news / sources / news-lemonde.fr.php /
3825201 6 years ago
1 contributor
174 lines | 7.744kb
<?php
include_once( 'news-constants.php' );

$rss_content = file_get_contents(NEWS_RSS_LEMONDE);
$xml = simplexml_load_string($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>';
  $cpt=0;
  foreach ($xml->channel->item as $item) {
    $article = array();
    $article['title'] = $item->title;
    $article['link'] = $item->link;
    $orgStrings = array('?xtor=RSS-3208');
    $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;">';
    echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
    echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;';
    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 $article['title'].'&nbsp;&nbsp;';
    echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
    $cpt++;
    if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
      break;
    }
  }
}
echo '</div><!-- ./panel-body -->';
echo '</div><!-- ./panel panel-default -->';
echo '</div><!-- ./col-md-6 -->';
echo '<div class="col-md-6">';
echo '<div class="panel panel-default">';
echo '<div class="panel-body">';
echo '<a name="article-top"></a><div id="article-current"></div>';
$cpt=0;
foreach ($articles as $article ) {
  $cpt_prev=$cpt-1;
  $cpt_next=$cpt+1;
  echo '<!-- ==================== article '.$cpt.'============== -->';
  echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
  echo "<hr>";
  echo "<a name=\"article-$cpt\">";
  $article_content = file_get_contents($article['link']);
  $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('article');
  $article_only="";
  if( isset($articles[0]) ) {
    $article_only=DOMinnerHTML($articles[0]);
  } else {
    $article_only = "Extraction Failed";
    break;
  }
  $figure="";
  
  $SEARCH_SUB='<span class="ea_article">édition abonné</span>';
  $pos_start=strpos($article_content, $SEARCH_SUB);
  if($pos_start) {
    $figures = $doc->getElementsByTagName('figure');
    $figure = DOMinnerHTML($figures[0]);
    $re = '/<img (.+?) class="illu lazy-retina" data-src="(.+?)" (.+?) data-lazyload="true" alt="(.+?)" title="(.+?)" (.+?)>/';
    preg_match($re, $figure, $array);
    $figure = '<img width="100%" src="'.$array[2].'"><figcaption>'.$array[4].'<br>&nbsp;<em>('.$array[5].')</em></figcaption>';

    $article_abonne = str_replace("www.lemonde.fr", "abonnes.lemonde.fr", $article['link']);
    $article_content = file_get_contents($article_abonne);
    $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('article');
    $article_only="";
    if( isset($articles[0]) ) {
      $article_only=DOMinnerHTML($articles[0]);
    } else {
      $article_only = "Extraction Failed for Subscribed Article";
      break;
    }
  }
 
  $orgStrings = array(' href="/');
  $newStrings = array(' href="https://www.lemonde.fr/');
  $article_only  = str_replace($orgStrings, $newStrings, $article_only);
  
  $orgStrings = array( 
      '<p class="txt3 description-article" itemprop="description">',
      '<p class="bloc_signature">',
      '<span id="publisher" itemprop="Publisher" data-source="LE MONDE">',
      '<div class="toolbar"></div>',
      '<div id="articleBody" class="contenu_article js_article_body" itemprop="articleBody">',
      '<h4>',
      '</h4>',
      '<h1 class="tt2" itemprop="Headline">',
      '</h1>',
      '<h2 class="taille_courante">',
      '<h2 class="intertitre">',
      '</h2>',
      '<span>Le Monde</span>',
      '<figure class="illustration_haut   " style="width: 534px">',
      '<img width="534" data-lazyload="false" src="',
      ' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"' );
  $newStrings = array( 
      '<p>',
      '<p>',
      '<span>',
      '',
      '<div>',
      '<h5>',
      '</h5>',
      '<h3>',
      '</h3>',
      '<h4>',
      '<h4>',
      '</h4>',
      'Le Monde',
      '<figure>',
      '<img width="100%" src="',
      '' );

  $article_only = str_replace($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 = '/<a target="_blank" onclick="return false;" (.+?)">(.+?)<\/a>/';
  //$temp = preg_replace($re, '\\3', $article_only);
  //$article_only = $temp;
  $re = '/<a class="lien_interne rub"(.+?)>(.+?)<\/a>/';
  $temp = preg_replace($re, '\\2', $article_only);
  $article_only = $temp;
  $re = '/<img (.+?)data-src="(.+?)"(.+?)data-lazyload="true"(.+?)>/';
  $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only); 
  $article_only = $temp;
  $re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
  $temp = preg_replace($re, '', $article_only); 
  $article_only = $temp;
  echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
  echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
  echo '<div class="extract-content" id="'.$cpt.'">'.$figure.$article_only.'</div>';
  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;';
  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;';
  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;';
  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
  $cpt++;
  if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
      break;
  }
}
?>