { |
| 2264 | ctxt->errNo = XML_ERR_LT_IN_ATTRIBUTE; |
| 2265 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2266 | ctxt->sax->error(ctxt->userData, |
| 2267 | "Unescaped '<' not allowed in attributes values\n"); |
| 2268 | ctxt->wellFormed = 0; |
| 2269 | ctxt->disableSAX = 1; |
| 2270 | } else if (RAW != limit) { |
| 2271 | ctxt->errNo = XML_ERR_ATTRIBUTE_NOT_FINISHED; |
| 2272 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2273 | ctxt->sax->error(ctxt->userData, "AttValue: ' expected\n"); |
| 2274 | ctxt->wellFormed = 0; |
| 2275 | ctxt->disableSAX = 1; |
| 2276 | } else |
| 2277 | NEXT; |
| 2278 | return(buf); |
| 2279 | } |
| 2280 | |
| 2281 | /** |
| 2282 | * xmlParseSystemLiteral: |
| 2283 | * @ctxt: an XML parser context |
| 2284 | * |
| 2285 | * parse an XML Literal |
| 2286 | * |
| 2287 | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
| 2288 | * |
| 2289 | * Returns the SystemLiteral parsed or NULL |
| 2290 | */ |
| 2291 | |
| 2292 | xmlChar * |
| 2293 | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
| 2294 | xmlChar *buf = NULL; |
| 2295 | int len = 0; |
| 2296 | int size = XML_PARSER_BUFFER_SIZE; |
| 2297 | int cur, l; |
| 2298 | xmlChar stop; |
| 2299 | int state = ctxt->instate; |
| 2300 | int count = 0; |
| 2301 | |
| 2302 | SHRINK; |
| 2303 | if (RAW == '"') { |
| 2304 | NEXT; |
| 2305 | stop = '"'; |
| 2306 | } else if (RAW == '\'') { |
| 2307 | NEXT; |
| 2308 | stop = '\''; |
| 2309 | } else { |
| 2310 | ctxt->errNo = XML_ERR_LITERAL_NOT_STARTED; |
| 2311 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2312 | ctxt->sax->error(ctxt->userData, |
| 2313 | "SystemLiteral \" or ' expected\n"); |
| 2314 | ctxt->wellFormed = 0; |
| 2315 | ctxt->disableSAX = 1; |
| 2316 | return(NULL); |
| 2317 | } |
| 2318 | |
| 2319 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 2320 | if (buf == NULL) { |
| 2321 | xmlGenericError(xmlGenericErrorContext, |
| 2322 | "malloc of %d byte failed\n", size); |
| 2323 | return(NULL); |
| 2324 | } |
| 2325 | ctxt->instate = XML_PARSER_SYSTEM_LITERAL; |
| 2326 | cur = CUR_CHAR(l); |
| 2327 | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
| 2328 | if (len + 5 >= size) { |
| 2329 | size *= 2; |
| 2330 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 2331 | if (buf == NULL) { |
| 2332 | xmlGenericError(xmlGenericErrorContext, |
| 2333 | "realloc of %d byte failed\n", size); |
| 2334 | ctxt->instate = (xmlParserInputState) state; |
| 2335 | return(NULL); |
| 2336 | } |
| 2337 | } |
| 2338 | count++; |
| 2339 | if (count > 50) { |
| 2340 | GROW; |
| 2341 | count = 0; |
| 2342 | } |
| 2343 | COPY_BUF(l,buf,len,cur); |
| 2344 | NEXTL(l); |
| 2345 | cur = CUR_CHAR(l); |
| 2346 | if (cur == 0) { |
| 2347 | GROW; |
| 2348 | SHRINK; |
| 2349 | cur = CUR_CHAR(l); |
| 2350 | } |
| 2351 | } |
| 2352 | buf[len] = 0; |
| 2353 | ctxt->instate = (xmlParserInputState) state; |
| 2354 | if (!IS_CHAR(cur)) { |
| 2355 | ctxt->errNo = XML_ERR_LITERAL_NOT_FINISHED; |
| 2356 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2357 | ctxt->sax->error(ctxt->userData, "Unfinished SystemLiteral\n"); |
| 2358 | ctxt->wellFormed = 0; |
| 2359 | ctxt->disableSAX = 1; |
| 2360 | } else { |
| 2361 | NEXT; |
| 2362 | } |
| 2363 | return(buf); |
| 2364 | } |
| 2365 | |
| 2366 | /** |
| 2367 | * xmlParsePubidLiteral: |
| 2368 | * @ctxt: an XML parser context |
| 2369 | * |
| 2370 | * parse an XML public literal |
| 2371 | * |
| 2372 | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
| 2373 | * |
| 2374 | * Returns the PubidLiteral parsed or NULL. |
| 2375 | */ |
| 2376 | |
| 2377 | xmlChar * |
| 2378 | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
| 2379 | xmlChar *buf = NULL; |
| 2380 | int len = 0; |
| 2381 | int size = XML_PARSER_BUFFER_SIZE; |
| 2382 | xmlChar cur; |
| 2383 | xmlChar stop; |
| 2384 | int count = 0; |
| 2385 | |
| 2386 | SHRINK; |
| 2387 | if (RAW == '"') { |
| 2388 | NEXT; |
| 2389 | stop = '"'; |
| 2390 | } else if (RAW == '\'') { |
| 2391 | NEXT; |
| 2392 | stop = '\''; |
| 2393 | } else { |
| 2394 | ctxt->errNo = XML_ERR_LITERAL_NOT_STARTED; |
| 2395 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2396 | ctxt->sax->error(ctxt->userData, |
| 2397 | "SystemLiteral \" or ' expected\n"); |
| 2398 | ctxt->wellFormed = 0; |
| 2399 | ctxt->disableSAX = 1; |
| 2400 | return(NULL); |
| 2401 | } |
| 2402 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 2403 | if (buf == NULL) { |
| 2404 | xmlGenericError(xmlGenericErrorContext, |
| 2405 | "malloc of %d byte failed\n", size); |
| 2406 | return(NULL); |
| 2407 | } |
| 2408 | cur = CUR; |
| 2409 | while ((IS_PUBIDCHAR(cur)) && (cur != stop)) { /* checked */ |
| 2410 | if (len + 1 >= size) { |
| 2411 | size *= 2; |
| 2412 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 2413 | if (buf == NULL) { |
| 2414 | xmlGenericError(xmlGenericErrorContext, |
| 2415 | "realloc of %d byte failed\n", size); |
| 2416 | return(NULL); |
| 2417 | } |
| 2418 | } |
| 2419 | buf[len++] = cur; |
| 2420 | count++; |
| 2421 | if (count > 50) { |
| 2422 | GROW; |
| 2423 | count = 0; |
| 2424 | } |
| 2425 | NEXT; |
| 2426 | cur = CUR; |
| 2427 | if (cur == 0) { |
| 2428 | GROW; |
| 2429 | SHRINK; |
| 2430 | cur = CUR; |
| 2431 | } |
| 2432 | } |
| 2433 | buf[len] = 0; |
| 2434 | if (cur != stop) { |
| 2435 | ctxt->errNo = XML_ERR_LITERAL_NOT_FINISHED; |
| 2436 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2437 | ctxt->sax->error(ctxt->userData, "Unfinished PubidLiteral\n"); |
| 2438 | ctxt->wellFormed = 0; |
| 2439 | ctxt->disableSAX = 1; |
| 2440 | } else { |
| 2441 | NEXT; |
| 2442 | } |
| 2443 | return(buf); |
| 2444 | } |
| 2445 | |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2446 | void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 2447 | /** |
| 2448 | * xmlParseCharData: |
| 2449 | * @ctxt: an XML parser context |
| 2450 | * @cdata: int indicating whether we are within a CDATA section |
| 2451 | * |
| 2452 | * parse a CharData section. |
| 2453 | * if we are within a CDATA section ']]>' marks an end of section. |
| 2454 | * |
| 2455 | * The right angle bracket (>) may be represented using the string ">", |
| 2456 | * and must, for compatibility, be escaped using ">" or a character |
| 2457 | * reference when it appears in the string "]]>" in content, when that |
| 2458 | * string is not marking the end of a CDATA section. |
| 2459 | * |
| 2460 | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
| 2461 | */ |
| 2462 | |
| 2463 | void |
| 2464 | xmlParseCharData(xmlParserCtxtPtr ctxt, int cdata) { |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2465 | const xmlChar *in; |
| 2466 | int nbchar = 0; |
Daniel Veillard | 5058211 | 2001-03-26 22:52:16 +0000 | [diff] [blame] | 2467 | int line = ctxt->input->line; |
| 2468 | int col = ctxt->input->col; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2469 | |
| 2470 | SHRINK; |
| 2471 | GROW; |
| 2472 | /* |
| 2473 | * Accelerated common case where input don't need to be |
| 2474 | * modified before passing it to the handler. |
| 2475 | */ |
| 2476 | if ((ctxt->token == 0) && (!cdata)) { |
| 2477 | in = ctxt->input->cur; |
| 2478 | do { |
Daniel Veillard | 3ed155f | 2001-04-29 19:56:59 +0000 | [diff] [blame] | 2479 | get_more: |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2480 | while (((*in >= 0x20) && (*in != '<') && |
| 2481 | (*in != '&') && (*in <= 0x7F)) || (*in == 0x09)) |
| 2482 | in++; |
| 2483 | if (*in == 0xA) { |
| 2484 | ctxt->input->line++; |
Daniel Veillard | 3ed155f | 2001-04-29 19:56:59 +0000 | [diff] [blame] | 2485 | in++; |
| 2486 | while (*in == 0xA) { |
| 2487 | ctxt->input->line++; |
| 2488 | in++; |
| 2489 | } |
| 2490 | goto get_more; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2491 | } |
| 2492 | nbchar = in - ctxt->input->cur; |
Daniel Veillard | 80f3257 | 2001-03-07 19:45:40 +0000 | [diff] [blame] | 2493 | if (nbchar > 0) { |
| 2494 | if (IS_BLANK(*ctxt->input->cur) && |
| 2495 | areBlanks(ctxt, ctxt->input->cur, nbchar)) { |
| 2496 | if (ctxt->sax->ignorableWhitespace != NULL) |
| 2497 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
| 2498 | ctxt->input->cur, nbchar); |
| 2499 | } else { |
| 2500 | if (ctxt->sax->characters != NULL) |
| 2501 | ctxt->sax->characters(ctxt->userData, |
| 2502 | ctxt->input->cur, nbchar); |
| 2503 | } |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2504 | } |
| 2505 | ctxt->input->cur = in; |
| 2506 | if (*in == 0xD) { |
| 2507 | in++; |
| 2508 | if (*in == 0xA) { |
| 2509 | ctxt->input->cur = in; |
| 2510 | in++; |
| 2511 | ctxt->input->line++; |
| 2512 | continue; /* while */ |
| 2513 | } |
| 2514 | in--; |
| 2515 | } |
Daniel Veillard | 80f3257 | 2001-03-07 19:45:40 +0000 | [diff] [blame] | 2516 | if (*in == '<') { |
| 2517 | return; |
| 2518 | } |
| 2519 | if (*in == '&') { |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2520 | return; |
| 2521 | } |
| 2522 | SHRINK; |
| 2523 | GROW; |
| 2524 | in = ctxt->input->cur; |
| 2525 | } while ((*in >= 0x20) && (*in <= 0x7F)); |
| 2526 | nbchar = 0; |
| 2527 | } |
Daniel Veillard | 5058211 | 2001-03-26 22:52:16 +0000 | [diff] [blame] | 2528 | ctxt->input->line = line; |
| 2529 | ctxt->input->col = col; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 2530 | xmlParseCharDataComplex(ctxt, cdata); |
| 2531 | } |
| 2532 | |
| 2533 | void |
| 2534 | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 2535 | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
| 2536 | int nbchar = 0; |
| 2537 | int cur, l; |
| 2538 | int count = 0; |
| 2539 | |
| 2540 | SHRINK; |
| 2541 | GROW; |
| 2542 | cur = CUR_CHAR(l); |
| 2543 | while (((cur != '<') || (ctxt->token == '<')) && /* checked */ |
| 2544 | ((cur != '&') || (ctxt->token == '&')) && |
| 2545 | (IS_CHAR(cur))) /* test also done in xmlCurrentChar() */ { |
| 2546 | if ((cur == ']') && (NXT(1) == ']') && |
| 2547 | (NXT(2) == '>')) { |
| 2548 | if (cdata) break; |
| 2549 | else { |
| 2550 | ctxt->errNo = XML_ERR_MISPLACED_CDATA_END; |
| 2551 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2552 | ctxt->sax->error(ctxt->userData, |
| 2553 | "Sequence ']]>' not allowed in content\n"); |
| 2554 | /* Should this be relaxed ??? I see a "must here */ |
| 2555 | ctxt->wellFormed = 0; |
| 2556 | ctxt->disableSAX = 1; |
| 2557 | } |
| 2558 | } |
| 2559 | COPY_BUF(l,buf,nbchar,cur); |
| 2560 | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
| 2561 | /* |
| 2562 | * Ok the segment is to be consumed as chars. |
| 2563 | */ |
| 2564 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
| 2565 | if (areBlanks(ctxt, buf, nbchar)) { |
| 2566 | if (ctxt->sax->ignorableWhitespace != NULL) |
| 2567 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
| 2568 | buf, nbchar); |
| 2569 | } else { |
| 2570 | if (ctxt->sax->characters != NULL) |
| 2571 | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
| 2572 | } |
| 2573 | } |
| 2574 | nbchar = 0; |
| 2575 | } |
| 2576 | count++; |
| 2577 | if (count > 50) { |
| 2578 | GROW; |
| 2579 | count = 0; |
| 2580 | } |
| 2581 | NEXTL(l); |
| 2582 | cur = CUR_CHAR(l); |
| 2583 | } |
| 2584 | if (nbchar != 0) { |
| 2585 | /* |
| 2586 | * Ok the segment is to be consumed as chars. |
| 2587 | */ |
| 2588 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
| 2589 | if (areBlanks(ctxt, buf, nbchar)) { |
| 2590 | if (ctxt->sax->ignorableWhitespace != NULL) |
| 2591 | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
| 2592 | } else { |
| 2593 | if (ctxt->sax->characters != NULL) |
| 2594 | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
| 2595 | } |
| 2596 | } |
| 2597 | } |
| 2598 | } |
| 2599 | |
| 2600 | /** |
| 2601 | * xmlParseExternalID: |
| 2602 | * @ctxt: an XML parser context |
| 2603 | * @publicID: a xmlChar** receiving PubidLiteral |
| 2604 | * @strict: indicate whether we should restrict parsing to only |
| 2605 | * production [75], see NOTE below |
| 2606 | * |
| 2607 | * Parse an External ID or a Public ID |
| 2608 | * |
| 2609 | * NOTE: Productions [75] and [83] interract badly since [75] can generate |
| 2610 | * 'PUBLIC' S PubidLiteral S SystemLiteral |
| 2611 | * |
| 2612 | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
| 2613 | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
| 2614 | * |
| 2615 | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
| 2616 | * |
| 2617 | * Returns the function returns SystemLiteral and in the second |
| 2618 | * case publicID receives PubidLiteral, is strict is off |
| 2619 | * it is possible to return NULL and have publicID set. |
| 2620 | */ |
| 2621 | |
| 2622 | xmlChar * |
| 2623 | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
| 2624 | xmlChar *URI = NULL; |
| 2625 | |
| 2626 | SHRINK; |
Daniel Veillard | 146c912 | 2001-03-22 15:22:27 +0000 | [diff] [blame] | 2627 | |
| 2628 | *publicID = NULL; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 2629 | if ((RAW == 'S') && (NXT(1) == 'Y') && |
| 2630 | (NXT(2) == 'S') && (NXT(3) == 'T') && |
| 2631 | (NXT(4) == 'E') && (NXT(5) == 'M')) { |
| 2632 | SKIP(6); |
| 2633 | if (!IS_BLANK(CUR)) { |
| 2634 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 2635 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2636 | ctxt->sax->error(ctxt->userData, |
| 2637 | "Space required after 'SYSTEM'\n"); |
| 2638 | ctxt->wellFormed = 0; |
| 2639 | ctxt->disableSAX = 1; |
| 2640 | } |
| 2641 | SKIP_BLANKS; |
| 2642 | URI = xmlParseSystemLiteral(ctxt); |
| 2643 | if (URI == NULL) { |
| 2644 | ctxt->errNo = XML_ERR_URI_REQUIRED; |
| 2645 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2646 | ctxt->sax->error(ctxt->userData, |
| 2647 | "xmlParseExternalID: SYSTEM, no URI\n"); |
| 2648 | ctxt->wellFormed = 0; |
| 2649 | ctxt->disableSAX = 1; |
| 2650 | } |
| 2651 | } else if ((RAW == 'P') && (NXT(1) == 'U') && |
| 2652 | (NXT(2) == 'B') && (NXT(3) == 'L') && |
| 2653 | (NXT(4) == 'I') && (NXT(5) == 'C')) { |
| 2654 | SKIP(6); |
| 2655 | if (!IS_BLANK(CUR)) { |
| 2656 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 2657 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2658 | ctxt->sax->error(ctxt->userData, |
| 2659 | "Space required after 'PUBLIC'\n"); |
| 2660 | ctxt->wellFormed = 0; |
| 2661 | ctxt->disableSAX = 1; |
| 2662 | } |
| 2663 | SKIP_BLANKS; |
| 2664 | *publicID = xmlParsePubidLiteral(ctxt); |
| 2665 | if (*publicID == NULL) { |
| 2666 | ctxt->errNo = XML_ERR_PUBID_REQUIRED; |
| 2667 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2668 | ctxt->sax->error(ctxt->userData, |
| 2669 | "xmlParseExternalID: PUBLIC, no Public Identifier\n"); |
| 2670 | ctxt->wellFormed = 0; |
| 2671 | ctxt->disableSAX = 1; |
| 2672 | } |
| 2673 | if (strict) { |
| 2674 | /* |
| 2675 | * We don't handle [83] so "S SystemLiteral" is required. |
| 2676 | */ |
| 2677 | if (!IS_BLANK(CUR)) { |
| 2678 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 2679 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2680 | ctxt->sax->error(ctxt->userData, |
| 2681 | "Space required after the Public Identifier\n"); |
| 2682 | ctxt->wellFormed = 0; |
| 2683 | ctxt->disableSAX = 1; |
| 2684 | } |
| 2685 | } else { |
| 2686 | /* |
| 2687 | * We handle [83] so we return immediately, if |
| 2688 | * "S SystemLiteral" is not detected. From a purely parsing |
| 2689 | * point of view that's a nice mess. |
| 2690 | */ |
| 2691 | const xmlChar *ptr; |
| 2692 | GROW; |
| 2693 | |
| 2694 | ptr = CUR_PTR; |
| 2695 | if (!IS_BLANK(*ptr)) return(NULL); |
| 2696 | |
| 2697 | while (IS_BLANK(*ptr)) ptr++; /* TODO: dangerous, fix ! */ |
| 2698 | if ((*ptr != '\'') && (*ptr != '"')) return(NULL); |
| 2699 | } |
| 2700 | SKIP_BLANKS; |
| 2701 | URI = xmlParseSystemLiteral(ctxt); |
| 2702 | if (URI == NULL) { |
| 2703 | ctxt->errNo = XML_ERR_URI_REQUIRED; |
| 2704 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2705 | ctxt->sax->error(ctxt->userData, |
| 2706 | "xmlParseExternalID: PUBLIC, no URI\n"); |
| 2707 | ctxt->wellFormed = 0; |
| 2708 | ctxt->disableSAX = 1; |
| 2709 | } |
| 2710 | } |
| 2711 | return(URI); |
| 2712 | } |
| 2713 | |
| 2714 | /** |
| 2715 | * xmlParseComment: |
| 2716 | * @ctxt: an XML parser context |
| 2717 | * |
| 2718 | * Skip an XML (SGML) comment <!-- .... --> |
| 2719 | * The spec says that "For compatibility, the string "--" (double-hyphen) |
| 2720 | * must not occur within comments. " |
| 2721 | * |
| 2722 | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
| 2723 | */ |
| 2724 | void |
| 2725 | xmlParseComment(xmlParserCtxtPtr ctxt) { |
| 2726 | xmlChar *buf = NULL; |
| 2727 | int len; |
| 2728 | int size = XML_PARSER_BUFFER_SIZE; |
| 2729 | int q, ql; |
| 2730 | int r, rl; |
| 2731 | int cur, l; |
| 2732 | xmlParserInputState state; |
| 2733 | xmlParserInputPtr input = ctxt->input; |
| 2734 | int count = 0; |
| 2735 | |
| 2736 | /* |
| 2737 | * Check that there is a comment right here. |
| 2738 | */ |
| 2739 | if ((RAW != '<') || (NXT(1) != '!') || |
| 2740 | (NXT(2) != '-') || (NXT(3) != '-')) return; |
| 2741 | |
| 2742 | state = ctxt->instate; |
| 2743 | ctxt->instate = XML_PARSER_COMMENT; |
| 2744 | SHRINK; |
| 2745 | SKIP(4); |
| 2746 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 2747 | if (buf == NULL) { |
| 2748 | xmlGenericError(xmlGenericErrorContext, |
| 2749 | "malloc of %d byte failed\n", size); |
| 2750 | ctxt->instate = state; |
| 2751 | return; |
| 2752 | } |
| 2753 | q = CUR_CHAR(ql); |
| 2754 | NEXTL(ql); |
| 2755 | r = CUR_CHAR(rl); |
| 2756 | NEXTL(rl); |
| 2757 | cur = CUR_CHAR(l); |
| 2758 | len = 0; |
| 2759 | while (IS_CHAR(cur) && /* checked */ |
| 2760 | ((cur != '>') || |
| 2761 | (r != '-') || (q != '-'))) { |
| 2762 | if ((r == '-') && (q == '-') && (len > 1)) { |
| 2763 | ctxt->errNo = XML_ERR_HYPHEN_IN_COMMENT; |
| 2764 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2765 | ctxt->sax->error(ctxt->userData, |
| 2766 | "Comment must not contain '--' (double-hyphen)`\n"); |
| 2767 | ctxt->wellFormed = 0; |
| 2768 | ctxt->disableSAX = 1; |
| 2769 | } |
| 2770 | if (len + 5 >= size) { |
| 2771 | size *= 2; |
| 2772 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 2773 | if (buf == NULL) { |
| 2774 | xmlGenericError(xmlGenericErrorContext, |
| 2775 | "realloc of %d byte failed\n", size); |
| 2776 | ctxt->instate = state; |
| 2777 | return; |
| 2778 | } |
| 2779 | } |
| 2780 | COPY_BUF(ql,buf,len,q); |
| 2781 | q = r; |
| 2782 | ql = rl; |
| 2783 | r = cur; |
| 2784 | rl = l; |
| 2785 | |
| 2786 | count++; |
| 2787 | if (count > 50) { |
| 2788 | GROW; |
| 2789 | count = 0; |
| 2790 | } |
| 2791 | NEXTL(l); |
| 2792 | cur = CUR_CHAR(l); |
| 2793 | if (cur == 0) { |
| 2794 | SHRINK; |
| 2795 | GROW; |
| 2796 | cur = CUR_CHAR(l); |
| 2797 | } |
| 2798 | } |
| 2799 | buf[len] = 0; |
| 2800 | if (!IS_CHAR(cur)) { |
| 2801 | ctxt->errNo = XML_ERR_COMMENT_NOT_FINISHED; |
| 2802 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2803 | ctxt->sax->error(ctxt->userData, |
| 2804 | "Comment not terminated \n<!--%.50s\n", buf); |
| 2805 | ctxt->wellFormed = 0; |
| 2806 | ctxt->disableSAX = 1; |
| 2807 | xmlFree(buf); |
| 2808 | } else { |
| 2809 | if (input != ctxt->input) { |
| 2810 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 2811 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2812 | ctxt->sax->error(ctxt->userData, |
| 2813 | "Comment doesn't start and stop in the same entity\n"); |
| 2814 | ctxt->wellFormed = 0; |
| 2815 | ctxt->disableSAX = 1; |
| 2816 | } |
| 2817 | NEXT; |
| 2818 | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
| 2819 | (!ctxt->disableSAX)) |
| 2820 | ctxt->sax->comment(ctxt->userData, buf); |
| 2821 | xmlFree(buf); |
| 2822 | } |
| 2823 | ctxt->instate = state; |
| 2824 | } |
| 2825 | |
| 2826 | /** |
| 2827 | * xmlParsePITarget: |
| 2828 | * @ctxt: an XML parser context |
| 2829 | * |
| 2830 | * parse the name of a PI |
| 2831 | * |
| 2832 | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
| 2833 | * |
| 2834 | * Returns the PITarget name or NULL |
| 2835 | */ |
| 2836 | |
| 2837 | xmlChar * |
| 2838 | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
| 2839 | xmlChar *name; |
| 2840 | |
| 2841 | name = xmlParseName(ctxt); |
| 2842 | if ((name != NULL) && |
| 2843 | ((name[0] == 'x') || (name[0] == 'X')) && |
| 2844 | ((name[1] == 'm') || (name[1] == 'M')) && |
| 2845 | ((name[2] == 'l') || (name[2] == 'L'))) { |
| 2846 | int i; |
| 2847 | if ((name[0] == 'x') && (name[1] == 'm') && |
| 2848 | (name[2] == 'l') && (name[3] == 0)) { |
| 2849 | ctxt->errNo = XML_ERR_RESERVED_XML_NAME; |
| 2850 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2851 | ctxt->sax->error(ctxt->userData, |
| 2852 | "XML declaration allowed only at the start of the document\n"); |
| 2853 | ctxt->wellFormed = 0; |
| 2854 | ctxt->disableSAX = 1; |
| 2855 | return(name); |
| 2856 | } else if (name[3] == 0) { |
| 2857 | ctxt->errNo = XML_ERR_RESERVED_XML_NAME; |
| 2858 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2859 | ctxt->sax->error(ctxt->userData, "Invalid PI name\n"); |
| 2860 | ctxt->wellFormed = 0; |
| 2861 | ctxt->disableSAX = 1; |
| 2862 | return(name); |
| 2863 | } |
| 2864 | for (i = 0;;i++) { |
| 2865 | if (xmlW3CPIs[i] == NULL) break; |
| 2866 | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
| 2867 | return(name); |
| 2868 | } |
| 2869 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) { |
| 2870 | ctxt->errNo = XML_ERR_RESERVED_XML_NAME; |
| 2871 | ctxt->sax->warning(ctxt->userData, |
| 2872 | "xmlParsePItarget: invalid name prefix 'xml'\n"); |
| 2873 | } |
| 2874 | } |
| 2875 | return(name); |
| 2876 | } |
| 2877 | |
| 2878 | /** |
| 2879 | * xmlParsePI: |
| 2880 | * @ctxt: an XML parser context |
| 2881 | * |
| 2882 | * parse an XML Processing Instruction. |
| 2883 | * |
| 2884 | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
| 2885 | * |
| 2886 | * The processing is transfered to SAX once parsed. |
| 2887 | */ |
| 2888 | |
| 2889 | void |
| 2890 | xmlParsePI(xmlParserCtxtPtr ctxt) { |
| 2891 | xmlChar *buf = NULL; |
| 2892 | int len = 0; |
| 2893 | int size = XML_PARSER_BUFFER_SIZE; |
| 2894 | int cur, l; |
| 2895 | xmlChar *target; |
| 2896 | xmlParserInputState state; |
| 2897 | int count = 0; |
| 2898 | |
| 2899 | if ((RAW == '<') && (NXT(1) == '?')) { |
| 2900 | xmlParserInputPtr input = ctxt->input; |
| 2901 | state = ctxt->instate; |
| 2902 | ctxt->instate = XML_PARSER_PI; |
| 2903 | /* |
| 2904 | * this is a Processing Instruction. |
| 2905 | */ |
| 2906 | SKIP(2); |
| 2907 | SHRINK; |
| 2908 | |
| 2909 | /* |
| 2910 | * Parse the target name and check for special support like |
| 2911 | * namespace. |
| 2912 | */ |
| 2913 | target = xmlParsePITarget(ctxt); |
| 2914 | if (target != NULL) { |
| 2915 | if ((RAW == '?') && (NXT(1) == '>')) { |
| 2916 | if (input != ctxt->input) { |
| 2917 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 2918 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2919 | ctxt->sax->error(ctxt->userData, |
| 2920 | "PI declaration doesn't start and stop in the same entity\n"); |
| 2921 | ctxt->wellFormed = 0; |
| 2922 | ctxt->disableSAX = 1; |
| 2923 | } |
| 2924 | SKIP(2); |
| 2925 | |
| 2926 | /* |
| 2927 | * SAX: PI detected. |
| 2928 | */ |
| 2929 | if ((ctxt->sax) && (!ctxt->disableSAX) && |
| 2930 | (ctxt->sax->processingInstruction != NULL)) |
| 2931 | ctxt->sax->processingInstruction(ctxt->userData, |
| 2932 | target, NULL); |
| 2933 | ctxt->instate = state; |
| 2934 | xmlFree(target); |
| 2935 | return; |
| 2936 | } |
| 2937 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 2938 | if (buf == NULL) { |
| 2939 | xmlGenericError(xmlGenericErrorContext, |
| 2940 | "malloc of %d byte failed\n", size); |
| 2941 | ctxt->instate = state; |
| 2942 | return; |
| 2943 | } |
| 2944 | cur = CUR; |
| 2945 | if (!IS_BLANK(cur)) { |
| 2946 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 2947 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2948 | ctxt->sax->error(ctxt->userData, |
| 2949 | "xmlParsePI: PI %s space expected\n", target); |
| 2950 | ctxt->wellFormed = 0; |
| 2951 | ctxt->disableSAX = 1; |
| 2952 | } |
| 2953 | SKIP_BLANKS; |
| 2954 | cur = CUR_CHAR(l); |
| 2955 | while (IS_CHAR(cur) && /* checked */ |
| 2956 | ((cur != '?') || (NXT(1) != '>'))) { |
| 2957 | if (len + 5 >= size) { |
| 2958 | size *= 2; |
| 2959 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 2960 | if (buf == NULL) { |
| 2961 | xmlGenericError(xmlGenericErrorContext, |
| 2962 | "realloc of %d byte failed\n", size); |
| 2963 | ctxt->instate = state; |
| 2964 | return; |
| 2965 | } |
| 2966 | } |
| 2967 | count++; |
| 2968 | if (count > 50) { |
| 2969 | GROW; |
| 2970 | count = 0; |
| 2971 | } |
| 2972 | COPY_BUF(l,buf,len,cur); |
| 2973 | NEXTL(l); |
| 2974 | cur = CUR_CHAR(l); |
| 2975 | if (cur == 0) { |
| 2976 | SHRINK; |
| 2977 | GROW; |
| 2978 | cur = CUR_CHAR(l); |
| 2979 | } |
| 2980 | } |
| 2981 | buf[len] = 0; |
| 2982 | if (cur != '?') { |
| 2983 | ctxt->errNo = XML_ERR_PI_NOT_FINISHED; |
| 2984 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2985 | ctxt->sax->error(ctxt->userData, |
| 2986 | "xmlParsePI: PI %s never end ...\n", target); |
| 2987 | ctxt->wellFormed = 0; |
| 2988 | ctxt->disableSAX = 1; |
| 2989 | } else { |
| 2990 | if (input != ctxt->input) { |
| 2991 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 2992 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 2993 | ctxt->sax->error(ctxt->userData, |
| 2994 | "PI declaration doesn't start and stop in the same entity\n"); |
| 2995 | ctxt->wellFormed = 0; |
| 2996 | ctxt->disableSAX = 1; |
| 2997 | } |
| 2998 | SKIP(2); |
| 2999 | |
| 3000 | /* |
| 3001 | * SAX: PI detected. |
| 3002 | */ |
| 3003 | if ((ctxt->sax) && (!ctxt->disableSAX) && |
| 3004 | (ctxt->sax->processingInstruction != NULL)) |
| 3005 | ctxt->sax->processingInstruction(ctxt->userData, |
| 3006 | target, buf); |
| 3007 | } |
| 3008 | xmlFree(buf); |
| 3009 | xmlFree(target); |
| 3010 | } else { |
| 3011 | ctxt->errNo = XML_ERR_PI_NOT_STARTED; |
| 3012 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3013 | ctxt->sax->error(ctxt->userData, |
| 3014 | "xmlParsePI : no target name\n"); |
| 3015 | ctxt->wellFormed = 0; |
| 3016 | ctxt->disableSAX = 1; |
| 3017 | } |
| 3018 | ctxt->instate = state; |
| 3019 | } |
| 3020 | } |
| 3021 | |
| 3022 | /** |
| 3023 | * xmlParseNotationDecl: |
| 3024 | * @ctxt: an XML parser context |
| 3025 | * |
| 3026 | * parse a notation declaration |
| 3027 | * |
| 3028 | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
| 3029 | * |
| 3030 | * Hence there is actually 3 choices: |
| 3031 | * 'PUBLIC' S PubidLiteral |
| 3032 | * 'PUBLIC' S PubidLiteral S SystemLiteral |
| 3033 | * and 'SYSTEM' S SystemLiteral |
| 3034 | * |
| 3035 | * See the NOTE on xmlParseExternalID(). |
| 3036 | */ |
| 3037 | |
| 3038 | void |
| 3039 | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
| 3040 | xmlChar *name; |
| 3041 | xmlChar *Pubid; |
| 3042 | xmlChar *Systemid; |
| 3043 | |
| 3044 | if ((RAW == '<') && (NXT(1) == '!') && |
| 3045 | (NXT(2) == 'N') && (NXT(3) == 'O') && |
| 3046 | (NXT(4) == 'T') && (NXT(5) == 'A') && |
| 3047 | (NXT(6) == 'T') && (NXT(7) == 'I') && |
| 3048 | (NXT(8) == 'O') && (NXT(9) == 'N')) { |
| 3049 | xmlParserInputPtr input = ctxt->input; |
| 3050 | SHRINK; |
| 3051 | SKIP(10); |
| 3052 | if (!IS_BLANK(CUR)) { |
| 3053 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3054 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3055 | ctxt->sax->error(ctxt->userData, |
| 3056 | "Space required after '<!NOTATION'\n"); |
| 3057 | ctxt->wellFormed = 0; |
| 3058 | ctxt->disableSAX = 1; |
| 3059 | return; |
| 3060 | } |
| 3061 | SKIP_BLANKS; |
| 3062 | |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3063 | name = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3064 | if (name == NULL) { |
| 3065 | ctxt->errNo = XML_ERR_NOTATION_NOT_STARTED; |
| 3066 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3067 | ctxt->sax->error(ctxt->userData, |
| 3068 | "NOTATION: Name expected here\n"); |
| 3069 | ctxt->wellFormed = 0; |
| 3070 | ctxt->disableSAX = 1; |
| 3071 | return; |
| 3072 | } |
| 3073 | if (!IS_BLANK(CUR)) { |
| 3074 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3075 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3076 | ctxt->sax->error(ctxt->userData, |
| 3077 | "Space required after the NOTATION name'\n"); |
| 3078 | ctxt->wellFormed = 0; |
| 3079 | ctxt->disableSAX = 1; |
| 3080 | return; |
| 3081 | } |
| 3082 | SKIP_BLANKS; |
| 3083 | |
| 3084 | /* |
| 3085 | * Parse the IDs. |
| 3086 | */ |
| 3087 | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
| 3088 | SKIP_BLANKS; |
| 3089 | |
| 3090 | if (RAW == '>') { |
| 3091 | if (input != ctxt->input) { |
| 3092 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 3093 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3094 | ctxt->sax->error(ctxt->userData, |
| 3095 | "Notation declaration doesn't start and stop in the same entity\n"); |
| 3096 | ctxt->wellFormed = 0; |
| 3097 | ctxt->disableSAX = 1; |
| 3098 | } |
| 3099 | NEXT; |
| 3100 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 3101 | (ctxt->sax->notationDecl != NULL)) |
| 3102 | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
| 3103 | } else { |
| 3104 | ctxt->errNo = XML_ERR_NOTATION_NOT_FINISHED; |
| 3105 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3106 | ctxt->sax->error(ctxt->userData, |
| 3107 | "'>' required to close NOTATION declaration\n"); |
| 3108 | ctxt->wellFormed = 0; |
| 3109 | ctxt->disableSAX = 1; |
| 3110 | } |
| 3111 | xmlFree(name); |
| 3112 | if (Systemid != NULL) xmlFree(Systemid); |
| 3113 | if (Pubid != NULL) xmlFree(Pubid); |
| 3114 | } |
| 3115 | } |
| 3116 | |
| 3117 | /** |
| 3118 | * xmlParseEntityDecl: |
| 3119 | * @ctxt: an XML parser context |
| 3120 | * |
| 3121 | * parse <!ENTITY declarations |
| 3122 | * |
| 3123 | * [70] EntityDecl ::= GEDecl | PEDecl |
| 3124 | * |
| 3125 | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
| 3126 | * |
| 3127 | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
| 3128 | * |
| 3129 | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
| 3130 | * |
| 3131 | * [74] PEDef ::= EntityValue | ExternalID |
| 3132 | * |
| 3133 | * [76] NDataDecl ::= S 'NDATA' S Name |
| 3134 | * |
| 3135 | * [ VC: Notation Declared ] |
| 3136 | * The Name must match the declared name of a notation. |
| 3137 | */ |
| 3138 | |
| 3139 | void |
| 3140 | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
| 3141 | xmlChar *name = NULL; |
| 3142 | xmlChar *value = NULL; |
| 3143 | xmlChar *URI = NULL, *literal = NULL; |
| 3144 | xmlChar *ndata = NULL; |
| 3145 | int isParameter = 0; |
| 3146 | xmlChar *orig = NULL; |
| 3147 | |
| 3148 | GROW; |
| 3149 | if ((RAW == '<') && (NXT(1) == '!') && |
| 3150 | (NXT(2) == 'E') && (NXT(3) == 'N') && |
| 3151 | (NXT(4) == 'T') && (NXT(5) == 'I') && |
| 3152 | (NXT(6) == 'T') && (NXT(7) == 'Y')) { |
| 3153 | xmlParserInputPtr input = ctxt->input; |
| 3154 | ctxt->instate = XML_PARSER_ENTITY_DECL; |
| 3155 | SHRINK; |
| 3156 | SKIP(8); |
| 3157 | if (!IS_BLANK(CUR)) { |
| 3158 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3159 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3160 | ctxt->sax->error(ctxt->userData, |
| 3161 | "Space required after '<!ENTITY'\n"); |
| 3162 | ctxt->wellFormed = 0; |
| 3163 | ctxt->disableSAX = 1; |
| 3164 | } |
| 3165 | SKIP_BLANKS; |
| 3166 | |
| 3167 | if (RAW == '%') { |
| 3168 | NEXT; |
| 3169 | if (!IS_BLANK(CUR)) { |
| 3170 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3171 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3172 | ctxt->sax->error(ctxt->userData, |
| 3173 | "Space required after '%'\n"); |
| 3174 | ctxt->wellFormed = 0; |
| 3175 | ctxt->disableSAX = 1; |
| 3176 | } |
| 3177 | SKIP_BLANKS; |
| 3178 | isParameter = 1; |
| 3179 | } |
| 3180 | |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3181 | name = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3182 | if (name == NULL) { |
| 3183 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 3184 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3185 | ctxt->sax->error(ctxt->userData, "xmlParseEntityDecl: no name\n"); |
| 3186 | ctxt->wellFormed = 0; |
| 3187 | ctxt->disableSAX = 1; |
| 3188 | return; |
| 3189 | } |
| 3190 | if (!IS_BLANK(CUR)) { |
| 3191 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3192 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3193 | ctxt->sax->error(ctxt->userData, |
| 3194 | "Space required after the entity name\n"); |
| 3195 | ctxt->wellFormed = 0; |
| 3196 | ctxt->disableSAX = 1; |
| 3197 | } |
| 3198 | SKIP_BLANKS; |
| 3199 | |
| 3200 | /* |
| 3201 | * handle the various case of definitions... |
| 3202 | */ |
| 3203 | if (isParameter) { |
| 3204 | if ((RAW == '"') || (RAW == '\'')) { |
| 3205 | value = xmlParseEntityValue(ctxt, &orig); |
| 3206 | if (value) { |
| 3207 | if ((ctxt->sax != NULL) && |
| 3208 | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
| 3209 | ctxt->sax->entityDecl(ctxt->userData, name, |
| 3210 | XML_INTERNAL_PARAMETER_ENTITY, |
| 3211 | NULL, NULL, value); |
| 3212 | } |
| 3213 | } else { |
| 3214 | URI = xmlParseExternalID(ctxt, &literal, 1); |
| 3215 | if ((URI == NULL) && (literal == NULL)) { |
| 3216 | ctxt->errNo = XML_ERR_VALUE_REQUIRED; |
| 3217 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3218 | ctxt->sax->error(ctxt->userData, |
| 3219 | "Entity value required\n"); |
| 3220 | ctxt->wellFormed = 0; |
| 3221 | ctxt->disableSAX = 1; |
| 3222 | } |
| 3223 | if (URI) { |
| 3224 | xmlURIPtr uri; |
| 3225 | |
| 3226 | uri = xmlParseURI((const char *) URI); |
| 3227 | if (uri == NULL) { |
| 3228 | ctxt->errNo = XML_ERR_INVALID_URI; |
| 3229 | if ((ctxt->sax != NULL) && |
| 3230 | (!ctxt->disableSAX) && |
| 3231 | (ctxt->sax->error != NULL)) |
| 3232 | ctxt->sax->error(ctxt->userData, |
| 3233 | "Invalid URI: %s\n", URI); |
| 3234 | ctxt->wellFormed = 0; |
| 3235 | } else { |
| 3236 | if (uri->fragment != NULL) { |
| 3237 | ctxt->errNo = XML_ERR_URI_FRAGMENT; |
| 3238 | if ((ctxt->sax != NULL) && |
| 3239 | (!ctxt->disableSAX) && |
| 3240 | (ctxt->sax->error != NULL)) |
| 3241 | ctxt->sax->error(ctxt->userData, |
| 3242 | "Fragment not allowed: %s\n", URI); |
| 3243 | ctxt->wellFormed = 0; |
| 3244 | } else { |
| 3245 | if ((ctxt->sax != NULL) && |
| 3246 | (!ctxt->disableSAX) && |
| 3247 | (ctxt->sax->entityDecl != NULL)) |
| 3248 | ctxt->sax->entityDecl(ctxt->userData, name, |
| 3249 | XML_EXTERNAL_PARAMETER_ENTITY, |
| 3250 | literal, URI, NULL); |
| 3251 | } |
| 3252 | xmlFreeURI(uri); |
| 3253 | } |
| 3254 | } |
| 3255 | } |
| 3256 | } else { |
| 3257 | if ((RAW == '"') || (RAW == '\'')) { |
| 3258 | value = xmlParseEntityValue(ctxt, &orig); |
| 3259 | if ((ctxt->sax != NULL) && |
| 3260 | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
| 3261 | ctxt->sax->entityDecl(ctxt->userData, name, |
| 3262 | XML_INTERNAL_GENERAL_ENTITY, |
| 3263 | NULL, NULL, value); |
| 3264 | } else { |
| 3265 | URI = xmlParseExternalID(ctxt, &literal, 1); |
| 3266 | if ((URI == NULL) && (literal == NULL)) { |
| 3267 | ctxt->errNo = XML_ERR_VALUE_REQUIRED; |
| 3268 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3269 | ctxt->sax->error(ctxt->userData, |
| 3270 | "Entity value required\n"); |
| 3271 | ctxt->wellFormed = 0; |
| 3272 | ctxt->disableSAX = 1; |
| 3273 | } |
| 3274 | if (URI) { |
| 3275 | xmlURIPtr uri; |
| 3276 | |
| 3277 | uri = xmlParseURI((const char *)URI); |
| 3278 | if (uri == NULL) { |
| 3279 | ctxt->errNo = XML_ERR_INVALID_URI; |
| 3280 | if ((ctxt->sax != NULL) && |
| 3281 | (!ctxt->disableSAX) && |
| 3282 | (ctxt->sax->error != NULL)) |
| 3283 | ctxt->sax->error(ctxt->userData, |
| 3284 | "Invalid URI: %s\n", URI); |
| 3285 | ctxt->wellFormed = 0; |
| 3286 | } else { |
| 3287 | if (uri->fragment != NULL) { |
| 3288 | ctxt->errNo = XML_ERR_URI_FRAGMENT; |
| 3289 | if ((ctxt->sax != NULL) && |
| 3290 | (!ctxt->disableSAX) && |
| 3291 | (ctxt->sax->error != NULL)) |
| 3292 | ctxt->sax->error(ctxt->userData, |
| 3293 | "Fragment not allowed: %s\n", URI); |
| 3294 | ctxt->wellFormed = 0; |
| 3295 | } |
| 3296 | xmlFreeURI(uri); |
| 3297 | } |
| 3298 | } |
| 3299 | if ((RAW != '>') && (!IS_BLANK(CUR))) { |
| 3300 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3301 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3302 | ctxt->sax->error(ctxt->userData, |
| 3303 | "Space required before 'NDATA'\n"); |
| 3304 | ctxt->wellFormed = 0; |
| 3305 | ctxt->disableSAX = 1; |
| 3306 | } |
| 3307 | SKIP_BLANKS; |
| 3308 | if ((RAW == 'N') && (NXT(1) == 'D') && |
| 3309 | (NXT(2) == 'A') && (NXT(3) == 'T') && |
| 3310 | (NXT(4) == 'A')) { |
| 3311 | SKIP(5); |
| 3312 | if (!IS_BLANK(CUR)) { |
| 3313 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3314 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3315 | ctxt->sax->error(ctxt->userData, |
| 3316 | "Space required after 'NDATA'\n"); |
| 3317 | ctxt->wellFormed = 0; |
| 3318 | ctxt->disableSAX = 1; |
| 3319 | } |
| 3320 | SKIP_BLANKS; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3321 | ndata = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3322 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 3323 | (ctxt->sax->unparsedEntityDecl != NULL)) |
| 3324 | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
| 3325 | literal, URI, ndata); |
| 3326 | } else { |
| 3327 | if ((ctxt->sax != NULL) && |
| 3328 | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
| 3329 | ctxt->sax->entityDecl(ctxt->userData, name, |
| 3330 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
| 3331 | literal, URI, NULL); |
| 3332 | } |
| 3333 | } |
| 3334 | } |
| 3335 | SKIP_BLANKS; |
| 3336 | if (RAW != '>') { |
| 3337 | ctxt->errNo = XML_ERR_ENTITY_NOT_FINISHED; |
| 3338 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3339 | ctxt->sax->error(ctxt->userData, |
| 3340 | "xmlParseEntityDecl: entity %s not terminated\n", name); |
| 3341 | ctxt->wellFormed = 0; |
| 3342 | ctxt->disableSAX = 1; |
| 3343 | } else { |
| 3344 | if (input != ctxt->input) { |
| 3345 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 3346 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3347 | ctxt->sax->error(ctxt->userData, |
| 3348 | "Entity declaration doesn't start and stop in the same entity\n"); |
| 3349 | ctxt->wellFormed = 0; |
| 3350 | ctxt->disableSAX = 1; |
| 3351 | } |
| 3352 | NEXT; |
| 3353 | } |
| 3354 | if (orig != NULL) { |
| 3355 | /* |
| 3356 | * Ugly mechanism to save the raw entity value. |
| 3357 | */ |
| 3358 | xmlEntityPtr cur = NULL; |
| 3359 | |
| 3360 | if (isParameter) { |
| 3361 | if ((ctxt->sax != NULL) && |
| 3362 | (ctxt->sax->getParameterEntity != NULL)) |
| 3363 | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
| 3364 | } else { |
| 3365 | if ((ctxt->sax != NULL) && |
| 3366 | (ctxt->sax->getEntity != NULL)) |
| 3367 | cur = ctxt->sax->getEntity(ctxt->userData, name); |
| 3368 | } |
| 3369 | if (cur != NULL) { |
| 3370 | if (cur->orig != NULL) |
| 3371 | xmlFree(orig); |
| 3372 | else |
| 3373 | cur->orig = orig; |
| 3374 | } else |
| 3375 | xmlFree(orig); |
| 3376 | } |
| 3377 | if (name != NULL) xmlFree(name); |
| 3378 | if (value != NULL) xmlFree(value); |
| 3379 | if (URI != NULL) xmlFree(URI); |
| 3380 | if (literal != NULL) xmlFree(literal); |
| 3381 | if (ndata != NULL) xmlFree(ndata); |
| 3382 | } |
| 3383 | } |
| 3384 | |
| 3385 | /** |
| 3386 | * xmlParseDefaultDecl: |
| 3387 | * @ctxt: an XML parser context |
| 3388 | * @value: Receive a possible fixed default value for the attribute |
| 3389 | * |
| 3390 | * Parse an attribute default declaration |
| 3391 | * |
| 3392 | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
| 3393 | * |
| 3394 | * [ VC: Required Attribute ] |
| 3395 | * if the default declaration is the keyword #REQUIRED, then the |
| 3396 | * attribute must be specified for all elements of the type in the |
| 3397 | * attribute-list declaration. |
| 3398 | * |
| 3399 | * [ VC: Attribute Default Legal ] |
| 3400 | * The declared default value must meet the lexical constraints of |
| 3401 | * the declared attribute type c.f. xmlValidateAttributeDecl() |
| 3402 | * |
| 3403 | * [ VC: Fixed Attribute Default ] |
| 3404 | * if an attribute has a default value declared with the #FIXED |
| 3405 | * keyword, instances of that attribute must match the default value. |
| 3406 | * |
| 3407 | * [ WFC: No < in Attribute Values ] |
| 3408 | * handled in xmlParseAttValue() |
| 3409 | * |
| 3410 | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
| 3411 | * or XML_ATTRIBUTE_FIXED. |
| 3412 | */ |
| 3413 | |
| 3414 | int |
| 3415 | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
| 3416 | int val; |
| 3417 | xmlChar *ret; |
| 3418 | |
| 3419 | *value = NULL; |
| 3420 | if ((RAW == '#') && (NXT(1) == 'R') && |
| 3421 | (NXT(2) == 'E') && (NXT(3) == 'Q') && |
| 3422 | (NXT(4) == 'U') && (NXT(5) == 'I') && |
| 3423 | (NXT(6) == 'R') && (NXT(7) == 'E') && |
| 3424 | (NXT(8) == 'D')) { |
| 3425 | SKIP(9); |
| 3426 | return(XML_ATTRIBUTE_REQUIRED); |
| 3427 | } |
| 3428 | if ((RAW == '#') && (NXT(1) == 'I') && |
| 3429 | (NXT(2) == 'M') && (NXT(3) == 'P') && |
| 3430 | (NXT(4) == 'L') && (NXT(5) == 'I') && |
| 3431 | (NXT(6) == 'E') && (NXT(7) == 'D')) { |
| 3432 | SKIP(8); |
| 3433 | return(XML_ATTRIBUTE_IMPLIED); |
| 3434 | } |
| 3435 | val = XML_ATTRIBUTE_NONE; |
| 3436 | if ((RAW == '#') && (NXT(1) == 'F') && |
| 3437 | (NXT(2) == 'I') && (NXT(3) == 'X') && |
| 3438 | (NXT(4) == 'E') && (NXT(5) == 'D')) { |
| 3439 | SKIP(6); |
| 3440 | val = XML_ATTRIBUTE_FIXED; |
| 3441 | if (!IS_BLANK(CUR)) { |
| 3442 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3443 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3444 | ctxt->sax->error(ctxt->userData, |
| 3445 | "Space required after '#FIXED'\n"); |
| 3446 | ctxt->wellFormed = 0; |
| 3447 | ctxt->disableSAX = 1; |
| 3448 | } |
| 3449 | SKIP_BLANKS; |
| 3450 | } |
| 3451 | ret = xmlParseAttValue(ctxt); |
| 3452 | ctxt->instate = XML_PARSER_DTD; |
| 3453 | if (ret == NULL) { |
| 3454 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3455 | ctxt->sax->error(ctxt->userData, |
| 3456 | "Attribute default value declaration error\n"); |
| 3457 | ctxt->wellFormed = 0; |
| 3458 | ctxt->disableSAX = 1; |
| 3459 | } else |
| 3460 | *value = ret; |
| 3461 | return(val); |
| 3462 | } |
| 3463 | |
| 3464 | /** |
| 3465 | * xmlParseNotationType: |
| 3466 | * @ctxt: an XML parser context |
| 3467 | * |
| 3468 | * parse an Notation attribute type. |
| 3469 | * |
| 3470 | * Note: the leading 'NOTATION' S part has already being parsed... |
| 3471 | * |
| 3472 | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
| 3473 | * |
| 3474 | * [ VC: Notation Attributes ] |
| 3475 | * Values of this type must match one of the notation names included |
| 3476 | * in the declaration; all notation names in the declaration must be declared. |
| 3477 | * |
| 3478 | * Returns: the notation attribute tree built while parsing |
| 3479 | */ |
| 3480 | |
| 3481 | xmlEnumerationPtr |
| 3482 | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
| 3483 | xmlChar *name; |
| 3484 | xmlEnumerationPtr ret = NULL, last = NULL, cur; |
| 3485 | |
| 3486 | if (RAW != '(') { |
| 3487 | ctxt->errNo = XML_ERR_NOTATION_NOT_STARTED; |
| 3488 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3489 | ctxt->sax->error(ctxt->userData, |
| 3490 | "'(' required to start 'NOTATION'\n"); |
| 3491 | ctxt->wellFormed = 0; |
| 3492 | ctxt->disableSAX = 1; |
| 3493 | return(NULL); |
| 3494 | } |
| 3495 | SHRINK; |
| 3496 | do { |
| 3497 | NEXT; |
| 3498 | SKIP_BLANKS; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3499 | name = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3500 | if (name == NULL) { |
| 3501 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 3502 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3503 | ctxt->sax->error(ctxt->userData, |
| 3504 | "Name expected in NOTATION declaration\n"); |
| 3505 | ctxt->wellFormed = 0; |
| 3506 | ctxt->disableSAX = 1; |
| 3507 | return(ret); |
| 3508 | } |
| 3509 | cur = xmlCreateEnumeration(name); |
| 3510 | xmlFree(name); |
| 3511 | if (cur == NULL) return(ret); |
| 3512 | if (last == NULL) ret = last = cur; |
| 3513 | else { |
| 3514 | last->next = cur; |
| 3515 | last = cur; |
| 3516 | } |
| 3517 | SKIP_BLANKS; |
| 3518 | } while (RAW == '|'); |
| 3519 | if (RAW != ')') { |
| 3520 | ctxt->errNo = XML_ERR_NOTATION_NOT_FINISHED; |
| 3521 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3522 | ctxt->sax->error(ctxt->userData, |
| 3523 | "')' required to finish NOTATION declaration\n"); |
| 3524 | ctxt->wellFormed = 0; |
| 3525 | ctxt->disableSAX = 1; |
| 3526 | if ((last != NULL) && (last != ret)) |
| 3527 | xmlFreeEnumeration(last); |
| 3528 | return(ret); |
| 3529 | } |
| 3530 | NEXT; |
| 3531 | return(ret); |
| 3532 | } |
| 3533 | |
| 3534 | /** |
| 3535 | * xmlParseEnumerationType: |
| 3536 | * @ctxt: an XML parser context |
| 3537 | * |
| 3538 | * parse an Enumeration attribute type. |
| 3539 | * |
| 3540 | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
| 3541 | * |
| 3542 | * [ VC: Enumeration ] |
| 3543 | * Values of this type must match one of the Nmtoken tokens in |
| 3544 | * the declaration |
| 3545 | * |
| 3546 | * Returns: the enumeration attribute tree built while parsing |
| 3547 | */ |
| 3548 | |
| 3549 | xmlEnumerationPtr |
| 3550 | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
| 3551 | xmlChar *name; |
| 3552 | xmlEnumerationPtr ret = NULL, last = NULL, cur; |
| 3553 | |
| 3554 | if (RAW != '(') { |
| 3555 | ctxt->errNo = XML_ERR_ATTLIST_NOT_STARTED; |
| 3556 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3557 | ctxt->sax->error(ctxt->userData, |
| 3558 | "'(' required to start ATTLIST enumeration\n"); |
| 3559 | ctxt->wellFormed = 0; |
| 3560 | ctxt->disableSAX = 1; |
| 3561 | return(NULL); |
| 3562 | } |
| 3563 | SHRINK; |
| 3564 | do { |
| 3565 | NEXT; |
| 3566 | SKIP_BLANKS; |
| 3567 | name = xmlParseNmtoken(ctxt); |
| 3568 | if (name == NULL) { |
| 3569 | ctxt->errNo = XML_ERR_NMTOKEN_REQUIRED; |
| 3570 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3571 | ctxt->sax->error(ctxt->userData, |
| 3572 | "NmToken expected in ATTLIST enumeration\n"); |
| 3573 | ctxt->wellFormed = 0; |
| 3574 | ctxt->disableSAX = 1; |
| 3575 | return(ret); |
| 3576 | } |
| 3577 | cur = xmlCreateEnumeration(name); |
| 3578 | xmlFree(name); |
| 3579 | if (cur == NULL) return(ret); |
| 3580 | if (last == NULL) ret = last = cur; |
| 3581 | else { |
| 3582 | last->next = cur; |
| 3583 | last = cur; |
| 3584 | } |
| 3585 | SKIP_BLANKS; |
| 3586 | } while (RAW == '|'); |
| 3587 | if (RAW != ')') { |
| 3588 | ctxt->errNo = XML_ERR_ATTLIST_NOT_FINISHED; |
| 3589 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3590 | ctxt->sax->error(ctxt->userData, |
| 3591 | "')' required to finish ATTLIST enumeration\n"); |
| 3592 | ctxt->wellFormed = 0; |
| 3593 | ctxt->disableSAX = 1; |
| 3594 | return(ret); |
| 3595 | } |
| 3596 | NEXT; |
| 3597 | return(ret); |
| 3598 | } |
| 3599 | |
| 3600 | /** |
| 3601 | * xmlParseEnumeratedType: |
| 3602 | * @ctxt: an XML parser context |
| 3603 | * @tree: the enumeration tree built while parsing |
| 3604 | * |
| 3605 | * parse an Enumerated attribute type. |
| 3606 | * |
| 3607 | * [57] EnumeratedType ::= NotationType | Enumeration |
| 3608 | * |
| 3609 | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
| 3610 | * |
| 3611 | * |
| 3612 | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
| 3613 | */ |
| 3614 | |
| 3615 | int |
| 3616 | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
| 3617 | if ((RAW == 'N') && (NXT(1) == 'O') && |
| 3618 | (NXT(2) == 'T') && (NXT(3) == 'A') && |
| 3619 | (NXT(4) == 'T') && (NXT(5) == 'I') && |
| 3620 | (NXT(6) == 'O') && (NXT(7) == 'N')) { |
| 3621 | SKIP(8); |
| 3622 | if (!IS_BLANK(CUR)) { |
| 3623 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3624 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3625 | ctxt->sax->error(ctxt->userData, |
| 3626 | "Space required after 'NOTATION'\n"); |
| 3627 | ctxt->wellFormed = 0; |
| 3628 | ctxt->disableSAX = 1; |
| 3629 | return(0); |
| 3630 | } |
| 3631 | SKIP_BLANKS; |
| 3632 | *tree = xmlParseNotationType(ctxt); |
| 3633 | if (*tree == NULL) return(0); |
| 3634 | return(XML_ATTRIBUTE_NOTATION); |
| 3635 | } |
| 3636 | *tree = xmlParseEnumerationType(ctxt); |
| 3637 | if (*tree == NULL) return(0); |
| 3638 | return(XML_ATTRIBUTE_ENUMERATION); |
| 3639 | } |
| 3640 | |
| 3641 | /** |
| 3642 | * xmlParseAttributeType: |
| 3643 | * @ctxt: an XML parser context |
| 3644 | * @tree: the enumeration tree built while parsing |
| 3645 | * |
| 3646 | * parse the Attribute list def for an element |
| 3647 | * |
| 3648 | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
| 3649 | * |
| 3650 | * [55] StringType ::= 'CDATA' |
| 3651 | * |
| 3652 | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
| 3653 | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
| 3654 | * |
| 3655 | * Validity constraints for attribute values syntax are checked in |
| 3656 | * xmlValidateAttributeValue() |
| 3657 | * |
| 3658 | * [ VC: ID ] |
| 3659 | * Values of type ID must match the Name production. A name must not |
| 3660 | * appear more than once in an XML document as a value of this type; |
| 3661 | * i.e., ID values must uniquely identify the elements which bear them. |
| 3662 | * |
| 3663 | * [ VC: One ID per Element Type ] |
| 3664 | * No element type may have more than one ID attribute specified. |
| 3665 | * |
| 3666 | * [ VC: ID Attribute Default ] |
| 3667 | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
| 3668 | * |
| 3669 | * [ VC: IDREF ] |
| 3670 | * Values of type IDREF must match the Name production, and values |
| 3671 | * of type IDREFS must match Names; each IDREF Name must match the value |
| 3672 | * of an ID attribute on some element in the XML document; i.e. IDREF |
| 3673 | * values must match the value of some ID attribute. |
| 3674 | * |
| 3675 | * [ VC: Entity Name ] |
| 3676 | * Values of type ENTITY must match the Name production, values |
| 3677 | * of type ENTITIES must match Names; each Entity Name must match the |
| 3678 | * name of an unparsed entity declared in the DTD. |
| 3679 | * |
| 3680 | * [ VC: Name Token ] |
| 3681 | * Values of type NMTOKEN must match the Nmtoken production; values |
| 3682 | * of type NMTOKENS must match Nmtokens. |
| 3683 | * |
| 3684 | * Returns the attribute type |
| 3685 | */ |
| 3686 | int |
| 3687 | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
| 3688 | SHRINK; |
| 3689 | if ((RAW == 'C') && (NXT(1) == 'D') && |
| 3690 | (NXT(2) == 'A') && (NXT(3) == 'T') && |
| 3691 | (NXT(4) == 'A')) { |
| 3692 | SKIP(5); |
| 3693 | return(XML_ATTRIBUTE_CDATA); |
| 3694 | } else if ((RAW == 'I') && (NXT(1) == 'D') && |
| 3695 | (NXT(2) == 'R') && (NXT(3) == 'E') && |
| 3696 | (NXT(4) == 'F') && (NXT(5) == 'S')) { |
| 3697 | SKIP(6); |
| 3698 | return(XML_ATTRIBUTE_IDREFS); |
| 3699 | } else if ((RAW == 'I') && (NXT(1) == 'D') && |
| 3700 | (NXT(2) == 'R') && (NXT(3) == 'E') && |
| 3701 | (NXT(4) == 'F')) { |
| 3702 | SKIP(5); |
| 3703 | return(XML_ATTRIBUTE_IDREF); |
| 3704 | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
| 3705 | SKIP(2); |
| 3706 | return(XML_ATTRIBUTE_ID); |
| 3707 | } else if ((RAW == 'E') && (NXT(1) == 'N') && |
| 3708 | (NXT(2) == 'T') && (NXT(3) == 'I') && |
| 3709 | (NXT(4) == 'T') && (NXT(5) == 'Y')) { |
| 3710 | SKIP(6); |
| 3711 | return(XML_ATTRIBUTE_ENTITY); |
| 3712 | } else if ((RAW == 'E') && (NXT(1) == 'N') && |
| 3713 | (NXT(2) == 'T') && (NXT(3) == 'I') && |
| 3714 | (NXT(4) == 'T') && (NXT(5) == 'I') && |
| 3715 | (NXT(6) == 'E') && (NXT(7) == 'S')) { |
| 3716 | SKIP(8); |
| 3717 | return(XML_ATTRIBUTE_ENTITIES); |
| 3718 | } else if ((RAW == 'N') && (NXT(1) == 'M') && |
| 3719 | (NXT(2) == 'T') && (NXT(3) == 'O') && |
| 3720 | (NXT(4) == 'K') && (NXT(5) == 'E') && |
| 3721 | (NXT(6) == 'N') && (NXT(7) == 'S')) { |
| 3722 | SKIP(8); |
| 3723 | return(XML_ATTRIBUTE_NMTOKENS); |
| 3724 | } else if ((RAW == 'N') && (NXT(1) == 'M') && |
| 3725 | (NXT(2) == 'T') && (NXT(3) == 'O') && |
| 3726 | (NXT(4) == 'K') && (NXT(5) == 'E') && |
| 3727 | (NXT(6) == 'N')) { |
| 3728 | SKIP(7); |
| 3729 | return(XML_ATTRIBUTE_NMTOKEN); |
| 3730 | } |
| 3731 | return(xmlParseEnumeratedType(ctxt, tree)); |
| 3732 | } |
| 3733 | |
| 3734 | /** |
| 3735 | * xmlParseAttributeListDecl: |
| 3736 | * @ctxt: an XML parser context |
| 3737 | * |
| 3738 | * : parse the Attribute list def for an element |
| 3739 | * |
| 3740 | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
| 3741 | * |
| 3742 | * [53] AttDef ::= S Name S AttType S DefaultDecl |
| 3743 | * |
| 3744 | */ |
| 3745 | void |
| 3746 | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
| 3747 | xmlChar *elemName; |
| 3748 | xmlChar *attrName; |
| 3749 | xmlEnumerationPtr tree; |
| 3750 | |
| 3751 | if ((RAW == '<') && (NXT(1) == '!') && |
| 3752 | (NXT(2) == 'A') && (NXT(3) == 'T') && |
| 3753 | (NXT(4) == 'T') && (NXT(5) == 'L') && |
| 3754 | (NXT(6) == 'I') && (NXT(7) == 'S') && |
| 3755 | (NXT(8) == 'T')) { |
| 3756 | xmlParserInputPtr input = ctxt->input; |
| 3757 | |
| 3758 | SKIP(9); |
| 3759 | if (!IS_BLANK(CUR)) { |
| 3760 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3761 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3762 | ctxt->sax->error(ctxt->userData, |
| 3763 | "Space required after '<!ATTLIST'\n"); |
| 3764 | ctxt->wellFormed = 0; |
| 3765 | ctxt->disableSAX = 1; |
| 3766 | } |
| 3767 | SKIP_BLANKS; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3768 | elemName = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3769 | if (elemName == NULL) { |
| 3770 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 3771 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3772 | ctxt->sax->error(ctxt->userData, |
| 3773 | "ATTLIST: no name for Element\n"); |
| 3774 | ctxt->wellFormed = 0; |
| 3775 | ctxt->disableSAX = 1; |
| 3776 | return; |
| 3777 | } |
| 3778 | SKIP_BLANKS; |
| 3779 | GROW; |
| 3780 | while (RAW != '>') { |
| 3781 | const xmlChar *check = CUR_PTR; |
| 3782 | int type; |
| 3783 | int def; |
| 3784 | xmlChar *defaultValue = NULL; |
| 3785 | |
| 3786 | GROW; |
| 3787 | tree = NULL; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3788 | attrName = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3789 | if (attrName == NULL) { |
| 3790 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 3791 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3792 | ctxt->sax->error(ctxt->userData, |
| 3793 | "ATTLIST: no name for Attribute\n"); |
| 3794 | ctxt->wellFormed = 0; |
| 3795 | ctxt->disableSAX = 1; |
| 3796 | break; |
| 3797 | } |
| 3798 | GROW; |
| 3799 | if (!IS_BLANK(CUR)) { |
| 3800 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3801 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3802 | ctxt->sax->error(ctxt->userData, |
| 3803 | "Space required after the attribute name\n"); |
| 3804 | ctxt->wellFormed = 0; |
| 3805 | ctxt->disableSAX = 1; |
| 3806 | if (attrName != NULL) |
| 3807 | xmlFree(attrName); |
| 3808 | if (defaultValue != NULL) |
| 3809 | xmlFree(defaultValue); |
| 3810 | break; |
| 3811 | } |
| 3812 | SKIP_BLANKS; |
| 3813 | |
| 3814 | type = xmlParseAttributeType(ctxt, &tree); |
| 3815 | if (type <= 0) { |
| 3816 | if (attrName != NULL) |
| 3817 | xmlFree(attrName); |
| 3818 | if (defaultValue != NULL) |
| 3819 | xmlFree(defaultValue); |
| 3820 | break; |
| 3821 | } |
| 3822 | |
| 3823 | GROW; |
| 3824 | if (!IS_BLANK(CUR)) { |
| 3825 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3826 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3827 | ctxt->sax->error(ctxt->userData, |
| 3828 | "Space required after the attribute type\n"); |
| 3829 | ctxt->wellFormed = 0; |
| 3830 | ctxt->disableSAX = 1; |
| 3831 | if (attrName != NULL) |
| 3832 | xmlFree(attrName); |
| 3833 | if (defaultValue != NULL) |
| 3834 | xmlFree(defaultValue); |
| 3835 | if (tree != NULL) |
| 3836 | xmlFreeEnumeration(tree); |
| 3837 | break; |
| 3838 | } |
| 3839 | SKIP_BLANKS; |
| 3840 | |
| 3841 | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
| 3842 | if (def <= 0) { |
| 3843 | if (attrName != NULL) |
| 3844 | xmlFree(attrName); |
| 3845 | if (defaultValue != NULL) |
| 3846 | xmlFree(defaultValue); |
| 3847 | if (tree != NULL) |
| 3848 | xmlFreeEnumeration(tree); |
| 3849 | break; |
| 3850 | } |
| 3851 | |
| 3852 | GROW; |
| 3853 | if (RAW != '>') { |
| 3854 | if (!IS_BLANK(CUR)) { |
| 3855 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 3856 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3857 | ctxt->sax->error(ctxt->userData, |
| 3858 | "Space required after the attribute default value\n"); |
| 3859 | ctxt->wellFormed = 0; |
| 3860 | ctxt->disableSAX = 1; |
| 3861 | if (attrName != NULL) |
| 3862 | xmlFree(attrName); |
| 3863 | if (defaultValue != NULL) |
| 3864 | xmlFree(defaultValue); |
| 3865 | if (tree != NULL) |
| 3866 | xmlFreeEnumeration(tree); |
| 3867 | break; |
| 3868 | } |
| 3869 | SKIP_BLANKS; |
| 3870 | } |
| 3871 | if (check == CUR_PTR) { |
| 3872 | ctxt->errNo = XML_ERR_INTERNAL_ERROR; |
| 3873 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3874 | ctxt->sax->error(ctxt->userData, |
| 3875 | "xmlParseAttributeListDecl: detected internal error\n"); |
| 3876 | if (attrName != NULL) |
| 3877 | xmlFree(attrName); |
| 3878 | if (defaultValue != NULL) |
| 3879 | xmlFree(defaultValue); |
| 3880 | if (tree != NULL) |
| 3881 | xmlFreeEnumeration(tree); |
| 3882 | break; |
| 3883 | } |
| 3884 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 3885 | (ctxt->sax->attributeDecl != NULL)) |
| 3886 | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
| 3887 | type, def, defaultValue, tree); |
| 3888 | if (attrName != NULL) |
| 3889 | xmlFree(attrName); |
| 3890 | if (defaultValue != NULL) |
| 3891 | xmlFree(defaultValue); |
| 3892 | GROW; |
| 3893 | } |
| 3894 | if (RAW == '>') { |
| 3895 | if (input != ctxt->input) { |
| 3896 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 3897 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3898 | ctxt->sax->error(ctxt->userData, |
| 3899 | "Attribute list declaration doesn't start and stop in the same entity\n"); |
| 3900 | ctxt->wellFormed = 0; |
| 3901 | ctxt->disableSAX = 1; |
| 3902 | } |
| 3903 | NEXT; |
| 3904 | } |
| 3905 | |
| 3906 | xmlFree(elemName); |
| 3907 | } |
| 3908 | } |
| 3909 | |
| 3910 | /** |
| 3911 | * xmlParseElementMixedContentDecl: |
| 3912 | * @ctxt: an XML parser context |
| 3913 | * |
| 3914 | * parse the declaration for a Mixed Element content |
| 3915 | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
| 3916 | * |
| 3917 | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
| 3918 | * '(' S? '#PCDATA' S? ')' |
| 3919 | * |
| 3920 | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
| 3921 | * |
| 3922 | * [ VC: No Duplicate Types ] |
| 3923 | * The same name must not appear more than once in a single |
| 3924 | * mixed-content declaration. |
| 3925 | * |
| 3926 | * returns: the list of the xmlElementContentPtr describing the element choices |
| 3927 | */ |
| 3928 | xmlElementContentPtr |
| 3929 | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt) { |
| 3930 | xmlElementContentPtr ret = NULL, cur = NULL, n; |
| 3931 | xmlChar *elem = NULL; |
| 3932 | |
| 3933 | GROW; |
| 3934 | if ((RAW == '#') && (NXT(1) == 'P') && |
| 3935 | (NXT(2) == 'C') && (NXT(3) == 'D') && |
| 3936 | (NXT(4) == 'A') && (NXT(5) == 'T') && |
| 3937 | (NXT(6) == 'A')) { |
| 3938 | SKIP(7); |
| 3939 | SKIP_BLANKS; |
| 3940 | SHRINK; |
| 3941 | if (RAW == ')') { |
| 3942 | ctxt->entity = ctxt->input; |
| 3943 | NEXT; |
| 3944 | ret = xmlNewElementContent(NULL, XML_ELEMENT_CONTENT_PCDATA); |
| 3945 | if (RAW == '*') { |
| 3946 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
| 3947 | NEXT; |
| 3948 | } |
| 3949 | return(ret); |
| 3950 | } |
| 3951 | if ((RAW == '(') || (RAW == '|')) { |
| 3952 | ret = cur = xmlNewElementContent(NULL, XML_ELEMENT_CONTENT_PCDATA); |
| 3953 | if (ret == NULL) return(NULL); |
| 3954 | } |
| 3955 | while (RAW == '|') { |
| 3956 | NEXT; |
| 3957 | if (elem == NULL) { |
| 3958 | ret = xmlNewElementContent(NULL, XML_ELEMENT_CONTENT_OR); |
| 3959 | if (ret == NULL) return(NULL); |
| 3960 | ret->c1 = cur; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 3961 | if (cur != NULL) |
| 3962 | cur->parent = ret; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3963 | cur = ret; |
| 3964 | } else { |
| 3965 | n = xmlNewElementContent(NULL, XML_ELEMENT_CONTENT_OR); |
| 3966 | if (n == NULL) return(NULL); |
| 3967 | n->c1 = xmlNewElementContent(elem, XML_ELEMENT_CONTENT_ELEMENT); |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 3968 | if (n->c1 != NULL) |
| 3969 | n->c1->parent = n; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3970 | cur->c2 = n; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 3971 | if (n != NULL) |
| 3972 | n->parent = cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3973 | cur = n; |
| 3974 | xmlFree(elem); |
| 3975 | } |
| 3976 | SKIP_BLANKS; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 3977 | elem = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3978 | if (elem == NULL) { |
| 3979 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 3980 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 3981 | ctxt->sax->error(ctxt->userData, |
| 3982 | "xmlParseElementMixedContentDecl : Name expected\n"); |
| 3983 | ctxt->wellFormed = 0; |
| 3984 | ctxt->disableSAX = 1; |
| 3985 | xmlFreeElementContent(cur); |
| 3986 | return(NULL); |
| 3987 | } |
| 3988 | SKIP_BLANKS; |
| 3989 | GROW; |
| 3990 | } |
| 3991 | if ((RAW == ')') && (NXT(1) == '*')) { |
| 3992 | if (elem != NULL) { |
| 3993 | cur->c2 = xmlNewElementContent(elem, |
| 3994 | XML_ELEMENT_CONTENT_ELEMENT); |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 3995 | if (cur->c2 != NULL) |
| 3996 | cur->c2->parent = cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 3997 | xmlFree(elem); |
| 3998 | } |
| 3999 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
| 4000 | ctxt->entity = ctxt->input; |
| 4001 | SKIP(2); |
| 4002 | } else { |
| 4003 | if (elem != NULL) xmlFree(elem); |
| 4004 | xmlFreeElementContent(ret); |
| 4005 | ctxt->errNo = XML_ERR_MIXED_NOT_STARTED; |
| 4006 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4007 | ctxt->sax->error(ctxt->userData, |
| 4008 | "xmlParseElementMixedContentDecl : '|' or ')*' expected\n"); |
| 4009 | ctxt->wellFormed = 0; |
| 4010 | ctxt->disableSAX = 1; |
| 4011 | return(NULL); |
| 4012 | } |
| 4013 | |
| 4014 | } else { |
| 4015 | ctxt->errNo = XML_ERR_PCDATA_REQUIRED; |
| 4016 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4017 | ctxt->sax->error(ctxt->userData, |
| 4018 | "xmlParseElementMixedContentDecl : '#PCDATA' expected\n"); |
| 4019 | ctxt->wellFormed = 0; |
| 4020 | ctxt->disableSAX = 1; |
| 4021 | } |
| 4022 | return(ret); |
| 4023 | } |
| 4024 | |
| 4025 | /** |
| 4026 | * xmlParseElementChildrenContentDecl: |
| 4027 | * @ctxt: an XML parser context |
| 4028 | * |
| 4029 | * parse the declaration for a Mixed Element content |
| 4030 | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
| 4031 | * |
| 4032 | * |
| 4033 | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
| 4034 | * |
| 4035 | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
| 4036 | * |
| 4037 | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
| 4038 | * |
| 4039 | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
| 4040 | * |
| 4041 | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
| 4042 | * TODO Parameter-entity replacement text must be properly nested |
| 4043 | * with parenthetized groups. That is to say, if either of the |
| 4044 | * opening or closing parentheses in a choice, seq, or Mixed |
| 4045 | * construct is contained in the replacement text for a parameter |
| 4046 | * entity, both must be contained in the same replacement text. For |
| 4047 | * interoperability, if a parameter-entity reference appears in a |
| 4048 | * choice, seq, or Mixed construct, its replacement text should not |
| 4049 | * be empty, and neither the first nor last non-blank character of |
| 4050 | * the replacement text should be a connector (| or ,). |
| 4051 | * |
| 4052 | * returns: the tree of xmlElementContentPtr describing the element |
| 4053 | * hierarchy. |
| 4054 | */ |
| 4055 | xmlElementContentPtr |
| 4056 | #ifdef VMS |
| 4057 | xmlParseElementChildrenContentD |
| 4058 | #else |
| 4059 | xmlParseElementChildrenContentDecl |
| 4060 | #endif |
| 4061 | (xmlParserCtxtPtr ctxt) { |
| 4062 | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
| 4063 | xmlChar *elem; |
| 4064 | xmlChar type = 0; |
| 4065 | |
| 4066 | SKIP_BLANKS; |
| 4067 | GROW; |
| 4068 | if (RAW == '(') { |
| 4069 | /* Recurse on first child */ |
| 4070 | NEXT; |
| 4071 | SKIP_BLANKS; |
| 4072 | cur = ret = xmlParseElementChildrenContentDecl(ctxt); |
| 4073 | SKIP_BLANKS; |
| 4074 | GROW; |
| 4075 | } else { |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 4076 | elem = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4077 | if (elem == NULL) { |
| 4078 | ctxt->errNo = XML_ERR_ELEMCONTENT_NOT_STARTED; |
| 4079 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4080 | ctxt->sax->error(ctxt->userData, |
| 4081 | "xmlParseElementChildrenContentDecl : Name or '(' expected\n"); |
| 4082 | ctxt->wellFormed = 0; |
| 4083 | ctxt->disableSAX = 1; |
| 4084 | return(NULL); |
| 4085 | } |
| 4086 | cur = ret = xmlNewElementContent(elem, XML_ELEMENT_CONTENT_ELEMENT); |
| 4087 | GROW; |
| 4088 | if (RAW == '?') { |
| 4089 | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
| 4090 | NEXT; |
| 4091 | } else if (RAW == '*') { |
| 4092 | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
| 4093 | NEXT; |
| 4094 | } else if (RAW == '+') { |
| 4095 | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
| 4096 | NEXT; |
| 4097 | } else { |
| 4098 | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
| 4099 | } |
| 4100 | xmlFree(elem); |
| 4101 | GROW; |
| 4102 | } |
| 4103 | SKIP_BLANKS; |
| 4104 | SHRINK; |
| 4105 | while (RAW != ')') { |
| 4106 | /* |
| 4107 | * Each loop we parse one separator and one element. |
| 4108 | */ |
| 4109 | if (RAW == ',') { |
| 4110 | if (type == 0) type = CUR; |
| 4111 | |
| 4112 | /* |
| 4113 | * Detect "Name | Name , Name" error |
| 4114 | */ |
| 4115 | else if (type != CUR) { |
| 4116 | ctxt->errNo = XML_ERR_SEPARATOR_REQUIRED; |
| 4117 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4118 | ctxt->sax->error(ctxt->userData, |
| 4119 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
| 4120 | type); |
| 4121 | ctxt->wellFormed = 0; |
| 4122 | ctxt->disableSAX = 1; |
| 4123 | if ((op != NULL) && (op != ret)) |
| 4124 | xmlFreeElementContent(op); |
| 4125 | if ((last != NULL) && (last != ret) && |
| 4126 | (last != ret->c1) && (last != ret->c2)) |
| 4127 | xmlFreeElementContent(last); |
| 4128 | if (ret != NULL) |
| 4129 | xmlFreeElementContent(ret); |
| 4130 | return(NULL); |
| 4131 | } |
| 4132 | NEXT; |
| 4133 | |
| 4134 | op = xmlNewElementContent(NULL, XML_ELEMENT_CONTENT_SEQ); |
| 4135 | if (op == NULL) { |
| 4136 | xmlFreeElementContent(ret); |
| 4137 | return(NULL); |
| 4138 | } |
| 4139 | if (last == NULL) { |
| 4140 | op->c1 = ret; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4141 | if (ret != NULL) |
| 4142 | ret->parent = op; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4143 | ret = cur = op; |
| 4144 | } else { |
| 4145 | cur->c2 = op; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4146 | if (op != NULL) |
| 4147 | op->parent = cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4148 | op->c1 = last; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4149 | if (last != NULL) |
| 4150 | last->parent = op; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4151 | cur =op; |
| 4152 | last = NULL; |
| 4153 | } |
| 4154 | } else if (RAW == '|') { |
| 4155 | if (type == 0) type = CUR; |
| 4156 | |
| 4157 | /* |
| 4158 | * Detect "Name , Name | Name" error |
| 4159 | */ |
| 4160 | else if (type != CUR) { |
| 4161 | ctxt->errNo = XML_ERR_SEPARATOR_REQUIRED; |
| 4162 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4163 | ctxt->sax->error(ctxt->userData, |
| 4164 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
| 4165 | type); |
| 4166 | ctxt->wellFormed = 0; |
| 4167 | ctxt->disableSAX = 1; |
| 4168 | if ((op != NULL) && (op != ret) && (op != last)) |
| 4169 | xmlFreeElementContent(op); |
| 4170 | if ((last != NULL) && (last != ret) && |
| 4171 | (last != ret->c1) && (last != ret->c2)) |
| 4172 | xmlFreeElementContent(last); |
| 4173 | if (ret != NULL) |
| 4174 | xmlFreeElementContent(ret); |
| 4175 | return(NULL); |
| 4176 | } |
| 4177 | NEXT; |
| 4178 | |
| 4179 | op = xmlNewElementContent(NULL, XML_ELEMENT_CONTENT_OR); |
| 4180 | if (op == NULL) { |
| 4181 | if ((op != NULL) && (op != ret)) |
| 4182 | xmlFreeElementContent(op); |
| 4183 | if ((last != NULL) && (last != ret) && |
| 4184 | (last != ret->c1) && (last != ret->c2)) |
| 4185 | xmlFreeElementContent(last); |
| 4186 | if (ret != NULL) |
| 4187 | xmlFreeElementContent(ret); |
| 4188 | return(NULL); |
| 4189 | } |
| 4190 | if (last == NULL) { |
| 4191 | op->c1 = ret; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4192 | if (ret != NULL) |
| 4193 | ret->parent = op; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4194 | ret = cur = op; |
| 4195 | } else { |
| 4196 | cur->c2 = op; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4197 | if (op != NULL) |
| 4198 | op->parent = cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4199 | op->c1 = last; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4200 | if (last != NULL) |
| 4201 | last->parent = op; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4202 | cur =op; |
| 4203 | last = NULL; |
| 4204 | } |
| 4205 | } else { |
| 4206 | ctxt->errNo = XML_ERR_ELEMCONTENT_NOT_FINISHED; |
| 4207 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4208 | ctxt->sax->error(ctxt->userData, |
| 4209 | "xmlParseElementChildrenContentDecl : ',' '|' or ')' expected\n"); |
| 4210 | ctxt->wellFormed = 0; |
| 4211 | ctxt->disableSAX = 1; |
| 4212 | if ((op != NULL) && (op != ret)) |
| 4213 | xmlFreeElementContent(op); |
| 4214 | if ((last != NULL) && (last != ret) && |
| 4215 | (last != ret->c1) && (last != ret->c2)) |
| 4216 | xmlFreeElementContent(last); |
| 4217 | if (ret != NULL) |
| 4218 | xmlFreeElementContent(ret); |
| 4219 | return(NULL); |
| 4220 | } |
| 4221 | GROW; |
| 4222 | SKIP_BLANKS; |
| 4223 | GROW; |
| 4224 | if (RAW == '(') { |
| 4225 | /* Recurse on second child */ |
| 4226 | NEXT; |
| 4227 | SKIP_BLANKS; |
| 4228 | last = xmlParseElementChildrenContentDecl(ctxt); |
| 4229 | SKIP_BLANKS; |
| 4230 | } else { |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 4231 | elem = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4232 | if (elem == NULL) { |
| 4233 | ctxt->errNo = XML_ERR_ELEMCONTENT_NOT_STARTED; |
| 4234 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4235 | ctxt->sax->error(ctxt->userData, |
| 4236 | "xmlParseElementChildrenContentDecl : Name or '(' expected\n"); |
| 4237 | ctxt->wellFormed = 0; |
| 4238 | ctxt->disableSAX = 1; |
| 4239 | if ((op != NULL) && (op != ret)) |
| 4240 | xmlFreeElementContent(op); |
| 4241 | if ((last != NULL) && (last != ret) && |
| 4242 | (last != ret->c1) && (last != ret->c2)) |
| 4243 | xmlFreeElementContent(last); |
| 4244 | if (ret != NULL) |
| 4245 | xmlFreeElementContent(ret); |
| 4246 | return(NULL); |
| 4247 | } |
| 4248 | last = xmlNewElementContent(elem, XML_ELEMENT_CONTENT_ELEMENT); |
| 4249 | xmlFree(elem); |
| 4250 | if (RAW == '?') { |
| 4251 | last->ocur = XML_ELEMENT_CONTENT_OPT; |
| 4252 | NEXT; |
| 4253 | } else if (RAW == '*') { |
| 4254 | last->ocur = XML_ELEMENT_CONTENT_MULT; |
| 4255 | NEXT; |
| 4256 | } else if (RAW == '+') { |
| 4257 | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
| 4258 | NEXT; |
| 4259 | } else { |
| 4260 | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
| 4261 | } |
| 4262 | } |
| 4263 | SKIP_BLANKS; |
| 4264 | GROW; |
| 4265 | } |
| 4266 | if ((cur != NULL) && (last != NULL)) { |
| 4267 | cur->c2 = last; |
Daniel Veillard | dab4cb3 | 2001-04-20 13:03:48 +0000 | [diff] [blame] | 4268 | if (last != NULL) |
| 4269 | last->parent = cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4270 | } |
| 4271 | ctxt->entity = ctxt->input; |
| 4272 | NEXT; |
| 4273 | if (RAW == '?') { |
Daniel Veillard | e470df7 | 2001-04-18 21:41:07 +0000 | [diff] [blame] | 4274 | if (ret != NULL) |
| 4275 | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4276 | NEXT; |
| 4277 | } else if (RAW == '*') { |
Daniel Veillard | e470df7 | 2001-04-18 21:41:07 +0000 | [diff] [blame] | 4278 | if (ret != NULL) |
| 4279 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4280 | NEXT; |
| 4281 | } else if (RAW == '+') { |
Daniel Veillard | e470df7 | 2001-04-18 21:41:07 +0000 | [diff] [blame] | 4282 | if (ret != NULL) |
| 4283 | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4284 | NEXT; |
| 4285 | } |
| 4286 | return(ret); |
| 4287 | } |
| 4288 | |
| 4289 | /** |
| 4290 | * xmlParseElementContentDecl: |
| 4291 | * @ctxt: an XML parser context |
| 4292 | * @name: the name of the element being defined. |
| 4293 | * @result: the Element Content pointer will be stored here if any |
| 4294 | * |
| 4295 | * parse the declaration for an Element content either Mixed or Children, |
| 4296 | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
| 4297 | * |
| 4298 | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
| 4299 | * |
| 4300 | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
| 4301 | */ |
| 4302 | |
| 4303 | int |
| 4304 | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, xmlChar *name, |
| 4305 | xmlElementContentPtr *result) { |
| 4306 | |
| 4307 | xmlElementContentPtr tree = NULL; |
| 4308 | xmlParserInputPtr input = ctxt->input; |
| 4309 | int res; |
| 4310 | |
| 4311 | *result = NULL; |
| 4312 | |
| 4313 | if (RAW != '(') { |
| 4314 | ctxt->errNo = XML_ERR_ELEMCONTENT_NOT_STARTED; |
| 4315 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4316 | ctxt->sax->error(ctxt->userData, |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4317 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4318 | ctxt->wellFormed = 0; |
| 4319 | ctxt->disableSAX = 1; |
| 4320 | return(-1); |
| 4321 | } |
| 4322 | NEXT; |
| 4323 | GROW; |
| 4324 | SKIP_BLANKS; |
| 4325 | if ((RAW == '#') && (NXT(1) == 'P') && |
| 4326 | (NXT(2) == 'C') && (NXT(3) == 'D') && |
| 4327 | (NXT(4) == 'A') && (NXT(5) == 'T') && |
| 4328 | (NXT(6) == 'A')) { |
| 4329 | tree = xmlParseElementMixedContentDecl(ctxt); |
| 4330 | res = XML_ELEMENT_TYPE_MIXED; |
| 4331 | } else { |
| 4332 | tree = xmlParseElementChildrenContentDecl(ctxt); |
| 4333 | res = XML_ELEMENT_TYPE_ELEMENT; |
| 4334 | } |
| 4335 | if ((ctxt->entity != NULL) && (input != ctxt->entity)) { |
| 4336 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 4337 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4338 | ctxt->sax->error(ctxt->userData, |
| 4339 | "Element content declaration doesn't start and stop in the same entity\n"); |
| 4340 | ctxt->wellFormed = 0; |
| 4341 | ctxt->disableSAX = 1; |
| 4342 | } |
| 4343 | SKIP_BLANKS; |
| 4344 | *result = tree; |
| 4345 | return(res); |
| 4346 | } |
| 4347 | |
| 4348 | /** |
| 4349 | * xmlParseElementDecl: |
| 4350 | * @ctxt: an XML parser context |
| 4351 | * |
| 4352 | * parse an Element declaration. |
| 4353 | * |
| 4354 | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
| 4355 | * |
| 4356 | * [ VC: Unique Element Type Declaration ] |
| 4357 | * No element type may be declared more than once |
| 4358 | * |
| 4359 | * Returns the type of the element, or -1 in case of error |
| 4360 | */ |
| 4361 | int |
| 4362 | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
| 4363 | xmlChar *name; |
| 4364 | int ret = -1; |
| 4365 | xmlElementContentPtr content = NULL; |
| 4366 | |
| 4367 | GROW; |
| 4368 | if ((RAW == '<') && (NXT(1) == '!') && |
| 4369 | (NXT(2) == 'E') && (NXT(3) == 'L') && |
| 4370 | (NXT(4) == 'E') && (NXT(5) == 'M') && |
| 4371 | (NXT(6) == 'E') && (NXT(7) == 'N') && |
| 4372 | (NXT(8) == 'T')) { |
| 4373 | xmlParserInputPtr input = ctxt->input; |
| 4374 | |
| 4375 | SKIP(9); |
| 4376 | if (!IS_BLANK(CUR)) { |
| 4377 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 4378 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4379 | ctxt->sax->error(ctxt->userData, |
| 4380 | "Space required after 'ELEMENT'\n"); |
| 4381 | ctxt->wellFormed = 0; |
| 4382 | ctxt->disableSAX = 1; |
| 4383 | } |
| 4384 | SKIP_BLANKS; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 4385 | name = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4386 | if (name == NULL) { |
| 4387 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 4388 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4389 | ctxt->sax->error(ctxt->userData, |
| 4390 | "xmlParseElementDecl: no name for Element\n"); |
| 4391 | ctxt->wellFormed = 0; |
| 4392 | ctxt->disableSAX = 1; |
| 4393 | return(-1); |
| 4394 | } |
| 4395 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 4396 | xmlPopInput(ctxt); |
| 4397 | if (!IS_BLANK(CUR)) { |
| 4398 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 4399 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4400 | ctxt->sax->error(ctxt->userData, |
| 4401 | "Space required after the element name\n"); |
| 4402 | ctxt->wellFormed = 0; |
| 4403 | ctxt->disableSAX = 1; |
| 4404 | } |
| 4405 | SKIP_BLANKS; |
| 4406 | if ((RAW == 'E') && (NXT(1) == 'M') && |
| 4407 | (NXT(2) == 'P') && (NXT(3) == 'T') && |
| 4408 | (NXT(4) == 'Y')) { |
| 4409 | SKIP(5); |
| 4410 | /* |
| 4411 | * Element must always be empty. |
| 4412 | */ |
| 4413 | ret = XML_ELEMENT_TYPE_EMPTY; |
| 4414 | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
| 4415 | (NXT(2) == 'Y')) { |
| 4416 | SKIP(3); |
| 4417 | /* |
| 4418 | * Element is a generic container. |
| 4419 | */ |
| 4420 | ret = XML_ELEMENT_TYPE_ANY; |
| 4421 | } else if (RAW == '(') { |
| 4422 | ret = xmlParseElementContentDecl(ctxt, name, &content); |
| 4423 | } else { |
| 4424 | /* |
| 4425 | * [ WFC: PEs in Internal Subset ] error handling. |
| 4426 | */ |
| 4427 | if ((RAW == '%') && (ctxt->external == 0) && |
| 4428 | (ctxt->inputNr == 1)) { |
| 4429 | ctxt->errNo = XML_ERR_PEREF_IN_INT_SUBSET; |
| 4430 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4431 | ctxt->sax->error(ctxt->userData, |
| 4432 | "PEReference: forbidden within markup decl in internal subset\n"); |
| 4433 | } else { |
| 4434 | ctxt->errNo = XML_ERR_ELEMCONTENT_NOT_STARTED; |
| 4435 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4436 | ctxt->sax->error(ctxt->userData, |
| 4437 | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
| 4438 | } |
| 4439 | ctxt->wellFormed = 0; |
| 4440 | ctxt->disableSAX = 1; |
| 4441 | if (name != NULL) xmlFree(name); |
| 4442 | return(-1); |
| 4443 | } |
| 4444 | |
| 4445 | SKIP_BLANKS; |
| 4446 | /* |
| 4447 | * Pop-up of finished entities. |
| 4448 | */ |
| 4449 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 4450 | xmlPopInput(ctxt); |
| 4451 | SKIP_BLANKS; |
| 4452 | |
| 4453 | if (RAW != '>') { |
| 4454 | ctxt->errNo = XML_ERR_GT_REQUIRED; |
| 4455 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4456 | ctxt->sax->error(ctxt->userData, |
| 4457 | "xmlParseElementDecl: expected '>' at the end\n"); |
| 4458 | ctxt->wellFormed = 0; |
| 4459 | ctxt->disableSAX = 1; |
| 4460 | } else { |
| 4461 | if (input != ctxt->input) { |
| 4462 | ctxt->errNo = XML_ERR_ENTITY_BOUNDARY; |
| 4463 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4464 | ctxt->sax->error(ctxt->userData, |
| 4465 | "Element declaration doesn't start and stop in the same entity\n"); |
| 4466 | ctxt->wellFormed = 0; |
| 4467 | ctxt->disableSAX = 1; |
| 4468 | } |
| 4469 | |
| 4470 | NEXT; |
| 4471 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 4472 | (ctxt->sax->elementDecl != NULL)) |
| 4473 | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
| 4474 | content); |
| 4475 | } |
| 4476 | if (content != NULL) { |
| 4477 | xmlFreeElementContent(content); |
| 4478 | } |
| 4479 | if (name != NULL) { |
| 4480 | xmlFree(name); |
| 4481 | } |
| 4482 | } |
| 4483 | return(ret); |
| 4484 | } |
| 4485 | |
| 4486 | /** |
| 4487 | * xmlParseMarkupDecl: |
| 4488 | * @ctxt: an XML parser context |
| 4489 | * |
| 4490 | * parse Markup declarations |
| 4491 | * |
| 4492 | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
| 4493 | * NotationDecl | PI | Comment |
| 4494 | * |
| 4495 | * [ VC: Proper Declaration/PE Nesting ] |
| 4496 | * Parameter-entity replacement text must be properly nested with |
| 4497 | * markup declarations. That is to say, if either the first character |
| 4498 | * or the last character of a markup declaration (markupdecl above) is |
| 4499 | * contained in the replacement text for a parameter-entity reference, |
| 4500 | * both must be contained in the same replacement text. |
| 4501 | * |
| 4502 | * [ WFC: PEs in Internal Subset ] |
| 4503 | * In the internal DTD subset, parameter-entity references can occur |
| 4504 | * only where markup declarations can occur, not within markup declarations. |
| 4505 | * (This does not apply to references that occur in external parameter |
| 4506 | * entities or to the external subset.) |
| 4507 | */ |
| 4508 | void |
| 4509 | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
| 4510 | GROW; |
| 4511 | xmlParseElementDecl(ctxt); |
| 4512 | xmlParseAttributeListDecl(ctxt); |
| 4513 | xmlParseEntityDecl(ctxt); |
| 4514 | xmlParseNotationDecl(ctxt); |
| 4515 | xmlParsePI(ctxt); |
| 4516 | xmlParseComment(ctxt); |
| 4517 | /* |
| 4518 | * This is only for internal subset. On external entities, |
| 4519 | * the replacement is done before parsing stage |
| 4520 | */ |
| 4521 | if ((ctxt->external == 0) && (ctxt->inputNr == 1)) |
| 4522 | xmlParsePEReference(ctxt); |
| 4523 | ctxt->instate = XML_PARSER_DTD; |
| 4524 | } |
| 4525 | |
| 4526 | /** |
| 4527 | * xmlParseTextDecl: |
| 4528 | * @ctxt: an XML parser context |
| 4529 | * |
| 4530 | * parse an XML declaration header for external entities |
| 4531 | * |
| 4532 | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
| 4533 | * |
| 4534 | * Question: Seems that EncodingDecl is mandatory ? Is that a typo ? |
| 4535 | */ |
| 4536 | |
| 4537 | void |
| 4538 | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
| 4539 | xmlChar *version; |
| 4540 | |
| 4541 | /* |
| 4542 | * We know that '<?xml' is here. |
| 4543 | */ |
| 4544 | if ((RAW == '<') && (NXT(1) == '?') && |
| 4545 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 4546 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 4547 | SKIP(5); |
| 4548 | } else { |
| 4549 | ctxt->errNo = XML_ERR_XMLDECL_NOT_STARTED; |
| 4550 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4551 | ctxt->sax->error(ctxt->userData, |
| 4552 | "Text declaration '<?xml' required\n"); |
| 4553 | ctxt->wellFormed = 0; |
| 4554 | ctxt->disableSAX = 1; |
| 4555 | |
| 4556 | return; |
| 4557 | } |
| 4558 | |
| 4559 | if (!IS_BLANK(CUR)) { |
| 4560 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 4561 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4562 | ctxt->sax->error(ctxt->userData, |
| 4563 | "Space needed after '<?xml'\n"); |
| 4564 | ctxt->wellFormed = 0; |
| 4565 | ctxt->disableSAX = 1; |
| 4566 | } |
| 4567 | SKIP_BLANKS; |
| 4568 | |
| 4569 | /* |
| 4570 | * We may have the VersionInfo here. |
| 4571 | */ |
| 4572 | version = xmlParseVersionInfo(ctxt); |
| 4573 | if (version == NULL) |
| 4574 | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
| 4575 | ctxt->input->version = version; |
| 4576 | |
| 4577 | /* |
| 4578 | * We must have the encoding declaration |
| 4579 | */ |
| 4580 | if (!IS_BLANK(CUR)) { |
| 4581 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 4582 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4583 | ctxt->sax->error(ctxt->userData, "Space needed here\n"); |
| 4584 | ctxt->wellFormed = 0; |
| 4585 | ctxt->disableSAX = 1; |
| 4586 | } |
| 4587 | xmlParseEncodingDecl(ctxt); |
| 4588 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 4589 | /* |
| 4590 | * The XML REC instructs us to stop parsing right here |
| 4591 | */ |
| 4592 | return; |
| 4593 | } |
| 4594 | |
| 4595 | SKIP_BLANKS; |
| 4596 | if ((RAW == '?') && (NXT(1) == '>')) { |
| 4597 | SKIP(2); |
| 4598 | } else if (RAW == '>') { |
| 4599 | /* Deprecated old WD ... */ |
| 4600 | ctxt->errNo = XML_ERR_XMLDECL_NOT_FINISHED; |
| 4601 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4602 | ctxt->sax->error(ctxt->userData, |
| 4603 | "XML declaration must end-up with '?>'\n"); |
| 4604 | ctxt->wellFormed = 0; |
| 4605 | ctxt->disableSAX = 1; |
| 4606 | NEXT; |
| 4607 | } else { |
| 4608 | ctxt->errNo = XML_ERR_XMLDECL_NOT_FINISHED; |
| 4609 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4610 | ctxt->sax->error(ctxt->userData, |
| 4611 | "parsing XML declaration: '?>' expected\n"); |
| 4612 | ctxt->wellFormed = 0; |
| 4613 | ctxt->disableSAX = 1; |
| 4614 | MOVETO_ENDTAG(CUR_PTR); |
| 4615 | NEXT; |
| 4616 | } |
| 4617 | } |
| 4618 | |
| 4619 | /* |
| 4620 | * xmlParseConditionalSections |
| 4621 | * @ctxt: an XML parser context |
| 4622 | * |
| 4623 | * [61] conditionalSect ::= includeSect | ignoreSect |
| 4624 | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
| 4625 | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
| 4626 | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
| 4627 | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
| 4628 | */ |
| 4629 | |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4630 | static void |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4631 | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
| 4632 | SKIP(3); |
| 4633 | SKIP_BLANKS; |
| 4634 | if ((RAW == 'I') && (NXT(1) == 'N') && (NXT(2) == 'C') && |
| 4635 | (NXT(3) == 'L') && (NXT(4) == 'U') && (NXT(5) == 'D') && |
| 4636 | (NXT(6) == 'E')) { |
| 4637 | SKIP(7); |
| 4638 | SKIP_BLANKS; |
| 4639 | if (RAW != '[') { |
| 4640 | ctxt->errNo = XML_ERR_CONDSEC_INVALID; |
| 4641 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4642 | ctxt->sax->error(ctxt->userData, |
| 4643 | "XML conditional section '[' expected\n"); |
| 4644 | ctxt->wellFormed = 0; |
| 4645 | ctxt->disableSAX = 1; |
| 4646 | } else { |
| 4647 | NEXT; |
| 4648 | } |
| 4649 | if (xmlParserDebugEntities) { |
| 4650 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
| 4651 | xmlGenericError(xmlGenericErrorContext, |
| 4652 | "%s(%d): ", ctxt->input->filename, |
| 4653 | ctxt->input->line); |
| 4654 | xmlGenericError(xmlGenericErrorContext, |
| 4655 | "Entering INCLUDE Conditional Section\n"); |
| 4656 | } |
| 4657 | |
| 4658 | while ((RAW != 0) && ((RAW != ']') || (NXT(1) != ']') || |
| 4659 | (NXT(2) != '>'))) { |
| 4660 | const xmlChar *check = CUR_PTR; |
| 4661 | int cons = ctxt->input->consumed; |
| 4662 | int tok = ctxt->token; |
| 4663 | |
| 4664 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
| 4665 | xmlParseConditionalSections(ctxt); |
| 4666 | } else if (IS_BLANK(CUR)) { |
| 4667 | NEXT; |
| 4668 | } else if (RAW == '%') { |
| 4669 | xmlParsePEReference(ctxt); |
| 4670 | } else |
| 4671 | xmlParseMarkupDecl(ctxt); |
| 4672 | |
| 4673 | /* |
| 4674 | * Pop-up of finished entities. |
| 4675 | */ |
| 4676 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 4677 | xmlPopInput(ctxt); |
| 4678 | |
| 4679 | if ((CUR_PTR == check) && (cons == ctxt->input->consumed) && |
| 4680 | (tok == ctxt->token)) { |
| 4681 | ctxt->errNo = XML_ERR_EXT_SUBSET_NOT_FINISHED; |
| 4682 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4683 | ctxt->sax->error(ctxt->userData, |
| 4684 | "Content error in the external subset\n"); |
| 4685 | ctxt->wellFormed = 0; |
| 4686 | ctxt->disableSAX = 1; |
| 4687 | break; |
| 4688 | } |
| 4689 | } |
| 4690 | if (xmlParserDebugEntities) { |
| 4691 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
| 4692 | xmlGenericError(xmlGenericErrorContext, |
| 4693 | "%s(%d): ", ctxt->input->filename, |
| 4694 | ctxt->input->line); |
| 4695 | xmlGenericError(xmlGenericErrorContext, |
| 4696 | "Leaving INCLUDE Conditional Section\n"); |
| 4697 | } |
| 4698 | |
| 4699 | } else if ((RAW == 'I') && (NXT(1) == 'G') && (NXT(2) == 'N') && |
| 4700 | (NXT(3) == 'O') && (NXT(4) == 'R') && (NXT(5) == 'E')) { |
| 4701 | int state; |
| 4702 | int instate; |
| 4703 | int depth = 0; |
| 4704 | |
| 4705 | SKIP(6); |
| 4706 | SKIP_BLANKS; |
| 4707 | if (RAW != '[') { |
| 4708 | ctxt->errNo = XML_ERR_CONDSEC_INVALID; |
| 4709 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4710 | ctxt->sax->error(ctxt->userData, |
| 4711 | "XML conditional section '[' expected\n"); |
| 4712 | ctxt->wellFormed = 0; |
| 4713 | ctxt->disableSAX = 1; |
| 4714 | } else { |
| 4715 | NEXT; |
| 4716 | } |
| 4717 | if (xmlParserDebugEntities) { |
| 4718 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
| 4719 | xmlGenericError(xmlGenericErrorContext, |
| 4720 | "%s(%d): ", ctxt->input->filename, |
| 4721 | ctxt->input->line); |
| 4722 | xmlGenericError(xmlGenericErrorContext, |
| 4723 | "Entering IGNORE Conditional Section\n"); |
| 4724 | } |
| 4725 | |
| 4726 | /* |
| 4727 | * Parse up to the end of the conditionnal section |
| 4728 | * But disable SAX event generating DTD building in the meantime |
| 4729 | */ |
| 4730 | state = ctxt->disableSAX; |
| 4731 | instate = ctxt->instate; |
| 4732 | ctxt->disableSAX = 1; |
| 4733 | ctxt->instate = XML_PARSER_IGNORE; |
| 4734 | |
| 4735 | while (depth >= 0) { |
| 4736 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
| 4737 | depth++; |
| 4738 | SKIP(3); |
| 4739 | continue; |
| 4740 | } |
| 4741 | if ((RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
| 4742 | if (--depth >= 0) SKIP(3); |
| 4743 | continue; |
| 4744 | } |
| 4745 | NEXT; |
| 4746 | continue; |
| 4747 | } |
| 4748 | |
| 4749 | ctxt->disableSAX = state; |
| 4750 | ctxt->instate = instate; |
| 4751 | |
| 4752 | if (xmlParserDebugEntities) { |
| 4753 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
| 4754 | xmlGenericError(xmlGenericErrorContext, |
| 4755 | "%s(%d): ", ctxt->input->filename, |
| 4756 | ctxt->input->line); |
| 4757 | xmlGenericError(xmlGenericErrorContext, |
| 4758 | "Leaving IGNORE Conditional Section\n"); |
| 4759 | } |
| 4760 | |
| 4761 | } else { |
| 4762 | ctxt->errNo = XML_ERR_CONDSEC_INVALID; |
| 4763 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4764 | ctxt->sax->error(ctxt->userData, |
| 4765 | "XML conditional section INCLUDE or IGNORE keyword expected\n"); |
| 4766 | ctxt->wellFormed = 0; |
| 4767 | ctxt->disableSAX = 1; |
| 4768 | } |
| 4769 | |
| 4770 | if (RAW == 0) |
| 4771 | SHRINK; |
| 4772 | |
| 4773 | if (RAW == 0) { |
| 4774 | ctxt->errNo = XML_ERR_CONDSEC_NOT_FINISHED; |
| 4775 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4776 | ctxt->sax->error(ctxt->userData, |
| 4777 | "XML conditional section not closed\n"); |
| 4778 | ctxt->wellFormed = 0; |
| 4779 | ctxt->disableSAX = 1; |
| 4780 | } else { |
| 4781 | SKIP(3); |
| 4782 | } |
| 4783 | } |
| 4784 | |
| 4785 | /** |
| 4786 | * xmlParseExternalSubset: |
| 4787 | * @ctxt: an XML parser context |
| 4788 | * @ExternalID: the external identifier |
| 4789 | * @SystemID: the system identifier (or URL) |
| 4790 | * |
| 4791 | * parse Markup declarations from an external subset |
| 4792 | * |
| 4793 | * [30] extSubset ::= textDecl? extSubsetDecl |
| 4794 | * |
| 4795 | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
| 4796 | */ |
| 4797 | void |
| 4798 | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
| 4799 | const xmlChar *SystemID) { |
| 4800 | GROW; |
| 4801 | if ((RAW == '<') && (NXT(1) == '?') && |
| 4802 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 4803 | (NXT(4) == 'l')) { |
| 4804 | xmlParseTextDecl(ctxt); |
| 4805 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 4806 | /* |
| 4807 | * The XML REC instructs us to stop parsing right here |
| 4808 | */ |
| 4809 | ctxt->instate = XML_PARSER_EOF; |
| 4810 | return; |
| 4811 | } |
| 4812 | } |
| 4813 | if (ctxt->myDoc == NULL) { |
| 4814 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
| 4815 | } |
| 4816 | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL)) |
| 4817 | xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID); |
| 4818 | |
| 4819 | ctxt->instate = XML_PARSER_DTD; |
| 4820 | ctxt->external = 1; |
| 4821 | while (((RAW == '<') && (NXT(1) == '?')) || |
| 4822 | ((RAW == '<') && (NXT(1) == '!')) || |
| 4823 | IS_BLANK(CUR)) { |
| 4824 | const xmlChar *check = CUR_PTR; |
| 4825 | int cons = ctxt->input->consumed; |
| 4826 | int tok = ctxt->token; |
| 4827 | |
| 4828 | GROW; |
| 4829 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
| 4830 | xmlParseConditionalSections(ctxt); |
| 4831 | } else if (IS_BLANK(CUR)) { |
| 4832 | NEXT; |
| 4833 | } else if (RAW == '%') { |
| 4834 | xmlParsePEReference(ctxt); |
| 4835 | } else |
| 4836 | xmlParseMarkupDecl(ctxt); |
| 4837 | |
| 4838 | /* |
| 4839 | * Pop-up of finished entities. |
| 4840 | */ |
| 4841 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 4842 | xmlPopInput(ctxt); |
| 4843 | |
| 4844 | if ((CUR_PTR == check) && (cons == ctxt->input->consumed) && |
| 4845 | (tok == ctxt->token)) { |
| 4846 | ctxt->errNo = XML_ERR_EXT_SUBSET_NOT_FINISHED; |
| 4847 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4848 | ctxt->sax->error(ctxt->userData, |
| 4849 | "Content error in the external subset\n"); |
| 4850 | ctxt->wellFormed = 0; |
| 4851 | ctxt->disableSAX = 1; |
| 4852 | break; |
| 4853 | } |
| 4854 | } |
| 4855 | |
| 4856 | if (RAW != 0) { |
| 4857 | ctxt->errNo = XML_ERR_EXT_SUBSET_NOT_FINISHED; |
| 4858 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 4859 | ctxt->sax->error(ctxt->userData, |
| 4860 | "Extra content at the end of the document\n"); |
| 4861 | ctxt->wellFormed = 0; |
| 4862 | ctxt->disableSAX = 1; |
| 4863 | } |
| 4864 | |
| 4865 | } |
| 4866 | |
| 4867 | /** |
| 4868 | * xmlParseReference: |
| 4869 | * @ctxt: an XML parser context |
| 4870 | * |
| 4871 | * parse and handle entity references in content, depending on the SAX |
| 4872 | * interface, this may end-up in a call to character() if this is a |
| 4873 | * CharRef, a predefined entity, if there is no reference() callback. |
| 4874 | * or if the parser was asked to switch to that mode. |
| 4875 | * |
| 4876 | * [67] Reference ::= EntityRef | CharRef |
| 4877 | */ |
| 4878 | void |
| 4879 | xmlParseReference(xmlParserCtxtPtr ctxt) { |
| 4880 | xmlEntityPtr ent; |
| 4881 | xmlChar *val; |
| 4882 | if (RAW != '&') return; |
| 4883 | |
| 4884 | if (NXT(1) == '#') { |
| 4885 | int i = 0; |
| 4886 | xmlChar out[10]; |
| 4887 | int hex = NXT(2); |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4888 | int value = xmlParseCharRef(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4889 | |
| 4890 | if (ctxt->charset != XML_CHAR_ENCODING_UTF8) { |
| 4891 | /* |
| 4892 | * So we are using non-UTF-8 buffers |
| 4893 | * Check that the char fit on 8bits, if not |
| 4894 | * generate a CharRef. |
| 4895 | */ |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4896 | if (value <= 0xFF) { |
| 4897 | out[0] = value; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4898 | out[1] = 0; |
| 4899 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
| 4900 | (!ctxt->disableSAX)) |
| 4901 | ctxt->sax->characters(ctxt->userData, out, 1); |
| 4902 | } else { |
| 4903 | if ((hex == 'x') || (hex == 'X')) |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4904 | sprintf((char *)out, "#x%X", value); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4905 | else |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4906 | sprintf((char *)out, "#%d", value); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4907 | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
| 4908 | (!ctxt->disableSAX)) |
| 4909 | ctxt->sax->reference(ctxt->userData, out); |
| 4910 | } |
| 4911 | } else { |
| 4912 | /* |
| 4913 | * Just encode the value in UTF-8 |
| 4914 | */ |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 4915 | COPY_BUF(0 ,out, i, value); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 4916 | out[i] = 0; |
| 4917 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
| 4918 | (!ctxt->disableSAX)) |
| 4919 | ctxt->sax->characters(ctxt->userData, out, i); |
| 4920 | } |
| 4921 | } else { |
| 4922 | ent = xmlParseEntityRef(ctxt); |
| 4923 | if (ent == NULL) return; |
| 4924 | if ((ent->name != NULL) && |
| 4925 | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
| 4926 | xmlNodePtr list = NULL; |
| 4927 | int ret; |
| 4928 | |
| 4929 | |
| 4930 | /* |
| 4931 | * The first reference to the entity trigger a parsing phase |
| 4932 | * where the ent->children is filled with the result from |
| 4933 | * the parsing. |
| 4934 | */ |
| 4935 | if (ent->children == NULL) { |
| 4936 | xmlChar *value; |
| 4937 | value = ent->content; |
| 4938 | |
| 4939 | /* |
| 4940 | * Check that this entity is well formed |
| 4941 | */ |
| 4942 | if ((value != NULL) && |
| 4943 | (value[1] == 0) && (value[0] == '<') && |
| 4944 | (xmlStrEqual(ent->name, BAD_CAST "lt"))) { |
| 4945 | /* |
| 4946 | * DONE: get definite answer on this !!! |
| 4947 | * Lots of entity decls are used to declare a single |
| 4948 | * char |
| 4949 | * <!ENTITY lt "<"> |
| 4950 | * Which seems to be valid since |
| 4951 | * 2.4: The ampersand character (&) and the left angle |
| 4952 | * bracket (<) may appear in their literal form only |
| 4953 | * when used ... They are also legal within the literal |
| 4954 | * entity value of an internal entity declaration;i |
| 4955 | * see "4.3.2 Well-Formed Parsed Entities". |
| 4956 | * IMHO 2.4 and 4.3.2 are directly in contradiction. |
| 4957 | * Looking at the OASIS test suite and James Clark |
| 4958 | * tests, this is broken. However the XML REC uses |
| 4959 | * it. Is the XML REC not well-formed ???? |
| 4960 | * This is a hack to avoid this problem |
| 4961 | * |
| 4962 | * ANSWER: since lt gt amp .. are already defined, |
| 4963 | * this is a redefinition and hence the fact that the |
| 4964 | * contentis not well balanced is not a Wf error, this |
| 4965 | * is lousy but acceptable. |
| 4966 | */ |
| 4967 | list = xmlNewDocText(ctxt->myDoc, value); |
| 4968 | if (list != NULL) { |
| 4969 | if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) && |
| 4970 | (ent->children == NULL)) { |
| 4971 | ent->children = list; |
| 4972 | ent->last = list; |
| 4973 | list->parent = (xmlNodePtr) ent; |
| 4974 | } else { |
| 4975 | xmlFreeNodeList(list); |
| 4976 | } |
| 4977 | } else if (list != NULL) { |
| 4978 | xmlFreeNodeList(list); |
| 4979 | } |
| 4980 | } else { |
| 4981 | /* |
| 4982 | * 4.3.2: An internal general parsed entity is well-formed |
| 4983 | * if its replacement text matches the production labeled |
| 4984 | * content. |
| 4985 | */ |
| 4986 | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
| 4987 | ctxt->depth++; |
| 4988 | ret = xmlParseBalancedChunkMemory(ctxt->myDoc, |
| 4989 | ctxt->sax, NULL, ctxt->depth, |
| 4990 | value, &list); |
| 4991 | ctxt->depth--; |
| 4992 | } else if (ent->etype == |
| 4993 | XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
| 4994 | ctxt->depth++; |
| 4995 | ret = xmlParseExternalEntity(ctxt->myDoc, |
| 4996 | ctxt->sax, NULL, ctxt->depth, |
| 4997 | ent->URI, ent->ExternalID, &list); |
| 4998 | ctxt->depth--; |
| 4999 | } else { |
| 5000 | ret = -1; |
| 5001 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5002 | ctxt->sax->error(ctxt->userData, |
| 5003 | "Internal: invalid entity type\n"); |
| 5004 | } |
| 5005 | if (ret == XML_ERR_ENTITY_LOOP) { |
| 5006 | ctxt->errNo = XML_ERR_ENTITY_LOOP; |
| 5007 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5008 | ctxt->sax->error(ctxt->userData, |
| 5009 | "Detected entity reference loop\n"); |
| 5010 | ctxt->wellFormed = 0; |
| 5011 | ctxt->disableSAX = 1; |
| 5012 | } else if ((ret == 0) && (list != NULL)) { |
| 5013 | if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) && |
| 5014 | (ent->children == NULL)) { |
| 5015 | ent->children = list; |
| 5016 | while (list != NULL) { |
| 5017 | list->parent = (xmlNodePtr) ent; |
| 5018 | if (list->next == NULL) |
| 5019 | ent->last = list; |
| 5020 | list = list->next; |
| 5021 | } |
| 5022 | } else { |
| 5023 | xmlFreeNodeList(list); |
| 5024 | } |
| 5025 | } else if (ret > 0) { |
| 5026 | ctxt->errNo = ret; |
| 5027 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5028 | ctxt->sax->error(ctxt->userData, |
| 5029 | "Entity value required\n"); |
| 5030 | ctxt->wellFormed = 0; |
| 5031 | ctxt->disableSAX = 1; |
| 5032 | } else if (list != NULL) { |
| 5033 | xmlFreeNodeList(list); |
| 5034 | } |
| 5035 | } |
| 5036 | } |
| 5037 | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
| 5038 | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
| 5039 | /* |
| 5040 | * Create a node. |
| 5041 | */ |
| 5042 | ctxt->sax->reference(ctxt->userData, ent->name); |
| 5043 | return; |
| 5044 | } else if (ctxt->replaceEntities) { |
| 5045 | if ((ctxt->node != NULL) && (ent->children != NULL)) { |
| 5046 | /* |
| 5047 | * Seems we are generating the DOM content, do |
| 5048 | * a simple tree copy |
| 5049 | */ |
| 5050 | xmlNodePtr new; |
| 5051 | new = xmlCopyNodeList(ent->children); |
| 5052 | |
| 5053 | xmlAddChildList(ctxt->node, new); |
| 5054 | /* |
| 5055 | * This is to avoid a nasty side effect, see |
| 5056 | * characters() in SAX.c |
| 5057 | */ |
| 5058 | ctxt->nodemem = 0; |
| 5059 | ctxt->nodelen = 0; |
| 5060 | return; |
| 5061 | } else { |
| 5062 | /* |
| 5063 | * Probably running in SAX mode |
| 5064 | */ |
| 5065 | xmlParserInputPtr input; |
| 5066 | |
| 5067 | input = xmlNewEntityInputStream(ctxt, ent); |
| 5068 | xmlPushInput(ctxt, input); |
| 5069 | if ((ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) && |
| 5070 | (RAW == '<') && (NXT(1) == '?') && |
| 5071 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 5072 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 5073 | xmlParseTextDecl(ctxt); |
| 5074 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 5075 | /* |
| 5076 | * The XML REC instructs us to stop parsing right here |
| 5077 | */ |
| 5078 | ctxt->instate = XML_PARSER_EOF; |
| 5079 | return; |
| 5080 | } |
| 5081 | if (input->standalone == 1) { |
| 5082 | ctxt->errNo = XML_ERR_EXT_ENTITY_STANDALONE; |
| 5083 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5084 | ctxt->sax->error(ctxt->userData, |
| 5085 | "external parsed entities cannot be standalone\n"); |
| 5086 | ctxt->wellFormed = 0; |
| 5087 | ctxt->disableSAX = 1; |
| 5088 | } |
| 5089 | } |
| 5090 | return; |
| 5091 | } |
| 5092 | } |
| 5093 | } else { |
| 5094 | val = ent->content; |
| 5095 | if (val == NULL) return; |
| 5096 | /* |
| 5097 | * inline the entity. |
| 5098 | */ |
| 5099 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
| 5100 | (!ctxt->disableSAX)) |
| 5101 | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
| 5102 | } |
| 5103 | } |
| 5104 | } |
| 5105 | |
| 5106 | /** |
| 5107 | * xmlParseEntityRef: |
| 5108 | * @ctxt: an XML parser context |
| 5109 | * |
| 5110 | * parse ENTITY references declarations |
| 5111 | * |
| 5112 | * [68] EntityRef ::= '&' Name ';' |
| 5113 | * |
| 5114 | * [ WFC: Entity Declared ] |
| 5115 | * In a document without any DTD, a document with only an internal DTD |
| 5116 | * subset which contains no parameter entity references, or a document |
| 5117 | * with "standalone='yes'", the Name given in the entity reference |
| 5118 | * must match that in an entity declaration, except that well-formed |
| 5119 | * documents need not declare any of the following entities: amp, lt, |
| 5120 | * gt, apos, quot. The declaration of a parameter entity must precede |
| 5121 | * any reference to it. Similarly, the declaration of a general entity |
| 5122 | * must precede any reference to it which appears in a default value in an |
| 5123 | * attribute-list declaration. Note that if entities are declared in the |
| 5124 | * external subset or in external parameter entities, a non-validating |
| 5125 | * processor is not obligated to read and process their declarations; |
| 5126 | * for such documents, the rule that an entity must be declared is a |
| 5127 | * well-formedness constraint only if standalone='yes'. |
| 5128 | * |
| 5129 | * [ WFC: Parsed Entity ] |
| 5130 | * An entity reference must not contain the name of an unparsed entity |
| 5131 | * |
| 5132 | * Returns the xmlEntityPtr if found, or NULL otherwise. |
| 5133 | */ |
| 5134 | xmlEntityPtr |
| 5135 | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
| 5136 | xmlChar *name; |
| 5137 | xmlEntityPtr ent = NULL; |
| 5138 | |
| 5139 | GROW; |
| 5140 | |
| 5141 | if (RAW == '&') { |
| 5142 | NEXT; |
| 5143 | name = xmlParseName(ctxt); |
| 5144 | if (name == NULL) { |
| 5145 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 5146 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5147 | ctxt->sax->error(ctxt->userData, |
| 5148 | "xmlParseEntityRef: no name\n"); |
| 5149 | ctxt->wellFormed = 0; |
| 5150 | ctxt->disableSAX = 1; |
| 5151 | } else { |
| 5152 | if (RAW == ';') { |
| 5153 | NEXT; |
| 5154 | /* |
| 5155 | * Ask first SAX for entity resolution, otherwise try the |
| 5156 | * predefined set. |
| 5157 | */ |
| 5158 | if (ctxt->sax != NULL) { |
| 5159 | if (ctxt->sax->getEntity != NULL) |
| 5160 | ent = ctxt->sax->getEntity(ctxt->userData, name); |
| 5161 | if (ent == NULL) |
| 5162 | ent = xmlGetPredefinedEntity(name); |
| 5163 | } |
| 5164 | /* |
| 5165 | * [ WFC: Entity Declared ] |
| 5166 | * In a document without any DTD, a document with only an |
| 5167 | * internal DTD subset which contains no parameter entity |
| 5168 | * references, or a document with "standalone='yes'", the |
| 5169 | * Name given in the entity reference must match that in an |
| 5170 | * entity declaration, except that well-formed documents |
| 5171 | * need not declare any of the following entities: amp, lt, |
| 5172 | * gt, apos, quot. |
| 5173 | * The declaration of a parameter entity must precede any |
| 5174 | * reference to it. |
| 5175 | * Similarly, the declaration of a general entity must |
| 5176 | * precede any reference to it which appears in a default |
| 5177 | * value in an attribute-list declaration. Note that if |
| 5178 | * entities are declared in the external subset or in |
| 5179 | * external parameter entities, a non-validating processor |
| 5180 | * is not obligated to read and process their declarations; |
| 5181 | * for such documents, the rule that an entity must be |
| 5182 | * declared is a well-formedness constraint only if |
| 5183 | * standalone='yes'. |
| 5184 | */ |
| 5185 | if (ent == NULL) { |
| 5186 | if ((ctxt->standalone == 1) || |
| 5187 | ((ctxt->hasExternalSubset == 0) && |
| 5188 | (ctxt->hasPErefs == 0))) { |
| 5189 | ctxt->errNo = XML_ERR_UNDECLARED_ENTITY; |
| 5190 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5191 | ctxt->sax->error(ctxt->userData, |
| 5192 | "Entity '%s' not defined\n", name); |
| 5193 | ctxt->wellFormed = 0; |
| 5194 | ctxt->disableSAX = 1; |
| 5195 | } else { |
| 5196 | ctxt->errNo = XML_WAR_UNDECLARED_ENTITY; |
| 5197 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5198 | ctxt->sax->warning(ctxt->userData, |
| 5199 | "Entity '%s' not defined\n", name); |
| 5200 | } |
| 5201 | } |
| 5202 | |
| 5203 | /* |
| 5204 | * [ WFC: Parsed Entity ] |
| 5205 | * An entity reference must not contain the name of an |
| 5206 | * unparsed entity |
| 5207 | */ |
| 5208 | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
| 5209 | ctxt->errNo = XML_ERR_UNPARSED_ENTITY; |
| 5210 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5211 | ctxt->sax->error(ctxt->userData, |
| 5212 | "Entity reference to unparsed entity %s\n", name); |
| 5213 | ctxt->wellFormed = 0; |
| 5214 | ctxt->disableSAX = 1; |
| 5215 | } |
| 5216 | |
| 5217 | /* |
| 5218 | * [ WFC: No External Entity References ] |
| 5219 | * Attribute values cannot contain direct or indirect |
| 5220 | * entity references to external entities. |
| 5221 | */ |
| 5222 | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
| 5223 | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
| 5224 | ctxt->errNo = XML_ERR_ENTITY_IS_EXTERNAL; |
| 5225 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5226 | ctxt->sax->error(ctxt->userData, |
| 5227 | "Attribute references external entity '%s'\n", name); |
| 5228 | ctxt->wellFormed = 0; |
| 5229 | ctxt->disableSAX = 1; |
| 5230 | } |
| 5231 | /* |
| 5232 | * [ WFC: No < in Attribute Values ] |
| 5233 | * The replacement text of any entity referred to directly or |
| 5234 | * indirectly in an attribute value (other than "<") must |
| 5235 | * not contain a <. |
| 5236 | */ |
| 5237 | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
| 5238 | (ent != NULL) && |
| 5239 | (!xmlStrEqual(ent->name, BAD_CAST "lt")) && |
| 5240 | (ent->content != NULL) && |
| 5241 | (xmlStrchr(ent->content, '<'))) { |
| 5242 | ctxt->errNo = XML_ERR_LT_IN_ATTRIBUTE; |
| 5243 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5244 | ctxt->sax->error(ctxt->userData, |
| 5245 | "'<' in entity '%s' is not allowed in attributes values\n", name); |
| 5246 | ctxt->wellFormed = 0; |
| 5247 | ctxt->disableSAX = 1; |
| 5248 | } |
| 5249 | |
| 5250 | /* |
| 5251 | * Internal check, no parameter entities here ... |
| 5252 | */ |
| 5253 | else { |
| 5254 | switch (ent->etype) { |
| 5255 | case XML_INTERNAL_PARAMETER_ENTITY: |
| 5256 | case XML_EXTERNAL_PARAMETER_ENTITY: |
| 5257 | ctxt->errNo = XML_ERR_ENTITY_IS_PARAMETER; |
| 5258 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5259 | ctxt->sax->error(ctxt->userData, |
| 5260 | "Attempt to reference the parameter entity '%s'\n", name); |
| 5261 | ctxt->wellFormed = 0; |
| 5262 | ctxt->disableSAX = 1; |
| 5263 | break; |
| 5264 | default: |
| 5265 | break; |
| 5266 | } |
| 5267 | } |
| 5268 | |
| 5269 | /* |
| 5270 | * [ WFC: No Recursion ] |
| 5271 | * A parsed entity must not contain a recursive reference |
| 5272 | * to itself, either directly or indirectly. |
| 5273 | * Done somewhere else |
| 5274 | */ |
| 5275 | |
| 5276 | } else { |
| 5277 | ctxt->errNo = XML_ERR_ENTITYREF_SEMICOL_MISSING; |
| 5278 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5279 | ctxt->sax->error(ctxt->userData, |
| 5280 | "xmlParseEntityRef: expecting ';'\n"); |
| 5281 | ctxt->wellFormed = 0; |
| 5282 | ctxt->disableSAX = 1; |
| 5283 | } |
| 5284 | xmlFree(name); |
| 5285 | } |
| 5286 | } |
| 5287 | return(ent); |
| 5288 | } |
| 5289 | |
| 5290 | /** |
| 5291 | * xmlParseStringEntityRef: |
| 5292 | * @ctxt: an XML parser context |
| 5293 | * @str: a pointer to an index in the string |
| 5294 | * |
| 5295 | * parse ENTITY references declarations, but this version parses it from |
| 5296 | * a string value. |
| 5297 | * |
| 5298 | * [68] EntityRef ::= '&' Name ';' |
| 5299 | * |
| 5300 | * [ WFC: Entity Declared ] |
| 5301 | * In a document without any DTD, a document with only an internal DTD |
| 5302 | * subset which contains no parameter entity references, or a document |
| 5303 | * with "standalone='yes'", the Name given in the entity reference |
| 5304 | * must match that in an entity declaration, except that well-formed |
| 5305 | * documents need not declare any of the following entities: amp, lt, |
| 5306 | * gt, apos, quot. The declaration of a parameter entity must precede |
| 5307 | * any reference to it. Similarly, the declaration of a general entity |
| 5308 | * must precede any reference to it which appears in a default value in an |
| 5309 | * attribute-list declaration. Note that if entities are declared in the |
| 5310 | * external subset or in external parameter entities, a non-validating |
| 5311 | * processor is not obligated to read and process their declarations; |
| 5312 | * for such documents, the rule that an entity must be declared is a |
| 5313 | * well-formedness constraint only if standalone='yes'. |
| 5314 | * |
| 5315 | * [ WFC: Parsed Entity ] |
| 5316 | * An entity reference must not contain the name of an unparsed entity |
| 5317 | * |
| 5318 | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
| 5319 | * is updated to the current location in the string. |
| 5320 | */ |
| 5321 | xmlEntityPtr |
| 5322 | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
| 5323 | xmlChar *name; |
| 5324 | const xmlChar *ptr; |
| 5325 | xmlChar cur; |
| 5326 | xmlEntityPtr ent = NULL; |
| 5327 | |
| 5328 | if ((str == NULL) || (*str == NULL)) |
| 5329 | return(NULL); |
| 5330 | ptr = *str; |
| 5331 | cur = *ptr; |
| 5332 | if (cur == '&') { |
| 5333 | ptr++; |
| 5334 | cur = *ptr; |
| 5335 | name = xmlParseStringName(ctxt, &ptr); |
| 5336 | if (name == NULL) { |
| 5337 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 5338 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5339 | ctxt->sax->error(ctxt->userData, |
| 5340 | "xmlParseEntityRef: no name\n"); |
| 5341 | ctxt->wellFormed = 0; |
| 5342 | ctxt->disableSAX = 1; |
| 5343 | } else { |
| 5344 | if (*ptr == ';') { |
| 5345 | ptr++; |
| 5346 | /* |
| 5347 | * Ask first SAX for entity resolution, otherwise try the |
| 5348 | * predefined set. |
| 5349 | */ |
| 5350 | if (ctxt->sax != NULL) { |
| 5351 | if (ctxt->sax->getEntity != NULL) |
| 5352 | ent = ctxt->sax->getEntity(ctxt->userData, name); |
| 5353 | if (ent == NULL) |
| 5354 | ent = xmlGetPredefinedEntity(name); |
| 5355 | } |
| 5356 | /* |
| 5357 | * [ WFC: Entity Declared ] |
| 5358 | * In a document without any DTD, a document with only an |
| 5359 | * internal DTD subset which contains no parameter entity |
| 5360 | * references, or a document with "standalone='yes'", the |
| 5361 | * Name given in the entity reference must match that in an |
| 5362 | * entity declaration, except that well-formed documents |
| 5363 | * need not declare any of the following entities: amp, lt, |
| 5364 | * gt, apos, quot. |
| 5365 | * The declaration of a parameter entity must precede any |
| 5366 | * reference to it. |
| 5367 | * Similarly, the declaration of a general entity must |
| 5368 | * precede any reference to it which appears in a default |
| 5369 | * value in an attribute-list declaration. Note that if |
| 5370 | * entities are declared in the external subset or in |
| 5371 | * external parameter entities, a non-validating processor |
| 5372 | * is not obligated to read and process their declarations; |
| 5373 | * for such documents, the rule that an entity must be |
| 5374 | * declared is a well-formedness constraint only if |
| 5375 | * standalone='yes'. |
| 5376 | */ |
| 5377 | if (ent == NULL) { |
| 5378 | if ((ctxt->standalone == 1) || |
| 5379 | ((ctxt->hasExternalSubset == 0) && |
| 5380 | (ctxt->hasPErefs == 0))) { |
| 5381 | ctxt->errNo = XML_ERR_UNDECLARED_ENTITY; |
| 5382 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5383 | ctxt->sax->error(ctxt->userData, |
| 5384 | "Entity '%s' not defined\n", name); |
| 5385 | ctxt->wellFormed = 0; |
| 5386 | ctxt->disableSAX = 1; |
| 5387 | } else { |
| 5388 | ctxt->errNo = XML_WAR_UNDECLARED_ENTITY; |
| 5389 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5390 | ctxt->sax->warning(ctxt->userData, |
| 5391 | "Entity '%s' not defined\n", name); |
| 5392 | } |
| 5393 | } |
| 5394 | |
| 5395 | /* |
| 5396 | * [ WFC: Parsed Entity ] |
| 5397 | * An entity reference must not contain the name of an |
| 5398 | * unparsed entity |
| 5399 | */ |
| 5400 | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
| 5401 | ctxt->errNo = XML_ERR_UNPARSED_ENTITY; |
| 5402 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5403 | ctxt->sax->error(ctxt->userData, |
| 5404 | "Entity reference to unparsed entity %s\n", name); |
| 5405 | ctxt->wellFormed = 0; |
| 5406 | ctxt->disableSAX = 1; |
| 5407 | } |
| 5408 | |
| 5409 | /* |
| 5410 | * [ WFC: No External Entity References ] |
| 5411 | * Attribute values cannot contain direct or indirect |
| 5412 | * entity references to external entities. |
| 5413 | */ |
| 5414 | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
| 5415 | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
| 5416 | ctxt->errNo = XML_ERR_ENTITY_IS_EXTERNAL; |
| 5417 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5418 | ctxt->sax->error(ctxt->userData, |
| 5419 | "Attribute references external entity '%s'\n", name); |
| 5420 | ctxt->wellFormed = 0; |
| 5421 | ctxt->disableSAX = 1; |
| 5422 | } |
| 5423 | /* |
| 5424 | * [ WFC: No < in Attribute Values ] |
| 5425 | * The replacement text of any entity referred to directly or |
| 5426 | * indirectly in an attribute value (other than "<") must |
| 5427 | * not contain a <. |
| 5428 | */ |
| 5429 | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
| 5430 | (ent != NULL) && |
| 5431 | (!xmlStrEqual(ent->name, BAD_CAST "lt")) && |
| 5432 | (ent->content != NULL) && |
| 5433 | (xmlStrchr(ent->content, '<'))) { |
| 5434 | ctxt->errNo = XML_ERR_LT_IN_ATTRIBUTE; |
| 5435 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5436 | ctxt->sax->error(ctxt->userData, |
| 5437 | "'<' in entity '%s' is not allowed in attributes values\n", name); |
| 5438 | ctxt->wellFormed = 0; |
| 5439 | ctxt->disableSAX = 1; |
| 5440 | } |
| 5441 | |
| 5442 | /* |
| 5443 | * Internal check, no parameter entities here ... |
| 5444 | */ |
| 5445 | else { |
| 5446 | switch (ent->etype) { |
| 5447 | case XML_INTERNAL_PARAMETER_ENTITY: |
| 5448 | case XML_EXTERNAL_PARAMETER_ENTITY: |
| 5449 | ctxt->errNo = XML_ERR_ENTITY_IS_PARAMETER; |
| 5450 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5451 | ctxt->sax->error(ctxt->userData, |
| 5452 | "Attempt to reference the parameter entity '%s'\n", name); |
| 5453 | ctxt->wellFormed = 0; |
| 5454 | ctxt->disableSAX = 1; |
| 5455 | break; |
| 5456 | default: |
| 5457 | break; |
| 5458 | } |
| 5459 | } |
| 5460 | |
| 5461 | /* |
| 5462 | * [ WFC: No Recursion ] |
| 5463 | * A parsed entity must not contain a recursive reference |
| 5464 | * to itself, either directly or indirectly. |
| 5465 | * Done somewhwere else |
| 5466 | */ |
| 5467 | |
| 5468 | } else { |
| 5469 | ctxt->errNo = XML_ERR_ENTITYREF_SEMICOL_MISSING; |
| 5470 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5471 | ctxt->sax->error(ctxt->userData, |
| 5472 | "xmlParseEntityRef: expecting ';'\n"); |
| 5473 | ctxt->wellFormed = 0; |
| 5474 | ctxt->disableSAX = 1; |
| 5475 | } |
| 5476 | xmlFree(name); |
| 5477 | } |
| 5478 | } |
| 5479 | *str = ptr; |
| 5480 | return(ent); |
| 5481 | } |
| 5482 | |
| 5483 | /** |
| 5484 | * xmlParsePEReference: |
| 5485 | * @ctxt: an XML parser context |
| 5486 | * |
| 5487 | * parse PEReference declarations |
| 5488 | * The entity content is handled directly by pushing it's content as |
| 5489 | * a new input stream. |
| 5490 | * |
| 5491 | * [69] PEReference ::= '%' Name ';' |
| 5492 | * |
| 5493 | * [ WFC: No Recursion ] |
| 5494 | * A parsed entity must not contain a recursive |
| 5495 | * reference to itself, either directly or indirectly. |
| 5496 | * |
| 5497 | * [ WFC: Entity Declared ] |
| 5498 | * In a document without any DTD, a document with only an internal DTD |
| 5499 | * subset which contains no parameter entity references, or a document |
| 5500 | * with "standalone='yes'", ... ... The declaration of a parameter |
| 5501 | * entity must precede any reference to it... |
| 5502 | * |
| 5503 | * [ VC: Entity Declared ] |
| 5504 | * In a document with an external subset or external parameter entities |
| 5505 | * with "standalone='no'", ... ... The declaration of a parameter entity |
| 5506 | * must precede any reference to it... |
| 5507 | * |
| 5508 | * [ WFC: In DTD ] |
| 5509 | * Parameter-entity references may only appear in the DTD. |
| 5510 | * NOTE: misleading but this is handled. |
| 5511 | */ |
| 5512 | void |
| 5513 | xmlParsePEReference(xmlParserCtxtPtr ctxt) { |
| 5514 | xmlChar *name; |
| 5515 | xmlEntityPtr entity = NULL; |
| 5516 | xmlParserInputPtr input; |
| 5517 | |
| 5518 | if (RAW == '%') { |
| 5519 | NEXT; |
Daniel Veillard | 29631a8 | 2001-03-05 09:49:20 +0000 | [diff] [blame] | 5520 | name = xmlParseNameComplex(ctxt); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 5521 | if (name == NULL) { |
| 5522 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 5523 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5524 | ctxt->sax->error(ctxt->userData, |
| 5525 | "xmlParsePEReference: no name\n"); |
| 5526 | ctxt->wellFormed = 0; |
| 5527 | ctxt->disableSAX = 1; |
| 5528 | } else { |
| 5529 | if (RAW == ';') { |
| 5530 | NEXT; |
| 5531 | if ((ctxt->sax != NULL) && |
| 5532 | (ctxt->sax->getParameterEntity != NULL)) |
| 5533 | entity = ctxt->sax->getParameterEntity(ctxt->userData, |
| 5534 | name); |
| 5535 | if (entity == NULL) { |
| 5536 | /* |
| 5537 | * [ WFC: Entity Declared ] |
| 5538 | * In a document without any DTD, a document with only an |
| 5539 | * internal DTD subset which contains no parameter entity |
| 5540 | * references, or a document with "standalone='yes'", ... |
| 5541 | * ... The declaration of a parameter entity must precede |
| 5542 | * any reference to it... |
| 5543 | */ |
| 5544 | if ((ctxt->standalone == 1) || |
| 5545 | ((ctxt->hasExternalSubset == 0) && |
| 5546 | (ctxt->hasPErefs == 0))) { |
| 5547 | ctxt->errNo = XML_ERR_UNDECLARED_ENTITY; |
| 5548 | if ((!ctxt->disableSAX) && |
| 5549 | (ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5550 | ctxt->sax->error(ctxt->userData, |
| 5551 | "PEReference: %%%s; not found\n", name); |
| 5552 | ctxt->wellFormed = 0; |
| 5553 | ctxt->disableSAX = 1; |
| 5554 | } else { |
| 5555 | /* |
| 5556 | * [ VC: Entity Declared ] |
| 5557 | * In a document with an external subset or external |
| 5558 | * parameter entities with "standalone='no'", ... |
| 5559 | * ... The declaration of a parameter entity must precede |
| 5560 | * any reference to it... |
| 5561 | */ |
| 5562 | if ((!ctxt->disableSAX) && |
| 5563 | (ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5564 | ctxt->sax->warning(ctxt->userData, |
| 5565 | "PEReference: %%%s; not found\n", name); |
| 5566 | ctxt->valid = 0; |
| 5567 | } |
| 5568 | } else { |
| 5569 | /* |
| 5570 | * Internal checking in case the entity quest barfed |
| 5571 | */ |
| 5572 | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
| 5573 | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
| 5574 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5575 | ctxt->sax->warning(ctxt->userData, |
| 5576 | "Internal: %%%s; is not a parameter entity\n", name); |
| 5577 | } else { |
| 5578 | /* |
| 5579 | * TODO !!! |
| 5580 | * handle the extra spaces added before and after |
| 5581 | * c.f. http://www.w3.org/TR/REC-xml#as-PE |
| 5582 | */ |
| 5583 | input = xmlNewEntityInputStream(ctxt, entity); |
| 5584 | xmlPushInput(ctxt, input); |
| 5585 | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
| 5586 | (RAW == '<') && (NXT(1) == '?') && |
| 5587 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 5588 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 5589 | xmlParseTextDecl(ctxt); |
| 5590 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 5591 | /* |
| 5592 | * The XML REC instructs us to stop parsing |
| 5593 | * right here |
| 5594 | */ |
| 5595 | ctxt->instate = XML_PARSER_EOF; |
| 5596 | xmlFree(name); |
| 5597 | return; |
| 5598 | } |
| 5599 | } |
| 5600 | if (ctxt->token == 0) |
| 5601 | ctxt->token = ' '; |
| 5602 | } |
| 5603 | } |
| 5604 | ctxt->hasPErefs = 1; |
| 5605 | } else { |
| 5606 | ctxt->errNo = XML_ERR_ENTITYREF_SEMICOL_MISSING; |
| 5607 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5608 | ctxt->sax->error(ctxt->userData, |
| 5609 | "xmlParsePEReference: expecting ';'\n"); |
| 5610 | ctxt->wellFormed = 0; |
| 5611 | ctxt->disableSAX = 1; |
| 5612 | } |
| 5613 | xmlFree(name); |
| 5614 | } |
| 5615 | } |
| 5616 | } |
| 5617 | |
| 5618 | /** |
| 5619 | * xmlParseStringPEReference: |
| 5620 | * @ctxt: an XML parser context |
| 5621 | * @str: a pointer to an index in the string |
| 5622 | * |
| 5623 | * parse PEReference declarations |
| 5624 | * |
| 5625 | * [69] PEReference ::= '%' Name ';' |
| 5626 | * |
| 5627 | * [ WFC: No Recursion ] |
| 5628 | * A parsed entity must not contain a recursive |
| 5629 | * reference to itself, either directly or indirectly. |
| 5630 | * |
| 5631 | * [ WFC: Entity Declared ] |
| 5632 | * In a document without any DTD, a document with only an internal DTD |
| 5633 | * subset which contains no parameter entity references, or a document |
| 5634 | * with "standalone='yes'", ... ... The declaration of a parameter |
| 5635 | * entity must precede any reference to it... |
| 5636 | * |
| 5637 | * [ VC: Entity Declared ] |
| 5638 | * In a document with an external subset or external parameter entities |
| 5639 | * with "standalone='no'", ... ... The declaration of a parameter entity |
| 5640 | * must precede any reference to it... |
| 5641 | * |
| 5642 | * [ WFC: In DTD ] |
| 5643 | * Parameter-entity references may only appear in the DTD. |
| 5644 | * NOTE: misleading but this is handled. |
| 5645 | * |
| 5646 | * Returns the string of the entity content. |
| 5647 | * str is updated to the current value of the index |
| 5648 | */ |
| 5649 | xmlEntityPtr |
| 5650 | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
| 5651 | const xmlChar *ptr; |
| 5652 | xmlChar cur; |
| 5653 | xmlChar *name; |
| 5654 | xmlEntityPtr entity = NULL; |
| 5655 | |
| 5656 | if ((str == NULL) || (*str == NULL)) return(NULL); |
| 5657 | ptr = *str; |
| 5658 | cur = *ptr; |
| 5659 | if (cur == '%') { |
| 5660 | ptr++; |
| 5661 | cur = *ptr; |
| 5662 | name = xmlParseStringName(ctxt, &ptr); |
| 5663 | if (name == NULL) { |
| 5664 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 5665 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5666 | ctxt->sax->error(ctxt->userData, |
| 5667 | "xmlParseStringPEReference: no name\n"); |
| 5668 | ctxt->wellFormed = 0; |
| 5669 | ctxt->disableSAX = 1; |
| 5670 | } else { |
| 5671 | cur = *ptr; |
| 5672 | if (cur == ';') { |
| 5673 | ptr++; |
| 5674 | cur = *ptr; |
| 5675 | if ((ctxt->sax != NULL) && |
| 5676 | (ctxt->sax->getParameterEntity != NULL)) |
| 5677 | entity = ctxt->sax->getParameterEntity(ctxt->userData, |
| 5678 | name); |
| 5679 | if (entity == NULL) { |
| 5680 | /* |
| 5681 | * [ WFC: Entity Declared ] |
| 5682 | * In a document without any DTD, a document with only an |
| 5683 | * internal DTD subset which contains no parameter entity |
| 5684 | * references, or a document with "standalone='yes'", ... |
| 5685 | * ... The declaration of a parameter entity must precede |
| 5686 | * any reference to it... |
| 5687 | */ |
| 5688 | if ((ctxt->standalone == 1) || |
| 5689 | ((ctxt->hasExternalSubset == 0) && |
| 5690 | (ctxt->hasPErefs == 0))) { |
| 5691 | ctxt->errNo = XML_ERR_UNDECLARED_ENTITY; |
| 5692 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5693 | ctxt->sax->error(ctxt->userData, |
| 5694 | "PEReference: %%%s; not found\n", name); |
| 5695 | ctxt->wellFormed = 0; |
| 5696 | ctxt->disableSAX = 1; |
| 5697 | } else { |
| 5698 | /* |
| 5699 | * [ VC: Entity Declared ] |
| 5700 | * In a document with an external subset or external |
| 5701 | * parameter entities with "standalone='no'", ... |
| 5702 | * ... The declaration of a parameter entity must |
| 5703 | * precede any reference to it... |
| 5704 | */ |
| 5705 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5706 | ctxt->sax->warning(ctxt->userData, |
| 5707 | "PEReference: %%%s; not found\n", name); |
| 5708 | ctxt->valid = 0; |
| 5709 | } |
| 5710 | } else { |
| 5711 | /* |
| 5712 | * Internal checking in case the entity quest barfed |
| 5713 | */ |
| 5714 | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
| 5715 | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
| 5716 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5717 | ctxt->sax->warning(ctxt->userData, |
| 5718 | "Internal: %%%s; is not a parameter entity\n", name); |
| 5719 | } |
| 5720 | } |
| 5721 | ctxt->hasPErefs = 1; |
| 5722 | } else { |
| 5723 | ctxt->errNo = XML_ERR_ENTITYREF_SEMICOL_MISSING; |
| 5724 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5725 | ctxt->sax->error(ctxt->userData, |
| 5726 | "xmlParseStringPEReference: expecting ';'\n"); |
| 5727 | ctxt->wellFormed = 0; |
| 5728 | ctxt->disableSAX = 1; |
| 5729 | } |
| 5730 | xmlFree(name); |
| 5731 | } |
| 5732 | } |
| 5733 | *str = ptr; |
| 5734 | return(entity); |
| 5735 | } |
| 5736 | |
| 5737 | /** |
| 5738 | * xmlParseDocTypeDecl: |
| 5739 | * @ctxt: an XML parser context |
| 5740 | * |
| 5741 | * parse a DOCTYPE declaration |
| 5742 | * |
| 5743 | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
| 5744 | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
| 5745 | * |
| 5746 | * [ VC: Root Element Type ] |
| 5747 | * The Name in the document type declaration must match the element |
| 5748 | * type of the root element. |
| 5749 | */ |
| 5750 | |
| 5751 | void |
| 5752 | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
| 5753 | xmlChar *name = NULL; |
| 5754 | xmlChar *ExternalID = NULL; |
| 5755 | xmlChar *URI = NULL; |
| 5756 | |
| 5757 | /* |
| 5758 | * We know that '<!DOCTYPE' has been detected. |
| 5759 | */ |
| 5760 | SKIP(9); |
| 5761 | |
| 5762 | SKIP_BLANKS; |
| 5763 | |
| 5764 | /* |
| 5765 | * Parse the DOCTYPE name. |
| 5766 | */ |
| 5767 | name = xmlParseName(ctxt); |
| 5768 | if (name == NULL) { |
| 5769 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 5770 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5771 | ctxt->sax->error(ctxt->userData, |
| 5772 | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
| 5773 | ctxt->wellFormed = 0; |
| 5774 | ctxt->disableSAX = 1; |
| 5775 | } |
| 5776 | ctxt->intSubName = name; |
| 5777 | |
| 5778 | SKIP_BLANKS; |
| 5779 | |
| 5780 | /* |
| 5781 | * Check for SystemID and ExternalID |
| 5782 | */ |
| 5783 | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
| 5784 | |
| 5785 | if ((URI != NULL) || (ExternalID != NULL)) { |
| 5786 | ctxt->hasExternalSubset = 1; |
| 5787 | } |
| 5788 | ctxt->extSubURI = URI; |
| 5789 | ctxt->extSubSystem = ExternalID; |
| 5790 | |
| 5791 | SKIP_BLANKS; |
| 5792 | |
| 5793 | /* |
| 5794 | * Create and update the internal subset. |
| 5795 | */ |
| 5796 | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
| 5797 | (!ctxt->disableSAX)) |
| 5798 | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
| 5799 | |
| 5800 | /* |
| 5801 | * Is there any internal subset declarations ? |
| 5802 | * they are handled separately in xmlParseInternalSubset() |
| 5803 | */ |
| 5804 | if (RAW == '[') |
| 5805 | return; |
| 5806 | |
| 5807 | /* |
| 5808 | * We should be at the end of the DOCTYPE declaration. |
| 5809 | */ |
| 5810 | if (RAW != '>') { |
| 5811 | ctxt->errNo = XML_ERR_DOCTYPE_NOT_FINISHED; |
| 5812 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5813 | ctxt->sax->error(ctxt->userData, "DOCTYPE unproperly terminated\n"); |
| 5814 | ctxt->wellFormed = 0; |
| 5815 | ctxt->disableSAX = 1; |
| 5816 | } |
| 5817 | NEXT; |
| 5818 | } |
| 5819 | |
| 5820 | /** |
| 5821 | * xmlParseInternalsubset: |
| 5822 | * @ctxt: an XML parser context |
| 5823 | * |
| 5824 | * parse the internal subset declaration |
| 5825 | * |
| 5826 | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
| 5827 | */ |
| 5828 | |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 5829 | static void |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 5830 | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
| 5831 | /* |
| 5832 | * Is there any DTD definition ? |
| 5833 | */ |
| 5834 | if (RAW == '[') { |
| 5835 | ctxt->instate = XML_PARSER_DTD; |
| 5836 | NEXT; |
| 5837 | /* |
| 5838 | * Parse the succession of Markup declarations and |
| 5839 | * PEReferences. |
| 5840 | * Subsequence (markupdecl | PEReference | S)* |
| 5841 | */ |
| 5842 | while (RAW != ']') { |
| 5843 | const xmlChar *check = CUR_PTR; |
| 5844 | int cons = ctxt->input->consumed; |
| 5845 | |
| 5846 | SKIP_BLANKS; |
| 5847 | xmlParseMarkupDecl(ctxt); |
| 5848 | xmlParsePEReference(ctxt); |
| 5849 | |
| 5850 | /* |
| 5851 | * Pop-up of finished entities. |
| 5852 | */ |
| 5853 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 5854 | xmlPopInput(ctxt); |
| 5855 | |
| 5856 | if ((CUR_PTR == check) && (cons == ctxt->input->consumed)) { |
| 5857 | ctxt->errNo = XML_ERR_INTERNAL_ERROR; |
| 5858 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5859 | ctxt->sax->error(ctxt->userData, |
| 5860 | "xmlParseInternalSubset: error detected in Markup declaration\n"); |
| 5861 | ctxt->wellFormed = 0; |
| 5862 | ctxt->disableSAX = 1; |
| 5863 | break; |
| 5864 | } |
| 5865 | } |
| 5866 | if (RAW == ']') { |
| 5867 | NEXT; |
| 5868 | SKIP_BLANKS; |
| 5869 | } |
| 5870 | } |
| 5871 | |
| 5872 | /* |
| 5873 | * We should be at the end of the DOCTYPE declaration. |
| 5874 | */ |
| 5875 | if (RAW != '>') { |
| 5876 | ctxt->errNo = XML_ERR_DOCTYPE_NOT_FINISHED; |
| 5877 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5878 | ctxt->sax->error(ctxt->userData, "DOCTYPE unproperly terminated\n"); |
| 5879 | ctxt->wellFormed = 0; |
| 5880 | ctxt->disableSAX = 1; |
| 5881 | } |
| 5882 | NEXT; |
| 5883 | } |
| 5884 | |
| 5885 | /** |
| 5886 | * xmlParseAttribute: |
| 5887 | * @ctxt: an XML parser context |
| 5888 | * @value: a xmlChar ** used to store the value of the attribute |
| 5889 | * |
| 5890 | * parse an attribute |
| 5891 | * |
| 5892 | * [41] Attribute ::= Name Eq AttValue |
| 5893 | * |
| 5894 | * [ WFC: No External Entity References ] |
| 5895 | * Attribute values cannot contain direct or indirect entity references |
| 5896 | * to external entities. |
| 5897 | * |
| 5898 | * [ WFC: No < in Attribute Values ] |
| 5899 | * The replacement text of any entity referred to directly or indirectly in |
| 5900 | * an attribute value (other than "<") must not contain a <. |
| 5901 | * |
| 5902 | * [ VC: Attribute Value Type ] |
| 5903 | * The attribute must have been declared; the value must be of the type |
| 5904 | * declared for it. |
| 5905 | * |
| 5906 | * [25] Eq ::= S? '=' S? |
| 5907 | * |
| 5908 | * With namespace: |
| 5909 | * |
| 5910 | * [NS 11] Attribute ::= QName Eq AttValue |
| 5911 | * |
| 5912 | * Also the case QName == xmlns:??? is handled independently as a namespace |
| 5913 | * definition. |
| 5914 | * |
| 5915 | * Returns the attribute name, and the value in *value. |
| 5916 | */ |
| 5917 | |
| 5918 | xmlChar * |
| 5919 | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
| 5920 | xmlChar *name, *val; |
| 5921 | |
| 5922 | *value = NULL; |
| 5923 | name = xmlParseName(ctxt); |
| 5924 | if (name == NULL) { |
| 5925 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 5926 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5927 | ctxt->sax->error(ctxt->userData, "error parsing attribute name\n"); |
| 5928 | ctxt->wellFormed = 0; |
| 5929 | ctxt->disableSAX = 1; |
| 5930 | return(NULL); |
| 5931 | } |
| 5932 | |
| 5933 | /* |
| 5934 | * read the value |
| 5935 | */ |
| 5936 | SKIP_BLANKS; |
| 5937 | if (RAW == '=') { |
| 5938 | NEXT; |
| 5939 | SKIP_BLANKS; |
| 5940 | val = xmlParseAttValue(ctxt); |
| 5941 | ctxt->instate = XML_PARSER_CONTENT; |
| 5942 | } else { |
| 5943 | ctxt->errNo = XML_ERR_ATTRIBUTE_WITHOUT_VALUE; |
| 5944 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5945 | ctxt->sax->error(ctxt->userData, |
| 5946 | "Specification mandate value for attribute %s\n", name); |
| 5947 | ctxt->wellFormed = 0; |
| 5948 | ctxt->disableSAX = 1; |
| 5949 | xmlFree(name); |
| 5950 | return(NULL); |
| 5951 | } |
| 5952 | |
| 5953 | /* |
| 5954 | * Check that xml:lang conforms to the specification |
| 5955 | * No more registered as an error, just generate a warning now |
| 5956 | * since this was deprecated in XML second edition |
| 5957 | */ |
| 5958 | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
| 5959 | if (!xmlCheckLanguageID(val)) { |
| 5960 | if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL)) |
| 5961 | ctxt->sax->warning(ctxt->userData, |
| 5962 | "Malformed value for xml:lang : %s\n", val); |
| 5963 | } |
| 5964 | } |
| 5965 | |
| 5966 | /* |
| 5967 | * Check that xml:space conforms to the specification |
| 5968 | */ |
| 5969 | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
| 5970 | if (xmlStrEqual(val, BAD_CAST "default")) |
| 5971 | *(ctxt->space) = 0; |
| 5972 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
| 5973 | *(ctxt->space) = 1; |
| 5974 | else { |
| 5975 | ctxt->errNo = XML_ERR_ATTRIBUTE_WITHOUT_VALUE; |
| 5976 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 5977 | ctxt->sax->error(ctxt->userData, |
| 5978 | "Invalid value for xml:space : \"%s\", \"default\" or \"preserve\" expected\n", |
| 5979 | val); |
| 5980 | ctxt->wellFormed = 0; |
| 5981 | ctxt->disableSAX = 1; |
| 5982 | } |
| 5983 | } |
| 5984 | |
| 5985 | *value = val; |
| 5986 | return(name); |
| 5987 | } |
| 5988 | |
| 5989 | /** |
| 5990 | * xmlParseStartTag: |
| 5991 | * @ctxt: an XML parser context |
| 5992 | * |
| 5993 | * parse a start of tag either for rule element or |
| 5994 | * EmptyElement. In both case we don't parse the tag closing chars. |
| 5995 | * |
| 5996 | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
| 5997 | * |
| 5998 | * [ WFC: Unique Att Spec ] |
| 5999 | * No attribute name may appear more than once in the same start-tag or |
| 6000 | * empty-element tag. |
| 6001 | * |
| 6002 | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
| 6003 | * |
| 6004 | * [ WFC: Unique Att Spec ] |
| 6005 | * No attribute name may appear more than once in the same start-tag or |
| 6006 | * empty-element tag. |
| 6007 | * |
| 6008 | * With namespace: |
| 6009 | * |
| 6010 | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
| 6011 | * |
| 6012 | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
| 6013 | * |
| 6014 | * Returns the element name parsed |
| 6015 | */ |
| 6016 | |
| 6017 | xmlChar * |
| 6018 | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
| 6019 | xmlChar *name; |
| 6020 | xmlChar *attname; |
| 6021 | xmlChar *attvalue; |
| 6022 | const xmlChar **atts = NULL; |
| 6023 | int nbatts = 0; |
| 6024 | int maxatts = 0; |
| 6025 | int i; |
| 6026 | |
| 6027 | if (RAW != '<') return(NULL); |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6028 | NEXT1; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6029 | |
| 6030 | name = xmlParseName(ctxt); |
| 6031 | if (name == NULL) { |
| 6032 | ctxt->errNo = XML_ERR_NAME_REQUIRED; |
| 6033 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6034 | ctxt->sax->error(ctxt->userData, |
| 6035 | "xmlParseStartTag: invalid element name\n"); |
| 6036 | ctxt->wellFormed = 0; |
| 6037 | ctxt->disableSAX = 1; |
| 6038 | return(NULL); |
| 6039 | } |
| 6040 | |
| 6041 | /* |
| 6042 | * Now parse the attributes, it ends up with the ending |
| 6043 | * |
| 6044 | * (S Attribute)* S? |
| 6045 | */ |
| 6046 | SKIP_BLANKS; |
| 6047 | GROW; |
| 6048 | |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6049 | while ((RAW != '>') && |
| 6050 | ((RAW != '/') || (NXT(1) != '>')) && |
| 6051 | (IS_CHAR(RAW))) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6052 | const xmlChar *q = CUR_PTR; |
| 6053 | int cons = ctxt->input->consumed; |
| 6054 | |
| 6055 | attname = xmlParseAttribute(ctxt, &attvalue); |
| 6056 | if ((attname != NULL) && (attvalue != NULL)) { |
| 6057 | /* |
| 6058 | * [ WFC: Unique Att Spec ] |
| 6059 | * No attribute name may appear more than once in the same |
| 6060 | * start-tag or empty-element tag. |
| 6061 | */ |
| 6062 | for (i = 0; i < nbatts;i += 2) { |
| 6063 | if (xmlStrEqual(atts[i], attname)) { |
| 6064 | ctxt->errNo = XML_ERR_ATTRIBUTE_REDEFINED; |
| 6065 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6066 | ctxt->sax->error(ctxt->userData, |
| 6067 | "Attribute %s redefined\n", |
| 6068 | attname); |
| 6069 | ctxt->wellFormed = 0; |
| 6070 | ctxt->disableSAX = 1; |
| 6071 | xmlFree(attname); |
| 6072 | xmlFree(attvalue); |
| 6073 | goto failed; |
| 6074 | } |
| 6075 | } |
| 6076 | |
| 6077 | /* |
| 6078 | * Add the pair to atts |
| 6079 | */ |
| 6080 | if (atts == NULL) { |
| 6081 | maxatts = 10; |
| 6082 | atts = (const xmlChar **) xmlMalloc(maxatts * sizeof(xmlChar *)); |
| 6083 | if (atts == NULL) { |
| 6084 | xmlGenericError(xmlGenericErrorContext, |
| 6085 | "malloc of %ld byte failed\n", |
| 6086 | maxatts * (long)sizeof(xmlChar *)); |
| 6087 | return(NULL); |
| 6088 | } |
| 6089 | } else if (nbatts + 4 > maxatts) { |
| 6090 | maxatts *= 2; |
| 6091 | atts = (const xmlChar **) xmlRealloc((void *) atts, |
| 6092 | maxatts * sizeof(xmlChar *)); |
| 6093 | if (atts == NULL) { |
| 6094 | xmlGenericError(xmlGenericErrorContext, |
| 6095 | "realloc of %ld byte failed\n", |
| 6096 | maxatts * (long)sizeof(xmlChar *)); |
| 6097 | return(NULL); |
| 6098 | } |
| 6099 | } |
| 6100 | atts[nbatts++] = attname; |
| 6101 | atts[nbatts++] = attvalue; |
| 6102 | atts[nbatts] = NULL; |
| 6103 | atts[nbatts + 1] = NULL; |
| 6104 | } else { |
| 6105 | if (attname != NULL) |
| 6106 | xmlFree(attname); |
| 6107 | if (attvalue != NULL) |
| 6108 | xmlFree(attvalue); |
| 6109 | } |
| 6110 | |
| 6111 | failed: |
| 6112 | |
| 6113 | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
| 6114 | break; |
| 6115 | if (!IS_BLANK(RAW)) { |
| 6116 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 6117 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6118 | ctxt->sax->error(ctxt->userData, |
| 6119 | "attributes construct error\n"); |
| 6120 | ctxt->wellFormed = 0; |
| 6121 | ctxt->disableSAX = 1; |
| 6122 | } |
| 6123 | SKIP_BLANKS; |
| 6124 | if ((cons == ctxt->input->consumed) && (q == CUR_PTR)) { |
| 6125 | ctxt->errNo = XML_ERR_INTERNAL_ERROR; |
| 6126 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6127 | ctxt->sax->error(ctxt->userData, |
| 6128 | "xmlParseStartTag: problem parsing attributes\n"); |
| 6129 | ctxt->wellFormed = 0; |
| 6130 | ctxt->disableSAX = 1; |
| 6131 | break; |
| 6132 | } |
| 6133 | GROW; |
| 6134 | } |
| 6135 | |
| 6136 | /* |
| 6137 | * SAX: Start of Element ! |
| 6138 | */ |
| 6139 | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
| 6140 | (!ctxt->disableSAX)) |
| 6141 | ctxt->sax->startElement(ctxt->userData, name, atts); |
| 6142 | |
| 6143 | if (atts != NULL) { |
| 6144 | for (i = 0;i < nbatts;i++) xmlFree((xmlChar *) atts[i]); |
| 6145 | xmlFree((void *) atts); |
| 6146 | } |
| 6147 | return(name); |
| 6148 | } |
| 6149 | |
| 6150 | /** |
| 6151 | * xmlParseEndTag: |
| 6152 | * @ctxt: an XML parser context |
| 6153 | * |
| 6154 | * parse an end of tag |
| 6155 | * |
| 6156 | * [42] ETag ::= '</' Name S? '>' |
| 6157 | * |
| 6158 | * With namespace |
| 6159 | * |
| 6160 | * [NS 9] ETag ::= '</' QName S? '>' |
| 6161 | */ |
| 6162 | |
| 6163 | void |
| 6164 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
| 6165 | xmlChar *name; |
| 6166 | xmlChar *oldname; |
| 6167 | |
| 6168 | GROW; |
| 6169 | if ((RAW != '<') || (NXT(1) != '/')) { |
| 6170 | ctxt->errNo = XML_ERR_LTSLASH_REQUIRED; |
| 6171 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6172 | ctxt->sax->error(ctxt->userData, "xmlParseEndTag: '</' not found\n"); |
| 6173 | ctxt->wellFormed = 0; |
| 6174 | ctxt->disableSAX = 1; |
| 6175 | return; |
| 6176 | } |
| 6177 | SKIP(2); |
| 6178 | |
| 6179 | name = xmlParseName(ctxt); |
| 6180 | |
| 6181 | /* |
| 6182 | * We should definitely be at the ending "S? '>'" part |
| 6183 | */ |
| 6184 | GROW; |
| 6185 | SKIP_BLANKS; |
| 6186 | if ((!IS_CHAR(RAW)) || (RAW != '>')) { |
| 6187 | ctxt->errNo = XML_ERR_GT_REQUIRED; |
| 6188 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6189 | ctxt->sax->error(ctxt->userData, "End tag : expected '>'\n"); |
| 6190 | ctxt->wellFormed = 0; |
| 6191 | ctxt->disableSAX = 1; |
| 6192 | } else |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6193 | NEXT1; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6194 | |
| 6195 | /* |
| 6196 | * [ WFC: Element Type Match ] |
| 6197 | * The Name in an element's end-tag must match the element type in the |
| 6198 | * start-tag. |
| 6199 | * |
| 6200 | */ |
| 6201 | if ((name == NULL) || (ctxt->name == NULL) || |
| 6202 | (!xmlStrEqual(name, ctxt->name))) { |
| 6203 | ctxt->errNo = XML_ERR_TAG_NAME_MISMATCH; |
| 6204 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) { |
| 6205 | if ((name != NULL) && (ctxt->name != NULL)) { |
| 6206 | ctxt->sax->error(ctxt->userData, |
| 6207 | "Opening and ending tag mismatch: %s and %s\n", |
| 6208 | ctxt->name, name); |
| 6209 | } else if (ctxt->name != NULL) { |
| 6210 | ctxt->sax->error(ctxt->userData, |
| 6211 | "Ending tag eror for: %s\n", ctxt->name); |
| 6212 | } else { |
| 6213 | ctxt->sax->error(ctxt->userData, |
| 6214 | "Ending tag error: internal error ???\n"); |
| 6215 | } |
| 6216 | |
| 6217 | } |
| 6218 | ctxt->wellFormed = 0; |
| 6219 | ctxt->disableSAX = 1; |
| 6220 | } |
| 6221 | |
| 6222 | /* |
| 6223 | * SAX: End of Tag |
| 6224 | */ |
| 6225 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
| 6226 | (!ctxt->disableSAX)) |
| 6227 | ctxt->sax->endElement(ctxt->userData, name); |
| 6228 | |
| 6229 | if (name != NULL) |
| 6230 | xmlFree(name); |
| 6231 | oldname = namePop(ctxt); |
| 6232 | spacePop(ctxt); |
| 6233 | if (oldname != NULL) { |
| 6234 | #ifdef DEBUG_STACK |
| 6235 | xmlGenericError(xmlGenericErrorContext,"Close: popped %s\n", oldname); |
| 6236 | #endif |
| 6237 | xmlFree(oldname); |
| 6238 | } |
| 6239 | return; |
| 6240 | } |
| 6241 | |
| 6242 | /** |
| 6243 | * xmlParseCDSect: |
| 6244 | * @ctxt: an XML parser context |
| 6245 | * |
| 6246 | * Parse escaped pure raw content. |
| 6247 | * |
| 6248 | * [18] CDSect ::= CDStart CData CDEnd |
| 6249 | * |
| 6250 | * [19] CDStart ::= '<![CDATA[' |
| 6251 | * |
| 6252 | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
| 6253 | * |
| 6254 | * [21] CDEnd ::= ']]>' |
| 6255 | */ |
| 6256 | void |
| 6257 | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
| 6258 | xmlChar *buf = NULL; |
| 6259 | int len = 0; |
| 6260 | int size = XML_PARSER_BUFFER_SIZE; |
| 6261 | int r, rl; |
| 6262 | int s, sl; |
| 6263 | int cur, l; |
| 6264 | int count = 0; |
| 6265 | |
| 6266 | if ((NXT(0) == '<') && (NXT(1) == '!') && |
| 6267 | (NXT(2) == '[') && (NXT(3) == 'C') && |
| 6268 | (NXT(4) == 'D') && (NXT(5) == 'A') && |
| 6269 | (NXT(6) == 'T') && (NXT(7) == 'A') && |
| 6270 | (NXT(8) == '[')) { |
| 6271 | SKIP(9); |
| 6272 | } else |
| 6273 | return; |
| 6274 | |
| 6275 | ctxt->instate = XML_PARSER_CDATA_SECTION; |
| 6276 | r = CUR_CHAR(rl); |
| 6277 | if (!IS_CHAR(r)) { |
| 6278 | ctxt->errNo = XML_ERR_CDATA_NOT_FINISHED; |
| 6279 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6280 | ctxt->sax->error(ctxt->userData, |
| 6281 | "CData section not finished\n"); |
| 6282 | ctxt->wellFormed = 0; |
| 6283 | ctxt->disableSAX = 1; |
| 6284 | ctxt->instate = XML_PARSER_CONTENT; |
| 6285 | return; |
| 6286 | } |
| 6287 | NEXTL(rl); |
| 6288 | s = CUR_CHAR(sl); |
| 6289 | if (!IS_CHAR(s)) { |
| 6290 | ctxt->errNo = XML_ERR_CDATA_NOT_FINISHED; |
| 6291 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6292 | ctxt->sax->error(ctxt->userData, |
| 6293 | "CData section not finished\n"); |
| 6294 | ctxt->wellFormed = 0; |
| 6295 | ctxt->disableSAX = 1; |
| 6296 | ctxt->instate = XML_PARSER_CONTENT; |
| 6297 | return; |
| 6298 | } |
| 6299 | NEXTL(sl); |
| 6300 | cur = CUR_CHAR(l); |
| 6301 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 6302 | if (buf == NULL) { |
| 6303 | xmlGenericError(xmlGenericErrorContext, |
| 6304 | "malloc of %d byte failed\n", size); |
| 6305 | return; |
| 6306 | } |
| 6307 | while (IS_CHAR(cur) && |
| 6308 | ((r != ']') || (s != ']') || (cur != '>'))) { |
| 6309 | if (len + 5 >= size) { |
| 6310 | size *= 2; |
| 6311 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 6312 | if (buf == NULL) { |
| 6313 | xmlGenericError(xmlGenericErrorContext, |
| 6314 | "realloc of %d byte failed\n", size); |
| 6315 | return; |
| 6316 | } |
| 6317 | } |
| 6318 | COPY_BUF(rl,buf,len,r); |
| 6319 | r = s; |
| 6320 | rl = sl; |
| 6321 | s = cur; |
| 6322 | sl = l; |
| 6323 | count++; |
| 6324 | if (count > 50) { |
| 6325 | GROW; |
| 6326 | count = 0; |
| 6327 | } |
| 6328 | NEXTL(l); |
| 6329 | cur = CUR_CHAR(l); |
| 6330 | } |
| 6331 | buf[len] = 0; |
| 6332 | ctxt->instate = XML_PARSER_CONTENT; |
| 6333 | if (cur != '>') { |
| 6334 | ctxt->errNo = XML_ERR_CDATA_NOT_FINISHED; |
| 6335 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6336 | ctxt->sax->error(ctxt->userData, |
| 6337 | "CData section not finished\n%.50s\n", buf); |
| 6338 | ctxt->wellFormed = 0; |
| 6339 | ctxt->disableSAX = 1; |
| 6340 | xmlFree(buf); |
| 6341 | return; |
| 6342 | } |
| 6343 | NEXTL(l); |
| 6344 | |
| 6345 | /* |
| 6346 | * Ok the buffer is to be consumed as cdata. |
| 6347 | */ |
| 6348 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
| 6349 | if (ctxt->sax->cdataBlock != NULL) |
| 6350 | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
| 6351 | } |
| 6352 | xmlFree(buf); |
| 6353 | } |
| 6354 | |
| 6355 | /** |
| 6356 | * xmlParseContent: |
| 6357 | * @ctxt: an XML parser context |
| 6358 | * |
| 6359 | * Parse a content: |
| 6360 | * |
| 6361 | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
| 6362 | */ |
| 6363 | |
| 6364 | void |
| 6365 | xmlParseContent(xmlParserCtxtPtr ctxt) { |
| 6366 | GROW; |
| 6367 | while (((RAW != 0) || (ctxt->token != 0)) && |
| 6368 | ((RAW != '<') || (NXT(1) != '/'))) { |
| 6369 | const xmlChar *test = CUR_PTR; |
| 6370 | int cons = ctxt->input->consumed; |
Daniel Veillard | 04be4f5 | 2001-03-26 21:23:53 +0000 | [diff] [blame] | 6371 | int tok = ctxt->token; |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6372 | const xmlChar *cur = ctxt->input->cur; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6373 | |
| 6374 | /* |
| 6375 | * Handle possible processed charrefs. |
| 6376 | */ |
| 6377 | if (ctxt->token != 0) { |
| 6378 | xmlParseCharData(ctxt, 0); |
| 6379 | } |
| 6380 | /* |
| 6381 | * First case : a Processing Instruction. |
| 6382 | */ |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6383 | else if ((*cur == '<') && (cur[1] == '?')) { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6384 | xmlParsePI(ctxt); |
| 6385 | } |
| 6386 | |
| 6387 | /* |
| 6388 | * Second case : a CDSection |
| 6389 | */ |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6390 | else if ((*cur == '<') && (NXT(1) == '!') && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6391 | (NXT(2) == '[') && (NXT(3) == 'C') && |
| 6392 | (NXT(4) == 'D') && (NXT(5) == 'A') && |
| 6393 | (NXT(6) == 'T') && (NXT(7) == 'A') && |
| 6394 | (NXT(8) == '[')) { |
| 6395 | xmlParseCDSect(ctxt); |
| 6396 | } |
| 6397 | |
| 6398 | /* |
| 6399 | * Third case : a comment |
| 6400 | */ |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6401 | else if ((*cur == '<') && (NXT(1) == '!') && |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6402 | (NXT(2) == '-') && (NXT(3) == '-')) { |
| 6403 | xmlParseComment(ctxt); |
| 6404 | ctxt->instate = XML_PARSER_CONTENT; |
| 6405 | } |
| 6406 | |
| 6407 | /* |
| 6408 | * Fourth case : a sub-element. |
| 6409 | */ |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6410 | else if (*cur == '<') { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6411 | xmlParseElement(ctxt); |
| 6412 | } |
| 6413 | |
| 6414 | /* |
| 6415 | * Fifth case : a reference. If if has not been resolved, |
| 6416 | * parsing returns it's Name, create the node |
| 6417 | */ |
| 6418 | |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6419 | else if (*cur == '&') { |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6420 | xmlParseReference(ctxt); |
| 6421 | } |
| 6422 | |
| 6423 | /* |
| 6424 | * Last case, text. Note that References are handled directly. |
| 6425 | */ |
| 6426 | else { |
| 6427 | xmlParseCharData(ctxt, 0); |
| 6428 | } |
| 6429 | |
| 6430 | GROW; |
| 6431 | /* |
| 6432 | * Pop-up of finished entities. |
| 6433 | */ |
| 6434 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 6435 | xmlPopInput(ctxt); |
| 6436 | SHRINK; |
| 6437 | |
| 6438 | if ((cons == ctxt->input->consumed) && (test == CUR_PTR) && |
| 6439 | (tok == ctxt->token)) { |
| 6440 | ctxt->errNo = XML_ERR_INTERNAL_ERROR; |
| 6441 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6442 | ctxt->sax->error(ctxt->userData, |
| 6443 | "detected an error in element content\n"); |
| 6444 | ctxt->wellFormed = 0; |
| 6445 | ctxt->disableSAX = 1; |
| 6446 | ctxt->instate = XML_PARSER_EOF; |
| 6447 | break; |
| 6448 | } |
| 6449 | } |
| 6450 | } |
| 6451 | |
| 6452 | /** |
| 6453 | * xmlParseElement: |
| 6454 | * @ctxt: an XML parser context |
| 6455 | * |
| 6456 | * parse an XML element, this is highly recursive |
| 6457 | * |
| 6458 | * [39] element ::= EmptyElemTag | STag content ETag |
| 6459 | * |
| 6460 | * [ WFC: Element Type Match ] |
| 6461 | * The Name in an element's end-tag must match the element type in the |
| 6462 | * start-tag. |
| 6463 | * |
| 6464 | * [ VC: Element Valid ] |
| 6465 | * An element is valid if there is a declaration matching elementdecl |
| 6466 | * where the Name matches the element type and one of the following holds: |
| 6467 | * - The declaration matches EMPTY and the element has no content. |
| 6468 | * - The declaration matches children and the sequence of child elements |
| 6469 | * belongs to the language generated by the regular expression in the |
| 6470 | * content model, with optional white space (characters matching the |
| 6471 | * nonterminal S) between each pair of child elements. |
| 6472 | * - The declaration matches Mixed and the content consists of character |
| 6473 | * data and child elements whose types match names in the content model. |
| 6474 | * - The declaration matches ANY, and the types of any child elements have |
| 6475 | * been declared. |
| 6476 | */ |
| 6477 | |
| 6478 | void |
| 6479 | xmlParseElement(xmlParserCtxtPtr ctxt) { |
| 6480 | const xmlChar *openTag = CUR_PTR; |
| 6481 | xmlChar *name; |
| 6482 | xmlChar *oldname; |
| 6483 | xmlParserNodeInfo node_info; |
| 6484 | xmlNodePtr ret; |
| 6485 | |
| 6486 | /* Capture start position */ |
| 6487 | if (ctxt->record_info) { |
| 6488 | node_info.begin_pos = ctxt->input->consumed + |
| 6489 | (CUR_PTR - ctxt->input->base); |
| 6490 | node_info.begin_line = ctxt->input->line; |
| 6491 | } |
| 6492 | |
| 6493 | if (ctxt->spaceNr == 0) |
| 6494 | spacePush(ctxt, -1); |
| 6495 | else |
| 6496 | spacePush(ctxt, *ctxt->space); |
| 6497 | |
| 6498 | name = xmlParseStartTag(ctxt); |
| 6499 | if (name == NULL) { |
| 6500 | spacePop(ctxt); |
| 6501 | return; |
| 6502 | } |
| 6503 | namePush(ctxt, name); |
| 6504 | ret = ctxt->node; |
| 6505 | |
| 6506 | /* |
| 6507 | * [ VC: Root Element Type ] |
| 6508 | * The Name in the document type declaration must match the element |
| 6509 | * type of the root element. |
| 6510 | */ |
| 6511 | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
| 6512 | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
| 6513 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
| 6514 | |
| 6515 | /* |
| 6516 | * Check for an Empty Element. |
| 6517 | */ |
| 6518 | if ((RAW == '/') && (NXT(1) == '>')) { |
| 6519 | SKIP(2); |
| 6520 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
| 6521 | (!ctxt->disableSAX)) |
| 6522 | ctxt->sax->endElement(ctxt->userData, name); |
| 6523 | oldname = namePop(ctxt); |
| 6524 | spacePop(ctxt); |
| 6525 | if (oldname != NULL) { |
| 6526 | #ifdef DEBUG_STACK |
| 6527 | xmlGenericError(xmlGenericErrorContext,"Close: popped %s\n", oldname); |
| 6528 | #endif |
| 6529 | xmlFree(oldname); |
| 6530 | } |
| 6531 | if ( ret != NULL && ctxt->record_info ) { |
| 6532 | node_info.end_pos = ctxt->input->consumed + |
| 6533 | (CUR_PTR - ctxt->input->base); |
| 6534 | node_info.end_line = ctxt->input->line; |
| 6535 | node_info.node = ret; |
| 6536 | xmlParserAddNodeInfo(ctxt, &node_info); |
| 6537 | } |
| 6538 | return; |
| 6539 | } |
| 6540 | if (RAW == '>') { |
Daniel Veillard | 21a0f91 | 2001-02-25 19:54:14 +0000 | [diff] [blame] | 6541 | NEXT1; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 6542 | } else { |
| 6543 | ctxt->errNo = XML_ERR_GT_REQUIRED; |
| 6544 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6545 | ctxt->sax->error(ctxt->userData, |
| 6546 | "Couldn't find end of Start Tag\n%.30s\n", |
| 6547 | openTag); |
| 6548 | ctxt->wellFormed = 0; |
| 6549 | ctxt->disableSAX = 1; |
| 6550 | |
| 6551 | /* |
| 6552 | * end of parsing of this node. |
| 6553 | */ |
| 6554 | nodePop(ctxt); |
| 6555 | oldname = namePop(ctxt); |
| 6556 | spacePop(ctxt); |
| 6557 | if (oldname != NULL) { |
| 6558 | #ifdef DEBUG_STACK |
| 6559 | xmlGenericError(xmlGenericErrorContext,"Close: popped %s\n", oldname); |
| 6560 | #endif |
| 6561 | xmlFree(oldname); |
| 6562 | } |
| 6563 | |
| 6564 | /* |
| 6565 | * Capture end position and add node |
| 6566 | */ |
| 6567 | if ( ret != NULL && ctxt->record_info ) { |
| 6568 | node_info.end_pos = ctxt->input->consumed + |
| 6569 | (CUR_PTR - ctxt->input->base); |
| 6570 | node_info.end_line = ctxt->input->line; |
| 6571 | node_info.node = ret; |
| 6572 | xmlParserAddNodeInfo(ctxt, &node_info); |
| 6573 | } |
| 6574 | return; |
| 6575 | } |
| 6576 | |
| 6577 | /* |
| 6578 | * Parse the content of the element: |
| 6579 | */ |
| 6580 | xmlParseContent(ctxt); |
| 6581 | if (!IS_CHAR(RAW)) { |
| 6582 | ctxt->errNo = XML_ERR_TAG_NOT_FINISED; |
| 6583 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6584 | ctxt->sax->error(ctxt->userData, |
| 6585 | "Premature end of data in tag %.30s\n", openTag); |
| 6586 | ctxt->wellFormed = 0; |
| 6587 | ctxt->disableSAX = 1; |
| 6588 | |
| 6589 | /* |
| 6590 | * end of parsing of this node. |
| 6591 | */ |
| 6592 | nodePop(ctxt); |
| 6593 | oldname = namePop(ctxt); |
| 6594 | spacePop(ctxt); |
| 6595 | if (oldname != NULL) { |
| 6596 | #ifdef DEBUG_STACK |
| 6597 | xmlGenericError(xmlGenericErrorContext,"Close: popped %s\n", oldname); |
| 6598 | #endif |
| 6599 | xmlFree(oldname); |
| 6600 | } |
| 6601 | return; |
| 6602 | } |
| 6603 | |
| 6604 | /* |
| 6605 | * parse the end of tag: '</' should be here. |
| 6606 | */ |
| 6607 | xmlParseEndTag(ctxt); |
| 6608 | |
| 6609 | /* |
| 6610 | * Capture end position and add node |
| 6611 | */ |
| 6612 | if ( ret != NULL && ctxt->record_info ) { |
| 6613 | node_info.end_pos = ctxt->input->consumed + |
| 6614 | (CUR_PTR - ctxt->input->base); |
| 6615 | node_info.end_line = ctxt->input->line; |
| 6616 | node_info.node = ret; |
| 6617 | xmlParserAddNodeInfo(ctxt, &node_info); |
| 6618 | } |
| 6619 | } |
| 6620 | |
| 6621 | /** |
| 6622 | * xmlParseVersionNum: |
| 6623 | * @ctxt: an XML parser context |
| 6624 | * |
| 6625 | * parse the XML version value. |
| 6626 | * |
| 6627 | * [26] VersionNum ::= ([a-zA-Z0-9_.:] | '-')+ |
| 6628 | * |
| 6629 | * Returns the string giving the XML version number, or NULL |
| 6630 | */ |
| 6631 | xmlChar * |
| 6632 | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
| 6633 | xmlChar *buf = NULL; |
| 6634 | int len = 0; |
| 6635 | int size = 10; |
| 6636 | xmlChar cur; |
| 6637 | |
| 6638 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 6639 | if (buf == NULL) { |
| 6640 | xmlGenericError(xmlGenericErrorContext, |
| 6641 | "malloc of %d byte failed\n", size); |
| 6642 | return(NULL); |
| 6643 | } |
| 6644 | cur = CUR; |
| 6645 | while (((cur >= 'a') && (cur <= 'z')) || |
| 6646 | ((cur >= 'A') && (cur <= 'Z')) || |
| 6647 | ((cur >= '0') && (cur <= '9')) || |
| 6648 | (cur == '_') || (cur == '.') || |
| 6649 | (cur == ':') || (cur == '-')) { |
| 6650 | if (len + 1 >= size) { |
| 6651 | size *= 2; |
| 6652 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 6653 | if (buf == NULL) { |
| 6654 | xmlGenericError(xmlGenericErrorContext, |
| 6655 | "realloc of %d byte failed\n", size); |
| 6656 | return(NULL); |
| 6657 | } |
| 6658 | } |
| 6659 | buf[len++] = cur; |
| 6660 | NEXT; |
| 6661 | cur=CUR; |
| 6662 | } |
| 6663 | buf[len] = 0; |
| 6664 | return(buf); |
| 6665 | } |
| 6666 | |
| 6667 | /** |
| 6668 | * xmlParseVersionInfo: |
| 6669 | * @ctxt: an XML parser context |
| 6670 | * |
| 6671 | * parse the XML version. |
| 6672 | * |
| 6673 | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
| 6674 | * |
| 6675 | * [25] Eq ::= S? '=' S? |
| 6676 | * |
| 6677 | * Returns the version string, e.g. "1.0" |
| 6678 | */ |
| 6679 | |
| 6680 | xmlChar * |
| 6681 | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
| 6682 | xmlChar *version = NULL; |
| 6683 | const xmlChar *q; |
| 6684 | |
| 6685 | if ((RAW == 'v') && (NXT(1) == 'e') && |
| 6686 | (NXT(2) == 'r') && (NXT(3) == 's') && |
| 6687 | (NXT(4) == 'i') && (NXT(5) == 'o') && |
| 6688 | (NXT(6) == 'n')) { |
| 6689 | SKIP(7); |
| 6690 | SKIP_BLANKS; |
| 6691 | if (RAW != '=') { |
| 6692 | ctxt->errNo = XML_ERR_EQUAL_REQUIRED; |
| 6693 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6694 | ctxt->sax->error(ctxt->userData, |
| 6695 | "xmlParseVersionInfo : expected '='\n"); |
| 6696 | ctxt->wellFormed = 0; |
| 6697 | ctxt->disableSAX = 1; |
| 6698 | return(NULL); |
| 6699 | } |
| 6700 | NEXT; |
| 6701 | SKIP_BLANKS; |
| 6702 | if (RAW == '"') { |
| 6703 | NEXT; |
| 6704 | q = CUR_PTR; |
| 6705 | version = xmlParseVersionNum(ctxt); |
| 6706 | if (RAW != '"') { |
| 6707 | ctxt->errNo = XML_ERR_STRING_NOT_CLOSED; |
| 6708 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6709 | ctxt->sax->error(ctxt->userData, |
| 6710 | "String not closed\n%.50s\n", q); |
| 6711 | ctxt->wellFormed = 0; |
| 6712 | ctxt->disableSAX = 1; |
| 6713 | } else |
| 6714 | NEXT; |
| 6715 | } else if (RAW == '\''){ |
| 6716 | NEXT; |
| 6717 | q = CUR_PTR; |
| 6718 | version = xmlParseVersionNum(ctxt); |
| 6719 | if (RAW != '\'') { |
| 6720 | ctxt->errNo = XML_ERR_STRING_NOT_CLOSED; |
| 6721 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6722 | ctxt->sax->error(ctxt->userData, |
| 6723 | "String not closed\n%.50s\n", q); |
| 6724 | ctxt->wellFormed = 0; |
| 6725 | ctxt->disableSAX = 1; |
| 6726 | } else |
| 6727 | NEXT; |
| 6728 | } else { |
| 6729 | ctxt->errNo = XML_ERR_STRING_NOT_STARTED; |
| 6730 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6731 | ctxt->sax->error(ctxt->userData, |
| 6732 | "xmlParseVersionInfo : expected ' or \"\n"); |
| 6733 | ctxt->wellFormed = 0; |
| 6734 | ctxt->disableSAX = 1; |
| 6735 | } |
| 6736 | } |
| 6737 | return(version); |
| 6738 | } |
| 6739 | |
| 6740 | /** |
| 6741 | * xmlParseEncName: |
| 6742 | * @ctxt: an XML parser context |
| 6743 | * |
| 6744 | * parse the XML encoding name |
| 6745 | * |
| 6746 | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
| 6747 | * |
| 6748 | * Returns the encoding name value or NULL |
| 6749 | */ |
| 6750 | xmlChar * |
| 6751 | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
| 6752 | xmlChar *buf = NULL; |
| 6753 | int len = 0; |
| 6754 | int size = 10; |
| 6755 | xmlChar cur; |
| 6756 | |
| 6757 | cur = CUR; |
| 6758 | if (((cur >= 'a') && (cur <= 'z')) || |
| 6759 | ((cur >= 'A') && (cur <= 'Z'))) { |
| 6760 | buf = (xmlChar *) xmlMalloc(size * sizeof(xmlChar)); |
| 6761 | if (buf == NULL) { |
| 6762 | xmlGenericError(xmlGenericErrorContext, |
| 6763 | "malloc of %d byte failed\n", size); |
| 6764 | return(NULL); |
| 6765 | } |
| 6766 | |
| 6767 | buf[len++] = cur; |
| 6768 | NEXT; |
| 6769 | cur = CUR; |
| 6770 | while (((cur >= 'a') && (cur <= 'z')) || |
| 6771 | ((cur >= 'A') && (cur <= 'Z')) || |
| 6772 | ((cur >= '0') && (cur <= '9')) || |
| 6773 | (cur == '.') || (cur == '_') || |
| 6774 | (cur == '-')) { |
| 6775 | if (len + 1 >= size) { |
| 6776 | size *= 2; |
| 6777 | buf = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
| 6778 | if (buf == NULL) { |
| 6779 | xmlGenericError(xmlGenericErrorContext, |
| 6780 | "realloc of %d byte failed\n", size); |
| 6781 | return(NULL); |
| 6782 | } |
| 6783 | } |
| 6784 | buf[len++] = cur; |
| 6785 | NEXT; |
| 6786 | cur = CUR; |
| 6787 | if (cur == 0) { |
| 6788 | SHRINK; |
| 6789 | GROW; |
| 6790 | cur = CUR; |
| 6791 | } |
| 6792 | } |
| 6793 | buf[len] = 0; |
| 6794 | } else { |
| 6795 | ctxt->errNo = XML_ERR_ENCODING_NAME; |
| 6796 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6797 | ctxt->sax->error(ctxt->userData, "Invalid XML encoding name\n"); |
| 6798 | ctxt->wellFormed = 0; |
| 6799 | ctxt->disableSAX = 1; |
| 6800 | } |
| 6801 | return(buf); |
| 6802 | } |
| 6803 | |
| 6804 | /** |
| 6805 | * xmlParseEncodingDecl: |
| 6806 | * @ctxt: an XML parser context |
| 6807 | * |
| 6808 | * parse the XML encoding declaration |
| 6809 | * |
| 6810 | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
| 6811 | * |
| 6812 | * this setups the conversion filters. |
| 6813 | * |
| 6814 | * Returns the encoding value or NULL |
| 6815 | */ |
| 6816 | |
| 6817 | xmlChar * |
| 6818 | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
| 6819 | xmlChar *encoding = NULL; |
| 6820 | const xmlChar *q; |
| 6821 | |
| 6822 | SKIP_BLANKS; |
| 6823 | if ((RAW == 'e') && (NXT(1) == 'n') && |
| 6824 | (NXT(2) == 'c') && (NXT(3) == 'o') && |
| 6825 | (NXT(4) == 'd') && (NXT(5) == 'i') && |
| 6826 | (NXT(6) == 'n') && (NXT(7) == 'g')) { |
| 6827 | SKIP(8); |
| 6828 | SKIP_BLANKS; |
| 6829 | if (RAW != '=') { |
| 6830 | ctxt->errNo = XML_ERR_EQUAL_REQUIRED; |
| 6831 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6832 | ctxt->sax->error(ctxt->userData, |
| 6833 | "xmlParseEncodingDecl : expected '='\n"); |
| 6834 | ctxt->wellFormed = 0; |
| 6835 | ctxt->disableSAX = 1; |
| 6836 | return(NULL); |
| 6837 | } |
| 6838 | NEXT; |
| 6839 | SKIP_BLANKS; |
| 6840 | if (RAW == '"') { |
| 6841 | NEXT; |
| 6842 | q = CUR_PTR; |
| 6843 | encoding = xmlParseEncName(ctxt); |
| 6844 | if (RAW != '"') { |
| 6845 | ctxt->errNo = XML_ERR_STRING_NOT_CLOSED; |
| 6846 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6847 | ctxt->sax->error(ctxt->userData, |
| 6848 | "String not closed\n%.50s\n", q); |
| 6849 | ctxt->wellFormed = 0; |
| 6850 | ctxt->disableSAX = 1; |
| 6851 | } else |
| 6852 | NEXT; |
| 6853 | } else if (RAW == '\''){ |
| 6854 | NEXT; |
| 6855 | q = CUR_PTR; |
| 6856 | encoding = xmlParseEncName(ctxt); |
| 6857 | if (RAW != '\'') { |
| 6858 | ctxt->errNo = XML_ERR_STRING_NOT_CLOSED; |
| 6859 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6860 | ctxt->sax->error(ctxt->userData, |
| 6861 | "String not closed\n%.50s\n", q); |
| 6862 | ctxt->wellFormed = 0; |
| 6863 | ctxt->disableSAX = 1; |
| 6864 | } else |
| 6865 | NEXT; |
| 6866 | } else if (RAW == '"'){ |
| 6867 | ctxt->errNo = XML_ERR_STRING_NOT_STARTED; |
| 6868 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6869 | ctxt->sax->error(ctxt->userData, |
| 6870 | "xmlParseEncodingDecl : expected ' or \"\n"); |
| 6871 | ctxt->wellFormed = 0; |
| 6872 | ctxt->disableSAX = 1; |
| 6873 | } |
| 6874 | if (encoding != NULL) { |
| 6875 | xmlCharEncoding enc; |
| 6876 | xmlCharEncodingHandlerPtr handler; |
| 6877 | |
| 6878 | if (ctxt->input->encoding != NULL) |
| 6879 | xmlFree((xmlChar *) ctxt->input->encoding); |
| 6880 | ctxt->input->encoding = encoding; |
| 6881 | |
| 6882 | enc = xmlParseCharEncoding((const char *) encoding); |
| 6883 | /* |
| 6884 | * registered set of known encodings |
| 6885 | */ |
| 6886 | if (enc != XML_CHAR_ENCODING_ERROR) { |
| 6887 | xmlSwitchEncoding(ctxt, enc); |
| 6888 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 6889 | xmlFree(encoding); |
| 6890 | return(NULL); |
| 6891 | } |
| 6892 | } else { |
| 6893 | /* |
| 6894 | * fallback for unknown encodings |
| 6895 | */ |
| 6896 | handler = xmlFindCharEncodingHandler((const char *) encoding); |
| 6897 | if (handler != NULL) { |
| 6898 | xmlSwitchToEncoding(ctxt, handler); |
| 6899 | } else { |
| 6900 | ctxt->errNo = XML_ERR_UNSUPPORTED_ENCODING; |
| 6901 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6902 | ctxt->sax->error(ctxt->userData, |
| 6903 | "Unsupported encoding %s\n", encoding); |
| 6904 | return(NULL); |
| 6905 | } |
| 6906 | } |
| 6907 | } |
| 6908 | } |
| 6909 | return(encoding); |
| 6910 | } |
| 6911 | |
| 6912 | /** |
| 6913 | * xmlParseSDDecl: |
| 6914 | * @ctxt: an XML parser context |
| 6915 | * |
| 6916 | * parse the XML standalone declaration |
| 6917 | * |
| 6918 | * [32] SDDecl ::= S 'standalone' Eq |
| 6919 | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
| 6920 | * |
| 6921 | * [ VC: Standalone Document Declaration ] |
| 6922 | * TODO The standalone document declaration must have the value "no" |
| 6923 | * if any external markup declarations contain declarations of: |
| 6924 | * - attributes with default values, if elements to which these |
| 6925 | * attributes apply appear in the document without specifications |
| 6926 | * of values for these attributes, or |
| 6927 | * - entities (other than amp, lt, gt, apos, quot), if references |
| 6928 | * to those entities appear in the document, or |
| 6929 | * - attributes with values subject to normalization, where the |
| 6930 | * attribute appears in the document with a value which will change |
| 6931 | * as a result of normalization, or |
| 6932 | * - element types with element content, if white space occurs directly |
| 6933 | * within any instance of those types. |
| 6934 | * |
| 6935 | * Returns 1 if standalone, 0 otherwise |
| 6936 | */ |
| 6937 | |
| 6938 | int |
| 6939 | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
| 6940 | int standalone = -1; |
| 6941 | |
| 6942 | SKIP_BLANKS; |
| 6943 | if ((RAW == 's') && (NXT(1) == 't') && |
| 6944 | (NXT(2) == 'a') && (NXT(3) == 'n') && |
| 6945 | (NXT(4) == 'd') && (NXT(5) == 'a') && |
| 6946 | (NXT(6) == 'l') && (NXT(7) == 'o') && |
| 6947 | (NXT(8) == 'n') && (NXT(9) == 'e')) { |
| 6948 | SKIP(10); |
| 6949 | SKIP_BLANKS; |
| 6950 | if (RAW != '=') { |
| 6951 | ctxt->errNo = XML_ERR_EQUAL_REQUIRED; |
| 6952 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6953 | ctxt->sax->error(ctxt->userData, |
| 6954 | "XML standalone declaration : expected '='\n"); |
| 6955 | ctxt->wellFormed = 0; |
| 6956 | ctxt->disableSAX = 1; |
| 6957 | return(standalone); |
| 6958 | } |
| 6959 | NEXT; |
| 6960 | SKIP_BLANKS; |
| 6961 | if (RAW == '\''){ |
| 6962 | NEXT; |
| 6963 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
| 6964 | standalone = 0; |
| 6965 | SKIP(2); |
| 6966 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
| 6967 | (NXT(2) == 's')) { |
| 6968 | standalone = 1; |
| 6969 | SKIP(3); |
| 6970 | } else { |
| 6971 | ctxt->errNo = XML_ERR_STANDALONE_VALUE; |
| 6972 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6973 | ctxt->sax->error(ctxt->userData, |
| 6974 | "standalone accepts only 'yes' or 'no'\n"); |
| 6975 | ctxt->wellFormed = 0; |
| 6976 | ctxt->disableSAX = 1; |
| 6977 | } |
| 6978 | if (RAW != '\'') { |
| 6979 | ctxt->errNo = XML_ERR_STRING_NOT_CLOSED; |
| 6980 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6981 | ctxt->sax->error(ctxt->userData, "String not closed\n"); |
| 6982 | ctxt->wellFormed = 0; |
| 6983 | ctxt->disableSAX = 1; |
| 6984 | } else |
| 6985 | NEXT; |
| 6986 | } else if (RAW == '"'){ |
| 6987 | NEXT; |
| 6988 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
| 6989 | standalone = 0; |
| 6990 | SKIP(2); |
| 6991 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
| 6992 | (NXT(2) == 's')) { |
| 6993 | standalone = 1; |
| 6994 | SKIP(3); |
| 6995 | } else { |
| 6996 | ctxt->errNo = XML_ERR_STANDALONE_VALUE; |
| 6997 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 6998 | ctxt->sax->error(ctxt->userData, |
| 6999 | "standalone accepts only 'yes' or 'no'\n"); |
| 7000 | ctxt->wellFormed = 0; |
| 7001 | ctxt->disableSAX = 1; |
| 7002 | } |
| 7003 | if (RAW != '"') { |
| 7004 | ctxt->errNo = XML_ERR_STRING_NOT_CLOSED; |
| 7005 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7006 | ctxt->sax->error(ctxt->userData, "String not closed\n"); |
| 7007 | ctxt->wellFormed = 0; |
| 7008 | ctxt->disableSAX = 1; |
| 7009 | } else |
| 7010 | NEXT; |
| 7011 | } else { |
| 7012 | ctxt->errNo = XML_ERR_STRING_NOT_STARTED; |
| 7013 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7014 | ctxt->sax->error(ctxt->userData, |
| 7015 | "Standalone value not found\n"); |
| 7016 | ctxt->wellFormed = 0; |
| 7017 | ctxt->disableSAX = 1; |
| 7018 | } |
| 7019 | } |
| 7020 | return(standalone); |
| 7021 | } |
| 7022 | |
| 7023 | /** |
| 7024 | * xmlParseXMLDecl: |
| 7025 | * @ctxt: an XML parser context |
| 7026 | * |
| 7027 | * parse an XML declaration header |
| 7028 | * |
| 7029 | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
| 7030 | */ |
| 7031 | |
| 7032 | void |
| 7033 | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
| 7034 | xmlChar *version; |
| 7035 | |
| 7036 | /* |
| 7037 | * We know that '<?xml' is here. |
| 7038 | */ |
| 7039 | SKIP(5); |
| 7040 | |
| 7041 | if (!IS_BLANK(RAW)) { |
| 7042 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 7043 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7044 | ctxt->sax->error(ctxt->userData, "Blank needed after '<?xml'\n"); |
| 7045 | ctxt->wellFormed = 0; |
| 7046 | ctxt->disableSAX = 1; |
| 7047 | } |
| 7048 | SKIP_BLANKS; |
| 7049 | |
| 7050 | /* |
| 7051 | * We should have the VersionInfo here. |
| 7052 | */ |
| 7053 | version = xmlParseVersionInfo(ctxt); |
| 7054 | if (version == NULL) |
| 7055 | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
| 7056 | ctxt->version = xmlStrdup(version); |
| 7057 | xmlFree(version); |
| 7058 | |
| 7059 | /* |
| 7060 | * We may have the encoding declaration |
| 7061 | */ |
| 7062 | if (!IS_BLANK(RAW)) { |
| 7063 | if ((RAW == '?') && (NXT(1) == '>')) { |
| 7064 | SKIP(2); |
| 7065 | return; |
| 7066 | } |
| 7067 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 7068 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7069 | ctxt->sax->error(ctxt->userData, "Blank needed here\n"); |
| 7070 | ctxt->wellFormed = 0; |
| 7071 | ctxt->disableSAX = 1; |
| 7072 | } |
| 7073 | xmlParseEncodingDecl(ctxt); |
| 7074 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 7075 | /* |
| 7076 | * The XML REC instructs us to stop parsing right here |
| 7077 | */ |
| 7078 | return; |
| 7079 | } |
| 7080 | |
| 7081 | /* |
| 7082 | * We may have the standalone status. |
| 7083 | */ |
| 7084 | if ((ctxt->input->encoding != NULL) && (!IS_BLANK(RAW))) { |
| 7085 | if ((RAW == '?') && (NXT(1) == '>')) { |
| 7086 | SKIP(2); |
| 7087 | return; |
| 7088 | } |
| 7089 | ctxt->errNo = XML_ERR_SPACE_REQUIRED; |
| 7090 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7091 | ctxt->sax->error(ctxt->userData, "Blank needed here\n"); |
| 7092 | ctxt->wellFormed = 0; |
| 7093 | ctxt->disableSAX = 1; |
| 7094 | } |
| 7095 | SKIP_BLANKS; |
| 7096 | ctxt->input->standalone = xmlParseSDDecl(ctxt); |
| 7097 | |
| 7098 | SKIP_BLANKS; |
| 7099 | if ((RAW == '?') && (NXT(1) == '>')) { |
| 7100 | SKIP(2); |
| 7101 | } else if (RAW == '>') { |
| 7102 | /* Deprecated old WD ... */ |
| 7103 | ctxt->errNo = XML_ERR_XMLDECL_NOT_FINISHED; |
| 7104 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7105 | ctxt->sax->error(ctxt->userData, |
| 7106 | "XML declaration must end-up with '?>'\n"); |
| 7107 | ctxt->wellFormed = 0; |
| 7108 | ctxt->disableSAX = 1; |
| 7109 | NEXT; |
| 7110 | } else { |
| 7111 | ctxt->errNo = XML_ERR_XMLDECL_NOT_FINISHED; |
| 7112 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7113 | ctxt->sax->error(ctxt->userData, |
| 7114 | "parsing XML declaration: '?>' expected\n"); |
| 7115 | ctxt->wellFormed = 0; |
| 7116 | ctxt->disableSAX = 1; |
| 7117 | MOVETO_ENDTAG(CUR_PTR); |
| 7118 | NEXT; |
| 7119 | } |
| 7120 | } |
| 7121 | |
| 7122 | /** |
| 7123 | * xmlParseMisc: |
| 7124 | * @ctxt: an XML parser context |
| 7125 | * |
| 7126 | * parse an XML Misc* optionnal field. |
| 7127 | * |
| 7128 | * [27] Misc ::= Comment | PI | S |
| 7129 | */ |
| 7130 | |
| 7131 | void |
| 7132 | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
| 7133 | while (((RAW == '<') && (NXT(1) == '?')) || |
| 7134 | ((RAW == '<') && (NXT(1) == '!') && |
| 7135 | (NXT(2) == '-') && (NXT(3) == '-')) || |
| 7136 | IS_BLANK(CUR)) { |
| 7137 | if ((RAW == '<') && (NXT(1) == '?')) { |
| 7138 | xmlParsePI(ctxt); |
| 7139 | } else if (IS_BLANK(CUR)) { |
| 7140 | NEXT; |
| 7141 | } else |
| 7142 | xmlParseComment(ctxt); |
| 7143 | } |
| 7144 | } |
| 7145 | |
| 7146 | /** |
| 7147 | * xmlParseDocument: |
| 7148 | * @ctxt: an XML parser context |
| 7149 | * |
| 7150 | * parse an XML document (and build a tree if using the standard SAX |
| 7151 | * interface). |
| 7152 | * |
| 7153 | * [1] document ::= prolog element Misc* |
| 7154 | * |
| 7155 | * [22] prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? |
| 7156 | * |
| 7157 | * Returns 0, -1 in case of error. the parser context is augmented |
| 7158 | * as a result of the parsing. |
| 7159 | */ |
| 7160 | |
| 7161 | int |
| 7162 | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
| 7163 | xmlChar start[4]; |
| 7164 | xmlCharEncoding enc; |
| 7165 | |
| 7166 | xmlInitParser(); |
| 7167 | |
| 7168 | GROW; |
| 7169 | |
| 7170 | /* |
| 7171 | * SAX: beginning of the document processing. |
| 7172 | */ |
| 7173 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
| 7174 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
| 7175 | |
| 7176 | /* |
| 7177 | * Get the 4 first bytes and decode the charset |
| 7178 | * if enc != XML_CHAR_ENCODING_NONE |
| 7179 | * plug some encoding conversion routines. |
| 7180 | */ |
| 7181 | start[0] = RAW; |
| 7182 | start[1] = NXT(1); |
| 7183 | start[2] = NXT(2); |
| 7184 | start[3] = NXT(3); |
| 7185 | enc = xmlDetectCharEncoding(start, 4); |
| 7186 | if (enc != XML_CHAR_ENCODING_NONE) { |
| 7187 | xmlSwitchEncoding(ctxt, enc); |
| 7188 | } |
| 7189 | |
| 7190 | |
| 7191 | if (CUR == 0) { |
| 7192 | ctxt->errNo = XML_ERR_DOCUMENT_EMPTY; |
| 7193 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7194 | ctxt->sax->error(ctxt->userData, "Document is empty\n"); |
| 7195 | ctxt->wellFormed = 0; |
| 7196 | ctxt->disableSAX = 1; |
| 7197 | } |
| 7198 | |
| 7199 | /* |
| 7200 | * Check for the XMLDecl in the Prolog. |
| 7201 | */ |
| 7202 | GROW; |
| 7203 | if ((RAW == '<') && (NXT(1) == '?') && |
| 7204 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 7205 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 7206 | |
| 7207 | /* |
| 7208 | * Note that we will switch encoding on the fly. |
| 7209 | */ |
| 7210 | xmlParseXMLDecl(ctxt); |
| 7211 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 7212 | /* |
| 7213 | * The XML REC instructs us to stop parsing right here |
| 7214 | */ |
| 7215 | return(-1); |
| 7216 | } |
| 7217 | ctxt->standalone = ctxt->input->standalone; |
| 7218 | SKIP_BLANKS; |
| 7219 | } else { |
| 7220 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
| 7221 | } |
| 7222 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
| 7223 | ctxt->sax->startDocument(ctxt->userData); |
| 7224 | |
| 7225 | /* |
| 7226 | * The Misc part of the Prolog |
| 7227 | */ |
| 7228 | GROW; |
| 7229 | xmlParseMisc(ctxt); |
| 7230 | |
| 7231 | /* |
| 7232 | * Then possibly doc type declaration(s) and more Misc |
| 7233 | * (doctypedecl Misc*)? |
| 7234 | */ |
| 7235 | GROW; |
| 7236 | if ((RAW == '<') && (NXT(1) == '!') && |
| 7237 | (NXT(2) == 'D') && (NXT(3) == 'O') && |
| 7238 | (NXT(4) == 'C') && (NXT(5) == 'T') && |
| 7239 | (NXT(6) == 'Y') && (NXT(7) == 'P') && |
| 7240 | (NXT(8) == 'E')) { |
| 7241 | |
| 7242 | ctxt->inSubset = 1; |
| 7243 | xmlParseDocTypeDecl(ctxt); |
| 7244 | if (RAW == '[') { |
| 7245 | ctxt->instate = XML_PARSER_DTD; |
| 7246 | xmlParseInternalSubset(ctxt); |
| 7247 | } |
| 7248 | |
| 7249 | /* |
| 7250 | * Create and update the external subset. |
| 7251 | */ |
| 7252 | ctxt->inSubset = 2; |
| 7253 | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
| 7254 | (!ctxt->disableSAX)) |
| 7255 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
| 7256 | ctxt->extSubSystem, ctxt->extSubURI); |
| 7257 | ctxt->inSubset = 0; |
| 7258 | |
| 7259 | |
| 7260 | ctxt->instate = XML_PARSER_PROLOG; |
| 7261 | xmlParseMisc(ctxt); |
| 7262 | } |
| 7263 | |
| 7264 | /* |
| 7265 | * Time to start parsing the tree itself |
| 7266 | */ |
| 7267 | GROW; |
| 7268 | if (RAW != '<') { |
| 7269 | ctxt->errNo = XML_ERR_DOCUMENT_EMPTY; |
| 7270 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7271 | ctxt->sax->error(ctxt->userData, |
| 7272 | "Start tag expected, '<' not found\n"); |
| 7273 | ctxt->wellFormed = 0; |
| 7274 | ctxt->disableSAX = 1; |
| 7275 | ctxt->instate = XML_PARSER_EOF; |
| 7276 | } else { |
| 7277 | ctxt->instate = XML_PARSER_CONTENT; |
| 7278 | xmlParseElement(ctxt); |
| 7279 | ctxt->instate = XML_PARSER_EPILOG; |
| 7280 | |
| 7281 | |
| 7282 | /* |
| 7283 | * The Misc part at the end |
| 7284 | */ |
| 7285 | xmlParseMisc(ctxt); |
| 7286 | |
| 7287 | if (RAW != 0) { |
| 7288 | ctxt->errNo = XML_ERR_DOCUMENT_END; |
| 7289 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7290 | ctxt->sax->error(ctxt->userData, |
| 7291 | "Extra content at the end of the document\n"); |
| 7292 | ctxt->wellFormed = 0; |
| 7293 | ctxt->disableSAX = 1; |
| 7294 | } |
| 7295 | ctxt->instate = XML_PARSER_EOF; |
| 7296 | } |
| 7297 | |
| 7298 | /* |
| 7299 | * SAX: end of the document processing. |
| 7300 | */ |
| 7301 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL) && |
| 7302 | (!ctxt->disableSAX)) |
| 7303 | ctxt->sax->endDocument(ctxt->userData); |
| 7304 | |
| 7305 | if (! ctxt->wellFormed) return(-1); |
| 7306 | return(0); |
| 7307 | } |
| 7308 | |
| 7309 | /** |
| 7310 | * xmlParseExtParsedEnt: |
| 7311 | * @ctxt: an XML parser context |
| 7312 | * |
| 7313 | * parse a genreral parsed entity |
| 7314 | * An external general parsed entity is well-formed if it matches the |
| 7315 | * production labeled extParsedEnt. |
| 7316 | * |
| 7317 | * [78] extParsedEnt ::= TextDecl? content |
| 7318 | * |
| 7319 | * Returns 0, -1 in case of error. the parser context is augmented |
| 7320 | * as a result of the parsing. |
| 7321 | */ |
| 7322 | |
| 7323 | int |
| 7324 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
| 7325 | xmlChar start[4]; |
| 7326 | xmlCharEncoding enc; |
| 7327 | |
| 7328 | xmlDefaultSAXHandlerInit(); |
| 7329 | |
| 7330 | GROW; |
| 7331 | |
| 7332 | /* |
| 7333 | * SAX: beginning of the document processing. |
| 7334 | */ |
| 7335 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
| 7336 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
| 7337 | |
| 7338 | /* |
| 7339 | * Get the 4 first bytes and decode the charset |
| 7340 | * if enc != XML_CHAR_ENCODING_NONE |
| 7341 | * plug some encoding conversion routines. |
| 7342 | */ |
| 7343 | start[0] = RAW; |
| 7344 | start[1] = NXT(1); |
| 7345 | start[2] = NXT(2); |
| 7346 | start[3] = NXT(3); |
| 7347 | enc = xmlDetectCharEncoding(start, 4); |
| 7348 | if (enc != XML_CHAR_ENCODING_NONE) { |
| 7349 | xmlSwitchEncoding(ctxt, enc); |
| 7350 | } |
| 7351 | |
| 7352 | |
| 7353 | if (CUR == 0) { |
| 7354 | ctxt->errNo = XML_ERR_DOCUMENT_EMPTY; |
| 7355 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7356 | ctxt->sax->error(ctxt->userData, "Document is empty\n"); |
| 7357 | ctxt->wellFormed = 0; |
| 7358 | ctxt->disableSAX = 1; |
| 7359 | } |
| 7360 | |
| 7361 | /* |
| 7362 | * Check for the XMLDecl in the Prolog. |
| 7363 | */ |
| 7364 | GROW; |
| 7365 | if ((RAW == '<') && (NXT(1) == '?') && |
| 7366 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 7367 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 7368 | |
| 7369 | /* |
| 7370 | * Note that we will switch encoding on the fly. |
| 7371 | */ |
| 7372 | xmlParseXMLDecl(ctxt); |
| 7373 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 7374 | /* |
| 7375 | * The XML REC instructs us to stop parsing right here |
| 7376 | */ |
| 7377 | return(-1); |
| 7378 | } |
| 7379 | SKIP_BLANKS; |
| 7380 | } else { |
| 7381 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
| 7382 | } |
| 7383 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
| 7384 | ctxt->sax->startDocument(ctxt->userData); |
| 7385 | |
| 7386 | /* |
| 7387 | * Doing validity checking on chunk doesn't make sense |
| 7388 | */ |
| 7389 | ctxt->instate = XML_PARSER_CONTENT; |
| 7390 | ctxt->validate = 0; |
| 7391 | ctxt->loadsubset = 0; |
| 7392 | ctxt->depth = 0; |
| 7393 | |
| 7394 | xmlParseContent(ctxt); |
| 7395 | |
| 7396 | if ((RAW == '<') && (NXT(1) == '/')) { |
| 7397 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 7398 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7399 | ctxt->sax->error(ctxt->userData, |
| 7400 | "chunk is not well balanced\n"); |
| 7401 | ctxt->wellFormed = 0; |
| 7402 | ctxt->disableSAX = 1; |
| 7403 | } else if (RAW != 0) { |
| 7404 | ctxt->errNo = XML_ERR_EXTRA_CONTENT; |
| 7405 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7406 | ctxt->sax->error(ctxt->userData, |
| 7407 | "extra content at the end of well balanced chunk\n"); |
| 7408 | ctxt->wellFormed = 0; |
| 7409 | ctxt->disableSAX = 1; |
| 7410 | } |
| 7411 | |
| 7412 | /* |
| 7413 | * SAX: end of the document processing. |
| 7414 | */ |
| 7415 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL) && |
| 7416 | (!ctxt->disableSAX)) |
| 7417 | ctxt->sax->endDocument(ctxt->userData); |
| 7418 | |
| 7419 | if (! ctxt->wellFormed) return(-1); |
| 7420 | return(0); |
| 7421 | } |
| 7422 | |
| 7423 | /************************************************************************ |
| 7424 | * * |
| 7425 | * Progressive parsing interfaces * |
| 7426 | * * |
| 7427 | ************************************************************************/ |
| 7428 | |
| 7429 | /** |
| 7430 | * xmlParseLookupSequence: |
| 7431 | * @ctxt: an XML parser context |
| 7432 | * @first: the first char to lookup |
| 7433 | * @next: the next char to lookup or zero |
| 7434 | * @third: the next char to lookup or zero |
| 7435 | * |
| 7436 | * Try to find if a sequence (first, next, third) or just (first next) or |
| 7437 | * (first) is available in the input stream. |
| 7438 | * This function has a side effect of (possibly) incrementing ctxt->checkIndex |
| 7439 | * to avoid rescanning sequences of bytes, it DOES change the state of the |
| 7440 | * parser, do not use liberally. |
| 7441 | * |
| 7442 | * Returns the index to the current parsing point if the full sequence |
| 7443 | * is available, -1 otherwise. |
| 7444 | */ |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 7445 | static int |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7446 | xmlParseLookupSequence(xmlParserCtxtPtr ctxt, xmlChar first, |
| 7447 | xmlChar next, xmlChar third) { |
| 7448 | int base, len; |
| 7449 | xmlParserInputPtr in; |
| 7450 | const xmlChar *buf; |
| 7451 | |
| 7452 | in = ctxt->input; |
| 7453 | if (in == NULL) return(-1); |
| 7454 | base = in->cur - in->base; |
| 7455 | if (base < 0) return(-1); |
| 7456 | if (ctxt->checkIndex > base) |
| 7457 | base = ctxt->checkIndex; |
| 7458 | if (in->buf == NULL) { |
| 7459 | buf = in->base; |
| 7460 | len = in->length; |
| 7461 | } else { |
| 7462 | buf = in->buf->buffer->content; |
| 7463 | len = in->buf->buffer->use; |
| 7464 | } |
| 7465 | /* take into account the sequence length */ |
| 7466 | if (third) len -= 2; |
| 7467 | else if (next) len --; |
| 7468 | for (;base < len;base++) { |
| 7469 | if (buf[base] == first) { |
| 7470 | if (third != 0) { |
| 7471 | if ((buf[base + 1] != next) || |
| 7472 | (buf[base + 2] != third)) continue; |
| 7473 | } else if (next != 0) { |
| 7474 | if (buf[base + 1] != next) continue; |
| 7475 | } |
| 7476 | ctxt->checkIndex = 0; |
| 7477 | #ifdef DEBUG_PUSH |
| 7478 | if (next == 0) |
| 7479 | xmlGenericError(xmlGenericErrorContext, |
| 7480 | "PP: lookup '%c' found at %d\n", |
| 7481 | first, base); |
| 7482 | else if (third == 0) |
| 7483 | xmlGenericError(xmlGenericErrorContext, |
| 7484 | "PP: lookup '%c%c' found at %d\n", |
| 7485 | first, next, base); |
| 7486 | else |
| 7487 | xmlGenericError(xmlGenericErrorContext, |
| 7488 | "PP: lookup '%c%c%c' found at %d\n", |
| 7489 | first, next, third, base); |
| 7490 | #endif |
| 7491 | return(base - (in->cur - in->base)); |
| 7492 | } |
| 7493 | } |
| 7494 | ctxt->checkIndex = base; |
| 7495 | #ifdef DEBUG_PUSH |
| 7496 | if (next == 0) |
| 7497 | xmlGenericError(xmlGenericErrorContext, |
| 7498 | "PP: lookup '%c' failed\n", first); |
| 7499 | else if (third == 0) |
| 7500 | xmlGenericError(xmlGenericErrorContext, |
| 7501 | "PP: lookup '%c%c' failed\n", first, next); |
| 7502 | else |
| 7503 | xmlGenericError(xmlGenericErrorContext, |
| 7504 | "PP: lookup '%c%c%c' failed\n", first, next, third); |
| 7505 | #endif |
| 7506 | return(-1); |
| 7507 | } |
| 7508 | |
| 7509 | /** |
| 7510 | * xmlParseTryOrFinish: |
| 7511 | * @ctxt: an XML parser context |
| 7512 | * @terminate: last chunk indicator |
| 7513 | * |
| 7514 | * Try to progress on parsing |
| 7515 | * |
| 7516 | * Returns zero if no parsing was possible |
| 7517 | */ |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 7518 | static int |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7519 | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
| 7520 | int ret = 0; |
| 7521 | int avail; |
| 7522 | xmlChar cur, next; |
| 7523 | |
| 7524 | #ifdef DEBUG_PUSH |
| 7525 | switch (ctxt->instate) { |
| 7526 | case XML_PARSER_EOF: |
| 7527 | xmlGenericError(xmlGenericErrorContext, |
| 7528 | "PP: try EOF\n"); break; |
| 7529 | case XML_PARSER_START: |
| 7530 | xmlGenericError(xmlGenericErrorContext, |
| 7531 | "PP: try START\n"); break; |
| 7532 | case XML_PARSER_MISC: |
| 7533 | xmlGenericError(xmlGenericErrorContext, |
| 7534 | "PP: try MISC\n");break; |
| 7535 | case XML_PARSER_COMMENT: |
| 7536 | xmlGenericError(xmlGenericErrorContext, |
| 7537 | "PP: try COMMENT\n");break; |
| 7538 | case XML_PARSER_PROLOG: |
| 7539 | xmlGenericError(xmlGenericErrorContext, |
| 7540 | "PP: try PROLOG\n");break; |
| 7541 | case XML_PARSER_START_TAG: |
| 7542 | xmlGenericError(xmlGenericErrorContext, |
| 7543 | "PP: try START_TAG\n");break; |
| 7544 | case XML_PARSER_CONTENT: |
| 7545 | xmlGenericError(xmlGenericErrorContext, |
| 7546 | "PP: try CONTENT\n");break; |
| 7547 | case XML_PARSER_CDATA_SECTION: |
| 7548 | xmlGenericError(xmlGenericErrorContext, |
| 7549 | "PP: try CDATA_SECTION\n");break; |
| 7550 | case XML_PARSER_END_TAG: |
| 7551 | xmlGenericError(xmlGenericErrorContext, |
| 7552 | "PP: try END_TAG\n");break; |
| 7553 | case XML_PARSER_ENTITY_DECL: |
| 7554 | xmlGenericError(xmlGenericErrorContext, |
| 7555 | "PP: try ENTITY_DECL\n");break; |
| 7556 | case XML_PARSER_ENTITY_VALUE: |
| 7557 | xmlGenericError(xmlGenericErrorContext, |
| 7558 | "PP: try ENTITY_VALUE\n");break; |
| 7559 | case XML_PARSER_ATTRIBUTE_VALUE: |
| 7560 | xmlGenericError(xmlGenericErrorContext, |
| 7561 | "PP: try ATTRIBUTE_VALUE\n");break; |
| 7562 | case XML_PARSER_DTD: |
| 7563 | xmlGenericError(xmlGenericErrorContext, |
| 7564 | "PP: try DTD\n");break; |
| 7565 | case XML_PARSER_EPILOG: |
| 7566 | xmlGenericError(xmlGenericErrorContext, |
| 7567 | "PP: try EPILOG\n");break; |
| 7568 | case XML_PARSER_PI: |
| 7569 | xmlGenericError(xmlGenericErrorContext, |
| 7570 | "PP: try PI\n");break; |
| 7571 | case XML_PARSER_IGNORE: |
| 7572 | xmlGenericError(xmlGenericErrorContext, |
| 7573 | "PP: try IGNORE\n");break; |
| 7574 | } |
| 7575 | #endif |
| 7576 | |
| 7577 | while (1) { |
| 7578 | /* |
| 7579 | * Pop-up of finished entities. |
| 7580 | */ |
| 7581 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 7582 | xmlPopInput(ctxt); |
| 7583 | |
| 7584 | if (ctxt->input ==NULL) break; |
| 7585 | if (ctxt->input->buf == NULL) |
| 7586 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
| 7587 | else |
| 7588 | avail = ctxt->input->buf->buffer->use - (ctxt->input->cur - ctxt->input->base); |
| 7589 | if (avail < 1) |
| 7590 | goto done; |
| 7591 | switch (ctxt->instate) { |
| 7592 | case XML_PARSER_EOF: |
| 7593 | /* |
| 7594 | * Document parsing is done ! |
| 7595 | */ |
| 7596 | goto done; |
| 7597 | case XML_PARSER_START: |
| 7598 | /* |
| 7599 | * Very first chars read from the document flow. |
| 7600 | */ |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7601 | if (avail < 2) |
| 7602 | goto done; |
| 7603 | |
| 7604 | cur = ctxt->input->cur[0]; |
| 7605 | next = ctxt->input->cur[1]; |
| 7606 | if (cur == 0) { |
| 7607 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
| 7608 | ctxt->sax->setDocumentLocator(ctxt->userData, |
| 7609 | &xmlDefaultSAXLocator); |
| 7610 | ctxt->errNo = XML_ERR_DOCUMENT_EMPTY; |
| 7611 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7612 | ctxt->sax->error(ctxt->userData, "Document is empty\n"); |
| 7613 | ctxt->wellFormed = 0; |
| 7614 | ctxt->disableSAX = 1; |
| 7615 | ctxt->instate = XML_PARSER_EOF; |
| 7616 | #ifdef DEBUG_PUSH |
| 7617 | xmlGenericError(xmlGenericErrorContext, |
| 7618 | "PP: entering EOF\n"); |
| 7619 | #endif |
| 7620 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
| 7621 | ctxt->sax->endDocument(ctxt->userData); |
| 7622 | goto done; |
| 7623 | } |
| 7624 | if ((cur == '<') && (next == '?')) { |
| 7625 | /* PI or XML decl */ |
| 7626 | if (avail < 5) return(ret); |
| 7627 | if ((!terminate) && |
| 7628 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
| 7629 | return(ret); |
| 7630 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
| 7631 | ctxt->sax->setDocumentLocator(ctxt->userData, |
| 7632 | &xmlDefaultSAXLocator); |
| 7633 | if ((ctxt->input->cur[2] == 'x') && |
| 7634 | (ctxt->input->cur[3] == 'm') && |
| 7635 | (ctxt->input->cur[4] == 'l') && |
| 7636 | (IS_BLANK(ctxt->input->cur[5]))) { |
| 7637 | ret += 5; |
| 7638 | #ifdef DEBUG_PUSH |
| 7639 | xmlGenericError(xmlGenericErrorContext, |
| 7640 | "PP: Parsing XML Decl\n"); |
| 7641 | #endif |
| 7642 | xmlParseXMLDecl(ctxt); |
| 7643 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
| 7644 | /* |
| 7645 | * The XML REC instructs us to stop parsing right |
| 7646 | * here |
| 7647 | */ |
| 7648 | ctxt->instate = XML_PARSER_EOF; |
| 7649 | return(0); |
| 7650 | } |
| 7651 | ctxt->standalone = ctxt->input->standalone; |
| 7652 | if ((ctxt->encoding == NULL) && |
| 7653 | (ctxt->input->encoding != NULL)) |
| 7654 | ctxt->encoding = xmlStrdup(ctxt->input->encoding); |
| 7655 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
| 7656 | (!ctxt->disableSAX)) |
| 7657 | ctxt->sax->startDocument(ctxt->userData); |
| 7658 | ctxt->instate = XML_PARSER_MISC; |
| 7659 | #ifdef DEBUG_PUSH |
| 7660 | xmlGenericError(xmlGenericErrorContext, |
| 7661 | "PP: entering MISC\n"); |
| 7662 | #endif |
| 7663 | } else { |
| 7664 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
| 7665 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
| 7666 | (!ctxt->disableSAX)) |
| 7667 | ctxt->sax->startDocument(ctxt->userData); |
| 7668 | ctxt->instate = XML_PARSER_MISC; |
| 7669 | #ifdef DEBUG_PUSH |
| 7670 | xmlGenericError(xmlGenericErrorContext, |
| 7671 | "PP: entering MISC\n"); |
| 7672 | #endif |
| 7673 | } |
| 7674 | } else { |
| 7675 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
| 7676 | ctxt->sax->setDocumentLocator(ctxt->userData, |
| 7677 | &xmlDefaultSAXLocator); |
| 7678 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
| 7679 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
| 7680 | (!ctxt->disableSAX)) |
| 7681 | ctxt->sax->startDocument(ctxt->userData); |
| 7682 | ctxt->instate = XML_PARSER_MISC; |
| 7683 | #ifdef DEBUG_PUSH |
| 7684 | xmlGenericError(xmlGenericErrorContext, |
| 7685 | "PP: entering MISC\n"); |
| 7686 | #endif |
| 7687 | } |
| 7688 | break; |
| 7689 | case XML_PARSER_MISC: |
| 7690 | SKIP_BLANKS; |
| 7691 | if (ctxt->input->buf == NULL) |
| 7692 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
| 7693 | else |
| 7694 | avail = ctxt->input->buf->buffer->use - (ctxt->input->cur - ctxt->input->base); |
| 7695 | if (avail < 2) |
| 7696 | goto done; |
| 7697 | cur = ctxt->input->cur[0]; |
| 7698 | next = ctxt->input->cur[1]; |
| 7699 | if ((cur == '<') && (next == '?')) { |
| 7700 | if ((!terminate) && |
| 7701 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
| 7702 | goto done; |
| 7703 | #ifdef DEBUG_PUSH |
| 7704 | xmlGenericError(xmlGenericErrorContext, |
| 7705 | "PP: Parsing PI\n"); |
| 7706 | #endif |
| 7707 | xmlParsePI(ctxt); |
| 7708 | } else if ((cur == '<') && (next == '!') && |
| 7709 | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
| 7710 | if ((!terminate) && |
| 7711 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) |
| 7712 | goto done; |
| 7713 | #ifdef DEBUG_PUSH |
| 7714 | xmlGenericError(xmlGenericErrorContext, |
| 7715 | "PP: Parsing Comment\n"); |
| 7716 | #endif |
| 7717 | xmlParseComment(ctxt); |
| 7718 | ctxt->instate = XML_PARSER_MISC; |
| 7719 | } else if ((cur == '<') && (next == '!') && |
| 7720 | (ctxt->input->cur[2] == 'D') && (ctxt->input->cur[3] == 'O') && |
| 7721 | (ctxt->input->cur[4] == 'C') && (ctxt->input->cur[5] == 'T') && |
| 7722 | (ctxt->input->cur[6] == 'Y') && (ctxt->input->cur[7] == 'P') && |
| 7723 | (ctxt->input->cur[8] == 'E')) { |
| 7724 | if ((!terminate) && |
| 7725 | (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0)) |
| 7726 | goto done; |
| 7727 | #ifdef DEBUG_PUSH |
| 7728 | xmlGenericError(xmlGenericErrorContext, |
| 7729 | "PP: Parsing internal subset\n"); |
| 7730 | #endif |
| 7731 | ctxt->inSubset = 1; |
| 7732 | xmlParseDocTypeDecl(ctxt); |
| 7733 | if (RAW == '[') { |
| 7734 | ctxt->instate = XML_PARSER_DTD; |
| 7735 | #ifdef DEBUG_PUSH |
| 7736 | xmlGenericError(xmlGenericErrorContext, |
| 7737 | "PP: entering DTD\n"); |
| 7738 | #endif |
| 7739 | } else { |
| 7740 | /* |
| 7741 | * Create and update the external subset. |
| 7742 | */ |
| 7743 | ctxt->inSubset = 2; |
| 7744 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 7745 | (ctxt->sax->externalSubset != NULL)) |
| 7746 | ctxt->sax->externalSubset(ctxt->userData, |
| 7747 | ctxt->intSubName, ctxt->extSubSystem, |
| 7748 | ctxt->extSubURI); |
| 7749 | ctxt->inSubset = 0; |
| 7750 | ctxt->instate = XML_PARSER_PROLOG; |
| 7751 | #ifdef DEBUG_PUSH |
| 7752 | xmlGenericError(xmlGenericErrorContext, |
| 7753 | "PP: entering PROLOG\n"); |
| 7754 | #endif |
| 7755 | } |
| 7756 | } else if ((cur == '<') && (next == '!') && |
| 7757 | (avail < 9)) { |
| 7758 | goto done; |
| 7759 | } else { |
| 7760 | ctxt->instate = XML_PARSER_START_TAG; |
| 7761 | #ifdef DEBUG_PUSH |
| 7762 | xmlGenericError(xmlGenericErrorContext, |
| 7763 | "PP: entering START_TAG\n"); |
| 7764 | #endif |
| 7765 | } |
| 7766 | break; |
| 7767 | case XML_PARSER_IGNORE: |
| 7768 | xmlGenericError(xmlGenericErrorContext, |
| 7769 | "PP: internal error, state == IGNORE"); |
| 7770 | ctxt->instate = XML_PARSER_DTD; |
| 7771 | #ifdef DEBUG_PUSH |
| 7772 | xmlGenericError(xmlGenericErrorContext, |
| 7773 | "PP: entering DTD\n"); |
| 7774 | #endif |
| 7775 | break; |
| 7776 | case XML_PARSER_PROLOG: |
| 7777 | SKIP_BLANKS; |
| 7778 | if (ctxt->input->buf == NULL) |
| 7779 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
| 7780 | else |
| 7781 | avail = ctxt->input->buf->buffer->use - (ctxt->input->cur - ctxt->input->base); |
| 7782 | if (avail < 2) |
| 7783 | goto done; |
| 7784 | cur = ctxt->input->cur[0]; |
| 7785 | next = ctxt->input->cur[1]; |
| 7786 | if ((cur == '<') && (next == '?')) { |
| 7787 | if ((!terminate) && |
| 7788 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
| 7789 | goto done; |
| 7790 | #ifdef DEBUG_PUSH |
| 7791 | xmlGenericError(xmlGenericErrorContext, |
| 7792 | "PP: Parsing PI\n"); |
| 7793 | #endif |
| 7794 | xmlParsePI(ctxt); |
| 7795 | } else if ((cur == '<') && (next == '!') && |
| 7796 | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
| 7797 | if ((!terminate) && |
| 7798 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) |
| 7799 | goto done; |
| 7800 | #ifdef DEBUG_PUSH |
| 7801 | xmlGenericError(xmlGenericErrorContext, |
| 7802 | "PP: Parsing Comment\n"); |
| 7803 | #endif |
| 7804 | xmlParseComment(ctxt); |
| 7805 | ctxt->instate = XML_PARSER_PROLOG; |
| 7806 | } else if ((cur == '<') && (next == '!') && |
| 7807 | (avail < 4)) { |
| 7808 | goto done; |
| 7809 | } else { |
| 7810 | ctxt->instate = XML_PARSER_START_TAG; |
| 7811 | #ifdef DEBUG_PUSH |
| 7812 | xmlGenericError(xmlGenericErrorContext, |
| 7813 | "PP: entering START_TAG\n"); |
| 7814 | #endif |
| 7815 | } |
| 7816 | break; |
| 7817 | case XML_PARSER_EPILOG: |
| 7818 | SKIP_BLANKS; |
| 7819 | if (ctxt->input->buf == NULL) |
| 7820 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
| 7821 | else |
| 7822 | avail = ctxt->input->buf->buffer->use - (ctxt->input->cur - ctxt->input->base); |
| 7823 | if (avail < 2) |
| 7824 | goto done; |
| 7825 | cur = ctxt->input->cur[0]; |
| 7826 | next = ctxt->input->cur[1]; |
| 7827 | if ((cur == '<') && (next == '?')) { |
| 7828 | if ((!terminate) && |
| 7829 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
| 7830 | goto done; |
| 7831 | #ifdef DEBUG_PUSH |
| 7832 | xmlGenericError(xmlGenericErrorContext, |
| 7833 | "PP: Parsing PI\n"); |
| 7834 | #endif |
| 7835 | xmlParsePI(ctxt); |
| 7836 | ctxt->instate = XML_PARSER_EPILOG; |
| 7837 | } else if ((cur == '<') && (next == '!') && |
| 7838 | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
| 7839 | if ((!terminate) && |
| 7840 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) |
| 7841 | goto done; |
| 7842 | #ifdef DEBUG_PUSH |
| 7843 | xmlGenericError(xmlGenericErrorContext, |
| 7844 | "PP: Parsing Comment\n"); |
| 7845 | #endif |
| 7846 | xmlParseComment(ctxt); |
| 7847 | ctxt->instate = XML_PARSER_EPILOG; |
| 7848 | } else if ((cur == '<') && (next == '!') && |
| 7849 | (avail < 4)) { |
| 7850 | goto done; |
| 7851 | } else { |
| 7852 | ctxt->errNo = XML_ERR_DOCUMENT_END; |
| 7853 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7854 | ctxt->sax->error(ctxt->userData, |
| 7855 | "Extra content at the end of the document\n"); |
| 7856 | ctxt->wellFormed = 0; |
| 7857 | ctxt->disableSAX = 1; |
| 7858 | ctxt->instate = XML_PARSER_EOF; |
| 7859 | #ifdef DEBUG_PUSH |
| 7860 | xmlGenericError(xmlGenericErrorContext, |
| 7861 | "PP: entering EOF\n"); |
| 7862 | #endif |
| 7863 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL) && |
| 7864 | (!ctxt->disableSAX)) |
| 7865 | ctxt->sax->endDocument(ctxt->userData); |
| 7866 | goto done; |
| 7867 | } |
| 7868 | break; |
| 7869 | case XML_PARSER_START_TAG: { |
| 7870 | xmlChar *name, *oldname; |
| 7871 | |
| 7872 | if ((avail < 2) && (ctxt->inputNr == 1)) |
| 7873 | goto done; |
| 7874 | cur = ctxt->input->cur[0]; |
| 7875 | if (cur != '<') { |
| 7876 | ctxt->errNo = XML_ERR_DOCUMENT_EMPTY; |
| 7877 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7878 | ctxt->sax->error(ctxt->userData, |
| 7879 | "Start tag expect, '<' not found\n"); |
| 7880 | ctxt->wellFormed = 0; |
| 7881 | ctxt->disableSAX = 1; |
| 7882 | ctxt->instate = XML_PARSER_EOF; |
| 7883 | #ifdef DEBUG_PUSH |
| 7884 | xmlGenericError(xmlGenericErrorContext, |
| 7885 | "PP: entering EOF\n"); |
| 7886 | #endif |
| 7887 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL) && |
| 7888 | (!ctxt->disableSAX)) |
| 7889 | ctxt->sax->endDocument(ctxt->userData); |
| 7890 | goto done; |
| 7891 | } |
| 7892 | if ((!terminate) && |
| 7893 | (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0)) |
| 7894 | goto done; |
| 7895 | if (ctxt->spaceNr == 0) |
| 7896 | spacePush(ctxt, -1); |
| 7897 | else |
| 7898 | spacePush(ctxt, *ctxt->space); |
| 7899 | name = xmlParseStartTag(ctxt); |
| 7900 | if (name == NULL) { |
| 7901 | spacePop(ctxt); |
| 7902 | ctxt->instate = XML_PARSER_EOF; |
| 7903 | #ifdef DEBUG_PUSH |
| 7904 | xmlGenericError(xmlGenericErrorContext, |
| 7905 | "PP: entering EOF\n"); |
| 7906 | #endif |
| 7907 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL) && |
| 7908 | (!ctxt->disableSAX)) |
| 7909 | ctxt->sax->endDocument(ctxt->userData); |
| 7910 | goto done; |
| 7911 | } |
| 7912 | namePush(ctxt, xmlStrdup(name)); |
| 7913 | |
| 7914 | /* |
| 7915 | * [ VC: Root Element Type ] |
| 7916 | * The Name in the document type declaration must match |
| 7917 | * the element type of the root element. |
| 7918 | */ |
| 7919 | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
| 7920 | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
| 7921 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
| 7922 | |
| 7923 | /* |
| 7924 | * Check for an Empty Element. |
| 7925 | */ |
| 7926 | if ((RAW == '/') && (NXT(1) == '>')) { |
| 7927 | SKIP(2); |
| 7928 | if ((ctxt->sax != NULL) && |
| 7929 | (ctxt->sax->endElement != NULL) && (!ctxt->disableSAX)) |
| 7930 | ctxt->sax->endElement(ctxt->userData, name); |
| 7931 | xmlFree(name); |
| 7932 | oldname = namePop(ctxt); |
| 7933 | spacePop(ctxt); |
| 7934 | if (oldname != NULL) { |
| 7935 | #ifdef DEBUG_STACK |
| 7936 | xmlGenericError(xmlGenericErrorContext,"Close: popped %s\n", oldname); |
| 7937 | #endif |
| 7938 | xmlFree(oldname); |
| 7939 | } |
| 7940 | if (ctxt->name == NULL) { |
| 7941 | ctxt->instate = XML_PARSER_EPILOG; |
| 7942 | #ifdef DEBUG_PUSH |
| 7943 | xmlGenericError(xmlGenericErrorContext, |
| 7944 | "PP: entering EPILOG\n"); |
| 7945 | #endif |
| 7946 | } else { |
| 7947 | ctxt->instate = XML_PARSER_CONTENT; |
| 7948 | #ifdef DEBUG_PUSH |
| 7949 | xmlGenericError(xmlGenericErrorContext, |
| 7950 | "PP: entering CONTENT\n"); |
| 7951 | #endif |
| 7952 | } |
| 7953 | break; |
| 7954 | } |
| 7955 | if (RAW == '>') { |
| 7956 | NEXT; |
| 7957 | } else { |
| 7958 | ctxt->errNo = XML_ERR_GT_REQUIRED; |
| 7959 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 7960 | ctxt->sax->error(ctxt->userData, |
| 7961 | "Couldn't find end of Start Tag %s\n", |
| 7962 | name); |
| 7963 | ctxt->wellFormed = 0; |
| 7964 | ctxt->disableSAX = 1; |
| 7965 | |
| 7966 | /* |
| 7967 | * end of parsing of this node. |
| 7968 | */ |
| 7969 | nodePop(ctxt); |
| 7970 | oldname = namePop(ctxt); |
| 7971 | spacePop(ctxt); |
| 7972 | if (oldname != NULL) { |
| 7973 | #ifdef DEBUG_STACK |
| 7974 | xmlGenericError(xmlGenericErrorContext,"Close: popped %s\n", oldname); |
| 7975 | #endif |
| 7976 | xmlFree(oldname); |
| 7977 | } |
| 7978 | } |
| 7979 | xmlFree(name); |
| 7980 | ctxt->instate = XML_PARSER_CONTENT; |
| 7981 | #ifdef DEBUG_PUSH |
| 7982 | xmlGenericError(xmlGenericErrorContext, |
| 7983 | "PP: entering CONTENT\n"); |
| 7984 | #endif |
| 7985 | break; |
| 7986 | } |
| 7987 | case XML_PARSER_CONTENT: { |
| 7988 | const xmlChar *test; |
| 7989 | int cons; |
Daniel Veillard | 04be4f5 | 2001-03-26 21:23:53 +0000 | [diff] [blame] | 7990 | int tok; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7991 | |
| 7992 | /* |
| 7993 | * Handle preparsed entities and charRef |
| 7994 | */ |
| 7995 | if (ctxt->token != 0) { |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 7996 | xmlChar current[2] = { 0 , 0 } ; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7997 | |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 7998 | current[0] = (xmlChar) ctxt->token; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 7999 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 8000 | (ctxt->sax->characters != NULL)) |
Daniel Veillard | 56a4cb8 | 2001-03-24 17:00:36 +0000 | [diff] [blame] | 8001 | ctxt->sax->characters(ctxt->userData, current, 1); |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 8002 | ctxt->token = 0; |
| 8003 | } |
| 8004 | if ((avail < 2) && (ctxt->inputNr == 1)) |
| 8005 | goto done; |
| 8006 | cur = ctxt->input->cur[0]; |
| 8007 | next = ctxt->input->cur[1]; |
| 8008 | |
| 8009 | test = CUR_PTR; |
| 8010 | cons = ctxt->input->consumed; |
| 8011 | tok = ctxt->token; |
| 8012 | if ((cur == '<') && (next == '?')) { |
| 8013 | if ((!terminate) && |
| 8014 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
| 8015 | goto done; |
| 8016 | #ifdef DEBUG_PUSH |
| 8017 | xmlGenericError(xmlGenericErrorContext, |
| 8018 | "PP: Parsing PI\n"); |
| 8019 | #endif |
| 8020 | xmlParsePI(ctxt); |
| 8021 | } else if ((cur == '<') && (next == '!') && |
| 8022 | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
| 8023 | if ((!terminate) && |
| 8024 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) |
| 8025 | goto done; |
| 8026 | #ifdef DEBUG_PUSH |
| 8027 | xmlGenericError(xmlGenericErrorContext, |
| 8028 | "PP: Parsing Comment\n"); |
| 8029 | #endif |
| 8030 | xmlParseComment(ctxt); |
| 8031 | ctxt->instate = XML_PARSER_CONTENT; |
| 8032 | } else if ((cur == '<') && (ctxt->input->cur[1] == '!') && |
| 8033 | (ctxt->input->cur[2] == '[') && (NXT(3) == 'C') && |
| 8034 | (ctxt->input->cur[4] == 'D') && (NXT(5) == 'A') && |
| 8035 | (ctxt->input->cur[6] == 'T') && (NXT(7) == 'A') && |
| 8036 | (ctxt->input->cur[8] == '[')) { |
| 8037 | SKIP(9); |
| 8038 | ctxt->instate = XML_PARSER_CDATA_SECTION; |
| 8039 | #ifdef DEBUG_PUSH |
| 8040 | xmlGenericError(xmlGenericErrorContext, |
| 8041 | "PP: entering CDATA_SECTION\n"); |
| 8042 | #endif |
| 8043 | break; |
| 8044 | } else if ((cur == '<') && (next == '!') && |
| 8045 | (avail < 9)) { |
| 8046 | goto done; |
| 8047 | } else if ((cur == '<') && (next == '/')) { |
| 8048 | ctxt->instate = XML_PARSER_END_TAG; |
| 8049 | #ifdef DEBUG_PUSH |
| 8050 | xmlGenericError(xmlGenericErrorContext, |
| 8051 | "PP: entering END_TAG\n"); |
| 8052 | #endif |
| 8053 | break; |
| 8054 | } else if (cur == '<') { |
| 8055 | ctxt->instate = XML_PARSER_START_TAG; |
| 8056 | #ifdef DEBUG_PUSH |
| 8057 | xmlGenericError(xmlGenericErrorContext, |
| 8058 | "PP: entering START_TAG\n"); |
| 8059 | #endif |
| 8060 | break; |
| 8061 | } else if (cur == '&') { |
| 8062 | if ((!terminate) && |
| 8063 | (xmlParseLookupSequence(ctxt, ';', 0, 0) < 0)) |
| 8064 | goto done; |
| 8065 | #ifdef DEBUG_PUSH |
| 8066 | xmlGenericError(xmlGenericErrorContext, |
| 8067 | "PP: Parsing Reference\n"); |
| 8068 | #endif |
| 8069 | xmlParseReference(ctxt); |
| 8070 | } else { |
| 8071 | /* TODO Avoid the extra copy, handle directly !!! */ |
| 8072 | /* |
| 8073 | * Goal of the following test is: |
| 8074 | * - minimize calls to the SAX 'character' callback |
| 8075 | * when they are mergeable |
| 8076 | * - handle an problem for isBlank when we only parse |
| 8077 | * a sequence of blank chars and the next one is |
| 8078 | * not available to check against '<' presence. |
| 8079 | * - tries to homogenize the differences in SAX |
| 8080 | * callbacks beween the push and pull versions |
| 8081 | * of the parser. |
| 8082 | */ |
| 8083 | if ((ctxt->inputNr == 1) && |
| 8084 | (avail < XML_PARSER_BIG_BUFFER_SIZE)) { |
| 8085 | if ((!terminate) && |
| 8086 | (xmlParseLookupSequence(ctxt, '<', 0, 0) < 0)) |
| 8087 | goto done; |
| 8088 | } |
| 8089 | ctxt->checkIndex = 0; |
| 8090 | #ifdef DEBUG_PUSH |
| 8091 | xmlGenericError(xmlGenericErrorContext, |
| 8092 | "PP: Parsing char data\n"); |
| 8093 | #endif |
| 8094 | xmlParseCharData(ctxt, 0); |
| 8095 | } |
| 8096 | /* |
| 8097 | * Pop-up of finished entities. |
| 8098 | */ |
| 8099 | while ((RAW == 0) && (ctxt->inputNr > 1)) |
| 8100 | xmlPopInput(ctxt); |
| 8101 | if ((cons == ctxt->input->consumed) && (test == CUR_PTR) && |
| 8102 | (tok == ctxt->token)) { |
| 8103 | ctxt->errNo = XML_ERR_INTERNAL_ERROR; |
| 8104 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 8105 | ctxt->sax->error(ctxt->userData, |
| 8106 | "detected an error in element content\n"); |
| 8107 | ctxt->wellFormed = 0; |
| 8108 | ctxt->disableSAX = 1; |
| 8109 | ctxt->instate = XML_PARSER_EOF; |
| 8110 | break; |
| 8111 | } |
| 8112 | break; |
| 8113 | } |
| 8114 | case XML_PARSER_CDATA_SECTION: { |
| 8115 | /* |
| 8116 | * The Push mode need to have the SAX callback for |
| 8117 | * cdataBlock merge back contiguous callbacks. |
| 8118 | */ |
| 8119 | int base; |
| 8120 | |
| 8121 | base = xmlParseLookupSequence(ctxt, ']', ']', '>'); |
| 8122 | if (base < 0) { |
| 8123 | if (avail >= XML_PARSER_BIG_BUFFER_SIZE + 2) { |
| 8124 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
| 8125 | if (ctxt->sax->cdataBlock != NULL) |
| 8126 | ctxt->sax->cdataBlock(ctxt->userData, ctxt->input->cur, |
| 8127 | XML_PARSER_BIG_BUFFER_SIZE); |
| 8128 | } |
| 8129 | SKIP(XML_PARSER_BIG_BUFFER_SIZE); |
| 8130 | ctxt->checkIndex = 0; |
| 8131 | } |
| 8132 | goto done; |
| 8133 | } else { |
| 8134 | if ((ctxt->sax != NULL) && (base > 0) && |
| 8135 | (!ctxt->disableSAX)) { |
| 8136 | if (ctxt->sax->cdataBlock != NULL) |
| 8137 | ctxt->sax->cdataBlock(ctxt->userData, |
| 8138 | ctxt->input->cur, base); |
| 8139 | } |
| 8140 | SKIP(base + 3); |
| 8141 | ctxt->checkIndex = 0; |
| 8142 | ctxt->instate = XML_PARSER_CONTENT; |
| 8143 | #ifdef DEBUG_PUSH |
| 8144 | xmlGenericError(xmlGenericErrorContext, |
| 8145 | "PP: entering CONTENT\n"); |
| 8146 | #endif |
| 8147 | } |
| 8148 | break; |
| 8149 | } |
| 8150 | case XML_PARSER_END_TAG: |
| 8151 | if (avail < 2) |
| 8152 | goto done; |
| 8153 | if ((!terminate) && |
| 8154 | (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0)) |
| 8155 | goto done; |
| 8156 | xmlParseEndTag(ctxt); |
| 8157 | if (ctxt->name == NULL) { |
| 8158 | ctxt->instate = XML_PARSER_EPILOG; |
| 8159 | #ifdef DEBUG_PUSH |
| 8160 | xmlGenericError(xmlGenericErrorContext, |
| 8161 | "PP: entering EPILOG\n"); |
| 8162 | #endif |
| 8163 | } else { |
| 8164 | ctxt->instate = XML_PARSER_CONTENT; |
| 8165 | #ifdef DEBUG_PUSH |
| 8166 | xmlGenericError(xmlGenericErrorContext, |
| 8167 | "PP: entering CONTENT\n"); |
| 8168 | #endif |
| 8169 | } |
| 8170 | break; |
| 8171 | case XML_PARSER_DTD: { |
| 8172 | /* |
| 8173 | * Sorry but progressive parsing of the internal subset |
| 8174 | * is not expected to be supported. We first check that |
| 8175 | * the full content of the internal subset is available and |
| 8176 | * the parsing is launched only at that point. |
| 8177 | * Internal subset ends up with "']' S? '>'" in an unescaped |
| 8178 | * section and not in a ']]>' sequence which are conditional |
| 8179 | * sections (whoever argued to keep that crap in XML deserve |
| 8180 | * a place in hell !). |
| 8181 | */ |
| 8182 | int base, i; |
| 8183 | xmlChar *buf; |
| 8184 | xmlChar quote = 0; |
| 8185 | |
| 8186 | base = ctxt->input->cur - ctxt->input->base; |
| 8187 | if (base < 0) return(0); |
| 8188 | if (ctxt->checkIndex > base) |
| 8189 | base = ctxt->checkIndex; |
| 8190 | buf = ctxt->input->buf->buffer->content; |
| 8191 | for (;(unsigned int) base < ctxt->input->buf->buffer->use; |
| 8192 | base++) { |
| 8193 | if (quote != 0) { |
| 8194 | if (buf[base] == quote) |
| 8195 | quote = 0; |
| 8196 | continue; |
| 8197 | } |
| 8198 | if (buf[base] == '"') { |
| 8199 | quote = '"'; |
| 8200 | continue; |
| 8201 | } |
| 8202 | if (buf[base] == '\'') { |
| 8203 | quote = '\''; |
| 8204 | continue; |
| 8205 | } |
| 8206 | if (buf[base] == ']') { |
| 8207 | if ((unsigned int) base +1 >= |
| 8208 | ctxt->input->buf->buffer->use) |
| 8209 | break; |
| 8210 | if (buf[base + 1] == ']') { |
| 8211 | /* conditional crap, skip both ']' ! */ |
| 8212 | base++; |
| 8213 | continue; |
| 8214 | } |
| 8215 | for (i = 0; |
| 8216 | (unsigned int) base + i < ctxt->input->buf->buffer->use; |
| 8217 | i++) { |
| 8218 | if (buf[base + i] == '>') |
| 8219 | goto found_end_int_subset; |
| 8220 | } |
| 8221 | break; |
| 8222 | } |
| 8223 | } |
| 8224 | /* |
| 8225 | * We didn't found the end of the Internal subset |
| 8226 | */ |
| 8227 | if (quote == 0) |
| 8228 | ctxt->checkIndex = base; |
| 8229 | #ifdef DEBUG_PUSH |
| 8230 | if (next == 0) |
| 8231 | xmlGenericError(xmlGenericErrorContext, |
| 8232 | "PP: lookup of int subset end filed\n"); |
| 8233 | #endif |
| 8234 | goto done; |
| 8235 | |
| 8236 | found_end_int_subset: |
| 8237 | xmlParseInternalSubset(ctxt); |
| 8238 | ctxt->inSubset = 2; |
| 8239 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
| 8240 | (ctxt->sax->externalSubset != NULL)) |
| 8241 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
| 8242 | ctxt->extSubSystem, ctxt->extSubURI); |
| 8243 | ctxt->inSubset = 0; |
| 8244 | ctxt->instate = XML_PARSER_PROLOG; |
| 8245 | ctxt->checkIndex = 0; |
| 8246 | #ifdef DEBUG_PUSH |
| 8247 | xmlGenericError(xmlGenericErrorContext, |
| 8248 | "PP: entering PROLOG\n"); |
| 8249 | #endif |
| 8250 | break; |
| 8251 | } |
| 8252 | case XML_PARSER_COMMENT: |
| 8253 | xmlGenericError(xmlGenericErrorContext, |
| 8254 | "PP: internal error, state == COMMENT\n"); |
| 8255 | ctxt->instate = XML_PARSER_CONTENT; |
| 8256 | #ifdef DEBUG_PUSH |
| 8257 | xmlGenericError(xmlGenericErrorContext, |
| 8258 | "PP: entering CONTENT\n"); |
| 8259 | #endif |
| 8260 | break; |
| 8261 | case XML_PARSER_PI: |
| 8262 | xmlGenericError(xmlGenericErrorContext, |
| 8263 | "PP: internal error, state == PI\n"); |
| 8264 | ctxt->instate = XML_PARSER_CONTENT; |
| 8265 | #ifdef DEBUG_PUSH |
| 8266 | xmlGenericError(xmlGenericErrorContext, |
| 8267 | "PP: entering CONTENT\n"); |
| 8268 | #endif |
| 8269 | break; |
| 8270 | case XML_PARSER_ENTITY_DECL: |
| 8271 | xmlGenericError(xmlGenericErrorContext, |
| 8272 | "PP: internal error, state == ENTITY_DECL\n"); |
| 8273 | ctxt->instate = XML_PARSER_DTD; |
| 8274 | #ifdef DEBUG_PUSH |
| 8275 | xmlGenericError(xmlGenericErrorContext, |
| 8276 | "PP: entering DTD\n"); |
| 8277 | #endif |
| 8278 | break; |
| 8279 | case XML_PARSER_ENTITY_VALUE: |
| 8280 | xmlGenericError(xmlGenericErrorContext, |
| 8281 | "PP: internal error, state == ENTITY_VALUE\n"); |
| 8282 | ctxt->instate = XML_PARSER_CONTENT; |
| 8283 | #ifdef DEBUG_PUSH |
| 8284 | xmlGenericError(xmlGenericErrorContext, |
| 8285 | "PP: entering DTD\n"); |
| 8286 | #endif |
| 8287 | break; |
| 8288 | case XML_PARSER_ATTRIBUTE_VALUE: |
| 8289 | xmlGenericError(xmlGenericErrorContext, |
| 8290 | "PP: internal error, state == ATTRIBUTE_VALUE\n"); |
| 8291 | ctxt->instate = XML_PARSER_START_TAG; |
| 8292 | #ifdef DEBUG_PUSH |
| 8293 | xmlGenericError(xmlGenericErrorContext, |
| 8294 | "PP: entering START_TAG\n"); |
| 8295 | #endif |
| 8296 | break; |
| 8297 | case XML_PARSER_SYSTEM_LITERAL: |
| 8298 | xmlGenericError(xmlGenericErrorContext, |
| 8299 | "PP: internal error, state == SYSTEM_LITERAL\n"); |
| 8300 | ctxt->instate = XML_PARSER_START_TAG; |
| 8301 | #ifdef DEBUG_PUSH |
| 8302 | xmlGenericError(xmlGenericErrorContext, |
| 8303 | "PP: entering START_TAG\n"); |
| 8304 | #endif |
| 8305 | break; |
| 8306 | } |
| 8307 | } |
| 8308 | done: |
| 8309 | #ifdef DEBUG_PUSH |
| 8310 | xmlGenericError(xmlGenericErrorContext, "PP: done %d\n", ret); |
| 8311 | #endif |
| 8312 | return(ret); |
| 8313 | } |
| 8314 | |
| 8315 | /** |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 8316 | * xmlParseChunk: |
| 8317 | * @ctxt: an XML parser context |
| 8318 | * @chunk: an char array |
| 8319 | * @size: the size in byte of the chunk |
| 8320 | * @terminate: last chunk indicator |
| 8321 | * |
| 8322 | * Parse a Chunk of memory |
| 8323 | * |
| 8324 | * Returns zero if no error, the xmlParserErrors otherwise. |
| 8325 | */ |
| 8326 | int |
| 8327 | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
| 8328 | int terminate) { |
| 8329 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
| 8330 | (ctxt->input->buf != NULL) && (ctxt->instate != XML_PARSER_EOF)) { |
| 8331 | int base = ctxt->input->base - ctxt->input->buf->buffer->content; |
| 8332 | int cur = ctxt->input->cur - ctxt->input->base; |
| 8333 | |
| 8334 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
| 8335 | ctxt->input->base = ctxt->input->buf->buffer->content + base; |
| 8336 | ctxt->input->cur = ctxt->input->base + cur; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 8337 | ctxt->input->end = |
| 8338 | &ctxt->input->buf->buffer->content[ctxt->input->buf->buffer->use]; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 8339 | #ifdef DEBUG_PUSH |
| 8340 | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
| 8341 | #endif |
| 8342 | |
| 8343 | if ((terminate) || (ctxt->input->buf->buffer->use > 80)) |
| 8344 | xmlParseTryOrFinish(ctxt, terminate); |
| 8345 | } else if (ctxt->instate != XML_PARSER_EOF) { |
| 8346 | if ((ctxt->input != NULL) && ctxt->input->buf != NULL) { |
| 8347 | xmlParserInputBufferPtr in = ctxt->input->buf; |
| 8348 | if ((in->encoder != NULL) && (in->buffer != NULL) && |
| 8349 | (in->raw != NULL)) { |
| 8350 | int nbchars; |
| 8351 | |
| 8352 | nbchars = xmlCharEncInFunc(in->encoder, in->buffer, in->raw); |
| 8353 | if (nbchars < 0) { |
| 8354 | xmlGenericError(xmlGenericErrorContext, |
| 8355 | "xmlParseChunk: encoder error\n"); |
| 8356 | return(XML_ERR_INVALID_ENCODING); |
| 8357 | } |
| 8358 | } |
| 8359 | } |
| 8360 | } |
| 8361 | xmlParseTryOrFinish(ctxt, terminate); |
| 8362 | if (terminate) { |
| 8363 | /* |
| 8364 | * Check for termination |
| 8365 | */ |
| 8366 | if ((ctxt->instate != XML_PARSER_EOF) && |
| 8367 | (ctxt->instate != XML_PARSER_EPILOG)) { |
| 8368 | ctxt->errNo = XML_ERR_DOCUMENT_END; |
| 8369 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 8370 | ctxt->sax->error(ctxt->userData, |
| 8371 | "Extra content at the end of the document\n"); |
| 8372 | ctxt->wellFormed = 0; |
| 8373 | ctxt->disableSAX = 1; |
| 8374 | } |
| 8375 | if (ctxt->instate != XML_PARSER_EOF) { |
| 8376 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL) && |
| 8377 | (!ctxt->disableSAX)) |
| 8378 | ctxt->sax->endDocument(ctxt->userData); |
| 8379 | } |
| 8380 | ctxt->instate = XML_PARSER_EOF; |
| 8381 | } |
| 8382 | return((xmlParserErrors) ctxt->errNo); |
| 8383 | } |
| 8384 | |
| 8385 | /************************************************************************ |
| 8386 | * * |
| 8387 | * I/O front end functions to the parser * |
| 8388 | * * |
| 8389 | ************************************************************************/ |
| 8390 | |
| 8391 | /** |
| 8392 | * xmlStopParser: |
| 8393 | * @ctxt: an XML parser context |
| 8394 | * |
| 8395 | * Blocks further parser processing |
| 8396 | */ |
| 8397 | void |
| 8398 | xmlStopParser(xmlParserCtxtPtr ctxt) { |
| 8399 | ctxt->instate = XML_PARSER_EOF; |
| 8400 | if (ctxt->input != NULL) |
| 8401 | ctxt->input->cur = BAD_CAST""; |
| 8402 | } |
| 8403 | |
| 8404 | /** |
| 8405 | * xmlCreatePushParserCtxt: |
| 8406 | * @sax: a SAX handler |
| 8407 | * @user_data: The user data returned on SAX callbacks |
| 8408 | * @chunk: a pointer to an array of chars |
| 8409 | * @size: number of chars in the array |
| 8410 | * @filename: an optional file name or URI |
| 8411 | * |
| 8412 | * Create a parser context for using the XML parser in push mode |
| 8413 | * To allow content encoding detection, @size should be >= 4 |
| 8414 | * The value of @filename is used for fetching external entities |
| 8415 | * and error/warning reports. |
| 8416 | * |
| 8417 | * Returns the new parser context or NULL |
| 8418 | */ |
| 8419 | xmlParserCtxtPtr |
| 8420 | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
| 8421 | const char *chunk, int size, const char *filename) { |
| 8422 | xmlParserCtxtPtr ctxt; |
| 8423 | xmlParserInputPtr inputStream; |
| 8424 | xmlParserInputBufferPtr buf; |
| 8425 | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
| 8426 | |
| 8427 | /* |
| 8428 | * plug some encoding conversion routines |
| 8429 | */ |
| 8430 | if ((chunk != NULL) && (size >= 4)) |
| 8431 | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
| 8432 | |
| 8433 | buf = xmlAllocParserInputBuffer(enc); |
| 8434 | if (buf == NULL) return(NULL); |
| 8435 | |
| 8436 | ctxt = xmlNewParserCtxt(); |
| 8437 | if (ctxt == NULL) { |
| 8438 | xmlFree(buf); |
| 8439 | return(NULL); |
| 8440 | } |
| 8441 | if (sax != NULL) { |
| 8442 | if (ctxt->sax != &xmlDefaultSAXHandler) |
| 8443 | xmlFree(ctxt->sax); |
| 8444 | ctxt->sax = (xmlSAXHandlerPtr) xmlMalloc(sizeof(xmlSAXHandler)); |
| 8445 | if (ctxt->sax == NULL) { |
| 8446 | xmlFree(buf); |
| 8447 | xmlFree(ctxt); |
| 8448 | return(NULL); |
| 8449 | } |
| 8450 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandler)); |
| 8451 | if (user_data != NULL) |
| 8452 | ctxt->userData = user_data; |
| 8453 | } |
| 8454 | if (filename == NULL) { |
| 8455 | ctxt->directory = NULL; |
| 8456 | } else { |
| 8457 | ctxt->directory = xmlParserGetDirectory(filename); |
| 8458 | } |
| 8459 | |
| 8460 | inputStream = xmlNewInputStream(ctxt); |
| 8461 | if (inputStream == NULL) { |
| 8462 | xmlFreeParserCtxt(ctxt); |
| 8463 | return(NULL); |
| 8464 | } |
| 8465 | |
| 8466 | if (filename == NULL) |
| 8467 | inputStream->filename = NULL; |
| 8468 | else |
| 8469 | inputStream->filename = xmlMemStrdup(filename); |
| 8470 | inputStream->buf = buf; |
| 8471 | inputStream->base = inputStream->buf->buffer->content; |
| 8472 | inputStream->cur = inputStream->buf->buffer->content; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 8473 | inputStream->end = |
| 8474 | &inputStream->buf->buffer->content[inputStream->buf->buffer->use]; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 8475 | if (enc != XML_CHAR_ENCODING_NONE) { |
| 8476 | xmlSwitchEncoding(ctxt, enc); |
| 8477 | } |
| 8478 | |
| 8479 | inputPush(ctxt, inputStream); |
| 8480 | |
| 8481 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
| 8482 | (ctxt->input->buf != NULL)) { |
| 8483 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
| 8484 | #ifdef DEBUG_PUSH |
| 8485 | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
| 8486 | #endif |
| 8487 | } |
| 8488 | |
| 8489 | return(ctxt); |
| 8490 | } |
| 8491 | |
| 8492 | /** |
| 8493 | * xmlCreateIOParserCtxt: |
| 8494 | * @sax: a SAX handler |
| 8495 | * @user_data: The user data returned on SAX callbacks |
| 8496 | * @ioread: an I/O read function |
| 8497 | * @ioclose: an I/O close function |
| 8498 | * @ioctx: an I/O handler |
| 8499 | * @enc: the charset encoding if known |
| 8500 | * |
| 8501 | * Create a parser context for using the XML parser with an existing |
| 8502 | * I/O stream |
| 8503 | * |
| 8504 | * Returns the new parser context or NULL |
| 8505 | */ |
| 8506 | xmlParserCtxtPtr |
| 8507 | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
| 8508 | xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
| 8509 | void *ioctx, xmlCharEncoding enc) { |
| 8510 | xmlParserCtxtPtr ctxt; |
| 8511 | xmlParserInputPtr inputStream; |
| 8512 | xmlParserInputBufferPtr buf; |
| 8513 | |
| 8514 | buf = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, enc); |
| 8515 | if (buf == NULL) return(NULL); |
| 8516 | |
| 8517 | ctxt = xmlNewParserCtxt(); |
| 8518 | if (ctxt == NULL) { |
| 8519 | xmlFree(buf); |
| 8520 | return(NULL); |
| 8521 | } |
| 8522 | if (sax != NULL) { |
| 8523 | if (ctxt->sax != &xmlDefaultSAXHandler) |
| 8524 | xmlFree(ctxt->sax); |
| 8525 | ctxt->sax = (xmlSAXHandlerPtr) xmlMalloc(sizeof(xmlSAXHandler)); |
| 8526 | if (ctxt->sax == NULL) { |
| 8527 | xmlFree(buf); |
| 8528 | xmlFree(ctxt); |
| 8529 | return(NULL); |
| 8530 | } |
| 8531 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandler)); |
| 8532 | if (user_data != NULL) |
| 8533 | ctxt->userData = user_data; |
| 8534 | } |
| 8535 | |
| 8536 | inputStream = xmlNewIOInputStream(ctxt, buf, enc); |
| 8537 | if (inputStream == NULL) { |
| 8538 | xmlFreeParserCtxt(ctxt); |
| 8539 | return(NULL); |
| 8540 | } |
| 8541 | inputPush(ctxt, inputStream); |
| 8542 | |
| 8543 | return(ctxt); |
| 8544 | } |
| 8545 | |
| 8546 | /************************************************************************ |
| 8547 | * * |
| 8548 | * Front ends when parsing a Dtd * |
| 8549 | * * |
| 8550 | ************************************************************************/ |
| 8551 | |
| 8552 | /** |
| 8553 | * xmlIOParseDTD: |
| 8554 | * @sax: the SAX handler block or NULL |
| 8555 | * @input: an Input Buffer |
| 8556 | * @enc: the charset encoding if known |
| 8557 | * |
| 8558 | * Load and parse a DTD |
| 8559 | * |
| 8560 | * Returns the resulting xmlDtdPtr or NULL in case of error. |
| 8561 | * @input will be freed at parsing end. |
| 8562 | */ |
| 8563 | |
| 8564 | xmlDtdPtr |
| 8565 | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
| 8566 | xmlCharEncoding enc) { |
| 8567 | xmlDtdPtr ret = NULL; |
| 8568 | xmlParserCtxtPtr ctxt; |
| 8569 | xmlParserInputPtr pinput = NULL; |
| 8570 | |
| 8571 | if (input == NULL) |
| 8572 | return(NULL); |
| 8573 | |
| 8574 | ctxt = xmlNewParserCtxt(); |
| 8575 | if (ctxt == NULL) { |
| 8576 | return(NULL); |
| 8577 | } |
| 8578 | |
| 8579 | /* |
| 8580 | * Set-up the SAX context |
| 8581 | */ |
| 8582 | if (sax != NULL) { |
| 8583 | if (ctxt->sax != NULL) |
| 8584 | xmlFree(ctxt->sax); |
| 8585 | ctxt->sax = sax; |
| 8586 | ctxt->userData = NULL; |
| 8587 | } |
| 8588 | |
| 8589 | /* |
| 8590 | * generate a parser input from the I/O handler |
| 8591 | */ |
| 8592 | |
| 8593 | pinput = xmlNewIOInputStream(ctxt, input, enc); |
| 8594 | if (pinput == NULL) { |
| 8595 | if (sax != NULL) ctxt->sax = NULL; |
| 8596 | xmlFreeParserCtxt(ctxt); |
| 8597 | return(NULL); |
| 8598 | } |
| 8599 | |
| 8600 | /* |
| 8601 | * plug some encoding conversion routines here. |
| 8602 | */ |
| 8603 | xmlPushInput(ctxt, pinput); |
| 8604 | |
| 8605 | pinput->filename = NULL; |
| 8606 | pinput->line = 1; |
| 8607 | pinput->col = 1; |
| 8608 | pinput->base = ctxt->input->cur; |
| 8609 | pinput->cur = ctxt->input->cur; |
| 8610 | pinput->free = NULL; |
| 8611 | |
| 8612 | /* |
| 8613 | * let's parse that entity knowing it's an external subset. |
| 8614 | */ |
| 8615 | ctxt->inSubset = 2; |
| 8616 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
| 8617 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
| 8618 | BAD_CAST "none", BAD_CAST "none"); |
| 8619 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
| 8620 | |
| 8621 | if (ctxt->myDoc != NULL) { |
| 8622 | if (ctxt->wellFormed) { |
| 8623 | ret = ctxt->myDoc->extSubset; |
| 8624 | ctxt->myDoc->extSubset = NULL; |
| 8625 | } else { |
| 8626 | ret = NULL; |
| 8627 | } |
| 8628 | xmlFreeDoc(ctxt->myDoc); |
| 8629 | ctxt->myDoc = NULL; |
| 8630 | } |
| 8631 | if (sax != NULL) ctxt->sax = NULL; |
| 8632 | xmlFreeParserCtxt(ctxt); |
| 8633 | |
| 8634 | return(ret); |
| 8635 | } |
| 8636 | |
| 8637 | /** |
| 8638 | * xmlSAXParseDTD: |
| 8639 | * @sax: the SAX handler block |
| 8640 | * @ExternalID: a NAME* containing the External ID of the DTD |
| 8641 | * @SystemID: a NAME* containing the URL to the DTD |
| 8642 | * |
| 8643 | * Load and parse an external subset. |
| 8644 | * |
| 8645 | * Returns the resulting xmlDtdPtr or NULL in case of error. |
| 8646 | */ |
| 8647 | |
| 8648 | xmlDtdPtr |
| 8649 | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
| 8650 | const xmlChar *SystemID) { |
| 8651 | xmlDtdPtr ret = NULL; |
| 8652 | xmlParserCtxtPtr ctxt; |
| 8653 | xmlParserInputPtr input = NULL; |
| 8654 | xmlCharEncoding enc; |
| 8655 | |
| 8656 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
| 8657 | |
| 8658 | ctxt = xmlNewParserCtxt(); |
| 8659 | if (ctxt == NULL) { |
| 8660 | return(NULL); |
| 8661 | } |
| 8662 | |
| 8663 | /* |
| 8664 | * Set-up the SAX context |
| 8665 | */ |
| 8666 | if (sax != NULL) { |
| 8667 | if (ctxt->sax != NULL) |
| 8668 | xmlFree(ctxt->sax); |
| 8669 | ctxt->sax = sax; |
| 8670 | ctxt->userData = NULL; |
| 8671 | } |
| 8672 | |
| 8673 | /* |
| 8674 | * Ask the Entity resolver to load the damn thing |
| 8675 | */ |
| 8676 | |
| 8677 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
| 8678 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, SystemID); |
| 8679 | if (input == NULL) { |
| 8680 | if (sax != NULL) ctxt->sax = NULL; |
| 8681 | xmlFreeParserCtxt(ctxt); |
| 8682 | return(NULL); |
| 8683 | } |
| 8684 | |
| 8685 | /* |
| 8686 | * plug some encoding conversion routines here. |
| 8687 | */ |
| 8688 | xmlPushInput(ctxt, input); |
| 8689 | enc = xmlDetectCharEncoding(ctxt->input->cur, 4); |
| 8690 | xmlSwitchEncoding(ctxt, enc); |
| 8691 | |
| 8692 | if (input->filename == NULL) |
| 8693 | input->filename = (char *) xmlStrdup(SystemID); |
| 8694 | input->line = 1; |
| 8695 | input->col = 1; |
| 8696 | input->base = ctxt->input->cur; |
| 8697 | input->cur = ctxt->input->cur; |
| 8698 | input->free = NULL; |
| 8699 | |
| 8700 | /* |
| 8701 | * let's parse that entity knowing it's an external subset. |
| 8702 | */ |
| 8703 | ctxt->inSubset = 2; |
| 8704 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
| 8705 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
| 8706 | ExternalID, SystemID); |
| 8707 | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
| 8708 | |
| 8709 | if (ctxt->myDoc != NULL) { |
| 8710 | if (ctxt->wellFormed) { |
| 8711 | ret = ctxt->myDoc->extSubset; |
| 8712 | ctxt->myDoc->extSubset = NULL; |
| 8713 | } else { |
| 8714 | ret = NULL; |
| 8715 | } |
| 8716 | xmlFreeDoc(ctxt->myDoc); |
| 8717 | ctxt->myDoc = NULL; |
| 8718 | } |
| 8719 | if (sax != NULL) ctxt->sax = NULL; |
| 8720 | xmlFreeParserCtxt(ctxt); |
| 8721 | |
| 8722 | return(ret); |
| 8723 | } |
| 8724 | |
| 8725 | /** |
| 8726 | * xmlParseDTD: |
| 8727 | * @ExternalID: a NAME* containing the External ID of the DTD |
| 8728 | * @SystemID: a NAME* containing the URL to the DTD |
| 8729 | * |
| 8730 | * Load and parse an external subset. |
| 8731 | * |
| 8732 | * Returns the resulting xmlDtdPtr or NULL in case of error. |
| 8733 | */ |
| 8734 | |
| 8735 | xmlDtdPtr |
| 8736 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
| 8737 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
| 8738 | } |
| 8739 | |
| 8740 | /************************************************************************ |
| 8741 | * * |
| 8742 | * Front ends when parsing an Entity * |
| 8743 | * * |
| 8744 | ************************************************************************/ |
| 8745 | |
| 8746 | /** |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 8747 | * xmlParseCtxtExternalEntity: |
| 8748 | * @ctx: the existing parsing context |
| 8749 | * @URL: the URL for the entity to load |
| 8750 | * @ID: the System ID for the entity to load |
| 8751 | * @list: the return value for the set of parsed nodes |
| 8752 | * |
| 8753 | * Parse an external general entity within an existing parsing context |
| 8754 | * An external general parsed entity is well-formed if it matches the |
| 8755 | * production labeled extParsedEnt. |
| 8756 | * |
| 8757 | * [78] extParsedEnt ::= TextDecl? content |
| 8758 | * |
| 8759 | * Returns 0 if the entity is well formed, -1 in case of args problem and |
| 8760 | * the parser error code otherwise |
| 8761 | */ |
| 8762 | |
| 8763 | int |
| 8764 | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctx, const xmlChar *URL, |
| 8765 | const xmlChar *ID, xmlNodePtr *list) { |
| 8766 | xmlParserCtxtPtr ctxt; |
| 8767 | xmlDocPtr newDoc; |
| 8768 | xmlSAXHandlerPtr oldsax = NULL; |
| 8769 | int ret = 0; |
| 8770 | |
| 8771 | if (ctx->depth > 40) { |
| 8772 | return(XML_ERR_ENTITY_LOOP); |
| 8773 | } |
| 8774 | |
| 8775 | if (list != NULL) |
| 8776 | *list = NULL; |
| 8777 | if ((URL == NULL) && (ID == NULL)) |
| 8778 | return(-1); |
| 8779 | if (ctx->myDoc == NULL) /* @@ relax but check for dereferences */ |
| 8780 | return(-1); |
| 8781 | |
| 8782 | |
| 8783 | ctxt = xmlCreateEntityParserCtxt(URL, ID, NULL); |
| 8784 | if (ctxt == NULL) return(-1); |
| 8785 | ctxt->userData = ctxt; |
| 8786 | oldsax = ctxt->sax; |
| 8787 | ctxt->sax = ctx->sax; |
| 8788 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
| 8789 | if (newDoc == NULL) { |
| 8790 | xmlFreeParserCtxt(ctxt); |
| 8791 | return(-1); |
| 8792 | } |
| 8793 | if (ctx->myDoc != NULL) { |
| 8794 | newDoc->intSubset = ctx->myDoc->intSubset; |
| 8795 | newDoc->extSubset = ctx->myDoc->extSubset; |
| 8796 | } |
| 8797 | if (ctx->myDoc->URL != NULL) { |
| 8798 | newDoc->URL = xmlStrdup(ctx->myDoc->URL); |
| 8799 | } |
| 8800 | newDoc->children = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
| 8801 | if (newDoc->children == NULL) { |
| 8802 | ctxt->sax = oldsax; |
| 8803 | xmlFreeParserCtxt(ctxt); |
| 8804 | newDoc->intSubset = NULL; |
| 8805 | newDoc->extSubset = NULL; |
| 8806 | xmlFreeDoc(newDoc); |
| 8807 | return(-1); |
| 8808 | } |
| 8809 | nodePush(ctxt, newDoc->children); |
| 8810 | if (ctx->myDoc == NULL) { |
| 8811 | ctxt->myDoc = newDoc; |
| 8812 | } else { |
| 8813 | ctxt->myDoc = ctx->myDoc; |
| 8814 | newDoc->children->doc = ctx->myDoc; |
| 8815 | } |
| 8816 | |
| 8817 | /* |
| 8818 | * Parse a possible text declaration first |
| 8819 | */ |
| 8820 | GROW; |
| 8821 | if ((RAW == '<') && (NXT(1) == '?') && |
| 8822 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 8823 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 8824 | xmlParseTextDecl(ctxt); |
| 8825 | } |
| 8826 | |
| 8827 | /* |
| 8828 | * Doing validity checking on chunk doesn't make sense |
| 8829 | */ |
| 8830 | ctxt->instate = XML_PARSER_CONTENT; |
| 8831 | ctxt->validate = ctx->validate; |
| 8832 | ctxt->loadsubset = ctx->loadsubset; |
| 8833 | ctxt->depth = ctx->depth + 1; |
| 8834 | ctxt->replaceEntities = ctx->replaceEntities; |
| 8835 | if (ctxt->validate) { |
| 8836 | ctxt->vctxt.error = ctx->vctxt.error; |
| 8837 | ctxt->vctxt.warning = ctx->vctxt.warning; |
| 8838 | /* Allocate the Node stack */ |
| 8839 | ctxt->vctxt.nodeTab = (xmlNodePtr *) xmlMalloc(4 * sizeof(xmlNodePtr)); |
| 8840 | if (ctxt->vctxt.nodeTab == NULL) { |
| 8841 | xmlGenericError(xmlGenericErrorContext, |
| 8842 | "xmlParseCtxtExternalEntity: out of memory\n"); |
| 8843 | ctxt->validate = 0; |
| 8844 | ctxt->vctxt.error = NULL; |
| 8845 | ctxt->vctxt.warning = NULL; |
| 8846 | } else { |
| 8847 | ctxt->vctxt.nodeNr = 0; |
| 8848 | ctxt->vctxt.nodeMax = 4; |
| 8849 | ctxt->vctxt.node = NULL; |
| 8850 | } |
| 8851 | } else { |
| 8852 | ctxt->vctxt.error = NULL; |
| 8853 | ctxt->vctxt.warning = NULL; |
| 8854 | } |
| 8855 | |
| 8856 | xmlParseContent(ctxt); |
| 8857 | |
| 8858 | if ((RAW == '<') && (NXT(1) == '/')) { |
| 8859 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 8860 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 8861 | ctxt->sax->error(ctxt->userData, |
| 8862 | "chunk is not well balanced\n"); |
| 8863 | ctxt->wellFormed = 0; |
| 8864 | ctxt->disableSAX = 1; |
| 8865 | } else if (RAW != 0) { |
| 8866 | ctxt->errNo = XML_ERR_EXTRA_CONTENT; |
| 8867 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 8868 | ctxt->sax->error(ctxt->userData, |
| 8869 | "extra content at the end of well balanced chunk\n"); |
| 8870 | ctxt->wellFormed = 0; |
| 8871 | ctxt->disableSAX = 1; |
| 8872 | } |
| 8873 | if (ctxt->node != newDoc->children) { |
| 8874 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 8875 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 8876 | ctxt->sax->error(ctxt->userData, |
| 8877 | "chunk is not well balanced\n"); |
| 8878 | ctxt->wellFormed = 0; |
| 8879 | ctxt->disableSAX = 1; |
| 8880 | } |
| 8881 | |
| 8882 | if (!ctxt->wellFormed) { |
| 8883 | if (ctxt->errNo == 0) |
| 8884 | ret = 1; |
| 8885 | else |
| 8886 | ret = ctxt->errNo; |
| 8887 | } else { |
| 8888 | if (list != NULL) { |
| 8889 | xmlNodePtr cur; |
| 8890 | |
| 8891 | /* |
| 8892 | * Return the newly created nodeset after unlinking it from |
| 8893 | * they pseudo parent. |
| 8894 | */ |
| 8895 | cur = newDoc->children->children; |
| 8896 | *list = cur; |
| 8897 | while (cur != NULL) { |
| 8898 | cur->parent = NULL; |
| 8899 | cur = cur->next; |
| 8900 | } |
| 8901 | newDoc->children->children = NULL; |
| 8902 | } |
| 8903 | ret = 0; |
| 8904 | } |
| 8905 | ctxt->sax = oldsax; |
| 8906 | xmlFreeParserCtxt(ctxt); |
| 8907 | newDoc->intSubset = NULL; |
| 8908 | newDoc->extSubset = NULL; |
| 8909 | xmlFreeDoc(newDoc); |
| 8910 | |
| 8911 | return(ret); |
| 8912 | } |
| 8913 | |
| 8914 | /** |
| 8915 | * xmlParseExternalEntity: |
| 8916 | * @doc: the document the chunk pertains to |
| 8917 | * @sax: the SAX handler bloc (possibly NULL) |
| 8918 | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
| 8919 | * @depth: Used for loop detection, use 0 |
| 8920 | * @URL: the URL for the entity to load |
| 8921 | * @ID: the System ID for the entity to load |
| 8922 | * @list: the return value for the set of parsed nodes |
| 8923 | * |
| 8924 | * Parse an external general entity |
| 8925 | * An external general parsed entity is well-formed if it matches the |
| 8926 | * production labeled extParsedEnt. |
| 8927 | * |
| 8928 | * [78] extParsedEnt ::= TextDecl? content |
| 8929 | * |
| 8930 | * Returns 0 if the entity is well formed, -1 in case of args problem and |
| 8931 | * the parser error code otherwise |
| 8932 | */ |
| 8933 | |
| 8934 | int |
| 8935 | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
| 8936 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *list) { |
| 8937 | xmlParserCtxtPtr ctxt; |
| 8938 | xmlDocPtr newDoc; |
| 8939 | xmlSAXHandlerPtr oldsax = NULL; |
| 8940 | int ret = 0; |
| 8941 | |
| 8942 | if (depth > 40) { |
| 8943 | return(XML_ERR_ENTITY_LOOP); |
| 8944 | } |
| 8945 | |
| 8946 | |
| 8947 | |
| 8948 | if (list != NULL) |
| 8949 | *list = NULL; |
| 8950 | if ((URL == NULL) && (ID == NULL)) |
| 8951 | return(-1); |
| 8952 | if (doc == NULL) /* @@ relax but check for dereferences */ |
| 8953 | return(-1); |
| 8954 | |
| 8955 | |
| 8956 | ctxt = xmlCreateEntityParserCtxt(URL, ID, NULL); |
| 8957 | if (ctxt == NULL) return(-1); |
| 8958 | ctxt->userData = ctxt; |
| 8959 | if (sax != NULL) { |
| 8960 | oldsax = ctxt->sax; |
| 8961 | ctxt->sax = sax; |
| 8962 | if (user_data != NULL) |
| 8963 | ctxt->userData = user_data; |
| 8964 | } |
| 8965 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
| 8966 | if (newDoc == NULL) { |
| 8967 | xmlFreeParserCtxt(ctxt); |
| 8968 | return(-1); |
| 8969 | } |
| 8970 | if (doc != NULL) { |
| 8971 | newDoc->intSubset = doc->intSubset; |
| 8972 | newDoc->extSubset = doc->extSubset; |
| 8973 | } |
| 8974 | if (doc->URL != NULL) { |
| 8975 | newDoc->URL = xmlStrdup(doc->URL); |
| 8976 | } |
| 8977 | newDoc->children = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
| 8978 | if (newDoc->children == NULL) { |
| 8979 | if (sax != NULL) |
| 8980 | ctxt->sax = oldsax; |
| 8981 | xmlFreeParserCtxt(ctxt); |
| 8982 | newDoc->intSubset = NULL; |
| 8983 | newDoc->extSubset = NULL; |
| 8984 | xmlFreeDoc(newDoc); |
| 8985 | return(-1); |
| 8986 | } |
| 8987 | nodePush(ctxt, newDoc->children); |
| 8988 | if (doc == NULL) { |
| 8989 | ctxt->myDoc = newDoc; |
| 8990 | } else { |
| 8991 | ctxt->myDoc = doc; |
| 8992 | newDoc->children->doc = doc; |
| 8993 | } |
| 8994 | |
| 8995 | /* |
| 8996 | * Parse a possible text declaration first |
| 8997 | */ |
| 8998 | GROW; |
| 8999 | if ((RAW == '<') && (NXT(1) == '?') && |
| 9000 | (NXT(2) == 'x') && (NXT(3) == 'm') && |
| 9001 | (NXT(4) == 'l') && (IS_BLANK(NXT(5)))) { |
| 9002 | xmlParseTextDecl(ctxt); |
| 9003 | } |
| 9004 | |
| 9005 | /* |
| 9006 | * Doing validity checking on chunk doesn't make sense |
| 9007 | */ |
| 9008 | ctxt->instate = XML_PARSER_CONTENT; |
| 9009 | ctxt->validate = 0; |
Daniel Veillard | e470df7 | 2001-04-18 21:41:07 +0000 | [diff] [blame] | 9010 | ctxt->external = 2; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9011 | ctxt->loadsubset = 0; |
| 9012 | ctxt->depth = depth; |
| 9013 | |
| 9014 | xmlParseContent(ctxt); |
| 9015 | |
| 9016 | if ((RAW == '<') && (NXT(1) == '/')) { |
| 9017 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 9018 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 9019 | ctxt->sax->error(ctxt->userData, |
| 9020 | "chunk is not well balanced\n"); |
| 9021 | ctxt->wellFormed = 0; |
| 9022 | ctxt->disableSAX = 1; |
| 9023 | } else if (RAW != 0) { |
| 9024 | ctxt->errNo = XML_ERR_EXTRA_CONTENT; |
| 9025 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 9026 | ctxt->sax->error(ctxt->userData, |
| 9027 | "extra content at the end of well balanced chunk\n"); |
| 9028 | ctxt->wellFormed = 0; |
| 9029 | ctxt->disableSAX = 1; |
| 9030 | } |
| 9031 | if (ctxt->node != newDoc->children) { |
| 9032 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 9033 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 9034 | ctxt->sax->error(ctxt->userData, |
| 9035 | "chunk is not well balanced\n"); |
| 9036 | ctxt->wellFormed = 0; |
| 9037 | ctxt->disableSAX = 1; |
| 9038 | } |
| 9039 | |
| 9040 | if (!ctxt->wellFormed) { |
| 9041 | if (ctxt->errNo == 0) |
| 9042 | ret = 1; |
| 9043 | else |
| 9044 | ret = ctxt->errNo; |
| 9045 | } else { |
| 9046 | if (list != NULL) { |
| 9047 | xmlNodePtr cur; |
| 9048 | |
| 9049 | /* |
| 9050 | * Return the newly created nodeset after unlinking it from |
| 9051 | * they pseudo parent. |
| 9052 | */ |
| 9053 | cur = newDoc->children->children; |
| 9054 | *list = cur; |
| 9055 | while (cur != NULL) { |
| 9056 | cur->parent = NULL; |
| 9057 | cur = cur->next; |
| 9058 | } |
| 9059 | newDoc->children->children = NULL; |
| 9060 | } |
| 9061 | ret = 0; |
| 9062 | } |
| 9063 | if (sax != NULL) |
| 9064 | ctxt->sax = oldsax; |
| 9065 | xmlFreeParserCtxt(ctxt); |
| 9066 | newDoc->intSubset = NULL; |
| 9067 | newDoc->extSubset = NULL; |
| 9068 | xmlFreeDoc(newDoc); |
| 9069 | |
| 9070 | return(ret); |
| 9071 | } |
| 9072 | |
| 9073 | /** |
Daniel Veillard | e020c3a | 2001-03-21 18:06:15 +0000 | [diff] [blame] | 9074 | * xmlParseBalancedChunkMemory: |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9075 | * @doc: the document the chunk pertains to |
| 9076 | * @sax: the SAX handler bloc (possibly NULL) |
| 9077 | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
| 9078 | * @depth: Used for loop detection, use 0 |
| 9079 | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
| 9080 | * @list: the return value for the set of parsed nodes |
| 9081 | * |
| 9082 | * Parse a well-balanced chunk of an XML document |
| 9083 | * called by the parser |
| 9084 | * The allowed sequence for the Well Balanced Chunk is the one defined by |
| 9085 | * the content production in the XML grammar: |
| 9086 | * |
| 9087 | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
| 9088 | * |
| 9089 | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
| 9090 | * the parser error code otherwise |
| 9091 | */ |
| 9092 | |
| 9093 | int |
| 9094 | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
| 9095 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *list) { |
| 9096 | xmlParserCtxtPtr ctxt; |
| 9097 | xmlDocPtr newDoc; |
| 9098 | xmlSAXHandlerPtr oldsax = NULL; |
| 9099 | int size; |
| 9100 | int ret = 0; |
| 9101 | |
| 9102 | if (depth > 40) { |
| 9103 | return(XML_ERR_ENTITY_LOOP); |
| 9104 | } |
| 9105 | |
| 9106 | |
| 9107 | if (list != NULL) |
| 9108 | *list = NULL; |
| 9109 | if (string == NULL) |
| 9110 | return(-1); |
| 9111 | |
| 9112 | size = xmlStrlen(string); |
| 9113 | |
| 9114 | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
| 9115 | if (ctxt == NULL) return(-1); |
| 9116 | ctxt->userData = ctxt; |
| 9117 | if (sax != NULL) { |
| 9118 | oldsax = ctxt->sax; |
| 9119 | ctxt->sax = sax; |
| 9120 | if (user_data != NULL) |
| 9121 | ctxt->userData = user_data; |
| 9122 | } |
| 9123 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
| 9124 | if (newDoc == NULL) { |
| 9125 | xmlFreeParserCtxt(ctxt); |
| 9126 | return(-1); |
| 9127 | } |
| 9128 | if (doc != NULL) { |
| 9129 | newDoc->intSubset = doc->intSubset; |
| 9130 | newDoc->extSubset = doc->extSubset; |
| 9131 | } |
| 9132 | newDoc->children = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
| 9133 | if (newDoc->children == NULL) { |
| 9134 | if (sax != NULL) |
| 9135 | ctxt->sax = oldsax; |
| 9136 | xmlFreeParserCtxt(ctxt); |
| 9137 | newDoc->intSubset = NULL; |
| 9138 | newDoc->extSubset = NULL; |
| 9139 | xmlFreeDoc(newDoc); |
| 9140 | return(-1); |
| 9141 | } |
| 9142 | nodePush(ctxt, newDoc->children); |
| 9143 | if (doc == NULL) { |
| 9144 | ctxt->myDoc = newDoc; |
| 9145 | } else { |
| 9146 | ctxt->myDoc = doc; |
| 9147 | newDoc->children->doc = doc; |
| 9148 | } |
| 9149 | ctxt->instate = XML_PARSER_CONTENT; |
| 9150 | ctxt->depth = depth; |
| 9151 | |
| 9152 | /* |
| 9153 | * Doing validity checking on chunk doesn't make sense |
| 9154 | */ |
| 9155 | ctxt->validate = 0; |
| 9156 | ctxt->loadsubset = 0; |
| 9157 | |
| 9158 | xmlParseContent(ctxt); |
| 9159 | |
| 9160 | if ((RAW == '<') && (NXT(1) == '/')) { |
| 9161 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 9162 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 9163 | ctxt->sax->error(ctxt->userData, |
| 9164 | "chunk is not well balanced\n"); |
| 9165 | ctxt->wellFormed = 0; |
| 9166 | ctxt->disableSAX = 1; |
| 9167 | } else if (RAW != 0) { |
| 9168 | ctxt->errNo = XML_ERR_EXTRA_CONTENT; |
| 9169 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 9170 | ctxt->sax->error(ctxt->userData, |
| 9171 | "extra content at the end of well balanced chunk\n"); |
| 9172 | ctxt->wellFormed = 0; |
| 9173 | ctxt->disableSAX = 1; |
| 9174 | } |
| 9175 | if (ctxt->node != newDoc->children) { |
| 9176 | ctxt->errNo = XML_ERR_NOT_WELL_BALANCED; |
| 9177 | if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL)) |
| 9178 | ctxt->sax->error(ctxt->userData, |
| 9179 | "chunk is not well balanced\n"); |
| 9180 | ctxt->wellFormed = 0; |
| 9181 | ctxt->disableSAX = 1; |
| 9182 | } |
| 9183 | |
| 9184 | if (!ctxt->wellFormed) { |
| 9185 | if (ctxt->errNo == 0) |
| 9186 | ret = 1; |
| 9187 | else |
| 9188 | ret = ctxt->errNo; |
| 9189 | } else { |
| 9190 | if (list != NULL) { |
| 9191 | xmlNodePtr cur; |
| 9192 | |
| 9193 | /* |
| 9194 | * Return the newly created nodeset after unlinking it from |
| 9195 | * they pseudo parent. |
| 9196 | */ |
| 9197 | cur = newDoc->children->children; |
| 9198 | *list = cur; |
| 9199 | while (cur != NULL) { |
| 9200 | cur->parent = NULL; |
| 9201 | cur = cur->next; |
| 9202 | } |
| 9203 | newDoc->children->children = NULL; |
| 9204 | } |
| 9205 | ret = 0; |
| 9206 | } |
| 9207 | if (sax != NULL) |
| 9208 | ctxt->sax = oldsax; |
| 9209 | xmlFreeParserCtxt(ctxt); |
| 9210 | newDoc->intSubset = NULL; |
| 9211 | newDoc->extSubset = NULL; |
| 9212 | xmlFreeDoc(newDoc); |
| 9213 | |
| 9214 | return(ret); |
| 9215 | } |
| 9216 | |
| 9217 | /** |
| 9218 | * xmlSAXParseEntity: |
| 9219 | * @sax: the SAX handler block |
| 9220 | * @filename: the filename |
| 9221 | * |
| 9222 | * parse an XML external entity out of context and build a tree. |
| 9223 | * It use the given SAX function block to handle the parsing callback. |
| 9224 | * If sax is NULL, fallback to the default DOM tree building routines. |
| 9225 | * |
| 9226 | * [78] extParsedEnt ::= TextDecl? content |
| 9227 | * |
| 9228 | * This correspond to a "Well Balanced" chunk |
| 9229 | * |
| 9230 | * Returns the resulting document tree |
| 9231 | */ |
| 9232 | |
| 9233 | xmlDocPtr |
| 9234 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
| 9235 | xmlDocPtr ret; |
| 9236 | xmlParserCtxtPtr ctxt; |
| 9237 | char *directory = NULL; |
| 9238 | |
| 9239 | ctxt = xmlCreateFileParserCtxt(filename); |
| 9240 | if (ctxt == NULL) { |
| 9241 | return(NULL); |
| 9242 | } |
| 9243 | if (sax != NULL) { |
| 9244 | if (ctxt->sax != NULL) |
| 9245 | xmlFree(ctxt->sax); |
| 9246 | ctxt->sax = sax; |
| 9247 | ctxt->userData = NULL; |
| 9248 | } |
| 9249 | |
| 9250 | if ((ctxt->directory == NULL) && (directory == NULL)) |
| 9251 | directory = xmlParserGetDirectory(filename); |
| 9252 | |
| 9253 | xmlParseExtParsedEnt(ctxt); |
| 9254 | |
| 9255 | if (ctxt->wellFormed) |
| 9256 | ret = ctxt->myDoc; |
| 9257 | else { |
| 9258 | ret = NULL; |
| 9259 | xmlFreeDoc(ctxt->myDoc); |
| 9260 | ctxt->myDoc = NULL; |
| 9261 | } |
| 9262 | if (sax != NULL) |
| 9263 | ctxt->sax = NULL; |
| 9264 | xmlFreeParserCtxt(ctxt); |
| 9265 | |
| 9266 | return(ret); |
| 9267 | } |
| 9268 | |
| 9269 | /** |
| 9270 | * xmlParseEntity: |
| 9271 | * @filename: the filename |
| 9272 | * |
| 9273 | * parse an XML external entity out of context and build a tree. |
| 9274 | * |
| 9275 | * [78] extParsedEnt ::= TextDecl? content |
| 9276 | * |
| 9277 | * This correspond to a "Well Balanced" chunk |
| 9278 | * |
| 9279 | * Returns the resulting document tree |
| 9280 | */ |
| 9281 | |
| 9282 | xmlDocPtr |
| 9283 | xmlParseEntity(const char *filename) { |
| 9284 | return(xmlSAXParseEntity(NULL, filename)); |
| 9285 | } |
| 9286 | |
| 9287 | /** |
| 9288 | * xmlCreateEntityParserCtxt: |
| 9289 | * @URL: the entity URL |
| 9290 | * @ID: the entity PUBLIC ID |
| 9291 | * @base: a posible base for the target URI |
| 9292 | * |
| 9293 | * Create a parser context for an external entity |
| 9294 | * Automatic support for ZLIB/Compress compressed document is provided |
| 9295 | * by default if found at compile-time. |
| 9296 | * |
| 9297 | * Returns the new parser context or NULL |
| 9298 | */ |
| 9299 | xmlParserCtxtPtr |
| 9300 | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
| 9301 | const xmlChar *base) { |
| 9302 | xmlParserCtxtPtr ctxt; |
| 9303 | xmlParserInputPtr inputStream; |
| 9304 | char *directory = NULL; |
| 9305 | xmlChar *uri; |
| 9306 | |
| 9307 | ctxt = xmlNewParserCtxt(); |
| 9308 | if (ctxt == NULL) { |
| 9309 | return(NULL); |
| 9310 | } |
| 9311 | |
| 9312 | uri = xmlBuildURI(URL, base); |
| 9313 | |
| 9314 | if (uri == NULL) { |
| 9315 | inputStream = xmlLoadExternalEntity((char *)URL, (char *)ID, ctxt); |
| 9316 | if (inputStream == NULL) { |
| 9317 | xmlFreeParserCtxt(ctxt); |
| 9318 | return(NULL); |
| 9319 | } |
| 9320 | |
| 9321 | inputPush(ctxt, inputStream); |
| 9322 | |
| 9323 | if ((ctxt->directory == NULL) && (directory == NULL)) |
| 9324 | directory = xmlParserGetDirectory((char *)URL); |
| 9325 | if ((ctxt->directory == NULL) && (directory != NULL)) |
| 9326 | ctxt->directory = directory; |
| 9327 | } else { |
| 9328 | inputStream = xmlLoadExternalEntity((char *)uri, (char *)ID, ctxt); |
| 9329 | if (inputStream == NULL) { |
| 9330 | xmlFree(uri); |
| 9331 | xmlFreeParserCtxt(ctxt); |
| 9332 | return(NULL); |
| 9333 | } |
| 9334 | |
| 9335 | inputPush(ctxt, inputStream); |
| 9336 | |
| 9337 | if ((ctxt->directory == NULL) && (directory == NULL)) |
| 9338 | directory = xmlParserGetDirectory((char *)uri); |
| 9339 | if ((ctxt->directory == NULL) && (directory != NULL)) |
| 9340 | ctxt->directory = directory; |
| 9341 | xmlFree(uri); |
| 9342 | } |
| 9343 | |
| 9344 | return(ctxt); |
| 9345 | } |
| 9346 | |
| 9347 | /************************************************************************ |
| 9348 | * * |
| 9349 | * Front ends when parsing from a file * |
| 9350 | * * |
| 9351 | ************************************************************************/ |
| 9352 | |
| 9353 | /** |
| 9354 | * xmlCreateFileParserCtxt: |
| 9355 | * @filename: the filename |
| 9356 | * |
| 9357 | * Create a parser context for a file content. |
| 9358 | * Automatic support for ZLIB/Compress compressed document is provided |
| 9359 | * by default if found at compile-time. |
| 9360 | * |
| 9361 | * Returns the new parser context or NULL |
| 9362 | */ |
| 9363 | xmlParserCtxtPtr |
| 9364 | xmlCreateFileParserCtxt(const char *filename) |
| 9365 | { |
| 9366 | xmlParserCtxtPtr ctxt; |
| 9367 | xmlParserInputPtr inputStream; |
| 9368 | xmlParserInputBufferPtr buf; |
| 9369 | char *directory = NULL; |
| 9370 | |
| 9371 | buf = xmlParserInputBufferCreateFilename(filename, XML_CHAR_ENCODING_NONE); |
| 9372 | if (buf == NULL) { |
| 9373 | return(NULL); |
| 9374 | } |
| 9375 | |
| 9376 | ctxt = xmlNewParserCtxt(); |
| 9377 | if (ctxt == NULL) { |
| 9378 | if (xmlDefaultSAXHandler.error != NULL) { |
| 9379 | xmlDefaultSAXHandler.error(NULL, "out of memory\n"); |
| 9380 | } |
| 9381 | return(NULL); |
| 9382 | } |
| 9383 | |
| 9384 | inputStream = xmlNewInputStream(ctxt); |
| 9385 | if (inputStream == NULL) { |
| 9386 | xmlFreeParserCtxt(ctxt); |
| 9387 | return(NULL); |
| 9388 | } |
| 9389 | |
| 9390 | inputStream->filename = xmlMemStrdup(filename); |
| 9391 | inputStream->buf = buf; |
| 9392 | inputStream->base = inputStream->buf->buffer->content; |
| 9393 | inputStream->cur = inputStream->buf->buffer->content; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 9394 | inputStream->end = |
| 9395 | &inputStream->buf->buffer->content[inputStream->buf->buffer->use]; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9396 | |
| 9397 | inputPush(ctxt, inputStream); |
| 9398 | if ((ctxt->directory == NULL) && (directory == NULL)) |
| 9399 | directory = xmlParserGetDirectory(filename); |
| 9400 | if ((ctxt->directory == NULL) && (directory != NULL)) |
| 9401 | ctxt->directory = directory; |
| 9402 | |
| 9403 | return(ctxt); |
| 9404 | } |
| 9405 | |
| 9406 | /** |
| 9407 | * xmlSAXParseFile: |
| 9408 | * @sax: the SAX handler block |
| 9409 | * @filename: the filename |
| 9410 | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
| 9411 | * documents |
| 9412 | * |
| 9413 | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
| 9414 | * compressed document is provided by default if found at compile-time. |
| 9415 | * It use the given SAX function block to handle the parsing callback. |
| 9416 | * If sax is NULL, fallback to the default DOM tree building routines. |
| 9417 | * |
| 9418 | * Returns the resulting document tree |
| 9419 | */ |
| 9420 | |
| 9421 | xmlDocPtr |
| 9422 | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
| 9423 | int recovery) { |
| 9424 | xmlDocPtr ret; |
| 9425 | xmlParserCtxtPtr ctxt; |
| 9426 | char *directory = NULL; |
| 9427 | |
| 9428 | ctxt = xmlCreateFileParserCtxt(filename); |
| 9429 | if (ctxt == NULL) { |
| 9430 | return(NULL); |
| 9431 | } |
| 9432 | if (sax != NULL) { |
| 9433 | if (ctxt->sax != NULL) |
| 9434 | xmlFree(ctxt->sax); |
| 9435 | ctxt->sax = sax; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9436 | } |
| 9437 | |
| 9438 | if ((ctxt->directory == NULL) && (directory == NULL)) |
| 9439 | directory = xmlParserGetDirectory(filename); |
| 9440 | if ((ctxt->directory == NULL) && (directory != NULL)) |
| 9441 | ctxt->directory = (char *) xmlStrdup((xmlChar *) directory); |
| 9442 | |
| 9443 | xmlParseDocument(ctxt); |
| 9444 | |
| 9445 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
| 9446 | else { |
| 9447 | ret = NULL; |
| 9448 | xmlFreeDoc(ctxt->myDoc); |
| 9449 | ctxt->myDoc = NULL; |
| 9450 | } |
| 9451 | if (sax != NULL) |
| 9452 | ctxt->sax = NULL; |
| 9453 | xmlFreeParserCtxt(ctxt); |
| 9454 | |
| 9455 | return(ret); |
| 9456 | } |
| 9457 | |
| 9458 | /** |
| 9459 | * xmlRecoverDoc: |
| 9460 | * @cur: a pointer to an array of xmlChar |
| 9461 | * |
| 9462 | * parse an XML in-memory document and build a tree. |
| 9463 | * In the case the document is not Well Formed, a tree is built anyway |
| 9464 | * |
| 9465 | * Returns the resulting document tree |
| 9466 | */ |
| 9467 | |
| 9468 | xmlDocPtr |
| 9469 | xmlRecoverDoc(xmlChar *cur) { |
| 9470 | return(xmlSAXParseDoc(NULL, cur, 1)); |
| 9471 | } |
| 9472 | |
| 9473 | /** |
| 9474 | * xmlParseFile: |
| 9475 | * @filename: the filename |
| 9476 | * |
| 9477 | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
| 9478 | * compressed document is provided by default if found at compile-time. |
| 9479 | * |
| 9480 | * Returns the resulting document tree |
| 9481 | */ |
| 9482 | |
| 9483 | xmlDocPtr |
| 9484 | xmlParseFile(const char *filename) { |
| 9485 | return(xmlSAXParseFile(NULL, filename, 0)); |
| 9486 | } |
| 9487 | |
| 9488 | /** |
| 9489 | * xmlRecoverFile: |
| 9490 | * @filename: the filename |
| 9491 | * |
| 9492 | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
| 9493 | * compressed document is provided by default if found at compile-time. |
| 9494 | * In the case the document is not Well Formed, a tree is built anyway |
| 9495 | * |
| 9496 | * Returns the resulting document tree |
| 9497 | */ |
| 9498 | |
| 9499 | xmlDocPtr |
| 9500 | xmlRecoverFile(const char *filename) { |
| 9501 | return(xmlSAXParseFile(NULL, filename, 1)); |
| 9502 | } |
| 9503 | |
| 9504 | |
| 9505 | /** |
| 9506 | * xmlSetupParserForBuffer: |
| 9507 | * @ctxt: an XML parser context |
| 9508 | * @buffer: a xmlChar * buffer |
| 9509 | * @filename: a file name |
| 9510 | * |
| 9511 | * Setup the parser context to parse a new buffer; Clears any prior |
| 9512 | * contents from the parser context. The buffer parameter must not be |
| 9513 | * NULL, but the filename parameter can be |
| 9514 | */ |
| 9515 | void |
| 9516 | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
| 9517 | const char* filename) |
| 9518 | { |
| 9519 | xmlParserInputPtr input; |
| 9520 | |
| 9521 | input = xmlNewInputStream(ctxt); |
| 9522 | if (input == NULL) { |
| 9523 | perror("malloc"); |
| 9524 | xmlFree(ctxt); |
| 9525 | return; |
| 9526 | } |
| 9527 | |
| 9528 | xmlClearParserCtxt(ctxt); |
| 9529 | if (filename != NULL) |
| 9530 | input->filename = xmlMemStrdup(filename); |
| 9531 | input->base = buffer; |
| 9532 | input->cur = buffer; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 9533 | input->end = &buffer[xmlStrlen(buffer)]; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9534 | inputPush(ctxt, input); |
| 9535 | } |
| 9536 | |
| 9537 | /** |
| 9538 | * xmlSAXUserParseFile: |
| 9539 | * @sax: a SAX handler |
| 9540 | * @user_data: The user data returned on SAX callbacks |
| 9541 | * @filename: a file name |
| 9542 | * |
| 9543 | * parse an XML file and call the given SAX handler routines. |
| 9544 | * Automatic support for ZLIB/Compress compressed document is provided |
| 9545 | * |
| 9546 | * Returns 0 in case of success or a error number otherwise |
| 9547 | */ |
| 9548 | int |
| 9549 | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
| 9550 | const char *filename) { |
| 9551 | int ret = 0; |
| 9552 | xmlParserCtxtPtr ctxt; |
| 9553 | |
| 9554 | ctxt = xmlCreateFileParserCtxt(filename); |
| 9555 | if (ctxt == NULL) return -1; |
| 9556 | if (ctxt->sax != &xmlDefaultSAXHandler) |
| 9557 | xmlFree(ctxt->sax); |
| 9558 | ctxt->sax = sax; |
| 9559 | if (user_data != NULL) |
| 9560 | ctxt->userData = user_data; |
| 9561 | |
| 9562 | xmlParseDocument(ctxt); |
| 9563 | |
| 9564 | if (ctxt->wellFormed) |
| 9565 | ret = 0; |
| 9566 | else { |
| 9567 | if (ctxt->errNo != 0) |
| 9568 | ret = ctxt->errNo; |
| 9569 | else |
| 9570 | ret = -1; |
| 9571 | } |
| 9572 | if (sax != NULL) |
| 9573 | ctxt->sax = NULL; |
| 9574 | xmlFreeParserCtxt(ctxt); |
| 9575 | |
| 9576 | return ret; |
| 9577 | } |
| 9578 | |
| 9579 | /************************************************************************ |
| 9580 | * * |
| 9581 | * Front ends when parsing from memory * |
| 9582 | * * |
| 9583 | ************************************************************************/ |
| 9584 | |
| 9585 | /** |
| 9586 | * xmlCreateMemoryParserCtxt: |
| 9587 | * @buffer: a pointer to a char array |
| 9588 | * @size: the size of the array |
| 9589 | * |
| 9590 | * Create a parser context for an XML in-memory document. |
| 9591 | * |
| 9592 | * Returns the new parser context or NULL |
| 9593 | */ |
| 9594 | xmlParserCtxtPtr |
| 9595 | xmlCreateMemoryParserCtxt(char *buffer, int size) { |
| 9596 | xmlParserCtxtPtr ctxt; |
| 9597 | xmlParserInputPtr input; |
| 9598 | xmlParserInputBufferPtr buf; |
| 9599 | |
| 9600 | if (buffer == NULL) |
| 9601 | return(NULL); |
| 9602 | if (size <= 0) |
| 9603 | return(NULL); |
| 9604 | |
| 9605 | ctxt = xmlNewParserCtxt(); |
| 9606 | if (ctxt == NULL) |
| 9607 | return(NULL); |
| 9608 | |
| 9609 | buf = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
| 9610 | if (buf == NULL) return(NULL); |
| 9611 | |
| 9612 | input = xmlNewInputStream(ctxt); |
| 9613 | if (input == NULL) { |
| 9614 | xmlFreeParserCtxt(ctxt); |
| 9615 | return(NULL); |
| 9616 | } |
| 9617 | |
| 9618 | input->filename = NULL; |
| 9619 | input->buf = buf; |
| 9620 | input->base = input->buf->buffer->content; |
| 9621 | input->cur = input->buf->buffer->content; |
Daniel Veillard | 48b2f89 | 2001-02-25 16:11:03 +0000 | [diff] [blame] | 9622 | input->end = &input->buf->buffer->content[input->buf->buffer->use]; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9623 | |
| 9624 | inputPush(ctxt, input); |
| 9625 | return(ctxt); |
| 9626 | } |
| 9627 | |
| 9628 | /** |
| 9629 | * xmlSAXParseMemory: |
| 9630 | * @sax: the SAX handler block |
| 9631 | * @buffer: an pointer to a char array |
| 9632 | * @size: the size of the array |
| 9633 | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
| 9634 | * documents |
| 9635 | * |
| 9636 | * parse an XML in-memory block and use the given SAX function block |
| 9637 | * to handle the parsing callback. If sax is NULL, fallback to the default |
| 9638 | * DOM tree building routines. |
| 9639 | * |
| 9640 | * Returns the resulting document tree |
| 9641 | */ |
| 9642 | xmlDocPtr |
| 9643 | xmlSAXParseMemory(xmlSAXHandlerPtr sax, char *buffer, int size, int recovery) { |
| 9644 | xmlDocPtr ret; |
| 9645 | xmlParserCtxtPtr ctxt; |
| 9646 | |
| 9647 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
| 9648 | if (ctxt == NULL) return(NULL); |
| 9649 | if (sax != NULL) { |
| 9650 | ctxt->sax = sax; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9651 | } |
| 9652 | |
| 9653 | xmlParseDocument(ctxt); |
| 9654 | |
| 9655 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
| 9656 | else { |
| 9657 | ret = NULL; |
| 9658 | xmlFreeDoc(ctxt->myDoc); |
| 9659 | ctxt->myDoc = NULL; |
| 9660 | } |
| 9661 | if (sax != NULL) |
| 9662 | ctxt->sax = NULL; |
| 9663 | xmlFreeParserCtxt(ctxt); |
| 9664 | |
| 9665 | return(ret); |
| 9666 | } |
| 9667 | |
| 9668 | /** |
| 9669 | * xmlParseMemory: |
| 9670 | * @buffer: an pointer to a char array |
| 9671 | * @size: the size of the array |
| 9672 | * |
| 9673 | * parse an XML in-memory block and build a tree. |
| 9674 | * |
| 9675 | * Returns the resulting document tree |
| 9676 | */ |
| 9677 | |
| 9678 | xmlDocPtr xmlParseMemory(char *buffer, int size) { |
| 9679 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
| 9680 | } |
| 9681 | |
| 9682 | /** |
| 9683 | * xmlRecoverMemory: |
| 9684 | * @buffer: an pointer to a char array |
| 9685 | * @size: the size of the array |
| 9686 | * |
| 9687 | * parse an XML in-memory block and build a tree. |
| 9688 | * In the case the document is not Well Formed, a tree is built anyway |
| 9689 | * |
| 9690 | * Returns the resulting document tree |
| 9691 | */ |
| 9692 | |
| 9693 | xmlDocPtr xmlRecoverMemory(char *buffer, int size) { |
| 9694 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
| 9695 | } |
| 9696 | |
| 9697 | /** |
| 9698 | * xmlSAXUserParseMemory: |
| 9699 | * @sax: a SAX handler |
| 9700 | * @user_data: The user data returned on SAX callbacks |
| 9701 | * @buffer: an in-memory XML document input |
| 9702 | * @size: the length of the XML document in bytes |
| 9703 | * |
| 9704 | * A better SAX parsing routine. |
| 9705 | * parse an XML in-memory buffer and call the given SAX handler routines. |
| 9706 | * |
| 9707 | * Returns 0 in case of success or a error number otherwise |
| 9708 | */ |
| 9709 | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
| 9710 | char *buffer, int size) { |
| 9711 | int ret = 0; |
| 9712 | xmlParserCtxtPtr ctxt; |
| 9713 | xmlSAXHandlerPtr oldsax = NULL; |
| 9714 | |
| 9715 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
| 9716 | if (ctxt == NULL) return -1; |
| 9717 | if (sax != NULL) { |
| 9718 | oldsax = ctxt->sax; |
| 9719 | ctxt->sax = sax; |
| 9720 | } |
Daniel Veillard | 30211a0 | 2001-04-26 09:33:18 +0000 | [diff] [blame] | 9721 | if (user_data != NULL) |
| 9722 | ctxt->userData = user_data; |
Owen Taylor | 3473f88 | 2001-02-23 17:55:21 +0000 | [diff] [blame] | 9723 | |
| 9724 | xmlParseDocument(ctxt); |
| 9725 | |
| 9726 | if (ctxt->wellFormed) |
| 9727 | ret = 0; |
| 9728 | else { |
| 9729 | if (ctxt->errNo != 0) |
| 9730 | ret = ctxt->errNo; |
| 9731 | else |
| 9732 | ret = -1; |
| 9733 | } |
| 9734 | if (sax != NULL) { |
| 9735 | ctxt->sax = oldsax; |
| 9736 | } |
| 9737 | xmlFreeParserCtxt(ctxt); |
| 9738 | |
| 9739 | return ret; |
| 9740 | } |
| 9741 | |
| 9742 | /** |
| 9743 | * xmlCreateDocParserCtxt: |
| 9744 | * @cur: a pointer to an array of xmlChar |
| 9745 | * |
| 9746 | * Creates a parser context for an XML in-memory document. |
| 9747 | * |
| 9748 | * Returns the new parser context or NULL |
| 9749 | */ |
| 9750 | xmlParserCtxtPtr |
| 9751 | xmlCreateDocParserCtxt(xmlChar *cur) { |
| 9752 | int len; |
| 9753 | |
| 9754 | if (cur == NULL) |
| 9755 | return(NULL); |
| 9756 | len = xmlStrlen(cur); |
| 9757 | return(xmlCreateMemoryParserCtxt((char *)cur, len)); |
| 9758 | } |
| 9759 | |
| 9760 | /** |
| 9761 | * xmlSAXParseDoc: |
| 9762 | * @sax: the SAX handler block |
| 9763 | * @cur: a pointer to an array of xmlChar |
| 9764 | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
| 9765 | * documents |
| 9766 | * |
| 9767 | * parse an XML in-memory document and build a tree. |
| 9768 | * It use the given SAX function block to handle the parsing callback. |
| 9769 | * If sax is NULL, fallback to the default DOM tree building routines. |
| 9770 | * |
| 9771 | * Returns the resulting document tree |
| 9772 | */ |
| 9773 | |
| 9774 | xmlDocPtr |
| 9775 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, xmlChar *cur, int recovery) { |
| 9776 | xmlDocPtr ret; |
| 9777 | xmlParserCtxtPtr ctxt; |
| 9778 | |
| 9779 | if (cur == NULL) return(NULL); |
| 9780 | |
| 9781 | |
| 9782 | ctxt = xmlCreateDocParserCtxt(cur); |
| 9783 | if (ctxt == NULL) return(NULL); |
| 9784 | if (sax != NULL) { |
| 9785 | ctxt->sax = sax; |
| 9786 | ctxt->userData = NULL; |
| 9787 | } |
| 9788 | |
| 9789 | xmlParseDocument(ctxt); |
| 9790 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
| 9791 | else { |
| 9792 | ret = NULL; |
| 9793 | xmlFreeDoc(ctxt->myDoc); |
| 9794 | ctxt->myDoc = NULL; |
| 9795 | } |
| 9796 | if (sax != NULL) |
| 9797 | ctxt->sax = NULL; |
| 9798 | xmlFreeParserCtxt(ctxt); |
| 9799 | |
| 9800 | return(ret); |
| 9801 | } |
| 9802 | |
| 9803 | /** |
| 9804 | * xmlParseDoc: |
| 9805 | * @cur: a pointer to an array of xmlChar |
| 9806 | * |
| 9807 | * parse an XML in-memory document and build a tree. |
| 9808 | * |
| 9809 | * Returns the resulting document tree |
| 9810 | */ |
| 9811 | |
| 9812 | xmlDocPtr |
| 9813 | xmlParseDoc(xmlChar *cur) { |
| 9814 | return(xmlSAXParseDoc(NULL, cur, 0)); |
| 9815 | } |
| 9816 | |
| 9817 | |
| 9818 | /************************************************************************ |
| 9819 | * * |
| 9820 | * Miscellaneous * |
| 9821 | * * |
| 9822 | ************************************************************************/ |
| 9823 | |
| 9824 | #ifdef LIBXML_XPATH_ENABLED |
| 9825 | #include <libxml/xpath.h> |
| 9826 | #endif |
| 9827 | |
| 9828 | static int xmlParserInitialized = 0; |
| 9829 | |
| 9830 | /** |
| 9831 | * xmlInitParser: |
| 9832 | * |
| 9833 | * Initialization function for the XML parser. |
| 9834 | * This is not reentrant. Call once before processing in case of |
| 9835 | * use in multithreaded programs. |
| 9836 | */ |
| 9837 | |
| 9838 | void |
| 9839 | xmlInitParser(void) { |
| 9840 | if (xmlParserInitialized) return; |
| 9841 | |
| 9842 | xmlInitCharEncodingHandlers(); |
| 9843 | xmlInitializePredefinedEntities(); |
| 9844 | xmlDefaultSAXHandlerInit(); |
| 9845 | xmlRegisterDefaultInputCallbacks(); |
| 9846 | xmlRegisterDefaultOutputCallbacks(); |
| 9847 | #ifdef LIBXML_HTML_ENABLED |
| 9848 | htmlInitAutoClose(); |
| 9849 | htmlDefaultSAXHandlerInit(); |
| 9850 | #endif |
| 9851 | #ifdef LIBXML_XPATH_ENABLED |
| 9852 | xmlXPathInit(); |
| 9853 | #endif |
| 9854 | xmlParserInitialized = 1; |
| 9855 | } |
| 9856 | |
| 9857 | /** |
| 9858 | * xmlCleanupParser: |
| 9859 | * |
| 9860 | * Cleanup function for the XML parser. It tries to reclaim all |
| 9861 | * parsing related global memory allocated for the parser processing. |
| 9862 | * It doesn't deallocate any document related memory. Calling this |
| 9863 | * function should not prevent reusing the parser. |
| 9864 | */ |
| 9865 | |
| 9866 | void |
| 9867 | xmlCleanupParser(void) { |
| 9868 | xmlParserInitialized = 0; |
| 9869 | xmlCleanupCharEncodingHandlers(); |
| 9870 | xmlCleanupPredefinedEntities(); |
| 9871 | } |
| 9872 | |
| 9873 | /** |
| 9874 | * xmlPedanticParserDefault: |
| 9875 | * @val: int 0 or 1 |
| 9876 | * |
| 9877 | * Set and return the previous value for enabling pedantic warnings. |
| 9878 | * |
| 9879 | * Returns the last value for 0 for no substitution, 1 for substitution. |
| 9880 | */ |
| 9881 | |
| 9882 | int |
| 9883 | xmlPedanticParserDefault(int val) { |
| 9884 | int old = xmlPedanticParserDefaultValue; |
| 9885 | |
| 9886 | xmlPedanticParserDefaultValue = val; |
| 9887 | return(old); |
| 9888 | } |
| 9889 | |
| 9890 | /** |
| 9891 | * xmlSubstituteEntitiesDefault: |
| 9892 | * @val: int 0 or 1 |
| 9893 | * |
| 9894 | * Set and return the previous value for default entity support. |
| 9895 | * Initially the parser always keep entity references instead of substituting |
| 9896 | * entity values in the output. This function has to be used to change the |
| 9897 | * default parser behaviour |
| 9898 | * SAX::subtituteEntities() has to be used for changing that on a file by |
| 9899 | * file basis. |
| 9900 | * |
| 9901 | * Returns the last value for 0 for no substitution, 1 for substitution. |
| 9902 | */ |
| 9903 | |
| 9904 | int |
| 9905 | xmlSubstituteEntitiesDefault(int val) { |
| 9906 | int old = xmlSubstituteEntitiesDefaultValue; |
| 9907 | |
| 9908 | xmlSubstituteEntitiesDefaultValue = val; |
| 9909 | return(old); |
| 9910 | } |
| 9911 | |
| 9912 | /** |
| 9913 | * xmlKeepBlanksDefault: |
| 9914 | * @val: int 0 or 1 |
| 9915 | * |
| 9916 | * Set and return the previous value for default blanks text nodes support. |
| 9917 | * The 1.x version of the parser used an heuristic to try to detect |
| 9918 | * ignorable white spaces. As a result the SAX callback was generating |
| 9919 | * ignorableWhitespace() callbacks instead of characters() one, and when |
| 9920 | * using the DOM output text nodes containing those blanks were not generated. |
| 9921 | * The 2.x and later version will switch to the XML standard way and |
| 9922 | * ignorableWhitespace() are only generated when running the parser in |
| 9923 | * validating mode and when the current element doesn't allow CDATA or |
| 9924 | * mixed content. |
| 9925 | * This function is provided as a way to force the standard behaviour |
| 9926 | * on 1.X libs and to switch back to the old mode for compatibility when |
| 9927 | * running 1.X client code on 2.X . Upgrade of 1.X code should be done |
| 9928 | * by using xmlIsBlankNode() commodity function to detect the "empty" |
| 9929 | * nodes generated. |
| 9930 | * This value also affect autogeneration of indentation when saving code |
| 9931 | * if blanks sections are kept, indentation is not generated. |
| 9932 | * |
| 9933 | * Returns the last value for 0 for no substitution, 1 for substitution. |
| 9934 | */ |
| 9935 | |
| 9936 | int |
| 9937 | xmlKeepBlanksDefault(int val) { |
| 9938 | int old = xmlKeepBlanksDefaultValue; |
| 9939 | |
| 9940 | xmlKeepBlanksDefaultValue = val; |
| 9941 | xmlIndentTreeOutput = !val; |
| 9942 | return(old); |
| 9943 | } |
| 9944 | |