/**
* Execute Item Processor
*
* @access public
* @param Feed $feed
* @param Item $item
* @return bool
*/
public function execute(Feed $feed, Item $item)
{
if ($this->config->getContentFiltering(true)) {
$filter = Filter::html($item->getContent(), $feed->getSiteUrl());
$filter->setConfig($this->config);
$item->setContent($filter->execute());
} else {
Logger::setMessage(get_called_class() . ': Content filtering disabled');
}
}
function download($url)
{
if ((bool) Config\get('debug_mode')) {
Logger::enable();
}
$client = Client::getInstance();
$client->setUserAgent(Config\HTTP_USER_AGENT);
$client->enablePassthroughMode();
$client->execute($url);
Config\write_debug();
}
/**
* 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();
}
/**
* 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
*
* @access private
* @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));
}
/**
* Parse the document.
*
* @return \PicoFeed\Parser\Feed
*/
public function execute()
{
Logger::setMessage(get_called_class() . ': begin parsing');
$xml = XmlParser::getSimpleXml($this->content);
if ($xml === false) {
Logger::setMessage(get_called_class() . ': Applying XML workarounds');
$this->content = Filter::normalizeData($this->content);
$xml = XmlParser::getSimpleXml($this->content);
if ($xml === false) {
Logger::setMessage(get_called_class() . ': XML parsing error');
Logger::setMessage(XmlParser::getErrors());
throw new MalformedXmlException('XML parsing error');
}
}
$this->used_namespaces = $xml->getNamespaces(true);
$xml = $this->registerSupportedNamespaces($xml);
$feed = new Feed();
$this->findFeedUrl($xml, $feed);
$this->checkFeedUrl($feed);
$this->findSiteUrl($xml, $feed);
$this->checkSiteUrl($feed);
$this->findFeedTitle($xml, $feed);
$this->findFeedDescription($xml, $feed);
$this->findFeedLanguage($xml, $feed);
$this->findFeedId($xml, $feed);
$this->findFeedDate($xml, $feed);
$this->findFeedLogo($xml, $feed);
$this->findFeedIcon($xml, $feed);
foreach ($this->getItemsTree($xml) as $entry) {
$entry = $this->registerSupportedNamespaces($entry);
$item = new Item();
$item->xml = $entry;
$item->namespaces = $this->used_namespaces;
$this->findItemAuthor($xml, $entry, $item);
$this->findItemUrl($entry, $item);
$this->checkItemUrl($feed, $item);
$this->findItemTitle($entry, $item);
$this->findItemContent($entry, $item);
// Id generation can use the item url/title/content (order is important)
$this->findItemId($entry, $item, $feed);
$this->findItemDate($entry, $item, $feed);
$this->findItemEnclosure($entry, $item, $feed);
$this->findItemLanguage($entry, $item, $feed);
$this->itemPostProcessor->execute($feed, $item);
$feed->items[] = $item;
}
Logger::setMessage(get_called_class() . PHP_EOL . $feed);
return $feed;
}
/**
* 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);
}
/**
* 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');
}
}
/**
* 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;
}
function write_debug()
{
if ((bool) get('debug_mode')) {
file_put_contents(DEBUG_FILENAME, implode(PHP_EOL, Logger::getMessages()));
}
}
/**
* Find feed urls inside a HTML document
*
* @access public
* @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;
}
/**
* 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 "' . $html_encoding . '"');
}
function cleanup($feed_id, array $items_in_feed)
{
if (!empty($items_in_feed)) {
$db = Database::getInstance('db');
$removed_items = $db->table('items')->columns('id')->notin('id', $items_in_feed)->eq('status', 'removed')->eq('feed_id', $feed_id)->desc('updated')->findAllByColumn('id');
// Keep a buffer of 2 items
// It's workaround for buggy feeds (cache issue with some Wordpress plugins)
if (is_array($removed_items)) {
$items_to_remove = array_slice($removed_items, 2);
if (!empty($items_to_remove)) {
$nb_items = count($items_to_remove);
Logger::setMessage('There is ' . $nb_items . ' items to remove');
// Handle the case when there is a huge number of items to remove
// Sqlite have a limit of 1000 sql variables by default
// Avoid the error message "too many SQL variables"
// We remove old items by batch of 500 items
$chunks = array_chunk($items_to_remove, 500);
foreach ($chunks as $chunk) {
$db->table('items')->in('id', $chunk)->eq('status', 'removed')->eq('feed_id', $feed_id)->remove();
}
}
}
}
}
/**
* Download and check if a resource exists.
*
* @param string $url
* URL
*
* @return \PicoFeed\Client Client instance
*/
public function download($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;
}
/**
* Constructor.
*
* @param \PicoFeed\Config\Config $config Config class instance
*/
public function __construct(Config $config = null)
{
$this->config = $config ?: new Config();
Logger::setTimezone($this->config->getTimezone());
}
