1 contributor
<?php
include_once( 'news-constants.php' );
$rss_content = file_get_contents(NEWS_RSS_STAR);
$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'] = "star";
$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> '.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> '.PHP_EOL;
echo $article['title'].' ';
echo '<div id="nav-source" style="display:inline;"><a href="'.$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('div');
$article_only = "";
foreach ( $articles as $node) {
$classname = $node->getAttribute('class');
if($classname == 'article-body-items' ) {
$article_only = DOMinnerHTML($node);
break;
} else if($classname == 'article-content-container' ) {
$article_only = DOMinnerHTML($node);
break;
}
}
if( 0 == strlen($article_only)) {
$article_only = "Extraction failed";
}
$article_only = re_remove($article_only, '/<div class="side-box__button"><span><span><a href="\/emails\.html">SIGN UP<\/a><\/span><\/span><\/div>/');
$article_only = re_remove($article_only, '/<span class="subline-icon-color-headlines">NEW NEWSLETTER<\/span><span>HEADLINES<\/span>/');
$article_only = re_remove($article_only, '/<li class="report-error"><a><i class="staricon staricon-error-reporting "><\/i>Report an error<\/a><\/li>/');
$article_only = re_remove($article_only, '/<li class="about-the"><span><span><a href="\/about\/aboutus\.html"><i class="staricon staricon-trust-icon "><\/i>About <em>The Star<\/em><\/a><\/span><\/span><\/li>/');
$article_only = re_remove($article_only, '/<div class="banner-inner-content__left side-box__title title-color-headlines"><div class="about-text">TOP STORIES, DELIVERED TO YOUR INBOX.<\/div><\/div>/');
$article_only = re_remove($article_only, '/<li class="journalistic-standards"><span><span><a href="\/about\/statementofprinciples.html"><i class="staricon staricon-journalistic-standards "><\/i>Journalistic Standards<\/a><\/span><\/span><\/li>/');
$article_only = re_remove($article_only, '/<div class="article-continued"><div class="article-continued__heading">Article Continued Below<\/div><\/div>/');
$article_only = re_remove($article_only, '/<div class="alpha__big-box__body"><div class="big-box"><\/div><\/div>/');
//$article_only = re_remove($article_only, '//');
//$article_only = re_remove($article_only, '//');
$article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> '.PHP_EOL;
echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div> '.PHP_EOL;
echo '<button id="html-btn" onclick="printHTML()" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button>'.PHP_EOL;
echo '<button id="pdf-btn" onclick="printPDF()" style="display:inline;"><img src="img/pdf.png" width="24px" height="24px"></button>'.PHP_EOL;
echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div> '.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.'">'.'<h1>'.$article['title'].'</h1>'.$article_only.'</div>';
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-up fa-2x"></i></a></div> '.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> '.PHP_EOL;
echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div> '.PHP_EOL;
echo '<div id="nav-next" onclick="onArticle('.$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;
}
}
?>