diff --git a/bridges/FolhaDeSaoPauloBridge.php b/bridges/FolhaDeSaoPauloBridge.php new file mode 100644 index 00000000..acd8d259 --- /dev/null +++ b/bridges/FolhaDeSaoPauloBridge.php @@ -0,0 +1,51 @@ + array( + 'name' => 'Feed sub-URL', + 'type' => 'text', + 'title' => 'Select the sub-feed (see https://www1.folha.uol.com.br/feed/)', + 'exampleValue' => 'emcimadahora/rss091.xml', + ) + ) + ); + + protected function parseItem($item){ + $item = parent::parseItem($item); + + $articleHTMLContent = getSimpleHTMLDOMCached($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; + } + } else { + Debug::log('???: ' . $item['uri']); + } + + 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); + $this->collectExpandableDatas($feed_url); + } +}