- add HTTP_URL_STRIP_ALL constant
[m6w6/ext-http] / http_url_api.c
index 2fde306a74154dee0f3ecdedb6b2fd08967b80ea..827c7b6a2c5d6c1db62c459deb9af71ed7aaa011 100644 (file)
     | modification, are permitted provided that the conditions mentioned |
     | in the accompanying LICENSE file are met.                          |
     +--------------------------------------------------------------------+
-    | Copyright (c) 2004-2005, Michael Wallner <mike@php.net>            |
+    | Copyright (c) 2004-2006, Michael Wallner <mike@php.net>            |
     +--------------------------------------------------------------------+
 */
 
 /* $Id$ */
 
-#ifdef HAVE_CONFIG_H
-#      include "config.h"
-#endif
-
+#define HTTP_WANT_SAPI
 #define HTTP_WANT_NETDB
 #include "php_http.h"
 
-#include "SAPI.h"
 #include "zend_ini.h"
 #include "php_output.h"
 #include "ext/standard/php_string.h"
 
 #include "php_http_api.h"
+#include "php_http_querystring_api.h"
 #include "php_http_url_api.h"
 
+static inline char *localhostname(void)
+{
+       char hostname[1024] = {0};
+       
+#ifdef PHP_WIN32
+       if (SUCCESS == gethostname(hostname, lenof(hostname))) {
+               return estrdup(hostname);
+       }
+#elif defined(HAVE_UNISTD_H)
+       if (SUCCESS == gethostname(hostname, lenof(hostname))) {
+               size_t hlen = strlen(hostname);
+               
+               if (hlen <= lenof(hostname) - lenof("(none)")) {
+                       hostname[hlen++] = '.';
+                       if (SUCCESS == getdomainname(&hostname[hlen], lenof(hostname) - hlen)) {
+                               if (!strcmp(&hostname[hlen], "(none)")) {
+                                       hostname[hlen - 1] = '\0';
+                               }
+                               return estrdup(hostname);
+                       }
+               }
+       }
+#endif
+       return estrdup("localhost");
+}
+
+PHP_MINIT_FUNCTION(http_url)
+{
+       HTTP_LONG_CONSTANT("HTTP_URL_REPLACE", HTTP_URL_REPLACE);
+       HTTP_LONG_CONSTANT("HTTP_URL_JOIN_PATH", HTTP_URL_JOIN_PATH);
+       HTTP_LONG_CONSTANT("HTTP_URL_JOIN_QUERY", HTTP_URL_JOIN_QUERY);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_USER", HTTP_URL_STRIP_USER);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PASS", HTTP_URL_STRIP_PASS);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_AUTH", HTTP_URL_STRIP_AUTH);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PORT", HTTP_URL_STRIP_PORT);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_PATH", HTTP_URL_STRIP_PATH);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_QUERY", HTTP_URL_STRIP_QUERY);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_FRAGMENT", HTTP_URL_STRIP_FRAGMENT);
+       HTTP_LONG_CONSTANT("HTTP_URL_STRIP_ALL", HTTP_URL_STRIP_ALL);
+       return SUCCESS;
+}
+
 PHP_HTTP_API char *_http_absolute_url(const char *url TSRMLS_DC)
 {
-       char *abs = estrdup(url);
-       php_url *purl = php_url_parse(abs);
+       char *abs = NULL;
+       php_url *purl = NULL;
        
-       STR_SET(abs, NULL);
+       if (url) {
+               purl = php_url_parse(abs = estrdup(url));
+               STR_SET(abs, NULL);
+               if (!purl) {
+                       http_error_ex(HE_WARNING, HTTP_E_URL, "Could not parse URL (%s)", url);
+                       return NULL;
+               }
+       }
+       
+       http_build_url(0, purl, NULL, NULL, &abs, NULL);
        
        if (purl) {
-               http_build_url(purl, NULL, NULL, &abs, NULL);
                php_url_free(purl);
-       } else {
-               http_error_ex(HE_WARNING, HTTP_E_URL, "Could not parse URL (%s)", url);
        }
        
        return abs;
 }
 
-/* {{{ void http_build_url(const php_url *, const php_url *, php_url **, char **, size_t *) */
-PHP_HTTP_API void _http_build_url(const php_url *old_url, const php_url *new_url, php_url **url_ptr, char **url_str, size_t *url_len TSRMLS_DC)
+/* {{{ void http_build_url(int flags, const php_url *, const php_url *, php_url **, char **, size_t *) */
+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)
 {
-#if defined(PHP_WIN32) || defined(HAVE_NETDB_H)
+#ifdef HTTP_HAVE_NETDB
        struct servent *se;
 #endif
-       php_url *url = emalloc(sizeof(php_url));
+       php_url *url = ecalloc(1, sizeof(php_url));
 
+#define __URLSET(u,n) \
+       ((u)&&(u)->n)
 #define __URLCPY(n) \
-       url->n = (new_url&&new_url->n) ? estrdup(new_url->n) : ((old_url&&old_url->n) ? estrdup(old_url->n) : NULL)
-       url->port = (new_url&&new_url->port) ? new_url->port : ((old_url) ? old_url->port : 0);
+       url->n = __URLSET(new_url,n) ? estrdup(new_url->n) : (__URLSET(old_url,n) ? estrdup(old_url->n) : NULL)
+       
+       if (!(flags & HTTP_URL_STRIP_PORT)) {
+               url->port = (new_url&&new_url->port) ? new_url->port : ((old_url) ? old_url->port : 0);
+       }
+       if ((!(flags & HTTP_URL_STRIP_AUTH)) && (!(flags & HTTP_URL_STRIP_USER))) {
+               __URLCPY(user);
+       }
+       if ((!(flags & HTTP_URL_STRIP_AUTH)) && (!(flags & HTTP_URL_STRIP_PASS))) {
+               __URLCPY(pass);
+       }
+       
        __URLCPY(scheme);
-       __URLCPY(user);
-       __URLCPY(pass);
        __URLCPY(host);
-       __URLCPY(path);
-       __URLCPY(query);
-       __URLCPY(fragment);
+       
+       if (!(flags & HTTP_URL_STRIP_PATH)) {
+               if ((flags & HTTP_URL_JOIN_PATH) && __URLSET(old_url, path) && __URLSET(new_url, path) && *new_url->path != '/') {
+                       size_t old_path_len = strlen(old_url->path), new_path_len = strlen(new_url->path);
+                       
+                       url->path = ecalloc(1, old_path_len + new_path_len + 1 + 1);
+                       
+                       strcat(url->path, old_url->path);
+                       if (url->path[old_path_len - 1] != '/') {
+                               php_dirname(url->path, old_path_len);
+                               strcat(url->path, "/");
+                       }
+                       strcat(url->path, new_url->path);
+               } else {
+                       __URLCPY(path);
+               }
+       }
+       if (!(flags & HTTP_URL_STRIP_QUERY)) {
+               if ((flags & HTTP_URL_JOIN_QUERY) && __URLSET(new_url, query) && __URLSET(old_url, query)) {
+                       zval qarr, qstr;
+                       
+                       INIT_PZVAL(&qstr);
+                       INIT_PZVAL(&qarr);
+                       array_init(&qarr);
+                       
+                       ZVAL_STRING(&qstr, old_url->query, 0);
+                       http_querystring_modify(&qarr, &qstr);
+                       ZVAL_STRING(&qstr, new_url->query, 0);
+                       http_querystring_modify(&qarr, &qstr);
+                       
+                       ZVAL_NULL(&qstr);
+                       http_querystring_update(&qarr, &qstr);
+                       url->query = Z_STRVAL(qstr);
+                       zval_dtor(&qarr);
+               } else {
+                       __URLCPY(query);
+               }
+       }
+       if (!(flags & HTTP_URL_STRIP_FRAGMENT)) {
+               __URLCPY(fragment);
+       }
        
        if (!url->scheme) {
+               zval *https = http_get_server_var("HTTPS");
+               if (https && !strcasecmp(Z_STRVAL_P(https), "ON")) {
+                       url->scheme = estrndup("https", lenof("https"));
+               } else
                switch (url->port)
                {
                        case 443:
                                url->scheme = estrndup("https", lenof("https"));
                        break;
 
-#if !defined(PHP_WIN32) && !defined(HAVE_NETDB_H)
+#ifndef HTTP_HAVE_NETDB
                        default:
 #endif
                        case 80:
                                url->scheme = estrndup("http", lenof("http"));
                        break;
                        
-#if defined(PHP_WIN32) || defined(HAVE_NETDB_H)
+#ifdef HTTP_HAVE_NETDB
                        default:
                                if ((se = getservbyport(htons(url->port), "tcp")) && se->s_name) {
                                        url->scheme = estrdup(se->s_name);
@@ -96,12 +193,12 @@ PHP_HTTP_API void _http_build_url(const php_url *old_url, const php_url *new_url
                                (zhost = http_get_server_var("SERVER_NAME")))) && Z_STRLEN_P(zhost)) {
                        url->host = estrndup(Z_STRVAL_P(zhost), Z_STRLEN_P(zhost));
                } else {
-                       url->host = estrndup("localhost", lenof("localhost"));
+                       url->host = localhostname();
                }
        }
        
        if (!url->path) {
-               if (SG(request_info).request_uri && *SG(request_info).request_uri) {
+               if (SG(request_info).request_uri && SG(request_info).request_uri[0]) {
                        const char *q = strchr(SG(request_info).request_uri, '?');
                        
                        if (q) {
@@ -112,34 +209,60 @@ PHP_HTTP_API void _http_build_url(const php_url *old_url, const php_url *new_url
                } else {
                        url->path = estrndup("/", 1);
                }
-       } else if (*url->path != '/') {
-               if (SG(request_info).request_uri && *SG(request_info).request_uri) {
-                       const char *q = strchr(SG(request_info).request_uri, '?');
-                       char *uri, *path;
-                       size_t len;
+       } else if (url->path[0] != '/' && SG(request_info).request_uri && SG(request_info).request_uri[0]) {
+               size_t ulen = strlen(SG(request_info).request_uri);
+               size_t plen = strlen(url->path);
+               char *path;
+               
+               if (SG(request_info).request_uri[ulen-1] != '/') {
+                       for (--ulen; ulen && SG(request_info).request_uri[ulen - 1] != '/'; --ulen);
+               }
+               
+               path = emalloc(ulen + plen + 1);
+               memcpy(path, SG(request_info).request_uri, ulen);
+               memcpy(path + ulen, url->path, plen);
+               path[ulen + plen] = '\0';
+               STR_SET(url->path, path);
+       }
+       /* replace directory references if path is not a single slash */
+       if (url->path[0] && (url->path[0] != '/' || url->path[1])) {
+               char *ptr, *end = url->path + strlen(url->path) + 1;
                        
-                       if (q) {
-                               uri = estrndup(SG(request_info).request_uri, len = q - SG(request_info).request_uri);
-                       } else {
-                               uri = estrndup(SG(request_info).request_uri, len = strlen(SG(request_info).request_uri));
+               for (ptr = strstr(url->path, "/."); ptr; ptr = strstr(ptr, "/.")) {
+                       switch (ptr[2])
+                       {
+                               case '\0':
+                                       ptr[1] = '\0';
+                               break;
+                               
+                               case '/':
+                                       memmove(&ptr[1], &ptr[3], end - &ptr[3]);
+                               break;
+                                       
+                               case '.':
+                                       if (ptr[3] == '/') {
+                                               char *pos = &ptr[4];
+                                               while (ptr != url->path) {
+                                                       if (*--ptr == '/') {
+                                                               break;
+                                                       }
+                                               }
+                                               memmove(&ptr[1], pos, end - pos);
+                                       }
+                               break;
+                               
+                               default:
+                                       /* something else */
+                                       ++ptr;
+                               break;
                        }
-                       
-                       php_dirname(uri, len);
-                       spprintf(&path, 0, "%s/%s", uri, url->path);
-                       efree(uri);
-                       STR_SET(url->path, path);
-               } else {
-                       char *uri;
-                       
-                       spprintf(&uri, 0, "/%s", url->path);
-                       STR_SET(url->path, uri);
                }
        }
        
        if (url->port) {
                if (    ((url->port == 80) && !strcmp(url->scheme, "http"))
                        ||      ((url->port ==443) && !strcmp(url->scheme, "https"))
-#if defined(PHP_WIN32) || defined(HAVE_NETDB_H)
+#ifdef HTTP_HAVE_NETDB
                        ||      ((se = getservbyname(url->scheme, "tcp")) && se->s_port && 
                                        (url->port == ntohs(se->s_port)))
 #endif
@@ -169,9 +292,9 @@ PHP_HTTP_API void _http_build_url(const php_url *old_url, const php_url *new_url
                strlcat(*url_str, url->host, HTTP_URL_MAXLEN);
                
                if (url->port) {
-                       char port_str[6] = {0};
+                       char port_str[8] = {0};
                        
-                       snprintf(port_str, 5, "%d", (int) url->port);
+                       snprintf(port_str, lenof(port_str), "%d", (int) url->port);
                        strlcat(*url_str, ":", HTTP_URL_MAXLEN);
                        strlcat(*url_str, port_str, HTTP_URL_MAXLEN);
                }
@@ -260,6 +383,7 @@ PHP_HTTP_API STATUS _http_urlencode_hash_recursive(HashTable *ht, phpstr *str, c
                phpstr new_prefix;
                
                if (!data || !*data) {
+                       phpstr_dtor(str);
                        return FAILURE;
                }
                
@@ -277,14 +401,14 @@ PHP_HTTP_API STATUS _http_urlencode_hash_recursive(HashTable *ht, phpstr *str, c
                        phpstr_init(&new_prefix);
                        if (prefix && prefix_len) {
                                phpstr_append(&new_prefix, prefix, prefix_len);
-                               phpstr_appends(&new_prefix, "[");
+                               phpstr_appends(&new_prefix, "%5B");
                        }
                        
                        phpstr_append(&new_prefix, encoded_key, encoded_len);
                        efree(encoded_key);
                        
                        if (prefix && prefix_len) {
-                               phpstr_appends(&new_prefix, "]");
+                               phpstr_appends(&new_prefix, "%5D");
                        }
                        phpstr_fix(&new_prefix);
                }
@@ -296,16 +420,11 @@ PHP_HTTP_API STATUS _http_urlencode_hash_recursive(HashTable *ht, phpstr *str, c
                        --ht->nApplyCount;
                        if (SUCCESS != status) {
                                phpstr_dtor(&new_prefix);
+                               phpstr_dtor(str);
                                return FAILURE;
                        }
                } else {
-                       zval *val;
-                       
-                       ALLOC_ZVAL(val);
-                       *val = **data;
-                       INIT_PZVAL(val);
-                       zval_copy_ctor(val);
-                       convert_to_string(val);
+                       zval *val = zval_copy(IS_STRING, *data);
                        
                        if (PHPSTR_LEN(str)) {
                                phpstr_append(str, arg_sep, arg_sep_len);
@@ -322,7 +441,7 @@ PHP_HTTP_API STATUS _http_urlencode_hash_recursive(HashTable *ht, phpstr *str, c
                                efree(encoded_val);
                        }
                        
-                       zval_ptr_dtor(&val);
+                       zval_free(&val);
                }
                phpstr_dtor(&new_prefix);
        }