root/trunk/Phergie/Plugin/Url.php @ 145

<?php

/**
* @see Phergie_Plugin_Abstract_Base
*/
require_once 'Phergie/Plugin/Abstract/Base.php';

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

    /**
    * Url cache to prevent spamming, especially with multiple bots on the same channel
    */
    protected $urlCache = array();
    protected $tinyCache = array();

    /**
    * The time in seconds to store the cached entries
    * Setting it to 0 or below disables the cache expiration
    */
    protected $expire = 1800;

    /**
    * The number of entries to keep in the cache at one time per channel
    * Setting it to 0 or below disables the cache limit
    */
    protected $limit = 10;

    /**
    * Initializes settings
    *
    * @return void
    */
    public function init()
    {
        $format = $this->getPluginIni('format');
        if ($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_all('#(https?://(?:[a-z0-9_-]+\.)+[a-z0-9]{1,6}[^\s]*)#is', $this->event->getArgument(1), $matches, PREG_SET_ORDER)) {
            $titleLength = $this->getPluginIni('title_length');
            foreach ($matches as $m) {
                $url = rtrim($m[1], '), ].?!');
                // Convert url
                $tinyUrl = $this->tinyUrl($url);

                // Prevent spamfest
                if($this->checkURLCache($url, $tinyUrl))
                    return;

                $opts = array('http' =>
                    array(
                        'timeout' => 4,
                        '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'
                    )
                );
                $context = stream_context_create($opts);

                if ($page = @fopen($url, 'r', false, $context)) {
                    $data = stream_get_meta_data($page);
                    foreach ($data['wrapper_data'] as $header) {
                        if (preg_match('/^Content-Type: ([^;]+)/', $header, $match)
                            && ! preg_match('#^(text/x?html|application/xhtml+xml)$#', $match[1])) {
                            $title = $match[1];
                        }
                    }
                    if (!isset ($title)) {
                        $content = '';
                        while ($chunk = fread($page, 512)) {
                            $content .= $chunk;
                            if (preg_match('#<title[^>]*>([^<]*)#is', $content, $m)) {
                                $content .= fread($page, 256);
                                preg_match('#<title[^>]*>([^<]*)#is', $content, $m);
                                $title = preg_replace('#\s+#', ' ', $m[1]);
                                $title = $this->decode($title, $titleLength);
                                break;
                            }
                            if (preg_match('#</head>|<body#i', $content)) {
                                break;
                            }
                        }
                    }
                    fclose($page);
                }

                if (!isset($title)) {
                    if ($tinyUrl === $url) {
                        unset($tinyUrl, $url);
                        return;
                    }
                    $title = '[ No Title ]';
                }

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

                // Update cache
                $this->updateURLCache($url, $tinyUrl);
                unset($title, $tinyUrl, $title);
            }
        }
    }

    /**
    * Checks a given URL and TinyURL against the cache to verify if they were
    * previously posted on the channel.
    *
    * @param string $url The URL to check against
    * @param string $tiny The TinyURL to check against
    * @return bool
    */
    protected function checkURLCache($url, $tiny) {
        $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));
        $cache = array('url' => $this->urlCache[$source][$url], 'tiny' => $this->tinyCache[$source][$tiny]);

        $expire = $this->expire;
        // If cache expiration is enabled, check to see if the given url has expired in the cache
        // If expire is disabled, simply check to see if the url is listed
        if (($expire > 0 && (($cache['url'] + $expire) > time() || ($cache['tiny'] + $expire) > time())) ||
            ($expire <= 0 && (isset($cache['url']) || isset($cache['tiny'])))) {
            unset($cache, $url, $tiny, $expire);
            return true;
        }
        unset($cache, $url, $tiny, $expire);
        return false;
    }

    /**
    * Updates the cache and adds the given URL and TinyURL to the cache. It
    * also handles cleaning the cache of old entries as well.
    *
    * @param string $url The URL to add to the cache
    * @param string $tiny The TinyURL to add to the cache
    * @return bool
    */
    protected function updateURLCache($url, $tiny) {
        $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));
        $time = time();

        // Handle the URL cache and remove old entries that surpass the limit if enabled
        $this->urlCache[$source][$url] = $time;
        if ($this->limit > 0 && count($this->urlCache[$source]) > $this->limit) {
            asort($this->urlCache[$source], SORT_NUMERIC);
            array_shift($this->urlCache[$source]);
        }

        // Handle the TinyURL cache and remove old entries that surpass the limit if enabled
        $this->tinyCache[$source][$url] = $time;
        if ($this->limit > 0 && count($this->tinyCache[$source]) > $this->limit) {
            asort($this->tinyCache[$source], SORT_NUMERIC);
            array_shift($this->tinyCache[$source]);
        }
        unset($url, $tiny, $time);
    }

    /**
    * 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 = $this->decodeTranslit($str);
        if($trim > 0) {
            $out = substr($out, 0, $trim) . (strlen($out) > $trim ? '...' : '');
        }
        return $out;
    }
}