~m-baert/planet-drupal/trunk

<?php

define('PLANET_ADMIN', 'administer planet');

function planet_node_info() {
        return array('planet' => array('name' => t('planet'), 'base' => 'planet'));
} 

function planet_perm() {
	return array(PLANET_ADMIN);
}

function planet_help($section){
	switch($section){
		case 'admin/help/planet':
			$output = '<p>Planet is an aggregator that allows you to aggregate the blogs for users in a given role (e.g. staff) and associate content with the users rather than as a detached feed. This provides the benefit of showing avatars with content, providing per-user aggregation of planet content in addition to blog content, etc.</p>';
			$output .= '<p>To use planet, go to admin/settings/planet and note the following sections:</p>';
			$output .= '<ul>';
			$output .= '<li><strong>Recent Items</strong>. This is a queue of items that have recently been pulled into the site. The convert column allows you to turn them into bona fide blog posts. This is useful if you want to feature blog posts front and center and keep planet as a secondary content view. Conversion allows you to float site-relevant content to the mainstream while providing a view of more general content from your members.</li>';
			$output .= '<li><strong>General Settings</strong>. The auto-publish value is a string that the module looks for in categories (if possible) or within a post itself. Posts that include the value will be auto-converted into blog entries. The role to select bloggers from lets you narrow the user list for when you\'re adding a feed and associating it with a user. A common setting will be to create a staff role and use this for planet.</li>';
			$output .= '<li><strong>Feeds</strong>. This section lets you add a new feed. Give it a title, select an author, provide the feed url, and you\'re off. You\'ll have to manually refresh it or wait for a cron run for items to be imported.</li>';
			$output .= '<li><strong>Feeds</strong>. This section lists current feeds, when they were last updated, how many items they have, and it allows you to edit, refresh, or freeze them. Freezing is a quick way to temporarily suspend updates from the given feed.</li>';
			$output .= '</ul>';
			return $output;
		case 'admin/modules#description':
			return t('Aggregates RSS feeds and faciliates their association with site users who belong to a given role.');
	}
}

function planet_node_name(){
	return 'planet';
}

function planet_access($op, $node){
	global $user;

	if ($op == 'create') {
		return user_access('edit own blog') && $user->uid;
	}

	if ($op == 'update' || $op == 'delete') {
		if (user_access('edit own blog') && ($user->uid == $node->uid)) {
			return TRUE;
		}
	}
}

function planet_menu($may_cache){
       global $user;
	$items = array();
	if($may_cache){
	
	}
	
	$items[] = array('path' => 'admin/settings/planet', 'title' => t('planet settings'),
      'callback' => '_planet_settings', 'access' => user_access('administer nodes'),
      'type' => MENU_NORMAL_ITEM);


    if(arg(0) == 'planet' && is_numeric(arg(1))){
      $items[] = array('path' => 'planet/' . arg(1), 'title' => 'planet', 'callback' => 'planet_page_user', 'callback arguments' => array(arg(1)));
    }
    
    if(arg(3) == 'refresh' && is_numeric(arg(4))){
    	$items[] = array('path' => 'admin/settings/planet/refresh/' . arg(4), 'title' => t('planet refresh'),
      		'callback' => 'planet_call_refresh', 'access' => user_access('administer nodes'),
      		'type' => MENU_CALLBACK);
    }
    
    if(is_numeric(arg(4)) && (arg(3) == 'freeze' || arg(3) == 'unfreeze')){
    	$items[] = array('path' => 'admin/settings/planet/' . arg(3). '/' . arg(4), 'title' => t('planet freeze'),
      		'callback' => 'planet_toggle_frozen', 'access' => user_access('administer nodes'),
      		'type' => MENU_CALLBACK);
    }
    
    if(is_numeric(arg(4)) && arg(3) == 'convert'){
    	$items[] = array('path' => 'admin/settings/planet/convert/' . arg(4), 'title' => t('planet convert'),
      		'callback' => 'planet_convert_to_blog', 'access' => user_access('administer nodes'),
      		'type' => MENU_CALLBACK);
    }

    $items[] = array('path' => 'planet', 'title' => t('Planet'),
      'callback' => 'planet_page_last', 'access' => user_access('access content'),
      'type' => MENU_CALLBACK);
      
    $items[] = array('path' => 'planet/feed', 'title' => t('Planet'),
      'callback' => 'planet_feed', 'access' => user_access('access content'),
      'type' => MENU_CALLBACK);
      
	return $items;
}

function planet_call_refresh(){
	$title = planet_refresh();
	watchdog('planet', 'Feed "' . $title . '" refreshed.');
	drupal_set_message('Feed "' . $title . '" refreshed.');
	drupal_goto('admin/settings/planet');
}

function planet_toggle_frozen(){
	$fid = intval(arg(4));
	db_query('UPDATE {planet_feeds} SET frozen = %d WHERE fid = %d', arg(3) == 'unfreeze' ? 0 : 1, $fid);
	drupal_set_message('Feed ' . (arg(3) == 'unfreeze' ? 'un' : '') . 'frozen.');
	drupal_goto('admin/settings/planet');
}

function _planet_settings(){
	if($_POST['edit']){	
		$edit = $_POST['edit'];
		
		if($_POST['op'] == 'Delete' && intval($edit['fid']) > 0){
			db_query('DELETE FROM {planet_feeds} WHERE fid = %d', intval($edit['fid']));
			drupal_set_message('Deleted feed.');
		}
		else{		
			if(isset($edit['fid']) && intval($edit['fid']) == 0){
				db_query('INSERT INTO {planet_feeds} (uid, title, link, image, checked, frozen) VALUES(%d, "%s", "%s", "%s", 0, 0)', $edit['uid'], $edit['title'], $edit['link'], $edit['image']);
				drupal_set_message('Added new feed.');		
			}
			else if($edit['fid'] && intval($edit['fid']) > 0){						
				db_query('UPDATE {planet_feeds} SET uid = %d, title="%s", link = "%s", image="%s" WHERE fid=%d', $edit['uid'], $edit['title'], $edit['link'], $edit['image'], $edit['fid']);
				drupal_set_message('Edited "' . $edit['title'] . '" feed.');			
			}
			else{
				if($edit['planet_author_roles']){						
					variable_set('planet_author_roles', $edit['planet_author_roles']);
				}
				if($edit['planet_auto_publish']){
					variable_set('planet_auto_publish', $edit['planet_auto_publish']);
				}
				drupal_set_message('Edited general planet settings.');							
			}			
		}	
		drupal_goto('admin/settings/planet');	
	}
	else{		
		$fid = intval(arg(3));
		if($fid > 0){			
			$edit = db_fetch_array(db_query('SELECT * FROM {planet_feeds} WHERE fid = %d', $fid));
			$output .= planet_feed_form($edit, true);
		}
		else{
		
			$roles = array();

			$result = db_query('SELECT rid, name FROM {role}');
			while($role = db_fetch_object($result)){
				$roles[$role->rid] = $role->name;
			}

			$output .= '<h2>Recent Items</h2>';
			$output .= planet_queue();
			
			$form = array();

			$form['general'] = array('#type' => 'fieldset', '#title' => t('General Settings'));
		
			$form['general']['planet_auto_publish'] = array('#type' => 'textfield', '#title' => t('Auto-publish text'), '#value' => variable_get('planet_auto_publish', ''), '#size' => 30, '#maxlength' => 30, '#description' => t('If this snippet appears in the aggregated item, it will be auto-published.'));
			$form['general']['planet_author_roles'] = array('#type' => 'select', '#title' => 'Role to select authors from', '#options' => $roles, '#default_value' => variable_get('planet_author_roles', 2), '#description' =>t('Select the role from which blog authors should be selected on the feed creation screen.'));
			$form['general']['submit'] = array('#type' => 'submit', '#value' => 'Adjust Settings');
			
			$output .= drupal_get_form('settings', $form);
			//$output .= $form;

			$output .= planet_feed_form($edit);
		
			$result = db_query('SELECT *, (UNIX_TIMESTAMP(NOW()) - checked) _checked FROM {planet_feeds}');
			$result = db_query('SELECT COUNT(f.uid) cnt, f.*, (UNIX_TIMESTAMP(NOW()) - checked) _checked FROM planet_feeds f LEFT OUTER JOIN planet_items i ON i.fid = f.fid GROUP BY f.uid;');
			$rows = array();
			$headers = array('Feed', 'Items', 'Edit', 'Last checked', 'Refresh', 'Freeze');
			while($feed = db_fetch_object($result)){
				$checked = intval($feed->_checked / 60) . ' minutes';
				if($feed->_checked % 60 > 0){
					$checked .= ', ' . $feed->_checked % 60 . ' seconds';
				}
				$checked .= ' ago';
				array_push($rows, array(
					$feed->title,
					$feed->cnt,
					l('edit', 'admin/settings/planet/' . intval($feed->fid)),
					$checked,
					l('refresh', 'admin/settings/planet/refresh/' . intval($feed->fid)),
					l($feed->frozen ? 'unfreeze' : 'freeze', 'admin/settings/planet/' . ($feed->frozen ? 'unfreeze/' : 'freeze/') . intval($feed->fid))					
					)
				);
			}
			$output .= '<h2>Feeds</h2>';
			$output .= theme('table', $headers, $rows);			
		}
		print theme('page', $output);		
	}
}

function planet_feed_form($edit = array(), $individual = false){
	$uids = array();
	$result = db_query('SELECT u.uid, u.name FROM {users} u, {role} r, {users_roles} ur WHERE u.uid = ur.uid AND ur.rid = r.rid AND r.rid = %d ORDER BY u.name ASC', variable_get('planet_author_roles',2));
	while($f_user = db_fetch_object($result)){
		$uids[$f_user->uid] = $f_user->name;
	}		
	
	$form = array();
	$form['feeds'] = array('#type' => 'fieldset', '#title' => 'Feeds');	
	$form['feeds']['fid'] = array('#type' => 'hidden', '#value' => $edit['title']);
	$form['feeds']['title'] = array('#type' => 'textfield', '#title' => t('Title'), '#value' => $edit['title'], '#size' => 40, '#maxlength' => 40);
	$form['feeds']['uid'] = array('#type' => 'select', '#title' => t('Original author'), '#value' => $edit['uid'], '#options' => $uids, '#description' => t('Select a user to associate this feed with'));
	$form['feeds']['link'] = array('#type' => 'textfield', '#title' => t('URL'), '#value' => $edit['link'], '#size' => 40, '#maxlength' => 80);
	$form['feeds']['submit'] = array('#type' => 'submit', '#value' => $edit['fid'] > 0 ? 'Edit' : 'Add' . ' Feed');
	$output .= drupal_get_form('settings',$form);
	
	if($individual){
		$output .= planet_queue(intval($edit['uid']));
	}
	
	return $output;
}

function planet_queue($uid = null){
	$rows = array();
	$headers = array('Title', 'Created', 'Convert');
	
	if($uid > 0){
		$result = pager_query(db_rewrite_sql('SELECT nid, title, created FROM {node} WHERE type = "planet" AND uid = ' . intval($uid) . ' ORDER BY created DESC'), variable_get('default_nodes_main', 10), 0, NULL);
	}
	else{
		$result = pager_query(db_rewrite_sql('SELECT nid, title, created FROM {node} WHERE type = "planet" ORDER BY created DESC'), variable_get('default_nodes_main', 10), 0, NULL);	
	}
	
	while($node = db_fetch_object($result)){
		array_push($rows, 	array(
								l($node->title, 'node/' . $node->nid),
								format_date($node->created, 'custom', 'Y-m-d H:i'),
								l('convert', 'admin/settings/planet/convert/' . $node->nid)
							)
		);
	}
	$output .= theme('table',$headers, $rows);
	$output .= theme('pager', NULL, variable_get('default_nodes_main', 10));
	return $output;
}


function planet_cron(){
	$result = db_query('SELECT fid FROM {planet_feeds} WHERE frozen = 0');
	while($feed = db_fetch_object($result)){
		$title = planet_refresh($feed->fid);
		watchdog('planet', 'Cron updated feed "' . $title . '".');
	}
}

function planet_refresh($fid = null){
	if(!$fid){
		$fid = intval(arg(4));
	}
	
	$feed = db_fetch_object(db_query('SELECT * FROM {planet_feeds} WHERE fid = %d', $fid));
	
	$headers = array();
	$result = planet_http_request($feed->link, $headers, 15);	
	
	switch ($result->code) {
		case 304:
			drupal_set_message(t('No new content syndicated from %site.', array('%site' => '<em>'. $feed->title .'</em>')));
			break;

		case 301:
			if ($result->redirect_url) {
				$feed->link = $result->redirect_url;
				watchdog('planet', t('Updated URL for feed %title to %url.', array('%title' => '<em>'. $feed->title .'</em>', '%url' => '<em>'. $feed->url .'</em>')), WATCHDOG_NOTICE, l(t('view'), 'planet/'.$feed->fid));
				db_query("UPDATE {planet_feeds} SET link = '%s' WHERE fid = %d", $feed->link, $feed->fid);
			}
			break;

		case 200:
		case 302:
		case 307:
			$xml_tree = planet_parse_xml($result->data);

			if ($xml_tree['parser_error']) {
				watchdog('planet', t('Failed to parse RSS feed %site: %error at line %line.', array('%site' => '<em>'. $feed->title .'</em>', '%error' => $xml_tree['parser_error'], '%line' => $xml_tree['parser_line'])), WATCHDOG_ERROR);
				drupal_set_message(t('Failed to parse RSS feed %site: %error at line %line.', array('%site' => '<em>'. $feed->title .'</em>', '%error' => $xml_tree['parser_error'], '%line' => $xml_tree['parser_line'])), 'error');
				break;
			}
			else {
				drupal_set_message('Parsing feed '. $feed->title .' took '. $xml_tree['parser_time'] .' seconds.');
			}

			if (planet_parse_items($xml_tree, $feed) !== false){
				if ($result->headers['Last-Modified']) {
					$modified = strtotime($result->headers['Last-Modified']);
				}

				/*
				** Prepare data:
				*/
				if ($xml_tree['RSS']) { // RSS 0.91, 0.92, 2.0
					$root = &$xml_tree['RSS'][0];
					$channel = &$root['CHANNEL'][0];
					$image = &$channel['IMAGE'][0];
					$description = &$channel['DESCRIPTION'][0]['VALUE'];
					$link = &$channel['LINK'][0]['VALUE'];
				}
				else if ($xml_tree['RDF:RDF']) {
					$root = &$xml_tree['RDF:RDF'][0];
					$channel = &$root['CHANNEL'][0];
					$image = &$root['IMAGE'][0];
					$description = &$channel['DESCRIPTION'][0]['VALUE'];
					$link = &$channel['LINK'][0]['VALUE'];
				}
				else if ($xml_tree['FEED']) { // Atom 0.3, 1.0
					$root = &$xml_tree['FEED'][0];
					$channel = &$root;
					$image = &$channel['LOGO'][0]['VALUE'];
					$description = ($channel['TAGLINE'][0]['VALUE'] ? $channel['TAGLINE'][0]['VALUE'] : '');
					// TODO: remove this Atom hack when we have field mapping or at least specialized parsers in place
					if (count($channel['LINK']) > 1) {
						$link = $feed->link;
						foreach ($channel['LINK'] as $l) {
							if ($l['REL'] == 'alternate') {
								$link = $l['HREF'];
							}
						}
					}
					else {
						$link = $channel['LINK'][0]['HREF'];
					}
				}
				else if ($xml_tree['CHANNEL']) { // RSS 1.1
					$root = &$xml_tree['CHANNEL'][0];
					$channel = &$root;
					$image = &$channel['IMAGE'][0];
					$description = &$channel['DESCRIPTION'][0]['VALUE'];
					$link = &$channel['LINK'][0]['VALUE'];
				}
				else if ($xml_tree['OPML']) {
					$root = &$xml_tree['OPML'][0];
					$channel = &$root;
					$image = NULL;
					$description = NULL;
					$link = NULL;
				}
				else {
					// unsupported format
					break;
				}

				if (!$feed->uid) {
					if ($channel['AUTHOR'][0]['VALUE']) {
						$feed->uid = $channel['AUTHOR'][0]['VALUE'];
					}
					if ($channel['AUTHOR'][0]['NAME'][0]['VALUE']) {
						$feed->uid = $channel['AUTHOR'][0]['NAME'][0]['VALUE'];
					}
					else if ($channel['DC:CREATOR']) {
						$feed->uid = $channel['DC:CREATOR'][0]['VALUE'];
					}
					else {
						$feed->uid = '';
					}
				}

				/*
				** Generate image link
				*/
				if (!$feed->image && $image['LINK'] && $image['URL'] && $image['TITLE']) {
					if (strlen($image['TITLE'][0]['VALUE']) > 250) {
						$image['TITLE'][0]['VALUE'] = trim(substr($image['TITLE'][0]['VALUE'], 0, 250)).'...';
					}
					$feed->image = '<a href="'. $image['LINK'][0]['VALUE'] .'" class="planet_logo_link"><img src="'. $image['URL'][0]['VALUE'] .'" class="planet_logo" alt="'. $image['TITLE'][0]['VALUE'] .'" /></a>';
				}

				/*
				** Update the feed data:
				*/
				$feed->checked = time();
				$feed->link = $link;
				$feed->etag = $result->headers['ETag'];
				$feed->modified = $modified;
				if ($feed->body == '' && $description/* && valid_input_data($description)*/) {
					$feed->body = $feed->teaser = $description;
				}
				$feed->rss_data = &$xml_tree;

				/*
				** Taxonomy module doesn't add taxonomy terms at load time... so we have to do it by hand :((
				*/
				$terms = module_invoke('taxonomy', 'node_get_terms', $feed->nid, 'tid');
				$feed->taxonomy = array();
				foreach ($terms as $tid => $term) {
					if ($term->tid) {
						$feed->taxonomy[] = $term->tid;
					}
				}
			}
		default:			
	}	
	
	
	db_query('UPDATE {planet_feeds} SET checked = %d WHERE fid = %d', time(), $fid);
	return $feed->title;		
	//print theme('page', 'refreshing ' . $fid . '.');// and got ' . print_r($feed, 1));
}

/**
 * Let admin users convert nodes to blog posts.
 */
function planet_convert_to_blog($nid = NULL){
	if(user_access('administer nodes')){
		if ($nid == NULL && is_numeric(arg(4))) {
			$nid = intval(arg(4));
		}
		$node = node_load($nid);
		if($node->type == 'planet'){
			//Unpublish planet item
			$node->status = 0;
			node_save($node);
			
			//Now reset some fields and resave as a blog node.
			$node->type = 'blog';
			$node->nid = null;
			$node->comment = 2;
			$node->status = 1;
			$nid = node_save($node);
			
			//And redirect to the new node.	
			$url = $node->path ? $node->path : 'node/' . $nid;
			drupal_set_message(t('The feed item has been converted.'));
			drupal_goto($url);
			//drupal_goto('admin/feeds/queue');
		}
	}
}


/**
 * Private function; Parse HTTP headers from data retreived with cURL
 * from: http://pl2.php.net/manual/en/function.curl-setopt.php#42009
 */
function planet_parse_response($response){
	/*
	 ***original code extracted from examples at
	 ***http://www.webreference.com/programming
													 /php/cookbook/chap11/1/3.html

	 ***returns an array in the following format which varies depending on headers returned

			 [0] => the HTTP error or response code such as 404
			 [1] => Array
			 (
					 [Server] => Microsoft-IIS/5.0
					 [Date] => Wed, 28 Apr 2004 23:29:20 GMT
					 [X-Powered-By] => ASP.NET
					 [Connection] => close
					 [Set-Cookie] => COOKIESTUFF
					 [Expires] => Thu, 01 Dec 1994 16:00:00 GMT
					 [Content-Type] => text/html
					 [Content-Length] => 4040
			 )
			 [2] => Response body (string)
	*/

	do {
		list($response_headers, $response) = explode("\r\n\r\n", $response, 2);
		$response_header_lines = explode("\r\n", $response_headers);

		// first line of headers is the HTTP response code
		$http_response_line = array_shift($response_header_lines);
		if (preg_match('@^HTTP/[0-9]\.[0-9] ([0-9]{3})@', $http_response_line, $matches)) {
			$response_code = $matches[1];
		}
		else {
			$response_code = "Error";
		}
	}
	while (substr($response_code, 0, 1) == "1");

	$response_body = $response;
				
	// put the rest of the headers in an array
	$response_header_array = array();
	foreach ($response_header_lines as $header_line) {
		list($header, $value) = explode(':', $header_line, 2);
		$response_header_array[$header] = trim($value);
	}

	return array($response_code, $response_header_array, $response_body, $http_response_line);
}

/**
 * Private function; Gets data from given URL :)
 */
function planet_http_request($url, $headers = array(), $timeout = 15, $method = 'GET', $data = NULL, $follow = 3) {
	if (!function_exists('curl_init')) {
		return drupal_http_request($url, $headers, $method, $data, $follow);
	}

	// convert headers array to format used by cURL
	$temp = array();
	foreach ($headers as $header => $value) {
		$temp[] = $header .': '. $value;
	}
	$headers = $temp;

	$result = new StdClass();

	$ch = curl_init();
	curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
	curl_setopt($ch, CURLOPT_URL, $url);
	curl_setopt($ch, CURLOPT_HEADER, 1);
	curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
	curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 0);
	curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);

	$data = curl_exec($ch);
	$info = curl_getinfo($ch);

	curl_close($ch);
	unset($ch);

	$response = planet_parse_response($data);
	$result->code = $response[0];
	$result->headers = $response[1];
	$result->data = $response[2];
	$error = $response[3];
	switch ($code) {
		case 200: // OK
		case 304: // Not modified
			break;
		case 301: // Moved permanently
		case 302: // Moved temporarily
		case 307: // Moved temporarily
			$location = $result->headers['Location'];

			if ($follow) {
				$result = planet_http_request($result->headers['Location'], $headers, $timeout, $method, $data, --$follow);
				$result->redirect_code = $result->code;
			}
			$result->redirect_url = $location;
			break;
		default:
			$result->error = $error;
			break;
	}

	$result->code = $response[0];
	return $result;
}

/**
 * Private function; Checks a news feed for new items.
 */


/**
 * Private function;
 * Parse the W3C date/time format, a subset of ISO 8601. PHP date parsing
 * functions do not handle this format.
 * See http://www.w3.org/TR/NOTE-datetime for more information.
 * Origionally from MagpieRSS (http://magpierss.sourceforge.net/).
 *
 * @param $date_str A string with a potentially W3C DTF date.
 * @return A timestamp if parsed successfully or -1 if not.
 */
function planet_parse_w3cdtf($date_str) {
	if (preg_match('/(\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2})(:(\d{2}))?(?:([-+])(\d{2}):?(\d{2})|(Z))?/', $date_str, $match)) {
		list($year, $month, $day, $hours, $minutes, $seconds) = array($match[1], $match[2], $match[3], $match[4], $match[5], $match[6]);
		// calc epoch for current date assuming GMT
		$epoch = gmmktime($hours, $minutes, $seconds, $month, $day, $year);
		if ($match[10] != 'Z') { // Z is zulu time, aka GMT
			list($tz_mod, $tz_hour, $tz_min) = array($match[8], $match[9], $match[10]);
			// zero out the variables
			if (!$tz_hour) {
				$tz_hour = 0;
			}
			if (!$tz_min) {
				$tz_min = 0;
			}
			$offset_secs = (($tz_hour * 60) + $tz_min) * 60;
			// is timezone ahead of GMT?	then subtract offset
			if ($tz_mod == '+') {
				$offset_secs *= -1;
			}
			$epoch += $offset_secs;
		}
		return $epoch;
	}
	else {
		return -1;
	}
}

/**
 * Private function;
 * from: http://pl2.php.net/manual/en/function.html-entity-decode.php#51055
 * Used as callback function for preg_replace_all() to decode numeric entities to UTF-8 chars
 *
 * @param $ord Number
 * @return UTF-8 string
 */
function planet_replace_num_entity($ord) {
	$ord = $ord[1];
	if (preg_match('/^x([0-9a-f]+)$/i', $ord, $match)) {
		$ord = hexdec($match[1]);
	}
	else {
		$ord = intval($ord);
	}

	$no_bytes = 0;
	$byte = array();

	if ($ord == 128) {
		return chr(226).chr(130).chr(172);
	}
	else if($ord == 129) {
		return chr(239).chr(191).chr(189);
	}
	else if($ord == 130) {
		return chr(226).chr(128).chr(154);
	}
	else if($ord == 131) {
		return chr(198).chr(146);
	}
	else if($ord == 132) {
		return chr(226).chr(128).chr(158);
	}
	else if($ord == 133) {
		return chr(226).chr(128).chr(166);
	}
	else if($ord == 134) {
		return chr(226).chr(128).chr(160);
	}
	else if($ord == 135) {
		return chr(226).chr(128).chr(161);
	}
	else if($ord == 136) {
		return chr(203).chr(134);
	}
	else if($ord == 137) {
		return chr(226).chr(128).chr(176);
	}
	else if($ord == 138) {
		return chr(197).chr(160);
	}
	else if($ord == 139) {
		return chr(226).chr(128).chr(185);
	}
	else if($ord == 140) {
		return chr(197).chr(146);
	}
	else if($ord == 141) {
		return chr(239).chr(191).chr(189);
	}
	else if($ord == 142) {
		return chr(197).chr(189);
	}
	else if($ord == 143) {
		return chr(239).chr(191).chr(189);
	}
	else if($ord == 144) {
		return chr(239).chr(191).chr(189);
	}
	else if($ord == 145) {
		return chr(226).chr(128).chr(152);
	}
	else if($ord == 146) {
		return chr(226).chr(128).chr(153);
	}
	else if($ord == 147) {
		return chr(226).chr(128).chr(156);
	}
	else if($ord == 148) {
		return chr(226).chr(128).chr(157);
	}
	else if($ord == 149) {
		return chr(226).chr(128).chr(162);
	}
	else if($ord == 150) {
		return chr(226).chr(128).chr(147);
	}
	else if($ord == 151) {
		return chr(226).chr(128).chr(148);
	}
	else if($ord == 152) {
		return chr(203).chr(156);
	}
	else if($ord == 153) {
		return chr(226).chr(132).chr(162);
	}
	else if($ord == 154) {
		return chr(197).chr(161);
	}
	else if($ord == 155) {
		return chr(226).chr(128).chr(186);
	}
	else if($ord == 156) {
		return chr(197).chr(147);
	}
	else if($ord == 157) {
		return chr(239).chr(191).chr(189);
	}
	else if($ord == 158) {
		return chr(197).chr(190);
	}
	else if($ord == 159) {
		return chr(197).chr(184);
	}
	else if($ord == 160) {
		return chr(194).chr(160);
	}

	if ($ord < 128) {
		return chr($ord);
	}
	else if ($ord < 2048) {
		$no_bytes = 2;
	}
	else if ($ord < 65536) {
		$no_bytes = 3;
	}
	else if ($ord < 1114112) {
		$no_bytes = 4;
	}
	else {
		return;
	}

	switch ($no_bytes) {
		case 2:
			$prefix = array(31, 192);
			break;

		case 3:
			$prefix = array(15, 224);
			break;

		case 4:
			$prefix = array(7, 240);
			break;
	}

	for ($i = 0; $i < $no_bytes; $i++) {
		$byte[$no_bytes - $i - 1] = (($ord & (63 * pow(2, 6 * $i))) / pow(2, 6 * $i)) & 63 | 128;
	}

	$byte[0] = ($byte[0] & $prefix[0]) | $prefix[1];

	$ret = '';
	for ($i = 0; $i < $no_bytes; $i++) {
		$ret .= chr($byte[$i]);
	}

	return $ret;
}

/**
 * Private function; Convert named entities to UTF-8 characters
 * from: http://pl2.php.net/manual/en/function.html-entity-decode.php#51722
 */
function planet_replace_name_entities(&$text) {
	static $ttr;
	if (!$ttr) {
		$trans_tbl = get_html_translation_table(HTML_ENTITIES);
		foreach ($trans_tbl as $k => $v) {
			$ttr[$v] = utf8_encode($k);
		}
		$ttr['&apos;'] = "'";
	}
	return strtr($text, $ttr);
}

/**
 * Private function; Convert all entities to UTF-8 characters
 */
function planet_replace_entities(&$text) {
	$result = planet_replace_name_entities($text);
	return preg_replace_callback('/&#([0-9a-fx]+);/mi', 'planet_replace_num_entity', $result);
}

/**
 * Private function; Clone object function to stay compatible with both php4 and php5
 * from: Drupal 4.7CVS
 * TODO: remove after moving to Drupal 4.7
 */
function planet_clone($object) {
	return version_compare(phpversion(), '5.0') < 0 ? $object : clone($object);
}

/**
 * Private function; Convert relative URLs
 */
function planet_convert_relative_urls(&$data, $base_url) {
	$src = '%( href| src)="(?!\w+://)/?([^"]*)"%';
	$dst = '$1="'. trim($base_url, '/') .'/$2"';
	return preg_replace($src, $dst, $data);
}

/**
 * Private function; Creates nodes from data found in given xml_tree
 */
function planet_parse_items(&$xml_tree, &$feed) {

	if ($xml_tree['RSS']) { // RSS 0.91, 0.92, 2.0
		$items = &$xml_tree['RSS'][0]['CHANNEL'][0]['ITEM'];
		$link_field = 'VALUE';
	}
	else if ($xml_tree['RDF:RDF']) {
		$items = &$xml_tree['RDF:RDF'][0]['ITEM'];
		$link_field = 'VALUE';
	}
	else if ($xml_tree['FEED']) { // Atom 0.3, 1.0
		$items = &$xml_tree['FEED'][0]['ENTRY'];
		$link_field = 'HREF';
	}
	else if ($xml_tree['CHANNEL']) { // RSS 1.1
		$items = &$xml_tree['CHANNEL'][0]['ITEMS'][0]['ITEM'];
		$link_field = 'VALUE';
	}
	else {
		// unsupported format
		$items = array();
		return false;
	}

	/*
	** We reverse the array such that we store the first item last,
	** and the last item first. In the database, the newest item
	** should be at the top.
	*/
	$items_added = 0;
	$auto_publish = variable_get('planet_auto_publish', '');				


	for ($index = count($items) - 1; $index >= 0; $index--) {
		$item = &$items[$index];
	//print '<pre>' . print_r($item, 1) . '</pre>';
		$teaser = NULL;
		$body = NULL;

		// Description field is needed early for case when no title is specified
		if ($item['DESCRIPTION']) { // RSS 0.91, 0.92, 1.0, 1.1, 2.0
			$body = &$item['DESCRIPTION'][0]['VALUE'];
		}
		else if ($item['SUMMARY']) { // Atom 0.3, 1.0
			$body = &$item['SUMMARY'][0]['VALUE'];
		}
		
		if ($item['CONTENT']) { // Atom 0.3, 1.0
			if (strlen($body) < strlen($item['CONTENT'][0]['VALUE'])) {
				if ($body) {
					$teaser = $body;
				}
				$body = &$item['CONTENT'][0]['VALUE'];
			}
		}
		else if ($item['CONTENT:ENCODED']) { // Don't know where it came from but it can be found in RSS 2.0 feeds
			if (strlen($body) < strlen($item['CONTENT:ENCODED'][0]['VALUE'])) {
				if ($body) {
					$teaser = $body;
				}
				$body = &$item['CONTENT:ENCODED'][0]['VALUE'];
			}
		}
		
		/*
		** Resolve the item's title. If no title is found, we use
		** up to 40 characters of the description ending at a word
		** boundary but not splitting potential entities.
		*/
		if (!($title = $item['TITLE'][0]['VALUE'])) {
			$title = preg_replace('/^(.*)[^\w;&].*?$/', "\\1", truncate_utf8($body, 40));
		}

		// If title was "escaped" then it may still contain entities, becuase each & from entity was also escabet to &amp; before
		// TODO: the same for content?
		if ($item['TITLE'][0]['MODE'] == 'escaped') {
			$title = feeds_replace_entities($title);
		}
		$title = strip_tags($title);

		/*
		** Resolve the items link.
		*/
		if ($item['LINK']) {
			// TODO: remove this Atom hack when we have field mapping or at least specialized parsers in place
			if (count($item['LINK']) > 1) {
				$link = $feed->link;
				foreach ($item['LINK'] as $temp) {
					if ($temp['REL'] == 'alternate') {
						$link = $temp[$link_field];
					}
				}
			}
			else {
				$link = $item['LINK'][0][$link_field];
			}
		}
		elseif ($item['GUID'] && (strncmp($item['GUID'][0][$link_field], 'http://', 7) == 0) && $item['GUID'][0]['ISPERMALINK'] != 'false') {
			$link = $item['GUID'][0][$link_field];
		}
		else {
			$link = $feed->link;
		}

		/*
		** Resolve the items source.
		*/
		if ($item['SOURCE'][0]['VALUE'] && $item['SOURCE'][0]['URL']) { // RSS 2.0
			$source_title = &$item['SOURCE'][0]['VALUE'];
			$source_link = &$item['SOURCE'][0]['URL'];
		}
		else if ($item['SOURCE'] || $item['ATOM:SOURCE']) { // ATOM 1.0
			if ($item['SOURCE'][0]['TITLE']) $source_title = &$item['SOURCE'][0]['TITLE'][0]['VALUE'];
			else if ($item['SOURCE'][0]['ATOM:TITLE']) $source_title = &$item['SOURCE'][0]['ATOM:TITLE'][0]['VALUE'];
			if ($item['SOURCE'][0]['LINK']) $source_link = &$item['SOURCE'][0]['LINK'][0]['VALUE'];
			else if ($item['SOURCE'][0]['ATOM:LINK']) $source_link = &$item['SOURCE'][0]['ATOM:LINK'][0]['VALUE'];
		}
		else {
			$source_title = '';
			$source_link = '';
		}

		/*
		** Try to resolve and parse the item's publication date.	If no
		** date is found, we use the current date instead.
		*/
		// TODO: find nicer way for handling namespaces ;)
		if ($item['PUBDATE']) $date = $item['PUBDATE'][0]['VALUE'];												// RSS 2.0
		else if ($item['DC:DATE']) $date = $item['DC:DATE'][0]['VALUE'];									 // Dublin core
		else if ($item['DATE']) $date = $item['DATE'][0]['VALUE'];												 // Dublin core
		else if ($item['DCTERMS:ISSUED']) $date = $item['DCTERMS:ISSUED'][0]['VALUE'];		 // Dublin core
		else if ($item['ISSUED']) $date = $item['ISSUED'][0]['VALUE'];										 // Dublin core
		else if ($item['DCTERMS:CREATED']) $date = $item['DCTERMS:CREATED'][0]['VALUE'];	 // Dublin core
		else if ($item['CREATED']) $date = $item['CREATED'][0]['VALUE'];									 // Dublin core
		else if ($item['DCTERMS:MODIFIED']) $date = $item['DCTERMS:MODIFIED'][0]['VALUE']; // Dublin core
		else if ($item['MODIFIED']) $date = $item['MODIFIED'][0]['VALUE'];								 // Dublin core
		else if ($item['ATOM:UPDATED']) $date = $item['ATOM:UPDATED'][0]['VALUE'];				 // Atom
		else if ($item['UPDATED']) $date = $item['UPDATED'][0]['VALUE'];									 // Atom
		else $date = 'now';

		if ($feed->item_date_source == FEEDS_ITEM_DATE_SNIFFED && $date) {
			$timestamp = strtotime($date); // strtotime() returns -1 on failure
			if ($timestamp < 0) {
				$timestamp = planet_parse_w3cdtf($date); // also returns -1 on failure
				if ($timestamp < 0) {
					$timestamp = time(); // better than nothing
				}
			}
		}
		else {
			$timestamp = time();
		}

		// Ignore items older than allowed for feed
		if ($timestamp < $time_horizont) {
			continue;
		}

		/*
		** Save this item. Try to avoid duplicate entries as much as
		** possible. If we find a duplicate entry, we resolve it and
		** pass along it's ID such that we can update it if needed.
		*/
		// Try to use RSS:GUID/ATOM:ID as unique identifier
		$guid = '';
		if ($item['GUID'][0]['VALUE']) { // RSS 2.0
			$guid = $item['GUID'][0]['VALUE'];
		}
		else if ($item['ATOM:ID'][0]['VALUE']) { // ATOM 0.3, 1.0
			$guid = $item['ATOM:ID'][0]['VALUE'];
		}
		else if ($item['ID'][0]['VALUE']) { // ATOM 0.3, 1.0
			$guid = $item['ID'][0]['VALUE'];
		}
		else{
			// feed may contain duplicated links for different items, so we try to generate unique ID for each item
			$guid = md5("$title - . " . $feed->fid);
		}
		// TODO: is there anyway to check if DC:IDENTIFIER is unique?
		//			 http://dublincore.org/documents/usageguide/elements.shtml says it can be non-unique so useles for us :(			

		$entry = NULL;
		if ($guid && strlen($guid) > 0) {
			$entry = db_fetch_object(db_query("SELECT nid FROM {planet_items} WHERE guid = '%s' AND fid = %d", $guid, $feed->fid));
		}
		else if ($link && $link != $feed->link && $link != $feed->url) {
			$entry = db_fetch_object(db_query("SELECT nid FROM {planet_items} WHERE guid = '%s' AND fid = %d", $link, $feed->fid));
		}
		else {
			$entry = db_fetch_object(db_query("SELECT ai.nid AS nid FROM {node} n, {planet_items} ai WHERE ai.fid = %d AND ai.nid = n.nid AND n.title = '%s'", $feed->fid, $title));
		}
		
		//print $guid . '<br />';
		//print $entry->nid . '<br />';
		// Ignore items already existing in database and not allowed to be updated

		//Fields to update in either case
		$entry->changed = strtotime($date);
		$entry->title = $title;
		$entry->body = $body;
		$entry->body = planet_convert_relative_urls($body, $link);
		$entry->teaser = node_teaser($entry->body);		

		// Check for an auto-publish string in the post and publish if applicable.
		$has_category = false;
		if($auto_publish != ''){
			$has_category = false;
			if($item['CATEGORY']){
				foreach($item['CATEGORY'] as $cat){
					if($cat['VALUE'] == $auto_publish){
						$has_category = true;
					}	
				}	
			}
		}
		//end auto-publish

		//Fields to set if it's a new item.		
		if(!isset($entry->nid)){						
			$entry->type = 'planet';
			if($has_category == true || preg_match('/' . $auto_publish . '/',$entry->body)){							
				$entry->type = 'blog';
			}			

			$options = variable_get('node_options_planet', array());

			$entry->uid = $feed->uid;
			$entry->status = 1;
			$entry->moderate = 0;
			$entry->promote = in_array('promote', $options) ? 1 : 0;
			$entry->sticky = in_array('sticky', $options) ? 1 : 0;
			$entry->comment = in_array('comment', $options) ? 2 : 0;
			$entry->format = 3;
			$entry->created = strtotime($date);
			
			$terms = module_invoke('taxonomy', 'node_get_terms', $edit->nid, 'tid');
			foreach ($terms as $tid => $term) {
				if ($term->tid) {
					$edit->taxonomy[] = $term->tid;
				}
			}
			node_save($entry);
			db_query('INSERT INTO {planet_items} (fid, nid, guid, created) VALUES(%d, %d, "%s", UNIX_TIMESTAMP(NOW()))', $feed->fid, $entry->nid, $guid);							
			watchdog('planet', 'Adding ' . $title);
			drupal_set_message('Adding ' . $title);
		}	
		else{
			if($entry->status == 1){
				if($has_category == true || preg_match('/' . $auto_publish . '/',$entry->body)){							
					$entry->type = 'blog';
				}
				node_save($entry);
				watchdog('planet', 'Updating ' . $title);
				drupal_set_message('Updating ' . $title);
			}
		}
	}

	return $items_added;
}


/**
 * Private function; parses given XML data and returns array
 */
function planet_parse_xml(&$data) {
	global $xml_tree, $xml_paths, $xml_path_cur;
	$xml_tree = array();
	$xml_paths[] = &$xml_tree;
	$xml_path_cur = 0;

	$_start = microtime();

	// Some feeds already use CDATA but in "wrong way": http://www.rocketboom.com/vlog/quicktime_daily_enclosures.xml (ie. <description> something <CDATA soemthing else></description>
	$data = trim(str_replace(array('<![CDATA[', ']]>'), '', $data));
	
	// Add CDATA around content which may contain (x)html data, and is not contained in CDATA yet
	$src = array(
		'%(<(link|content|content:encoded|description|title|summary|info|tagline|copyright|source|itunes:summary|media:text|text)(?>[^<]*(?<!/)>)(?!<!\[CDATA\[))(.*)(</\2>)%sUS',
		'%24:(\d\d:\d\d)%' // workaround buggy hour format in feeds
		/*'%(<(\w+)(?>[^<]*type=")(?:text/html|application/xhtml\+xml|html|xhtml")(?>[^<]*(?<!/)>)(?!<!\[CDATA\[))(.*)(</\2>)%sUS'*/
		);
	$dst = array(
		'$1<![CDATA[$3]]>$4',
		'00:$1'
		);
	$data = preg_replace($src, $dst, $data);

	// parse the data:
	$xml_parser = drupal_xml_parser_create($data);
	if ($xml_parser == NULL) {
		return $xml_tree;
	}

	xml_set_element_handler($xml_parser, 'planet_element_start', 'planet_element_end');
	xml_set_character_data_handler($xml_parser, 'planet_element_data');
	xml_parser_set_option($xml_parser, XML_OPTION_CASE_FOLDING, 1);
	xml_parser_set_option($xml_parser, XML_OPTION_SKIP_WHITE, 1);
	if (!xml_parse($xml_parser, $data, 1)) {
		$xml_tree['parser_error'] = xml_error_string(xml_get_error_code($xml_parser));
		$xml_tree['parser_line'] = xml_get_current_line_number($xml_parser);
	}
	else {
			 unset($xml_tree['parser_error']);
			 unset($xml_tree['parser_line']);
	}
	xml_parser_free($xml_parser);

	$_end = microtime();

	list($sec, $usec) = explode(' ', $_start);
	$_start = $sec + $usec;
	list($sec, $usec) = explode(' ', $_end);
	$xml_tree['parser_time'] = ($sec + $usec) - $_start;

	return $xml_tree;
}

/**
 * Private call-back function used by the XML parser.
 */
function planet_element_start($parser, $name, $attributes) {
	global $xml_tree, $xml_paths, $xml_path_cur;

	$temp = &$xml_paths[$xml_path_cur++];
	$temp[$name][] = $attributes;
	$xml_paths[$xml_path_cur] = &$temp[$name][count($temp[$name])-1];
}

/**
 * Private call-back function used by the XML parser.
 */
function planet_element_end($parser, $name) {
	global $xml_tree, $xml_paths, $xml_path_cur;

	$temp = &$xml_paths[$xml_path_cur];
	array_pop($xml_paths);
	$xml_path_cur--;
	if (isset($temp['VALUE'])) {
		$temp['VALUE'] = trim(planet_replace_entities($temp['VALUE']));
	}
}

/**
 * Private call-back function used by the XML parser.
 */
function planet_element_data($parser, $data) {
	global $xml_tree, $xml_paths, $xml_path_cur;

	$temp = trim($data);
	if (strlen($temp) > 0) {
		$temp = &$xml_paths[$xml_path_cur];
		$temp['VALUE'] .= $data;
	}
}

function planet_page_last() {
  global $user;

  $output = '';

  $result = pager_query(db_rewrite_sql("SELECT n.nid, n.created FROM {node} n WHERE n.type = 'planet' AND n.status = 1 ORDER BY n.created DESC"), variable_get('default_nodes_main', 10));

  while ($node = db_fetch_object($result)) {
	$node = node_load($node->nid);
	$node->type = 'blog';
	$output .= node_view($node);
  }
  $output .= theme('pager', NULL, variable_get('default_nodes_main', 10));
  $output .= theme('xml_icon', url('blog/feed'));

  drupal_add_link(array('rel' => 'alternate',
                        'type' => 'application/rss+xml',
                        'title' => t('RSS - blogs'),
                        'href' => url("blog/feed")));
  print theme('page', $output);
}

function planet_feed() {
  $result = db_query_range(db_rewrite_sql("SELECT n.nid, n.title, n.body teaser, n.created, u.name, u.uid FROM {node} n INNER JOIN {users} u ON n.uid = u.uid WHERE n.type = 'planet' AND n.status = 1 ORDER BY n.created DESC"), 0, 15);
  //$channel['title'] = variable_get('site_name', 'drupal') .' blogs';
  $channel['title'] = 'Planet Blog';
  $channel['link'] = url('planet', NULL, NULL, TRUE);
  $channel['description'] = 'Planet feed';
  node_feed($result, $channel);
}

/**
 * Implementation of hook_user().
 */
function planet_user($type, &$edit, &$user) {
  if ($type == 'view' && user_access('edit own blog', $user)) {
    $items[] = array('title' => t('Planet'),
      'value' => l(t('view recent planet entries'), "planet/$user->uid", array('title' => t("Read %username's latest planet entries.", array('%username' => $user->name)))),
      'class' => 'planet',
    );
    return array(t('History') => $items);
  }
}

/**
 * Menu callback; displays a Drupal page containing recent planet entries.
 */
function planet_page($a = NULL, $b = NULL) {

  if (is_numeric($a)) { // $a is a user ID
    if ($b == 'feed') {
      return planet_feed_user($a);
    }
    else {
      return planet_page_user($a);
    }
  }
  else if ($a == 'feed') {
    return planet_feed_last();
  }
  else {
    return planet_page_last();
  }
}

function planet_page_user($uid) {
  global $user;

  $account = user_load(array((is_numeric($uid) ? 'uid' : 'name') => $uid, 'status' => 1));

  if ($account->uid) {
    drupal_set_title($title = t("%name's planet", array('%name' => $account->name)));

    if ($output) {
      $output = '<ul>'. $output .'</ul>';
    }
    else {
      $output = '';
    }
     $result = pager_query(db_rewrite_sql("SELECT n.nid, n.sticky, n.created FROM {node} n WHERE type = 'planet' AND n.uid = %d AND n.status = 1 ORDER BY n.sticky DESC, n.created DESC"), variable_get('default_nodes_main', 10), 0, NULL, $account->uid);
    while ($node = db_fetch_object($result)) {
      $output .= node_view(node_load($node->nid), 1);
    }
    $output .= theme('pager', NULL, variable_get('default_nodes_main', 10));
    $output .= theme('feed_icon', url("planet/$account->uid/feed"));

    drupal_add_link(array('rel' => 'alternate',
                          'type' => 'application/rss+xml',
                          'title' => t('RSS - %title', array('%title' => $title)),
                          'href' => url("planet/$account->uid/feed")));
    return $output;
  }
  else{
    drupal_not_found();
  }
}

function planet_load($node){
	$additions = db_fetch_object(db_query('SELECT guid FROM {planet_items} WHERE nid = %d', $node->nid));
	return $additions;
}

function planet_form(&$node, &$param){
	$form = array();
	$form['title'] = array('#type' => 'textfield', '#title' => 'Title', '#value' => $node->title, '#size' => 30, '#maxlength' => 80);
	$form['body'] = array('#type' => 'textarea', '#title' => 'Body', '#value' => $node->body);
	return $form;
}


?>