lib/json.c

   1 /*
   2  * Copyright (c) 2009, 2010, 2011, 2012, 2014, 2015 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 <ctype.h>
  22 #include <errno.h>
  23 #include <float.h>
  24 #include <limits.h>
  25 #include <string.h>
  26
  27 #include "dynamic-string.h"
  28 #include "hash.h"
  29 #include "shash.h"
  30 #include "unicode.h"
  31 #include "util.h"
  32
  33 /* The type of a JSON token. */
  34 enum json_token_type {
  35     T_EOF = 0,
  36     T_BEGIN_ARRAY = '[',
  37     T_END_ARRAY = ']',
  38     T_BEGIN_OBJECT = '{',
  39     T_END_OBJECT = '}',
  40     T_NAME_SEPARATOR = ':',
  41     T_VALUE_SEPARATOR = ',',
  42     T_FALSE = UCHAR_MAX + 1,
  43     T_NULL,
  44     T_TRUE,
  45     T_INTEGER,
  46     T_REAL,
  47     T_STRING
  48 };
  49
  50 /* A JSON token.
  51  *
  52  * RFC 4627 doesn't define a lexical structure for JSON but I believe this to
  53  * be compliant with the standard.
  54  */
  55 struct json_token {
  56     enum json_token_type type;
  57     union {
  58         double real;
  59         long long int integer;
  60         const char *string;
  61     } u;
  62 };
  63
  64 enum json_lex_state {
  65     JSON_LEX_START,             /* Not inside a token. */
  66     JSON_LEX_NUMBER,            /* Reading a number. */
  67     JSON_LEX_KEYWORD,           /* Reading a keyword. */
  68     JSON_LEX_STRING,            /* Reading a quoted string. */
  69     JSON_LEX_ESCAPE             /* In a quoted string just after a "\". */
  70 };
  71
  72 enum json_parse_state {
  73     JSON_PARSE_START,           /* Beginning of input. */
  74     JSON_PARSE_END,             /* End of input. */
  75
  76     /* Objects. */
  77     JSON_PARSE_OBJECT_INIT,     /* Expecting '}' or an object name. */
  78     JSON_PARSE_OBJECT_NAME,     /* Expecting an object name. */
  79     JSON_PARSE_OBJECT_COLON,    /* Expecting ':'. */
  80     JSON_PARSE_OBJECT_VALUE,    /* Expecting an object value. */
  81     JSON_PARSE_OBJECT_NEXT,     /* Expecting ',' or '}'. */
  82
  83     /* Arrays. */
  84     JSON_PARSE_ARRAY_INIT,      /* Expecting ']' or a value. */
  85     JSON_PARSE_ARRAY_VALUE,     /* Expecting a value. */
  86     JSON_PARSE_ARRAY_NEXT       /* Expecting ',' or ']'. */
  87 };
  88
  89 struct json_parser_node {
  90     struct json *json;
  91 };
  92
  93 /* A JSON parser. */
  94 struct json_parser {
  95     int flags;
  96
  97     /* Lexical analysis. */
  98     enum json_lex_state lex_state;
  99     struct ds buffer;           /* Buffer for accumulating token text. */
 100     int line_number;
 101     int column_number;
 102     int byte_number;
 103
 104     /* Parsing. */
 105     enum json_parse_state parse_state;
 106 #define JSON_MAX_HEIGHT 1000
 107     struct json_parser_node *stack;
 108     size_t height, allocated_height;
 109     char *member_name;
 110
 111     /* Parse status. */
 112     bool done;
 113     char *error;                /* Error message, if any, null if none yet. */
 114 };
 115
 116 static struct json *json_create(enum json_type type);
 117 static void json_parser_input(struct json_parser *, struct json_token *);
 118
 119 static void json_error(struct json_parser *p, const char *format, ...)
 120     OVS_PRINTF_FORMAT(2, 3);
 121 \f
 122 const char *
 123 json_type_to_string(enum json_type type)
 124 {
 125     switch (type) {
 126     case JSON_NULL:
 127         return "null";
 128
 129     case JSON_FALSE:
 130         return "false";
 131
 132     case JSON_TRUE:
 133         return "true";
 134
 135     case JSON_OBJECT:
 136         return "object";
 137
 138     case JSON_ARRAY:
 139         return "array";
 140
 141     case JSON_INTEGER:
 142     case JSON_REAL:
 143         return "number";
 144
 145     case JSON_STRING:
 146         return "string";
 147
 148     case JSON_N_TYPES:
 149     default:
 150         return "<invalid>";
 151     }
 152 }
 153 \f
 154 /* Functions for manipulating struct json. */
 155
 156 struct json *
 157 json_null_create(void)
 158 {
 159     return json_create(JSON_NULL);
 160 }
 161
 162 struct json *
 163 json_boolean_create(bool b)
 164 {
 165     return json_create(b ? JSON_TRUE : JSON_FALSE);
 166 }
 167
 168 struct json *
 169 json_string_create_nocopy(char *s)
 170 {
 171     struct json *json = json_create(JSON_STRING);
 172     json->u.string = s;
 173     return json;
 174 }
 175
 176 struct json *
 177 json_string_create(const char *s)
 178 {
 179     return json_string_create_nocopy(xstrdup(s));
 180 }
 181
 182 struct json *
 183 json_array_create_empty(void)
 184 {
 185     struct json *json = json_create(JSON_ARRAY);
 186     json->u.array.elems = NULL;
 187     json->u.array.n = 0;
 188     json->u.array.n_allocated = 0;
 189     return json;
 190 }
 191
 192 void
 193 json_array_add(struct json *array_, struct json *element)
 194 {
 195     struct json_array *array = json_array(array_);
 196     if (array->n >= array->n_allocated) {
 197         array->elems = x2nrealloc(array->elems, &array->n_allocated,
 198                                   sizeof *array->elems);
 199     }
 200     array->elems[array->n++] = element;
 201 }
 202
 203 void
 204 json_array_trim(struct json *array_)
 205 {
 206     struct json_array *array = json_array(array_);
 207     if (array->n < array->n_allocated){
 208         array->n_allocated = array->n;
 209         array->elems = xrealloc(array->elems, array->n * sizeof *array->elems);
 210     }
 211 }
 212
 213 struct json *
 214 json_array_create(struct json **elements, size_t n)
 215 {
 216     struct json *json = json_create(JSON_ARRAY);
 217     json->u.array.elems = elements;
 218     json->u.array.n = n;
 219     json->u.array.n_allocated = n;
 220     return json;
 221 }
 222
 223 struct json *
 224 json_array_create_1(struct json *elem0)
 225 {
 226     struct json **elems = xmalloc(sizeof *elems);
 227     elems[0] = elem0;
 228     return json_array_create(elems, 1);
 229 }
 230
 231 struct json *
 232 json_array_create_2(struct json *elem0, struct json *elem1)
 233 {
 234     struct json **elems = xmalloc(2 * sizeof *elems);
 235     elems[0] = elem0;
 236     elems[1] = elem1;
 237     return json_array_create(elems, 2);
 238 }
 239
 240 struct json *
 241 json_array_create_3(struct json *elem0, struct json *elem1, struct json *elem2)
 242 {
 243     struct json **elems = xmalloc(3 * sizeof *elems);
 244     elems[0] = elem0;
 245     elems[1] = elem1;
 246     elems[2] = elem2;
 247     return json_array_create(elems, 3);
 248 }
 249
 250 struct json *
 251 json_object_create(void)
 252 {
 253     struct json *json = json_create(JSON_OBJECT);
 254     json->u.object = xmalloc(sizeof *json->u.object);
 255     shash_init(json->u.object);
 256     return json;
 257 }
 258
 259 struct json *
 260 json_integer_create(long long int integer)
 261 {
 262     struct json *json = json_create(JSON_INTEGER);
 263     json->u.integer = integer;
 264     return json;
 265 }
 266
 267 struct json *
 268 json_real_create(double real)
 269 {
 270     struct json *json = json_create(JSON_REAL);
 271     json->u.real = real;
 272     return json;
 273 }
 274
 275 void
 276 json_object_put(struct json *json, const char *name, struct json *value)
 277 {
 278     json_destroy(shash_replace(json->u.object, name, value));
 279 }
 280
 281 void
 282 json_object_put_string(struct json *json, const char *name, const char *value)
 283 {
 284     json_object_put(json, name, json_string_create(value));
 285 }
 286
 287 const char *
 288 json_string(const struct json *json)
 289 {
 290     ovs_assert(json->type == JSON_STRING);
 291     return json->u.string;
 292 }
 293
 294 struct json_array *
 295 json_array(const struct json *json)
 296 {
 297     ovs_assert(json->type == JSON_ARRAY);
 298     return CONST_CAST(struct json_array *, &json->u.array);
 299 }
 300
 301 struct shash *
 302 json_object(const struct json *json)
 303 {
 304     ovs_assert(json->type == JSON_OBJECT);
 305     return CONST_CAST(struct shash *, json->u.object);
 306 }
 307
 308 bool
 309 json_boolean(const struct json *json)
 310 {
 311     ovs_assert(json->type == JSON_TRUE || json->type == JSON_FALSE);
 312     return json->type == JSON_TRUE;
 313 }
 314
 315 double
 316 json_real(const struct json *json)
 317 {
 318     ovs_assert(json->type == JSON_REAL || json->type == JSON_INTEGER);
 319     return json->type == JSON_REAL ? json->u.real : json->u.integer;
 320 }
 321
 322 int64_t
 323 json_integer(const struct json *json)
 324 {
 325     ovs_assert(json->type == JSON_INTEGER);
 326     return json->u.integer;
 327 }
 328 \f
 329 static void json_destroy_object(struct shash *object);
 330 static void json_destroy_array(struct json_array *array);
 331
 332 /* Frees 'json' and everything it points to, recursively. */
 333 void
 334 json_destroy(struct json *json)
 335 {
 336     if (json) {
 337         switch (json->type) {
 338         case JSON_OBJECT:
 339             json_destroy_object(json->u.object);
 340             break;
 341
 342         case JSON_ARRAY:
 343             json_destroy_array(&json->u.array);
 344             break;
 345
 346         case JSON_STRING:
 347             free(json->u.string);
 348             break;
 349
 350         case JSON_NULL:
 351         case JSON_FALSE:
 352         case JSON_TRUE:
 353         case JSON_INTEGER:
 354         case JSON_REAL:
 355             break;
 356
 357         case JSON_N_TYPES:
 358             OVS_NOT_REACHED();
 359         }
 360         free(json);
 361     }
 362 }
 363
 364 static void
 365 json_destroy_object(struct shash *object)
 366 {
 367     struct shash_node *node, *next;
 368
 369     SHASH_FOR_EACH_SAFE (node, next, object) {
 370         struct json *value = node->data;
 371
 372         json_destroy(value);
 373         shash_delete(object, node);
 374     }
 375     shash_destroy(object);
 376     free(object);
 377 }
 378
 379 static void
 380 json_destroy_array(struct json_array *array)
 381 {
 382     size_t i;
 383
 384     for (i = 0; i < array->n; i++) {
 385         json_destroy(array->elems[i]);
 386     }
 387     free(array->elems);
 388 }
 389 \f
 390 static struct json *json_clone_object(const struct shash *object);
 391 static struct json *json_clone_array(const struct json_array *array);
 392
 393 /* Returns a deep copy of 'json'. */
 394 struct json *
 395 json_clone(const struct json *json)
 396 {
 397     switch (json->type) {
 398     case JSON_OBJECT:
 399         return json_clone_object(json->u.object);
 400
 401     case JSON_ARRAY:
 402         return json_clone_array(&json->u.array);
 403
 404     case JSON_STRING:
 405         return json_string_create(json->u.string);
 406
 407     case JSON_NULL:
 408     case JSON_FALSE:
 409     case JSON_TRUE:
 410         return json_create(json->type);
 411
 412     case JSON_INTEGER:
 413         return json_integer_create(json->u.integer);
 414
 415     case JSON_REAL:
 416         return json_real_create(json->u.real);
 417
 418     case JSON_N_TYPES:
 419     default:
 420         OVS_NOT_REACHED();
 421     }
 422 }
 423
 424 static struct json *
 425 json_clone_object(const struct shash *object)
 426 {
 427     struct shash_node *node;
 428     struct json *json;
 429
 430     json = json_object_create();
 431     SHASH_FOR_EACH (node, object) {
 432         struct json *value = node->data;
 433         json_object_put(json, node->name, json_clone(value));
 434     }
 435     return json;
 436 }
 437
 438 static struct json *
 439 json_clone_array(const struct json_array *array)
 440 {
 441     struct json **elems;
 442     size_t i;
 443
 444     elems = xmalloc(array->n * sizeof *elems);
 445     for (i = 0; i < array->n; i++) {
 446         elems[i] = json_clone(array->elems[i]);
 447     }
 448     return json_array_create(elems, array->n);
 449 }
 450 \f
 451 static size_t
 452 json_hash_object(const struct shash *object, size_t basis)
 453 {
 454     const struct shash_node **nodes;
 455     size_t n, i;
 456
 457     nodes = shash_sort(object);
 458     n = shash_count(object);
 459     for (i = 0; i < n; i++) {
 460         const struct shash_node *node = nodes[i];
 461         basis = hash_string(node->name, basis);
 462         basis = json_hash(node->data, basis);
 463     }
 464     free(nodes);
 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         OVS_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         OVS_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((unsigned char) *cp)) {
 625             json_error(p, "leading zeros not allowed");
 626             return;
 627         }
 628     } else if (isdigit((unsigned char) *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((unsigned char) *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((unsigned char) *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((unsigned char) *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((unsigned char) *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((unsigned char) *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     bool ok;
 750
 751     if (cp + 4 > end) {
 752         return "quoted string ends within \\u escape";
 753     }
 754
 755     value = hexits_value(cp, 4, &ok);
 756     if (!ok) {
 757         return "malformed \\u escape";
 758     }
 759     if (!value) {
 760         return "null bytes not supported in quoted strings";
 761     }
 762     *valuep = value;
 763     return NULL;
 764 }
 765
 766 static const char *
 767 json_lex_unicode(const char *cp, const char *end, struct ds *out)
 768 {
 769     const char *error;
 770     int c0, c1;
 771
 772     error = json_lex_4hex(cp, end, &c0);
 773     if (error) {
 774         ds_clear(out);
 775         ds_put_cstr(out, error);
 776         return NULL;
 777     }
 778     cp += 4;
 779     if (!uc_is_leading_surrogate(c0)) {
 780         ds_put_utf8(out, c0);
 781         return cp;
 782     }
 783
 784     if (cp + 2 > end || *cp++ != '\\' || *cp++ != 'u') {
 785         ds_clear(out);
 786         ds_put_cstr(out, "malformed escaped surrogate pair");
 787         return NULL;
 788     }
 789
 790     error = json_lex_4hex(cp, end, &c1);
 791     if (error) {
 792         ds_clear(out);
 793         ds_put_cstr(out, error);
 794         return NULL;
 795     }
 796     cp += 4;
 797     if (!uc_is_trailing_surrogate(c1)) {
 798         ds_clear(out);
 799         ds_put_cstr(out, "second half of escaped surrogate pair is not "
 800                     "trailing surrogate");
 801         return NULL;
 802     }
 803
 804     ds_put_utf8(out, utf16_decode_surrogate_pair(c0, c1));
 805     return cp;
 806 }
 807
 808 bool
 809 json_string_unescape(const char *in, size_t in_len, char **outp)
 810 {
 811     const char *end = in + in_len;
 812     bool ok = false;
 813     struct ds out;
 814
 815     ds_init(&out);
 816     ds_reserve(&out, in_len);
 817     while (in < end) {
 818         if (*in == '"') {
 819             ds_clear(&out);
 820             ds_put_cstr(&out, "quoted string may not include unescaped \"");
 821             goto exit;
 822         }
 823         if (*in != '\\') {
 824             ds_put_char(&out, *in++);
 825             continue;
 826         }
 827
 828         in++;
 829         if (in >= end) {
 830             /* The JSON parser will never trigger this message, because its
 831              * lexer will never pass in a string that ends in a single
 832              * backslash, but json_string_unescape() has other callers that
 833              * are not as careful.*/
 834             ds_put_cstr(&out, "quoted string may not end with backslash");
 835             goto exit;
 836         }
 837         switch (*in++) {
 838         case '"': case '\\': case '/':
 839             ds_put_char(&out, in[-1]);
 840             break;
 841
 842         case 'b':
 843             ds_put_char(&out, '\b');
 844             break;
 845
 846         case 'f':
 847             ds_put_char(&out, '\f');
 848             break;
 849
 850         case 'n':
 851             ds_put_char(&out, '\n');
 852             break;
 853
 854         case 'r':
 855             ds_put_char(&out, '\r');
 856             break;
 857
 858         case 't':
 859             ds_put_char(&out, '\t');
 860             break;
 861
 862         case 'u':
 863             in = json_lex_unicode(in, end, &out);
 864             if (!in) {
 865                 goto exit;
 866             }
 867             break;
 868
 869         default:
 870             ds_clear(&out);
 871             ds_put_format(&out, "bad escape \\%c", in[-1]);
 872             goto exit;
 873         }
 874     }
 875     ok = true;
 876
 877 exit:
 878     *outp = ds_cstr(&out);
 879     return ok;
 880 }
 881
 882 void
 883 json_string_escape(const char *in, struct ds *out)
 884 {
 885     struct json json = {
 886         .type = JSON_STRING,
 887         .u.string = CONST_CAST(char *, in),
 888     };
 889     json_to_ds(&json, 0, out);
 890 }
 891
 892 static void
 893 json_parser_input_string(struct json_parser *p, const char *s)
 894 {
 895     struct json_token token;
 896
 897     token.type = T_STRING;
 898     token.u.string = s;
 899     json_parser_input(p, &token);
 900 }
 901
 902 static void
 903 json_lex_string(struct json_parser *p)
 904 {
 905     const char *raw = ds_cstr(&p->buffer);
 906     if (!strchr(raw, '\\')) {
 907         json_parser_input_string(p, raw);
 908     } else {
 909         char *cooked;
 910
 911         if (json_string_unescape(raw, strlen(raw), &cooked)) {
 912             json_parser_input_string(p, cooked);
 913         } else {
 914             json_error(p, "%s", cooked);
 915         }
 916
 917         free(cooked);
 918     }
 919 }
 920
 921 static bool
 922 json_lex_input(struct json_parser *p, unsigned char c)
 923 {
 924     struct json_token token;
 925
 926     switch (p->lex_state) {
 927     case JSON_LEX_START:
 928         switch (c) {
 929         case ' ': case '\t': case '\n': case '\r':
 930             /* Nothing to do. */
 931             return true;
 932
 933         case 'a': case 'b': case 'c': case 'd': case 'e':
 934         case 'f': case 'g': case 'h': case 'i': case 'j':
 935         case 'k': case 'l': case 'm': case 'n': case 'o':
 936         case 'p': case 'q': case 'r': case 's': case 't':
 937         case 'u': case 'v': case 'w': case 'x': case 'y':
 938         case 'z':
 939             p->lex_state = JSON_LEX_KEYWORD;
 940             break;
 941
 942         case '[': case '{': case ']': case '}': case ':': case ',':
 943             token.type = c;
 944             json_parser_input(p, &token);
 945             return true;
 946
 947         case '-':
 948         case '0': case '1': case '2': case '3': case '4':
 949         case '5': case '6': case '7': case '8': case '9':
 950             p->lex_state = JSON_LEX_NUMBER;
 951             break;
 952
 953         case '"':
 954             p->lex_state = JSON_LEX_STRING;
 955             return true;
 956
 957         default:
 958             if (isprint(c)) {
 959                 json_error(p, "invalid character '%c'", c);
 960             } else {
 961                 json_error(p, "invalid character U+%04x", c);
 962             }
 963             return true;
 964         }
 965         break;
 966
 967     case JSON_LEX_KEYWORD:
 968         if (!isalpha((unsigned char) c)) {
 969             json_lex_keyword(p);
 970             return false;
 971         }
 972         break;
 973
 974     case JSON_LEX_NUMBER:
 975         if (!strchr(".0123456789eE-+", c)) {
 976             json_lex_number(p);
 977             return false;
 978         }
 979         break;
 980
 981     case JSON_LEX_STRING:
 982         if (c == '\\') {
 983             p->lex_state = JSON_LEX_ESCAPE;
 984         } else if (c == '"') {
 985             json_lex_string(p);
 986             return true;
 987         } else if (c < 0x20) {
 988             json_error(p, "U+%04X must be escaped in quoted string", c);
 989             return true;
 990         }
 991         break;
 992
 993     case JSON_LEX_ESCAPE:
 994         p->lex_state = JSON_LEX_STRING;
 995         break;
 996
 997     default:
 998         abort();
 999     }
1000     ds_put_char(&p->buffer, c);
1001     return true;
1002 }
1003 \f
1004 /* Parsing. */
1005
1006 /* Parses 'string' as a JSON object or array and returns a newly allocated
1007  * 'struct json'.  The caller must free the returned structure with
1008  * json_destroy() when it is no longer needed.
1009  *
1010  * 'string' must be encoded in UTF-8.
1011  *
1012  * If 'string' is valid JSON, then the returned 'struct json' will be either an
1013  * object (JSON_OBJECT) or an array (JSON_ARRAY).
1014  *
1015  * If 'string' is not valid JSON, then the returned 'struct json' will be a
1016  * string (JSON_STRING) that describes the particular error encountered during
1017  * parsing.  (This is an acceptable means of error reporting because at its top
1018  * level JSON must be either an object or an array; a bare string is not
1019  * valid.) */
1020 struct json *
1021 json_from_string(const char *string)
1022 {
1023     struct json_parser *p = json_parser_create(JSPF_TRAILER);
1024     json_parser_feed(p, string, strlen(string));
1025     return json_parser_finish(p);
1026 }
1027
1028 /* Reads the file named 'file_name', parses its contents as a JSON object or
1029  * array, and returns a newly allocated 'struct json'.  The caller must free
1030  * the returned structure with json_destroy() when it is no longer needed.
1031  *
1032  * The file must be encoded in UTF-8.
1033  *
1034  * See json_from_string() for return value semantics.
1035  */
1036 struct json *
1037 json_from_file(const char *file_name)
1038 {
1039     struct json *json;
1040     FILE *stream;
1041
1042     stream = fopen(file_name, "r");
1043     if (!stream) {
1044         return json_string_create_nocopy(
1045             xasprintf("error opening \"%s\": %s", file_name,
1046                       ovs_strerror(errno)));
1047     }
1048     json = json_from_stream(stream);
1049     fclose(stream);
1050
1051     return json;
1052 }
1053
1054 /* Parses the contents of 'stream' as a JSON object or array, and returns a
1055  * newly allocated 'struct json'.  The caller must free the returned structure
1056  * with json_destroy() when it is no longer needed.
1057  *
1058  * The file must be encoded in UTF-8.
1059  *
1060  * See json_from_string() for return value semantics.
1061  */
1062 struct json *
1063 json_from_stream(FILE *stream)
1064 {
1065     struct json_parser *p;
1066     struct json *json;
1067
1068     p = json_parser_create(JSPF_TRAILER);
1069     for (;;) {
1070         char buffer[BUFSIZ];
1071         size_t n;
1072
1073         n = fread(buffer, 1, sizeof buffer, stream);
1074         if (!n || json_parser_feed(p, buffer, n) != n) {
1075             break;
1076         }
1077     }
1078     json = json_parser_finish(p);
1079
1080     if (ferror(stream)) {
1081         json_destroy(json);
1082         json = json_string_create_nocopy(
1083             xasprintf("error reading JSON stream: %s", ovs_strerror(errno)));
1084     }
1085
1086     return json;
1087 }
1088
1089 struct json_parser *
1090 json_parser_create(int flags)
1091 {
1092     struct json_parser *p = xzalloc(sizeof *p);
1093     p->flags = flags;
1094     return p;
1095 }
1096
1097 size_t
1098 json_parser_feed(struct json_parser *p, const char *input, size_t n)
1099 {
1100     size_t i;
1101     for (i = 0; !p->done && i < n; ) {
1102         if (json_lex_input(p, input[i])) {
1103             p->byte_number++;
1104             if (input[i] == '\n') {
1105                 p->column_number = 0;
1106                 p->line_number++;
1107             } else {
1108                 p->column_number++;
1109             }
1110             i++;
1111         }
1112     }
1113     return i;
1114 }
1115
1116 bool
1117 json_parser_is_done(const struct json_parser *p)
1118 {
1119     return p->done;
1120 }
1121
1122 struct json *
1123 json_parser_finish(struct json_parser *p)
1124 {
1125     struct json *json;
1126
1127     switch (p->lex_state) {
1128     case JSON_LEX_START:
1129         break;
1130
1131     case JSON_LEX_STRING:
1132     case JSON_LEX_ESCAPE:
1133         json_error(p, "unexpected end of input in quoted string");
1134         break;
1135
1136     case JSON_LEX_NUMBER:
1137     case JSON_LEX_KEYWORD:
1138         json_lex_input(p, ' ');
1139         break;
1140     }
1141
1142     if (p->parse_state == JSON_PARSE_START) {
1143         json_error(p, "empty input stream");
1144     } else if (p->parse_state != JSON_PARSE_END) {
1145         json_error(p, "unexpected end of input");
1146     }
1147
1148     if (!p->error) {
1149         ovs_assert(p->height == 1);
1150         ovs_assert(p->stack[0].json != NULL);
1151         json = p->stack[--p->height].json;
1152     } else {
1153         json = json_string_create_nocopy(p->error);
1154         p->error = NULL;
1155     }
1156
1157     json_parser_abort(p);
1158
1159     return json;
1160 }
1161
1162 void
1163 json_parser_abort(struct json_parser *p)
1164 {
1165     if (p) {
1166         ds_destroy(&p->buffer);
1167         if (p->height) {
1168             json_destroy(p->stack[0].json);
1169         }
1170         free(p->stack);
1171         free(p->member_name);
1172         free(p->error);
1173         free(p);
1174     }
1175 }
1176
1177 static struct json_parser_node *
1178 json_parser_top(struct json_parser *p)
1179 {
1180     return &p->stack[p->height - 1];
1181 }
1182
1183 static void
1184 json_parser_put_value(struct json_parser *p, struct json *value)
1185 {
1186     struct json_parser_node *node = json_parser_top(p);
1187     if (node->json->type == JSON_OBJECT) {
1188         json_object_put(node->json, p->member_name, value);
1189         free(p->member_name);
1190         p->member_name = NULL;
1191     } else if (node->json->type == JSON_ARRAY) {
1192         json_array_add(node->json, value);
1193     } else {
1194         OVS_NOT_REACHED();
1195     }
1196 }
1197
1198 static void
1199 json_parser_push(struct json_parser *p,
1200                  struct json *new_json, enum json_parse_state new_state)
1201 {
1202     if (p->height < JSON_MAX_HEIGHT) {
1203         struct json_parser_node *node;
1204
1205         if (p->height >= p->allocated_height) {
1206             p->stack = x2nrealloc(p->stack, &p->allocated_height,
1207                                   sizeof *p->stack);
1208         }
1209
1210         if (p->height > 0) {
1211             json_parser_put_value(p, new_json);
1212         }
1213
1214         node = &p->stack[p->height++];
1215         node->json = new_json;
1216         p->parse_state = new_state;
1217     } else {
1218         json_destroy(new_json);
1219         json_error(p, "input exceeds maximum nesting depth %d",
1220                    JSON_MAX_HEIGHT);
1221     }
1222 }
1223
1224 static void
1225 json_parser_push_object(struct json_parser *p)
1226 {
1227     json_parser_push(p, json_object_create(), JSON_PARSE_OBJECT_INIT);
1228 }
1229
1230 static void
1231 json_parser_push_array(struct json_parser *p)
1232 {
1233     json_parser_push(p, json_array_create_empty(), JSON_PARSE_ARRAY_INIT);
1234 }
1235
1236 static void
1237 json_parse_value(struct json_parser *p, struct json_token *token,
1238                  enum json_parse_state next_state)
1239 {
1240     struct json *value;
1241
1242     switch (token->type) {
1243     case T_FALSE:
1244         value = json_boolean_create(false);
1245         break;
1246
1247     case T_NULL:
1248         value = json_null_create();
1249         break;
1250
1251     case T_TRUE:
1252         value = json_boolean_create(true);
1253         break;
1254
1255     case '{':
1256         json_parser_push_object(p);
1257         return;
1258
1259     case '[':
1260         json_parser_push_array(p);
1261         return;
1262
1263     case T_INTEGER:
1264         value = json_integer_create(token->u.integer);
1265         break;
1266
1267     case T_REAL:
1268         value = json_real_create(token->u.real);
1269         break;
1270
1271     case T_STRING:
1272         value = json_string_create(token->u.string);
1273         break;
1274
1275     case T_EOF:
1276     case '}':
1277     case ']':
1278     case ':':
1279     case ',':
1280     default:
1281         json_error(p, "syntax error expecting value");
1282         return;
1283     }
1284
1285     json_parser_put_value(p, value);
1286     p->parse_state = next_state;
1287 }
1288
1289 static void
1290 json_parser_pop(struct json_parser *p)
1291 {
1292     struct json_parser_node *node;
1293
1294     /* Conserve memory. */
1295     node = json_parser_top(p);
1296     if (node->json->type == JSON_ARRAY) {
1297         json_array_trim(node->json);
1298     }
1299
1300     /* Pop off the top-of-stack. */
1301     if (p->height == 1) {
1302         p->parse_state = JSON_PARSE_END;
1303         if (!(p->flags & JSPF_TRAILER)) {
1304             p->done = true;
1305         }
1306     } else {
1307         p->height--;
1308         node = json_parser_top(p);
1309         if (node->json->type == JSON_ARRAY) {
1310             p->parse_state = JSON_PARSE_ARRAY_NEXT;
1311         } else if (node->json->type == JSON_OBJECT) {
1312             p->parse_state = JSON_PARSE_OBJECT_NEXT;
1313         } else {
1314             OVS_NOT_REACHED();
1315         }
1316     }
1317 }
1318
1319 static void
1320 json_parser_input(struct json_parser *p, struct json_token *token)
1321 {
1322     switch (p->parse_state) {
1323     case JSON_PARSE_START:
1324         if (token->type == '{') {
1325             json_parser_push_object(p);
1326         } else if (token->type == '[') {
1327             json_parser_push_array(p);
1328         } else {
1329             json_error(p, "syntax error at beginning of input");
1330         }
1331         break;
1332
1333     case JSON_PARSE_END:
1334         json_error(p, "trailing garbage at end of input");
1335         break;
1336
1337     case JSON_PARSE_OBJECT_INIT:
1338         if (token->type == '}') {
1339             json_parser_pop(p);
1340             break;
1341         }
1342         /* Fall through. */
1343     case JSON_PARSE_OBJECT_NAME:
1344         if (token->type == T_STRING) {
1345             p->member_name = xstrdup(token->u.string);
1346             p->parse_state = JSON_PARSE_OBJECT_COLON;
1347         } else {
1348             json_error(p, "syntax error parsing object expecting string");
1349         }
1350         break;
1351
1352     case JSON_PARSE_OBJECT_COLON:
1353         if (token->type == ':') {
1354             p->parse_state = JSON_PARSE_OBJECT_VALUE;
1355         } else {
1356             json_error(p, "syntax error parsing object expecting ':'");
1357         }
1358         break;
1359
1360     case JSON_PARSE_OBJECT_VALUE:
1361         json_parse_value(p, token, JSON_PARSE_OBJECT_NEXT);
1362         break;
1363
1364     case JSON_PARSE_OBJECT_NEXT:
1365         if (token->type == ',') {
1366             p->parse_state = JSON_PARSE_OBJECT_NAME;
1367         } else if (token->type == '}') {
1368             json_parser_pop(p);
1369         } else {
1370             json_error(p, "syntax error expecting '}' or ','");
1371         }
1372         break;
1373
1374     case JSON_PARSE_ARRAY_INIT:
1375         if (token->type == ']') {
1376             json_parser_pop(p);
1377             break;
1378         }
1379         /* Fall through. */
1380     case JSON_PARSE_ARRAY_VALUE:
1381         json_parse_value(p, token, JSON_PARSE_ARRAY_NEXT);
1382         break;
1383
1384     case JSON_PARSE_ARRAY_NEXT:
1385         if (token->type == ',') {
1386             p->parse_state = JSON_PARSE_ARRAY_VALUE;
1387         } else if (token->type == ']') {
1388             json_parser_pop(p);
1389         } else {
1390             json_error(p, "syntax error expecting ']' or ','");
1391         }
1392         break;
1393
1394     default:
1395         abort();
1396     }
1397
1398     p->lex_state = JSON_LEX_START;
1399     ds_clear(&p->buffer);
1400 }
1401
1402 static struct json *
1403 json_create(enum json_type type)
1404 {
1405     struct json *json = xmalloc(sizeof *json);
1406     json->type = type;
1407     return json;
1408 }
1409
1410 static void
1411 json_error(struct json_parser *p, const char *format, ...)
1412 {
1413     if (!p->error) {
1414         struct ds msg;
1415         va_list args;
1416
1417         ds_init(&msg);
1418         ds_put_format(&msg, "line %d, column %d, byte %d: ",
1419                       p->line_number, p->column_number, p->byte_number);
1420         va_start(args, format);
1421         ds_put_format_valist(&msg, format, args);
1422         va_end(args);
1423
1424         p->error = ds_steal_cstr(&msg);
1425
1426         p->done = true;
1427     }
1428 }
1429 \f
1430 #define SPACES_PER_LEVEL 2
1431
1432 struct json_serializer {
1433     struct ds *ds;
1434     int depth;
1435     int flags;
1436 };
1437
1438 static void json_serialize(const struct json *, struct json_serializer *);
1439 static void json_serialize_object(const struct shash *object,
1440                                   struct json_serializer *);
1441 static void json_serialize_array(const struct json_array *,
1442                                  struct json_serializer *);
1443 static void json_serialize_string(const char *, struct ds *);
1444
1445 /* Converts 'json' to a string in JSON format, encoded in UTF-8, and returns
1446  * that string.  The caller is responsible for freeing the returned string,
1447  * with free(), when it is no longer needed.
1448  *
1449  * If 'flags' contains JSSF_PRETTY, the output is pretty-printed with each
1450  * nesting level introducing an additional indentation.  Otherwise, the
1451  * returned string does not contain any new-line characters.
1452  *
1453  * If 'flags' contains JSSF_SORT, members of objects in the output are sorted
1454  * in bytewise lexicographic order for reproducibility.  Otherwise, members of
1455  * objects are output in an indeterminate order.
1456  *
1457  * The returned string is valid JSON only if 'json' represents an array or an
1458  * object, since a bare literal does not satisfy the JSON grammar. */
1459 char *
1460 json_to_string(const struct json *json, int flags)
1461 {
1462     struct ds ds;
1463
1464     ds_init(&ds);
1465     json_to_ds(json, flags, &ds);
1466     return ds_steal_cstr(&ds);
1467 }
1468
1469 /* Same as json_to_string(), but the output is appended to 'ds'. */
1470 void
1471 json_to_ds(const struct json *json, int flags, struct ds *ds)
1472 {
1473     struct json_serializer s;
1474
1475     s.ds = ds;
1476     s.depth = 0;
1477     s.flags = flags;
1478     json_serialize(json, &s);
1479 }
1480
1481 static void
1482 json_serialize(const struct json *json, struct json_serializer *s)
1483 {
1484     struct ds *ds = s->ds;
1485
1486     switch (json->type) {
1487     case JSON_NULL:
1488         ds_put_cstr(ds, "null");
1489         break;
1490
1491     case JSON_FALSE:
1492         ds_put_cstr(ds, "false");
1493         break;
1494
1495     case JSON_TRUE:
1496         ds_put_cstr(ds, "true");
1497         break;
1498
1499     case JSON_OBJECT:
1500         json_serialize_object(json->u.object, s);
1501         break;
1502
1503     case JSON_ARRAY:
1504         json_serialize_array(&json->u.array, s);
1505         break;
1506
1507     case JSON_INTEGER:
1508         ds_put_format(ds, "%lld", json->u.integer);
1509         break;
1510
1511     case JSON_REAL:
1512         ds_put_format(ds, "%.*g", DBL_DIG, json->u.real);
1513         break;
1514
1515     case JSON_STRING:
1516         json_serialize_string(json->u.string, ds);
1517         break;
1518
1519     case JSON_N_TYPES:
1520     default:
1521         OVS_NOT_REACHED();
1522     }
1523 }
1524
1525 static void
1526 indent_line(struct json_serializer *s)
1527 {
1528     if (s->flags & JSSF_PRETTY) {
1529         ds_put_char(s->ds, '\n');
1530         ds_put_char_multiple(s->ds, ' ', SPACES_PER_LEVEL * s->depth);
1531     }
1532 }
1533
1534 static void
1535 json_serialize_object_member(size_t i, const struct shash_node *node,
1536                              struct json_serializer *s)
1537 {
1538     struct ds *ds = s->ds;
1539
1540     if (i) {
1541         ds_put_char(ds, ',');
1542         indent_line(s);
1543     }
1544
1545     json_serialize_string(node->name, ds);
1546     ds_put_char(ds, ':');
1547     if (s->flags & JSSF_PRETTY) {
1548         ds_put_char(ds, ' ');
1549     }
1550     json_serialize(node->data, s);
1551 }
1552
1553 static void
1554 json_serialize_object(const struct shash *object, struct json_serializer *s)
1555 {
1556     struct ds *ds = s->ds;
1557
1558     ds_put_char(ds, '{');
1559
1560     s->depth++;
1561     indent_line(s);
1562
1563     if (s->flags & JSSF_SORT) {
1564         const struct shash_node **nodes;
1565         size_t n, i;
1566
1567         nodes = shash_sort(object);
1568         n = shash_count(object);
1569         for (i = 0; i < n; i++) {
1570             json_serialize_object_member(i, nodes[i], s);
1571         }
1572         free(nodes);
1573     } else {
1574         struct shash_node *node;
1575         size_t i;
1576
1577         i = 0;
1578         SHASH_FOR_EACH (node, object) {
1579             json_serialize_object_member(i++, node, s);
1580         }
1581     }
1582
1583     ds_put_char(ds, '}');
1584     s->depth--;
1585 }
1586
1587 static void
1588 json_serialize_array(const struct json_array *array, struct json_serializer *s)
1589 {
1590     struct ds *ds = s->ds;
1591     size_t i;
1592
1593     ds_put_char(ds, '[');
1594     s->depth++;
1595
1596     if (array->n > 0) {
1597         indent_line(s);
1598
1599         for (i = 0; i < array->n; i++) {
1600             if (i) {
1601                 ds_put_char(ds, ',');
1602                 indent_line(s);
1603             }
1604             json_serialize(array->elems[i], s);
1605         }
1606     }
1607
1608     s->depth--;
1609     ds_put_char(ds, ']');
1610 }
1611
1612 static void
1613 json_serialize_string(const char *string, struct ds *ds)
1614 {
1615     uint8_t c;
1616
1617     ds_put_char(ds, '"');
1618     while ((c = *string++) != '\0') {
1619         switch (c) {
1620         case '"':
1621             ds_put_cstr(ds, "\\\"");
1622             break;
1623
1624         case '\\':
1625             ds_put_cstr(ds, "\\\\");
1626             break;
1627
1628         case '\b':
1629             ds_put_cstr(ds, "\\b");
1630             break;
1631
1632         case '\f':
1633             ds_put_cstr(ds, "\\f");
1634             break;
1635
1636         case '\n':
1637             ds_put_cstr(ds, "\\n");
1638             break;
1639
1640         case '\r':
1641             ds_put_cstr(ds, "\\r");
1642             break;
1643
1644         case '\t':
1645             ds_put_cstr(ds, "\\t");
1646             break;
1647
1648         default:
1649             if (c >= 32) {
1650                 ds_put_char(ds, c);
1651             } else {
1652                 ds_put_format(ds, "\\u%04x", c);
1653             }
1654             break;
1655         }
1656     }
1657     ds_put_char(ds, '"');
1658 }