Anthony Liguori | 4a5fcab | 2009-11-11 10:39:23 -0600 | [diff] [blame] | 1 | /* |
| 2 | * JSON Parser |
| 3 | * |
| 4 | * Copyright IBM, Corp. 2009 |
| 5 | * |
| 6 | * Authors: |
| 7 | * Anthony Liguori <aliguori@us.ibm.com> |
| 8 | * |
| 9 | * This work is licensed under the terms of the GNU LGPL, version 2.1 or later. |
| 10 | * See the COPYING.LIB file in the top-level directory. |
| 11 | * |
| 12 | */ |
| 13 | |
| 14 | #include <stdbool.h> |
| 15 | |
| 16 | #include "qemu-common.h" |
| 17 | #include "qstring.h" |
| 18 | #include "qint.h" |
| 19 | #include "qdict.h" |
| 20 | #include "qlist.h" |
| 21 | #include "qfloat.h" |
| 22 | #include "qbool.h" |
| 23 | #include "json-parser.h" |
| 24 | #include "json-lexer.h" |
| 25 | |
| 26 | typedef struct JSONParserContext |
| 27 | { |
| 28 | } JSONParserContext; |
| 29 | |
| 30 | #define BUG_ON(cond) assert(!(cond)) |
| 31 | |
| 32 | /** |
| 33 | * TODO |
| 34 | * |
| 35 | * 0) make errors meaningful again |
| 36 | * 1) add geometry information to tokens |
| 37 | * 3) should we return a parsed size? |
| 38 | * 4) deal with premature EOI |
| 39 | */ |
| 40 | |
| 41 | static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap); |
| 42 | |
| 43 | /** |
| 44 | * Token manipulators |
| 45 | * |
| 46 | * tokens are dictionaries that contain a type, a string value, and geometry information |
| 47 | * about a token identified by the lexer. These are routines that make working with |
| 48 | * these objects a bit easier. |
| 49 | */ |
| 50 | static const char *token_get_value(QObject *obj) |
| 51 | { |
| 52 | return qdict_get_str(qobject_to_qdict(obj), "token"); |
| 53 | } |
| 54 | |
| 55 | static JSONTokenType token_get_type(QObject *obj) |
| 56 | { |
| 57 | return qdict_get_int(qobject_to_qdict(obj), "type"); |
| 58 | } |
| 59 | |
| 60 | static int token_is_operator(QObject *obj, char op) |
| 61 | { |
| 62 | const char *val; |
| 63 | |
| 64 | if (token_get_type(obj) != JSON_OPERATOR) { |
| 65 | return 0; |
| 66 | } |
| 67 | |
| 68 | val = token_get_value(obj); |
| 69 | |
| 70 | return (val[0] == op) && (val[1] == 0); |
| 71 | } |
| 72 | |
| 73 | static int token_is_keyword(QObject *obj, const char *value) |
| 74 | { |
| 75 | if (token_get_type(obj) != JSON_KEYWORD) { |
| 76 | return 0; |
| 77 | } |
| 78 | |
| 79 | return strcmp(token_get_value(obj), value) == 0; |
| 80 | } |
| 81 | |
| 82 | static int token_is_escape(QObject *obj, const char *value) |
| 83 | { |
| 84 | if (token_get_type(obj) != JSON_ESCAPE) { |
| 85 | return 0; |
| 86 | } |
| 87 | |
| 88 | return (strcmp(token_get_value(obj), value) == 0); |
| 89 | } |
| 90 | |
| 91 | /** |
| 92 | * Error handler |
| 93 | */ |
| 94 | static void parse_error(JSONParserContext *ctxt, QObject *token, const char *msg, ...) |
| 95 | { |
| 96 | fprintf(stderr, "parse error: %s\n", msg); |
| 97 | } |
| 98 | |
| 99 | /** |
| 100 | * String helpers |
| 101 | * |
| 102 | * These helpers are used to unescape strings. |
| 103 | */ |
| 104 | static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length) |
| 105 | { |
| 106 | if (wchar <= 0x007F) { |
| 107 | BUG_ON(buffer_length < 2); |
| 108 | |
| 109 | buffer[0] = wchar & 0x7F; |
| 110 | buffer[1] = 0; |
| 111 | } else if (wchar <= 0x07FF) { |
| 112 | BUG_ON(buffer_length < 3); |
| 113 | |
| 114 | buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F); |
| 115 | buffer[1] = 0x80 | (wchar & 0x3F); |
| 116 | buffer[2] = 0; |
| 117 | } else { |
| 118 | BUG_ON(buffer_length < 4); |
| 119 | |
| 120 | buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F); |
| 121 | buffer[1] = 0x80 | ((wchar >> 6) & 0x3F); |
| 122 | buffer[2] = 0x80 | (wchar & 0x3F); |
| 123 | buffer[3] = 0; |
| 124 | } |
| 125 | } |
| 126 | |
| 127 | static int hex2decimal(char ch) |
| 128 | { |
| 129 | if (ch >= '0' && ch <= '9') { |
| 130 | return (ch - '0'); |
| 131 | } else if (ch >= 'a' && ch <= 'f') { |
| 132 | return 10 + (ch - 'a'); |
| 133 | } else if (ch >= 'A' && ch <= 'F') { |
| 134 | return 10 + (ch - 'A'); |
| 135 | } |
| 136 | |
| 137 | return -1; |
| 138 | } |
| 139 | |
| 140 | /** |
| 141 | * parse_string(): Parse a json string and return a QObject |
| 142 | * |
| 143 | * string |
| 144 | * "" |
| 145 | * " chars " |
| 146 | * chars |
| 147 | * char |
| 148 | * char chars |
| 149 | * char |
| 150 | * any-Unicode-character- |
| 151 | * except-"-or-\-or- |
| 152 | * control-character |
| 153 | * \" |
| 154 | * \\ |
| 155 | * \/ |
| 156 | * \b |
| 157 | * \f |
| 158 | * \n |
| 159 | * \r |
| 160 | * \t |
| 161 | * \u four-hex-digits |
| 162 | */ |
| 163 | static QString *qstring_from_escaped_str(JSONParserContext *ctxt, QObject *token) |
| 164 | { |
| 165 | const char *ptr = token_get_value(token); |
| 166 | QString *str; |
| 167 | int double_quote = 1; |
| 168 | |
| 169 | if (*ptr == '"') { |
| 170 | double_quote = 1; |
| 171 | } else { |
| 172 | double_quote = 0; |
| 173 | } |
| 174 | ptr++; |
| 175 | |
| 176 | str = qstring_new(); |
| 177 | while (*ptr && |
| 178 | ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) { |
| 179 | if (*ptr == '\\') { |
| 180 | ptr++; |
| 181 | |
| 182 | switch (*ptr) { |
| 183 | case '"': |
| 184 | qstring_append(str, "\""); |
| 185 | ptr++; |
| 186 | break; |
| 187 | case '\'': |
| 188 | qstring_append(str, "'"); |
| 189 | ptr++; |
| 190 | break; |
| 191 | case '\\': |
| 192 | qstring_append(str, "\\"); |
| 193 | ptr++; |
| 194 | break; |
| 195 | case '/': |
| 196 | qstring_append(str, "/"); |
| 197 | ptr++; |
| 198 | break; |
| 199 | case 'b': |
| 200 | qstring_append(str, "\b"); |
| 201 | ptr++; |
| 202 | break; |
| 203 | case 'n': |
| 204 | qstring_append(str, "\n"); |
| 205 | ptr++; |
| 206 | break; |
| 207 | case 'r': |
| 208 | qstring_append(str, "\r"); |
| 209 | ptr++; |
| 210 | break; |
| 211 | case 't': |
| 212 | qstring_append(str, "\t"); |
| 213 | ptr++; |
| 214 | break; |
| 215 | case 'u': { |
| 216 | uint16_t unicode_char = 0; |
| 217 | char utf8_char[4]; |
| 218 | int i = 0; |
| 219 | |
| 220 | ptr++; |
| 221 | |
| 222 | for (i = 0; i < 4; i++) { |
| 223 | if (qemu_isxdigit(*ptr)) { |
| 224 | unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4); |
| 225 | } else { |
| 226 | parse_error(ctxt, token, |
| 227 | "invalid hex escape sequence in string"); |
| 228 | goto out; |
| 229 | } |
| 230 | ptr++; |
| 231 | } |
| 232 | |
| 233 | wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char)); |
| 234 | qstring_append(str, utf8_char); |
| 235 | } break; |
| 236 | default: |
| 237 | parse_error(ctxt, token, "invalid escape sequence in string"); |
| 238 | goto out; |
| 239 | } |
| 240 | } else { |
| 241 | char dummy[2]; |
| 242 | |
| 243 | dummy[0] = *ptr++; |
| 244 | dummy[1] = 0; |
| 245 | |
| 246 | qstring_append(str, dummy); |
| 247 | } |
| 248 | } |
| 249 | |
| 250 | ptr++; |
| 251 | |
| 252 | return str; |
| 253 | |
| 254 | out: |
| 255 | QDECREF(str); |
| 256 | return NULL; |
| 257 | } |
| 258 | |
| 259 | /** |
| 260 | * Parsing rules |
| 261 | */ |
| 262 | static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap) |
| 263 | { |
| 264 | QObject *key, *token = NULL, *value, *peek; |
| 265 | QList *working = qlist_copy(*tokens); |
| 266 | |
| 267 | peek = qlist_peek(working); |
| 268 | key = parse_value(ctxt, &working, ap); |
| 269 | if (qobject_type(key) != QTYPE_QSTRING) { |
| 270 | parse_error(ctxt, peek, "key is not a string in object"); |
| 271 | goto out; |
| 272 | } |
| 273 | |
| 274 | token = qlist_pop(working); |
| 275 | if (!token_is_operator(token, ':')) { |
| 276 | parse_error(ctxt, token, "missing : in object pair"); |
| 277 | goto out; |
| 278 | } |
| 279 | |
| 280 | value = parse_value(ctxt, &working, ap); |
| 281 | if (value == NULL) { |
| 282 | parse_error(ctxt, token, "Missing value in dict"); |
| 283 | goto out; |
| 284 | } |
| 285 | |
| 286 | qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value); |
| 287 | |
| 288 | qobject_decref(token); |
| 289 | qobject_decref(key); |
| 290 | QDECREF(*tokens); |
| 291 | *tokens = working; |
| 292 | |
| 293 | return 0; |
| 294 | |
| 295 | out: |
| 296 | qobject_decref(token); |
| 297 | qobject_decref(key); |
| 298 | QDECREF(working); |
| 299 | |
| 300 | return -1; |
| 301 | } |
| 302 | |
| 303 | static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap) |
| 304 | { |
| 305 | QDict *dict = NULL; |
| 306 | QObject *token, *peek; |
| 307 | QList *working = qlist_copy(*tokens); |
| 308 | |
| 309 | token = qlist_pop(working); |
| 310 | if (!token_is_operator(token, '{')) { |
| 311 | goto out; |
| 312 | } |
| 313 | qobject_decref(token); |
| 314 | token = NULL; |
| 315 | |
| 316 | dict = qdict_new(); |
| 317 | |
| 318 | peek = qlist_peek(working); |
| 319 | if (!token_is_operator(peek, '}')) { |
| 320 | if (parse_pair(ctxt, dict, &working, ap) == -1) { |
| 321 | goto out; |
| 322 | } |
| 323 | |
| 324 | token = qlist_pop(working); |
| 325 | while (!token_is_operator(token, '}')) { |
| 326 | if (!token_is_operator(token, ',')) { |
| 327 | parse_error(ctxt, token, "expected separator in dict"); |
| 328 | goto out; |
| 329 | } |
| 330 | qobject_decref(token); |
| 331 | token = NULL; |
| 332 | |
| 333 | if (parse_pair(ctxt, dict, &working, ap) == -1) { |
| 334 | goto out; |
| 335 | } |
| 336 | |
| 337 | token = qlist_pop(working); |
| 338 | } |
| 339 | qobject_decref(token); |
| 340 | token = NULL; |
| 341 | } else { |
| 342 | token = qlist_pop(working); |
| 343 | qobject_decref(token); |
| 344 | token = NULL; |
| 345 | } |
| 346 | |
| 347 | QDECREF(*tokens); |
| 348 | *tokens = working; |
| 349 | |
| 350 | return QOBJECT(dict); |
| 351 | |
| 352 | out: |
| 353 | qobject_decref(token); |
| 354 | QDECREF(working); |
| 355 | QDECREF(dict); |
| 356 | return NULL; |
| 357 | } |
| 358 | |
| 359 | static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap) |
| 360 | { |
| 361 | QList *list = NULL; |
| 362 | QObject *token, *peek; |
| 363 | QList *working = qlist_copy(*tokens); |
| 364 | |
| 365 | token = qlist_pop(working); |
| 366 | if (!token_is_operator(token, '[')) { |
| 367 | goto out; |
| 368 | } |
| 369 | qobject_decref(token); |
| 370 | token = NULL; |
| 371 | |
| 372 | list = qlist_new(); |
| 373 | |
| 374 | peek = qlist_peek(working); |
| 375 | if (!token_is_operator(peek, ']')) { |
| 376 | QObject *obj; |
| 377 | |
| 378 | obj = parse_value(ctxt, &working, ap); |
| 379 | if (obj == NULL) { |
| 380 | parse_error(ctxt, token, "expecting value"); |
| 381 | goto out; |
| 382 | } |
| 383 | |
| 384 | qlist_append_obj(list, obj); |
| 385 | |
| 386 | token = qlist_pop(working); |
| 387 | while (!token_is_operator(token, ']')) { |
| 388 | if (!token_is_operator(token, ',')) { |
| 389 | parse_error(ctxt, token, "expected separator in list"); |
| 390 | goto out; |
| 391 | } |
| 392 | |
| 393 | qobject_decref(token); |
| 394 | token = NULL; |
| 395 | |
| 396 | obj = parse_value(ctxt, &working, ap); |
| 397 | if (obj == NULL) { |
| 398 | parse_error(ctxt, token, "expecting value"); |
| 399 | goto out; |
| 400 | } |
| 401 | |
| 402 | qlist_append_obj(list, obj); |
| 403 | |
| 404 | token = qlist_pop(working); |
| 405 | } |
| 406 | |
| 407 | qobject_decref(token); |
| 408 | token = NULL; |
| 409 | } else { |
| 410 | token = qlist_pop(working); |
| 411 | qobject_decref(token); |
| 412 | token = NULL; |
| 413 | } |
| 414 | |
| 415 | QDECREF(*tokens); |
| 416 | *tokens = working; |
| 417 | |
| 418 | return QOBJECT(list); |
| 419 | |
| 420 | out: |
| 421 | qobject_decref(token); |
| 422 | QDECREF(working); |
| 423 | QDECREF(list); |
| 424 | return NULL; |
| 425 | } |
| 426 | |
| 427 | static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens) |
| 428 | { |
| 429 | QObject *token, *ret; |
| 430 | QList *working = qlist_copy(*tokens); |
| 431 | |
| 432 | token = qlist_pop(working); |
| 433 | |
| 434 | if (token_get_type(token) != JSON_KEYWORD) { |
| 435 | goto out; |
| 436 | } |
| 437 | |
| 438 | if (token_is_keyword(token, "true")) { |
| 439 | ret = QOBJECT(qbool_from_int(true)); |
| 440 | } else if (token_is_keyword(token, "false")) { |
| 441 | ret = QOBJECT(qbool_from_int(false)); |
| 442 | } else { |
| 443 | parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token)); |
| 444 | goto out; |
| 445 | } |
| 446 | |
| 447 | qobject_decref(token); |
| 448 | QDECREF(*tokens); |
| 449 | *tokens = working; |
| 450 | |
| 451 | return ret; |
| 452 | |
| 453 | out: |
| 454 | qobject_decref(token); |
| 455 | QDECREF(working); |
| 456 | |
| 457 | return NULL; |
| 458 | } |
| 459 | |
| 460 | static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap) |
| 461 | { |
| 462 | QObject *token = NULL, *obj; |
| 463 | QList *working = qlist_copy(*tokens); |
| 464 | |
| 465 | if (ap == NULL) { |
| 466 | goto out; |
| 467 | } |
| 468 | |
| 469 | token = qlist_pop(working); |
| 470 | |
| 471 | if (token_is_escape(token, "%p")) { |
| 472 | obj = va_arg(*ap, QObject *); |
| 473 | } else if (token_is_escape(token, "%i")) { |
| 474 | obj = QOBJECT(qbool_from_int(va_arg(*ap, int))); |
| 475 | } else if (token_is_escape(token, "%d")) { |
| 476 | obj = QOBJECT(qint_from_int(va_arg(*ap, int))); |
| 477 | } else if (token_is_escape(token, "%ld")) { |
| 478 | obj = QOBJECT(qint_from_int(va_arg(*ap, long))); |
| 479 | } else if (token_is_escape(token, "%lld")) { |
| 480 | obj = QOBJECT(qint_from_int(va_arg(*ap, long long))); |
| 481 | } else if (token_is_escape(token, "%s")) { |
| 482 | obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *))); |
| 483 | } else if (token_is_escape(token, "%f")) { |
| 484 | obj = QOBJECT(qfloat_from_double(va_arg(*ap, double))); |
| 485 | } else { |
| 486 | goto out; |
| 487 | } |
| 488 | |
| 489 | qobject_decref(token); |
| 490 | QDECREF(*tokens); |
| 491 | *tokens = working; |
| 492 | |
| 493 | return obj; |
| 494 | |
| 495 | out: |
| 496 | qobject_decref(token); |
| 497 | QDECREF(working); |
| 498 | |
| 499 | return NULL; |
| 500 | } |
| 501 | |
| 502 | static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens) |
| 503 | { |
| 504 | QObject *token, *obj; |
| 505 | QList *working = qlist_copy(*tokens); |
| 506 | |
| 507 | token = qlist_pop(working); |
| 508 | switch (token_get_type(token)) { |
| 509 | case JSON_STRING: |
| 510 | obj = QOBJECT(qstring_from_escaped_str(ctxt, token)); |
| 511 | break; |
| 512 | case JSON_INTEGER: |
| 513 | obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10))); |
| 514 | break; |
| 515 | case JSON_FLOAT: |
| 516 | /* FIXME dependent on locale */ |
| 517 | obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL))); |
| 518 | break; |
| 519 | default: |
| 520 | goto out; |
| 521 | } |
| 522 | |
| 523 | qobject_decref(token); |
| 524 | QDECREF(*tokens); |
| 525 | *tokens = working; |
| 526 | |
| 527 | return obj; |
| 528 | |
| 529 | out: |
| 530 | qobject_decref(token); |
| 531 | QDECREF(working); |
| 532 | |
| 533 | return NULL; |
| 534 | } |
| 535 | |
| 536 | static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap) |
| 537 | { |
| 538 | QObject *obj; |
| 539 | |
| 540 | obj = parse_object(ctxt, tokens, ap); |
| 541 | if (obj == NULL) { |
| 542 | obj = parse_array(ctxt, tokens, ap); |
| 543 | } |
| 544 | if (obj == NULL) { |
| 545 | obj = parse_escape(ctxt, tokens, ap); |
| 546 | } |
| 547 | if (obj == NULL) { |
| 548 | obj = parse_keyword(ctxt, tokens); |
| 549 | } |
| 550 | if (obj == NULL) { |
| 551 | obj = parse_literal(ctxt, tokens); |
| 552 | } |
| 553 | |
| 554 | return obj; |
| 555 | } |
| 556 | |
| 557 | QObject *json_parser_parse(QList *tokens, va_list *ap) |
| 558 | { |
| 559 | JSONParserContext ctxt = {}; |
| 560 | QList *working = qlist_copy(tokens); |
| 561 | QObject *result; |
| 562 | |
| 563 | result = parse_value(&ctxt, &working, ap); |
| 564 | |
| 565 | QDECREF(working); |
| 566 | |
| 567 | return result; |
| 568 | } |