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) 2004-2007, 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_GETHOSTNAME)
  36         if (SUCCESS == gethostname(hostname, lenof(hostname))) {
  37 #       if defined(HAVE_GETDOMAINNAME)
  38                 size_t hlen = strlen(hostname);
  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 #       endif
  49                 if (strcmp(hostname, "(none)")) {
  50                         return estrdup(hostname);
  51                 }
  52         }
  53 #endif
  54         return estrdup("localhost");
  55 }
  56
  57 PHP_MINIT_FUNCTION(http_url)
  58 {
  59         HTTP_LONG_CONSTANT("HTTP_URL_REPLACE", HTTP_URL_REPLACE);
  60         HTTP_LONG_CONSTANT("HTTP_URL_JOIN_PATH", HTTP_URL_JOIN_PATH);
  61         HTTP_LONG_CONSTANT("HTTP_URL_JOIN_QUERY", HTTP_URL_JOIN_QUERY);
  62         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_USER", HTTP_URL_STRIP_USER);
  63         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PASS", HTTP_URL_STRIP_PASS);
  64         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_AUTH", HTTP_URL_STRIP_AUTH);
  65         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PORT", HTTP_URL_STRIP_PORT);
  66         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PATH", HTTP_URL_STRIP_PATH);
  67         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_QUERY", HTTP_URL_STRIP_QUERY);
  68         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_FRAGMENT", HTTP_URL_STRIP_FRAGMENT);
  69         HTTP_LONG_CONSTANT("HTTP_URL_STRIP_ALL", HTTP_URL_STRIP_ALL);
  70         return SUCCESS;
  71 }
  72
  73 PHP_HTTP_API char *_http_absolute_url(const char *url TSRMLS_DC)
  74 {
  75         char *abs = NULL;
  76         php_url *purl = NULL;
  77
  78         if (url) {
  79                 purl = php_url_parse(abs = estrdup(url));
  80                 STR_SET(abs, NULL);
  81                 if (!purl) {
  82                         http_error_ex(HE_WARNING, HTTP_E_URL, "Could not parse URL (%s)", url);
  83                         return NULL;
  84                 }
  85         }
  86
  87         http_build_url(0, purl, NULL, NULL, &abs, NULL);
  88
  89         if (purl) {
  90                 php_url_free(purl);
  91         }
  92
  93         return abs;
  94 }
  95
  96 /* {{{ void http_build_url(int flags, const php_url *, const php_url *, php_url **, char **, size_t *) */
  97 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)
  98 {
  99 #if defined(HAVE_GETSERVBYPORT) || defined(HAVE_GETSERVBYNAME)
 100         struct servent *se;
 101 #endif
 102         php_url *url = ecalloc(1, sizeof(php_url));
 103
 104 #define __URLSET(u,n) \
 105         ((u)&&(u)->n)
 106 #define __URLCPY(n) \
 107         url->n = __URLSET(new_url,n) ? estrdup(new_url->n) : (__URLSET(old_url,n) ? estrdup(old_url->n) : NULL)
 108
 109         if (!(flags & HTTP_URL_STRIP_PORT)) {
 110                 url->port = (new_url&&new_url->port) ? new_url->port : ((old_url) ? old_url->port : 0);
 111         }
 112         if (!(flags & HTTP_URL_STRIP_USER)) {
 113                 __URLCPY(user);
 114         }
 115         if (!(flags & HTTP_URL_STRIP_PASS)) {
 116                 __URLCPY(pass);
 117         }
 118
 119         __URLCPY(scheme);
 120         __URLCPY(host);
 121
 122         if (!(flags & HTTP_URL_STRIP_PATH)) {
 123                 if ((flags & HTTP_URL_JOIN_PATH) && __URLSET(old_url, path) && __URLSET(new_url, path) && *new_url->path != '/') {
 124                         size_t old_path_len = strlen(old_url->path), new_path_len = strlen(new_url->path);
 125
 126                         url->path = ecalloc(1, old_path_len + new_path_len + 1 + 1);
 127
 128                         strcat(url->path, old_url->path);
 129                         if (url->path[old_path_len - 1] != '/') {
 130                                 php_dirname(url->path, old_path_len);
 131                                 strcat(url->path, "/");
 132                         }
 133                         strcat(url->path, new_url->path);
 134                 } else {
 135                         __URLCPY(path);
 136                 }
 137         }
 138         if (!(flags & HTTP_URL_STRIP_QUERY)) {
 139                 if ((flags & HTTP_URL_JOIN_QUERY) && __URLSET(new_url, query) && __URLSET(old_url, query)) {
 140                         zval qarr, qstr;
 141
 142                         INIT_PZVAL(&qstr);
 143                         INIT_PZVAL(&qarr);
 144                         array_init(&qarr);
 145
 146                         ZVAL_STRING(&qstr, old_url->query, 0);
 147                         http_querystring_modify(&qarr, &qstr);
 148                         ZVAL_STRING(&qstr, new_url->query, 0);
 149                         http_querystring_modify(&qarr, &qstr);
 150
 151                         ZVAL_NULL(&qstr);
 152                         http_querystring_update(&qarr, &qstr);
 153                         url->query = Z_STRVAL(qstr);
 154                         zval_dtor(&qarr);
 155                 } else {
 156                         __URLCPY(query);
 157                 }
 158         }
 159         if (!(flags & HTTP_URL_STRIP_FRAGMENT)) {
 160                 __URLCPY(fragment);
 161         }
 162
 163         if (!url->scheme) {
 164                 zval *https = http_get_server_var("HTTPS", 1);
 165                 if (https && !strcasecmp(Z_STRVAL_P(https), "ON")) {
 166                         url->scheme = estrndup("https", lenof("https"));
 167                 } else switch (url->port) {
 168                         case 443:
 169                                 url->scheme = estrndup("https", lenof("https"));
 170                                 break;
 171
 172 #ifndef HAVE_GETSERVBYPORT
 173                         default:
 174 #endif
 175                         case 80:
 176                                 url->scheme = estrndup("http", lenof("http"));
 177                                 break;
 178
 179 #ifdef HAVE_GETSERVBYPORT
 180                         default:
 181                                 if ((se = getservbyport(htons(url->port), "tcp")) && se->s_name) {
 182                                         url->scheme = estrdup(se->s_name);
 183                                 } else {
 184                                         url->scheme = estrndup("http", lenof("http"));
 185                                 }
 186                                 break;
 187 #endif
 188                 }
 189         }
 190
 191         if (!url->host) {
 192                 zval *zhost;
 193
 194                 if ((((zhost = http_get_server_var("HTTP_HOST", 1)) ||
 195                                 (zhost = http_get_server_var("SERVER_NAME", 1)))) && Z_STRLEN_P(zhost)) {
 196                         url->host = estrndup(Z_STRVAL_P(zhost), Z_STRLEN_P(zhost));
 197                 } else {
 198                         url->host = localhostname();
 199                 }
 200         }
 201
 202         if (!url->path) {
 203                 if (SG(request_info).request_uri && SG(request_info).request_uri[0]) {
 204                         const char *q = strchr(SG(request_info).request_uri, '?');
 205
 206                         if (q) {
 207                                 url->path = estrndup(SG(request_info).request_uri, q - SG(request_info).request_uri);
 208                         } else {
 209                                 url->path = estrdup(SG(request_info).request_uri);
 210                         }
 211                 } else {
 212                         url->path = estrndup("/", 1);
 213                 }
 214         } else if (url->path[0] != '/') {
 215                 if (SG(request_info).request_uri && SG(request_info).request_uri[0]) {
 216                         size_t ulen = strlen(SG(request_info).request_uri);
 217                         size_t plen = strlen(url->path);
 218                         char *path;
 219
 220                         if (SG(request_info).request_uri[ulen-1] != '/') {
 221                                 for (--ulen; ulen && SG(request_info).request_uri[ulen - 1] != '/'; --ulen);
 222                         }
 223
 224                         path = emalloc(ulen + plen + 1);
 225                         memcpy(path, SG(request_info).request_uri, ulen);
 226                         memcpy(path + ulen, url->path, plen);
 227                         path[ulen + plen] = '\0';
 228                         STR_SET(url->path, path);
 229                 } else {
 230                         size_t plen = strlen(url->path);
 231                         char *path = emalloc(plen + 1 + 1);
 232
 233                         path[0] = '/';
 234                         memcpy(&path[1], url->path, plen + 1);
 235                         STR_SET(url->path, path);
 236                 }
 237         }
 238         /* replace directory references if path is not a single slash */
 239         if (url->path[0] && (url->path[0] != '/' || url->path[1])) {
 240                 char *ptr, *end = url->path + strlen(url->path) + 1;
 241
 242                 for (ptr = strstr(url->path, "/."); ptr; ptr = strstr(ptr, "/.")) {
 243                         switch (ptr[2]) {
 244                                 case '\0':
 245                                         ptr[1] = '\0';
 246                                         break;
 247
 248                                 case '/':
 249                                         memmove(&ptr[1], &ptr[3], end - &ptr[3]);
 250                                         break;
 251
 252                                 case '.':
 253                                         if (ptr[3] == '/') {
 254                                                 char *pos = &ptr[4];
 255                                                 while (ptr != url->path) {
 256                                                         if (*--ptr == '/') {
 257                                                                 break;
 258                                                         }
 259                                                 }
 260                                                 memmove(&ptr[1], pos, end - pos);
 261                                                 break;
 262                                         } else if (!ptr[3]) {
 263                                                 /* .. at the end */
 264                                                 ptr[1] = '\0';
 265                                         }
 266                                         /* fallthrough */
 267
 268                                 default:
 269                                         /* something else */
 270                                         ++ptr;
 271                                         break;
 272                         }
 273                 }
 274         }
 275
 276         if (url->port) {
 277                 if (    ((url->port == 80) && !strcmp(url->scheme, "http"))
 278                         ||      ((url->port ==443) && !strcmp(url->scheme, "https"))
 279 #ifdef HAVE_GETSERVBYNAME
 280                         ||      ((se = getservbyname(url->scheme, "tcp")) && se->s_port &&
 281                                         (url->port == ntohs(se->s_port)))
 282 #endif
 283                 ) {
 284                         url->port = 0;
 285                 }
 286         }
 287
 288         if (url_str) {
 289                 size_t len;
 290
 291                 *url_str = emalloc(HTTP_URL_MAXLEN + 1);
 292
 293                 **url_str = '\0';
 294                 strlcat(*url_str, url->scheme, HTTP_URL_MAXLEN);
 295                 strlcat(*url_str, "://", HTTP_URL_MAXLEN);
 296
 297                 if (url->user && *url->user) {
 298                         strlcat(*url_str, url->user, HTTP_URL_MAXLEN);
 299                         if (url->pass && *url->pass) {
 300                                 strlcat(*url_str, ":", HTTP_URL_MAXLEN);
 301                                 strlcat(*url_str, url->pass, HTTP_URL_MAXLEN);
 302                         }
 303                         strlcat(*url_str, "@", HTTP_URL_MAXLEN);
 304                 }
 305
 306                 strlcat(*url_str, url->host, HTTP_URL_MAXLEN);
 307
 308                 if (url->port) {
 309                         char port_str[8];
 310
 311                         snprintf(port_str, sizeof(port_str), "%d", (int) url->port);
 312                         strlcat(*url_str, ":", HTTP_URL_MAXLEN);
 313                         strlcat(*url_str, port_str, HTTP_URL_MAXLEN);
 314                 }
 315
 316                 strlcat(*url_str, url->path, HTTP_URL_MAXLEN);
 317
 318                 if (url->query && *url->query) {
 319                         strlcat(*url_str, "?", HTTP_URL_MAXLEN);
 320                         strlcat(*url_str, url->query, HTTP_URL_MAXLEN);
 321                 }
 322
 323                 if (url->fragment && *url->fragment) {
 324                         strlcat(*url_str, "#", HTTP_URL_MAXLEN);
 325                         strlcat(*url_str, url->fragment, HTTP_URL_MAXLEN);
 326                 }
 327
 328                 if (HTTP_URL_MAXLEN == (len = strlen(*url_str))) {
 329                         http_error(HE_NOTICE, HTTP_E_URL, "Length of URL exceeds HTTP_URL_MAXLEN");
 330                 }
 331                 if (url_len) {
 332                         *url_len = len;
 333                 }
 334         }
 335
 336         if (url_ptr) {
 337                 *url_ptr = url;
 338         } else {
 339                 php_url_free(url);
 340         }
 341 }
 342 /* }}} */
 343
 344 /* {{{ STATUS http_urlencode_hash_ex(HashTable *, zend_bool, char *, size_t, char **, size_t *) */
 345 PHP_HTTP_API STATUS _http_urlencode_hash_ex(HashTable *hash, zend_bool override_argsep,
 346         char *pre_encoded_data, size_t pre_encoded_len,
 347         char **encoded_data, size_t *encoded_len TSRMLS_DC)
 348 {
 349         char *arg_sep;
 350         size_t arg_sep_len;
 351         phpstr *qstr = phpstr_new();
 352
 353         if (override_argsep || !(arg_sep_len = strlen(arg_sep = INI_STR("arg_separator.output")))) {
 354                 arg_sep = HTTP_URL_ARGSEP;
 355                 arg_sep_len = lenof(HTTP_URL_ARGSEP);
 356         }
 357
 358         if (pre_encoded_len && pre_encoded_data) {
 359                 phpstr_append(qstr, pre_encoded_data, pre_encoded_len);
 360         }
 361
 362         if (SUCCESS != http_urlencode_hash_recursive(hash, qstr, arg_sep, arg_sep_len, NULL, 0)) {
 363                 phpstr_free(&qstr);
 364                 return FAILURE;
 365         }
 366
 367         phpstr_data(qstr, encoded_data, encoded_len);
 368         phpstr_free(&qstr);
 369
 370         return SUCCESS;
 371 }
 372 /* }}} */
 373
 374 /* {{{ http_urlencode_hash_recursive */
 375 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)
 376 {
 377         HashKey key = initHashKey(0);
 378         zval **data = NULL;
 379         HashPosition pos;
 380
 381         if (!ht || !str) {
 382                 http_error(HE_WARNING, HTTP_E_INVALID_PARAM, "Invalid parameters");
 383                 return FAILURE;
 384         }
 385         if (ht->nApplyCount > 0) {
 386                 return SUCCESS;
 387         }
 388
 389         FOREACH_HASH_KEYVAL(pos, ht, key, data) {
 390                 char *encoded_key;
 391                 int encoded_len;
 392                 phpstr new_prefix;
 393
 394                 if (!data || !*data) {
 395                         phpstr_dtor(str);
 396                         return FAILURE;
 397                 }
 398
 399                 if (key.type == HASH_KEY_IS_STRING) {
 400                         if (!*key.str) {
 401                                 /* only public properties */
 402                                 continue;
 403                         }
 404                         if (key.len && key.str[key.len - 1] == '\0') {
 405                                 --key.len;
 406                         }
 407                         encoded_key = php_url_encode(key.str, key.len, &encoded_len);
 408                 } else {
 409                         encoded_len = spprintf(&encoded_key, 0, "%ld", key.num);
 410                 }
 411
 412                 {
 413                         phpstr_init(&new_prefix);
 414                         if (prefix && prefix_len) {
 415                                 phpstr_append(&new_prefix, prefix, prefix_len);
 416                                 phpstr_appends(&new_prefix, "%5B");
 417                         }
 418
 419                         phpstr_append(&new_prefix, encoded_key, encoded_len);
 420                         efree(encoded_key);
 421
 422                         if (prefix && prefix_len) {
 423                                 phpstr_appends(&new_prefix, "%5D");
 424                         }
 425                         phpstr_fix(&new_prefix);
 426                 }
 427
 428                 if (Z_TYPE_PP(data) == IS_ARRAY || Z_TYPE_PP(data) == IS_OBJECT) {
 429                         STATUS status;
 430                         ++ht->nApplyCount;
 431                         status = http_urlencode_hash_recursive(HASH_OF(*data), str, arg_sep, arg_sep_len, PHPSTR_VAL(&new_prefix), PHPSTR_LEN(&new_prefix));
 432                         --ht->nApplyCount;
 433                         if (SUCCESS != status) {
 434                                 phpstr_dtor(&new_prefix);
 435                                 phpstr_dtor(str);
 436                                 return FAILURE;
 437                         }
 438                 } else {
 439                         zval *val = zval_copy(IS_STRING, *data);
 440
 441                         if (PHPSTR_LEN(str)) {
 442                                 phpstr_append(str, arg_sep, arg_sep_len);
 443                         }
 444                         phpstr_append(str, PHPSTR_VAL(&new_prefix), PHPSTR_LEN(&new_prefix));
 445                         phpstr_appends(str, "=");
 446
 447                         if (Z_STRLEN_P(val) && Z_STRVAL_P(val)) {
 448                                 char *encoded_val;
 449                                 int encoded_len;
 450
 451                                 encoded_val = php_url_encode(Z_STRVAL_P(val), Z_STRLEN_P(val), &encoded_len);
 452                                 phpstr_append(str, encoded_val, encoded_len);
 453                                 efree(encoded_val);
 454                         }
 455
 456                         zval_free(&val);
 457                 }
 458                 phpstr_dtor(&new_prefix);
 459         }
 460         return SUCCESS;
 461 }
 462 /* }}} */
 463
 464 /*
 465  * Local variables:
 466  * tab-width: 4
 467  * c-basic-offset: 4
 468  * End:
 469  * vim600: noet sw=4 ts=4 fdm=marker
 470  * vim<600: noet sw=4 ts=4
 471  */
 472