1 contributor
<?php
include_once( 'news-constants.php' );
$rss_content = file_get_contents(NEWS_RSS_LATIMES);
$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['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>';
$cpt=0;
foreach ($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;
$article['image']=$item->mediacontent['url'];
$articles[$cpt] = $article;
echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
echo '<img src="'.$article['image'].'" style="display:block;"><br>';
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> ';
echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
echo $article['title'].' ';
echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
$cpt++;
if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
break;
}
}
}
echo '</div><!-- ./panel-body -->';
echo '</div><!-- ./panel panel-default -->';
echo '</div><!-- ./col-md-6 -->';
echo '<div class="col-md-6">';
echo '<div class="panel panel-default">';
echo '<div class="panel-body">';
echo '<a name="article-top"></a><div id="article-current"></div>';
$cpt=0;
foreach ($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;
$doc->loadHTML($article_content);
$articles = $doc->getElementsByTagName('article');
$article_only="";
foreach ( $articles as $node) {
$article_only = DOMinnerHTML($node);
}
/*
$orgStrings = array(' href="/');
$newStrings = array(' href="https://www.latimes.com/');
$article_only = str_replace($orgStrings, $newStrings, $article_only);
$orgStrings = array(' src="/');
$newStrings = array(' src="https://www.latimes.com/');
$article_only = str_replace($orgStrings, $newStrings, $article_only);
*/
//Clean some sections
//$article_only = re_remove($article_only, '//');
$article_only = re_remove($article_only, '/<a class="button button-icon button-icon-hover-accent " href=(.+?)> <i class="fa fa-share button-icon-inset-accent"><\/i> <\/a>/');
$article_only = re_remove($article_only, '/<button class="button button-icon button-icon-hover-accent (.+?)> <i class="fa fa-(.+?) button-icon-inset-accent"><\/i> <\/button>/');
$article_only = re_remove($article_only, '/<div class="spaced spaced-sm spaced-bottom hidden-desktop hidden-tablet"><\/div>/');
$article_only = re_remove($article_only, '/<div class="spaced spaced-md spaced-bottom hidden-mobile"><\/div>/');
$article_only = re_remove($article_only, '/<div class="ad-caption-text align-center uppercase">Advertisement<\/div> /');
$article_only = re_remove($article_only, '/<div class="desktop-nativo mobile-yieldmo inline-ad-arrow "> <i class="fa fa-angle-down flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="desktop-nativo mobile-yieldmo inline-ad-arrow less-spacing hidden-desktop hidden-mobile"> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="desktop-nativo mobile-yieldmo inline-ad-arrow less-spacing hidden-desktop hidden-tablet "> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-dfp inline-ad-arrow hidden-desktop"> <i class="fa fa-angle-down flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-dfp inline-ad-arrow less-spacing hidden-desktop hidden-mobile"> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-dfp inline-ad-arrow less-spacing hidden-desktop hidden-tablet "> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-nativo inline-ad-arrow hidden-desktop"> <i class="fa fa-angle-down flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-nativo inline-ad-arrow less-spacing hidden-desktop hidden-mobile"> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-nativo inline-ad-arrow less-spacing hidden-desktop hidden-tablet "> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-yieldmo inline-ad-arrow hidden-desktop"> <i class="fa fa-angle-down flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-yieldmo inline-ad-arrow less-spacing hidden-desktop hidden-mobile"> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
$article_only = re_remove($article_only, '/<div class="mobile-yieldmo inline-ad-arrow less-spacing hidden-desktop hidden-tablet "> <i class="fa fa-angle-up flex-container-column flex-row-reverse"><\/i> <\/div>/');
//Some little replacements
$re = '/<div class="preview-text">/';
$article_only = preg_replace($re, '<div style="display:none;">', $article_only);
$re = '/<div class="flex-container-row ">/';
$article_only = preg_replace($re, '<div style="display:none;">', $article_only);
$re = '/<div class="flex-container-column">/';
$article_only = preg_replace($re, '<div style="display:none;">', $article_only);
$re = '/<li class="collection-item flex-container-row align-items-start collection-item-border-bottom ">/';
$article_only = preg_replace($re, '<li>', $article_only);
$re = '/<div class="flex">/';
$article_only = preg_replace($re, '<div class="flex" style="display:none;">', $article_only);
$re='/<ul class="tag-list-wrapper list-unstyled flex-container-row align-left">/';
$article_only = preg_replace($re, '<ul style="display:none;">', $article_only);
$re = '/<div class="mobile-yieldmo inline-ad hidden-desktop" data-inline-ad-count="4"> <div class="wrapper clearfix col pb-feature pb-layout-item pb-f-ads-yieldmo" data-pb-name="Yieldmo Ad" data-pb-curated="curated" id=""> <div id="(.+?)"><\/div> <script>window.serviceCallbacks.push(.+?)<\/script>/';
$article_only = preg_replace($re, '<div><div>', $article_only);
$article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
$article_only = re_remove($article_only, '/<span class="sharebar align-right flex"> <\/span>/');
$article_only = re_remove($article_only, '/ "data-arctrack-item-page": "",/');
$article_only = re_remove($article_only, '/ "data-arctrack-item-producer": "lanews",/');
$article_only = re_remove($article_only, '/ "data-arctrack-item-videoid": "(.+?)",/');
$article_only = re_remove($article_only, '/ "data-arctrack-item-classification": "(.+?)",/');
$article_only = re_remove($article_only, '/ "data-arctrack-item-name": "(.+?)",/');
$article_only = re_remove($article_only, '/"data-arctrack-item-autoplay": "false"/');
$article_only = re_remove($article_only, '/ type: "video",/');
$article_only = re_remove($article_only, '/ system: "Goldfish",/');
$article_only = re_remove($article_only, '/ url: "http:\/\/www.latimes.com\/resizer\/(.+?)",/');
$article_only = re_remove($article_only, '/ headline: "(.+?)",/');
$article_only = re_remove($article_only, '/ brightcoveVideoId: "(.+?)",/');
$article_only = re_remove($article_only, '/ goldfishVideoId: "(.+?)",/');
$article_only = re_remove($article_only, '/ urlId: "(.+?)",/');
$article_only = re_remove($article_only, '/ site: "",/');
$article_only = re_remove($article_only, '/ date: "(.+?)",/');
$article_only = re_remove($article_only, '/ duration: "(.+?)",/');
$article_only = re_remove($article_only, '/ description: "(.+?)",/');
$article_only = re_remove($article_only, '/ rand: "(.+?)",/');
$article_only = re_remove($article_only, '/ analyticsData:/');
$article_only = re_remove($article_only, '/ const videos(.+?) = \[/');
$article_only = re_remove($article_only, '/^ },$/m');
$article_only = re_remove($article_only, '/^ {$/m');
$article_only = re_remove($article_only, '/^}$/m');
$article_only = re_remove($article_only, '/^{$/m');
$article_only = re_remove($article_only, '/ sharebar: "(.+?)",/');
$article_only = re_remove($article_only, '/<div class="goldfish-info" data-script-path="diodxcszlbmwk" data-environment="prod"><\/div>/');
$article_only = re_remove($article_only, '/<div class="brightcove-info" data-account="(.+?)" data-player="(.+?)"><\/div>/');
$article_only = re_remove($article_only, '/<div class="logo logo-full logo-src-1 header-logo"><\/div>/');
$article_only = re_remove($article_only, '/<div class="card flex-container-column align-items-stretch"> <div class="card-header flex-container-row align-items-center"> <\/div> <\/div>/');
$article_only = re_remove($article_only, '/<div id="pb-(.+?)"><\/div>/');
$article_only = re_remove($article_only, '/<button disabled name="loading" class="button-icon button-accent button-rounded button-lg button-loading"> <i class="fa fa-circle-o-notch fa-spin fa-fw"><\/i> <\/button>/');
$article_only = re_remove($article_only, '/<button name="error" aria-label="error" type="submit" class="button-icon button-error button-lg button-rounded"> <i class="fa fa-exclamation"><\/i> <\/button>/');
$article_only = re_remove($article_only, '/<div> <h6 class="flex capitalized">Today\'s Headlines Newsletter<\/h6> <\/div>/');
$article_only = re_remove($article_only, '/ <div class="caption-text">Weekdays<\/div>/');
$article_only = re_remove($article_only, '/<div class="align-center"> <button name="show-comments" (.+?)> Be the first to comment <\/button> <button name="hide-comments" (.+?)>Hide Comments<\/button> <\/div>/');
$re = '/<div class="wrapper clearfix full pb-feature pb-layout-item pb-f-article-video-playlist" data-pb-name="Video Playlist" data-pb-curated="curated" id="(.+?)">/';
$article_only = preg_replace($re, '<div>', $article_only);
$article_only = re_remove($article_only, '/<div class="trb_cm_so" data-role="cm_container"> <div data-role="comments" data-comments-id="(.+?)" class="so_comments hidden" data-sitename="lanewsprod" data-content-id="lanews(.+?)" data-type="promo-comment" data-publisher="tronc"><\/div> <\/div>/');
$article_only = re_remove($article_only, '/<div class="wrapper clearfix full pb-feature pb-layout-item pb-f-article-comments" data-pb-name="Article Comments" data-pb-curated="curated" id="(.+?)"> <noscript>Please enable JavaScript to view the <a href="http:\/\/solidopinion.com\/">comments powered by SolidOpinion.<\/a>/');
$re = '/<\/noscript> <\/div>/';
$article_only = preg_replace($re, '<div></div>', $article_only);
$article_only = re_remove($article_only, '/<form data-cm-aid="(.+?)" data-captcha-site-key=(.+?)>(.+?)<\/form>/');
//$article_only = re_remove($article_only, '//');
//$article_only = re_remove($article_only, '//');
//$article_only = re_remove($article_only, '//');
$article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> ';
echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div> ';
echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div> ';
echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> ';
echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div> ';
echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div> ';
echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div> ';
echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
$cpt++;
if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
break;
}
}
?>