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

View file

@ -2,7 +2,7 @@
define("RSS_PREFIX", "http://feeds.gawker.com/");
define("RSS_SUFFIX", "/full");
class Gawker extends RssExpander{
class GawkerBridge extends RssExpander{
public function loadMetadatas() {
@ -28,7 +28,7 @@ class Gawker extends RssExpander{
$this->name = $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);
}
@ -37,10 +37,10 @@ class Gawker extends RssExpander{
$item['uri'] = trim($newsItem->link);
$item['title'] = trim($newsItem->title);
$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 {
// 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($item['uri']));
if(is_object($articlePage)) {
$content = $articlePage->find('.post-content', 0);
@ -51,14 +51,14 @@ class Gawker extends RssExpander{
$item['author'] = $authorLink->innertext;
// 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
$item['content'] = $content->innertext;
} else {
throw new Exception("cache content for ".$item['uri']." is NOT a Simple DOM parser object !");
}
} 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
$this->remove_from_cache($item['url']);
$item['content'] = $e->getMessage();

View file

@ -2,7 +2,7 @@
define("SEXE", "http://sexes.blogs.liberation.fr");
define("SEXE_FEED", "http://sexes.blogs.liberation.fr/feeds/");
class Les400Culs extends RssExpander{
class Les400CulsBridge extends RssExpander{
public function loadMetadatas() {
@ -21,14 +21,14 @@ class Les400Culs extends RssExpander{
protected function parseRSSItem($newsItem) {
$item = array();
$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)) {
$item['uri'] = (string) $newsItem->link;
} else {
$item['uri'] = (string) $newsItem->guid;
}
// 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']));
// $content = $articlePage->find('.post-container', 0);

View file

@ -24,10 +24,10 @@ class TheOatmealBridge extends RssExpander{
protected function collect_RSS_2_0_data($rssContent) {
$rssContent->registerXPathNamespace("dc", "http://purl.org/dc/elements/1.1/");
$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);
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);
}
}
@ -39,10 +39,10 @@ class TheOatmealBridge extends RssExpander{
$rdf = $newsItem->children($namespaces['rdf']);
$item = array();
$item['title'] = trim($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;
// 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']));
$content = $articlePage->find('#comic', 0);
@ -51,10 +51,10 @@ class TheOatmealBridge extends RssExpander{
}
$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['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;
}

View file

@ -1,7 +1,7 @@
<?php
define('WORLD_OF_TANKS', 'http://worldoftanks.eu/');
define('NEWS', '/news/');
class WorldOfTanks extends HttpCachingBridgeAbstract{
class WorldOfTanksBridge extends HttpCachingBridgeAbstract{
private $lang = "fr";
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']."/";
}
$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
$this->name = $html->find('title', 0)->innertext;
foreach($html->find('.b-imgblock_ico') as $infoLink) {
@ -57,7 +57,7 @@ class WorldOfTanks extends HttpCachingBridgeAbstract{
$item = array();
$item['uri'] = WORLD_OF_TANKS.$infoLink->href;
// 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']));
$content = $articlePage->find('.l-content', 0);
HTMLSanitizer::defaultImageSrcTo($content, WORLD_OF_TANKS);

View file

@ -1,240 +1,4 @@
<?php
/**
* All bridge logic
* Note : adapter are store in other place
*/
interface BridgeInterface{
public function collectData(array $param);
public function getCacheDuration();
public function loadMetadatas();
public function getName();
public function getURI();
}
abstract class BridgeAbstract implements BridgeInterface{
protected $cache;
protected $items = array();
public $name = "Unnamed bridge";
public $uri = "";
public $description = 'No description provided';
public $maintainer = 'No maintainer';
public $useProxy = true;
public $parameters = array();
/**
* Launch probative exception
*/
protected function returnError($message, $code){
throw new \HttpException($message, $code);
}
protected function returnClientError($message){
$this->returnError($message, 400);
}
protected function returnServerError($message){
$this->returnError($message, 500);
}
/**
* Return datas stored in the bridge
* @return mixed
*/
public function getDatas(){
return $this->items;
}
/**
* Defined datas with parameters depending choose bridge
* Note : you can define a cache before with "setCache"
* @param array $param $_REQUEST, $_GET, $_POST, or array with bridge expected paramters
*/
public function setDatas(array $param){
if( !is_null($this->cache) ){
$this->cache->prepare($param);
$time = $this->cache->getTime();
}
else{
$time = false; // No cache ? No time !
}
if( $time !== false && ( time() - $this->getCacheDuration() < $time ) ){ // Cache file has not expired. Serve it.
$this->items = $this->cache->loadData();
}
else{
$this->collectData($param);
if( !is_null($this->cache) ){ // Cache defined ? We go to refresh is memory :D
$this->cache->saveData($this->getDatas());
}
}
}
/**
* Define default bridge name
*/
public function getName(){
return $this->name;
}
/**
* Define default bridge URI
*/
public function getURI(){
return $this->uri;
}
/**
* Define default duraction for cache
*/
public function getCacheDuration(){
return 3600;
}
/**
* Defined cache object to use
*/
public function setCache(\CacheAbstract $cache){
$this->cache = $cache;
return $this;
}
public function message($text) {
if(!file_exists('DEBUG')){
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)){
$content=@file_get_contents($url, $use_include_path, $context, $offset);
}else{
$content=@file_get_contents($url, $use_include_path, $context, $offset,$maxlen);
}
if($content===false){
$this->message('Cant\'t download '.$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){
$content=$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.
* 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
* separately.
* 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 {
/**
* Maintain locally cached versions of pages to download to avoid multiple doiwnloads.
* A file name is generated by replacing all "/" by "_", and the file is saved below this bridge cache
* @param url url to cache
* @return content of file as string
*/
public function get_cached($url) {
$simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url);
// TODO build this from the variable given to Cache
$pageCacheDir = __DIR__ . '/../cache/'."pages/";
$filename = $pageCacheDir.$simplified_url;
if (substr($filename, -1) == '/') {
$filename = $filename."index.html";
}
if(file_exists($filename)) {
// $this->message("loading cached file from ".$filename." for page at url ".$url);
// TODO touch file and its parent, and try to do neighbour deletion
$this->refresh_in_cache($pageCacheDir, $filename);
$content=file_get_contents($filename);
} else {
// $this->message("we have no local copy of ".$url." Downloading to ".$filename);
$dir = substr($filename, 0, strrpos($filename, '/'));
if(!is_dir($dir)) {
// $this->message("creating directories for ".$dir);
mkdir($dir, 0777, true);
}
$content=$this->getContents($url);
if($content!==false){
file_put_contents($filename,$content);
}
}
return $content;
}
public function get_cached_time($url) {
$simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url);
// TODO build this from the variable given to Cache
$pageCacheDir = __DIR__ . '/../cache/'."pages/";
$filename = $pageCacheDir.$simplified_url;
if (substr($filename, -1) == '/') {
$filename = $filename."index.html";
}
if(!file_exists($filename)) {
$this->get_cached($url);
}
return filectime($filename);
}
private function refresh_in_cache($pageCacheDir, $filename) {
$currentPath = $filename;
while(!$pageCacheDir==$currentPath) {
touch($currentPath);
$currentPath = dirname($currentPath);
}
}
public function remove_from_cache($url) {
$simplified_url = str_replace(["http://", "https://", "?", "&", "="], ["", "", "/", "/", "/"], $url);
// TODO build this from the variable given to Cache
$pageCacheDir = __DIR__ . '/../cache/'."pages/";
$filename = realpath($pageCacheDir.$simplified_url);
$this->message("removing from cache \"".$filename."\" WELL, NOT REALLY");
// filename is NO GOOD
// unlink($filename);
}
}
class Bridge {
static protected $dirBridge;
@ -249,13 +13,10 @@ class Bridge{
* @return true if it is an instantiable bridge, false otherwise.
*/
static public function isInstantiable($nameBridge){
$re = new ReflectionClass($nameBridge);
return $re->IsInstantiable();
}
/**
* Create a new bridge object
* @param string $nameBridge Defined bridge name you want use
@ -263,7 +24,11 @@ class Bridge{
*/
static public function create($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.');
$message = <<<EOD
'nameBridge' must start with one uppercase character followed or not by
alphanumeric or dash characters!
EOD;
throw new \InvalidArgumentException($message);
}
$nameBridge = $nameBridge . 'Bridge';
@ -278,7 +43,7 @@ class Bridge{
if(Bridge::isInstantiable($nameBridge)){
return new $nameBridge();
} else {
return FALSE;
return false;
}
}
@ -309,7 +74,6 @@ class Bridge{
* @return array List of the bridges
*/
static public function listBridges(){
$pathDirBridge = self::getDir();
$listBridge = array();
$dirFiles = scandir($pathDirBridge);
@ -324,46 +88,291 @@ class Bridge{
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($whitelist[0]) === '*')
return TRUE;
else
return FALSE;
static public function isWhitelisted($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 getCacheDuration();
public function loadMetadatas();
public function getName();
public function getURI();
}
abstract class BridgeAbstract implements BridgeInterface {
protected $cache;
protected $items = array();
public $name = 'Unnamed bridge';
public $uri = '';
public $description = 'No description provided';
public $maintainer = 'No maintainer';
public $useProxy = true;
public $parameters = array();
protected function returnError($message, $code){
throw new \HttpException($message, $code);
}
protected function returnClientError($message){
$this->returnError($message, 400);
}
protected function returnServerError($message){
$this->returnError($message, 500);
}
/**
* Return items stored in the bridge
* @return mixed
*/
public function getDatas(){
return $this->items;
}
/**
* Defined datas with parameters depending choose bridge
* Note : you can define a cache with "setCache"
* @param array $param $_REQUEST, $_GET, $_POST, or array with expected
* bridge paramters
*/
public function setDatas(array $param){
if(!is_null($this->cache)){
$this->cache->prepare($param);
$time = $this->cache->getTime();
} else {
$time = false;
}
if($time !== false && (time() - $this->getCacheDuration() < $time)){
$this->items = $this->cache->loadData();
} else {
$this->collectData($param);
if(!is_null($this->cache)){
$this->cache->saveData($this->getDatas());
}
}
}
public function getName(){
return $this->name;
}
public function getURI(){
return $this->uri;
}
public function getCacheDuration(){
return 3600;
}
public function setCache(\CacheAbstract $cache){
$this->cache = $cache;
}
public function debugMessage($text){
if(!file_exists('DEBUG')) {
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)){
$content = @file_get_contents($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
, $defaultSpanText);
}
}
/**
* Extension of BridgeAbstract allowing caching of files downloaded over http.
* TODO allow file cache invalidation by touching files on access, and removing
* files/directories which have not been touched since ... a long time
*/
abstract class HttpCachingBridgeAbstract extends BridgeAbstract {
/**
* Maintain locally cached versions of pages to download, to avoid multiple downloads.
* @param url url to cache
* @return content of the file as string
*/
public function get_cached($url){
// TODO build this from the variable given to Cache
$cacheDir = __DIR__ . '/../cache/pages/';
$filepath = $this->buildCacheFilePath($url, $cacheDir);
if(file_exists($filepath)){
$this->debugMessage('loading cached file from ' . $filepath . ' for page at url ' . $url);
// TODO touch file and its parent, and try to do neighbour deletion
$this->refresh_in_cache($cacheDir, $filepath);
$content = file_get_contents($filepath);
} else {
$this->debugMessage('we have no local copy of ' . $url . ' Downloading to ' . $filepath);
$dir = substr($filepath, 0, strrpos($filepath, '/'));
if(!is_dir($dir)){
$this->debugMessage('creating directories for ' . $dir);
mkdir($dir, 0777, true);
}
$content = $this->getContents($url);
if($content !== false){
file_put_contents($filepath, $content);
}
}
return $content;
}
public function get_cached_time($url){
// TODO build this from the variable given to Cache
$cacheDir = __DIR__ . '/../cache/pages/';
$filepath = $this->buildCacheFilePath($url, $cacheDir);
if(!file_exists($filepath)){
$this->get_cached($url);
}
return filectime($filepath);
}
private function refresh_in_cache($cacheDir, $filepath){
$currentPath = $filepath;
while(!$cacheDir == $currentPath){
touch($currentPath);
$currentPath = dirname($currentPath);
}
}
private function buildCacheFilePath($url, $cacheDir){
$simplified_url = str_replace(
['http://', 'https://', '?', '&', '='],
['', '', '/', '/', '/'],
$url);
if(substr($cacheDir, -1) !== '/'){
$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);
}
}
abstract class RssExpander extends HttpCachingBridgeAbstract {
public $name;
public $uri;
public $description;
public function collectExpandableDatas(array $param, $name){
if(empty($name)){
$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
$content=$this->getContents($name) or
$this->returnServerError('Could not request '.$name);
$this->debugMessage('Loading from ' . $param['url']);
/* 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);
// $this->message("loaded RSS from ".$param['url']);
$this->debugMessage('loaded RSS from ' . $param['url']);
// 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);
}
protected function collect_RSS_2_0_data($rssContent){
$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);
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);
}
}
@ -390,5 +399,3 @@ abstract class RssExpander extends HttpCachingBridgeAbstract{
return $this->description;
}
}