| 1 | #define _POSIX_C_SOURCE 200112L |
| 2 | #include <stdio.h> |
| 3 | #include <stdbool.h> |
| 4 | #include <stdlib.h> |
| 5 | #include <string.h> |
| 6 | #include <ctype.h> |
| 7 | #include <errno.h> |
| 8 | #include "pdjson.h" |
| 9 | |
| 10 | #define JSON_FLAG_ERROR (1u << 0) |
| 11 | #define JSON_FLAG_STREAMING (1u << 1) |
| 12 | |
| 13 | #define json_error(json, format, ...) \ |
| 14 | if (!(json->flags & JSON_FLAG_ERROR)) { \ |
| 15 | json->flags |= JSON_FLAG_ERROR; \ |
| 16 | snprintf(json->errmsg, sizeof(json->errmsg), \ |
| 17 | "error: %lu: " format, \ |
| 18 | (unsigned long) json->lineno, \ |
| 19 | __VA_ARGS__); \ |
| 20 | } \ |
| 21 | |
| 22 | #define STACK_INC 4 |
| 23 | |
| 24 | #if defined(_MSC_VER) || defined(__MINGW32__) |
| 25 | #define strerror_r(err, buf, len) strerror_s(buf, len, err) |
| 26 | #endif |
| 27 | |
| 28 | const char *json_typename[] = { |
| 29 | [JSON_ERROR] = "ERROR" , |
| 30 | [JSON_DONE] = "DONE" , |
| 31 | [JSON_OBJECT] = "OBJECT" , |
| 32 | [JSON_OBJECT_END] = "OBJECT_END" , |
| 33 | [JSON_ARRAY] = "ARRAY" , |
| 34 | [JSON_ARRAY_END] = "ARRAY_END" , |
| 35 | [JSON_STRING] = "STRING" , |
| 36 | [JSON_NUMBER] = "NUMBER" , |
| 37 | [JSON_TRUE] = "TRUE" , |
| 38 | [JSON_FALSE] = "FALSE" , |
| 39 | [JSON_NULL] = "NULL" , |
| 40 | }; |
| 41 | |
| 42 | struct json_stack { |
| 43 | enum json_type type; |
| 44 | long count; |
| 45 | }; |
| 46 | |
| 47 | static void json_error_s(json_stream *json, int err) |
| 48 | { |
| 49 | char errbuf[1024] = {0}; |
| 50 | strerror_r(err, errbuf, sizeof(errbuf)); |
| 51 | json_error(json, "%s" , errbuf); |
| 52 | } |
| 53 | |
| 54 | static enum json_type |
| 55 | push(json_stream *json, enum json_type type) |
| 56 | { |
| 57 | json->stack_top++; |
| 58 | |
| 59 | if (json->stack_top >= json->stack_size) { |
| 60 | struct json_stack *stack; |
| 61 | stack = json->alloc.realloc(json->stack, |
| 62 | (json->stack_size + STACK_INC) * sizeof(*json->stack)); |
| 63 | if (stack == NULL) { |
| 64 | json_error_s(json, errno); |
| 65 | return JSON_ERROR; |
| 66 | } |
| 67 | |
| 68 | json->stack_size += STACK_INC; |
| 69 | json->stack = stack; |
| 70 | } |
| 71 | |
| 72 | json->stack[json->stack_top].type = type; |
| 73 | json->stack[json->stack_top].count = 0; |
| 74 | |
| 75 | return type; |
| 76 | } |
| 77 | |
| 78 | static enum json_type |
| 79 | pop(json_stream *json, int c, enum json_type expected) |
| 80 | { |
| 81 | if (json->stack == NULL || json->stack[json->stack_top].type != expected) { |
| 82 | json_error(json, "unexpected byte, '%c'" , c); |
| 83 | return JSON_ERROR; |
| 84 | } |
| 85 | json->stack_top--; |
| 86 | return expected == JSON_ARRAY ? JSON_ARRAY_END : JSON_OBJECT_END; |
| 87 | } |
| 88 | |
| 89 | static int buffer_peek(struct json_source *source) |
| 90 | { |
| 91 | if (source->position < source->source.buffer.length) |
| 92 | return source->source.buffer.buffer[source->position]; |
| 93 | else |
| 94 | return EOF; |
| 95 | } |
| 96 | |
| 97 | static int buffer_get(struct json_source *source) |
| 98 | { |
| 99 | int c = source->peek(source); |
| 100 | source->position++; |
| 101 | return c; |
| 102 | } |
| 103 | |
| 104 | static int stream_get(struct json_source *source) |
| 105 | { |
| 106 | source->position++; |
| 107 | return fgetc(source->source.stream.stream); |
| 108 | } |
| 109 | |
| 110 | static int stream_peek(struct json_source *source) |
| 111 | { |
| 112 | int c = fgetc(source->source.stream.stream); |
| 113 | ungetc(c, source->source.stream.stream); |
| 114 | return c; |
| 115 | } |
| 116 | |
| 117 | static void init(json_stream *json) |
| 118 | { |
| 119 | json->lineno = 1; |
| 120 | json->flags = JSON_FLAG_STREAMING; |
| 121 | json->errmsg[0] = '\0'; |
| 122 | json->ntokens = 0; |
| 123 | json->next = 0; |
| 124 | |
| 125 | json->stack = NULL; |
| 126 | json->stack_top = (size_t)(-1); |
| 127 | json->stack_size = 0; |
| 128 | |
| 129 | json->data.string = NULL; |
| 130 | json->data.string_size = 0; |
| 131 | json->data.string_fill = 0; |
| 132 | json->source.position = 0; |
| 133 | |
| 134 | json->alloc.malloc = malloc; |
| 135 | json->alloc.realloc = realloc; |
| 136 | json->alloc.free = free; |
| 137 | } |
| 138 | |
| 139 | static enum json_type |
| 140 | is_match(json_stream *json, const char *pattern, enum json_type type) |
| 141 | { |
| 142 | for (const char *p = pattern; *p; p++) |
| 143 | if (*p != json->source.get(&json->source)) |
| 144 | return JSON_ERROR; |
| 145 | return type; |
| 146 | } |
| 147 | |
| 148 | static int pushchar(json_stream *json, int c) |
| 149 | { |
| 150 | if (json->data.string_fill == json->data.string_size) { |
| 151 | size_t size = json->data.string_size * 2; |
| 152 | char *buffer = json->alloc.realloc(json->data.string, size); |
| 153 | if (buffer == NULL) { |
| 154 | json_error_s(json, errno); |
| 155 | return -1; |
| 156 | } else { |
| 157 | json->data.string_size = size; |
| 158 | json->data.string = buffer; |
| 159 | } |
| 160 | } |
| 161 | json->data.string[json->data.string_fill++] = (char)(c); |
| 162 | return 0; |
| 163 | } |
| 164 | |
| 165 | static int init_string(json_stream *json) |
| 166 | { |
| 167 | json->data.string_fill = 0; |
| 168 | if (json->data.string == NULL) { |
| 169 | json->data.string_size = 1024; |
| 170 | json->data.string = json->alloc.malloc(json->data.string_size); |
| 171 | if (json->data.string == NULL) { |
| 172 | json_error_s(json, errno); |
| 173 | return -1; |
| 174 | } |
| 175 | } |
| 176 | json->data.string[0] = '\0'; |
| 177 | return 0; |
| 178 | } |
| 179 | |
| 180 | static int encode_utf8(json_stream *json, unsigned long c) |
| 181 | { |
| 182 | if (c < 0x80UL) { |
| 183 | return pushchar(json, c); |
| 184 | } else if (c < 0x0800UL) { |
| 185 | return !((pushchar(json, (c >> 6 & 0x1F) | 0xC0) == 0) && |
| 186 | (pushchar(json, (c >> 0 & 0x3F) | 0x80) == 0)); |
| 187 | } else if (c < 0x010000UL) { |
| 188 | if (c >= 0xd800 && c <= 0xdfff) { |
| 189 | json_error(json, "invalid codepoint %06lx" , c); |
| 190 | return -1; |
| 191 | } |
| 192 | return !((pushchar(json, (c >> 12 & 0x0F) | 0xE0) == 0) && |
| 193 | (pushchar(json, (c >> 6 & 0x3F) | 0x80) == 0) && |
| 194 | (pushchar(json, (c >> 0 & 0x3F) | 0x80) == 0)); |
| 195 | } else if (c < 0x110000UL) { |
| 196 | return !((pushchar(json, (c >> 18 & 0x07) | 0xF0) == 0) && |
| 197 | (pushchar(json, (c >> 12 & 0x3F) | 0x80) == 0) && |
| 198 | (pushchar(json, (c >> 6 & 0x3F) | 0x80) == 0) && |
| 199 | (pushchar(json, (c >> 0 & 0x3F) | 0x80) == 0)); |
| 200 | } else { |
| 201 | json_error(json, "can't encode UTF-8 for %06lx" , c); |
| 202 | return -1; |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | static int hexchar(int c) |
| 207 | { |
| 208 | switch (c) { |
| 209 | case '0': return 0; |
| 210 | case '1': return 1; |
| 211 | case '2': return 2; |
| 212 | case '3': return 3; |
| 213 | case '4': return 4; |
| 214 | case '5': return 5; |
| 215 | case '6': return 6; |
| 216 | case '7': return 7; |
| 217 | case '8': return 8; |
| 218 | case '9': return 9; |
| 219 | case 'a': |
| 220 | case 'A': return 10; |
| 221 | case 'b': |
| 222 | case 'B': return 11; |
| 223 | case 'c': |
| 224 | case 'C': return 12; |
| 225 | case 'd': |
| 226 | case 'D': return 13; |
| 227 | case 'e': |
| 228 | case 'E': return 14; |
| 229 | case 'f': |
| 230 | case 'F': return 15; |
| 231 | default: |
| 232 | return -1; |
| 233 | } |
| 234 | } |
| 235 | |
| 236 | static long |
| 237 | read_unicode_cp(json_stream *json) |
| 238 | { |
| 239 | long cp = 0; |
| 240 | int shift = 12; |
| 241 | |
| 242 | for (size_t i = 0; i < 4; i++) { |
| 243 | int c = json->source.get(&json->source); |
| 244 | int hc; |
| 245 | |
| 246 | if (c == EOF) { |
| 247 | json_error(json, "%s" , "unterminated string literal in unicode" ); |
| 248 | return -1; |
| 249 | } else if ((hc = hexchar(c)) == -1) { |
| 250 | json_error(json, "bad escape unicode byte, '%c'" , c); |
| 251 | return -1; |
| 252 | } |
| 253 | |
| 254 | cp += hc * (1 << shift); |
| 255 | shift -= 4; |
| 256 | } |
| 257 | |
| 258 | |
| 259 | return cp; |
| 260 | } |
| 261 | |
| 262 | static int read_unicode(json_stream *json) |
| 263 | { |
| 264 | long cp, h, l; |
| 265 | |
| 266 | if ((cp = read_unicode_cp(json)) == -1) { |
| 267 | return -1; |
| 268 | } |
| 269 | |
| 270 | if (cp >= 0xd800 && cp <= 0xdbff) { |
| 271 | /* This is the high portion of a surrogate pair; we need to read the |
| 272 | * lower portion to get the codepoint |
| 273 | */ |
| 274 | h = cp; |
| 275 | |
| 276 | int c = json->source.get(&json->source); |
| 277 | if (c == EOF) { |
| 278 | json_error(json, "%s" , "unterminated string literal in unicode" ); |
| 279 | return -1; |
| 280 | } else if (c != '\\') { |
| 281 | json_error(json, "invalid continuation for surrogate pair: '%c', " |
| 282 | "expected '\\'" , c); |
| 283 | return -1; |
| 284 | } |
| 285 | |
| 286 | c = json->source.get(&json->source); |
| 287 | if (c == EOF) { |
| 288 | json_error(json, "%s" , "unterminated string literal in unicode" ); |
| 289 | return -1; |
| 290 | } else if (c != 'u') { |
| 291 | json_error(json, "invalid continuation for surrogate pair: '%c', " |
| 292 | "expected 'u'" , c); |
| 293 | return -1; |
| 294 | } |
| 295 | |
| 296 | if ((l = read_unicode_cp(json)) == -1) { |
| 297 | return -1; |
| 298 | } |
| 299 | |
| 300 | if (l < 0xdc00 || l > 0xdfff) { |
| 301 | json_error(json, "invalid surrogate pair continuation \\u%04lx out " |
| 302 | "of range (dc00-dfff)" , l); |
| 303 | return -1; |
| 304 | } |
| 305 | |
| 306 | cp = ((h - 0xd800) * 0x400) + ((l - 0xdc00) + 0x10000); |
| 307 | } else if (cp >= 0xdc00 && cp <= 0xdfff) { |
| 308 | json_error(json, "dangling surrogate \\u%04lx" , cp); |
| 309 | return -1; |
| 310 | } |
| 311 | |
| 312 | return encode_utf8(json, cp); |
| 313 | } |
| 314 | |
| 315 | int read_escaped(json_stream *json) |
| 316 | { |
| 317 | int c = json->source.get(&json->source); |
| 318 | if (c == EOF) { |
| 319 | json_error(json, "%s" , "unterminated string literal in escape" ); |
| 320 | return -1; |
| 321 | } else if (c == 'u') { |
| 322 | if (read_unicode(json) != 0) |
| 323 | return -1; |
| 324 | } else { |
| 325 | switch (c) { |
| 326 | case '\\': |
| 327 | case 'b': |
| 328 | case 'f': |
| 329 | case 'n': |
| 330 | case 'r': |
| 331 | case 't': |
| 332 | case '/': |
| 333 | case '"': |
| 334 | { |
| 335 | const char *codes = "\\bfnrt/\"" ; |
| 336 | char *p = strchr(codes, c); |
| 337 | if (pushchar(json, "\\\b\f\n\r\t/\"" [p - codes]) != 0) |
| 338 | return -1; |
| 339 | } |
| 340 | break; |
| 341 | default: |
| 342 | json_error(json, "bad escaped byte, '%c'" , c); |
| 343 | return -1; |
| 344 | } |
| 345 | } |
| 346 | return 0; |
| 347 | } |
| 348 | |
| 349 | static int |
| 350 | char_needs_escaping(int c) |
| 351 | { |
| 352 | if ((c >= 0) && (c < 0x20 || c == 0x22 || c == 0x5c)) { |
| 353 | return 1; |
| 354 | } |
| 355 | |
| 356 | return 0; |
| 357 | } |
| 358 | |
| 359 | static int |
| 360 | utf8_seq_length(char byte) |
| 361 | { |
| 362 | unsigned char u = (unsigned char) byte; |
| 363 | if (u < 0x80) return 1; |
| 364 | |
| 365 | if (0x80 <= u && u <= 0xBF) |
| 366 | { |
| 367 | // second, third or fourth byte of a multi-byte |
| 368 | // sequence, i.e. a "continuation byte" |
| 369 | return 0; |
| 370 | } |
| 371 | else if (u == 0xC0 || u == 0xC1) |
| 372 | { |
| 373 | // overlong encoding of an ASCII byte |
| 374 | return 0; |
| 375 | } |
| 376 | else if (0xC2 <= u && u <= 0xDF) |
| 377 | { |
| 378 | // 2-byte sequence |
| 379 | return 2; |
| 380 | } |
| 381 | else if (0xE0 <= u && u <= 0xEF) |
| 382 | { |
| 383 | // 3-byte sequence |
| 384 | return 3; |
| 385 | } |
| 386 | else if (0xF0 <= u && u <= 0xF4) |
| 387 | { |
| 388 | // 4-byte sequence |
| 389 | return 4; |
| 390 | } |
| 391 | else |
| 392 | { |
| 393 | // u >= 0xF5 |
| 394 | // Restricted (start of 4-, 5- or 6-byte sequence) or invalid UTF-8 |
| 395 | return 0; |
| 396 | } |
| 397 | } |
| 398 | |
| 399 | static int |
| 400 | is_legal_utf8(const unsigned char *bytes, int length) |
| 401 | { |
| 402 | if (0 == bytes || 0 == length) return 0; |
| 403 | |
| 404 | unsigned char a; |
| 405 | const unsigned char* srcptr = bytes + length; |
| 406 | switch (length) |
| 407 | { |
| 408 | default: |
| 409 | return 0; |
| 410 | // Everything else falls through when true. |
| 411 | case 4: |
| 412 | if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return 0; |
| 413 | case 3: |
| 414 | if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return 0; |
| 415 | case 2: |
| 416 | a = (*--srcptr); |
| 417 | switch (*bytes) |
| 418 | { |
| 419 | case 0xE0: |
| 420 | if (a < 0xA0 || a > 0xBF) return 0; |
| 421 | break; |
| 422 | case 0xED: |
| 423 | if (a < 0x80 || a > 0x9F) return 0; |
| 424 | break; |
| 425 | case 0xF0: |
| 426 | if (a < 0x90 || a > 0xBF) return 0; |
| 427 | break; |
| 428 | case 0xF4: |
| 429 | if (a < 0x80 || a > 0x8F) return 0; |
| 430 | break; |
| 431 | default: |
| 432 | if (a < 0x80 || a > 0xBF) return 0; |
| 433 | } |
| 434 | case 1: |
| 435 | if (*bytes >= 0x80 && *bytes < 0xC2) return 0; |
| 436 | } |
| 437 | return *bytes <= 0xF4; |
| 438 | } |
| 439 | |
| 440 | static int |
| 441 | read_utf8(json_stream* json, int next_char) |
| 442 | { |
| 443 | int count = utf8_seq_length((char)(next_char)); |
| 444 | if (!count) |
| 445 | { |
| 446 | json_error(json, "%s" , "Bad character." ); |
| 447 | return -1; |
| 448 | } |
| 449 | |
| 450 | char buffer[4]; |
| 451 | buffer[0] = (char)(next_char); |
| 452 | for (int i = 1; i < count; ++i) |
| 453 | { |
| 454 | buffer[i] = (char)(json->source.get(&json->source)); |
| 455 | } |
| 456 | |
| 457 | if (!is_legal_utf8((unsigned char*) buffer, count)) |
| 458 | { |
| 459 | json_error(json, "%s" , "No legal UTF8 found" ); |
| 460 | return -1; |
| 461 | } |
| 462 | |
| 463 | for (int i = 0; i < count; ++i) |
| 464 | { |
| 465 | if (pushchar(json, buffer[i]) != 0) |
| 466 | return -1; |
| 467 | } |
| 468 | return 0; |
| 469 | } |
| 470 | |
| 471 | static enum json_type |
| 472 | read_string(json_stream *json) |
| 473 | { |
| 474 | if (init_string(json) != 0) |
| 475 | return JSON_ERROR; |
| 476 | while (1) { |
| 477 | int c = json->source.get(&json->source); |
| 478 | if (c == EOF) { |
| 479 | json_error(json, "%s" , "unterminated string literal" ); |
| 480 | return JSON_ERROR; |
| 481 | } else if (c == '"') { |
| 482 | if (pushchar(json, '\0') == 0) |
| 483 | return JSON_STRING; |
| 484 | else |
| 485 | return JSON_ERROR; |
| 486 | } else if (c == '\\') { |
| 487 | if (read_escaped(json) != 0) |
| 488 | return JSON_ERROR; |
| 489 | } else if ((unsigned) c >= 0x80) { |
| 490 | if (read_utf8(json, c) != 0) |
| 491 | return JSON_ERROR; |
| 492 | } else { |
| 493 | if (char_needs_escaping(c)) { |
| 494 | json_error(json, "%s" , "unescaped control character in string" ); |
| 495 | return JSON_ERROR; |
| 496 | } |
| 497 | |
| 498 | if (pushchar(json, c) != 0) |
| 499 | return JSON_ERROR; |
| 500 | } |
| 501 | } |
| 502 | return JSON_ERROR; |
| 503 | } |
| 504 | |
| 505 | static int |
| 506 | is_digit(int c) |
| 507 | { |
| 508 | return c >= 48 /*0*/ && c <= 57 /*9*/; |
| 509 | } |
| 510 | |
| 511 | static int |
| 512 | read_digits(json_stream *json) |
| 513 | { |
| 514 | unsigned nread = 0; |
| 515 | while (is_digit(json->source.peek(&json->source))) { |
| 516 | if (pushchar(json, json->source.get(&json->source)) != 0) |
| 517 | return -1; |
| 518 | |
| 519 | nread++; |
| 520 | } |
| 521 | |
| 522 | if (nread == 0) { |
| 523 | return -1; |
| 524 | } |
| 525 | |
| 526 | return 0; |
| 527 | } |
| 528 | |
| 529 | static enum json_type |
| 530 | read_number(json_stream *json, int c) |
| 531 | { |
| 532 | if (pushchar(json, c) != 0) |
| 533 | return JSON_ERROR; |
| 534 | if (c == '-') { |
| 535 | c = json->source.get(&json->source); |
| 536 | if (is_digit(c)) { |
| 537 | return read_number(json, c); |
| 538 | } else { |
| 539 | json_error(json, "unexpected byte, '%c'" , c); |
| 540 | } |
| 541 | } else if (strchr("123456789" , c) != NULL) { |
| 542 | c = json->source.peek(&json->source); |
| 543 | if (is_digit(c)) { |
| 544 | if (read_digits(json) != 0) |
| 545 | return JSON_ERROR; |
| 546 | } |
| 547 | } |
| 548 | /* Up to decimal or exponent has been read. */ |
| 549 | c = json->source.peek(&json->source); |
| 550 | if (strchr(".eE" , c) == NULL) { |
| 551 | if (pushchar(json, '\0') != 0) |
| 552 | return JSON_ERROR; |
| 553 | else |
| 554 | return JSON_NUMBER; |
| 555 | } |
| 556 | if (c == '.') { |
| 557 | json->source.get(&json->source); // consume . |
| 558 | if (pushchar(json, c) != 0) |
| 559 | return JSON_ERROR; |
| 560 | if (read_digits(json) != 0) |
| 561 | return JSON_ERROR; |
| 562 | } |
| 563 | /* Check for exponent. */ |
| 564 | c = json->source.peek(&json->source); |
| 565 | if (c == 'e' || c == 'E') { |
| 566 | json->source.get(&json->source); // consume e/E |
| 567 | if (pushchar(json, c) != 0) |
| 568 | return JSON_ERROR; |
| 569 | c = json->source.peek(&json->source); |
| 570 | if (c == '+' || c == '-') { |
| 571 | json->source.get(&json->source); // consume |
| 572 | if (pushchar(json, c) != 0) |
| 573 | return JSON_ERROR; |
| 574 | if (read_digits(json) != 0) |
| 575 | return JSON_ERROR; |
| 576 | } else if (is_digit(c)) { |
| 577 | if (read_digits(json) != 0) |
| 578 | return JSON_ERROR; |
| 579 | } else { |
| 580 | json_error(json, "unexpected byte in number, '%c'" , c); |
| 581 | return JSON_ERROR; |
| 582 | } |
| 583 | } |
| 584 | if (pushchar(json, '\0') != 0) |
| 585 | return JSON_ERROR; |
| 586 | else |
| 587 | return JSON_NUMBER; |
| 588 | } |
| 589 | |
| 590 | static int |
| 591 | json_isspace(int c) |
| 592 | { |
| 593 | switch (c) { |
| 594 | case 0x09: |
| 595 | case 0x0a: |
| 596 | case 0x0d: |
| 597 | case 0x20: |
| 598 | return 1; |
| 599 | } |
| 600 | |
| 601 | return 0; |
| 602 | } |
| 603 | |
| 604 | /* Returns the next non-whitespace character in the stream. */ |
| 605 | static int next(json_stream *json) |
| 606 | { |
| 607 | int c; |
| 608 | while (json_isspace(c = json->source.get(&json->source))) |
| 609 | if (c == '\n') |
| 610 | json->lineno++; |
| 611 | return c; |
| 612 | } |
| 613 | |
| 614 | static enum json_type |
| 615 | read_value(json_stream *json, int c) |
| 616 | { |
| 617 | json->ntokens++; |
| 618 | switch (c) { |
| 619 | case EOF: |
| 620 | json_error(json, "%s" , "unexpected end of data" ); |
| 621 | return JSON_ERROR; |
| 622 | case '{': |
| 623 | return push(json, JSON_OBJECT); |
| 624 | case '[': |
| 625 | return push(json, JSON_ARRAY); |
| 626 | case '"': |
| 627 | return read_string(json); |
| 628 | case 'n': |
| 629 | return is_match(json, "ull" , JSON_NULL); |
| 630 | case 'f': |
| 631 | return is_match(json, "alse" , JSON_FALSE); |
| 632 | case 't': |
| 633 | return is_match(json, "rue" , JSON_TRUE); |
| 634 | case '0': |
| 635 | case '1': |
| 636 | case '2': |
| 637 | case '3': |
| 638 | case '4': |
| 639 | case '5': |
| 640 | case '6': |
| 641 | case '7': |
| 642 | case '8': |
| 643 | case '9': |
| 644 | case '-': |
| 645 | if (init_string(json) != 0) |
| 646 | return JSON_ERROR; |
| 647 | return read_number(json, c); |
| 648 | default: |
| 649 | json_error(json, "unexpected byte, '%c'" , c); |
| 650 | return JSON_ERROR; |
| 651 | } |
| 652 | } |
| 653 | |
| 654 | enum json_type json_peek(json_stream *json) |
| 655 | { |
| 656 | enum json_type next = json_next(json); |
| 657 | json->next = next; |
| 658 | return next; |
| 659 | } |
| 660 | |
| 661 | enum json_type json_next(json_stream *json) |
| 662 | { |
| 663 | if (json->flags & JSON_FLAG_ERROR) |
| 664 | return JSON_ERROR; |
| 665 | if (json->next != 0) { |
| 666 | enum json_type next = json->next; |
| 667 | json->next = 0; |
| 668 | return next; |
| 669 | } |
| 670 | if (json->ntokens > 0 && json->stack_top == (size_t)-1) { |
| 671 | int c; |
| 672 | |
| 673 | do { |
| 674 | c = json->source.peek(&json->source); |
| 675 | if (json_isspace(c)) { |
| 676 | c = json->source.get(&json->source); |
| 677 | } |
| 678 | } while (json_isspace(c)); |
| 679 | |
| 680 | if (!(json->flags & JSON_FLAG_STREAMING) && c != EOF) { |
| 681 | return JSON_ERROR; |
| 682 | } |
| 683 | |
| 684 | return JSON_DONE; |
| 685 | } |
| 686 | int c = next(json); |
| 687 | if (json->stack_top == (size_t)-1) |
| 688 | return read_value(json, c); |
| 689 | if (json->stack[json->stack_top].type == JSON_ARRAY) { |
| 690 | if (json->stack[json->stack_top].count == 0) { |
| 691 | if (c == ']') { |
| 692 | return pop(json, c, JSON_ARRAY); |
| 693 | } |
| 694 | json->stack[json->stack_top].count++; |
| 695 | return read_value(json, c); |
| 696 | } else if (c == ',') { |
| 697 | json->stack[json->stack_top].count++; |
| 698 | return read_value(json, next(json)); |
| 699 | } else if (c == ']') { |
| 700 | return pop(json, c, JSON_ARRAY); |
| 701 | } else { |
| 702 | json_error(json, "unexpected byte, '%c'" , c); |
| 703 | return JSON_ERROR; |
| 704 | } |
| 705 | } else if (json->stack[json->stack_top].type == JSON_OBJECT) { |
| 706 | if (json->stack[json->stack_top].count == 0) { |
| 707 | if (c == '}') { |
| 708 | return pop(json, c, JSON_OBJECT); |
| 709 | } |
| 710 | |
| 711 | /* No property value pairs yet. */ |
| 712 | enum json_type value = read_value(json, c); |
| 713 | if (value != JSON_STRING) { |
| 714 | json_error(json, "%s" , "expected property name or '}'" ); |
| 715 | return JSON_ERROR; |
| 716 | } else { |
| 717 | json->stack[json->stack_top].count++; |
| 718 | return value; |
| 719 | } |
| 720 | } else if ((json->stack[json->stack_top].count % 2) == 0) { |
| 721 | /* Expecting comma followed by property name. */ |
| 722 | if (c != ',' && c != '}') { |
| 723 | json_error(json, "%s" , "expected ',' or '}'" ); |
| 724 | return JSON_ERROR; |
| 725 | } else if (c == '}') { |
| 726 | return pop(json, c, JSON_OBJECT); |
| 727 | } else { |
| 728 | enum json_type value = read_value(json, next(json)); |
| 729 | if (value != JSON_STRING) { |
| 730 | json_error(json, "%s" , "expected property name" ); |
| 731 | return JSON_ERROR; |
| 732 | } else { |
| 733 | json->stack[json->stack_top].count++; |
| 734 | return value; |
| 735 | } |
| 736 | } |
| 737 | } else if ((json->stack[json->stack_top].count % 2) == 1) { |
| 738 | /* Expecting colon followed by value. */ |
| 739 | if (c != ':') { |
| 740 | json_error(json, "%s" , "expected ':' after property name" ); |
| 741 | return JSON_ERROR; |
| 742 | } else { |
| 743 | json->stack[json->stack_top].count++; |
| 744 | return read_value(json, next(json)); |
| 745 | } |
| 746 | } |
| 747 | } |
| 748 | json_error(json, "%s" , "invalid parser state" ); |
| 749 | return JSON_ERROR; |
| 750 | } |
| 751 | |
| 752 | void json_reset(json_stream *json) |
| 753 | { |
| 754 | json->stack_top = (size_t)(-1); |
| 755 | json->ntokens = 0; |
| 756 | json->flags &= ~JSON_FLAG_ERROR; |
| 757 | json->errmsg[0] = '\0'; |
| 758 | } |
| 759 | |
| 760 | const char *json_get_string(json_stream *json, size_t *length) |
| 761 | { |
| 762 | if (length != NULL) |
| 763 | *length = json->data.string_fill; |
| 764 | if (json->data.string == NULL) |
| 765 | return "" ; |
| 766 | else |
| 767 | return json->data.string; |
| 768 | } |
| 769 | |
| 770 | double json_get_number(json_stream *json) |
| 771 | { |
| 772 | char *p = json->data.string; |
| 773 | return p == NULL ? 0 : strtod(p, NULL); |
| 774 | } |
| 775 | |
| 776 | const char *json_get_error(json_stream *json) |
| 777 | { |
| 778 | return json->flags & JSON_FLAG_ERROR ? json->errmsg : NULL; |
| 779 | } |
| 780 | |
| 781 | size_t json_get_lineno(json_stream *json) |
| 782 | { |
| 783 | return json->lineno; |
| 784 | } |
| 785 | |
| 786 | size_t json_get_position(json_stream *json) |
| 787 | { |
| 788 | return json->source.position; |
| 789 | } |
| 790 | |
| 791 | size_t json_get_depth(json_stream *json) |
| 792 | { |
| 793 | return json->stack_top + 1; |
| 794 | } |
| 795 | |
| 796 | void json_open_buffer(json_stream *json, const void *buffer, size_t size) |
| 797 | { |
| 798 | init(json); |
| 799 | json->source.get = buffer_get; |
| 800 | json->source.peek = buffer_peek; |
| 801 | json->source.source.buffer.buffer = buffer; |
| 802 | json->source.source.buffer.length = size; |
| 803 | } |
| 804 | |
| 805 | void json_open_string(json_stream *json, const char *string) |
| 806 | { |
| 807 | json_open_buffer(json, string, strlen(string)); |
| 808 | } |
| 809 | |
| 810 | void json_open_stream(json_stream *json, FILE * stream) |
| 811 | { |
| 812 | init(json); |
| 813 | json->source.get = stream_get; |
| 814 | json->source.peek = stream_peek; |
| 815 | json->source.source.stream.stream = stream; |
| 816 | } |
| 817 | |
| 818 | static int user_get(struct json_source *json) |
| 819 | { |
| 820 | return json->source.user.get(json->source.user.ptr); |
| 821 | } |
| 822 | |
| 823 | static int user_peek(struct json_source *json) |
| 824 | { |
| 825 | return json->source.user.peek(json->source.user.ptr); |
| 826 | } |
| 827 | |
| 828 | void json_open_user(json_stream *json, json_user_io get, json_user_io peek, void *user) |
| 829 | { |
| 830 | init(json); |
| 831 | json->source.get = user_get; |
| 832 | json->source.peek = user_peek; |
| 833 | json->source.source.user.ptr = user; |
| 834 | json->source.source.user.get = get; |
| 835 | json->source.source.user.peek = peek; |
| 836 | } |
| 837 | |
| 838 | void json_set_allocator(json_stream *json, json_allocator *a) |
| 839 | { |
| 840 | json->alloc = *a; |
| 841 | } |
| 842 | |
| 843 | void json_set_streaming(json_stream *json, bool streaming) |
| 844 | { |
| 845 | if (streaming) |
| 846 | json->flags |= JSON_FLAG_STREAMING; |
| 847 | else |
| 848 | json->flags &= ~JSON_FLAG_STREAMING; |
| 849 | } |
| 850 | |
| 851 | void json_close(json_stream *json) |
| 852 | { |
| 853 | json->alloc.free(json->stack); |
| 854 | json->alloc.free(json->data.string); |
| 855 | } |
| 856 | |