| 1 | #include <stdio.h> |
| 2 | #include <stdbool.h> |
| 3 | #include <stdlib.h> |
| 4 | #include <string.h> |
| 5 | #include <ctype.h> |
| 6 | #include <errno.h> |
| 7 | #include "pd_json.h" |
| 8 | |
| 9 | #define json_error(json, format, ...) \ |
| 10 | if (!json->error) { \ |
| 11 | json->error = 1; \ |
| 12 | snprintf(json->errmsg, sizeof(json->errmsg), \ |
| 13 | "error: %lu: " format, \ |
| 14 | (unsigned long) json->lineno, \ |
| 15 | __VA_ARGS__); \ |
| 16 | } \ |
| 17 | |
| 18 | #define STACK_INC 4 |
| 19 | |
| 20 | static enum json_type |
| 21 | push(json_stream *json, enum json_type type) |
| 22 | { |
| 23 | json->stack_top++; |
| 24 | |
| 25 | if (json->stack_top >= json->stack_size) { |
| 26 | struct json_stack *stack; |
| 27 | stack = json->alloc.realloc(json->stack, |
| 28 | (json->stack_size + STACK_INC) * sizeof(*json->stack)); |
| 29 | if (stack == NULL) { |
| 30 | json_error(json, "%s" , strerror(errno)); |
| 31 | return JSON_ERROR; |
| 32 | } |
| 33 | |
| 34 | json->stack_size += STACK_INC; |
| 35 | json->stack = stack; |
| 36 | } |
| 37 | |
| 38 | json->stack[json->stack_top].type = type; |
| 39 | json->stack[json->stack_top].count = 0; |
| 40 | |
| 41 | return type; |
| 42 | } |
| 43 | |
| 44 | static enum json_type |
| 45 | pop(json_stream *json, int c, enum json_type expected) |
| 46 | { |
| 47 | if (json->stack == NULL || json->stack[json->stack_top].type != expected) { |
| 48 | json_error(json, "unexpected byte, '%c'" , c); |
| 49 | json->alloc.free(json->stack); |
| 50 | return JSON_ERROR; |
| 51 | } |
| 52 | json->stack_top--; |
| 53 | return expected == JSON_ARRAY ? JSON_ARRAY_END : JSON_OBJECT_END; |
| 54 | } |
| 55 | |
| 56 | static void pop_all(json_stream *json) |
| 57 | { |
| 58 | json->alloc.free(json->stack); |
| 59 | } |
| 60 | |
| 61 | static int buffer_peek(struct json_source *source) |
| 62 | { |
| 63 | if (source->position < source->source.buffer.length) |
| 64 | return source->source.buffer.buffer[source->position]; |
| 65 | else |
| 66 | return EOF; |
| 67 | } |
| 68 | |
| 69 | static int buffer_get(struct json_source *source) |
| 70 | { |
| 71 | int c = source->peek(source); |
| 72 | source->position++; |
| 73 | return c; |
| 74 | } |
| 75 | |
| 76 | static int stream_get(struct json_source *source) |
| 77 | { |
| 78 | source->position++; |
| 79 | return fgetc(source->source.stream.stream); |
| 80 | } |
| 81 | |
| 82 | static int stream_peek(struct json_source *source) |
| 83 | { |
| 84 | int c = fgetc(source->source.stream.stream); |
| 85 | ungetc(c, source->source.stream.stream); |
| 86 | return c; |
| 87 | } |
| 88 | |
| 89 | static void init(json_stream *json) |
| 90 | { |
| 91 | json->lineno = 1; |
| 92 | json->error = 0; |
| 93 | json->errmsg[0] = '\0'; |
| 94 | json->ntokens = 0; |
| 95 | json->next = 0; |
| 96 | json->streaming = true; |
| 97 | |
| 98 | json->stack = NULL; |
| 99 | json->stack_top = -1; |
| 100 | json->stack_size = 0; |
| 101 | |
| 102 | json->data.string = NULL; |
| 103 | json->data.string_size = 0; |
| 104 | json->data.string_fill = 0; |
| 105 | json->source.position = 0; |
| 106 | |
| 107 | json->alloc.malloc = malloc; |
| 108 | json->alloc.realloc = realloc; |
| 109 | json->alloc.free = free; |
| 110 | } |
| 111 | |
| 112 | static enum json_type |
| 113 | is_match(json_stream *json, const char *pattern, enum json_type type) |
| 114 | { |
| 115 | for (const char *p = pattern; *p; p++) |
| 116 | if (*p != json->source.get(&json->source)) |
| 117 | return JSON_ERROR; |
| 118 | return type; |
| 119 | } |
| 120 | |
| 121 | static int pushchar(json_stream *json, int c) |
| 122 | { |
| 123 | if (json->data.string_fill == json->data.string_size) { |
| 124 | size_t size = json->data.string_size * 2; |
| 125 | char *buffer = json->alloc.realloc(json->data.string, size); |
| 126 | if (buffer == NULL) { |
| 127 | json_error(json, "%s" , strerror(errno)); |
| 128 | return -1; |
| 129 | } else { |
| 130 | json->data.string_size = size; |
| 131 | json->data.string = buffer; |
| 132 | } |
| 133 | } |
| 134 | json->data.string[json->data.string_fill++] = c; |
| 135 | return 0; |
| 136 | } |
| 137 | |
| 138 | static int init_string(json_stream *json) |
| 139 | { |
| 140 | json->data.string_fill = 0; |
| 141 | if (json->data.string == NULL) { |
| 142 | json->data.string_size = 1024; |
| 143 | json->data.string = json->alloc.malloc(json->data.string_size); |
| 144 | if (json->data.string == NULL) { |
| 145 | json_error(json, "%s" , strerror(errno)); |
| 146 | return -1; |
| 147 | } |
| 148 | } |
| 149 | json->data.string[0] = '\0'; |
| 150 | return 0; |
| 151 | } |
| 152 | |
| 153 | static int encode_utf8(json_stream *json, unsigned long c) |
| 154 | { |
| 155 | if (c < 0x80UL) { |
| 156 | return pushchar(json, c); |
| 157 | } else if (c < 0x0800UL) { |
| 158 | return !((pushchar(json, (c >> 6 & 0x1F) | 0xC0) == 0) && |
| 159 | (pushchar(json, (c >> 0 & 0x3F) | 0x80) == 0)); |
| 160 | } else if (c < 0x010000UL) { |
| 161 | if (c >= 0xd800 && c <= 0xdfff) { |
| 162 | json_error(json, "invalid codepoint %06lx" , c); |
| 163 | return -1; |
| 164 | } |
| 165 | return !((pushchar(json, (c >> 12 & 0x0F) | 0xE0) == 0) && |
| 166 | (pushchar(json, (c >> 6 & 0x3F) | 0x80) == 0) && |
| 167 | (pushchar(json, (c >> 0 & 0x3F) | 0x80) == 0)); |
| 168 | } else if (c < 0x110000UL) { |
| 169 | return !((pushchar(json, (c >> 18 & 0x07) | 0xF0) == 0) && |
| 170 | (pushchar(json, (c >> 12 & 0x3F) | 0x80) == 0) && |
| 171 | (pushchar(json, (c >> 6 & 0x3F) | 0x80) == 0) && |
| 172 | (pushchar(json, (c >> 0 & 0x3F) | 0x80) == 0)); |
| 173 | } else { |
| 174 | json_error(json, "can't encode UTF-8 for %06lx" , c); |
| 175 | return -1; |
| 176 | } |
| 177 | } |
| 178 | |
| 179 | static int hexchar(int c) |
| 180 | { |
| 181 | switch (c) { |
| 182 | case '0': return 0; |
| 183 | case '1': return 1; |
| 184 | case '2': return 2; |
| 185 | case '3': return 3; |
| 186 | case '4': return 4; |
| 187 | case '5': return 5; |
| 188 | case '6': return 6; |
| 189 | case '7': return 7; |
| 190 | case '8': return 8; |
| 191 | case '9': return 9; |
| 192 | case 'a': |
| 193 | case 'A': return 10; |
| 194 | case 'b': |
| 195 | case 'B': return 11; |
| 196 | case 'c': |
| 197 | case 'C': return 12; |
| 198 | case 'd': |
| 199 | case 'D': return 13; |
| 200 | case 'e': |
| 201 | case 'E': return 14; |
| 202 | case 'f': |
| 203 | case 'F': return 15; |
| 204 | default: |
| 205 | return -1; |
| 206 | } |
| 207 | } |
| 208 | |
| 209 | static long |
| 210 | read_unicode_cp(json_stream *json) |
| 211 | { |
| 212 | long cp = 0; |
| 213 | int shift = 12; |
| 214 | |
| 215 | for (size_t i = 0; i < 4; i++) { |
| 216 | int c = json->source.get(&json->source); |
| 217 | int hc; |
| 218 | |
| 219 | if (c == EOF) { |
| 220 | json_error(json, "%s" , "unterminated string literal in unicode" ); |
| 221 | return -1; |
| 222 | } else if ((hc = hexchar(c)) == -1) { |
| 223 | json_error(json, "bad escape unicode byte, '%c'" , c); |
| 224 | return -1; |
| 225 | } |
| 226 | |
| 227 | cp += hc * (1 << shift); |
| 228 | shift -= 4; |
| 229 | } |
| 230 | |
| 231 | |
| 232 | return cp; |
| 233 | } |
| 234 | |
| 235 | static int read_unicode(json_stream *json) |
| 236 | { |
| 237 | long cp, h, l; |
| 238 | |
| 239 | if ((cp = read_unicode_cp(json)) == -1) { |
| 240 | return -1; |
| 241 | } |
| 242 | |
| 243 | if (cp >= 0xd800 && cp <= 0xdbff) { |
| 244 | /* This is the high portion of a surrogate pair; we need to read the |
| 245 | * lower portion to get the codepoint |
| 246 | */ |
| 247 | h = cp; |
| 248 | |
| 249 | int c = json->source.get(&json->source); |
| 250 | if (c == EOF) { |
| 251 | json_error(json, "%s" , "unterminated string literal in unicode" ); |
| 252 | return -1; |
| 253 | } else if (c != '\\') { |
| 254 | json_error(json, "invalid continuation for surrogate pair: '%c', " |
| 255 | "expected '\\'" , c); |
| 256 | return -1; |
| 257 | } |
| 258 | |
| 259 | c = json->source.get(&json->source); |
| 260 | if (c == EOF) { |
| 261 | json_error(json, "%s" , "unterminated string literal in unicode" ); |
| 262 | return -1; |
| 263 | } else if (c != 'u') { |
| 264 | json_error(json, "invalid continuation for surrogate pair: '%c', " |
| 265 | "expected 'u'" , c); |
| 266 | return -1; |
| 267 | } |
| 268 | |
| 269 | if ((l = read_unicode_cp(json)) == -1) { |
| 270 | return -1; |
| 271 | } |
| 272 | |
| 273 | if (l < 0xdc00 || l > 0xdfff) { |
| 274 | json_error(json, "invalid surrogate pair continuation \\u%04lx out " |
| 275 | "of range (dc00-dfff)" , l); |
| 276 | return -1; |
| 277 | } |
| 278 | |
| 279 | cp = ((h - 0xd800) * 0x400) + ((l - 0xdc00) + 0x10000); |
| 280 | } else if (cp >= 0xdc00 && cp <= 0xdfff) { |
| 281 | json_error(json, "dangling surrogate \\u%04lx" , cp); |
| 282 | return -1; |
| 283 | } |
| 284 | |
| 285 | return encode_utf8(json, cp); |
| 286 | } |
| 287 | |
| 288 | int read_escaped(json_stream *json) |
| 289 | { |
| 290 | int c = json->source.get(&json->source); |
| 291 | if (c == EOF) { |
| 292 | json_error(json, "%s" , "unterminated string literal in escape" ); |
| 293 | return -1; |
| 294 | } else if (c == 'u') { |
| 295 | if (read_unicode(json) != 0) |
| 296 | return -1; |
| 297 | } else { |
| 298 | switch (c) { |
| 299 | case '\\': |
| 300 | case 'b': |
| 301 | case 'f': |
| 302 | case 'n': |
| 303 | case 'r': |
| 304 | case 't': |
| 305 | case '/': |
| 306 | case '"': |
| 307 | { |
| 308 | const char *codes = "\\bfnrt/\"" ; |
| 309 | char *p = strchr(codes, c); |
| 310 | if (pushchar(json, "\\\b\f\n\r\t/\"" [p - codes]) != 0) |
| 311 | return -1; |
| 312 | } |
| 313 | break; |
| 314 | default: |
| 315 | json_error(json, "bad escaped byte, '%c'" , c); |
| 316 | return -1; |
| 317 | } |
| 318 | } |
| 319 | return 0; |
| 320 | } |
| 321 | |
| 322 | static int |
| 323 | char_needs_escaping(int c) |
| 324 | { |
| 325 | if ((c >= 0) && (c < 0x20 || c == 0x22 || c == 0x5c)) { |
| 326 | return 1; |
| 327 | } |
| 328 | |
| 329 | return 0; |
| 330 | } |
| 331 | |
| 332 | static int |
| 333 | utf8_seq_length(char byte) |
| 334 | { |
| 335 | unsigned char u = (unsigned char) byte; |
| 336 | if (u < 0x80) return 1; |
| 337 | |
| 338 | if (0x80 <= u && u <= 0xBF) |
| 339 | { |
| 340 | // second, third or fourth byte of a multi-byte |
| 341 | // sequence, i.e. a "continuation byte" |
| 342 | return 0; |
| 343 | } |
| 344 | else if (u == 0xC0 || u == 0xC1) |
| 345 | { |
| 346 | // overlong encoding of an ASCII byte |
| 347 | return 0; |
| 348 | } |
| 349 | else if (0xC2 <= u && u <= 0xDF) |
| 350 | { |
| 351 | // 2-byte sequence |
| 352 | return 2; |
| 353 | } |
| 354 | else if (0xE0 <= u && u <= 0xEF) |
| 355 | { |
| 356 | // 3-byte sequence |
| 357 | return 3; |
| 358 | } |
| 359 | else if (0xF0 <= u && u <= 0xF4) |
| 360 | { |
| 361 | // 4-byte sequence |
| 362 | return 4; |
| 363 | } |
| 364 | else |
| 365 | { |
| 366 | // u >= 0xF5 |
| 367 | // Restricted (start of 4-, 5- or 6-byte sequence) or invalid UTF-8 |
| 368 | return 0; |
| 369 | } |
| 370 | } |
| 371 | |
| 372 | static int |
| 373 | is_legal_utf8(const unsigned char *bytes, int length) |
| 374 | { |
| 375 | if (0 == bytes || 0 == length) return 0; |
| 376 | |
| 377 | unsigned char a; |
| 378 | const unsigned char* srcptr = bytes + length; |
| 379 | switch (length) |
| 380 | { |
| 381 | default: |
| 382 | return 0; |
| 383 | // Everything else falls through when true. |
| 384 | case 4: |
| 385 | if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return 0; |
| 386 | case 3: |
| 387 | if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return 0; |
| 388 | case 2: |
| 389 | a = (*--srcptr); |
| 390 | switch (*bytes) |
| 391 | { |
| 392 | case 0xE0: |
| 393 | if (a < 0xA0 || a > 0xBF) return 0; |
| 394 | break; |
| 395 | case 0xED: |
| 396 | if (a < 0x80 || a > 0x9F) return 0; |
| 397 | break; |
| 398 | case 0xF0: |
| 399 | if (a < 0x90 || a > 0xBF) return 0; |
| 400 | break; |
| 401 | case 0xF4: |
| 402 | if (a < 0x80 || a > 0x8F) return 0; |
| 403 | break; |
| 404 | default: |
| 405 | if (a < 0x80 || a > 0xBF) return 0; |
| 406 | } |
| 407 | case 1: |
| 408 | if (*bytes >= 0x80 && *bytes < 0xC2) return 0; |
| 409 | } |
| 410 | return *bytes <= 0xF4; |
| 411 | } |
| 412 | |
| 413 | static int |
| 414 | read_utf8(json_stream* json, int next_char) |
| 415 | { |
| 416 | int count = utf8_seq_length(next_char); |
| 417 | if (!count) |
| 418 | { |
| 419 | json_error(json, "%s" , "Bad character." ); |
| 420 | return -1; |
| 421 | } |
| 422 | |
| 423 | char buffer[4]; |
| 424 | buffer[0] = next_char; |
| 425 | for (int i = 1; i < count; ++i) |
| 426 | { |
| 427 | buffer[i] = json->source.get(&json->source);; |
| 428 | } |
| 429 | |
| 430 | if (!is_legal_utf8((unsigned char*) buffer, count)) |
| 431 | { |
| 432 | json_error(json, "%s" , "No legal UTF8 found" ); |
| 433 | return -1; |
| 434 | } |
| 435 | |
| 436 | for (int i = 0; i < count; ++i) |
| 437 | { |
| 438 | if (pushchar(json, buffer[i]) != 0) |
| 439 | return -1; |
| 440 | } |
| 441 | return 0; |
| 442 | } |
| 443 | |
| 444 | static enum json_type |
| 445 | read_string(json_stream *json) |
| 446 | { |
| 447 | if (init_string(json) != 0) |
| 448 | return JSON_ERROR; |
| 449 | while (1) { |
| 450 | int c = json->source.get(&json->source); |
| 451 | if (c == EOF) { |
| 452 | json_error(json, "%s" , "unterminated string literal" ); |
| 453 | return JSON_ERROR; |
| 454 | } else if (c == '"') { |
| 455 | if (pushchar(json, '\0') == 0) |
| 456 | return JSON_STRING; |
| 457 | else |
| 458 | return JSON_ERROR; |
| 459 | } else if (c == '\\') { |
| 460 | if (read_escaped(json) != 0) |
| 461 | return JSON_ERROR; |
| 462 | } else if ((unsigned) c >= 0x80) { |
| 463 | if (read_utf8(json, c) != 0) |
| 464 | return JSON_ERROR; |
| 465 | } else { |
| 466 | if (char_needs_escaping(c)) { |
| 467 | json_error(json, "%s:%u" , "unescaped control character in string" , (unsigned)c); |
| 468 | return JSON_ERROR; |
| 469 | } |
| 470 | |
| 471 | if (pushchar(json, c) != 0) |
| 472 | return JSON_ERROR; |
| 473 | } |
| 474 | } |
| 475 | return JSON_ERROR; |
| 476 | } |
| 477 | |
| 478 | static int |
| 479 | is_digit(int c) |
| 480 | { |
| 481 | return c >= 48 /*0*/ && c <= 57 /*9*/; |
| 482 | } |
| 483 | |
| 484 | static int |
| 485 | read_digits(json_stream *json) |
| 486 | { |
| 487 | unsigned nread = 0; |
| 488 | while (is_digit(json->source.peek(&json->source))) { |
| 489 | if (pushchar(json, json->source.get(&json->source)) != 0) |
| 490 | return -1; |
| 491 | |
| 492 | nread++; |
| 493 | } |
| 494 | |
| 495 | if (nread == 0) { |
| 496 | return -1; |
| 497 | } |
| 498 | |
| 499 | return 0; |
| 500 | } |
| 501 | |
| 502 | static enum json_type |
| 503 | read_number(json_stream *json, int c) |
| 504 | { |
| 505 | if (pushchar(json, c) != 0) |
| 506 | return JSON_ERROR; |
| 507 | if (c == '-') { |
| 508 | c = json->source.get(&json->source); |
| 509 | if (is_digit(c)) { |
| 510 | return read_number(json, c); |
| 511 | } else { |
| 512 | json_error(json, "unexpected byte, '%c'" , c); |
| 513 | } |
| 514 | } else if (strchr("123456789" , c) != NULL) { |
| 515 | c = json->source.peek(&json->source); |
| 516 | if (is_digit(c)) { |
| 517 | if (read_digits(json) != 0) |
| 518 | return JSON_ERROR; |
| 519 | } |
| 520 | } |
| 521 | /* Up to decimal or exponent has been read. */ |
| 522 | c = json->source.peek(&json->source); |
| 523 | if (strchr(".eE" , c) == NULL) { |
| 524 | if (pushchar(json, '\0') != 0) |
| 525 | return JSON_ERROR; |
| 526 | else |
| 527 | return JSON_NUMBER; |
| 528 | } |
| 529 | if (c == '.') { |
| 530 | json->source.get(&json->source); // consume . |
| 531 | if (pushchar(json, c) != 0) |
| 532 | return JSON_ERROR; |
| 533 | if (read_digits(json) != 0) |
| 534 | return JSON_ERROR; |
| 535 | } |
| 536 | /* Check for exponent. */ |
| 537 | c = json->source.peek(&json->source); |
| 538 | if (c == 'e' || c == 'E') { |
| 539 | json->source.get(&json->source); // consume e/E |
| 540 | if (pushchar(json, c) != 0) |
| 541 | return JSON_ERROR; |
| 542 | c = json->source.peek(&json->source); |
| 543 | if (c == '+' || c == '-') { |
| 544 | json->source.get(&json->source); // consume |
| 545 | if (pushchar(json, c) != 0) |
| 546 | return JSON_ERROR; |
| 547 | if (read_digits(json) != 0) |
| 548 | return JSON_ERROR; |
| 549 | } else if (is_digit(c)) { |
| 550 | if (read_digits(json) != 0) |
| 551 | return JSON_ERROR; |
| 552 | } else { |
| 553 | json_error(json, "unexpected byte in number, '%c'" , c); |
| 554 | return JSON_ERROR; |
| 555 | } |
| 556 | } |
| 557 | if (pushchar(json, '\0') != 0) |
| 558 | return JSON_ERROR; |
| 559 | else |
| 560 | return JSON_NUMBER; |
| 561 | } |
| 562 | |
| 563 | static int |
| 564 | json_isspace(int c) |
| 565 | { |
| 566 | switch (c) { |
| 567 | case 0x09: |
| 568 | case 0x0a: |
| 569 | case 0x0d: |
| 570 | case 0x20: |
| 571 | return 1; |
| 572 | } |
| 573 | |
| 574 | return 0; |
| 575 | } |
| 576 | |
| 577 | /* Returns the next non-whitespace character in the stream. */ |
| 578 | static int next(json_stream *json) |
| 579 | { |
| 580 | int c; |
| 581 | while (json_isspace(c = json->source.get(&json->source))) |
| 582 | if (c == '\n') |
| 583 | json->lineno++; |
| 584 | return c; |
| 585 | } |
| 586 | |
| 587 | static enum json_type |
| 588 | read_value(json_stream *json, int c) |
| 589 | { |
| 590 | json->ntokens++; |
| 591 | switch (c) { |
| 592 | case EOF: |
| 593 | json_error(json, "%s" , "unexpected end of data" ); |
| 594 | return JSON_ERROR; |
| 595 | case '{': |
| 596 | return push(json, JSON_OBJECT); |
| 597 | case '[': |
| 598 | return push(json, JSON_ARRAY); |
| 599 | case '"': |
| 600 | return read_string(json); |
| 601 | case 'n': |
| 602 | return is_match(json, "ull" , JSON_NULL); |
| 603 | case 'f': |
| 604 | return is_match(json, "alse" , JSON_FALSE); |
| 605 | case 't': |
| 606 | return is_match(json, "rue" , JSON_TRUE); |
| 607 | case '0': |
| 608 | case '1': |
| 609 | case '2': |
| 610 | case '3': |
| 611 | case '4': |
| 612 | case '5': |
| 613 | case '6': |
| 614 | case '7': |
| 615 | case '8': |
| 616 | case '9': |
| 617 | case '-': |
| 618 | if (init_string(json) != 0) |
| 619 | return JSON_ERROR; |
| 620 | return read_number(json, c); |
| 621 | default: |
| 622 | json_error(json, "unexpected byte, '%c'" , c); |
| 623 | return JSON_ERROR; |
| 624 | } |
| 625 | } |
| 626 | |
| 627 | enum json_type json_peek(json_stream *json) |
| 628 | { |
| 629 | enum json_type next = json_next(json); |
| 630 | json->next = next; |
| 631 | return next; |
| 632 | } |
| 633 | |
| 634 | enum json_type json_next(json_stream *json) |
| 635 | { |
| 636 | if (json->error) |
| 637 | return JSON_ERROR; |
| 638 | if (json->next != 0) { |
| 639 | enum json_type next = json->next; |
| 640 | json->next = 0; |
| 641 | return next; |
| 642 | } |
| 643 | if (json->ntokens > 0 && json->stack_top == (size_t)-1) { |
| 644 | int c; |
| 645 | |
| 646 | do { |
| 647 | c = json->source.peek(&json->source); |
| 648 | if (json_isspace(c)) { |
| 649 | c = json->source.get(&json->source); |
| 650 | } |
| 651 | } while (json_isspace(c)); |
| 652 | if (!json->streaming && c != EOF) { |
| 653 | return JSON_ERROR; |
| 654 | } |
| 655 | return JSON_DONE; |
| 656 | } |
| 657 | int c = next(json); |
| 658 | if (json->stack == NULL) |
| 659 | return read_value(json, c); |
| 660 | if (json->stack[json->stack_top].type == JSON_ARRAY) { |
| 661 | if (json->stack[json->stack_top].count == 0) { |
| 662 | if (c == ']') { |
| 663 | return pop(json, c, JSON_ARRAY); |
| 664 | } |
| 665 | json->stack[json->stack_top].count++; |
| 666 | return read_value(json, c); |
| 667 | } else if (c == ',') { |
| 668 | json->stack[json->stack_top].count++; |
| 669 | return read_value(json, next(json)); |
| 670 | } else if (c == ']') { |
| 671 | return pop(json, c, JSON_ARRAY); |
| 672 | } else { |
| 673 | json_error(json, "unexpected byte, '%c'" , c); |
| 674 | return JSON_ERROR; |
| 675 | } |
| 676 | } else if (json->stack[json->stack_top].type == JSON_OBJECT) { |
| 677 | if (json->stack[json->stack_top].count == 0) { |
| 678 | if (c == '}') { |
| 679 | return pop(json, c, JSON_OBJECT); |
| 680 | } |
| 681 | |
| 682 | /* No property value pairs yet. */ |
| 683 | enum json_type value = read_value(json, c); |
| 684 | if (value != JSON_STRING) { |
| 685 | json_error(json, "%s" , "expected property name or '}'" ); |
| 686 | return JSON_ERROR; |
| 687 | } else { |
| 688 | json->stack[json->stack_top].count++; |
| 689 | return value; |
| 690 | } |
| 691 | } else if ((json->stack[json->stack_top].count % 2) == 0) { |
| 692 | /* Expecting comma followed by property name. */ |
| 693 | if (c != ',' && c != '}') { |
| 694 | json_error(json, "%s" , "expected ',' or '}'" ); |
| 695 | return JSON_ERROR; |
| 696 | } else if (c == '}') { |
| 697 | return pop(json, c, JSON_OBJECT); |
| 698 | } else { |
| 699 | enum json_type value = read_value(json, next(json)); |
| 700 | if (value != JSON_STRING) { |
| 701 | json_error(json, "%s" , "expected property name" ); |
| 702 | return JSON_ERROR; |
| 703 | } else { |
| 704 | json->stack[json->stack_top].count++; |
| 705 | return value; |
| 706 | } |
| 707 | } |
| 708 | } else if ((json->stack[json->stack_top].count % 2) == 1) { |
| 709 | /* Expecting colon followed by value. */ |
| 710 | if (c != ':') { |
| 711 | json_error(json, "%s" , "expected ':' after property name" ); |
| 712 | return JSON_ERROR; |
| 713 | } else { |
| 714 | json->stack[json->stack_top].count++; |
| 715 | return read_value(json, next(json)); |
| 716 | } |
| 717 | } |
| 718 | } |
| 719 | json_error(json, "%s" , "invalid parser state" ); |
| 720 | return JSON_ERROR; |
| 721 | } |
| 722 | |
| 723 | void json_reset(json_stream *json) |
| 724 | { |
| 725 | pop_all(json); |
| 726 | json->ntokens = 0; |
| 727 | json->error = 0; |
| 728 | json->errmsg[0] = '\0'; |
| 729 | } |
| 730 | |
| 731 | const char *json_get_string(json_stream *json, size_t *length) |
| 732 | { |
| 733 | if (length != NULL) |
| 734 | *length = json->data.string_fill; |
| 735 | if (json->data.string == NULL) |
| 736 | return "" ; |
| 737 | else |
| 738 | return json->data.string; |
| 739 | } |
| 740 | |
| 741 | double json_get_number(json_stream *json) |
| 742 | { |
| 743 | char *p = json->data.string; |
| 744 | return p == NULL ? 0 : strtod(p, NULL); |
| 745 | } |
| 746 | |
| 747 | const char *json_get_error(json_stream *json) |
| 748 | { |
| 749 | return json->error ? json->errmsg : NULL; |
| 750 | } |
| 751 | |
| 752 | size_t json_get_lineno(json_stream *json) |
| 753 | { |
| 754 | return json->lineno; |
| 755 | } |
| 756 | |
| 757 | size_t json_get_position(json_stream *json) |
| 758 | { |
| 759 | return json->source.position; |
| 760 | } |
| 761 | |
| 762 | size_t json_get_depth(json_stream *json) |
| 763 | { |
| 764 | return json->stack_top + 1; |
| 765 | } |
| 766 | |
| 767 | void json_open_buffer(json_stream *json, const void *buffer, size_t size) |
| 768 | { |
| 769 | init(json); |
| 770 | json->source.get = buffer_get; |
| 771 | json->source.peek = buffer_peek; |
| 772 | json->source.source.buffer.buffer = buffer; |
| 773 | json->source.source.buffer.length = size; |
| 774 | } |
| 775 | |
| 776 | void json_open_string(json_stream *json, const char *string) |
| 777 | { |
| 778 | json_open_buffer(json, string, strlen(string)); |
| 779 | } |
| 780 | |
| 781 | void json_open_stream(json_stream *json, FILE * stream) |
| 782 | { |
| 783 | init(json); |
| 784 | json->source.get = stream_get; |
| 785 | json->source.peek = stream_peek; |
| 786 | json->source.source.stream.stream = stream; |
| 787 | } |
| 788 | |
| 789 | void json_set_allocator(json_stream *json, json_allocator *a) |
| 790 | { |
| 791 | json->alloc = *a; |
| 792 | } |
| 793 | |
| 794 | void json_set_streaming(json_stream *json, bool streaming) |
| 795 | { |
| 796 | json->streaming = streaming; |
| 797 | } |
| 798 | |
| 799 | void json_close(json_stream *json) |
| 800 | { |
| 801 | pop_all(json); |
| 802 | json->alloc.free(json->data.string); |
| 803 | } |
| 804 | |