.
*
* @version $Id: index.php 68 2007-07-26 16:42:19Z epsi $
* @author Stephane Daury: http://stephane.daury.org/
* @license GNU Public License: http://www.gnu.org/licenses/gpl.html
*/
new parseMe();
class parseMe {
/**
* Admin-defined configurable options.
* Associative array containing the application-level settings.
* @var array
*/
private $_config = array(
'appName' => 'parseMe',
'cache' => array(
'directory' => './cache',
'timeToLive' => '30 minutes',
'secret' => ''
),
'xmlSourceList' => array(
'tekArtist' => 'http://feeds.feedburner.com/tekArtist',
'Digg: Tech.' => 'http://digg.com/rss/containertechnology.xml',
'Slashdot' => 'http://rss.slashdot.org/Slashdot/slashdot',
'Google News' => 'http://news.google.com/?output=rss'
),
'limits' => array(
'increment' => 5,
'max' => 25,
'default' => 5,
'description' => 1024
),
'order' => 'feed',
'googleMobile' => true
);
/**
* User-agent for connections, updated in __construct().
* @var string
*/
private $_userAgent = 'parseMe';
/**
* Cookie name for source preferences.
* @var string
*/
private $_cookieDefaultSources = 'parseMe_defaultSource';
/**
* Cookie name for items-per-page preferences.
* @var string
*/
private $_cookieDefaultLimit = 'parseMe_defaultLimit';
/**
* Cookie name for sorting preferences.
* @var string
*/
private $_cookieDefaultOrder = 'parseMe_defaultOrder';
/**
* Cookie name for Google Mobile Gateway preferences.
* @var string
*/
private $_cookieDefaultGoogle = 'parseMe_defaultGoogle';
/**
* Cache file modification time (Epoch).
* @var integer
*/
private $_cacheMtime = 0;
/**
* Associative array containing the unified data to display, regardless
* of the feed(s) type.
* @var array
*/
private $_feedData = array();
/**
* date() format string used to generate keys for $this->_feedData
* @var string
*/
private $_feedDataKeyFormat = 'Y-m-d H:i:s';
/**
* Valid sorting options
* @var array
*/
private $_orderTypes = array(
'feed' => 'By feeds',
'desc' => 'Newest first',
'asc' => 'Oldest first'
);
/**
* parseMe homepage, for credits
* @var string
*/
private $_parseMeLink = 'http://www.tekartist.org/labs/parseme/';
/**
* Initializes custom user-agent and assigns values
*
* @param array Admin-defined application settings
* @return void
*/
public function __construct() {
$this->_userAgent = $this->_config['appName'].'('.$_SERVER['SCRIPT_URI'].')';
$this->_cacheMtime = time();
$this->_init();
}
/**
* Gets the user or application default xml sources.
*
* @return array Array keys in $this->_config['xmlSourceList']
*/
private function _getDefaultSources(){
$xmlSourceList = $this->_config['xmlSourceList'];
$sourceKeys = array_keys($xmlSourceList);
$defaultSource = $sourceKeys[0];
if(isset($_COOKIE[$this->_cookieDefaultSources])){
$userCookieSources = unserialize(base64_decode($_COOKIE[$this->_cookieDefaultSources]));
if(is_array($userCookieSources)){
foreach($userCookieSources as $thisSource){
if(isset($xmlSourceList[$thisSource])){
$userDefaultSources[] = $thisSource;
}
}
}
}
if(count($userDefaultSources) > 0){
return $userDefaultSources;
}else{
return array($defaultSource);
}
}
/**
* Gets the user or application default items-per-page limit.
*
* @return integer
*/
private function _getDefaultLimit(){
$limit = intval($_COOKIE[$this->_cookieDefaultLimit]);
if($limit > 0 && $limit <= $this->_config['limits']['max']){
return $limit;
}
else{
return $this->_config['limits']['default'];
}
}
/**
* Gets the user or application default sort option.
*
* @return string Array key in $this->_orderTypes.
*/
private function _getDefaultOrder(){
$order = $_COOKIE[$this->_cookieDefaultOrder];
if(isset($this->_orderTypes[$order])){
return $order;
}
else{
return $this->_config['order'];
}
}
/**
* Gets the user or application default Google Mobile Gateway pref.
*
* @return boolean
*/
private function _getDefaultGoogle(){
$useGoogle = unserialize(base64_decode($_COOKIE[$this->_cookieDefaultGoogle]));
if( isset($_COOKIE[$this->_cookieDefaultGoogle])
&& ($useGoogle === true || $useGoogle === false)){
return $useGoogle;
}
else{
return $this->_config['googleMobile'];
}
}
/**
* Generates fake date-based keys to be used with $this->_feedData,
* when an item does not have a publication date.
*
* @return string Date as defined by $this->_feedDataKeyFormat
*/
private function _bogusDateKey(){
return date($this->_feedDataKeyFormat,rand(strtotime('yesterday'), time())).' [bogus]';
}
/**
* Feed parsing decisional wrapper.
*
* @param object SimpleXML object (http://php.net/simplexml)
* @return void
*/
private function _parseFeed($xml){
if(isset($xml->channel)){
$this->_parseRSS($xml);
}
elseif(isset($xml->entry)){
$this->_parseAtom($xml);
}
}
/**
* RSS handling (http://en.wikipedia.org/wiki/RSS_%28file_format%29).
*
* @param object SimpleXML object (http://php.net/simplexml)
* @return void
*/
private function _parseRSS($xml){
$nameSpaces = $xml->getDocNamespaces(TRUE);
$descLimit = $this->_config['limits']['description'];
$feedTitle = $xml->channel->title;
$feedLink = $xml->channel->link;
if(isset($xml->item)){
$feedEntries = $xml->item;
}
elseif(isset($xml->channel->item)){
$feedEntries = $xml->channel->item;
}
else{
$feedEntries = false;
}
if(is_object($feedEntries)){
$keyFormat = $this->_feedDataKeyFormat;
foreach($feedEntries as $entry){
if(isset($entry->pubDate)){
$key = date($keyFormat, strtotime($entry->pubDate));
if(isset($this->_feedData[$key])){
$key = $this->_bogusDateKey();
}
}
elseif(isset($nameSpaces['dc'])){
// argh, custom namespaces kinda suck w/ simplexml...
$entry->registerXPathNamespace('dc', $nameSpaces['dc']);
$xpath = $entry->xpath('dc:date');
if(list( , $date) = each($xpath[0])){
$key = date($keyFormat, strtotime($date));
if(isset($this->_feedData[$key])){
$key = $this->_bogusDateKey();
}
}
}
else{
$key = $this->_bogusDateKey();
}
$data = array();
$data['feedTitle'] = strip_tags($feedTitle);
$data['feedLink'] = strip_tags($feedLink);
$data['itemTitle'] = strip_tags($entry->title);
$data['itemLink'] = strip_tags($entry->link);
$itemDesc = strip_tags(str_replace("\n",
' ',
trim($entry->description)));
if(strlen($itemDesc) > $descLimit){
$itemDesc = substr($itemDesc, 0, $descLimit).'...';
}
$data['itemDesc'] = $itemDesc;
$this->_feedData[$key] = $data;
}
}
}
/**
* Atom handling (http://en.wikipedia.org/wiki/Atom_%28standard%29).
*
* @param object SimpleXML object (http://php.net/simplexml)
* @return void
*/
private function _parseAtom($xml){
$descLimit = $this->_config['limits']['description'];
$feedLinkAttributes = $xml->link->attributes();
$feedTitle = $xml->title;
$feedLink = $feedLinkAttributes['href'];
if(isset($xml->entry)){
$feedEntries = $xml->entry;
}
else{
$feedEntries = false;
}
if(is_object($feedEntries)){
$key = $this->_feedDataKeyFormat;
foreach($feedEntries as $entry){
if(isset($entry->pubDate)){
$key = date($keyFormat, strtotime($entry->pubDate));
if(isset($this->_feedData[$key])){
$key = $this->_bogusDateKey();
}
}
else{
$key = $this->_bogusDateKey();
}
$entryAttributes = $entry->link->attributes();
$data = array();
$data['feedTitle'] = strip_tags($feedTitle);
$data['feedLink'] = strip_tags($feedLink);
$data['itemTitle'] = strip_tags($entry->title);
$data['itemLink'] = strip_tags($entryAttributes['href']);
$itemDesc = strip_tags(str_replace("\n",
' ',
trim($entry->content)));
if(strlen($itemDesc) > $descLimit){
$itemDesc = substr($itemDesc, 0, $descLimit).'...';
}
$data['itemDesc'] = $itemDesc;
$this->_feedData[$key] = $data;
}
}
}
/**
* Fetches the feed(s), populates and caches $this->_feedData if needed, or
* acquires $this->_feedData from existing cache file.
*
* @param array Array keys in $this->_config['xmlSourceList']
* @return boolean True on success, false on error
*/
private function _getData($sourceChoices){
$returnVal = false;
$secret = $this->_config['cache']['secret'];
$scriptName = basename($_SERVER['PHP_SELF']);
if($secret == ''){
print " Sorry, but you have not defined a new secret key for secure cache filename generation.
\n"
. " Please be sure to set a random string for \$_config['cache']['secret'] in {$scriptName}.
\n";
}
else{
$cacheDir = $this->_config['cache']['directory'];
if(!function_exists('curl_init')){
print " Sorry, but your PHP install is not configured .
\n"
. " Please create a writable directory at {$cacheDir}.
\n";
}
elseif(!file_exists($cacheDir)){
print " Sorry, but the cache directory cannot be found.
\n"
. " Please create a writable directory at {$cacheDir}.
\n";
}
elseif(!is_dir($cacheDir)){
print " Sorry, but the cache location is not a directory.
\n"
. " Please create a writable directory at {$cacheDir}.
\n";
}
elseif(!is_writable($cacheDir)){
print " Sorry, but the cache directory is not writable.
\n"
. " Please make the directory writable: {$cacheDir}.
\n";
}
else{
$feedData = array();
$cacheFileName = sha1(serialize($sourceChoices).$secret);
$cacheDestination = "{$cacheDir}/.{$cacheFileName}.cache";
if( ( ! file_exists($cacheDestination))
|| (filemtime($cacheDestination) < strtotime($this->_config['cache']['timeToLive'].' ago')) ){
foreach($sourceChoices as $name){
$link = $this->_config['xmlSourceList'][$name];
$cleanName = htmlspecialchars($name);
if(function_exists('curl_init')){
// try curl first
$curlRes = curl_init($link);
curl_setopt_array($curlRes, array(
CURLOPT_RETURNTRANSFER => 1,
CURLOPT_USERAGENT => $this->_userAgent
));
if(($xmlSource = @curl_exec($curlRes)) === false){
print " Sorry, but \"{$cleanName}\" cannot be reached at this time. (curl)
\n";
}
}
else{
// or use URL-aware fopen wrappers
ini_set('user_agent', $this->_userAgent);
if(($xmlSource = @file_get_contents($link)) === false){
print " Sorry, but \"{$cleanName}\" cannot be reached at this time.
\n";
}
}
if( !isset($xmlSource) || (strlen($xmlSource) == 0) ){
print " Nothing to parse for \"{$cleanName}\".
\n";
}
elseif(($xml = @simplexml_load_string(trim($xmlSource))) === false){
print " Sorry, but \"{$cleanName}\" cannot be parsed at this time.
\n";
}
else{
$this->_parseFeed($xml);
}
}
if(count($this->_feedData) == 0){
print " Sorry, but every feed failed, and there was nothing to cache or display.
\n";
}
else{
@file_put_contents($cacheDestination,serialize($this->_feedData));
$returnVal = true;
}
}
else{
if(($cache = unserialize(@file_get_contents($cacheDestination))) === false){
print " Sorry, but \"{$cacheFileName}\" cannot be unserialized at this time.
\n";
}
else{
$this->_feedData = $cache;
$this->_cacheMtime = filemtime($cacheDestination);
$returnVal = true;
}
}
}
}
return $returnVal;
}
/**
* Main processing function.
*
* @return void
*/
private function _init(){
$xmlSourceList = $this->_config['xmlSourceList'];
$cacheDir = $this->_config['cache']['directory'];
$limitIncrement = $this->_config['limits']['increment'];
$maxLimit = $this->_config['limits']['max'];
$defaultLimit = $this->_config['limits']['default'];
$defaultOrder = $this->_config['order'];
$defaultOffset = 0;
$cookieLifespan = strtotime('5 years');
$googleMobileUrl = 'http://www.google.com/gwt/n?_gwt_noimg=1&u=';
// Get and clean user input
$sentChoices = isset($_GET['src']) ? $_GET['src'] : array();
$limit = isset($_GET['lmt']) ? intval($_GET['lmt']) : $defaultLimit;
$offset = isset($_GET['ofst']) ? intval($_GET['ofst']) : $defaultOffset;
$order = isset($_GET['order']) ? $_GET['order'] : $defaultOrder;
$useGoogle = intval($_GET['gmg']) > 0 ? true : false;
$saveDefault = intval($_GET['sv']) > 0 ? true : false;
if(!is_array($sentChoices)){
$sentChoices = array();
}
if( (count($sentChoices) == 0)
|| (count($sentChoices) > count($xmlSourceList)) ){
// No or too many choices sent, get default (user | app)
$sourceChoices = $this->_getDefaultSources();
}
else{
$CleanSources = array();
foreach($sentChoices as $thisSource){
if(isset($xmlSourceList[$thisSource])){
// Add as safe source
$CleanSources[] = $thisSource;
}
}
if(count($CleanSources) == 0){
// All sent sources are invalid, get default (user | app)
$sourceChoices = $this->_getDefaultSources();
}
else{
// We have at leats one clean source, move on
$sourceChoices = $CleanSources;
if($saveDefault > 0){
// Save as a new user default if requested.
setcookie(
$this->_cookieDefaultSources,
base64_encode(serialize($sourceChoices)),
$cookieLifespan,
'/'
);
}
}
unset($CleanSources);
}
if( ( ! isset($_GET['lmt'])) || ($limit < $limitIncrement) || ($limit > $maxLimit) ){
// Invalid limit sent, get default (user | app)
$limit = $this->_getDefaultLimit();
}
elseif($saveDefault > 0){
// Save as a new user default if requested.
setcookie($this->_cookieDefaultLimit, $limit, $cookieLifespan, '/');
}
if( ($offset < 1) || ($offset > 100) ){
// Invalid offset sent, use default
$offset = $defaultOffset;
}
if( ( ! isset($_GET['order'])) || ( ! isset($this->_orderTypes[$order]))){
// Invalid order sent, get default (user|app)
$order = $this->_getDefaultOrder();
}
elseif($saveDefault > 0){
// Save as a new user default if requested.
setcookie($this->_cookieDefaultOrder, $order, $cookieLifespan, '/');
}
if( ! isset($_GET['gmg'])){
// No Google Mobile prefs sent at all, get default (user|app)
$useGoogle = $this->_getDefaultGoogle();
}
elseif($saveDefault > 0){
// Save as a new user default if requested.
setcookie(
$this->_cookieDefaultGoogle,
base64_encode(serialize($useGoogle)),
$cookieLifespan,
'/'
);
}
// Set page title and header
$title = (count($sourceChoices) == 1)
? htmlspecialchars($sourceChoices[0])
: 'Mixed Sources';
// Print head block w/ defined and clean info
print '<'.'?'.'xml version="1.0"?'.">\n"
. "\n"
. "\n"
. " \n"
. " {$this->_config['appName']} » {$title}\n"
. " \n"
. " \n"
. " \n"
. " \n"
. " {$title}
\n";
if($this->_getData($sourceChoices)){
$feedData = $this->_feedData;
if($order == 'asc'){
ksort($feedData);
}
elseif($order == 'desc'){
krsort($feedData);
}
if(count($feedData) > 0){
print " \n";
$i = 0;
foreach($feedData as $date => $entry){
if($i >= $offset){
printf(
' - %s
'."\n",
(
$useGoogle
? $googleMobileUrl.rawurlencode($entry['itemLink'])
: $entry['itemLink']
),
$entry['itemTitle']
);
printf(
' - %s: %s
'."\n",
(
$useGoogle
? $googleMobileUrl.rawurlencode($entry['feedLink'])
: $entry['feedLink']
),
$entry['feedTitle'],
$entry['itemDesc']
);
if(($i + 1) >= ($offset + $limit)) break;
}
$i++;
}
print "
\n";
}
// Deal with the paging toolbar
if(count($feedData) > $limit){
print " | \n";
$link = $_SERVER['REQUEST_URI'];
if(strstr($link, 'ofst=') === false) $link .= '?';
// "Previous" link
if($offset > 0){
if(strstr($link, 'ofst=')) $link = preg_replace('/(\&ofst=\d{1,2})/','',$link);
$link .= '&ofst='.($offset - $limit);
print " « previous\n";
}
// "Next" link
if(count($feedData) > ($limit + $offset)){
if($offset > 0) print " | \n";
if(strstr($link, 'ofst=')) $link = preg_replace('/(\&ofst=\d{1,2})/','',$link);
$link .= '&ofst='.($limit + $offset);
print " next »\n";
}
print " |
\n";
}
// Print the cache specs
$cacheMtime = $this->_cacheMtime;
$parseMeLink = $this->_parseMeLink;
if($useGoogle){
$parseMeLink = $googleMobileUrl.rawurlencode($parseMeLink);
}
print " \n"
. " \n";
}
print " \n"
. "\n";
}
}
?>