@@ -0,0 +1,266 @@

+<?php

+include_once( 'news-constants.php' );

+

+$rss_content = http_get_contents(NEWS_RSS_LOBS);

+$xml = simplexml_load_string($rss_content);

+if ($xml === false) {

+  echo 'Failed to read RSS';

+} else {

+  $channel = array();

+  $channel['title'] = $xml->channel->title;

+  $channel['link'] = $xml->channel->link;

+  $channel['description'] = $xml->channel->description;

+  $channel['pubDate'] = $xml->channel->pubDate;

+  $channel['timestamp'] = strtotime($xml->channel->pubDate);

+  //echo '<h4>' . $channel['title'] . '</h4>';

+  echo '<h4>' . $channel['title'].'<button id="html-btn" onclick="printHTMLAll(\''.$channel['title'].'\')" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button></h4>'.PHP_EOL;

+  $cpt=0;

+  foreach ($xml->channel->item as $item) {

+    $article = array();

+    $article['title'] = $item->title;

+    $article['link'] = $item->link;

+    $orgStrings = array('?xtor=RSS-13');

+    $newStrings = array('');

+    $article['link']  = str_replace($orgStrings, $newStrings, $article['link']);

+    $article['pubDate'] = $item->pubDate;

+    $article['timestamp'] = strtotime($item->pubDate);

+    $article['description'] = $item->description;

+    //$article['image'] = $item->enclosure['url'];

+    $articles[$cpt] = $article; 

+

+    echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">'.PHP_EOL;

+    //echo '<div class="img-menu"><img src="'.$article['image'].'" style="display:inline;" width="100%"></div><br>'.PHP_EOL;

+    echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;'.PHP_EOL;

+    echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div>&nbsp;';

+    echo $article['title'].'  '.PHP_EOL;

+    echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;

+    $cpt++;

+    if( $cpt > $NEWS_RSS_MAX_ITEMS ) {

+      break;

+    }

+  }

+}

+echo '</div><!-- ./col-4 -->';

+echo '<div id="article-display" class="col-6">'.PHP_EOL;

+echo '<a name="article-top"></a><div id="article-current"></div>';

+echo '</div><!-- ./col-6 -->'.PHP_EOL;

+$cpt=0;

+foreach ($articles as $article ) {

+  $cpt_prev=$cpt-1;

+  $cpt_next=$cpt+1;

+  warning($article['link']);

+  echo PHP_EOL.PHP_EOL."<!-- ==================== article '.$cpt.'============== -->".PHP_EOL;

+  echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";

+  echo "<hr><a name=\"article-$cpt\">\n";

+  $article_content = file_get_contents($article['link']);

+  $article_content_utf8 = mb_convert_encoding($article_content, 'HTML-ENTITIES', "UTF-8");

+  $article_content = $article_content_utf8;

+  $doc = new DOMDocument();

+  $doc->preserveWhiteSpace = false;

+  $doc->formatOutput       = true;

+  $libxml_previous_state = libxml_use_internal_errors(true);

+  $doc->loadHTML($article_content);

+  libxml_clear_errors();

+  libxml_use_internal_errors($libxml_previous_state);

+  $articles = $doc->getElementsByTagName('div');

+  $article_only = "";

+  foreach ( $articles as $node) {

+    $classname = $node->getAttribute('id');

+    if($classname == 'ObsArticle-body' ) {

+      $article_only = DOMinnerHTML($node);

+      break;

+    }

+  }

+  if( 0 == strlen($article_only)) {

+    $article_only = "Extraction failed";

+  }

+

+  $metas = $doc->getElementsByTagName('meta');

+  $chapo = "";

+  foreach ( $metas as $node) {

+    $classname = $node->getAttribute('property');

+    if(0==strcmp($classname,'og:description') ) {

+      $chapo = $node->getAttribute('content');

+    }else if(0==strcmp($classname,'og:image') ) {

+      $article['image'] = $node->getAttribute('content');

+    }

+  }

+

+  $isPortfolio=false;

+  if( 0 == strlen($article_only) ) {

+    //<div class="layout layout--portfolio">

+    $node = getElementByClass($doc, 'div', 'layout layout--portfolio', $offset = 0);

+    if( false !== $node ) {

+      //Article is a portfolio

+      $article_only = DOMinnerHTML($node);

+      $isPortfolio=true;

+    }

+  }

+

+  $orgStrings = array( ' href="/',

+    ' src="/sites/',

+    '-src="/sites/',

+    '<img src=');

+  $newStrings = array( ' href="http://www.slate.fr/',

+    ' src="http://www.slate.fr/sites/',

+    '-src="http://www.slate.fr/sites/',

+    '<img width="100%" src=' );

+  $article_only  = str_replace($orgStrings, $newStrings, $article_only);

+  

+  $orgStrings = array( '<div class="row">',

+    '<div class="col-md-2">',

+    '<div class="col-md-8">',

+    '<div class="container width_wrap">',

+    '<div class="col-left">',

+    '<div class="article-header__inner col-sm-10 col-sm-offset-1 col-md-8 col-md-offset-2">',

+    '<div class="content-left col-md-8 col-auto">',

+    '<p class="article-header__infos article-infos">',

+    '<div class="article-thumb-wrapper">',

+    '<div class="article-thumb">',

+    '<figure class="preload article-image">',

+    '<div class="article-content row">',

+    '<h3> </h3>',

+    '<h1>',

+    '</h1>',

+    '<p style="text-align:center">' );

+  $newStrings = array('<div>',

+    '<div>',

+    '<div>',

+    '<div>',

+    '<div>',

+    '<div>',

+    '<div>',

+    '<p>',

+    '<div>',

+    '<div>',

+    '<figure>',

+    '<div>',

+    '',

+    '<h4>',

+    '</h4>',

+    '<p>');

+  $article_only = str_replace($orgStrings, $newStrings, $article_only);

+  

+  $temp = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);

+  $article_only = $temp;

+  $temp = preg_replace('/\s\s+/', ' ', $article_only);

+  $article_only = $temp;

+  $re = '/ style="height:(.+?)width: 640px;">/';

+  $temp = preg_replace($re, '>', $article_only); 

+  $article_only = $temp;

+  $re = '/<script>\(function \(\) \{var sasCallOptions = \{ siteId: 59629(.+?)<\/script>/';

+  $temp = preg_replace($re, '', $article_only); 

+  $article_only = $temp;

+  $re = '/<img data-file-id=(.+?)src="(.+?)"(.+?)>/';

+  $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only); 

+  $article_only = $temp;

+  $re='/<div class="media_embed" height="(.+?)" width="(.+?)">/';

+  $temp = preg_replace($re, '<div>', $article_only); 

+  $article_only = $temp;

+  $re='/<iframe allowfullscreen="" frameborder="0" height="(.+?)" src="(.+?)" width="(.+?)"><\/iframe>/';

+  $temp = preg_replace($re, '<iframe frameborder="0" width="100%" src="\\2"></iframe>', $article_only); 

+  $article_only = $temp;

+  $re='/<img width="100%" src="(.+?)" style="height:(.+?)width: (.+?)">/';

+  $temp = preg_replace($re, '<img width="100%" src="\\1">', $article_only); 

+  $article_only = $temp;

+  $re='/<div class="icon icon--(.+?)"> (.+)/';

+  $temp = preg_replace($re, '<div>', $article_only); 

+  $article_only = $temp;

+  $re='/<div class="article-header__breadcrumb">/';

+  $temp = preg_replace($re, '<div>', $article_only); 

+  $article_only = $temp;

+

+  $re='/<img class="image image--full lazyload" data-full-src=/';

+  $temp = preg_replace($re, '<img width="100%" src=', $article_only); 

+  $article_only = $temp;

+

+  $re='/<span class="sharing-btn__numbers">(.+)<\/span>/';

+  $temp = preg_replace($re, '', $article_only); 

+  $article_only = $temp;

+  

+  $re='/<span class="sharing-btn__bg"><\/span>/';

+  $temp = preg_replace($re, '', $article_only); 

+  $article_only = $temp;

+

+  $re='/<script data-cfasync="false" src=(.+)<\/script><script>(.+?)<\/script>/';

+  $temp = preg_replace($re, '', $article_only); 

+  $article_only = $temp;

+

+  //Clean so called social

+  $re='/<a data-share="(fb|li|tw)" data-url="(.+)">  <div>/';

+  $temp = preg_replace($re, '<a><div>', $article_only); 

+  $article_only = $temp;

+  $re='/<a class="sharing-btn sharing-btn--circle sharing-btn--whatsapp" href="whatsapp:(.+)>  <div>/';

+  $temp = preg_replace($re, '<a><div>', $article_only); 

+  $article_only = $temp;

+  $re='/<a href="fb-messenger:(.+) class="sharing-btn sharing-btn--circle sharing-btn--messenger" target="_blank">  <div>/';

+  $temp = preg_replace($re, '<a><div>', $article_only); 

+  $article_only = $temp;

+  $re='/<a href="https:\/\/www\.facebook\.com\/dialog\/send(.+)target="_blank">  <div>/';

+  $temp = preg_replace($re, '<a><div>', $article_only); 

+  $article_only = $temp;

+  $re='/<div id="inreadbis" class="ad-wrapper"><span id="wrapper_inread"><div(.+)<\/div><\/span><\/div>/';

+  $temp = preg_replace($re, '', $article_only); 

+  $article_only = $temp;

+  $re='/<\/div> <\/a> <a><div>/';

+  $temp = preg_replace($re, '',$article_only);

+  $article_only = $temp;

+

+  //Remove Emptylines

+  $temp = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);

+  $article_only = $temp;

+  $temp = preg_replace('/\s\s+/', ' ', $article_only);

+  $article_only = $temp;

+

+  

+  $re='/<iframe allow="autoplay; encrypted-media" allowfullscreen="" frameborder="0" height="(.+?)" src="https:\/\/www\.youtube\.com\/embed\/(.+?)" width="(.+?)"><\/iframe>/';

+  $temp = preg_replace($re, '<iframe allow="autoplay; encrypted-media" allowfullscreen="" frameborder="0" width="100%" src="https://www.youtube.com/embed/\\2"></iframe>', $article_only); 

+  $article_only = $temp;

+

+  $SEARCH='<div class="sharing-tools sharing-tools--align-center">';

+  $posend = strpos($article_only,$SEARCH);

+  if($posend) {

+    $temp = substr( $article_only, 0, $posend );

+    $article_only = $temp;

+  }

+

+  //Portfolio specific

+  if($isPortfolio) {

+    $re = '/<img class="cover parallax-enable fit lazyload" data-speed="2" data-full-src="(.+?)" alt="(.+?)">/';

+    $temp = preg_replace($re, '<img src="\\1" alt="\\2">', $article_only );

+    $article_only = $temp;

+    $re = '/" width="(.+)" height="(.+)">/';

+    $temp = preg_replace($re, '>', $article_only );

+    $article_only = $temp;

+  }

+  //Force HTTPS

+  $article_only = add_https($article_only, 'www.twitter.com');

+  $article_only = add_https($article_only, 'www.facebook.com');

+  $article_only = add_https($article_only, 'pbs.twimg.com');

+  $article_only = add_https($article_only, 'schema.org');

+  $article_only = add_https($article_only, 'www.w3.org');

+  $article_only = add_https($article_only, 'www.slate.fr');

+  echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<button id="html-btn" onclick="printHTML()" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button>'.PHP_EOL;

+  echo '<button id="pdf-btn" onclick="printPDF()" style="display:inline;"><img src="img/pdf.png" width="24px" height="24px"></button>'.PHP_EOL;

+  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;

+  echo '<div class="extract-content" id="'.$cpt.'">'.PHP_EOL;

+  echo '<h1>'.$article['title'].'</h1>'.PHP_EOL;

+  echo '<img src="'.$article['image'].'">'.PHP_EOL;

+  echo '<p><em>'.$chapo.'</em></p>'.PHP_EOL;

+  echo $article_only;

+  echo '</div>'.PHP_EOL;

+  echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;

+  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;

+  $cpt++;

+  if( $cpt > $NEWS_RSS_MAX_ITEMS ) {

+      break;

+  }

+}

+?>

+

@@ -10,6 +10,7 @@ define('NEWS_RSS_NUMERAMA', 'https://www.numerama.com/rss/news.rss');

 define('NEWS_RSS_HUFFINGTONPOSTFR', 'https://www.huffingtonpost.fr/feeds/index.xml');

 define('NEWS_RSS_FRANCEINFO', 'https://www.francetvinfo.fr/titres.rss');

 define('NEWS_RSS_CONSPIRACY', 'http://www.conspiracywatch.info/feed');

+define('NEWS_RSS_LOBS', 'http://www.nouvelobs.com/rss.xml');

 $array_title=array(

   "lemonde" => "LeMonde.fr",

@@ -20,6 +21,7 @@ $array_title=array(

   "vicefr" => "Vice.fr",

   "franceinfo" => "FranceInfo",

   "lesinrocks" => "LesInrocks.fr",

+  "lobs" => "L'Obs",

   "conspiracy" => "Conspiracy Watch"

 );

 $array_url=array(

@@ -31,6 +33,7 @@ $array_url=array(

   "vicefr" => NEWS_RSS_VICEFR,

   "franceinfo" => NEWS_RSS_FRANCEINFO,

   "lesinrocks" => NEWS_RSS_LESINROCKS,

+  "lobs" => NEWS_RSS_LOBS,

   "conspiracy" => NEWS_RSS_CONSPIRACY

 );