Merge pull request #5386 from annando/dfrn-unknown-owner

[friendica.git] / src / Util / Network.php

diff --git a/src/Util/Network.php b/src/Util/Network.php
index c9955f800f40252d588504590016d266f2024a36..0e65145b78340caca37bc49f583a9d62c441df0e 100644 (file)
--- a/src/Util/Network.php
+++ b/src/Util/Network.php
@@ -18,12 +18,13 @@ use DomXPath;
 class Network
 {
        /**
-        * @brief Curl wrapper
+        * Curl wrapper
         *
         * If binary flag is true, return binary results.
         * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
         * to preserve cookies from one request to the next.
         *
+        * @brief Curl wrapper
         * @param string  $url            URL to fetch
         * @param boolean $binary         default false
         *                                TRUE if asked to return binary results (file download)
@@ -36,7 +37,31 @@ class Network
         */
        public static function fetchUrl($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
        {
-               $ret = self::curl(
+               $ret = self::fetchUrlFull($url, $binary, $redirects, $timeout, $accept_content, $cookiejar);
+
+               return $ret['body'];
+       }
+
+       /**
+        * Curl wrapper with array of return values.
+        *
+        * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
+        * all the information collected during the fetch.
+        *
+        * @brief Curl wrapper with array of return values.
+        * @param string  $url            URL to fetch
+        * @param boolean $binary         default false
+        *                                TRUE if asked to return binary results (file download)
+        * @param integer $redirects      The recursion counter for internal use - default 0
+        * @param integer $timeout        Timeout in seconds, default system config value or 60 seconds
+        * @param string  $accept_content supply Accept: header with 'accept_content' as the value
+        * @param string  $cookiejar      Path to cookie jar file
+        *
+        * @return array With all relevant information, 'body' contains the actual fetched content.
+        */
+       public static function fetchUrlFull($url, $binary = false, &$redirects = 0, $timeout = 0, $accept_content = null, $cookiejar = 0)
+       {
+               return self::curl(
                        $url,
                        $binary,
                        $redirects,
@@ -45,8 +70,6 @@ class Network
                        'cookiejar'=>$cookiejar
                        ]
                );
-
-               return($ret['body']);
        }
 
        /**
@@ -79,6 +102,18 @@ class Network
 
                $a = get_app();
 
+               $parts = parse_url($url);
+               $path_parts = explode('/', defaults($parts, 'path', ''));
+               foreach ($path_parts as $part) {
+                       if (strlen($part) <> mb_strlen($part)) {
+                               $parts2[] = rawurlencode($part);
+                       } else {
+                               $parts2[] = $part;
+                       }
+               }
+               $parts['path'] =  implode('/', $parts2);
+               $url = self::unparseURL($parts);
+
                if (self::isUrlBlocked($url)) {
                        logger('domain of ' . $url . ' is blocked', LOGGER_DATA);
                        return $ret;
@@ -217,7 +252,7 @@ class Network
 
                        $newurl = $curl_info['redirect_url'];
 
-                       if (($new_location_info['path'] == '') && ( $new_location_info['host'] != '')) {
+                       if (empty($new_location_info['path']) && !empty($new_location_info['host'])) {
                                $newurl = $new_location_info['scheme'] . '://' . $new_location_info['host'] . $old_location_info['path'];
                        }
 
@@ -229,6 +264,11 @@ class Network
                        if (strpos($newurl, '/') === 0) {
                                $newurl = $old_location_info["scheme"]."://".$old_location_info["host"].$newurl;
                        }
+                       $old_location_query = @parse_url($url, PHP_URL_QUERY);
+
+                       if ($old_location_query != '') {
+                               $newurl .= '?' . $old_location_query;
+                       }
 
                        if (filter_var($newurl, FILTER_VALIDATE_URL)) {
                                $redirects++;
@@ -372,7 +412,7 @@ class Network
                        $matches = [];
                        $new_location_info = @parse_url($curl_info['redirect_url']);
                        $old_location_info = @parse_url($curl_info['url']);
-       
+
                        preg_match('/(Location:|URI:)(.*?)\n/', $header, $matches);
                        $newurl = trim(array_pop($matches));
 
@@ -429,7 +469,7 @@ class Network
                /// @TODO Really suppress function outcomes? Why not find them + debug them?
                $h = @parse_url($url);
 
-               if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME + DNS_PTR) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
+               if ((is_array($h)) && (@dns_get_record($h['host'], DNS_A + DNS_CNAME) || filter_var($h['host'], FILTER_VALIDATE_IP) )) {
                        return $url;
                }
 
@@ -454,7 +494,10 @@ class Network
 
                $h = substr($addr, strpos($addr, '@') + 1);
 
-               if (($h) && (dns_get_record($h, DNS_A + DNS_CNAME + DNS_PTR + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
+               if (($h) && (dns_get_record($h, DNS_A + DNS_MX) || filter_var($h, FILTER_VALIDATE_IP) )) {
+                       return true;
+               }
+               if (($h) && dns_get_record($h, DNS_CNAME + DNS_MX)) {
                        return true;
                }
                return false;
@@ -613,7 +656,7 @@ class Network
        public static function stripTrackingQueryParams($url)
        {
                $urldata = parse_url($url);
-               if (is_string($urldata["query"])) {
+               if (!empty($urldata["query"])) {
                        $query = $urldata["query"];
                        parse_str($query, $querydata);
 
@@ -797,14 +840,35 @@ class Network
                        return "";
                }
 
+               if (empty($parts1["scheme"])) {
+                       $parts1["scheme"] = '';
+               }
+               if (empty($parts2["scheme"])) {
+                       $parts2["scheme"] = '';
+               }
+
                if ($parts1["scheme"] != $parts2["scheme"]) {
                        return "";
                }
 
+               if (empty($parts1["host"])) {
+                       $parts1["host"] = '';
+               }
+               if (empty($parts2["host"])) {
+                       $parts2["host"] = '';
+               }
+
                if ($parts1["host"] != $parts2["host"]) {
                        return "";
                }
 
+               if (empty($parts1["port"])) {
+                       $parts1["port"] = '';
+               }
+               if (empty($parts2["port"])) {
+                       $parts2["port"] = '';
+               }
+
                if ($parts1["port"] != $parts2["port"]) {
                        return "";
                }
@@ -815,14 +879,21 @@ class Network
                        $match .= ":".$parts1["port"];
                }
 
+               if (empty($parts1["path"])) {
+                       $parts1["path"] = '';
+               }
+               if (empty($parts2["path"])) {
+                       $parts2["path"] = '';
+               }
+
                $pathparts1 = explode("/", $parts1["path"]);
                $pathparts2 = explode("/", $parts2["path"]);
 
                $i = 0;
                $path = "";
                do {
-                       $path1 = $pathparts1[$i];
-                       $path2 = $pathparts2[$i];
+                       $path1 = defaults($pathparts1, $i, '');
+                       $path2 = defaults($pathparts2, $i, '');
 
                        if ($path1 == $path2) {
                                $path .= $path1."/";