Thierry Strudel | 62874b3 | 2017-02-23 15:43:51 -0800 | [diff] [blame] | 1 | /* |
| 2 | * $Id: json_tokener.c,v 1.20 2006/07/25 03:24:50 mclark Exp $ |
| 3 | * |
| 4 | * Copyright (c) 2004, 2005 Metaparadigm Pte. Ltd. |
| 5 | * Michael Clark <michael@metaparadigm.com> |
| 6 | * |
| 7 | * This library is free software; you can redistribute it and/or modify |
| 8 | * it under the terms of the MIT license. See COPYING for details. |
| 9 | * |
| 10 | * |
| 11 | * Copyright (c) 2008-2009 Yahoo! Inc. All rights reserved. |
| 12 | * The copyrights to the contents of this file are licensed under the MIT License |
| 13 | * (http://www.opensource.org/licenses/mit-license.php) |
| 14 | */ |
| 15 | |
| 16 | #include "config.h" |
| 17 | |
| 18 | #include <math.h> |
| 19 | #include <stdio.h> |
| 20 | #include <stdlib.h> |
| 21 | #include <stddef.h> |
| 22 | #include <ctype.h> |
| 23 | #include <string.h> |
| 24 | #include <limits.h> |
| 25 | |
| 26 | #include "debug.h" |
| 27 | #include "printbuf.h" |
| 28 | #include "arraylist.h" |
| 29 | #include "json_inttypes.h" |
| 30 | #include "json_object.h" |
| 31 | #include "json_tokener.h" |
| 32 | #include "json_util.h" |
| 33 | |
| 34 | #ifdef HAVE_LOCALE_H |
| 35 | #include <locale.h> |
| 36 | #endif /* HAVE_LOCALE_H */ |
| 37 | |
| 38 | #define jt_hexdigit(x) (((x) <= '9') ? (x) - '0' : ((x) & 7) + 9) |
| 39 | |
| 40 | #if !HAVE_STRDUP && defined(_MSC_VER) |
| 41 | /* MSC has the version as _strdup */ |
| 42 | # define strdup _strdup |
| 43 | #elif !HAVE_STRDUP |
| 44 | # error You do not have strdup on your system. |
| 45 | #endif /* HAVE_STRDUP */ |
| 46 | |
| 47 | #if !HAVE_STRNCASECMP && defined(_MSC_VER) |
| 48 | /* MSC has the version as _strnicmp */ |
| 49 | # define strncasecmp _strnicmp |
| 50 | #elif !HAVE_STRNCASECMP |
| 51 | # error You do not have strncasecmp on your system. |
| 52 | #endif /* HAVE_STRNCASECMP */ |
| 53 | |
| 54 | /* Use C99 NAN by default; if not available, nan("") should work too. */ |
| 55 | #ifndef NAN |
| 56 | #define NAN nan("") |
| 57 | #endif /* !NAN */ |
| 58 | |
| 59 | static const char json_null_str[] = "null"; |
| 60 | static const int json_null_str_len = sizeof(json_null_str) - 1; |
| 61 | static const char json_inf_str[] = "Infinity"; |
| 62 | static const int json_inf_str_len = sizeof(json_inf_str) - 1; |
| 63 | static const char json_nan_str[] = "NaN"; |
| 64 | static const int json_nan_str_len = sizeof(json_nan_str) - 1; |
| 65 | static const char json_true_str[] = "true"; |
| 66 | static const int json_true_str_len = sizeof(json_true_str) - 1; |
| 67 | static const char json_false_str[] = "false"; |
| 68 | static const int json_false_str_len = sizeof(json_false_str) - 1; |
| 69 | |
| 70 | static const char* json_tokener_errors[] = { |
| 71 | "success", |
| 72 | "continue", |
| 73 | "nesting too deep", |
| 74 | "unexpected end of data", |
| 75 | "unexpected character", |
| 76 | "null expected", |
| 77 | "boolean expected", |
| 78 | "number expected", |
| 79 | "array value separator ',' expected", |
| 80 | "quoted object property name expected", |
| 81 | "object property name separator ':' expected", |
| 82 | "object value separator ',' expected", |
| 83 | "invalid string sequence", |
| 84 | "expected comment", |
| 85 | "buffer size overflow" |
| 86 | }; |
| 87 | |
| 88 | const char *json_tokener_error_desc(enum json_tokener_error jerr) |
| 89 | { |
| 90 | int jerr_int = (int)jerr; |
| 91 | if (jerr_int < 0 || jerr_int >= (int)(sizeof(json_tokener_errors) / sizeof(json_tokener_errors[0]))) |
| 92 | return "Unknown error, invalid json_tokener_error value passed to json_tokener_error_desc()"; |
| 93 | return json_tokener_errors[jerr]; |
| 94 | } |
| 95 | |
| 96 | enum json_tokener_error json_tokener_get_error(json_tokener *tok) |
| 97 | { |
| 98 | return tok->err; |
| 99 | } |
| 100 | |
| 101 | /* Stuff for decoding unicode sequences */ |
| 102 | #define IS_HIGH_SURROGATE(uc) (((uc) & 0xFC00) == 0xD800) |
| 103 | #define IS_LOW_SURROGATE(uc) (((uc) & 0xFC00) == 0xDC00) |
| 104 | #define DECODE_SURROGATE_PAIR(hi,lo) ((((hi) & 0x3FF) << 10) + ((lo) & 0x3FF) + 0x10000) |
| 105 | static unsigned char utf8_replacement_char[3] = { 0xEF, 0xBF, 0xBD }; |
| 106 | |
| 107 | struct json_tokener* json_tokener_new_ex(int depth) |
| 108 | { |
| 109 | struct json_tokener *tok; |
| 110 | |
| 111 | tok = (struct json_tokener*)calloc(1, sizeof(struct json_tokener)); |
| 112 | if (!tok) return NULL; |
| 113 | tok->stack = (struct json_tokener_srec *)calloc(depth, sizeof(struct json_tokener_srec)); |
| 114 | if (!tok->stack) { |
| 115 | free(tok); |
| 116 | return NULL; |
| 117 | } |
| 118 | tok->pb = printbuf_new(); |
| 119 | tok->max_depth = depth; |
| 120 | json_tokener_reset(tok); |
| 121 | return tok; |
| 122 | } |
| 123 | |
| 124 | struct json_tokener* json_tokener_new(void) |
| 125 | { |
| 126 | return json_tokener_new_ex(JSON_TOKENER_DEFAULT_DEPTH); |
| 127 | } |
| 128 | |
| 129 | void json_tokener_free(struct json_tokener *tok) |
| 130 | { |
| 131 | json_tokener_reset(tok); |
| 132 | if (tok->pb) printbuf_free(tok->pb); |
| 133 | if (tok->stack) free(tok->stack); |
| 134 | free(tok); |
| 135 | } |
| 136 | |
| 137 | static void json_tokener_reset_level(struct json_tokener *tok, int depth) |
| 138 | { |
| 139 | tok->stack[depth].state = json_tokener_state_eatws; |
| 140 | tok->stack[depth].saved_state = json_tokener_state_start; |
| 141 | json_object_put(tok->stack[depth].current); |
| 142 | tok->stack[depth].current = NULL; |
| 143 | free(tok->stack[depth].obj_field_name); |
| 144 | tok->stack[depth].obj_field_name = NULL; |
| 145 | } |
| 146 | |
| 147 | void json_tokener_reset(struct json_tokener *tok) |
| 148 | { |
| 149 | int i; |
| 150 | if (!tok) |
| 151 | return; |
| 152 | |
| 153 | for(i = tok->depth; i >= 0; i--) |
| 154 | json_tokener_reset_level(tok, i); |
| 155 | tok->depth = 0; |
| 156 | tok->err = json_tokener_success; |
| 157 | } |
| 158 | |
| 159 | struct json_object* json_tokener_parse(const char *str) |
| 160 | { |
| 161 | enum json_tokener_error jerr_ignored; |
| 162 | struct json_object* obj; |
| 163 | obj = json_tokener_parse_verbose(str, &jerr_ignored); |
| 164 | return obj; |
| 165 | } |
| 166 | |
| 167 | struct json_object* json_tokener_parse_verbose(const char *str, enum json_tokener_error *error) |
| 168 | { |
| 169 | struct json_tokener* tok; |
| 170 | struct json_object* obj; |
| 171 | |
| 172 | tok = json_tokener_new(); |
| 173 | if (!tok) |
| 174 | return NULL; |
| 175 | obj = json_tokener_parse_ex(tok, str, -1); |
| 176 | *error = tok->err; |
| 177 | if(tok->err != json_tokener_success) { |
| 178 | if (obj != NULL) |
| 179 | json_object_put(obj); |
| 180 | obj = NULL; |
| 181 | } |
| 182 | |
| 183 | json_tokener_free(tok); |
| 184 | return obj; |
| 185 | } |
| 186 | |
| 187 | #define state tok->stack[tok->depth].state |
| 188 | #define saved_state tok->stack[tok->depth].saved_state |
| 189 | #define current tok->stack[tok->depth].current |
| 190 | #define obj_field_name tok->stack[tok->depth].obj_field_name |
| 191 | |
| 192 | /* Optimization: |
| 193 | * json_tokener_parse_ex() consumed a lot of CPU in its main loop, |
| 194 | * iterating character-by character. A large performance boost is |
| 195 | * achieved by using tighter loops to locally handle units such as |
| 196 | * comments and strings. Loops that handle an entire token within |
| 197 | * their scope also gather entire strings and pass them to |
| 198 | * printbuf_memappend() in a single call, rather than calling |
| 199 | * printbuf_memappend() one char at a time. |
| 200 | * |
| 201 | * PEEK_CHAR() and ADVANCE_CHAR() macros are used for code that is |
| 202 | * common to both the main loop and the tighter loops. |
| 203 | */ |
| 204 | |
| 205 | /* PEEK_CHAR(dest, tok) macro: |
| 206 | * Peeks at the current char and stores it in dest. |
| 207 | * Returns 1 on success, sets tok->err and returns 0 if no more chars. |
| 208 | * Implicit inputs: str, len vars |
| 209 | */ |
| 210 | #define PEEK_CHAR(dest, tok) \ |
| 211 | (((tok)->char_offset == len) ? \ |
| 212 | (((tok)->depth == 0 && state == json_tokener_state_eatws && saved_state == json_tokener_state_finish) ? \ |
| 213 | (((tok)->err = json_tokener_success), 0) \ |
| 214 | : \ |
| 215 | (((tok)->err = json_tokener_continue), 0) \ |
| 216 | ) : \ |
| 217 | (((dest) = *str), 1) \ |
| 218 | ) |
| 219 | |
| 220 | /* ADVANCE_CHAR() macro: |
| 221 | * Incrementes str & tok->char_offset. |
| 222 | * For convenience of existing conditionals, returns the old value of c (0 on eof) |
| 223 | * Implicit inputs: c var |
| 224 | */ |
| 225 | #define ADVANCE_CHAR(str, tok) \ |
| 226 | ( ++(str), ((tok)->char_offset)++, c) |
| 227 | |
| 228 | |
| 229 | /* End optimization macro defs */ |
| 230 | |
| 231 | |
| 232 | struct json_object* json_tokener_parse_ex(struct json_tokener *tok, |
| 233 | const char *str, int len) |
| 234 | { |
| 235 | struct json_object *obj = NULL; |
| 236 | char c = '\1'; |
| 237 | #ifdef HAVE_SETLOCALE |
| 238 | char *oldlocale=NULL, *tmplocale; |
| 239 | |
| 240 | tmplocale = setlocale(LC_NUMERIC, NULL); |
| 241 | if (tmplocale) oldlocale = strdup(tmplocale); |
| 242 | setlocale(LC_NUMERIC, "C"); |
| 243 | #endif |
| 244 | |
| 245 | tok->char_offset = 0; |
| 246 | tok->err = json_tokener_success; |
| 247 | |
| 248 | /* this interface is presently not 64-bit clean due to the int len argument |
| 249 | and the internal printbuf interface that takes 32-bit int len arguments |
| 250 | so the function limits the maximum string size to INT32_MAX (2GB). |
| 251 | If the function is called with len == -1 then strlen is called to check |
| 252 | the string length is less than INT32_MAX (2GB) */ |
| 253 | if ((len < -1) || (len == -1 && strlen(str) > INT32_MAX)) { |
| 254 | tok->err = json_tokener_error_size; |
| 255 | return NULL; |
| 256 | } |
| 257 | |
| 258 | while (PEEK_CHAR(c, tok)) { |
| 259 | |
| 260 | redo_char: |
| 261 | switch(state) { |
| 262 | |
| 263 | case json_tokener_state_eatws: |
| 264 | /* Advance until we change state */ |
| 265 | while (isspace((int)c)) { |
| 266 | if ((!ADVANCE_CHAR(str, tok)) || (!PEEK_CHAR(c, tok))) |
| 267 | goto out; |
| 268 | } |
| 269 | if(c == '/' && !(tok->flags & JSON_TOKENER_STRICT)) { |
| 270 | printbuf_reset(tok->pb); |
| 271 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 272 | state = json_tokener_state_comment_start; |
| 273 | } else { |
| 274 | state = saved_state; |
| 275 | goto redo_char; |
| 276 | } |
| 277 | break; |
| 278 | |
| 279 | case json_tokener_state_start: |
| 280 | switch(c) { |
| 281 | case '{': |
| 282 | state = json_tokener_state_eatws; |
| 283 | saved_state = json_tokener_state_object_field_start; |
| 284 | current = json_object_new_object(); |
| 285 | break; |
| 286 | case '[': |
| 287 | state = json_tokener_state_eatws; |
| 288 | saved_state = json_tokener_state_array; |
| 289 | current = json_object_new_array(); |
| 290 | break; |
| 291 | case 'I': |
| 292 | case 'i': |
| 293 | state = json_tokener_state_inf; |
| 294 | printbuf_reset(tok->pb); |
| 295 | tok->st_pos = 0; |
| 296 | goto redo_char; |
| 297 | case 'N': |
| 298 | case 'n': |
| 299 | state = json_tokener_state_null; // or NaN |
| 300 | printbuf_reset(tok->pb); |
| 301 | tok->st_pos = 0; |
| 302 | goto redo_char; |
| 303 | case '\'': |
| 304 | if (tok->flags & JSON_TOKENER_STRICT) { |
| 305 | /* in STRICT mode only double-quote are allowed */ |
| 306 | tok->err = json_tokener_error_parse_unexpected; |
| 307 | goto out; |
| 308 | } |
| 309 | case '"': |
| 310 | state = json_tokener_state_string; |
| 311 | printbuf_reset(tok->pb); |
| 312 | tok->quote_char = c; |
| 313 | break; |
| 314 | case 'T': |
| 315 | case 't': |
| 316 | case 'F': |
| 317 | case 'f': |
| 318 | state = json_tokener_state_boolean; |
| 319 | printbuf_reset(tok->pb); |
| 320 | tok->st_pos = 0; |
| 321 | goto redo_char; |
| 322 | #if defined(__GNUC__) |
| 323 | case '0' ... '9': |
| 324 | #else |
| 325 | case '0': |
| 326 | case '1': |
| 327 | case '2': |
| 328 | case '3': |
| 329 | case '4': |
| 330 | case '5': |
| 331 | case '6': |
| 332 | case '7': |
| 333 | case '8': |
| 334 | case '9': |
| 335 | #endif |
| 336 | case '-': |
| 337 | state = json_tokener_state_number; |
| 338 | printbuf_reset(tok->pb); |
| 339 | tok->is_double = 0; |
| 340 | goto redo_char; |
| 341 | default: |
| 342 | tok->err = json_tokener_error_parse_unexpected; |
| 343 | goto out; |
| 344 | } |
| 345 | break; |
| 346 | |
| 347 | case json_tokener_state_finish: |
| 348 | if(tok->depth == 0) goto out; |
| 349 | obj = json_object_get(current); |
| 350 | json_tokener_reset_level(tok, tok->depth); |
| 351 | tok->depth--; |
| 352 | goto redo_char; |
| 353 | |
| 354 | case json_tokener_state_inf: /* aka starts with 'i' */ |
| 355 | { |
| 356 | int size_inf; |
| 357 | int is_negative = 0; |
| 358 | |
| 359 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 360 | size_inf = json_min(tok->st_pos+1, json_inf_str_len); |
| 361 | char *infbuf = tok->pb->buf; |
| 362 | if (*infbuf == '-') |
| 363 | { |
| 364 | infbuf++; |
| 365 | is_negative = 1; |
| 366 | } |
| 367 | if ((!(tok->flags & JSON_TOKENER_STRICT) && |
| 368 | strncasecmp(json_inf_str, infbuf, size_inf) == 0) || |
| 369 | (strncmp(json_inf_str, infbuf, size_inf) == 0) |
| 370 | ) |
| 371 | { |
| 372 | if (tok->st_pos == json_inf_str_len) |
| 373 | { |
| 374 | current = json_object_new_double(is_negative ? -INFINITY : INFINITY); |
| 375 | saved_state = json_tokener_state_finish; |
| 376 | state = json_tokener_state_eatws; |
| 377 | goto redo_char; |
| 378 | } |
| 379 | } else { |
| 380 | tok->err = json_tokener_error_parse_unexpected; |
| 381 | goto out; |
| 382 | } |
| 383 | tok->st_pos++; |
| 384 | } |
| 385 | break; |
| 386 | case json_tokener_state_null: /* aka starts with 'n' */ |
| 387 | { |
| 388 | int size; |
| 389 | int size_nan; |
| 390 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 391 | size = json_min(tok->st_pos+1, json_null_str_len); |
| 392 | size_nan = json_min(tok->st_pos+1, json_nan_str_len); |
| 393 | if((!(tok->flags & JSON_TOKENER_STRICT) && |
| 394 | strncasecmp(json_null_str, tok->pb->buf, size) == 0) |
| 395 | || (strncmp(json_null_str, tok->pb->buf, size) == 0) |
| 396 | ) { |
| 397 | if (tok->st_pos == json_null_str_len) { |
| 398 | current = NULL; |
| 399 | saved_state = json_tokener_state_finish; |
| 400 | state = json_tokener_state_eatws; |
| 401 | goto redo_char; |
| 402 | } |
| 403 | } |
| 404 | else if ((!(tok->flags & JSON_TOKENER_STRICT) && |
| 405 | strncasecmp(json_nan_str, tok->pb->buf, size_nan) == 0) || |
| 406 | (strncmp(json_nan_str, tok->pb->buf, size_nan) == 0) |
| 407 | ) |
| 408 | { |
| 409 | if (tok->st_pos == json_nan_str_len) |
| 410 | { |
| 411 | current = json_object_new_double(NAN); |
| 412 | saved_state = json_tokener_state_finish; |
| 413 | state = json_tokener_state_eatws; |
| 414 | goto redo_char; |
| 415 | } |
| 416 | } else { |
| 417 | tok->err = json_tokener_error_parse_null; |
| 418 | goto out; |
| 419 | } |
| 420 | tok->st_pos++; |
| 421 | } |
| 422 | break; |
| 423 | |
| 424 | case json_tokener_state_comment_start: |
| 425 | if(c == '*') { |
| 426 | state = json_tokener_state_comment; |
| 427 | } else if(c == '/') { |
| 428 | state = json_tokener_state_comment_eol; |
| 429 | } else { |
| 430 | tok->err = json_tokener_error_parse_comment; |
| 431 | goto out; |
| 432 | } |
| 433 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 434 | break; |
| 435 | |
| 436 | case json_tokener_state_comment: |
| 437 | { |
| 438 | /* Advance until we change state */ |
| 439 | const char *case_start = str; |
| 440 | while(c != '*') { |
| 441 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 442 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 443 | goto out; |
| 444 | } |
| 445 | } |
| 446 | printbuf_memappend_fast(tok->pb, case_start, 1+str-case_start); |
| 447 | state = json_tokener_state_comment_end; |
| 448 | } |
| 449 | break; |
| 450 | |
| 451 | case json_tokener_state_comment_eol: |
| 452 | { |
| 453 | /* Advance until we change state */ |
| 454 | const char *case_start = str; |
| 455 | while(c != '\n') { |
| 456 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 457 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 458 | goto out; |
| 459 | } |
| 460 | } |
| 461 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 462 | MC_DEBUG("json_tokener_comment: %s\n", tok->pb->buf); |
| 463 | state = json_tokener_state_eatws; |
| 464 | } |
| 465 | break; |
| 466 | |
| 467 | case json_tokener_state_comment_end: |
| 468 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 469 | if(c == '/') { |
| 470 | MC_DEBUG("json_tokener_comment: %s\n", tok->pb->buf); |
| 471 | state = json_tokener_state_eatws; |
| 472 | } else { |
| 473 | state = json_tokener_state_comment; |
| 474 | } |
| 475 | break; |
| 476 | |
| 477 | case json_tokener_state_string: |
| 478 | { |
| 479 | /* Advance until we change state */ |
| 480 | const char *case_start = str; |
| 481 | while(1) { |
| 482 | if(c == tok->quote_char) { |
| 483 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 484 | current = json_object_new_string_len(tok->pb->buf, tok->pb->bpos); |
| 485 | saved_state = json_tokener_state_finish; |
| 486 | state = json_tokener_state_eatws; |
| 487 | break; |
| 488 | } else if(c == '\\') { |
| 489 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 490 | saved_state = json_tokener_state_string; |
| 491 | state = json_tokener_state_string_escape; |
| 492 | break; |
| 493 | } |
| 494 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 495 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 496 | goto out; |
| 497 | } |
| 498 | } |
| 499 | } |
| 500 | break; |
| 501 | |
| 502 | case json_tokener_state_string_escape: |
| 503 | switch(c) { |
| 504 | case '"': |
| 505 | case '\\': |
| 506 | case '/': |
| 507 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 508 | state = saved_state; |
| 509 | break; |
| 510 | case 'b': |
| 511 | case 'n': |
| 512 | case 'r': |
| 513 | case 't': |
| 514 | case 'f': |
| 515 | if(c == 'b') printbuf_memappend_fast(tok->pb, "\b", 1); |
| 516 | else if(c == 'n') printbuf_memappend_fast(tok->pb, "\n", 1); |
| 517 | else if(c == 'r') printbuf_memappend_fast(tok->pb, "\r", 1); |
| 518 | else if(c == 't') printbuf_memappend_fast(tok->pb, "\t", 1); |
| 519 | else if(c == 'f') printbuf_memappend_fast(tok->pb, "\f", 1); |
| 520 | state = saved_state; |
| 521 | break; |
| 522 | case 'u': |
| 523 | tok->ucs_char = 0; |
| 524 | tok->st_pos = 0; |
| 525 | state = json_tokener_state_escape_unicode; |
| 526 | break; |
| 527 | default: |
| 528 | tok->err = json_tokener_error_parse_string; |
| 529 | goto out; |
| 530 | } |
| 531 | break; |
| 532 | |
| 533 | case json_tokener_state_escape_unicode: |
| 534 | { |
| 535 | unsigned int got_hi_surrogate = 0; |
| 536 | |
| 537 | /* Handle a 4-byte sequence, or two sequences if a surrogate pair */ |
| 538 | while(1) { |
| 539 | if(strchr(json_hex_chars, c)) { |
| 540 | tok->ucs_char += ((unsigned int)jt_hexdigit(c) << ((3-tok->st_pos++)*4)); |
| 541 | if(tok->st_pos == 4) { |
| 542 | unsigned char unescaped_utf[4]; |
| 543 | |
| 544 | if (got_hi_surrogate) { |
| 545 | if (IS_LOW_SURROGATE(tok->ucs_char)) { |
| 546 | /* Recalculate the ucs_char, then fall thru to process normally */ |
| 547 | tok->ucs_char = DECODE_SURROGATE_PAIR(got_hi_surrogate, tok->ucs_char); |
| 548 | } else { |
| 549 | /* Hi surrogate was not followed by a low surrogate */ |
| 550 | /* Replace the hi and process the rest normally */ |
| 551 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 552 | } |
| 553 | got_hi_surrogate = 0; |
| 554 | } |
| 555 | |
| 556 | if (tok->ucs_char < 0x80) { |
| 557 | unescaped_utf[0] = tok->ucs_char; |
| 558 | printbuf_memappend_fast(tok->pb, (char*)unescaped_utf, 1); |
| 559 | } else if (tok->ucs_char < 0x800) { |
| 560 | unescaped_utf[0] = 0xc0 | (tok->ucs_char >> 6); |
| 561 | unescaped_utf[1] = 0x80 | (tok->ucs_char & 0x3f); |
| 562 | printbuf_memappend_fast(tok->pb, (char*)unescaped_utf, 2); |
| 563 | } else if (IS_HIGH_SURROGATE(tok->ucs_char)) { |
| 564 | /* Got a high surrogate. Remember it and look for the |
| 565 | * the beginning of another sequence, which should be the |
| 566 | * low surrogate. |
| 567 | */ |
| 568 | got_hi_surrogate = tok->ucs_char; |
| 569 | /* Not at end, and the next two chars should be "\u" */ |
| 570 | if ((tok->char_offset+1 != len) && |
| 571 | (tok->char_offset+2 != len) && |
| 572 | (str[1] == '\\') && |
| 573 | (str[2] == 'u')) |
| 574 | { |
| 575 | /* Advance through the 16 bit surrogate, and move on to the |
| 576 | * next sequence. The next step is to process the following |
| 577 | * characters. |
| 578 | */ |
| 579 | if( !ADVANCE_CHAR(str, tok) || !ADVANCE_CHAR(str, tok) ) { |
| 580 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 581 | } |
| 582 | /* Advance to the first char of the next sequence and |
| 583 | * continue processing with the next sequence. |
| 584 | */ |
| 585 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 586 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 587 | goto out; |
| 588 | } |
| 589 | tok->ucs_char = 0; |
| 590 | tok->st_pos = 0; |
| 591 | continue; /* other json_tokener_state_escape_unicode */ |
| 592 | } else { |
| 593 | /* Got a high surrogate without another sequence following |
| 594 | * it. Put a replacement char in for the hi surrogate |
| 595 | * and pretend we finished. |
| 596 | */ |
| 597 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 598 | } |
| 599 | } else if (IS_LOW_SURROGATE(tok->ucs_char)) { |
| 600 | /* Got a low surrogate not preceded by a high */ |
| 601 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 602 | } else if (tok->ucs_char < 0x10000) { |
| 603 | unescaped_utf[0] = 0xe0 | (tok->ucs_char >> 12); |
| 604 | unescaped_utf[1] = 0x80 | ((tok->ucs_char >> 6) & 0x3f); |
| 605 | unescaped_utf[2] = 0x80 | (tok->ucs_char & 0x3f); |
| 606 | printbuf_memappend_fast(tok->pb, (char*)unescaped_utf, 3); |
| 607 | } else if (tok->ucs_char < 0x110000) { |
| 608 | unescaped_utf[0] = 0xf0 | ((tok->ucs_char >> 18) & 0x07); |
| 609 | unescaped_utf[1] = 0x80 | ((tok->ucs_char >> 12) & 0x3f); |
| 610 | unescaped_utf[2] = 0x80 | ((tok->ucs_char >> 6) & 0x3f); |
| 611 | unescaped_utf[3] = 0x80 | (tok->ucs_char & 0x3f); |
| 612 | printbuf_memappend_fast(tok->pb, (char*)unescaped_utf, 4); |
| 613 | } else { |
| 614 | /* Don't know what we got--insert the replacement char */ |
| 615 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 616 | } |
| 617 | state = saved_state; |
| 618 | break; |
| 619 | } |
| 620 | } else { |
| 621 | tok->err = json_tokener_error_parse_string; |
| 622 | goto out; |
| 623 | } |
| 624 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 625 | if (got_hi_surrogate) /* Clean up any pending chars */ |
| 626 | printbuf_memappend_fast(tok->pb, (char*)utf8_replacement_char, 3); |
| 627 | goto out; |
| 628 | } |
| 629 | } |
| 630 | } |
| 631 | break; |
| 632 | |
| 633 | case json_tokener_state_boolean: |
| 634 | { |
| 635 | int size1, size2; |
| 636 | printbuf_memappend_fast(tok->pb, &c, 1); |
| 637 | size1 = json_min(tok->st_pos+1, json_true_str_len); |
| 638 | size2 = json_min(tok->st_pos+1, json_false_str_len); |
| 639 | if((!(tok->flags & JSON_TOKENER_STRICT) && |
| 640 | strncasecmp(json_true_str, tok->pb->buf, size1) == 0) |
| 641 | || (strncmp(json_true_str, tok->pb->buf, size1) == 0) |
| 642 | ) { |
| 643 | if(tok->st_pos == json_true_str_len) { |
| 644 | current = json_object_new_boolean(1); |
| 645 | saved_state = json_tokener_state_finish; |
| 646 | state = json_tokener_state_eatws; |
| 647 | goto redo_char; |
| 648 | } |
| 649 | } else if((!(tok->flags & JSON_TOKENER_STRICT) && |
| 650 | strncasecmp(json_false_str, tok->pb->buf, size2) == 0) |
| 651 | || (strncmp(json_false_str, tok->pb->buf, size2) == 0)) { |
| 652 | if(tok->st_pos == json_false_str_len) { |
| 653 | current = json_object_new_boolean(0); |
| 654 | saved_state = json_tokener_state_finish; |
| 655 | state = json_tokener_state_eatws; |
| 656 | goto redo_char; |
| 657 | } |
| 658 | } else { |
| 659 | tok->err = json_tokener_error_parse_boolean; |
| 660 | goto out; |
| 661 | } |
| 662 | tok->st_pos++; |
| 663 | } |
| 664 | break; |
| 665 | |
| 666 | case json_tokener_state_number: |
| 667 | { |
| 668 | /* Advance until we change state */ |
| 669 | const char *case_start = str; |
| 670 | int case_len=0; |
| 671 | while(c && strchr(json_number_chars, c)) { |
| 672 | ++case_len; |
| 673 | if(c == '.' || c == 'e' || c == 'E') |
| 674 | tok->is_double = 1; |
| 675 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 676 | printbuf_memappend_fast(tok->pb, case_start, case_len); |
| 677 | goto out; |
| 678 | } |
| 679 | } |
| 680 | if (case_len>0) |
| 681 | printbuf_memappend_fast(tok->pb, case_start, case_len); |
| 682 | |
| 683 | // Check for -Infinity |
| 684 | if (tok->pb->buf[0] == '-' && case_len == 1 && |
| 685 | (c == 'i' || c == 'I')) |
| 686 | { |
| 687 | state = json_tokener_state_inf; |
| 688 | goto redo_char; |
| 689 | } |
| 690 | } |
| 691 | { |
| 692 | int64_t num64; |
| 693 | double numd; |
| 694 | if (!tok->is_double && json_parse_int64(tok->pb->buf, &num64) == 0) { |
| 695 | if (num64 && tok->pb->buf[0]=='0' && (tok->flags & JSON_TOKENER_STRICT)) { |
| 696 | /* in strict mode, number must not start with 0 */ |
| 697 | tok->err = json_tokener_error_parse_number; |
| 698 | goto out; |
| 699 | } |
| 700 | current = json_object_new_int64(num64); |
| 701 | } |
| 702 | else if(tok->is_double && json_parse_double(tok->pb->buf, &numd) == 0) |
| 703 | { |
| 704 | current = json_object_new_double_s(numd, tok->pb->buf); |
| 705 | } else { |
| 706 | tok->err = json_tokener_error_parse_number; |
| 707 | goto out; |
| 708 | } |
| 709 | saved_state = json_tokener_state_finish; |
| 710 | state = json_tokener_state_eatws; |
| 711 | goto redo_char; |
| 712 | } |
| 713 | break; |
| 714 | |
| 715 | case json_tokener_state_array_after_sep: |
| 716 | case json_tokener_state_array: |
| 717 | if(c == ']') { |
| 718 | if (state == json_tokener_state_array_after_sep && |
| 719 | (tok->flags & JSON_TOKENER_STRICT)) |
| 720 | { |
| 721 | tok->err = json_tokener_error_parse_unexpected; |
| 722 | goto out; |
| 723 | } |
| 724 | saved_state = json_tokener_state_finish; |
| 725 | state = json_tokener_state_eatws; |
| 726 | } else { |
| 727 | if(tok->depth >= tok->max_depth-1) { |
| 728 | tok->err = json_tokener_error_depth; |
| 729 | goto out; |
| 730 | } |
| 731 | state = json_tokener_state_array_add; |
| 732 | tok->depth++; |
| 733 | json_tokener_reset_level(tok, tok->depth); |
| 734 | goto redo_char; |
| 735 | } |
| 736 | break; |
| 737 | |
| 738 | case json_tokener_state_array_add: |
| 739 | json_object_array_add(current, obj); |
| 740 | saved_state = json_tokener_state_array_sep; |
| 741 | state = json_tokener_state_eatws; |
| 742 | goto redo_char; |
| 743 | |
| 744 | case json_tokener_state_array_sep: |
| 745 | if(c == ']') { |
| 746 | saved_state = json_tokener_state_finish; |
| 747 | state = json_tokener_state_eatws; |
| 748 | } else if(c == ',') { |
| 749 | saved_state = json_tokener_state_array_after_sep; |
| 750 | state = json_tokener_state_eatws; |
| 751 | } else { |
| 752 | tok->err = json_tokener_error_parse_array; |
| 753 | goto out; |
| 754 | } |
| 755 | break; |
| 756 | |
| 757 | case json_tokener_state_object_field_start: |
| 758 | case json_tokener_state_object_field_start_after_sep: |
| 759 | if(c == '}') { |
| 760 | if (state == json_tokener_state_object_field_start_after_sep && |
| 761 | (tok->flags & JSON_TOKENER_STRICT)) |
| 762 | { |
| 763 | tok->err = json_tokener_error_parse_unexpected; |
| 764 | goto out; |
| 765 | } |
| 766 | saved_state = json_tokener_state_finish; |
| 767 | state = json_tokener_state_eatws; |
| 768 | } else if (c == '"' || c == '\'') { |
| 769 | tok->quote_char = c; |
| 770 | printbuf_reset(tok->pb); |
| 771 | state = json_tokener_state_object_field; |
| 772 | } else { |
| 773 | tok->err = json_tokener_error_parse_object_key_name; |
| 774 | goto out; |
| 775 | } |
| 776 | break; |
| 777 | |
| 778 | case json_tokener_state_object_field: |
| 779 | { |
| 780 | /* Advance until we change state */ |
| 781 | const char *case_start = str; |
| 782 | while(1) { |
| 783 | if(c == tok->quote_char) { |
| 784 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 785 | obj_field_name = strdup(tok->pb->buf); |
| 786 | saved_state = json_tokener_state_object_field_end; |
| 787 | state = json_tokener_state_eatws; |
| 788 | break; |
| 789 | } else if(c == '\\') { |
| 790 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 791 | saved_state = json_tokener_state_object_field; |
| 792 | state = json_tokener_state_string_escape; |
| 793 | break; |
| 794 | } |
| 795 | if (!ADVANCE_CHAR(str, tok) || !PEEK_CHAR(c, tok)) { |
| 796 | printbuf_memappend_fast(tok->pb, case_start, str-case_start); |
| 797 | goto out; |
| 798 | } |
| 799 | } |
| 800 | } |
| 801 | break; |
| 802 | |
| 803 | case json_tokener_state_object_field_end: |
| 804 | if(c == ':') { |
| 805 | saved_state = json_tokener_state_object_value; |
| 806 | state = json_tokener_state_eatws; |
| 807 | } else { |
| 808 | tok->err = json_tokener_error_parse_object_key_sep; |
| 809 | goto out; |
| 810 | } |
| 811 | break; |
| 812 | |
| 813 | case json_tokener_state_object_value: |
| 814 | if(tok->depth >= tok->max_depth-1) { |
| 815 | tok->err = json_tokener_error_depth; |
| 816 | goto out; |
| 817 | } |
| 818 | state = json_tokener_state_object_value_add; |
| 819 | tok->depth++; |
| 820 | json_tokener_reset_level(tok, tok->depth); |
| 821 | goto redo_char; |
| 822 | |
| 823 | case json_tokener_state_object_value_add: |
| 824 | json_object_object_add(current, obj_field_name, obj); |
| 825 | free(obj_field_name); |
| 826 | obj_field_name = NULL; |
| 827 | saved_state = json_tokener_state_object_sep; |
| 828 | state = json_tokener_state_eatws; |
| 829 | goto redo_char; |
| 830 | |
| 831 | case json_tokener_state_object_sep: |
| 832 | if(c == '}') { |
| 833 | saved_state = json_tokener_state_finish; |
| 834 | state = json_tokener_state_eatws; |
| 835 | } else if(c == ',') { |
| 836 | saved_state = json_tokener_state_object_field_start_after_sep; |
| 837 | state = json_tokener_state_eatws; |
| 838 | } else { |
| 839 | tok->err = json_tokener_error_parse_object_value_sep; |
| 840 | goto out; |
| 841 | } |
| 842 | break; |
| 843 | |
| 844 | } |
| 845 | if (!ADVANCE_CHAR(str, tok)) |
| 846 | goto out; |
| 847 | } /* while(POP_CHAR) */ |
| 848 | |
| 849 | out: |
| 850 | if (c && |
| 851 | (state == json_tokener_state_finish) && |
| 852 | (tok->depth == 0) && |
| 853 | (tok->flags & JSON_TOKENER_STRICT)) { |
| 854 | /* unexpected char after JSON data */ |
| 855 | tok->err = json_tokener_error_parse_unexpected; |
| 856 | } |
| 857 | if (!c) { /* We hit an eof char (0) */ |
| 858 | if(state != json_tokener_state_finish && |
| 859 | saved_state != json_tokener_state_finish) |
| 860 | tok->err = json_tokener_error_parse_eof; |
| 861 | } |
| 862 | |
| 863 | #ifdef HAVE_SETLOCALE |
| 864 | setlocale(LC_NUMERIC, oldlocale); |
| 865 | if (oldlocale) free(oldlocale); |
| 866 | #endif |
| 867 | |
| 868 | if (tok->err == json_tokener_success) |
| 869 | { |
| 870 | json_object *ret = json_object_get(current); |
| 871 | int ii; |
| 872 | |
| 873 | /* Partially reset, so we parse additional objects on subsequent calls. */ |
| 874 | for(ii = tok->depth; ii >= 0; ii--) |
| 875 | json_tokener_reset_level(tok, ii); |
| 876 | return ret; |
| 877 | } |
| 878 | |
| 879 | MC_DEBUG("json_tokener_parse_ex: error %s at offset %d\n", |
| 880 | json_tokener_errors[tok->err], tok->char_offset); |
| 881 | return NULL; |
| 882 | } |
| 883 | |
| 884 | void json_tokener_set_flags(struct json_tokener *tok, int flags) |
| 885 | { |
| 886 | tok->flags = flags; |
| 887 | } |