1 contributor
<?php
include_once( 'news-constants.php' );
$rss_content = file_get_contents(NEWS_RSS_LEMONDE);
$xml = simplexml_load_string($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>';
echo '<h4>' . $channel['title'].'<button id="html-btn" onclick="printHTMLAll(\''.$channel['title'].'\')" 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;
$orgStrings = array('?xtor=RSS-3208');
$newStrings = array('');
$article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
$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> ';
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 -->'.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 PHP_EOL.PHP_EOL.'<!-- ==================== article '.$cpt.'============== -->'.PHP_EOL;
echo "<div class=\"article\" id=\"article-$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);
$articles = $doc->getElementsByTagName('article');
$article_only="";
if( isset($articles[0]) ) {
$article_only=DOMinnerHTML($articles[0]);
} else {
$article_only = "Extraction Failed";
ERROR("article($cpt) : $article_only : ".$article['link']);
}
$figure="";
$SEARCH_SUB='<span class="ea_article">édition abonné</span>';
$pos_start=strpos($article_content, $SEARCH_SUB);
if(!$pos_start) {
//Second Test
DEBUG("article($cpt) : Non Abonne 1");
$SEARCH_SUB='<p class="article__status"><span class="icon__premium"></span>Article réservé aux abonnés</p>';
$pos_start=strpos($article_content, $SEARCH_SUB);
if(!$pos_start) {
DEBUG("article($cpt) : Non Abonne 2");
} else {
DEBUG("article($cpt) : Abonne 2");
}
}
if($pos_start) {
try {
$figures = $doc->getElementsByTagName('figure');
if( NULL === $figures[0] ) {
DEBUG("article($cpt) : No Image");
} else {
$figure = DOMinnerHTML($figures[0]);
$re = '/<img (.+?) class="illu lazy-retina" data-src="(.+?)" (.+?) data-lazyload="true" alt="(.+?)" title="(.+?)" (.+?)>/';
preg_match($re, $figure, $array);
if(count($array) >= 5 ) {
$figure = '<img width="100%" src="'.$array[2].'"><figcaption>'.$array[4].'<br> <em>('.$array[5].')</em></figcaption>';
}
}
} catch(Exception $e) {
ERROR("article($cpt) : Exception".$e->getMessage());
}
$article_abonne = str_replace("www.lemonde.fr", "abonnes.lemonde.fr", $article['link']);
$article_content = file_get_contents($article_abonne);
$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);
$articles = $doc->getElementsByTagName('article');
$article_only="";
if( isset($articles[0]) ) {
$article_only=DOMinnerHTML($articles[0]);
} else {
$article_only = "Extraction Failed for Subscribed Article";
break;
}
}
$orgStrings = array(' href="/');
$newStrings = array(' href="https://www.lemonde.fr/');
$article_only = str_replace($orgStrings, $newStrings, $article_only);
//Remove Blank lines
$temp = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
$article_only = $temp;
$temp = preg_replace('/\s\s+/', ' ', $article_only);
$article_only = $temp;
//Remove Social section
$re = '/<li class="meta__icon"> <a class="js-social"(.+)<\/span> <\/li>/s';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
//Cleanup end-of-article extraction
$re = '/<section class="article__comments">(.+)commentaires <\/a> <\/section>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<li><a class="insert__teaser" (.+)<\/a><\/li>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<p class="insert__title">Les plus lus<\/p>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<li> <a href="http(.+)Recherche<\/a> <\/li>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<div class="bd__header"> <span>Annonces automobiles<\/span> <span>avec La Centrale<\/span> <\/div>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<div class="bd__header"> <span>Annonces immobilières<\/span> <span>avec Logic-Immo<\/span> <\/div>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<div class="bd__header"> <span>Codes promos<\/span> <span>avec Global Savings Group<\/span> <\/div>/';
$article_only = preg_replace($re, '', $article_only);
$re = '/<div class="bd__footer"> <a class="bd__btn" href="https:\/\/codepromo\.lemonde\.fr\/">Toutes les offres<\/a> <\/div>/';
$article_only = preg_replace($re, '', $article_only);
$re = '/<li><a href="https:\/\/codepromo(.+?)<\/a><\/li>/';
$article_only = preg_replace($re, '', $article_only);
$re = '/<p class="article__siblings-title">Dans la même rubrique<\/p>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<footer class=" old__article-footer">(.+)<\/footer>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<a href="https:\/\/www.lemonde.fr\/services\/"><h4 class="area__title area__title--bd">Services<\/h4><\/a>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<a class="insert__action button button--dark" target="_blank" data-target="jelec-sidebar" href="https:\/\/journal.lemonde.fr\/">Lire le journal numérique<\/a>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$re = '/<section class="friend"><div (.+?)<\/div> <\/section>/';
$article_only = preg_replace($re, '', $article_only);
$re = '/<aside class="aside__article old__aside"><section class="friend"><div class=""> <div id="pave_haut" class="dfp-slot dfp__slot friend--around dfp__sticky dfp__slot--pave dfp__slot--desktop dfp-unloaded" data-format="pave_haut" aria-hidden="true"><\/div> <\/div> <\/section><section class="friend"><div id="lig_lemonde_article_teaser_side"><\/div> <\/section><\/aside>/';
$article_only = preg_replace($re, '', $article_only);
//Remove Thumbnail
$re = '/<img src="\/thumbnail\/journal\/(.+) class="insert__media" alt="">/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
$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);
$temp = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
$article_only = $temp;
$temp = preg_replace('/\s\s+/', ' ', $article_only);
$article_only = $temp;
$re = '/<a class="lien_interne rub"(.+?)>(.+?)<\/a>/';
$temp = preg_replace($re, '\\2', $article_only);
$article_only = $temp;
$re = '/<img (.+?)data-src="(.+?)"(.+?)data-lazyload="true"(.+?)>/';
$temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
$article_only = $temp;
$re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
$temp = preg_replace($re, '', $article_only);
$article_only = $temp;
echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> '.PHP_EOL;
echo '<div id="nav-source" style="display:inline;"><a href="'.$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.'">'.$figure.$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-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;
echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
$cpt++;
if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
break;
}
}
?>