Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 1 | /* |
| 2 | * entities.c : implementation for the XML entities handking |
| 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
| 6 | * $Id$ |
| 7 | */ |
| 8 | |
| 9 | #include <stdio.h> |
Seth Alves | e7f12e6 | 1998-10-01 20:51:15 +0000 | [diff] [blame] | 10 | #include <stdlib.h> |
Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 11 | #include <string.h> |
| 12 | #include "entities.h" |
| 13 | |
| 14 | /* |
| 15 | * The XML predefined entities. |
| 16 | */ |
| 17 | |
| 18 | struct xmlPredefinedEntityValue { |
| 19 | const char *name; |
| 20 | const char *value; |
| 21 | }; |
| 22 | struct xmlPredefinedEntityValue xmlPredefinedEntityValues[] = { |
| 23 | { "lt", "<" }, |
| 24 | { "gt", ">" }, |
| 25 | { "apos", "'" }, |
| 26 | { "quot", "\"" }, |
| 27 | { "amp", "&" } |
| 28 | }; |
| 29 | |
| 30 | xmlEntitiesTablePtr xmlPredefinedEntities = NULL; |
| 31 | |
| 32 | /* |
| 33 | * A buffer used for converting entities to their equivalent and back. |
| 34 | */ |
| 35 | static int buffer_size = 0; |
| 36 | static CHAR *buffer = NULL; |
| 37 | |
| 38 | void growBuffer(void) { |
| 39 | buffer_size *= 2; |
| 40 | buffer = (CHAR *) realloc(buffer, buffer_size * sizeof(CHAR)); |
| 41 | if (buffer == NULL) { |
| 42 | perror("realloc failed"); |
| 43 | exit(1); |
| 44 | } |
| 45 | } |
| 46 | |
| 47 | /* |
| 48 | * xmlFreeEntity : clean-up an entity record. |
| 49 | */ |
| 50 | |
| 51 | void xmlFreeEntity(xmlEntityPtr entity) { |
| 52 | if (entity == NULL) return; |
| 53 | |
| 54 | if (entity->name != NULL) |
| 55 | free((char *) entity->name); |
| 56 | if (entity->ExternalID != NULL) |
| 57 | free((char *) entity->ExternalID); |
| 58 | if (entity->SystemID != NULL) |
| 59 | free((char *) entity->SystemID); |
| 60 | if (entity->content != NULL) |
| 61 | free((char *) entity->content); |
| 62 | memset(entity, -1, sizeof(xmlEntity)); |
| 63 | } |
| 64 | |
| 65 | /* |
| 66 | * xmlAddDocEntity : register a new entity for an entities table. |
| 67 | * |
| 68 | * TODO !!! We should check here that the combination of type |
| 69 | * ExternalID and SystemID is valid. |
| 70 | */ |
| 71 | static void xmlAddEntity(xmlEntitiesTablePtr table, const CHAR *name, int type, |
| 72 | const CHAR *ExternalID, const CHAR *SystemID, CHAR *content) { |
| 73 | int i; |
| 74 | xmlEntityPtr cur; |
| 75 | int len; |
| 76 | |
| 77 | for (i = 0;i < table->nb_entities;i++) { |
| 78 | cur = &table->table[i]; |
| 79 | if (!xmlStrcmp(cur->name, name)) { |
| 80 | /* |
| 81 | * The entity is already defined in this Dtd, the spec says to NOT |
| 82 | * override it ... Is it worth a Warning ??? !!! |
| 83 | */ |
| 84 | return; |
| 85 | } |
| 86 | } |
| 87 | if (table->nb_entities >= table->max_entities) { |
| 88 | /* |
| 89 | * need more elements. |
| 90 | */ |
| 91 | table->max_entities *= 2; |
| 92 | table->table = (xmlEntityPtr) |
| 93 | realloc(table->table, table->max_entities * sizeof(xmlEntity)); |
| 94 | if (table->table) { |
| 95 | perror("realloc failed"); |
| 96 | exit(1); |
| 97 | } |
| 98 | } |
| 99 | cur = &table->table[table->nb_entities]; |
| 100 | cur->name = xmlStrdup(name); |
| 101 | for (len = 0;name[0] != 0;name++)len++; |
| 102 | cur->len = len; |
| 103 | cur->type = type; |
| 104 | if (ExternalID != NULL) |
| 105 | cur->ExternalID = xmlStrdup(ExternalID); |
| 106 | else |
| 107 | cur->ExternalID = NULL; |
| 108 | if (SystemID != NULL) |
| 109 | cur->SystemID = xmlStrdup(SystemID); |
| 110 | else |
| 111 | cur->SystemID = NULL; |
| 112 | if (content != NULL) |
| 113 | cur->content = xmlStrdup(content); |
| 114 | else |
| 115 | cur->content = NULL; |
| 116 | table->nb_entities++; |
| 117 | } |
| 118 | |
| 119 | /* |
| 120 | * Set up xmlPredefinedEntities from xmlPredefinedEntityValues. |
| 121 | */ |
| 122 | void xmlInitializePredefinedEntities(void) { |
| 123 | int i; |
| 124 | CHAR name[50]; |
| 125 | CHAR value[50]; |
| 126 | const char *in; |
| 127 | CHAR *out; |
| 128 | |
| 129 | if (xmlPredefinedEntities != NULL) return; |
| 130 | |
| 131 | xmlPredefinedEntities = xmlCreateEntitiesTable(); |
| 132 | for (i = 0;i < sizeof(xmlPredefinedEntityValues) / |
| 133 | sizeof(xmlPredefinedEntityValues[0]);i++) { |
| 134 | in = xmlPredefinedEntityValues[i].name; |
| 135 | out = &name[0]; |
| 136 | for (;(*out++ = (CHAR) *in);)in++; |
| 137 | in = xmlPredefinedEntityValues[i].value; |
| 138 | out = &value[0]; |
| 139 | for (;(*out++ = (CHAR) *in);)in++; |
| 140 | xmlAddEntity(xmlPredefinedEntities, (const CHAR *) &name[0], |
Daniel Veillard | 25940b7 | 1998-10-29 05:51:30 +0000 | [diff] [blame] | 141 | XML_INTERNAL_PREDEFINED_ENTITY, NULL, NULL, |
Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 142 | &value[0]); |
| 143 | } |
| 144 | } |
| 145 | |
Daniel Veillard | ccb0963 | 1998-10-27 06:21:04 +0000 | [diff] [blame] | 146 | /** |
| 147 | * xmlGetPredefinedEntity: |
| 148 | * @name: the entity name |
| 149 | * |
| 150 | * Check whether this name is an predefined entity. |
| 151 | * |
| 152 | * return values: NULL if not, othervise the entity |
| 153 | */ |
| 154 | xmlEntityPtr |
| 155 | xmlGetPredefinedEntity(const CHAR *name) { |
| 156 | int i; |
| 157 | xmlEntityPtr cur; |
| 158 | |
| 159 | if (xmlPredefinedEntities == NULL) |
| 160 | xmlInitializePredefinedEntities(); |
| 161 | for (i = 0;i < xmlPredefinedEntities->nb_entities;i++) { |
| 162 | cur = &xmlPredefinedEntities->table[i]; |
| 163 | if (!xmlStrcmp(cur->name, name)) return(cur); |
| 164 | } |
| 165 | return(NULL); |
| 166 | } |
| 167 | |
| 168 | |
Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 169 | |
| 170 | /* |
| 171 | * xmlAddDtdEntity : register a new entity for this DTD. |
| 172 | */ |
| 173 | void xmlAddDtdEntity(xmlDocPtr doc, const CHAR *name, int type, |
| 174 | const CHAR *ExternalID, const CHAR *SystemID, CHAR *content) { |
| 175 | xmlEntitiesTablePtr table; |
| 176 | |
| 177 | if (doc->dtd == NULL) { |
| 178 | fprintf(stderr, "xmlAddDtdEntity: document without Dtd !\n"); |
| 179 | return; |
| 180 | } |
| 181 | table = (xmlEntitiesTablePtr) doc->dtd->entities; |
| 182 | if (table == NULL) { |
| 183 | table = xmlCreateEntitiesTable(); |
| 184 | doc->dtd->entities = table; |
| 185 | } |
| 186 | xmlAddEntity(table, name, type, ExternalID, SystemID, content); |
| 187 | } |
| 188 | |
| 189 | /* |
| 190 | * xmlAddDocEntity : register a new entity for this document. |
| 191 | */ |
| 192 | void xmlAddDocEntity(xmlDocPtr doc, const CHAR *name, int type, |
| 193 | const CHAR *ExternalID, const CHAR *SystemID, CHAR *content) { |
| 194 | xmlEntitiesTablePtr table; |
| 195 | |
| 196 | table = (xmlEntitiesTablePtr) doc->entities; |
| 197 | if (table == NULL) { |
| 198 | table = xmlCreateEntitiesTable(); |
| 199 | doc->entities = table; |
| 200 | } |
| 201 | xmlAddEntity(doc->entities, name, type, ExternalID, SystemID, content); |
| 202 | } |
| 203 | |
| 204 | /* |
| 205 | * xmlGetDtdEntity : do an entity lookup in the Dtd entity hash table and |
| 206 | * returns the corrsponding entity, if found, NULL otherwise. |
| 207 | */ |
| 208 | xmlEntityPtr xmlGetDtdEntity(xmlDocPtr doc, const CHAR *name) { |
| 209 | int i; |
| 210 | xmlEntityPtr cur; |
| 211 | xmlEntitiesTablePtr table; |
| 212 | |
| 213 | if ((doc->dtd != NULL) && (doc->dtd->entities != NULL)) { |
| 214 | table = (xmlEntitiesTablePtr) doc->dtd->entities; |
| 215 | for (i = 0;i < table->nb_entities;i++) { |
| 216 | cur = &table->table[i]; |
| 217 | if (!xmlStrcmp(cur->name, name)) return(cur); |
| 218 | } |
| 219 | } |
| 220 | return(NULL); |
| 221 | } |
| 222 | |
| 223 | /* |
| 224 | * xmlGetDocEntity : do an entity lookup in the document entity hash table and |
| 225 | * returns the corrsponding entity, otherwise a lookup is done |
| 226 | * in the predefined entities too. |
| 227 | */ |
| 228 | xmlEntityPtr xmlGetDocEntity(xmlDocPtr doc, const CHAR *name) { |
| 229 | int i; |
| 230 | xmlEntityPtr cur; |
| 231 | xmlEntitiesTablePtr table; |
| 232 | |
| 233 | if (doc->entities != NULL) { |
| 234 | table = (xmlEntitiesTablePtr) doc->entities; |
| 235 | for (i = 0;i < table->nb_entities;i++) { |
| 236 | cur = &table->table[i]; |
| 237 | if (!xmlStrcmp(cur->name, name)) return(cur); |
| 238 | } |
| 239 | } |
| 240 | if (xmlPredefinedEntities == NULL) |
| 241 | xmlInitializePredefinedEntities(); |
| 242 | table = xmlPredefinedEntities; |
| 243 | for (i = 0;i < table->nb_entities;i++) { |
| 244 | cur = &table->table[i]; |
| 245 | if (!xmlStrcmp(cur->name, name)) return(cur); |
| 246 | } |
| 247 | |
| 248 | return(NULL); |
| 249 | } |
| 250 | |
| 251 | /* |
| 252 | * xmlEncodeEntities : do a global encoding of a string, replacing the |
Daniel Veillard | 0ba4d53 | 1998-11-01 19:34:31 +0000 | [diff] [blame] | 253 | * predefined entities and non ASCII values with their |
| 254 | * entities and CharRef counterparts. |
| 255 | * TODO !!!! Once moved to UTF-8 internal encoding, the encoding of non-ascii |
| 256 | * get erroneous. |
Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 257 | */ |
| 258 | CHAR *xmlEncodeEntities(xmlDocPtr doc, const CHAR *input) { |
| 259 | const CHAR *cur = input; |
| 260 | CHAR *out = buffer; |
| 261 | |
| 262 | if (buffer == NULL) { |
| 263 | buffer_size = 1000; |
| 264 | buffer = (CHAR *) malloc(buffer_size * sizeof(CHAR)); |
| 265 | if (buffer == NULL) { |
| 266 | perror("malloc failed"); |
| 267 | exit(1); |
| 268 | } |
| 269 | out = buffer; |
| 270 | } |
| 271 | while (*cur != '\0') { |
| 272 | if (out - buffer > buffer_size - 100) { |
| 273 | int index = out - buffer; |
| 274 | |
| 275 | growBuffer(); |
| 276 | out = &buffer[index]; |
| 277 | } |
| 278 | |
| 279 | /* |
| 280 | * By default one have to encode at least '<', '>', '"' and '&' ! |
Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 281 | */ |
| 282 | if (*cur == '<') { |
| 283 | *out++ = '&'; |
| 284 | *out++ = 'l'; |
| 285 | *out++ = 't'; |
| 286 | *out++ = ';'; |
| 287 | } else if (*cur == '>') { |
| 288 | *out++ = '&'; |
| 289 | *out++ = 'g'; |
| 290 | *out++ = 't'; |
| 291 | *out++ = ';'; |
| 292 | } else if (*cur == '&') { |
| 293 | *out++ = '&'; |
| 294 | *out++ = 'a'; |
| 295 | *out++ = 'm'; |
| 296 | *out++ = 'p'; |
| 297 | *out++ = ';'; |
| 298 | } else if (*cur == '"') { |
| 299 | *out++ = '&'; |
| 300 | *out++ = 'q'; |
| 301 | *out++ = 'u'; |
| 302 | *out++ = 'o'; |
| 303 | *out++ = 't'; |
| 304 | *out++ = ';'; |
| 305 | } else if (*cur == '\'') { |
| 306 | *out++ = '&'; |
| 307 | *out++ = 'a'; |
| 308 | *out++ = 'p'; |
| 309 | *out++ = 'o'; |
| 310 | *out++ = 's'; |
| 311 | *out++ = ';'; |
Daniel Veillard | 0ba4d53 | 1998-11-01 19:34:31 +0000 | [diff] [blame] | 312 | #ifndef USE_UTF_8 |
| 313 | } else if ((sizeof(CHAR) == 1) && (*cur >= 0x80)) { |
| 314 | char buf[10], *ptr; |
Daniel Veillard | da4d3c4 | 1998-11-04 20:07:05 +0000 | [diff] [blame^] | 315 | #ifdef HAVE_SNPRINTF |
| 316 | snprintf(buf, 9, "&#%d;", *cur); |
| 317 | #else |
| 318 | sprintf(buf, "&#%d;", *cur); |
| 319 | #endif |
Daniel Veillard | 0ba4d53 | 1998-11-01 19:34:31 +0000 | [diff] [blame] | 320 | ptr = buf; |
| 321 | while (*ptr != 0) *out++ = *ptr++; |
| 322 | #endif |
Daniel Veillard | 260a68f | 1998-08-13 03:39:55 +0000 | [diff] [blame] | 323 | } else { |
| 324 | /* |
| 325 | * default case, just copy ! |
| 326 | */ |
| 327 | *out++ = *cur; |
| 328 | } |
| 329 | cur++; |
| 330 | } |
| 331 | *out++ = 0; |
| 332 | return(buffer); |
| 333 | } |
| 334 | |
| 335 | /* |
| 336 | * xmlCreateEntitiesTable : create and initialize an enmpty hash table |
| 337 | */ |
| 338 | xmlEntitiesTablePtr xmlCreateEntitiesTable(void) { |
| 339 | xmlEntitiesTablePtr ret; |
| 340 | |
| 341 | ret = (xmlEntitiesTablePtr) |
| 342 | malloc(sizeof(xmlEntitiesTable)); |
| 343 | if (ret == NULL) { |
| 344 | fprintf(stderr, "xmlCreateEntitiesTable : malloc(%d) failed\n", |
| 345 | sizeof(xmlEntitiesTable)); |
| 346 | return(NULL); |
| 347 | } |
| 348 | ret->max_entities = XML_MIN_ENTITIES_TABLE; |
| 349 | ret->nb_entities = 0; |
| 350 | ret->table = (xmlEntityPtr ) |
| 351 | malloc(ret->max_entities * sizeof(xmlEntity)); |
| 352 | if (ret == NULL) { |
| 353 | fprintf(stderr, "xmlCreateEntitiesTable : malloc(%d) failed\n", |
| 354 | ret->max_entities * sizeof(xmlEntity)); |
| 355 | free(ret); |
| 356 | return(NULL); |
| 357 | } |
| 358 | return(ret); |
| 359 | } |
| 360 | |
| 361 | /* |
| 362 | * xmlFreeEntitiesTable : clean up and free an entities hash table. |
| 363 | */ |
| 364 | void xmlFreeEntitiesTable(xmlEntitiesTablePtr table) { |
| 365 | int i; |
| 366 | |
| 367 | if (table == NULL) return; |
| 368 | |
| 369 | for (i = 0;i < table->nb_entities;i++) { |
| 370 | xmlFreeEntity(&table->table[i]); |
| 371 | } |
| 372 | free(table->table); |
| 373 | free(table); |
| 374 | } |
| 375 | |
| 376 | /* |
| 377 | * Dump the content of an entity table to the document output. |
| 378 | */ |
| 379 | void xmlDumpEntitiesTable(xmlEntitiesTablePtr table) { |
| 380 | int i; |
| 381 | xmlEntityPtr cur; |
| 382 | |
| 383 | if (table == NULL) return; |
| 384 | |
| 385 | for (i = 0;i < table->nb_entities;i++) { |
| 386 | cur = &table->table[i]; |
| 387 | switch (cur->type) { |
| 388 | case XML_INTERNAL_GENERAL_ENTITY: |
| 389 | xmlBufferWriteChar("<!ENTITY "); |
| 390 | xmlBufferWriteCHAR(cur->name); |
| 391 | xmlBufferWriteChar(" \""); |
| 392 | xmlBufferWriteCHAR(cur->content); |
| 393 | xmlBufferWriteChar("\">\n"); |
| 394 | break; |
| 395 | case XML_EXTERNAL_GENERAL_PARSED_ENTITY: |
| 396 | xmlBufferWriteChar("<!ENTITY "); |
| 397 | xmlBufferWriteCHAR(cur->name); |
| 398 | if (cur->ExternalID != NULL) { |
| 399 | xmlBufferWriteChar(" PUBLIC \""); |
| 400 | xmlBufferWriteCHAR(cur->ExternalID); |
| 401 | xmlBufferWriteChar("\" \""); |
| 402 | xmlBufferWriteCHAR(cur->SystemID); |
| 403 | xmlBufferWriteChar("\""); |
| 404 | } else { |
| 405 | xmlBufferWriteChar(" SYSTEM \""); |
| 406 | xmlBufferWriteCHAR(cur->SystemID); |
| 407 | xmlBufferWriteChar("\""); |
| 408 | } |
| 409 | xmlBufferWriteChar(">\n"); |
| 410 | break; |
| 411 | case XML_EXTERNAL_GENERAL_UNPARSED_ENTITY: |
| 412 | xmlBufferWriteChar("<!ENTITY "); |
| 413 | xmlBufferWriteCHAR(cur->name); |
| 414 | if (cur->ExternalID != NULL) { |
| 415 | xmlBufferWriteChar(" PUBLIC \""); |
| 416 | xmlBufferWriteCHAR(cur->ExternalID); |
| 417 | xmlBufferWriteChar("\" \""); |
| 418 | xmlBufferWriteCHAR(cur->SystemID); |
| 419 | xmlBufferWriteChar("\""); |
| 420 | } else { |
| 421 | xmlBufferWriteChar(" SYSTEM \""); |
| 422 | xmlBufferWriteCHAR(cur->SystemID); |
| 423 | xmlBufferWriteChar("\""); |
| 424 | } |
| 425 | if (cur->content != NULL) { /* Should be true ! */ |
| 426 | xmlBufferWriteChar(" NDATA "); |
| 427 | xmlBufferWriteCHAR(cur->content); |
| 428 | } |
| 429 | xmlBufferWriteChar(">\n"); |
| 430 | break; |
| 431 | case XML_INTERNAL_PARAMETER_ENTITY: |
| 432 | xmlBufferWriteChar("<!ENTITY % "); |
| 433 | xmlBufferWriteCHAR(cur->name); |
| 434 | xmlBufferWriteChar(" \""); |
| 435 | xmlBufferWriteCHAR(cur->content); |
| 436 | xmlBufferWriteChar("\">\n"); |
| 437 | break; |
| 438 | case XML_EXTERNAL_PARAMETER_ENTITY: |
| 439 | xmlBufferWriteChar("<!ENTITY % "); |
| 440 | xmlBufferWriteCHAR(cur->name); |
| 441 | if (cur->ExternalID != NULL) { |
| 442 | xmlBufferWriteChar(" PUBLIC \""); |
| 443 | xmlBufferWriteCHAR(cur->ExternalID); |
| 444 | xmlBufferWriteChar("\" \""); |
| 445 | xmlBufferWriteCHAR(cur->SystemID); |
| 446 | xmlBufferWriteChar("\""); |
| 447 | } else { |
| 448 | xmlBufferWriteChar(" SYSTEM \""); |
| 449 | xmlBufferWriteCHAR(cur->SystemID); |
| 450 | xmlBufferWriteChar("\""); |
| 451 | } |
| 452 | xmlBufferWriteChar(">\n"); |
| 453 | break; |
| 454 | default: |
| 455 | fprintf(stderr, |
| 456 | "xmlDumpEntitiesTable: internal: unknown type %d\n", |
| 457 | cur->type); |
| 458 | } |
| 459 | } |
| 460 | } |