@@ -7,4 +7,5 @@ php news-latimes.php > latimes.html 2> /dev/null

 php news-cna.php > cna.html 2> /dev/null

 php news-aje.php > aje.html 2> /dev/null

 php news-bbc.php > bbc.html 2> /dev/null

+php news-smh.php > smh.html 2> /dev/null

 scp *html rss.php kawi.fr:/var/www/html/newsen/sources/ &> /dev/null

@@ -4,3 +4,8 @@ php news-nyt.php > nyt.html 2> /dev/null

 php news-wapo.php > wapo.html 2> /dev/null

 php news-guardian.php > guardian.html 2> /dev/null

 php news-latimes.php > latimes.html 2> /dev/null

+php news-cna.php > cna.html 2> /dev/null

+php news-aje.php > aje.html 2> /dev/null

+php news-bbc.php > bbc.html 2> /dev/null

+php news-smh.php > smh.html 2> /dev/null

+#scp *html rss.php kawi.fr:/var/www/html/newsen/sources/ &> /dev/null

@@ -0,0 +1,114 @@

+<?php

+include_once( 'news-constants.php' );

+

+$rss_content = file_get_contents(NEWS_RSS_SMH);

+$xml = simplexml_load_string($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;

+    $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;">';

+    //echo '<img id="img-list" 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>&nbsp;&nbsp;';

+    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'].'  ';

+    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="";

+  if( isset( $articles[0] ) ) {

+    $article_only=DOMinnerHTML($articles[0]);

+  }

+  

+  //Clean some sections

+	//$article_only = re_remove($article_only, '//');

+	$article_only = re_remove($article_only, '/<header class="_2qhpu"><h2 class="_2qhpu">Most Viewed in (.+)<\/h2><\/header>/');

+	$article_only = re_remove($article_only, '/<div><section aria-live="polite" aria-busy="true" class="_2zaYr" style="height:380px"><div class="_29X_Z"><span class="_2wzgv D5idv _3lVFK"><span class="_29Qt8"><\/span><span class="_3qqDc">Loading<\/span><\/span><\/div><\/section><\/div>/');

+	$article_only = re_remove($article_only, '/<div class="_8Cum8"><a class="_4oUp_" href="(.+?)">License this article<\/a><\/div>/');

+	$article_only = re_remove($article_only, '/<p class="_3MW1W">Our new podcast series from the team behind Phoebe\'s Fall<\/p>/');

+	$article_only = re_remove($article_only, '/<p class="_29xCU"><svg class="QMY9O" height="18px" width="18px" aria-hidden="true" focusable="false"><use xlink:href="#icon-headphones"><\/use><\/svg>View episodes<\/p>/');

+	$article_only = re_remove($article_only, '/<p class="ZybjO">A relationship banned under traditional law.<\/p>/');

+	$article_only = re_remove($article_only, '/<section class="aKWhn noPrint"><div class="dLGbc"><\/div><\/section>/');

+	$article_only = re_remove($article_only, '/<button class="_30X7v _2-NUy">(.+)<\/button>/');

+	$article_only = re_remove($article_only, '/<script type="application\/ld\+json">(.+)<\/script>/');

+	$article_only = re_remove($article_only, '/<div id="adspot-N-6x2-pos1" class="_1-uzy"><\/div>/');

+	$article_only = re_remove($article_only, '/<div id="adspot-N-300x164-pos1" class="_1-uzy"><\/div>/');

+	$article_only = re_remove($article_only, '/<div class="noPrint"><\/div>/');

+	$article_only = re_remove($article_only, '/<script src="https:\/\/ad.doubleclick.net\/(.+)<\/script>/');

+	$article_only = re_remove($article_only, '/<div class="ymInT _3uJpn noPrint"><span class="_2wzgv D5idv _3lVFK"><span class="_29Qt8"><\/span><span class="_3qqDc">Loading<\/span><\/span><\/div>/');

+	$article_only = re_remove($article_only, '/<div class="_1lC_G"><\/div>/');

+	$article_only = re_remove($article_only, '/<div class="_1H_qO"><\/div>/');

+	$article_only = re_remove($article_only, '/<p> <\/p>/');

+

+  //Some little replacements

+//	$re = '/<\/div><\/div><\/div><\/section><\/aside><section class="_1ysFk"><div class="_1665V undefined">/';

+//	$article_only = preg_replace($re, '</section></aside><section class="_1ysFk"><div class="_1665V undefined">', $article_only);

+  

+  //Finally remove empty lines

+	$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;';

+  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;';

+  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;';

+  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="'.$cpt.'">'.'<h1>'.$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>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';

+  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;';

+  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;';

+  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;';

+  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;

+  }

+}

+?>

+

+

@@ -6,6 +6,7 @@ define('NEWS_RSS_LATIMES', 'http://www.latimes.com/world/rss2.0.xml');

 define('NEWS_RSS_CNA', 'https://www.channelnewsasia.com/rssfeeds/8395986');

 define('NEWS_RSS_BBC', 'http://feeds.bbci.co.uk/news/rss.xml');

 define('NEWS_RSS_AJE', 'https://www.aljazeera.com/xml/rss/all.xml');

+define('NEWS_RSS_SMH', 'https://www.smh.com.au/rss/feed.xml');

 $array_title=array(

   "nyt" => "New York Times",

@@ -14,7 +15,8 @@ $array_title=array(

   "latimes" => "LA Times",

   "cna" => "Channel NewsAsia",

   "bbc" => "BBC Top Stories",

-  "aje" => "AlJazeera English"

+  "aje" => "AlJazeera English",

+  "smh" => "Sydney Morning Herald"

 );

 $array_url=array(

   "nyt" => NEWS_RSS_NYT,

@@ -23,6 +25,7 @@ $array_url=array(

   "latimes" => NEWS_RSS_LATIMES,

   "cna" => NEWS_RSS_CNA,

   "bbc" => NEWS_RSS_BBC,

-  "aje" => NEWS_RSS_AJE

+  "aje" => NEWS_RSS_AJE,

+  "smh" => NEWS_RSS_SMH

 );

 ?>