forked from blallo/rss-bridge
[Bridge] Cleanup file
- Remove unnecessary documentation
- Update/Clarify documentation where necessary
- Remove empty lines
- Put 'else' between closing and opening curly braces
- Make sure curly braces start right after closing brace on functions '(){...'
- Start lines with '.' and use proper indentation when using multi-line string combinations
- Add spaces for function definitions/calls and assignments
- Add space before opening curly brace after class definition 'class xyz {'
This commit is contained in:
logmanoriginal
parent
a43703d32d
commit
429126e18a
1 changed files with 145 additions and 162 deletions
307
lib/Bridge.php
307
lib/Bridge.php
|
|
@ -1,10 +1,5 @@
|
|||
<?php
|
||||
/**
|
||||
* All bridge logic
|
||||
* Note : adapter are store in other place
|
||||
*/
|
||||
|
||||
interface BridgeInterface{
|
||||
interface BridgeInterface {
|
||||
public function collectData(array $param);
|
||||
public function getCacheDuration();
|
||||
public function loadMetadatas();
|
||||
|
|
@ -12,7 +7,7 @@ interface BridgeInterface{
|
|||
public function getURI();
|
||||
}
|
||||
|
||||
abstract class BridgeAbstract implements BridgeInterface{
|
||||
abstract class BridgeAbstract implements BridgeInterface {
|
||||
|
||||
protected $cache;
|
||||
protected $items = array();
|
||||
|
|
@ -24,9 +19,6 @@ abstract class BridgeAbstract implements BridgeInterface{
|
|||
public $useProxy = true;
|
||||
public $parameters = array();
|
||||
|
||||
/**
|
||||
* Launch probative exception
|
||||
*/
|
||||
protected function returnError($message, $code){
|
||||
throw new \HttpException($message, $code);
|
||||
}
|
||||
|
|
@ -40,202 +32,194 @@ abstract class BridgeAbstract implements BridgeInterface{
|
|||
}
|
||||
|
||||
/**
|
||||
* Return datas stored in the bridge
|
||||
* 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 before with "setCache"
|
||||
* Note : you can define a cache with "setCache"
|
||||
* @param array $param $_REQUEST, $_GET, $_POST, or array with bridge expected paramters
|
||||
*/
|
||||
public function setDatas(array $param){
|
||||
if( !is_null($this->cache) ){
|
||||
if(!is_null($this->cache)){
|
||||
$this->cache->prepare($param);
|
||||
$time = $this->cache->getTime();
|
||||
}
|
||||
else{
|
||||
$time = false; // No cache ? No time !
|
||||
} else {
|
||||
$time = false;
|
||||
}
|
||||
|
||||
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();
|
||||
}
|
||||
else{
|
||||
} else {
|
||||
$this->collectData($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());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 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;
|
||||
};
|
||||
public function message($text){
|
||||
if(!file_exists('DEBUG')) {
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
if(is_null($maxlen)){
|
||||
$content=@file_get_contents($url, $use_include_path, $context, $offset);
|
||||
}else{
|
||||
$content=@file_get_contents($url, $use_include_path, $context, $offset,$maxlen);
|
||||
}
|
||||
$backtrace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 3);
|
||||
$calling = $backtrace[2];
|
||||
$message = $calling['file'] . ':'
|
||||
. $calling['line'] . ' class '
|
||||
. get_class($this) . '->'
|
||||
. $calling['function'] . ' - '
|
||||
. $text;
|
||||
|
||||
if($content===false){
|
||||
$this->message('Cant\'t download '.$url );
|
||||
}
|
||||
return $content;
|
||||
error_log($message);
|
||||
}
|
||||
|
||||
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);
|
||||
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, $defaultSpanText);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Extension of BridgeAbstract allowing caching of files downloaded over http files.
|
||||
* This is specially useful for sites from Gawker or Liberation networks, which allow pages excerpts top be viewed together on index, while full pages have to be downloaded
|
||||
* separately.
|
||||
* This class mainly provides a get_cached method which will will download the file from its remote location.
|
||||
* TODO allow file cache invalidation by touching files on access, and removing files/directories which have not been touched since ... a long time
|
||||
* After all, rss-bridge is not respaw, isn't it ?
|
||||
* Extension of BridgeAbstract allowing caching of files downloaded over http.
|
||||
* TODO allow file cache invalidation by touching files on access, and removing
|
||||
* files/directories which have not been touched since ... a long time
|
||||
*/
|
||||
abstract class HttpCachingBridgeAbstract extends BridgeAbstract {
|
||||
|
||||
/**
|
||||
* Maintain locally cached versions of pages to download to avoid multiple doiwnloads.
|
||||
* A file name is generated by replacing all "/" by "_", and the file is saved below this bridge cache
|
||||
* Maintain locally cached versions of pages to download, to avoid multiple downloads.
|
||||
* @param url url to cache
|
||||
* @return content of file as string
|
||||
* @return content of the file as string
|
||||
*/
|
||||
public function get_cached($url) {
|
||||
public function get_cached($url){
|
||||
$simplified_url = str_replace(['http://', 'https://', '?', '&', '='], ['', '', '/', '/', '/'], $url);
|
||||
|
||||
// TODO build this from the variable given to Cache
|
||||
$pageCacheDir = __DIR__ . '/../cache/'.'pages/';
|
||||
$filename = $pageCacheDir.$simplified_url;
|
||||
if (substr($filename, -1) == '/') {
|
||||
$filename = $filename.'index.html';
|
||||
$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 '.$url);
|
||||
|
||||
if(file_exists($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);
|
||||
$content = file_get_contents($filename);
|
||||
} else {
|
||||
$this->message('we have no local copy of '.$url.' Downloading to '.$filename);
|
||||
$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);
|
||||
|
||||
if(!is_dir($dir)){
|
||||
$this->message('creating directories for ' . $dir);
|
||||
mkdir($dir, 0777, true);
|
||||
}
|
||||
$content=$this->getContents($url);
|
||||
|
||||
$content = $this->getContents($url);
|
||||
if($content!==false){
|
||||
file_put_contents($filename,$content);
|
||||
file_put_contents($filename,$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
|
||||
$pageCacheDir = __DIR__ . '/../cache/'.'pages/';
|
||||
$filename = $pageCacheDir.$simplified_url;
|
||||
if (substr($filename, -1) == '/') {
|
||||
$filename = $filename.'index.html';
|
||||
$pageCacheDir = __DIR__ . '/../cache/pages/';
|
||||
$filename = $pageCacheDir . $simplified_url;
|
||||
|
||||
if(substr($filename, -1) == '/'){
|
||||
$filename = $filename . 'index.html';
|
||||
}
|
||||
if(!file_exists($filename)) {
|
||||
|
||||
if(!file_exists($filename)){
|
||||
$this->get_cached($url);
|
||||
}
|
||||
|
||||
return filectime($filename);
|
||||
}
|
||||
|
||||
private function refresh_in_cache($pageCacheDir, $filename) {
|
||||
private function refresh_in_cache($pageCacheDir, $filename){
|
||||
$currentPath = $filename;
|
||||
while(!$pageCacheDir==$currentPath) {
|
||||
while(!$pageCacheDir == $currentPath){
|
||||
touch($currentPath);
|
||||
$currentPath = dirname($currentPath);
|
||||
}
|
||||
}
|
||||
|
||||
public function remove_from_cache($url) {
|
||||
public function remove_from_cache($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);
|
||||
}
|
||||
|
||||
// TODO build this from the variable given to Cache
|
||||
$pageCacheDir = __DIR__ . '/../cache/pages/';
|
||||
$filename = realpath($pageCacheDir . $simplified_url);
|
||||
$this->message('removing from cache \'' . $filename . '\' WELL, NOT REALLY');
|
||||
// unlink($filename);
|
||||
}
|
||||
}
|
||||
|
||||
class Bridge{
|
||||
class Bridge {
|
||||
|
||||
static protected $dirBridge;
|
||||
|
||||
|
|
@ -248,46 +232,43 @@ class 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) {
|
||||
|
||||
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-]*$@', $nameBridge)){
|
||||
if(!preg_match('@^[A-Z][a-zA-Z0-9-]*$@', $nameBridge)){
|
||||
throw new \InvalidArgumentException('Name bridge must be at least one uppercase follow or not by alphanumeric or dash characters.');
|
||||
}
|
||||
|
||||
$nameBridge=$nameBridge.'Bridge';
|
||||
$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 '.$pathBridge);
|
||||
if(!file_exists($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)) {
|
||||
if(Bridge::isInstantiable($nameBridge)){
|
||||
return new $nameBridge();
|
||||
} else {
|
||||
return FALSE;
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
static public function setDir($dirBridge){
|
||||
if( !is_string($dirBridge) ){
|
||||
if(!is_string($dirBridge)){
|
||||
throw new \InvalidArgumentException('Dir bridge must be a string.');
|
||||
}
|
||||
|
||||
if( !file_exists($dirBridge) ){
|
||||
if(!file_exists($dirBridge)){
|
||||
throw new \Exception('Dir bridge does not exist.');
|
||||
}
|
||||
|
||||
|
|
@ -297,7 +278,7 @@ class Bridge{
|
|||
static public function getDir(){
|
||||
$dirBridge = self::$dirBridge;
|
||||
|
||||
if( is_null($dirBridge) ){
|
||||
if(is_null($dirBridge)){
|
||||
throw new \LogicException(__CLASS__ . ' class need to know bridge path !');
|
||||
}
|
||||
|
||||
|
|
@ -308,68 +289,72 @@ class Bridge{
|
|||
* Lists the available bridges.
|
||||
* @return array List of the bridges
|
||||
*/
|
||||
static public function listBridges() {
|
||||
|
||||
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[] = $out[1];
|
||||
if($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($whitelist[0]) === '*')
|
||||
return TRUE;
|
||||
else
|
||||
return FALSE;
|
||||
}
|
||||
|
||||
static function isWhitelisted($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;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
abstract class RssExpander extends HttpCachingBridgeAbstract{
|
||||
abstract class RssExpander extends HttpCachingBridgeAbstract {
|
||||
|
||||
public function collectExpandableDatas(array $param, $name){
|
||||
if (empty($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->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);
|
||||
|
||||
$rssContent = simplexml_load_string($content);
|
||||
$this->message('loaded RSS from '.$param['url']);
|
||||
$this->message('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($rssContent) {
|
||||
protected function collect_RSS_2_0_data($rssContent){
|
||||
$rssContent = $rssContent->channel[0];
|
||||
$this->message('RSS content is ===========\n'.var_export($rssContent, true).'===========');
|
||||
$this->message('RSS content is ===========\n' . var_export($rssContent, true) . '===========');
|
||||
$this->load_RSS_2_0_feed_data($rssContent);
|
||||
foreach($rssContent->item as $item) {
|
||||
$this->message('parsing item '.var_export($item, true));
|
||||
foreach($rssContent->item as $item){
|
||||
$this->message('parsing item ' . var_export($item, true));
|
||||
$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', $item->pubDate)->getTimestamp();
|
||||
}
|
||||
|
||||
// TODO set title, link, description, language, and so on
|
||||
protected function load_RSS_2_0_feed_data($rssContent) {
|
||||
protected function load_RSS_2_0_feed_data($rssContent){
|
||||
$this->name = trim($rssContent->title);
|
||||
$this->uri = trim($rssContent->link);
|
||||
$this->description = trim($rssContent->description);
|
||||
|
|
@ -382,9 +367,7 @@ abstract class RssExpander extends HttpCachingBridgeAbstract{
|
|||
*/
|
||||
abstract protected function parseRSSItem($item);
|
||||
|
||||
public function getDescription() {
|
||||
public function getDescription(){
|
||||
return $this->description;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
|
|
|||
Loading…
Reference in a new issue