xs_json.h 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511
  1. /* copyright (c) 2022 - 2023 grunfink / MIT license */
  2. #ifndef _XS_JSON_H
  3. #define _XS_JSON_H
  4. xs_str *xs_json_dumps_pp(const xs_val *data, int indent);
  5. #define xs_json_dumps(data) xs_json_dumps_pp(data, 0)
  6. xs_val *xs_json_loads(const xs_str *json);
  7. #ifdef XS_IMPLEMENTATION
  8. /** IMPLEMENTATION **/
  9. /** JSON dumps **/
  10. static xs_str *_xs_json_dumps_str(xs_str *s, const char *data)
  11. /* dumps a string in JSON format */
  12. {
  13. unsigned char c;
  14. s = xs_str_cat(s, "\"");
  15. while ((c = *data)) {
  16. if (c == '\n')
  17. s = xs_str_cat(s, "\\n");
  18. else
  19. if (c == '\r')
  20. s = xs_str_cat(s, "\\r");
  21. else
  22. if (c == '\t')
  23. s = xs_str_cat(s, "\\t");
  24. else
  25. if (c == '\\')
  26. s = xs_str_cat(s, "\\\\");
  27. else
  28. if (c == '"')
  29. s = xs_str_cat(s, "\\\"");
  30. else
  31. if (c < 32) {
  32. char tmp[10];
  33. snprintf(tmp, sizeof(tmp), "\\u%04x", (unsigned int) c);
  34. s = xs_str_cat(s, tmp);
  35. }
  36. else
  37. s = xs_append_m(s, data, 1);
  38. data++;
  39. }
  40. s = xs_str_cat(s, "\"");
  41. return s;
  42. }
  43. static xs_str *_xs_json_indent(xs_str *s, int level, int indent)
  44. /* adds indentation */
  45. {
  46. if (indent) {
  47. int n;
  48. s = xs_str_cat(s, "\n");
  49. for (n = 0; n < level * indent; n++)
  50. s = xs_str_cat(s, " ");
  51. }
  52. return s;
  53. }
  54. static xs_str *_xs_json_dumps(xs_str *s, const xs_val *s_data, int level, int indent)
  55. /* dumps partial data as JSON */
  56. {
  57. int c = 0;
  58. xs_val *v;
  59. xs_val *data = (xs_val *)s_data;
  60. switch (xs_type(data)) {
  61. case XSTYPE_NULL:
  62. s = xs_str_cat(s, "null");
  63. break;
  64. case XSTYPE_TRUE:
  65. s = xs_str_cat(s, "true");
  66. break;
  67. case XSTYPE_FALSE:
  68. s = xs_str_cat(s, "false");
  69. break;
  70. case XSTYPE_NUMBER:
  71. s = xs_str_cat(s, xs_number_str(data));
  72. break;
  73. case XSTYPE_LIST:
  74. s = xs_str_cat(s, "[");
  75. while (xs_list_iter(&data, &v)) {
  76. if (c != 0)
  77. s = xs_str_cat(s, ",");
  78. s = _xs_json_indent(s, level + 1, indent);
  79. s = _xs_json_dumps(s, v, level + 1, indent);
  80. c++;
  81. }
  82. s = _xs_json_indent(s, level, indent);
  83. s = xs_str_cat(s, "]");
  84. break;
  85. case XSTYPE_DICT:
  86. s = xs_str_cat(s, "{");
  87. xs_str *k;
  88. while (xs_dict_iter(&data, &k, &v)) {
  89. if (c != 0)
  90. s = xs_str_cat(s, ",");
  91. s = _xs_json_indent(s, level + 1, indent);
  92. s = _xs_json_dumps_str(s, k);
  93. s = xs_str_cat(s, ":");
  94. if (indent)
  95. s = xs_str_cat(s, " ");
  96. s = _xs_json_dumps(s, v, level + 1, indent);
  97. c++;
  98. }
  99. s = _xs_json_indent(s, level, indent);
  100. s = xs_str_cat(s, "}");
  101. break;
  102. case XSTYPE_STRING:
  103. s = _xs_json_dumps_str(s, data);
  104. break;
  105. default:
  106. break;
  107. }
  108. return s;
  109. }
  110. xs_str *xs_json_dumps_pp(const xs_val *data, int indent)
  111. /* dumps a piece of data as JSON */
  112. {
  113. xstype t = xs_type(data);
  114. xs_str *s = NULL;
  115. if (t == XSTYPE_LIST || t == XSTYPE_DICT) {
  116. s = xs_str_new(NULL);
  117. s = _xs_json_dumps(s, data, 0, indent);
  118. }
  119. return s;
  120. }
  121. /** JSON loads **/
  122. /* this code comes mostly from the Minimum Profit Text Editor (MPDM) */
  123. typedef enum {
  124. JS_ERROR = -1,
  125. JS_INCOMPLETE,
  126. JS_OCURLY,
  127. JS_OBRACK,
  128. JS_CCURLY,
  129. JS_CBRACK,
  130. JS_COMMA,
  131. JS_COLON,
  132. JS_VALUE,
  133. JS_STRING,
  134. JS_INTEGER,
  135. JS_REAL,
  136. JS_TRUE,
  137. JS_FALSE,
  138. JS_NULL,
  139. JS_ARRAY,
  140. JS_OBJECT
  141. } js_type;
  142. static xs_val *_xs_json_loads_lexer(const char **json, js_type *t)
  143. {
  144. char c;
  145. const char *s = *json;
  146. xs_val *v = NULL;
  147. /* skip blanks */
  148. while (*s == L' ' || *s == L'\t' || *s == L'\n' || *s == L'\r')
  149. s++;
  150. c = *s++;
  151. if (c == '{')
  152. *t = JS_OCURLY;
  153. else
  154. if (c == '}')
  155. *t = JS_CCURLY;
  156. else
  157. if (c == '[')
  158. *t = JS_OBRACK;
  159. else
  160. if (c == ']')
  161. *t = JS_CBRACK;
  162. else
  163. if (c == ',')
  164. *t = JS_COMMA;
  165. else
  166. if (c == ':')
  167. *t = JS_COLON;
  168. else
  169. if (c == '"') {
  170. *t = JS_STRING;
  171. v = xs_str_new(NULL);
  172. while ((c = *s) != '"' && c != '\0') {
  173. char tmp[5];
  174. int cp, i;
  175. if (c == '\\') {
  176. s++;
  177. c = *s;
  178. switch (c) {
  179. case 'n': c = '\n'; break;
  180. case 'r': c = '\r'; break;
  181. case 't': c = '\t'; break;
  182. case 'u': /* Unicode codepoint as an hex char */
  183. s++;
  184. memcpy(tmp, s, 4);
  185. s += 3;
  186. tmp[4] = '\0';
  187. sscanf(tmp, "%04x", &i);
  188. if (i >= 0xd800 && i <= 0xdfff) {
  189. /* it's a surrogate pair */
  190. cp = (i & 0x3ff) << 10;
  191. /* skip to the next value */
  192. s += 3;
  193. memcpy(tmp, s, 4);
  194. s += 3;
  195. sscanf(tmp, "%04x", &i);
  196. cp |= (i & 0x3ff);
  197. cp += 0x10000;
  198. }
  199. else
  200. cp = i;
  201. /* replace dangerous control codes with their visual representations */
  202. if (cp >= '\0' && cp < ' ' && !strchr("\r\n\t", cp))
  203. cp += 0x2400;
  204. v = xs_utf8_enc(v, cp);
  205. c = '\0';
  206. break;
  207. }
  208. }
  209. if (c)
  210. v = xs_append_m(v, &c, 1);
  211. s++;
  212. }
  213. if (c != '\0')
  214. s++;
  215. }
  216. else
  217. if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
  218. xs *vn = NULL;
  219. *t = JS_INTEGER;
  220. vn = xs_str_new(NULL);
  221. vn = xs_append_m(vn, &c, 1);
  222. while (((c = *s) >= '0' && c <= '9') || c == '.') {
  223. if (c == '.')
  224. *t = JS_REAL;
  225. vn = xs_append_m(vn, &c, 1);
  226. s++;
  227. }
  228. /* convert to XSTYPE_NUMBER */
  229. v = xs_number_new(atof(vn));
  230. }
  231. else
  232. if (c == 't' && strncmp(s, "rue", 3) == 0) {
  233. s += 3;
  234. *t = JS_TRUE;
  235. v = xs_val_new(XSTYPE_TRUE);
  236. }
  237. else
  238. if (c == 'f' && strncmp(s, "alse", 4) == 0) {
  239. s += 4;
  240. *t = JS_FALSE;
  241. v = xs_val_new(XSTYPE_FALSE);
  242. }
  243. else
  244. if (c == 'n' && strncmp(s, "ull", 3) == 0) {
  245. s += 3;
  246. *t = JS_NULL;
  247. v = xs_val_new(XSTYPE_NULL);
  248. }
  249. else
  250. *t = JS_ERROR;
  251. *json = s;
  252. return v;
  253. }
  254. static xs_list *_xs_json_loads_array(const char **json, js_type *t);
  255. static xs_dict *_xs_json_loads_object(const char **json, js_type *t);
  256. static xs_val *_xs_json_loads_value(const char **json, js_type *t, xs_val *v)
  257. /* parses a JSON value */
  258. {
  259. if (*t == JS_OBRACK)
  260. v = _xs_json_loads_array(json, t);
  261. else
  262. if (*t == JS_OCURLY)
  263. v = _xs_json_loads_object(json, t);
  264. if (*t >= JS_VALUE)
  265. *t = JS_VALUE;
  266. else
  267. *t = JS_ERROR;
  268. return v;
  269. }
  270. static xs_list *_xs_json_loads_array(const char **json, js_type *t)
  271. /* parses a JSON array */
  272. {
  273. const char *s = *json;
  274. xs *v;
  275. xs_list *l;
  276. js_type tt;
  277. l = xs_list_new();
  278. *t = JS_INCOMPLETE;
  279. v = _xs_json_loads_lexer(&s, &tt);
  280. if (tt == JS_CBRACK)
  281. *t = JS_ARRAY;
  282. else {
  283. v = _xs_json_loads_value(&s, &tt, v);
  284. if (tt == JS_VALUE) {
  285. l = xs_list_append(l, v);
  286. while (*t == JS_INCOMPLETE) {
  287. xs_free(_xs_json_loads_lexer(&s, &tt));
  288. if (tt == JS_CBRACK)
  289. *t = JS_ARRAY;
  290. else
  291. if (tt == JS_COMMA) {
  292. xs *v2;
  293. v2 = _xs_json_loads_lexer(&s, &tt);
  294. v2 = _xs_json_loads_value(&s, &tt, v2);
  295. if (tt == JS_VALUE)
  296. l = xs_list_append(l, v2);
  297. else
  298. *t = JS_ERROR;
  299. }
  300. else
  301. *t = JS_ERROR;
  302. }
  303. }
  304. else
  305. *t = JS_ERROR;
  306. }
  307. if (*t == JS_ERROR)
  308. l = xs_free(l);
  309. *json = s;
  310. return l;
  311. }
  312. static xs_dict *_xs_json_loads_object(const char **json, js_type *t)
  313. /* parses a JSON object */
  314. {
  315. const char *s = *json;
  316. xs *k1;
  317. xs_dict *d;
  318. js_type tt;
  319. d = xs_dict_new();
  320. *t = JS_INCOMPLETE;
  321. k1 = _xs_json_loads_lexer(&s, &tt);
  322. if (tt == JS_CCURLY)
  323. *t = JS_OBJECT;
  324. else
  325. if (tt == JS_STRING) {
  326. xs_free(_xs_json_loads_lexer(&s, &tt));
  327. if (tt == JS_COLON) {
  328. xs *v1;
  329. v1 = _xs_json_loads_lexer(&s, &tt);
  330. v1 = _xs_json_loads_value(&s, &tt, v1);
  331. if (tt == JS_VALUE) {
  332. d = xs_dict_append(d, k1, v1);
  333. while (*t == JS_INCOMPLETE) {
  334. xs_free(_xs_json_loads_lexer(&s, &tt));
  335. if (tt == JS_CCURLY)
  336. *t = JS_OBJECT;
  337. else
  338. if (tt == JS_COMMA) {
  339. xs *k = _xs_json_loads_lexer(&s, &tt);
  340. if (tt == JS_STRING) {
  341. xs_free(_xs_json_loads_lexer(&s, &tt));
  342. if (tt == JS_COLON) {
  343. xs *v;
  344. v = _xs_json_loads_lexer(&s, &tt);
  345. v = _xs_json_loads_value(&s, &tt, v);
  346. if (tt == JS_VALUE)
  347. d = xs_dict_append(d, k, v);
  348. else
  349. *t = JS_ERROR;
  350. }
  351. else
  352. *t = JS_ERROR;
  353. }
  354. else
  355. *t = JS_ERROR;
  356. }
  357. else
  358. *t = JS_ERROR;
  359. }
  360. }
  361. else
  362. *t = JS_ERROR;
  363. }
  364. else
  365. *t = JS_ERROR;
  366. }
  367. else
  368. *t = JS_ERROR;
  369. if (*t == JS_ERROR)
  370. d = xs_free(d);
  371. *json = s;
  372. return d;
  373. }
  374. xs_val *xs_json_loads(const xs_str *json)
  375. /* loads a string in JSON format and converts to a multiple data */
  376. {
  377. xs_val *v = NULL;
  378. js_type t;
  379. xs_free(_xs_json_loads_lexer(&json, &t));
  380. if (t == JS_OBRACK)
  381. v = _xs_json_loads_array(&json, &t);
  382. else
  383. if (t == JS_OCURLY)
  384. v = _xs_json_loads_object(&json, &t);
  385. else
  386. t = JS_ERROR;
  387. return v;
  388. }
  389. #endif /* XS_IMPLEMENTATION */
  390. #endif /* _XS_JSON_H */