test_parse.c 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394
  1. #include <stdio.h>
  2. #include <stdlib.h>
  3. #include <stddef.h>
  4. #include <string.h>
  5. #include <assert.h>
  6. #include "json.h"
  7. #include "json_tokener.h"
  8. #include "json_visit.h"
  9. static void test_basic_parse(void);
  10. static void test_verbose_parse(void);
  11. static void test_incremental_parse(void);
  12. int main(void)
  13. {
  14. MC_SET_DEBUG(1);
  15. static const char separator[] = "==================================";
  16. test_basic_parse();
  17. puts(separator);
  18. test_verbose_parse();
  19. puts(separator);
  20. test_incremental_parse();
  21. puts(separator);
  22. }
  23. static json_c_visit_userfunc clear_serializer;
  24. static void do_clear_serializer(json_object *jso);
  25. static void single_basic_parse(const char *test_string, int clear_serializer)
  26. {
  27. json_object *new_obj;
  28. new_obj = json_tokener_parse(test_string);
  29. if (clear_serializer)
  30. do_clear_serializer(new_obj);
  31. printf("new_obj.to_string(%s)=%s\n", test_string, json_object_to_json_string(new_obj));
  32. json_object_put(new_obj);
  33. }
  34. static void test_basic_parse()
  35. {
  36. single_basic_parse("\"\003\"", 0);
  37. single_basic_parse("/* hello */\"foo\"", 0);
  38. single_basic_parse("// hello\n\"foo\"", 0);
  39. single_basic_parse("\"foo\"blue", 0);
  40. single_basic_parse("\"\\u0041\\u0042\\u0043\"", 0);
  41. // Test with a "short" high surrogate
  42. single_basic_parse("[9,'\\uDAD", 0);
  43. single_basic_parse("null", 0);
  44. single_basic_parse("NaN", 0);
  45. single_basic_parse("-NaN", 0); /* non-sensical, returns null */
  46. single_basic_parse("Inf", 0); /* must use full string, returns null */
  47. single_basic_parse("inf", 0); /* must use full string, returns null */
  48. single_basic_parse("Infinity", 0);
  49. single_basic_parse("infinity", 0);
  50. single_basic_parse("-Infinity", 0);
  51. single_basic_parse("-infinity", 0);
  52. single_basic_parse("{ \"min\": Infinity, \"max\": -Infinity}", 0);
  53. single_basic_parse("Infinity!", 0);
  54. single_basic_parse("Infinitynull", 0);
  55. single_basic_parse("InfinityXXXX", 0);
  56. single_basic_parse("-Infinitynull", 0);
  57. single_basic_parse("-InfinityXXXX", 0);
  58. single_basic_parse("Infinoodle", 0);
  59. single_basic_parse("InfinAAA", 0);
  60. single_basic_parse("-Infinoodle", 0);
  61. single_basic_parse("-InfinAAA", 0);
  62. single_basic_parse("True", 0);
  63. single_basic_parse("False", 0);
  64. single_basic_parse("12", 0);
  65. single_basic_parse("12.3", 0);
  66. single_basic_parse("12.3.4", 0); /* non-sensical, returns null */
  67. /* was returning (int)2015 before patch, should return null */
  68. single_basic_parse("2015-01-15", 0);
  69. /* ...but this works. It's rather inconsistent, and a future major release
  70. * should change the behavior so it either always returns null when extra
  71. * bytes are present (preferred), or always return object created from as much
  72. * as was able to be parsed.
  73. */
  74. single_basic_parse("12.3xxx", 0);
  75. single_basic_parse("{\"FoO\" : -12.3E512}", 0);
  76. single_basic_parse("{\"FoO\" : -12.3E51.2}", 0); /* non-sensical, returns null */
  77. single_basic_parse("[\"\\n\"]", 0);
  78. single_basic_parse("[\"\\nabc\\n\"]", 0);
  79. single_basic_parse("[null]", 0);
  80. single_basic_parse("[]", 0);
  81. single_basic_parse("[false]", 0);
  82. single_basic_parse("[\"abc\",null,\"def\",12]", 0);
  83. single_basic_parse("{}", 0);
  84. single_basic_parse("{ \"foo\": \"bar\" }", 0);
  85. single_basic_parse("{ \"foo\": \"bar\", \"baz\": null, \"bool0\": true }", 0);
  86. single_basic_parse("{ \"foo\": [null, \"foo\"] }", 0);
  87. single_basic_parse("{ \"abc\": 12, \"foo\": \"bar\", \"bool0\": false, \"bool1\": true, \"arr\": [ 1, 2, 3, null, 5 ] }", 0);
  88. single_basic_parse("{ \"abc\": \"blue\nred\\ngreen\" }", 0);
  89. // Clear serializer for these tests so we see the actual parsed value.
  90. single_basic_parse("[0e]", 1);
  91. single_basic_parse("[0e+]", 1);
  92. single_basic_parse("[0e+-1]", 1);
  93. single_basic_parse("[18446744073709551616]", 1);
  94. }
  95. // Clear the re-serialization information that the tokener
  96. // saves to ensure that the output reflects the actual
  97. // values we parsed, rather than just the original input.
  98. static void do_clear_serializer(json_object *jso)
  99. {
  100. json_c_visit(jso, 0, clear_serializer, NULL);
  101. }
  102. static int clear_serializer(json_object *jso, int flags,
  103. json_object *parent_jso,
  104. const char *jso_key,
  105. size_t *jso_index, void *userarg)
  106. {
  107. if (jso)
  108. json_object_set_serializer(jso, NULL, NULL, NULL);
  109. return JSON_C_VISIT_RETURN_CONTINUE;
  110. }
  111. static void test_verbose_parse()
  112. {
  113. json_object *new_obj;
  114. enum json_tokener_error error = json_tokener_success;
  115. new_obj = json_tokener_parse_verbose("{ foo }", &error);
  116. assert (error == json_tokener_error_parse_object_key_name);
  117. assert (new_obj == NULL);
  118. new_obj = json_tokener_parse("{ foo }");
  119. assert (new_obj == NULL);
  120. new_obj = json_tokener_parse("foo");
  121. assert (new_obj == NULL);
  122. new_obj = json_tokener_parse_verbose("foo", &error);
  123. assert (new_obj == NULL);
  124. /* b/c the string starts with 'f' parsing return a boolean error */
  125. assert (error == json_tokener_error_parse_boolean);
  126. puts("json_tokener_parse_versbose() OK");
  127. }
  128. struct incremental_step {
  129. const char *string_to_parse;
  130. int length;
  131. int char_offset;
  132. enum json_tokener_error expected_error;
  133. int reset_tokener;
  134. } incremental_steps[] = {
  135. /* Check that full json messages can be parsed, both w/ and w/o a reset */
  136. { "{ \"foo\": 123 }", -1, -1, json_tokener_success, 0 },
  137. { "{ \"foo\": 456 }", -1, -1, json_tokener_success, 1 },
  138. { "{ \"foo\": 789 }", -1, -1, json_tokener_success, 1 },
  139. /* Check a basic incremental parse */
  140. { "{ \"foo", -1, -1, json_tokener_continue, 0 },
  141. { "\": {\"bar", -1, -1, json_tokener_continue, 0 },
  142. { "\":13}}", -1, -1, json_tokener_success, 1 },
  143. /* Check that json_tokener_reset actually resets */
  144. { "{ \"foo", -1, -1, json_tokener_continue, 1 },
  145. { ": \"bar\"}", -1, 0, json_tokener_error_parse_unexpected, 1 },
  146. /* Check incremental parsing with trailing characters */
  147. { "{ \"foo", -1, -1, json_tokener_continue, 0 },
  148. { "\": {\"bar", -1, -1, json_tokener_continue, 0 },
  149. { "\":13}}XXXX", 10, 6, json_tokener_success, 0 },
  150. { "XXXX", 4, 0, json_tokener_error_parse_unexpected, 1 },
  151. /* Check that trailing characters can change w/o a reset */
  152. { "{\"x\": 123 }\"X\"", -1, 11, json_tokener_success, 0 },
  153. { "\"Y\"", -1, -1, json_tokener_success, 1 },
  154. /* To stop parsing a number we need to reach a non-digit, e.g. a \0 */
  155. { "1", 1, 1, json_tokener_continue, 0 },
  156. /* This should parse as the number 12, since it continues the "1" */
  157. { "2", 2, 1, json_tokener_success, 0 },
  158. { "12{", 3, 2, json_tokener_success, 1 },
  159. /* Similar tests for other kinds of objects: */
  160. /* These could all return success immediately, since regardless of
  161. what follows the false/true/null token we *will* return a json object,
  162. but it currently doesn't work that way. hmm... */
  163. { "false", 5, 5, json_tokener_continue, 1 },
  164. { "false", 6, 5, json_tokener_success, 1 },
  165. { "true", 4, 4, json_tokener_continue, 1 },
  166. { "true", 5, 4, json_tokener_success, 1 },
  167. { "null", 4, 4, json_tokener_continue, 1 },
  168. { "null", 5, 4, json_tokener_success, 1 },
  169. { "Infinity", 9, 8, json_tokener_success, 1 },
  170. { "infinity", 9, 8, json_tokener_success, 1 },
  171. { "-infinity", 10, 9, json_tokener_success, 1 },
  172. { "infinity", 9, 0, json_tokener_error_parse_unexpected, 3 },
  173. { "-infinity", 10, 1, json_tokener_error_parse_unexpected, 3 },
  174. { "inf", 3, 3, json_tokener_continue, 0 },
  175. { "inity", 6, 5, json_tokener_success, 1 },
  176. { "-inf", 4, 4, json_tokener_continue, 0 },
  177. { "inity", 6, 5, json_tokener_success, 1 },
  178. { "i", 1, 1, json_tokener_continue, 0 },
  179. { "n", 1, 1, json_tokener_continue, 0 },
  180. { "f", 1, 1, json_tokener_continue, 0 },
  181. { "i", 1, 1, json_tokener_continue, 0 },
  182. { "n", 1, 1, json_tokener_continue, 0 },
  183. { "i", 1, 1, json_tokener_continue, 0 },
  184. { "t", 1, 1, json_tokener_continue, 0 },
  185. { "y", 1, 1, json_tokener_continue, 0 },
  186. { "", 1, 0, json_tokener_success, 1 },
  187. { "-", 1, 1, json_tokener_continue, 0 },
  188. { "inf", 3, 3, json_tokener_continue, 0 },
  189. { "ini", 3, 3, json_tokener_continue, 0 },
  190. { "ty", 3, 2, json_tokener_success, 1 },
  191. { "-", 1, 1, json_tokener_continue, 0 },
  192. { "i", 1, 1, json_tokener_continue, 0 },
  193. { "nfini", 5, 5, json_tokener_continue, 0 },
  194. { "ty", 3, 2, json_tokener_success, 1 },
  195. { "-i", 2, 2, json_tokener_continue, 0 },
  196. { "nfinity", 8, 7, json_tokener_success, 1 },
  197. { "InfinityX", 10, 8, json_tokener_success, 0 },
  198. { "X", 1, 0, json_tokener_error_parse_unexpected, 1 },
  199. { "Infinity1234", 13, 8, json_tokener_success, 0 },
  200. { "1234", 5, 4, json_tokener_success, 1 },
  201. { "Infinity9999", 8, 8, json_tokener_continue, 0 },
  202. /* returns the Infinity loaded up by the previous call: */
  203. { "1234", 5, 0, json_tokener_success, 0 },
  204. { "1234", 5, 4, json_tokener_success, 1 },
  205. /* offset=1 because "n" is the start of "null". hmm... */
  206. { "noodle", 7, 1, json_tokener_error_parse_null, 1 },
  207. /* offset=2 because "na" is the start of "nan". hmm... */
  208. { "naodle", 7, 2, json_tokener_error_parse_null, 1 },
  209. /* offset=2 because "tr" is the start of "true". hmm... */
  210. { "track", 6, 2, json_tokener_error_parse_boolean, 1 },
  211. /* Although they may initially look like they should fail,
  212. the next few tests check that parsing multiple sequential
  213. json objects in the input works as expected */
  214. { "null123", 9, 4, json_tokener_success, 0 },
  215. { "null123" + 4, 4, 3, json_tokener_success, 1 },
  216. { "nullx", 5, 4, json_tokener_success, 0 },
  217. { "nullx" + 4, 2, 0, json_tokener_error_parse_unexpected, 1 },
  218. { "{\"a\":1}{\"b\":2}",15, 7, json_tokener_success, 0 },
  219. { "{\"a\":1}{\"b\":2}" + 7,
  220. 8, 7, json_tokener_success, 1 },
  221. /* Some bad formatting. Check we get the correct error status */
  222. { "2015-01-15", 10, 4, json_tokener_error_parse_number, 1 },
  223. /* Strings have a well defined end point, so we can stop at the quote */
  224. { "\"blue\"", -1, -1, json_tokener_success, 0 },
  225. /* Check each of the escape sequences defined by the spec */
  226. { "\"\\\"\"", -1, -1, json_tokener_success, 0 },
  227. { "\"\\\\\"", -1, -1, json_tokener_success, 0 },
  228. { "\"\\b\"", -1, -1, json_tokener_success, 0 },
  229. { "\"\\f\"", -1, -1, json_tokener_success, 0 },
  230. { "\"\\n\"", -1, -1, json_tokener_success, 0 },
  231. { "\"\\r\"", -1, -1, json_tokener_success, 0 },
  232. { "\"\\t\"", -1, -1, json_tokener_success, 0 },
  233. { "\"\\/\"", -1, -1, json_tokener_success, 0 },
  234. // Escaping a forward slash is optional
  235. { "\"/\"", -1, -1, json_tokener_success, 0 },
  236. { "[1,2,3]", -1, -1, json_tokener_success, 0 },
  237. /* This behaviour doesn't entirely follow the json spec, but until we have
  238. a way to specify how strict to be we follow Postel's Law and be liberal
  239. in what we accept (up to a point). */
  240. { "[1,2,3,]", -1, -1, json_tokener_success, 0 },
  241. { "[1,2,,3,]", -1, 5, json_tokener_error_parse_unexpected, 0 },
  242. { "[1,2,3,]", -1, 7, json_tokener_error_parse_unexpected, 3 },
  243. { "{\"a\":1,}", -1, 7, json_tokener_error_parse_unexpected, 3 },
  244. { NULL, -1, -1, json_tokener_success, 0 },
  245. };
  246. static void test_incremental_parse()
  247. {
  248. json_object *new_obj;
  249. enum json_tokener_error jerr;
  250. struct json_tokener *tok;
  251. const char *string_to_parse;
  252. int ii;
  253. int num_ok, num_error;
  254. num_ok = 0;
  255. num_error = 0;
  256. printf("Starting incremental tests.\n");
  257. printf("Note: quotes and backslashes seen in the output here are literal values passed\n");
  258. printf(" to the parse functions. e.g. this is 4 characters: \"\\f\"\n");
  259. string_to_parse = "{ \"foo"; /* } */
  260. printf("json_tokener_parse(%s) ... ", string_to_parse);
  261. new_obj = json_tokener_parse(string_to_parse);
  262. if (new_obj == NULL) puts("got error as expected");
  263. /* test incremental parsing in various forms */
  264. tok = json_tokener_new();
  265. for (ii = 0; incremental_steps[ii].string_to_parse != NULL; ii++)
  266. {
  267. int this_step_ok = 0;
  268. struct incremental_step *step = &incremental_steps[ii];
  269. int length = step->length;
  270. int expected_char_offset = step->char_offset;
  271. if (step->reset_tokener & 2)
  272. json_tokener_set_flags(tok, JSON_TOKENER_STRICT);
  273. else
  274. json_tokener_set_flags(tok, 0);
  275. if (length == -1)
  276. length = strlen(step->string_to_parse);
  277. if (expected_char_offset == -1)
  278. expected_char_offset = length;
  279. printf("json_tokener_parse_ex(tok, %-12s, %3d) ... ",
  280. step->string_to_parse, length);
  281. new_obj = json_tokener_parse_ex(tok, step->string_to_parse, length);
  282. jerr = json_tokener_get_error(tok);
  283. if (step->expected_error != json_tokener_success)
  284. {
  285. if (new_obj != NULL)
  286. printf("ERROR: invalid object returned: %s\n",
  287. json_object_to_json_string(new_obj));
  288. else if (jerr != step->expected_error)
  289. printf("ERROR: got wrong error: %s\n",
  290. json_tokener_error_desc(jerr));
  291. else if (tok->char_offset != expected_char_offset)
  292. printf("ERROR: wrong char_offset %d != expected %d\n",
  293. tok->char_offset,
  294. expected_char_offset);
  295. else
  296. {
  297. printf("OK: got correct error: %s\n",
  298. json_tokener_error_desc(jerr));
  299. this_step_ok = 1;
  300. }
  301. }
  302. else
  303. {
  304. if (new_obj == NULL &&
  305. !(step->length >= 4 &&
  306. strncmp(step->string_to_parse, "null", 4) == 0))
  307. printf("ERROR: expected valid object, instead: %s\n",
  308. json_tokener_error_desc(jerr));
  309. else if (tok->char_offset != expected_char_offset)
  310. printf("ERROR: wrong char_offset %d != expected %d\n",
  311. tok->char_offset,
  312. expected_char_offset);
  313. else
  314. {
  315. printf("OK: got object of type [%s]: %s\n",
  316. json_type_to_name(json_object_get_type(new_obj)),
  317. json_object_to_json_string(new_obj));
  318. this_step_ok = 1;
  319. }
  320. }
  321. if (new_obj)
  322. json_object_put(new_obj);
  323. if (step->reset_tokener & 1)
  324. json_tokener_reset(tok);
  325. if (this_step_ok)
  326. num_ok++;
  327. else
  328. num_error++;
  329. }
  330. json_tokener_free(tok);
  331. printf("End Incremental Tests OK=%d ERROR=%d\n", num_ok, num_error);
  332. }