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 estrndup("localhost", lenof("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         HTTP_LONG_CONSTANT("HTTP_URL_FROM_ENV", HTTP_URL_FROM_ENV);
  71         return SUCCESS;
  72 }
  73
  74 PHP_HTTP_API char *_http_absolute_url_ex(const char *url, int flags TSRMLS_DC)
  75 {
  76         char *abs = NULL;
  77         php_url *purl = NULL;
  78
  79         if (url) {
  80                 purl = php_url_parse(abs = estrdup(url));
  81                 STR_SET(abs, NULL);
  82                 if (!purl) {
  83                         http_error_ex(HE_WARNING, HTTP_E_URL, "Could not parse URL (%s)", url);
  84                         return NULL;
  85                 }
  86         }
  87
  88         http_build_url(flags, purl, NULL, NULL, &abs, NULL);
  89
  90         if (purl) {
  91                 php_url_free(purl);
  92         }
  93
  94         return abs;
  95 }
  96
  97 /* {{{ void http_build_url(int flags, const php_url *, const php_url *, php_url **, char **, size_t *) */
  98 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)
  99 {
 100 #if defined(HAVE_GETSERVBYPORT) || defined(HAVE_GETSERVBYNAME)
 101         struct servent *se;
 102 #endif
 103         php_url *url = ecalloc(1, sizeof(php_url));
 104
 105 #define __URLSET(u,n) \
 106         ((u)&&(u)->n)
 107 #define __URLCPY(n) \
 108         url->n = __URLSET(new_url,n) ? estrdup(new_url->n) : (__URLSET(old_url,n) ? estrdup(old_url->n) : NULL)
 109
 110         if (!(flags & HTTP_URL_STRIP_PORT)) {
 111                 url->port = __URLSET(new_url, port) ? new_url->port : ((old_url) ? old_url->port : 0);
 112         }
 113         if (!(flags & HTTP_URL_STRIP_USER)) {
 114                 __URLCPY(user);
 115         }
 116         if (!(flags & HTTP_URL_STRIP_PASS)) {
 117                 __URLCPY(pass);
 118         }
 119
 120         __URLCPY(scheme);
 121         __URLCPY(host);
 122
 123         if (!(flags & HTTP_URL_STRIP_PATH)) {
 124                 if ((flags & HTTP_URL_JOIN_PATH) && __URLSET(old_url, path) && __URLSET(new_url, path) && *new_url->path != '/') {
 125                         size_t old_path_len = strlen(old_url->path), new_path_len = strlen(new_url->path);
 126
 127                         url->path = ecalloc(1, old_path_len + new_path_len + 1 + 1);
 128
 129                         strcat(url->path, old_url->path);
 130                         if (url->path[old_path_len - 1] != '/') {
 131                                 php_dirname(url->path, old_path_len);
 132                                 strcat(url->path, "/");
 133                         }
 134                         strcat(url->path, new_url->path);
 135                 } else {
 136                         __URLCPY(path);
 137                 }
 138         }
 139         if (!(flags & HTTP_URL_STRIP_QUERY)) {
 140                 if ((flags & HTTP_URL_JOIN_QUERY) && __URLSET(new_url, query) && __URLSET(old_url, query)) {
 141                         zval qarr, qstr;
 142
 143                         INIT_PZVAL(&qstr);
 144                         INIT_PZVAL(&qarr);
 145                         array_init(&qarr);
 146
 147                         ZVAL_STRING(&qstr, old_url->query, 0);
 148                         http_querystring_modify(&qarr, &qstr);
 149                         ZVAL_STRING(&qstr, new_url->query, 0);
 150                         http_querystring_modify(&qarr, &qstr);
 151
 152                         ZVAL_NULL(&qstr);
 153                         http_querystring_update(&qarr, &qstr);
 154                         url->query = Z_STRVAL(qstr);
 155                         zval_dtor(&qarr);
 156                 } else {
 157                         __URLCPY(query);
 158                 }
 159         }
 160         if (!(flags & HTTP_URL_STRIP_FRAGMENT)) {
 161                 __URLCPY(fragment);
 162         }
 163
 164         if (!url->scheme) {
 165                 if (flags & HTTP_URL_FROM_ENV) {
 166                         zval *https = http_get_server_var("HTTPS", 1);
 167                         if (https && !strcasecmp(Z_STRVAL_P(https), "ON")) {
 168                                 url->scheme = estrndup("https", lenof("https"));
 169                         } else switch (url->port) {
 170                                 case 443:
 171                                         url->scheme = estrndup("https", lenof("https"));
 172                                         break;
 173
 174 #ifndef HAVE_GETSERVBYPORT
 175                                 default:
 176 #endif
 177                                 case 80:
 178                                         url->scheme = estrndup("http", lenof("http"));
 179                                         break;
 180
 181 #ifdef HAVE_GETSERVBYPORT
 182                                 default:
 183                                         if ((se = getservbyport(htons(url->port), "tcp")) && se->s_name) {
 184                                                 url->scheme = estrdup(se->s_name);
 185                                         } else {
 186                                                 url->scheme = estrndup("http", lenof("http"));
 187                                         }
 188                                         break;
 189 #endif
 190                         }
 191                 } else {
 192                         url->scheme = estrndup("http", lenof("http"));
 193                 }
 194         }
 195
 196         if (!url->host) {
 197                 if (flags & HTTP_URL_FROM_ENV) {
 198                         zval *zhost;
 199
 200                         if ((((zhost = http_get_server_var("HTTP_HOST", 1)) ||
 201                                         (zhost = http_get_server_var("SERVER_NAME", 1)))) && Z_STRLEN_P(zhost)) {
 202                                 url->host = estrndup(Z_STRVAL_P(zhost), Z_STRLEN_P(zhost));
 203                         } else {
 204                                 url->host = localhostname();
 205                         }
 206                 } else {
 207                         url->host = estrndup("localhost", lenof("localhost"));
 208                 }
 209         }
 210
 211         if (!url->path) {
 212                 if ((flags & HTTP_URL_FROM_ENV) && SG(request_info).request_uri && SG(request_info).request_uri[0]) {
 213                         const char *q = strchr(SG(request_info).request_uri, '?');
 214
 215                         if (q) {
 216                                 url->path = estrndup(SG(request_info).request_uri, q - SG(request_info).request_uri);
 217                         } else {
 218                                 url->path = estrdup(SG(request_info).request_uri);
 219                         }
 220                 } else {
 221                         url->path = estrndup("/", 1);
 222                 }
 223         } else if (url->path[0] != '/') {
 224                 if ((flags & HTTP_URL_FROM_ENV) && SG(request_info).request_uri && SG(request_info).request_uri[0]) {
 225                         size_t ulen = strlen(SG(request_info).request_uri);
 226                         size_t plen = strlen(url->path);
 227                         char *path;
 228
 229                         if (SG(request_info).request_uri[ulen-1] != '/') {
 230                                 for (--ulen; ulen && SG(request_info).request_uri[ulen - 1] != '/'; --ulen);
 231                         }
 232
 233                         path = emalloc(ulen + plen + 1);
 234                         memcpy(path, SG(request_info).request_uri, ulen);
 235                         memcpy(path + ulen, url->path, plen);
 236                         path[ulen + plen] = '\0';
 237                         STR_SET(url->path, path);
 238                 } else {
 239                         size_t plen = strlen(url->path);
 240                         char *path = emalloc(plen + 1 + 1);
 241
 242                         path[0] = '/';
 243                         memcpy(&path[1], url->path, plen + 1);
 244                         STR_SET(url->path, path);
 245                 }
 246         }
 247         /* replace directory references if path is not a single slash */
 248         if (url->path[0] && (url->path[0] != '/' || url->path[1])) {
 249                 char *ptr, *end = url->path + strlen(url->path) + 1;
 250
 251                 for (ptr = strstr(url->path, "/."); ptr; ptr = strstr(ptr, "/.")) {
 252                         switch (ptr[2]) {
 253                                 case '\0':
 254                                         ptr[1] = '\0';
 255                                         break;
 256
 257                                 case '/':
 258                                         memmove(&ptr[1], &ptr[3], end - &ptr[3]);
 259                                         break;
 260
 261                                 case '.':
 262                                         if (ptr[3] == '/') {
 263                                                 char *pos = &ptr[4];
 264                                                 while (ptr != url->path) {
 265                                                         if (*--ptr == '/') {
 266                                                                 break;
 267                                                         }
 268                                                 }
 269                                                 memmove(&ptr[1], pos, end - pos);
 270                                                 break;
 271                                         } else if (!ptr[3]) {
 272                                                 /* .. at the end */
 273                                                 ptr[1] = '\0';
 274                                         }
 275                                         /* fallthrough */
 276
 277                                 default:
 278                                         /* something else */
 279                                         ++ptr;
 280                                         break;
 281                         }
 282                 }
 283         }
 284
 285         if (url->port) {
 286                 if (    ((url->port == 80) && !strcmp(url->scheme, "http"))
 287                         ||      ((url->port ==443) && !strcmp(url->scheme, "https"))
 288 #ifdef HAVE_GETSERVBYNAME
 289                         ||      ((se = getservbyname(url->scheme, "tcp")) && se->s_port &&
 290                                         (url->port == ntohs(se->s_port)))
 291 #endif
 292                 ) {
 293                         url->port = 0;
 294                 }
 295         }
 296
 297         if (url_str) {
 298                 size_t len;
 299
 300                 *url_str = emalloc(HTTP_URL_MAXLEN + 1);
 301
 302                 **url_str = '\0';
 303                 strlcat(*url_str, url->scheme, HTTP_URL_MAXLEN);
 304                 strlcat(*url_str, "://", HTTP_URL_MAXLEN);
 305
 306                 if (url->user && *url->user) {
 307                         strlcat(*url_str, url->user, HTTP_URL_MAXLEN);
 308                         if (url->pass && *url->pass) {
 309                                 strlcat(*url_str, ":", HTTP_URL_MAXLEN);
 310                                 strlcat(*url_str, url->pass, HTTP_URL_MAXLEN);
 311                         }
 312                         strlcat(*url_str, "@", HTTP_URL_MAXLEN);
 313                 }
 314
 315                 strlcat(*url_str, url->host, HTTP_URL_MAXLEN);
 316
 317                 if (url->port) {
 318                         char port_str[8];
 319
 320                         snprintf(port_str, sizeof(port_str), "%d", (int) url->port);
 321                         strlcat(*url_str, ":", HTTP_URL_MAXLEN);
 322                         strlcat(*url_str, port_str, HTTP_URL_MAXLEN);
 323                 }
 324
 325                 strlcat(*url_str, url->path, HTTP_URL_MAXLEN);
 326
 327                 if (url->query && *url->query) {
 328                         strlcat(*url_str, "?", HTTP_URL_MAXLEN);
 329                         strlcat(*url_str, url->query, HTTP_URL_MAXLEN);
 330                 }
 331
 332                 if (url->fragment && *url->fragment) {
 333                         strlcat(*url_str, "#", HTTP_URL_MAXLEN);
 334                         strlcat(*url_str, url->fragment, HTTP_URL_MAXLEN);
 335                 }
 336
 337                 if (HTTP_URL_MAXLEN == (len = strlen(*url_str))) {
 338                         http_error(HE_NOTICE, HTTP_E_URL, "Length of URL exceeds HTTP_URL_MAXLEN");
 339                 }
 340                 if (url_len) {
 341                         *url_len = len;
 342                 }
 343         }
 344
 345         if (url_ptr) {
 346                 *url_ptr = url;
 347         } else {
 348                 php_url_free(url);
 349         }
 350 }
 351 /* }}} */
 352
 353 /* {{{ STATUS http_urlencode_hash_ex(HashTable *, zend_bool, char *, size_t, char **, size_t *) */
 354 PHP_HTTP_API STATUS _http_urlencode_hash_ex(HashTable *hash, zend_bool override_argsep,
 355         char *pre_encoded_data, size_t pre_encoded_len,
 356         char **encoded_data, size_t *encoded_len TSRMLS_DC)
 357 {
 358         char *arg_sep;
 359         size_t arg_sep_len;
 360         phpstr *qstr = phpstr_new();
 361
 362         if (override_argsep || !(arg_sep_len = strlen(arg_sep = INI_STR("arg_separator.output")))) {
 363                 arg_sep = HTTP_URL_ARGSEP;
 364                 arg_sep_len = lenof(HTTP_URL_ARGSEP);
 365         }
 366
 367         if (pre_encoded_len && pre_encoded_data) {
 368                 phpstr_append(qstr, pre_encoded_data, pre_encoded_len);
 369         }
 370
 371         if (SUCCESS != http_urlencode_hash_recursive(hash, qstr, arg_sep, arg_sep_len, NULL, 0)) {
 372                 phpstr_free(&qstr);
 373                 return FAILURE;
 374         }
 375
 376         phpstr_data(qstr, encoded_data, encoded_len);
 377         phpstr_free(&qstr);
 378
 379         return SUCCESS;
 380 }
 381 /* }}} */
 382
 383 /* {{{ http_urlencode_hash_recursive */
 384 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)
 385 {
 386         HashKey key = initHashKey(0);
 387         zval **data = NULL;
 388         HashPosition pos;
 389
 390         if (!ht || !str) {
 391                 http_error(HE_WARNING, HTTP_E_INVALID_PARAM, "Invalid parameters");
 392                 return FAILURE;
 393         }
 394         if (ht->nApplyCount > 0) {
 395                 return SUCCESS;
 396         }
 397
 398         FOREACH_HASH_KEYVAL(pos, ht, key, data) {
 399                 char *encoded_key;
 400                 int encoded_len;
 401                 phpstr new_prefix;
 402
 403                 if (!data || !*data) {
 404                         phpstr_dtor(str);
 405                         return FAILURE;
 406                 }
 407
 408                 if (key.type == HASH_KEY_IS_STRING) {
 409                         if (!*key.str) {
 410                                 /* only public properties */
 411                                 continue;
 412                         }
 413                         if (key.len && key.str[key.len - 1] == '\0') {
 414                                 --key.len;
 415                         }
 416                         encoded_key = php_url_encode(key.str, key.len, &encoded_len);
 417                 } else {
 418                         encoded_len = spprintf(&encoded_key, 0, "%ld", key.num);
 419                 }
 420
 421                 {
 422                         phpstr_init(&new_prefix);
 423                         if (prefix && prefix_len) {
 424                                 phpstr_append(&new_prefix, prefix, prefix_len);
 425                                 phpstr_appends(&new_prefix, "%5B");
 426                         }
 427
 428                         phpstr_append(&new_prefix, encoded_key, encoded_len);
 429                         efree(encoded_key);
 430
 431                         if (prefix && prefix_len) {
 432                                 phpstr_appends(&new_prefix, "%5D");
 433                         }
 434                         phpstr_fix(&new_prefix);
 435                 }
 436
 437                 if (Z_TYPE_PP(data) == IS_ARRAY || Z_TYPE_PP(data) == IS_OBJECT) {
 438                         STATUS status;
 439                         ++ht->nApplyCount;
 440                         status = http_urlencode_hash_recursive(HASH_OF(*data), str, arg_sep, arg_sep_len, PHPSTR_VAL(&new_prefix), PHPSTR_LEN(&new_prefix));
 441                         --ht->nApplyCount;
 442                         if (SUCCESS != status) {
 443                                 phpstr_dtor(&new_prefix);
 444                                 phpstr_dtor(str);
 445                                 return FAILURE;
 446                         }
 447                 } else {
 448                         zval *val = zval_copy(IS_STRING, *data);
 449
 450                         if (PHPSTR_LEN(str)) {
 451                                 phpstr_append(str, arg_sep, arg_sep_len);
 452                         }
 453                         phpstr_append(str, PHPSTR_VAL(&new_prefix), PHPSTR_LEN(&new_prefix));
 454                         phpstr_appends(str, "=");
 455
 456                         if (Z_STRLEN_P(val) && Z_STRVAL_P(val)) {
 457                                 char *encoded_val;
 458                                 int encoded_len;
 459
 460                                 encoded_val = php_url_encode(Z_STRVAL_P(val), Z_STRLEN_P(val), &encoded_len);
 461                                 phpstr_append(str, encoded_val, encoded_len);
 462                                 efree(encoded_val);
 463                         }
 464
 465                         zval_free(&val);
 466                 }
 467                 phpstr_dtor(&new_prefix);
 468         }
 469         return SUCCESS;
 470 }
 471 /* }}} */
 472
 473 /*
 474  * Local variables:
 475  * tab-width: 4
 476  * c-basic-offset: 4
 477  * End:
 478  * vim600: noet sw=4 ts=4 fdm=marker
 479  * vim<600: noet sw=4 ts=4
 480  */
 481