xs_json.h 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556
  1. /* copyright (c) 2022 - 2025 grunfink et al. / MIT license */
  2. #ifndef _XS_JSON_H
  3. #define _XS_JSON_H
  4. int xs_json_dump(const xs_val *data, int indent, FILE *f);
  5. xs_str *xs_json_dumps(const xs_val *data, int indent);
  6. xs_val *xs_json_load_full(FILE *f, int maxdepth);
  7. xs_val *xs_json_loads_full(const xs_str *json, int maxdepth);
  8. #define xs_json_load(f) xs_json_load_full(f, MAX_JSON_DEPTH)
  9. #define xs_json_loads(s) xs_json_loads_full(s, MAX_JSON_DEPTH)
  10. xstype xs_json_load_type(FILE *f);
  11. int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c);
  12. int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c);
  13. xs_list *xs_json_load_array(FILE *f, int maxdepth);
  14. xs_dict *xs_json_load_object(FILE *f, int maxdepth);
  15. #ifdef XS_IMPLEMENTATION
  16. /** IMPLEMENTATION **/
  17. /** JSON dumps **/
  18. static void _xs_json_dump_str(const char *data, FILE *f)
  19. /* dumps a string in JSON format */
  20. {
  21. unsigned char c;
  22. fputs("\"", f);
  23. while ((c = *data)) {
  24. if (c == '\n')
  25. fputs("\\n", f);
  26. else
  27. if (c == '\r')
  28. fputs("\\r", f);
  29. else
  30. if (c == '\t')
  31. fputs("\\t", f);
  32. else
  33. if (c == '\\')
  34. fputs("\\\\", f);
  35. else
  36. if (c == '"')
  37. fputs("\\\"", f);
  38. else
  39. if (c < 32)
  40. fprintf(f, "\\u%04x", (unsigned int) c);
  41. else
  42. fputc(c, f);
  43. data++;
  44. }
  45. fputs("\"", f);
  46. }
  47. static void _xs_json_indent(int level, int indent, FILE *f)
  48. /* adds indentation */
  49. {
  50. if (indent) {
  51. int n;
  52. fputc('\n', f);
  53. for (n = 0; n < level * indent; n++)
  54. fputc(' ', f);
  55. }
  56. }
  57. static void _xs_json_dump(const xs_val *data, int level, int indent, FILE *f)
  58. /* dumps partial data as JSON */
  59. {
  60. int c = 0;
  61. const xs_val *v;
  62. switch (xs_type(data)) {
  63. case XSTYPE_NULL:
  64. fputs("null", f);
  65. break;
  66. case XSTYPE_TRUE:
  67. fputs("true", f);
  68. break;
  69. case XSTYPE_FALSE:
  70. fputs("false", f);
  71. break;
  72. case XSTYPE_NUMBER:
  73. fputs(xs_number_str(data), f);
  74. break;
  75. case XSTYPE_LIST:
  76. fputc('[', f);
  77. xs_list_foreach(data, v) {
  78. if (c != 0)
  79. fputc(',', f);
  80. _xs_json_indent(level + 1, indent, f);
  81. _xs_json_dump(v, level + 1, indent, f);
  82. c++;
  83. }
  84. _xs_json_indent(level, indent, f);
  85. fputc(']', f);
  86. break;
  87. case XSTYPE_DICT:
  88. fputc('{', f);
  89. const xs_str *k;
  90. xs_dict_foreach(data, k, v) {
  91. if (c != 0)
  92. fputc(',', f);
  93. _xs_json_indent(level + 1, indent, f);
  94. _xs_json_dump_str(k, f);
  95. fputc(':', f);
  96. if (indent)
  97. fputc(' ', f);
  98. _xs_json_dump(v, level + 1, indent, f);
  99. c++;
  100. }
  101. _xs_json_indent(level, indent, f);
  102. fputc('}', f);
  103. break;
  104. case XSTYPE_STRING:
  105. _xs_json_dump_str(data, f);
  106. break;
  107. default:
  108. break;
  109. }
  110. }
  111. xs_str *xs_json_dumps(const xs_val *data, int indent)
  112. /* dumps data as a JSON string */
  113. {
  114. xs_str *s = NULL;
  115. size_t sz;
  116. FILE *f;
  117. if ((f = open_memstream(&s, &sz)) != NULL) {
  118. int r = xs_json_dump(data, indent, f);
  119. fclose(f);
  120. if (!r)
  121. s = xs_free(s);
  122. }
  123. return s;
  124. }
  125. int xs_json_dump(const xs_val *data, int indent, FILE *f)
  126. /* dumps data into a file as JSON */
  127. {
  128. xstype t = xs_type(data);
  129. if (t == XSTYPE_LIST || t == XSTYPE_DICT) {
  130. _xs_json_dump(data, 0, indent, f);
  131. return 1;
  132. }
  133. return 0;
  134. }
  135. /** JSON loads **/
  136. typedef enum {
  137. JS_ERROR = -1,
  138. JS_INCOMPLETE,
  139. JS_OCURLY,
  140. JS_OBRACK,
  141. JS_CCURLY,
  142. JS_CBRACK,
  143. JS_COMMA,
  144. JS_COLON,
  145. JS_VALUE,
  146. JS_STRING,
  147. JS_NUMBER,
  148. JS_TRUE,
  149. JS_FALSE,
  150. JS_NULL,
  151. JS_ARRAY,
  152. JS_OBJECT
  153. } js_type;
  154. static xs_val *_xs_json_load_lexer(FILE *f, js_type *t)
  155. {
  156. int c;
  157. xs_val *v = NULL;
  158. int offset;
  159. *t = JS_ERROR;
  160. /* skip blanks */
  161. while ((c = fgetc(f)) == ' ' || c == '\t' || c == '\n' || c == '\r');
  162. if (c == '{')
  163. *t = JS_OCURLY;
  164. else
  165. if (c == '}')
  166. *t = JS_CCURLY;
  167. else
  168. if (c == '[')
  169. *t = JS_OBRACK;
  170. else
  171. if (c == ']')
  172. *t = JS_CBRACK;
  173. else
  174. if (c == ',')
  175. *t = JS_COMMA;
  176. else
  177. if (c == ':')
  178. *t = JS_COLON;
  179. else
  180. if (c == '"') {
  181. *t = JS_STRING;
  182. v = xs_str_new(NULL);
  183. offset = 0;
  184. while ((c = fgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
  185. if (c == '\\') {
  186. unsigned int cp = fgetc(f);
  187. switch (cp) {
  188. case 'n': cp = '\n'; break;
  189. case 'r': cp = '\r'; break;
  190. case 't': cp = '\t'; break;
  191. case 'u': /* Unicode codepoint as an hex char */
  192. if (fscanf(f, "%04x", &cp) != 1) {
  193. *t = JS_ERROR;
  194. break;
  195. }
  196. if (xs_is_surrogate(cp)) {
  197. /* \u must follow */
  198. if (fgetc(f) != '\\' || fgetc(f) != 'u') {
  199. *t = JS_ERROR;
  200. break;
  201. }
  202. unsigned int p2;
  203. if (fscanf(f, "%04x", &p2) != 1) {
  204. *t = JS_ERROR;
  205. break;
  206. }
  207. cp = xs_surrogate_dec(cp, p2);
  208. }
  209. /* replace dangerous control codes with their visual representations */
  210. if (cp < ' ' && !strchr("\r\n\t", cp))
  211. cp += 0x2400;
  212. break;
  213. }
  214. v = xs_utf8_insert(v, cp, &offset);
  215. }
  216. else {
  217. char cc = c;
  218. v = xs_insert_m(v, offset, &cc, 1);
  219. if (!xs_is_string(v)) {
  220. *t = JS_ERROR;
  221. break;
  222. }
  223. offset++;
  224. }
  225. }
  226. if (c == EOF)
  227. *t = JS_ERROR;
  228. }
  229. else
  230. if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
  231. double d;
  232. ungetc(c, f);
  233. if (fscanf(f, "%lf", &d) == 1) {
  234. *t = JS_NUMBER;
  235. v = xs_number_new(d);
  236. }
  237. }
  238. else
  239. if (c == 't') {
  240. if (fgetc(f) == 'r' && fgetc(f) == 'u' && fgetc(f) == 'e') {
  241. *t = JS_TRUE;
  242. v = xs_val_new(XSTYPE_TRUE);
  243. }
  244. }
  245. else
  246. if (c == 'f') {
  247. if (fgetc(f) == 'a' && fgetc(f) == 'l' &&
  248. fgetc(f) == 's' && fgetc(f) == 'e') {
  249. *t = JS_FALSE;
  250. v = xs_val_new(XSTYPE_FALSE);
  251. }
  252. }
  253. else
  254. if (c == 'n') {
  255. if (fgetc(f) == 'u' && fgetc(f) == 'l' && fgetc(f) == 'l') {
  256. *t = JS_NULL;
  257. v = xs_val_new(XSTYPE_NULL);
  258. }
  259. }
  260. if (*t == JS_ERROR)
  261. v = xs_free(v);
  262. return v;
  263. }
  264. int xs_json_load_array_iter(FILE *f, xs_val **value, xstype *pt, int *c)
  265. /* loads the next scalar value from the JSON stream */
  266. /* if the value ahead is compound, value is NULL and pt is set */
  267. {
  268. js_type t;
  269. *value = _xs_json_load_lexer(f, &t);
  270. if (t == JS_ERROR)
  271. return -1;
  272. if (t == JS_CBRACK)
  273. return 0;
  274. if (*c > 0) {
  275. if (t == JS_COMMA)
  276. *value = _xs_json_load_lexer(f, &t);
  277. else
  278. return -1;
  279. }
  280. if (*value == NULL) {
  281. /* possible compound type ahead */
  282. if (t == JS_OBRACK)
  283. *pt = XSTYPE_LIST;
  284. else
  285. if (t == JS_OCURLY)
  286. *pt = XSTYPE_DICT;
  287. else
  288. return -1;
  289. }
  290. *c = *c + 1;
  291. return 1;
  292. }
  293. xs_list *xs_json_load_array(FILE *f, int maxdepth)
  294. /* loads a full JSON array (after the initial OBRACK) */
  295. {
  296. xstype t;
  297. xs_list *l = xs_list_new();
  298. int c = 0;
  299. for (;;) {
  300. xs *v = NULL;
  301. int r = xs_json_load_array_iter(f, &v, &t, &c);
  302. if (r == -1)
  303. l = xs_free(l);
  304. if (r == 1) {
  305. /* partial load? */
  306. if (v == NULL && maxdepth != 0) {
  307. if (t == XSTYPE_LIST)
  308. v = xs_json_load_array(f, maxdepth - 1);
  309. else
  310. if (t == XSTYPE_DICT)
  311. v = xs_json_load_object(f, maxdepth - 1);
  312. }
  313. /* still null? fail */
  314. if (v == NULL) {
  315. l = xs_free(l);
  316. break;
  317. }
  318. l = xs_list_append(l, v);
  319. }
  320. else
  321. break;
  322. }
  323. return l;
  324. }
  325. int xs_json_load_object_iter(FILE *f, xs_str **key, xs_val **value, xstype *pt, int *c)
  326. /* loads the next key and scalar value from the JSON stream */
  327. /* if the value ahead is compound, value is NULL and pt is set */
  328. {
  329. js_type t;
  330. *key = _xs_json_load_lexer(f, &t);
  331. if (t == JS_ERROR)
  332. return -1;
  333. if (t == JS_CCURLY)
  334. return 0;
  335. if (*c > 0) {
  336. if (t == JS_COMMA)
  337. *key = _xs_json_load_lexer(f, &t);
  338. else
  339. return -1;
  340. }
  341. if (t != JS_STRING)
  342. return -1;
  343. xs_free(_xs_json_load_lexer(f, &t));
  344. if (t != JS_COLON)
  345. return -1;
  346. *value = _xs_json_load_lexer(f, &t);
  347. if (*value == NULL) {
  348. /* possible complex type ahead */
  349. if (t == JS_OBRACK)
  350. *pt = XSTYPE_LIST;
  351. else
  352. if (t == JS_OCURLY)
  353. *pt = XSTYPE_DICT;
  354. else
  355. return -1;
  356. }
  357. *c = *c + 1;
  358. return 1;
  359. }
  360. xs_dict *xs_json_load_object(FILE *f, int maxdepth)
  361. /* loads a full JSON object (after the initial OCURLY) */
  362. {
  363. xstype t;
  364. xs_dict *d = xs_dict_new();
  365. int c = 0;
  366. for (;;) {
  367. xs *k = NULL;
  368. xs *v = NULL;
  369. int r = xs_json_load_object_iter(f, &k, &v, &t, &c);
  370. if (r == -1)
  371. d = xs_free(d);
  372. if (r == 1) {
  373. /* partial load? */
  374. if (v == NULL && maxdepth != 0) {
  375. if (t == XSTYPE_LIST)
  376. v = xs_json_load_array(f, maxdepth - 1);
  377. else
  378. if (t == XSTYPE_DICT)
  379. v = xs_json_load_object(f, maxdepth - 1);
  380. }
  381. /* still null? fail */
  382. if (v == NULL) {
  383. d = xs_free(d);
  384. break;
  385. }
  386. d = xs_dict_append(d, k, v);
  387. }
  388. else
  389. break;
  390. }
  391. return d;
  392. }
  393. xs_val *xs_json_loads_full(const xs_str *json, int maxdepth)
  394. /* loads a string in JSON format and converts to a multiple data */
  395. {
  396. FILE *f;
  397. xs_val *v = NULL;
  398. if ((f = fmemopen((char *)json, strlen(json), "r")) != NULL) {
  399. v = xs_json_load_full(f, maxdepth);
  400. fclose(f);
  401. }
  402. return v;
  403. }
  404. xstype xs_json_load_type(FILE *f)
  405. /* identifies the type of a JSON stream */
  406. {
  407. xstype t = XSTYPE_NULL;
  408. js_type jt;
  409. xs_free(_xs_json_load_lexer(f, &jt));
  410. if (jt == JS_OBRACK)
  411. t = XSTYPE_LIST;
  412. else
  413. if (jt == JS_OCURLY)
  414. t = XSTYPE_DICT;
  415. return t;
  416. }
  417. xs_val *xs_json_load_full(FILE *f, int maxdepth)
  418. /* loads a JSON file */
  419. {
  420. xs_val *v = NULL;
  421. xstype t = xs_json_load_type(f);
  422. if (t == XSTYPE_LIST)
  423. v = xs_json_load_array(f, maxdepth);
  424. else
  425. if (t == XSTYPE_DICT)
  426. v = xs_json_load_object(f, maxdepth);
  427. return v;
  428. }
  429. #endif /* XS_IMPLEMENTATION */
  430. #endif /* _XS_JSON_H */