Adding FranceInfoTv ・ e7d0c7f ・ Gitprep

- Adding FranceInfoTv;
- Browse files
- ycawidro commited on 2018-01-18
- 1 parent 6e0bf19
  
  commit e7d0c7fa3ebf0fafc2bd4c3e96cac5b3497f5bb2

Showing 4 changed files with 147 additions and 4 deletions

sources/gennews.jessie.sh

...	...	@@ -9,3 +9,4 @@ php news-numerama.fr.php > numerama.html 2> /dev/null
9	9	php news-huffingtonpost.fr.php > huffingtonpostfr.html 2> /dev/null
10	10	php news-lepoint.fr.php > lepointfr.html 2> /dev/null
11	11	php news-lexpress.fr.php > lexpressfr.html 2> /dev/null
	12	+php news-franceinfo.php > franceinfo.html 2> /dev/null

sources/gennews.wheezy.sh

View

...	...	@@ -9,3 +9,4 @@ php news-numerama.fr.php > numerama.html 2> /dev/null
9	9	php news-huffingtonpost.fr.php > huffingtonpostfr.html 2> /dev/null
10	10	php news-lepoint.fr.php > lepointfr.html 2> /dev/null
11	11	php news-lexpress.fr.php > lexpressfr.html 2> /dev/null
	12	+php news-franceinfo.php > franceinfo.html 2> /dev/null

+140

sources/news-franceinfo.php

View

...	...	@@ -0,0 +1,140 @@
	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 > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+echo '</div><!-- ./panel-body -->';
	42	+echo '</div><!-- ./panel panel-default -->';
	43	+echo '</div><!-- ./col-md-6 -->';
	44	+echo '<div class="col-md-6">';
	45	+echo '<div class="panel panel-default">';
	46	+echo '<div class="panel-body">';
	47	+echo '<a name="article-top"></a><div id="article-current"></div>';
	48	+$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	+ $article_content = file_get_contents($article['link']);
	57	+ $doc = new DOMDocument();
	58	+ $doc->preserveWhiteSpace = false;
	59	+ $doc->formatOutput = true;
	60	+ $doc->loadHTML($article_content);
	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.lemonde.fr/');
	69	+ $newStrings = array(' href="https://www.francetvinfo.fr/');
	70	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	71	+
	72	+ $orgStrings = array(
	73	+ '<p class="txt3 description-article" itemprop="description">',
	74	+ '<p class="bloc_signature">',
	75	+ '<span id="publisher" itemprop="Publisher" data-source="LE MONDE">',
	76	+ '<div class="toolbar"></div>',
	77	+ '<div id="articleBody" class="contenu_article js_article_body" itemprop="articleBody">',
	78	+ '<h4>',
	79	+ '</h4>',
	80	+ '<h1 class="tt2" itemprop="Headline">',
	81	+ '</h1>',
	82	+ '<h2 class="taille_courante">',
	83	+ '<h2 class="intertitre">',
	84	+ '</h2>',
	85	+ '<span>Le Monde</span>',
	86	+ '<figure class="illustration_haut " style="width: 534px">',
	87	+ '<img width="534" data-lazyload="false" src="',
	88	+ ' onload="lmd.pic(this);" onerror="lmd.pic(this);" class="lazy-retina"' );
	89	+ $newStrings = array(
	90	+ '<p>',
	91	+ '<p>',
	92	+ '<span>',
	93	+ '',
	94	+ '<div>',
	95	+ '<h5>',
	96	+ '</h5>',
	97	+ '<h3>',
	98	+ '</h3>',
	99	+ '<h4>',
	100	+ '<h4>',
	101	+ '</h4>',
	102	+ 'Le Monde',
	103	+ '<figure>',
	104	+ '<img width="100%" src="',
	105	+ '' );
	106	+
	107	+ $article_only = str_replace($orgStrings, $newStrings, $article_only);
	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 = $temp;
	113	+ //$re = '/<a target="_blank" onclick="return false;" (.+?)">(.+?)<\/a>/';
	114	+ //$temp = preg_replace($re, '\\3', $article_only);
	115	+ //$article_only = $temp;
	116	+ $re = '/<a class="lien_interne rub"(.+?)>(.+?)<\/a>/';
	117	+ $temp = preg_replace($re, '\\2', $article_only);
	118	+ $article_only = $temp;
	119	+ $re = '/<img (.+?)data-src="(.+?)"(.+?)data-lazyload="true"(.+?)>/';
	120	+ $temp = preg_replace($re, '<img width="100%" src="\\2">', $article_only);
	121	+ $article_only = $temp;
	122	+ $re = '/<a href="http:\/\/ad.apsalar.com\/api\/v1\/ad(.+?)target="_blank(.+?)<\/a>/';
	123	+ $temp = preg_replace($re, '', $article_only);
	124	+ $article_only = $temp;
	125	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	126	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	127	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	128	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
	129	+ echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
	130	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     ';
	131	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>     ';
	132	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     ';
	133	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     ';
	134	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
	135	+ $cpt++;
	136	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	137	+ break;
	138	+ }
	139	+}
	140	+?>

+5 -4

sources/rss.php

View

@@ -8,8 +8,9 @@ define('NEWS_RSS_NUMERAMA', 'https://www.numerama.com/rss/news.rss');
 define('NEWS_RSS_HUFFINGTONPOSTFR', 'https://www.huffingtonpost.fr/feeds/index.xml');
 define('NEWS_RSS_LEPOINTFR', 'http://www.lepoint.fr/24h-infos/rss.xml');
 define('NEWS_RSS_LEXPRESSFR', 'https://www.lexpress.fr/rss/alaune.xml');
-define('NEWS_RSS_BUZZFEEDFR', 'https://www.buzzfeed.com/index.xml?country=fr');
+define('NEWS_RSS_FRANCEINFO', 'https://www.francetvinfo.fr/titres.rss');
 
+/*
 define('SRC_NEWS_RSS_LEMONDE', 'sources/lemonde.html');
 define('SRC_NEWS_RSS_LIBERATION', 'sources/liberation.html');
 define('SRC_NEWS_RSS_SLATEFR', 'sources/slatefr.html');
@@ -20,7 +21,7 @@ define('SRC_NEWS_RSS_HUFFINGTONPOSTFR', 'sources/huffingtonpostfr.html');
 define('SRC_NEWS_RSS_LEPOINTFR', 'sources/lepointfr.html');
 define('SRC_NEWS_RSS_LEXPRESSFR', 'sources/lexpressfr.html');
 define('SRC_NEWS_RSS_BUZZFEEDFR', 'sources/buzzfeedfr.html');
-
+ */
 $array_title=array(
   "lemonde" => "LeMonde.fr",
   "liberation" => "Liberation.fr",
@@ -30,7 +31,7 @@ $array_title=array(
   "lepointfr" => "LePoint.fr",
   "lexpressfr" => "L'Express.fr",
   "vicefr" => "Vice.fr",
-  "buzzfeedfr" => "Buzzfeed.com fr",
+  "franceinfo" => "FranceInfo",
   "lesinrocks" => "LesInrocks.fr"
 );
 $array_url=array(
@@ -42,7 +43,7 @@ $array_url=array(
   "lepointfr" => NEWS_RSS_LEPOINTFR,
   "lexpressfr" => NEWS_RSS_LEXPRESSFR,
   "vicefr" => NEWS_RSS_VICEFR,
-  "buzzfeedfr" => NEWS_RSS_BUZZFEEDFR,
+  "franceinfo" => NEWS_RSS_FRANCEINFO,
   "lesinrocks" => NEWS_RSS_LESINROCKS
 );
 ?>

...	...	@@ -8,8 +8,9 @@ define('NEWS_RSS_NUMERAMA', 'https://www.numerama.com/rss/news.rss');
8	8	define('NEWS_RSS_HUFFINGTONPOSTFR', 'https://www.huffingtonpost.fr/feeds/index.xml');
9	9	define('NEWS_RSS_LEPOINTFR', 'http://www.lepoint.fr/24h-infos/rss.xml');
10	10	define('NEWS_RSS_LEXPRESSFR', 'https://www.lexpress.fr/rss/alaune.xml');
11		-define('NEWS_RSS_BUZZFEEDFR', 'https://www.buzzfeed.com/index.xml?country=fr');
	11	+define('NEWS_RSS_FRANCEINFO', 'https://www.francetvinfo.fr/titres.rss');
12	12
	13	+/*
13	14	define('SRC_NEWS_RSS_LEMONDE', 'sources/lemonde.html');
14	15	define('SRC_NEWS_RSS_LIBERATION', 'sources/liberation.html');
15	16	define('SRC_NEWS_RSS_SLATEFR', 'sources/slatefr.html');
...	...	@@ -20,7 +21,7 @@ define('SRC_NEWS_RSS_HUFFINGTONPOSTFR', 'sources/huffingtonpostfr.html');
20	21	define('SRC_NEWS_RSS_LEPOINTFR', 'sources/lepointfr.html');
21	22	define('SRC_NEWS_RSS_LEXPRESSFR', 'sources/lexpressfr.html');
22	23	define('SRC_NEWS_RSS_BUZZFEEDFR', 'sources/buzzfeedfr.html');
23		-
	24	+ */
24	25	$array_title=array(
25	26	"lemonde" => "LeMonde.fr",
26	27	"liberation" => "Liberation.fr",
...	...	@@ -30,7 +31,7 @@ $array_title=array(
30	31	"lepointfr" => "LePoint.fr",
31	32	"lexpressfr" => "L'Express.fr",
32	33	"vicefr" => "Vice.fr",
33		- "buzzfeedfr" => "Buzzfeed.com fr",
	34	+ "franceinfo" => "FranceInfo",
34	35	"lesinrocks" => "LesInrocks.fr"
35	36	);
36	37	$array_url=array(
...	...	@@ -42,7 +43,7 @@ $array_url=array(
42	43	"lepointfr" => NEWS_RSS_LEPOINTFR,
43	44	"lexpressfr" => NEWS_RSS_LEXPRESSFR,
44	45	"vicefr" => NEWS_RSS_VICEFR,
45		- "buzzfeedfr" => NEWS_RSS_BUZZFEEDFR,
	46	+ "franceinfo" => NEWS_RSS_FRANCEINFO,
46	47	"lesinrocks" => NEWS_RSS_LESINROCKS
47	48	);
48	49	?>