xs_url.h 6.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260
  1. /* copyright (c) 2022 - 2025 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_dict *xs_url_vars(const char *str);
  7. xs_dict *xs_multipart_form_data(const char *payload, int p_size, const char *header);
  8. #ifdef XS_IMPLEMENTATION
  9. xs_str *xs_url_dec(const char *str)
  10. /* decodes an URL */
  11. {
  12. xs_str *s = xs_str_new(NULL);
  13. while (*str) {
  14. if (*str == '%') {
  15. unsigned int i;
  16. if (sscanf(str + 1, "%02x", &i) == 1) {
  17. unsigned char uc = i;
  18. if (!xs_is_string((char *)&uc))
  19. break;
  20. s = xs_append_m(s, (char *)&uc, 1);
  21. str += 2;
  22. }
  23. }
  24. else
  25. if (*str == '+')
  26. s = xs_append_m(s, " ", 1);
  27. else
  28. s = xs_append_m(s, str, 1);
  29. str++;
  30. }
  31. return s;
  32. }
  33. xs_str *xs_url_enc(const char *str)
  34. /* URL-encodes a string (RFC 3986) */
  35. {
  36. xs_str *s = xs_str_new(NULL);
  37. while (*str) {
  38. if (isalnum(*str) || strchr("-._~", *str)) {
  39. s = xs_append_m(s, str, 1);
  40. }
  41. else {
  42. char tmp[8];
  43. snprintf(tmp, sizeof(tmp), "%%%02X", (unsigned char)*str);
  44. s = xs_append_m(s, tmp, 3);
  45. }
  46. str++;
  47. }
  48. return s;
  49. }
  50. xs_dict *xs_url_vars(const char *str)
  51. /* parse url variables */
  52. {
  53. xs_dict *vars;
  54. vars = xs_dict_new();
  55. if (xs_is_string(str)) {
  56. /* split by arguments */
  57. xs *args = xs_split(str, "&");
  58. const xs_val *v;
  59. xs_list_foreach(args, v) {
  60. xs *dv = xs_url_dec(v);
  61. xs *kv = xs_split_n(dv, "=", 1);
  62. if (xs_list_len(kv) == 2) {
  63. const char *key = xs_list_get(kv, 0);
  64. const char *pv = xs_dict_get(vars, key);
  65. if (!xs_is_null(pv)) {
  66. /* there is a previous value: convert to a list and append */
  67. xs *vlist = NULL;
  68. if (xs_type(pv) == XSTYPE_LIST)
  69. vlist = xs_dup(pv);
  70. else {
  71. vlist = xs_list_new();
  72. vlist = xs_list_append(vlist, pv);
  73. }
  74. vlist = xs_list_append(vlist, xs_list_get(kv, 1));
  75. vars = xs_dict_set(vars, key, vlist);
  76. }
  77. else {
  78. /* ends with []? force to always be a list */
  79. if (xs_endswith(key, "[]")) {
  80. xs *vlist = xs_list_new();
  81. vlist = xs_list_append(vlist, xs_list_get(kv, 1));
  82. vars = xs_dict_append(vars, key, vlist);
  83. }
  84. else
  85. vars = xs_dict_append(vars, key, xs_list_get(kv, 1));
  86. }
  87. }
  88. }
  89. }
  90. return vars;
  91. }
  92. xs_dict *xs_multipart_form_data(const char *payload, int p_size, const char *header)
  93. /* parses a multipart/form-data payload */
  94. {
  95. xs *boundary = NULL;
  96. int offset = 0;
  97. int bsz;
  98. char *p;
  99. /* build the boundary string */
  100. {
  101. xs *l1 = xs_split(header, "=");
  102. if (xs_list_len(l1) != 2)
  103. return NULL;
  104. xs *t_boundary = xs_dup(xs_list_get(l1, 1));
  105. /* Tokodon sends the boundary header with double quotes surrounded */
  106. if (xs_between("\"", t_boundary, "\"") != 0)
  107. t_boundary = xs_strip_chars_i(t_boundary, "\"");
  108. boundary = xs_fmt("--%s", t_boundary);
  109. }
  110. bsz = strlen(boundary);
  111. xs_dict *p_vars = xs_dict_new();
  112. /* iterate searching the boundaries */
  113. while ((p = xs_memmem(payload + offset, p_size - offset, boundary, bsz)) != NULL) {
  114. xs *s1 = NULL;
  115. xs *l1 = NULL;
  116. const char *vn = NULL;
  117. const char *fn = NULL;
  118. const char *ct = NULL;
  119. char *q;
  120. int po, ps;
  121. /* final boundary? */
  122. p += bsz;
  123. if (p[0] == '-' && p[1] == '-')
  124. break;
  125. /* skip the \r\n */
  126. p += 2;
  127. /* Tokodon sends also a Content-Type headers,
  128. let's use it to determine the file type */
  129. do {
  130. if (p[0] == 13 && p[1] == 10)
  131. break;
  132. q = strchr(p, '\r');
  133. /* unexpected formatting, fail immediately */
  134. if (q == NULL)
  135. return p_vars;
  136. s1 = xs_realloc(NULL, q - p + 1);
  137. memcpy(s1, p, q - p);
  138. s1[q - p] = '\0';
  139. if (xs_startswith(s1, "Content-Disposition") || xs_startswith(s1, "content-disposition")) {
  140. /* split by " like a primitive man */
  141. l1 = xs_split(s1, "\"");
  142. /* get the variable name */
  143. vn = xs_list_get(l1, 1);
  144. /* is it an attached file? */
  145. if (xs_list_len(l1) >= 4 && strcmp(xs_list_get(l1, 2), "; filename=") == 0) {
  146. /* get the file name */
  147. fn = xs_list_get(l1, 3);
  148. }
  149. }
  150. else
  151. if (xs_startswith(s1, "Content-Type") || xs_startswith(s1, "content-type")) {
  152. l1 = xs_split(s1, ":");
  153. if (xs_list_len(l1) >= 2) {
  154. ct = xs_lstrip_chars_i(xs_dup(xs_list_get(l1, 1)), " ");
  155. }
  156. }
  157. p += (q - p);
  158. p += 2; // Skip /r/n
  159. } while (1);
  160. /* find the start of the part content */
  161. if ((p = xs_memmem(p, p_size - (p - payload), "\r\n", 2)) == NULL)
  162. break;
  163. p += 2; // Skip empty line
  164. /* find the next boundary */
  165. if ((q = xs_memmem(p, p_size - (p - payload), boundary, bsz)) == NULL)
  166. break;
  167. po = p - payload;
  168. ps = q - p - 2; /* - 2 because the final \r\n */
  169. /* is it a filename? */
  170. if (fn != NULL) {
  171. /* p_var value is a list */
  172. /* if filename has no extension and content-type is image, attach extension to the filename */
  173. if (strchr(fn, '.') == NULL && ct && xs_startswith(ct, "image/")) {
  174. char *ext = strchr(ct, '/');
  175. ext++;
  176. fn = xs_str_cat(xs_str_new(""), fn, ".", ext);
  177. }
  178. xs *l1 = xs_list_new();
  179. xs *vpo = xs_number_new(po);
  180. xs *vps = xs_number_new(ps);
  181. l1 = xs_list_append(l1, fn);
  182. l1 = xs_list_append(l1, vpo);
  183. l1 = xs_list_append(l1, vps);
  184. p_vars = xs_dict_append(p_vars, vn, l1);
  185. }
  186. else {
  187. /* regular variable; just copy */
  188. xs *vc = xs_realloc(NULL, ps + 1);
  189. memcpy(vc, payload + po, ps);
  190. vc[ps] = '\0';
  191. p_vars = xs_dict_append(p_vars, vn, vc);
  192. }
  193. /* move on */
  194. offset = q - payload;
  195. }
  196. return p_vars;
  197. }
  198. #endif /* XS_IMPLEMENTATION */
  199. #endif /* XS_URL_H */