json-parser.c 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593
  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. #include "qemu/osdep.h"
  14. #include "qemu/cutils.h"
  15. #include "qapi/error.h"
  16. #include "qemu-common.h"
  17. #include "qapi/qmp/types.h"
  18. #include "qapi/qmp/json-parser.h"
  19. #include "qapi/qmp/json-lexer.h"
  20. #include "qapi/qmp/json-streamer.h"
  21. typedef struct JSONParserContext
  22. {
  23. Error *err;
  24. JSONToken *current;
  25. GQueue *buf;
  26. } JSONParserContext;
  27. #define BUG_ON(cond) assert(!(cond))
  28. /**
  29. * TODO
  30. *
  31. * 0) make errors meaningful again
  32. * 1) add geometry information to tokens
  33. * 3) should we return a parsed size?
  34. * 4) deal with premature EOI
  35. */
  36. static QObject *parse_value(JSONParserContext *ctxt, va_list *ap);
  37. /**
  38. * Error handler
  39. */
  40. static void GCC_FMT_ATTR(3, 4) parse_error(JSONParserContext *ctxt,
  41. JSONToken *token, const char *msg, ...)
  42. {
  43. va_list ap;
  44. char message[1024];
  45. va_start(ap, msg);
  46. vsnprintf(message, sizeof(message), msg, ap);
  47. va_end(ap);
  48. if (ctxt->err) {
  49. error_free(ctxt->err);
  50. ctxt->err = NULL;
  51. }
  52. error_setg(&ctxt->err, "JSON parse error, %s", message);
  53. }
  54. /**
  55. * String helpers
  56. *
  57. * These helpers are used to unescape strings.
  58. */
  59. static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length)
  60. {
  61. if (wchar <= 0x007F) {
  62. BUG_ON(buffer_length < 2);
  63. buffer[0] = wchar & 0x7F;
  64. buffer[1] = 0;
  65. } else if (wchar <= 0x07FF) {
  66. BUG_ON(buffer_length < 3);
  67. buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F);
  68. buffer[1] = 0x80 | (wchar & 0x3F);
  69. buffer[2] = 0;
  70. } else {
  71. BUG_ON(buffer_length < 4);
  72. buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F);
  73. buffer[1] = 0x80 | ((wchar >> 6) & 0x3F);
  74. buffer[2] = 0x80 | (wchar & 0x3F);
  75. buffer[3] = 0;
  76. }
  77. }
  78. static int hex2decimal(char ch)
  79. {
  80. if (ch >= '0' && ch <= '9') {
  81. return (ch - '0');
  82. } else if (ch >= 'a' && ch <= 'f') {
  83. return 10 + (ch - 'a');
  84. } else if (ch >= 'A' && ch <= 'F') {
  85. return 10 + (ch - 'A');
  86. }
  87. return -1;
  88. }
  89. /**
  90. * parse_string(): Parse a json string and return a QObject
  91. *
  92. * string
  93. * ""
  94. * " chars "
  95. * chars
  96. * char
  97. * char chars
  98. * char
  99. * any-Unicode-character-
  100. * except-"-or-\-or-
  101. * control-character
  102. * \"
  103. * \\
  104. * \/
  105. * \b
  106. * \f
  107. * \n
  108. * \r
  109. * \t
  110. * \u four-hex-digits
  111. */
  112. static QString *qstring_from_escaped_str(JSONParserContext *ctxt,
  113. JSONToken *token)
  114. {
  115. const char *ptr = token->str;
  116. QString *str;
  117. int double_quote = 1;
  118. if (*ptr == '"') {
  119. double_quote = 1;
  120. } else {
  121. double_quote = 0;
  122. }
  123. ptr++;
  124. str = qstring_new();
  125. while (*ptr &&
  126. ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) {
  127. if (*ptr == '\\') {
  128. ptr++;
  129. switch (*ptr) {
  130. case '"':
  131. qstring_append(str, "\"");
  132. ptr++;
  133. break;
  134. case '\'':
  135. qstring_append(str, "'");
  136. ptr++;
  137. break;
  138. case '\\':
  139. qstring_append(str, "\\");
  140. ptr++;
  141. break;
  142. case '/':
  143. qstring_append(str, "/");
  144. ptr++;
  145. break;
  146. case 'b':
  147. qstring_append(str, "\b");
  148. ptr++;
  149. break;
  150. case 'f':
  151. qstring_append(str, "\f");
  152. ptr++;
  153. break;
  154. case 'n':
  155. qstring_append(str, "\n");
  156. ptr++;
  157. break;
  158. case 'r':
  159. qstring_append(str, "\r");
  160. ptr++;
  161. break;
  162. case 't':
  163. qstring_append(str, "\t");
  164. ptr++;
  165. break;
  166. case 'u': {
  167. uint16_t unicode_char = 0;
  168. char utf8_char[4];
  169. int i = 0;
  170. ptr++;
  171. for (i = 0; i < 4; i++) {
  172. if (qemu_isxdigit(*ptr)) {
  173. unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4);
  174. } else {
  175. parse_error(ctxt, token,
  176. "invalid hex escape sequence in string");
  177. goto out;
  178. }
  179. ptr++;
  180. }
  181. wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char));
  182. qstring_append(str, utf8_char);
  183. } break;
  184. default:
  185. parse_error(ctxt, token, "invalid escape sequence in string");
  186. goto out;
  187. }
  188. } else {
  189. char dummy[2];
  190. dummy[0] = *ptr++;
  191. dummy[1] = 0;
  192. qstring_append(str, dummy);
  193. }
  194. }
  195. return str;
  196. out:
  197. QDECREF(str);
  198. return NULL;
  199. }
  200. /* Note: the token object returned by parser_context_peek_token or
  201. * parser_context_pop_token is deleted as soon as parser_context_pop_token
  202. * is called again.
  203. */
  204. static JSONToken *parser_context_pop_token(JSONParserContext *ctxt)
  205. {
  206. g_free(ctxt->current);
  207. assert(!g_queue_is_empty(ctxt->buf));
  208. ctxt->current = g_queue_pop_head(ctxt->buf);
  209. return ctxt->current;
  210. }
  211. static JSONToken *parser_context_peek_token(JSONParserContext *ctxt)
  212. {
  213. assert(!g_queue_is_empty(ctxt->buf));
  214. return g_queue_peek_head(ctxt->buf);
  215. }
  216. static JSONParserContext *parser_context_new(GQueue *tokens)
  217. {
  218. JSONParserContext *ctxt;
  219. if (!tokens) {
  220. return NULL;
  221. }
  222. ctxt = g_malloc0(sizeof(JSONParserContext));
  223. ctxt->buf = tokens;
  224. return ctxt;
  225. }
  226. /* to support error propagation, ctxt->err must be freed separately */
  227. static void parser_context_free(JSONParserContext *ctxt)
  228. {
  229. if (ctxt) {
  230. while (!g_queue_is_empty(ctxt->buf)) {
  231. parser_context_pop_token(ctxt);
  232. }
  233. g_free(ctxt->current);
  234. g_queue_free(ctxt->buf);
  235. g_free(ctxt);
  236. }
  237. }
  238. /**
  239. * Parsing rules
  240. */
  241. static int parse_pair(JSONParserContext *ctxt, QDict *dict, va_list *ap)
  242. {
  243. QObject *key = NULL, *value;
  244. JSONToken *peek, *token;
  245. peek = parser_context_peek_token(ctxt);
  246. if (peek == NULL) {
  247. parse_error(ctxt, NULL, "premature EOI");
  248. goto out;
  249. }
  250. key = parse_value(ctxt, ap);
  251. if (!key || qobject_type(key) != QTYPE_QSTRING) {
  252. parse_error(ctxt, peek, "key is not a string in object");
  253. goto out;
  254. }
  255. token = parser_context_pop_token(ctxt);
  256. if (token == NULL) {
  257. parse_error(ctxt, NULL, "premature EOI");
  258. goto out;
  259. }
  260. if (token->type != JSON_COLON) {
  261. parse_error(ctxt, token, "missing : in object pair");
  262. goto out;
  263. }
  264. value = parse_value(ctxt, ap);
  265. if (value == NULL) {
  266. parse_error(ctxt, token, "Missing value in dict");
  267. goto out;
  268. }
  269. qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value);
  270. qobject_decref(key);
  271. return 0;
  272. out:
  273. qobject_decref(key);
  274. return -1;
  275. }
  276. static QObject *parse_object(JSONParserContext *ctxt, va_list *ap)
  277. {
  278. QDict *dict = NULL;
  279. JSONToken *token, *peek;
  280. token = parser_context_pop_token(ctxt);
  281. assert(token && token->type == JSON_LCURLY);
  282. dict = qdict_new();
  283. peek = parser_context_peek_token(ctxt);
  284. if (peek == NULL) {
  285. parse_error(ctxt, NULL, "premature EOI");
  286. goto out;
  287. }
  288. if (peek->type != JSON_RCURLY) {
  289. if (parse_pair(ctxt, dict, ap) == -1) {
  290. goto out;
  291. }
  292. token = parser_context_pop_token(ctxt);
  293. if (token == NULL) {
  294. parse_error(ctxt, NULL, "premature EOI");
  295. goto out;
  296. }
  297. while (token->type != JSON_RCURLY) {
  298. if (token->type != JSON_COMMA) {
  299. parse_error(ctxt, token, "expected separator in dict");
  300. goto out;
  301. }
  302. if (parse_pair(ctxt, dict, ap) == -1) {
  303. goto out;
  304. }
  305. token = parser_context_pop_token(ctxt);
  306. if (token == NULL) {
  307. parse_error(ctxt, NULL, "premature EOI");
  308. goto out;
  309. }
  310. }
  311. } else {
  312. (void)parser_context_pop_token(ctxt);
  313. }
  314. return QOBJECT(dict);
  315. out:
  316. QDECREF(dict);
  317. return NULL;
  318. }
  319. static QObject *parse_array(JSONParserContext *ctxt, va_list *ap)
  320. {
  321. QList *list = NULL;
  322. JSONToken *token, *peek;
  323. token = parser_context_pop_token(ctxt);
  324. assert(token && token->type == JSON_LSQUARE);
  325. list = qlist_new();
  326. peek = parser_context_peek_token(ctxt);
  327. if (peek == NULL) {
  328. parse_error(ctxt, NULL, "premature EOI");
  329. goto out;
  330. }
  331. if (peek->type != JSON_RSQUARE) {
  332. QObject *obj;
  333. obj = parse_value(ctxt, ap);
  334. if (obj == NULL) {
  335. parse_error(ctxt, token, "expecting value");
  336. goto out;
  337. }
  338. qlist_append_obj(list, obj);
  339. token = parser_context_pop_token(ctxt);
  340. if (token == NULL) {
  341. parse_error(ctxt, NULL, "premature EOI");
  342. goto out;
  343. }
  344. while (token->type != JSON_RSQUARE) {
  345. if (token->type != JSON_COMMA) {
  346. parse_error(ctxt, token, "expected separator in list");
  347. goto out;
  348. }
  349. obj = parse_value(ctxt, ap);
  350. if (obj == NULL) {
  351. parse_error(ctxt, token, "expecting value");
  352. goto out;
  353. }
  354. qlist_append_obj(list, obj);
  355. token = parser_context_pop_token(ctxt);
  356. if (token == NULL) {
  357. parse_error(ctxt, NULL, "premature EOI");
  358. goto out;
  359. }
  360. }
  361. } else {
  362. (void)parser_context_pop_token(ctxt);
  363. }
  364. return QOBJECT(list);
  365. out:
  366. QDECREF(list);
  367. return NULL;
  368. }
  369. static QObject *parse_keyword(JSONParserContext *ctxt)
  370. {
  371. JSONToken *token;
  372. token = parser_context_pop_token(ctxt);
  373. assert(token && token->type == JSON_KEYWORD);
  374. if (!strcmp(token->str, "true")) {
  375. return QOBJECT(qbool_from_bool(true));
  376. } else if (!strcmp(token->str, "false")) {
  377. return QOBJECT(qbool_from_bool(false));
  378. } else if (!strcmp(token->str, "null")) {
  379. return QOBJECT(qnull());
  380. }
  381. parse_error(ctxt, token, "invalid keyword '%s'", token->str);
  382. return NULL;
  383. }
  384. static QObject *parse_escape(JSONParserContext *ctxt, va_list *ap)
  385. {
  386. JSONToken *token;
  387. if (ap == NULL) {
  388. return NULL;
  389. }
  390. token = parser_context_pop_token(ctxt);
  391. assert(token && token->type == JSON_ESCAPE);
  392. if (!strcmp(token->str, "%p")) {
  393. return va_arg(*ap, QObject *);
  394. } else if (!strcmp(token->str, "%i")) {
  395. return QOBJECT(qbool_from_bool(va_arg(*ap, int)));
  396. } else if (!strcmp(token->str, "%d")) {
  397. return QOBJECT(qnum_from_int(va_arg(*ap, int)));
  398. } else if (!strcmp(token->str, "%ld")) {
  399. return QOBJECT(qnum_from_int(va_arg(*ap, long)));
  400. } else if (!strcmp(token->str, "%lld") ||
  401. !strcmp(token->str, "%I64d")) {
  402. return QOBJECT(qnum_from_int(va_arg(*ap, long long)));
  403. } else if (!strcmp(token->str, "%u")) {
  404. return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned int)));
  405. } else if (!strcmp(token->str, "%lu")) {
  406. return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long)));
  407. } else if (!strcmp(token->str, "%llu") ||
  408. !strcmp(token->str, "%I64u")) {
  409. return QOBJECT(qnum_from_uint(va_arg(*ap, unsigned long long)));
  410. } else if (!strcmp(token->str, "%s")) {
  411. return QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
  412. } else if (!strcmp(token->str, "%f")) {
  413. return QOBJECT(qnum_from_double(va_arg(*ap, double)));
  414. }
  415. return NULL;
  416. }
  417. static QObject *parse_literal(JSONParserContext *ctxt)
  418. {
  419. JSONToken *token;
  420. token = parser_context_pop_token(ctxt);
  421. assert(token);
  422. switch (token->type) {
  423. case JSON_STRING:
  424. return QOBJECT(qstring_from_escaped_str(ctxt, token));
  425. case JSON_INTEGER: {
  426. /*
  427. * Represent JSON_INTEGER as QNUM_I64 if possible, else as
  428. * QNUM_U64, else as QNUM_DOUBLE. Note that qemu_strtoi64()
  429. * and qemu_strtou64() fail with ERANGE when it's not
  430. * possible.
  431. *
  432. * qnum_get_int() will then work for any signed 64-bit
  433. * JSON_INTEGER, qnum_get_uint() for any unsigned 64-bit
  434. * integer, and qnum_get_double() both for any JSON_INTEGER
  435. * and any JSON_FLOAT (with precision loss for integers beyond
  436. * 53 bits)
  437. */
  438. int ret;
  439. int64_t value;
  440. uint64_t uvalue;
  441. ret = qemu_strtoi64(token->str, NULL, 10, &value);
  442. if (!ret) {
  443. return QOBJECT(qnum_from_int(value));
  444. }
  445. assert(ret == -ERANGE);
  446. if (token->str[0] != '-') {
  447. ret = qemu_strtou64(token->str, NULL, 10, &uvalue);
  448. if (!ret) {
  449. return QOBJECT(qnum_from_uint(uvalue));
  450. }
  451. assert(ret == -ERANGE);
  452. }
  453. /* fall through to JSON_FLOAT */
  454. }
  455. case JSON_FLOAT:
  456. /* FIXME dependent on locale; a pervasive issue in QEMU */
  457. /* FIXME our lexer matches RFC 7159 in forbidding Inf or NaN,
  458. * but those might be useful extensions beyond JSON */
  459. return QOBJECT(qnum_from_double(strtod(token->str, NULL)));
  460. default:
  461. abort();
  462. }
  463. }
  464. static QObject *parse_value(JSONParserContext *ctxt, va_list *ap)
  465. {
  466. JSONToken *token;
  467. token = parser_context_peek_token(ctxt);
  468. if (token == NULL) {
  469. parse_error(ctxt, NULL, "premature EOI");
  470. return NULL;
  471. }
  472. switch (token->type) {
  473. case JSON_LCURLY:
  474. return parse_object(ctxt, ap);
  475. case JSON_LSQUARE:
  476. return parse_array(ctxt, ap);
  477. case JSON_ESCAPE:
  478. return parse_escape(ctxt, ap);
  479. case JSON_INTEGER:
  480. case JSON_FLOAT:
  481. case JSON_STRING:
  482. return parse_literal(ctxt);
  483. case JSON_KEYWORD:
  484. return parse_keyword(ctxt);
  485. default:
  486. parse_error(ctxt, token, "expecting value");
  487. return NULL;
  488. }
  489. }
  490. QObject *json_parser_parse(GQueue *tokens, va_list *ap)
  491. {
  492. return json_parser_parse_err(tokens, ap, NULL);
  493. }
  494. QObject *json_parser_parse_err(GQueue *tokens, va_list *ap, Error **errp)
  495. {
  496. JSONParserContext *ctxt = parser_context_new(tokens);
  497. QObject *result;
  498. if (!ctxt) {
  499. return NULL;
  500. }
  501. result = parse_value(ctxt, ap);
  502. error_propagate(errp, ctxt->err);
  503. parser_context_free(ctxt);
  504. return result;
  505. }