<?php
/**
 * Content component: SearchEngineOptimizer
 *
 * $Id$
 *
 * @author gERD Schaufelberger <gerd@php-tools.net>
 * @license PHP License
 * @package WB
 * @subpackage content
 */

/**
 * Load base class
 */
WBClass::load('WBContent');

/**
 * Content component: SearchEngineOptimizer
 *
 * Display robots.txt and sitemap
 *
 * @version 0.4.0
 * @package WB
 * @subpackage content
 */
class WBContent_SearchEngineOptimizer extends WBContent
{
    /**
     * my parameter list
     *
     * - 'process' select what to display (sitemap, robots.txt)
     * - 'page' whether to add pages to sitemap (0,1)
     * - 'blog' add blog articles to sitemap (blog/%day/%title)
     * - 'blogcategory' list of blog categories (optional)
     * - 'gallery' add gallery pages to sitemap (gallery/%obscureid/%title)
     * - 'vfs' add media files page to sitemap (vfs/%obscureid/%name)
     * - 'vfsmimemajor' limit to file types - don't care for mimetype if empty
     *
     * @var array
     */
    protected $config = array(
            'process'       =>  'sitemap',
            'page'          =>  1,
            'blog'          =>  '', // blog/%day/%title
            'blogcategory'  =>  array(),
            'gallery'       =>  '', // gallery/%obscureid/%title
            'vfs'           =>  '', // vfs/%obscureid/%name
            'vfsmimemajor'  =>  array('video', 'image')
        );

    /**
     * url template for google sitemap
     *
     * @var string
     */
    protected $url = "<url><loc>http://[[SERVER]][[SERVICE_HTML]]%s</loc><priority>%1.2f</priority></url>\n";

    /**
     * list of found or not to add (to sitemap) pages
     * @var array
     */
    protected $pages    =   array(
                                    '/robot'        =>  true,
                                    '/robots.txt'   =>  true,
                                    '/sitemap'      =>  true,
                                    '/sitemap.xml'  =>  true
                                );
	/**
	 * current menu level
	 * @var int
	 */
	protected $level	=	0;

    /**
     * vfs gallery
     * @var WBVFS_Gallery
     */
    protected $gallery;

    /**
     * Table access
     * @var WBDatasource_Table
     */
    private $table;

    /**
     * config loader
     * @var WBConfig_Loader
     */
    private $configLoader;

    /**
     * run
     *
     * run component
     *
     * @return array parameter list
     */
    public function run()
    {
        switch (strtolower($this->config['process'])) {
            case 'robots.txt';
            case 'robots';
                $this->displayRobots();
                break;

            case 'sitemap':
            default:
                $this->displaySitemap();
                break;
        }
        return $this->config;
    }

    /**
     * create standard robots.txt
     *
     * @link http://www.robotstxt.org/
     * @uses display()
     */
    protected function displayRobots()
    {
        $cnt    =   <<<EOT
User-agent: *
Disallow: [[SERVICE_CSS]]
Disallow: [[SERVICE_JAVASCRIPT]]
Allow: [[SERVICE_JAVASCRIPT]]WB/player.swf
EOT;
        $this->display($cnt, 'text/plain');
    }

    /**
     * create standard google sitemap
     *
     * @link http://www.sitemaps.org/
     * @uses display()
     */
    protected function displaySitemap()
    {
		// reset locale settings to make sure decimal point works properly
		setlocale(LC_ALL, null);

        $cnt    =   <<<EOT
<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:image="http://www.google.com/schemas/sitemap-image/1.1"
        xmlns:video="http://www.google.com/schemas/sitemap-video/1.1">

EOT;

        $this->addPage2Sitemap($cnt);
        $this->addBlog2Sitemap($cnt);
        $this->addGallery2Sitemap($cnt);
        $this->addVFS2Sitemap($cnt);

        $cnt    .=  "</urlset>\n";
        $this->display($cnt, 'text/xml');
    }

    /**
     * add blog article pages
     *
     * In case there are special blog article pages, list recent 100 article URLs
     *
     * See config option 'blog. If this parameter is empty, nothing will be added to
     * the sitemap. Otherwise path to each article is used. The placeholder %id and
     * %title are replaced with article's id and title.
     *
     * @param string $cnt
     */
    protected function addBlog2Sitemap(&$cnt)
    {
        if (empty($this->config['blog'])) {
            return;
        }

        // basic blog path
        $path   =   $this->config['blog'];
        $path   =   str_replace(array('%id', '%title'), '', $path);
        $path   =   trim($path, '/');

        $level  =   count(explode('/', $path));

        // priority of blog article pages
        $prio   =   1 - ($level / 10);
        $path   =   $this->config['blog'];

        // fetch recent 100 acticles
        if (!$this->table) {
            $this->table    =   WBClass::create('WBDatasource_Table');
        }
        $opt    =   array(
                        'limit'     =>  500,
                        'column'    =>  array('title', 'created')
                        );
        $clause =   array();

        if (!empty($this->config['blogcategory'])) {
            if (!is_array($this->config['blogcategory'])) {
                $this->config['blogcategory']   =   array($this->config['blogcategory']);
            }
            $clause[]   =   array(
                                'field'     =>  $this->table->getIdentifier('blogcategory'),
                                'relation'  =>  'in',
                                'value'     =>  $this->config['blogcategory']
                            );
        }

        $list   =   $this->table->get('blog', null, null, $clause, $opt);
        $search =   array(
                        '%id',
                        '%day',
                        '%title'
                    );
        foreach ($list as $l) {
            $replace    =   array(
                                $l['id'],
                                date('Y-m-d', strtotime($l['created'])),
                                urlencode($l['title'])
                            );
            $p      =   str_replace($search, $replace, $path);
            $cnt    .=  sprintf($this->url, $p, $prio);
        }
        WBDatasource_Table::flushCache();
    }

    /**
     * add list of galleries to sitemap
     *
     * In case there automatically created special pages for gallery listings,
     * load list of user galleries and add them to sitemap
     *
     * @param string $cnt
     */
    protected function addGallery2Sitemap(&$cnt)
    {
        if (empty($this->config['gallery'])) {
            return;
        }

        // basic blog path
        $path   =   $this->config['gallery'];
        $path   =   str_replace(array('%obscureid', '%title'), '', $path);
        $path   =   trim($path, '/');

        $level  =   count(explode('/', $path));

        // priority of gallery list page
        $prio   =   1 - ($level / 10);
        $path   =   $this->config['gallery'];

        $options        =   array();
        $this->gallery  =   WBClass::create('WBVFS_Gallery', $options);

        $search =   array(
                        '%obscureid',
                        '%title'
                    );
        $list   =   $this->gallery->getList();
        foreach ($list as $l) {
            $replace    =   array(
                                $l['obscureid'],
                                urlencode($l['title'])
                            );
            $p      =   str_replace($search, $replace, $path);
            $cnt    .=  sprintf($this->url, $p, $prio);
        }
        WBDatasource_Table::flushCache();
    }

    /**
     * add media files
     *
     * @param string $cnt
     */
    protected function addVFS2Sitemap(&$cnt)
    {
        if (empty($this->config['vfs'])) {
            return;
        }

        if (!$this->table) {
            $this->table    =   WBClass::create('WBDatasource_Table');
        }

        $opt    =   array(
                        'limit'     =>  500,
                        'order'     =>  array(
                                            'field' =>  'created',
                                            'asc'   =>  0
                                        )
                        );
        $clause     =   array();

        // description is required
        $clause[]   =   array(
                            'field'     =>  'description',
                            'relation'  =>  'not',
                            'value'     =>  ''
                        );
        // just a view mime-types
        if (!empty($this->config['vfsmimemajor'])) {
            if (!is_array($this->config['mimemajor'])) {
                $this->config['mimemajor']  =   array($this->config['mimemajor']);
            }

            $clause[]   =   array(
                                    'field'     =>  'mimemajor',
                                    'relation'  =>  'in',
                                    'value'     =>  $this->config['vfsmimemajor']
                                );
        }
        $list   =   $this->table->getIds('vfsfile', null, $clause, $opt);
        $search =   array(
                        '%obscureid',
                        '%name'
                    );

        $file   =   WBClass::create('WBVFS_File');
        /** @var $file WBVFS_File */

        foreach ($list as $l) {
            $file   =   $file->loadById($l);
            if (!$file->isOK()) {
                continue;
            }

            /**
             * @todo check for file references
             */

            $major  =   $file->getMime(WBVFS_File::MIME_MAJOR);

            // add vfs file view page
            $replace    =   array(
                                $file->getObscureId(),
                                urlencode($file->getName())
                            );

            $m      =   array('<url>');
            $m[]    =   sprintf('  <loc>http://[[SERVER]][[SERVICE_HTML]]%s</loc>', str_replace($search, $replace, $this->config['vfs']));
            $m[]    =   sprintf('  <%s:%1$s>', $major);
            $m[]    =   sprintf('    <%s:title><![CDATA[%s]]></%1$s:title>', $major, $file->getName());
            switch ($major) {
                case 'image':
                    $m[]    =   sprintf('    <%s:loc>http://[[SERVER]][[SERVICE_FILE]]%s</%1$s:loc>', $major, $file->getUri());
                    $m[]    =   sprintf('    <%s:caption><![CDATA[%s]]></%1$s:caption>', $major, $file->getDescription());
                    break;

                case 'video':
                    $created    =   strtotime($file->getCreated());
                    $m[]    =   sprintf('    <%s:publication_date>%s</%1$s:publication_date>', $major, strftime('%Y-%m-%dT%H:%M:%S+00:00', $created));
                    $m[]    =   sprintf('    <%s:content_loc>http://[[SERVER]][[SERVICE_FILE]]%s</%1$s:content_loc>', $major, $file->getUri());
                    $m[]    =   sprintf('    <%s:player_loc allow_embed="yes" autoplay="autostart=true">http://[[SERVER]][[SERVICE_JAVASCRIPT]]WB/player.swf?file=[[SERVICE_FILE]]%s</%1$s:player_loc>', $major, $file->getUri());
                    $m[]    =   sprintf('    <%s:thumbnail_loc>http://[[SERVER]][[SERVICE_FILE]]%s</%1$s:thumbnail_loc>', $major, $file->getUri('image'));
                    $m[]    =   sprintf('    <%s:description><![CDATA[%s]]></%1$s:description>', $major, $file->getDescription());
                    $m[]    =   sprintf('    <%s:duration>%d</%1$s:duration>', $major, round($file->getInfo('seconds')));
                    $m[]    =   sprintf('    <%s:view_count>%d</%1$s:view_count>', $major, $file->getViews());
                    break;

                default:
                    continue;
                    break;
            }

            $m[]    =   sprintf('  </%s:%1$s>', $major);
            $m[]    =   '</url>';
            $m[]    =   '';
            $cnt    .=  implode("\n", $m);
        }
        WBDatasource_Table::flushCache();
    }

    /**
     * add all configured pages
     *
     * @todo implement switch for pages stored in database
     * @param string $cnt
     * @todo move old code to WBConfigLoader_File
     */
    protected function addPage2Sitemap(&$cnt)
    {
        if (empty($this->config['page']) || intval($this->config['page']) < 1) {
            return;
        }

        $loader =   WBParam::get('wb/config/loader/service/site', 'File');
        if ($loader != 'File') {
            $this->configLoader =   WBClass::create('WBConfig_Loader_' . WBParam::get('wb/config/loader/service/site', 'File'));
            $this->addPages2Sitemap($cnt);
            return;
        }

        $cnt    .=  sprintf($this->url, '', 1);
        // add pages from page configuration folders
        $etcDir     =   WBParam::get('wb/dir/base')
                    .   '/'
                    .   WBParam::get( 'wb/dir/config', 'etc' );

        if (is_dir($etcDir . '/site/page')) {
            $this->addPageDir2Sitemap($cnt, $etcDir . '/site/page');
        }
        $this->addPageDir2Sitemap($cnt, $etcDir . '-default/site/page');
    }

    /**
     * list pages in sitemap
     *
     * Recursive method to load find all page configurations and add them
     * to sitemap.
     *
     * @param string $cnt
     * @param string $path
     */
    private function addPages2Sitemap(&$cnt, $path = '')
    {
        $path   =   trim($path, '/');
        $branch =   array_pop(explode('/', $path));
        if (strncmp('__', $branch, 2) == 0) {
            return;
        }
        if (isset($this->pages[$path])) {
            return;
        }
        $this->pages['/' . $path] =   true;
        $prio   =   1.1 - (count(explode('/', $path)) / 10);
        $cnt    .=  sprintf($this->url, $path, $prio);

        // try to find children
        if (!empty($path)) {
            $path   =   '/' . $path;
        }
        $children   =   $this->configLoader->ls('site/page' . $path);
        if (empty($children)) {
            return;
        }

        foreach ($children as $child) {
            $child  =   substr($child, 10);
            $this->addPages2Sitemap($cnt, $child);
        }
    }

    /**
     * add pages to sitemap
     *
     * Recursively scans folder for page definitions. It also makes sure, that each page
     * will be added just once.
     *
     * @uses display()
     * @param string $cnt
     * @param string $dir
     * @param string $prefix
     * @todo move to WBConfig_Loader_File
     */
    protected function addPageDir2Sitemap(&$cnt, $dir, $prefix = '')
    {
        $iter   =   new DirectoryIterator($dir);
        foreach ($iter as $page) {
            if ($page->isDot()) {
                continue;
            }

            if ($page->isDir()) {
				++$this->level;
                $this->addPageDir2Sitemap($cnt, $page->getPathname(), $prefix . $page->getFilename() . '/');
				--$this->level;
                continue;
            }

            $name   =   explode('.', $page->getFilename());
            if (array_pop($name) != 'xml') {
                continue;
            }
            $name   =   implode('.', $name);

            // skip index and default pages
            if (strncmp($name, '__', 2) == 0) {
                continue;
            }
            $name   =   $prefix . $name;

            // make sure each page will be added only once
            if (isset($this->pages[$name])) {
                continue;
            }
            $this->pages[$name] =   true;
			$prio	=	1 - ($this->level / 10);
            $cnt    .=  sprintf($this->url, $name, $prio);
        }
    }

    /**
     * Display
     *
     * CAUTION: This method does not return!
     *
     * @param string $content
     * @param string $mime
     */
    protected function display($content, $mime)
    {
        $res  =   WBClass::create('WBResponse');
        /** @var $res WBResponse */
        $res->add($content);

        $res->addHeader('Content-Type', $mime);
        $res->send($this->req);
        exit(0);
    }
}
?>