1
0

xs_url.h 8.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335
  1. /* copyright (c) 2022 - 2026 grunfink et al. / MIT license */
  2. #ifndef _XS_URL_H
  3. #define _XS_URL_H
  4. xs_str *xs_url_dec(const char *str);
  5. xs_str *xs_url_enc(const char *str);
  6. xs_str *xs_url_dec_emoji(const char *str);
  7. xs_dict *xs_url_vars(const char *str);
  8. xs_dict *xs_multipart_form_data(const char *payload, int p_size, const char *header);
  9. #ifdef XS_IMPLEMENTATION
  10. char *xs_url_dec_in(char *str, int qs)
  11. {
  12. char *w = str;
  13. char *r;
  14. for (r = str; *r != '\0'; r++) {
  15. switch (*r) {
  16. case '%': {
  17. unsigned hex;
  18. if (!r[1] || !r[2])
  19. return NULL;
  20. if (sscanf(r + 1, "%2x", &hex) != 1)
  21. return NULL;
  22. *w++ = hex;
  23. r += 2;
  24. break;
  25. }
  26. case '+':
  27. if (qs) {
  28. *w++ = ' ';
  29. break;
  30. }
  31. /* fall-through */
  32. default:
  33. *w++ = *r;
  34. }
  35. }
  36. *w++ = '\0';
  37. return str;
  38. }
  39. xs_str *xs_url_dec(const char *str)
  40. /* decodes an URL */
  41. {
  42. xs_str *s = xs_str_new(NULL);
  43. while (*str) {
  44. if (!xs_is_string(str))
  45. break;
  46. if (*str == '%') {
  47. unsigned int i;
  48. if (sscanf(str + 1, "%02x", &i) == 1) {
  49. unsigned char uc = i;
  50. if (!xs_is_string((char *)&uc))
  51. break;
  52. s = xs_append_m(s, (char *)&uc, 1);
  53. str += 2;
  54. }
  55. }
  56. else
  57. if (*str == '+')
  58. s = xs_append_m(s, " ", 1);
  59. else
  60. s = xs_append_m(s, str, 1);
  61. str++;
  62. }
  63. return s;
  64. }
  65. xs_str *xs_url_dec_emoji(const char *str)
  66. /* decodes an URL, returns NULL if not every char is an encoded hex */
  67. {
  68. xs_str *s = xs_str_new(NULL);
  69. while (*str) {
  70. if (!xs_is_string(str))
  71. break;
  72. if (*str == '%') {
  73. unsigned int i;
  74. if (sscanf(str + 1, "%02x", &i) == 1) {
  75. unsigned char uc = i;
  76. if (!xs_is_string((char *)&uc))
  77. break;
  78. s = xs_append_m(s, (char *)&uc, 1);
  79. str += 2;
  80. }
  81. }
  82. else
  83. return NULL;
  84. str++;
  85. }
  86. return s;
  87. }
  88. xs_str *xs_url_enc(const char *str)
  89. /* URL-encodes a string (RFC 3986) */
  90. {
  91. xs_str *s = xs_str_new(NULL);
  92. while (*str) {
  93. if (isalnum(*str) || strchr("-._~", *str)) {
  94. s = xs_append_m(s, str, 1);
  95. }
  96. else {
  97. char tmp[8];
  98. snprintf(tmp, sizeof(tmp), "%%%02X", (unsigned char)*str);
  99. s = xs_append_m(s, tmp, 3);
  100. }
  101. str++;
  102. }
  103. return s;
  104. }
  105. xs_dict *xs_url_vars(const char *str)
  106. /* parse url variables */
  107. {
  108. xs_dict *vars;
  109. vars = xs_dict_new();
  110. if (xs_is_string(str)) {
  111. xs *dup = xs_dup(str);
  112. char *k;
  113. char *saveptr;
  114. for (k = strtok_r(dup, "&", &saveptr);
  115. k;
  116. k = strtok_r(NULL, "&", &saveptr)) {
  117. char *v = strchr(k, '=');
  118. if (!v)
  119. continue;
  120. *v++ = '\0';
  121. k = xs_url_dec_in(k, 1);
  122. v = xs_url_dec_in(v, 1);
  123. if (!xs_is_string(k) || !xs_is_string(v))
  124. continue;
  125. const char *pv = xs_dict_get(vars, k);
  126. if (!xs_is_null(pv)) {
  127. /* there is a previous value: convert to a list and append */
  128. xs *vlist = NULL;
  129. if (xs_type(pv) == XSTYPE_LIST)
  130. vlist = xs_dup(pv);
  131. else {
  132. vlist = xs_list_new();
  133. vlist = xs_list_append(vlist, pv);
  134. }
  135. vlist = xs_list_append(vlist, v);
  136. vars = xs_dict_set(vars, k, vlist);
  137. }
  138. else {
  139. /* ends with []? force to always be a list */
  140. if (xs_endswith(k, "[]")) {
  141. xs *vlist = xs_list_new();
  142. vlist = xs_list_append(vlist, v);
  143. vars = xs_dict_append(vars, k, vlist);
  144. }
  145. else
  146. vars = xs_dict_append(vars, k, v);
  147. }
  148. }
  149. }
  150. return vars;
  151. }
  152. xs_dict *xs_multipart_form_data(const char *payload, int p_size, const char *header)
  153. /* parses a multipart/form-data payload */
  154. {
  155. xs *boundary = NULL;
  156. int offset = 0;
  157. int bsz;
  158. char *p;
  159. /* build the boundary string */
  160. {
  161. xs *l1 = xs_split(header, "=");
  162. if (xs_list_len(l1) != 2)
  163. return NULL;
  164. xs *t_boundary = xs_dup(xs_list_get(l1, 1));
  165. /* Tokodon sends the boundary header with double quotes surrounded */
  166. if (xs_between("\"", t_boundary, "\"") != 0)
  167. t_boundary = xs_strip_chars_i(t_boundary, "\"");
  168. boundary = xs_fmt("--%s", t_boundary);
  169. }
  170. bsz = strlen(boundary);
  171. xs_dict *p_vars = xs_dict_new();
  172. /* iterate searching the boundaries */
  173. while ((p = xs_memmem(payload + offset, p_size - offset, boundary, bsz)) != NULL) {
  174. xs *vn = NULL;
  175. xs *fn = NULL;
  176. xs *ct = NULL;
  177. char *q;
  178. int po, ps;
  179. /* final boundary? */
  180. p += bsz;
  181. if ((p - payload) + 2 > p_size || (p[0] == '-' && p[1] == '-'))
  182. break;
  183. /* skip the \r\n */
  184. p += 2;
  185. /* Tokodon sends also a Content-Type headers,
  186. let's use it to determine the file type */
  187. do {
  188. xs *s1 = NULL;
  189. xs *l1 = NULL;
  190. if (p[0] == '\r' && p[1] == '\n')
  191. break;
  192. q = memchr(p, '\r', p_size - (p - payload));
  193. /* unexpected formatting, fail immediately */
  194. if (q == NULL)
  195. return p_vars;
  196. s1 = xs_realloc(NULL, q - p + 1);
  197. memcpy(s1, p, q - p);
  198. s1[q - p] = '\0';
  199. if (xs_startswith(s1, "Content-Disposition") || xs_startswith(s1, "content-disposition")) {
  200. /* split by " like a primitive man */
  201. l1 = xs_split(s1, "\"");
  202. /* get the variable name */
  203. vn = xs_dup(xs_list_get(l1, 1));
  204. /* is it an attached file? */
  205. if (xs_list_len(l1) >= 4 && strcmp(xs_list_get(l1, 2), "; filename=") == 0) {
  206. /* get the file name */
  207. fn = xs_dup(xs_list_get(l1, 3));
  208. }
  209. }
  210. else
  211. if (xs_startswith(s1, "Content-Type") || xs_startswith(s1, "content-type")) {
  212. l1 = xs_split(s1, ":");
  213. if (xs_list_len(l1) >= 2) {
  214. ct = xs_lstrip_chars_i(xs_dup(xs_list_get(l1, 1)), " ");
  215. }
  216. }
  217. p += (q - p);
  218. p += 2; // Skip /r/n
  219. } while (1);
  220. /* find the start of the part content */
  221. if ((p = xs_memmem(p, p_size - (p - payload), "\r\n", 2)) == NULL)
  222. break;
  223. p += 2; // Skip empty line
  224. /* find the next boundary */
  225. if ((q = xs_memmem(p, p_size - (p - payload), boundary, bsz)) == NULL)
  226. break;
  227. po = p - payload;
  228. ps = q - p - 2; /* - 2 because the final \r\n */
  229. /* is it a filename? */
  230. if (fn != NULL) {
  231. /* p_var value is a list */
  232. /* if filename has no extension and content-type is image, attach extension to the filename */
  233. if (strchr(fn, '.') == NULL && ct && xs_startswith(ct, "image/")) {
  234. char *ext = strchr(ct, '/');
  235. ext++;
  236. fn = xs_str_cat(xs_str_new(""), fn, ".", ext);
  237. }
  238. xs *l1 = xs_list_new();
  239. xs *vpo = xs_number_new(po);
  240. xs *vps = xs_number_new(ps);
  241. l1 = xs_list_append(l1, fn);
  242. l1 = xs_list_append(l1, vpo);
  243. l1 = xs_list_append(l1, vps);
  244. if (xs_is_string(vn))
  245. p_vars = xs_dict_append(p_vars, vn, l1);
  246. }
  247. else {
  248. /* regular variable; just copy */
  249. if (ps > 0) {
  250. xs *vc = xs_realloc(NULL, ps + 1);
  251. memcpy(vc, payload + po, ps);
  252. vc[ps] = '\0';
  253. if (xs_is_string(vn) && xs_is_string(vc))
  254. p_vars = xs_dict_append(p_vars, vn, vc);
  255. }
  256. }
  257. /* move on */
  258. offset = q - payload;
  259. }
  260. return p_vars;
  261. }
  262. #endif /* XS_IMPLEMENTATION */
  263. #endif /* XS_URL_H */