Daniel Veillard | 01791d5 | 1998-07-24 19:24:09 +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> |
| 10 | #include <malloc.h> |
| 11 | #include <strings.h> |
| 12 | #include "xml_entities.h" |
| 13 | |
| 14 | /* |
| 15 | * A buffer used for converting entities to their equivalent and back. |
| 16 | */ |
| 17 | static CHAR *buffer = NULL; |
| 18 | static int buffer_size = 0; |
| 19 | |
| 20 | void growBuffer(void) { |
| 21 | buffer_size *= 2; |
| 22 | buffer = (CHAR *) realloc(buffer, buffer_size * sizeof(CHAR)); |
| 23 | if (buffer == NULL) { |
| 24 | perror("realloc failed"); |
| 25 | exit(1); |
| 26 | } |
| 27 | } |
| 28 | |
| 29 | /* |
| 30 | * xmlFreeEntity : clean-up an entity record. |
| 31 | */ |
| 32 | |
| 33 | void xmlFreeEntity(xmlEntityPtr entity) { |
| 34 | if (entity == NULL) return; |
| 35 | |
| 36 | if (entity->value != NULL) free(entity->value); |
| 37 | entity->value = NULL; |
| 38 | if (entity->id != NULL) |
| 39 | free((char *) entity->id); |
| 40 | } |
| 41 | |
| 42 | /* |
| 43 | * xmlAddDocEntity : register a new entity for an entities table. |
| 44 | */ |
| 45 | static void xmlAddEntity(xmlEntitiesTablePtr table, CHAR *value, |
| 46 | const CHAR *id) { |
| 47 | int i; |
| 48 | xmlEntityPtr cur; |
| 49 | |
| 50 | for (i = 0;i < table->nb_entities;i++) { |
| 51 | cur = &table->table[i]; |
| 52 | if (!xmlStrcmp(cur->id, id)) { |
| 53 | free(cur->value); |
| 54 | cur->value = xmlStrdup(value); |
| 55 | } |
| 56 | } |
| 57 | if (table->nb_entities >= table->max_entities) { |
| 58 | /* |
| 59 | * need more elements. |
| 60 | */ |
| 61 | table->max_entities *= 2; |
| 62 | table->table = (xmlEntityPtr) |
| 63 | realloc(table->table, table->max_entities * sizeof(xmlEntity)); |
| 64 | if (table->table) { |
| 65 | perror("realloc failed"); |
| 66 | exit(1); |
| 67 | } |
| 68 | } |
| 69 | cur = &table->table[table->nb_entities]; |
| 70 | cur->value = xmlStrdup(value); |
| 71 | cur->id = xmlStrdup(id); |
| 72 | table->nb_entities++; |
| 73 | } |
| 74 | |
| 75 | |
| 76 | /* |
| 77 | * xmlAddDtdEntity : register a new entity for this document. |
| 78 | */ |
| 79 | void xmlAddDtdEntity(xmlDtdPtr dtd, CHAR *value, const CHAR *id) { |
| 80 | xmlEntitiesTablePtr table; |
| 81 | |
| 82 | table = (xmlEntitiesTablePtr) dtd->entities; |
| 83 | if (table == NULL) { |
| 84 | table = xmlCreateEntitiesTable(); |
| 85 | dtd->entities = table; |
| 86 | } |
| 87 | xmlAddEntity(table, value, id); |
| 88 | } |
| 89 | |
| 90 | /* |
| 91 | * xmlAddDocEntity : register a new entity for this document. |
| 92 | */ |
| 93 | void xmlAddDocEntity(xmlDocPtr doc, CHAR *value, const CHAR *id) { |
| 94 | xmlEntitiesTablePtr table; |
| 95 | |
| 96 | table = (xmlEntitiesTablePtr) doc->entities; |
| 97 | if (table == NULL) { |
| 98 | table = xmlCreateEntitiesTable(); |
| 99 | doc->entities = table; |
| 100 | } |
| 101 | xmlAddEntity(table, value, id); |
| 102 | } |
| 103 | |
| 104 | /* |
| 105 | * xmlGetEntity : do an entity lookup in the hash table and |
| 106 | * returns the corrsponding CHAR *, if found, zero otherwise. |
| 107 | */ |
| 108 | CHAR *xmlGetEntity(xmlDocPtr doc, const CHAR *id) { |
| 109 | int i; |
| 110 | xmlEntityPtr cur; |
| 111 | xmlEntitiesTablePtr table; |
| 112 | |
| 113 | if (doc->entities == NULL) return(0); |
| 114 | table = (xmlEntitiesTablePtr) doc->entities; |
| 115 | for (i = 0;i < table->nb_entities;i++) { |
| 116 | cur = &table->table[i]; |
| 117 | if (!xmlStrcmp(cur->id, id)) return(cur->value); |
| 118 | } |
| 119 | return(NULL); |
| 120 | } |
| 121 | |
| 122 | /* |
| 123 | * xmlReadEntities : read an entity. |
| 124 | */ |
| 125 | const CHAR *xmlReadEntity(xmlDocPtr doc, const CHAR **input) { |
| 126 | static CHAR *entity = NULL; |
| 127 | static int entity_size = 100; |
| 128 | const CHAR *cur = *input; |
| 129 | |
| 130 | if (entity == NULL) { |
| 131 | entity = (CHAR *) malloc(entity_size * sizeof(CHAR)); |
| 132 | if (entity == NULL) { |
| 133 | fprintf(stderr, "xmlReadEntity : cannot allocate %d bytes\n", |
| 134 | entity_size * sizeof(CHAR)); |
| 135 | return(NULL); |
| 136 | } |
| 137 | } |
| 138 | if (*cur == '&') { |
| 139 | cur++; |
| 140 | if (*cur == '#') { |
| 141 | /* TODO !!!! |
| 142 | fprintf(stderr, "Character reference not yet implemented\n"); */ |
| 143 | } else { |
| 144 | /* TODO !!!! |
| 145 | fprintf(stderr, "Entity search not yet implemented\n"); */ |
| 146 | } |
| 147 | } |
| 148 | |
| 149 | /* |
| 150 | * The few predefined entities. |
| 151 | */ |
| 152 | if ((cur[0] == 'a') && (cur[1] == 'm') && (cur[2] == 'p') && |
| 153 | (cur[3] == ';')) { |
| 154 | entity[0] = '%'; |
| 155 | entity[1] = 0; |
| 156 | cur += 3; |
| 157 | *input = cur; |
| 158 | return(entity); |
| 159 | } else if ((cur[0] == 'q') && (cur[1] == 'u') && (cur[2] == 'o') && |
| 160 | (cur[3] == 't') && (cur[4] == ';')) { |
| 161 | entity[0] = '"'; |
| 162 | entity[1] = 0; |
| 163 | cur += 4; |
| 164 | *input = cur; |
| 165 | return(entity); |
| 166 | } else if ((cur[0] == 'a') && (cur[1] == 'p') && (cur[2] == 'o') && |
| 167 | (cur[3] == 's') && (cur[4] == ';')) { |
| 168 | entity[0] = '\''; |
| 169 | entity[1] = 0; |
| 170 | cur += 4; |
| 171 | *input = cur; |
| 172 | return(entity); |
| 173 | } else if ((cur[0] == 'l') && (cur[1] == 't') && (cur[2] == ';')) { |
| 174 | entity[0] = '<'; |
| 175 | entity[1] = 0; |
| 176 | cur += 2; |
| 177 | *input = cur; |
| 178 | return(entity); |
| 179 | } else if ((cur[0] == 'g') && (cur[1] == 't') && (cur[2] == ';')) { |
| 180 | entity[0] = '>'; |
| 181 | entity[1] = 0; |
| 182 | cur += 2; |
| 183 | *input = cur; |
| 184 | return(entity); |
| 185 | } |
| 186 | |
| 187 | return(NULL); |
| 188 | } |
| 189 | |
| 190 | /* |
| 191 | * xmlDecodeEntities : do a global entities lookup on a input string |
| 192 | * and returns a duplicate after the entities substitution. |
| 193 | */ |
| 194 | CHAR *xmlDecodeEntities(xmlDocPtr doc, const CHAR *input, int len) { |
| 195 | const CHAR *cur = input; |
| 196 | CHAR *out = buffer; |
| 197 | int i; |
| 198 | |
| 199 | if (buffer == NULL) { |
| 200 | buffer_size = 1000; |
| 201 | buffer = (CHAR *) malloc(buffer_size * sizeof(CHAR)); |
| 202 | if (buffer == NULL) { |
| 203 | perror("malloc failed"); |
| 204 | exit(1); |
| 205 | } |
| 206 | out = buffer; |
| 207 | } |
| 208 | for (i = 0;(*cur != 0) && (cur - input < len);cur++) { |
| 209 | if (*cur == '&') { |
| 210 | const CHAR *entity = xmlReadEntity(doc, &cur); |
| 211 | if (entity != NULL) |
| 212 | while (*entity != 0) { |
| 213 | *out++ = *entity++; |
| 214 | i++; |
| 215 | if (i + 10 > buffer_size) { |
| 216 | int index = out - buffer; |
| 217 | |
| 218 | growBuffer(); |
| 219 | out = &buffer[index]; |
| 220 | } |
| 221 | } |
| 222 | } else if (*cur == '%') { |
| 223 | /* TODO !!!!! |
| 224 | fprintf(stderr, " \n"); */ |
| 225 | } else { |
| 226 | *out++ = *cur; |
| 227 | i++; |
| 228 | } |
| 229 | |
| 230 | if (i + 10 > buffer_size) { |
| 231 | int index = out - buffer; |
| 232 | |
| 233 | growBuffer(); |
| 234 | out = &buffer[index]; |
| 235 | } |
| 236 | } |
| 237 | *out++ = 0; |
| 238 | return(buffer); |
| 239 | } |
| 240 | |
| 241 | /* |
| 242 | * xmlEncodeEntities : do a global encoding of a string, replacing the |
| 243 | * basic values with their entities form. |
| 244 | */ |
| 245 | CHAR *xmlEncodeEntities(xmlDocPtr doc, const CHAR *input) { |
| 246 | const CHAR *cur = input; |
| 247 | CHAR *out = buffer; |
| 248 | |
| 249 | if (buffer == NULL) { |
| 250 | buffer_size = 1000; |
| 251 | buffer = (CHAR *) malloc(buffer_size * sizeof(CHAR)); |
| 252 | if (buffer == NULL) { |
| 253 | perror("malloc failed"); |
| 254 | exit(1); |
| 255 | } |
| 256 | out = buffer; |
| 257 | } |
| 258 | while (*cur != '\0') { |
| 259 | if (out - buffer > buffer_size - 100) { |
| 260 | int index = out - buffer; |
| 261 | |
| 262 | growBuffer(); |
| 263 | out = &buffer[index]; |
| 264 | } |
| 265 | |
| 266 | /* |
| 267 | * By default one have to encode at least '<', '>', '"' and '&' ! |
| 268 | * One could try a better encoding using the entities defined and |
| 269 | * used as a compression code !!!. |
| 270 | */ |
| 271 | if (*cur == '<') { |
| 272 | *out++ = '&'; |
| 273 | *out++ = 'l'; |
| 274 | *out++ = 't'; |
| 275 | *out++ = ';'; |
| 276 | } else if (*cur == '>') { |
| 277 | *out++ = '&'; |
| 278 | *out++ = 'g'; |
| 279 | *out++ = 't'; |
| 280 | *out++ = ';'; |
| 281 | } else if (*cur == '&') { |
| 282 | *out++ = '&'; |
| 283 | *out++ = 'a'; |
| 284 | *out++ = 'm'; |
| 285 | *out++ = 'p'; |
| 286 | *out++ = ';'; |
| 287 | } else if (*cur == '"') { |
| 288 | *out++ = '&'; |
| 289 | *out++ = 'q'; |
| 290 | *out++ = 'u'; |
| 291 | *out++ = 'o'; |
| 292 | *out++ = 't'; |
| 293 | *out++ = ';'; |
| 294 | } else if (*cur == '\'') { |
| 295 | *out++ = '&'; |
| 296 | *out++ = 'a'; |
| 297 | *out++ = 'p'; |
| 298 | *out++ = 'o'; |
| 299 | *out++ = 's'; |
| 300 | *out++ = ';'; |
| 301 | } else { |
| 302 | /* |
| 303 | * default case, just copy ! |
| 304 | */ |
| 305 | *out++ = *cur; |
| 306 | } |
| 307 | cur++; |
| 308 | } |
| 309 | *out++ = 0; |
| 310 | return(buffer); |
| 311 | } |
| 312 | |
| 313 | /* |
| 314 | * xmlCreateEntitiesTable : create and initialize an enmpty hash table |
| 315 | */ |
| 316 | xmlEntitiesTablePtr xmlCreateEntitiesTable(void) { |
| 317 | xmlEntitiesTablePtr ret; |
| 318 | |
| 319 | ret = (xmlEntitiesTablePtr) |
| 320 | malloc(sizeof(xmlEntitiesTable)); |
| 321 | if (ret == NULL) { |
| 322 | fprintf(stderr, "xmlCreateEntitiesTable : malloc(%d) failed\n", |
| 323 | sizeof(xmlEntitiesTable)); |
| 324 | return(NULL); |
| 325 | } |
| 326 | ret->max_entities = XML_MIN_ENTITIES_TABLE; |
| 327 | ret->nb_entities = 0; |
| 328 | ret->table = (xmlEntityPtr ) |
| 329 | malloc(ret->max_entities * sizeof(xmlEntity)); |
| 330 | if (ret == NULL) { |
| 331 | fprintf(stderr, "xmlCreateEntitiesTable : malloc(%d) failed\n", |
| 332 | ret->max_entities * sizeof(xmlEntity)); |
| 333 | free(ret); |
| 334 | return(NULL); |
| 335 | } |
| 336 | return(ret); |
| 337 | } |
| 338 | |
| 339 | /* |
| 340 | * xmlFreeEntitiesTable : clean up and free an entities hash table. |
| 341 | */ |
| 342 | void xmlFreeEntitiesTable(xmlEntitiesTablePtr table) { |
| 343 | int i; |
| 344 | |
| 345 | if (table == NULL) return; |
| 346 | |
| 347 | for (i = 0;i < table->nb_entities;i++) { |
| 348 | xmlFreeEntity(&table->table[i]); |
| 349 | } |
| 350 | free(table->table); |
| 351 | free(table); |
| 352 | } |
| 353 | |