From 281eaacaeb769e9b3a97e81953a4b6c4e8a0b52f Mon Sep 17 00:00:00 2001 From: Dreckiger-Dan Date: Sat, 23 Mar 2019 16:22:44 +0100 Subject: [PATCH] [HeiseBridge] Add new bridge (#744) --- bridges/HeiseBridge.php | 75 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 75 insertions(+) create mode 100644 bridges/HeiseBridge.php diff --git a/bridges/HeiseBridge.php b/bridges/HeiseBridge.php new file mode 100644 index 00000000..1d9d8025 --- /dev/null +++ b/bridges/HeiseBridge.php @@ -0,0 +1,75 @@ + array( + 'name' => 'Category', + 'type' => 'list', + 'values' => array( + 'Alle News' + => 'https://www.heise.de/newsticker/heise-atom.xml', + 'Top-News' + => 'https://www.heise.de/newsticker/heise-top-atom.xml', + 'Internet-Störungen' + => 'https://www.heise.de/netze/netzwerk-tools/imonitor-internet-stoerungen/feed/aktuelle-meldungen/', + 'Alle News von heise Developer' + => 'https://www.heise.de/developer/rss/news-atom.xml' + ) + ), + 'limit' => array( + 'name' => 'Limit', + 'type' => 'number', + 'required' => false, + 'title' => 'Specify number of full articles to return', + 'defaultValue' => 5 + ) + )); + const LIMIT = 5; + + public function collectData() { + $this->collectExpandableDatas( + $this->getInput('category'), + $this->getInput('limit') ?: static::LIMIT + ); + } + + protected function parseItem($feedItem) { + $item = parent::parseItem($feedItem); + $uri = $item['uri']; + + do { + $article = getSimpleHTMLDOMCached($uri) + or returnServerError('Could not open article: ' . $uri); + + $article = defaultLinkTo($article, $uri); + $item = $this->addArticleToItem($item, $article); + + if($next = $article->find('.pagination a[rel="next"]', 0)) + $uri = $next->href; + } while ($next); + + return $item; + } + + private function addArticleToItem($item, $article) { + if($author = $article->find('[itemprop="author"]', 0)) + $item['author'] = $author->plaintext; + + $content = $article->find('div[class*="article-content"]', 0); + + foreach($content->find('p, h3, ul, table, pre, img') as $element) { + $item['content'] .= $element; + } + + foreach($content->find('img') as $img) { + $item['enclosures'][] = $img->src; + } + + return $item; + } +}