boyska/rss-bridge
1
0
Fork 0
forked from blallo/rss-bridge
Code Releases Activity

bridges: use BridgeAbstract::getSimpleHTMLDOM

Browse source 
instead of BridgeAbstract::file_get_html

Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
This commit is contained in:
Pierre Mazière 2016-07-08 19:06:35 +02:00
parent f43bbda83e
commit 3c0d13c1bb
121 changed files with 1212 additions and 396 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
bridges/ABCTabsBridge.php
View file

@ -15,9 +15,9 @@ class ABCTabsBridge extends BridgeAbstract{
public function collectData(array $param){
$html = '';
$html = $this->file_get_html('http://www.abc-tabs.com/tablatures/nouveautes.html') or $this->returnClientError('No results for this query.');
$html = $this->getSimpleHTMLDOM('http://www.abc-tabs.com/tablatures/nouveautes.html') or $this->returnClientError('No results for this query.');
$table = $html->find('table#myTable', 0)->children(1);
foreach ($table->find('tr') as $tab)
{
$item = new \Item();

2
bridges/AcrimedBridge.php
View file

@ -29,7 +29,7 @@ class AcrimedBridge extends RssExpander{
$item->title = trim($newsItem->title);
$item->timestamp = strtotime($dc->date);
$articlePage = $this->file_get_html($newsItem->link);
$articlePage = $this->getSimpleHTMLDOM($newsItem->link);
$article = $hs->sanitize($articlePage->find('article.article1', 0)->innertext);
$article = HTMLSanitizer::defaultImageSrcTo($article, "http://www.acrimed.org/");

8
bridges/AllocineFRBridge.php
View file

@ -9,7 +9,7 @@ class AllocineFRBridge extends BridgeAbstract{
$this->description = "Bridge for allocine.fr";
$this->update = '2016-08-17';
$this->parameters[] =
$this->parameters[] =
'[
{
"name" : "category",
@ -18,7 +18,7 @@ class AllocineFRBridge extends BridgeAbstract{
"required" : true,
"exampleValue" : "Faux Raccord",
"title" : "Select your category",
"values" :
"values" :
[
{
"name" : "Faux Raccord",
@ -64,12 +64,12 @@ class AllocineFRBridge extends BridgeAbstract{
// Update bridge name to match selection
$this->name .= ' : ' . $category;
$html = $this->file_get_html($this->uri) or $this->returnServerError("Could not request {$this->uri}!");
$html = $this->getSimpleTMLOM($this->uri) or $this->returnServerError("Could not request {$this->uri}!");
foreach($html->find('figure.media-meta-fig') as $element)
{
$item = new Item();
$title = $element->find('div.titlebar h3.title a', 0);
$content = trim($element->innertext);
$figCaption = strpos($content, $category);

4
bridges/AnimeUltimeBridge.php
View file

@ -67,7 +67,7 @@ class AnimeUltimeBridge extends BridgeAbstract {
//Retrive page contents
$website = 'http://www.anime-ultime.net/';
$url = $website.'history-0-1/'.$requestFilter;
$html = $this->file_get_html($url) or $this->returnServerError('Could not request Anime-Ultime: '.$url);
$html = $this->getSimpleHTMLDOM($url) or $this->returnServerError('Could not request Anime-Ultime: '.$url);
//Relases are sorted by day : process each day individually
foreach ($html->find('div.history', 0)->find('h3') as $daySection) {
@ -110,7 +110,7 @@ class AnimeUltimeBridge extends BridgeAbstract {
$item->content = $item_description;
$this->items[] = $item;
$processedOK++;
//Stop processing once limit is reached
if ($processedOK >= 10)
return;

2
bridges/BandcampBridge.php
View file

@ -26,7 +26,7 @@ class BandcampBridge extends BridgeAbstract{
$html = '';
if (isset($param['tag'])) {
$this->request = $param['tag'];
$html = $this->file_get_html('http://bandcamp.com/tag/'.urlencode($this->request).'?sort_field=date') or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM('http://bandcamp.com/tag/'.urlencode($this->request).'?sort_field=date') or $this->returnServerError('No results for this query.');
}
else {
$this->returnClientError('You must specify tag (/tag/...)');

8
bridges/BastaBridge.php
View file

@ -7,14 +7,14 @@ class BastaBridge extends BridgeAbstract{
$this->description = "Returns the newest articles.";
$this->update = '2016-08-17';
}
public function collectData(array $param){
// Replaces all relative image URLs by absolute URLs. Relative URLs always start with 'local/'!
function ReplaceImageUrl($content){
return preg_replace('/src=["\']{1}([^"\']+)/ims', 'src=\'http://www.bastamag.net/$1\'', $content);
}
$html = $this->file_get_html('http://www.bastamag.net/spip.php?page=backend') or $this->returnServerError('Could not request Bastamag.');
$html = $this->getSimpleHTMLDOM('http://www.bastamag.net/spip.php?page=backend') or $this->returnServerError('Could not request Bastamag.');
$limit = 0;
foreach($html->find('item') as $element) {
@ -34,4 +34,4 @@ class BastaBridge extends BridgeAbstract{
return 3600*2; // 2 hours
}
}
?>
?>

4
bridges/BlaguesDeMerdeBridge.php
View file

@ -12,8 +12,8 @@ class BlaguesDeMerdeBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://www.blaguesdemerde.fr/') or $this->returnServerError('Could not request BDM.');
$html = $this->getSimpleHTMLDOM('http://www.blaguesdemerde.fr/') or $this->returnServerError('Could not request BDM.');
foreach($html->find('article.joke_contener') as $element) {
$item = new Item();
$temp = $element->find('a');

14
bridges/BooruprojectBridge.php
View file

@ -32,29 +32,29 @@ class BooruprojectBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0; $tags = '';
if (!empty($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (!empty($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
$page = $page - 1;
$page = $page * 20;
}
if (!empty($param['t'])) {
$tags = '&tags='.urlencode($param['t']);
if (!empty($param['t'])) {
$tags = '&tags='.urlencode($param['t']);
}
if (empty($param['i'])) {
$this->returnServerError('Please enter a ***.booru.org instance.');
}
$html = $this->file_get_html("http://".$param['i'].".booru.org/index.php?page=post&s=list&pid=".$page.$tags) or $this->returnServerError('Could not request Booruproject.');
$html = $this->getSimpleHTMLDOM("http://".$param['i'].".booru.org/index.php?page=post&s=list&pid=".$page.$tags) or $this->returnServerError('Could not request Booruproject.');
foreach($html->find('div[class=content] span') as $element) {
$item = new \Item();
$item->uri = 'http://'.$param['i'].'.booru.org/'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->find('a', 0)->getAttribute('id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->find('a', 0)->getAttribute('id'));
$item->timestamp = time();
$item->tags = $element->find('img', 0)->getAttribute('title');
$item->title = 'Booruproject '.$param['i'].' | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $element->find('img', 0)->src . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

7
bridges/CADBridge.php
View file

@ -9,7 +9,7 @@ class CADBridge extends BridgeAbstract{
}
private function CADExtractContent($url) {
$html3 = $this->file_get_html($url);
$html3 = $this->getSimpleHTMLDOM($url);
// The request might fail due to missing https support or wrong URL
if($html3 == false)
@ -27,7 +27,6 @@ class CADBridge extends BridgeAbstract{
default:
return 'Daily comic not released yet';
}
$img = implode ($url2[0]);
$html3->clear();
unset ($html3);
@ -45,7 +44,7 @@ class CADBridge extends BridgeAbstract{
return $string;
}
$html = $this->file_get_html('http://cdn2.cad-comic.com/rss.xml') or $this->returnServerError('Could not request CAD.');
$html = $this->getSimpleHTMLDOM('http://cdn2.cad-comic.com/rss.xml') or $this->returnServerError('Could not request CAD.');
$limit = 0;
foreach($html->find('item') as $element) {
@ -67,4 +66,4 @@ class CADBridge extends BridgeAbstract{
return 3600*2; // 2 hours
}
}
?>
?>

4
bridges/CNETBridge.php
View file

@ -51,7 +51,7 @@ class CNETBridge extends BridgeAbstract {
$this->topicName = $param['topic'];
$pageUrl = 'http://www.cnet.com/'.(empty($this->topicName) ? '' : 'topics/'.$this->topicName.'/');
$html = $this->file_get_html($pageUrl) or $this->returnServerError('Could not request CNET: '.$pageUrl);
$html = $this->getSimpleHTMLDOM($pageUrl) or $this->returnServerError('Could not request CNET: '.$pageUrl);
$limit = 0;
foreach($html->find('div.assetBody') as $element) {
@ -64,7 +64,7 @@ class CNETBridge extends BridgeAbstract {
if (!empty($article_title) && !empty($article_uri) && strpos($article_uri, '/news/') !== false) {
$article_html = $this->file_get_html($article_uri) or $this->returnServerError('Could not request CNET: '.$article_uri);
$article_html = $this->getSimpleHTMLDOM($article_uri) or $this->returnServerError('Could not request CNET: '.$article_uri);
$article_content = trim(CleanArticle(ExtractFromDelimiters($article_html, '<div class="articleContent', '<footer>')));

56
bridges/CoinDeskBridge.php Normal file
View file

@ -0,0 +1,56 @@
<?php
class CoinDeskBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "mitsukarenai";
$this->name = "CoinDesk";
$this->uri = "http://www.coindesk.com/";
$this->description = "Returns the 5 newest posts from CoinDesk (full text)";
$this->update = "2014-05-30";
}
public function collectData(array $param){
function CoinDeskStripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function CoinDeskExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.single-content', 0)->innertext;
$text = strip_tags($text, '<p><a><img>');
return $text;
}
$html = $this->getSimpleHTMLDOM('http://www.coindesk.com/feed/atom/') or $this->returnError('Could not request CoinDesk.', 404);
$limit = 0;
foreach($html->find('entry') as $element) {
if($limit < 5) {
$item = new \Item();
$item->title = CoinDeskStripCDATA($element->find('title', 0)->innertext);
$item->author = $element->find('author', 0)->plaintext;
$item->uri = $element->find('link', 0)->href;
$item->timestamp = strtotime($element->find('published', 0)->plaintext);
$item->content = CoinDeskExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'CoinDesk';
}
public function getURI(){
return 'http://www.coindesk.com/';
}
public function getCacheDuration(){
return 1800; // 30min
}
}

2
bridges/CollegeDeFranceBridge.php
View file

@ -33,7 +33,7 @@ class CollegeDeFranceBridge extends BridgeAbstract{
* </a>
* </li>
*/
$html = $this->file_get_html('http://www.college-de-france.fr/components/search-audiovideo.jsp?fulltext=&siteid=1156951719600&lang=FR&type=all') or $this->returnServerError('Could not request CollegeDeFrance.');
$html = $this->getSimpleHTMLDOM('http://www.college-de-france.fr/components/search-audiovideo.jsp?fulltext=&siteid=1156951719600&lang=FR&type=all') or $this->returnServerError('Could not request CollegeDeFrance.');
foreach($html->find('a[data-target]') as $element) {
$item = new \Item();
$item->title = $element->find('.title', 0)->plaintext;

4
bridges/CommonDreamsBridge.php
View file

@ -10,7 +10,7 @@ class CommonDreamsBridge extends BridgeAbstract{
}
private function CommonDreamsExtractContent($url) {
$html3 = $this->file_get_html($url);
$html3 = $this->getSimpleHTMLDOM($url);
$text = $html3->find('div[class=field--type-text-with-summary]', 0)->innertext;
$html3->clear();
unset ($html3);
@ -25,7 +25,7 @@ class CommonDreamsBridge extends BridgeAbstract{
return $string;
}
$html = $this->file_get_html('http://www.commondreams.org/rss.xml') or $this->returnServerError('Could not request CommonDreams.');
$html = $this->getSimpleHTMLDOM('http://www.commondreams.org/rss.xml') or $this->returnServerError('Could not request CommonDreams.');
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 4) {

8
bridges/CopieDoubleBridge.php
View file

@ -13,9 +13,9 @@ class CopieDoubleBridge extends BridgeAbstract{
public function collectData(array $param){
$html = $this->file_get_html('http://www.copie-double.com/') or $this->returnServerError('Could not request CopieDouble.');
$html = $this->getSimpleHTMLDOM('http://www.copie-double.com/') or $this->returnServerError('Could not request CopieDouble.');
$table = $html->find('table table', 2);
foreach($table->find('tr') as $element)
{
$td = $element->find('td', 0);
@ -23,7 +23,7 @@ class CopieDoubleBridge extends BridgeAbstract{
if($td->class == "couleur_1")
{
$item = new Item();
$title = $td->innertext;
$pos = strpos($title, "<a");
$title = substr($title, 0, $pos);
@ -33,7 +33,7 @@ class CopieDoubleBridge extends BridgeAbstract{
{
$a=$element->find("a", 0);
$item->uri = "http://www.copie-double.com" . $a->href;
$content = str_replace('src="/', 'src="http://www.copie-double.com/',$element->find("td", 0)->innertext);
$content = str_replace('href="/', 'href="http://www.copie-double.com/',$content);
$item->content = $content;

34
bridges/CourrierInternationalBridge.php
View file

@ -12,49 +12,49 @@ class CourrierInternationalBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = '';
$html = $this->file_get_html('http://www.courrierinternational.com/') or $this->returnServerError('Error.');
$html = $this->getSimpleHTMLDOM('http://www.courrierinternational.com/') or $this->returnServerError('Error.');
$element = $html->find("article");
$article_count = 1;
$article_count = 1;
foreach($element as $article) {
$item = new \Item();
$item->uri = $article->parent->getAttribute("href");
if(strpos($item->uri, "http") === FALSE) {
$item->uri = "http://courrierinternational.fr/".$item->uri;
}
$page = $this->file_get_html($item->uri);
$page = $this->getSimpleHTMLDOM($item->uri);
$cleaner = new HTMLSanitizer();
$item->content = $cleaner->sanitize($page->find("div.article-text")[0]);
$item->title = strip_tags($article->find(".title")[0]);
$dateTime = date_parse($page->find("time")[0]);
$item->timestamp = mktime(
$dateTime['hour'],
$dateTime['minute'],
$dateTime['second'],
$dateTime['month'],
$dateTime['day'],
$dateTime['hour'],
$dateTime['minute'],
$dateTime['second'],
$dateTime['month'],
$dateTime['day'],
$dateTime['year']
);
$this->items[] = $item;
$article_count ++;
if($article_count > 5) break;
}

8
bridges/CpasbienBridge.php
View file

@ -15,7 +15,7 @@ function content_get_html($contents, $maxLen=-1, $lowercase = true, $forceTagsCl
}
class CpasbienBridge extends HttpCachingBridgeAbstract{
private $request;
public function loadMetadatas() {
@ -44,7 +44,7 @@ class CpasbienBridge extends HttpCachingBridgeAbstract{
$html = '';
if (isset($param['q'])) { /* keyword search mode */
$this->request = str_replace(" ","-",trim($param['q']));
$html = $this->file_get_html($this->uri.'/recherche/'.urlencode($this->request).'.html') or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM($this->uri.'/recherche/'.urlencode($this->request).'.html') or $this->returnServerError('No results for this query.');
}
else {
$this->returnClientError('You must specify a keyword (?q=...).');
@ -53,7 +53,7 @@ class CpasbienBridge extends HttpCachingBridgeAbstract{
foreach ($html->find('#gauche',0)->find('div') as $episode) {
if ($episode->getAttribute('class')=='ligne0' || $episode->getAttribute('class')=='ligne1')
{
$htmlepisode=content_get_html($this->get_cached($episode->find('a', 0)->getAttribute('href')));
$item = new \Item();
@ -65,7 +65,7 @@ class CpasbienBridge extends HttpCachingBridgeAbstract{
$item->content = $textefiche->text();
}
else {
$item->content = $htmlepisode->find('#textefiche', 0)->find('p',0)->text();
$item->content = $htmlepisode->find('#textefiche', 0)->find('p',0)->text();
}
$item->id = $episode->find('a', 0)->getAttribute('href');

2
bridges/CryptomeBridge.php
View file

@ -28,7 +28,7 @@ class CryptomeBridge extends BridgeAbstract{
// If you want HTTPS access instead, uncomment the following line:
//$link = 'https://secure.netsolhost.com/cryptome.org/';
$html = $this->file_get_html($link) or $this->returnServerError('Could not request Cryptome.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request Cryptome.');
if (!empty($param['n'])) { /* number of documents */
$num = min(max(1, $param['n']+0), $num);
}

12
bridges/DailymotionBridge.php
View file

@ -47,14 +47,14 @@ class DailymotionBridge extends BridgeAbstract{
function getMetadata($id) {
$metadata=array();
$html2 = file_get_html('http://www.dailymotion.com/video/'.$id) or $this->returnServerError('Could not request Dailymotion.');
$html2 = $this->getSimpleHTMLDOM('http://www.dailymotion.com/video/'.$id) or $this->returnServerError('Could not request Dailymotion.');
$metadata['title'] = $html2->find('meta[property=og:title]', 0)->getAttribute('content');
$metadata['timestamp'] = strtotime($html2->find('meta[property=video:release_date]', 0)->getAttribute('content') );
$metadata['thumbnailUri'] = $html2->find('meta[property=og:image]', 0)->getAttribute('content');
$metadata['uri'] = $html2->find('meta[property=og:url]', 0)->getAttribute('content');
return $metadata;
}
}
$html = '';
@ -63,15 +63,15 @@ class DailymotionBridge extends BridgeAbstract{
if (isset($param['u'])) { // user timeline mode
$this->request = $param['u'];
$html = $this->file_get_html('http://www.dailymotion.com/user/'.urlencode($this->request).'/1') or $this->returnServerError('Could not request Dailymotion.');
$html = $this->getSimpleHTMLDOM('http://www.dailymotion.com/user/'.urlencode($this->request).'/1') or $this->returnServerError('Could not request Dailymotion.');
}
else if (isset($param['p'])) { // playlist mode
$this->request = strtok($param['p'], '_');
$html = $this->file_get_html('http://www.dailymotion.com/playlist/'.urlencode($this->request).'') or $this->returnServerError('Could not request Dailymotion.');
$html = $this->getSimpleHTMLDOM('http://www.dailymotion.com/playlist/'.urlencode($this->request).'') or $this->returnServerError('Could not request Dailymotion.');
}
else if (isset($param['s'])) { // search mode
$this->request = $param['s']; $page = 1; if (isset($param['pa'])) $page = (int)preg_replace("/[^0-9]/",'', $param['pa']);
$html = $this->file_get_html('http://www.dailymotion.com/search/'.urlencode($this->request).'/'.$page.'') or $this->returnServerError('Could not request Dailymotion.');
$this->request = $param['s']; $page = 1; if (isset($param['pa'])) $page = (int)preg_replace("/[^0-9]/",'', $param['pa']);
$html = $this->getSimpleHTMLDOM('http://www.dailymotion.com/search/'.urlencode($this->request).'/'.$page.'') or $this->returnServerError('Could not request Dailymotion.');
}
else {
$this->returnClientError('You must either specify a Dailymotion username (?u=...) or a playlist id (?p=...) or search (?s=...)');

14
bridges/DanbooruBridge.php
View file

@ -25,23 +25,23 @@ class DanbooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 1;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://donmai.us/posts?&page=$page&tags=$tags") or $this->returnServerError('Could not request Danbooru.');
$html = $this->getSimpleHTMLDOM("http://donmai.us/posts?&page=$page&tags=$tags") or $this->returnServerError('Could not request Danbooru.');
foreach($html->find('div[id=posts] article') as $element) {
$item = new \Item();
$item->uri = 'http://donmai.us'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('data-id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('data-id'));
$item->timestamp = time();
$thumbnailUri = 'http://donmai.us'.$element->find('img', 0)->src;
$item->tags = $element->find('img', 0)->getAttribute('alt');
$item->title = 'Danbooru | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

2
bridges/DansTonChatBridge.php
View file

@ -15,7 +15,7 @@ class DansTonChatBridge extends BridgeAbstract{
$html = '';
$link = 'http://danstonchat.com/latest.html';
$html = $this->file_get_html($link) or $this->returnServerError('Could not request DansTonChat.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request DansTonChat.');
foreach($html->find('div.item') as $element) {
$item = new \Item();

8
bridges/DauphineLibereBridge.php
View file

@ -82,7 +82,7 @@ class DauphineLibereBridge extends BridgeAbstract {
}
private function ExtractContent($url, $context) {
$html2 = $this->file_get_html($url, false, $context);
$html2 = $this->getSimpleHTMLDOM($url,false,$context);
$text = $html2->find('div.column', 0)->innertext;
$text = preg_replace('@<script[^>]*?>.*?</script>@si', '', $text);
return $text;
@ -102,10 +102,10 @@ class DauphineLibereBridge extends BridgeAbstract {
if (isset($param['u'])) { /* user timeline mode */
$this->request = $param['u'];
$html = $this->file_get_html('http://www.ledauphine.com/'.$this->request.'/rss', false, $context) or $this->returnServerError('Could not request DauphineLibere.');
$html = $this->getSimpleHTMLDOM('http://www.ledauphine.com/'.$this->request.'/rss',false,$context) or $this->returnServerError('Could not request DauphineLibere.');
}
else {
$html = $this->file_get_html('http://www.ledauphine.com/rss', false, $context) or $this->returnServerError('Could not request DauphineLibere.');
$html = $this->getSimpleHTMLDOM('http://www.ledauphine.com/rss',false,$context) or $this->returnServerError('Could not request DauphineLibere.');
}
$limit = 0;
@ -126,4 +126,4 @@ class DauphineLibereBridge extends BridgeAbstract {
return 3600*2; // 2 hours
}
}
?>
?>

6
bridges/DeveloppezDotComBridge.php
View file

@ -15,7 +15,7 @@ class DeveloppezDotComBridge extends BridgeAbstract{
return $string;
}
// F***ing quotes from Microsoft Word badly encoded, here was the trick:
// F***ing quotes from Microsoft Word badly encoded, here was the trick:
// http://stackoverflow.com/questions/1262038/how-to-replace-microsoft-encoded-quotes-in-php
private function convert_smart_quotes($string)
{
@ -35,14 +35,14 @@ class DeveloppezDotComBridge extends BridgeAbstract{
}
private function DeveloppezDotComExtractContent($url) {
$articleHTMLContent = $this->file_get_html($url);
$articleHTMLContent = $this->getSimpleHTMLDOM($url);
$text = $this->convert_smart_quotes($articleHTMLContent->find('div.content', 0)->innertext);
$text = utf8_encode($text);
return trim($text);
}
public function collectData(array $param){
$rssFeed = $this->file_get_html('http://www.developpez.com/index/rss') or $this->returnServerError('Could not request http://www.developpez.com/index/rss');
$rssFeed = $this->getSimpleHTMLDOM('http://www.developpez.com/index/rss') or $this->returnServerError('Could not request http://www.developpez.com/index/rss');
$limit = 0;
foreach($rssFeed->find('item') as $element) {

2
bridges/DilbertBridge.php
View file

@ -13,7 +13,7 @@ class DilbertBridge extends BridgeAbstract {
public function collectData(array $param) {
$html = $this->file_get_html($this->getURI()) or $this->returnServerError('Could not request Dilbert: '.$this->getURI());
$html = $this->getSimpleHTMLDOM($this->getURI()) or $this->returnServerError('Could not request Dilbert: '.$this->getURI());
foreach ($html->find('section.comic-item') as $element) {

14
bridges/DollbooruBridge.php
View file

@ -26,25 +26,25 @@ class DollbooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://dollbooru.org/post/list/$tags/$page") or $this->returnServerError('Could not request Dollbooru.');
$html = $this->getSimpleHTMLDOM("http://dollbooru.org/post/list/$tags/$page") or $this->returnServerError('Could not request Dollbooru.');
foreach($html->find('div[class=shm-image-list] a') as $element) {
$item = new \Item();
$item->uri = 'http://dollbooru.org'.$element->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('data-post-id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('data-post-id'));
$item->timestamp = time();
$thumbnailUri = 'http://dollbooru.org'.$element->find('img', 0)->src;
$item->tags = $element->getAttribute('data-tags');
$item->title = 'Dollbooru | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

2
bridges/DuckDuckGoBridge.php
View file

@ -22,7 +22,7 @@ class DuckDuckGoBridge extends BridgeAbstract{
$html = '';
$link = 'http://duckduckgo.com/html/?q='.$param[u].'+sort:date';
$html = $this->file_get_html($link) or $this->returnServerError('Could not request DuckDuckGo.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request DuckDuckGo.');
foreach($html->find('div.results_links') as $element) {
$item = new \Item();

4
bridges/EZTVBridge.php
View file

@ -40,11 +40,11 @@ class EZTVBridge extends BridgeAbstract{
$this->returnClientError('You must provide a list of ID (?i=showID1,showID2,...)');
// Loop on show ids
$showList = explode(",",$param['i']);
$showList = explode(",",$param['i']);
foreach($showList as $showID){
// Get show page
$html = $this->file_get_html('https://eztv.ch/shows/'.rawurlencode($showID).'/') or $this->returnServerError('Could not request EZTV for id "'.$showID.'"');
$html = $this->getSimpleHTMLDOM('https://eztv.ch/shows/'.rawurlencode($showID).'/') or $this->returnServerError('Could not request EZTV for id "'.$showID.'"');
// Loop on each element that look like an episode entry...
foreach($html->find('.forum_header_border') as $element) {

4
bridges/EliteDangerousGalnetBridge.php
View file

@ -13,7 +13,7 @@ class EliteDangerousGalnetBridge extends BridgeAbstract
public function collectData(array $param)
{
$html = $this->file_get_html('https://community.elitedangerous.com/galnet') or $this->returnServerError('Error while downloading the website content');
$html = $this->getSimpleHTMLDOM('https://community.elitedangerous.com/galnet') or $this->returnServerError('Error while downloading the website content');
foreach($html->find('div.article') as $element) {
$item = new Item();
@ -40,4 +40,4 @@ class EliteDangerousGalnetBridge extends BridgeAbstract
{
return 3600 * 2; // 2 hours
}
}
}

6
bridges/ElsevierBridge.php
View file

@ -33,7 +33,7 @@ class ElsevierBridge extends BridgeAbstract{
$time = $article->find('.article-info', 0);
if($time){
$timestring = trim($time->plaintext);
/*
/*
The format depends on the age of an article:
- Available online 29 July 2016
- July 2016
@ -63,7 +63,7 @@ class ElsevierBridge extends BridgeAbstract{
public function collectData(array $param){
$uri = 'http://www.journals.elsevier.com/' . $param['j'] . '/recent-articles/';
$html = file_get_html($uri) or $this->returnServerError('No results for Elsevier journal '.$param['j']);
$html = $this->getSimpleHTMLDOM($uri) or $this->returnServerError('No results for Elsevier journal '.$param['j']);
foreach($html->find('.pod-listing') as $article){
$item = new \Item();
@ -80,4 +80,4 @@ class ElsevierBridge extends BridgeAbstract{
return 43200; // 12h
}
}
?>
?>

56
bridges/FSBridge.php Normal file
View file

@ -0,0 +1,56 @@
<?php
class FSBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "qwertygc";
$this->name = "Futurasciences";
$this->uri = "http://www.futura-sciences.com";
$this->description = "Returns the 5 newest posts from FS (full text)";
$this->update = "03/11/2015";
}
public function collectData(array $param){
function FS_StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function FS_ExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.fiche-actualite', 0)->innertext;
$text = preg_replace('@<script[^>]*?>.*?</script>@si', '', $text);
return $text;
}
$html = $this->getSimpleHTMLDOM('http://www.futura-sciences.com/rss/actualites.xml') or $this->returnError('Could not request Futura Sciences.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 5) {
$item = new \Item();
$item->title = FS_StripCDATA($element->find('title', 0)->innertext);
$item->uri = FS_StripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = FS_ExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'Futura Sciences';
}
public function getURI(){
return 'http://www.futura-sciences.com/';
}
public function getCacheDuration(){
return 3600; // 1 hour
// return 0; // 1 hour
}
}

6
bridges/FacebookBridge.php
View file

@ -109,9 +109,9 @@ class FacebookBridge extends BridgeAbstract{
if (is_null($html)) {
if (isset($param['u'])) {
if (!strpos($param['u'], "/")) {
$html = $this->file_get_html('https://www.facebook.com/'.urlencode($param['u']).'?_fb_noscript=1') or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM('https://www.facebook.com/'.urlencode($param['u']).'?_fb_noscript=1') or $this->returnServerError('No results for this query.');
} else {
$html = $this->file_get_html('https://www.facebook.com/pages/'.$param['u'].'?_fb_noscript=1') or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM('https://www.facebook.com/pages/'.$param['u'].'?_fb_noscript=1') or $this->returnServerError('No results for this query.');
}
} else {
$this->returnClientError('You must specify a Facebook username.');
@ -155,7 +155,7 @@ class FacebookBridge extends BridgeAbstract{
$this->name = $author;
foreach($element->children() as $post) {
$item = new \Item();
if (count($post->find('abbr')) > 0) {

2
bridges/FierPandaBridge.php
View file

@ -14,7 +14,7 @@ Class FierPandaBridge extends BridgeAbstract{
public function collectData(array $param){
$link = 'http://www.fier-panda.fr/';
$html = $this->file_get_html($link) or $this->returnServerError('Could not request Fier Panda.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request Fier Panda.');
foreach($html->find('div.container-content article') as $element) {
$item = new \Item();

2
bridges/FlickrExploreBridge.php
View file

@ -12,7 +12,7 @@ class FlickrExploreBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('https://www.flickr.com/explore') or $this->returnServerError('Could not request Flickr.');
$html = $this->getSimpleHTMLDOM('https://www.flickr.com/explore') or $this->returnServerError('Could not request Flickr.');
foreach($html->find('.photo-list-photo-view') as $element) {
// Get the styles

8
bridges/FlickrTagBridge.php
View file

@ -27,16 +27,16 @@ class FlickrTagBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://www.flickr.com/search/?q=vendee&s=rec') or $this->returnServerError('Could not request Flickr.');
$html = $this->getSimpleHTMLDOM('http://www.flickr.com/search/?q=vendee&s=rec') or $this->returnServerError('Could not request Flickr.');
if (isset($param['q'])) { /* keyword search mode */
$this->request = $param['q'];
$html = $this->file_get_html('http://www.flickr.com/search/?q='.urlencode($this->request).'&s=rec') or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM('http://www.flickr.com/search/?q='.urlencode($this->request).'&s=rec') or $this->returnServerError('No results for this query.');
}
elseif (isset($param['u'])) { /* user timeline mode */
$this->request = $param['u'];
$html = $this->file_get_html('http://www.flickr.com/photos/'.urlencode($this->request).'/') or $this->returnServerError('Requested username can\'t be found.');
$html = $this->getSimpleHTMLDOM('http://www.flickr.com/photos/'.urlencode($this->request).'/') or $this->returnServerError('Requested username can\'t be found.');
}
else {
$this->returnClientError('You must specify a keyword or a Flickr username.');
}

16
bridges/FootitoBridge.php
View file

@ -12,11 +12,11 @@ class FootitoBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://www.footito.fr/') or $this->returnServerError('Could not request Footito.');
$html = $this->getSimpleHTMLDOM('http://www.footito.fr/') or $this->returnServerError('Could not request Footito.');
foreach($html->find('div.post') as $element) {
$item = new Item();
$content = trim($element->innertext);
$content = str_replace("<img", "<img style='float : left;'", $content );
$content = str_replace("class=\"logo\"", "style='float : left;'", $content );
@ -27,17 +27,17 @@ class FootitoBridge extends BridgeAbstract{
$content = str_replace("class=\"report-abuse-button\"", "style='display : none;'", $content );
$content = str_replace("class=\"reaction clearfix\"", "style='margin : 10px 0px; padding : 5px; border-bottom : 1px #DDD solid;'", $content );
$content = str_replace("class=\"infos\"", "style='font-size : 0.7em;'", $content );
$item->content = $content;
$title = $element->find('.contenu .texte ', 0)->plaintext;
$item->title = $title;
$info = $element->find('div.infos', 0);
$item->timestamp = strtotime($info->find('time', 0)->datetime);
$item->author = $info->find('a.auteur', 0)->plaintext;
$this->items[] = $item;
}
}

4
bridges/FourchanBridge.php
View file

@ -32,7 +32,7 @@ class FourchanBridge extends BridgeAbstract{
$this->returnClientError('You must specify the thread URL.');
$url = 'https://boards.4chan.org'.$thread['path'].'';
$html = $this->file_get_html($url) or $this->returnServerError("Could not request 4chan, thread not found");
$html = $this->getSimpleHTMLDOM($url) or $this->returnServerError("Could not request 4chan, thread not found");
foreach($html->find('div.postContainer') as $element) {
$item = new \Item();
@ -59,7 +59,7 @@ class FourchanBridge extends BridgeAbstract{
}
$this->items = array_reverse($this->items);
}
public function getCacheDuration(){
return 300; // 5min
}

69
bridges/FrandroidBridge.php Normal file
View file

@ -0,0 +1,69 @@
<?php
class FrandroidBridge extends BridgeAbstract
{
public function loadMetadatas() {
$this->maintainer = "Daiyousei";
$this->name = "Frandroid";
$this->uri = "http://www.frandroid.com/";
$this->description = "Returns the RSS feed from Frandroid (full text articles)";
$this->update = "2015-03-05";
}
public function collectData(array $param)
{
function FrandroidStripCDATA($string)
{
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function FrandroidExtractContent($url)
{
$html2 = $this->getSimpleHTMLDOM($url);
$html3 = $html2->find('div.post-content', 0);
$html3->find('div.no-sidebar-ad-top', 0)->outertext = '';
$ret = $html3->find('div.shortcode-container');
foreach ($ret as $value) {
$value->outertext = '';
}
$html3->find('div#hrr-link', 0)->outertext = '';
$text = $html3->innertext;
$text = strip_tags($text, '<h1><span><h2><p><b><a><blockquote><img><em><ul><ol>');
return $text;
}
$html = $this->getSimpleHTMLDOM('http://feeds.feedburner.com/Frandroid?format=xml') or $this->returnError('Could not request Frandroid.', 404);
$limit = 0;
foreach ($html->find('item') as $element) {
if ($limit < 5) {
$item = new \Item();
$item->title = FrandroidStripCDATA($element->find('title', 0)->innertext);
$item->uri = FrandroidStripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = FrandroidExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName()
{
return 'Frandroid';
}
public function getURI()
{
return 'http://www.frandroid.com/';
}
public function getCacheDuration()
{
return 300; // 5min
}
}

4
bridges/FuturaSciencesBridge.php
View file

@ -172,13 +172,13 @@ class FuturaSciencesBridge extends BridgeAbstract {
$this->returnClientError('Invalid "feed" parameter.'.$url);
$url = $this->getURI().'rss/'.$param['feed'].'.xml';
$html = $this->file_get_html($url) or $this->returnServerError('Could not request Futura-Sciences: '.$url);
$html = $this->getSimpleHTMLDOM($url) or $this->returnServerError('Could not request Futura-Sciences: '.$url);
$limit = 0;
foreach($html->find('item') as $element) {
if ($limit < 10) {
$article_url = str_replace('#xtor=RSS-8', '', StripCDATA($element->find('guid', 0)->plaintext));
$article = $this->file_get_html($article_url) or $this->returnServerError('Could not request Futura-Sciences: '.$article_url);
$article = $this->getSimpleHTMLDOM($article_url) or $this->returnServerError('Could not request Futura-Sciences: '.$article_url);
$contents = $article->find('div.content', 0)->innertext;
foreach (array(

6
bridges/GBAtempBridge.php
View file

@ -74,7 +74,7 @@ class GBAtempBridge extends BridgeAbstract {
}
private function fetch_post_content($uri, $site_url) {
$html = $this->file_get_html($uri) or $this->returnServerError('Could not request GBAtemp: '.$uri);
$html = $this->getSimpleHTMLDOM($uri) or $this->returnServerError('Could not request GBAtemp: '.$uri);
$content = $html->find('div.messageContent', 0)->innertext;
return $this->cleanup_post_content($content, $site_url);
}
@ -91,7 +91,7 @@ class GBAtempBridge extends BridgeAbstract {
} else $this->returnClientError('The provided type filter is invalid. Expecting N, R, T, or F.');
} else $this->returnClientError('Please provide a type filter. Expecting N, R, T, or F.');
$html = $this->file_get_html($this->uri) or $this->returnServerError('Could not request GBAtemp.');
$html = $this->getSimpleHTMLDOM($this->uri) or $this->returnServerError('Could not request GBAtemp.');
if ($typeFilter == 'N') {
foreach ($html->find('li[class=news_item full]') as $newsItem) {
@ -106,7 +106,7 @@ class GBAtempBridge extends BridgeAbstract {
foreach ($html->find('li.portal_review') as $reviewItem) {
$url = $this->uri.$reviewItem->find('a', 0)->href;
$title = $reviewItem->find('span.review_title', 0)->plaintext;
$content = $this->file_get_html($url) or $this->returnServerError('Could not request GBAtemp: '.$uri);
$content = $this->getSimpleHTMLDOM($url) or $this->returnServerError('Could not request GBAtemp: '.$uri);
$author = $content->find('a.username', 0)->plaintext;
$time = intval($this->ExtractFromDelimiters($content->find('abbr.DateTime', 0)->outertext, 'data-time="', '"'));
$intro = '<p><b>'.($content->find('div#review_intro', 0)->plaintext).'</b></p>';

14
bridges/GelbooruBridge.php
View file

@ -26,27 +26,27 @@ class GelbooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
$page = $page - 1;
$page = $page * 63;
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://gelbooru.com/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Gelbooru.');
$html = $this->getSimpleHTMLDOM("http://gelbooru.com/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Gelbooru.');
foreach($html->find('div[class=content] span') as $element) {
$item = new \Item();
$item->uri = 'http://gelbooru.com/'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->timestamp = time();
$thumbnailUri = $element->find('img', 0)->src;
$item->tags = $element->find('img', 0)->getAttribute('alt');
$item->title = 'Gelbooru | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

22
bridges/GiphyBridge.php
View file

@ -30,10 +30,10 @@ class GiphyBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = '';
$html = '';
$base_url = 'http://giphy.com';
if (isset($param['s'])) { /* keyword search mode */
$html = $this->file_get_html($base_url.'/search/'.urlencode($param['s'].'/')) or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM($base_url.'/search/'.urlencode($param['s'].'/')) or $this->returnServerError('No results for this query.');
}
else {
$this->returnClientError('You must specify a search worf (?s=...).');
@ -43,22 +43,22 @@ class GiphyBridge extends BridgeAbstract{
if (isset($param['n'])) {
$max = (integer) $param['n'];
}
$limit = 0;
$kw = urlencode($param['s']);
foreach($html->find('div.hoverable-gif') as $entry) {
if($limit < $max) {
$node = $entry->first_child();
$href = $node->getAttribute('href');
$html2 = $this->file_get_html($base_url . $href) or $this->returnServerError('No results for this query.');
$node = $entry->first_child();
$href = $node->getAttribute('href');
$html2 = $this->getSimpleHTMLDOM($base_url . $href) or $this->returnServerError('No results for this query.');
$figure = $html2->getElementByTagName('figure');
$img = $figure->firstChild();
$caption = $figure->lastChild();
$item = new \Item();
$item->id = $img->getAttribute('data-gif_id');
$item->uri = $img->getAttribute('data-bitly_gif_url');
$item->uri = $img->getAttribute('data-bitly_gif_url');
$item->username = 'Giphy - '.ucfirst($kw);
$title = $caption->innertext();
$title = preg_replace('/\s+/', ' ',$title);
@ -74,8 +74,8 @@ class GiphyBridge extends BridgeAbstract{
'<a href="'.$item->uri.'">'
.'<img src="'.$img->getAttribute('src').'" width="'.$img->getAttribute('data-original-width').'" height="'.$img->getAttribute('data-original-height').'" />'
.'</a>';
$this->items[] = $item;
$this->items[] = $item;
$limit++;
}
}

2
bridges/GithubIssueBridge.php
View file

@ -34,7 +34,7 @@ class GithubIssueBridge extends BridgeAbstract{
public function collectData(array $param){
$uri = 'https://github.com/'.$param['u'].'/'.$param['p'].'/issues/'.$param['i'];
$html = file_get_html($uri)
$html = $this->getSimpleHTMLDOM($uri)
or $this->returnServerError('No results for Github Issue '.$param['i'].' in project '.$param['u'].'/'.$param['p']);
foreach($html->find('.js-comment-container') as $comment){

2
bridges/GitlabCommitsBridge.php
View file

@ -44,7 +44,7 @@ class GitlabCommitsBridge extends BridgeAbstract{
$uri.='master';
}
$html = file_get_html($uri)
$html = $this->getSimpleHTMLDOM($uri)
or $this->returnServerError('No results for Gitlab Commits of project '.$param['uri'].'/'.$param['u'].'/'.$param['p']);

4
bridges/GizmodoFRBridge.php
View file

@ -14,7 +14,7 @@ class GizmodoFRBridge extends BridgeAbstract{
public function collectData(array $param){
function GizmodoFRExtractContent($url) {
$articleHTMLContent = $this->file_get_html($url);
$articleHTMLContent = $this->getSimpleHTMLDOM($url);
$text = $articleHTMLContent->find('div.entry-thumbnail', 0)->innertext;
$text = $text.$articleHTMLContent->find('div.entry-excerpt', 0)->innertext;
$text = $text.$articleHTMLContent->find('div.entry-content', 0)->innertext;
@ -26,7 +26,7 @@ class GizmodoFRBridge extends BridgeAbstract{
return $text;
}
$rssFeed = $this->file_get_html('http://www.gizmodo.fr/feed') or $this->returnServerError('Could not request http://www.gizmodo.fr/feed');
$rssFeed = $this->getSimpleHTMLDOM('http://www.gizmodo.fr/feed') or $this->returnServerError('Could not request http://www.gizmodo.fr/feed');
$limit = 0;
foreach($rssFeed->find('item') as $element) {

4
bridges/GooglePlusPostBridge.php
View file

@ -33,8 +33,8 @@ class GooglePlusPostBridge extends BridgeAbstract
$this->request = $param['username'];
// get content parsed
// $html = $this->file_get_html(__DIR__ . '/../posts2.html'
$html = $this->file_get_html(self::GOOGLE_PLUS_BASE_URL . urlencode($this->request) . '/posts'
// $html = $this->getSimpleHTMLDOM(__DIR__ . '/../posts2.html'
$html = $this->getSimpleHTMLDOM(self::GOOGLE_PLUS_BASE_URL . urlencode($this->request) . '/posts'
// force language
, false, stream_context_create(array('http'=> array(
'header' => 'Accept-Language: fr,fr-be,fr-fr;q=0.8,en;q=0.4,en-us;q=0.2;*' . "\r\n"

4
bridges/GoogleSearchBridge.php
View file

@ -35,7 +35,7 @@ class GoogleSearchBridge extends BridgeAbstract{
if (isset($param['q'])) { /* keyword search mode */
$this->request = $param['q'];
$html = $this->file_get_html('https://www.google.com/search?q=' . urlencode($this->request) . '&num=100&complete=0&tbs=qdr:y,sbd:1') or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM('https://www.google.com/search?q=' . urlencode($this->request) . '&num=100&complete=0&tbs=qdr:y,sbd:1') or $this->returnServerError('No results for this query.');
}
else{
$this->returnClientError('You must specify a keyword (?q=...).');
@ -45,7 +45,7 @@ class GoogleSearchBridge extends BridgeAbstract{
if( !is_null($emIsRes) ){
foreach($emIsRes->find('li[class=g]') as $element) {
$item = new Item();
// Extract direct URL from google href (eg. /url?q=...)
$t = $element->find('a[href]',0)->href;
$item->uri = ''.$t;

2
bridges/GuruMedBridge.php
View file

@ -16,7 +16,7 @@ class GuruMedBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://gurumed.org/feed') or $this->returnServerError('Could not request Gurumed.');
$html = $this->getSimpleHTMLDOM('http://gurumed.org/feed') or $this->returnServerError('Could not request Gurumed.');
$limit = 0;
foreach($html->find('item') as $element) {

2
bridges/HDWallpapersBridge.php
View file

@ -48,7 +48,7 @@ class HDWallpapersBridge extends BridgeAbstract {
for ($page = 1; $page <= $lastpage; $page++) {
$link = $baseUri.'/'.$category.'/page/'.$page;
$html = $this->file_get_html($link) or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('No results for this query.');
if ($page === 1) {
preg_match('/page\/(\d+)$/', $html->find('.pagination a', -2)->href, $matches);

2
bridges/HentaiHavenBridge.php
View file

@ -12,7 +12,7 @@ class HentaiHavenBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://hentaihaven.org/') or $this->returnServerError('Could not request Hentai Haven.');
$html = $this->getSimpleHTMLDOM('http://hentaihaven.org/') or $this->returnServerError('Could not request Hentai Haven.');
foreach($html->find('div.zoe-grid') as $element) {
$item = new \Item();
$item->uri = $element->find('div.brick-content h3 a', 0)->href;

2
bridges/IdenticaBridge.php
View file

@ -25,7 +25,7 @@ class IdenticaBridge extends BridgeAbstract{
$html = '';
if (isset($param['u'])) { /* user timeline mode */
$this->request = $param['u'];
$html = $this->file_get_html('https://identi.ca/'.urlencode($this->request)) or $this->returnServerError('Requested username can\'t be found.');
$html = $this->getSimpleHTMLDOM('https://identi.ca/'.urlencode($this->request)) or $this->returnServerError('Requested username can\'t be found.');
}
else {
$this->returnClientError('You must specify an Identica username (?u=...).');

18
bridges/InstagramBridge.php
View file

@ -25,35 +25,35 @@ class InstagramBridge extends BridgeAbstract{
$html = '';
if (isset($param['u'])) { /* user timeline mode */
$this->request = $param['u'];
$html = $this->file_get_html('http://instagram.com/'.urlencode($this->request)) or $this->returnServerError('Could not request Instagram.');
$html = $this->getSimpleHTMLDOM('http://instagram.com/'.urlencode($this->request)) or $this->returnServerError('Could not request Instagram.');
}
else {
$this->returnClientError('You must specify a Instagram username (?u=...).');
}
$innertext = null;
foreach($html->find('script') as $script)
{
if ('' === $script->innertext) {
continue;
}
$pos = strpos(trim($script->innertext), 'window._sharedData');
if (0 !== $pos)
{
continue;
}
$innertext = $script->innertext;
break;
}
$json = trim(substr($innertext, $pos+18), ' =;');
$data = json_decode($json);
$userMedia = $data->entry_data->ProfilePage[0]->user->media->nodes;
foreach($userMedia as $media)
@ -70,7 +70,7 @@ class InstagramBridge extends BridgeAbstract{
}
$item->timestamp = $media->date;
$this->items[] = $item;
}
}

4
bridges/JapanExpoBridge.php
View file

@ -59,7 +59,7 @@ class JapanExpoBridge extends BridgeAbstract{
};
$link = 'http://www.japan-expo-paris.com/fr/actualites';
$html = $this->file_get_html($link) or $this->returnServerError('Could not request JapanExpo: '.$link);
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request JapanExpo: '.$link);
$fullcontent = (!empty($param['mode']) && $param['mode'] == 'full');
$count = 0;
@ -73,7 +73,7 @@ class JapanExpoBridge extends BridgeAbstract{
if ($fullcontent) {
if ($count < 5) {
$article_html = $this->file_get_html($url) or $this->returnServerError('Could not request JapanExpo: '.$url);
$article_html = $this->getSimpleHTMLDOM($url) or $this->returnServerError('Could not request JapanExpo: '.$url);
$header = $article_html->find('header.pageHeadBox', 0);
$timestamp = strtotime($header->find('time', 0)->datetime);
$title_html = $header->find('div.section', 0)->next_sibling();

14
bridges/KonachanBridge.php
View file

@ -25,18 +25,18 @@ class KonachanBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 1;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://konachan.com/post?page=$page&tags=$tags") or $this->returnServerError('Could not request Konachan.');
$html = $this->getSimpleHTMLDOM("http://konachan.com/post?page=$page&tags=$tags") or $this->returnServerError('Could not request Konachan.');
$input_json = explode('Post.register(', $html);
foreach($input_json as $element)
$data[] = preg_replace('/}\)(.*)/', '}', $element);
unset($data[0]);
foreach($data as $datai) {
$json = json_decode($datai, TRUE);
$item = new \Item();
@ -45,7 +45,7 @@ class KonachanBridge extends BridgeAbstract{
$item->timestamp = $json['created_at'];
$item->imageUri = $json['file_url'];
$item->title = 'Konachan | '.$json['id'];
$item->content = '<a href="' . $item->imageUri . '"><img src="' . $json['preview_url'] . '" /></a><br>Tags: '.$json['tags'];
$item->content = '<a href="' . $item->imageUri . '"><img src="' . $json['preview_url'] . '" /></a><br>Tags: '.$json['tags'];
$this->items[] = $item;
}
}

4
bridges/KoreusBridge.php
View file

@ -16,14 +16,14 @@ class KoreusBridge extends BridgeAbstract{
}
private function KoreusExtractContent($url) {
$html2 = $this->file_get_html($url);
$html2 = $this->getSimpeHTMLDOM($url);
$text = $html2->find('p[class=itemText]', 0)->innertext;
$text = utf8_encode(preg_replace('/(Sur le m.+?)+$/i','',$text));
return $text;
}
public function collectData(array $param){
$html = $this->file_get_html('http://feeds.feedburner.com/Koreus-articles') or $this->returnServerError('Could not request Koreus.');
$html = $this->getSimpleHTMLDOM('http://feeds.feedburner.com/Koreus-articles') or $this->returnServerError('Could not request Koreus.');
$limit = 0;
foreach($html->find('item') as $element) {

24
bridges/LeBonCoinBridge.php
View file

@ -134,10 +134,10 @@ class LeBonCoinBridge extends BridgeAbstract{
"type" : "list",
"values" : [
{ "name" : "---- Select ----", "value" : "" },
{ "name" : "", "value" : "" },
{ "name" : "EMPLOI", "value" : "_emploi_" },
{ "name" : "", "value" : "" },
{ "name" : "VEHICULES", "value" : "_vehicules_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Voitures", "value" : "voitures" },
@ -149,14 +149,14 @@ class LeBonCoinBridge extends BridgeAbstract{
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Equipement Caravaning", "value" : "equipement_caravaning" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Nautisme", "value" : "nautisme" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Equipement Nautisme", "value" : "equipement_nautisme" },
{ "name" : "", "value" : "" },
{ "name" : "IMMOBILIER", "value" : "_immobilier_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Ventes immobilieres", "value" : "ventes_immobilieres" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Locations", "value" : "locations" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Colocations", "value" : "colocations" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Bureaux &amp; Commerces", "value" : "bureaux_commerces" },
{ "name" : "", "value" : "" },
{ "name" : "VACANCES", "value" : "_vacances_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Locations gites", "value" : "locations_gites" },
@ -164,14 +164,14 @@ class LeBonCoinBridge extends BridgeAbstract{
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Campings", "value" : "campings" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Hôtels", "value" : "hotels" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Hébergements insolites", "value" : "hebergements_insolites" },
{ "name" : "", "value" : "" },
{ "name" : "MULTIMEDIA", "value" : "_multimedia_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Informatique", "value" : "informatique" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Consoles & Jeux vidéo", "value" : "consoles_jeux_video" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Image & Son", "value" : "image_son" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Téléphonie", "value" : "telephonie" },
{ "name" : "", "value" : "" },
{ "name" : "LOISIRS", "value" : "_loisirs_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;DVD / Films", "value" : "dvd_films" },
@ -184,7 +184,7 @@ class LeBonCoinBridge extends BridgeAbstract{
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Collection", "value" : "collection" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Jeux & Jouets", "value" : "jeux_jouets" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Vins & Gastronomie", "value" : "vins_gastronomie" },
{ "name" : "", "value" : "" },
{ "name" : "MATERIEL PROFESSIONNEL", "value" : "_materiel_professionnel_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Materiel Agricole", "value" : "materiel_agricole" },
@ -196,7 +196,7 @@ class LeBonCoinBridge extends BridgeAbstract{
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Fournitures de Bureau", "value" : "fournitures_de_bureau" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Commerces & Marchés", "value" : "commerces_marches" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Matériel médical", "value" : "materiel_medical" },
{ "name" : "", "value" : "" },
{ "name" : "SERVICES", "value" : "_services_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Prestations de services", "value" : "prestations_de_services" },
@ -204,7 +204,7 @@ class LeBonCoinBridge extends BridgeAbstract{
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Evénements", "value" : "evenements" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Cours particuliers", "value" : "cours_particuliers" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Covoiturage", "value" : "covoiturage" },
{ "name" : "", "value" : "" },
{ "name" : "MAISON", "value" : "_maison_" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Ameublement", "value" : "ameublement" },
@ -220,7 +220,7 @@ class LeBonCoinBridge extends BridgeAbstract{
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Montres & Bijoux", "value" : "montres_bijoux" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Equipement bébé", "value" : "equipement_bebe" },
{ "name" : "&nbsp;&nbsp;&nbsp;&nbsp;Vêtements bébé", "value" : "vetements_bebe" },
{ "name" : "", "value" : "" },
{ "name" : "AUTRES", "value" : "autres" }
]
@ -239,7 +239,7 @@ class LeBonCoinBridge extends BridgeAbstract{
else {
$link = 'http://www.leboncoin.fr/' . $param['c'] . '/offres/' . $param['r'] . '/?f=a&th=1&q=' . urlencode($param['k']);
}
$html = $this->file_get_html($link) or $this->returnServerError('Could not request LeBonCoin.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request LeBonCoin.');
$list = $html->find('.tabsContent', 0);
if($list === NULL) {
@ -275,4 +275,4 @@ class LeBonCoinBridge extends BridgeAbstract{
$this->items[] = $item;
}
}
}
}

4
bridges/LeJournalDuGeekBridge.php
View file

@ -16,7 +16,7 @@ class LeJournalDuGeekBridge extends BridgeAbstract{
}
private function LeJournalDuGeekExtractContent($url) {
$articleHTMLContent = $this->file_get_html($url);
$articleHTMLContent = $this->getSimpleHMLDOM($url);
$text = $articleHTMLContent->find('div.post-content', 0)->innertext;
foreach($articleHTMLContent->find('a.more') as $element) {
@ -38,7 +38,7 @@ class LeJournalDuGeekBridge extends BridgeAbstract{
}
public function collectData(array $param){
$rssFeed = $this->file_get_html('http://www.journaldugeek.com/rss') or $this->returnServerError('Could not request http://www.journaldugeek.com/rss');
$rssFeed = $this->getSimpleHTMLDOM('http://www.journaldugeek.com/rss') or $this->returnServerError('Could not request http://www.journaldugeek.com/rss');
$limit = 0;
foreach($rssFeed->find('item') as $element) {

4
bridges/LeMondeInformatiqueBridge.php
View file

@ -34,7 +34,7 @@ class LeMondeInformatiqueBridge extends BridgeAbstract {
}
$feedUrl = 'http://www.lemondeinformatique.fr/rss/rss.xml';
$html = $this->file_get_html($feedUrl) or $this->returnServerError('Could not request LeMondeInformatique: '.$feedUrl);
$html = $this->getSimpleHTMLDOM($feedUrl) or $this->returnServerError('Could not request LeMondeInformatique: '.$feedUrl);
$limit = 0;
foreach($html->find('item') as $element) {
@ -44,7 +44,7 @@ class LeMondeInformatiqueBridge extends BridgeAbstract {
$article_uri = $element->innertext;
$article_uri = substr($article_uri, strpos($article_uri, '<link>') + 6);
$article_uri = substr($article_uri, 0, strpos($article_uri, '</link>'));
$article_html = $this->file_get_html($article_uri) or $this->returnServerError('Could not request LeMondeInformatique: '.$article_uri);
$article_html = $this->getSimpleHTMLDOM($article_uri) or $this->returnServerError('Could not request LeMondeInformatique: '.$article_uri);
$article_content = CleanArticle($article_html->find('div#article', 0)->innertext);
$article_title = $article_html->find('h1.cleanprint-title', 0)->plaintext;

55
bridges/LeMotDuJourBridge.php Normal file
View file

@ -0,0 +1,55 @@
<?php
class LeMotDuJourBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "qwertygc";
$this->name = "LeMotDuJour Bridge";
$this->uri = "http://www.lemotdujour.com/";
$this->description = "Returns the newest articles.";
$this->update = "2014-05-25";
}
public function collectData(array $param){
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function ExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.single-contenu', 0)->innertext;
return $text;
}
$html = $this->getSimpleHTMLDOM('http://feeds2.feedburner.com/lemotdujour/lemotdujour') or $this->returnError('Could not request LeMotDuJour.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 10) {
$item = new \Item();
$item->title = StripCDATA($element->find('title', 0)->innertext);
$item->uri = StripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = ExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'LeMotDuJour Bridge';
}
public function getURI(){
return 'http://lemotdujour.com/';
}
public function getCacheDuration(){
return 3600*2; // 2 hours
// return 0; // 2 hours
}
}

16
bridges/LesJoiesDuCodeBridge.php
View file

@ -12,14 +12,14 @@ class LesJoiesDuCodeBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://lesjoiesducode.fr/') or $this->returnServerError('Could not request LesJoiesDuCode.');
$html = $this->getSimpleHTMLDOM('http://lesjoiesducode.fr/') or $this->returnServerError('Could not request LesJoiesDuCode.');
foreach($html->find('div.blog-post') as $element) {
$item = new Item();
$temp = $element->find('h1 a', 0);
$titre = html_entity_decode($temp->innertext);
$url = $temp->href;
$temp = $element->find('div.blog-post-content', 0);
// retrieve .gif instead of static .jpg
@ -29,21 +29,21 @@ class LesJoiesDuCodeBridge extends BridgeAbstract{
$image->src = $img_src;
}
$content = $temp->innertext;
$auteur = $temp->find('i', 0);
$pos = strpos($auteur->innertext, "by");
if($pos > 0)
{
$auteur = trim(str_replace("*/", "", substr($auteur->innertext, ($pos + 2))));
$item->author = $auteur;
}
$item->content .= trim($content);
$item->uri = $url;
$item->title = trim($titre);
$this->items[] = $item;
}
}

4
bridges/LichessBridge.php
View file

@ -13,7 +13,7 @@ class LichessBridge extends BridgeAbstract
public function collectData(array $param)
{
$xml_feed = $this->file_get_html('http://fr.lichess.org/blog.atom') or $this->returnServerError('Could not retrieve Lichess blog feed.');
$xml_feed = $this->getSimpleHTMLDOM('http://fr.lichess.org/blog.atom') or $this->returnServerError('Could not retrieve Lichess blog feed.');
$posts_loaded = 0;
foreach($xml_feed->find('entry') as $entry)
@ -37,7 +37,7 @@ class LichessBridge extends BridgeAbstract
private function retrieve_lichess_post($blog_post_uri)
{
$blog_post_html = $this->file_get_html($blog_post_uri);
$blog_post_html = $this->getSimpleHTMLDOM($blog_post_uri);
$blog_post_div = $blog_post_html->find('#lichess_blog', 0);
$post_chapo = $blog_post_div->find('.shortlede', 0)->innertext;

2
bridges/LinkedInCompany.php
View file

@ -22,7 +22,7 @@ class LinkedInCompany extends BridgeAbstract{
$html = '';
$link = 'https://www.linkedin.com/company/'.$param[c];
$html = $this->file_get_html($link) or $this->returnServerError('Could not request LinkedIn.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request LinkedIn.');
foreach($html->find('//*[@id="my-feed-post"]/li') as $element) {
$title = $element->find('span.share-body', 0)->innertext;

14
bridges/LolibooruBridge.php
View file

@ -26,18 +26,18 @@ class LolibooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 1; $tags = '';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://lolibooru.moe/post?page=$page&tags=$tags") or $this->returnServerError('Could not request Lolibooru.');
$html = $this->getSimpleHTMLDOM("http://lolibooru.moe/post?page=$page&tags=$tags") or $this->returnServerError('Could not request Lolibooru.');
$input_json = explode('Post.register(', $html);
foreach($input_json as $element)
$data[] = preg_replace('/}\)(.*)/', '}', $element);
unset($data[0]);
foreach($data as $datai) {
$json = json_decode($datai, TRUE);
$item = new \Item();
@ -46,7 +46,7 @@ class LolibooruBridge extends BridgeAbstract{
$item->timestamp = $json['created_at'];
$item->imageUri = $json['file_url'];
$item->title = 'Lolibooru | '.$json['id'];
$item->content = '<a href="' . $item->imageUri . '"><img src="' . $json['preview_url'] . '" /></a><br>Tags: '.$json['tags'];
$item->content = '<a href="' . $item->imageUri . '"><img src="' . $json['preview_url'] . '" /></a><br>Tags: '.$json['tags'];
$this->items[] = $item;
}
}

60
bridges/MalikiBridge.php Normal file
View file

@ -0,0 +1,60 @@
<?php
class MalikiBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "mitsukarenai";
$this->name = "Maliki";
$this->uri = "http://www.maliki.com/";
$this->description = "Returns Maliki's newest strips";
$this->update = "2014-05-30";
}
public function collectData(array $param){
$html = $this->getSimpleHTMLDOM('http://www.maliki.com/') or $this->returnError('Could not request Maliki.', 404);
$count=0;
$latest=1; $latest_title="";
$latest = $html->find('div.conteneur_page a', 1)->href;
$latest_title = $html->find('div.conteneur_page img', 0)->title;
function MalikiExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = 'http://www.maliki.com/'.$html2->find('img', 0)->src;
$text = '<img alt="" src="'.$text.'"/><br>'.$html2->find('div.imageetnews', 0)->plaintext;
return $text;
}
$item = new \Item();
$item->uri = 'http://www.maliki.com/'.$latest;
$item->title = $latest_title;
$item->timestamp = time();
$item->content = MalikiExtractContent($item->uri);
$this->items[] = $item;
foreach($html->find('div.boite_strip') as $element) {
if(!empty($element->find('a',0)->href) and $count < 3) {
$item = new \Item();
$item->uri = 'http://www.maliki.com/'.$element->find('a',0)->href;
$item->title = $element->find('img',0)->title;
$item->timestamp = strtotime(str_replace('/', '-', $element->find('span.stylepetit', 0)->innertext));
$item->content = MalikiExtractContent($item->uri);
$this->items[] = $item;
$count++;
}
}
}
public function getName(){
return 'Maliki';
}
public function getURI(){
return 'http://www.maliki.com/';
}
public function getCacheDuration(){
return 86400*6; // 6 days
}
}

58
bridges/MemoLinuxBridge.php Normal file
View file

@ -0,0 +1,58 @@
<?php
class MemoLinuxBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "qwertygc";
$this->name = "MemoLinux";
$this->uri = "http://memo-linux.com/";
$this->description = "Returns the 10 newest posts from MemoLinux (full text)";
$this->update = "2015-01-30";
}
public function collectData(array $param){
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function ExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.entry-content', 0)->innertext;
$text = preg_replace('@<script[^>]*?>.*?</script>@si', '', $text);
$text = preg_replace('@<div[^>]*?>.*?</div>@si', '', $text);
$text = preg_replace("/<h1.*/", '', $text);
return $text;
}
$html = $this->getSimpleHTMLDOM('http://memo-linux.com/feed/') or $this->returnError('Could not request MemoLinux.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 10) {
$item = new \Item();
$item->title = StripCDATA($element->find('title', 0)->innertext);
$item->uri = StripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = ExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'MemoLinux';
}
public function getURI(){
return 'http://memo-linux.com/feed/';
}
public function getCacheDuration(){
return 3600*12; // 12 hours
}
}

14
bridges/MilbooruBridge.php
View file

@ -25,25 +25,25 @@ class MilbooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://sheslostcontrol.net/moe/shimmie/index.php?q=/post/list/$tags/$page") or $this->returnServerError('Could not request Milbooru.');
$html = $this->getSimpleHTMLDOM("http://sheslostcontrol.net/moe/shimmie/index.php?q=/post/list/$tags/$page") or $this->returnServerError('Could not request Milbooru.');
foreach($html->find('div[class=shm-image-list] span[class=thumb]') as $element) {
$item = new \Item();
$item->uri = 'http://sheslostcontrol.net/moe/shimmie/'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->find('a', 0)->getAttribute('data-post-id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->find('a', 0)->getAttribute('data-post-id'));
$item->timestamp = time();
$thumbnailUri = 'http://sheslostcontrol.net/moe/shimmie/'.$element->find('img', 0)->src;
$item->tags = $element->find('a', 0)->getAttribute('data-tags');
$item->title = 'Milbooru | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

4
bridges/MondeDiploBridge.php
View file

@ -9,8 +9,8 @@ class MondeDiploBridge extends BridgeAbstract{
$this->update = '2016-08-17';
}
public function collectData(array $param){
$html = $this->file_get_html($this->uri) or $this->returnServerError('Could not request MondeDiplo. for : ' . $link);
public function collectData(array $param){
$html = $this->getSimpleHTMLDOM($this->uri) or $this->returnServerError('Could not request MondeDiplo. for : ' . $link);
foreach($html->find('div.unarticle') as $article) {
$element = $article->parent();

4
bridges/MsnMondeBridge.php
View file

@ -10,13 +10,13 @@ class MsnMondeBridge extends BridgeAbstract{
}
private function MsnMondeExtractContent($url, &$item) {
$html2 = $this->file_get_html($url);
$html2 = $this->getSimpleHTLDOM($url);
$item->content = $html2->find('#content', 0)->find('article', 0)->find('section', 0)->plaintext;
$item->timestamp = strtotime($html2->find('.authorinfo-txt', 0)->find('time', 0)->datetime);
}
public function collectData(array $param){
$html = $this->file_get_html($this->uri) or $this->returnServerError('Could not request MsnMonde.');
$html = $this->getSimpleHTMLDOM($this->uri) or $this->returnServerError('Could not request MsnMonde.');
$limit = 0;
foreach($html->find('.smalla') as $article) {
if($limit < 10) {

14
bridges/MspabooruBridge.php
View file

@ -26,27 +26,27 @@ class MspabooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
$page = $page - 1;
$page = $page * 50;
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://mspabooru.com/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Mspabooru.');
$html = $this->getSimpleHTMLDOM("http://mspabooru.com/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Mspabooru.');
foreach($html->find('div[class=content] span') as $element) {
$item = new \Item();
$item->uri = 'http://mspabooru.com/'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->timestamp = time();
$thumbnailUri = $element->find('img', 0)->src;
$item->tags = $element->find('img', 0)->getAttribute('alt');
$item->title = 'Mspabooru | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

6
bridges/NakedSecurityBridge.php
View file

@ -36,7 +36,7 @@ class NakedSecurityBridge extends BridgeAbstract {
}
$feedUrl = 'https://feeds.feedburner.com/nakedsecurity?format=xml';
$html = $this->file_get_html($feedUrl) or $this->returnServerError('Could not request '.$this->getName().': '.$feedUrl);
$html = $this->getSimpleHTMLDOM($feedUrl) or $this->returnServerError('Could not request '.$this->getName().': '.$feedUrl);
$limit = 0;
foreach ($html->find('item') as $element) {
@ -44,7 +44,7 @@ class NakedSecurityBridge extends BridgeAbstract {
//Retrieve article Uri and get that page
$article_uri = $element->find('guid', 0)->plaintext;
$article_html = $this->file_get_html($article_uri) or $this->returnServerError('Could not request '.$this->getName().': '.$article_uri);
$article_html = $this->getSimpleHTMLDOM($article_uri) or $this->returnServerError('Could not request '.$this->getName().': '.$article_uri);
//Build article contents from corresponding elements
$article_title = trim($element->find('title', 0)->plaintext);
@ -68,4 +68,4 @@ class NakedSecurityBridge extends BridgeAbstract {
}
}
}
}
}

4
bridges/NasaApodBridge.php
View file

@ -13,7 +13,7 @@ class NasaApodBridge extends BridgeAbstract{
public function collectData(array $param) {
$html = $this->file_get_html('http://apod.nasa.gov/apod/archivepix.html') or $this->returnServerError('Error while downloading the website content');
$html = $this->getSimpleHTMLDOM('http://apod.nasa.gov/apod/archivepix.html') or $this->returnServerError('Error while downloading the website content');
$list = explode("<br>", $html->find('b', 0)->innertext);
for($i = 0; $i < 3;$i++)
@ -25,7 +25,7 @@ class NasaApodBridge extends BridgeAbstract{
$uri = 'http://apod.nasa.gov/apod/'.$uri_page;
$item->uri = $uri;
$picture_html = $this->file_get_html($uri);
$picture_html = $this->getSimpleHTMLDOM($uri);
$picture_html_string = $picture_html->innertext;
//Extract image and explanation

4
bridges/NeuviemeArtBridge.php
View file

@ -20,7 +20,7 @@ class NeuviemeArtBridge extends BridgeAbstract {
}
$feedUrl = 'http://www.9emeart.fr/9emeart.rss';
$html = $this->file_get_html($feedUrl) or $this->returnServerError('Could not request 9eme Art: '.$feedUrl);
$html = $this->getSimpleHTMLDOM($feedUrl) or $this->returnServerError('Could not request 9eme Art: '.$feedUrl);
$limit = 0;
foreach ($html->find('item') as $element) {
@ -28,7 +28,7 @@ class NeuviemeArtBridge extends BridgeAbstract {
//Retrieve article Uri and get that page
$article_uri = $element->find('guid', 0)->plaintext;
$article_html = $this->file_get_html($article_uri) or $this->returnServerError('Could not request 9eme Art: '.$article_uri);
$article_html = $this->getSimpleHTMLDOM($article_uri) or $this->returnServerError('Could not request 9eme Art: '.$article_uri);
//Build article contents from corresponding elements
$article_title = trim($element->find('title', 0)->plaintext);

4
bridges/NextInpactBridge.php
View file

@ -16,7 +16,7 @@ class NextInpactBridge extends BridgeAbstract {
}
private function ExtractContent($url) {
$html2 = $this->file_get_html($url);
$html2 = $this->getSimpleHTLDOM($url);
$text = '<p><em>'.$html2->find('span.sub_title', 0)->innertext.'</em></p>'
.'<p><img src="'.$html2->find('div.container_main_image_article', 0)->find('img.dedicated',0)->src.'" alt="-" /></p>'
.'<div>'.$html2->find('div[itemprop=articleBody]', 0)->innertext.'</div>';
@ -27,7 +27,7 @@ class NextInpactBridge extends BridgeAbstract {
}
public function collectData(array $param) {
$html = $this->file_get_html('http://www.nextinpact.com/rss/news.xml') or $this->returnServerError('Could not request NextInpact.');
$html = $this->getSimpleHTMLDOM('http://www.nextinpact.com/rss/news.xml') or $this->returnServerError('Could not request NextInpact.');
$limit = 0;
foreach($html->find('item') as $element) {

8
bridges/NiceMatinBridge.php
View file

@ -10,21 +10,21 @@ class NiceMatinBridge extends BridgeAbstract{
}
private function NiceMatinExtractContent($url) {
$html = $this->file_get_html($url);
$html = $this->getSimpleHTMLDOM($url);
if(!$html)
$this->returnServerError('Could not acquire content from url: ' . $url . '!');
$content = $html->find('article', 0);
if(!$content)
$this->returnServerError('Could not find \'section\'!');
$text = preg_replace('#<script(.*?)>(.*?)</script>#is', '', $content->innertext);
$text = strip_tags($text, '<p><a><img>');
return $text;
}
public function collectData(array $param){
$html = $this->file_get_html('http://www.nicematin.com/derniere-minute/rss') or $this->returnServerError('Could not request NiceMatin.');
$html = $this->getSimpleHTMLDOM('http://www.nicematin.com/derniere-minute/rss') or $this->returnServerError('Could not request NiceMatin.');
$limit = 0;
foreach($html->find('item') as $element) {

2
bridges/NovelUpdatesBridge.php
View file

@ -26,7 +26,7 @@ class NovelUpdatesBridge extends BridgeAbstract{
if(strpos($thread['path'], 'series/') === FALSE)
$this->returnClientError('You must specify the novel URL.');
$url = 'http://www.novelupdates.com'.$thread['path'].'';
$fullhtml = $this->file_get_html($url) or $this->returnServerError("Could not request NovelUpdates, novel not found");
$fullhtml = $this->getSimpleHTMLDOM($url) or $this->returnServerError("Could not request NovelUpdates, novel not found");
$this->request = $fullhtml->find('h4.seriestitle', 0)->plaintext;
// dirty fix for nasty simpledom bug: https://github.com/sebsauvage/rss-bridge/issues/259
// forcefully removes tbody

1
bridges/NumeramaBridge.php
View file

@ -42,7 +42,6 @@ class NumeramaBridge extends BridgeAbstract{
$limit++;
}
}
}
public function getCacheDuration() {

4
bridges/OpenClassroomsBridge.php
View file

@ -64,11 +64,11 @@ class OpenClassroomsBridge extends BridgeAbstract{
{
$this->returnServerError('Error: You must chose a category.');
}
$html = '';
$link = 'https://openclassrooms.com/courses?categories='.$param['u'].'&title=&sort=updatedAt+desc';
$html = $this->file_get_html($link) or $this->returnServerError('Could not request OpenClassrooms.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request OpenClassrooms.');
foreach($html->find('.courseListItem') as $element) {
$item = new \Item();

56
bridges/OpenTheoryBridge.php Normal file
View file

@ -0,0 +1,56 @@
<?php
class OpenTheoryBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "qwertygc";
$this->name = "Opentheory";
$this->uri = "http://open1theory.com";
$this->description = "Returns the 5 newest posts from OpenTheory (full text)";
$this->update = "02-08-2014";
}
public function collectData(array $param){
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function ExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.entry-content', 0)->innertext;
$text = preg_replace('@<script[^>]*?>.*?</script>@si', '', $text);
return $text;
}
$html = $this->getSimpleHTMLDOM('http://open1theory.com/feed') or $this->returnError('Could not request OpenTheory.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 5) {
$item = new \Item();
$item->title = StripCDATA($element->find('title', 0)->innertext);
$item->uri = StripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = ExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'OpenTheory';
}
public function getURI(){
return 'http://open1theory.com/feed';
}
public function getCacheDuration(){
return 3600; // 1 hour
// return 0; // 1 hour
}
}

18
bridges/ParuVenduImmoBridge.php
View file

@ -47,7 +47,7 @@ class ParuVenduImmoBridge extends BridgeAbstract
$appartment = '&tbApp=1&tbDup=1&tbChb=1&tbLof=1&tbAtl=1&tbPla=1';
$maison = '&tbMai=1&tbVil=1&tbCha=1&tbPro=1&tbHot=1&tbMou=1&tbFer=1';
$link = $this->uri.'/immobilier/annonceimmofo/liste/listeAnnonces?tt=1'.$appartment.$maison;
if (isset($param['minarea'])) {
$this->request .= ' '.$param['minarea'].' m2';
$link .= '&sur0='.urlencode($param['minarea']);
@ -56,39 +56,39 @@ class ParuVenduImmoBridge extends BridgeAbstract
if (isset($param['maxprice'])) {
$link .= '&px1='.urlencode($param['maxprice']);
}
if (isset($param['pa'])) {
$link .= '&pa='.urlencode($param['pa']);
}
if (isset($param['lo'])) {
$this->request .= ' In: '.$param['lo'];
$link .= '&lo='.urlencode($param['lo']);
}
$html = $this->file_get_html($link) or $this->returnServerError('Could not request paruvendu.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request paruvendu.');
foreach($html->find('div.annonce a') as $element) {
if (!$element->title) {
continue;
}
$img ='';
foreach($element->find('span.img img') as $img) {
if ($img->original) {
$img = '<img src="'.$img->original.'" />';
}
}
$desc = $element->find('span.desc')[0]->innertext;
$desc = str_replace("voir l'annonce", '', $desc);
$price = $element->find('span.price')[0]->innertext;
list($href) = explode('#', $element->href);
$item = new \Item();
$item->uri = $this->uri.$href;
$item->title = $element->title;

2
bridges/PickyWallpapersBridge.php
View file

@ -55,7 +55,7 @@ class PickyWallpapersBridge extends BridgeAbstract {
for ($page = 1; $page <= $lastpage; $page++) {
$link = $baseUri.'/'.$this->resolution.'/'.$this->category.'/'.(!empty($this->subcategory)?$this->subcategory.'/':'').'page-'.$page.'/';
$html = $this->file_get_html($link) or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('No results for this query.');
if ($page === 1) {
preg_match('/page-(\d+)\/$/', $html->find('.pages li a', -2)->href, $matches);

36
bridges/PinterestBridge.php
View file

@ -38,7 +38,7 @@ class PinterestBridge extends BridgeAbstract{
public function collectData(array $param){
$html = '';
if (isset($param['u']) || isset($param['b'])) {
if (empty($param['u']))
{
$this->returnClientError('You must specify a Pinterest username (?u=...).');
@ -48,33 +48,33 @@ class PinterestBridge extends BridgeAbstract{
{
$this->returnClientError('You must specify a Pinterest board for this username (?b=...).');
}
$this->username = $param['u'];
$this->board = $param['b'];
$html = $this->file_get_html($this->getURI().'/'.urlencode($this->username).'/'.urlencode($this->board)) or $this->returnServerError('Username and/or board not found');
$html = $this->getSimpleHTMLDOM($this->getURI().'/'.urlencode($this->username).'/'.urlencode($this->board)) or $this->returnServerError('Username and/or board not found');
} else if (isset($param['q']))
{
$this->query = $param['q'];
$html = $this->file_get_html($this->getURI().'/search/?q='.urlencode($this->query)) or $this->returnServerError('Could not request Pinterest.');
$html = $this->getSimpleHTMLDOM($this->getURI().'/search/?q='.urlencode($this->query)) or $this->returnServerError('Could not request Pinterest.');
}
else {
$this->returnClientError('You must specify a Pinterest username and a board name (?u=...&b=...).');
}
foreach($html->find('div.pinWrapper') as $div)
{
$a = $div->find('a.pinImageWrapper',0);
$img = $a->find('img', 0);
$item = new \Item();
$item->uri = $this->getURI().$a->getAttribute('href');
$item->content = '<img src="' . htmlentities(str_replace('/236x/', '/736x/', $img->getAttribute('src'))) . '" alt="" />';
if (isset($this->query))
{
$avatar = $div->find('div.creditImg', 0)->find('img', 0);
@ -84,25 +84,25 @@ class PinterestBridge extends BridgeAbstract{
$username = $div->find('div.creditName', 0);
$board = $div->find('div.creditTitle', 0);
$item->username =$username->innertext;
$item->username =$username->innertext;
$item->fullname = $board->innertext;
$item->avatar = $avatar;
$item->content .= '<br /><img align="left" style="margin: 2px 4px;" src="'.htmlentities($item->avatar).'" /> <strong>'.$item->username.'</strong>';
$item->content .= '<br />'.$item->fullname;
}
$item->title = $img->getAttribute('alt');
//$item->timestamp = $media->created_time;
$this->items[] = $item;
}
}
public function getName(){
if (isset($this->query))
{
return $this->query .' - Pinterest';

4
bridges/PlanetLibreBridge.php
View file

@ -10,13 +10,13 @@ class PlanetLibreBridge extends BridgeAbstract{
}
private function PlanetLibreExtractContent($url){
$html2 = $this->file_get_html($url);
$html2 = $this->geSimpleHTMLDOM($url);
$text = $html2->find('div[class="post-text"]', 0)->innertext;
return $text;
}
public function collectData(array $param){
$html = $this->file_get_html('http://www.planet-libre.org/') or $this->returnServerError('Could not request PlanetLibre.');
$html = $this->getSimpleHTMLDOM('http://www.planet-libre.org/') or $this->returnServerError('Could not request PlanetLibre.');
$limit = 0;
foreach($html->find('div.post') as $element) {
if($limit < 5) {

2
bridges/ProjectMGameBridge.php
View file

@ -14,7 +14,7 @@ class ProjectMGameBridge extends BridgeAbstract{
public function collectData(array $param){
$html = '';
$html = $this->file_get_html('http://projectmgame.com/en/') or $this->returnServerError('Error while downloading the Project M homepage');
$html = $this->getSimpleHTMLDOM('http://projectmgame.com/en/') or $this->returnServerError('Error while downloading the Project M homepage');
foreach($html->find('article') as $article) {
$item = new \Item();

2
bridges/RTBFBridge.php
View file

@ -25,7 +25,7 @@ class RTBFBridge extends BridgeAbstract {
$count = 0;
if (isset($param['c'])) {
$html = $this->file_get_html('http://www.rtbf.be/auvio/emissions/detail?id='.$param['c']) or $this->returnServerError('Could not request RTBF.');
$html = $this->getSimpleHTMLDOM('http://www.rtbf.be/auvio/emissions/detail?id='.$param['c']) or $this->returnServerError('Could not request RTBF.');
foreach($html->find('section[id!=widget-ml-avoiraussi-] .rtbf-media-grid article') as $element) {
if($count < $limit) {

53
bridges/RaymondBridge.php Normal file
View file

@ -0,0 +1,53 @@
<?php
class RaymondBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "pit-fgfjiudghdf";
$this->name = "Raymond";
$this->uri = "http://www.raymond.cc";
$this->description = "Returns the 3 newest posts from Raymond.cc (full text)";
$this->update = "2014-05-26";
}
public function collectData(array $param){
function raymondStripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function raymondExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.entry-content', 0)->innertext;
$text = preg_replace('/class="ad".*/', '', $text);
$text = strip_tags($text, '<p><a><i><strong><em><img>');
$text = str_replace('(adsbygoogle = window.adsbygoogle || []).push({});', '', $text);
return $text;
}
$html = $this->getSimpleHTMLDOM('http://www.raymond.cc/blog/feed') or $this->returnError('Could not request raymond.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 3) {
$item = new \Item();
$item->title = raymondStripCDATA($element->find('title', 0)->innertext);
$item->uri = raymondStripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = raymondExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'raymond';
}
public function getURI(){
return 'http://www.raymond.cc/blog';
}
public function getCacheDuration(){
return 3600*12; // 12 hour
}
}

2
bridges/Releases3DSBridge.php
View file

@ -66,7 +66,7 @@ class Releases3DSBridge extends BridgeAbstract {
//Retrieve cover art and short desc from IGN?
$ignResult = false; $ignDescription = ''; $ignLink = ''; $ignDate = time(); $ignCoverArt = '';
$ignSearchUrl = 'http://www.ign.com/search?q='.urlencode($name);
if ($ignResult = $this->file_get_html($ignSearchUrl)) {
if ($ignResult = $this->getSimpleHTMLDOM($ignSearchUrl)) {
$ignCoverArt = $ignResult->find('div.search-item-media', 0)->find('img', 0)->src;
$ignDesc = $ignResult->find('div.search-item-description', 0)->plaintext;
$ignLink = $ignResult->find('div.search-item-sub-title', 0)->find('a', 1)->href;

4
bridges/ReporterreBridge.php
View file

@ -10,7 +10,7 @@ class ReporterreBridge extends BridgeAbstract{
}
private function ExtractContentReporterre($url) {
$html2 = $this->file_get_html($url);
$html2 = $this->getSimpleHTMLDOM($url);
foreach($html2->find('div[style=text-align:justify]') as $e) {
$text = $e->outertext;
@ -27,7 +27,7 @@ class ReporterreBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('http://www.reporterre.net/spip.php?page=backend') or $this->returnServerError('Could not request Reporterre.');
$html = $this->getSimpleHTMLDOM('http://www.reporterre.net/spip.php?page=backend') or $this->returnServerError('Could not request Reporterre.');
$limit = 0;
foreach($html->find('item') as $element) {

2
bridges/Rue89Bridge.php
View file

@ -22,7 +22,7 @@ class Rue89Bridge extends BridgeAbstract{
public function collectData(array $param){
$html = $this->file_get_html('http://api.rue89.nouvelobs.com/feed') or $this->returnServerError('Could not request Rue89.');
$html = $this->getSimpleHTMLDOM('http://api.rue89.nouvelobs.com/feed') or $this->returnServerError('Could not request Rue89.');
$limit = 0;
foreach($html->find('item') as $element) {

14
bridges/Rule34Bridge.php
View file

@ -25,27 +25,27 @@ class Rule34Bridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
$page = $page - 1;
$page = $page * 50;
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://rule34.xxx/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Rule34.');
$html = $this->getSimpleHTMLDOM("http://rule34.xxx/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Rule34.');
foreach($html->find('div[class=content] span') as $element) {
$item = new \Item();
$item->uri = 'http://rule34.xxx/'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->timestamp = time();
$thumbnailUri = $element->find('img', 0)->src;
$item->tags = $element->find('img', 0)->getAttribute('alt');
$item->title = 'Rule34 | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

14
bridges/Rule34pahealBridge.php
View file

@ -26,25 +26,25 @@ class Rule34pahealBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://rule34.paheal.net/post/list/$tags/$page") or $this->returnServerError('Could not request Rule34paheal.');
$html = $this->getSimpleHTMLDOM("http://rule34.paheal.net/post/list/$tags/$page") or $this->returnServerError('Could not request Rule34paheal.');
foreach($html->find('div[class=shm-image-list] div[class=shm-thumb]') as $element) {
$item = new \Item();
$item->uri = 'http://rule34.paheal.net'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->find('img', 0)->getAttribute('id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->find('img', 0)->getAttribute('id'));
$item->timestamp = time();
$thumbnailUri = $element->find('img', 0)->src;
$item->tags = $element->getAttribute('data-tags');
$item->title = 'Rule34paheal | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

14
bridges/SafebooruBridge.php
View file

@ -26,27 +26,27 @@ class SafebooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 0;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
$page = $page - 1;
$page = $page * 40;
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://safebooru.org/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Safebooru.');
$html = $this->getSimpleHTMLDOM("http://safebooru.org/index.php?page=post&s=list&tags=$tags&pid=$page") or $this->returnServerError('Could not request Safebooru.');
foreach($html->find('div[class=content] span') as $element) {
$item = new \Item();
$item->uri = 'http://safebooru.org/'.$element->find('a', 0)->href;
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->postid = (int)preg_replace("/[^0-9]/",'', $element->getAttribute('id'));
$item->timestamp = time();
$thumbnailUri = $element->find('img', 0)->src;
$item->tags = $element->find('img', 0)->getAttribute('alt');
$item->title = 'Safebooru | '.$item->postid;
$item->content = '<a href="' . $item->uri . '"><img src="' . $thumbnailUri . '" /></a><br>Tags: '.$item->tags;
$this->items[] = $item;
$this->items[] = $item;
}
}

14
bridges/SakugabooruBridge.php
View file

@ -25,18 +25,18 @@ class SakugabooruBridge extends BridgeAbstract{
public function collectData(array $param){
$page = 1;$tags='';
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
if (isset($param['p'])) {
$page = (int)preg_replace("/[^0-9]/",'', $param['p']);
}
if (isset($param['t'])) {
$tags = urlencode($param['t']);
if (isset($param['t'])) {
$tags = urlencode($param['t']);
}
$html = $this->file_get_html("http://sakuga.yshi.org/post?page=$page&tags=$tags") or $this->returnServerError('Could not request Sakugabooru.');
$html = $this->getSimpleHTMLDOM("http://sakuga.yshi.org/post?page=$page&tags=$tags") or $this->returnServerError('Could not request Sakugabooru.');
$input_json = explode('Post.register(', $html);
foreach($input_json as $element)
$data[] = preg_replace('/}\)(.*)/', '}', $element);
unset($data[0]);
foreach($data as $datai) {
$json = json_decode($datai, TRUE);
$item = new \Item();
@ -45,7 +45,7 @@ class SakugabooruBridge extends BridgeAbstract{
$item->timestamp = $json['created_at'];
$item->imageUri = $json['file_url'];
$item->title = 'Sakugabooru | '.$json['id'];
$item->content = '<a href="' . $item->imageUri . '"><img src="' . $json['preview_url'] . '" /></a><br>Tags: '.$json['tags'];
$item->content = '<a href="' . $item->imageUri . '"><img src="' . $json['preview_url'] . '" /></a><br>Tags: '.$json['tags'];
$this->items[] = $item;
}
}

54
bridges/ScilogsBridge.php Normal file
View file

@ -0,0 +1,54 @@
<?php
class ScilogsBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "qwertygc";
$this->name = "Scilogs Bridge";
$this->uri = "http://www.scilogs.fr/";
$this->description = "Returns the newest articles.";
$this->update = "2014-05-25";
}
public function collectData(array $param){
function ScilogsStripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function ScilogsExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.entrybody', 0)->innertext;
return $text;
}
$html = $this->getSimpleHTMLDOM('http://www.scilogs.fr/?wpmu-feed=posts') or $this->returnError('Could not request Scilogs.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 10) {
$item = new \Item();
$item->title = ScilogsStripCDATA($element->find('title', 0)->innertext);
$item->uri = ScilogsStripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = ScilogsExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'Scilogs Bridge';
}
public function getURI(){
return 'http://scilogs.fr/';
}
public function getCacheDuration(){
return 3600*2; // 2 hours
}
}

14
bridges/ScmbBridge.php
View file

@ -10,20 +10,20 @@ class ScmbBridge extends BridgeAbstract{
$this->update = '2016-08-17';
}
public function collectData(array $param){
$html = '';
$html = $this->file_get_html('http://secouchermoinsbete.fr/') or $this->returnServerError('Could not request Se Coucher Moins Bete.');
$html = $this->getSimpleHTMLDOM('http://secouchermoinsbete.fr/') or $this->returnServerError('Could not request Se Coucher Moins Bete.');
foreach($html->find('article') as $article) {
$item = new \Item();
$item->uri = 'http://secouchermoinsbete.fr'.$article->find('p.summary a',0)->href;
$item->title = $article->find('header h1 a',0)->innertext;
$article->find('span.read-more',0)->outertext=''; // remove text "En savoir plus" from anecdote content
$content = $article->find('p.summary a',0)->innertext;
$content =substr($content,0,strlen($content)-17); // remove superfluous spaces at the end
// get publication date
$str_date = $article->find('time',0)->datetime;
list($date, $time) = explode(' ', $str_date);
@ -31,8 +31,8 @@ class ScmbBridge extends BridgeAbstract{
list($h, $i) = explode(':', $time);
$timestamp = mktime($h,$i,0,$m,$d,$y);
$item->timestamp = $timestamp;
$item->content = $content;
$this->items[] = $item;
}

6
bridges/ScoopItBridge.php
View file

@ -24,9 +24,9 @@ class ScoopItBridge extends BridgeAbstract{
if ($param['u'] != '') {
$this->request = $param['u'];
$link = 'http://scoop.it/search?q=' .urlencode($this->request);
$html = $this->file_get_html($link) or $this->returnServerError('Could not request ScoopIt. for : ' . $link);
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request ScoopIt. for : ' . $link);
foreach($html->find('div.post-view') as $element) {
$item = new Item();
$item->uri = $element->find('a', 0)->href;

55
bridges/SegfaultMintBridge.php Normal file
View file

@ -0,0 +1,55 @@
<?php
class SegfaultMintBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "qwertygc";
$this->name = "SegfaultMint";
$this->uri = "http://segfault.linuxmint.com/";
$this->description = "Returns the 5 newest posts from SegfaultMint (full text)";
$this->update = "2014-07-05";
}
public function collectData(array $param){
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function ExtractContent($url) {
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find('div.post-bodycopy', 0)->innertext;
$text = preg_replace('@<script[^>]*?>.*?</script>@si', '', $text);
return $text;
}
$html = $this->getSimpleHTMLDOM('http://segfault.linuxmint.com/feed/') or $this->returnError('Could not request segfault.', 404);
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 5) {
$item = new \Item();
$item->title = StripCDATA($element->find('title', 0)->innertext);
$item->uri = StripCDATA($element->find('guid', 0)->plaintext);
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = ExtractContent($item->uri);
$this->items[] = $item;
$limit++;
}
}
}
public function getName(){
return 'Segfault Mint';
}
public function getURI(){
return 'http://segfault.linuxmint.com/feed/';
}
public function getCacheDuration(){
return 3600*24; // 24 hours
}
}

14
bridges/Sexactu.php
View file

@ -15,10 +15,10 @@ class Sexactu extends BridgeAbstract{
$find = array('janvier', 'février', 'mars', 'avril', 'mai', 'juin', 'juillet', 'août', 'septembre', 'novembre', 'décembre');
$replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December');
$html = $this->file_get_html($this->getURI()) or $this->returnServerError('Could not request '.$this->getURI());
$html = $this->getSimpleHTMLDOM($this->getURI()) or $this->returnServerError('Could not request '.$this->getURI());
foreach($html->find('.content-holder') as $contentHolder) {
// only use first list as second one only contains pages numbers
// only use first list as second one only contains pages numbers
$articles = $contentHolder->find('ul', 0);
foreach($articles->find('li') as $element) {
// if you ask about that method_exists, there seems to be a bug in simple html dom
@ -38,7 +38,7 @@ $replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July',
$dateText = $titleTimestamp->innertext;
$dateText = substr($dateText, strpos($dateText,',')+1);
$dateText = str_replace($find, $replace, strtolower($dateText));
$date = strtotime($dateText);
$date = strtotime($dateText);
$item->timestamp = $date;
$item->author = "Maïa Mazaurette";
@ -50,9 +50,9 @@ $replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July',
$item->content = $elementText->innertext;
$this->items[] = $item;
}
}
}
}
}
@ -64,7 +64,7 @@ $replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July',
public function getCacheDuration(){
return 7200; // 2h hours
}
private function correctCase($str) {
$sentences=explode('.', mb_strtolower($str, "UTF-8"));
$str="";
@ -73,7 +73,7 @@ $replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July',
{
//upper case first char
$sentence=ucfirst(trim($sentence));
//append sentence to output
$str=$str.$sep.$sentence;
$sep=". ";

4
bridges/SiliconBridge.php
View file

@ -20,7 +20,7 @@ class SiliconBridge extends BridgeAbstract {
}
$feedUrl = 'http://www.silicon.fr/feed';
$html = $this->file_get_html($feedUrl) or $this->returnServerError('Could not request Silicon: '.$feedUrl);
$html = $this->getSimpleHTMLDOM($feedUrl) or $this->returnServerError('Could not request Silicon: '.$feedUrl);
$limit = 0;
foreach($html->find('item') as $element) {
@ -30,7 +30,7 @@ class SiliconBridge extends BridgeAbstract {
$article_uri = $element->innertext;
$article_uri = substr($article_uri, strpos($article_uri, '<link>') + 6);
$article_uri = substr($article_uri, 0, strpos($article_uri, '</link>'));
$article_html = $this->file_get_html($article_uri) or $this->returnServerError('Could not request Silicon: '.$article_uri);
$article_html = $this->getSimpleHTMLDOM($article_uri) or $this->returnServerError('Could not request Silicon: '.$article_uri);
//Build article contents from corresponding elements
$thumbnailUri = $element->find('enclosure', 0)->url;

2
bridges/StripeAPIChangeLogBridge.php
View file

@ -16,7 +16,7 @@ class StripeAPIChangeLogBridge extends BridgeAbstract{
}
public function collectData(array $param){
$html = $this->file_get_html('https://stripe.com/docs/upgrades')
$html = $this->getSimpleHTMLDOM('https://stripe.com/docs/upgrades')
or $this->returnServerError('No results for Stripe API Changelog');

4
bridges/SuperbWallpapersBridge.php
View file

@ -47,12 +47,12 @@ class SuperbWallpapersBridge extends BridgeAbstract {
// Get last page number
$link = $baseUri.'/'.$this->category.'/9999.html';
$html = $this->file_get_html($link);
$html = $this->getSimpleHTMLDOM($link);
$lastpage = min($html->find('.paging .cpage', 0)->innertext(), ceil($max/36));
for ($page = 1; $page <= $lastpage; $page++) {
$link = $baseUri.'/'.$this->category.'/'.$page.'.html';
$html = $this->file_get_html($link) or $this->returnServerError('No results for this query.');
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('No results for this query.');
foreach($html->find('.wpl .i a') as $element) {
$thumbnail = $element->find('img', 0);

4
bridges/T411Bridge.php
View file

@ -36,7 +36,7 @@ class T411Bridge extends BridgeAbstract {
//Retrieve torrent listing from search results, which does not contain torrent description
$url = $this->uri.'torrents/search/?'.$param['search'].'&order=added&type=desc';
$html = $this->file_get_html($url) or $this->returnServerError('Could not request t411: '.$url);
$html = $this->getSimpleHTMLDOM($url) or $this->returnServerError('Could not request t411: '.$url);
$results = $html->find('table.results', 0);
if (is_null($results))
$this->returnServerError('No results from t411: '.$url);
@ -57,7 +57,7 @@ class T411Bridge extends BridgeAbstract {
$item_date = strtotime($element->find('dd', 0)->plaintext);
//Retrieve full description from torrent page
if ($item_html = $this->file_get_html($item_uri)) {
if ($item_html = $this->getSimpleHTMLDOM($item_uri)) {
//Retrieve data from page contents
$item_desc = $item_html->find('div.description', 0);

6
bridges/TagBoardBridge.php
View file

@ -23,8 +23,8 @@ class TagBoardBridge extends BridgeAbstract{
$html = '';
$this->request = $param['u'];
$link = 'https://post-cache.tagboard.com/search/' .$this->request;
$html = $this->file_get_html($link) or $this->returnServerError('Could not request TagBoard for : ' . $link);
$html = $this->getSimpleHTMLDOM($link) or $this->returnServerError('Could not request TagBoard for : ' . $link);
$parsed_json = json_decode($html);
foreach($parsed_json->{'posts'} as $element) {
@ -49,4 +49,4 @@ class TagBoardBridge extends BridgeAbstract{
return 21600; // 6 hours
}
}

Some files were not shown because too many files have changed in this diff Show more