json-parser.c 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590
  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, *token = NULL, *value, *peek;
  244. QList *working = qlist_copy(*tokens);
  245. peek = qlist_peek(working);
  246. key = parse_value(ctxt, &working, ap);
  247. if (!key || qobject_type(key) != QTYPE_QSTRING) {
  248. parse_error(ctxt, peek, "key is not a string in object");
  249. goto out;
  250. }
  251. token = qlist_pop(working);
  252. if (!token_is_operator(token, ':')) {
  253. parse_error(ctxt, token, "missing : in object pair");
  254. goto out;
  255. }
  256. value = parse_value(ctxt, &working, ap);
  257. if (value == NULL) {
  258. parse_error(ctxt, token, "Missing value in dict");
  259. goto out;
  260. }
  261. qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value);
  262. qobject_decref(token);
  263. qobject_decref(key);
  264. QDECREF(*tokens);
  265. *tokens = working;
  266. return 0;
  267. out:
  268. qobject_decref(token);
  269. qobject_decref(key);
  270. QDECREF(working);
  271. return -1;
  272. }
  273. static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  274. {
  275. QDict *dict = NULL;
  276. QObject *token, *peek;
  277. QList *working = qlist_copy(*tokens);
  278. token = qlist_pop(working);
  279. if (!token_is_operator(token, '{')) {
  280. goto out;
  281. }
  282. qobject_decref(token);
  283. token = NULL;
  284. dict = qdict_new();
  285. peek = qlist_peek(working);
  286. if (!token_is_operator(peek, '}')) {
  287. if (parse_pair(ctxt, dict, &working, ap) == -1) {
  288. goto out;
  289. }
  290. token = qlist_pop(working);
  291. while (!token_is_operator(token, '}')) {
  292. if (!token_is_operator(token, ',')) {
  293. parse_error(ctxt, token, "expected separator in dict");
  294. goto out;
  295. }
  296. qobject_decref(token);
  297. token = NULL;
  298. if (parse_pair(ctxt, dict, &working, ap) == -1) {
  299. goto out;
  300. }
  301. token = qlist_pop(working);
  302. }
  303. qobject_decref(token);
  304. token = NULL;
  305. } else {
  306. token = qlist_pop(working);
  307. qobject_decref(token);
  308. token = NULL;
  309. }
  310. QDECREF(*tokens);
  311. *tokens = working;
  312. return QOBJECT(dict);
  313. out:
  314. qobject_decref(token);
  315. QDECREF(working);
  316. QDECREF(dict);
  317. return NULL;
  318. }
  319. static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  320. {
  321. QList *list = NULL;
  322. QObject *token, *peek;
  323. QList *working = qlist_copy(*tokens);
  324. token = qlist_pop(working);
  325. if (!token_is_operator(token, '[')) {
  326. goto out;
  327. }
  328. qobject_decref(token);
  329. token = NULL;
  330. list = qlist_new();
  331. peek = qlist_peek(working);
  332. if (!token_is_operator(peek, ']')) {
  333. QObject *obj;
  334. obj = parse_value(ctxt, &working, ap);
  335. if (obj == NULL) {
  336. parse_error(ctxt, token, "expecting value");
  337. goto out;
  338. }
  339. qlist_append_obj(list, obj);
  340. token = qlist_pop(working);
  341. while (!token_is_operator(token, ']')) {
  342. if (!token_is_operator(token, ',')) {
  343. parse_error(ctxt, token, "expected separator in list");
  344. goto out;
  345. }
  346. qobject_decref(token);
  347. token = NULL;
  348. obj = parse_value(ctxt, &working, ap);
  349. if (obj == NULL) {
  350. parse_error(ctxt, token, "expecting value");
  351. goto out;
  352. }
  353. qlist_append_obj(list, obj);
  354. token = qlist_pop(working);
  355. }
  356. qobject_decref(token);
  357. token = NULL;
  358. } else {
  359. token = qlist_pop(working);
  360. qobject_decref(token);
  361. token = NULL;
  362. }
  363. QDECREF(*tokens);
  364. *tokens = working;
  365. return QOBJECT(list);
  366. out:
  367. qobject_decref(token);
  368. QDECREF(working);
  369. QDECREF(list);
  370. return NULL;
  371. }
  372. static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens)
  373. {
  374. QObject *token, *ret;
  375. QList *working = qlist_copy(*tokens);
  376. token = qlist_pop(working);
  377. if (token_get_type(token) != JSON_KEYWORD) {
  378. goto out;
  379. }
  380. if (token_is_keyword(token, "true")) {
  381. ret = QOBJECT(qbool_from_int(true));
  382. } else if (token_is_keyword(token, "false")) {
  383. ret = QOBJECT(qbool_from_int(false));
  384. } else {
  385. parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token));
  386. goto out;
  387. }
  388. qobject_decref(token);
  389. QDECREF(*tokens);
  390. *tokens = working;
  391. return ret;
  392. out:
  393. qobject_decref(token);
  394. QDECREF(working);
  395. return NULL;
  396. }
  397. static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  398. {
  399. QObject *token = NULL, *obj;
  400. QList *working = qlist_copy(*tokens);
  401. if (ap == NULL) {
  402. goto out;
  403. }
  404. token = qlist_pop(working);
  405. if (token_is_escape(token, "%p")) {
  406. obj = va_arg(*ap, QObject *);
  407. } else if (token_is_escape(token, "%i")) {
  408. obj = QOBJECT(qbool_from_int(va_arg(*ap, int)));
  409. } else if (token_is_escape(token, "%d")) {
  410. obj = QOBJECT(qint_from_int(va_arg(*ap, int)));
  411. } else if (token_is_escape(token, "%ld")) {
  412. obj = QOBJECT(qint_from_int(va_arg(*ap, long)));
  413. } else if (token_is_escape(token, "%lld") ||
  414. token_is_escape(token, "%I64d")) {
  415. obj = QOBJECT(qint_from_int(va_arg(*ap, long long)));
  416. } else if (token_is_escape(token, "%s")) {
  417. obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
  418. } else if (token_is_escape(token, "%f")) {
  419. obj = QOBJECT(qfloat_from_double(va_arg(*ap, double)));
  420. } else {
  421. goto out;
  422. }
  423. qobject_decref(token);
  424. QDECREF(*tokens);
  425. *tokens = working;
  426. return obj;
  427. out:
  428. qobject_decref(token);
  429. QDECREF(working);
  430. return NULL;
  431. }
  432. static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens)
  433. {
  434. QObject *token, *obj;
  435. QList *working = qlist_copy(*tokens);
  436. token = qlist_pop(working);
  437. switch (token_get_type(token)) {
  438. case JSON_STRING:
  439. obj = QOBJECT(qstring_from_escaped_str(ctxt, token));
  440. break;
  441. case JSON_INTEGER:
  442. obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10)));
  443. break;
  444. case JSON_FLOAT:
  445. /* FIXME dependent on locale */
  446. obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL)));
  447. break;
  448. default:
  449. goto out;
  450. }
  451. qobject_decref(token);
  452. QDECREF(*tokens);
  453. *tokens = working;
  454. return obj;
  455. out:
  456. qobject_decref(token);
  457. QDECREF(working);
  458. return NULL;
  459. }
  460. static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap)
  461. {
  462. QObject *obj;
  463. obj = parse_object(ctxt, tokens, ap);
  464. if (obj == NULL) {
  465. obj = parse_array(ctxt, tokens, ap);
  466. }
  467. if (obj == NULL) {
  468. obj = parse_escape(ctxt, tokens, ap);
  469. }
  470. if (obj == NULL) {
  471. obj = parse_keyword(ctxt, tokens);
  472. }
  473. if (obj == NULL) {
  474. obj = parse_literal(ctxt, tokens);
  475. }
  476. return obj;
  477. }
  478. QObject *json_parser_parse(QList *tokens, va_list *ap)
  479. {
  480. return json_parser_parse_err(tokens, ap, NULL);
  481. }
  482. QObject *json_parser_parse_err(QList *tokens, va_list *ap, Error **errp)
  483. {
  484. JSONParserContext ctxt = {};
  485. QList *working = qlist_copy(tokens);
  486. QObject *result;
  487. result = parse_value(&ctxt, &working, ap);
  488. QDECREF(working);
  489. error_propagate(errp, ctxt.err);
  490. return result;
  491. }