newsFR / sources / news-lepoint.fr.php /
Yanik Cawidrone Added News parsers
625fcf5 6 years ago
1 contributor
170 lines | 7.619kb
<?php
include_once( 'news-constants.php' );

$rss_content = http_get_contents(NEWS_RSS_LEPOINTFR);
$xml = simplexml_load_string($rss_content);
if ($xml === false) {
  echo 'Failed to read RSS';
  exit(0);
} 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-221');
    $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><!-- ./col-4 -->';
echo '<div class="col-6">';
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;
  echo '<!-- ==================== article '.$cpt.'============== -->';
  echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
  echo "<hr>";
  echo "<a name=\"article-$cpt\">";
  $article_content = http_get_contents($article['link']);

  $SEARCH_SUB1='<h1 class="art-titre list-view">';
  $pos_start = strpos($article_content, $SEARCH_SUB1);
  $pos_stop = false;
  if(!$pos_start) {
    $SEARCH_SUB1='<h2 class="art-chapeau">';
    $pos_start = strpos($article_content, $SEARCH_SUB1);
  }
  if($pos_start) {
    $SEARCH_SUB1='<h3 class="header-meme-sujet">';
    $pos_stop = strpos($article_content, $SEARCH_SUB1);
    if(!$pos_stop) {
      $SEARCH_SUB1='<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>';
      $pos_stop = strpos($article_content, $SEARCH_SUB1);
      if(!$pos_stop) {
        $SEARCH_SUB1='</article>';
        $pos_stop = strpos($article_content, $SEARCH_SUB1);
      }
    }
  }

  if( $pos_start && $pos_stop ) {
    $article_only = substr($article_content, $pos_start, $pos_stop - $pos_start);
  } else {
    $article_only="";
  }
  
  $orgStrings = array( ' href="/',
    ' src="/sites/',
    '<img src=',
    ' data-src="/');
  $newStrings = array( ' href="https://www.lepoint.fr/',
    ' src="https://www.lepoint.fr/sites/',
    '<img width="100%" src=',
    ' data-src="https://www.lepoint.fr/' );
  $article_only  = str_replace($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">',
    '<h3> </h3>',
    '<h1>',
    '</h1>',
    '<p style="text-align:center">',
    '<div class="image-wrap panoramique bg-undefined">',
    '<div class="placement-left">' );
  $newStrings = array('<div>',
    '<div>',
    '<div>',
    '<div>',
    '<div>',
    '',
    '<h4>',
    '</h4>',
    '<p>',
    '<div>',
    '<div>');
  $article_only = str_replace($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 class="cboxElement" (.+?)><img (.+?)><\/a>/';
  $temp = preg_replace($re, '<img \\2>', $article_only);
  $article_only = $temp;
  $re = '/<noscript>(.+?)<\/noscript>/ms';
  $temp = preg_replace($re, '', $article_only);
  $article_only = $temp;
  
  $orgStrings = array( 'http://www.lepoint.fr/' );
  $newStrings = array( 'https://www.lepoint.fr/' );
  $article_only  = str_replace($orgStrings, $newStrings, $article_only);

  $re = '/<img class="w100" alt="(.+?)" data-lazyload="loading" onclick="(.+?)" src="(.+?)" title="(.+?)" data-src="(.+?)" \/>/';
  $temp = preg_replace($re, '<img width="100%" src="\\5">', $article_only);
  $article_only = $temp;

  $re = '/<img class="" src data-lazyload="loading" data-src="(.+?)" data-alt="(.+?)" width="(.+?)" height="(.+?)" alt title="(.+?)"\/>/ms';
  $temp = preg_replace($re, '<img width="100%" src="\\1">', $article_only);
  $article_only = $temp;

  $re = '/<img src data-lazyload="loading" data-src="(.+?)" data-alt="(.+?)" alt="(.+?)" width="(.+?)" height="(.+?)" \/>/';
  $temp = preg_replace($re, '<img width="100%" src="\\1">', $article_only);
  $article_only = $temp;

  $re = '/<img class="w100" alt="" data-lazyload="loading" onclick="(.+?)" src="(.+?)" title="" data-src="(.+?)" \/>/';
  $temp = preg_replace($re, '<img width="100%" src="\\3">', $article_only);
  $article_only = $temp;

  $re = '/<a href="(.+?)" class="button-type pat little-button-mob" onclick="(.+?)">Abonnez-vous<\/a>/';
  $temp = preg_replace($re, '', $article_only);
  $article_only = $temp;
  
  $re = '/<a onclick="(.+?)" data-shadowbox="#authform" href="(.+?)" id="connect_viafreemium" class="w100 block txtcenter br-4-all bg-grey-medium-blue pas font-black bold br-4-all upper">Identifiez-vous<\/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="'.$cpt.'">'.$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;
  }
}
?>