src/objson/json_parser.c

   1 /*
   2 Copyright (C) 2005  Georgia Public Library Service
   3 Bill Erickson <highfalutin@gmail.com>
   4
   5 This program is free software; you can redistribute it and/or
   6 modify it under the terms of the GNU General Public License
   7 as published by the Free Software Foundation; either version 2
   8 of the License, or (at your option) any later version.
   9
  10 This program is distributed in the hope that it will be useful,
  11 but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 GNU General Public License for more details.
  14 */
  15
  16
  17 #include "json_parser.h"
  18
  19 /* keep a copy of the length of the current json string so we don't
  20  * have to calculate it in each function
  21  */
  22 int current_strlen; /* XXX need to move this into the function params for thread support */
  23
  24 object* json_parse_string(char* string) {
  25
  26         if(string == NULL) return NULL;
  27
  28         current_strlen = strlen(string);
  29
  30         if(current_strlen == 0)
  31                 return NULL;
  32
  33         object* obj = new_object(NULL);
  34         unsigned long index = 0;
  35
  36         int status = _json_parse_string(string, &index, obj);
  37         if(!status)
  38                 return obj;
  39
  40         if(status == -2)
  41                 return NULL;
  42
  43         return NULL;
  44 }
  45
  46
  47 int _json_parse_string(char* string, unsigned long* index, object* obj) {
  48         assert(string && index && *index < current_strlen);
  49
  50         int status = 0; /* return code from parsing routines */
  51         char* classname = NULL; /* object class hint */
  52         json_eat_ws(string, index, 1); /* remove leading whitespace */
  53
  54         char c = string[*index];
  55
  56         /* remove any leading comments */
  57         if( c == '/' ) {
  58
  59                 while(1) {
  60                         (*index)++; /* move to second comment char */
  61                         status = json_eat_comment(string, index, &classname, 1);
  62                         if(status) return status;
  63
  64                         json_eat_ws(string, index, 1);
  65                         c = string[*index];
  66                         if(c != '/')
  67                                 break;
  68                 }
  69         }
  70
  71         json_eat_ws(string, index, 1); /* remove leading whitespace */
  72
  73         if(*index >= current_strlen)
  74                 return -2;
  75
  76         switch(c) {
  77
  78                 /* json string */
  79                 case '"':
  80                         (*index)++;
  81                         status = json_parse_json_string(string, index, obj);
  82                         break;
  83
  84                 /* json array */
  85                 case '[':
  86                         (*index)++;
  87                         status = json_parse_json_array(string, index, obj);
  88                         break;
  89
  90                 /* json object */
  91                 case '{':
  92                         (*index)++;
  93                         status = json_parse_json_object(string, index, obj);
  94                         break;
  95
  96                 /* NULL */
  97                 case 'n':
  98                 case 'N':
  99                         status = json_parse_json_null(string, index, obj);
 100                         break;
 101
 102
 103                 /* true, false */
 104                 case 'f':
 105                 case 'F':
 106                 case 't':
 107                 case 'T':
 108                         status = json_parse_json_bool(string, index, obj);
 109                         break;
 110
 111                 default:
 112                         if(is_number(c) || c == '.' || c == '-') { /* are we a number? */
 113                                 status = json_parse_json_number(string, index, obj);
 114                                 if(status) return status;
 115                                 break;
 116                         }
 117
 118                         (*index)--;
 119                         /* we should never get here */
 120                         return json_handle_error(string, index, "_json_parse_string() final switch clause");
 121         }
 122
 123         if(status) return status;
 124
 125         json_eat_ws(string, index, 1);
 126
 127         if( *index < current_strlen ) {
 128                 /* remove any trailing comments */
 129                 c = string[*index];
 130                 if( c == '/' ) {
 131                         (*index)++;
 132                         status = json_eat_comment(string, index, NULL, 0);
 133                         if(status) return status;
 134                 }
 135         }
 136
 137         if(classname){
 138                 obj->set_class(obj, classname);
 139                 free(classname);
 140         }
 141
 142         return 0;
 143 }
 144
 145
 146 int json_parse_json_null(char* string, unsigned long* index, object* obj) {
 147
 148         if(*index >= (current_strlen - 3)) {
 149                 return json_handle_error(string, index,
 150                         "_parse_json_string(): invalid null" );
 151         }
 152
 153         if(!strncasecmp(string + (*index), "null", 4)) {
 154                 (*index) += 4;
 155                 obj->is_null = 1;
 156                 return 0;
 157         } else {
 158                 return json_handle_error(string, index,
 159                         "_parse_json_string(): invalid null" );
 160         }
 161 }
 162
 163 /* should be at the first character of the bool at this point */
 164 int json_parse_json_bool(char* string, unsigned long* index, object* obj) {
 165         assert(string && obj && *index < current_strlen);
 166
 167         char* ret = "json_parse_json_bool(): truncated bool";
 168
 169         if( *index >= (current_strlen - 5))
 170                 return json_handle_error(string, index, ret);
 171
 172         if(!strncasecmp( string + (*index), "false", 5)) {
 173                 (*index) += 5;
 174                 return 0;
 175         }
 176
 177         if( *index >= (current_strlen - 4))
 178                 return json_handle_error(string, index, ret);
 179
 180         if(!strncasecmp( string + (*index), "true", 4)) {
 181                 (*index) += 4;
 182                 return 0;
 183         }
 184
 185         return json_handle_error(string, index, ret);
 186 }
 187
 188
 189 /* expecting the first character of the number */
 190 int json_parse_json_number(char* string, unsigned long* index, object* obj) {
 191         assert(string && obj && *index < current_strlen);
 192
 193         growing_buffer* buf = buffer_init(64);
 194         char c = string[*index];
 195
 196         int done = 0;
 197         int dot_seen = 0;
 198
 199         /* negative number? */
 200         if(c == '-') { buffer_add(buf, "-"); (*index)++; }
 201
 202         while(*index < current_strlen) {
 203
 204                 if(is_number(c))
 205                         buffer_add_char(buf, c);
 206
 207                 else if( c == '.' ) {
 208                         if(dot_seen) {
 209                                 return json_handle_error(string, index,
 210                                         "json_parse_json_number(): malformed json number");
 211                         }
 212                         dot_seen = 1;
 213                         buffer_add_char(buf, c);
 214                 } else {
 215                         done = 1; break;
 216                 }
 217                 (*index)++;
 218                 c = string[*index];
 219                 if(done) break;
 220         }
 221
 222         if(dot_seen) {
 223                 obj->is_double = 1;
 224                 obj->double_value = strtod(buf->buf, NULL);
 225                 buffer_free(buf);
 226                 return 0;
 227
 228         } else {
 229                 obj->is_number = 1;
 230                 obj->num_value = atol(buf->buf);
 231                 buffer_free(buf);
 232                 return 0;
 233         }
 234 }
 235
 236 /* index should point to the character directly following the '['.  when done
 237  * index will point to the character directly following the ']' character
 238  */
 239 int json_parse_json_array(char* string, unsigned long* index, object* obj) {
 240         assert(string && obj && index && *index < current_strlen);
 241
 242         int status;
 243         int in_parse = 0; /* true if this array already contains one item */
 244         obj->is_array = 1;
 245         while(*index < current_strlen) {
 246
 247                 json_eat_ws(string, index, 1);
 248
 249                 if(string[*index] == ']') {
 250                         (*index)++;
 251                         break;
 252                 }
 253
 254                 if(in_parse) {
 255                         json_eat_ws(string, index, 1);
 256                         if(string[*index] != ',') {
 257                                 return json_handle_error(string, index,
 258                                         "json_parse_json_array(): array not followed by a ','");
 259                         }
 260                         (*index)++;
 261                         json_eat_ws(string, index, 1);
 262                 }
 263
 264                 object* item = new_object(NULL);
 265                 status = _json_parse_string(string, index, item);
 266
 267                 if(status) return status;
 268                 obj->push(obj, item);
 269                 in_parse = 1;
 270         }
 271
 272         return 0;
 273 }
 274
 275
 276 /* index should point to the character directly following the '{'.  when done
 277  * index will point to the character directly following the '}'
 278  */
 279 int json_parse_json_object(char* string, unsigned long* index, object* obj) {
 280         assert(string && obj && index && *index < current_strlen);
 281
 282         obj->is_hash = 1;
 283         int status;
 284         int in_parse = 0; /* true if we've already added one item to this object */
 285
 286         while(*index < current_strlen) {
 287
 288                 json_eat_ws(string, index, 1);
 289
 290                 if(string[*index] == '}') {
 291                         (*index)++;
 292                         break;
 293                 }
 294
 295                 if(in_parse) {
 296                         if(string[*index] != ',') {
 297                                 return json_handle_error(string, index,
 298                                         "json_parse_json_object(): object missing ',' betweenn elements" );
 299                         }
 300                         (*index)++;
 301                         json_eat_ws(string, index, 1);
 302                 }
 303
 304                 /* first we grab the hash key */
 305                 object* key_obj = new_object(NULL);
 306                 status = _json_parse_string(string, index, key_obj);
 307                 if(status) return status;
 308
 309                 if(!key_obj->is_string) {
 310                         return json_handle_error(string, index,
 311                                 "_json_parse_json_object(): hash key not a string");
 312                 }
 313
 314                 char* key = key_obj->string_data;
 315
 316                 json_eat_ws(string, index, 1);
 317
 318                 if(string[*index] != ':') {
 319                         return json_handle_error(string, index,
 320                                 "json_parse_json_object(): hash key not followed by ':' character");
 321                 }
 322
 323                 (*index)++;
 324
 325                 /* now grab the value object */
 326                 json_eat_ws(string, index, 1);
 327                 object* value_obj = new_object(NULL);
 328                 status = _json_parse_string(string, index, value_obj);
 329                 if(status) return status;
 330
 331                 /* put the data into the object and continue */
 332                 obj->add_key(obj, key, value_obj);
 333                 free_object(key_obj);
 334                 in_parse = 1;
 335         }
 336         return 0;
 337 }
 338
 339
 340
 341 /* when done, index will point to the character after the closing quote */
 342 int json_parse_json_string(char* string, unsigned long* index, object* obj) {
 343         assert(string && index && *index < current_strlen);
 344
 345         int in_escape = 0;
 346         int done = 0;
 347         growing_buffer* buf = buffer_init(64);
 348
 349         while(*index < current_strlen) {
 350
 351                 char c = string[*index];
 352
 353                 switch(c) {
 354
 355                         case '\\':
 356                                 if(in_escape) {
 357                                         buffer_add(buf, "\\");
 358                                         in_escape = 0;
 359                                 } else
 360                                         in_escape = 1;
 361                                 break;
 362
 363                         case '"':
 364                                 if(in_escape) {
 365                                         buffer_add(buf, "\"");
 366                                         in_escape = 0;
 367                                 } else
 368                                         done = 1;
 369                                 break;
 370
 371                         case 't':
 372                                 if(in_escape) {
 373                                         buffer_add(buf,"\t");
 374                                         in_escape = 0;
 375                                 } else
 376                                         buffer_add_char(buf, c);
 377                                 break;
 378
 379                         case 'b':
 380                                 if(in_escape) {
 381                                         buffer_add(buf,"\b");
 382                                         in_escape = 0;
 383                                 } else
 384                                         buffer_add_char(buf, c);
 385                                 break;
 386
 387                         case 'f':
 388                                 if(in_escape) {
 389                                         buffer_add(buf,"\f");
 390                                         in_escape = 0;
 391                                 } else
 392                                         buffer_add_char(buf, c);
 393                                 break;
 394
 395                         case 'r':
 396                                 if(in_escape) {
 397                                         buffer_add(buf,"\r");
 398                                         in_escape = 0;
 399                                 } else
 400                                         buffer_add_char(buf, c);
 401                                 break;
 402
 403                         case 'n':
 404                                 if(in_escape) {
 405                                         buffer_add(buf,"\n");
 406                                         in_escape = 0;
 407                                 } else
 408                                         buffer_add_char(buf, c);
 409                                 break;
 410
 411                         case 'u':
 412                                 if(in_escape) {
 413                                         (*index)++;
 414
 415                                         if(*index >= (current_strlen - 4)) {
 416                                                 return json_handle_error(string, index,
 417                                                         "json_parse_json_string(): truncated escaped unicode"); }
 418
 419                                         char buff[5];
 420                                         memset(buff,0,5);
 421                                         memcpy(buff, string + (*index), 4);
 422
 423
 424                                         /* ----------------------------------------------------------------------- */
 425                                         /* ----------------------------------------------------------------------- */
 426                                         /* The following chunk was borrowed with permission from
 427                                                 json-c http://oss.metaparadigm.com/json-c/ */
 428                                         unsigned char utf_out[3];
 429                                         memset(utf_out,0,3);
 430
 431                                         #define hexdigit(x) ( ((x) <= '9') ? (x) - '0' : ((x) & 7) + 9)
 432
 433                                         unsigned int ucs_char =
 434                                                 (hexdigit(string[*index] ) << 12) +
 435                                                 (hexdigit(string[*index + 1]) << 8) +
 436                                                 (hexdigit(string[*index + 2]) << 4) +
 437                                                 hexdigit(string[*index + 3]);
 438
 439                                         if (ucs_char < 0x80) {
 440                                                 utf_out[0] = ucs_char;
 441                                                 buffer_add(buf, utf_out);
 442
 443                                         } else if (ucs_char < 0x800) {
 444                                                 utf_out[0] = 0xc0 | (ucs_char >> 6);
 445                                                 utf_out[1] = 0x80 | (ucs_char & 0x3f);
 446                                                 buffer_add(buf, utf_out);
 447
 448                                         } else {
 449                                                 utf_out[0] = 0xe0 | (ucs_char >> 12);
 450                                                 utf_out[1] = 0x80 | ((ucs_char >> 6) & 0x3f);
 451                                                 utf_out[2] = 0x80 | (ucs_char & 0x3f);
 452                                                 buffer_add(buf, utf_out);
 453                                         }
 454                                         /* ----------------------------------------------------------------------- */
 455                                         /* ----------------------------------------------------------------------- */
 456
 457                                         (*index) += 3;
 458                                         in_escape = 0;
 459
 460                                 } else {
 461
 462                                         buffer_add_char(buf, c);
 463                                 }
 464
 465                                 break;
 466
 467                         default:
 468                                 buffer_add_char(buf, c);
 469                 }
 470
 471                 (*index)++;
 472                 if(done) break;
 473         }
 474
 475         obj->set_string(obj, buf->buf);
 476         buffer_free(buf);
 477         return 0;
 478 }
 479
 480
 481 void json_eat_ws(char* string, unsigned long* index, int eat_all) {
 482         assert(string && index);
 483         if(*index >= current_strlen)
 484                 return;
 485
 486         if( eat_all ) { /* removes newlines, etc */
 487                 while(string[*index] == ' '     ||
 488                                 string[*index] == '\n'  ||
 489                                 string[*index] == '\t')
 490                         (*index)++;
 491         }
 492
 493         else
 494                 while(string[*index] == ' ') (*index)++;
 495 }
 496
 497
 498 /* index should be at the '*' character at the beginning of the comment.
 499  * when done, index will point to the first character after the final /
 500  */
 501 int json_eat_comment(char* string, unsigned long* index, char** buffer, int parse_class) {
 502         assert(string && index && *index < current_strlen);
 503
 504         if(string[*index] != '*' && string[*index] != '/' )
 505                 return json_handle_error(string, index,
 506                         "json_eat_comment(): invalid character after /");
 507
 508         /* chop out any // style comments */
 509         if(string[*index] == '/') {
 510                 (*index)++;
 511                 char c = string[*index];
 512                 while(*index < current_strlen) {
 513                         (*index)++;
 514                         if(c == '\n')
 515                                 return 0;
 516                         c = string[*index];
 517                 }
 518                 return 0;
 519         }
 520
 521         (*index)++;
 522
 523         int on_star                     = 0; /* true if we just saw a '*' character */
 524
 525         /* we're just past the '*' */
 526         if(!parse_class) { /* we're not concerned with class hints */
 527                 while(*index < current_strlen) {
 528                         if(string[*index] == '/') {
 529                                 if(on_star) {
 530                                         (*index)++;
 531                                         return 0;
 532                                 }
 533                         }
 534
 535                         if(string[*index] == '*') on_star = 1;
 536                         else on_star = 0;
 537
 538                         (*index)++;
 539                 }
 540                 return 0;
 541         }
 542
 543
 544
 545         growing_buffer* buf = buffer_init(64);
 546
 547         int first_dash          = 0;
 548         int second_dash = 0;
 549         int third_dash          = 0;
 550         int fourth_dash = 0;
 551
 552         int in_hint                     = 0;
 553         int done                                = 0;
 554
 555         /*--S hint--*/   /* <-- Hints  look like this */
 556         /*--E hint--*/
 557
 558         while(*index < current_strlen) {
 559                 char c = string[*index];
 560
 561                 switch(c) {
 562
 563                         case '-':
 564                                 on_star = 0;
 565                                 if(third_dash)                  fourth_dash = 1;
 566                                 else if(in_hint)                third_dash      = 1;
 567                                 else if(first_dash)     second_dash = 1;
 568                                 else                                            first_dash = 1;
 569                                 break;
 570
 571                         case 'S':
 572                                 on_star = 0;
 573                                 if(second_dash && !in_hint) {
 574                                         (*index)++;
 575                                         json_eat_ws(string, index, 1);
 576                                         (*index)--; /* this will get incremented at the bottom of the loop */
 577                                         in_hint = 1;
 578                                         break;
 579                                 }
 580
 581                         case 'E':
 582                                 on_star = 0;
 583                                 if(second_dash && !in_hint) {
 584                                         (*index)++;
 585                                         json_eat_ws(string, index, 1);
 586                                         (*index)--; /* this will get incremented at the bottom of the loop */
 587                                         in_hint = 1;
 588                                         break;
 589                                 }
 590
 591                         case '*':
 592                                 on_star = 1;
 593                                 break;
 594
 595                         case '/':
 596                                 if(on_star)
 597                                         done = 1;
 598                                 else
 599                                 on_star = 0;
 600                                 break;
 601
 602                         default:
 603                                 on_star = 0;
 604                                 if(in_hint)
 605                                         buffer_add_char(buf, c);
 606                 }
 607
 608                 (*index)++;
 609                 if(done) break;
 610         }
 611
 612         if( buf->n_used > 0 && buffer)
 613                 *buffer = buffer_data(buf);
 614
 615         buffer_free(buf);
 616         return 0;
 617 }
 618
 619 int is_number(char c) {
 620         switch(c) {
 621                 case '0':
 622                 case '1':
 623                 case '2':
 624                 case '3':
 625                 case '4':
 626                 case '5':
 627                 case '6':
 628                 case '7':
 629                 case '8':
 630                 case '9':
 631                         return 1;
 632         }
 633         return 0;
 634 }
 635
 636 int json_handle_error(char* string, unsigned long* index, char* err_msg) {
 637
 638         char buf[60];
 639         memset(buf, 0, 60);
 640
 641         if(*index > 30)
 642                 strncpy( buf, string + (*index - 30), 59 );
 643         else
 644                 strncpy( buf, string, 59 );
 645
 646         fprintf(stderr,
 647                         "\nError parsing json string at charracter %c "
 648                         "(code %d) and index %ld\nMsg:\t%s\nNear:\t%s\n\n",
 649                         string[*index], string[*index], *index, err_msg, buf );
 650         return -1;
 651 }
 652
 653