1 contributor
<?php
include_once( 'news-constants.php' );
$rss_content = file_get_contents(NEWS_RSS_FRANCEINFO);
$xml = simplexml_load_string($rss_content);
if ($xml === false) {
echo 'Failed to read RSS';
} else {
$channel = array();
$channel['name'] = "franceinfo.html";
$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>' . $channel['title'].'<button id="html-btn" onclick="printHTMLAll(\''.$channel['name'].'\')" 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;
$article['link'] = re_remove($article['link'], '/#xtor=RSS-(.+)/');
$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> '.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 $article['title'].' '.PHP_EOL;
echo '<div id="nav-source" style="display:inline;"><a href="'.$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">';
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;
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']);
$article_content_utf8 = mb_convert_encoding($article_content, 'HTML-ENTITIES', "UTF-8");
$doc = new DOMDocument();
$doc->preserveWhiteSpace = false;
$doc->formatOutput = true;
$libxml_previous_state = libxml_use_internal_errors(true);
$doc->loadHTML($article_content_utf8);
libxml_clear_errors();
libxml_use_internal_errors($libxml_previous_state);
$articles = $doc->getElementsByTagName('article');
$article_only="";
foreach ( $articles as $node) {
$article_only = DOMinnerHTML($node);
}
$orgStrings = array(' href="/');
$newStrings = array(' href="https://www.francetvinfo.fr/');
$article_only = str_replace($orgStrings, $newStrings, $article_only);
$orgStrings = array(' src="/');
$newStrings = array(' src="https://www.francetvinfo.fr/');
$article_only = str_replace($orgStrings, $newStrings, $article_only);
$orgStrings = array(' data-src="/');
$newStrings = array(' data-src="https://www.francetvinfo.fr/');
$article_only = str_replace($orgStrings, $newStrings, $article_only);
$orgStrings = array(
'<p class="txt3 description-article" itemprop="description">',
'<p class="bloc_signature">',
'<span id="publisher" itemprop="Publisher" data-source="LE MONDE">',
'<div class="toolbar"></div>',
'<div id="articleBody" class="contenu_article js_article_body" itemprop="articleBody">',
'<h4>',
'</h4>',
'<h1 class="tt2" itemprop="Headline">',
'</h1>',
'<h2 class="taille_courante">',
'<h2 class="intertitre">',
'</h2>',
'<span>Le Monde</span>',
'<figure class="illustration_haut " style="width: 534px">',
'<img width="534" data-lazyload="false" src="',
' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"' );
$newStrings = array(
'<p>',
'<p>',
'<span>',
'',
'<div>',
'<h5>',
'</h5>',
'<h3>',
'</h3>',
'<h4>',
'<h4>',
'</h4>',
'Le Monde',
'<figure>',
'<img width="100%" src="',
'' );
$article_only = str_replace($orgStrings, $newStrings, $article_only);
$article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
$article_only = preg_replace('/\s\s+/', ' ', $article_only);
$re = '/<a class="lien_interne rub"(.+?)>(.+?)<\/a>/';
$article_only = preg_replace($re, '\\2', $article_only);
$re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
$article_only = preg_replace($re, '', $article_only);
$re = '/<aside class="a-lire-aussi">/';
$article_only = preg_replace($re, '<aside class="a-lire-aussi" style="display: none;">', $article_only);
$re = '/<img (.+?)data-src="(.+?)">/';
$article_only = preg_replace($re, '<img src="\\2">', $article_only);
$re = '/<iframe width="(.+?)" height="(.+?)" src="(.+?)" data-src="(.+?)" (.+?)><\/iframe>/';
$article_only = preg_replace($re, '<iframe src="\\4"></iframe>', $article_only);
$article_only = re_remove($article_only, '/<span class="avatar">(.+?)class="avatar"><\/span>/');
$article_only = re_remove($article_only, '/<a class="avatar" (.+?)class="avatar"><\/a>/');
$article_only = re_remove($article_only, '/onclick="FTVi\.countClick\(\'(.+?)\)(;"|")/');
$article_only = re_remove($article_only, '/<span class="hastag">Votre avis est précieux\.<\/span><p>Aidez-nous à améliorer notre site en répondant à notre questionnaire\.<\/p>/');
$article_only = re_remove($article_only, '/<a href="https:\/\/franceinfo\.typeform\.com\/to\/g7CQZs" target="_blank"><div class="button">je donne mon avis<\/div><\/a>/');
$article_only = re_remove($article_only, '/<div class="content-img"><\/div/');
$article_only = re_remove($article_only, '/<li><a rel="tag" href="(.+?)" >(.+?)<\/a><\/li>/');
$article_only = re_remove($article_only, '/<div class="pub-pave" id="dfp_middle"><\/div> <script>\(function\(\)(.+?)<\/script>/');
$article_only = re_remove($article_only, '/<script>document\.addEventListener\(\'FTVi::ready\'(.+?)<\/script>/');
$article_only = re_remove($article_only, '/<span class="title">Sujets associés<\/span>/');
//$article_only = re_remove($article_only, '//');
//
$article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
$artcile_only = preg_replace('/\s\s+/', ' ', $article_only);
// '<span class="avatar">'
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> '.PHP_EOL;
echo '<div id="nav-source" style="display:inline;"><a href="'.$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.'">'.$article_only.'</div>'.PHP_EOL;
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> '.PHP_EOL;
echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div> '.PHP_EOL;
echo '<div id="nav-source" style="display:inline;"><a href="'.$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;
}
}
?>