OpenSRF/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                 } else {
 214                         done = 1; break;
 215                 }
 216                 (*index)++;
 217                 c = string[*index];
 218                 if(done) break;
 219         }
 220
 221         if(dot_seen) {
 222                 obj->is_double = 1;
 223                 obj->double_value = strtod(buf->buf, NULL);
 224                 buffer_free(buf);
 225                 return 0;
 226
 227         } else {
 228                 obj->is_number = 1;
 229                 obj->num_value = atol(buf->buf);
 230                 buffer_free(buf);
 231                 return 0;
 232         }
 233 }
 234
 235 /* index should point to the character directly following the '['.  when done
 236  * index will point to the character directly following the ']' character
 237  */
 238 int json_parse_json_array(char* string, unsigned long* index, object* obj) {
 239         assert(string && obj && index && *index < current_strlen);
 240
 241         int status;
 242         int in_parse = 0; /* true if this array already contains one item */
 243         obj->is_array = 1;
 244         while(*index < current_strlen) {
 245
 246                 json_eat_ws(string, index, 1);
 247
 248                 if(string[*index] == ']') {
 249                         (*index)++;
 250                         break;
 251                 }
 252
 253                 if(in_parse) {
 254                         json_eat_ws(string, index, 1);
 255                         if(string[*index] != ',') {
 256                                 return json_handle_error(string, index,
 257                                         "json_parse_json_array(): array not followed by a ','");
 258                         }
 259                         (*index)++;
 260                         json_eat_ws(string, index, 1);
 261                 }
 262
 263                 object* item = new_object(NULL);
 264                 status = _json_parse_string(string, index, item);
 265
 266                 if(status) return status;
 267                 obj->push(obj, item);
 268                 in_parse = 1;
 269         }
 270
 271         return 0;
 272 }
 273
 274
 275 /* index should point to the character directly following the '{'.  when done
 276  * index will point to the character directly following the '}'
 277  */
 278 int json_parse_json_object(char* string, unsigned long* index, object* obj) {
 279         assert(string && obj && index && *index < current_strlen);
 280
 281         obj->is_hash = 1;
 282         int status;
 283         int in_parse = 0; /* true if we've already added one item to this object */
 284
 285         while(*index < current_strlen) {
 286
 287                 json_eat_ws(string, index, 1);
 288
 289                 if(string[*index] == '}') {
 290                         (*index)++;
 291                         break;
 292                 }
 293
 294                 if(in_parse) {
 295                         if(string[*index] != ',') {
 296                                 return json_handle_error(string, index,
 297                                         "json_parse_json_object(): object missing ',' betweenn elements" );
 298                         }
 299                         (*index)++;
 300                         json_eat_ws(string, index, 1);
 301                 }
 302
 303                 /* first we grab the hash key */
 304                 object* key_obj = new_object(NULL);
 305                 status = _json_parse_string(string, index, key_obj);
 306                 if(status) return status;
 307
 308                 if(!key_obj->is_string) {
 309                         return json_handle_error(string, index,
 310                                 "_json_parse_json_object(): hash key not a string");
 311                 }
 312
 313                 char* key = key_obj->string_data;
 314
 315                 json_eat_ws(string, index, 1);
 316
 317                 if(string[*index] != ':') {
 318                         return json_handle_error(string, index,
 319                                 "json_parse_json_object(): hash key not followed by ':' character");
 320                 }
 321
 322                 (*index)++;
 323
 324                 /* now grab the value object */
 325                 json_eat_ws(string, index, 1);
 326                 object* value_obj = new_object(NULL);
 327                 status = _json_parse_string(string, index, value_obj);
 328                 if(status) return status;
 329
 330                 /* put the data into the object and continue */
 331                 obj->add_key(obj, key, value_obj);
 332                 free_object(key_obj);
 333                 in_parse = 1;
 334         }
 335         return 0;
 336 }
 337
 338
 339
 340 /* when done, index will point to the character after the closing quote */
 341 int json_parse_json_string(char* string, unsigned long* index, object* obj) {
 342         assert(string && index && *index < current_strlen);
 343
 344         int in_escape = 0;
 345         int done = 0;
 346         growing_buffer* buf = buffer_init(64);
 347
 348         while(*index < current_strlen) {
 349
 350                 char c = string[*index];
 351
 352                 switch(c) {
 353
 354                         case '\\':
 355                                 if(in_escape) {
 356                                         buffer_add(buf, "\\");
 357                                         in_escape = 0;
 358                                 } else
 359                                         in_escape = 1;
 360                                 break;
 361
 362                         case '"':
 363                                 if(in_escape) {
 364                                         buffer_add(buf, "\"");
 365                                         in_escape = 0;
 366                                 } else
 367                                         done = 1;
 368                                 break;
 369
 370                         case 't':
 371                                 if(in_escape) {
 372                                         buffer_add(buf,"\t");
 373                                         in_escape = 0;
 374                                 } else
 375                                         buffer_add_char(buf, c);
 376                                 break;
 377
 378                         case 'b':
 379                                 if(in_escape) {
 380                                         buffer_add(buf,"\b");
 381                                         in_escape = 0;
 382                                 } else
 383                                         buffer_add_char(buf, c);
 384                                 break;
 385
 386                         case 'f':
 387                                 if(in_escape) {
 388                                         buffer_add(buf,"\f");
 389                                         in_escape = 0;
 390                                 } else
 391                                         buffer_add_char(buf, c);
 392                                 break;
 393
 394                         case 'r':
 395                                 if(in_escape) {
 396                                         buffer_add(buf,"\r");
 397                                         in_escape = 0;
 398                                 } else
 399                                         buffer_add_char(buf, c);
 400                                 break;
 401
 402                         case 'n':
 403                                 if(in_escape) {
 404                                         buffer_add(buf,"\n");
 405                                         in_escape = 0;
 406                                 } else
 407                                         buffer_add_char(buf, c);
 408                                 break;
 409
 410                         case 'u':
 411                                 if(in_escape) {
 412                                         (*index)++;
 413
 414                                         if(*index >= (current_strlen - 4)) {
 415                                                 return json_handle_error(string, index,
 416                                                         "json_parse_json_string(): truncated escaped unicode"); }
 417
 418                                         char buff[5];
 419                                         memset(buff,0,5);
 420                                         memcpy(buff, string + (*index), 4);
 421
 422
 423                                         /* ----------------------------------------------------------------------- */
 424                                         /* ----------------------------------------------------------------------- */
 425                                         /* The following chunk was borrowed with permission from
 426                                                 json-c http://oss.metaparadigm.com/json-c/ */
 427                                         unsigned char utf_out[3];
 428                                         memset(utf_out,0,3);
 429
 430                                         #define hexdigit(x) ( ((x) <= '9') ? (x) - '0' : ((x) & 7) + 9)
 431
 432                                         unsigned int ucs_char =
 433                                                 (hexdigit(string[*index] ) << 12) +
 434                                                 (hexdigit(string[*index + 1]) << 8) +
 435                                                 (hexdigit(string[*index + 2]) << 4) +
 436                                                 hexdigit(string[*index + 3]);
 437
 438                                         if (ucs_char < 0x80) {
 439                                                 utf_out[0] = ucs_char;
 440                                                 buffer_add(buf, utf_out);
 441
 442                                         } else if (ucs_char < 0x800) {
 443                                                 utf_out[0] = 0xc0 | (ucs_char >> 6);
 444                                                 utf_out[1] = 0x80 | (ucs_char & 0x3f);
 445                                                 buffer_add(buf, utf_out);
 446
 447                                         } else {
 448                                                 utf_out[0] = 0xe0 | (ucs_char >> 12);
 449                                                 utf_out[1] = 0x80 | ((ucs_char >> 6) & 0x3f);
 450                                                 utf_out[2] = 0x80 | (ucs_char & 0x3f);
 451                                                 buffer_add(buf, utf_out);
 452                                         }
 453                                         /* ----------------------------------------------------------------------- */
 454                                         /* ----------------------------------------------------------------------- */
 455
 456                                         (*index) += 3;
 457                                         in_escape = 0;
 458
 459                                 } else {
 460
 461                                         buffer_add_char(buf, c);
 462                                 }
 463
 464                                 break;
 465
 466                         default:
 467                                 buffer_add_char(buf, c);
 468                 }
 469
 470                 (*index)++;
 471                 if(done) break;
 472         }
 473
 474         obj->set_string(obj, buf->buf);
 475         buffer_free(buf);
 476         return 0;
 477 }
 478
 479
 480 void json_eat_ws(char* string, unsigned long* index, int eat_all) {
 481         assert(string && index);
 482         if(*index >= current_strlen)
 483                 return;
 484
 485         if( eat_all ) { /* removes newlines, etc */
 486                 while(string[*index] == ' '     ||
 487                                 string[*index] == '\n'  ||
 488                                 string[*index] == '\t')
 489                         (*index)++;
 490         }
 491
 492         else
 493                 while(string[*index] == ' ') (*index)++;
 494 }
 495
 496
 497 /* index should be at the '*' character at the beginning of the comment.
 498  * when done, index will point to the first character after the final /
 499  */
 500 int json_eat_comment(char* string, unsigned long* index, char** buffer, int parse_class) {
 501         assert(string && index && *index < current_strlen);
 502
 503         if(string[*index] != '*' && string[*index] != '/' )
 504                 return json_handle_error(string, index,
 505                         "json_eat_comment(): invalid character after /");
 506
 507         /* chop out any // style comments */
 508         if(string[*index] == '/') {
 509                 (*index)++;
 510                 char c = string[*index];
 511                 while(*index < current_strlen) {
 512                         (*index)++;
 513                         if(c == '\n')
 514                                 return 0;
 515                         c = string[*index];
 516                 }
 517                 return 0;
 518         }
 519
 520         (*index)++;
 521
 522         int on_star                     = 0; /* true if we just saw a '*' character */
 523
 524         /* we're just past the '*' */
 525         if(!parse_class) { /* we're not concerned with class hints */
 526                 while(*index < current_strlen) {
 527                         if(string[*index] == '/') {
 528                                 if(on_star) {
 529                                         (*index)++;
 530                                         return 0;
 531                                 }
 532                         }
 533
 534                         if(string[*index] == '*') on_star = 1;
 535                         else on_star = 0;
 536
 537                         (*index)++;
 538                 }
 539                 return 0;
 540         }
 541
 542
 543
 544         growing_buffer* buf = buffer_init(64);
 545
 546         int first_dash          = 0;
 547         int second_dash = 0;
 548         int third_dash          = 0;
 549         int fourth_dash = 0;
 550
 551         int in_hint                     = 0;
 552         int done                                = 0;
 553
 554         /*--S hint--*/   /* <-- Hints  look like this */
 555         /*--E hint--*/
 556
 557         while(*index < current_strlen) {
 558                 char c = string[*index];
 559
 560                 switch(c) {
 561
 562                         case '-':
 563                                 on_star = 0;
 564                                 if(third_dash)                  fourth_dash = 1;
 565                                 else if(in_hint)                third_dash      = 1;
 566                                 else if(first_dash)     second_dash = 1;
 567                                 else                                            first_dash = 1;
 568                                 break;
 569
 570                         case 'S':
 571                                 on_star = 0;
 572                                 if(second_dash && !in_hint) {
 573                                         (*index)++;
 574                                         json_eat_ws(string, index, 1);
 575                                         (*index)--; /* this will get incremented at the bottom of the loop */
 576                                         in_hint = 1;
 577                                         break;
 578                                 }
 579
 580                         case 'E':
 581                                 on_star = 0;
 582                                 if(second_dash && !in_hint) {
 583                                         (*index)++;
 584                                         json_eat_ws(string, index, 1);
 585                                         (*index)--; /* this will get incremented at the bottom of the loop */
 586                                         in_hint = 1;
 587                                         break;
 588                                 }
 589
 590                         case '*':
 591                                 on_star = 1;
 592                                 break;
 593
 594                         case '/':
 595                                 if(on_star)
 596                                         done = 1;
 597                                 else
 598                                 on_star = 0;
 599                                 break;
 600
 601                         default:
 602                                 on_star = 0;
 603                                 if(in_hint)
 604                                         buffer_add_char(buf, c);
 605                 }
 606
 607                 (*index)++;
 608                 if(done) break;
 609         }
 610
 611         if( buf->n_used > 0 && buffer)
 612                 *buffer = buffer_data(buf);
 613
 614         buffer_free(buf);
 615         return 0;
 616 }
 617
 618 int is_number(char c) {
 619         switch(c) {
 620                 case '0':
 621                 case '1':
 622                 case '2':
 623                 case '3':
 624                 case '4':
 625                 case '5':
 626                 case '6':
 627                 case '7':
 628                 case '8':
 629                 case '9':
 630                         return 1;
 631         }
 632         return 0;
 633 }
 634
 635 int json_handle_error(char* string, unsigned long* index, char* err_msg) {
 636
 637         char buf[60];
 638         memset(buf, 0, 60);
 639
 640         if(*index > 30)
 641                 strncpy( buf, string + (*index - 30), 59 );
 642         else
 643                 strncpy( buf, string, 59 );
 644
 645         fprintf(stderr,
 646                         "\nError parsing json string at charracter %c "
 647                         "(code %d) and index %ld\nMsg:\t%s\nNear:\t%s\n\n",
 648                         string[*index], string[*index], *index, err_msg, buf );
 649         return -1;
 650 }
 651
 652