newsFR / sources / news-huffingtonpostfr.php /
231f665 6 years ago
1 contributor
205 lines | 10.633kb
<?php
include_once( 'news-constants.php' );

$rss_content = file_get_contents(NEWS_RSS_HUFFINGTONPOSTFR);
$orgStrings = array( 'media:content');
$newStrings = array( 'mediacontent');
$rss_content  = str_replace($orgStrings, $newStrings, $rss_content);
$xml = simplexml_load_string($rss_content);
if ($xml === false) {
  echo 'Failed to read RSS';
} else {
  $channel = array();
  $channel['name'] = "huffingtonpostfr.html";
  $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>' . $channel['title'].'<button id="html-btn" onclick="printHTMLAll(\''.$channel['name'].'\')" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button>'.PHP_EOL;
  echo '<button id="epub-btn" onclick="printEPUBAll(\''.$channel['name'].'.epub\')" style="display:inline;"><img src="img/epub.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;
    $article['pubDate'] = $item->pubDate;
    $article['timestamp'] = strtotime($item->pubDate);
    $article['description'] = $item->description;
    $SEARCH_SUB1='<p><img width="1024" height="';
    $pos_start = strpos($item->description, $SEARCH_SUB1);
    $pos_start += strlen($SEARCH_SUB1);
    $SEARCH_SUB2="\"";
    $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
    $height=substr($item->description, $pos_start, $pos_stop - $pos_start);
    $article['width']=1024;
    $article['height']=$height;
    $SEARCH_SUB1='src="';
    $pos_start = strpos($item->description, $SEARCH_SUB1);
    $pos_start += strlen($SEARCH_SUB1);
    $SEARCH_SUB2="\"";
    $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
    $article['image']=$item->mediacontent['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 $article['title'].'&nbsp;&nbsp;'.PHP_EOL;
    echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><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">';
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 PHP_EOL.PHP_EOL."<!-- ==================== article '.$cpt.'============== -->".PHP_EOL;
  echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
  echo "<hr>";
  echo "<a name=\"article-$cpt\">\n";
  $article_content = file_get_contents($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');
  foreach ( $articles as $node) {
    $article_only = DOMinnerHTML($node);
  }
  
  $orgStrings = array( '<img src="',
    'style="width: 680px"',
    '<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>',
    ' width="500" height="281" frameborder="0" ');
  $newStrings = array( '<img width="100%" src="',
    'style="width: 100%"',
    '',
    ' width="100%" frameborder="0" ');
  $article_only  = str_replace($orgStrings, $newStrings, $article_only);
  
  $SEARCH_SUB1='<p><strong>Lire aussi :</strong></p>';
  $pos_stop = strpos($article_only, $SEARCH_SUB1);
  if($pos_stop) {
    $temp = substr($article_only, 0, $pos_stop);
    $article_only = $temp."</blockquote>";
  } else {
    //End of article cleanup
    $SEARCHSUB='<h3 class="related-entries__title">LIRE AUSSI</h3>';
    $pos_stop=false;
    $pos_stop = strpos($article_only, $SEARCHSUB);
    if($pos_stop) {
      $temp = substr($article_only, 0, $pos_stop);
      $article_only = $temp;
    } else {
      $SEARCHSUB='<h3 class="recirc__title trending">';
      $pos_stop=false;
      $pos_stop = strpos($article_only, $SEARCHSUB);
      if($pos_stop) {
        $temp = substr($article_only, 0, $pos_stop);
        $article_only = $temp;
      } else {
        DEBUG("Article($cpt) : No Match");
      }
    }
  }

  $re = '/alt="(.+?)" width="(.+?)" height="(.+?)"/';
  $temp = preg_replace($re, 'width="100%"', $article_only);
  $article_only = $temp;
  $re = '/alt="" width="(.+?)" height="(.+?)"/';
  $temp = preg_replace($re, 'width="100%"', $article_only);
  $article_only = $temp;
  $re = '/<iframe width="500" height="(.+?)"/';
  $temp = preg_replace($re, '<iframe width="100%"', $article_only);
  $article_only = $temp;
  $re = '/sizes="\(max-width: (.+?)\) 100vw, (.+?)"/';
  $temp = preg_replace($re, 'sizes="(max-width: 100%) 100vw,100%"', $article_only);
  $article_only = $temp;
  $re = '/<figure id="attachment_(.+?)" style="width: (.+?)" class="wp-caption aligncenter">/';
  $temp = preg_replace($re, '<figure id="attachment_\\1" style="width: 100%" class="wp-caption aligncenter">', $article_only);
  $article_only = $temp;
  $re = '/<img class="aligncenter size-full wp-image-(.+?)"/';
  $temp = preg_replace($re, '<img ', $article_only);
  $article_only = $temp;
  $re = '/<iframe (.+?) height="(.+?)" (.+?) width="(.+?)">/';
  $temp = preg_replace($re, '<iframe \\1 \\3 width="100%">', $article_only);
  $article_only = $temp;
  $re = '/<img class="image__src"/';
  $temp = preg_replace($re, '<img class="image__src" width="100%"', $article_only);
  $article_only = $temp;

  $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;
  $orgStrings = array( 
    '<div id="entry_paragraph_1" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>',
    '<div id="entry_paragraph_2" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>',
    '<div id="entry_paragraph_3" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>',
    '<div id="entry_paragraph_4" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>');
  $newStrings = array( 
    '',
    '',
    '',
    '');
  $article_only  = str_replace($orgStrings, $newStrings, $article_only);

  // Cleanup "Social"
  $re='/<script id="share-bar__social-mob" (.+)<\/script>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<li class="(facebook|twitter|gplus|pinterest|linkedin|email) share-bar__item">(.+)<\/a><\/li>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<li class="(facebook|twitter|snapchat|instagram) follow-us-networks-item " (.+?)<\/span> <\/a> <\/li>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<div class="social-overlay__qr">(.+)<\/a><\/div>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<script type="text\/javascript"> \(function\(\)\{ var dfp_placement_slug(.+)<\/script>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<form class="newsletter-signup" method="POST">(.+)<\/form>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<div class="follow-us__header"> <h3>SUIVEZ-NOUS<\/h3> <\/div>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<div class="follow-us__description"> <p>Avec la newsletter quotidienne du HuffPost(.+) target="_blank">En savoir plus<\/a> <\/p> <\/div>/';
  $article_only = preg_replace($re, '', $article_only);
  $re='/<li class="(facebook|instagram) follow-us-networks-item even" (.+) <\/a> <\/li>/';
  $article_only = preg_replace($re, '', $article_only);
  //Clean Empty lines
  $article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
  $article_only = preg_replace('/\s\s+/', ' ', $article_only);



  
  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="'.$cpt.'">';
  echo '<h4>'.$article['title'].'</h4>';
  echo $article_only.'</div>'.PHP_EOL;
  echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
  echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
  echo '<div id="nav-source" style="display:inline;"><a href="'.$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('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
  echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
  $cpt++;
  if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
      break;
  }
}
?>