newsFR / sources / news-liberation.php /
df587cf 6 years ago
1 contributor
273 lines | 13.031kb
<?php
include_once( 'news-constants.php' );

$rss_content = file_get_contents(NEWS_RSS_LIBERATION);
$xml = simplexml_load_string($rss_content);
if ($xml === false) {
  echo 'Failed to read RSS';
} else {
  $channel = array();
  $channel['title'] = $xml->title;
  $channel['pubDate'] = $xml->updated;
  $channel['timestamp'] = strtotime($xml->updated);
  //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->entry as $item)
  {
    $article = array();
    $article['title'] = $item->title;
    $article['link'] = $item->link['href'];
    $orgStrings = array('?xtor=rss-450');
    $newStrings = array('');
    $article['link']  = str_replace($orgStrings, $newStrings, $article['link']);
    $article['pubDate'] = $item->updated;
    $article['timestamp'] = strtotime($item->updated);
    $article['description'] = $item->summary;
    $article['image']="img/liberation.png";
    foreach ($item->link as $item2 )
    {
      if( ( 0 == strcmp( $item2['type'], "image/jpeg") ) ||
          ( 0 == strcmp( $item2['type'], "image/png") ) ) {
        $pos_start=strpos($item2['href'],"?modified_at");
        if( $pos_start ) {
          $image_link=substr($item2['href'],0,$pos_start);
          $article['image']=substr($item2['href'],0,$pos_start);
        }
      }
    }

    $articles[$cpt] = $article;

    echo "<div onclick=\"onArticle($cpt)\" style=\"display:inline;\">".PHP_EOL;
    echo '<div class="img-menu"><img width="100%" src="'.$article['image'].'" style="display:inline;"></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>'.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>';
    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;
  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\"></a>";
  $article_content = file_get_contents($article['link']);
  $article_content_utf8 = mb_convert_encoding($article_content, 'HTML-ENTITIES', "UTF-8");
  $doc = new DOMDocument();
  $doc->preserveWhiteSpace = false;
  $doc->formatOutput       = true;
  $libxml_previous_state = libxml_use_internal_errors(true);
  $doc->loadHTML($article_content_utf8);
  libxml_clear_errors();
  libxml_use_internal_errors($libxml_previous_state);
  $articles = $doc->getElementsByTagName('article');
  $article_only="";
  foreach ( $articles as $node) {
    $article_only = DOMinnerHTML($node);
  }


  $SEARCH_SUB='<aside class="aside-column side-flow-bloc width-padded">';
  $pos_stop = strpos($article_only, $SEARCH_SUB);
  $article_only = substr($article_only, 0, $pos_stop);
 
  if(0==strcmp($article_only,"")) {
    $SEARCH_SUB='itemtype="http://schema.org/NewsArticle">';
    $pos_start=strpos($article_content_utf8, $SEARCH_SUB);
    if($pos_start) {
      $article_only = substr($article_content_utf8, $pos_start + strlen($SEARCH_SUB));
      $SEARCH_SUB='<aside class="outbrain">';
      $pos_stop=strpos($article_only, $SEARCH_SUB);
      if($pos_stop) {
        $article_only = substr($article_only, 0, $pos_stop);
      } else {
        $article_only = "Oups";
      }
    } else {
      $article_only = "";
      $articles = $doc->getElementsByTagName('div');
      foreach ( $articles as $node) {
        if(0==strcmp("answer-body",$node->getAttribute('class'))) {
          $article_only = "<h3>".$article['title']."</h3>".DOMinnerHTML($node);
        }else if(0==strcmp("live-content",$node->getAttribute('class'))) {
          //$article_only = "<h3>".$article['title']."</h3>".DOMinnerHTML($node);
          $temp = DOMinnerHTML($node);
          $article_only = $article_only.$temp;
        }
      }
    }
  }
  DEBUG("Article($cpt) : length = ".strlen($article_only));
  $isCheckNews=false;
  if( 0 == strlen($article_only)) {
    $articles = $doc->getElementsByTagName('article');
    $article_only="";
    foreach ( $articles as $node) {
      $article_only = DOMinnerHTML($node);
    }
    $isCheckNews=true;
  }

  $orgStrings = array( ' href="/' );
  $newStrings = array( ' href="https://www.liberation.fr/' );
  $article_only  = str_replace($orgStrings, $newStrings, $article_only);
  
  $orgStrings = array( '<p lang="en" dir="ltr">',
      '<p dir="ltr">',
      '<h4>',
      '</h4>',
      '<h1 class="article-headline">',
      '</h1>',
      '<h2 class="article-standfirst read-left-padding">',
      '<h2 class="article-standfirst">',
      '<h2 class="intertitre">',
      '</h2>',
      '<figure class="article-image article-header-image"><a role="button" class="figure-zoom">',
      '<figure class="article-image article-header-image"><a role="button" class="figure-zoom js-figure-zoom">',
      '<img width="534" data-lazyload="false" src="',
      '<i class="zoom-icon"> <svg class="icon " role="img" width="22" height="22"><use xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="#icon-zoom"></use></svg></i>',
      ' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"',
      ' width="960">',
      '<img class="live-image"',
      '<a class="slug"',
      '<figcaption class="legende" data-caption="',
      '<figcaption class="read-left-padding caption">',
      '</figcaption>' );
  $newStrings = array( '<p>',
      '<p>',
      '<h5>',
      '</h5>',
      '<h3>',
      '</h3>',
      '<h4>',
      '<h4>',
      '<h4>',
      '</h4>',
      '<figure>',
      '<figure>',
      '<img width="100%" src="',
      '',
      '',
      ' width="100%">',
      '<img width="100%"',
      '<a',
      '<em>',
      '<em>',
      '</em>' );

  $article_only = str_replace($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="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;
  $re = '/<span class="share">(.+?)<\/span>/';
  $temp = preg_replace($re, '', $article_only); 
  $article_only = $temp;
  $orgStrings = array( '<ul class="article-rel-list">','<li class="article-rel-item">  </li>','</ul>', '<a href=');
  $newStrings = array( '','','', '<a target="new" href=' );
  $article_only = str_replace($orgStrings, $newStrings, $article_only);

  //Checknews Specific
  if($isCheckNews) {
    $re='/<\/form>/';
    $temp = preg_replace($re, "", $article_only);
    $article_only = $temp;
    $re='/<form class="checknews-form" method="GET" action="\/checknews\/search\/" name="checknews_box">(.+)<\/a>/';
    $temp = preg_replace($re, "", $article_only);
    $article_only = $temp;
    /*
    $re='//';
    $temp = preg_replace($re, "", $article_only);
    $article_only = $temp;
     */
    $re='/<li class="answer no-tags">(.+?)<\/li>/';
    $temp = preg_replace($re, "", $article_only);
    $article_only = $temp;
    $re='/<li class="answer no-tags">(.+)<ul class="checknews-tags">/';
    $temp = preg_replace($re, "<li><ul>", $article_only);
    $article_only = $temp;
    $re='/<p class="btn" href="https:\/\/www\.liberation\.fr\/checknews\/(.+)Voir la réponse<br><span>(.+)<\/span> <\/p> <\/div> <\/a> <\/li>/';
    $article_only = preg_replace($re, "</div></a></li>", $article_only);
    $re='/<a class="btn about-button" href="https:\/\/www\.liberation\.fr\/checknews\/voter\/"> Accédez à toutes les questions <\/a>/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<script type="text\/javascript" src="https:\/\/statics\.liberation\.fr\/newsite\/(.+)<\/script>/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<div class="checknews-header"> <div class="title"> <h3>(Nos dernières réponses|Le vote des lecteurs)<\/h3> <\/div> <\/div>/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<p> Voici les cinq questions qui sont en tête des votes.<br> Vous aussi, donnez votre avis et choisissez les sujets que la rédaction va traiter. <\/p>/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<\/div><\/a><\/li> <li><ul>/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<ul id="question" class="checknews-answers vote">/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<ul class="checknews-answers">/';
    $article_only = preg_replace($re, "", $article_only);
    $re='/<div class="checknews-box last-content"> <li><ul>/';
    $article_only = preg_replace($re, "<div><li>", $article_only);
    $re='/ target="_blank">/';
    $article_only = preg_replace($re, ">", $article_only);
    $re='/<\/div><\/a><\/li>/';
    $article_only = preg_replace($re, "", $article_only);
    $article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
    $article_only = preg_replace('/\s\s+/', ' ', $article_only);
    $re='/<div class="checknews-box last-content"> <li><ul>/';
    $article_only = preg_replace($re, "<div>", $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;

  //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, 'videos.senat.fr');
  $article_only = add_https($article_only, '9w4t.mjt.lu');
  $article_only = add_https($article_only, 'www.lefigaro.fr');
  $article_only = add_https($article_only, 'www.leparisien.fr');
  $article_only = add_https($article_only, 'www.lepoint.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="'.$cpt.'">'.$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-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;
  echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
  $cpt++;
  if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
      break;
  }
}
?>