json-parser.c

   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     return str;
 251
 252 out:
 253     QDECREF(str);
 254     return NULL;
 255 }
 256
 257 /**
 258  * Parsing rules
 259  */
 260 static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap)
 261 {
 262     QObject *key, *token = NULL, *value, *peek;
 263     QList *working = qlist_copy(*tokens);
 264
 265     peek = qlist_peek(working);
 266     key = parse_value(ctxt, &working, ap);
 267     if (qobject_type(key) != QTYPE_QSTRING) {
 268         parse_error(ctxt, peek, "key is not a string in object");
 269         goto out;
 270     }
 271
 272     token = qlist_pop(working);
 273     if (!token_is_operator(token, ':')) {
 274         parse_error(ctxt, token, "missing : in object pair");
 275         goto out;
 276     }
 277
 278     value = parse_value(ctxt, &working, ap);
 279     if (value == NULL) {
 280         parse_error(ctxt, token, "Missing value in dict");
 281         goto out;
 282     }
 283
 284     qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value);
 285
 286     qobject_decref(token);
 287     qobject_decref(key);
 288     QDECREF(*tokens);
 289     *tokens = working;
 290
 291     return 0;
 292
 293 out:
 294     qobject_decref(token);
 295     qobject_decref(key);
 296     QDECREF(working);
 297
 298     return -1;
 299 }
 300
 301 static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 302 {
 303     QDict *dict = NULL;
 304     QObject *token, *peek;
 305     QList *working = qlist_copy(*tokens);
 306
 307     token = qlist_pop(working);
 308     if (!token_is_operator(token, '{')) {
 309         goto out;
 310     }
 311     qobject_decref(token);
 312     token = NULL;
 313
 314     dict = qdict_new();
 315
 316     peek = qlist_peek(working);
 317     if (!token_is_operator(peek, '}')) {
 318         if (parse_pair(ctxt, dict, &working, ap) == -1) {
 319             goto out;
 320         }
 321
 322         token = qlist_pop(working);
 323         while (!token_is_operator(token, '}')) {
 324             if (!token_is_operator(token, ',')) {
 325                 parse_error(ctxt, token, "expected separator in dict");
 326                 goto out;
 327             }
 328             qobject_decref(token);
 329             token = NULL;
 330
 331             if (parse_pair(ctxt, dict, &working, ap) == -1) {
 332                 goto out;
 333             }
 334
 335             token = qlist_pop(working);
 336         }
 337         qobject_decref(token);
 338         token = NULL;
 339     } else {
 340         token = qlist_pop(working);
 341         qobject_decref(token);
 342         token = NULL;
 343     }
 344
 345     QDECREF(*tokens);
 346     *tokens = working;
 347
 348     return QOBJECT(dict);
 349
 350 out:
 351     qobject_decref(token);
 352     QDECREF(working);
 353     QDECREF(dict);
 354     return NULL;
 355 }
 356
 357 static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 358 {
 359     QList *list = NULL;
 360     QObject *token, *peek;
 361     QList *working = qlist_copy(*tokens);
 362
 363     token = qlist_pop(working);
 364     if (!token_is_operator(token, '[')) {
 365         goto out;
 366     }
 367     qobject_decref(token);
 368     token = NULL;
 369
 370     list = qlist_new();
 371
 372     peek = qlist_peek(working);
 373     if (!token_is_operator(peek, ']')) {
 374         QObject *obj;
 375
 376         obj = parse_value(ctxt, &working, ap);
 377         if (obj == NULL) {
 378             parse_error(ctxt, token, "expecting value");
 379             goto out;
 380         }
 381
 382         qlist_append_obj(list, obj);
 383
 384         token = qlist_pop(working);
 385         while (!token_is_operator(token, ']')) {
 386             if (!token_is_operator(token, ',')) {
 387                 parse_error(ctxt, token, "expected separator in list");
 388                 goto out;
 389             }
 390
 391             qobject_decref(token);
 392             token = NULL;
 393
 394             obj = parse_value(ctxt, &working, ap);
 395             if (obj == NULL) {
 396                 parse_error(ctxt, token, "expecting value");
 397                 goto out;
 398             }
 399
 400             qlist_append_obj(list, obj);
 401
 402             token = qlist_pop(working);
 403         }
 404
 405         qobject_decref(token);
 406         token = NULL;
 407     } else {
 408         token = qlist_pop(working);
 409         qobject_decref(token);
 410         token = NULL;
 411     }
 412
 413     QDECREF(*tokens);
 414     *tokens = working;
 415
 416     return QOBJECT(list);
 417
 418 out:
 419     qobject_decref(token);
 420     QDECREF(working);
 421     QDECREF(list);
 422     return NULL;
 423 }
 424
 425 static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens)
 426 {
 427     QObject *token, *ret;
 428     QList *working = qlist_copy(*tokens);
 429
 430     token = qlist_pop(working);
 431
 432     if (token_get_type(token) != JSON_KEYWORD) {
 433         goto out;
 434     }
 435
 436     if (token_is_keyword(token, "true")) {
 437         ret = QOBJECT(qbool_from_int(true));
 438     } else if (token_is_keyword(token, "false")) {
 439         ret = QOBJECT(qbool_from_int(false));
 440     } else {
 441         parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token));
 442         goto out;
 443     }
 444
 445     qobject_decref(token);
 446     QDECREF(*tokens);
 447     *tokens = working;
 448
 449     return ret;
 450
 451 out:
 452     qobject_decref(token);
 453     QDECREF(working);
 454
 455     return NULL;
 456 }
 457
 458 static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 459 {
 460     QObject *token = NULL, *obj;
 461     QList *working = qlist_copy(*tokens);
 462
 463     if (ap == NULL) {
 464         goto out;
 465     }
 466
 467     token = qlist_pop(working);
 468
 469     if (token_is_escape(token, "%p")) {
 470         obj = va_arg(*ap, QObject *);
 471     } else if (token_is_escape(token, "%i")) {
 472         obj = QOBJECT(qbool_from_int(va_arg(*ap, int)));
 473     } else if (token_is_escape(token, "%d")) {
 474         obj = QOBJECT(qint_from_int(va_arg(*ap, int)));
 475     } else if (token_is_escape(token, "%ld")) {
 476         obj = QOBJECT(qint_from_int(va_arg(*ap, long)));
 477     } else if (token_is_escape(token, "%lld") ||
 478                token_is_escape(token, "%I64d")) {
 479         obj = QOBJECT(qint_from_int(va_arg(*ap, long long)));
 480     } else if (token_is_escape(token, "%s")) {
 481         obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
 482     } else if (token_is_escape(token, "%f")) {
 483         obj = QOBJECT(qfloat_from_double(va_arg(*ap, double)));
 484     } else {
 485         goto out;
 486     }
 487
 488     qobject_decref(token);
 489     QDECREF(*tokens);
 490     *tokens = working;
 491
 492     return obj;
 493
 494 out:
 495     qobject_decref(token);
 496     QDECREF(working);
 497
 498     return NULL;
 499 }
 500
 501 static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens)
 502 {
 503     QObject *token, *obj;
 504     QList *working = qlist_copy(*tokens);
 505
 506     token = qlist_pop(working);
 507     switch (token_get_type(token)) {
 508     case JSON_STRING:
 509         obj = QOBJECT(qstring_from_escaped_str(ctxt, token));
 510         break;
 511     case JSON_INTEGER:
 512         obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10)));
 513         break;
 514     case JSON_FLOAT:
 515         /* FIXME dependent on locale */
 516         obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL)));
 517         break;
 518     default:
 519         goto out;
 520     }
 521
 522     qobject_decref(token);
 523     QDECREF(*tokens);
 524     *tokens = working;
 525
 526     return obj;
 527
 528 out:
 529     qobject_decref(token);
 530     QDECREF(working);
 531
 532     return NULL;
 533 }
 534
 535 static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 536 {
 537     QObject *obj;
 538
 539     obj = parse_object(ctxt, tokens, ap);
 540     if (obj == NULL) {
 541         obj = parse_array(ctxt, tokens, ap);
 542     }
 543     if (obj == NULL) {
 544         obj = parse_escape(ctxt, tokens, ap);
 545     }
 546     if (obj == NULL) {
 547         obj = parse_keyword(ctxt, tokens);
 548     }
 549     if (obj == NULL) {
 550         obj = parse_literal(ctxt, tokens);
 551     }
 552
 553     return obj;
 554 }
 555
 556 QObject *json_parser_parse(QList *tokens, va_list *ap)
 557 {
 558     JSONParserContext ctxt = {};
 559     QList *working = qlist_copy(tokens);
 560     QObject *result;
 561
 562     result = parse_value(&ctxt, &working, ap);
 563
 564     QDECREF(working);
 565
 566     return result;
 567 }