<?php
/**
 * Dictionary: URL
 *
 * $Id$
 *
 * @author gERD Schaufelberger <gerd@php-tools.net>
 * @license PHP License
 * @package WB
 * @subpackage base
 */

WBClass::load('WBDictionary'
        , 'WBString');

/**
 * Dictionary: URL
 *
 * Manage URLs
 *
 * @version 0.2.3
 * @package WB
 * @subpackage
 */
class WBDictionary_URL extends WBDictionary
{
    /**
     * Table access
     * @var WBDatasource_Table
     */
    protected $table;

    /**
     * additional data to populate record
     *
     * This array will be extended by primary key of vfsfile
     *
     * @see init()
     * @var array
     */
    protected $mergeFields   =   array(
                                            'title',
                                            'description',
                                            'headlines'
                                            );
    static private $wordCache   =   array();

    /**
     * 2nd constructor
     *
     */
    public function init()
    {
        $this->table            =   WBClass::create('WBDatasource_Table');
        $this->mergeFields[]    =   $this->table->getIdentifier('vfsfile');
    }

    /**
     * find entry in dictionary
     *
     * Actual look up data in dictionary
     *
     * @param array $data
     * @return array|null
     */
    protected function find($data)
    {
        $cacheId    =   $this->getCacheId($data);
        $cache      =   $this->getCache($cacheId);
        if (!empty($cache)) {
            return $cache;
        }

        $clause =   array();
        foreach ($data as $field => $value) {
            $clause[]   =   array(
                                    'field' =>  $field,
                                    'value' =>  $value
                                );
        }

        $row    =   $this->table->get('url', null, null, $clause);
        if (0 == count($row)) {
            return null;
        }

        $this->setCache($cacheId, $row[0]);
        return $row[0];
    }

    /**
     * load record by id
     *
     * Restore dictionary object from table
     *
     * @param string $id
     */
    public function load($id)
    {
        if ($this->id == $id) {
            return;
        }
        $data   =   $this->table->get('url', $id);
        if (count($data) != 1) {
            WBClass::load('WBException_Argument');
            throw new WBException_Argument('Failed to load dictionary record URL by id ' . $id . '!', 1, __CLASS__);
        }
        $data       =   $data[0];
        $this->data =   $data;
        $this->id   =   $data['id'];
        $this->word =   $data['path'];

        switch ($data['protocol']) {
            case 'mailto':
                $this->word =   sprintf('%s@%s', $data['path'], $data['host']);
                break;

            case 'self':
                break;

            case 'javascript':
                $this->word =   sprintf('javascript:%s', $data['path']);
                break;

            default:
                $this->word =   sprintf('%s://%s%s', $data['protocol'], $data['host'], $data['path']);
                break;

        }

        $cacheId    =   $this->getCacheId($data);
        $this->setCache($cacheId, $data);
    }

    /**
     * extract dictionary data from word
     *
     * Well, in URL dictionary, a word is a URL. This method extracts
     * protocol, hostname and path from URL. URL also may be e-mail addresses
     *
     * @see include/WB/WBDictionary#explode()
     * @param string $word
     * @return array
     */
    protected function explode($word)
    {
        $word   =   trim($word);
        if (empty($word)) {
            return $this->explodeEmpty();
        }

        if (0 == strncmp('javascript:', $word, 11)) {
            return $this->explodeJavascript($word);
        }
        if (strstr($word, '@')) {
            return $this->explodeEMail($word);
        }
        return $this->explodeUrl($word);
    }

    /**
     * Convert Empty String to URL-Data
     *
     * @return array
     */
    private function explodeEmpty()
    {
        $url    =   array(
                            'protocol'  =>  '',
                            'host'      =>  '',
                            'path'      =>  ''
                    );
        return $url;
    }

    /**
     * Convert javascript Call to URL-Data
     *
     * @param string $word
     * @return array
     */
    private function explodeJavascript($word)
    {
        $word   =   substr($word, 11);
        $url    =   array(
                            'protocol'  =>  'javascript',
                            'host'      =>  '',
                            'path'      =>  $word
                    );
        return $url;
    }

    /**
     * convert e-mail address to URL data
     *
     * @todo validate e-mail address properly
     * @param string $word
     * @return array
     */
    private function explodeEMail($word)
    {
        $word   =   explode('@', $word, 2);
        $url    =   array(
                            'protocol'  =>  'mailto',
                            'host'      =>  $word[1],
                            'path'      =>  $word[0]
                    );
        return $url;
    }

    /**
     * explode standard URLs
     *
     *
     *
     * @todo check for username and password in URL
     * @see explodeServiceUrl()
     * @param string $word
     * @return array
     */
    private function explodeUrl($word)
    {
        $url    =   array(
                            'protocol'  =>  '[[PROTOCOL]]',
                            'host'      =>  '[[SERVER]]',
                            'path'      =>  '/'
                    );

        $word   =   WBString::replaceSuperPlaceholders($word);

        // prepend SERVER to local URLs
        if ('/' == $word[0]) {
            $word   =   WBString::replaceSuperPlaceholders('[[PROTOCOL]]://[[SERVER]]') .  $word;
        }

        // lazy www.-url
        if (0 == strncmp('www.', $word, 4)) {
            $word   =   'http://' . $word;
        }

        // URL points to this site?
        $self   =   WBString::replaceSuperPlaceholders('[[PROTOCOL]]://[[SERVER]][[DOCROOT]]');
        if (strncmp($self, $word, strlen($self)) == 0) {
            $word   =   substr($word, strlen($self));
            return $this->explodeServiceUrl($word);
        }

        if (preg_match('|^(\\w+)://([^/]*)|', $word, $match)) {
            $url['protocol']    =   $match[1];
            $url['host']        =   $match[2];
            $word               =   substr($word, strlen($url['protocol'] . '://' . $url['host']));
            if (0 === $word || empty($word)) {
                $word = '/';
            }
        }
        else {
            $url['protocol']    =   'string';
            $url['host']        =   '';
        }

        $url['path']    =   $word;
        return $url;
    }

    /**
     * explode rest of path from URL for "local" services
     *
     * Local services are definde in main configuration in secrtion service
     * and mapped to strings like [[SERVICE_HTML]]. This method tries to
     * find any [[SERVICE_ path and replaces the URL path with the
     * corresponding super placeholder. Therefore no real URLs or pathes must
     * be saved in URL dictionary for well knwon services.
     *
     * @param $word
     * @return array
     */
    protected function explodeServiceUrl($word)
    {
        $url    =   array(
                            'protocol'  =>  'self',
                            'host'      =>  '',
                            'path'      =>  '/'
                    );
        if (empty($word)) {
            return $url;
        }

        $sph                =   WBString::getSuperPlaceHolder();

        // docroot must be removed
        $docroot            =   $sph['map']['docroot'];
        $docrootLen         =   strlen($docroot);

        $found  =   false;

        foreach ($sph['replace'] as $i => $r) {
            if (strncmp($sph['search'][$i], '[[SERVICE_', 10) != 0) {
                continue;
            }

            // is URL within docroot?
            if (!strncmp($r, $docroot, $docrootLen) == 0) {
                continue;
            }

            $r  =   substr($r, $docrootLen);
            $rt =   $r;
            // the base URL also works without slash at the end
            if (strlen($word) < strlen($r)) {
                $rt =   rtrim($rt, '/');
            }
            if (strncmp($rt, $word, strlen($rt)) != 0) {
                continue;
            }

            // remove actual path and use placeholder
            $word   =   substr($word, strlen($r));
            $word   =   $sph['search'][$i] . $word;
            $found  =   true;
            break;
        }

        // try SELF_NO_LANG if SERVICE_ failed
        if (!$found) {
            $r  =   substr($sph['map']['self_no_lang'], $docrootLen);
            if (strncmp($r, $word, strlen($r)) == 0) {
                $word   =   substr($word, strlen($r));
                $word   =   '[[SELF_NO_LANG]]' . $word;
            }
        }

        $url['path']    =   $word;
        return $url;
    }

    /**
     * add record to dictionary
     *
     * @see include/WB/WBDictionary#insert()
     * @param array record to save
     * @return string record id
     */
    protected function insert($data)
    {
        $cacheId    =   $this->getCacheId($data);
        $id         =   $this->table->save('url', '__new', $data);
        $data['id'] =   $id;
        $this->setCache($cacheId, $data);
        return $id;
    }

    /**
     * save additional data to dictionary record
     *
     * This is used to populate the dictionary record
     *
     * @see populate()
     * @see $mergeFields
     * @param array $data
     * @return array
     */
    protected function merge($save)
    {
        $this->table->save('url', $this->id, $save);
        return $save;
    }

    /**
     * automatical populate function
     *
     * Surf to URL and extract title, description as well as
     *
     * @param string $id
     * @see include/WB/WBDictionary#autoPopulate()
     */
    public function autoPopulate($id = null)
    {
        if ($id) {
            $this->load($id);
        }

        $url    =   $this->word;
        $tmp    =   null;
        switch ($this->data['protocol']) {

            case 'self':
                return;
                // @todo build propper URL
                $url    =   WBString::replaceSuperPlaceholders('http://[[SERVER]]' . $url);
                // fall through

            case 'http':
            case 'https':
                $tmp    =   $this->download($url);
                break;

            default:
                // don't do anything for other protocols
                break;

        }

        if (!$tmp) {
            return;
        }

        $mime   =   $tmp->getMimeType();
        if ($mime[0] != 'text' && $mime[1] != 'xml') {
            return;
        }

        $save   =   array();
        $cnt    =   file_get_contents($tmp->realpath());

        if (preg_match( '#<title>(.*)</title>#iU', $cnt, $match)) {
            $save['title']  =   trim($match[1]);
        }

        if (preg_match_all( '#<(h1|h2|h3|h4)[^>]*>(.*)</(\\1)>#i', $cnt, $matches, PREG_SET_ORDER)) {
            $blurb  =   array();
            foreach ($matches as $match) {
                $blurb[]    =   str_replace("\n", ' ', strip_tags($match[2]));
            }
            $save['headlines']  =   implode("\n", $blurb);
        }

        $this->populate($save);

        // schedule for taking screen shot of web site
        $queue  =   array(
                        $this->table->getIdentifier('urlscreenshotqueue')  =>  $this->id
                    );
        $this->table->save('urlscreenshotqueue', '__new', $queue);
    }

    /**
     * download to temporary file
     *
     *
     *
     * @param string $url
     * @return WBFile
     */
    protected function download($url)
    {
        $tmp    =   WBClass::create('WBFile');
        $tmp->tempnam('url');

        WBClass::load('WBStream');
        $fin    =   WBStream::open($url, 'r');
        $fout   =   fopen($tmp->realpath(), 'w');

        while (!feof($fin)) {
            $buffer =   fgets($fin, 8192);
            fputs($fout, $buffer, strlen($buffer));
        }

        fclose($fin);
        fclose($fout);
        return $tmp;
    }

    /**
     * Build cacheId from data
     *
     * @return string
     */
    private function getCacheId($data)
    {
        return sprintf('%s://%s/%s', $data['protocol'], $data['host'], $data['path']);
    }

    /**
     * Get cached data
     *
     * @param string $cacheId
     * @return array|null
     */
    private function getCache($cacheId)
    {
        if (isset(self::$wordCache[$cacheId])) {
            return self::$wordCache[$cacheId];
        }
        return null;
    }

    /**
     * store in cache
     *
     * @param string $cacheId
     * @param array $data
     */
    private function setCache($cacheId, $data)
    {
        self::$wordCache[$cacheId]  =   $data;
    }
}