newsEN / sources / news-nykrmag.php /
390ba05 6 years ago
1 contributor
122 lines | 9.066kb
<?php
include_once( 'news-constants.php' );

$rss_content = file_get_contents(NEWS_RSS_NYKR_MAG);
$orgStrings = array( 'media:thumbnail');
$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'] = "nykrmag";
  $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>' . $channel['title'].PHP_EOL;
  echo '<button id="html-btn" onclick="printHTMLAll(\''.$channel['name'].'.html\')" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button>'.PHP_EOL;
  echo '<button id="pdf-btn" onclick="printPDFAll(\''.$channel['name'].'.pdf\')" style="display:inline;"><img src="img/pdf.png" width="24px" height="24px"></button>'.PHP_EOL;
  echo '</h4>'.PHP_EOL;
  $cpt=0;
  foreach ($xml->channel->item as $item) {
    $article = array();
    $article['title'] = $item->title;
    $article['link'] = $item->link;
    $SEARCHREF="?ref=rss";
    //Remove reference from links
    $posref=strpos($article['link'], $SEARCHREF);
    if( $posref ) {
      $article['link'] = substr($article['link'],0,$posref);
    }
    $article['pubDate'] = $item->pubDate;
    $article['timestamp'] = strtotime($item->pubDate);
    $article['description'] = $item->description;
    $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;'.PHP_EOL;
    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>'.PHP_EOL;
    $cpt++;
    if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
      break;
    }
  }
}

echo '</div><!-- ./col-4 -->'.PHP_EOL;
echo '<div id="article-display" class="col-6">'.PHP_EOL;
echo '<a name="article-top"></a><div id="article-current"></div>'.PHP_EOL;
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\">";
  $article_content = file_get_contents($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);

  //Article is in <div class="article-body-items"
  $articles = $doc->getElementsByTagName('article');
  if( null != $articles ) {
    $article_only = DOMinnerHTML($articles[0]);
  }
  if( 0 == strlen($article_only)) {
    $article_only = "Extraction failed";
  }

  $article_only = re_remove($article_only, '/<li data-track-type="socialShare" data-track-location="default" data-source=(.+?)<\/svg><\/i><\/div><\/a><\/li>/');
  $article_only = re_remove($article_only, '/<h3 class="RecirculationMostPopular__title___(.+?)">Most Popular<\/h3>/');
  $article_only = re_remove($article_only, '/<a class="Link__link___(.+?) RecirculationMostPopular__more___(.+?) RecirculationMostPopular__more___(.+?)" href="\/popular\?intcid=most-popular"><span>our thirty most popular<\/span><\/a>/');
  $article_only = re_remove($article_only, '/<h5 class="RecirculationRecommendedStories__header___(.+?)">Recommended Stories<\/h5>/');
  $article_only = re_remove($article_only, '/<span class="RecirculationMostPopular__rubric___(.+?)">(.+?)<\/span><a class="Link__link___(.+?)  " href="(.+?)"><h5 class="RecirculationMostPopular__hed___(.+?)">(.+?)<\/h5><\/a><div class="Byline__theLatest___(.+?)"><p class="Byline__by___(.+?)">(.+?)<a class="Link__link___(.+?)" title="(.+?)" rel="author" href="(.+?)">(.+?)<\/a><\/p><\/div>/');
  $article_only = re_remove($article_only, '/<div class="Avatar__avatar___(.+?) ArticleContributors__bylineAvatar___(.+?)"><a class="Link__link___(.+?)" href="(.+?)"><\/picture><\/div><\/a><\/div>/');
  $article_only = re_remove($article_only, '/<div class="Avatar__avatar___(.+?) ArticleContributors__bioAvatar___(.+?)"><a class="Link__link___(.+?)" href="(.+?)"><\/picture><\/div><\/a><\/div>/');
  $article_only = re_remove($article_only, '/<div data-cm-unit="show-failsafe"><div data-cns="cns-slot-end-of-article" class="ArticleFooter__adEndArticle___(.+?)"><\/div><\/div>/');
  $article_only = re_remove($article_only, '/<p class="NewsletterSubscribe__newsletterDek___(.+?)">(.+?)<\/p>/');
  $article_only = re_remove($article_only, '/<a class="Link__link___(.+?) ArticleContributors__contributorBioLink___(.+?)" href="(.+?)">Read more(.+?)<\/a>/');
  $article_only = re_remove($article_only, '/<input type="email" autocomplete="on" class="Input__input___(.+?)" name="" placeholder="Enter your e-mail"><button class="Input__inputButton___(.+?)" type="submit">Submit<\/button>/');
  $article_only = re_remove($article_only, '/<div class="SectionBreak SectionBreak__sectionBreak___(.+?)"><div class="" data-type="callout" data-callout="interactive"><div class=""><iframe class="IframeEmbed__iframe___(.+?) IframeEmbed__noBorder___(.+?)" height="(.+?)" scrolling="auto" src="http:\/\/audm\.herokuapp\.com\/player-embed(.+?)" title="" width="(.+?)"><\/iframe><\/div><\/div><\/div>/');
  $article_only = re_remove($article_only, '/<button class="Button__button___(.+?) Button__seeMore___(.+?)  ArticleLedeImage__button___(.+?)"><div class="Button__content___(.+?)"><i class="Button__icon___(.+?) Button__left___(.+?)"><svg width="18" height="18" viewbox="0 0 18 18" xmlns="http:\/\/www.w3.org\/2000\/svg"><path (.+?)><\/path><\/svg><\/i><\/div><\/button>/');
  //$article_only = re_remove($article_only, '//');
  //$article_only = preg_replace('//','',$article_only);
  $article_only = preg_replace('/<div class="RecirculationRecommendedStories__story___(.+?)" data-track-type="recircItem" data-track-location="mod-related" data-service="tbd" data-source="(.+?)" data-track-position="(.+?)">/','<div style="display: none;">',$article_only);
  $article_only = preg_replace('/<li class="RecirculationMostPopular__item___(.+?)" data-track-type="recircItem" data-track-location="mod-most-popular" data-source="(.+?)" data-track-position="(.+?)">/', '<li style="display: none;">', $article_only);
  $article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $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="'.$cpt.'">'.$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;'.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;
  }
}
?>