root/trunk/Phergie/Plugin/Url.php @ 59

<?php

/**
* @see Phergie_Event_Handler
*/
require_once 'Phergie/Event/Handler.php';

/**
* Monitors incoming messages for instances of URLs and responds with messages
* containing relevant information about detected URLs.
*/
class Phergie_Plugin_Url extends Phergie_Event_Handler
{
    /**
    * Links output format
    *
    * Can use the variables %title% and %link% in it to display page titles
    * and links
    *
    * @var string
    */
    protected $format = '%title% [ %link% ]';

    /**
    * Initializes settings
    *
    * @return void
    */
    public function init()
    {
        $format = $this->getIni('format');
        if ($format) {
            $this->format = $format;
        }
    }

    /**
    * Checks an incoming message for the presence of a URL and, if one is
    * found, responds with its title if it is an HTML document and the
    * TinyURL equivalent of its original URL if it meets length requirements.
    *
    * @return void
    */
    public function onPrivmsg()
    {
        // URL Match
        if (preg_match('#(https?://(?:[a-z0-9_-]+\.)+[a-z]{2,6}[^\s]*)#is', $event->getArgument(1), $m)) {
            $url = rtrim($m[1], '), ]');
            // @todo if image or something, output content type
            $tinyUrl = $this->tinyUrl($url);
            $title = '';
            $titleLength = $this->getIni('title_length');
            if ($page = @fopen($url, 'r')) {
                $content = '';
                while ($chunk = fread($page, 512)) {
                    $content .= $chunk;
                    if (preg_match('#<title[^>]*>([^<]*)#is', $content, $m)) {
                        $title = $this->decode($m[1], $titleLength);
                        break;
                    }
                    if (preg_match('#</head>|<body#i', $content)) {
                        break;
                    }
                }
                fclose($page);
            }

            $this->doPrivmsg(
                $this->getSource(),
                str_replace(
                    array('%title%', '%link%'),
                    array($title, $tinyUrl),
                    $this->format
                )
            );
        }
    }

    /**
    * Transliterates a UTF-8 string into corresponding ASCII characters and
    * truncates and appends an ellipsis to the string if it exceeds a given
    * length.
    *
    * @param string $str String to decode
    * @param int $trim Maximum string length, optional
    * @return string
    */
    protected function decode($str, $trim=null)
    {
        $out = utf8_decode($str);
        $out = html_entity_decode($out, ENT_QUOTES);
        $out = strtr($out, 'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', 'AAAAAAACEEEEIIIIDNOOOOOOUUUUYPYaaaaaaaceeeeiiiidnoooooouuuuypy');
        $out = preg_replace('{[^a-z0-9&|"#\'\{\}()§^!°\[\]$*¨µ£%´`~=+:/;.,?><\\ _-]}i', '', $out);
        if($trim > 0) {
            $out = substr($out, 0, $trim) . (strlen($out) > $trim ? '...' : '');
        }
        return $out;
    }
}