lib/json.c

   1 /*
   2  * Copyright (c) 2009 Nicira Networks.
   3  *
   4  * Licensed under the Apache License, Version 2.0 (the "License");
   5  * you may not use this file except in compliance with the License.
   6  * You may obtain a copy of the License at:
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  */
  16
  17 #include <config.h>
  18
  19 #include "json.h"
  20
  21 #include <assert.h>
  22 #include <ctype.h>
  23 #include <errno.h>
  24 #include <float.h>
  25 #include <limits.h>
  26 #include <math.h>
  27 #include <string.h>
  28
  29 #include "dynamic-string.h"
  30 #include "hash.h"
  31 #include "shash.h"
  32 #include "unicode.h"
  33 #include "util.h"
  34
  35 /* The type of a JSON token. */
  36 enum json_token_type {
  37     T_EOF = 0,
  38     T_BEGIN_ARRAY = '[',
  39     T_END_ARRAY = ']',
  40     T_BEGIN_OBJECT = '{',
  41     T_END_OBJECT = '}',
  42     T_NAME_SEPARATOR = ':',
  43     T_VALUE_SEPARATOR = ',',
  44     T_FALSE = UCHAR_MAX + 1,
  45     T_NULL,
  46     T_TRUE,
  47     T_INTEGER,
  48     T_REAL,
  49     T_STRING
  50 };
  51
  52 /* A JSON token.
  53  *
  54  * RFC 4627 doesn't define a lexical structure for JSON but I believe this to
  55  * be compliant with the standard.
  56  */
  57 struct json_token {
  58     enum json_token_type type;
  59     union {
  60         double real;
  61         long long int integer;
  62         const char *string;
  63     } u;
  64 };
  65
  66 enum json_lex_state {
  67     JSON_LEX_START,             /* Not inside a token. */
  68     JSON_LEX_NUMBER,            /* Reading a number. */
  69     JSON_LEX_KEYWORD,           /* Reading a keyword. */
  70     JSON_LEX_STRING,            /* Reading a quoted string. */
  71     JSON_LEX_ESCAPE             /* In a quoted string just after a "\". */
  72 };
  73
  74 enum json_parse_state {
  75     JSON_PARSE_START,           /* Beginning of input. */
  76     JSON_PARSE_END,             /* End of input. */
  77
  78     /* Objects. */
  79     JSON_PARSE_OBJECT_INIT,     /* Expecting '}' or an object name. */
  80     JSON_PARSE_OBJECT_NAME,     /* Expecting an object name. */
  81     JSON_PARSE_OBJECT_COLON,    /* Expecting ':'. */
  82     JSON_PARSE_OBJECT_VALUE,    /* Expecting an object value. */
  83     JSON_PARSE_OBJECT_NEXT,     /* Expecting ',' or '}'. */
  84
  85     /* Arrays. */
  86     JSON_PARSE_ARRAY_INIT,      /* Expecting ']' or a value. */
  87     JSON_PARSE_ARRAY_VALUE,     /* Expecting a value. */
  88     JSON_PARSE_ARRAY_NEXT       /* Expecting ',' or ']'. */
  89 };
  90
  91 struct json_parser_node {
  92     struct json *json;
  93 };
  94
  95 /* A JSON parser. */
  96 struct json_parser {
  97     int flags;
  98
  99     /* Lexical analysis. */
 100     enum json_lex_state lex_state;
 101     struct ds buffer;           /* Buffer for accumulating token text. */
 102     int line_number;
 103     int column_number;
 104     int byte_number;
 105
 106     /* Parsing. */
 107     enum json_parse_state parse_state;
 108 #define JSON_MAX_HEIGHT 1000
 109     struct json_parser_node *stack;
 110     size_t height, allocated_height;
 111     char *member_name;
 112
 113     /* Parse status. */
 114     bool done;
 115     char *error;                /* Error message, if any, null if none yet. */
 116 };
 117
 118 static struct json *json_create(enum json_type type);
 119 static void json_parser_input(struct json_parser *, struct json_token *);
 120
 121 static void json_error(struct json_parser *p, const char *format, ...)
 122     PRINTF_FORMAT(2, 3);
 123 \f
 124 const char *
 125 json_type_to_string(enum json_type type)
 126 {
 127     switch (type) {
 128     case JSON_NULL:
 129         return "null";
 130
 131     case JSON_FALSE:
 132         return "false";
 133
 134     case JSON_TRUE:
 135         return "true";
 136
 137     case JSON_OBJECT:
 138         return "object";
 139
 140     case JSON_ARRAY:
 141         return "array";
 142
 143     case JSON_INTEGER:
 144     case JSON_REAL:
 145         return "number";
 146
 147     case JSON_STRING:
 148         return "string";
 149
 150     case JSON_N_TYPES:
 151     default:
 152         return "<invalid>";
 153     }
 154 }
 155 \f
 156 /* Functions for manipulating struct json. */
 157
 158 struct json *
 159 json_null_create(void)
 160 {
 161     return json_create(JSON_NULL);
 162 }
 163
 164 struct json *
 165 json_boolean_create(bool b)
 166 {
 167     return json_create(b ? JSON_TRUE : JSON_FALSE);
 168 }
 169
 170 struct json *
 171 json_string_create_nocopy(char *s)
 172 {
 173     struct json *json = json_create(JSON_STRING);
 174     json->u.string = s;
 175     return json;
 176 }
 177
 178 struct json *
 179 json_string_create(const char *s)
 180 {
 181     return json_string_create_nocopy(xstrdup(s));
 182 }
 183
 184 struct json *
 185 json_array_create_empty(void)
 186 {
 187     struct json *json = json_create(JSON_ARRAY);
 188     json->u.array.elems = NULL;
 189     json->u.array.n = 0;
 190     json->u.array.n_allocated = 0;
 191     return json;
 192 }
 193
 194 void
 195 json_array_add(struct json *array_, struct json *element)
 196 {
 197     struct json_array *array = json_array(array_);
 198     if (array->n >= array->n_allocated) {
 199         array->elems = x2nrealloc(array->elems, &array->n_allocated,
 200                                   sizeof *array->elems);
 201     }
 202     array->elems[array->n++] = element;
 203 }
 204
 205 void
 206 json_array_trim(struct json *array_)
 207 {
 208     struct json_array *array = json_array(array_);
 209     if (array->n < array->n_allocated){
 210         array->n_allocated = array->n;
 211         array->elems = xrealloc(array->elems, array->n * sizeof *array->elems);
 212     }
 213 }
 214
 215 struct json *
 216 json_array_create(struct json **elements, size_t n)
 217 {
 218     struct json *json = json_create(JSON_ARRAY);
 219     json->u.array.elems = elements;
 220     json->u.array.n = n;
 221     json->u.array.n_allocated = n;
 222     return json;
 223 }
 224
 225 struct json *
 226 json_array_create_1(struct json *elem0)
 227 {
 228     struct json **elems = xmalloc(sizeof *elems);
 229     elems[0] = elem0;
 230     return json_array_create(elems, 1);
 231 }
 232
 233 struct json *
 234 json_array_create_2(struct json *elem0, struct json *elem1)
 235 {
 236     struct json **elems = xmalloc(2 * sizeof *elems);
 237     elems[0] = elem0;
 238     elems[1] = elem1;
 239     return json_array_create(elems, 2);
 240 }
 241
 242 struct json *
 243 json_array_create_3(struct json *elem0, struct json *elem1, struct json *elem2)
 244 {
 245     struct json **elems = xmalloc(3 * sizeof *elems);
 246     elems[0] = elem0;
 247     elems[1] = elem1;
 248     elems[2] = elem2;
 249     return json_array_create(elems, 3);
 250 }
 251
 252 struct json *
 253 json_object_create(void)
 254 {
 255     struct json *json = json_create(JSON_OBJECT);
 256     json->u.object = xmalloc(sizeof *json->u.object);
 257     shash_init(json->u.object);
 258     return json;
 259 }
 260
 261 struct json *
 262 json_integer_create(long long int integer)
 263 {
 264     struct json *json = json_create(JSON_INTEGER);
 265     json->u.integer = integer;
 266     return json;
 267 }
 268
 269 struct json *
 270 json_real_create(double real)
 271 {
 272     struct json *json = json_create(JSON_REAL);
 273     json->u.real = real;
 274     return json;
 275 }
 276
 277 void
 278 json_object_put(struct json *json, const char *name, struct json *value)
 279 {
 280     shash_add(json->u.object, name, value);
 281 }
 282
 283 void
 284 json_object_put_string(struct json *json, const char *name, const char *value)
 285 {
 286     json_object_put(json, name, json_string_create(value));
 287 }
 288
 289 const char *
 290 json_string(const struct json *json)
 291 {
 292     assert(json->type == JSON_STRING);
 293     return json->u.string;
 294 }
 295
 296 struct json_array *
 297 json_array(const struct json *json)
 298 {
 299     assert(json->type == JSON_ARRAY);
 300     return (struct json_array *) &json->u.array;
 301 }
 302
 303 struct shash *
 304 json_object(const struct json *json)
 305 {
 306     assert(json->type == JSON_OBJECT);
 307     return (struct shash *) json->u.object;
 308 }
 309
 310 bool
 311 json_boolean(const struct json *json)
 312 {
 313     assert(json->type == JSON_TRUE || json->type == JSON_FALSE);
 314     return json->type == JSON_TRUE;
 315 }
 316
 317 double
 318 json_real(const struct json *json)
 319 {
 320     assert(json->type == JSON_REAL || json->type == JSON_INTEGER);
 321     return json->type == JSON_REAL ? json->u.real : json->u.integer;
 322 }
 323
 324 int64_t
 325 json_integer(const struct json *json)
 326 {
 327     assert(json->type == JSON_INTEGER);
 328     return json->u.integer;
 329 }
 330 \f
 331 static void json_destroy_object(struct shash *object);
 332 static void json_destroy_array(struct json_array *array);
 333
 334 /* Frees 'json' and everything it points to, recursively. */
 335 void
 336 json_destroy(struct json *json)
 337 {
 338     if (json) {
 339         switch (json->type) {
 340         case JSON_OBJECT:
 341             json_destroy_object(json->u.object);
 342             break;
 343
 344         case JSON_ARRAY:
 345             json_destroy_array(&json->u.array);
 346             break;
 347
 348         case JSON_STRING:
 349             free(json->u.string);
 350             break;
 351
 352         case JSON_NULL:
 353         case JSON_FALSE:
 354         case JSON_TRUE:
 355         case JSON_INTEGER:
 356         case JSON_REAL:
 357             break;
 358
 359         case JSON_N_TYPES:
 360             NOT_REACHED();
 361         }
 362         free(json);
 363     }
 364 }
 365
 366 static void
 367 json_destroy_object(struct shash *object)
 368 {
 369     struct shash_node *node, *next;
 370
 371     SHASH_FOR_EACH_SAFE (node, next, object) {
 372         struct json *value = node->data;
 373
 374         json_destroy(value);
 375         shash_delete(object, node);
 376     }
 377     shash_destroy(object);
 378     free(object);
 379 }
 380
 381 static void
 382 json_destroy_array(struct json_array *array)
 383 {
 384     size_t i;
 385
 386     for (i = 0; i < array->n; i++) {
 387         json_destroy(array->elems[i]);
 388     }
 389     free(array->elems);
 390 }
 391 \f
 392 static struct json *json_clone_object(const struct shash *object);
 393 static struct json *json_clone_array(const struct json_array *array);
 394
 395 /* Returns a deep copy of 'json'. */
 396 struct json *
 397 json_clone(const struct json *json)
 398 {
 399     switch (json->type) {
 400     case JSON_OBJECT:
 401         return json_clone_object(json->u.object);
 402
 403     case JSON_ARRAY:
 404         return json_clone_array(&json->u.array);
 405
 406     case JSON_STRING:
 407         return json_string_create(json->u.string);
 408
 409     case JSON_NULL:
 410     case JSON_FALSE:
 411     case JSON_TRUE:
 412         return json_create(json->type);
 413
 414     case JSON_INTEGER:
 415         return json_integer_create(json->u.integer);
 416
 417     case JSON_REAL:
 418         return json_real_create(json->u.real);
 419
 420     case JSON_N_TYPES:
 421     default:
 422         NOT_REACHED();
 423     }
 424 }
 425
 426 static struct json *
 427 json_clone_object(const struct shash *object)
 428 {
 429     struct shash_node *node;
 430     struct json *json;
 431
 432     json = json_object_create();
 433     SHASH_FOR_EACH (node, object) {
 434         struct json *value = node->data;
 435         json_object_put(json, node->name, json_clone(value));
 436     }
 437     return json;
 438 }
 439
 440 static struct json *
 441 json_clone_array(const struct json_array *array)
 442 {
 443     struct json **elems;
 444     size_t i;
 445
 446     elems = xmalloc(array->n * sizeof *elems);
 447     for (i = 0; i < array->n; i++) {
 448         elems[i] = json_clone(array->elems[i]);
 449     }
 450     return json_array_create(elems, array->n);
 451 }
 452 \f
 453 static size_t
 454 json_hash_object(const struct shash *object, size_t basis)
 455 {
 456     const struct shash_node **nodes;
 457     size_t n, i;
 458
 459     nodes = shash_sort(object);
 460     n = shash_count(object);
 461     for (i = 0; i < n; i++) {
 462         const struct shash_node *node = nodes[i];
 463         basis = hash_string(node->name, basis);
 464         basis = json_hash(node->data, basis);
 465     }
 466     return basis;
 467 }
 468
 469 static size_t
 470 json_hash_array(const struct json_array *array, size_t basis)
 471 {
 472     size_t i;
 473
 474     basis = hash_int(array->n, basis);
 475     for (i = 0; i < array->n; i++) {
 476         basis = json_hash(array->elems[i], basis);
 477     }
 478     return basis;
 479 }
 480
 481 size_t
 482 json_hash(const struct json *json, size_t basis)
 483 {
 484     switch (json->type) {
 485     case JSON_OBJECT:
 486         return json_hash_object(json->u.object, basis);
 487
 488     case JSON_ARRAY:
 489         return json_hash_array(&json->u.array, basis);
 490
 491     case JSON_STRING:
 492         return hash_string(json->u.string, basis);
 493
 494     case JSON_NULL:
 495     case JSON_FALSE:
 496     case JSON_TRUE:
 497         return hash_int(json->type << 8, basis);
 498
 499     case JSON_INTEGER:
 500         return hash_int(json->u.integer, basis);
 501
 502     case JSON_REAL:
 503         return hash_double(json->u.real, basis);
 504
 505     case JSON_N_TYPES:
 506     default:
 507         NOT_REACHED();
 508     }
 509 }
 510
 511 static bool
 512 json_equal_object(const struct shash *a, const struct shash *b)
 513 {
 514     struct shash_node *a_node;
 515
 516     if (shash_count(a) != shash_count(b)) {
 517         return false;
 518     }
 519
 520     SHASH_FOR_EACH (a_node, a) {
 521         struct shash_node *b_node = shash_find(b, a_node->name);
 522         if (!b_node || !json_equal(a_node->data, b_node->data)) {
 523             return false;
 524         }
 525     }
 526
 527     return true;
 528 }
 529
 530 static bool
 531 json_equal_array(const struct json_array *a, const struct json_array *b)
 532 {
 533     size_t i;
 534
 535     if (a->n != b->n) {
 536         return false;
 537     }
 538
 539     for (i = 0; i < a->n; i++) {
 540         if (!json_equal(a->elems[i], b->elems[i])) {
 541             return false;
 542         }
 543     }
 544
 545     return true;
 546 }
 547
 548 bool
 549 json_equal(const struct json *a, const struct json *b)
 550 {
 551     if (a->type != b->type) {
 552         return false;
 553     }
 554
 555     switch (a->type) {
 556     case JSON_OBJECT:
 557         return json_equal_object(a->u.object, b->u.object);
 558
 559     case JSON_ARRAY:
 560         return json_equal_array(&a->u.array, &b->u.array);
 561
 562     case JSON_STRING:
 563         return !strcmp(a->u.string, b->u.string);
 564
 565     case JSON_NULL:
 566     case JSON_FALSE:
 567     case JSON_TRUE:
 568         return true;
 569
 570     case JSON_INTEGER:
 571         return a->u.integer == b->u.integer;
 572
 573     case JSON_REAL:
 574         return a->u.real == b->u.real;
 575
 576     case JSON_N_TYPES:
 577     default:
 578         NOT_REACHED();
 579     }
 580 }
 581 \f
 582 /* Lexical analysis. */
 583
 584 static void
 585 json_lex_keyword(struct json_parser *p)
 586 {
 587     struct json_token token;
 588     const char *s;
 589
 590     s = ds_cstr(&p->buffer);
 591     if (!strcmp(s, "false")) {
 592         token.type = T_FALSE;
 593     } else if (!strcmp(s, "true")) {
 594         token.type = T_TRUE;
 595     } else if (!strcmp(s, "null")) {
 596         token.type = T_NULL;
 597     } else {
 598         json_error(p, "invalid keyword '%s'", s);
 599         return;
 600     }
 601     json_parser_input(p, &token);
 602 }
 603
 604 static void
 605 json_lex_number(struct json_parser *p)
 606 {
 607     const char *cp = ds_cstr(&p->buffer);
 608     unsigned long long int significand = 0;
 609     int sig_digits = 0;
 610     bool imprecise = false;
 611     bool negative = false;
 612     int pow10 = 0;
 613
 614     /* Leading minus sign. */
 615     if (*cp == '-') {
 616         negative = true;
 617         cp++;
 618     }
 619
 620     /* At least one integer digit, but 0 may not be used as a leading digit for
 621      * a longer number. */
 622     significand = 0;
 623     sig_digits = 0;
 624     if (*cp == '0') {
 625         cp++;
 626         if (isdigit(*cp)) {
 627             json_error(p, "leading zeros not allowed");
 628             return;
 629         }
 630     } else if (isdigit(*cp)) {
 631         do {
 632             if (significand <= ULLONG_MAX / 10) {
 633                 significand = significand * 10 + (*cp - '0');
 634                 sig_digits++;
 635             } else {
 636                 pow10++;
 637                 if (*cp != '0') {
 638                     imprecise = true;
 639                 }
 640             }
 641             cp++;
 642         } while (isdigit(*cp));
 643     } else {
 644         json_error(p, "'-' must be followed by digit");
 645         return;
 646     }
 647
 648     /* Optional fraction. */
 649     if (*cp == '.') {
 650         cp++;
 651         if (!isdigit(*cp)) {
 652             json_error(p, "decimal point must be followed by digit");
 653             return;
 654         }
 655         do {
 656             if (significand <= ULLONG_MAX / 10) {
 657                 significand = significand * 10 + (*cp - '0');
 658                 sig_digits++;
 659                 pow10--;
 660             } else if (*cp != '0') {
 661                 imprecise = true;
 662             }
 663             cp++;
 664         } while (isdigit(*cp));
 665     }
 666
 667     /* Optional exponent. */
 668     if (*cp == 'e' || *cp == 'E') {
 669         bool negative_exponent = false;
 670         int exponent;
 671
 672         cp++;
 673         if (*cp == '+') {
 674             cp++;
 675         } else if (*cp == '-') {
 676             negative_exponent = true;
 677             cp++;
 678         }
 679
 680         if (!isdigit(*cp)) {
 681             json_error(p, "exponent must contain at least one digit");
 682             return;
 683         }
 684
 685         exponent = 0;
 686         do {
 687             if (exponent >= INT_MAX / 10) {
 688                 json_error(p, "exponent outside valid range");
 689                 return;
 690             }
 691             exponent = exponent * 10 + (*cp - '0');
 692             cp++;
 693         } while (isdigit(*cp));
 694
 695         if (negative_exponent) {
 696             pow10 -= exponent;
 697         } else {
 698             pow10 += exponent;
 699         }
 700     }
 701
 702     if (*cp != '\0') {
 703         json_error(p, "syntax error in number");
 704         return;
 705     }
 706
 707     /* Figure out number.
 708      *
 709      * We suppress negative zeros as a matter of policy. */
 710     if (!significand) {
 711         struct json_token token;
 712         token.type = T_INTEGER;
 713         token.u.integer = 0;
 714         json_parser_input(p, &token);
 715         return;
 716     }
 717
 718     if (!imprecise) {
 719         while (pow10 > 0 && significand < ULLONG_MAX / 10) {
 720             significand *= 10;
 721             sig_digits++;
 722             pow10--;
 723         }
 724         while (pow10 < 0 && significand % 10 == 0) {
 725             significand /= 10;
 726             sig_digits--;
 727             pow10++;
 728         }
 729         if (pow10 == 0
 730             && significand <= (negative
 731                                ? (unsigned long long int) LLONG_MAX + 1
 732                                : LLONG_MAX)) {
 733             struct json_token token;
 734             token.type = T_INTEGER;
 735             token.u.integer = negative ? -significand : significand;
 736             json_parser_input(p, &token);
 737             return;
 738         }
 739     }
 740
 741     if (pow10 + sig_digits <= DBL_MAX_10_EXP) {
 742         struct json_token token;
 743         token.type = T_REAL;
 744         token.u.real = significand * pow(10.0, pow10);
 745         if (token.u.real <= DBL_MAX) {
 746             if (negative && token.u.real) {
 747                 token.u.real = -token.u.real;
 748             }
 749             json_parser_input(p, &token);
 750             return;
 751         }
 752     }
 753     json_error(p, "number outside valid range");
 754 }
 755
 756 static bool
 757 json_lex_4hex(struct json_parser *p, const char *cp, int *valuep)
 758 {
 759     int value, i;
 760
 761     value = 0;
 762     for (i = 0; i < 4; i++) {
 763         unsigned char c = *cp++;
 764         if (!isxdigit(c)) {
 765             json_error(p, "malformed \\u escape");
 766             return false;
 767         }
 768         value = (value << 4) | hexit_value(c);
 769     }
 770     if (!value) {
 771         json_error(p, "null bytes not supported in quoted strings");
 772         return false;
 773     }
 774     *valuep = value;
 775     return true;
 776 }
 777
 778 static const char *
 779 json_lex_unicode(struct json_parser *p, const char *cp, struct ds *s)
 780 {
 781     int c0, c1;
 782
 783     if (!json_lex_4hex(p, cp, &c0)) {
 784         return NULL;
 785     }
 786     cp += 4;
 787     if (!uc_is_leading_surrogate(c0)) {
 788         ds_put_utf8(s, c0);
 789         return cp;
 790     }
 791
 792     if (*cp++ != '\\' || *cp++ != 'u') {
 793         json_error(p, "malformed escaped surrogate pair");
 794         return NULL;
 795     }
 796
 797     if (!json_lex_4hex(p, cp, &c1)) {
 798         return NULL;
 799     }
 800     cp += 4;
 801     if (!uc_is_trailing_surrogate(c1)) {
 802         json_error(p, "second half of escaped surrogate pair is not "
 803                    "trailing surrogate");
 804         return NULL;
 805     }
 806
 807     ds_put_utf8(s, utf16_decode_surrogate_pair(c0, c1));
 808     return cp;
 809 }
 810
 811 static void
 812 json_lex_string(struct json_parser *p)
 813 {
 814     struct json_token token;
 815     const char *cp;
 816     struct ds s;
 817
 818     cp = ds_cstr(&p->buffer);
 819     if (!strchr(cp, '\\')) {
 820         token.type = T_STRING;
 821         token.u.string = cp;
 822         json_parser_input(p, &token);
 823         return;
 824     }
 825
 826     ds_init(&s);
 827     ds_reserve(&s, strlen(cp));
 828     while (*cp != '\0') {
 829         if (*cp != '\\') {
 830             ds_put_char(&s, *cp++);
 831             continue;
 832         }
 833
 834         cp++;
 835         switch (*cp++) {
 836         case '"': case '\\': case '/':
 837             ds_put_char(&s, cp[-1]);
 838             break;
 839
 840         case 'b':
 841             ds_put_char(&s, '\b');
 842             break;
 843
 844         case 'f':
 845             ds_put_char(&s, '\f');
 846             break;
 847
 848         case 'n':
 849             ds_put_char(&s, '\n');
 850             break;
 851
 852         case 'r':
 853             ds_put_char(&s, '\r');
 854             break;
 855
 856         case 't':
 857             ds_put_char(&s, '\t');
 858             break;
 859
 860         case 'u':
 861             cp = json_lex_unicode(p, cp, &s);
 862             if (!cp) {
 863                 goto exit;
 864             }
 865             break;
 866
 867         default:
 868             json_error(p, "bad escape \\%c", cp[-1]);
 869             goto exit;
 870         }
 871     }
 872
 873     token.type = T_STRING;
 874     token.u.string = ds_cstr(&s);
 875     json_parser_input(p, &token);
 876
 877 exit:
 878     ds_destroy(&s);
 879     return;
 880 }
 881
 882 static bool
 883 json_lex_input(struct json_parser *p, unsigned char c)
 884 {
 885     struct json_token token;
 886
 887     p->byte_number++;
 888     if (c == '\n') {
 889         p->column_number = 0;
 890         p->line_number++;
 891     } else {
 892         p->column_number++;
 893     }
 894
 895     switch (p->lex_state) {
 896     case JSON_LEX_START:
 897         switch (c) {
 898         case ' ': case '\t': case '\n': case '\r':
 899             /* Nothing to do. */
 900             return true;
 901
 902         case 'a': case 'b': case 'c': case 'd': case 'e':
 903         case 'f': case 'g': case 'h': case 'i': case 'j':
 904         case 'k': case 'l': case 'm': case 'n': case 'o':
 905         case 'p': case 'q': case 'r': case 's': case 't':
 906         case 'u': case 'v': case 'w': case 'x': case 'y':
 907         case 'z':
 908             p->lex_state = JSON_LEX_KEYWORD;
 909             break;
 910
 911         case '[': case '{': case ']': case '}': case ':': case ',':
 912             token.type = c;
 913             json_parser_input(p, &token);
 914             return true;
 915
 916         case '-':
 917         case '0': case '1': case '2': case '3': case '4':
 918         case '5': case '6': case '7': case '8': case '9':
 919             p->lex_state = JSON_LEX_NUMBER;
 920             break;
 921
 922         case '"':
 923             p->lex_state = JSON_LEX_STRING;
 924             return true;
 925
 926         default:
 927             if (isprint(c)) {
 928                 json_error(p, "invalid character '%c'", c);
 929             } else {
 930                 json_error(p, "invalid character U+%04x", c);
 931             }
 932             return true;
 933         }
 934         break;
 935
 936     case JSON_LEX_KEYWORD:
 937         if (!isalpha((unsigned char) c)) {
 938             json_lex_keyword(p);
 939             return false;
 940         }
 941         break;
 942
 943     case JSON_LEX_NUMBER:
 944         if (!strchr(".0123456789eE-+", c)) {
 945             json_lex_number(p);
 946             return false;
 947         }
 948         break;
 949
 950     case JSON_LEX_STRING:
 951         if (c == '\\') {
 952             p->lex_state = JSON_LEX_ESCAPE;
 953         } else if (c == '"') {
 954             json_lex_string(p);
 955             return true;
 956         } else if (c < 0x20) {
 957             json_error(p, "U+%04X must be escaped in quoted string", c);
 958             return true;
 959         }
 960         break;
 961
 962     case JSON_LEX_ESCAPE:
 963         p->lex_state = JSON_LEX_STRING;
 964         break;
 965
 966     default:
 967         abort();
 968     }
 969     ds_put_char(&p->buffer, c);
 970     return true;
 971 }
 972 \f
 973 /* Parsing. */
 974
 975 /* Parses 'string' as a JSON object or array and returns a newly allocated
 976  * 'struct json'.  The caller must free the returned structure with
 977  * json_destroy() when it is no longer needed.
 978  *
 979  * 'string' must be encoded in UTF-8.
 980  *
 981  * If 'string' is valid JSON, then the returned 'struct json' will be either an
 982  * object (JSON_OBJECT) or an array (JSON_ARRAY).
 983  *
 984  * If 'string' is not valid JSON, then the returned 'struct json' will be a
 985  * string (JSON_STRING) that describes the particular error encountered during
 986  * parsing.  (This is an acceptable means of error reporting because at its top
 987  * level JSON must be either an object or an array; a bare string is not
 988  * valid.) */
 989 struct json *
 990 json_from_string(const char *string)
 991 {
 992     struct json_parser *p = json_parser_create(JSPF_TRAILER);
 993     json_parser_feed(p, string, strlen(string));
 994     return json_parser_finish(p);
 995 }
 996
 997 /* Reads the file named 'file_name', parses its contents as a JSON object or
 998  * array, and returns a newly allocated 'struct json'.  The caller must free
 999  * the returned structure with json_destroy() when it is no longer needed.
1000  *
1001  * The file must be encoded in UTF-8.
1002  *
1003  * See json_from_string() for return value semantics.
1004  */
1005 struct json *
1006 json_from_file(const char *file_name)
1007 {
1008     struct json_parser *p;
1009     struct json *json;
1010     FILE *stream;
1011
1012     /* Open file. */
1013     stream = fopen(file_name, "r");
1014     if (!stream) {
1015         return json_string_create_nocopy(
1016             xasprintf("error opening \"%s\": %s", file_name, strerror(errno)));
1017     }
1018
1019     /* Read and parse file. */
1020     p = json_parser_create(JSPF_TRAILER);
1021     for (;;) {
1022         char buffer[BUFSIZ];
1023         size_t n;
1024
1025         n = fread(buffer, 1, sizeof buffer, stream);
1026         if (!n || json_parser_feed(p, buffer, n) != n) {
1027             break;
1028         }
1029     }
1030     json = json_parser_finish(p);
1031
1032     /* Close file and check for I/O errors. */
1033     if (ferror(stream)) {
1034         json_destroy(json);
1035         json = json_string_create_nocopy(
1036             xasprintf("error reading \"%s\": %s", file_name, strerror(errno)));
1037     }
1038     fclose(stream);
1039
1040     return json;
1041 }
1042
1043 struct json_parser *
1044 json_parser_create(int flags)
1045 {
1046     struct json_parser *p = xzalloc(sizeof *p);
1047     p->flags = flags;
1048     return p;
1049 }
1050
1051 size_t
1052 json_parser_feed(struct json_parser *p, const char *input, size_t n)
1053 {
1054     size_t i;
1055     for (i = 0; !p->done && i < n; ) {
1056         if (json_lex_input(p, input[i])) {
1057             i++;
1058         }
1059     }
1060     return i;
1061 }
1062
1063 bool
1064 json_parser_is_done(const struct json_parser *p)
1065 {
1066     return p->done;
1067 }
1068
1069 struct json *
1070 json_parser_finish(struct json_parser *p)
1071 {
1072     struct json *json;
1073
1074     switch (p->lex_state) {
1075     case JSON_LEX_START:
1076         break;
1077
1078     case JSON_LEX_STRING:
1079     case JSON_LEX_ESCAPE:
1080         json_error(p, "unexpected end of input in quoted string");
1081         break;
1082
1083     case JSON_LEX_NUMBER:
1084     case JSON_LEX_KEYWORD:
1085         json_lex_input(p, ' ');
1086         break;
1087     }
1088
1089     if (p->parse_state == JSON_PARSE_START) {
1090         json_error(p, "empty input stream");
1091     } else if (p->parse_state != JSON_PARSE_END) {
1092         json_error(p, "unexpected end of input");
1093     }
1094
1095     if (!p->error) {
1096         assert(p->height == 1);
1097         assert(p->stack[0].json != NULL);
1098         json = p->stack[--p->height].json;
1099     } else {
1100         json = json_string_create_nocopy(p->error);
1101         p->error = NULL;
1102     }
1103
1104     json_parser_abort(p);
1105
1106     return json;
1107 }
1108
1109 void
1110 json_parser_abort(struct json_parser *p)
1111 {
1112     if (p) {
1113         ds_destroy(&p->buffer);
1114         if (p->height) {
1115             json_destroy(p->stack[0].json);
1116         }
1117         free(p->stack);
1118         free(p->member_name);
1119         free(p->error);
1120         free(p);
1121     }
1122 }
1123
1124 static struct json_parser_node *
1125 json_parser_top(struct json_parser *p)
1126 {
1127     return &p->stack[p->height - 1];
1128 }
1129
1130 static void
1131 json_parser_put_value(struct json_parser *p, struct json *value)
1132 {
1133     struct json_parser_node *node = json_parser_top(p);
1134     if (node->json->type == JSON_OBJECT) {
1135         json_object_put(node->json, p->member_name, value);
1136         free(p->member_name);
1137         p->member_name = NULL;
1138     } else if (node->json->type == JSON_ARRAY) {
1139         json_array_add(node->json, value);
1140     } else {
1141         NOT_REACHED();
1142     }
1143 }
1144
1145 static struct json_parser_node *
1146 json_parser_push(struct json_parser *p,
1147                  struct json *new_json, enum json_parse_state new_state)
1148 {
1149     if (p->height < JSON_MAX_HEIGHT) {
1150         struct json_parser_node *node;
1151
1152         if (p->height >= p->allocated_height) {
1153             p->stack = x2nrealloc(p->stack, &p->allocated_height,
1154                                   sizeof *p->stack);
1155         }
1156
1157         if (p->height > 0) {
1158             json_parser_put_value(p, new_json);
1159         }
1160
1161         node = &p->stack[p->height++];
1162         node->json = new_json;
1163         p->parse_state = new_state;
1164         return node;
1165     } else {
1166         json_error(p, "input exceeds maximum nesting depth %d",
1167                    JSON_MAX_HEIGHT);
1168         return NULL;
1169     }
1170 }
1171
1172 static void
1173 json_parser_push_object(struct json_parser *p)
1174 {
1175     json_parser_push(p, json_object_create(), JSON_PARSE_OBJECT_INIT);
1176 }
1177
1178 static void
1179 json_parser_push_array(struct json_parser *p)
1180 {
1181     json_parser_push(p, json_array_create_empty(), JSON_PARSE_ARRAY_INIT);
1182 }
1183
1184 static void
1185 json_parse_value(struct json_parser *p, struct json_token *token,
1186                  enum json_parse_state next_state)
1187 {
1188     struct json *value;
1189
1190     switch (token->type) {
1191     case T_FALSE:
1192         value = json_boolean_create(false);
1193         break;
1194
1195     case T_NULL:
1196         value = json_null_create();
1197         break;
1198
1199     case T_TRUE:
1200         value = json_boolean_create(true);
1201         break;
1202
1203     case '{':
1204         json_parser_push_object(p);
1205         return;
1206
1207     case '[':
1208         json_parser_push_array(p);
1209         return;
1210
1211     case T_INTEGER:
1212         value = json_integer_create(token->u.integer);
1213         break;
1214
1215     case T_REAL:
1216         value = json_real_create(token->u.real);
1217         break;
1218
1219     case T_STRING:
1220         value = json_string_create(token->u.string);
1221         break;
1222
1223     case T_EOF:
1224     case '}':
1225     case ']':
1226     case ':':
1227     case ',':
1228     default:
1229         json_error(p, "syntax error expecting value");
1230         return;
1231     }
1232
1233     json_parser_put_value(p, value);
1234     p->parse_state = next_state;
1235 }
1236
1237 static void
1238 json_parser_pop(struct json_parser *p)
1239 {
1240     struct json_parser_node *node;
1241
1242     /* Conserve memory. */
1243     node = json_parser_top(p);
1244     if (node->json->type == JSON_ARRAY) {
1245         json_array_trim(node->json);
1246     }
1247
1248     /* Pop off the top-of-stack. */
1249     if (p->height == 1) {
1250         p->parse_state = JSON_PARSE_END;
1251         if (!(p->flags & JSPF_TRAILER)) {
1252             p->done = true;
1253         }
1254     } else {
1255         p->height--;
1256         node = json_parser_top(p);
1257         if (node->json->type == JSON_ARRAY) {
1258             p->parse_state = JSON_PARSE_ARRAY_NEXT;
1259         } else if (node->json->type == JSON_OBJECT) {
1260             p->parse_state = JSON_PARSE_OBJECT_NEXT;
1261         } else {
1262             NOT_REACHED();
1263         }
1264     }
1265 }
1266
1267 static void
1268 json_parser_input(struct json_parser *p, struct json_token *token)
1269 {
1270     switch (p->parse_state) {
1271     case JSON_PARSE_START:
1272         if (token->type == '{') {
1273             json_parser_push_object(p);
1274         } else if (token->type == '[') {
1275             json_parser_push_array(p);
1276         } else {
1277             json_error(p, "syntax error at beginning of input");
1278         }
1279         break;
1280
1281     case JSON_PARSE_END:
1282         json_error(p, "trailing garbage at end of input");
1283         break;
1284
1285     case JSON_PARSE_OBJECT_INIT:
1286         if (token->type == '}') {
1287             json_parser_pop(p);
1288             break;
1289         }
1290         /* Fall through. */
1291     case JSON_PARSE_OBJECT_NAME:
1292         if (token->type == T_STRING) {
1293             p->member_name = xstrdup(token->u.string);
1294             p->parse_state = JSON_PARSE_OBJECT_COLON;
1295         } else {
1296             json_error(p, "syntax error parsing object expecting string");
1297         }
1298         break;
1299
1300     case JSON_PARSE_OBJECT_COLON:
1301         if (token->type == ':') {
1302             p->parse_state = JSON_PARSE_OBJECT_VALUE;
1303         } else {
1304             json_error(p, "syntax error parsing object expecting ':'");
1305         }
1306         break;
1307
1308     case JSON_PARSE_OBJECT_VALUE:
1309         json_parse_value(p, token, JSON_PARSE_OBJECT_NEXT);
1310         break;
1311
1312     case JSON_PARSE_OBJECT_NEXT:
1313         if (token->type == ',') {
1314             p->parse_state = JSON_PARSE_OBJECT_NAME;
1315         } else if (token->type == '}') {
1316             json_parser_pop(p);
1317         } else {
1318             json_error(p, "syntax error expecting '}' or ','");
1319         }
1320         break;
1321
1322     case JSON_PARSE_ARRAY_INIT:
1323         if (token->type == ']') {
1324             json_parser_pop(p);
1325             break;
1326         }
1327         /* Fall through. */
1328     case JSON_PARSE_ARRAY_VALUE:
1329         json_parse_value(p, token, JSON_PARSE_ARRAY_NEXT);
1330         break;
1331
1332     case JSON_PARSE_ARRAY_NEXT:
1333         if (token->type == ',') {
1334             p->parse_state = JSON_PARSE_ARRAY_VALUE;
1335         } else if (token->type == ']') {
1336             json_parser_pop(p);
1337         } else {
1338             json_error(p, "syntax error expecting ']' or ','");
1339         }
1340         break;
1341
1342     default:
1343         abort();
1344     }
1345
1346     p->lex_state = JSON_LEX_START;
1347     ds_clear(&p->buffer);
1348 }
1349
1350 static struct json *
1351 json_create(enum json_type type)
1352 {
1353     struct json *json = xmalloc(sizeof *json);
1354     json->type = type;
1355     return json;
1356 }
1357
1358 static void
1359 json_error(struct json_parser *p, const char *format, ...)
1360 {
1361     if (!p->error) {
1362         struct ds msg;
1363         va_list args;
1364
1365         ds_init(&msg);
1366         ds_put_format(&msg, "line %d, column %d, byte %d: ",
1367                       p->line_number, p->column_number, p->byte_number);
1368         va_start(args, format);
1369         ds_put_format_valist(&msg, format, args);
1370         va_end(args);
1371
1372         p->error = ds_steal_cstr(&msg);
1373
1374         p->done = true;
1375     }
1376 }
1377 \f
1378 #define SPACES_PER_LEVEL 2
1379
1380 struct json_serializer {
1381     struct ds ds;
1382     int depth;
1383     int flags;
1384 };
1385
1386 static void json_to_ds(const struct json *, struct json_serializer *);
1387 static void json_object_to_ds(const struct shash *object,
1388                               struct json_serializer *);
1389 static void json_array_to_ds(const struct json_array *,
1390                              struct json_serializer *);
1391 static void json_string_to_ds(const char *string, struct ds *);
1392
1393 /* Converts 'json' to a string in JSON format, encoded in UTF-8, and returns
1394  * that string.  The caller is responsible for freeing the returned string,
1395  * with free(), when it is no longer needed.
1396  *
1397  * If 'flags' contains JSSF_PRETTY, the output is pretty-printed with each
1398  * nesting level introducing an additional indentation.  Otherwise, the
1399  * returned string does not contain any new-line characters.
1400  *
1401  * If 'flags' contains JSSF_SORT, members of objects in the output are sorted
1402  * in bytewise lexicographic order for reproducibility.  Otherwise, members of
1403  * objects are output in an indeterminate order.
1404  *
1405  * The returned string is valid JSON only if 'json' represents an array or an
1406  * object, since a bare literal does not satisfy the JSON grammar. */
1407 char *
1408 json_to_string(const struct json *json, int flags)
1409 {
1410     struct json_serializer s;
1411     ds_init(&s.ds);
1412     s.depth = 0;
1413     s.flags = flags;
1414     json_to_ds(json, &s);
1415     return ds_steal_cstr(&s.ds);
1416 }
1417
1418 static void
1419 json_to_ds(const struct json *json, struct json_serializer *s)
1420 {
1421     struct ds *ds = &s->ds;
1422
1423     switch (json->type) {
1424     case JSON_NULL:
1425         ds_put_cstr(ds, "null");
1426         break;
1427
1428     case JSON_FALSE:
1429         ds_put_cstr(ds, "false");
1430         break;
1431
1432     case JSON_TRUE:
1433         ds_put_cstr(ds, "true");
1434         break;
1435
1436     case JSON_OBJECT:
1437         json_object_to_ds(json->u.object, s);
1438         break;
1439
1440     case JSON_ARRAY:
1441         json_array_to_ds(&json->u.array, s);
1442         break;
1443
1444     case JSON_INTEGER:
1445         ds_put_format(ds, "%lld", json->u.integer);
1446         break;
1447
1448     case JSON_REAL:
1449         ds_put_format(ds, "%.*g", DBL_DIG, json->u.real);
1450         break;
1451
1452     case JSON_STRING:
1453         json_string_to_ds(json->u.string, ds);
1454         break;
1455
1456     case JSON_N_TYPES:
1457     default:
1458         NOT_REACHED();
1459     }
1460 }
1461
1462 static void
1463 indent_line(struct json_serializer *s)
1464 {
1465     if (s->flags & JSSF_PRETTY) {
1466         ds_put_char(&s->ds, '\n');
1467         ds_put_char_multiple(&s->ds, ' ', SPACES_PER_LEVEL * s->depth);
1468     }
1469 }
1470
1471 static void
1472 json_object_member_to_ds(size_t i, const struct shash_node *node,
1473                          struct json_serializer *s)
1474 {
1475     struct ds *ds = &s->ds;
1476
1477     if (i) {
1478         ds_put_char(ds, ',');
1479         indent_line(s);
1480     }
1481
1482     json_string_to_ds(node->name, ds);
1483     ds_put_char(ds, ':');
1484     if (s->flags & JSSF_PRETTY) {
1485         ds_put_char(ds, ' ');
1486     }
1487     json_to_ds(node->data, s);
1488 }
1489
1490 static void
1491 json_object_to_ds(const struct shash *object, struct json_serializer *s)
1492 {
1493     struct ds *ds = &s->ds;
1494
1495     ds_put_char(ds, '{');
1496
1497     s->depth++;
1498     indent_line(s);
1499
1500     if (s->flags & JSSF_SORT) {
1501         const struct shash_node **nodes;
1502         size_t n, i;
1503
1504         nodes = shash_sort(object);
1505         n = shash_count(object);
1506         for (i = 0; i < n; i++) {
1507             json_object_member_to_ds(i, nodes[i], s);
1508         }
1509         free(nodes);
1510     } else {
1511         struct shash_node *node;
1512         size_t i;
1513
1514         i = 0;
1515         SHASH_FOR_EACH (node, object) {
1516             json_object_member_to_ds(i++, node, s);
1517         }
1518     }
1519
1520     ds_put_char(ds, '}');
1521     s->depth--;
1522 }
1523
1524 static void
1525 json_array_to_ds(const struct json_array *array, struct json_serializer *s)
1526 {
1527     struct ds *ds = &s->ds;
1528     size_t i;
1529
1530     ds_put_char(ds, '[');
1531     s->depth++;
1532
1533     if (array->n > 0) {
1534         indent_line(s);
1535
1536         for (i = 0; i < array->n; i++) {
1537             if (i) {
1538                 ds_put_char(ds, ',');
1539                 indent_line(s);
1540             }
1541             json_to_ds(array->elems[i], s);
1542         }
1543     }
1544
1545     s->depth--;
1546     ds_put_char(ds, ']');
1547 }
1548
1549 static void
1550 json_string_to_ds(const char *string, struct ds *ds)
1551 {
1552     uint8_t c;
1553
1554     ds_put_char(ds, '"');
1555     while ((c = *string++) != '\0') {
1556         switch (c) {
1557         case '"':
1558             ds_put_cstr(ds, "\\\"");
1559             break;
1560
1561         case '\\':
1562             ds_put_cstr(ds, "\\\\");
1563             break;
1564
1565         case '\b':
1566             ds_put_cstr(ds, "\\b");
1567             break;
1568
1569         case '\f':
1570             ds_put_cstr(ds, "\\f");
1571             break;
1572
1573         case '\n':
1574             ds_put_cstr(ds, "\\n");
1575             break;
1576
1577         case '\r':
1578             ds_put_cstr(ds, "\\r");
1579             break;
1580
1581         case '\t':
1582             ds_put_cstr(ds, "\\t");
1583             break;
1584
1585         default:
1586             if (c >= 32) {
1587                 ds_put_char(ds, c);
1588             } else {
1589                 ds_put_format(ds, "\\u%04x", c);
1590             }
1591             break;
1592         }
1593     }
1594     ds_put_char(ds, '"');
1595 }