Changeset 163

Timestamp:

03/11/08 14:29:06 (5 years ago)

Author:

Seldaek

Message:

Url: Should fix the choking bug on very slowly loading pages (thanks scoates for the proof of concept)

Files:

• trunk/Phergie/Plugin/Url.php (modified) (7 diffs)

trunk/Phergie/Plugin/Url.php

@@ -21 +21 @@
     */
     protected $format = '%nick%: %title% [ %link% ]';
+    protected $titleOnlyFormat = '%title%';
 
         /**
@@ -50 +51 @@
         if ($format) {
             $this->format = $format;
+        }
+        $titleOnlyFormat = $this->getPluginIni('title_only_format');
+        if ($titleOnlyFormat) {
+            $this->titleOnlyFormat = $titleOnlyFormat;
         }
     }
@@ -76 +81 @@
                 $opts = array('http' =>
                     array(
-                        'timeout' => 4,
+                        'timeout' => 3.5,
                         'method' => 'GET',
                         'user_agent' => 'Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.12) Gecko/20080201 Firefox/2.0.0.12'
@@ -93 +98 @@
                     if (!isset ($title)) {
                         $content = '';
-                        while ($chunk = fread($page, 512)) {
+                        $tstamp = time()+4;
+                        while ($chunk = fread($page, 1)) {
                             $content .= $chunk;
+                            // Check for timeout
+                            if(time() > $tstamp)
+                                break;
+                            // Skip checks unless we read 100 bytes, no need to kill the cpu
+                            if (!(strlen($content) % 100)) {
+                                continue;
+                            }
+                            // Try to read title
                             if (preg_match('#<title[^>]*>([^<]*)#is', $content, $m)) {
-                                $content .= fread($page, 256);
+                                // Start another loop to grab some more data in order to be sure we have the complete title
+                                $loop = 120;
+                                        while (($chunk = fread($page, 1)) && $loop--) {
+                                            $content .= $chunk;
+                                            // Check for timeout
+                                            if(time() > $tstamp)
+                                                break;
+                                        }
                                 preg_match('#<title[^>]*>([^<]*)#is', $content, $m);
                                 $title = preg_replace('#\s+#', ' ', $m[1]);
-                                $title = $this->decode($title, $titleLength);
+                                $title = trim($this->decode($title, $titleLength));
                                 break;
                             }
+                            // Title won't appear beyond that point so stop parsing
                             if (preg_match('#</head>|<body#i', $content)) {
                                 break;
@@ -110 +132 @@
                 }
 
-                if (!isset($title)) {
+                if (!isset($title) || empty($title)) {
                         if ($tinyUrl === $url) {
                                 unset($tinyUrl, $url);
@@ -118 +140 @@
                 }
 
+
+                // Send url and title
                 $this->doPrivmsg(
                     $this->event->getSource(),
@@ -148 +172 @@
         // and minimize the size of the cache for less cache bloat.
         $url = dechex(crc32($url)); $tiny = dechex(crc32($tiny));
-        $cache = array('url' => $this->urlCache[$source][$url], 'tiny' => $this->tinyCache[$source][$tiny]);
+        $cache = array
+        (
+                        'url' => isset($this->urlCache[$source][$url]) ? $this->urlCache[$source][$url] : null,
+                'tiny' => isset($this->tinyCache[$source][$tiny]) ? $this->tinyCache[$source][$tiny] : null
+        );
 
         $expire = $this->expire;