lib/json.c

   1 /*
   2  * Copyright (c) 2009, 2010, 2011, 2012 Nicira, Inc.
   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 <string.h>
  27
  28 #include "dynamic-string.h"
  29 #include "hash.h"
  30 #include "shash.h"
  31 #include "unicode.h"
  32 #include "util.h"
  33
  34 /* The type of a JSON token. */
  35 enum json_token_type {
  36     T_EOF = 0,
  37     T_BEGIN_ARRAY = '[',
  38     T_END_ARRAY = ']',
  39     T_BEGIN_OBJECT = '{',
  40     T_END_OBJECT = '}',
  41     T_NAME_SEPARATOR = ':',
  42     T_VALUE_SEPARATOR = ',',
  43     T_FALSE = UCHAR_MAX + 1,
  44     T_NULL,
  45     T_TRUE,
  46     T_INTEGER,
  47     T_REAL,
  48     T_STRING
  49 };
  50
  51 /* A JSON token.
  52  *
  53  * RFC 4627 doesn't define a lexical structure for JSON but I believe this to
  54  * be compliant with the standard.
  55  */
  56 struct json_token {
  57     enum json_token_type type;
  58     union {
  59         double real;
  60         long long int integer;
  61         const char *string;
  62     } u;
  63 };
  64
  65 enum json_lex_state {
  66     JSON_LEX_START,             /* Not inside a token. */
  67     JSON_LEX_NUMBER,            /* Reading a number. */
  68     JSON_LEX_KEYWORD,           /* Reading a keyword. */
  69     JSON_LEX_STRING,            /* Reading a quoted string. */
  70     JSON_LEX_ESCAPE             /* In a quoted string just after a "\". */
  71 };
  72
  73 enum json_parse_state {
  74     JSON_PARSE_START,           /* Beginning of input. */
  75     JSON_PARSE_END,             /* End of input. */
  76
  77     /* Objects. */
  78     JSON_PARSE_OBJECT_INIT,     /* Expecting '}' or an object name. */
  79     JSON_PARSE_OBJECT_NAME,     /* Expecting an object name. */
  80     JSON_PARSE_OBJECT_COLON,    /* Expecting ':'. */
  81     JSON_PARSE_OBJECT_VALUE,    /* Expecting an object value. */
  82     JSON_PARSE_OBJECT_NEXT,     /* Expecting ',' or '}'. */
  83
  84     /* Arrays. */
  85     JSON_PARSE_ARRAY_INIT,      /* Expecting ']' or a value. */
  86     JSON_PARSE_ARRAY_VALUE,     /* Expecting a value. */
  87     JSON_PARSE_ARRAY_NEXT       /* Expecting ',' or ']'. */
  88 };
  89
  90 struct json_parser_node {
  91     struct json *json;
  92 };
  93
  94 /* A JSON parser. */
  95 struct json_parser {
  96     int flags;
  97
  98     /* Lexical analysis. */
  99     enum json_lex_state lex_state;
 100     struct ds buffer;           /* Buffer for accumulating token text. */
 101     int line_number;
 102     int column_number;
 103     int byte_number;
 104
 105     /* Parsing. */
 106     enum json_parse_state parse_state;
 107 #define JSON_MAX_HEIGHT 1000
 108     struct json_parser_node *stack;
 109     size_t height, allocated_height;
 110     char *member_name;
 111
 112     /* Parse status. */
 113     bool done;
 114     char *error;                /* Error message, if any, null if none yet. */
 115 };
 116
 117 static struct json *json_create(enum json_type type);
 118 static void json_parser_input(struct json_parser *, struct json_token *);
 119
 120 static void json_error(struct json_parser *p, const char *format, ...)
 121     PRINTF_FORMAT(2, 3);
 122 \f
 123 const char *
 124 json_type_to_string(enum json_type type)
 125 {
 126     switch (type) {
 127     case JSON_NULL:
 128         return "null";
 129
 130     case JSON_FALSE:
 131         return "false";
 132
 133     case JSON_TRUE:
 134         return "true";
 135
 136     case JSON_OBJECT:
 137         return "object";
 138
 139     case JSON_ARRAY:
 140         return "array";
 141
 142     case JSON_INTEGER:
 143     case JSON_REAL:
 144         return "number";
 145
 146     case JSON_STRING:
 147         return "string";
 148
 149     case JSON_N_TYPES:
 150     default:
 151         return "<invalid>";
 152     }
 153 }
 154 \f
 155 /* Functions for manipulating struct json. */
 156
 157 struct json *
 158 json_null_create(void)
 159 {
 160     return json_create(JSON_NULL);
 161 }
 162
 163 struct json *
 164 json_boolean_create(bool b)
 165 {
 166     return json_create(b ? JSON_TRUE : JSON_FALSE);
 167 }
 168
 169 struct json *
 170 json_string_create_nocopy(char *s)
 171 {
 172     struct json *json = json_create(JSON_STRING);
 173     json->u.string = s;
 174     return json;
 175 }
 176
 177 struct json *
 178 json_string_create(const char *s)
 179 {
 180     return json_string_create_nocopy(xstrdup(s));
 181 }
 182
 183 struct json *
 184 json_array_create_empty(void)
 185 {
 186     struct json *json = json_create(JSON_ARRAY);
 187     json->u.array.elems = NULL;
 188     json->u.array.n = 0;
 189     json->u.array.n_allocated = 0;
 190     return json;
 191 }
 192
 193 void
 194 json_array_add(struct json *array_, struct json *element)
 195 {
 196     struct json_array *array = json_array(array_);
 197     if (array->n >= array->n_allocated) {
 198         array->elems = x2nrealloc(array->elems, &array->n_allocated,
 199                                   sizeof *array->elems);
 200     }
 201     array->elems[array->n++] = element;
 202 }
 203
 204 void
 205 json_array_trim(struct json *array_)
 206 {
 207     struct json_array *array = json_array(array_);
 208     if (array->n < array->n_allocated){
 209         array->n_allocated = array->n;
 210         array->elems = xrealloc(array->elems, array->n * sizeof *array->elems);
 211     }
 212 }
 213
 214 struct json *
 215 json_array_create(struct json **elements, size_t n)
 216 {
 217     struct json *json = json_create(JSON_ARRAY);
 218     json->u.array.elems = elements;
 219     json->u.array.n = n;
 220     json->u.array.n_allocated = n;
 221     return json;
 222 }
 223
 224 struct json *
 225 json_array_create_1(struct json *elem0)
 226 {
 227     struct json **elems = xmalloc(sizeof *elems);
 228     elems[0] = elem0;
 229     return json_array_create(elems, 1);
 230 }
 231
 232 struct json *
 233 json_array_create_2(struct json *elem0, struct json *elem1)
 234 {
 235     struct json **elems = xmalloc(2 * sizeof *elems);
 236     elems[0] = elem0;
 237     elems[1] = elem1;
 238     return json_array_create(elems, 2);
 239 }
 240
 241 struct json *
 242 json_array_create_3(struct json *elem0, struct json *elem1, struct json *elem2)
 243 {
 244     struct json **elems = xmalloc(3 * sizeof *elems);
 245     elems[0] = elem0;
 246     elems[1] = elem1;
 247     elems[2] = elem2;
 248     return json_array_create(elems, 3);
 249 }
 250
 251 struct json *
 252 json_object_create(void)
 253 {
 254     struct json *json = json_create(JSON_OBJECT);
 255     json->u.object = xmalloc(sizeof *json->u.object);
 256     shash_init(json->u.object);
 257     return json;
 258 }
 259
 260 struct json *
 261 json_integer_create(long long int integer)
 262 {
 263     struct json *json = json_create(JSON_INTEGER);
 264     json->u.integer = integer;
 265     return json;
 266 }
 267
 268 struct json *
 269 json_real_create(double real)
 270 {
 271     struct json *json = json_create(JSON_REAL);
 272     json->u.real = real;
 273     return json;
 274 }
 275
 276 void
 277 json_object_put(struct json *json, const char *name, struct json *value)
 278 {
 279     json_destroy(shash_replace(json->u.object, name, value));
 280 }
 281
 282 void
 283 json_object_put_string(struct json *json, const char *name, const char *value)
 284 {
 285     json_object_put(json, name, json_string_create(value));
 286 }
 287
 288 const char *
 289 json_string(const struct json *json)
 290 {
 291     assert(json->type == JSON_STRING);
 292     return json->u.string;
 293 }
 294
 295 struct json_array *
 296 json_array(const struct json *json)
 297 {
 298     assert(json->type == JSON_ARRAY);
 299     return CONST_CAST(struct json_array *, &json->u.array);
 300 }
 301
 302 struct shash *
 303 json_object(const struct json *json)
 304 {
 305     assert(json->type == JSON_OBJECT);
 306     return CONST_CAST(struct shash *, json->u.object);
 307 }
 308
 309 bool
 310 json_boolean(const struct json *json)
 311 {
 312     assert(json->type == JSON_TRUE || json->type == JSON_FALSE);
 313     return json->type == JSON_TRUE;
 314 }
 315
 316 double
 317 json_real(const struct json *json)
 318 {
 319     assert(json->type == JSON_REAL || json->type == JSON_INTEGER);
 320     return json->type == JSON_REAL ? json->u.real : json->u.integer;
 321 }
 322
 323 int64_t
 324 json_integer(const struct json *json)
 325 {
 326     assert(json->type == JSON_INTEGER);
 327     return json->u.integer;
 328 }
 329 \f
 330 static void json_destroy_object(struct shash *object);
 331 static void json_destroy_array(struct json_array *array);
 332
 333 /* Frees 'json' and everything it points to, recursively. */
 334 void
 335 json_destroy(struct json *json)
 336 {
 337     if (json) {
 338         switch (json->type) {
 339         case JSON_OBJECT:
 340             json_destroy_object(json->u.object);
 341             break;
 342
 343         case JSON_ARRAY:
 344             json_destroy_array(&json->u.array);
 345             break;
 346
 347         case JSON_STRING:
 348             free(json->u.string);
 349             break;
 350
 351         case JSON_NULL:
 352         case JSON_FALSE:
 353         case JSON_TRUE:
 354         case JSON_INTEGER:
 355         case JSON_REAL:
 356             break;
 357
 358         case JSON_N_TYPES:
 359             NOT_REACHED();
 360         }
 361         free(json);
 362     }
 363 }
 364
 365 static void
 366 json_destroy_object(struct shash *object)
 367 {
 368     struct shash_node *node, *next;
 369
 370     SHASH_FOR_EACH_SAFE (node, next, object) {
 371         struct json *value = node->data;
 372
 373         json_destroy(value);
 374         shash_delete(object, node);
 375     }
 376     shash_destroy(object);
 377     free(object);
 378 }
 379
 380 static void
 381 json_destroy_array(struct json_array *array)
 382 {
 383     size_t i;
 384
 385     for (i = 0; i < array->n; i++) {
 386         json_destroy(array->elems[i]);
 387     }
 388     free(array->elems);
 389 }
 390 \f
 391 static struct json *json_clone_object(const struct shash *object);
 392 static struct json *json_clone_array(const struct json_array *array);
 393
 394 /* Returns a deep copy of 'json'. */
 395 struct json *
 396 json_clone(const struct json *json)
 397 {
 398     switch (json->type) {
 399     case JSON_OBJECT:
 400         return json_clone_object(json->u.object);
 401
 402     case JSON_ARRAY:
 403         return json_clone_array(&json->u.array);
 404
 405     case JSON_STRING:
 406         return json_string_create(json->u.string);
 407
 408     case JSON_NULL:
 409     case JSON_FALSE:
 410     case JSON_TRUE:
 411         return json_create(json->type);
 412
 413     case JSON_INTEGER:
 414         return json_integer_create(json->u.integer);
 415
 416     case JSON_REAL:
 417         return json_real_create(json->u.real);
 418
 419     case JSON_N_TYPES:
 420     default:
 421         NOT_REACHED();
 422     }
 423 }
 424
 425 static struct json *
 426 json_clone_object(const struct shash *object)
 427 {
 428     struct shash_node *node;
 429     struct json *json;
 430
 431     json = json_object_create();
 432     SHASH_FOR_EACH (node, object) {
 433         struct json *value = node->data;
 434         json_object_put(json, node->name, json_clone(value));
 435     }
 436     return json;
 437 }
 438
 439 static struct json *
 440 json_clone_array(const struct json_array *array)
 441 {
 442     struct json **elems;
 443     size_t i;
 444
 445     elems = xmalloc(array->n * sizeof *elems);
 446     for (i = 0; i < array->n; i++) {
 447         elems[i] = json_clone(array->elems[i]);
 448     }
 449     return json_array_create(elems, array->n);
 450 }
 451 \f
 452 static size_t
 453 json_hash_object(const struct shash *object, size_t basis)
 454 {
 455     const struct shash_node **nodes;
 456     size_t n, i;
 457
 458     nodes = shash_sort(object);
 459     n = shash_count(object);
 460     for (i = 0; i < n; i++) {
 461         const struct shash_node *node = nodes[i];
 462         basis = hash_string(node->name, basis);
 463         basis = json_hash(node->data, basis);
 464     }
 465     return basis;
 466 }
 467
 468 static size_t
 469 json_hash_array(const struct json_array *array, size_t basis)
 470 {
 471     size_t i;
 472
 473     basis = hash_int(array->n, basis);
 474     for (i = 0; i < array->n; i++) {
 475         basis = json_hash(array->elems[i], basis);
 476     }
 477     return basis;
 478 }
 479
 480 size_t
 481 json_hash(const struct json *json, size_t basis)
 482 {
 483     switch (json->type) {
 484     case JSON_OBJECT:
 485         return json_hash_object(json->u.object, basis);
 486
 487     case JSON_ARRAY:
 488         return json_hash_array(&json->u.array, basis);
 489
 490     case JSON_STRING:
 491         return hash_string(json->u.string, basis);
 492
 493     case JSON_NULL:
 494     case JSON_FALSE:
 495     case JSON_TRUE:
 496         return hash_int(json->type << 8, basis);
 497
 498     case JSON_INTEGER:
 499         return hash_int(json->u.integer, basis);
 500
 501     case JSON_REAL:
 502         return hash_double(json->u.real, basis);
 503
 504     case JSON_N_TYPES:
 505     default:
 506         NOT_REACHED();
 507     }
 508 }
 509
 510 static bool
 511 json_equal_object(const struct shash *a, const struct shash *b)
 512 {
 513     struct shash_node *a_node;
 514
 515     if (shash_count(a) != shash_count(b)) {
 516         return false;
 517     }
 518
 519     SHASH_FOR_EACH (a_node, a) {
 520         struct shash_node *b_node = shash_find(b, a_node->name);
 521         if (!b_node || !json_equal(a_node->data, b_node->data)) {
 522             return false;
 523         }
 524     }
 525
 526     return true;
 527 }
 528
 529 static bool
 530 json_equal_array(const struct json_array *a, const struct json_array *b)
 531 {
 532     size_t i;
 533
 534     if (a->n != b->n) {
 535         return false;
 536     }
 537
 538     for (i = 0; i < a->n; i++) {
 539         if (!json_equal(a->elems[i], b->elems[i])) {
 540             return false;
 541         }
 542     }
 543
 544     return true;
 545 }
 546
 547 bool
 548 json_equal(const struct json *a, const struct json *b)
 549 {
 550     if (a->type != b->type) {
 551         return false;
 552     }
 553
 554     switch (a->type) {
 555     case JSON_OBJECT:
 556         return json_equal_object(a->u.object, b->u.object);
 557
 558     case JSON_ARRAY:
 559         return json_equal_array(&a->u.array, &b->u.array);
 560
 561     case JSON_STRING:
 562         return !strcmp(a->u.string, b->u.string);
 563
 564     case JSON_NULL:
 565     case JSON_FALSE:
 566     case JSON_TRUE:
 567         return true;
 568
 569     case JSON_INTEGER:
 570         return a->u.integer == b->u.integer;
 571
 572     case JSON_REAL:
 573         return a->u.real == b->u.real;
 574
 575     case JSON_N_TYPES:
 576     default:
 577         NOT_REACHED();
 578     }
 579 }
 580 \f
 581 /* Lexical analysis. */
 582
 583 static void
 584 json_lex_keyword(struct json_parser *p)
 585 {
 586     struct json_token token;
 587     const char *s;
 588
 589     s = ds_cstr(&p->buffer);
 590     if (!strcmp(s, "false")) {
 591         token.type = T_FALSE;
 592     } else if (!strcmp(s, "true")) {
 593         token.type = T_TRUE;
 594     } else if (!strcmp(s, "null")) {
 595         token.type = T_NULL;
 596     } else {
 597         json_error(p, "invalid keyword '%s'", s);
 598         return;
 599     }
 600     json_parser_input(p, &token);
 601 }
 602
 603 static void
 604 json_lex_number(struct json_parser *p)
 605 {
 606     const char *cp = ds_cstr(&p->buffer);
 607     unsigned long long int significand = 0;
 608     struct json_token token;
 609     bool imprecise = false;
 610     bool negative = false;
 611     int pow10 = 0;
 612
 613     /* Leading minus sign. */
 614     if (*cp == '-') {
 615         negative = true;
 616         cp++;
 617     }
 618
 619     /* At least one integer digit, but 0 may not be used as a leading digit for
 620      * a longer number. */
 621     significand = 0;
 622     if (*cp == '0') {
 623         cp++;
 624         if (isdigit(*cp)) {
 625             json_error(p, "leading zeros not allowed");
 626             return;
 627         }
 628     } else if (isdigit(*cp)) {
 629         do {
 630             if (significand <= ULLONG_MAX / 10) {
 631                 significand = significand * 10 + (*cp - '0');
 632             } else {
 633                 pow10++;
 634                 if (*cp != '0') {
 635                     imprecise = true;
 636                 }
 637             }
 638             cp++;
 639         } while (isdigit(*cp));
 640     } else {
 641         json_error(p, "'-' must be followed by digit");
 642         return;
 643     }
 644
 645     /* Optional fraction. */
 646     if (*cp == '.') {
 647         cp++;
 648         if (!isdigit(*cp)) {
 649             json_error(p, "decimal point must be followed by digit");
 650             return;
 651         }
 652         do {
 653             if (significand <= ULLONG_MAX / 10) {
 654                 significand = significand * 10 + (*cp - '0');
 655                 pow10--;
 656             } else if (*cp != '0') {
 657                 imprecise = true;
 658             }
 659             cp++;
 660         } while (isdigit(*cp));
 661     }
 662
 663     /* Optional exponent. */
 664     if (*cp == 'e' || *cp == 'E') {
 665         bool negative_exponent = false;
 666         int exponent;
 667
 668         cp++;
 669         if (*cp == '+') {
 670             cp++;
 671         } else if (*cp == '-') {
 672             negative_exponent = true;
 673             cp++;
 674         }
 675
 676         if (!isdigit(*cp)) {
 677             json_error(p, "exponent must contain at least one digit");
 678             return;
 679         }
 680
 681         exponent = 0;
 682         do {
 683             if (exponent >= INT_MAX / 10) {
 684                 json_error(p, "exponent outside valid range");
 685                 return;
 686             }
 687             exponent = exponent * 10 + (*cp - '0');
 688             cp++;
 689         } while (isdigit(*cp));
 690
 691         if (negative_exponent) {
 692             pow10 -= exponent;
 693         } else {
 694             pow10 += exponent;
 695         }
 696     }
 697
 698     if (*cp != '\0') {
 699         json_error(p, "syntax error in number");
 700         return;
 701     }
 702
 703     /* Figure out number.
 704      *
 705      * We suppress negative zeros as a matter of policy. */
 706     if (!significand) {
 707         token.type = T_INTEGER;
 708         token.u.integer = 0;
 709         json_parser_input(p, &token);
 710         return;
 711     }
 712
 713     if (!imprecise) {
 714         while (pow10 > 0 && significand < ULLONG_MAX / 10) {
 715             significand *= 10;
 716             pow10--;
 717         }
 718         while (pow10 < 0 && significand % 10 == 0) {
 719             significand /= 10;
 720             pow10++;
 721         }
 722         if (pow10 == 0
 723             && significand <= (negative
 724                                ? (unsigned long long int) LLONG_MAX + 1
 725                                : LLONG_MAX)) {
 726             token.type = T_INTEGER;
 727             token.u.integer = negative ? -significand : significand;
 728             json_parser_input(p, &token);
 729             return;
 730         }
 731     }
 732
 733     token.type = T_REAL;
 734     if (!str_to_double(ds_cstr(&p->buffer), &token.u.real)) {
 735         json_error(p, "number outside valid range");
 736         return;
 737     }
 738     /* Suppress negative zero. */
 739     if (token.u.real == 0) {
 740         token.u.real = 0;
 741     }
 742     json_parser_input(p, &token);
 743 }
 744
 745 static const char *
 746 json_lex_4hex(const char *cp, const char *end, int *valuep)
 747 {
 748     unsigned int value;
 749
 750     if (cp + 4 > end) {
 751         return "quoted string ends within \\u escape";
 752     }
 753
 754     value = hexits_value(cp, 4, NULL);
 755     if (value == UINT_MAX) {
 756         return "malformed \\u escape";
 757     }
 758     if (!value) {
 759         return "null bytes not supported in quoted strings";
 760     }
 761     *valuep = value;
 762     return NULL;
 763 }
 764
 765 static const char *
 766 json_lex_unicode(const char *cp, const char *end, struct ds *out)
 767 {
 768     const char *error;
 769     int c0, c1;
 770
 771     error = json_lex_4hex(cp, end, &c0);
 772     if (error) {
 773         ds_clear(out);
 774         ds_put_cstr(out, error);
 775         return NULL;
 776     }
 777     cp += 4;
 778     if (!uc_is_leading_surrogate(c0)) {
 779         ds_put_utf8(out, c0);
 780         return cp;
 781     }
 782
 783     if (cp + 2 > end || *cp++ != '\\' || *cp++ != 'u') {
 784         ds_clear(out);
 785         ds_put_cstr(out, "malformed escaped surrogate pair");
 786         return NULL;
 787     }
 788
 789     error = json_lex_4hex(cp, end, &c1);
 790     if (error) {
 791         ds_clear(out);
 792         ds_put_cstr(out, error);
 793         return NULL;
 794     }
 795     cp += 4;
 796     if (!uc_is_trailing_surrogate(c1)) {
 797         ds_clear(out);
 798         ds_put_cstr(out, "second half of escaped surrogate pair is not "
 799                     "trailing surrogate");
 800         return NULL;
 801     }
 802
 803     ds_put_utf8(out, utf16_decode_surrogate_pair(c0, c1));
 804     return cp;
 805 }
 806
 807 bool
 808 json_string_unescape(const char *in, size_t in_len, char **outp)
 809 {
 810     const char *end = in + in_len;
 811     bool ok = false;
 812     struct ds out;
 813
 814     ds_init(&out);
 815     ds_reserve(&out, in_len);
 816     if (in_len > 0 && in[in_len - 1] == '\\') {
 817         ds_put_cstr(&out, "quoted string may not end with backslash");
 818         goto exit;
 819     }
 820     while (in < end) {
 821         if (*in == '"') {
 822             ds_clear(&out);
 823             ds_put_cstr(&out, "quoted string may not include unescaped \"");
 824             goto exit;
 825         }
 826         if (*in != '\\') {
 827             ds_put_char(&out, *in++);
 828             continue;
 829         }
 830
 831         in++;
 832         switch (*in++) {
 833         case '"': case '\\': case '/':
 834             ds_put_char(&out, in[-1]);
 835             break;
 836
 837         case 'b':
 838             ds_put_char(&out, '\b');
 839             break;
 840
 841         case 'f':
 842             ds_put_char(&out, '\f');
 843             break;
 844
 845         case 'n':
 846             ds_put_char(&out, '\n');
 847             break;
 848
 849         case 'r':
 850             ds_put_char(&out, '\r');
 851             break;
 852
 853         case 't':
 854             ds_put_char(&out, '\t');
 855             break;
 856
 857         case 'u':
 858             in = json_lex_unicode(in, end, &out);
 859             if (!in) {
 860                 goto exit;
 861             }
 862             break;
 863
 864         default:
 865             ds_clear(&out);
 866             ds_put_format(&out, "bad escape \\%c", in[-1]);
 867             goto exit;
 868         }
 869     }
 870     ok = true;
 871
 872 exit:
 873     *outp = ds_cstr(&out);
 874     return ok;
 875 }
 876
 877 static void
 878 json_parser_input_string(struct json_parser *p, const char *s)
 879 {
 880     struct json_token token;
 881
 882     token.type = T_STRING;
 883     token.u.string = s;
 884     json_parser_input(p, &token);
 885 }
 886
 887 static void
 888 json_lex_string(struct json_parser *p)
 889 {
 890     const char *raw = ds_cstr(&p->buffer);
 891     if (!strchr(raw, '\\')) {
 892         json_parser_input_string(p, raw);
 893     } else {
 894         char *cooked;
 895
 896         if (json_string_unescape(raw, strlen(raw), &cooked)) {
 897             json_parser_input_string(p, cooked);
 898         } else {
 899             json_error(p, "%s", cooked);
 900         }
 901
 902         free(cooked);
 903     }
 904 }
 905
 906 static bool
 907 json_lex_input(struct json_parser *p, unsigned char c)
 908 {
 909     struct json_token token;
 910
 911     switch (p->lex_state) {
 912     case JSON_LEX_START:
 913         switch (c) {
 914         case ' ': case '\t': case '\n': case '\r':
 915             /* Nothing to do. */
 916             return true;
 917
 918         case 'a': case 'b': case 'c': case 'd': case 'e':
 919         case 'f': case 'g': case 'h': case 'i': case 'j':
 920         case 'k': case 'l': case 'm': case 'n': case 'o':
 921         case 'p': case 'q': case 'r': case 's': case 't':
 922         case 'u': case 'v': case 'w': case 'x': case 'y':
 923         case 'z':
 924             p->lex_state = JSON_LEX_KEYWORD;
 925             break;
 926
 927         case '[': case '{': case ']': case '}': case ':': case ',':
 928             token.type = c;
 929             json_parser_input(p, &token);
 930             return true;
 931
 932         case '-':
 933         case '0': case '1': case '2': case '3': case '4':
 934         case '5': case '6': case '7': case '8': case '9':
 935             p->lex_state = JSON_LEX_NUMBER;
 936             break;
 937
 938         case '"':
 939             p->lex_state = JSON_LEX_STRING;
 940             return true;
 941
 942         default:
 943             if (isprint(c)) {
 944                 json_error(p, "invalid character '%c'", c);
 945             } else {
 946                 json_error(p, "invalid character U+%04x", c);
 947             }
 948             return true;
 949         }
 950         break;
 951
 952     case JSON_LEX_KEYWORD:
 953         if (!isalpha((unsigned char) c)) {
 954             json_lex_keyword(p);
 955             return false;
 956         }
 957         break;
 958
 959     case JSON_LEX_NUMBER:
 960         if (!strchr(".0123456789eE-+", c)) {
 961             json_lex_number(p);
 962             return false;
 963         }
 964         break;
 965
 966     case JSON_LEX_STRING:
 967         if (c == '\\') {
 968             p->lex_state = JSON_LEX_ESCAPE;
 969         } else if (c == '"') {
 970             json_lex_string(p);
 971             return true;
 972         } else if (c < 0x20) {
 973             json_error(p, "U+%04X must be escaped in quoted string", c);
 974             return true;
 975         }
 976         break;
 977
 978     case JSON_LEX_ESCAPE:
 979         p->lex_state = JSON_LEX_STRING;
 980         break;
 981
 982     default:
 983         abort();
 984     }
 985     ds_put_char(&p->buffer, c);
 986     return true;
 987 }
 988 \f
 989 /* Parsing. */
 990
 991 /* Parses 'string' as a JSON object or array and returns a newly allocated
 992  * 'struct json'.  The caller must free the returned structure with
 993  * json_destroy() when it is no longer needed.
 994  *
 995  * 'string' must be encoded in UTF-8.
 996  *
 997  * If 'string' is valid JSON, then the returned 'struct json' will be either an
 998  * object (JSON_OBJECT) or an array (JSON_ARRAY).
 999  *
1000  * If 'string' is not valid JSON, then the returned 'struct json' will be a
1001  * string (JSON_STRING) that describes the particular error encountered during
1002  * parsing.  (This is an acceptable means of error reporting because at its top
1003  * level JSON must be either an object or an array; a bare string is not
1004  * valid.) */
1005 struct json *
1006 json_from_string(const char *string)
1007 {
1008     struct json_parser *p = json_parser_create(JSPF_TRAILER);
1009     json_parser_feed(p, string, strlen(string));
1010     return json_parser_finish(p);
1011 }
1012
1013 /* Reads the file named 'file_name', parses its contents as a JSON object or
1014  * array, and returns a newly allocated 'struct json'.  The caller must free
1015  * the returned structure with json_destroy() when it is no longer needed.
1016  *
1017  * The file must be encoded in UTF-8.
1018  *
1019  * See json_from_string() for return value semantics.
1020  */
1021 struct json *
1022 json_from_file(const char *file_name)
1023 {
1024     struct json *json;
1025     FILE *stream;
1026
1027     stream = fopen(file_name, "r");
1028     if (!stream) {
1029         return json_string_create_nocopy(
1030             xasprintf("error opening \"%s\": %s", file_name, strerror(errno)));
1031     }
1032     json = json_from_stream(stream);
1033     fclose(stream);
1034
1035     return json;
1036 }
1037
1038 /* Parses the contents of 'stream' as a JSON object or array, and returns a
1039  * newly allocated 'struct json'.  The caller must free the returned structure
1040  * with json_destroy() when it is no longer needed.
1041  *
1042  * The file must be encoded in UTF-8.
1043  *
1044  * See json_from_string() for return value semantics.
1045  */
1046 struct json *
1047 json_from_stream(FILE *stream)
1048 {
1049     struct json_parser *p;
1050     struct json *json;
1051
1052     p = json_parser_create(JSPF_TRAILER);
1053     for (;;) {
1054         char buffer[BUFSIZ];
1055         size_t n;
1056
1057         n = fread(buffer, 1, sizeof buffer, stream);
1058         if (!n || json_parser_feed(p, buffer, n) != n) {
1059             break;
1060         }
1061     }
1062     json = json_parser_finish(p);
1063
1064     if (ferror(stream)) {
1065         json_destroy(json);
1066         json = json_string_create_nocopy(
1067             xasprintf("error reading JSON stream: %s", strerror(errno)));
1068     }
1069
1070     return json;
1071 }
1072
1073 struct json_parser *
1074 json_parser_create(int flags)
1075 {
1076     struct json_parser *p = xzalloc(sizeof *p);
1077     p->flags = flags;
1078     return p;
1079 }
1080
1081 size_t
1082 json_parser_feed(struct json_parser *p, const char *input, size_t n)
1083 {
1084     size_t i;
1085     for (i = 0; !p->done && i < n; ) {
1086         if (json_lex_input(p, input[i])) {
1087             p->byte_number++;
1088             if (input[i] == '\n') {
1089                 p->column_number = 0;
1090                 p->line_number++;
1091             } else {
1092                 p->column_number++;
1093             }
1094             i++;
1095         }
1096     }
1097     return i;
1098 }
1099
1100 bool
1101 json_parser_is_done(const struct json_parser *p)
1102 {
1103     return p->done;
1104 }
1105
1106 struct json *
1107 json_parser_finish(struct json_parser *p)
1108 {
1109     struct json *json;
1110
1111     switch (p->lex_state) {
1112     case JSON_LEX_START:
1113         break;
1114
1115     case JSON_LEX_STRING:
1116     case JSON_LEX_ESCAPE:
1117         json_error(p, "unexpected end of input in quoted string");
1118         break;
1119
1120     case JSON_LEX_NUMBER:
1121     case JSON_LEX_KEYWORD:
1122         json_lex_input(p, ' ');
1123         break;
1124     }
1125
1126     if (p->parse_state == JSON_PARSE_START) {
1127         json_error(p, "empty input stream");
1128     } else if (p->parse_state != JSON_PARSE_END) {
1129         json_error(p, "unexpected end of input");
1130     }
1131
1132     if (!p->error) {
1133         assert(p->height == 1);
1134         assert(p->stack[0].json != NULL);
1135         json = p->stack[--p->height].json;
1136     } else {
1137         json = json_string_create_nocopy(p->error);
1138         p->error = NULL;
1139     }
1140
1141     json_parser_abort(p);
1142
1143     return json;
1144 }
1145
1146 void
1147 json_parser_abort(struct json_parser *p)
1148 {
1149     if (p) {
1150         ds_destroy(&p->buffer);
1151         if (p->height) {
1152             json_destroy(p->stack[0].json);
1153         }
1154         free(p->stack);
1155         free(p->member_name);
1156         free(p->error);
1157         free(p);
1158     }
1159 }
1160
1161 static struct json_parser_node *
1162 json_parser_top(struct json_parser *p)
1163 {
1164     return &p->stack[p->height - 1];
1165 }
1166
1167 static void
1168 json_parser_put_value(struct json_parser *p, struct json *value)
1169 {
1170     struct json_parser_node *node = json_parser_top(p);
1171     if (node->json->type == JSON_OBJECT) {
1172         json_object_put(node->json, p->member_name, value);
1173         free(p->member_name);
1174         p->member_name = NULL;
1175     } else if (node->json->type == JSON_ARRAY) {
1176         json_array_add(node->json, value);
1177     } else {
1178         NOT_REACHED();
1179     }
1180 }
1181
1182 static void
1183 json_parser_push(struct json_parser *p,
1184                  struct json *new_json, enum json_parse_state new_state)
1185 {
1186     if (p->height < JSON_MAX_HEIGHT) {
1187         struct json_parser_node *node;
1188
1189         if (p->height >= p->allocated_height) {
1190             p->stack = x2nrealloc(p->stack, &p->allocated_height,
1191                                   sizeof *p->stack);
1192         }
1193
1194         if (p->height > 0) {
1195             json_parser_put_value(p, new_json);
1196         }
1197
1198         node = &p->stack[p->height++];
1199         node->json = new_json;
1200         p->parse_state = new_state;
1201     } else {
1202         json_destroy(new_json);
1203         json_error(p, "input exceeds maximum nesting depth %d",
1204                    JSON_MAX_HEIGHT);
1205     }
1206 }
1207
1208 static void
1209 json_parser_push_object(struct json_parser *p)
1210 {
1211     json_parser_push(p, json_object_create(), JSON_PARSE_OBJECT_INIT);
1212 }
1213
1214 static void
1215 json_parser_push_array(struct json_parser *p)
1216 {
1217     json_parser_push(p, json_array_create_empty(), JSON_PARSE_ARRAY_INIT);
1218 }
1219
1220 static void
1221 json_parse_value(struct json_parser *p, struct json_token *token,
1222                  enum json_parse_state next_state)
1223 {
1224     struct json *value;
1225
1226     switch (token->type) {
1227     case T_FALSE:
1228         value = json_boolean_create(false);
1229         break;
1230
1231     case T_NULL:
1232         value = json_null_create();
1233         break;
1234
1235     case T_TRUE:
1236         value = json_boolean_create(true);
1237         break;
1238
1239     case '{':
1240         json_parser_push_object(p);
1241         return;
1242
1243     case '[':
1244         json_parser_push_array(p);
1245         return;
1246
1247     case T_INTEGER:
1248         value = json_integer_create(token->u.integer);
1249         break;
1250
1251     case T_REAL:
1252         value = json_real_create(token->u.real);
1253         break;
1254
1255     case T_STRING:
1256         value = json_string_create(token->u.string);
1257         break;
1258
1259     case T_EOF:
1260     case '}':
1261     case ']':
1262     case ':':
1263     case ',':
1264     default:
1265         json_error(p, "syntax error expecting value");
1266         return;
1267     }
1268
1269     json_parser_put_value(p, value);
1270     p->parse_state = next_state;
1271 }
1272
1273 static void
1274 json_parser_pop(struct json_parser *p)
1275 {
1276     struct json_parser_node *node;
1277
1278     /* Conserve memory. */
1279     node = json_parser_top(p);
1280     if (node->json->type == JSON_ARRAY) {
1281         json_array_trim(node->json);
1282     }
1283
1284     /* Pop off the top-of-stack. */
1285     if (p->height == 1) {
1286         p->parse_state = JSON_PARSE_END;
1287         if (!(p->flags & JSPF_TRAILER)) {
1288             p->done = true;
1289         }
1290     } else {
1291         p->height--;
1292         node = json_parser_top(p);
1293         if (node->json->type == JSON_ARRAY) {
1294             p->parse_state = JSON_PARSE_ARRAY_NEXT;
1295         } else if (node->json->type == JSON_OBJECT) {
1296             p->parse_state = JSON_PARSE_OBJECT_NEXT;
1297         } else {
1298             NOT_REACHED();
1299         }
1300     }
1301 }
1302
1303 static void
1304 json_parser_input(struct json_parser *p, struct json_token *token)
1305 {
1306     switch (p->parse_state) {
1307     case JSON_PARSE_START:
1308         if (token->type == '{') {
1309             json_parser_push_object(p);
1310         } else if (token->type == '[') {
1311             json_parser_push_array(p);
1312         } else {
1313             json_error(p, "syntax error at beginning of input");
1314         }
1315         break;
1316
1317     case JSON_PARSE_END:
1318         json_error(p, "trailing garbage at end of input");
1319         break;
1320
1321     case JSON_PARSE_OBJECT_INIT:
1322         if (token->type == '}') {
1323             json_parser_pop(p);
1324             break;
1325         }
1326         /* Fall through. */
1327     case JSON_PARSE_OBJECT_NAME:
1328         if (token->type == T_STRING) {
1329             p->member_name = xstrdup(token->u.string);
1330             p->parse_state = JSON_PARSE_OBJECT_COLON;
1331         } else {
1332             json_error(p, "syntax error parsing object expecting string");
1333         }
1334         break;
1335
1336     case JSON_PARSE_OBJECT_COLON:
1337         if (token->type == ':') {
1338             p->parse_state = JSON_PARSE_OBJECT_VALUE;
1339         } else {
1340             json_error(p, "syntax error parsing object expecting ':'");
1341         }
1342         break;
1343
1344     case JSON_PARSE_OBJECT_VALUE:
1345         json_parse_value(p, token, JSON_PARSE_OBJECT_NEXT);
1346         break;
1347
1348     case JSON_PARSE_OBJECT_NEXT:
1349         if (token->type == ',') {
1350             p->parse_state = JSON_PARSE_OBJECT_NAME;
1351         } else if (token->type == '}') {
1352             json_parser_pop(p);
1353         } else {
1354             json_error(p, "syntax error expecting '}' or ','");
1355         }
1356         break;
1357
1358     case JSON_PARSE_ARRAY_INIT:
1359         if (token->type == ']') {
1360             json_parser_pop(p);
1361             break;
1362         }
1363         /* Fall through. */
1364     case JSON_PARSE_ARRAY_VALUE:
1365         json_parse_value(p, token, JSON_PARSE_ARRAY_NEXT);
1366         break;
1367
1368     case JSON_PARSE_ARRAY_NEXT:
1369         if (token->type == ',') {
1370             p->parse_state = JSON_PARSE_ARRAY_VALUE;
1371         } else if (token->type == ']') {
1372             json_parser_pop(p);
1373         } else {
1374             json_error(p, "syntax error expecting ']' or ','");
1375         }
1376         break;
1377
1378     default:
1379         abort();
1380     }
1381
1382     p->lex_state = JSON_LEX_START;
1383     ds_clear(&p->buffer);
1384 }
1385
1386 static struct json *
1387 json_create(enum json_type type)
1388 {
1389     struct json *json = xmalloc(sizeof *json);
1390     json->type = type;
1391     return json;
1392 }
1393
1394 static void
1395 json_error(struct json_parser *p, const char *format, ...)
1396 {
1397     if (!p->error) {
1398         struct ds msg;
1399         va_list args;
1400
1401         ds_init(&msg);
1402         ds_put_format(&msg, "line %d, column %d, byte %d: ",
1403                       p->line_number, p->column_number, p->byte_number);
1404         va_start(args, format);
1405         ds_put_format_valist(&msg, format, args);
1406         va_end(args);
1407
1408         p->error = ds_steal_cstr(&msg);
1409
1410         p->done = true;
1411     }
1412 }
1413 \f
1414 #define SPACES_PER_LEVEL 2
1415
1416 struct json_serializer {
1417     struct ds *ds;
1418     int depth;
1419     int flags;
1420 };
1421
1422 static void json_serialize(const struct json *, struct json_serializer *);
1423 static void json_serialize_object(const struct shash *object,
1424                                   struct json_serializer *);
1425 static void json_serialize_array(const struct json_array *,
1426                                  struct json_serializer *);
1427 static void json_serialize_string(const char *, struct ds *);
1428
1429 /* Converts 'json' to a string in JSON format, encoded in UTF-8, and returns
1430  * that string.  The caller is responsible for freeing the returned string,
1431  * with free(), when it is no longer needed.
1432  *
1433  * If 'flags' contains JSSF_PRETTY, the output is pretty-printed with each
1434  * nesting level introducing an additional indentation.  Otherwise, the
1435  * returned string does not contain any new-line characters.
1436  *
1437  * If 'flags' contains JSSF_SORT, members of objects in the output are sorted
1438  * in bytewise lexicographic order for reproducibility.  Otherwise, members of
1439  * objects are output in an indeterminate order.
1440  *
1441  * The returned string is valid JSON only if 'json' represents an array or an
1442  * object, since a bare literal does not satisfy the JSON grammar. */
1443 char *
1444 json_to_string(const struct json *json, int flags)
1445 {
1446     struct ds ds;
1447
1448     ds_init(&ds);
1449     json_to_ds(json, flags, &ds);
1450     return ds_steal_cstr(&ds);
1451 }
1452
1453 /* Same as json_to_string(), but the output is appended to 'ds'. */
1454 void
1455 json_to_ds(const struct json *json, int flags, struct ds *ds)
1456 {
1457     struct json_serializer s;
1458
1459     s.ds = ds;
1460     s.depth = 0;
1461     s.flags = flags;
1462     json_serialize(json, &s);
1463 }
1464
1465 static void
1466 json_serialize(const struct json *json, struct json_serializer *s)
1467 {
1468     struct ds *ds = s->ds;
1469
1470     switch (json->type) {
1471     case JSON_NULL:
1472         ds_put_cstr(ds, "null");
1473         break;
1474
1475     case JSON_FALSE:
1476         ds_put_cstr(ds, "false");
1477         break;
1478
1479     case JSON_TRUE:
1480         ds_put_cstr(ds, "true");
1481         break;
1482
1483     case JSON_OBJECT:
1484         json_serialize_object(json->u.object, s);
1485         break;
1486
1487     case JSON_ARRAY:
1488         json_serialize_array(&json->u.array, s);
1489         break;
1490
1491     case JSON_INTEGER:
1492         ds_put_format(ds, "%lld", json->u.integer);
1493         break;
1494
1495     case JSON_REAL:
1496         ds_put_format(ds, "%.*g", DBL_DIG, json->u.real);
1497         break;
1498
1499     case JSON_STRING:
1500         json_serialize_string(json->u.string, ds);
1501         break;
1502
1503     case JSON_N_TYPES:
1504     default:
1505         NOT_REACHED();
1506     }
1507 }
1508
1509 static void
1510 indent_line(struct json_serializer *s)
1511 {
1512     if (s->flags & JSSF_PRETTY) {
1513         ds_put_char(s->ds, '\n');
1514         ds_put_char_multiple(s->ds, ' ', SPACES_PER_LEVEL * s->depth);
1515     }
1516 }
1517
1518 static void
1519 json_serialize_object_member(size_t i, const struct shash_node *node,
1520                              struct json_serializer *s)
1521 {
1522     struct ds *ds = s->ds;
1523
1524     if (i) {
1525         ds_put_char(ds, ',');
1526         indent_line(s);
1527     }
1528
1529     json_serialize_string(node->name, ds);
1530     ds_put_char(ds, ':');
1531     if (s->flags & JSSF_PRETTY) {
1532         ds_put_char(ds, ' ');
1533     }
1534     json_serialize(node->data, s);
1535 }
1536
1537 static void
1538 json_serialize_object(const struct shash *object, struct json_serializer *s)
1539 {
1540     struct ds *ds = s->ds;
1541
1542     ds_put_char(ds, '{');
1543
1544     s->depth++;
1545     indent_line(s);
1546
1547     if (s->flags & JSSF_SORT) {
1548         const struct shash_node **nodes;
1549         size_t n, i;
1550
1551         nodes = shash_sort(object);
1552         n = shash_count(object);
1553         for (i = 0; i < n; i++) {
1554             json_serialize_object_member(i, nodes[i], s);
1555         }
1556         free(nodes);
1557     } else {
1558         struct shash_node *node;
1559         size_t i;
1560
1561         i = 0;
1562         SHASH_FOR_EACH (node, object) {
1563             json_serialize_object_member(i++, node, s);
1564         }
1565     }
1566
1567     ds_put_char(ds, '}');
1568     s->depth--;
1569 }
1570
1571 static void
1572 json_serialize_array(const struct json_array *array, struct json_serializer *s)
1573 {
1574     struct ds *ds = s->ds;
1575     size_t i;
1576
1577     ds_put_char(ds, '[');
1578     s->depth++;
1579
1580     if (array->n > 0) {
1581         indent_line(s);
1582
1583         for (i = 0; i < array->n; i++) {
1584             if (i) {
1585                 ds_put_char(ds, ',');
1586                 indent_line(s);
1587             }
1588             json_serialize(array->elems[i], s);
1589         }
1590     }
1591
1592     s->depth--;
1593     ds_put_char(ds, ']');
1594 }
1595
1596 static void
1597 json_serialize_string(const char *string, struct ds *ds)
1598 {
1599     uint8_t c;
1600
1601     ds_put_char(ds, '"');
1602     while ((c = *string++) != '\0') {
1603         switch (c) {
1604         case '"':
1605             ds_put_cstr(ds, "\\\"");
1606             break;
1607
1608         case '\\':
1609             ds_put_cstr(ds, "\\\\");
1610             break;
1611
1612         case '\b':
1613             ds_put_cstr(ds, "\\b");
1614             break;
1615
1616         case '\f':
1617             ds_put_cstr(ds, "\\f");
1618             break;
1619
1620         case '\n':
1621             ds_put_cstr(ds, "\\n");
1622             break;
1623
1624         case '\r':
1625             ds_put_cstr(ds, "\\r");
1626             break;
1627
1628         case '\t':
1629             ds_put_cstr(ds, "\\t");
1630             break;
1631
1632         default:
1633             if (c >= 32) {
1634                 ds_put_char(ds, c);
1635             } else {
1636                 ds_put_format(ds, "\\u%04x", c);
1637             }
1638             break;
1639         }
1640     }
1641     ds_put_char(ds, '"');
1642 }