newsFR / sources / news-lobs.php /
31bd7ab 6 years ago
1 contributor
136 lines | 6.87kb
<?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 $article['title'].'&nbsp;&nbsp;'.PHP_EOL;
    echo '<div id="nav-source" style="display:inline;"><a href="'.$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">'.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');
    }
  }
  $orgStrings = array(' href="/');
  $newStrings = array(' href="https://www.nouvelobs.com/');
  $article_only  = str_replace($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;

  $article_only = re_remove($article_only, '/ <footer class="ObsArticle-footer(.+?)">(.+?)<\/footer>/');

  //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;

  
  //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.nouvelobs.com');
  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;
  }
}
?>