/**
* Parse the OPML file.
*
* @return array|false
*/
public function execute()
{
Logger::setMessage(get_called_class() . ': start importation');
$xml = XmlParser::getSimpleXml(trim($this->content));
if ($xml === false || $xml->getName() !== 'opml' || !isset($xml->body)) {
Logger::setMessage(get_called_class() . ': OPML tag not found or malformed XML document');
return false;
}
$this->parseEntries($xml->body);
Logger::setMessage(get_called_class() . ': ' . count($this->items) . ' subscriptions found');
return $this->items;
}
/**
* Prepare stream context.
*
* @return array
*/
private function prepareContext()
{
$context = array('http' => array('method' => 'GET', 'protocol_version' => 1.1, 'timeout' => $this->timeout, 'max_redirects' => $this->max_redirects));
if ($this->proxy_hostname) {
Logger::setMessage(get_called_class() . ' Proxy: ' . $this->proxy_hostname . ':' . $this->proxy_port);
$context['http']['proxy'] = 'tcp://' . $this->proxy_hostname . ':' . $this->proxy_port;
$context['http']['request_fulluri'] = true;
if ($this->proxy_username) {
Logger::setMessage(get_called_class() . ' Proxy credentials: Yes');
} else {
Logger::setMessage(get_called_class() . ' Proxy credentials: No');
}
}
$context['http']['header'] = implode("\r\n", $this->prepareHeaders());
return $context;
}
/**
* Parse HTTP headers.
*
* @static
*
* @param array $lines List of headers
*
* @return array
*/
public static function parse(array $lines)
{
$status = 0;
$headers = array();
foreach ($lines as $line) {
if (strpos($line, 'HTTP/1') === 0) {
$headers = array();
$status = (int) substr($line, 9, 3);
} elseif (strpos($line, ': ') !== false) {
list($name, $value) = explode(': ', $line);
if ($value) {
$headers[trim($name)] = trim($value);
}
}
}
Logger::setMessage(get_called_class() . ' HTTP status code: ' . $status);
foreach ($headers as $name => $value) {
Logger::setMessage(get_called_class() . ' HTTP header: ' . $name . ' => ' . $value);
}
return array($status, new self($headers));
}
/**
* Find feed urls inside a HTML document.
*
* @param string $url Website url
* @param string $html HTML content
*
* @return array List of feed links
*/
public function find($url, $html)
{
Logger::setMessage(get_called_class() . ': Try to discover subscriptions');
$dom = XmlParser::getHtmlDocument($html);
$xpath = new DOMXPath($dom);
$links = array();
$queries = array('//link[@type="application/rss+xml"]', '//link[@type="application/atom+xml"]');
foreach ($queries as $query) {
$nodes = $xpath->query($query);
foreach ($nodes as $node) {
$link = $node->getAttribute('href');
if (!empty($link)) {
$feedUrl = new Url($link);
$siteUrl = new Url($url);
$links[] = $feedUrl->getAbsoluteUrl($feedUrl->isRelativeUrl() ? $siteUrl->getBaseUrl() : '');
}
}
}
Logger::setMessage(get_called_class() . ': ' . implode(', ', $links));
return $links;
}
/**
* Filter HTML for entry content.
*
* @param Feed $feed Feed object
* @param Item $item Item object
*/
public function filterItemContent(Feed $feed, Item $item)
{
if ($this->isFilteringEnabled()) {
$filter = Filter::html($item->getContent(), $feed->getSiteUrl());
$filter->setConfig($this->config);
$item->content = $filter->execute();
} else {
Logger::setMessage(get_called_class() . ': Content filtering disabled');
}
}
/**
* Execute curl context.
*/
private function executeContext()
{
$ch = $this->prepareContext();
curl_exec($ch);
Logger::setMessage(get_called_class() . ' cURL total time: ' . curl_getinfo($ch, CURLINFO_TOTAL_TIME));
Logger::setMessage(get_called_class() . ' cURL dns lookup time: ' . curl_getinfo($ch, CURLINFO_NAMELOOKUP_TIME));
Logger::setMessage(get_called_class() . ' cURL connect time: ' . curl_getinfo($ch, CURLINFO_CONNECT_TIME));
Logger::setMessage(get_called_class() . ' cURL speed download: ' . curl_getinfo($ch, CURLINFO_SPEED_DOWNLOAD));
Logger::setMessage(get_called_class() . ' cURL effective url: ' . curl_getinfo($ch, CURLINFO_EFFECTIVE_URL));
$curl_errno = curl_errno($ch);
if ($curl_errno) {
Logger::setMessage(get_called_class() . ' cURL error: ' . curl_error($ch));
curl_close($ch);
$this->handleError($curl_errno);
}
// Update the url if there where redirects
$this->url = curl_getinfo($ch, CURLINFO_EFFECTIVE_URL);
curl_close($ch);
}
/**
* Normalize encoding and strip head tag.
*/
public function prepareHtml()
{
$html_encoding = XmlParser::getEncodingFromMetaTag($this->html);
$this->html = Encoding::convert($this->html, $html_encoding ?: $this->encoding);
$this->html = Filter::stripHeadTags($this->html);
Logger::setMessage(get_called_class() . ': HTTP Encoding "' . $this->encoding . '" ; HTML Encoding "' . $html_encoding . '"');
}
/**
* Download and check if a resource exists.
*
* @param string $url URL
*
* @return \AsteFeed\Client Client instance
*/
public function download($url)
{
$client = Client::getInstance();
$client->setConfig($this->config);
Logger::setMessage(get_called_class() . ' Download => ' . $url);
try {
$client->execute($url);
} catch (ClientException $e) {
Logger::setMessage(get_called_class() . ' Download Failed => ' . $e->getMessage());
}
return $client;
}
/**
* Load a rule file from the defined folder.
*
* @param string $folder Rule directory
* @param array $files List of possible file names
*
* @return array
*/
public function loadRuleFile($folder, array $files)
{
foreach ($files as $file) {
$filename = $folder . '/' . $file . '.php';
if (file_exists($filename)) {
Logger::setMessage(get_called_class() . ' Load rule: ' . $file);
return include $filename;
}
}
return array();
}
/**
* Return false if the node should not be removed.
*
* @param DomDocument $dom
* @param DomNode $node
*
* @return bool
*/
public function shouldRemove(DomDocument $dom, $node)
{
$document_length = strlen($dom->textContent);
$node_length = strlen($node->textContent);
if ($document_length === 0) {
return true;
}
$ratio = $node_length * 100 / $document_length;
if ($ratio >= 90) {
Logger::setMessage(get_called_class() . ': Should not remove this node (' . $node->nodeName . ') ratio: ' . $ratio . '%');
return false;
}
return true;
}
/**
* Handle not modified response.
*
* @param array $response Client response
*/
public function handleNotModifiedResponse(array $response)
{
if ($response['status'] == 304) {
$this->is_modified = false;
} elseif ($response['status'] == 200) {
$this->is_modified = $this->hasBeenModified($response, $this->etag, $this->last_modified);
$this->etag = $this->getHeader($response, 'ETag');
$this->last_modified = $this->getHeader($response, 'Last-Modified');
}
if ($this->is_modified === false) {
Logger::setMessage(get_called_class() . ' Resource not modified');
}
}