1
0
Fork 0
forked from blallo/rss-bridge

Merge pull request #136 from Riduidel/TheOatmeal

[new] The oatmeal
This commit is contained in:
Mitsu 2015-07-30 11:42:21 +02:00
commit ce89ed0fc5
3 changed files with 67 additions and 4 deletions

View file

@ -25,7 +25,7 @@ abstract class RssExpander extends HttpCachingBridgeAbstract{
$this->collect_RSS_2_0_data($rssContent);
}
private function collect_RSS_2_0_data($rssContent) {
protected function collect_RSS_2_0_data($rssContent) {
$rssContent = $rssContent->channel[0];
// $this->message("RSS content is ===========\n".var_export($rssContent, true)."===========");
$this->load_RSS_2_0_feed_data($rssContent);

View file

@ -0,0 +1,62 @@
<?php
/**
*
* @name The Oatmeal
* @description Un petit site de dessins assez rigolos
* @update 20/02/201403/07/2015
*/
require_once 'bridges/RssExpander.php';
define("THE_OATMEAL", "http://theoatmeal.com/");
define("RSS", "http://feeds.feedburner.com/oatmealfeed");
class TheOatmealBridge extends RssExpander{
public function collectData(array $param){
$param['url'] = RSS;
parent::collectData($param);
}
/**
* Since the oatmeal produces a weird RSS feed, I have to fix it by loading the items separatly from the feed infos
*/
protected function collect_RSS_2_0_data($rssContent) {
$rssContent->registerXPathNamespace("dc", "http://purl.org/dc/elements/1.1/");
$rssHeaderContent = $rssContent->channel[0];
// $this->message("RSS content is ===========\n".var_export($rssHeaderContent, true)."===========");
$this->load_RSS_2_0_feed_data($rssHeaderContent);
foreach($rssContent->item as $item) {
$this->message("parsing item ".var_export($item, true));
$this->items[] = $this->parseRSSItem($item);
}
}
protected function parseRSSItem($newsItem) {
$namespaces = $newsItem->getNameSpaces(true);
$dc = $newsItem->children($namespaces['dc']);
$rdf = $newsItem->children($namespaces['rdf']);
$item = new Item();
$item->title = trim($newsItem->title);
$this->message("browsing Oatmeal item ".var_export($newsItem, true));
$item->uri=$newsItem->attributes($namespaces['rdf'])->about;
// now load that uri from cache
$this->message("now loading page ".$item->uri);
$articlePage = str_get_html($this->get_cached($item->uri));
$content = $articlePage->find('#comic', 0);
if($content==null) {
$content = $articlePage->find('#blog');
}
$item->content = $content->innertext;
$this->message("dc content is ".var_export($dc, true));
$item->name = $dc->creator;
$item->timestamp = DateTime::createFromFormat(DateTime::ISO8601, $dc->date)->getTimestamp();
$this->message("writtem by ".$item->name." on ".$item->timestamp);
return $item;
}
public function getCacheDuration(){
return 7200; // 2h hours
}
}

View file

@ -13,7 +13,8 @@ TODO :
date_default_timezone_set('UTC');
error_reporting(0);
//ini_set('display_errors','1'); error_reporting(E_ALL); // For debugging only.
//ini_set('display_errors','1');
//error_reporting(E_ALL); // For debugging only.
// extensions check
if (!extension_loaded('openssl'))
@ -220,7 +221,7 @@ $formats = Format::searchInformation();
<header>
<h1>RSS-Bridge</h1>
<h2>·Reconnecting the Web·</h2>
<h2>·Reconnecting the Web·</h2>
</header>
<?php
$activeFoundBridgeCount = 0;