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         json_eat_ws(string, &index, 1); /* remove leading whitespace */
  37         if(index == current_strlen) return NULL;
  38
  39         int status = _json_parse_string(string, &index, obj);
  40         if(!status) return obj;
  41
  42         if(status == -2)
  43                 return NULL;
  44
  45         return NULL;
  46 }
  47
  48
  49 int _json_parse_string(char* string, unsigned long* index, object* obj) {
  50         assert(string && index && *index < current_strlen);
  51
  52         int status = 0; /* return code from parsing routines */
  53         char* classname = NULL; /* object class hint */
  54         json_eat_ws(string, index, 1); /* remove leading whitespace */
  55
  56         char c = string[*index];
  57
  58         /* remove any leading comments */
  59         if( c == '/' ) {
  60
  61                 while(1) {
  62                         (*index)++; /* move to second comment char */
  63                         status = json_eat_comment(string, index, &classname, 1);
  64                         if(status) return status;
  65
  66                         json_eat_ws(string, index, 1);
  67                         c = string[*index];
  68                         if(c != '/')
  69                                 break;
  70                 }
  71         }
  72
  73         json_eat_ws(string, index, 1); /* remove leading whitespace */
  74
  75         if(*index >= current_strlen)
  76                 return -2;
  77
  78         switch(c) {
  79
  80                 /* json string */
  81                 case '"':
  82                         (*index)++;
  83                         status = json_parse_json_string(string, index, obj);
  84                         break;
  85
  86                 /* json array */
  87                 case '[':
  88                         (*index)++;
  89                         status = json_parse_json_array(string, index, obj);
  90                         break;
  91
  92                 /* json object */
  93                 case '{':
  94                         (*index)++;
  95                         status = json_parse_json_object(string, index, obj);
  96                         break;
  97
  98                 /* NULL */
  99                 case 'n':
 100                 case 'N':
 101                         status = json_parse_json_null(string, index, obj);
 102                         break;
 103
 104
 105                 /* true, false */
 106                 case 'f':
 107                 case 'F':
 108                 case 't':
 109                 case 'T':
 110                         status = json_parse_json_bool(string, index, obj);
 111                         break;
 112
 113                 default:
 114                         if(is_number(c) || c == '.' || c == '-') { /* are we a number? */
 115                                 status = json_parse_json_number(string, index, obj);
 116                                 if(status) return status;
 117                                 break;
 118                         }
 119
 120                         (*index)--;
 121                         /* we should never get here */
 122                         return json_handle_error(string, index, "_json_parse_string() final switch clause");
 123         }
 124
 125         if(status) return status;
 126
 127         json_eat_ws(string, index, 1);
 128
 129         if( *index < current_strlen ) {
 130                 /* remove any trailing comments */
 131                 c = string[*index];
 132                 if( c == '/' ) {
 133                         (*index)++;
 134                         status = json_eat_comment(string, index, NULL, 0);
 135                         if(status) return status;
 136                 }
 137         }
 138
 139         if(classname){
 140                 obj->set_class(obj, classname);
 141                 free(classname);
 142         }
 143
 144         return 0;
 145 }
 146
 147
 148 int json_parse_json_null(char* string, unsigned long* index, object* obj) {
 149
 150         if(*index >= (current_strlen - 3)) {
 151                 return json_handle_error(string, index,
 152                         "_parse_json_string(): invalid null" );
 153         }
 154
 155         if(!strncasecmp(string + (*index), "null", 4)) {
 156                 (*index) += 4;
 157                 obj->is_null = 1;
 158                 return 0;
 159         } else {
 160                 return json_handle_error(string, index,
 161                         "_parse_json_string(): invalid null" );
 162         }
 163 }
 164
 165 /* should be at the first character of the bool at this point */
 166 int json_parse_json_bool(char* string, unsigned long* index, object* obj) {
 167         assert(string && obj && *index < current_strlen);
 168
 169         char* ret = "json_parse_json_bool(): truncated bool";
 170
 171         if( *index >= (current_strlen - 5))
 172                 return json_handle_error(string, index, ret);
 173
 174         if(!strncasecmp( string + (*index), "false", 5)) {
 175                 (*index) += 5;
 176                 obj->bool_value = 0;
 177                 obj->is_bool = 1;
 178                 obj->is_null = 0;
 179                 return 0;
 180         }
 181
 182         if( *index >= (current_strlen - 4))
 183                 return json_handle_error(string, index, ret);
 184
 185         if(!strncasecmp( string + (*index), "true", 4)) {
 186                 (*index) += 4;
 187                 obj->bool_value = 1;
 188                 obj->is_bool = 1;
 189                 obj->is_null = 0;
 190                 return 0;
 191         }
 192
 193         return json_handle_error(string, index, ret);
 194 }
 195
 196
 197 /* expecting the first character of the number */
 198 int json_parse_json_number(char* string, unsigned long* index, object* obj) {
 199         assert(string && obj && *index < current_strlen);
 200
 201         growing_buffer* buf = buffer_init(64);
 202         char c = string[*index];
 203
 204         int done = 0;
 205         int dot_seen = 0;
 206
 207         /* negative number? */
 208         if(c == '-') { buffer_add(buf, "-"); (*index)++; }
 209
 210         while(*index < current_strlen) {
 211
 212                 if(is_number(c))
 213                         buffer_add_char(buf, c);
 214
 215                 else if( c == '.' ) {
 216                         if(dot_seen) {
 217                                 return json_handle_error(string, index,
 218                                         "json_parse_json_number(): malformed json number");
 219                         }
 220                         dot_seen = 1;
 221                         buffer_add_char(buf, c);
 222                 } else {
 223                         done = 1; break;
 224                 }
 225                 (*index)++;
 226                 c = string[*index];
 227                 if(done) break;
 228         }
 229
 230         if(dot_seen) {
 231                 obj->is_double = 1;
 232                 obj->is_null = 0;
 233                 obj->double_value = strtod(buf->buf, NULL);
 234                 buffer_free(buf);
 235                 return 0;
 236
 237         } else {
 238                 obj->is_number = 1;
 239                 obj->is_null = 0;
 240                 obj->num_value = atol(buf->buf);
 241                 buffer_free(buf);
 242                 return 0;
 243         }
 244 }
 245
 246 /* index should point to the character directly following the '['.  when done
 247  * index will point to the character directly following the ']' character
 248  */
 249 int json_parse_json_array(char* string, unsigned long* index, object* obj) {
 250         assert(string && obj && index && *index < current_strlen);
 251
 252         int status;
 253         int in_parse = 0; /* true if this array already contains one item */
 254         obj->is_array = 1;
 255         obj->is_null = 0;
 256         while(*index < current_strlen) {
 257
 258                 json_eat_ws(string, index, 1);
 259
 260                 if(string[*index] == ']') {
 261                         (*index)++;
 262                         break;
 263                 }
 264
 265                 if(in_parse) {
 266                         json_eat_ws(string, index, 1);
 267                         if(string[*index] != ',') {
 268                                 return json_handle_error(string, index,
 269                                         "json_parse_json_array(): array not followed by a ','");
 270                         }
 271                         (*index)++;
 272                         json_eat_ws(string, index, 1);
 273                 }
 274
 275                 object* item = new_object(NULL);
 276                 status = _json_parse_string(string, index, item);
 277
 278                 if(status) return status;
 279                 obj->push(obj, item);
 280                 in_parse = 1;
 281         }
 282
 283         return 0;
 284 }
 285
 286
 287 /* index should point to the character directly following the '{'.  when done
 288  * index will point to the character directly following the '}'
 289  */
 290 int json_parse_json_object(char* string, unsigned long* index, object* obj) {
 291         assert(string && obj && index && *index < current_strlen);
 292
 293         obj->is_hash = 1;
 294         obj->is_null = 0;
 295         int status;
 296         int in_parse = 0; /* true if we've already added one item to this object */
 297
 298         while(*index < current_strlen) {
 299
 300                 json_eat_ws(string, index, 1);
 301
 302                 if(string[*index] == '}') {
 303                         (*index)++;
 304                         break;
 305                 }
 306
 307                 if(in_parse) {
 308                         if(string[*index] != ',') {
 309                                 return json_handle_error(string, index,
 310                                         "json_parse_json_object(): object missing ',' betweenn elements" );
 311                         }
 312                         (*index)++;
 313                         json_eat_ws(string, index, 1);
 314                 }
 315
 316                 /* first we grab the hash key */
 317                 object* key_obj = new_object(NULL);
 318                 status = _json_parse_string(string, index, key_obj);
 319                 if(status) return status;
 320
 321                 if(!key_obj->is_string) {
 322                         return json_handle_error(string, index,
 323                                 "_json_parse_json_object(): hash key not a string");
 324                 }
 325
 326                 char* key = key_obj->string_data;
 327
 328                 json_eat_ws(string, index, 1);
 329
 330                 if(string[*index] != ':') {
 331                         return json_handle_error(string, index,
 332                                 "json_parse_json_object(): hash key not followed by ':' character");
 333                 }
 334
 335                 (*index)++;
 336
 337                 /* now grab the value object */
 338                 json_eat_ws(string, index, 1);
 339                 object* value_obj = new_object(NULL);
 340                 status = _json_parse_string(string, index, value_obj);
 341                 if(status) return status;
 342
 343                 /* put the data into the object and continue */
 344                 obj->add_key(obj, key, value_obj);
 345                 free_object(key_obj);
 346                 in_parse = 1;
 347         }
 348
 349         return 0;
 350 }
 351
 352
 353
 354 /* when done, index will point to the character after the closing quote */
 355 int json_parse_json_string(char* string, unsigned long* index, object* obj) {
 356         assert(string && index && *index < current_strlen);
 357
 358         int in_escape = 0;
 359         int done = 0;
 360         growing_buffer* buf = buffer_init(64);
 361
 362         while(*index < current_strlen) {
 363
 364                 char c = string[*index];
 365
 366                 switch(c) {
 367
 368                         case '\\':
 369                                 if(in_escape) {
 370                                         buffer_add(buf, "\\");
 371                                         in_escape = 0;
 372                                 } else
 373                                         in_escape = 1;
 374                                 break;
 375
 376                         case '"':
 377                                 if(in_escape) {
 378                                         buffer_add(buf, "\"");
 379                                         in_escape = 0;
 380                                 } else
 381                                         done = 1;
 382                                 break;
 383
 384                         case 't':
 385                                 if(in_escape) {
 386                                         buffer_add(buf,"\t");
 387                                         in_escape = 0;
 388                                 } else
 389                                         buffer_add_char(buf, c);
 390                                 break;
 391
 392                         case 'b':
 393                                 if(in_escape) {
 394                                         buffer_add(buf,"\b");
 395                                         in_escape = 0;
 396                                 } else
 397                                         buffer_add_char(buf, c);
 398                                 break;
 399
 400                         case 'f':
 401                                 if(in_escape) {
 402                                         buffer_add(buf,"\f");
 403                                         in_escape = 0;
 404                                 } else
 405                                         buffer_add_char(buf, c);
 406                                 break;
 407
 408                         case 'r':
 409                                 if(in_escape) {
 410                                         buffer_add(buf,"\r");
 411                                         in_escape = 0;
 412                                 } else
 413                                         buffer_add_char(buf, c);
 414                                 break;
 415
 416                         case 'n':
 417                                 if(in_escape) {
 418                                         buffer_add(buf,"\n");
 419                                         in_escape = 0;
 420                                 } else
 421                                         buffer_add_char(buf, c);
 422                                 break;
 423
 424                         case 'u':
 425                                 if(in_escape) {
 426                                         (*index)++;
 427
 428                                         if(*index >= (current_strlen - 4)) {
 429                                                 return json_handle_error(string, index,
 430                                                         "json_parse_json_string(): truncated escaped unicode"); }
 431
 432                                         char buff[5];
 433                                         memset(buff,0,5);
 434                                         memcpy(buff, string + (*index), 4);
 435
 436
 437                                         /* ----------------------------------------------------------------------- */
 438                                         /* ----------------------------------------------------------------------- */
 439                                         /* The following chunk was borrowed with permission from
 440                                                 json-c http://oss.metaparadigm.com/json-c/ */
 441                                         unsigned char utf_out[3];
 442                                         memset(utf_out,0,3);
 443
 444                                         #define hexdigit(x) ( ((x) <= '9') ? (x) - '0' : ((x) & 7) + 9)
 445
 446                                         unsigned int ucs_char =
 447                                                 (hexdigit(string[*index] ) << 12) +
 448                                                 (hexdigit(string[*index + 1]) << 8) +
 449                                                 (hexdigit(string[*index + 2]) << 4) +
 450                                                 hexdigit(string[*index + 3]);
 451
 452                                         if (ucs_char < 0x80) {
 453                                                 utf_out[0] = ucs_char;
 454                                                 buffer_add(buf, utf_out);
 455
 456                                         } else if (ucs_char < 0x800) {
 457                                                 utf_out[0] = 0xc0 | (ucs_char >> 6);
 458                                                 utf_out[1] = 0x80 | (ucs_char & 0x3f);
 459                                                 buffer_add(buf, utf_out);
 460
 461                                         } else {
 462                                                 utf_out[0] = 0xe0 | (ucs_char >> 12);
 463                                                 utf_out[1] = 0x80 | ((ucs_char >> 6) & 0x3f);
 464                                                 utf_out[2] = 0x80 | (ucs_char & 0x3f);
 465                                                 buffer_add(buf, utf_out);
 466                                         }
 467                                         /* ----------------------------------------------------------------------- */
 468                                         /* ----------------------------------------------------------------------- */
 469
 470                                         (*index) += 3;
 471                                         in_escape = 0;
 472
 473                                 } else {
 474
 475                                         buffer_add_char(buf, c);
 476                                 }
 477
 478                                 break;
 479
 480                         default:
 481                                 buffer_add_char(buf, c);
 482                 }
 483
 484                 (*index)++;
 485                 if(done) break;
 486         }
 487
 488         obj->set_string(obj, buf->buf);
 489         buffer_free(buf);
 490         return 0;
 491 }
 492
 493
 494 void json_eat_ws(char* string, unsigned long* index, int eat_all) {
 495         assert(string && index);
 496         if(*index >= current_strlen)
 497                 return;
 498
 499         if( eat_all ) { /* removes newlines, etc */
 500                 while(string[*index] == ' '     ||
 501                                 string[*index] == '\n'  ||
 502                                 string[*index] == '\t')
 503                         (*index)++;
 504         }
 505
 506         else
 507                 while(string[*index] == ' ') (*index)++;
 508 }
 509
 510
 511 /* index should be at the '*' character at the beginning of the comment.
 512  * when done, index will point to the first character after the final /
 513  */
 514 int json_eat_comment(char* string, unsigned long* index, char** buffer, int parse_class) {
 515         assert(string && index && *index < current_strlen);
 516
 517         if(string[*index] != '*' && string[*index] != '/' )
 518                 return json_handle_error(string, index,
 519                         "json_eat_comment(): invalid character after /");
 520
 521         /* chop out any // style comments */
 522         if(string[*index] == '/') {
 523                 (*index)++;
 524                 char c = string[*index];
 525                 while(*index < current_strlen) {
 526                         (*index)++;
 527                         if(c == '\n')
 528                                 return 0;
 529                         c = string[*index];
 530                 }
 531                 return 0;
 532         }
 533
 534         (*index)++;
 535
 536         int on_star                     = 0; /* true if we just saw a '*' character */
 537
 538         /* we're just past the '*' */
 539         if(!parse_class) { /* we're not concerned with class hints */
 540                 while(*index < current_strlen) {
 541                         if(string[*index] == '/') {
 542                                 if(on_star) {
 543                                         (*index)++;
 544                                         return 0;
 545                                 }
 546                         }
 547
 548                         if(string[*index] == '*') on_star = 1;
 549                         else on_star = 0;
 550
 551                         (*index)++;
 552                 }
 553                 return 0;
 554         }
 555
 556
 557
 558         growing_buffer* buf = buffer_init(64);
 559
 560         int first_dash          = 0;
 561         int second_dash = 0;
 562         int third_dash          = 0;
 563         int fourth_dash = 0;
 564
 565         int in_hint                     = 0;
 566         int done                                = 0;
 567
 568         /*--S hint--*/   /* <-- Hints  look like this */
 569         /*--E hint--*/
 570
 571         while(*index < current_strlen) {
 572                 char c = string[*index];
 573
 574                 switch(c) {
 575
 576                         case '-':
 577                                 on_star = 0;
 578                                 if(third_dash)                  fourth_dash = 1;
 579                                 else if(in_hint)                third_dash      = 1;
 580                                 else if(first_dash)     second_dash = 1;
 581                                 else                                            first_dash = 1;
 582                                 break;
 583
 584                         case 'S':
 585                                 on_star = 0;
 586                                 if(second_dash && !in_hint) {
 587                                         (*index)++;
 588                                         json_eat_ws(string, index, 1);
 589                                         (*index)--; /* this will get incremented at the bottom of the loop */
 590                                         in_hint = 1;
 591                                         break;
 592                                 }
 593
 594                         case 'E':
 595                                 on_star = 0;
 596                                 if(second_dash && !in_hint) {
 597                                         (*index)++;
 598                                         json_eat_ws(string, index, 1);
 599                                         (*index)--; /* this will get incremented at the bottom of the loop */
 600                                         in_hint = 1;
 601                                         break;
 602                                 }
 603
 604                         case '*':
 605                                 on_star = 1;
 606                                 break;
 607
 608                         case '/':
 609                                 if(on_star)
 610                                         done = 1;
 611                                 else
 612                                 on_star = 0;
 613                                 break;
 614
 615                         default:
 616                                 on_star = 0;
 617                                 if(in_hint)
 618                                         buffer_add_char(buf, c);
 619                 }
 620
 621                 (*index)++;
 622                 if(done) break;
 623         }
 624
 625         if( buf->n_used > 0 && buffer)
 626                 *buffer = buffer_data(buf);
 627
 628         buffer_free(buf);
 629         return 0;
 630 }
 631
 632 int is_number(char c) {
 633         switch(c) {
 634                 case '0':
 635                 case '1':
 636                 case '2':
 637                 case '3':
 638                 case '4':
 639                 case '5':
 640                 case '6':
 641                 case '7':
 642                 case '8':
 643                 case '9':
 644                         return 1;
 645         }
 646         return 0;
 647 }
 648
 649 int json_handle_error(char* string, unsigned long* index, char* err_msg) {
 650
 651         char buf[60];
 652         memset(buf, 0, 60);
 653
 654         if(*index > 30)
 655                 strncpy( buf, string + (*index - 30), 59 );
 656         else
 657                 strncpy( buf, string, 59 );
 658
 659         fprintf(stderr,
 660                         "\nError parsing json string at charracter %c "
 661                         "(code %d) and index %ld\nMsg:\t%s\nNear:\t%s\n\n",
 662                         string[*index], string[*index], *index, err_msg, buf );
 663         return -1;
 664 }
 665
 666