Added News parsers ・ 625fcf5 ・ Gitprep

- Added News parsers;
- Browse files
- Yanik Cawidrone commited on 2019-03-07
- 1 parent 5c3c367
  
  commit 625fcf522a79fc00eb1fc117871fd65812632294

Showing 15 changed files with 2059 additions and 0 deletions

+19

sources/create-json.php

...	...	@@ -0,0 +1,19 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$numNews = count($array_title);
	5	+$cpt = 1;
	6	+echo '['.PHP_EOL;
	7	+foreach ($array_title as $key => $value) {
	8	+ echo '{"name": "'.$value.'", "image": "'.$key.'", "html": "sources/'.$key.'.html"}';
	9	+ if($cpt<$numNews) {
	10	+ echo ','.PHP_EOL;
	11	+ } else {
	12	+ echo PHP_EOL;
	13	+ }
	14	+
	15	+ $cpt=$cpt+1;
	16	+}
	17	+echo ']'.PHP_EOL;
	18	+
	19	+?>

+15

sources/gennews.sh

View

...	...	@@ -0,0 +1,15 @@
	1	+#!/bin/bash
	2	+cd /var/www/html/news2/sources
	3	+php create-json.php > news.json 2> /dev/null
	4	+exit 0
	5	+php news-lemonde.fr.php > lemonde.html 2> /dev/null
	6	+php news-liberation.fr.php > liberation.html 2> /dev/null
	7	+php news-slate.fr.php > slatefr.html 2> /dev/null
	8	+php news-vice.fr.php > vicefr.html 2> /dev/null
	9	+php news-lesinrocks.fr.php > lesinrocks.html 2> /dev/null
	10	+php news-numerama.fr.php > numerama.html 2> /dev/null
	11	+php news-huffingtonpost.fr.php > huffingtonpostfr.html 2> /dev/null
	12	+php news-lepoint.fr.php > lepointfr.html 2> /dev/null
	13	+php news-lexpress.fr.php > lexpressfr.html 2> /dev/null
	14	+php news-franceinfo.php > franceinfo.html 2> /dev/null
	15	+php news-conspiracy.php > conspiracy.html 2> /dev/null

+94

sources/news-conspiracy.php

View

...	...	@@ -0,0 +1,94 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = http_get_contents(NEWS_RSS_CONSPIRACY);
	5	+$rss_content = str_replace("<content:encoded>","<contentEncoded>",$rss_content);
	6	+$rss_content = str_replace("</content:encoded>","</contentEncoded>",$rss_content);
	7	+$articles = array();
	8	+$xml = simplexml_load_string($rss_content);
	9	+if ($xml === false) {
	10	+ echo 'Failed to read RSS';
	11	+} else {
	12	+ $channel = array();
	13	+ $channel['title'] = $xml->channel->title;
	14	+ $channel['link'] = $xml->channel->link;
	15	+ $channel['description'] = $xml->channel->description;
	16	+ $channel['pubDate'] = $xml->channel->pubDate;
	17	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	18	+ echo '<h4>' . $channel['title'] . '</h4>';
	19	+ $cpt=0;
	20	+ foreach ($xml->channel->item as $item) {
	21	+ $article = array();
	22	+ $article['title'] = $item->title;
	23	+ $article['link'] = $item->link;
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $article['content'] = $item->contentEncoded;
	28	+ //Extract image
	29	+ $re = '/<img class="size-medium wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
	30	+ preg_match($re, $article['content'], $matches);
	31	+ if( isset($matches[2]) ) {
	32	+ $article['image'] = $matches[2];
	33	+ if ( 0 == strlen($article['image']) ) {
	34	+ $re = '/<img class="aligncenter wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
	35	+ preg_match($re, $article['content'], $matches);
	36	+ $article['image'] = $matches[2];
	37	+ }
	38	+ } else {
	39	+ $article['image'] = "";
	40	+ }
	41	+
	42	+ $articles[$cpt]=$article;
	43	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	44	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
	45	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	46	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	47	+ echo $article['title'].'  ';
	48	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	49	+ $cpt++;
	50	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	51	+ break;
	52	+ }
	53	+ }
	54	+}
	55	+echo '</div><!-- ./col-4 -->';
	56	+echo '<div class="col-6">';
	57	+echo '<a name="article-top"></a><div id="article-current"></div>';
	58	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	59	+$cpt=0;
	60	+foreach ($articles as $article ) {
	61	+ $cpt_prev=$cpt-1;
	62	+ $cpt_next=$cpt+1;
	63	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	64	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	65	+ echo "<hr>";
	66	+ echo "<a name=\"article-$cpt\">";
	67	+ $article_only = $article['content'];
	68	+
	69	+ $article_only = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	70	+ $article_only = preg_replace('/\s\s+/', ' ', $article_only);
	71	+ $article_only = preg_replace('/<div (.+?) class="wp-video">/', '<div class="wp-video">', $article_only);
	72	+ $article_only = preg_replace('/<video class="wp-video-shortcode" id="(.+?)" (.+?) preload="metadata" controls="controls">/', '<video class="wp-video-shortcode" id="\1" preload="metadata" controls="controls">', $article_only);
	73	+ $article_only = preg_replace('/<iframe (.+?) src="(.+?)" (.+?)><\/iframe>/', '<iframe src="\2"></iframe>', $article_only);
	74	+ $article_only = preg_replace('/<img (.+?) width="(.+?)" height="(.+?)" (.+?)>/', '<img \1 \4>', $article_only);
	75	+ $article_only = preg_replace('/<img (.+?) srcset="(.+?)" sizes="(.+?)" \/>/', '<img \1>', $article_only);
	76	+ $article_only = preg_replace('/<div id="(.+?)" style="width: (.+?)" (.+?)>/', '<div id="\1" \3>', $article_only);
	77	+ $article_only = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	78	+
	79	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	80	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	81	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	82	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	83	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
	84	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> ';
	85	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     ';
	86	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	87	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	88	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	89	+ $cpt++;
	90	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	91	+ break;
	92	+ }
	93	+}
	94	+?>

+47

sources/news-constants.php

View

...	...	@@ -0,0 +1,47 @@
	1	+<?php
	2	+include_once( 'rss.php' );
	3	+$NEWS_RSS_MAX_ITEMS = 10;
	4	+
	5	+function DOMinnerHTML(DOMNode $element)
	6	+{
	7	+ $innerHTML = "";
	8	+ $children = $element->childNodes;
	9	+
	10	+ foreach ($children as $child)
	11	+ {
	12	+ $innerHTML .= $element->ownerDocument->saveHTML($child);
	13	+ }
	14	+
	15	+ return $innerHTML;
	16	+}
	17	+
	18	+function http_get_contents($url, $opts = [])
	19	+{
	20	+ $ch = curl_init();
	21	+ curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
	22	+ curl_setopt($ch, CURLOPT_TIMEOUT, 10);
	23	+ curl_setopt($ch, CURLOPT_USERAGENT, "Wget 1");
	24	+ curl_setopt($ch, CURLOPT_URL, $url);
	25	+ if(is_array($opts) && $opts) {
	26	+ foreach($opts as $key => $val) {
	27	+ curl_setopt($ch, $key, $val);
	28	+ }
	29	+ }
	30	+ curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
	31	+ if(FALSE === ($retval = curl_exec($ch))) {
	32	+ error_log(curl_error($ch));
	33	+ } else {
	34	+ return $retval;
	35	+ }
	36	+}
	37	+
	38	+function re_remove($source, $regex) {
	39	+ return preg_replace($regex, '', $source);
	40	+}
	41	+
	42	+function add_https($source, $seed) {
	43	+ $orgStrings = array( "http://$seed/" );
	44	+ $newStrings = array( "https://$seed/" );
	45	+ return str_replace($orgStrings, $newStrings, $source);
	46	+}
	47	+?>

+136

sources/news-franceinfo.php

View

...	...	@@ -0,0 +1,136 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_FRANCEINFO);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->channel->title;
	11	+ $channel['link'] = $xml->channel->link;
	12	+ $channel['description'] = $xml->channel->description;
	13	+ $channel['pubDate'] = $xml->channel->pubDate;
	14	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	15	+ echo '<h4>' . $channel['title'] . '</h4>';
	16	+ $cpt=0;
	17	+ foreach ($xml->channel->item as $item) {
	18	+ $article = array();
	19	+ $article['title'] = $item->title;
	20	+ $article['link'] = $item->link;
	21	+ $orgStrings = array('?xtor=RSS-3208');
	22	+ $newStrings = array('');
	23	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $article['image'] = $item->enclosure['url'];
	28	+ $articles[$cpt] = $article;
	29	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	30	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
	31	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	33	+ echo $article['title'].'  ';
	34	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	35	+ $cpt++;
	36	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+echo '</div><!-- ./col-4 -->';
	42	+echo '<div class="col-6">';
	43	+echo '<a name="article-top"></a><div id="article-current"></div>';
	44	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	45	+$cpt=0;
	46	+foreach ($articles as $article ) {
	47	+ $cpt_prev=$cpt-1;
	48	+ $cpt_next=$cpt+1;
	49	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	50	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	51	+ echo "<hr>";
	52	+ echo "<a name=\"article-$cpt\">";
	53	+ $article_content = file_get_contents($article['link']);
	54	+ $doc = new DOMDocument();
	55	+ $doc->preserveWhiteSpace = false;
	56	+ $doc->formatOutput = true;
	57	+ $libxml_previous_state = libxml_use_internal_errors(true);
	58	+ $doc->loadHTML($article_content);
	59	+ libxml_clear_errors();
	60	+ libxml_use_internal_errors($libxml_previous_state);
	61	+ $articles = $doc->getElementsByTagName('article');
	62	+ $article_only="";
	63	+ foreach ( $articles as $node) {
	64	+ $article_only = DOMinnerHTML($node);
	65	+ }
	66	+
	67	+ $orgStrings = array(' href="/');
	68	+ $newStrings = array(' href="https://www.francetvinfo.fr/');
	69	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	70	+
	71	+ $orgStrings = array(
	72	+ '<p class="txt3 description-article" itemprop="description">',
	73	+ '<p class="bloc_signature">',
	74	+ '<span id="publisher" itemprop="Publisher" data-source="LE MONDE">',
	75	+ '<div class="toolbar"></div>',
	76	+ '<div id="articleBody" class="contenu_article js_article_body" itemprop="articleBody">',
	77	+ '<h4>',
	78	+ '</h4>',
	79	+ '<h1 class="tt2" itemprop="Headline">',
	80	+ '</h1>',
	81	+ '<h2 class="taille_courante">',
	82	+ '<h2 class="intertitre">',
	83	+ '</h2>',
	84	+ '<span>Le Monde</span>',
	85	+ '<figure class="illustration_haut " style="width: 534px">',
	86	+ '<img width="534" data-lazyload="false" src="',
	87	+ ' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"' );
	88	+ $newStrings = array(
	89	+ '<p>',
	90	+ '<p>',
	91	+ '<span>',
	92	+ '',
	93	+ '<div>',
	94	+ '<h5>',
	95	+ '</h5>',
	96	+ '<h3>',
	97	+ '</h3>',
	98	+ '<h4>',
	99	+ '<h4>',
	100	+ '</h4>',
	101	+ 'Le Monde',
	102	+ '<figure>',
	103	+ '<img width="100%" src="',
	104	+ '' );
	105	+
	106	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	107	+
	108	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	109	+ $article_only = $temp;
	110	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	111	+ $article_only = $temp;
	112	+ $re = '/<a class="lien_interne rub"(.+?)>(.+?)<\/a>/';
	113	+ $temp = preg_replace($re, '\\2', $article_only);
	114	+ $article_only = $temp;
	115	+ $re = '/<img (.+?)data-src="(.+?)"(.+?)data-lazyload="true"(.+?)>/';
	116	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	117	+ $article_only = $temp;
	118	+ $re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
	119	+ $temp = preg_replace($re, '', $article_only);
	120	+ $article_only = $temp;
	121	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	122	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	123	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	124	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	125	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
	126	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	127	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>     ';
	128	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	129	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	130	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	131	+ $cpt++;
	132	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	133	+ break;
	134	+ }
	135	+}
	136	+?>

+200

sources/news-huffingtonpost.fr.php

View

...	...	@@ -0,0 +1,200 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_HUFFINGTONPOSTFR);
	5	+$orgStrings = array( 'media:content');
	6	+$newStrings = array( 'mediacontent');
	7	+$rss_content = str_replace($orgStrings, $newStrings, $rss_content);
	8	+$xml = simplexml_load_string($rss_content);
	9	+if ($xml === false) {
	10	+ echo 'Failed to read RSS';
	11	+} else {
	12	+ $channel = array();
	13	+ $channel['title'] = $xml->channel->title;
	14	+ $channel['link'] = $xml->channel->link;
	15	+ $channel['description'] = $xml->channel->description;
	16	+ $channel['pubDate'] = $xml->channel->pubDate;
	17	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	18	+ echo '<h4>' . $channel['title'] . '</h4>';
	19	+ $cpt=0;
	20	+ foreach ($xml->channel->item as $item) {
	21	+ $article = array();
	22	+ $article['title'] = $item->title;
	23	+ $article['link'] = $item->link;
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $SEARCH_SUB1='<p><img width="1024" height="';
	28	+ $pos_start = strpos($item->description, $SEARCH_SUB1);
	29	+ $pos_start += strlen($SEARCH_SUB1);
	30	+ $SEARCH_SUB2="\"";
	31	+ $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
	32	+ $height=substr($item->description, $pos_start, $pos_stop - $pos_start);
	33	+ $article['width']=1024;
	34	+ $article['height']=$height;
	35	+ $SEARCH_SUB1='src="';
	36	+ $pos_start = strpos($item->description, $SEARCH_SUB1);
	37	+ $pos_start += strlen($SEARCH_SUB1);
	38	+ $SEARCH_SUB2="\"";
	39	+ $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
	40	+ $article['image']=$item->mediacontent['url'];
	41	+ $articles[$cpt] = $article;
	42	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">'.PHP_EOL;
	43	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>'.PHP_EOL;
	44	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  '.PHP_EOL;
	45	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	46	+ echo $article['title'].'  '.PHP_EOL;
	47	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;
	48	+ $cpt++;
	49	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	50	+ break;
	51	+ }
	52	+ }
	53	+}
	54	+echo '</div><!-- ./col-4 -->';
	55	+echo '<div class="col-6">';
	56	+echo '<a name="article-top"></a><div id="article-current"></div>';
	57	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	58	+$cpt=0;
	59	+foreach ($articles as $article ) {
	60	+ $cpt_prev=$cpt-1;
	61	+ $cpt_next=$cpt+1;
	62	+ echo PHP_EOL.PHP_EOL."<!-- ==================== article '.$cpt.'============== -->".PHP_EOL;
	63	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	64	+ echo "<hr>";
	65	+ echo "<a name=\"article-$cpt\">\n";
	66	+ $article_content = file_get_contents($article['link']);
	67	+ $doc = new DOMDocument();
	68	+ $doc->preserveWhiteSpace = false;
	69	+ $doc->formatOutput = true;
	70	+ $libxml_previous_state = libxml_use_internal_errors(true);
	71	+ $doc->loadHTML($article_content);
	72	+ libxml_clear_errors();
	73	+ libxml_use_internal_errors($libxml_previous_state);
	74	+ $articles = $doc->getElementsByTagName('article');
	75	+ foreach ( $articles as $node) {
	76	+ $article_only = DOMinnerHTML($node);
	77	+ }
	78	+
	79	+ $orgStrings = array( '<img src="',
	80	+ 'style="width: 680px"',
	81	+ '<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>',
	82	+ ' width="500" height="281" frameborder="0" ');
	83	+ $newStrings = array( '<img width="100%" src="',
	84	+ 'style="width: 100%"',
	85	+ '',
	86	+ ' width="100%" frameborder="0" ');
	87	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	88	+
	89	+ $SEARCH_SUB1='<p><strong>Lire aussi :</strong></p>';
	90	+ $pos_stop = strpos($article_only, $SEARCH_SUB1);
	91	+ if($pos_stop) {
	92	+ $temp = substr($article_only, 0, $pos_stop);
	93	+ $article_only = $temp."</blockquote>";
	94	+ } else {
	95	+ //End of article cleanup
	96	+ $SEARCHSUB='<h3 class="related-entries__title">LIRE AUSSI</h3>';
	97	+ $pos_stop=false;
	98	+ $pos_stop = strpos($article_only, $SEARCHSUB);
	99	+ if($pos_stop) {
	100	+ $temp = substr($article_only, 0, $pos_stop);
	101	+ $article_only = $temp;
	102	+ } else {
	103	+ $SEARCHSUB='<h3 class="recirc__title trending">';
	104	+ $pos_stop=false;
	105	+ $pos_stop = strpos($article_only, $SEARCHSUB);
	106	+ if($pos_stop) {
	107	+ $temp = substr($article_only, 0, $pos_stop);
	108	+ $article_only = $temp;
	109	+ } else {
	110	+ DEBUG("Article($cpt) : No Match");
	111	+ }
	112	+ }
	113	+ }
	114	+
	115	+ $re = '/alt="(.+?)" width="(.+?)" height="(.+?)"/';
	116	+ $temp = preg_replace($re, 'width="100%"', $article_only);
	117	+ $article_only = $temp;
	118	+ $re = '/alt="" width="(.+?)" height="(.+?)"/';
	119	+ $temp = preg_replace($re, 'width="100%"', $article_only);
	120	+ $article_only = $temp;
	121	+ $re = '/<iframe width="500" height="(.+?)"/';
	122	+ $temp = preg_replace($re, '<iframe width="100%"', $article_only);
	123	+ $article_only = $temp;
	124	+ $re = '/sizes="$max-width: (.+?)$ 100vw, (.+?)"/';
	125	+ $temp = preg_replace($re, 'sizes="(max-width: 100%) 100vw,100%"', $article_only);
	126	+ $article_only = $temp;
	127	+ $re = '/<figure id="attachment_(.+?)" style="width: (.+?)" class="wp-caption aligncenter">/';
	128	+ $temp = preg_replace($re, '<figure id="attachment_\\1" style="width: 100%" class="wp-caption aligncenter">', $article_only);
	129	+ $article_only = $temp;
	130	+ $re = '/<img class="aligncenter size-full wp-image-(.+?)"/';
	131	+ $temp = preg_replace($re, '<img ', $article_only);
	132	+ $article_only = $temp;
	133	+ $re = '/<iframe (.+?) height="(.+?)" (.+?) width="(.+?)">/';
	134	+ $temp = preg_replace($re, '<iframe \\1 \\3 width="100%">', $article_only);
	135	+ $article_only = $temp;
	136	+ $re = '/<img class="image__src"/';
	137	+ $temp = preg_replace($re, '<img class="image__src" width="100%"', $article_only);
	138	+ $article_only = $temp;
	139	+
	140	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	141	+ $article_only = $temp;
	142	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	143	+ $article_only = $temp;
	144	+ $orgStrings = array(
	145	+ '<div id="entry_paragraph_1" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>',
	146	+ '<div id="entry_paragraph_2" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>',
	147	+ '<div id="entry_paragraph_3" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>',
	148	+ '<div id="entry_paragraph_4" class="entry-body--paragraph-ad"> <div class="ad_spot "> </div> </div>');
	149	+ $newStrings = array(
	150	+ '',
	151	+ '',
	152	+ '',
	153	+ '');
	154	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	155	+
	156	+ // Cleanup "Social"
	157	+ $re='/<script id="share-bar__social-mob" (.+)<\/script>/';
	158	+ $article_only = preg_replace($re, '', $article_only);
	159	+ $re='/<li class="(facebook\|twitter\|gplus\|pinterest\|linkedin\|email) share-bar__item">(.+)<\/a><\/li>/';
	160	+ $article_only = preg_replace($re, '', $article_only);
	161	+ $re='/<li class="(facebook\|twitter\|snapchat\|instagram) follow-us-networks-item " (.+?)<\/span> <\/a> <\/li>/';
	162	+ $article_only = preg_replace($re, '', $article_only);
	163	+ $re='/<div class="social-overlay__qr">(.+)<\/a><\/div>/';
	164	+ $article_only = preg_replace($re, '', $article_only);
	165	+ $re='/<script type="text\/javascript"> $function\($\{ var dfp_placement_slug(.+)<\/script>/';
	166	+ $article_only = preg_replace($re, '', $article_only);
	167	+ $re='/<form class="newsletter-signup" method="POST">(.+)<\/form>/';
	168	+ $article_only = preg_replace($re, '', $article_only);
	169	+ $re='/<div class="follow-us__header"> <h3>SUIVEZ-NOUS<\/h3> <\/div>/';
	170	+ $article_only = preg_replace($re, '', $article_only);
	171	+ $re='/<div class="follow-us__description"> <p>Avec la newsletter quotidienne du HuffPost(.+) target="_blank">En savoir plus<\/a> <\/p> <\/div>/';
	172	+ $article_only = preg_replace($re, '', $article_only);
	173	+ $re='/<li class="(facebook\|instagram) follow-us-networks-item even" (.+) <\/a> <\/li>/';
	174	+ $article_only = preg_replace($re, '', $article_only);
	175	+ //Clean Empty lines
	176	+ $article_only = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	177	+ $article_only = preg_replace('/\s\s+/', ' ', $article_only);
	178	+
	179	+
	180	+
	181	+
	182	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	183	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	184	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	185	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
	186	+ echo '<div class="extract-content" id="'.$cpt.'">';
	187	+ echo '<h4>'.$article['title'].'</h4>';
	188	+ echo $article_only.'</div>'.PHP_EOL;
	189	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	190	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>     '.PHP_EOL;
	191	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	192	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	193	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
	194	+ echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
	195	+ $cpt++;
	196	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	197	+ break;
	198	+ }
	199	+}
	200	+?>

+249

sources/news-lemonde.fr.php

View

...	...	@@ -0,0 +1,249 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_LEMONDE);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->channel->title;
	11	+ $channel['link'] = $xml->channel->link;
	12	+ $channel['description'] = $xml->channel->description;
	13	+ $channel['pubDate'] = $xml->channel->pubDate;
	14	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	15	+ echo '<h4>' . $channel['title'] . '</h4>';
	16	+ $cpt=0;
	17	+ foreach ($xml->channel->item as $item) {
	18	+ $article = array();
	19	+ $article['title'] = $item->title;
	20	+ $article['link'] = $item->link;
	21	+ $orgStrings = array('?xtor=RSS-3208');
	22	+ $newStrings = array('');
	23	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $article['image'] = $item->enclosure['url'];
	28	+ $articles[$cpt] = $article;
	29	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">'.PHP_EOL;
	30	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>'.PHP_EOL;
	31	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  '.PHP_EOL;
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	33	+ echo $article['title'].'  '.PHP_EOL;
	34	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;
	35	+ $cpt++;
	36	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+echo '</div><!-- ./col-4 -->'.PHP_EOL;
	42	+echo '<div class="col-6">'.PHP_EOL;
	43	+echo '<a name="article-top"></a><div id="article-current"></div>'.PHP_EOL;
	44	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	45	+
	46	+$cpt=0;
	47	+foreach ($articles as $article ) {
	48	+ $cpt_prev=$cpt-1;
	49	+ $cpt_next=$cpt+1;
	50	+ echo PHP_EOL.PHP_EOL.'<!-- ==================== article '.$cpt.'============== -->'.PHP_EOL;
	51	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	52	+ echo "<hr>";
	53	+ echo "<a name=\"article-$cpt\">";
	54	+ $article_content = file_get_contents($article['link']);
	55	+ $doc = new DOMDocument();
	56	+ $doc->preserveWhiteSpace = false;
	57	+ $doc->formatOutput = true;
	58	+ $libxml_previous_state = libxml_use_internal_errors(true);
	59	+ $doc->loadHTML($article_content);
	60	+ libxml_clear_errors();
	61	+ libxml_use_internal_errors($libxml_previous_state);
	62	+ $articles = $doc->getElementsByTagName('article');
	63	+ $article_only="";
	64	+ if( isset($articles[0]) ) {
	65	+ $article_only=DOMinnerHTML($articles[0]);
	66	+ } else {
	67	+ $article_only = "Extraction Failed";
	68	+ break;
	69	+ }
	70	+ $figure="";
	71	+
	72	+ $SEARCH_SUB='<span class="ea_article">édition abonné</span>';
	73	+ $pos_start=strpos($article_content, $SEARCH_SUB);
	74	+ if(!$pos_start) {
	75	+ //Second Test
	76	+ DEBUG("article($cpt) : Non Abonne 1");
	77	+ $SEARCH_SUB='<p class="article__status"><span class="icon__premium"></span>Article réservé aux abonnés</p>';
	78	+ $pos_start=strpos($article_content, $SEARCH_SUB);
	79	+ if(!$pos_start) {
	80	+ DEBUG("article($cpt) : Non Abonne 2");
	81	+ } else {
	82	+ DEBUG("article($cpt) : Abonne 2");
	83	+ }
	84	+ }
	85	+ if($pos_start) {
	86	+ try {
	87	+ $figures = $doc->getElementsByTagName('figure');
	88	+ if( NULL === $figures[0] ) {
	89	+ DEBUG("article($cpt) : No Image");
	90	+ } else {
	91	+ $figure = DOMinnerHTML($figures[0]);
	92	+ $re = '/<img (.+?) class="illu lazy-retina" data-src="(.+?)" (.+?) data-lazyload="true" alt="(.+?)" title="(.+?)" (.+?)>/';
	93	+ preg_match($re, $figure, $array);
	94	+ if(count($array) >= 5 ) {
	95	+ $figure = '<img width="100%" src="'.$array[2].'"><figcaption>'.$array[4].'<br> <em>('.$array[5].')</em></figcaption>';
	96	+ }
	97	+ }
	98	+ } catch(Exception $e) {
	99	+ ERROR("article($cpt) : Exception".$e->getMessage());
	100	+ }
	101	+
	102	+ $article_abonne = str_replace("www.lemonde.fr", "abonnes.lemonde.fr", $article['link']);
	103	+ $article_content = file_get_contents($article_abonne);
	104	+ $doc = new DOMDocument();
	105	+ $doc->preserveWhiteSpace = false;
	106	+ $doc->formatOutput = true;
	107	+ $libxml_previous_state = libxml_use_internal_errors(true);
	108	+ $doc->loadHTML($article_content);
	109	+ libxml_clear_errors();
	110	+ libxml_use_internal_errors($libxml_previous_state);
	111	+ $articles = $doc->getElementsByTagName('article');
	112	+ $article_only="";
	113	+ if( isset($articles[0]) ) {
	114	+ $article_only=DOMinnerHTML($articles[0]);
	115	+ } else {
	116	+ $article_only = "Extraction Failed for Subscribed Article";
	117	+ break;
	118	+ }
	119	+ }
	120	+
	121	+ $orgStrings = array(' href="/');
	122	+ $newStrings = array(' href="https://www.lemonde.fr/');
	123	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	124	+
	125	+ //Remove Blank lines
	126	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	127	+ $article_only = $temp;
	128	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	129	+ $article_only = $temp;
	130	+
	131	+ //Remove Social section
	132	+ $re = '/<li class="meta__icon"> <a class="js-social"(.+)<\/span> <\/li>/s';
	133	+ $temp = preg_replace($re, '', $article_only);
	134	+ $article_only = $temp;
	135	+
	136	+ //Cleanup end-of-article extraction
	137	+ $re = '/<section class="article__comments">(.+)commentaires <\/a> <\/section>/';
	138	+ $temp = preg_replace($re, '', $article_only);
	139	+ $article_only = $temp;
	140	+ $re = '/<li><a class="insert__teaser" (.+)<\/a><\/li>/';
	141	+ $temp = preg_replace($re, '', $article_only);
	142	+ $article_only = $temp;
	143	+ $re = '/<p class="insert__title">Les plus lus<\/p>/';
	144	+ $temp = preg_replace($re, '', $article_only);
	145	+ $article_only = $temp;
	146	+ $re = '/<li> <a href="http(.+)Recherche<\/a> <\/li>/';
	147	+ $temp = preg_replace($re, '', $article_only);
	148	+ $article_only = $temp;
	149	+ $re = '/<div class="bd__header"> <span>Annonces automobiles<\/span> <span>avec La Centrale<\/span> <\/div>/';
	150	+ $temp = preg_replace($re, '', $article_only);
	151	+ $article_only = $temp;
	152	+ $re = '/<div class="bd__header"> <span>Annonces immobilières<\/span> <span>avec Logic-Immo<\/span> <\/div>/';
	153	+ $temp = preg_replace($re, '', $article_only);
	154	+ $article_only = $temp;
	155	+ $re = '/<div class="bd__header"> <span>Codes promos<\/span> <span>avec Global Savings Group<\/span> <\/div>/';
	156	+ $article_only = preg_replace($re, '', $article_only);
	157	+ $re = '/<div class="bd__footer"> <a class="bd__btn" href="https:\/\/codepromo\.lemonde\.fr\/">Toutes les offres<\/a> <\/div>/';
	158	+ $article_only = preg_replace($re, '', $article_only);
	159	+ $re = '/<li><a href="https:\/\/codepromo(.+?)<\/a><\/li>/';
	160	+ $article_only = preg_replace($re, '', $article_only);
	161	+ $re = '/<p class="article__siblings-title">Dans la même rubrique<\/p>/';
	162	+ $temp = preg_replace($re, '', $article_only);
	163	+ $article_only = $temp;
	164	+ $re = '/<footer class=" old__article-footer">(.+)<\/footer>/';
	165	+ $temp = preg_replace($re, '', $article_only);
	166	+ $article_only = $temp;
	167	+ $re = '/<a href="https:\/\/www.lemonde.fr\/services\/"><h4 class="area__title area__title--bd">Services<\/h4><\/a>/';
	168	+ $temp = preg_replace($re, '', $article_only);
	169	+ $article_only = $temp;
	170	+ $re = '/<a class="insert__action button button--dark" target="_blank" data-target="jelec-sidebar" href="https:\/\/journal.lemonde.fr\/">Lire le journal numérique<\/a>/';
	171	+ $temp = preg_replace($re, '', $article_only);
	172	+ $article_only = $temp;
	173	+ $re = '/<section class="friend"><div (.+?)<\/div> <\/section>/';
	174	+ $article_only = preg_replace($re, '', $article_only);
	175	+ $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>/';
	176	+ $article_only = preg_replace($re, '', $article_only);
	177	+
	178	+ //Remove Thumbnail
	179	+ $re = '/<img src="\/thumbnail\/journal\/(.+) class="insert__media" alt="">/';
	180	+ $temp = preg_replace($re, '', $article_only);
	181	+ $article_only = $temp;
	182	+
	183	+ $orgStrings = array(
	184	+ '<p class="txt3 description-article" itemprop="description">',
	185	+ '<p class="bloc_signature">',
	186	+ '<span id="publisher" itemprop="Publisher" data-source="LE MONDE">',
	187	+ '<div class="toolbar"></div>',
	188	+ '<div id="articleBody" class="contenu_article js_article_body" itemprop="articleBody">',
	189	+ '<h4>',
	190	+ '</h4>',
	191	+ '<h1 class="tt2" itemprop="Headline">',
	192	+ '</h1>',
	193	+ '<h2 class="taille_courante">',
	194	+ '<h2 class="intertitre">',
	195	+ '</h2>',
	196	+ '<span>Le Monde</span>',
	197	+ '<figure class="illustration_haut " style="width: 534px">',
	198	+ '<img width="534" data-lazyload="false" src="',
	199	+ ' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"' );
	200	+ $newStrings = array(
	201	+ '<p>',
	202	+ '<p>',
	203	+ '<span>',
	204	+ '',
	205	+ '<div>',
	206	+ '<h5>',
	207	+ '</h5>',
	208	+ '<h3>',
	209	+ '</h3>',
	210	+ '<h4>',
	211	+ '<h4>',
	212	+ '</h4>',
	213	+ 'Le Monde',
	214	+ '<figure>',
	215	+ '<img width="100%" src="',
	216	+ '' );
	217	+
	218	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	219	+
	220	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	221	+ $article_only = $temp;
	222	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	223	+ $article_only = $temp;
	224	+ $re = '/<a class="lien_interne rub"(.+?)>(.+?)<\/a>/';
	225	+ $temp = preg_replace($re, '\\2', $article_only);
	226	+ $article_only = $temp;
	227	+ $re = '/<img (.+?)data-src="(.+?)"(.+?)data-lazyload="true"(.+?)>/';
	228	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	229	+ $article_only = $temp;
	230	+ $re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
	231	+ $temp = preg_replace($re, '', $article_only);
	232	+ $article_only = $temp;
	233	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	234	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	235	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	236	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
	237	+ echo '<div class="extract-content" id="'.$cpt.'">'.$figure.$article_only.'</div>'.PHP_EOL;
	238	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	239	+ 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;
	240	+ 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;
	241	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	242	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
	243	+ echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
	244	+ $cpt++;
	245	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	246	+ break;
	247	+ }
	248	+}
	249	+?>

+170

sources/news-lepoint.fr.php

View

...	...	@@ -0,0 +1,170 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = http_get_contents(NEWS_RSS_LEPOINTFR);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+ exit(0);
	9	+} else {
	10	+ $channel = array();
	11	+ $channel['title'] = $xml->channel->title;
	12	+ $channel['link'] = $xml->channel->link;
	13	+ $channel['description'] = $xml->channel->description;
	14	+ $channel['pubDate'] = $xml->channel->pubDate;
	15	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	16	+ echo '<h4>' . $channel['title'] . '</h4>';
	17	+ $cpt=0;
	18	+ foreach ($xml->channel->item as $item) {
	19	+ $article = array();
	20	+ $article['title'] = $item->title;
	21	+ $article['link'] = $item->link;
	22	+ $orgStrings = array('#xtor=RSS-221');
	23	+ $newStrings = array('');
	24	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	25	+ $article['pubDate'] = $item->pubDate;
	26	+ $article['timestamp'] = strtotime($item->pubDate);
	27	+ $article['description'] = $item->description;
	28	+ $article['image'] = $item->enclosure['url'];
	29	+ $articles[$cpt] = $article;
	30	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	31	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	33	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	34	+ echo $article['title'].'  ';
	35	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	36	+ $cpt++;
	37	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	38	+ break;
	39	+ }
	40	+ }
	41	+}
	42	+echo '</div><!-- ./col-4 -->';
	43	+echo '<div class="col-6">';
	44	+echo '<a name="article-top"></a><div id="article-current"></div>';
	45	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	46	+$cpt=0;
	47	+foreach ($articles as $article ) {
	48	+ $cpt_prev=$cpt-1;
	49	+ $cpt_next=$cpt+1;
	50	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	51	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	52	+ echo "<hr>";
	53	+ echo "<a name=\"article-$cpt\">";
	54	+ $article_content = http_get_contents($article['link']);
	55	+
	56	+ $SEARCH_SUB1='<h1 class="art-titre list-view">';
	57	+ $pos_start = strpos($article_content, $SEARCH_SUB1);
	58	+ $pos_stop = false;
	59	+ if(!$pos_start) {
	60	+ $SEARCH_SUB1='<h2 class="art-chapeau">';
	61	+ $pos_start = strpos($article_content, $SEARCH_SUB1);
	62	+ }
	63	+ if($pos_start) {
	64	+ $SEARCH_SUB1='<h3 class="header-meme-sujet">';
	65	+ $pos_stop = strpos($article_content, $SEARCH_SUB1);
	66	+ if(!$pos_stop) {
	67	+ $SEARCH_SUB1='<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>';
	68	+ $pos_stop = strpos($article_content, $SEARCH_SUB1);
	69	+ if(!$pos_stop) {
	70	+ $SEARCH_SUB1='</article>';
	71	+ $pos_stop = strpos($article_content, $SEARCH_SUB1);
	72	+ }
	73	+ }
	74	+ }
	75	+
	76	+ if( $pos_start && $pos_stop ) {
	77	+ $article_only = substr($article_content, $pos_start, $pos_stop - $pos_start);
	78	+ } else {
	79	+ $article_only="";
	80	+ }
	81	+
	82	+ $orgStrings = array( ' href="/',
	83	+ ' src="/sites/',
	84	+ '<img src=',
	85	+ ' data-src="/');
	86	+ $newStrings = array( ' href="https://www.lepoint.fr/',
	87	+ ' src="https://www.lepoint.fr/sites/',
	88	+ '<img width="100%" src=',
	89	+ ' data-src="https://www.lepoint.fr/' );
	90	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	91	+
	92	+ $orgStrings = array( '<div class="row">',
	93	+ '<div class="col-md-2">',
	94	+ '<div class="col-md-8">',
	95	+ '<div class="container width_wrap">',
	96	+ '<div class="col-left">',
	97	+ '<h3> </h3>',
	98	+ '<h1>',
	99	+ '</h1>',
	100	+ '<p style="text-align:center">',
	101	+ '<div class="image-wrap panoramique bg-undefined">',
	102	+ '<div class="placement-left">' );
	103	+ $newStrings = array('<div>',
	104	+ '<div>',
	105	+ '<div>',
	106	+ '<div>',
	107	+ '<div>',
	108	+ '',
	109	+ '<h4>',
	110	+ '</h4>',
	111	+ '<p>',
	112	+ '<div>',
	113	+ '<div>');
	114	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	115	+
	116	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	117	+ $article_only = $temp;
	118	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	119	+ $article_only = $temp;
	120	+ $re = '/<a class="cboxElement" (.+?)><img (.+?)><\/a>/';
	121	+ $temp = preg_replace($re, '<img \\2>', $article_only);
	122	+ $article_only = $temp;
	123	+ $re = '/<noscript>(.+?)<\/noscript>/ms';
	124	+ $temp = preg_replace($re, '', $article_only);
	125	+ $article_only = $temp;
	126	+
	127	+ $orgStrings = array( 'http://www.lepoint.fr/' );
	128	+ $newStrings = array( 'https://www.lepoint.fr/' );
	129	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	130	+
	131	+ $re = '/<img class="w100" alt="(.+?)" data-lazyload="loading" onclick="(.+?)" src="(.+?)" title="(.+?)" data-src="(.+?)" \/>/';
	132	+ $temp = preg_replace($re, '<img width="100%" src="\\5">', $article_only);
	133	+ $article_only = $temp;
	134	+
	135	+ $re = '/<img class="" src data-lazyload="loading" data-src="(.+?)" data-alt="(.+?)" width="(.+?)" height="(.+?)" alt title="(.+?)"\/>/ms';
	136	+ $temp = preg_replace($re, '<img width="100%" src="\\1">', $article_only);
	137	+ $article_only = $temp;
	138	+
	139	+ $re = '/<img src data-lazyload="loading" data-src="(.+?)" data-alt="(.+?)" alt="(.+?)" width="(.+?)" height="(.+?)" \/>/';
	140	+ $temp = preg_replace($re, '<img width="100%" src="\\1">', $article_only);
	141	+ $article_only = $temp;
	142	+
	143	+ $re = '/<img class="w100" alt="" data-lazyload="loading" onclick="(.+?)" src="(.+?)" title="" data-src="(.+?)" \/>/';
	144	+ $temp = preg_replace($re, '<img width="100%" src="\\3">', $article_only);
	145	+ $article_only = $temp;
	146	+
	147	+ $re = '/<a href="(.+?)" class="button-type pat little-button-mob" onclick="(.+?)">Abonnez-vous<\/a>/';
	148	+ $temp = preg_replace($re, '', $article_only);
	149	+ $article_only = $temp;
	150	+
	151	+ $re = '/<a onclick="(.+?)" data-shadowbox="#authform" href="(.+?)" id="connect_viafreemium" class="w100 block txtcenter br-4-all bg-grey-medium-blue pas font-black bold br-4-all upper">Identifiez-vous<\/a>/';
	152	+ $temp = preg_replace($re, '', $article_only);
	153	+ $article_only = $temp;
	154	+
	155	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	156	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	157	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	158	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	159	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
	160	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	161	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     ';
	162	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	163	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	164	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	165	+ $cpt++;
	166	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	167	+ break;
	168	+ }
	169	+}
	170	+?>

+165

sources/news-lesinrocks.fr.php

View

...	...	@@ -0,0 +1,165 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_LESINROCKS);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->channel->title;
	11	+ $channel['link'] = $xml->channel->link;
	12	+ $channel['description'] = $xml->channel->description;
	13	+ $channel['pubDate'] = $xml->channel->pubDate;
	14	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	15	+ echo '<h4>' . $channel['title'] . '</h4>';
	16	+ $cpt=0;
	17	+ foreach ($xml->channel->item as $item) {
	18	+ $article = array();
	19	+ $article['title'] = $item->title;
	20	+ $article['link'] = $item->link;
	21	+ $orgStrings = array('?xtor=RSS-3208');
	22	+ $newStrings = array('');
	23	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $article['image'] = $item->enclosure['url'];
	28	+ $articles[$cpt] = $article;
	29	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	30	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
	31	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	33	+ echo $article['title'].'  ';
	34	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	35	+ $cpt++;
	36	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+echo '</div><!-- ./col-4 -->';
	42	+echo '<div class="col-6">';
	43	+echo '<a name="article-top"></a><div id="article-current"></div>';
	44	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	45	+$cpt=0;
	46	+foreach ($articles as $article ) {
	47	+ $cpt_prev=$cpt-1;
	48	+ $cpt_next=$cpt+1;
	49	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	50	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	51	+ echo "<hr>";
	52	+ echo "<a name=\"article-$cpt\">";
	53	+ $article_content = file_get_contents($article['link']);
	54	+ $article_content_utf8 = mb_convert_encoding($article_content, 'HTML-ENTITIES', "UTF-8");
	55	+ $doc = new DOMDocument();
	56	+ $doc->preserveWhiteSpace = false;
	57	+ $doc->formatOutput = true;
	58	+ $libxml_previous_state = libxml_use_internal_errors(true);
	59	+ $doc->loadHTML($article_content_utf8);
	60	+ libxml_clear_errors();
	61	+ libxml_use_internal_errors($libxml_previous_state);
	62	+ $articles = $doc->getElementsByTagName('article');
	63	+ $article_only="";
	64	+ foreach ( $articles as $node) {
	65	+ $article_only = DOMinnerHTML($node);
	66	+ }
	67	+
	68	+ $orgStrings = array( ' href="/',
	69	+ '<img src=');
	70	+ $newStrings = array( ' href="http://www.lesinrocks.com/',
	71	+ '<img width="100%" src=' );
	72	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	73	+
	74	+ $orgStrings = array( '<div class="row">',
	75	+ '<div class="col-md-2">',
	76	+ '<div class="col-md-8">',
	77	+ '<div class="container width_wrap">',
	78	+ '<div class="col-left">',
	79	+ '<h3> </h3>',
	80	+ '<span>PAR</span>',
	81	+ '<span>Par</span>',
	82	+ '<p></p>',
	83	+ '<div style="padding-top: 56.25% ">',
	84	+ '<h1>',
	85	+ '</h1>',
	86	+ '<img width="100%" src="/img/auteur.png'
	87	+ /*
	88	+ '<div style="padding-top: 73.170731707317% ">'*/ );
	89	+ $newStrings = array('<div>',
	90	+ '<div>',
	91	+ '<div>',
	92	+ '<div>',
	93	+ '<div>',
	94	+ '',
	95	+ 'par ',
	96	+ 'par ',
	97	+ '',
	98	+ '<div>',
	99	+ '<h4>',
	100	+ '</h4>',
	101	+ '<img width="5%" src="img/lesinrocks.png'
	102	+ /*
	103	+ '<div>'*/);
	104	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	105	+
	106	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	107	+ $article_only = $temp;
	108	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	109	+ $article_only = $temp;
	110	+ $re='/<img class="alignnone size-full(.+?)" src="(.+?)"(.+?)>/';
	111	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	112	+ $article_only = $temp;
	113	+ $re='/<img class="aligncenter size-full(.+?)" src="(.+?)"(.+?)>/';
	114	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	115	+ $article_only = $temp;
	116	+ $re='/<div class="OUTBRAIN"(.+?)data-ob-template="lesinrocks"><\/div>/';
	117	+ $temp = preg_replace($re, '', $article_only);
	118	+ $article_only = $temp;
	119	+ $re='/<div style="padding-top:(.+?)">/';
	120	+ $temp = preg_replace($re, '<div>', $article_only);
	121	+ $article_only = $temp;
	122	+ $re='/<iframe width="(.+?)" height="(.+?)" /';
	123	+ $temp = preg_replace($re, '<iframe width="100%" ', $article_only);
	124	+ $article_only = $temp;
	125	+ $re='/ width="(.+?)" height="(.+?)" frameborder="0" /';
	126	+ $temp = preg_replace($re, ' width="100%" frameborder="0" ', $article_only);
	127	+ $article_only = $temp;
	128	+ $SEARCH='Prolongez votre lecture !';
	129	+ $posend = strpos($article_only,$SEARCH);
	130	+ if($posend) {
	131	+ $temp = substr( $article_only, 0, $posend );
	132	+ $article_only = $temp;
	133	+ }
	134	+ $SEARCH='<div class="native-ad-in-content">';
	135	+ $posend = strpos($article_only,$SEARCH);
	136	+ if($posend) {
	137	+ $temp = substr( $article_only, 0, $posend );
	138	+ $article_only = $temp;
	139	+ }
	140	+ $re='/<a href="(.+?)" class="inner_(.+?)"> <svg(.+?)><\/use><\/svg><\/a>/';
	141	+ $temp = preg_replace($re, '', $article_only);
	142	+ $article_only = $temp;
	143	+ $orgStrings = array( '<div class="author"> <img width="100%"',
	144	+ 'par <br>',
	145	+ '<div class="single-cta"> <a href="http://abonnement.lesinrocks.com/" class="cta-header cta-abonnement" onclick="return trackClick(this, \'abonnement_article_avatar\');">Abonnez-vous<br>à partir de <span>1€</span></a> </div>' );
	146	+ $newStrings = array( '<div class="author"> <img width="5%"',
	147	+ 'par ',
	148	+ '' );
	149	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	150	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	151	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	152	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	153	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	154	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
	155	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	156	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     ';
	157	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	158	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	159	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	160	+ $cpt++;
	161	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	162	+ break;
	163	+ }
	164	+}
	165	+?>

+123

sources/news-lexpress.fr.php

View

...	...	@@ -0,0 +1,123 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = http_get_contents(NEWS_RSS_LEXPRESSFR);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->channel->title;
	11	+ $channel['link'] = $xml->channel->link;
	12	+ $channel['description'] = $xml->channel->description;
	13	+ $channel['pubDate'] = $xml->channel->pubDate;
	14	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	15	+ echo '<h4>' . $channel['title'] . '</h4>';
	16	+ $cpt=0;
	17	+ foreach ($xml->channel->item as $item) {
	18	+ $article = array();
	19	+ $article['title'] = $item->title;
	20	+ $article['link'] = $item->link;
	21	+ $orgStrings = array('#xtor=RSS-221');
	22	+ $newStrings = array('');
	23	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $orgStrings = array( 'w_200,h_134,' );
	28	+ $newStrings = array( 'w_2000,h_1125,' );
	29	+ $article['image'] = str_replace($orgStrings, $newStrings, $item->enclosure['url']);
	30	+
	31	+ $articles[$cpt] = $article;
	32	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	33	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
	34	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	35	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	36	+ echo $article['title'].'  ';
	37	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	38	+ $cpt++;
	39	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	40	+ break;
	41	+ }
	42	+ }
	43	+}
	44	+echo '</div><!-- ./col-4 -->';
	45	+echo '<div class="col-6">';
	46	+echo '<a name="article-top"></a><div id="article-current"></div>';
	47	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	48	+$cpt=0;
	49	+foreach ($articles as $article ) {
	50	+ $cpt_prev=$cpt-1;
	51	+ $cpt_next=$cpt+1;
	52	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	53	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	54	+ echo "<hr>";
	55	+ echo "<a name=\"article-$cpt\">";
	56	+ $article_content = http_get_contents($article['link']);
	57	+ $doc = new DOMDocument();
	58	+ $doc->preserveWhiteSpace = false;
	59	+ $doc->formatOutput = true;
	60	+ $libxml_previous_state = libxml_use_internal_errors(true);
	61	+ $doc->loadHTML($article_content);
	62	+ libxml_clear_errors();
	63	+ libxml_use_internal_errors($libxml_previous_state);
	64	+ $articles = $doc->getElementsByTagName('div');
	65	+ $article_only="";
	66	+ foreach ( $articles as $node) {
	67	+ if(0==strcmp("article_container",$node->getAttribute('class'))) {
	68	+ $article_only = DOMinnerHTML($node);
	69	+ }
	70	+ }
	71	+
	72	+ $orgStrings = array( ' href="/',
	73	+ ' src="/sites/',
	74	+ '<img src=',
	75	+ ' data-src="/');
	76	+ $newStrings = array( ' href="https://www.lexpress.fr/',
	77	+ ' src="https://www.lexpress.fr/sites/',
	78	+ '<img width="100%" src=',
	79	+ ' data-src="https://www.lexpress.fr/' );
	80	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	81	+
	82	+ $orgStrings = array( '<div class="row">',
	83	+ '<div class="col-md-2">',
	84	+ '<div class="col-md-8">',
	85	+ '<div class="container width_wrap">',
	86	+ '<div class="col-left">',
	87	+ '<h3> </h3>',
	88	+ '<h1>',
	89	+ '</h1>',
	90	+ '<p style="text-align:center">',
	91	+ '<div class="image-wrap panoramique bg-undefined">',
	92	+ '<div class="placement-left">' );
	93	+ $newStrings = array('<div>',
	94	+ '<div>',
	95	+ '<div>',
	96	+ '<div>',
	97	+ '<div>',
	98	+ '',
	99	+ '<h4>',
	100	+ '</h4>',
	101	+ '<p>',
	102	+ '<div>',
	103	+ '<div>');
	104	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	105	+ $orgStrings = array( 'http://www.lexpress.fr/' );
	106	+ $newStrings = array( 'https://www.lexpress.fr/' );
	107	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	108	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	109	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	110	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	111	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	112	+ echo '<div class="extract-content" id="'.$cpt.'"><h3>'.$article['title'].'</h3><img width="100%" src="'.$article['image'].'">'.$article_only.'</div>';
	113	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	114	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     ';
	115	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	116	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	117	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	118	+ $cpt++;
	119	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	120	+ break;
	121	+ }
	122	+}
	123	+?>

+269

sources/news-liberation.fr.php

View

...	...	@@ -0,0 +1,269 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_LIBERATION);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->title;
	11	+ $channel['pubDate'] = $xml->updated;
	12	+ $channel['timestamp'] = strtotime($xml->updated);
	13	+ echo '<h4>' . $channel['title'] . '</h4>';
	14	+ $cpt=0;
	15	+ foreach ($xml->entry as $item)
	16	+ {
	17	+ $article = array();
	18	+ $article['title'] = $item->title;
	19	+ $article['link'] = $item->link['href'];
	20	+ $orgStrings = array('?xtor=rss-450');
	21	+ $newStrings = array('');
	22	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	23	+ $article['pubDate'] = $item->updated;
	24	+ $article['timestamp'] = strtotime($item->updated);
	25	+ $article['description'] = $item->summary;
	26	+ $article['image']="img/liberation.png";
	27	+ foreach ($item->link as $item2 )
	28	+ {
	29	+ if( ( 0 == strcmp( $item2['type'], "image/jpeg") ) \|\|
	30	+ ( 0 == strcmp( $item2['type'], "image/png") ) ) {
	31	+ $pos_start=strpos($item2['href'],"?modified_at");
	32	+ if( $pos_start ) {
	33	+ $image_link=substr($item2['href'],0,$pos_start);
	34	+ $article['image']=substr($item2['href'],0,$pos_start);
	35	+ }
	36	+ }
	37	+ }
	38	+
	39	+ $articles[$cpt] = $article;
	40	+ echo "<div onclick=\"onArticle($cpt)\" style=\"display:inline;\">".PHP_EOL;
	41	+ echo '<img width="100%" src="'.$article['image'].'" style="display:inline;"><br>'.PHP_EOL;
	42	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>'.PHP_EOL;
	43	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div>';
	44	+ echo $article['title'].PHP_EOL;
	45	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;
	46	+ $cpt++;
	47	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	48	+ break;
	49	+ }
	50	+ }
	51	+}
	52	+echo '</div><!-- ./col-4 -->';
	53	+echo '<div class="col-6">';
	54	+echo '<a name="article-top"></a><div id="article-current"></div>';
	55	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	56	+$cpt=0;
	57	+foreach ($articles as $article ) {
	58	+ $cpt_prev=$cpt-1;
	59	+ $cpt_next=$cpt+1;
	60	+ echo PHP_EOL.PHP_EOL.'<!-- ==================== article '.$cpt.'============== -->'.PHP_EOL;
	61	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	62	+ echo "<hr>";
	63	+ echo "<a name=\"article-$cpt\"></a>";
	64	+ $article_content = file_get_contents($article['link']);
	65	+ $article_content_utf8 = mb_convert_encoding($article_content, 'HTML-ENTITIES', "UTF-8");
	66	+ $doc = new DOMDocument();
	67	+ $doc->preserveWhiteSpace = false;
	68	+ $doc->formatOutput = true;
	69	+ $libxml_previous_state = libxml_use_internal_errors(true);
	70	+ $doc->loadHTML($article_content_utf8);
	71	+ libxml_clear_errors();
	72	+ libxml_use_internal_errors($libxml_previous_state);
	73	+ $articles = $doc->getElementsByTagName('article');
	74	+ $article_only="";
	75	+ foreach ( $articles as $node) {
	76	+ $article_only = DOMinnerHTML($node);
	77	+ }
	78	+
	79	+
	80	+ $SEARCH_SUB='<aside class="aside-column side-flow-bloc width-padded">';
	81	+ $pos_stop = strpos($article_only, $SEARCH_SUB);
	82	+ $article_only = substr($article_only, 0, $pos_stop);
	83	+
	84	+ if(0==strcmp($article_only,"")) {
	85	+ $SEARCH_SUB='itemtype="http://schema.org/NewsArticle">';
	86	+ $pos_start=strpos($article_content_utf8, $SEARCH_SUB);
	87	+ if($pos_start) {
	88	+ $article_only = substr($article_content_utf8, $pos_start + strlen($SEARCH_SUB));
	89	+ $SEARCH_SUB='<aside class="outbrain">';
	90	+ $pos_stop=strpos($article_only, $SEARCH_SUB);
	91	+ if($pos_stop) {
	92	+ $article_only = substr($article_only, 0, $pos_stop);
	93	+ } else {
	94	+ $article_only = "Oups";
	95	+ }
	96	+ } else {
	97	+ $article_only = "";
	98	+ $articles = $doc->getElementsByTagName('div');
	99	+ foreach ( $articles as $node) {
	100	+ if(0==strcmp("answer-body",$node->getAttribute('class'))) {
	101	+ $article_only = "<h3>".$article['title']."</h3>".DOMinnerHTML($node);
	102	+ }else if(0==strcmp("live-content",$node->getAttribute('class'))) {
	103	+ //$article_only = "<h3>".$article['title']."</h3>".DOMinnerHTML($node);
	104	+ $temp = DOMinnerHTML($node);
	105	+ $article_only = $article_only.$temp;
	106	+ }
	107	+ }
	108	+ }
	109	+ }
	110	+ DEBUG("Article($cpt) : length = ".strlen($article_only));
	111	+ $isCheckNews=false;
	112	+ if( 0 == strlen($article_only)) {
	113	+ $articles = $doc->getElementsByTagName('article');
	114	+ $article_only="";
	115	+ foreach ( $articles as $node) {
	116	+ $article_only = DOMinnerHTML($node);
	117	+ }
	118	+ $isCheckNews=true;
	119	+ }
	120	+
	121	+ $orgStrings = array( ' href="/' );
	122	+ $newStrings = array( ' href="https://www.liberation.fr/' );
	123	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	124	+
	125	+ $orgStrings = array( '<p lang="en" dir="ltr">',
	126	+ '<p dir="ltr">',
	127	+ '<h4>',
	128	+ '</h4>',
	129	+ '<h1 class="article-headline">',
	130	+ '</h1>',
	131	+ '<h2 class="article-standfirst read-left-padding">',
	132	+ '<h2 class="article-standfirst">',
	133	+ '<h2 class="intertitre">',
	134	+ '</h2>',
	135	+ '<figure class="article-image article-header-image"><a role="button" class="figure-zoom">',
	136	+ '<figure class="article-image article-header-image"><a role="button" class="figure-zoom js-figure-zoom">',
	137	+ '<img width="534" data-lazyload="false" src="',
	138	+ '<i class="zoom-icon"> <svg class="icon " role="img" width="22" height="22"><use xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="#icon-zoom"></use></svg></i>',
	139	+ ' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"',
	140	+ ' width="960">',
	141	+ '<img class="live-image"',
	142	+ '<a class="slug"',
	143	+ '<figcaption class="legende" data-caption="',
	144	+ '<figcaption class="read-left-padding caption">',
	145	+ '</figcaption>' );
	146	+ $newStrings = array( '<p>',
	147	+ '<p>',
	148	+ '<h5>',
	149	+ '</h5>',
	150	+ '<h3>',
	151	+ '</h3>',
	152	+ '<h4>',
	153	+ '<h4>',
	154	+ '<h4>',
	155	+ '</h4>',
	156	+ '<figure>',
	157	+ '<figure>',
	158	+ '<img width="100%" src="',
	159	+ '',
	160	+ '',
	161	+ ' width="100%">',
	162	+ '<img width="100%"',
	163	+ '<a',
	164	+ '<em>',
	165	+ '<em>',
	166	+ '</em>' );
	167	+
	168	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	169	+
	170	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	171	+ $article_only = $temp;
	172	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	173	+ $article_only = $temp;
	174	+ $re = '/<a class="lien_interne rub"(.+)>(.+)<\/a>/';
	175	+ $temp = preg_replace($re, '\\2', $article_only);
	176	+ $article_only = $temp;
	177	+ $re = '/<img (.+)data-src="(.+?)"(.+?)data-lazyload="true"(.+?)>/';
	178	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	179	+ $article_only = $temp;
	180	+ $re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
	181	+ $temp = preg_replace($re, '', $article_only);
	182	+ $article_only = $temp;
	183	+ $re = '/<span class="share">(.+?)<\/span>/';
	184	+ $temp = preg_replace($re, '', $article_only);
	185	+ $article_only = $temp;
	186	+ $orgStrings = array( '<ul class="article-rel-list">','<li class="article-rel-item"> </li>','</ul>', '<a href=');
	187	+ $newStrings = array( '','','', '<a target="new" href=' );
	188	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	189	+
	190	+ //Checknews Specific
	191	+ if($isCheckNews) {
	192	+ $re='/<\/form>/';
	193	+ $temp = preg_replace($re, "", $article_only);
	194	+ $article_only = $temp;
	195	+ $re='/<form class="checknews-form" method="GET" action="\/checknews\/search\/" name="checknews_box">(.+)<\/a>/';
	196	+ $temp = preg_replace($re, "", $article_only);
	197	+ $article_only = $temp;
	198	+ /*
	199	+ $re='//';
	200	+ $temp = preg_replace($re, "", $article_only);
	201	+ $article_only = $temp;
	202	+ */
	203	+ $re='/<li class="answer no-tags">(.+?)<\/li>/';
	204	+ $temp = preg_replace($re, "", $article_only);
	205	+ $article_only = $temp;
	206	+ $re='/<li class="answer no-tags">(.+)<ul class="checknews-tags">/';
	207	+ $temp = preg_replace($re, "<li><ul>", $article_only);
	208	+ $article_only = $temp;
	209	+ $re='/<p class="btn" href="https:\/\/www\.liberation\.fr\/checknews\/(.+)Voir la réponse<br><span>(.+)<\/span> <\/p> <\/div> <\/a> <\/li>/';
	210	+ $article_only = preg_replace($re, "</div></a></li>", $article_only);
	211	+ $re='/<a class="btn about-button" href="https:\/\/www\.liberation\.fr\/checknews\/voter\/"> Accédez à toutes les questions <\/a>/';
	212	+ $article_only = preg_replace($re, "", $article_only);
	213	+ $re='/<script type="text\/javascript" src="https:\/\/statics\.liberation\.fr\/newsite\/(.+)<\/script>/';
	214	+ $article_only = preg_replace($re, "", $article_only);
	215	+ $re='/<div class="checknews-header"> <div class="title"> <h3>(Nos dernières réponses\|Le vote des lecteurs)<\/h3> <\/div> <\/div>/';
	216	+ $article_only = preg_replace($re, "", $article_only);
	217	+ $re='/<p> Voici les cinq questions qui sont en tête des votes.<br> Vous aussi, donnez votre avis et choisissez les sujets que la rédaction va traiter. <\/p>/';
	218	+ $article_only = preg_replace($re, "", $article_only);
	219	+ $re='/<\/div><\/a><\/li> <li><ul>/';
	220	+ $article_only = preg_replace($re, "", $article_only);
	221	+ $re='/<ul id="question" class="checknews-answers vote">/';
	222	+ $article_only = preg_replace($re, "", $article_only);
	223	+ $re='/<ul class="checknews-answers">/';
	224	+ $article_only = preg_replace($re, "", $article_only);
	225	+ $re='/<div class="checknews-box last-content"> <li><ul>/';
	226	+ $article_only = preg_replace($re, "<div><li>", $article_only);
	227	+ $re='/ target="_blank">/';
	228	+ $article_only = preg_replace($re, ">", $article_only);
	229	+ $re='/<\/div><\/a><\/li>/';
	230	+ $article_only = preg_replace($re, "", $article_only);
	231	+ $article_only = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	232	+ $article_only = preg_replace('/\s\s+/', ' ', $article_only);
	233	+ $re='/<div class="checknews-box last-content"> <li><ul>/';
	234	+ $article_only = preg_replace($re, "<div>", $article_only);
	235	+
	236	+ }
	237	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	238	+ $article_only = $temp;
	239	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	240	+ $article_only = $temp;
	241	+
	242	+ //Force HTTPS
	243	+ $article_only = add_https($article_only, 'www.twitter.com');
	244	+ $article_only = add_https($article_only, 'www.facebook.com');
	245	+ $article_only = add_https($article_only, 'pbs.twimg.com');
	246	+ $article_only = add_https($article_only, 'schema.org');
	247	+ //$article_only = add_https($article_only, 'www.w3.org');
	248	+ $article_only = add_https($article_only, 'videos.senat.fr');
	249	+ $article_only = add_https($article_only, '9w4t.mjt.lu');
	250	+ $article_only = add_https($article_only, 'www.lefigaro.fr');
	251	+ $article_only = add_https($article_only, 'www.leparisien.fr');
	252	+ $article_only = add_https($article_only, 'www.lepoint.fr');
	253	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	254	+ 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>     ';
	255	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	256	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
	257	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>'.PHP_EOL;
	258	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	259	+ 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;
	260	+ 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;
	261	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	262	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
	263	+ echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
	264	+ $cpt++;
	265	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	266	+ break;
	267	+ }
	268	+}
	269	+?>

+130

sources/news-numerama.fr.php

View

...	...	@@ -0,0 +1,130 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_NUMERAMA);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->channel->title;
	11	+ $channel['link'] = $xml->channel->link;
	12	+ $channel['description'] = $xml->channel->description;
	13	+ $channel['pubDate'] = $xml->channel->pubDate;
	14	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	15	+ echo '<h4>' . $channel['title'] . '</h4>';
	16	+ $cpt=0;
	17	+ foreach ($xml->channel->item as $item) {
	18	+ $article = array();
	19	+ $article['title'] = $item->title;
	20	+ $article['link'] = $item->link;
	21	+ $article['pubDate'] = $item->pubDate;
	22	+ $article['timestamp'] = strtotime($item->pubDate);
	23	+ $article['description'] = $item->description;
	24	+ $SEARCH_SUB1='<p><img width="1024" height="';
	25	+ $pos_start = strpos($item->description, $SEARCH_SUB1);
	26	+ $pos_start += strlen($SEARCH_SUB1);
	27	+ $SEARCH_SUB2="\"";
	28	+ $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
	29	+ $height=substr($item->description, $pos_start, $pos_stop - $pos_start);
	30	+ $article['width']=1024;
	31	+ $article['height']=$height;
	32	+ $SEARCH_SUB1='src="';
	33	+ $pos_start = strpos($item->description, $SEARCH_SUB1);
	34	+ $pos_start += strlen($SEARCH_SUB1);
	35	+ $SEARCH_SUB2="\"";
	36	+ $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
	37	+ $article['image']=substr($item->description, $pos_start, $pos_stop - $pos_start);
	38	+ $articles[$cpt] = $article;
	39	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	40	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
	41	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	42	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	43	+ echo $article['title'].'  ';
	44	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	45	+ $cpt++;
	46	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	47	+ break;
	48	+ }
	49	+ }
	50	+}
	51	+echo '</div><!-- ./col-4 -->';
	52	+echo '<div class="col-6">';
	53	+echo '<a name="article-top"></a><div id="article-current"></div>';
	54	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	55	+$cpt=0;
	56	+foreach ($articles as $article ) {
	57	+ $cpt_prev=$cpt-1;
	58	+ $cpt_next=$cpt+1;
	59	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	60	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	61	+ echo "<hr>";
	62	+ echo "<a name=\"article-$cpt\">";
	63	+ $article_content = file_get_contents($article['link']);
	64	+ $SEARCH_SUB1='<article class="post-content span8 tablet12 phone12">';
	65	+ $pos_start = strpos($article_content, $SEARCH_SUB1);
	66	+ $pos_start += strlen($SEARCH_SUB1);
	67	+ $SEARCH_SUB2='<div id="bottom_ad"';
	68	+ $pos_stop = strpos($article_content, $SEARCH_SUB2);
	69	+ $article_only = substr($article_content, $pos_start, $pos_stop - $pos_start);
	70	+
	71	+ $orgStrings = array(' src="//www.numerama.com/');
	72	+ $newStrings = array(' src="https://www.numerama.com/');
	73	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	74	+ $orgStrings = array(' srcset="//www.numerama.com/');
	75	+ $newStrings = array(' srcset="https://www.numerama.com/');
	76	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	77	+ $orgStrings = array(' href="/');
	78	+ $newStrings = array(' href="https://www.numerama.com/');
	79	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	80	+
	81	+ $orgStrings = array( ' width="500" height="281" frameborder="0" ');
	82	+ $newStrings = array( ' width="100%" frameborder="0" ');
	83	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	84	+ $orgStrings = array( 'style="width: 680px"');
	85	+ $newStrings = array( 'style="width: 100%"');
	86	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	87	+ $orgStrings = array( '<img src="');
	88	+ $newStrings = array( '<img width="100%" src="');
	89	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	90	+ $re = '/alt="(.+?)" width="(.+?)" height="(.+?)"/';
	91	+ $temp = preg_replace($re, 'width="100%"', $article_only);
	92	+ $article_only = $temp;
	93	+ $re = '/alt="" width="(.+?)" height="(.+?)"/';
	94	+ $temp = preg_replace($re, 'width="100%"', $article_only);
	95	+ $article_only = $temp;
	96	+ $re = '/<iframe width="500" height="(.+?)"/';
	97	+ $temp = preg_replace($re, '<iframe width="100%"', $article_only);
	98	+ $article_only = $temp;
	99	+ $re = '/sizes="$max-width: (.+?)$ 100vw, (.+?)"/';
	100	+ $temp = preg_replace($re, 'sizes="(max-width: 100%) 100vw,100%"', $article_only);
	101	+ $article_only = $temp;
	102	+ $re = '/<figure id="attachment_(.+?)" style="width: (.+?)" class="wp-caption aligncenter">/';
	103	+ $temp = preg_replace($re, '<figure id="attachment_\\1" style="width: 100%" class="wp-caption aligncenter">', $article_only);
	104	+ $article_only = $temp;
	105	+ $re = '/<img class="aligncenter size-full wp-image-(.+?)"/';
	106	+ $temp = preg_replace($re, '<img ', $article_only);
	107	+ $article_only = $temp;
	108	+
	109	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	110	+ $article_only = $temp;
	111	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	112	+ $article_only = $temp;
	113	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	114	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	115	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	116	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	117	+ echo '<div class="extract-content" id="'.$cpt.'">';
	118	+ echo '<h4>'.$article['title'].'</h4><img width="100%" src="'.$article['image'].'" style="display:inline;"><br>';
	119	+ echo $article_only.'</div>';
	120	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	121	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>     ';
	122	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	123	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	124	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	125	+ $cpt++;
	126	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	127	+ break;
	128	+ }
	129	+}
	130	+?>

+250

sources/news-slate.fr.php

View

...	...	@@ -0,0 +1,250 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = http_get_contents(NEWS_RSS_SLATEFR);
	5	+$xml = simplexml_load_string($rss_content);
	6	+if ($xml === false) {
	7	+ echo 'Failed to read RSS';
	8	+} else {
	9	+ $channel = array();
	10	+ $channel['title'] = $xml->channel->title;
	11	+ $channel['link'] = $xml->channel->link;
	12	+ $channel['description'] = $xml->channel->description;
	13	+ $channel['pubDate'] = $xml->channel->pubDate;
	14	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	15	+ echo '<h4>' . $channel['title'] . '</h4>';
	16	+ $cpt=0;
	17	+ foreach ($xml->channel->item as $item) {
	18	+ $article = array();
	19	+ $article['title'] = $item->title;
	20	+ $article['link'] = $item->link;
	21	+ $orgStrings = array('?xtor=RSS-3208');
	22	+ $newStrings = array('');
	23	+ $article['link'] = str_replace($orgStrings, $newStrings, $article['link']);
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $article['image'] = $item->enclosure['url'];
	28	+ $articles[$cpt] = $article;
	29	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">'.PHP_EOL;
	30	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>'.PHP_EOL;
	31	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  '.PHP_EOL;
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	33	+ echo $article['title'].'  '.PHP_EOL;
	34	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;
	35	+ $cpt++;
	36	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+echo '</div><!-- ./col-4 -->';
	42	+echo '<div class="col-6">';
	43	+echo '<a name="article-top"></a><div id="article-current"></div>';
	44	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	45	+$cpt=0;
	46	+foreach ($articles as $article ) {
	47	+ $cpt_prev=$cpt-1;
	48	+ $cpt_next=$cpt+1;
	49	+ echo PHP_EOL.PHP_EOL."<!-- ==================== article '.$cpt.'============== -->".PHP_EOL;
	50	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	51	+ echo "<hr><a name=\"article-$cpt\">\n";
	52	+ $article_content = file_get_contents($article['link']);
	53	+ $doc = new DOMDocument();
	54	+ $doc->preserveWhiteSpace = false;
	55	+ $doc->formatOutput = true;
	56	+ $libxml_previous_state = libxml_use_internal_errors(true);
	57	+ $doc->loadHTML($article_content);
	58	+ libxml_clear_errors();
	59	+ libxml_use_internal_errors($libxml_previous_state);
	60	+ $articles = $doc->getElementsByTagName('article');
	61	+ $article_only="";
	62	+ foreach ( $articles as $node) {
	63	+ $article_only = DOMinnerHTML($node);
	64	+ }
	65	+ DEBUG("Article($cpt) : length = ".strlen($article_only));
	66	+ $isPortfolio=false;
	67	+ if( 0 == strlen($article_only) ) {
	68	+ //<div class="layout layout--portfolio">
	69	+ $node = getElementByClass($doc, 'div', 'layout layout--portfolio', $offset = 0);
	70	+ if( false !== $node ) {
	71	+ //Article is a portfolio
	72	+ $article_only = DOMinnerHTML($node);
	73	+ $isPortfolio=true;
	74	+ }
	75	+ }
	76	+
	77	+ $orgStrings = array( ' href="/',
	78	+ ' src="/sites/',
	79	+ '-src="/sites/',
	80	+ '<img src=');
	81	+ $newStrings = array( ' href="http://www.slate.fr/',
	82	+ ' src="http://www.slate.fr/sites/',
	83	+ '-src="http://www.slate.fr/sites/',
	84	+ '<img width="100%" src=' );
	85	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	86	+
	87	+ $orgStrings = array( '<div class="row">',
	88	+ '<div class="col-md-2">',
	89	+ '<div class="col-md-8">',
	90	+ '<div class="container width_wrap">',
	91	+ '<div class="col-left">',
	92	+ '<div class="article-header__inner col-sm-10 col-sm-offset-1 col-md-8 col-md-offset-2">',
	93	+ '<div class="content-left col-md-8 col-auto">',
	94	+ '<p class="article-header__infos article-infos">',
	95	+ '<div class="article-thumb-wrapper">',
	96	+ '<div class="article-thumb">',
	97	+ '<figure class="preload article-image">',
	98	+ '<div class="article-content row">',
	99	+ '<h3> </h3>',
	100	+ '<h1>',
	101	+ '</h1>',
	102	+ '<p style="text-align:center">' );
	103	+ $newStrings = array('<div>',
	104	+ '<div>',
	105	+ '<div>',
	106	+ '<div>',
	107	+ '<div>',
	108	+ '<div>',
	109	+ '<div>',
	110	+ '<p>',
	111	+ '<div>',
	112	+ '<div>',
	113	+ '<figure>',
	114	+ '<div>',
	115	+ '',
	116	+ '<h4>',
	117	+ '</h4>',
	118	+ '<p>');
	119	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	120	+
	121	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	122	+ $article_only = $temp;
	123	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	124	+ $article_only = $temp;
	125	+ $re = '/ style="height:(.+?)width: 640px;">/';
	126	+ $temp = preg_replace($re, '>', $article_only);
	127	+ $article_only = $temp;
	128	+ $re = '/<script>$function \($ \{var sasCallOptions = \{ siteId: 59629(.+?)<\/script>/';
	129	+ $temp = preg_replace($re, '', $article_only);
	130	+ $article_only = $temp;
	131	+ $re = '/<img data-file-id=(.+?)src="(.+?)"(.+?)>/';
	132	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	133	+ $article_only = $temp;
	134	+ $re='/<div class="media_embed" height="(.+?)" width="(.+?)">/';
	135	+ $temp = preg_replace($re, '<div>', $article_only);
	136	+ $article_only = $temp;
	137	+ $re='/<iframe allowfullscreen="" frameborder="0" height="(.+?)" src="(.+?)" width="(.+?)"><\/iframe>/';
	138	+ $temp = preg_replace($re, '<iframe frameborder="0" width="100%" src="\\2"></iframe>', $article_only);
	139	+ $article_only = $temp;
	140	+ $re='/<img width="100%" src="(.+?)" style="height:(.+?)width: (.+?)">/';
	141	+ $temp = preg_replace($re, '<img width="100%" src="\\1">', $article_only);
	142	+ $article_only = $temp;
	143	+ $re='/<div class="icon icon--(.+?)"> (.+)/';
	144	+ $temp = preg_replace($re, '<div>', $article_only);
	145	+ $article_only = $temp;
	146	+ $re='/<div class="article-header__breadcrumb">/';
	147	+ $temp = preg_replace($re, '<div>', $article_only);
	148	+ $article_only = $temp;
	149	+
	150	+ $re='/<img class="image image--full lazyload" data-full-src=/';
	151	+ $temp = preg_replace($re, '<img width="100%" src=', $article_only);
	152	+ $article_only = $temp;
	153	+
	154	+ $re='/<span class="sharing-btn__numbers">(.+)<\/span>/';
	155	+ $temp = preg_replace($re, '', $article_only);
	156	+ $article_only = $temp;
	157	+
	158	+ $re='/<span class="sharing-btn__bg"><\/span>/';
	159	+ $temp = preg_replace($re, '', $article_only);
	160	+ $article_only = $temp;
	161	+
	162	+ $re='/<script data-cfasync="false" src=(.+)<\/script><script>(.+?)<\/script>/';
	163	+ $temp = preg_replace($re, '', $article_only);
	164	+ $article_only = $temp;
	165	+
	166	+ //Clean so called social
	167	+ $re='/<a data-share="(fb\|li\|tw)" data-url="(.+)"> <div>/';
	168	+ $temp = preg_replace($re, '<a><div>', $article_only);
	169	+ $article_only = $temp;
	170	+ $re='/<a class="sharing-btn sharing-btn--circle sharing-btn--whatsapp" href="whatsapp:(.+)> <div>/';
	171	+ $temp = preg_replace($re, '<a><div>', $article_only);
	172	+ $article_only = $temp;
	173	+ $re='/<a href="fb-messenger:(.+) class="sharing-btn sharing-btn--circle sharing-btn--messenger" target="_blank"> <div>/';
	174	+ $temp = preg_replace($re, '<a><div>', $article_only);
	175	+ $article_only = $temp;
	176	+ $re='/<a href="https:\/\/www\.facebook\.com\/dialog\/send(.+)target="_blank"> <div>/';
	177	+ $temp = preg_replace($re, '<a><div>', $article_only);
	178	+ $article_only = $temp;
	179	+ $re='/<div id="inreadbis" class="ad-wrapper"><span id="wrapper_inread"><div(.+)<\/div><\/span><\/div>/';
	180	+ $temp = preg_replace($re, '', $article_only);
	181	+ $article_only = $temp;
	182	+ $re='/<\/div> <\/a> <a><div>/';
	183	+ $temp = preg_replace($re, '',$article_only);
	184	+ $article_only = $temp;
	185	+
	186	+ //Remove Emptylines
	187	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	188	+ $article_only = $temp;
	189	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	190	+ $article_only = $temp;
	191	+
	192	+
	193	+ $re='/<iframe allow="autoplay; encrypted-media" allowfullscreen="" frameborder="0" height="(.+?)" src="https:\/\/www\.youtube\.com\/embed\/(.+?)" width="(.+?)"><\/iframe>/';
	194	+ $temp = preg_replace($re, '<iframe allow="autoplay; encrypted-media" allowfullscreen="" frameborder="0" width="100%" src="https://www.youtube.com/embed/\\2"></iframe>', $article_only);
	195	+ $article_only = $temp;
	196	+
	197	+ $SEARCH='<div class="sharing-tools sharing-tools--align-center">';
	198	+ $posend = strpos($article_only,$SEARCH);
	199	+ if($posend) {
	200	+ $temp = substr( $article_only, 0, $posend );
	201	+ $article_only = $temp;
	202	+ }
	203	+
	204	+ //Portfolio specific
	205	+ if($isPortfolio) {
	206	+ $re = '/<img class="cover parallax-enable fit lazyload" data-speed="2" data-full-src="(.+?)" alt="(.+?)">/';
	207	+ $temp = preg_replace($re, '<img src="\\1" alt="\\2">', $article_only );
	208	+ $article_only = $temp;
	209	+ $re = '/" width="(.+)" height="(.+)">/';
	210	+ $temp = preg_replace($re, '>', $article_only );
	211	+ $article_only = $temp;
	212	+ }
	213	+ //Force HTTPS
	214	+ $article_only = add_https($article_only, 'www.twitter.com');
	215	+ $article_only = add_https($article_only, 'www.facebook.com');
	216	+ $article_only = add_https($article_only, 'pbs.twimg.com');
	217	+ $article_only = add_https($article_only, 'schema.org');
	218	+ $article_only = add_https($article_only, 'www.w3.org');
	219	+ $article_only = add_https($article_only, 'www.slate.fr');
	220	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	221	+ echo "\n";
	222	+ 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>     ';
	223	+ echo "\n";
	224	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	225	+ echo "\n";
	226	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	227	+ echo "\n";
	228	+ echo '<div class="extract-content" id="'.$cpt.'">';
	229	+ echo "\n";
	230	+ echo $article_only;
	231	+ echo "\n";
	232	+ echo '</div>';
	233	+ echo "\n";
	234	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	235	+ echo "\n";
	236	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     ';
	237	+ echo "\n";
	238	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	239	+ echo "\n";
	240	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	241	+ echo "\n";
	242	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	243	+ echo "\n";
	244	+ $cpt++;
	245	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	246	+ break;
	247	+ }
	248	+}
	249	+?>
	250	+

+115

sources/news-vice.fr.php

View

...	...	@@ -0,0 +1,115 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = http_get_contents(NEWS_RSS_VICEFR);
	5	+$rss_content = str_replace("<content:encoded>","<contentEncoded>",$rss_content);
	6	+$rss_content = str_replace("</content:encoded>","</contentEncoded>",$rss_content);
	7	+$articles = array();
	8	+$xml = simplexml_load_string($rss_content);
	9	+if ($xml === false) {
	10	+ echo 'Failed to read RSS';
	11	+} else {
	12	+ $channel = array();
	13	+ $channel['title'] = $xml->channel->title;
	14	+ $channel['link'] = $xml->channel->link;
	15	+ $channel['description'] = $xml->channel->description;
	16	+ $channel['pubDate'] = $xml->channel->pubDate;
	17	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	18	+ echo '<h4>' . $channel['title'] . '</h4>';
	19	+ $cpt=0;
	20	+ foreach ($xml->channel->item as $item) {
	21	+ $article = array();
	22	+ $article['title'] = $item->title;
	23	+ $article['link'] = $item->link;
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ $article['content'] = $item->contentEncoded;
	28	+ $article['image'] = $item->enclosure['url'];
	29	+ $articles[$cpt]=$article;
	30	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">'.PHP_EOL;
	31	+ echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>'.PHP_EOL;
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  '.PHP_EOL;
	33	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	34	+ echo $article['title'].'  '.PHP_EOL;
	35	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;
	36	+ $cpt++;
	37	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	38	+ break;
	39	+ }
	40	+ }
	41	+}
	42	+echo '</div><!-- ./col-4 -->';
	43	+echo '<div class="col-6">';
	44	+echo '<a name="article-top"></a><div id="article-current"></div>';
	45	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	46	+$cpt=0;
	47	+foreach ($articles as $article ) {
	48	+ $cpt_prev=$cpt-1;
	49	+ $cpt_next=$cpt+1;
	50	+ echo PHP_EOL.PHP_EOL.'<!-- ==================== article '.$cpt.'============== -->'.PHP_EOL;
	51	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	52	+ echo "<hr>";
	53	+ echo "<a name=\"article-$cpt\">";
	54	+ $article_only = $article['content'];
	55	+
	56	+ $temp = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	57	+ $article_only = $temp;
	58	+ $temp = preg_replace('/\s\s+/', ' ', $article_only);
	59	+ $article_only = $temp;
	60	+ $re='/<div data-iframely-id="(.+?)" data-embedded-url="https:\/\/www.youtube.com\/watch\?v=(.+?)"(.+?)>/';
	61	+ $newre='<div><iframe width="100%" src="https://www.youtube.com/embed/\\2?feature=oembed" frameborder="0" gesture="media" allowfullscreen></iframe>';
	62	+ $temp = preg_replace($re, $newre, $article_only);
	63	+ $article_only = $temp;
	64	+ $re='/<div data-iframely-id="(.+?)" data-embedded-url="https:\/\/www.youtube.com\/watch\?time_continue=(.+?)&v=(.+?)"(.+?)>/';
	65	+ $newre='<div><iframe width="100%" src="https://www.youtube.com/embed/\\3?feature=oembed&start=\\2" frameborder="0" gesture="media" allowfullscreen></iframe>';
	66	+ $temp = preg_replace($re, $newre, $article_only);
	67	+ $article_only = $temp;
	68	+ $re='/><iframe data-img="" data-iframely-url="(.+?)" style="(.+?)"><\/iframe>/';
	69	+ $newre='<a href="\\1" target="new">Video</a><iframe width="100%" src="https:\\1" frameborder="0" gesture="media" allowfullscreen></iframe>';
	70	+ $article_only = preg_replace($re, $newre, $article_only);
	71	+ $re = '/<div style="left:0;width:100%;height:0;position:relative;padding-bottom:56.2493%;" data-iframely-smart-iframe="true"/';
	72	+ $article_only = preg_replace($re, '<div>', $article_only);
	73	+
	74	+ $re='/<div style="max-width: (.+?)px;"/';
	75	+ $newre='<div ';
	76	+ $temp = preg_replace($re, $newre, $article_only);
	77	+ $article_only = $temp;
	78	+ $re='/<div style="left: 0; width: 100%; height: 0; position: relative; padding-bottom: 56.(.+?)" data-iframely-smart-iframe="true">/';
	79	+ $newre='<div>';
	80	+ $temp = preg_replace($re, $newre, $article_only);
	81	+ $article_only = $temp;
	82	+ $re='/<div data-iframely-id="(.+?)" data-embedded-url="https:\/\/www.youtube.com\/embed\/(.+?)"(.+?)>/';
	83	+ $newre='<div><iframe width="100%" src="https://www.youtube.com/embed/\\2" frameborder="0" gesture="media" allowfullscreen></iframe>';
	84	+ $temp = preg_replace($re, $newre, $article_only);
	85	+ $article_only = $temp;
	86	+ $re='/<div data-iframely-id="(.+?)" data-embedded-url="https:\/\/twitter.com\/(.+?)"(.+?)>/';
	87	+ $newre='<div><iframe width="100%" src="https://publish.twitter.com/oembed?https://twitter.com/\\2"></iframe>';
	88	+ $temp = preg_replace($re, $newre, $article_only);
	89	+ $article_only = $temp;
	90	+ $re='/<source media="$max-width: (.+?)$" /';
	91	+ $temp = preg_replace($re, '<source media="(max-width: 100%)" ', $article_only);
	92	+ $article_only = $temp;
	93	+ $re='/<source media="$min-width: (.+?)$" /';
	94	+ $temp = preg_replace($re, '<source media="(min-width: 100%)" ', $article_only);
	95	+ $article_only = $temp;
	96	+ $re='/<img src="(.+?)" alt="">/';
	97	+ $temp = preg_replace($re, '<img width="100%" src="\\1" alt="">" ', $article_only);
	98	+ $article_only = $temp;
	99	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	100	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	101	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	102	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
	103	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>'.PHP_EOL;
	104	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div> '.PHP_EOL;
	105	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     '.PHP_EOL;
	106	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	107	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	108	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
	109	+ echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
	110	+ $cpt++;
	111	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	112	+ break;
	113	+ }
	114	+}
	115	+?>

+77

sources/rss.php

View

...	...	@@ -0,0 +1,77 @@
	1	+<?php
	2	+$DEBUG=0;
	3	+define('NEWS_RSS_LEMONDE', 'https://www.lemonde.fr/rss/une.xml');
	4	+define('NEWS_RSS_LIBERATION', 'http://rss.liberation.fr/rss/latest/');
	5	+define('NEWS_RSS_SLATEFR', 'https://www.slate.fr/rss.xml');
	6	+define('NEWS_RSS_VICEFR', 'https://www.vice.com/fr/rss');
	7	+define('NEWS_RSS_LESINROCKS', 'https://www.lesinrocks.com/feeds/feed-a-la-une/');
	8	+define('NEWS_RSS_NUMERAMA', 'https://www.numerama.com/rss/news.rss');
	9	+define('NEWS_RSS_HUFFINGTONPOSTFR', 'https://www.huffingtonpost.fr/feeds/index.xml');
	10	+define('NEWS_RSS_LEPOINTFR', 'http://www.lepoint.fr/24h-infos/rss.xml');
	11	+define('NEWS_RSS_LEXPRESSFR', 'https://www.lexpress.fr/rss/alaune.xml');
	12	+define('NEWS_RSS_FRANCEINFO', 'https://www.francetvinfo.fr/titres.rss');
	13	+define('NEWS_RSS_CONSPIRACY', 'http://www.conspiracywatch.info/feed');
	14	+
	15	+$array_title=array(
	16	+ "lemonde" => "LeMonde.fr",
	17	+ "liberation" => "Liberation.fr",
	18	+ "slatefr" => "Slate.fr",
	19	+ "numerama" => "Numerama",
	20	+ "huffingtonpostfr" => "HuffingtonPost.fr",
	21	+ "lepointfr" => "LePoint.fr",
	22	+ "lexpressfr" => "L'Express.fr",
	23	+ "vicefr" => "Vice.fr",
	24	+ "franceinfo" => "FranceInfo",
	25	+ "lesinrocks" => "LesInrocks.fr",
	26	+ "conspiracy" => "Conspiracy Watch"
	27	+);
	28	+$array_url=array(
	29	+ "lemonde" => NEWS_RSS_LEMONDE,
	30	+ "liberation" => NEWS_RSS_LIBERATION,
	31	+ "slatefr" => NEWS_RSS_SLATEFR,
	32	+ "numerama" => NEWS_RSS_NUMERAMA,
	33	+ "huffingtonpostfr" => NEWS_RSS_HUFFINGTONPOSTFR,
	34	+ "lepointfr" => NEWS_RSS_LEPOINTFR,
	35	+ "lexpressfr" => NEWS_RSS_LEXPRESSFR,
	36	+ "vicefr" => NEWS_RSS_VICEFR,
	37	+ "franceinfo" => NEWS_RSS_FRANCEINFO,
	38	+ "lesinrocks" => NEWS_RSS_LESINROCKS,
	39	+ "conspiracy" => NEWS_RSS_CONSPIRACY
	40	+);
	41	+
	42	+function DEBUG($Message) {
	43	+ global $DEBUG;
	44	+ if($DEBUG) {
	45	+ $stderr = fopen('php://stderr', 'w');
	46	+ fwrite($stderr,"[DEBUG]".$Message.PHP_EOL);
	47	+ fclose($stderr);
	48	+ }
	49	+}
	50	+function WARNING($Message) {
	51	+ $stderr = fopen('php://stderr', 'w');
	52	+ fwrite($stderr,"[WARNING]".$Message.PHP_EOL);
	53	+ fclose($stderr);
	54	+}
	55	+function ERROR($Message) {
	56	+ $stderr = fopen('php://stderr', 'w');
	57	+ fwrite($stderr,"[ERROR]".$Message.PHP_EOL);
	58	+ fclose($stderr);
	59	+}
	60	+
	61	+function getElementByClass(&$parentNode, $tagName, $className, $offset = 0) {
	62	+ $response = false;
	63	+ $childNodeList = $parentNode->getElementsByTagName($tagName);
	64	+ $tagCount = 0;
	65	+ for ($i = 0; $i < $childNodeList->length; $i++) {
	66	+ $temp = $childNodeList->item($i);
	67	+ if (stripos($temp->getAttribute('class'), $className) !== false) {
	68	+ if ($tagCount == $offset) {
	69	+ $response = $temp;
	70	+ break;
	71	+ }
	72	+ $tagCount++;
	73	+ }
	74	+ }
	75	+ return $response;
	76	+}
	77	+?>