Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 1 | /* |
Michael Clark | 837240f | 2007-03-13 08:26:25 +0000 | [diff] [blame^] | 2 | * $Id: json_tokener.c,v 1.19 2006/01/30 23:07:57 mclark Exp $ |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 3 | * |
Michael Clark | f6a6e48 | 2007-03-13 08:26:23 +0000 | [diff] [blame] | 4 | * Copyright (c) 2004, 2005 Metaparadigm Pte. Ltd. |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 5 | * Michael Clark <michael@metaparadigm.com> |
| 6 | * |
Michael Clark | f6a6e48 | 2007-03-13 08:26:23 +0000 | [diff] [blame] | 7 | * This library is free software; you can redistribute it and/or modify |
| 8 | * it under the terms of the MIT license. See COPYING for details. |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 9 | * |
| 10 | */ |
| 11 | |
Michael Clark | 4504df7 | 2007-03-13 08:26:20 +0000 | [diff] [blame] | 12 | #include "config.h" |
| 13 | |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 14 | #include <stdio.h> |
| 15 | #include <stdlib.h> |
| 16 | #include <ctype.h> |
| 17 | #include <string.h> |
| 18 | |
| 19 | #include "bits.h" |
| 20 | #include "debug.h" |
| 21 | #include "printbuf.h" |
| 22 | #include "arraylist.h" |
| 23 | #include "json_object.h" |
| 24 | #include "json_tokener.h" |
| 25 | |
Michael Clark | 837240f | 2007-03-13 08:26:25 +0000 | [diff] [blame^] | 26 | #if !HAVE_STRNCASECMP && defined(_MSC_VER) |
| 27 | /* MSC has the version as _strnicmp */ |
| 28 | # define strncasecmp _strnicmp |
| 29 | #elif !HAVE_STRNCASECMP |
| 30 | # error You do not have strncasecmp on your system. |
| 31 | #endif /* HAVE_STRNCASECMP */ |
| 32 | |
| 33 | |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 34 | static struct json_object* json_tokener_do_parse(struct json_tokener *this); |
| 35 | |
| 36 | struct json_object* json_tokener_parse(char * s) |
| 37 | { |
| 38 | struct json_tokener tok; |
| 39 | struct json_object* obj; |
| 40 | |
| 41 | tok.source = s; |
| 42 | tok.pos = 0; |
| 43 | tok.pb = printbuf_new(); |
| 44 | obj = json_tokener_do_parse(&tok); |
| 45 | printbuf_free(tok.pb); |
| 46 | return obj; |
| 47 | } |
| 48 | |
Michael Clark | 4504df7 | 2007-03-13 08:26:20 +0000 | [diff] [blame] | 49 | #if !HAVE_STRNDUP |
| 50 | /* CAW: compliant version of strndup() */ |
| 51 | char* strndup(const char* str, size_t n) |
| 52 | { |
| 53 | if(str) { |
| 54 | size_t len = strlen(str); |
| 55 | size_t nn = min(len,n); |
| 56 | char* s = (char*)malloc(sizeof(char) * (nn + 1)); |
| 57 | |
| 58 | if(s) { |
| 59 | memcpy(s, str, nn); |
| 60 | s[nn] = '\0'; |
| 61 | } |
| 62 | |
| 63 | return s; |
| 64 | } |
| 65 | |
| 66 | return NULL; |
| 67 | } |
| 68 | #endif |
| 69 | |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 70 | static struct json_object* json_tokener_do_parse(struct json_tokener *this) |
| 71 | { |
| 72 | enum json_tokener_state state, saved_state; |
| 73 | enum json_tokener_error err = json_tokener_success; |
| 74 | struct json_object *current = NULL, *obj; |
| 75 | char *obj_field_name = NULL; |
| 76 | char quote_char; |
| 77 | int deemed_double, start_offset; |
Michael Clark | 4504df7 | 2007-03-13 08:26:20 +0000 | [diff] [blame] | 78 | char c; |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 79 | |
| 80 | state = json_tokener_state_eatws; |
| 81 | saved_state = json_tokener_state_start; |
| 82 | |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 83 | do { |
| 84 | c = this->source[this->pos]; |
| 85 | switch(state) { |
| 86 | |
| 87 | case json_tokener_state_eatws: |
| 88 | if(isspace(c)) { |
| 89 | this->pos++; |
| 90 | } else if(c == '/') { |
| 91 | state = json_tokener_state_comment_start; |
| 92 | start_offset = this->pos++; |
| 93 | } else { |
| 94 | state = saved_state; |
| 95 | } |
| 96 | break; |
| 97 | |
| 98 | case json_tokener_state_start: |
| 99 | switch(c) { |
| 100 | case '{': |
| 101 | state = json_tokener_state_eatws; |
| 102 | saved_state = json_tokener_state_object; |
| 103 | current = json_object_new_object(); |
| 104 | this->pos++; |
| 105 | break; |
| 106 | case '[': |
| 107 | state = json_tokener_state_eatws; |
| 108 | saved_state = json_tokener_state_array; |
| 109 | current = json_object_new_array(); |
| 110 | this->pos++; |
| 111 | break; |
| 112 | case 'N': |
| 113 | case 'n': |
| 114 | state = json_tokener_state_null; |
| 115 | start_offset = this->pos++; |
| 116 | break; |
| 117 | case '"': |
| 118 | case '\'': |
| 119 | quote_char = c; |
| 120 | printbuf_reset(this->pb); |
| 121 | state = json_tokener_state_string; |
| 122 | start_offset = ++this->pos; |
| 123 | break; |
| 124 | case 'T': |
| 125 | case 't': |
| 126 | case 'F': |
| 127 | case 'f': |
| 128 | state = json_tokener_state_boolean; |
| 129 | start_offset = this->pos++; |
| 130 | break; |
Michael Clark | 4504df7 | 2007-03-13 08:26:20 +0000 | [diff] [blame] | 131 | #if defined(__GNUC__) |
| 132 | case '0' ... '9': |
| 133 | #else |
| 134 | case '0': |
| 135 | case '1': |
| 136 | case '2': |
| 137 | case '3': |
| 138 | case '4': |
| 139 | case '5': |
| 140 | case '6': |
| 141 | case '7': |
| 142 | case '8': |
| 143 | case '9': |
| 144 | #endif |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 145 | case '-': |
| 146 | deemed_double = 0; |
| 147 | state = json_tokener_state_number; |
| 148 | start_offset = this->pos++; |
| 149 | break; |
| 150 | default: |
| 151 | err = json_tokener_error_parse_unexpected; |
| 152 | goto out; |
| 153 | } |
| 154 | break; |
| 155 | |
| 156 | case json_tokener_state_finish: |
| 157 | goto out; |
| 158 | |
| 159 | case json_tokener_state_null: |
| 160 | if(strncasecmp("null", this->source + start_offset, |
| 161 | this->pos - start_offset)) |
| 162 | return error_ptr(-json_tokener_error_parse_null); |
| 163 | if(this->pos - start_offset == 4) { |
| 164 | current = NULL; |
| 165 | saved_state = json_tokener_state_finish; |
| 166 | state = json_tokener_state_eatws; |
| 167 | } else { |
| 168 | this->pos++; |
| 169 | } |
| 170 | break; |
| 171 | |
| 172 | case json_tokener_state_comment_start: |
| 173 | if(c == '*') { |
| 174 | state = json_tokener_state_comment; |
| 175 | } else if(c == '/') { |
| 176 | state = json_tokener_state_comment_eol; |
| 177 | } else { |
| 178 | err = json_tokener_error_parse_comment; |
| 179 | goto out; |
| 180 | } |
| 181 | this->pos++; |
| 182 | break; |
| 183 | |
| 184 | case json_tokener_state_comment: |
| 185 | if(c == '*') state = json_tokener_state_comment_end; |
| 186 | this->pos++; |
| 187 | break; |
| 188 | |
| 189 | case json_tokener_state_comment_eol: |
| 190 | if(c == '\n') { |
| 191 | if(mc_get_debug()) { |
| 192 | char *tmp = strndup(this->source + start_offset, |
| 193 | this->pos - start_offset); |
| 194 | mc_debug("json_tokener_comment: %s\n", tmp); |
| 195 | free(tmp); |
| 196 | } |
| 197 | state = json_tokener_state_eatws; |
| 198 | } |
| 199 | this->pos++; |
| 200 | break; |
| 201 | |
| 202 | case json_tokener_state_comment_end: |
| 203 | if(c == '/') { |
| 204 | if(mc_get_debug()) { |
| 205 | char *tmp = strndup(this->source + start_offset, |
| 206 | this->pos - start_offset + 1); |
| 207 | mc_debug("json_tokener_comment: %s\n", tmp); |
| 208 | free(tmp); |
| 209 | } |
| 210 | state = json_tokener_state_eatws; |
| 211 | } else { |
| 212 | state = json_tokener_state_comment; |
| 213 | } |
| 214 | this->pos++; |
| 215 | break; |
| 216 | |
| 217 | case json_tokener_state_string: |
| 218 | if(c == quote_char) { |
| 219 | printbuf_memappend(this->pb, this->source + start_offset, |
| 220 | this->pos - start_offset); |
| 221 | current = json_object_new_string(this->pb->buf); |
| 222 | saved_state = json_tokener_state_finish; |
| 223 | state = json_tokener_state_eatws; |
| 224 | } else if(c == '\\') { |
| 225 | saved_state = json_tokener_state_string; |
| 226 | state = json_tokener_state_string_escape; |
| 227 | } |
| 228 | this->pos++; |
| 229 | break; |
| 230 | |
| 231 | case json_tokener_state_string_escape: |
| 232 | switch(c) { |
| 233 | case '"': |
| 234 | case '\\': |
| 235 | printbuf_memappend(this->pb, this->source + start_offset, |
| 236 | this->pos - start_offset - 1); |
| 237 | start_offset = this->pos++; |
| 238 | state = saved_state; |
| 239 | break; |
| 240 | case 'b': |
| 241 | case 'n': |
| 242 | case 'r': |
| 243 | case 't': |
| 244 | printbuf_memappend(this->pb, this->source + start_offset, |
| 245 | this->pos - start_offset - 1); |
| 246 | if(c == 'b') printbuf_memappend(this->pb, "\b", 1); |
| 247 | else if(c == 'n') printbuf_memappend(this->pb, "\n", 1); |
| 248 | else if(c == 'r') printbuf_memappend(this->pb, "\r", 1); |
| 249 | else if(c == 't') printbuf_memappend(this->pb, "\t", 1); |
| 250 | start_offset = ++this->pos; |
| 251 | state = saved_state; |
| 252 | break; |
| 253 | case 'u': |
| 254 | printbuf_memappend(this->pb, this->source + start_offset, |
| 255 | this->pos - start_offset - 1); |
| 256 | start_offset = ++this->pos; |
| 257 | state = json_tokener_state_escape_unicode; |
| 258 | break; |
| 259 | default: |
| 260 | err = json_tokener_error_parse_string; |
| 261 | goto out; |
| 262 | } |
| 263 | break; |
| 264 | |
| 265 | case json_tokener_state_escape_unicode: |
| 266 | if(strchr(json_hex_chars, c)) { |
| 267 | this->pos++; |
| 268 | if(this->pos - start_offset == 4) { |
| 269 | unsigned char utf_out[3]; |
| 270 | unsigned int ucs_char = |
| 271 | (hexdigit(*(this->source + start_offset)) << 12) + |
| 272 | (hexdigit(*(this->source + start_offset + 1)) << 8) + |
| 273 | (hexdigit(*(this->source + start_offset + 2)) << 4) + |
| 274 | hexdigit(*(this->source + start_offset + 3)); |
| 275 | if (ucs_char < 0x80) { |
| 276 | utf_out[0] = ucs_char; |
Michael Clark | 7b899b6 | 2007-03-13 08:26:21 +0000 | [diff] [blame] | 277 | printbuf_memappend(this->pb, (char*)utf_out, 1); |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 278 | } else if (ucs_char < 0x800) { |
| 279 | utf_out[0] = 0xc0 | (ucs_char >> 6); |
| 280 | utf_out[1] = 0x80 | (ucs_char & 0x3f); |
Michael Clark | 7b899b6 | 2007-03-13 08:26:21 +0000 | [diff] [blame] | 281 | printbuf_memappend(this->pb, (char*)utf_out, 2); |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 282 | } else { |
| 283 | utf_out[0] = 0xe0 | (ucs_char >> 12); |
| 284 | utf_out[1] = 0x80 | ((ucs_char >> 6) & 0x3f); |
| 285 | utf_out[2] = 0x80 | (ucs_char & 0x3f); |
Michael Clark | 7b899b6 | 2007-03-13 08:26:21 +0000 | [diff] [blame] | 286 | printbuf_memappend(this->pb, (char*)utf_out, 3); |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 287 | } |
| 288 | start_offset = this->pos; |
| 289 | state = saved_state; |
| 290 | } |
| 291 | } else { |
| 292 | err = json_tokener_error_parse_string; |
| 293 | goto out; |
| 294 | } |
| 295 | break; |
| 296 | |
| 297 | case json_tokener_state_boolean: |
| 298 | if(strncasecmp("true", this->source + start_offset, |
| 299 | this->pos - start_offset) == 0) { |
| 300 | if(this->pos - start_offset == 4) { |
| 301 | current = json_object_new_boolean(1); |
| 302 | saved_state = json_tokener_state_finish; |
| 303 | state = json_tokener_state_eatws; |
| 304 | } else { |
| 305 | this->pos++; |
| 306 | } |
| 307 | } else if(strncasecmp("false", this->source + start_offset, |
| 308 | this->pos - start_offset) == 0) { |
| 309 | if(this->pos - start_offset == 5) { |
| 310 | current = json_object_new_boolean(0); |
| 311 | saved_state = json_tokener_state_finish; |
| 312 | state = json_tokener_state_eatws; |
| 313 | } else { |
| 314 | this->pos++; |
| 315 | } |
| 316 | } else { |
| 317 | err = json_tokener_error_parse_boolean; |
| 318 | goto out; |
| 319 | } |
| 320 | break; |
| 321 | |
| 322 | case json_tokener_state_number: |
| 323 | if(!c || !strchr(json_number_chars, c)) { |
| 324 | int numi; |
| 325 | double numd; |
| 326 | char *tmp = strndup(this->source + start_offset, |
| 327 | this->pos - start_offset); |
| 328 | if(!deemed_double && sscanf(tmp, "%d", &numi) == 1) { |
| 329 | current = json_object_new_int(numi); |
| 330 | } else if(deemed_double && sscanf(tmp, "%lf", &numd) == 1) { |
| 331 | current = json_object_new_double(numd); |
| 332 | } else { |
| 333 | free(tmp); |
| 334 | err = json_tokener_error_parse_number; |
| 335 | goto out; |
| 336 | } |
| 337 | free(tmp); |
| 338 | saved_state = json_tokener_state_finish; |
| 339 | state = json_tokener_state_eatws; |
| 340 | } else { |
| 341 | if(c == '.' || c == 'e') deemed_double = 1; |
| 342 | this->pos++; |
| 343 | } |
| 344 | break; |
| 345 | |
| 346 | case json_tokener_state_array: |
| 347 | if(c == ']') { |
| 348 | this->pos++; |
| 349 | saved_state = json_tokener_state_finish; |
| 350 | state = json_tokener_state_eatws; |
| 351 | } else { |
| 352 | obj = json_tokener_do_parse(this); |
| 353 | if(is_error(obj)) { |
Michael Clark | 0370baa | 2007-03-13 08:26:22 +0000 | [diff] [blame] | 354 | err = -(enum json_tokener_error)obj; |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 355 | goto out; |
| 356 | } |
| 357 | json_object_array_add(current, obj); |
| 358 | saved_state = json_tokener_state_array_sep; |
| 359 | state = json_tokener_state_eatws; |
| 360 | } |
| 361 | break; |
| 362 | |
| 363 | case json_tokener_state_array_sep: |
| 364 | if(c == ']') { |
| 365 | this->pos++; |
| 366 | saved_state = json_tokener_state_finish; |
| 367 | state = json_tokener_state_eatws; |
| 368 | } else if(c == ',') { |
| 369 | this->pos++; |
| 370 | saved_state = json_tokener_state_array; |
| 371 | state = json_tokener_state_eatws; |
| 372 | } else { |
| 373 | json_object_put(current); |
| 374 | return error_ptr(-json_tokener_error_parse_array); |
| 375 | } |
| 376 | break; |
| 377 | |
| 378 | case json_tokener_state_object: |
| 379 | state = json_tokener_state_object_field_start; |
| 380 | start_offset = this->pos; |
| 381 | break; |
| 382 | |
| 383 | case json_tokener_state_object_field_start: |
| 384 | if(c == '}') { |
| 385 | this->pos++; |
| 386 | saved_state = json_tokener_state_finish; |
| 387 | state = json_tokener_state_eatws; |
| 388 | } else if (c == '"' || c == '\'') { |
| 389 | quote_char = c; |
| 390 | printbuf_reset(this->pb); |
| 391 | state = json_tokener_state_object_field; |
| 392 | start_offset = ++this->pos; |
Michael Clark | 0370baa | 2007-03-13 08:26:22 +0000 | [diff] [blame] | 393 | } else { |
| 394 | err = json_tokener_error_parse_object; |
| 395 | goto out; |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 396 | } |
| 397 | break; |
| 398 | |
| 399 | case json_tokener_state_object_field: |
| 400 | if(c == quote_char) { |
| 401 | printbuf_memappend(this->pb, this->source + start_offset, |
| 402 | this->pos - start_offset); |
| 403 | obj_field_name = strdup(this->pb->buf); |
| 404 | saved_state = json_tokener_state_object_field_end; |
| 405 | state = json_tokener_state_eatws; |
| 406 | } else if(c == '\\') { |
| 407 | saved_state = json_tokener_state_object_field; |
| 408 | state = json_tokener_state_string_escape; |
| 409 | } |
| 410 | this->pos++; |
| 411 | break; |
| 412 | |
| 413 | case json_tokener_state_object_field_end: |
| 414 | if(c == ':') { |
| 415 | this->pos++; |
| 416 | saved_state = json_tokener_state_object_value; |
| 417 | state = json_tokener_state_eatws; |
| 418 | } else { |
| 419 | return error_ptr(-json_tokener_error_parse_object); |
| 420 | } |
| 421 | break; |
| 422 | |
| 423 | case json_tokener_state_object_value: |
| 424 | obj = json_tokener_do_parse(this); |
| 425 | if(is_error(obj)) { |
Michael Clark | 0370baa | 2007-03-13 08:26:22 +0000 | [diff] [blame] | 426 | err = -(enum json_tokener_error)obj; |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 427 | goto out; |
| 428 | } |
| 429 | json_object_object_add(current, obj_field_name, obj); |
| 430 | free(obj_field_name); |
| 431 | obj_field_name = NULL; |
| 432 | saved_state = json_tokener_state_object_sep; |
| 433 | state = json_tokener_state_eatws; |
| 434 | break; |
| 435 | |
| 436 | case json_tokener_state_object_sep: |
| 437 | if(c == '}') { |
| 438 | this->pos++; |
| 439 | saved_state = json_tokener_state_finish; |
| 440 | state = json_tokener_state_eatws; |
| 441 | } else if(c == ',') { |
| 442 | this->pos++; |
| 443 | saved_state = json_tokener_state_object; |
| 444 | state = json_tokener_state_eatws; |
| 445 | } else { |
| 446 | err = json_tokener_error_parse_object; |
| 447 | goto out; |
| 448 | } |
| 449 | break; |
| 450 | |
| 451 | } |
| 452 | } while(c); |
| 453 | |
| 454 | if(state != json_tokener_state_finish && |
| 455 | saved_state != json_tokener_state_finish) |
| 456 | err = json_tokener_error_parse_eof; |
| 457 | |
| 458 | out: |
| 459 | free(obj_field_name); |
| 460 | if(err == json_tokener_success) return current; |
| 461 | mc_debug("json_tokener_do_parse: error=%d state=%d char=%c\n", |
| 462 | err, state, c); |
| 463 | json_object_put(current); |
Michael Clark | 0370baa | 2007-03-13 08:26:22 +0000 | [diff] [blame] | 464 | return error_ptr(-err); |
Michael Clark | f0d0888 | 2007-03-13 08:26:18 +0000 | [diff] [blame] | 465 | } |