[ 'name' => 'Feed sub-URL', 'type' => 'text', 'required' => true, 'title' => 'Select the sub-feed (see https://www1.folha.uol.com.br/feed/)', 'exampleValue' => 'emcimadahora/rss091.xml', ], 'amount' => [ 'name' => 'Amount of items to fetch', 'type' => 'number', 'defaultValue' => 15, ], 'deep_crawl' => [ 'name' => 'Deep Crawl', 'description' => 'Crawl each item "deeply", that is, return the article contents', 'type' => 'checkbox', 'defaultValue' => true, ], ] ]; protected function parseItem($item) { $item = parent::parseItem($item); if ($this->getInput('deep_crawl')) { $articleHTMLContent = getSimpleHTMLDOMCached($item['uri']); if ($articleHTMLContent) { foreach ($articleHTMLContent->find('div.c-news__body .is-hidden') as $toRemove) { $toRemove->innertext = ''; } $item_content = $articleHTMLContent->find('div.c-news__body', 0); if ($item_content) { $text = $item_content->innertext; $text = strip_tags($text, '

'); $item['content'] = $text; $item['uri'] = explode('*', $item['uri'])[1]; } } else { Debug::log('???: ' . $item['uri']); } } else { $item['uri'] = explode('*', $item['uri'])[1]; } return $item; } public function collectData() { $feed_input = $this->getInput('feed'); if (substr($feed_input, 0, strlen(self::URI)) === self::URI) { Debug::log('Input:: ' . $feed_input); $feed_url = $feed_input; } else { /* TODO: prepend `/` if missing */ $feed_url = self::URI . '/' . $this->getInput('feed'); } Debug::log('URL: ' . $feed_url); $limit = $this->getInput('amount'); $this->collectExpandableDatas($feed_url, $limit); } }