This commit is contained in:
logmanoriginal 2016-08-24 20:55:13 +02:00
commit 73ec10f9ca
6 changed files with 288 additions and 281 deletions

View file

@ -1,6 +1,6 @@
<?php <?php
define("FREENEWS_RSS", 'http://feeds.feedburner.com/Freenews-Freebox?format=xml'); define("FREENEWS_RSS", 'http://feeds.feedburner.com/Freenews-Freebox?format=xml');
class Freenews extends RssExpander { class FreenewsBridge extends RssExpander {
public function loadMetadatas() { public function loadMetadatas() {
@ -21,14 +21,14 @@ class Freenews extends RssExpander {
protected function parseRSSItem($newsItem) { protected function parseRSSItem($newsItem) {
$item = array(); $item = array();
$item['title'] = trim($newsItem->title); $item['title'] = trim($newsItem->title);
// $this->message("item has for title \"".$item['title']."\""); $this->debugMessage("item has for title \"".$item['title']."\"");
if(empty($newsItem->guid)) { if(empty($newsItem->guid)) {
$item['uri'] = (string) $newsItem->link; $item['uri'] = (string) $newsItem->link;
} else { } else {
$item['uri'] = (string) $newsItem->guid; $item['uri'] = (string) $newsItem->guid;
} }
// now load that uri from cache // now load that uri from cache
// $this->message("now loading page ".$item['uri']); $this->debugMessage("now loading page ".$item['uri']);
$articlePage = str_get_html($this->get_cached($item['uri'])); $articlePage = str_get_html($this->get_cached($item['uri']));
$content = $articlePage->find('.post-container', 0); $content = $articlePage->find('.post-container', 0);

View file

@ -2,7 +2,7 @@
define("RSS_PREFIX", "http://feeds.gawker.com/"); define("RSS_PREFIX", "http://feeds.gawker.com/");
define("RSS_SUFFIX", "/full"); define("RSS_SUFFIX", "/full");
class Gawker extends RssExpander{ class GawkerBridge extends RssExpander{
public function loadMetadatas() { public function loadMetadatas() {
@ -28,7 +28,7 @@ class Gawker extends RssExpander{
$this->name = $param['site']; $this->name = $param['site'];
$url = $this->toURI(strtolower($param['site'])); $url = $this->toURI(strtolower($param['site']));
} }
// $this->message("loading feed from ".$this->getURI()); $this->debugMessage("loading feed from ".$this->getURI());
parent::collectExpandableDatas($param, $url); parent::collectExpandableDatas($param, $url);
} }
@ -37,10 +37,10 @@ class Gawker extends RssExpander{
$item['uri'] = trim($newsItem->link); $item['uri'] = trim($newsItem->link);
$item['title'] = trim($newsItem->title); $item['title'] = trim($newsItem->title);
$item['timestamp'] = $this->RSS_2_0_time_to_timestamp($newsItem); $item['timestamp'] = $this->RSS_2_0_time_to_timestamp($newsItem);
// $this->message("///////////////////////////////////////////////////////////////////////////////////////\nprocessing item ".var_export($item, true)."\n\n\nbuilt from\n\n\n".var_export($newsItem, true)); $this->debugMessage("///////////////////////////////////////////////////////////////////////////////////////\nprocessing item ".var_export($item, true)."\n\n\nbuilt from\n\n\n".var_export($newsItem, true));
try { try {
// now load that uri from cache // now load that uri from cache
// $this->message("loading page ".$item['uri']); $this->debugMessage("loading page ".$item['uri']);
$articlePage = str_get_html($this->get_cached($item['uri'])); $articlePage = str_get_html($this->get_cached($item['uri']));
if(is_object($articlePage)) { if(is_object($articlePage)) {
$content = $articlePage->find('.post-content', 0); $content = $articlePage->find('.post-content', 0);
@ -51,14 +51,14 @@ class Gawker extends RssExpander{
$item['author'] = $authorLink->innertext; $item['author'] = $authorLink->innertext;
// TODO use author link href to fill the feed info // TODO use author link href to fill the feed info
} }
// $this->message("item quite loaded : ".var_export($item, true)); $this->debugMessage("item quite loaded : ".var_export($item, true));
// I set item content as last element, for easier var_export reading // I set item content as last element, for easier var_export reading
$item['content'] = $content->innertext; $item['content'] = $content->innertext;
} else { } else {
throw new Exception("cache content for ".$item['uri']." is NOT a Simple DOM parser object !"); throw new Exception("cache content for ".$item['uri']." is NOT a Simple DOM parser object !");
} }
} catch(Exception $e) { } catch(Exception $e) {
$this->message("obtaining ".$item['uri']." resulted in exception ".$e->getMessage().". Deleting cached page ..."); $this->debugMessage("obtaining ".$item['uri']." resulted in exception ".$e->getMessage().". Deleting cached page ...");
// maybe file is incorrect. it should be discarded from cache // maybe file is incorrect. it should be discarded from cache
$this->remove_from_cache($item['url']); $this->remove_from_cache($item['url']);
$item['content'] = $e->getMessage(); $item['content'] = $e->getMessage();

View file

@ -2,7 +2,7 @@
define("SEXE", "http://sexes.blogs.liberation.fr"); define("SEXE", "http://sexes.blogs.liberation.fr");
define("SEXE_FEED", "http://sexes.blogs.liberation.fr/feeds/"); define("SEXE_FEED", "http://sexes.blogs.liberation.fr/feeds/");
class Les400Culs extends RssExpander{ class Les400CulsBridge extends RssExpander{
public function loadMetadatas() { public function loadMetadatas() {
@ -21,14 +21,14 @@ class Les400Culs extends RssExpander{
protected function parseRSSItem($newsItem) { protected function parseRSSItem($newsItem) {
$item = array(); $item = array();
$item['title'] = trim((string) $newsItem->title); $item['title'] = trim((string) $newsItem->title);
// $this->message("browsing item ".var_export($newsItem, true)); $this->debugMessage("browsing item ".var_export($newsItem, true));
if(empty($newsItem->guid)) { if(empty($newsItem->guid)) {
$item['uri'] = (string) $newsItem->link; $item['uri'] = (string) $newsItem->link;
} else { } else {
$item['uri'] = (string) $newsItem->guid; $item['uri'] = (string) $newsItem->guid;
} }
// now load that uri from cache // now load that uri from cache
// $this->message("now loading page ".$item['uri']); $this->debugMessage("now loading page ".$item['uri']);
// $articlePage = str_get_html($this->get_cached($item['uri'])); // $articlePage = str_get_html($this->get_cached($item['uri']));
// $content = $articlePage->find('.post-container', 0); // $content = $articlePage->find('.post-container', 0);

View file

@ -24,10 +24,10 @@ class TheOatmealBridge extends RssExpander{
protected function collect_RSS_2_0_data($rssContent) { protected function collect_RSS_2_0_data($rssContent) {
$rssContent->registerXPathNamespace("dc", "http://purl.org/dc/elements/1.1/"); $rssContent->registerXPathNamespace("dc", "http://purl.org/dc/elements/1.1/");
$rssHeaderContent = $rssContent->channel[0]; $rssHeaderContent = $rssContent->channel[0];
// $this->message("RSS content is ===========\n".var_export($rssHeaderContent, true)."==========="); $this->debugMessage("RSS content is ===========\n".var_export($rssHeaderContent, true)."===========");
$this->load_RSS_2_0_feed_data($rssHeaderContent); $this->load_RSS_2_0_feed_data($rssHeaderContent);
foreach($rssContent->item as $item) { foreach($rssContent->item as $item) {
$this->message("parsing item ".var_export($item, true)); $this->debugMessage("parsing item ".var_export($item, true));
$this->items[] = $this->parseRSSItem($item); $this->items[] = $this->parseRSSItem($item);
} }
} }
@ -39,10 +39,10 @@ class TheOatmealBridge extends RssExpander{
$rdf = $newsItem->children($namespaces['rdf']); $rdf = $newsItem->children($namespaces['rdf']);
$item = array(); $item = array();
$item['title'] = trim($newsItem->title); $item['title'] = trim($newsItem->title);
$this->message("browsing Oatmeal item ".var_export($newsItem, true)); $this->debugMessage("browsing Oatmeal item ".var_export($newsItem, true));
$item['uri']=(string) $newsItem->attributes($namespaces['rdf'])->about; $item['uri']=(string) $newsItem->attributes($namespaces['rdf'])->about;
// now load that uri from cache // now load that uri from cache
$this->message("now loading page ".$item['uri']); $this->debugMessage("now loading page ".$item['uri']);
$articlePage = str_get_html($this->get_cached($item['uri'])); $articlePage = str_get_html($this->get_cached($item['uri']));
$content = $articlePage->find('#comic', 0); $content = $articlePage->find('#comic', 0);
@ -51,10 +51,10 @@ class TheOatmealBridge extends RssExpander{
} }
$item['content'] = $content->innertext; $item['content'] = $content->innertext;
$this->message("dc content is ".var_export($dc, true)); $this->debugMessage("dc content is ".var_export($dc, true));
$item['author'] = (string) $dc->creator; $item['author'] = (string) $dc->creator;
$item['timestamp'] = DateTime::createFromFormat(DateTime::ISO8601, $dc->date)->getTimestamp(); $item['timestamp'] = DateTime::createFromFormat(DateTime::ISO8601, $dc->date)->getTimestamp();
$this->message("writtem by ".$item['author']." on ".$item['timestamp']); $this->debugMessage("writtem by ".$item['author']." on ".$item['timestamp']);
return $item; return $item;
} }

View file

@ -1,7 +1,7 @@
<?php <?php
define('WORLD_OF_TANKS', 'http://worldoftanks.eu/'); define('WORLD_OF_TANKS', 'http://worldoftanks.eu/');
define('NEWS', '/news/'); define('NEWS', '/news/');
class WorldOfTanks extends HttpCachingBridgeAbstract{ class WorldOfTanksBridge extends HttpCachingBridgeAbstract{
private $lang = "fr"; private $lang = "fr";
public $uri = WORLD_OF_TANKS; public $uri = WORLD_OF_TANKS;
@ -45,7 +45,7 @@ class WorldOfTanks extends HttpCachingBridgeAbstract{
$this->uri = WORLD_OF_TANKS.$this->lang.NEWS.'pc-browser/'.$param['category']."/"; $this->uri = WORLD_OF_TANKS.$this->lang.NEWS.'pc-browser/'.$param['category']."/";
} }
$html = $this->getSimpleHTMLDOM($this->getURI()) or $this->returnServerError('Could not request '.$this->getURI()); $html = $this->getSimpleHTMLDOM($this->getURI()) or $this->returnServerError('Could not request '.$this->getURI());
$this->message("loaded HTML from ".$this->getURI()); $this->debugMessage("loaded HTML from ".$this->getURI());
// customize name // customize name
$this->name = $html->find('title', 0)->innertext; $this->name = $html->find('title', 0)->innertext;
foreach($html->find('.b-imgblock_ico') as $infoLink) { foreach($html->find('.b-imgblock_ico') as $infoLink) {
@ -57,7 +57,7 @@ class WorldOfTanks extends HttpCachingBridgeAbstract{
$item = array(); $item = array();
$item['uri'] = WORLD_OF_TANKS.$infoLink->href; $item['uri'] = WORLD_OF_TANKS.$infoLink->href;
// now load that uri from cache // now load that uri from cache
// $this->message("loading page ".$item['uri']); $this->debugMessage("loading page ".$item['uri']);
$articlePage = str_get_html($this->get_cached($item['uri'])); $articlePage = str_get_html($this->get_cached($item['uri']));
$content = $articlePage->find('.l-content', 0); $content = $articlePage->find('.l-content', 0);
HTMLSanitizer::defaultImageSrcTo($content, WORLD_OF_TANKS); HTMLSanitizer::defaultImageSrcTo($content, WORLD_OF_TANKS);

View file

@ -1,10 +1,108 @@
<?php <?php
/** class Bridge {
* All bridge logic
* Note : adapter are store in other place
*/
interface BridgeInterface{ static protected $dirBridge;
public function __construct(){
throw new \LogicException('Please use ' . __CLASS__ . '::create for new object.');
}
/**
* Checks if a bridge is an instantiable bridge.
* @param string $nameBridge name of the bridge that you want to use
* @return true if it is an instantiable bridge, false otherwise.
*/
static public function isInstantiable($nameBridge){
$re = new ReflectionClass($nameBridge);
return $re->IsInstantiable();
}
/**
* Create a new bridge object
* @param string $nameBridge Defined bridge name you want use
* @return Bridge object dedicated
*/
static public function create($nameBridge){
if(!preg_match('@^[A-Z][a-zA-Z0-9-]*$@', $nameBridge)){
$message = <<<EOD
'nameBridge' must start with one uppercase character followed or not by
alphanumeric or dash characters!
EOD;
throw new \InvalidArgumentException($message);
}
$nameBridge = $nameBridge . 'Bridge';
$pathBridge = self::getDir() . $nameBridge . '.php';
if(!file_exists($pathBridge)){
throw new \Exception('The bridge you looking for does not exist. It should be at path ' . $pathBridge);
}
require_once $pathBridge;
if(Bridge::isInstantiable($nameBridge)){
return new $nameBridge();
} else {
return false;
}
}
static public function setDir($dirBridge){
if(!is_string($dirBridge)){
throw new \InvalidArgumentException('Dir bridge must be a string.');
}
if(!file_exists($dirBridge)){
throw new \Exception('Dir bridge does not exist.');
}
self::$dirBridge = $dirBridge;
}
static public function getDir(){
$dirBridge = self::$dirBridge;
if(is_null($dirBridge)){
throw new \LogicException(__CLASS__ . ' class need to know bridge path !');
}
return $dirBridge;
}
/**
* Lists the available bridges.
* @return array List of the bridges
*/
static public function listBridges(){
$pathDirBridge = self::getDir();
$listBridge = array();
$dirFiles = scandir($pathDirBridge);
if($dirFiles !== false){
foreach($dirFiles as $fileName){
if(preg_match('@^([^.]+)Bridge\.php$@U', $fileName, $out)){
$listBridge[] = $out[1];
}
}
}
return $listBridge;
}
static public function isWhitelisted($whitelist, $name){
if(in_array($name, $whitelist)
or in_array($name . '.php', $whitelist)
or in_array($name . 'Bridge', $whitelist) // DEPRECATED
or in_array($name . 'Bridge.php', $whitelist) // DEPRECATED
or count($whitelist) === 1 and trim($whitelist[0]) === '*'){
return true;
} else {
return false;
}
}
}
interface BridgeInterface {
public function collectData(array $param); public function collectData(array $param);
public function getCacheDuration(); public function getCacheDuration();
public function loadMetadatas(); public function loadMetadatas();
@ -12,21 +110,18 @@ interface BridgeInterface{
public function getURI(); public function getURI();
} }
abstract class BridgeAbstract implements BridgeInterface{ abstract class BridgeAbstract implements BridgeInterface {
protected $cache; protected $cache;
protected $items = array(); protected $items = array();
public $name = "Unnamed bridge"; public $name = 'Unnamed bridge';
public $uri = ""; public $uri = '';
public $description = 'No description provided'; public $description = 'No description provided';
public $maintainer = 'No maintainer'; public $maintainer = 'No maintainer';
public $useProxy = true; public $useProxy = true;
public $parameters = array(); public $parameters = array();
/**
* Launch probative exception
*/
protected function returnError($message, $code){ protected function returnError($message, $code){
throw new \HttpException($message, $code); throw new \HttpException($message, $code);
} }
@ -40,340 +135,254 @@ abstract class BridgeAbstract implements BridgeInterface{
} }
/** /**
* Return datas stored in the bridge * Return items stored in the bridge
* @return mixed * @return mixed
*/ */
public function getDatas(){ public function getDatas(){
return $this->items; return $this->items;
} }
/** /**
* Defined datas with parameters depending choose bridge * Defined datas with parameters depending choose bridge
* Note : you can define a cache before with "setCache" * Note : you can define a cache with "setCache"
* @param array $param $_REQUEST, $_GET, $_POST, or array with bridge expected paramters * @param array $param $_REQUEST, $_GET, $_POST, or array with expected
* bridge paramters
*/ */
public function setDatas(array $param){ public function setDatas(array $param){
if( !is_null($this->cache) ){ if(!is_null($this->cache)){
$this->cache->prepare($param); $this->cache->prepare($param);
$time = $this->cache->getTime(); $time = $this->cache->getTime();
} } else {
else{ $time = false;
$time = false; // No cache ? No time !
} }
if( $time !== false && ( time() - $this->getCacheDuration() < $time ) ){ // Cache file has not expired. Serve it. if($time !== false && (time() - $this->getCacheDuration() < $time)){
$this->items = $this->cache->loadData(); $this->items = $this->cache->loadData();
} } else {
else{
$this->collectData($param); $this->collectData($param);
if( !is_null($this->cache) ){ // Cache defined ? We go to refresh is memory :D if(!is_null($this->cache)){
$this->cache->saveData($this->getDatas()); $this->cache->saveData($this->getDatas());
} }
} }
} }
/**
* Define default bridge name
*/
public function getName(){ public function getName(){
return $this->name; return $this->name;
} }
/**
* Define default bridge URI
*/
public function getURI(){ public function getURI(){
return $this->uri; return $this->uri;
} }
/**
* Define default duraction for cache
*/
public function getCacheDuration(){ public function getCacheDuration(){
return 3600; return 3600;
} }
/**
* Defined cache object to use
*/
public function setCache(\CacheAbstract $cache){ public function setCache(\CacheAbstract $cache){
$this->cache = $cache; $this->cache = $cache;
return $this;
} }
public function message($text) { public function debugMessage($text){
if(!file_exists('DEBUG')){ if(!file_exists('DEBUG')) {
return; return;
}
$backtrace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 3);
$calling = $backtrace[2];
$message = $calling["file"].":".$calling["line"]
." class ".get_class($this)."->".$calling["function"]
." - ".$text;
error_log($message);
}
protected function getContents($url,$use_include_path=false,$context=null,$offset=0,$maxlen=null){
$contextOptions = array(
'http' => array(
'user_agent'=>ini_get('user_agent')
),
);
if(defined('PROXY_URL') && $this->useProxy) {
$contextOptions['http']['proxy'] = PROXY_URL;
$contextOptions['http']['request_fulluri'] = true;
if(is_null($context)){
$context = stream_context_create($contextOptions);
} else {
$prevContext=$context;
if(!stream_context_set_option($context,$contextOptions)){
$context=$prevContext;
};
} }
}
if(is_null($maxlen)){ $backtrace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 3);
$content=@file_get_contents($url, $use_include_path, $context, $offset); $calling = $backtrace[2];
}else{ $message = $calling['file'] . ':'
$content=@file_get_contents($url, $use_include_path, $context, $offset,$maxlen); . $calling['line'] . ' class '
} . get_class($this) . '->'
. $calling['function'] . ' - '
. $text;
if($content===false){ error_log($message);
$this->message('Cant\'t download '.$url );
}
return $content;
} }
protected function getSimpleHTMLDOM($url, $use_include_path = false, $context=null, $offset = 0, $maxLen=null, $lowercase = true, $forceTagsClosed=true, $target_charset = DEFAULT_TARGET_CHARSET, $stripRN=true, $defaultBRText=DEFAULT_BR_TEXT, $defaultSpanText=DEFAULT_SPAN_TEXT){ protected function getContents($url
$content=$this->getContents($url,$use_include_path,$context,$offset,$maxLen); , $use_include_path = false
return str_get_html($content,$lowercase,$forceTagsClosed,$target_charset,$stripRN,$defaultBRText,$defaultSpanText); , $context = null
, $offset = 0
, $maxlen = null){
$contextOptions = array(
'http' => array(
'user_agent' => ini_get('user_agent')
),
);
if(defined('PROXY_URL') && $this->useProxy){
$contextOptions['http']['proxy'] = PROXY_URL;
$contextOptions['http']['request_fulluri'] = true;
if(is_null($context)){
$context = stream_context_create($contextOptions);
} else {
$prevContext=$context;
if(!stream_context_set_option($context, $contextOptions)){
$context = $prevContext;
}
}
}
if(is_null($maxlen)){
$content = @file_get_contents($url, $use_include_path, $context, $offset);
} else {
$content = @file_get_contents($url, $use_include_path, $context, $offset, $maxlen);
}
if($content === false)
$this->debugMessage('Cant\'t download ' . $url);
return $content;
} }
protected function getSimpleHTMLDOM($url
, $use_include_path = false
, $context = null
, $offset = 0
, $maxLen = null
, $lowercase = true
, $forceTagsClosed = true
, $target_charset = DEFAULT_TARGET_CHARSET
, $stripRN = true
, $defaultBRText = DEFAULT_BR_TEXT
, $defaultSpanText = DEFAULT_SPAN_TEXT){
$content = $this->getContents($url, $use_include_path, $context, $offset, $maxLen);
return str_get_html($content
, $lowercase
, $forceTagsClosed
, $target_charset
, $stripRN
, $defaultBRText
, $defaultSpanText);
}
} }
/** /**
* Extension of BridgeAbstract allowing caching of files downloaded over http files. * Extension of BridgeAbstract allowing caching of files downloaded over http.
* This is specially useful for sites from Gawker or Liberation networks, which allow pages excerpts top be viewed together on index, while full pages have to be downloaded * TODO allow file cache invalidation by touching files on access, and removing
* separately. * files/directories which have not been touched since ... a long time
* This class mainly provides a get_cached method which will will download the file from its remote location.
* TODO allow file cache invalidation by touching files on access, and removing files/directories which have not been touched since ... a long time
* After all, rss-bridge is not respaw, isn't it ?
*/ */
abstract class HttpCachingBridgeAbstract extends BridgeAbstract { abstract class HttpCachingBridgeAbstract extends BridgeAbstract {
/** /**
* Maintain locally cached versions of pages to download to avoid multiple doiwnloads. * Maintain locally cached versions of pages to download, to avoid multiple downloads.
* A file name is generated by replacing all "/" by "_", and the file is saved below this bridge cache
* @param url url to cache * @param url url to cache
* @return content of file as string * @return content of the file as string
*/ */
public function get_cached($url) { public function get_cached($url){
$simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url); // TODO build this from the variable given to Cache
// TODO build this from the variable given to Cache $cacheDir = __DIR__ . '/../cache/pages/';
$pageCacheDir = __DIR__ . '/../cache/'."pages/"; $filepath = $this->buildCacheFilePath($url, $cacheDir);
$filename = $pageCacheDir.$simplified_url;
if (substr($filename, -1) == '/') { if(file_exists($filepath)){
$filename = $filename."index.html"; $this->debugMessage('loading cached file from ' . $filepath . ' for page at url ' . $url);
} // TODO touch file and its parent, and try to do neighbour deletion
if(file_exists($filename)) { $this->refresh_in_cache($cacheDir, $filepath);
// $this->message("loading cached file from ".$filename." for page at url ".$url); $content = file_get_contents($filepath);
// TODO touch file and its parent, and try to do neighbour deletion } else {
$this->refresh_in_cache($pageCacheDir, $filename); $this->debugMessage('we have no local copy of ' . $url . ' Downloading to ' . $filepath);
$content=file_get_contents($filename); $dir = substr($filepath, 0, strrpos($filepath, '/'));
} else {
// $this->message("we have no local copy of ".$url." Downloading to ".$filename); if(!is_dir($dir)){
$dir = substr($filename, 0, strrpos($filename, '/')); $this->debugMessage('creating directories for ' . $dir);
if(!is_dir($dir)) {
// $this->message("creating directories for ".$dir);
mkdir($dir, 0777, true); mkdir($dir, 0777, true);
} }
$content=$this->getContents($url);
if($content!==false){ $content = $this->getContents($url);
file_put_contents($filename,$content); if($content !== false){
file_put_contents($filepath, $content);
} }
} }
return $content; return $content;
} }
public function get_cached_time($url) { public function get_cached_time($url){
$simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url);
// TODO build this from the variable given to Cache // TODO build this from the variable given to Cache
$pageCacheDir = __DIR__ . '/../cache/'."pages/"; $cacheDir = __DIR__ . '/../cache/pages/';
$filename = $pageCacheDir.$simplified_url; $filepath = $this->buildCacheFilePath($url, $cacheDir);
if (substr($filename, -1) == '/') {
$filename = $filename."index.html"; if(!file_exists($filepath)){
}
if(!file_exists($filename)) {
$this->get_cached($url); $this->get_cached($url);
} }
return filectime($filename);
return filectime($filepath);
} }
private function refresh_in_cache($pageCacheDir, $filename) { private function refresh_in_cache($cacheDir, $filepath){
$currentPath = $filename; $currentPath = $filepath;
while(!$pageCacheDir==$currentPath) { while(!$cacheDir == $currentPath){
touch($currentPath); touch($currentPath);
$currentPath = dirname($currentPath); $currentPath = dirname($currentPath);
} }
} }
public function remove_from_cache($url) { private function buildCacheFilePath($url, $cacheDir){
$simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url); $simplified_url = str_replace(
// TODO build this from the variable given to Cache ['http://', 'https://', '?', '&', '='],
$pageCacheDir = __DIR__ . '/../cache/'."pages/"; ['', '', '/', '/', '/'],
$filename = realpath($pageCacheDir.$simplified_url); $url);
$this->message("removing from cache \"".$filename."\" WELL, NOT REALLY");
// filename is NO GOOD if(substr($cacheDir, -1) !== '/'){
// unlink($filename); $cacheDir .= '/';
}
$filepath = $cacheDir . $simplified_url;
if(substr($filepath, -1) === '/'){
$filepath .= 'index.html';
}
return $filepath;
} }
public function remove_from_cache($url){
// TODO build this from the variable given to Cache
$cacheDir = __DIR__ . '/../cache/pages/';
$filepath = $this->buildCacheFilePath($url, $cacheDir);
$this->debugMessage('removing from cache \'' . $filepath . '\' WELL, NOT REALLY');
// unlink($filepath);
}
} }
class Bridge{ abstract class RssExpander extends HttpCachingBridgeAbstract {
static protected $dirBridge;
public function __construct(){
throw new \LogicException('Please use ' . __CLASS__ . '::create for new object.');
}
/**
* Checks if a bridge is an instantiable bridge.
* @param string $nameBridge name of the bridge that you want to use
* @return true if it is an instantiable bridge, false otherwise.
*/
static public function isInstantiable($nameBridge) {
$re = new ReflectionClass($nameBridge);
return $re->IsInstantiable();
}
/**
* Create a new bridge object
* @param string $nameBridge Defined bridge name you want use
* @return Bridge object dedicated
*/
static public function create($nameBridge){
if( !preg_match('@^[A-Z][a-zA-Z0-9-]*$@', $nameBridge)){
throw new \InvalidArgumentException('Name bridge must be at least one uppercase follow or not by alphanumeric or dash characters.');
}
$nameBridge=$nameBridge.'Bridge';
$pathBridge = self::getDir() . $nameBridge . '.php';
if( !file_exists($pathBridge) ){
throw new \Exception('The bridge you looking for does not exist. It should be at path '.$pathBridge);
}
require_once $pathBridge;
if(Bridge::isInstantiable($nameBridge)) {
return new $nameBridge();
} else {
return FALSE;
}
}
static public function setDir($dirBridge){
if( !is_string($dirBridge) ){
throw new \InvalidArgumentException('Dir bridge must be a string.');
}
if( !file_exists($dirBridge) ){
throw new \Exception('Dir bridge does not exist.');
}
self::$dirBridge = $dirBridge;
}
static public function getDir(){
$dirBridge = self::$dirBridge;
if( is_null($dirBridge) ){
throw new \LogicException(__CLASS__ . ' class need to know bridge path !');
}
return $dirBridge;
}
/**
* Lists the available bridges.
* @return array List of the bridges
*/
static public function listBridges() {
$pathDirBridge = self::getDir();
$listBridge = array();
$dirFiles = scandir($pathDirBridge);
if( $dirFiles !== false ){
foreach( $dirFiles as $fileName ) {
if( preg_match('@^([^.]+)Bridge\.php$@U', $fileName, $out) ){
$listBridge[] = $out[1];
}
}
}
return $listBridge;
}
static function isWhitelisted( $whitelist, $name ) {
if(in_array($name, $whitelist) or in_array($name.'.php', $whitelist) or
// DEPRECATED: the nameBridge notation will be removed in future releases
in_array($name.'Bridge', $whitelist) or in_array($name.'Bridge.php', $whitelist) or
count($whitelist) === 1 and trim($whitelist[0]) === '*')
return TRUE;
else
return FALSE;
}
}
abstract class RssExpander extends HttpCachingBridgeAbstract{
public $name;
public $uri;
public $description;
public function collectExpandableDatas(array $param, $name){ public function collectExpandableDatas(array $param, $name){
if (empty($name)) { if(empty($name)){
$this->returnServerError('There is no $name for this RSS expander'); $this->returnServerError('There is no $name for this RSS expander');
} }
// $this->message("Loading from ".$param['url']);
// Notice WE DO NOT use cache here on purpose : we want a fresh view of the RSS stream each time $this->debugMessage('Loading from ' . $param['url']);
$content=$this->getContents($name) or
$this->returnServerError('Could not request '.$name); /* Notice we do not use cache here on purpose:
* we want a fresh view of the RSS stream each time
*/
$content = $this->getContents($name) or $this->returnServerError('Could not request ' . $name);
$rssContent = simplexml_load_string($content); $rssContent = simplexml_load_string($content);
// $this->message("loaded RSS from ".$param['url']); $this->debugMessage('loaded RSS from ' . $param['url']);
// TODO insert RSS format detection // TODO insert RSS format detection
// we suppose for now, we have some RSS 2.0 // For now we always assume RSS 2.0
$this->collect_RSS_2_0_data($rssContent); $this->collect_RSS_2_0_data($rssContent);
} }
protected function collect_RSS_2_0_data($rssContent) { protected function collect_RSS_2_0_data($rssContent){
$rssContent = $rssContent->channel[0]; $rssContent = $rssContent->channel[0];
// $this->message("RSS content is ===========\n".var_export($rssContent, true)."==========="); $this->debugMessage('RSS content is ===========\n' . var_export($rssContent, true) . '===========');
$this->load_RSS_2_0_feed_data($rssContent); $this->load_RSS_2_0_feed_data($rssContent);
foreach($rssContent->item as $item) { foreach($rssContent->item as $item){
// $this->message("parsing item ".var_export($item, true)); $this->debugMessage('parsing item ' . var_export($item, true));
$this->items[] = $this->parseRSSItem($item); $this->items[] = $this->parseRSSItem($item);
} }
} }
protected function RSS_2_0_time_to_timestamp($item) { protected function RSS_2_0_time_to_timestamp($item){
return DateTime::createFromFormat('D, d M Y H:i:s e', $item->pubDate)->getTimestamp(); return DateTime::createFromFormat('D, d M Y H:i:s e', $item->pubDate)->getTimestamp();
} }
// TODO set title, link, description, language, and so on // TODO set title, link, description, language, and so on
protected function load_RSS_2_0_feed_data($rssContent) { protected function load_RSS_2_0_feed_data($rssContent){
$this->name = trim($rssContent->title); $this->name = trim($rssContent->title);
$this->uri = trim($rssContent->link); $this->uri = trim($rssContent->link);
$this->description = trim($rssContent->description); $this->description = trim($rssContent->description);
@ -386,9 +395,7 @@ abstract class RssExpander extends HttpCachingBridgeAbstract{
*/ */
abstract protected function parseRSSItem($item); abstract protected function parseRSSItem($item);
public function getDescription() { public function getDescription(){
return $this->description; return $this->description;
} }
} }