public function testRssGrabContent()
{
$reader = new Reader();
$reader->download('http://www.egscomics.com/rss.php');
$parser = $reader->getParser();
$this->assertTrue($parser !== false);
$parser->grabber = true;
$feed = $parser->execute();
$this->assertTrue(is_array($feed->items));
$this->assertTrue(strpos($feed->items[0]->content, '<img') >= 0);
}
public function testFeedsReportedAsNotWorking()
{
$reader = new Reader();
$reader->setContent(file_get_contents('tests/fixtures/cercle.psy.xml'));
$this->assertInstanceOf('PicoFeed\\Parsers\\Rss20', $reader->getParser());
$reader = new Reader();
$reader->setContent(file_get_contents('tests/fixtures/ezrss.it'));
$this->assertInstanceOf('PicoFeed\\Parsers\\Rss20', $reader->getParser());
$reader = new Reader();
$reader->setContent(file_get_contents('tests/fixtures/grotte_barbu.xml'));
$this->assertInstanceOf('PicoFeed\\Parsers\\Rss20', $reader->getParser());
}
protected function absorb(OutputInterface $output, $id, $url)
{
$config = new Config();
$config->setClientUserAgent('Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:11.0) Gecko/20100101 Firefox/11.0');
$reader = new Reader($config);
$reader->download($url);
$parser = $reader->getParser();
if ($parser === false) {
//$this->feedRepository->disableFeed($id);
return $this->writeErrors($output);
}
$feed = $parser->execute();
if ($feed === false) {
//$this->feedRepository->disableFeed($id);
return $this->writeErrors($output);
}
$data = ['lang' => $feed->getLanguage(), 'title' => $feed->getTitle(), 'lastUpdate' => $this->formatDateForMySQL($feed->getDate())];
$this->feedRepository->updateByPk($data, $id);
foreach ($feed->items as $item) {
$url = $item->getUrl();
$output->writeln('+ ' . $item->title);
$fullContent = file_get_contents($url);
$tidy = tidy_parse_string($fullContent, array(), 'UTF8');
$tidy->cleanRepair();
$html = $tidy->value;
$readability = new \Readability($html, $url);
$result = $readability->init();
if ($result) {
$content = $readability->getContent()->innerHTML;
$tidy = tidy_parse_string($content, array('indent' => true, 'show-body-only' => true), 'UTF8');
$tidy->cleanRepair();
$content = $tidy->value;
} else {
$output->writeln('unable to get full content');
$content = $item->getContent();
}
$data = ['feedId' => $id, 'remoteId' => $item->getId(), 'title' => $item->getTitle(), 'url' => $url, 'pubDate' => $this->formatDateForMySQL($item->getDate()), 'content' => $content, 'author' => $item->getAuthor()];
$this->postRepository->add($data, true);
}
}