/**
* Get a parser instance with a custom config
*
* @access public
* @param string $name Parser name
* @return \PicoFeed\Parser
*/
public function getParserInstance($name)
{
require_once __DIR__ . '/Parsers/' . ucfirst($name) . '.php';
$name = '\\PicoFeed\\Parsers\\' . $name;
$parser = new $name($this->content, $this->encoding);
$parser->setHashAlgo($this->config->getParserHashAlgo());
$parser->setTimezone($this->config->getTimezone());
$parser->setConfig($this->config);
return $parser;
}
/**
* Download the HTML content
*
* @access public
* @return HTML content
*/
public function download()
{
$client = Client::getInstance();
if ($this->config !== null) {
$client->setTimeout($this->config->getGrabberTimeout())->setUserAgent($this->config->getGrabberUserAgent())->setMaxRedirections($this->config->getMaxRedirections())->setMaxBodySize($this->config->getMaxBodySize())->setProxyHostname($this->config->getProxyHostname())->setProxyPort($this->config->getProxyPort())->setProxyUsername($this->config->getProxyUsername())->setProxyPassword($this->config->getProxyPassword());
}
$client->execute($this->url);
$this->html = $client->getContent();
$this->encoding = $client->getEncoding();
return $this->html;
}
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);
}
}
/**
* Set config object
*
* @access public
* @param \PicoFeed\Config $config Config instance
* @return \PicoFeed\Parse
*/
public function setConfig($config)
{
$this->config = $config;
if ($this->config !== null) {
$this->setIframeWhitelist($this->config->getFilterIframeWhitelist(array()));
$this->setIntegerAttributes($this->config->getFilterIntegerAttributes(array()));
$this->setAttributeOverrides($this->config->getFilterAttributeOverrides(array()));
$this->setRequiredAttributes($this->config->getFilterRequiredAttributes(array()));
$this->setMediaBlacklist($this->config->getFilterMediaBlacklist(array()));
$this->setMediaAttributes($this->config->getFilterMediaAttributes(array()));
$this->setSchemeWhitelist($this->config->getFilterSchemeWhitelist(array()));
$this->setBlacklistedTags($this->config->getFilterBlacklistedTags(array()));
$this->setWhitelistedTags($this->config->getFilterWhitelistedTags(array()));
}
return $this;
}
/**
* Set config object
*
* @access public
* @param \PicoFeed\Config $config Config instance
* @return \PicoFeed\Client
*/
public function setConfig($config)
{
$this->setTimeout($config->getGrabberTimeout());
$this->setUserAgent($config->getGrabberUserAgent());
$this->setMaxRedirections($config->getMaxRedirections());
$this->setMaxBodySize($config->getMaxBodySize());
$this->setProxyHostname($config->getProxyHostname());
$this->setProxyPort($config->getProxyPort());
$this->setProxyUsername($config->getProxyUsername());
$this->setProxyPassword($config->getProxyPassword());
return $this;
}