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     struct json_token token;
 610     int sig_digits = 0;
 611     bool imprecise = false;
 612     bool negative = false;
 613     int pow10 = 0;
 614
 615     /* Leading minus sign. */
 616     if (*cp == '-') {
 617         negative = true;
 618         cp++;
 619     }
 620
 621     /* At least one integer digit, but 0 may not be used as a leading digit for
 622      * a longer number. */
 623     significand = 0;
 624     sig_digits = 0;
 625     if (*cp == '0') {
 626         cp++;
 627         if (isdigit(*cp)) {
 628             json_error(p, "leading zeros not allowed");
 629             return;
 630         }
 631     } else if (isdigit(*cp)) {
 632         do {
 633             if (significand <= ULLONG_MAX / 10) {
 634                 significand = significand * 10 + (*cp - '0');
 635                 sig_digits++;
 636             } else {
 637                 pow10++;
 638                 if (*cp != '0') {
 639                     imprecise = true;
 640                 }
 641             }
 642             cp++;
 643         } while (isdigit(*cp));
 644     } else {
 645         json_error(p, "'-' must be followed by digit");
 646         return;
 647     }
 648
 649     /* Optional fraction. */
 650     if (*cp == '.') {
 651         cp++;
 652         if (!isdigit(*cp)) {
 653             json_error(p, "decimal point must be followed by digit");
 654             return;
 655         }
 656         do {
 657             if (significand <= ULLONG_MAX / 10) {
 658                 significand = significand * 10 + (*cp - '0');
 659                 sig_digits++;
 660                 pow10--;
 661             } else if (*cp != '0') {
 662                 imprecise = true;
 663             }
 664             cp++;
 665         } while (isdigit(*cp));
 666     }
 667
 668     /* Optional exponent. */
 669     if (*cp == 'e' || *cp == 'E') {
 670         bool negative_exponent = false;
 671         int exponent;
 672
 673         cp++;
 674         if (*cp == '+') {
 675             cp++;
 676         } else if (*cp == '-') {
 677             negative_exponent = true;
 678             cp++;
 679         }
 680
 681         if (!isdigit(*cp)) {
 682             json_error(p, "exponent must contain at least one digit");
 683             return;
 684         }
 685
 686         exponent = 0;
 687         do {
 688             if (exponent >= INT_MAX / 10) {
 689                 json_error(p, "exponent outside valid range");
 690                 return;
 691             }
 692             exponent = exponent * 10 + (*cp - '0');
 693             cp++;
 694         } while (isdigit(*cp));
 695
 696         if (negative_exponent) {
 697             pow10 -= exponent;
 698         } else {
 699             pow10 += exponent;
 700         }
 701     }
 702
 703     if (*cp != '\0') {
 704         json_error(p, "syntax error in number");
 705         return;
 706     }
 707
 708     /* Figure out number.
 709      *
 710      * We suppress negative zeros as a matter of policy. */
 711     if (!significand) {
 712         struct json_token token;
 713         token.type = T_INTEGER;
 714         token.u.integer = 0;
 715         json_parser_input(p, &token);
 716         return;
 717     }
 718
 719     if (!imprecise) {
 720         while (pow10 > 0 && significand < ULLONG_MAX / 10) {
 721             significand *= 10;
 722             sig_digits++;
 723             pow10--;
 724         }
 725         while (pow10 < 0 && significand % 10 == 0) {
 726             significand /= 10;
 727             sig_digits--;
 728             pow10++;
 729         }
 730         if (pow10 == 0
 731             && significand <= (negative
 732                                ? (unsigned long long int) LLONG_MAX + 1
 733                                : LLONG_MAX)) {
 734             token.type = T_INTEGER;
 735             token.u.integer = negative ? -significand : significand;
 736             json_parser_input(p, &token);
 737             return;
 738         }
 739     }
 740
 741     token.type = T_REAL;
 742     if (!str_to_double(ds_cstr(&p->buffer), &token.u.real)) {
 743         json_error(p, "number outside valid range");
 744         return;
 745     }
 746     /* Suppress negative zero. */
 747     if (token.u.real == 0) {
 748         token.u.real = 0;
 749     }
 750     json_parser_input(p, &token);
 751 }
 752
 753 static bool
 754 json_lex_4hex(struct json_parser *p, const char *cp, int *valuep)
 755 {
 756     int value, i;
 757
 758     value = 0;
 759     for (i = 0; i < 4; i++) {
 760         unsigned char c = *cp++;
 761         if (!isxdigit(c)) {
 762             json_error(p, "malformed \\u escape");
 763             return false;
 764         }
 765         value = (value << 4) | hexit_value(c);
 766     }
 767     if (!value) {
 768         json_error(p, "null bytes not supported in quoted strings");
 769         return false;
 770     }
 771     *valuep = value;
 772     return true;
 773 }
 774
 775 static const char *
 776 json_lex_unicode(struct json_parser *p, const char *cp, struct ds *s)
 777 {
 778     int c0, c1;
 779
 780     if (!json_lex_4hex(p, cp, &c0)) {
 781         return NULL;
 782     }
 783     cp += 4;
 784     if (!uc_is_leading_surrogate(c0)) {
 785         ds_put_utf8(s, c0);
 786         return cp;
 787     }
 788
 789     if (*cp++ != '\\' || *cp++ != 'u') {
 790         json_error(p, "malformed escaped surrogate pair");
 791         return NULL;
 792     }
 793
 794     if (!json_lex_4hex(p, cp, &c1)) {
 795         return NULL;
 796     }
 797     cp += 4;
 798     if (!uc_is_trailing_surrogate(c1)) {
 799         json_error(p, "second half of escaped surrogate pair is not "
 800                    "trailing surrogate");
 801         return NULL;
 802     }
 803
 804     ds_put_utf8(s, utf16_decode_surrogate_pair(c0, c1));
 805     return cp;
 806 }
 807
 808 static void
 809 json_lex_string(struct json_parser *p)
 810 {
 811     struct json_token token;
 812     const char *cp;
 813     struct ds s;
 814
 815     cp = ds_cstr(&p->buffer);
 816     if (!strchr(cp, '\\')) {
 817         token.type = T_STRING;
 818         token.u.string = cp;
 819         json_parser_input(p, &token);
 820         return;
 821     }
 822
 823     ds_init(&s);
 824     ds_reserve(&s, strlen(cp));
 825     while (*cp != '\0') {
 826         if (*cp != '\\') {
 827             ds_put_char(&s, *cp++);
 828             continue;
 829         }
 830
 831         cp++;
 832         switch (*cp++) {
 833         case '"': case '\\': case '/':
 834             ds_put_char(&s, cp[-1]);
 835             break;
 836
 837         case 'b':
 838             ds_put_char(&s, '\b');
 839             break;
 840
 841         case 'f':
 842             ds_put_char(&s, '\f');
 843             break;
 844
 845         case 'n':
 846             ds_put_char(&s, '\n');
 847             break;
 848
 849         case 'r':
 850             ds_put_char(&s, '\r');
 851             break;
 852
 853         case 't':
 854             ds_put_char(&s, '\t');
 855             break;
 856
 857         case 'u':
 858             cp = json_lex_unicode(p, cp, &s);
 859             if (!cp) {
 860                 goto exit;
 861             }
 862             break;
 863
 864         default:
 865             json_error(p, "bad escape \\%c", cp[-1]);
 866             goto exit;
 867         }
 868     }
 869
 870     token.type = T_STRING;
 871     token.u.string = ds_cstr(&s);
 872     json_parser_input(p, &token);
 873
 874 exit:
 875     ds_destroy(&s);
 876     return;
 877 }
 878
 879 static bool
 880 json_lex_input(struct json_parser *p, unsigned char c)
 881 {
 882     struct json_token token;
 883
 884     p->byte_number++;
 885     if (c == '\n') {
 886         p->column_number = 0;
 887         p->line_number++;
 888     } else {
 889         p->column_number++;
 890     }
 891
 892     switch (p->lex_state) {
 893     case JSON_LEX_START:
 894         switch (c) {
 895         case ' ': case '\t': case '\n': case '\r':
 896             /* Nothing to do. */
 897             return true;
 898
 899         case 'a': case 'b': case 'c': case 'd': case 'e':
 900         case 'f': case 'g': case 'h': case 'i': case 'j':
 901         case 'k': case 'l': case 'm': case 'n': case 'o':
 902         case 'p': case 'q': case 'r': case 's': case 't':
 903         case 'u': case 'v': case 'w': case 'x': case 'y':
 904         case 'z':
 905             p->lex_state = JSON_LEX_KEYWORD;
 906             break;
 907
 908         case '[': case '{': case ']': case '}': case ':': case ',':
 909             token.type = c;
 910             json_parser_input(p, &token);
 911             return true;
 912
 913         case '-':
 914         case '0': case '1': case '2': case '3': case '4':
 915         case '5': case '6': case '7': case '8': case '9':
 916             p->lex_state = JSON_LEX_NUMBER;
 917             break;
 918
 919         case '"':
 920             p->lex_state = JSON_LEX_STRING;
 921             return true;
 922
 923         default:
 924             if (isprint(c)) {
 925                 json_error(p, "invalid character '%c'", c);
 926             } else {
 927                 json_error(p, "invalid character U+%04x", c);
 928             }
 929             return true;
 930         }
 931         break;
 932
 933     case JSON_LEX_KEYWORD:
 934         if (!isalpha((unsigned char) c)) {
 935             json_lex_keyword(p);
 936             return false;
 937         }
 938         break;
 939
 940     case JSON_LEX_NUMBER:
 941         if (!strchr(".0123456789eE-+", c)) {
 942             json_lex_number(p);
 943             return false;
 944         }
 945         break;
 946
 947     case JSON_LEX_STRING:
 948         if (c == '\\') {
 949             p->lex_state = JSON_LEX_ESCAPE;
 950         } else if (c == '"') {
 951             json_lex_string(p);
 952             return true;
 953         } else if (c < 0x20) {
 954             json_error(p, "U+%04X must be escaped in quoted string", c);
 955             return true;
 956         }
 957         break;
 958
 959     case JSON_LEX_ESCAPE:
 960         p->lex_state = JSON_LEX_STRING;
 961         break;
 962
 963     default:
 964         abort();
 965     }
 966     ds_put_char(&p->buffer, c);
 967     return true;
 968 }
 969 \f
 970 /* Parsing. */
 971
 972 /* Parses 'string' as a JSON object or array and returns a newly allocated
 973  * 'struct json'.  The caller must free the returned structure with
 974  * json_destroy() when it is no longer needed.
 975  *
 976  * 'string' must be encoded in UTF-8.
 977  *
 978  * If 'string' is valid JSON, then the returned 'struct json' will be either an
 979  * object (JSON_OBJECT) or an array (JSON_ARRAY).
 980  *
 981  * If 'string' is not valid JSON, then the returned 'struct json' will be a
 982  * string (JSON_STRING) that describes the particular error encountered during
 983  * parsing.  (This is an acceptable means of error reporting because at its top
 984  * level JSON must be either an object or an array; a bare string is not
 985  * valid.) */
 986 struct json *
 987 json_from_string(const char *string)
 988 {
 989     struct json_parser *p = json_parser_create(JSPF_TRAILER);
 990     json_parser_feed(p, string, strlen(string));
 991     return json_parser_finish(p);
 992 }
 993
 994 /* Reads the file named 'file_name', parses its contents as a JSON object or
 995  * array, and returns a newly allocated 'struct json'.  The caller must free
 996  * the returned structure with json_destroy() when it is no longer needed.
 997  *
 998  * The file must be encoded in UTF-8.
 999  *
1000  * See json_from_string() for return value semantics.
1001  */
1002 struct json *
1003 json_from_file(const char *file_name)
1004 {
1005     struct json_parser *p;
1006     struct json *json;
1007     FILE *stream;
1008
1009     /* Open file. */
1010     stream = fopen(file_name, "r");
1011     if (!stream) {
1012         return json_string_create_nocopy(
1013             xasprintf("error opening \"%s\": %s", file_name, strerror(errno)));
1014     }
1015
1016     /* Read and parse file. */
1017     p = json_parser_create(JSPF_TRAILER);
1018     for (;;) {
1019         char buffer[BUFSIZ];
1020         size_t n;
1021
1022         n = fread(buffer, 1, sizeof buffer, stream);
1023         if (!n || json_parser_feed(p, buffer, n) != n) {
1024             break;
1025         }
1026     }
1027     json = json_parser_finish(p);
1028
1029     /* Close file and check for I/O errors. */
1030     if (ferror(stream)) {
1031         json_destroy(json);
1032         json = json_string_create_nocopy(
1033             xasprintf("error reading \"%s\": %s", file_name, strerror(errno)));
1034     }
1035     fclose(stream);
1036
1037     return json;
1038 }
1039
1040 struct json_parser *
1041 json_parser_create(int flags)
1042 {
1043     struct json_parser *p = xzalloc(sizeof *p);
1044     p->flags = flags;
1045     return p;
1046 }
1047
1048 size_t
1049 json_parser_feed(struct json_parser *p, const char *input, size_t n)
1050 {
1051     size_t i;
1052     for (i = 0; !p->done && i < n; ) {
1053         if (json_lex_input(p, input[i])) {
1054             i++;
1055         }
1056     }
1057     return i;
1058 }
1059
1060 bool
1061 json_parser_is_done(const struct json_parser *p)
1062 {
1063     return p->done;
1064 }
1065
1066 struct json *
1067 json_parser_finish(struct json_parser *p)
1068 {
1069     struct json *json;
1070
1071     switch (p->lex_state) {
1072     case JSON_LEX_START:
1073         break;
1074
1075     case JSON_LEX_STRING:
1076     case JSON_LEX_ESCAPE:
1077         json_error(p, "unexpected end of input in quoted string");
1078         break;
1079
1080     case JSON_LEX_NUMBER:
1081     case JSON_LEX_KEYWORD:
1082         json_lex_input(p, ' ');
1083         break;
1084     }
1085
1086     if (p->parse_state == JSON_PARSE_START) {
1087         json_error(p, "empty input stream");
1088     } else if (p->parse_state != JSON_PARSE_END) {
1089         json_error(p, "unexpected end of input");
1090     }
1091
1092     if (!p->error) {
1093         assert(p->height == 1);
1094         assert(p->stack[0].json != NULL);
1095         json = p->stack[--p->height].json;
1096     } else {
1097         json = json_string_create_nocopy(p->error);
1098         p->error = NULL;
1099     }
1100
1101     json_parser_abort(p);
1102
1103     return json;
1104 }
1105
1106 void
1107 json_parser_abort(struct json_parser *p)
1108 {
1109     if (p) {
1110         ds_destroy(&p->buffer);
1111         if (p->height) {
1112             json_destroy(p->stack[0].json);
1113         }
1114         free(p->stack);
1115         free(p->member_name);
1116         free(p->error);
1117         free(p);
1118     }
1119 }
1120
1121 static struct json_parser_node *
1122 json_parser_top(struct json_parser *p)
1123 {
1124     return &p->stack[p->height - 1];
1125 }
1126
1127 static void
1128 json_parser_put_value(struct json_parser *p, struct json *value)
1129 {
1130     struct json_parser_node *node = json_parser_top(p);
1131     if (node->json->type == JSON_OBJECT) {
1132         json_object_put(node->json, p->member_name, value);
1133         free(p->member_name);
1134         p->member_name = NULL;
1135     } else if (node->json->type == JSON_ARRAY) {
1136         json_array_add(node->json, value);
1137     } else {
1138         NOT_REACHED();
1139     }
1140 }
1141
1142 static struct json_parser_node *
1143 json_parser_push(struct json_parser *p,
1144                  struct json *new_json, enum json_parse_state new_state)
1145 {
1146     if (p->height < JSON_MAX_HEIGHT) {
1147         struct json_parser_node *node;
1148
1149         if (p->height >= p->allocated_height) {
1150             p->stack = x2nrealloc(p->stack, &p->allocated_height,
1151                                   sizeof *p->stack);
1152         }
1153
1154         if (p->height > 0) {
1155             json_parser_put_value(p, new_json);
1156         }
1157
1158         node = &p->stack[p->height++];
1159         node->json = new_json;
1160         p->parse_state = new_state;
1161         return node;
1162     } else {
1163         json_error(p, "input exceeds maximum nesting depth %d",
1164                    JSON_MAX_HEIGHT);
1165         return NULL;
1166     }
1167 }
1168
1169 static void
1170 json_parser_push_object(struct json_parser *p)
1171 {
1172     json_parser_push(p, json_object_create(), JSON_PARSE_OBJECT_INIT);
1173 }
1174
1175 static void
1176 json_parser_push_array(struct json_parser *p)
1177 {
1178     json_parser_push(p, json_array_create_empty(), JSON_PARSE_ARRAY_INIT);
1179 }
1180
1181 static void
1182 json_parse_value(struct json_parser *p, struct json_token *token,
1183                  enum json_parse_state next_state)
1184 {
1185     struct json *value;
1186
1187     switch (token->type) {
1188     case T_FALSE:
1189         value = json_boolean_create(false);
1190         break;
1191
1192     case T_NULL:
1193         value = json_null_create();
1194         break;
1195
1196     case T_TRUE:
1197         value = json_boolean_create(true);
1198         break;
1199
1200     case '{':
1201         json_parser_push_object(p);
1202         return;
1203
1204     case '[':
1205         json_parser_push_array(p);
1206         return;
1207
1208     case T_INTEGER:
1209         value = json_integer_create(token->u.integer);
1210         break;
1211
1212     case T_REAL:
1213         value = json_real_create(token->u.real);
1214         break;
1215
1216     case T_STRING:
1217         value = json_string_create(token->u.string);
1218         break;
1219
1220     case T_EOF:
1221     case '}':
1222     case ']':
1223     case ':':
1224     case ',':
1225     default:
1226         json_error(p, "syntax error expecting value");
1227         return;
1228     }
1229
1230     json_parser_put_value(p, value);
1231     p->parse_state = next_state;
1232 }
1233
1234 static void
1235 json_parser_pop(struct json_parser *p)
1236 {
1237     struct json_parser_node *node;
1238
1239     /* Conserve memory. */
1240     node = json_parser_top(p);
1241     if (node->json->type == JSON_ARRAY) {
1242         json_array_trim(node->json);
1243     }
1244
1245     /* Pop off the top-of-stack. */
1246     if (p->height == 1) {
1247         p->parse_state = JSON_PARSE_END;
1248         if (!(p->flags & JSPF_TRAILER)) {
1249             p->done = true;
1250         }
1251     } else {
1252         p->height--;
1253         node = json_parser_top(p);
1254         if (node->json->type == JSON_ARRAY) {
1255             p->parse_state = JSON_PARSE_ARRAY_NEXT;
1256         } else if (node->json->type == JSON_OBJECT) {
1257             p->parse_state = JSON_PARSE_OBJECT_NEXT;
1258         } else {
1259             NOT_REACHED();
1260         }
1261     }
1262 }
1263
1264 static void
1265 json_parser_input(struct json_parser *p, struct json_token *token)
1266 {
1267     switch (p->parse_state) {
1268     case JSON_PARSE_START:
1269         if (token->type == '{') {
1270             json_parser_push_object(p);
1271         } else if (token->type == '[') {
1272             json_parser_push_array(p);
1273         } else {
1274             json_error(p, "syntax error at beginning of input");
1275         }
1276         break;
1277
1278     case JSON_PARSE_END:
1279         json_error(p, "trailing garbage at end of input");
1280         break;
1281
1282     case JSON_PARSE_OBJECT_INIT:
1283         if (token->type == '}') {
1284             json_parser_pop(p);
1285             break;
1286         }
1287         /* Fall through. */
1288     case JSON_PARSE_OBJECT_NAME:
1289         if (token->type == T_STRING) {
1290             p->member_name = xstrdup(token->u.string);
1291             p->parse_state = JSON_PARSE_OBJECT_COLON;
1292         } else {
1293             json_error(p, "syntax error parsing object expecting string");
1294         }
1295         break;
1296
1297     case JSON_PARSE_OBJECT_COLON:
1298         if (token->type == ':') {
1299             p->parse_state = JSON_PARSE_OBJECT_VALUE;
1300         } else {
1301             json_error(p, "syntax error parsing object expecting ':'");
1302         }
1303         break;
1304
1305     case JSON_PARSE_OBJECT_VALUE:
1306         json_parse_value(p, token, JSON_PARSE_OBJECT_NEXT);
1307         break;
1308
1309     case JSON_PARSE_OBJECT_NEXT:
1310         if (token->type == ',') {
1311             p->parse_state = JSON_PARSE_OBJECT_NAME;
1312         } else if (token->type == '}') {
1313             json_parser_pop(p);
1314         } else {
1315             json_error(p, "syntax error expecting '}' or ','");
1316         }
1317         break;
1318
1319     case JSON_PARSE_ARRAY_INIT:
1320         if (token->type == ']') {
1321             json_parser_pop(p);
1322             break;
1323         }
1324         /* Fall through. */
1325     case JSON_PARSE_ARRAY_VALUE:
1326         json_parse_value(p, token, JSON_PARSE_ARRAY_NEXT);
1327         break;
1328
1329     case JSON_PARSE_ARRAY_NEXT:
1330         if (token->type == ',') {
1331             p->parse_state = JSON_PARSE_ARRAY_VALUE;
1332         } else if (token->type == ']') {
1333             json_parser_pop(p);
1334         } else {
1335             json_error(p, "syntax error expecting ']' or ','");
1336         }
1337         break;
1338
1339     default:
1340         abort();
1341     }
1342
1343     p->lex_state = JSON_LEX_START;
1344     ds_clear(&p->buffer);
1345 }
1346
1347 static struct json *
1348 json_create(enum json_type type)
1349 {
1350     struct json *json = xmalloc(sizeof *json);
1351     json->type = type;
1352     return json;
1353 }
1354
1355 static void
1356 json_error(struct json_parser *p, const char *format, ...)
1357 {
1358     if (!p->error) {
1359         struct ds msg;
1360         va_list args;
1361
1362         ds_init(&msg);
1363         ds_put_format(&msg, "line %d, column %d, byte %d: ",
1364                       p->line_number, p->column_number, p->byte_number);
1365         va_start(args, format);
1366         ds_put_format_valist(&msg, format, args);
1367         va_end(args);
1368
1369         p->error = ds_steal_cstr(&msg);
1370
1371         p->done = true;
1372     }
1373 }
1374 \f
1375 #define SPACES_PER_LEVEL 2
1376
1377 struct json_serializer {
1378     struct ds ds;
1379     int depth;
1380     int flags;
1381 };
1382
1383 static void json_to_ds(const struct json *, struct json_serializer *);
1384 static void json_object_to_ds(const struct shash *object,
1385                               struct json_serializer *);
1386 static void json_array_to_ds(const struct json_array *,
1387                              struct json_serializer *);
1388 static void json_string_to_ds(const char *string, struct ds *);
1389
1390 /* Converts 'json' to a string in JSON format, encoded in UTF-8, and returns
1391  * that string.  The caller is responsible for freeing the returned string,
1392  * with free(), when it is no longer needed.
1393  *
1394  * If 'flags' contains JSSF_PRETTY, the output is pretty-printed with each
1395  * nesting level introducing an additional indentation.  Otherwise, the
1396  * returned string does not contain any new-line characters.
1397  *
1398  * If 'flags' contains JSSF_SORT, members of objects in the output are sorted
1399  * in bytewise lexicographic order for reproducibility.  Otherwise, members of
1400  * objects are output in an indeterminate order.
1401  *
1402  * The returned string is valid JSON only if 'json' represents an array or an
1403  * object, since a bare literal does not satisfy the JSON grammar. */
1404 char *
1405 json_to_string(const struct json *json, int flags)
1406 {
1407     struct json_serializer s;
1408     ds_init(&s.ds);
1409     s.depth = 0;
1410     s.flags = flags;
1411     json_to_ds(json, &s);
1412     return ds_steal_cstr(&s.ds);
1413 }
1414
1415 static void
1416 json_to_ds(const struct json *json, struct json_serializer *s)
1417 {
1418     struct ds *ds = &s->ds;
1419
1420     switch (json->type) {
1421     case JSON_NULL:
1422         ds_put_cstr(ds, "null");
1423         break;
1424
1425     case JSON_FALSE:
1426         ds_put_cstr(ds, "false");
1427         break;
1428
1429     case JSON_TRUE:
1430         ds_put_cstr(ds, "true");
1431         break;
1432
1433     case JSON_OBJECT:
1434         json_object_to_ds(json->u.object, s);
1435         break;
1436
1437     case JSON_ARRAY:
1438         json_array_to_ds(&json->u.array, s);
1439         break;
1440
1441     case JSON_INTEGER:
1442         ds_put_format(ds, "%lld", json->u.integer);
1443         break;
1444
1445     case JSON_REAL:
1446         ds_put_format(ds, "%.*g", DBL_DIG, json->u.real);
1447         break;
1448
1449     case JSON_STRING:
1450         json_string_to_ds(json->u.string, ds);
1451         break;
1452
1453     case JSON_N_TYPES:
1454     default:
1455         NOT_REACHED();
1456     }
1457 }
1458
1459 static void
1460 indent_line(struct json_serializer *s)
1461 {
1462     if (s->flags & JSSF_PRETTY) {
1463         ds_put_char(&s->ds, '\n');
1464         ds_put_char_multiple(&s->ds, ' ', SPACES_PER_LEVEL * s->depth);
1465     }
1466 }
1467
1468 static void
1469 json_object_member_to_ds(size_t i, const struct shash_node *node,
1470                          struct json_serializer *s)
1471 {
1472     struct ds *ds = &s->ds;
1473
1474     if (i) {
1475         ds_put_char(ds, ',');
1476         indent_line(s);
1477     }
1478
1479     json_string_to_ds(node->name, ds);
1480     ds_put_char(ds, ':');
1481     if (s->flags & JSSF_PRETTY) {
1482         ds_put_char(ds, ' ');
1483     }
1484     json_to_ds(node->data, s);
1485 }
1486
1487 static void
1488 json_object_to_ds(const struct shash *object, struct json_serializer *s)
1489 {
1490     struct ds *ds = &s->ds;
1491
1492     ds_put_char(ds, '{');
1493
1494     s->depth++;
1495     indent_line(s);
1496
1497     if (s->flags & JSSF_SORT) {
1498         const struct shash_node **nodes;
1499         size_t n, i;
1500
1501         nodes = shash_sort(object);
1502         n = shash_count(object);
1503         for (i = 0; i < n; i++) {
1504             json_object_member_to_ds(i, nodes[i], s);
1505         }
1506         free(nodes);
1507     } else {
1508         struct shash_node *node;
1509         size_t i;
1510
1511         i = 0;
1512         SHASH_FOR_EACH (node, object) {
1513             json_object_member_to_ds(i++, node, s);
1514         }
1515     }
1516
1517     ds_put_char(ds, '}');
1518     s->depth--;
1519 }
1520
1521 static void
1522 json_array_to_ds(const struct json_array *array, struct json_serializer *s)
1523 {
1524     struct ds *ds = &s->ds;
1525     size_t i;
1526
1527     ds_put_char(ds, '[');
1528     s->depth++;
1529
1530     if (array->n > 0) {
1531         indent_line(s);
1532
1533         for (i = 0; i < array->n; i++) {
1534             if (i) {
1535                 ds_put_char(ds, ',');
1536                 indent_line(s);
1537             }
1538             json_to_ds(array->elems[i], s);
1539         }
1540     }
1541
1542     s->depth--;
1543     ds_put_char(ds, ']');
1544 }
1545
1546 static void
1547 json_string_to_ds(const char *string, struct ds *ds)
1548 {
1549     uint8_t c;
1550
1551     ds_put_char(ds, '"');
1552     while ((c = *string++) != '\0') {
1553         switch (c) {
1554         case '"':
1555             ds_put_cstr(ds, "\\\"");
1556             break;
1557
1558         case '\\':
1559             ds_put_cstr(ds, "\\\\");
1560             break;
1561
1562         case '\b':
1563             ds_put_cstr(ds, "\\b");
1564             break;
1565
1566         case '\f':
1567             ds_put_cstr(ds, "\\f");
1568             break;
1569
1570         case '\n':
1571             ds_put_cstr(ds, "\\n");
1572             break;
1573
1574         case '\r':
1575             ds_put_cstr(ds, "\\r");
1576             break;
1577
1578         case '\t':
1579             ds_put_cstr(ds, "\\t");
1580             break;
1581
1582         default:
1583             if (c >= 32) {
1584                 ds_put_char(ds, c);
1585             } else {
1586                 ds_put_format(ds, "\\u%04x", c);
1587             }
1588             break;
1589         }
1590     }
1591     ds_put_char(ds, '"');
1592 }