cuqmbr/rss-bridge
1
0
Fork 0
mirror of https://github.com/RSS-Bridge/rss-bridge.git synced 2025-04-23 15:06:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert "added a bridge for one Liberation blog that could easily be extended for others"

Browse Source 
This reverts commit 79e4e9fdea.
This commit is contained in:
Riduidel 2014-03-03 11:40:49 +01:00
parent 2d9d1cf1c9
commit b4696e589a
1 changed files with 0 additions and 98 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

98
bridges/Les400Culs.php
View File

@ -1,98 +0,0 @@
<?php
/**
*
* @name Sexactu
* @description Sexactu via rss-bridge
* @update 04/02/2014
*/
define("SEXE", "http://sexes.blogs.liberation.fr");
class Les400Culs extends BridgeAbstract{
public function collectData(array $param){
$html = file_get_html($this->getURI()) or $this->returnError('Could not request '.$this->getURI(), 404);
foreach($html->find('#alpha-inner') as $articles) {
foreach($articles->find('div.entry') as $article) {
$header = $article->find('h3.entry-header a', 0);
$content = $article->find('div.entry-content', 0);
$item = new Item();
$item->title = trim($header->innertext);
$item->uri = $header->href;
$item->name = "Agnès Girard";
// date is stored outside this node !
$dateHeader = $article->prev_sibling();
// http://stackoverflow.com/a/6239199/15619 (strtotime is typical amercian bullshit)
$item->timestamp = DateTime::createFromFormat('d/m/Y', $dateHeader->innertext)->getTimestamp();
$linkForMore = $content->find('p.entry-more-link a',0);
if($linkForMore==null) {
$item->content = $content->innertext;
} else {
$pageAddress = $linkForMore->href;
$articlePage = str_get_html($this->get_cached($linkForMore->href));
if($articlePage==null) {
$item->content = $content->innertext."\n<p>".$linkForMore->outertext."</p>";
} else {
// TODO use some caching there !
$fullContent = $articlePage->find('div.entry-content', 0);
$item->content = $fullContent->innertext;
}
}
$this->items[] = $item;
}
}
}
public function getName(){
return 'Les 400 Culs';
}
public function getURI(){
return SEXE;
}
public function getCacheDuration(){
return 7200; // 2h hours
}
public function getDescription(){
return "La planète sexe, vue et racontée par Agnès Giard. Et par rss-bridge";
}
/**
* Maintain locally cached versions of pages to download to avoid multiple doiwnloads.
* A file name is generated by replacing all "/" by "_", and the file is saved below this bridge cache
* @param url url to cache
* @return content of file as string
*/
public function get_cached($url) {
$simplified_url = str_replace(["http://", "https://", "?", "&"], ["", "", "/", "/"], $url);
$filename = __DIR__ . '/../cache/'."pages/".$simplified_url;
if (substr($filename, -1) == '/') {
$filename = $filename."index.html";
}
if(!file_exists($filename)) {
error_log("we have no local copy of ".$url." Downloading !");
$dir = substr($filename, 0, strrpos($filename, '/'));
if(!is_dir($dir)) {
mkdir($dir, 0777, true);
}
$this->download_remote($url, $filename);
}
return file_get_contents($filename);
}
public function download_remote($url , $save_path) {
$f = fopen( $save_path , 'w+');
$handle = fopen($url , "rb");
while (!feof($handle)) {
$contents = fread($handle, 8192);
fwrite($f , $contents);
}
fclose($handle);
fclose($f);
}
}