qobject/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 "qemu/osdep.h"
  15 #include "qemu/cutils.h"
  16 #include "qemu/unicode.h"
  17 #include "qapi/error.h"
  18 #include "qemu-common.h"
  19 #include "qapi/qmp/qbool.h"
  20 #include "qapi/qmp/qdict.h"
  21 #include "qapi/qmp/qlist.h"
  22 #include "qapi/qmp/qnull.h"
  23 #include "qapi/qmp/qnum.h"
  24 #include "qapi/qmp/qstring.h"
  25 #include "qapi/qmp/json-parser.h"
  26 #include "qapi/qmp/json-lexer.h"
  27 #include "qapi/qmp/json-streamer.h"
  28
  29 typedef struct JSONParserContext
  30 {
  31     Error *err;
  32     JSONToken *current;
  33     GQueue *buf;
  34 } JSONParserContext;
  35
  36 #define BUG_ON(cond) assert(!(cond))
  37
  38 /**
  39  * TODO
  40  *
  41  * 0) make errors meaningful again
  42  * 1) add geometry information to tokens
  43  * 3) should we return a parsed size?
  44  * 4) deal with premature EOI
  45  */
  46
  47 static QObject *parse_value(JSONParserContext *ctxt, va_list *ap);
  48
  49 /**
  50  * Error handler
  51  */
  52 static void GCC_FMT_ATTR(3, 4) parse_error(JSONParserContext *ctxt,
  53                                            JSONToken *token, const char *msg, ...)
  54 {
  55     va_list ap;
  56     char message[1024];
  57
  58     if (ctxt->err) {
  59         return;
  60     }
  61     va_start(ap, msg);
  62     vsnprintf(message, sizeof(message), msg, ap);
  63     va_end(ap);
  64     error_setg(&ctxt->err, "JSON parse error, %s", message);
  65 }
  66
  67 static int cvt4hex(const char *s)
  68 {
  69     int cp, i;
  70
  71     cp = 0;
  72     for (i = 0; i < 4; i++) {
  73         if (!qemu_isxdigit(s[i])) {
  74             return -1;
  75         }
  76         cp <<= 4;
  77         if (s[i] >= '0' && s[i] <= '9') {
  78             cp |= s[i] - '0';
  79         } else if (s[i] >= 'a' && s[i] <= 'f') {
  80             cp |= 10 + s[i] - 'a';
  81         } else if (s[i] >= 'A' && s[i] <= 'F') {
  82             cp |= 10 + s[i] - 'A';
  83         } else {
  84             return -1;
  85         }
  86     }
  87     return cp;
  88 }
  89
  90 /**
  91  * parse_string(): Parse a JSON string
  92  *
  93  * From RFC 8259 "The JavaScript Object Notation (JSON) Data
  94  * Interchange Format":
  95  *
  96  *    char = unescaped /
  97  *        escape (
  98  *            %x22 /          ; "    quotation mark  U+0022
  99  *            %x5C /          ; \    reverse solidus U+005C
 100  *            %x2F /          ; /    solidus         U+002F
 101  *            %x62 /          ; b    backspace       U+0008
 102  *            %x66 /          ; f    form feed       U+000C
 103  *            %x6E /          ; n    line feed       U+000A
 104  *            %x72 /          ; r    carriage return U+000D
 105  *            %x74 /          ; t    tab             U+0009
 106  *            %x75 4HEXDIG )  ; uXXXX                U+XXXX
 107  *    escape = %x5C              ; \
 108  *    quotation-mark = %x22      ; "
 109  *    unescaped = %x20-21 / %x23-5B / %x5D-10FFFF
 110  *
 111  * Extensions over RFC 8259:
 112  * - Extra escape sequence in strings:
 113  *   0x27 (apostrophe) is recognized after escape, too
 114  * - Single-quoted strings:
 115  *   Like double-quoted strings, except they're delimited by %x27
 116  *   (apostrophe) instead of %x22 (quotation mark), and can't contain
 117  *   unescaped apostrophe, but can contain unescaped quotation mark.
 118  *
 119  * Note:
 120  * - Encoding is modified UTF-8.
 121  * - Invalid Unicode characters are rejected.
 122  * - Control characters \x00..\x1F are rejected by the lexer.
 123  */
 124 static QString *parse_string(JSONParserContext *ctxt, JSONToken *token)
 125 {
 126     const char *ptr = token->str;
 127     QString *str;
 128     char quote;
 129     const char *beg;
 130     int cp, trailing;
 131     char *end;
 132     ssize_t len;
 133     char utf8_buf[5];
 134
 135     assert(*ptr == '"' || *ptr == '\'');
 136     quote = *ptr++;
 137     str = qstring_new();
 138
 139     while (*ptr != quote) {
 140         assert(*ptr);
 141         if (*ptr == '\\') {
 142             beg = ptr++;
 143             switch (*ptr++) {
 144             case '"':
 145                 qstring_append_chr(str, '"');
 146                 break;
 147             case '\'':
 148                 qstring_append_chr(str, '\'');
 149                 break;
 150             case '\\':
 151                 qstring_append_chr(str, '\\');
 152                 break;
 153             case '/':
 154                 qstring_append_chr(str, '/');
 155                 break;
 156             case 'b':
 157                 qstring_append_chr(str, '\b');
 158                 break;
 159             case 'f':
 160                 qstring_append_chr(str, '\f');
 161                 break;
 162             case 'n':
 163                 qstring_append_chr(str, '\n');
 164                 break;
 165             case 'r':
 166                 qstring_append_chr(str, '\r');
 167                 break;
 168             case 't':
 169                 qstring_append_chr(str, '\t');
 170                 break;
 171             case 'u':
 172                 cp = cvt4hex(ptr);
 173                 ptr += 4;
 174
 175                 /* handle surrogate pairs */
 176                 if (cp >= 0xD800 && cp <= 0xDBFF
 177                     && ptr[0] == '\\' && ptr[1] == 'u') {
 178                     /* leading surrogate followed by \u */
 179                     cp = 0x10000 + ((cp & 0x3FF) << 10);
 180                     trailing = cvt4hex(ptr + 2);
 181                     if (trailing >= 0xDC00 && trailing <= 0xDFFF) {
 182                         /* followed by trailing surrogate */
 183                         cp |= trailing & 0x3FF;
 184                         ptr += 6;
 185                     } else {
 186                         cp = -1; /* invalid */
 187                     }
 188                 }
 189
 190                 if (mod_utf8_encode(utf8_buf, sizeof(utf8_buf), cp) < 0) {
 191                     parse_error(ctxt, token,
 192                                 "%.*s is not a valid Unicode character",
 193                                 (int)(ptr - beg), beg);
 194                     goto out;
 195                 }
 196                 qstring_append(str, utf8_buf);
 197                 break;
 198             default:
 199                 parse_error(ctxt, token, "invalid escape sequence in string");
 200                 goto out;
 201             }
 202         } else {
 203             cp = mod_utf8_codepoint(ptr, 6, &end);
 204             if (cp < 0) {
 205                 parse_error(ctxt, token, "invalid UTF-8 sequence in string");
 206                 goto out;
 207             }
 208             ptr = end;
 209             len = mod_utf8_encode(utf8_buf, sizeof(utf8_buf), cp);
 210             assert(len >= 0);
 211             qstring_append(str, utf8_buf);
 212         }
 213     }
 214
 215     return str;
 216
 217 out:
 218     qobject_unref(str);
 219     return NULL;
 220 }
 221
 222 /* Note: the token object returned by parser_context_peek_token or
 223  * parser_context_pop_token is deleted as soon as parser_context_pop_token
 224  * is called again.
 225  */
 226 static JSONToken *parser_context_pop_token(JSONParserContext *ctxt)
 227 {
 228     g_free(ctxt->current);
 229     ctxt->current = g_queue_pop_head(ctxt->buf);
 230     return ctxt->current;
 231 }
 232
 233 static JSONToken *parser_context_peek_token(JSONParserContext *ctxt)
 234 {
 235     return g_queue_peek_head(ctxt->buf);
 236 }
 237
 238 /**
 239  * Parsing rules
 240  */
 241 static int parse_pair(JSONParserContext *ctxt, QDict *dict, va_list *ap)
 242 {
 243     QObject *value;
 244     QString *key = NULL;
 245     JSONToken *peek, *token;
 246
 247     peek = parser_context_peek_token(ctxt);
 248     if (peek == NULL) {
 249         parse_error(ctxt, NULL, "premature EOI");
 250         goto out;
 251     }
 252
 253     key = qobject_to(QString, parse_value(ctxt, ap));
 254     if (!key) {
 255         parse_error(ctxt, peek, "key is not a string in object");
 256         goto out;
 257     }
 258
 259     token = parser_context_pop_token(ctxt);
 260     if (token == NULL) {
 261         parse_error(ctxt, NULL, "premature EOI");
 262         goto out;
 263     }
 264
 265     if (token->type != JSON_COLON) {
 266         parse_error(ctxt, token, "missing : in object pair");
 267         goto out;
 268     }
 269
 270     value = parse_value(ctxt, ap);
 271     if (value == NULL) {
 272         parse_error(ctxt, token, "Missing value in dict");
 273         goto out;
 274     }
 275
 276     qdict_put_obj(dict, qstring_get_str(key), value);
 277
 278     qobject_unref(key);
 279
 280     return 0;
 281
 282 out:
 283     qobject_unref(key);
 284
 285     return -1;
 286 }
 287
 288 static QObject *parse_object(JSONParserContext *ctxt, va_list *ap)
 289 {
 290     QDict *dict = NULL;
 291     JSONToken *token, *peek;
 292
 293     token = parser_context_pop_token(ctxt);
 294     assert(token && token->type == JSON_LCURLY);
 295
 296     dict = qdict_new();
 297
 298     peek = parser_context_peek_token(ctxt);
 299     if (peek == NULL) {
 300         parse_error(ctxt, NULL, "premature EOI");
 301         goto out;
 302     }
 303
 304     if (peek->type != JSON_RCURLY) {
 305         if (parse_pair(ctxt, dict, ap) == -1) {
 306             goto out;
 307         }
 308
 309         token = parser_context_pop_token(ctxt);
 310         if (token == NULL) {
 311             parse_error(ctxt, NULL, "premature EOI");
 312             goto out;
 313         }
 314
 315         while (token->type != JSON_RCURLY) {
 316             if (token->type != JSON_COMMA) {
 317                 parse_error(ctxt, token, "expected separator in dict");
 318                 goto out;
 319             }
 320
 321             if (parse_pair(ctxt, dict, ap) == -1) {
 322                 goto out;
 323             }
 324
 325             token = parser_context_pop_token(ctxt);
 326             if (token == NULL) {
 327                 parse_error(ctxt, NULL, "premature EOI");
 328                 goto out;
 329             }
 330         }
 331     } else {
 332         (void)parser_context_pop_token(ctxt);
 333     }
 334
 335     return QOBJECT(dict);
 336
 337 out:
 338     qobject_unref(dict);
 339     return NULL;
 340 }
 341
 342 static QObject *parse_array(JSONParserContext *ctxt, va_list *ap)
 343 {
 344     QList *list = NULL;
 345     JSONToken *token, *peek;
 346
 347     token = parser_context_pop_token(ctxt);
 348     assert(token && token->type == JSON_LSQUARE);
 349
 350     list = qlist_new();
 351
 352     peek = parser_context_peek_token(ctxt);
 353     if (peek == NULL) {
 354         parse_error(ctxt, NULL, "premature EOI");
 355         goto out;
 356     }
 357
 358     if (peek->type != JSON_RSQUARE) {
 359         QObject *obj;
 360
 361         obj = parse_value(ctxt, ap);
 362         if (obj == NULL) {
 363             parse_error(ctxt, token, "expecting value");
 364             goto out;
 365         }
 366
 367         qlist_append_obj(list, obj);
 368
 369         token = parser_context_pop_token(ctxt);
 370         if (token == NULL) {
 371             parse_error(ctxt, NULL, "premature EOI");
 372             goto out;
 373         }
 374
 375         while (token->type != JSON_RSQUARE) {
 376             if (token->type != JSON_COMMA) {
 377                 parse_error(ctxt, token, "expected separator in list");
 378                 goto out;
 379             }
 380
 381             obj = parse_value(ctxt, ap);
 382             if (obj == NULL) {
 383                 parse_error(ctxt, token, "expecting value");
 384                 goto out;
 385             }
 386
 387             qlist_append_obj(list, obj);
 388
 389             token = parser_context_pop_token(ctxt);
 390             if (token == NULL) {
 391                 parse_error(ctxt, NULL, "premature EOI");
 392                 goto out;
 393             }
 394         }
 395     } else {
 396         (void)parser_context_pop_token(ctxt);
 397     }
 398
 399     return QOBJECT(list);
 400
 401 out:
 402     qobject_unref(list);
 403     return NULL;
 404 }
 405
 406 static QObject *parse_keyword(JSONParserContext *ctxt)
 407 {
 408     JSONToken *token;
 409
 410     token = parser_context_pop_token(ctxt);
 411     assert(token && token->type == JSON_KEYWORD);
 412
 413     if (!strcmp(token->str, "true")) {
 414         return QOBJECT(qbool_from_bool(true));
 415     } else if (!strcmp(token->str, "false")) {
 416         return QOBJECT(qbool_from_bool(false));
 417     } else if (!strcmp(token->str, "null")) {
 418         return QOBJECT(qnull());
 419     }
 420     parse_error(ctxt, token, "invalid keyword '%s'", token->str);
 421     return NULL;
 422 }
 423
 424 static QObject *parse_interpolation(JSONParserContext *ctxt, va_list *ap)
 425 {
 426     JSONToken *token;
 427
 428     token = parser_context_pop_token(ctxt);
 429     assert(token && token->type == JSON_INTERP);
 430
 431     if (!strcmp(token->str, "%p")) {
 432         return va_arg(*ap, QObject *);
 433     } else if (!strcmp(token->str, "%i")) {
 434         return QOBJECT(qbool_from_bool(va_arg(*ap, int)));
 435     } else if (!strcmp(token->str, "%d")) {
 436         return QOBJECT(qnum_from_int(va_arg(*ap, int)));
 437     } else if (!strcmp(token->str, "%ld")) {
 438         return QOBJECT(qnum_from_int(va_arg(*ap, long)));
 439     } else if (!strcmp(token->str, "%lld")) {
 440         return QOBJECT(qnum_from_int(va_arg(*ap, long long)));
 441     } else if (!strcmp(token->str, "%" PRId64)) {
 442         return QOBJECT(qnum_from_int(va_arg(*ap, int64_t)));
 443     } else if (!strcmp(token->str, "%u")) {
 444         return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned int)));
 445     } else if (!strcmp(token->str, "%lu")) {
 446         return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long)));
 447     } else if (!strcmp(token->str, "%llu")) {
 448         return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long long)));
 449     } else if (!strcmp(token->str, "%" PRIu64)) {
 450         return QOBJECT(qnum_from_uint(va_arg(*ap, uint64_t)));
 451     } else if (!strcmp(token->str, "%s")) {
 452         return QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
 453     } else if (!strcmp(token->str, "%f")) {
 454         return QOBJECT(qnum_from_double(va_arg(*ap, double)));
 455     }
 456     parse_error(ctxt, token, "invalid interpolation '%s'", token->str);
 457     return NULL;
 458 }
 459
 460 static QObject *parse_literal(JSONParserContext *ctxt)
 461 {
 462     JSONToken *token;
 463
 464     token = parser_context_pop_token(ctxt);
 465     assert(token);
 466
 467     switch (token->type) {
 468     case JSON_STRING:
 469         return QOBJECT(parse_string(ctxt, token));
 470     case JSON_INTEGER: {
 471         /*
 472          * Represent JSON_INTEGER as QNUM_I64 if possible, else as
 473          * QNUM_U64, else as QNUM_DOUBLE.  Note that qemu_strtoi64()
 474          * and qemu_strtou64() fail with ERANGE when it's not
 475          * possible.
 476          *
 477          * qnum_get_int() will then work for any signed 64-bit
 478          * JSON_INTEGER, qnum_get_uint() for any unsigned 64-bit
 479          * integer, and qnum_get_double() both for any JSON_INTEGER
 480          * and any JSON_FLOAT (with precision loss for integers beyond
 481          * 53 bits)
 482          */
 483         int ret;
 484         int64_t value;
 485         uint64_t uvalue;
 486
 487         ret = qemu_strtoi64(token->str, NULL, 10, &value);
 488         if (!ret) {
 489             return QOBJECT(qnum_from_int(value));
 490         }
 491         assert(ret == -ERANGE);
 492
 493         if (token->str[0] != '-') {
 494             ret = qemu_strtou64(token->str, NULL, 10, &uvalue);
 495             if (!ret) {
 496                 return QOBJECT(qnum_from_uint(uvalue));
 497             }
 498             assert(ret == -ERANGE);
 499         }
 500         /* fall through to JSON_FLOAT */
 501     }
 502     case JSON_FLOAT:
 503         /* FIXME dependent on locale; a pervasive issue in QEMU */
 504         /* FIXME our lexer matches RFC 7159 in forbidding Inf or NaN,
 505          * but those might be useful extensions beyond JSON */
 506         return QOBJECT(qnum_from_double(strtod(token->str, NULL)));
 507     default:
 508         abort();
 509     }
 510 }
 511
 512 static QObject *parse_value(JSONParserContext *ctxt, va_list *ap)
 513 {
 514     JSONToken *token;
 515
 516     token = parser_context_peek_token(ctxt);
 517     if (token == NULL) {
 518         parse_error(ctxt, NULL, "premature EOI");
 519         return NULL;
 520     }
 521
 522     switch (token->type) {
 523     case JSON_LCURLY:
 524         return parse_object(ctxt, ap);
 525     case JSON_LSQUARE:
 526         return parse_array(ctxt, ap);
 527     case JSON_INTERP:
 528         return parse_interpolation(ctxt, ap);
 529     case JSON_INTEGER:
 530     case JSON_FLOAT:
 531     case JSON_STRING:
 532         return parse_literal(ctxt);
 533     case JSON_KEYWORD:
 534         return parse_keyword(ctxt);
 535     default:
 536         parse_error(ctxt, token, "expecting value");
 537         return NULL;
 538     }
 539 }
 540
 541 QObject *json_parser_parse(GQueue *tokens, va_list *ap, Error **errp)
 542 {
 543     JSONParserContext ctxt = { .buf = tokens };
 544     QObject *result;
 545
 546     result = parse_value(&ctxt, ap);
 547     assert(ctxt.err || g_queue_is_empty(ctxt.buf));
 548
 549     error_propagate(errp, ctxt.err);
 550
 551     while (!g_queue_is_empty(ctxt.buf)) {
 552         parser_context_pop_token(&ctxt);
 553     }
 554     g_free(ctxt.current);
 555     g_queue_free(ctxt.buf);
 556
 557     return result;
 558 }