rss-bridge/lib/Bridge.php

397 lines
12 KiB
PHP
Raw Normal View History

2013-08-11 13:30:41 +02:00
<?php
interface BridgeInterface {
2013-08-11 13:30:41 +02:00
public function collectData(array $param);
public function getCacheDuration();
public function loadMetadatas();
public function getName();
public function getURI();
2013-08-11 13:30:41 +02:00
}
abstract class BridgeAbstract implements BridgeInterface {
2013-08-11 13:30:41 +02:00
protected $cache;
protected $items = array();
public $name = 'Unnamed bridge';
public $uri = '';
public $description = 'No description provided';
public $maintainer = 'No maintainer';
public $useProxy = true;
public $parameters = array();
2013-08-11 13:30:41 +02:00
protected function returnError($message, $code){
throw new \HttpException($message, $code);
}
protected function returnClientError($message){
$this->returnError($message, 400);
}
protected function returnServerError($message){
$this->returnError($message, 500);
}
2013-08-11 13:30:41 +02:00
/**
* Return items stored in the bridge
2013-08-11 13:30:41 +02:00
* @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
2013-08-11 13:30:41 +02:00
*/
public function setDatas(array $param){
if(!is_null($this->cache)){
2013-08-11 13:30:41 +02:00
$this->cache->prepare($param);
$time = $this->cache->getTime();
} else {
$time = false;
2013-08-11 13:30:41 +02:00
}
if($time !== false && (time() - $this->getCacheDuration() < $time)){
2013-08-11 13:30:41 +02:00
$this->items = $this->cache->loadData();
} else {
2013-08-11 13:30:41 +02:00
$this->collectData($param);
if(!is_null($this->cache)){
2013-08-11 13:30:41 +02:00
$this->cache->saveData($this->getDatas());
}
}
}
public function getName(){
return $this->name;
}
public function getURI(){
return $this->uri;
}
2013-08-11 13:30:41 +02:00
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
, $defaultSpanText);
}
2013-08-11 13:30:41 +02:00
}
/**
* Extension of BridgeAbstract allowing caching of files downloaded over http.
* TODO allow file cache invalidation by touching files on access, and removing
* files/directories which have not been touched since ... a long time
*/
abstract class HttpCachingBridgeAbstract extends BridgeAbstract {
/**
* Maintain locally cached versions of pages to download, to avoid multiple downloads.
* @param url url to cache
* @return content of the file as string
*/
public function get_cached($url){
$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->debugMessage('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->debugMessage('we have no local copy of ' . $url . ' Downloading to ' . $filename);
$dir = substr($filename, 0, strrpos($filename, '/'));
if(!is_dir($dir)){
$this->debugMessage('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->debugMessage('removing from cache \'' . $filename . '\' WELL, NOT REALLY');
// unlink($filename);
}
2013-08-11 13:30:41 +02:00
}
class Bridge {
2013-08-11 13:30:41 +02:00
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();
}
2013-08-11 13:30:41 +02:00
/**
* 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);
2013-08-11 13:30:41 +02:00
}
$nameBridge = $nameBridge . 'Bridge';
2013-08-11 13:30:41 +02:00
$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);
2013-08-11 13:30:41 +02:00
}
require_once $pathBridge;
if(Bridge::isInstantiable($nameBridge)){
return new $nameBridge();
} else {
return false;
}
2013-08-11 13:30:41 +02:00
}
static public function setDir($dirBridge){
if(!is_string($dirBridge)){
2013-08-11 13:30:41 +02:00
throw new \InvalidArgumentException('Dir bridge must be a string.');
}
if(!file_exists($dirBridge)){
2013-08-11 13:30:41 +02:00
throw new \Exception('Dir bridge does not exist.');
}
self::$dirBridge = $dirBridge;
}
static public function getDir(){
$dirBridge = self::$dirBridge;
if(is_null($dirBridge)){
2013-08-11 13:30:41 +02:00
throw new \LogicException(__CLASS__ . ' class need to know bridge path !');
}
return $dirBridge;
}
/**
* Lists the available bridges.
* @return array List of the bridges
2013-08-11 13:30:41 +02:00
*/
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];
}
}
}
2013-08-11 13:30:41 +02:00
return $listBridge;
}
2013-08-11 13:30:41 +02:00
static function isWhitelisted($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;
}
}
2014-05-21 19:09:37 +02:00
}
abstract class RssExpander extends HttpCachingBridgeAbstract {
public function collectExpandableDatas(array $param, $name){
if(empty($name)){
$this->returnServerError('There is no $name for this RSS expander');
}
$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->debugMessage('loaded RSS from ' . $param['url']);
// TODO insert RSS format detection
// 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];
$this->debugMessage('RSS content is ===========\n' . var_export($rssContent, true) . '===========');
$this->load_RSS_2_0_feed_data($rssContent);
foreach($rssContent->item as $item){
$this->debugMessage('parsing item ' . var_export($item, true));
$this->items[] = $this->parseRSSItem($item);
}
}
protected function RSS_2_0_time_to_timestamp($item){
return DateTime::createFromFormat('D, d M Y H:i:s e', $item->pubDate)->getTimestamp();
}
// TODO set title, link, description, language, and so on
protected function load_RSS_2_0_feed_data($rssContent){
$this->name = trim($rssContent->title);
$this->uri = trim($rssContent->link);
$this->description = trim($rssContent->description);
}
/**
* Method should return, from a source RSS item given by lastRSS, one of our Items objects
* @param $item the input rss item
* @return a RSS-Bridge Item, with (hopefully) the whole content)
*/
abstract protected function parseRSSItem($item);
public function getDescription(){
return $this->description;
}
}