xs_json.h 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658
  1. /* copyright (c) 2022 - 2026 grunfink et al. / MIT license */
  2. #ifndef _XS_JSON_H
  3. #define _XS_JSON_H
  4. #ifndef MAX_JSON_DEPTH
  5. #define MAX_JSON_DEPTH 32
  6. #endif
  7. typedef struct {
  8. union { FILE *f; const char *b; } src;
  9. const char *b_end;
  10. } json_src;
  11. void xs_json_dump_value(const xs_val *data, int level, int indent, FILE *f);
  12. int xs_json_dump(const xs_val *data, int indent, FILE *f);
  13. xs_str *xs_json_dumps(const xs_val *data, int indent);
  14. xs_val *xs_json_load_full(FILE *f, int maxdepth);
  15. xs_val *xs_json_loads_full(const xs_str *json, int maxdepth);
  16. #define xs_json_load(f) xs_json_load_full(f, MAX_JSON_DEPTH)
  17. #define xs_json_loads(s) xs_json_loads_full(s, MAX_JSON_DEPTH)
  18. xstype xs_json_load_type(json_src *f);
  19. int xs_json_load_array_iter(json_src *f, xs_val **value, xstype *pt, int *c);
  20. int xs_json_load_object_iter(json_src *f, xs_str **key, xs_val **value, xstype *pt, int *c);
  21. int xs_json_load_array(json_src *f, int maxdepth, xs_list **l);
  22. int xs_json_load_object(json_src *f, int maxdepth, xs_dict **d);
  23. #ifdef XS_IMPLEMENTATION
  24. /** IMPLEMENTATION **/
  25. #include "xs_unicode.h"
  26. /** JSON dumps **/
  27. static void _xs_json_dump_str(const char *data, FILE *f)
  28. /* dumps a string in JSON format */
  29. {
  30. unsigned char c;
  31. fputs("\"", f);
  32. while ((c = *data)) {
  33. if (c == '\n')
  34. fputs("\\n", f);
  35. else
  36. if (c == '\r')
  37. fputs("\\r", f);
  38. else
  39. if (c == '\t')
  40. fputs("\\t", f);
  41. else
  42. if (c == '\\')
  43. fputs("\\\\", f);
  44. else
  45. if (c == '"')
  46. fputs("\\\"", f);
  47. else
  48. if (c < 32)
  49. fprintf(f, "\\u%04x", (unsigned int) c);
  50. else
  51. fputc(c, f);
  52. data++;
  53. }
  54. fputs("\"", f);
  55. }
  56. static void _xs_json_indent(int level, int indent, FILE *f)
  57. /* adds indentation */
  58. {
  59. if (indent) {
  60. int n;
  61. fputc('\n', f);
  62. for (n = 0; n < level * indent; n++)
  63. fputc(' ', f);
  64. }
  65. }
  66. void xs_json_dump_value(const xs_val *data, int level, int indent, FILE *f)
  67. /* dumps partial data as JSON */
  68. {
  69. int c = 0;
  70. const xs_val *v;
  71. switch (xs_type(data)) {
  72. case XSTYPE_NULL:
  73. fputs("null", f);
  74. break;
  75. case XSTYPE_TRUE:
  76. fputs("true", f);
  77. break;
  78. case XSTYPE_FALSE:
  79. fputs("false", f);
  80. break;
  81. case XSTYPE_NUMBER:
  82. fputs(xs_number_str(data), f);
  83. break;
  84. case XSTYPE_LIST:
  85. fputc('[', f);
  86. xs_list_foreach(data, v) {
  87. if (c != 0)
  88. fputc(',', f);
  89. _xs_json_indent(level + 1, indent, f);
  90. xs_json_dump_value(v, level + 1, indent, f);
  91. c++;
  92. }
  93. _xs_json_indent(level, indent, f);
  94. fputc(']', f);
  95. break;
  96. case XSTYPE_DICT:
  97. fputc('{', f);
  98. const xs_str *k;
  99. xs_dict_foreach(data, k, v) {
  100. if (c != 0)
  101. fputc(',', f);
  102. _xs_json_indent(level + 1, indent, f);
  103. _xs_json_dump_str(k, f);
  104. fputc(':', f);
  105. if (indent)
  106. fputc(' ', f);
  107. xs_json_dump_value(v, level + 1, indent, f);
  108. c++;
  109. }
  110. _xs_json_indent(level, indent, f);
  111. fputc('}', f);
  112. break;
  113. case XSTYPE_STRING:
  114. _xs_json_dump_str(data, f);
  115. break;
  116. default:
  117. break;
  118. }
  119. }
  120. int xs_json_dump(const xs_val *data, int indent, FILE *f)
  121. /* dumps data into a file as JSON */
  122. {
  123. xstype t = xs_type(data);
  124. if (t == XSTYPE_LIST || t == XSTYPE_DICT) {
  125. xs_json_dump_value(data, 0, indent, f);
  126. return 1;
  127. }
  128. return 0;
  129. }
  130. xs_str *xs_json_dumps(const xs_val *data, int indent)
  131. /* dumps data as a JSON string */
  132. {
  133. xs_str *s = NULL;
  134. size_t sz;
  135. FILE *f;
  136. if ((f = open_memstream(&s, &sz)) != NULL) {
  137. int r = xs_json_dump(data, indent, f);
  138. fclose(f);
  139. if (!r)
  140. s = xs_free(s);
  141. }
  142. return s;
  143. }
  144. /** JSON loads **/
  145. typedef enum {
  146. JS_ERROR = -1,
  147. JS_INCOMPLETE,
  148. JS_OCURLY,
  149. JS_OBRACK,
  150. JS_CCURLY,
  151. JS_CBRACK,
  152. JS_COMMA,
  153. JS_COLON,
  154. JS_VALUE,
  155. JS_STRING,
  156. JS_NUMBER,
  157. JS_TRUE,
  158. JS_FALSE,
  159. JS_NULL,
  160. JS_ARRAY,
  161. JS_OBJECT
  162. } js_type;
  163. static int jgetc(json_src *f)
  164. /* calls fgetc on an underlying file, otherwise reads the buffer */
  165. {
  166. if (f->b_end == NULL)
  167. return fgetc(f->src.f);
  168. else
  169. if (f->src.b >= f->b_end)
  170. return EOF;
  171. else
  172. return *(f->src.b)++;
  173. }
  174. static int jungetc(int c, json_src *f)
  175. /* calls ungetc on an underlying file, otherwise decrements the buf ptr */
  176. /* this function will NOT mutate an underlying buffer */
  177. {
  178. if (f->b_end == NULL)
  179. return ungetc(c, f->src.f);
  180. else
  181. if (c == EOF)
  182. return EOF;
  183. else {
  184. f->src.b--;
  185. return c;
  186. }
  187. }
  188. #define jscanf(f, fmt, arg) _jscanf(f, fmt, fmt "%n", arg)
  189. static int _jscanf(json_src *restrict f, const char *restrict fmt, const char *restrict fmt2, void *arg)
  190. /* calls fscanf on an underlying file, otherwise reads the buffer. */
  191. {
  192. if (f->b_end == NULL)
  193. return fscanf(f->src.f, fmt, arg);
  194. else
  195. if (f->src.b >= f->b_end)
  196. return EOF;
  197. else {
  198. int n, b;
  199. n = sscanf(f->src.b, fmt2, arg, &b);
  200. if (n >= 0)
  201. f->src.b += b;
  202. return n;
  203. }
  204. }
  205. static xs_val *_xs_json_load_lexer(json_src *f, js_type *t)
  206. {
  207. int c;
  208. xs_val *v = NULL;
  209. int offset;
  210. *t = JS_ERROR;
  211. /* skip blanks */
  212. while ((c = jgetc(f)) == ' ' || c == '\t' || c == '\n' || c == '\r');
  213. if (c == '{')
  214. *t = JS_OCURLY;
  215. else
  216. if (c == '}')
  217. *t = JS_CCURLY;
  218. else
  219. if (c == '[')
  220. *t = JS_OBRACK;
  221. else
  222. if (c == ']')
  223. *t = JS_CBRACK;
  224. else
  225. if (c == ',')
  226. *t = JS_COMMA;
  227. else
  228. if (c == ':')
  229. *t = JS_COLON;
  230. else
  231. if (c == '"') {
  232. *t = JS_STRING;
  233. v = xs_str_new(NULL);
  234. offset = 0;
  235. while ((c = jgetc(f)) != '"' && c != EOF && *t != JS_ERROR) {
  236. if (c == '\\') {
  237. unsigned int cp = jgetc(f);
  238. switch (cp) {
  239. case 'n': cp = '\n'; break;
  240. case 'r': cp = '\r'; break;
  241. case 't': cp = '\t'; break;
  242. case '"': cp = '"'; break;
  243. case '\\': cp = '\\'; break;
  244. case '/': cp = '/'; break;
  245. case 'u': /* Unicode codepoint as an hex char */
  246. if (jscanf(f, "%04x", &cp) != 1) {
  247. *t = JS_ERROR;
  248. break;
  249. }
  250. if (xs_is_surrogate(cp)) {
  251. /* \u must follow */
  252. if (jgetc(f) != '\\' || jgetc(f) != 'u') {
  253. *t = JS_ERROR;
  254. break;
  255. }
  256. unsigned int p2;
  257. if (jscanf(f, "%04x", &p2) != 1) {
  258. *t = JS_ERROR;
  259. break;
  260. }
  261. cp = xs_surrogate_dec(cp, p2);
  262. }
  263. /* replace dangerous control codes with their visual representations */
  264. if (cp == 0 || (cp < ' ' && !strchr("\r\n\t", cp)))
  265. cp += 0x2400;
  266. break;
  267. default:
  268. *t = JS_ERROR;
  269. break;
  270. }
  271. v = xs_utf8_insert(v, cp, &offset);
  272. }
  273. else {
  274. if (c >= 0 && c < ' ') {
  275. v = xs_utf8_insert(v, c + 0x2400, &offset);
  276. }
  277. else {
  278. char cc = c;
  279. v = xs_insert_m(v, offset, &cc, 1);
  280. offset++;
  281. }
  282. if (!xs_is_string(v)) {
  283. *t = JS_ERROR;
  284. break;
  285. }
  286. }
  287. }
  288. if (c == EOF)
  289. *t = JS_ERROR;
  290. }
  291. else
  292. if (c == '-' || (c >= '0' && c <= '9') || c == '.') {
  293. double d;
  294. jungetc(c, f);
  295. if (jscanf(f, "%lf", &d) == 1) {
  296. *t = JS_NUMBER;
  297. v = xs_number_new(d);
  298. }
  299. }
  300. else
  301. if (c == 't') {
  302. if (jgetc(f) == 'r' && jgetc(f) == 'u' && jgetc(f) == 'e') {
  303. *t = JS_TRUE;
  304. v = xs_val_new(XSTYPE_TRUE);
  305. }
  306. }
  307. else
  308. if (c == 'f') {
  309. if (jgetc(f) == 'a' && jgetc(f) == 'l' &&
  310. jgetc(f) == 's' && jgetc(f) == 'e') {
  311. *t = JS_FALSE;
  312. v = xs_val_new(XSTYPE_FALSE);
  313. }
  314. }
  315. else
  316. if (c == 'n') {
  317. if (jgetc(f) == 'u' && jgetc(f) == 'l' && jgetc(f) == 'l') {
  318. *t = JS_NULL;
  319. v = xs_val_new(XSTYPE_NULL);
  320. }
  321. }
  322. if (*t == JS_ERROR)
  323. v = xs_free(v);
  324. return v;
  325. }
  326. int xs_json_load_array_iter(json_src *f, xs_val **value, xstype *pt, int *c)
  327. /* loads the next scalar value from the JSON stream */
  328. /* if the value ahead is compound, value is NULL and pt is set */
  329. {
  330. js_type t;
  331. *value = _xs_json_load_lexer(f, &t);
  332. if (t == JS_ERROR)
  333. return -1;
  334. if (t == JS_CBRACK)
  335. return 0;
  336. if (*c > 0) {
  337. if (t == JS_COMMA)
  338. *value = _xs_json_load_lexer(f, &t);
  339. else
  340. return -1;
  341. }
  342. if (*value == NULL) {
  343. /* possible compound type ahead */
  344. if (t == JS_OBRACK)
  345. *pt = XSTYPE_LIST;
  346. else
  347. if (t == JS_OCURLY)
  348. *pt = XSTYPE_DICT;
  349. else
  350. return -1;
  351. }
  352. else
  353. *pt = xs_type(*value);
  354. *c = *c + 1;
  355. return 1;
  356. }
  357. int xs_json_load_array(json_src *f, int maxdepth, xs_list **l)
  358. /* loads a full JSON array (after the initial OBRACK) */
  359. /* l can be NULL for the content to be dropped */
  360. {
  361. xstype t;
  362. int r = 0;
  363. int c = 0;
  364. for (;;) {
  365. xs *v = NULL;
  366. r = xs_json_load_array_iter(f, &v, &t, &c);
  367. if (r == 1) {
  368. /* partial load? */
  369. if (v == NULL && maxdepth != 0) {
  370. if (t == XSTYPE_LIST) {
  371. if (l)
  372. v = xs_list_new();
  373. r = xs_json_load_array(f, maxdepth - 1, &v);
  374. }
  375. else
  376. if (t == XSTYPE_DICT) {
  377. if (l)
  378. v = xs_dict_new();
  379. r = xs_json_load_object(f, maxdepth - 1, &v);
  380. }
  381. }
  382. /* error? */
  383. if (r < 0)
  384. break;
  385. if (l)
  386. *l = xs_list_append(*l, v);
  387. }
  388. else
  389. break;
  390. }
  391. if (r < 0 && l)
  392. *l = xs_free(*l);
  393. return r;
  394. }
  395. int xs_json_load_object_iter(json_src *f, xs_str **key, xs_val **value, xstype *pt, int *c)
  396. /* loads the next key and scalar value from the JSON stream */
  397. /* if the value ahead is compound, value is NULL and pt is set */
  398. {
  399. js_type t;
  400. *key = _xs_json_load_lexer(f, &t);
  401. if (t == JS_ERROR)
  402. return -1;
  403. if (t == JS_CCURLY)
  404. return 0;
  405. if (*c > 0) {
  406. if (t == JS_COMMA)
  407. *key = _xs_json_load_lexer(f, &t);
  408. else
  409. return -1;
  410. }
  411. if (t != JS_STRING)
  412. return -1;
  413. xs_free(_xs_json_load_lexer(f, &t));
  414. if (t != JS_COLON)
  415. return -1;
  416. *value = _xs_json_load_lexer(f, &t);
  417. if (*value == NULL) {
  418. /* possible complex type ahead */
  419. if (t == JS_OBRACK)
  420. *pt = XSTYPE_LIST;
  421. else
  422. if (t == JS_OCURLY)
  423. *pt = XSTYPE_DICT;
  424. else
  425. return -1;
  426. }
  427. else
  428. *pt = xs_type(*value);
  429. *c = *c + 1;
  430. return 1;
  431. }
  432. int xs_json_load_object(json_src *f, int maxdepth, xs_dict **d)
  433. /* loads a full JSON object (after the initial OCURLY) */
  434. /* d can be NULL for the content to be dropped */
  435. {
  436. xstype t;
  437. int r = 0;
  438. int c = 0;
  439. for (;;) {
  440. xs *k = NULL;
  441. xs *v = NULL;
  442. r = xs_json_load_object_iter(f, &k, &v, &t, &c);
  443. if (r == 1) {
  444. /* partial load? */
  445. if (v == NULL && maxdepth != 0) {
  446. if (t == XSTYPE_LIST) {
  447. if (d)
  448. v = xs_list_new();
  449. r = xs_json_load_array(f, maxdepth - 1, &v);
  450. }
  451. else
  452. if (t == XSTYPE_DICT) {
  453. if (d)
  454. v = xs_dict_new();
  455. r = xs_json_load_object(f, maxdepth - 1, &v);
  456. }
  457. }
  458. /* error? */
  459. if (r < 0)
  460. break;
  461. if (d)
  462. *d = xs_dict_append(*d, k, v);
  463. }
  464. else
  465. break;
  466. }
  467. if (r < 0 && d)
  468. *d = xs_free(*d);
  469. return r;
  470. }
  471. xstype xs_json_load_type(json_src *f)
  472. /* identifies the type of a JSON stream */
  473. {
  474. xstype t = XSTYPE_NULL;
  475. js_type jt;
  476. xs_free(_xs_json_load_lexer(f, &jt));
  477. if (jt == JS_OBRACK)
  478. t = XSTYPE_LIST;
  479. else
  480. if (jt == JS_OCURLY)
  481. t = XSTYPE_DICT;
  482. return t;
  483. }
  484. static xs_val *_xs_json_load(json_src *f, int maxdepth)
  485. {
  486. xs_val *v = NULL;
  487. xstype t = xs_json_load_type(f);
  488. if (t == XSTYPE_LIST) {
  489. v = xs_list_new();
  490. xs_json_load_array(f, maxdepth, &v);
  491. }
  492. else
  493. if (t == XSTYPE_DICT) {
  494. v = xs_dict_new();
  495. xs_json_load_object(f, maxdepth, &v);
  496. }
  497. return v;
  498. }
  499. xs_val *xs_json_load_full(FILE *fp, int maxdepth)
  500. /* loads a JSON file */
  501. {
  502. json_src f;
  503. f.src.f = fp;
  504. f.b_end = NULL;
  505. return _xs_json_load(&f, maxdepth);
  506. }
  507. xs_val *xs_json_loads_full(const xs_str *json, int maxdepth)
  508. /* loads a string in JSON format and converts to a multiple data */
  509. {
  510. json_src f;
  511. f.src.b = json;
  512. f.b_end = json+strlen(json);
  513. return _xs_json_load(&f, maxdepth);
  514. }
  515. #endif /* XS_IMPLEMENTATION */
  516. #endif /* _XS_JSON_H */