git.m6w6.name Git - m6w6/ext-http/blob - http_url_api.c

   1 /*
   2     +--------------------------------------------------------------------+
   3     | PECL :: http                                                       |
   4     +--------------------------------------------------------------------+
   5     | Redistribution and use in source and binary forms, with or without |
   6     | modification, are permitted provided that the conditions mentioned |
   7     | in the accompanying LICENSE file are met.                          |
   8     +--------------------------------------------------------------------+
   9     | Copyright (c) 2004-2006, Michael Wallner <mike@php.net>            |
  10     +--------------------------------------------------------------------+
  11 */
  12
  13 /* $Id$ */
  14
  15 #define HTTP_WANT_SAPI
  16 #define HTTP_WANT_NETDB
  17 #include "php_http.h"
  18
  19 #include "zend_ini.h"
  20 #include "php_output.h"
  21 #include "ext/standard/php_string.h"
  22
  23 #include "php_http_api.h"
  24 #include "php_http_querystring_api.h"
  25 #include "php_http_url_api.h"
  26
  27 static inline char *localhostname(void)
  28 {
  29         char hostname[1024] = {0};
  30
  31 #ifdef PHP_WIN32
  32         if (SUCCESS == gethostname(hostname, lenof(hostname))) {
  33                 return estrdup(hostname);
  34         }
  35 #elif defined(HAVE_UNISTD_H)
  36         if (SUCCESS == gethostname(hostname, lenof(hostname))) {
  37                 size_t hlen = strlen(hostname);
  38
  39                 if (hlen <= lenof(hostname) - lenof("(none)")) {
  40                         hostname[hlen++] = '.';
  41                         if (SUCCESS == getdomainname(&hostname[hlen], lenof(hostname) - hlen)) {
  42                                 if (!strcmp(&hostname[hlen], "(none)")) {
  43                                         hostname[hlen - 1] = '\0';
  44                                 }
  45                                 return estrdup(hostname);
  46                         }
  47                 }
  48         }
  49 #endif
  50         return estrdup("localhost");
  51 }
  52
  53 PHP_MINIT_FUNCTION(http_url)
  54 {
  55         HTTP_LONG_CONSTANT("HTTP_URL_REPLACE", HTTP_URL_REPLACE);
  56         HTTP_LONG_CONSTANT("HTTP_URL_JOIN_PATH", HTTP_URL_JOIN_PATH);
  57         HTTP_LONG_CONSTANT("HTTP_URL_JOIN_QUERY", HTTP_URL_JOIN_QUERY);
  58         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_USER", HTTP_URL_STRIP_USER);
  59         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PASS", HTTP_URL_STRIP_PASS);
  60         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_AUTH", HTTP_URL_STRIP_AUTH);
  61         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PORT", HTTP_URL_STRIP_PORT);
  62         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PATH", HTTP_URL_STRIP_PATH);
  63         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_QUERY", HTTP_URL_STRIP_QUERY);
  64         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_FRAGMENT", HTTP_URL_STRIP_FRAGMENT);
  65         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_ALL", HTTP_URL_STRIP_ALL);
  66         return SUCCESS;
  67 }
  68
  69 PHP_HTTP_API char *_http_absolute_url(const char *url TSRMLS_DC)
  70 {
  71         char *abs = NULL;
  72         php_url *purl = NULL;
  73
  74         if (url) {
  75                 purl = php_url_parse(abs = estrdup(url));
  76                 STR_SET(abs, NULL);
  77                 if (!purl) {
  78                         http_error_ex(HE_WARNING, HTTP_E_URL, "Could not parse URL (%s)", url);
  79                         return NULL;
  80                 }
  81         }
  82
  83         http_build_url(0, purl, NULL, NULL, &abs, NULL);
  84
  85         if (purl) {
  86                 php_url_free(purl);
  87         }
  88
  89         return abs;
  90 }
  91
  92 /* {{{ void http_build_url(int flags, const php_url *, const php_url *, php_url **, char **, size_t *) */
  93 PHP_HTTP_API void _http_build_url(int flags, const php_url *old_url, const php_url *new_url, php_url **url_ptr, char **url_str, size_t *url_len TSRMLS_DC)
  94 {
  95 #ifdef HTTP_HAVE_NETDB
  96         struct servent *se;
  97 #endif
  98         php_url *url = ecalloc(1, sizeof(php_url));
  99
 100 #define __URLSET(u,n) \
 101         ((u)&&(u)->n)
 102 #define __URLCPY(n) \
 103         url->n = __URLSET(new_url,n) ? estrdup(new_url->n) : (__URLSET(old_url,n) ? estrdup(old_url->n) : NULL)
 104
 105         if (!(flags & HTTP_URL_STRIP_PORT)) {
 106                 url->port = (new_url&&new_url->port) ? new_url->port : ((old_url) ? old_url->port : 0);
 107         }
 108         if (!(flags & HTTP_URL_STRIP_USER)) {
 109                 __URLCPY(user);
 110         }
 111         if (!(flags & HTTP_URL_STRIP_PASS)) {
 112                 __URLCPY(pass);
 113         }
 114
 115         __URLCPY(scheme);
 116         __URLCPY(host);
 117
 118         if (!(flags & HTTP_URL_STRIP_PATH)) {
 119                 if ((flags & HTTP_URL_JOIN_PATH) && __URLSET(old_url, path) && __URLSET(new_url, path) && *new_url->path != '/') {
 120                         size_t old_path_len = strlen(old_url->path), new_path_len = strlen(new_url->path);
 121
 122                         url->path = ecalloc(1, old_path_len + new_path_len + 1 + 1);
 123
 124                         strcat(url->path, old_url->path);
 125                         if (url->path[old_path_len - 1] != '/') {
 126                                 php_dirname(url->path, old_path_len);
 127                                 strcat(url->path, "/");
 128                         }
 129                         strcat(url->path, new_url->path);
 130                 } else {
 131                         __URLCPY(path);
 132                 }
 133         }
 134         if (!(flags & HTTP_URL_STRIP_QUERY)) {
 135                 if ((flags & HTTP_URL_JOIN_QUERY) && __URLSET(new_url, query) && __URLSET(old_url, query)) {
 136                         zval qarr, qstr;
 137
 138                         INIT_PZVAL(&qstr);
 139                         INIT_PZVAL(&qarr);
 140                         array_init(&qarr);
 141
 142                         ZVAL_STRING(&qstr, old_url->query, 0);
 143                         http_querystring_modify(&qarr, &qstr);
 144                         ZVAL_STRING(&qstr, new_url->query, 0);
 145                         http_querystring_modify(&qarr, &qstr);
 146
 147                         ZVAL_NULL(&qstr);
 148                         http_querystring_update(&qarr, &qstr);
 149                         url->query = Z_STRVAL(qstr);
 150                         zval_dtor(&qarr);
 151                 } else {
 152                         __URLCPY(query);
 153                 }
 154         }
 155         if (!(flags & HTTP_URL_STRIP_FRAGMENT)) {
 156                 __URLCPY(fragment);
 157         }
 158
 159         if (!url->scheme) {
 160                 zval *https = http_get_server_var("HTTPS");
 161                 if (https && !strcasecmp(Z_STRVAL_P(https), "ON")) {
 162                         url->scheme = estrndup("https", lenof("https"));
 163                 } else switch (url->port) {
 164                         case 443:
 165                                 url->scheme = estrndup("https", lenof("https"));
 166                                 break;
 167
 168 #ifndef HTTP_HAVE_NETDB
 169                         default:
 170 #endif
 171                         case 80:
 172                                 url->scheme = estrndup("http", lenof("http"));
 173                                 break;
 174
 175 #ifdef HTTP_HAVE_NETDB
 176                         default:
 177                                 if ((se = getservbyport(htons(url->port), "tcp")) && se->s_name) {
 178                                         url->scheme = estrdup(se->s_name);
 179                                 } else {
 180                                         url->scheme = estrndup("http", lenof("http"));
 181                                 }
 182                                 break;
 183 #endif
 184                 }
 185         }
 186
 187         if (!url->host) {
 188                 zval *zhost;
 189
 190                 if ((((zhost = http_get_server_var("HTTP_HOST")) ||
 191                                 (zhost = http_get_server_var("SERVER_NAME")))) && Z_STRLEN_P(zhost)) {
 192                         url->host = estrndup(Z_STRVAL_P(zhost), Z_STRLEN_P(zhost));
 193                 } else {
 194                         url->host = localhostname();
 195                 }
 196         }
 197
 198         if (!url->path) {
 199                 if (SG(request_info).request_uri && SG(request_info).request_uri[0]) {
 200                         const char *q = strchr(SG(request_info).request_uri, '?');
 201
 202                         if (q) {
 203                                 url->path = estrndup(SG(request_info).request_uri, q - SG(request_info).request_uri);
 204                         } else {
 205                                 url->path = estrdup(SG(request_info).request_uri);
 206                         }
 207                 } else {
 208                         url->path = estrndup("/", 1);
 209                 }
 210         } else if (url->path[0] != '/' && SG(request_info).request_uri && SG(request_info).request_uri[0]) {
 211                 size_t ulen = strlen(SG(request_info).request_uri);
 212                 size_t plen = strlen(url->path);
 213                 char *path;
 214
 215                 if (SG(request_info).request_uri[ulen-1] != '/') {
 216                         for (--ulen; ulen && SG(request_info).request_uri[ulen - 1] != '/'; --ulen);
 217                 }
 218
 219                 path = emalloc(ulen + plen + 1);
 220                 memcpy(path, SG(request_info).request_uri, ulen);
 221                 memcpy(path + ulen, url->path, plen);
 222                 path[ulen + plen] = '\0';
 223                 STR_SET(url->path, path);
 224         }
 225         /* replace directory references if path is not a single slash */
 226         if (url->path[0] && (url->path[0] != '/' || url->path[1])) {
 227                 char *ptr, *end = url->path + strlen(url->path) + 1;
 228
 229                 for (ptr = strstr(url->path, "/."); ptr; ptr = strstr(ptr, "/.")) {
 230                         switch (ptr[2]) {
 231                                 case '\0':
 232                                         ptr[1] = '\0';
 233                                         break;
 234
 235                                 case '/':
 236                                         memmove(&ptr[1], &ptr[3], end - &ptr[3]);
 237                                         break;
 238
 239                                 case '.':
 240                                         if (ptr[3] == '/') {
 241                                                 char *pos = &ptr[4];
 242                                                 while (ptr != url->path) {
 243                                                         if (*--ptr == '/') {
 244                                                                 break;
 245                                                         }
 246                                                 }
 247                                                 memmove(&ptr[1], pos, end - pos);
 248                                         }
 249                                         break;
 250
 251                                 default:
 252                                         /* something else */
 253                                         ++ptr;
 254                                         break;
 255                         }
 256                 }
 257         }
 258
 259         if (url->port) {
 260                 if (    ((url->port == 80) && !strcmp(url->scheme, "http"))
 261                         ||      ((url->port ==443) && !strcmp(url->scheme, "https"))
 262 #ifdef HTTP_HAVE_NETDB
 263                         ||      ((se = getservbyname(url->scheme, "tcp")) && se->s_port &&
 264                                         (url->port == ntohs(se->s_port)))
 265 #endif
 266                 ) {
 267                         url->port = 0;
 268                 }
 269         }
 270
 271         if (url_str) {
 272                 size_t len;
 273
 274                 *url_str = emalloc(HTTP_URL_MAXLEN + 1);
 275
 276                 **url_str = '\0';
 277                 strlcat(*url_str, url->scheme, HTTP_URL_MAXLEN);
 278                 strlcat(*url_str, "://", HTTP_URL_MAXLEN);
 279
 280                 if (url->user && *url->user) {
 281                         strlcat(*url_str, url->user, HTTP_URL_MAXLEN);
 282                         if (url->pass && *url->pass) {
 283                                 strlcat(*url_str, ":", HTTP_URL_MAXLEN);
 284                                 strlcat(*url_str, url->pass, HTTP_URL_MAXLEN);
 285                         }
 286                         strlcat(*url_str, "@", HTTP_URL_MAXLEN);
 287                 }
 288
 289                 strlcat(*url_str, url->host, HTTP_URL_MAXLEN);
 290
 291                 if (url->port) {
 292                         char port_str[8] = {0};
 293
 294                         snprintf(port_str, lenof(port_str), "%d", (int) url->port);
 295                         strlcat(*url_str, ":", HTTP_URL_MAXLEN);
 296                         strlcat(*url_str, port_str, HTTP_URL_MAXLEN);
 297                 }
 298
 299                 if (*url->path != '/') {
 300                         strlcat(*url_str, "/", HTTP_URL_MAXLEN);
 301                 }
 302                 strlcat(*url_str, url->path, HTTP_URL_MAXLEN);
 303
 304                 if (url->query && *url->query) {
 305                         strlcat(*url_str, "?", HTTP_URL_MAXLEN);
 306                         strlcat(*url_str, url->query, HTTP_URL_MAXLEN);
 307                 }
 308
 309                 if (url->fragment && *url->fragment) {
 310                         strlcat(*url_str, "#", HTTP_URL_MAXLEN);
 311                         strlcat(*url_str, url->fragment, HTTP_URL_MAXLEN);
 312                 }
 313
 314                 if (HTTP_URL_MAXLEN == (len = strlen(*url_str))) {
 315                         http_error(HE_NOTICE, HTTP_E_URL, "Length of URL exceeds HTTP_URL_MAXLEN");
 316                 }
 317                 if (url_len) {
 318                         *url_len = len;
 319                 }
 320         }
 321
 322         if (url_ptr) {
 323                 *url_ptr = url;
 324         } else {
 325                 php_url_free(url);
 326         }
 327 }
 328 /* }}} */
 329
 330 /* {{{ STATUS http_urlencode_hash_ex(HashTable *, zend_bool, char *, size_t, char **, size_t *) */
 331 PHP_HTTP_API STATUS _http_urlencode_hash_ex(HashTable *hash, zend_bool override_argsep,
 332         char *pre_encoded_data, size_t pre_encoded_len,
 333         char **encoded_data, size_t *encoded_len TSRMLS_DC)
 334 {
 335         char *arg_sep;
 336         size_t arg_sep_len;
 337         phpstr *qstr = phpstr_new();
 338
 339         if (override_argsep || !(arg_sep_len = strlen(arg_sep = INI_STR("arg_separator.output")))) {
 340                 arg_sep = HTTP_URL_ARGSEP;
 341                 arg_sep_len = lenof(HTTP_URL_ARGSEP);
 342         }
 343
 344         if (pre_encoded_len && pre_encoded_data) {
 345                 phpstr_append(qstr, pre_encoded_data, pre_encoded_len);
 346         }
 347
 348         if (SUCCESS != http_urlencode_hash_recursive(hash, qstr, arg_sep, arg_sep_len, NULL, 0)) {
 349                 phpstr_free(&qstr);
 350                 return FAILURE;
 351         }
 352
 353         phpstr_data(qstr, encoded_data, encoded_len);
 354         phpstr_free(&qstr);
 355
 356         return SUCCESS;
 357 }
 358 /* }}} */
 359
 360 /* {{{ http_urlencode_hash_recursive */
 361 PHP_HTTP_API STATUS _http_urlencode_hash_recursive(HashTable *ht, phpstr *str, const char *arg_sep, size_t arg_sep_len, const char *prefix, size_t prefix_len TSRMLS_DC)
 362 {
 363         char *key = NULL;
 364         uint len = 0;
 365         ulong idx = 0;
 366         zval **data = NULL;
 367         HashPosition pos;
 368
 369         if (!ht || !str) {
 370                 http_error(HE_WARNING, HTTP_E_INVALID_PARAM, "Invalid parameters");
 371                 return FAILURE;
 372         }
 373         if (ht->nApplyCount > 0) {
 374                 return SUCCESS;
 375         }
 376
 377         FOREACH_HASH_KEYLENVAL(pos, ht, key, len, idx, data) {
 378                 char *encoded_key;
 379                 int encoded_len;
 380                 phpstr new_prefix;
 381
 382                 if (!data || !*data) {
 383                         phpstr_dtor(str);
 384                         return FAILURE;
 385                 }
 386
 387                 if (key) {
 388                         if (len && key[len - 1] == '\0') {
 389                                 --len;
 390                         }
 391                         encoded_key = php_url_encode(key, len, &encoded_len);
 392                         key = NULL;
 393                 } else {
 394                         encoded_len = spprintf(&encoded_key, 0, "%ld", idx);
 395                 }
 396
 397                 {
 398                         phpstr_init(&new_prefix);
 399                         if (prefix && prefix_len) {
 400                                 phpstr_append(&new_prefix, prefix, prefix_len);
 401                                 phpstr_appends(&new_prefix, "%5B");
 402                         }
 403
 404                         phpstr_append(&new_prefix, encoded_key, encoded_len);
 405                         efree(encoded_key);
 406
 407                         if (prefix && prefix_len) {
 408                                 phpstr_appends(&new_prefix, "%5D");
 409                         }
 410                         phpstr_fix(&new_prefix);
 411                 }
 412
 413                 if (Z_TYPE_PP(data) == IS_ARRAY) {
 414                         STATUS status;
 415                         ++ht->nApplyCount;
 416                         status = http_urlencode_hash_recursive(Z_ARRVAL_PP(data), str, arg_sep, arg_sep_len, PHPSTR_VAL(&new_prefix), PHPSTR_LEN(&new_prefix));
 417                         --ht->nApplyCount;
 418                         if (SUCCESS != status) {
 419                                 phpstr_dtor(&new_prefix);
 420                                 phpstr_dtor(str);
 421                                 return FAILURE;
 422                         }
 423                 } else {
 424                         zval *val = zval_copy(IS_STRING, *data);
 425
 426                         if (PHPSTR_LEN(str)) {
 427                                 phpstr_append(str, arg_sep, arg_sep_len);
 428                         }
 429                         phpstr_append(str, PHPSTR_VAL(&new_prefix), PHPSTR_LEN(&new_prefix));
 430                         phpstr_appends(str, "=");
 431
 432                         if (Z_STRLEN_P(val) && Z_STRVAL_P(val)) {
 433                                 char *encoded_val;
 434                                 int encoded_len;
 435
 436                                 encoded_val = php_url_encode(Z_STRVAL_P(val), Z_STRLEN_P(val), &encoded_len);
 437                                 phpstr_append(str, encoded_val, encoded_len);
 438                                 efree(encoded_val);
 439                         }
 440
 441                         zval_free(&val);
 442                 }
 443                 phpstr_dtor(&new_prefix);
 444         }
 445         return SUCCESS;
 446 }
 447 /* }}} */
 448
 449 /*
 450  * Local variables:
 451  * tab-width: 4
 452  * c-basic-offset: 4
 453  * End:
 454  * vim600: noet sw=4 ts=4 fdm=marker
 455  * vim<600: noet sw=4 ts=4
 456  */
 457