From 9918a0eb4d68f0e24baf3a8ccccde035f3267092 Mon Sep 17 00:00:00 2001
From: franzz <francois@lutran.fr>
Date: Tue, 30 Aug 2016 19:26:48 +0200
Subject: [PATCH] rarbg parser

---
 standalone/rarbg.php | 146 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 146 insertions(+)
 create mode 100644 standalone/rarbg.php

diff --git a/standalone/rarbg.php b/standalone/rarbg.php
new file mode 100644
index 0000000..74cdfcb
--- /dev/null
+++ b/standalone/rarbg.php
@@ -0,0 +1,146 @@
+<?php
+
+class Settings
+{
+	const DEBUG = true;
+	const TEXT_ENC = 'UTF-8';
+}
+
+require_once '../class_management.php';
+require_once '../inc/rss.php';
+
+function getCurl($sUrl, $bHeader=false, $asPostData=array())
+{
+	$oCurl = curl_init();
+
+	curl_setopt($oCurl, CURLOPT_URL, $sUrl);
+	curl_setopt($oCurl, CURLOPT_HEADER, $bHeader);
+	if(!empty($asPostData))
+	{
+		curl_setopt($oCurl, CURLOPT_POST, count($asPostData));
+		curl_setopt($oCurl, CURLOPT_POSTFIELDS, $asPostData);
+	}
+
+	curl_setopt($oCurl, CURLOPT_RETURNTRANSFER, true);
+
+	//Fake browser
+	curl_setopt($oCurl, CURLOPT_USERAGENT, $_SERVER['HTTP_USER_AGENT']);
+	//if($this->sReferer!='') curl_setopt($oCurl, CURLOPT_REFERER, $this->sReferer);
+	curl_setopt($oCurl, CURLOPT_SSL_VERIFYPEER, false);
+	//if($bHeader) curl_setopt($oCurl, CURLOPT_FOLLOWLOCATION, true);
+
+	//Cookies
+	//$sCookies = $this->getCookies();
+	//if($sCookies!='') curl_setopt($oCurl, CURLOPT_COOKIE, $sCookies);
+	//curl_setopt($oCurl, CURLOPT_COOKIEJAR, $this->fCookieJar);
+	//curl_setopt($oCurl, CURLOPT_COOKIEFILE, $this->fCookieJar);
+
+	$sContent = curl_exec($oCurl);
+
+	curl_close($oCurl);
+	//$this->sReferer = $sUrl;
+	return $sContent;
+}
+
+function parseTop10()
+{
+	$asDesc = array
+	(
+			'title'=>'Rarbg RSS Feed',
+			'link'=>'',
+			'copyright'=>'Powered by Franzz. RSS Feed Generator under GPLv3 License',
+			'description'=>'Rarbg RSS Feed',
+			'language'=>'en',
+			'webmaster_mail'=>'franzz@gmail.com'
+	);
+	$oFeed = new Feed($asDesc);
+	
+	$sDomain = 'https://rarbg.to';
+	$oDom = new DOMDocument();
+	$sContent = getCurl($sDomain.'/torrents.php?category=14;17;42;44;45;46;47;48&search=&order=seeders&by=DESC');
+	//$sContent = file_get_contents('test.html');
+	@$oDom->loadHTML($sContent);
+	$oBody = $oDom->getElementsByTagName('body')->item(0);
+	$aoLines = $oDom->getElementsByTagName('tr');
+	foreach($aoLines as $oLine)
+	{
+		if($oLine->getAttribute('class')=='lista2')
+		{
+			$asArticle = array();
+			$aoCols = $oLine->getElementsByTagName('td');
+			foreach($aoCols as $iIndex=>$oCol)
+			{
+				
+				switch($iIndex)
+				{
+					case 1:
+						$aoLinks = $oCol->getElementsByTagName('a');
+						foreach($aoLinks as $oLink)
+						{
+							$sHref = $oLink->getAttribute('href');
+							switch(substr($sHref, 0, 9))
+							{
+								//Title & link
+								case '/torrent/':
+									$iCount = '|'.$sHref;
+									$asArticle['title'] = $oLink->nodeValue;
+									$sLinkPath = str_replace('#comments', '', $sHref);
+									$asArticle['link'] = $sDomain.$sLinkPath;
+									$asArticle['guid'] = substr($sLinkPath, -7);
+									$sHover = $oLink->getAttribute('onmouseover');//return overlib('<img src=\'//dyncdn.me/static/over/1c74b4cde890fc95be77ac253297c45e50a05f98.jpg\' border=0>')
+									preg_match('/'.preg_quote('src=\\\'').'(?P<url>.+)'.preg_quote('\\\' border').'/', $sHover, $asMatch);
+									$sImg = 'https:'.$asMatch['url'];
+									break;
+									//Author
+								case '/torrents': // /torrents.php?imdb=tt3647498
+									$sMovieSearchLink = $sHref;
+									break;
+							}
+						}
+						$aoSpans = $oLine->getElementsByTagName('span');
+						foreach($aoSpans as $oSpan)
+						{
+							if($oSpan->getAttribute('style')=='color:DarkSlateGray')
+							{
+								$sMovieType = $oSpan->nodeValue;
+								$sImdb = strstr($sMovieType, 'IMDB');
+								$asArticle['category'] = trim(strstr($sMovieType, 'IMDB', true));
+							}
+						}
+						break;
+					case 2:
+						$asArticle['pub_date'] = $oCol->nodeValue;
+						break;
+					case 3:
+						$sSize = $oCol->nodeValue;
+						break;
+					case 4:
+						$iSeeder = $oCol->nodeValue;
+						break;
+					case 5:
+						$iLeecher = $oCol->nodeValue;
+						break;
+					case 7:
+						$asArticle['author'] =  $oCol->nodeValue;
+						break;
+				}
+			}
+			$asArticle['description'] = '<img src="'.$sImg.'" />'.
+										'<br />'.$sMovieType.
+										'<br />'.$sImdb.
+										'<br />Size: '.$sSize.
+										'<br />S/L: <span style="color:green;">'.$iSeeder.'</span>/<span style="color:orange;">'.$iLeecher.'</span>'.
+										'<br /><a href="'.$sDomain.$sMovieSearchLink.'" target="_blank">Movie search</a>';
+			
+			//die(print_r($asArticle, true));
+			
+			$oFeed->addItem($asArticle);
+		}
+	}
+	
+	return $oFeed->getFeed();
+}
+
+echo parseTop10();
+
+?>
\ No newline at end of file