LesLibres.org/Rss-Bridge
2
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

[GizmodoBridge] Fix bridge (#1538)

Browse source 
* [GizmodoBridge] Update bridge
This commit is contained in:
Joseph 2020-05-17 18:35:34 +00:00 committed by GitHub
parent 8233497611
commit fa74d3728b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 64 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

80
bridges/GizmodoBridge.php
View file

@ -3,34 +3,78 @@ class GizmodoBridge extends FeedExpander {
const MAINTAINER = 'polopollo'; const MAINTAINER = 'polopollo';
const NAME = 'Gizmodo'; const NAME = 'Gizmodo';
const URI = 'http://gizmodo.com/'; const URI = 'https://gizmodo.com';
const CACHE_TIMEOUT = 1800; // 30min const CACHE_TIMEOUT = 1800; // 30min
const DESCRIPTION = 'Returns the newest posts from Gizmodo (full text).'; const DESCRIPTION = 'Returns the newest posts from Gizmodo.';
protected function parseItem($item) { protected function parseItem($item) {
$item = parent::parseItem($item); $item = parent::parseItem($item);
$articleHTMLContent = getSimpleHTMLDOMCached($item['uri']); $html = getSimpleHTMLDOMCached($item['uri'])
if(!$articleHTMLContent) { or returnServerError('Could not request: ' . $item['uri']);
$text = 'Could not load ' . $item['uri'];
} else {
$text = $articleHTMLContent->find('div.entry-content', 0)->innertext;
foreach($articleHTMLContent->find('pagespeed_iframe') as $element) {
$text .= '<p>link to a iframe (could be a video): <a href="'
. $element->src
. '">'
. $element->src
. '</a></p><br>';
}
$text = strip_tags($text, '<p><b><a><blockquote><img><em>'); $html = defaultLinkTo($html, $this->getURI());
} $this->stripTags($html);
$this->handleFigureTags($html);
$this->handleIframeTags($html);
// Get header image
$image = $html->find('meta[property="og:image"]', 0)->content;
$item['content'] = $html->find('div.js_post-content', 0)->innertext;
// Get categories
$categories = explode(',', $html->find('meta[name="keywords"]', 0)->content);
$item['categories'] = array_map('trim', $categories);
$item['enclosures'][] = $html->find('meta[property="og:image"]', 0)->content;
$item['content'] = $text;
return $item; return $item;
} }
public function collectData() { public function collectData() {
$this->collectExpandableDatas('http://feeds.gawker.com/gizmodo/full'); $this->collectExpandableDatas(self::URI . '/rss', 20);
}
private function stripTags($html) {
foreach ($html->find('aside') as $aside) {
$aside->outertext = '';
}
foreach ($html->find('div.ad-unit') as $div) {
$div->outertext = '';
}
foreach ($html->find('script') as $script) {
$script->outertext = '';
}
}
private function handleFigureTags($html) {
foreach ($html->find('figure') as $index => $figure) {
if (isset($figure->attr['data-id'])) {
$id = $figure->attr['data-id'];
$format = $figure->attr['data-format'];
} else {
$img = $figure->find('img', 0);
$id = $img->attr['data-chomp-id'];
$format = $img->attr['data-format'];
$figure->find('div.img-permalink-sub-wrapper', 0)->style = '';
}
$imageUrl = 'https://i.kinja-img.com/gawker-media/image/upload/' . $id . '.' . $format;
$figure->find('span', 0)->outertext = <<<EOD
<img src="{$imageUrl}">
EOD;
}
}
private function handleIframeTags($html) {
foreach($html->find('iframe') as $iframe) {
$iframe->src = urljoin($this->getURI(), $iframe->src);
}
} }
} }