Changeset 168

Timestamp:

03/12/08 23:23:54 (5 years ago)

Author:

Slynderdale

Message:

Added a couple bug fixes and improvements. Now when someone posts a url using the https scheme, it will check to see if openssl is loaded before processing that url, otherwise it skips it. Also originally, if someone posted a url that was in the cache, the URL would halt at that url even if there were others after it in the same message, now it simply skips the cached urls. Also the url is now cleaned, stripped of its scheme and anchors before checking and storing it to the cache.

Files:

• trunk/Phergie/Plugin/Url.php (modified) (8 diffs)

trunk/Phergie/Plugin/Url.php

@@ -78 +78 @@
             foreach ($matches as $m) {
                 $url = rtrim($m[1], '), ].?!');
+                $scheme = strtolower(parse_url($url, PHP_URL_SCHEME));
+                if ($scheme == 'https' && !extension_loaded('openssl'))
+                        continue;
+
                 // Convert url
                 $tinyUrl = $this->tinyUrl($url);
@@ -83 +87 @@
                                 // Prevent spamfest
                                 if($this->checkURLCache($url, $tinyUrl))
-                                        return;
+                                        continue;
 
                 $opts = array('http' =>
@@ -94 +98 @@
                 $context = stream_context_create($opts);
 
+                $this->debug("Opening Stream for $url");
                                 $previousHandler = set_error_handler(array($this, 'errorHandler'));
                 if ($page = fopen($url, 'r', false, $context)) {
+                        $this->debug("Opened Stream for $url");
                     $data = stream_get_meta_data($page);
                     foreach ($data['wrapper_data'] as $header) {
@@ -181 +187 @@
         $source = $this->event->getSource();
 
-        // Transform the URL and TinyURL into a HEX CRC32 checksum to prevent potential problems
-        // and minimize the size of the cache for less cache bloat.
-        $url = dechex(crc32($url)); $tiny = dechex(crc32($tiny));
+        /**
+        * Transform the URL and TinyURL into a HEX CRC32 checksum to prevent potential problems
+        * and minimize the size of the cache for less cache bloat.
+        */
+        $url = $this->getURLChecksum($url);  $tiny = $this->getURLChecksum($tiny);
         $cache = array
         (
@@ -191 +199 @@
 
         $expire = $this->expire;
-        // If cache expiration is enabled, check to see if the given url has expired in the cache
-        // If expire is disabled, simply check to see if the url is listed
+        /**
+        * If cache expiration is enabled, check to see if the given url has expired in the cache
+        * If expire is disabled, simply check to see if the url is listed
+        */
         if (($expire > 0 && (($cache['url'] + $expire) > time() || ($cache['tiny'] + $expire) > time())) ||
             ($expire <= 0 && (isset($cache['url']) || isset($cache['tiny'])))) {
@@ -213 +223 @@
         $source = $this->event->getSource();
 
-        // Transform the URL and TinyURL into a HEX CRC32 checksum to prevent potential problems
-        // and minimize the size of the cache for less cache bloat.
-        $url = dechex(crc32($url)); $tiny = dechex(crc32($tiny));
+        /**
+        * Transform the URL and TinyURL into a HEX CRC32 checksum to prevent potential problems
+        * and minimize the size of the cache for less cache bloat.
+        */
+        $url = $this->getURLChecksum($url);  $tiny = $this->getURLChecksum($tiny);
         $time = time();
 
@@ -252 +264 @@
 
         /**
-         * Custom error handler meant to handle 404 errors and such
-         */
+        * Custom error handler meant to handle 404 errors and such
+        */
         public function errorHandler($errno, $errstr, $errfile, $errline)
         {
@@ -263 +275 @@
             return false;
         }
+
+        /**
+        * Takes a url, parses and cleans the URL without of all the junk
+        * and then return the hex checksum of the url.
+        */
+        public function getURLChecksum($url) {
+                $parsed = parse_url($url);
+                if (is_array($parsed)) {
+                    $url = $parsed['host'];
+                    $url .= isset($parsed['port']) ? ':'.$parsed['port'] : '';
+                    if(isset($parsed['path']))
+                    {
+                        $url .= (substr($parsed['path'], 0, 1) == '/') ? $parsed['path'] : ('/'.$parsed['path']);
+                    }
+                    $url .= isset($parsed['query']) ? '?'.$parsed['query'] : '';
+                }
+                return dechex(crc32($url));
+        }
 }