2
0

json-parser.c 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650
  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 <stdarg.h>
  14. #include "qemu-common.h"
  15. #include "qstring.h"
  16. #include "qint.h"
  17. #include "qdict.h"
  18. #include "qlist.h"
  19. #include "qfloat.h"
  20. #include "qbool.h"
  21. #include "json-parser.h"
  22. #include "json-lexer.h"
  23. #include "qerror.h"
  24. typedef struct JSONParserContext
  25. {
  26. Error *err;
  27. } JSONParserContext;
  28. #define BUG_ON(cond) assert(!(cond))
  29. /**
  30. * TODO
  31. *
  32. * 0) make errors meaningful again
  33. * 1) add geometry information to tokens
  34. * 3) should we return a parsed size?
  35. * 4) deal with premature EOI
  36. */
  37. static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap);
  38. /**
  39. * Token manipulators
  40. *
  41. * tokens are dictionaries that contain a type, a string value, and geometry information
  42. * about a token identified by the lexer. These are routines that make working with
  43. * these objects a bit easier.
  44. */
  45. static const char *token_get_value(QObject *obj)
  46. {
  47. return qdict_get_str(qobject_to_qdict(obj), "token");
  48. }
  49. static JSONTokenType token_get_type(QObject *obj)
  50. {
  51. return qdict_get_int(qobject_to_qdict(obj), "type");
  52. }
  53. static int token_is_operator(QObject *obj, char op)
  54. {
  55. const char *val;
  56. if (token_get_type(obj) != JSON_OPERATOR) {
  57. return 0;
  58. }
  59. val = token_get_value(obj);
  60. return (val[0] == op) && (val[1] == 0);
  61. }
  62. static int token_is_keyword(QObject *obj, const char *value)
  63. {
  64. if (token_get_type(obj) != JSON_KEYWORD) {
  65. return 0;
  66. }
  67. return strcmp(token_get_value(obj), value) == 0;
  68. }
  69. static int token_is_escape(QObject *obj, const char *value)
  70. {
  71. if (token_get_type(obj) != JSON_ESCAPE) {
  72. return 0;
  73. }
  74. return (strcmp(token_get_value(obj), value) == 0);
  75. }
  76. /**
  77. * Error handler
  78. */
  79. static void GCC_FMT_ATTR(3, 4) parse_error(JSONParserContext *ctxt,
  80. QObject *token, const char *msg, ...)
  81. {
  82. va_list ap;
  83. char message[1024];
  84. va_start(ap, msg);
  85. vsnprintf(message, sizeof(message), msg, ap);
  86. va_end(ap);
  87. if (ctxt->err) {
  88. error_free(ctxt->err);
  89. ctxt->err = NULL;
  90. }
  91. error_set(&ctxt->err, QERR_JSON_PARSE_ERROR, message);
  92. }
  93. /**
  94. * String helpers
  95. *
  96. * These helpers are used to unescape strings.
  97. */
  98. static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length)
  99. {
  100. if (wchar <= 0x007F) {
  101. BUG_ON(buffer_length < 2);
  102. buffer[0] = wchar & 0x7F;
  103. buffer[1] = 0;
  104. } else if (wchar <= 0x07FF) {
  105. BUG_ON(buffer_length < 3);
  106. buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F);
  107. buffer[1] = 0x80 | (wchar & 0x3F);
  108. buffer[2] = 0;
  109. } else {
  110. BUG_ON(buffer_length < 4);
  111. buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F);
  112. buffer[1] = 0x80 | ((wchar >> 6) & 0x3F);
  113. buffer[2] = 0x80 | (wchar & 0x3F);
  114. buffer[3] = 0;
  115. }
  116. }
  117. static int hex2decimal(char ch)
  118. {
  119. if (ch >= '0' && ch <= '9') {
  120. return (ch - '0');
  121. } else if (ch >= 'a' && ch <= 'f') {
  122. return 10 + (ch - 'a');
  123. } else if (ch >= 'A' && ch <= 'F') {
  124. return 10 + (ch - 'A');
  125. }
  126. return -1;
  127. }
  128. /**
  129. * parse_string(): Parse a json string and return a QObject
  130. *
  131. * string
  132. * ""
  133. * " chars "
  134. * chars
  135. * char
  136. * char chars
  137. * char
  138. * any-Unicode-character-
  139. * except-"-or-\-or-
  140. * control-character
  141. * \"
  142. * \\
  143. * \/
  144. * \b
  145. * \f
  146. * \n
  147. * \r
  148. * \t
  149. * \u four-hex-digits
  150. */
  151. static QString *qstring_from_escaped_str(JSONParserContext *ctxt, QObject *token)
  152. {
  153. const char *ptr = token_get_value(token);
  154. QString *str;
  155. int double_quote = 1;
  156. if (*ptr == '"') {
  157. double_quote = 1;
  158. } else {
  159. double_quote = 0;
  160. }
  161. ptr++;
  162. str = qstring_new();
  163. while (*ptr &&
  164. ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) {
  165. if (*ptr == '\\') {
  166. ptr++;
  167. switch (*ptr) {
  168. case '"':
  169. qstring_append(str, "\"");
  170. ptr++;
  171. break;
  172. case '\'':
  173. qstring_append(str, "'");
  174. ptr++;
  175. break;
  176. case '\\':
  177. qstring_append(str, "\\");
  178. ptr++;
  179. break;
  180. case '/':
  181. qstring_append(str, "/");
  182. ptr++;
  183. break;
  184. case 'b':
  185. qstring_append(str, "\b");
  186. ptr++;
  187. break;
  188. case 'f':
  189. qstring_append(str, "\f");
  190. ptr++;
  191. break;
  192. case 'n':
  193. qstring_append(str, "\n");
  194. ptr++;
  195. break;
  196. case 'r':
  197. qstring_append(str, "\r");
  198. ptr++;
  199. break;
  200. case 't':
  201. qstring_append(str, "\t");
  202. ptr++;
  203. break;
  204. case 'u': {
  205. uint16_t unicode_char = 0;
  206. char utf8_char[4];
  207. int i = 0;
  208. ptr++;
  209. for (i = 0; i < 4; i++) {
  210. if (qemu_isxdigit(*ptr)) {
  211. unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4);
  212. } else {
  213. parse_error(ctxt, token,
  214. "invalid hex escape sequence in string");
  215. goto out;
  216. }
  217. ptr++;
  218. }
  219. wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char));
  220. qstring_append(str, utf8_char);
  221. } break;
  222. default:
  223. parse_error(ctxt, token, "invalid escape sequence in string");
  224. goto out;
  225. }
  226. } else {
  227. char dummy[2];
  228. dummy[0] = *ptr++;
  229. dummy[1] = 0;
  230. qstring_append(str, dummy);
  231. }
  232. }
  233. return str;
  234. out:
  235. QDECREF(str);
  236. return NULL;
  237. }
  238. /**
  239. * Parsing rules
  240. */
  241. static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap)
  242. {
  243. QObject *key = NULL, *token = NULL, *value, *peek;
  244. QList *working = qlist_copy(*tokens);
  245. peek = qlist_peek(working);
  246. if (peek == NULL) {
  247. parse_error(ctxt, NULL, "premature EOI");
  248. goto out;
  249. }
  250. key = parse_value(ctxt, &working, 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 = qlist_pop(working);
  256. if (token == NULL) {
  257. parse_error(ctxt, NULL, "premature EOI");
  258. goto out;
  259. }
  260. if (!token_is_operator(token, ':')) {
  261. parse_error(ctxt, token, "missing : in object pair");
  262. goto out;
  263. }
  264. value = parse_value(ctxt, &working, 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(token);
  271. qobject_decref(key);
  272. QDECREF(*tokens);
  273. *tokens = working;
  274. return 0;
  275. out:
  276. qobject_decref(token);
  277. qobject_decref(key);
  278. QDECREF(working);
  279. return -1;
  280. }
  281. static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  282. {
  283. QDict *dict = NULL;
  284. QObject *token, *peek;
  285. QList *working = qlist_copy(*tokens);
  286. token = qlist_pop(working);
  287. if (token == NULL) {
  288. goto out;
  289. }
  290. if (!token_is_operator(token, '{')) {
  291. goto out;
  292. }
  293. qobject_decref(token);
  294. token = NULL;
  295. dict = qdict_new();
  296. peek = qlist_peek(working);
  297. if (peek == NULL) {
  298. parse_error(ctxt, NULL, "premature EOI");
  299. goto out;
  300. }
  301. if (!token_is_operator(peek, '}')) {
  302. if (parse_pair(ctxt, dict, &working, ap) == -1) {
  303. goto out;
  304. }
  305. token = qlist_pop(working);
  306. if (token == NULL) {
  307. parse_error(ctxt, NULL, "premature EOI");
  308. goto out;
  309. }
  310. while (!token_is_operator(token, '}')) {
  311. if (!token_is_operator(token, ',')) {
  312. parse_error(ctxt, token, "expected separator in dict");
  313. goto out;
  314. }
  315. qobject_decref(token);
  316. token = NULL;
  317. if (parse_pair(ctxt, dict, &working, ap) == -1) {
  318. goto out;
  319. }
  320. token = qlist_pop(working);
  321. if (token == NULL) {
  322. parse_error(ctxt, NULL, "premature EOI");
  323. goto out;
  324. }
  325. }
  326. qobject_decref(token);
  327. token = NULL;
  328. } else {
  329. token = qlist_pop(working);
  330. qobject_decref(token);
  331. token = NULL;
  332. }
  333. QDECREF(*tokens);
  334. *tokens = working;
  335. return QOBJECT(dict);
  336. out:
  337. qobject_decref(token);
  338. QDECREF(working);
  339. QDECREF(dict);
  340. return NULL;
  341. }
  342. static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  343. {
  344. QList *list = NULL;
  345. QObject *token, *peek;
  346. QList *working = qlist_copy(*tokens);
  347. token = qlist_pop(working);
  348. if (token == NULL) {
  349. goto out;
  350. }
  351. if (!token_is_operator(token, '[')) {
  352. goto out;
  353. }
  354. qobject_decref(token);
  355. token = NULL;
  356. list = qlist_new();
  357. peek = qlist_peek(working);
  358. if (peek == NULL) {
  359. parse_error(ctxt, NULL, "premature EOI");
  360. goto out;
  361. }
  362. if (!token_is_operator(peek, ']')) {
  363. QObject *obj;
  364. obj = parse_value(ctxt, &working, ap);
  365. if (obj == NULL) {
  366. parse_error(ctxt, token, "expecting value");
  367. goto out;
  368. }
  369. qlist_append_obj(list, obj);
  370. token = qlist_pop(working);
  371. if (token == NULL) {
  372. parse_error(ctxt, NULL, "premature EOI");
  373. goto out;
  374. }
  375. while (!token_is_operator(token, ']')) {
  376. if (!token_is_operator(token, ',')) {
  377. parse_error(ctxt, token, "expected separator in list");
  378. goto out;
  379. }
  380. qobject_decref(token);
  381. token = NULL;
  382. obj = parse_value(ctxt, &working, ap);
  383. if (obj == NULL) {
  384. parse_error(ctxt, token, "expecting value");
  385. goto out;
  386. }
  387. qlist_append_obj(list, obj);
  388. token = qlist_pop(working);
  389. if (token == NULL) {
  390. parse_error(ctxt, NULL, "premature EOI");
  391. goto out;
  392. }
  393. }
  394. qobject_decref(token);
  395. token = NULL;
  396. } else {
  397. token = qlist_pop(working);
  398. qobject_decref(token);
  399. token = NULL;
  400. }
  401. QDECREF(*tokens);
  402. *tokens = working;
  403. return QOBJECT(list);
  404. out:
  405. qobject_decref(token);
  406. QDECREF(working);
  407. QDECREF(list);
  408. return NULL;
  409. }
  410. static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens)
  411. {
  412. QObject *token, *ret;
  413. QList *working = qlist_copy(*tokens);
  414. token = qlist_pop(working);
  415. if (token == NULL) {
  416. goto out;
  417. }
  418. if (token_get_type(token) != JSON_KEYWORD) {
  419. goto out;
  420. }
  421. if (token_is_keyword(token, "true")) {
  422. ret = QOBJECT(qbool_from_int(true));
  423. } else if (token_is_keyword(token, "false")) {
  424. ret = QOBJECT(qbool_from_int(false));
  425. } else {
  426. parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token));
  427. goto out;
  428. }
  429. qobject_decref(token);
  430. QDECREF(*tokens);
  431. *tokens = working;
  432. return ret;
  433. out:
  434. qobject_decref(token);
  435. QDECREF(working);
  436. return NULL;
  437. }
  438. static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  439. {
  440. QObject *token = NULL, *obj;
  441. QList *working = qlist_copy(*tokens);
  442. if (ap == NULL) {
  443. goto out;
  444. }
  445. token = qlist_pop(working);
  446. if (token == NULL) {
  447. goto out;
  448. }
  449. if (token_is_escape(token, "%p")) {
  450. obj = va_arg(*ap, QObject *);
  451. } else if (token_is_escape(token, "%i")) {
  452. obj = QOBJECT(qbool_from_int(va_arg(*ap, int)));
  453. } else if (token_is_escape(token, "%d")) {
  454. obj = QOBJECT(qint_from_int(va_arg(*ap, int)));
  455. } else if (token_is_escape(token, "%ld")) {
  456. obj = QOBJECT(qint_from_int(va_arg(*ap, long)));
  457. } else if (token_is_escape(token, "%lld") ||
  458. token_is_escape(token, "%I64d")) {
  459. obj = QOBJECT(qint_from_int(va_arg(*ap, long long)));
  460. } else if (token_is_escape(token, "%s")) {
  461. obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
  462. } else if (token_is_escape(token, "%f")) {
  463. obj = QOBJECT(qfloat_from_double(va_arg(*ap, double)));
  464. } else {
  465. goto out;
  466. }
  467. qobject_decref(token);
  468. QDECREF(*tokens);
  469. *tokens = working;
  470. return obj;
  471. out:
  472. qobject_decref(token);
  473. QDECREF(working);
  474. return NULL;
  475. }
  476. static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens)
  477. {
  478. QObject *token, *obj;
  479. QList *working = qlist_copy(*tokens);
  480. token = qlist_pop(working);
  481. if (token == NULL) {
  482. goto out;
  483. }
  484. switch (token_get_type(token)) {
  485. case JSON_STRING:
  486. obj = QOBJECT(qstring_from_escaped_str(ctxt, token));
  487. break;
  488. case JSON_INTEGER:
  489. obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10)));
  490. break;
  491. case JSON_FLOAT:
  492. /* FIXME dependent on locale */
  493. obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL)));
  494. break;
  495. default:
  496. goto out;
  497. }
  498. qobject_decref(token);
  499. QDECREF(*tokens);
  500. *tokens = working;
  501. return obj;
  502. out:
  503. qobject_decref(token);
  504. QDECREF(working);
  505. return NULL;
  506. }
  507. static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  508. {
  509. QObject *obj;
  510. obj = parse_object(ctxt, tokens, ap);
  511. if (obj == NULL) {
  512. obj = parse_array(ctxt, tokens, ap);
  513. }
  514. if (obj == NULL) {
  515. obj = parse_escape(ctxt, tokens, ap);
  516. }
  517. if (obj == NULL) {
  518. obj = parse_keyword(ctxt, tokens);
  519. }
  520. if (obj == NULL) {
  521. obj = parse_literal(ctxt, tokens);
  522. }
  523. return obj;
  524. }
  525. QObject *json_parser_parse(QList *tokens, va_list *ap)
  526. {
  527. return json_parser_parse_err(tokens, ap, NULL);
  528. }
  529. QObject *json_parser_parse_err(QList *tokens, va_list *ap, Error **errp)
  530. {
  531. JSONParserContext ctxt = {};
  532. QList *working;
  533. QObject *result;
  534. if (!tokens) {
  535. return NULL;
  536. }
  537. working = qlist_copy(tokens);
  538. result = parse_value(&ctxt, &working, ap);
  539. QDECREF(working);
  540. error_propagate(errp, ctxt.err);
  541. return result;
  542. }