Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 1 | /* |
| 2 | * pattern.c: Implemetation of selectors for nodes |
| 3 | * |
| 4 | * Reference: |
| 5 | * http://www.w3.org/TR/2001/REC-xmlschema-1-20010502/ |
| 6 | * to some extent |
| 7 | * http://www.w3.org/TR/1999/REC-xml-19991116 |
| 8 | * |
| 9 | * See Copyright for the status of this software. |
| 10 | * |
| 11 | * daniel@veillard.com |
| 12 | */ |
| 13 | |
| 14 | #define IN_LIBXML |
| 15 | #include "libxml.h" |
| 16 | |
| 17 | #include <string.h> |
| 18 | #include <libxml/xmlmemory.h> |
| 19 | #include <libxml/tree.h> |
| 20 | #include <libxml/hash.h> |
| 21 | #include <libxml/dict.h> |
| 22 | #include <libxml/xmlerror.h> |
| 23 | #include <libxml/parserInternals.h> |
| 24 | #include <libxml/pattern.h> |
| 25 | |
| 26 | #ifdef LIBXML_PATTERN_ENABLED |
| 27 | |
| 28 | #define ERROR(a, b, c, d) |
| 29 | #define ERROR5(a, b, c, d, e) |
| 30 | |
| 31 | /* |
| 32 | * Types are private: |
| 33 | */ |
| 34 | |
| 35 | typedef enum { |
| 36 | XML_OP_END=0, |
| 37 | XML_OP_ROOT, |
| 38 | XML_OP_ELEM, |
| 39 | XML_OP_CHILD, |
| 40 | XML_OP_ATTR, |
| 41 | XML_OP_PARENT, |
| 42 | XML_OP_ANCESTOR, |
| 43 | XML_OP_NS, |
| 44 | XML_OP_ALL |
| 45 | } xmlPatOp; |
| 46 | |
| 47 | |
| 48 | typedef struct _xmlStepOp xmlStepOp; |
| 49 | typedef xmlStepOp *xmlStepOpPtr; |
| 50 | struct _xmlStepOp { |
| 51 | xmlPatOp op; |
| 52 | const xmlChar *value; |
| 53 | const xmlChar *value2; |
| 54 | }; |
| 55 | |
| 56 | struct _xmlPattern { |
| 57 | void *data; /* the associated template */ |
| 58 | struct _xmlPattern *next; /* siblings */ |
| 59 | const xmlChar *pattern; /* the pattern */ |
| 60 | |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 61 | int nbStep; |
| 62 | int maxStep; |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 63 | xmlStepOpPtr steps; /* ops for computation */ |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 64 | }; |
| 65 | |
| 66 | typedef struct _xmlPatParserContext xmlPatParserContext; |
| 67 | typedef xmlPatParserContext *xmlPatParserContextPtr; |
| 68 | struct _xmlPatParserContext { |
| 69 | const xmlChar *cur; /* the current char being parsed */ |
| 70 | const xmlChar *base; /* the full expression */ |
| 71 | int error; /* error code */ |
| 72 | xmlDictPtr dict; /* the dictionnary if any */ |
| 73 | xmlPatternPtr comp; /* the result */ |
| 74 | xmlNodePtr elem; /* the current node if any */ |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 75 | const xmlChar **namespaces; /* the namespaces definitions */ |
| 76 | int nb_namespaces; /* the number of namespaces */ |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 77 | }; |
| 78 | |
| 79 | /************************************************************************ |
| 80 | * * |
| 81 | * Type functions * |
| 82 | * * |
| 83 | ************************************************************************/ |
| 84 | |
| 85 | /** |
| 86 | * xmlNewPattern: |
| 87 | * |
| 88 | * Create a new XSLT Pattern |
| 89 | * |
| 90 | * Returns the newly allocated xmlPatternPtr or NULL in case of error |
| 91 | */ |
| 92 | static xmlPatternPtr |
| 93 | xmlNewPattern(void) { |
| 94 | xmlPatternPtr cur; |
| 95 | |
| 96 | cur = (xmlPatternPtr) xmlMalloc(sizeof(xmlPattern)); |
| 97 | if (cur == NULL) { |
| 98 | ERROR(NULL, NULL, NULL, |
| 99 | "xmlNewPattern : malloc failed\n"); |
| 100 | return(NULL); |
| 101 | } |
| 102 | memset(cur, 0, sizeof(xmlPattern)); |
| 103 | cur->maxStep = 10; |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 104 | cur->steps = (xmlStepOpPtr) xmlMalloc(cur->maxStep * sizeof(xmlStepOp)); |
| 105 | if (cur->steps == NULL) { |
| 106 | xmlFree(cur); |
| 107 | ERROR(NULL, NULL, NULL, |
| 108 | "xmlNewPattern : malloc failed\n"); |
| 109 | return(NULL); |
| 110 | } |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 111 | return(cur); |
| 112 | } |
| 113 | |
| 114 | /** |
| 115 | * xmlFreePattern: |
| 116 | * @comp: an XSLT comp |
| 117 | * |
| 118 | * Free up the memory allocated by @comp |
| 119 | */ |
| 120 | void |
| 121 | xmlFreePattern(xmlPatternPtr comp) { |
| 122 | xmlStepOpPtr op; |
| 123 | int i; |
| 124 | |
| 125 | if (comp == NULL) |
| 126 | return; |
| 127 | if (comp->pattern != NULL) |
| 128 | xmlFree((xmlChar *)comp->pattern); |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 129 | if (comp->steps != NULL) { |
| 130 | for (i = 0;i < comp->nbStep;i++) { |
| 131 | op = &comp->steps[i]; |
| 132 | if (op->value != NULL) |
| 133 | xmlFree((xmlChar *) op->value); |
| 134 | if (op->value2 != NULL) |
| 135 | xmlFree((xmlChar *) op->value2); |
| 136 | } |
| 137 | xmlFree(comp->steps); |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 138 | } |
| 139 | memset(comp, -1, sizeof(xmlPattern)); |
| 140 | xmlFree(comp); |
| 141 | } |
| 142 | |
| 143 | /** |
| 144 | * xmlFreePatternList: |
| 145 | * @comp: an XSLT comp list |
| 146 | * |
| 147 | * Free up the memory allocated by all the elements of @comp |
| 148 | */ |
| 149 | void |
| 150 | xmlFreePatternList(xmlPatternPtr comp) { |
| 151 | xmlPatternPtr cur; |
| 152 | |
| 153 | while (comp != NULL) { |
| 154 | cur = comp; |
| 155 | comp = comp->next; |
| 156 | xmlFreePattern(cur); |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | /** |
| 161 | * xmlNewPatParserContext: |
| 162 | * @pattern: the pattern context |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 163 | * @dict: the inherited dictionnary or NULL |
| 164 | * @namespaces: the prefix definitions, array of [URI, prefix] or NULL |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 165 | * |
| 166 | * Create a new XML pattern parser context |
| 167 | * |
| 168 | * Returns the newly allocated xmlPatParserContextPtr or NULL in case of error |
| 169 | */ |
| 170 | static xmlPatParserContextPtr |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 171 | xmlNewPatParserContext(const xmlChar *pattern, xmlDictPtr dict, |
| 172 | const xmlChar **namespaces) { |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 173 | xmlPatParserContextPtr cur; |
| 174 | |
| 175 | if (pattern == NULL) |
| 176 | return(NULL); |
| 177 | |
| 178 | cur = (xmlPatParserContextPtr) xmlMalloc(sizeof(xmlPatParserContext)); |
| 179 | if (cur == NULL) { |
| 180 | ERROR(NULL, NULL, NULL, |
| 181 | "xmlNewPatParserContext : malloc failed\n"); |
| 182 | return(NULL); |
| 183 | } |
| 184 | memset(cur, 0, sizeof(xmlPatParserContext)); |
| 185 | cur->dict = dict; |
| 186 | cur->cur = pattern; |
| 187 | cur->base = pattern; |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 188 | if (namespaces != NULL) { |
| 189 | int i; |
| 190 | for (i = 0;namespaces[2 * i] != NULL;i++); |
| 191 | cur->nb_namespaces = i; |
| 192 | } else { |
| 193 | cur->nb_namespaces = 0; |
| 194 | } |
| 195 | cur->namespaces = namespaces; |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 196 | return(cur); |
| 197 | } |
| 198 | |
| 199 | /** |
| 200 | * xmlFreePatParserContext: |
| 201 | * @ctxt: an XSLT parser context |
| 202 | * |
| 203 | * Free up the memory allocated by @ctxt |
| 204 | */ |
| 205 | static void |
| 206 | xmlFreePatParserContext(xmlPatParserContextPtr ctxt) { |
| 207 | if (ctxt == NULL) |
| 208 | return; |
| 209 | memset(ctxt, -1, sizeof(xmlPatParserContext)); |
| 210 | xmlFree(ctxt); |
| 211 | } |
| 212 | |
| 213 | /** |
| 214 | * xmlPatternAdd: |
| 215 | * @comp: the compiled match expression |
| 216 | * @op: an op |
| 217 | * @value: the first value |
| 218 | * @value2: the second value |
| 219 | * |
| 220 | * Add an step to an XSLT Compiled Match |
| 221 | * |
| 222 | * Returns -1 in case of failure, 0 otherwise. |
| 223 | */ |
| 224 | static int |
| 225 | xmlPatternAdd(xmlPatParserContextPtr ctxt ATTRIBUTE_UNUSED, |
| 226 | xmlPatternPtr comp, |
| 227 | xmlPatOp op, xmlChar * value, xmlChar * value2) |
| 228 | { |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 229 | if (comp->nbStep >= comp->maxStep) { |
| 230 | xmlStepOpPtr temp; |
| 231 | temp = (xmlStepOpPtr) xmlRealloc(comp->steps, comp->maxStep * 2 * |
| 232 | sizeof(xmlStepOp)); |
| 233 | if (temp == NULL) { |
| 234 | ERROR(ctxt, NULL, NULL, |
| 235 | "xmlPatternAdd: realloc failed\n"); |
| 236 | return (-1); |
| 237 | } |
| 238 | comp->steps = temp; |
| 239 | comp->maxStep *= 2; |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 240 | } |
| 241 | comp->steps[comp->nbStep].op = op; |
| 242 | comp->steps[comp->nbStep].value = value; |
| 243 | comp->steps[comp->nbStep].value2 = value2; |
| 244 | comp->nbStep++; |
| 245 | return (0); |
| 246 | } |
| 247 | |
| 248 | #if 0 |
| 249 | /** |
| 250 | * xsltSwapTopPattern: |
| 251 | * @comp: the compiled match expression |
| 252 | * |
| 253 | * reverse the two top steps. |
| 254 | */ |
| 255 | static void |
| 256 | xsltSwapTopPattern(xmlPatternPtr comp) { |
| 257 | int i; |
| 258 | int j = comp->nbStep - 1; |
| 259 | |
| 260 | if (j > 0) { |
| 261 | register const xmlChar *tmp; |
| 262 | register xmlPatOp op; |
| 263 | i = j - 1; |
| 264 | tmp = comp->steps[i].value; |
| 265 | comp->steps[i].value = comp->steps[j].value; |
| 266 | comp->steps[j].value = tmp; |
| 267 | tmp = comp->steps[i].value2; |
| 268 | comp->steps[i].value2 = comp->steps[j].value2; |
| 269 | comp->steps[j].value2 = tmp; |
| 270 | op = comp->steps[i].op; |
| 271 | comp->steps[i].op = comp->steps[j].op; |
| 272 | comp->steps[j].op = op; |
| 273 | } |
| 274 | } |
| 275 | #endif |
| 276 | |
| 277 | /** |
| 278 | * xmlReversePattern: |
| 279 | * @comp: the compiled match expression |
| 280 | * |
| 281 | * reverse all the stack of expressions |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 282 | * |
| 283 | * returns 0 in case of success and -1 in case of error. |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 284 | */ |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 285 | static int |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 286 | xmlReversePattern(xmlPatternPtr comp) { |
| 287 | int i = 0; |
| 288 | int j = comp->nbStep - 1; |
| 289 | |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 290 | if (comp->nbStep >= comp->maxStep) { |
| 291 | xmlStepOpPtr temp; |
| 292 | temp = (xmlStepOpPtr) xmlRealloc(comp->steps, comp->maxStep * 2 * |
| 293 | sizeof(xmlStepOp)); |
| 294 | if (temp == NULL) { |
| 295 | ERROR(ctxt, NULL, NULL, |
| 296 | "xmlReversePattern: realloc failed\n"); |
| 297 | return (-1); |
| 298 | } |
| 299 | comp->steps = temp; |
| 300 | comp->maxStep *= 2; |
| 301 | } |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 302 | while (j > i) { |
| 303 | register const xmlChar *tmp; |
| 304 | register xmlPatOp op; |
| 305 | tmp = comp->steps[i].value; |
| 306 | comp->steps[i].value = comp->steps[j].value; |
| 307 | comp->steps[j].value = tmp; |
| 308 | tmp = comp->steps[i].value2; |
| 309 | comp->steps[i].value2 = comp->steps[j].value2; |
| 310 | comp->steps[j].value2 = tmp; |
| 311 | op = comp->steps[i].op; |
| 312 | comp->steps[i].op = comp->steps[j].op; |
| 313 | comp->steps[j].op = op; |
| 314 | j--; |
| 315 | i++; |
| 316 | } |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 317 | comp->steps[comp->nbStep].value = NULL; |
| 318 | comp->steps[comp->nbStep].value2 = NULL; |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 319 | comp->steps[comp->nbStep++].op = XML_OP_END; |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 320 | return(0); |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 321 | } |
| 322 | |
| 323 | /************************************************************************ |
| 324 | * * |
| 325 | * The interpreter for the precompiled patterns * |
| 326 | * * |
| 327 | ************************************************************************/ |
| 328 | |
| 329 | /** |
| 330 | * xmlPatMatch: |
| 331 | * @comp: the precompiled pattern |
| 332 | * @node: a node |
| 333 | * |
| 334 | * Test wether the node matches the pattern |
| 335 | * |
| 336 | * Returns 1 if it matches, 0 if it doesn't and -1 in case of failure |
| 337 | */ |
| 338 | static int |
| 339 | xmlPatMatch(xmlPatternPtr comp, xmlNodePtr node) { |
| 340 | int i; |
| 341 | xmlStepOpPtr step; |
| 342 | |
| 343 | if ((comp == NULL) || (node == NULL)) return(-1); |
| 344 | for (i = 0;i < comp->nbStep;i++) { |
| 345 | step = &comp->steps[i]; |
| 346 | switch (step->op) { |
| 347 | case XML_OP_END: |
| 348 | return(1); |
| 349 | case XML_OP_ROOT: |
| 350 | if ((node->type == XML_DOCUMENT_NODE) || |
| 351 | #ifdef LIBXML_DOCB_ENABLED |
| 352 | (node->type == XML_DOCB_DOCUMENT_NODE) || |
| 353 | #endif |
| 354 | (node->type == XML_HTML_DOCUMENT_NODE)) |
| 355 | continue; |
| 356 | return(0); |
| 357 | case XML_OP_ELEM: |
| 358 | if (node->type != XML_ELEMENT_NODE) |
| 359 | return(0); |
| 360 | if (step->value == NULL) |
| 361 | continue; |
| 362 | if (step->value[0] != node->name[0]) |
| 363 | return(0); |
| 364 | if (!xmlStrEqual(step->value, node->name)) |
| 365 | return(0); |
| 366 | |
| 367 | /* Namespace test */ |
| 368 | if (node->ns == NULL) { |
| 369 | if (step->value2 != NULL) |
| 370 | return(0); |
| 371 | } else if (node->ns->href != NULL) { |
| 372 | if (step->value2 == NULL) |
| 373 | return(0); |
| 374 | if (!xmlStrEqual(step->value2, node->ns->href)) |
| 375 | return(0); |
| 376 | } |
| 377 | continue; |
| 378 | case XML_OP_CHILD: { |
| 379 | xmlNodePtr lst; |
| 380 | |
| 381 | if ((node->type != XML_ELEMENT_NODE) && |
| 382 | (node->type != XML_DOCUMENT_NODE) && |
| 383 | #ifdef LIBXML_DOCB_ENABLED |
| 384 | (node->type != XML_DOCB_DOCUMENT_NODE) && |
| 385 | #endif |
| 386 | (node->type != XML_HTML_DOCUMENT_NODE)) |
| 387 | return(0); |
| 388 | |
| 389 | lst = node->children; |
| 390 | |
| 391 | if (step->value != NULL) { |
| 392 | while (lst != NULL) { |
| 393 | if ((lst->type == XML_ELEMENT_NODE) && |
| 394 | (step->value[0] == lst->name[0]) && |
| 395 | (xmlStrEqual(step->value, lst->name))) |
| 396 | break; |
| 397 | lst = lst->next; |
| 398 | } |
| 399 | if (lst != NULL) |
| 400 | continue; |
| 401 | } |
| 402 | return(0); |
| 403 | } |
| 404 | case XML_OP_ATTR: |
| 405 | if (node->type != XML_ATTRIBUTE_NODE) |
| 406 | return(0); |
| 407 | if (step->value != NULL) { |
| 408 | if (step->value[0] != node->name[0]) |
| 409 | return(0); |
| 410 | if (!xmlStrEqual(step->value, node->name)) |
| 411 | return(0); |
| 412 | } |
| 413 | /* Namespace test */ |
| 414 | if (node->ns == NULL) { |
| 415 | if (step->value2 != NULL) |
| 416 | return(0); |
| 417 | } else if (step->value2 != NULL) { |
| 418 | if (!xmlStrEqual(step->value2, node->ns->href)) |
| 419 | return(0); |
| 420 | } |
| 421 | continue; |
| 422 | case XML_OP_PARENT: |
| 423 | if ((node->type == XML_DOCUMENT_NODE) || |
| 424 | (node->type == XML_HTML_DOCUMENT_NODE) || |
| 425 | #ifdef LIBXML_DOCB_ENABLED |
| 426 | (node->type == XML_DOCB_DOCUMENT_NODE) || |
| 427 | #endif |
| 428 | (node->type == XML_NAMESPACE_DECL)) |
| 429 | return(0); |
| 430 | node = node->parent; |
| 431 | if (node == NULL) |
| 432 | return(0); |
| 433 | if (step->value == NULL) |
| 434 | continue; |
| 435 | if (step->value[0] != node->name[0]) |
| 436 | return(0); |
| 437 | if (!xmlStrEqual(step->value, node->name)) |
| 438 | return(0); |
| 439 | /* Namespace test */ |
| 440 | if (node->ns == NULL) { |
| 441 | if (step->value2 != NULL) |
| 442 | return(0); |
| 443 | } else if (node->ns->href != NULL) { |
| 444 | if (step->value2 == NULL) |
| 445 | return(0); |
| 446 | if (!xmlStrEqual(step->value2, node->ns->href)) |
| 447 | return(0); |
| 448 | } |
| 449 | continue; |
| 450 | case XML_OP_ANCESTOR: |
| 451 | /* TODO: implement coalescing of ANCESTOR/NODE ops */ |
| 452 | if (step->value == NULL) { |
| 453 | i++; |
| 454 | step = &comp->steps[i]; |
| 455 | if (step->op == XML_OP_ROOT) |
| 456 | return(1); |
| 457 | if (step->op != XML_OP_ELEM) |
| 458 | return(0); |
| 459 | if (step->value == NULL) |
| 460 | return(-1); |
| 461 | } |
| 462 | if (node == NULL) |
| 463 | return(0); |
| 464 | if ((node->type == XML_DOCUMENT_NODE) || |
| 465 | (node->type == XML_HTML_DOCUMENT_NODE) || |
| 466 | #ifdef LIBXML_DOCB_ENABLED |
| 467 | (node->type == XML_DOCB_DOCUMENT_NODE) || |
| 468 | #endif |
| 469 | (node->type == XML_NAMESPACE_DECL)) |
| 470 | return(0); |
| 471 | node = node->parent; |
| 472 | while (node != NULL) { |
| 473 | if (node == NULL) |
| 474 | return(0); |
| 475 | if ((node->type == XML_ELEMENT_NODE) && |
| 476 | (step->value[0] == node->name[0]) && |
| 477 | (xmlStrEqual(step->value, node->name))) { |
| 478 | /* Namespace test */ |
| 479 | if (node->ns == NULL) { |
| 480 | if (step->value2 == NULL) |
| 481 | break; |
| 482 | } else if (node->ns->href != NULL) { |
| 483 | if ((step->value2 != NULL) && |
| 484 | (xmlStrEqual(step->value2, node->ns->href))) |
| 485 | break; |
| 486 | } |
| 487 | } |
| 488 | node = node->parent; |
| 489 | } |
| 490 | if (node == NULL) |
| 491 | return(0); |
| 492 | continue; |
| 493 | case XML_OP_NS: |
| 494 | if (node->type != XML_ELEMENT_NODE) |
| 495 | return(0); |
| 496 | if (node->ns == NULL) { |
| 497 | if (step->value != NULL) |
| 498 | return(0); |
| 499 | } else if (node->ns->href != NULL) { |
| 500 | if (step->value == NULL) |
| 501 | return(0); |
| 502 | if (!xmlStrEqual(step->value, node->ns->href)) |
| 503 | return(0); |
| 504 | } |
| 505 | break; |
| 506 | case XML_OP_ALL: |
| 507 | if (node->type != XML_ELEMENT_NODE) |
| 508 | return(0); |
| 509 | break; |
| 510 | } |
| 511 | } |
| 512 | return(1); |
| 513 | } |
| 514 | |
| 515 | /************************************************************************ |
| 516 | * * |
| 517 | * Dedicated parser for templates * |
| 518 | * * |
| 519 | ************************************************************************/ |
| 520 | |
| 521 | #define TODO \ |
| 522 | xmlGenericError(xmlGenericErrorContext, \ |
| 523 | "Unimplemented block at %s:%d\n", \ |
| 524 | __FILE__, __LINE__); |
| 525 | #define CUR (*ctxt->cur) |
| 526 | #define SKIP(val) ctxt->cur += (val) |
| 527 | #define NXT(val) ctxt->cur[(val)] |
| 528 | #define CUR_PTR ctxt->cur |
| 529 | |
| 530 | #define SKIP_BLANKS \ |
Daniel Veillard | 427174f | 2003-12-10 10:42:59 +0000 | [diff] [blame] | 531 | while (IS_BLANK_CH(CUR)) NEXT |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 532 | |
| 533 | #define CURRENT (*ctxt->cur) |
| 534 | #define NEXT ((*ctxt->cur) ? ctxt->cur++: ctxt->cur) |
| 535 | |
| 536 | |
| 537 | #define PUSH(op, val, val2) \ |
| 538 | if (xmlPatternAdd(ctxt, ctxt->comp, (op), (val), (val2))) goto error; |
| 539 | |
| 540 | #define XSLT_ERROR(X) \ |
| 541 | { xsltError(ctxt, __FILE__, __LINE__, X); \ |
| 542 | ctxt->error = (X); return; } |
| 543 | |
| 544 | #define XSLT_ERROR0(X) \ |
| 545 | { xsltError(ctxt, __FILE__, __LINE__, X); \ |
| 546 | ctxt->error = (X); return(0); } |
| 547 | |
| 548 | #if 0 |
| 549 | /** |
| 550 | * xmlPatScanLiteral: |
| 551 | * @ctxt: the XPath Parser context |
| 552 | * |
| 553 | * Parse an XPath Litteral: |
| 554 | * |
| 555 | * [29] Literal ::= '"' [^"]* '"' |
| 556 | * | "'" [^']* "'" |
| 557 | * |
| 558 | * Returns the Literal parsed or NULL |
| 559 | */ |
| 560 | |
| 561 | static xmlChar * |
| 562 | xmlPatScanLiteral(xmlPatParserContextPtr ctxt) { |
| 563 | const xmlChar *q, *cur; |
| 564 | xmlChar *ret = NULL; |
| 565 | int val, len; |
| 566 | |
| 567 | SKIP_BLANKS; |
| 568 | if (CUR == '"') { |
| 569 | NEXT; |
| 570 | cur = q = CUR_PTR; |
| 571 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 572 | while ((IS_CHAR(val)) && (val != '"')) { |
| 573 | cur += len; |
| 574 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 575 | } |
| 576 | if (!IS_CHAR(val)) { |
| 577 | ctxt->error = 1; |
| 578 | return(NULL); |
| 579 | } else { |
| 580 | ret = xmlStrndup(q, cur - q); |
| 581 | } |
| 582 | cur += len; |
| 583 | CUR_PTR = cur; |
| 584 | } else if (CUR == '\'') { |
| 585 | NEXT; |
| 586 | cur = q = CUR_PTR; |
| 587 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 588 | while ((IS_CHAR(val)) && (val != '\'')) { |
| 589 | cur += len; |
| 590 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 591 | } |
| 592 | if (!IS_CHAR(val)) { |
| 593 | ctxt->error = 1; |
| 594 | return(NULL); |
| 595 | } else { |
| 596 | ret = xmlStrndup(q, cur - q); |
| 597 | } |
| 598 | cur += len; |
| 599 | CUR_PTR = cur; |
| 600 | } else { |
| 601 | /* XP_ERROR(XPATH_START_LITERAL_ERROR); */ |
| 602 | ctxt->error = 1; |
| 603 | return(NULL); |
| 604 | } |
| 605 | return(ret); |
| 606 | } |
| 607 | #endif |
| 608 | |
| 609 | /** |
| 610 | * xmlPatScanName: |
| 611 | * @ctxt: the XPath Parser context |
| 612 | * |
| 613 | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | |
| 614 | * CombiningChar | Extender |
| 615 | * |
| 616 | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
| 617 | * |
| 618 | * [6] Names ::= Name (S Name)* |
| 619 | * |
| 620 | * Returns the Name parsed or NULL |
| 621 | */ |
| 622 | |
| 623 | static xmlChar * |
| 624 | xmlPatScanName(xmlPatParserContextPtr ctxt) { |
| 625 | const xmlChar *q, *cur; |
| 626 | xmlChar *ret = NULL; |
| 627 | int val, len; |
| 628 | |
| 629 | SKIP_BLANKS; |
| 630 | |
| 631 | cur = q = CUR_PTR; |
| 632 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 633 | if (!IS_LETTER(val) && (val != '_') && (val != ':')) |
| 634 | return(NULL); |
| 635 | |
| 636 | while ((IS_LETTER(val)) || (IS_DIGIT(val)) || |
| 637 | (val == '.') || (val == '-') || |
| 638 | (val == '_') || |
| 639 | (IS_COMBINING(val)) || |
| 640 | (IS_EXTENDER(val))) { |
| 641 | cur += len; |
| 642 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 643 | } |
| 644 | ret = xmlStrndup(q, cur - q); |
| 645 | CUR_PTR = cur; |
| 646 | return(ret); |
| 647 | } |
| 648 | |
| 649 | /** |
| 650 | * xmlPatScanNCName: |
| 651 | * @ctxt: the XPath Parser context |
| 652 | * |
| 653 | * Parses a non qualified name |
| 654 | * |
| 655 | * Returns the Name parsed or NULL |
| 656 | */ |
| 657 | |
| 658 | static xmlChar * |
| 659 | xmlPatScanNCName(xmlPatParserContextPtr ctxt) { |
| 660 | const xmlChar *q, *cur; |
| 661 | xmlChar *ret = NULL; |
| 662 | int val, len; |
| 663 | |
| 664 | SKIP_BLANKS; |
| 665 | |
| 666 | cur = q = CUR_PTR; |
| 667 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 668 | if (!IS_LETTER(val) && (val != '_')) |
| 669 | return(NULL); |
| 670 | |
| 671 | while ((IS_LETTER(val)) || (IS_DIGIT(val)) || |
| 672 | (val == '.') || (val == '-') || |
| 673 | (val == '_') || |
| 674 | (IS_COMBINING(val)) || |
| 675 | (IS_EXTENDER(val))) { |
| 676 | cur += len; |
| 677 | val = xmlStringCurrentChar(NULL, cur, &len); |
| 678 | } |
| 679 | ret = xmlStrndup(q, cur - q); |
| 680 | CUR_PTR = cur; |
| 681 | return(ret); |
| 682 | } |
| 683 | |
| 684 | #if 0 |
| 685 | /** |
| 686 | * xmlPatScanQName: |
| 687 | * @ctxt: the XPath Parser context |
| 688 | * @prefix: the place to store the prefix |
| 689 | * |
| 690 | * Parse a qualified name |
| 691 | * |
| 692 | * Returns the Name parsed or NULL |
| 693 | */ |
| 694 | |
| 695 | static xmlChar * |
| 696 | xmlPatScanQName(xmlPatParserContextPtr ctxt, xmlChar **prefix) { |
| 697 | xmlChar *ret = NULL; |
| 698 | |
| 699 | *prefix = NULL; |
| 700 | ret = xmlPatScanNCName(ctxt); |
| 701 | if (CUR == ':') { |
| 702 | *prefix = ret; |
| 703 | NEXT; |
| 704 | ret = xmlPatScanNCName(ctxt); |
| 705 | } |
| 706 | return(ret); |
| 707 | } |
| 708 | #endif |
| 709 | |
| 710 | /** |
| 711 | * xmlCompileStepPattern: |
| 712 | * @ctxt: the compilation context |
| 713 | * |
| 714 | * Compile the Step Pattern and generates a precompiled |
| 715 | * form suitable for fast matching. |
| 716 | * |
| 717 | * [3] Step ::= '.' | NameTest |
| 718 | * [4] NameTest ::= QName | '*' | NCName ':' '*' |
| 719 | */ |
| 720 | |
| 721 | static void |
| 722 | xmlCompileStepPattern(xmlPatParserContextPtr ctxt) { |
| 723 | xmlChar *token = NULL; |
| 724 | xmlChar *name = NULL; |
| 725 | const xmlChar *URI = NULL; |
| 726 | xmlChar *URL = NULL; |
| 727 | |
| 728 | SKIP_BLANKS; |
| 729 | if (CUR == '.') { |
| 730 | NEXT; |
| 731 | PUSH(XML_OP_ELEM, NULL, NULL); |
| 732 | return; |
| 733 | } |
| 734 | name = xmlPatScanNCName(ctxt); |
| 735 | if (name == NULL) { |
| 736 | if (CUR == '*') { |
| 737 | NEXT; |
| 738 | PUSH(XML_OP_ALL, NULL, NULL); |
| 739 | return; |
| 740 | } else { |
| 741 | ERROR(NULL, NULL, NULL, |
| 742 | "xmlCompileStepPattern : Name expected\n"); |
| 743 | ctxt->error = 1; |
| 744 | return; |
| 745 | } |
| 746 | } |
| 747 | SKIP_BLANKS; |
| 748 | if (CUR == ':') { |
| 749 | NEXT; |
| 750 | if (CUR != ':') { |
| 751 | xmlChar *prefix = name; |
| 752 | xmlNsPtr ns; |
| 753 | |
| 754 | /* |
| 755 | * This is a namespace match |
| 756 | */ |
| 757 | token = xmlPatScanName(ctxt); |
| 758 | ns = xmlSearchNs(NULL, ctxt->elem, prefix); |
| 759 | if (ns == NULL) { |
| 760 | ERROR5(NULL, NULL, NULL, |
| 761 | "xmlCompileStepPattern : no namespace bound to prefix %s\n", |
| 762 | prefix); |
| 763 | ctxt->error = 1; |
| 764 | goto error; |
| 765 | } else { |
| 766 | URL = xmlStrdup(ns->href); |
| 767 | } |
| 768 | xmlFree(prefix); |
| 769 | if (token == NULL) { |
| 770 | if (CUR == '*') { |
| 771 | NEXT; |
| 772 | PUSH(XML_OP_NS, URL, NULL); |
| 773 | } else { |
| 774 | ERROR(NULL, NULL, NULL, |
| 775 | "xmlCompileStepPattern : Name expected\n"); |
| 776 | ctxt->error = 1; |
| 777 | goto error; |
| 778 | } |
| 779 | } else { |
| 780 | PUSH(XML_OP_ELEM, token, URL); |
| 781 | } |
| 782 | } else { |
| 783 | NEXT; |
| 784 | if (xmlStrEqual(token, (const xmlChar *) "child")) { |
| 785 | xmlFree(token); |
| 786 | token = xmlPatScanName(ctxt); |
| 787 | if (token == NULL) { |
| 788 | if (CUR == '*') { |
| 789 | NEXT; |
| 790 | PUSH(XML_OP_ALL, token, NULL); |
| 791 | return; |
| 792 | } else { |
| 793 | ERROR(NULL, NULL, NULL, |
| 794 | "xmlCompileStepPattern : QName expected\n"); |
| 795 | ctxt->error = 1; |
| 796 | goto error; |
| 797 | } |
| 798 | } |
| 799 | TODO |
| 800 | /* URI = xsltGetQNameURI(ctxt->elem, &token); */ |
| 801 | if (token == NULL) { |
| 802 | ctxt->error = 1; |
| 803 | goto error; |
| 804 | } else { |
| 805 | name = xmlStrdup(token); |
| 806 | if (URI != NULL) |
| 807 | URL = xmlStrdup(URI); |
| 808 | } |
| 809 | PUSH(XML_OP_CHILD, name, URL); |
| 810 | } else if (xmlStrEqual(token, (const xmlChar *) "attribute")) { |
| 811 | xmlFree(token); |
| 812 | token = xmlPatScanName(ctxt); |
| 813 | if (token == NULL) { |
| 814 | ERROR(NULL, NULL, NULL, |
| 815 | "xmlCompileStepPattern : QName expected\n"); |
| 816 | ctxt->error = 1; |
| 817 | goto error; |
| 818 | } |
| 819 | TODO |
| 820 | /* URI = xsltGetQNameURI(ctxt->elem, &token); */ |
| 821 | if (token == NULL) { |
| 822 | ctxt->error = 1; |
| 823 | goto error; |
| 824 | } else { |
| 825 | name = xmlStrdup(token); |
| 826 | if (URI != NULL) |
| 827 | URL = xmlStrdup(URI); |
| 828 | } |
| 829 | PUSH(XML_OP_ATTR, name, URL); |
| 830 | } else { |
| 831 | ERROR(NULL, NULL, NULL, |
| 832 | "xmlCompileStepPattern : 'child' or 'attribute' expected\n"); |
| 833 | ctxt->error = 1; |
| 834 | goto error; |
| 835 | } |
| 836 | xmlFree(token); |
| 837 | } |
| 838 | } else if (CUR == '*') { |
| 839 | NEXT; |
| 840 | PUSH(XML_OP_ALL, token, NULL); |
| 841 | } else { |
| 842 | if (name == NULL) { |
| 843 | ctxt->error = 1; |
| 844 | goto error; |
| 845 | } |
| 846 | PUSH(XML_OP_ELEM, name, NULL); |
| 847 | } |
| 848 | return; |
| 849 | error: |
| 850 | if (token != NULL) |
| 851 | xmlFree(token); |
| 852 | if (name != NULL) |
| 853 | xmlFree(name); |
| 854 | } |
| 855 | |
| 856 | /** |
| 857 | * xmlCompilePathPattern: |
| 858 | * @ctxt: the compilation context |
| 859 | * |
| 860 | * Compile the Path Pattern and generates a precompiled |
| 861 | * form suitable for fast matching. |
| 862 | * |
| 863 | * [5] Path ::= ('.//')? ( Step '/' )* ( Step | '@' NameTest ) |
| 864 | */ |
| 865 | static void |
| 866 | xmlCompilePathPattern(xmlPatParserContextPtr ctxt) { |
| 867 | SKIP_BLANKS; |
| 868 | if ((CUR == '/') && (NXT(1) == '/')) { |
| 869 | /* |
| 870 | * since we reverse the query |
| 871 | * a leading // can be safely ignored |
| 872 | */ |
| 873 | NEXT; |
| 874 | NEXT; |
| 875 | } else if ((CUR == '.') && (NXT(1) == '/') && (NXT(2) == '/')) { |
| 876 | /* |
| 877 | * a leading .// can be safely ignored |
| 878 | */ |
| 879 | NEXT; |
| 880 | NEXT; |
| 881 | NEXT; |
| 882 | } |
| 883 | if (CUR == '@') { |
| 884 | TODO |
| 885 | } else { |
| 886 | xmlCompileStepPattern(ctxt); |
| 887 | SKIP_BLANKS; |
| 888 | while (CUR == '/') { |
| 889 | if ((CUR == '/') && (NXT(1) == '/')) { |
| 890 | PUSH(XML_OP_ANCESTOR, NULL, NULL); |
| 891 | NEXT; |
| 892 | NEXT; |
| 893 | SKIP_BLANKS; |
| 894 | xmlCompileStepPattern(ctxt); |
| 895 | } else { |
| 896 | PUSH(XML_OP_PARENT, NULL, NULL); |
| 897 | NEXT; |
| 898 | SKIP_BLANKS; |
| 899 | if ((CUR != 0) || (CUR == '|')) { |
| 900 | xmlCompileStepPattern(ctxt); |
| 901 | } |
| 902 | } |
| 903 | } |
| 904 | } |
| 905 | error: |
| 906 | return; |
| 907 | } |
| 908 | /************************************************************************ |
| 909 | * * |
| 910 | * The public interfaces * |
| 911 | * * |
| 912 | ************************************************************************/ |
| 913 | |
| 914 | /** |
| 915 | * xmlPatterncompile: |
| 916 | * @pattern: the pattern to compile |
| 917 | * @dict: an optional dictionnary for interned strings |
| 918 | * @flags: compilation flags, undefined yet |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 919 | * @namespaces: the prefix definitions, array of [URI, prefix] or NULL |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 920 | * |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 921 | * Compile a pattern. |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 922 | * |
| 923 | * Returns the compiled for of the pattern or NULL in case of error |
| 924 | */ |
| 925 | xmlPatternPtr |
Daniel Veillard | 427174f | 2003-12-10 10:42:59 +0000 | [diff] [blame] | 926 | xmlPatterncompile(const xmlChar *pattern, xmlDict *dict, |
| 927 | int flags ATTRIBUTE_UNUSED, |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 928 | const xmlChar **namespaces) { |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 929 | xmlPatternPtr ret = NULL; |
| 930 | xmlPatParserContextPtr ctxt = NULL; |
| 931 | |
Daniel Veillard | ffa7b7e | 2003-12-05 16:10:21 +0000 | [diff] [blame] | 932 | ctxt = xmlNewPatParserContext(pattern, dict, namespaces); |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 933 | if (ctxt == NULL) goto error; |
| 934 | ret = xmlNewPattern(); |
| 935 | if (ret == NULL) goto error; |
| 936 | ctxt->comp = ret; |
| 937 | |
| 938 | xmlCompilePathPattern(ctxt); |
| 939 | xmlFreePatParserContext(ctxt); |
| 940 | |
Daniel Veillard | c7c9fb1 | 2005-01-12 21:04:15 +0000 | [diff] [blame] | 941 | if (xmlReversePattern(ret) < 0) |
| 942 | goto error; |
Daniel Veillard | b3de70c | 2003-12-02 22:32:15 +0000 | [diff] [blame] | 943 | return(ret); |
| 944 | error: |
| 945 | if (ctxt != NULL) xmlFreePatParserContext(ctxt); |
| 946 | if (ret != NULL) xmlFreePattern(ret); |
| 947 | return(NULL); |
| 948 | } |
| 949 | |
| 950 | /** |
| 951 | * xmlPatternMatch: |
| 952 | * @comp: the precompiled pattern |
| 953 | * @node: a node |
| 954 | * |
| 955 | * Test wether the node matches the pattern |
| 956 | * |
| 957 | * Returns 1 if it matches, 0 if it doesn't and -1 in case of failure |
| 958 | */ |
| 959 | int |
| 960 | xmlPatternMatch(xmlPatternPtr comp, xmlNodePtr node) |
| 961 | { |
| 962 | if ((comp == NULL) || (node == NULL)) |
| 963 | return(-1); |
| 964 | return(xmlPatMatch(comp, node)); |
| 965 | } |
| 966 | |
| 967 | #endif /* LIBXML_PATTERN_ENABLED */ |