/**
* Construct the final URL from location headers.
*
* @param array $headers List of HTTP response header
*/
private function setEffectiveUrl($headers)
{
foreach ($headers as $header) {
if (stripos($header, 'Location') === 0) {
list(, $value) = explode(': ', $header);
$this->url = Url::resolve($value, $this->url);
}
}
}
/**
* 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;
}
/**
* Check if the item url is correct.
*
* @param Feed $feed Feed object
* @param Item $item Item object
*/
public function checkItemUrl(Feed $feed, Item $item)
{
$item->url = Url::resolve($item->getUrl(), $feed->getSiteUrl());
}
/**
* Handle manually redirections when there is an open base dir restriction.
*
* @param string $location Redirected URL
*
* @return array
*/
private function handleRedirection($location)
{
$nb_redirects = 0;
$result = array();
$this->url = Url::resolve($location, $this->url);
$this->body = '';
$this->body_length = 0;
$this->response_headers = array();
$this->response_headers_count = 0;
while (true) {
++$nb_redirects;
if ($nb_redirects >= $this->max_redirects) {
throw new MaxRedirectException('Maximum number of redirections reached');
}
$result = $this->doRequest(false);
if ($this->isRedirection($result['status'])) {
$this->url = Url::resolve($result['headers']['Location'], $this->url);
$this->body = '';
$this->body_length = 0;
$this->response_headers = array();
$this->response_headers_count = 0;
} else {
break;
}
}
return $result;
}
/**
* Find the item enclosure.
*
* @param SimpleXMLElement $entry Feed item
* @param \AsteFeed\Parser\Item $item Item object
* @param \AsteFeed\Parser\Feed $feed Feed object
*/
public function findItemEnclosure(SimpleXMLElement $entry, Item $item, Feed $feed)
{
$enclosure = $this->findLink($entry, 'enclosure');
if ($enclosure) {
$item->enclosure_url = Url::resolve((string) $enclosure['href'], $feed->getSiteUrl());
$item->enclosure_type = (string) $enclosure['type'];
}
}
/**
* Get the parser.
*
* @return ParserInterface
*/
public function getParser()
{
$ruleLoader = new RuleLoader($this->config);
$rules = $ruleLoader->getRules($this->url);
if (!empty($rules['grabber'])) {
Logger::setMessage(get_called_class() . ': Parse content with rules');
foreach ($rules['grabber'] as $pattern => $rule) {
$url = new Url($this->url);
$sub_url = $url->getFullPath();
if (preg_match($pattern, $sub_url)) {
Logger::setMessage(get_called_class() . ': Matched url ' . $sub_url);
return new RuleParser($this->html, $rule);
}
}
} elseif ($this->enableCandidateParser) {
Logger::setMessage(get_called_class() . ': Parse content with candidates');
return new CandidateParser($this->html);
}
return;
}
/**
* Get the icon link for a website.
*
* @param string $website_link URL
* @param string $favicon_link optional URL
*
* @return string
*/
public function find($website_link, $favicon_link = '')
{
$website = new Url($website_link);
if ($favicon_link !== '') {
$icons = array($favicon_link);
} else {
$icons = $this->extract($this->download($website->getBaseUrl('/'))->getContent());
$icons[] = $website->getBaseUrl('/favicon.ico');
}
foreach ($icons as $icon_link) {
$icon_link = Url::resolve($icon_link, $website);
$resource = $this->download($icon_link);
$this->content = $resource->getContent();
$this->content_type = $resource->getContentType();
if ($this->content !== '') {
return $icon_link;
} elseif ($favicon_link !== '') {
return $this->find($website_link);
}
}
return '';
}
/**
* Find the item enclosure.
*
* @param SimpleXMLElement $entry Feed item
* @param \AsteFeed\Parser\Item $item Item object
* @param \AsteFeed\Parser\Feed $feed Feed object
*/
public function findItemEnclosure(SimpleXMLElement $entry, Item $item, Feed $feed)
{
if (isset($entry->enclosure)) {
$enclosure_url = XmlParser::getXPathResult($entry, 'feedburner:origEnclosureLink', $this->namespaces) ?: XmlParser::getXPathResult($entry, 'enclosure/@url');
$enclosure_type = XmlParser::getXPathResult($entry, 'enclosure/@type');
$item->enclosure_url = Url::resolve((string) current($enclosure_url), $feed->getSiteUrl());
$item->enclosure_type = (string) current($enclosure_type);
}
}
/**
* Convert all relative links to absolute url.
*
* @param string $tag Tag name
* @param string $attribute Attribute name
* @param string $value Attribute value
*
* @return bool
*/
public function rewriteAbsoluteUrl($tag, $attribute, &$value)
{
if ($this->isResource($attribute)) {
$value = Url::resolve($value, $this->website);
}
return true;
}
/**
* Called after XML parsing.
*
* @param string $content the content that should be filtered
*/
public function filterRules($content)
{
// the constructor should require a config, then this if can be removed
if ($this->config === null) {
$config = new Config();
} else {
$config = $this->config;
}
$loader = new RuleLoader($config);
$rules = $loader->getRules($this->website);
$url = new Url($this->website);
$sub_url = $url->getFullPath();
if (isset($rules['filter'])) {
foreach ($rules['filter'] as $pattern => $rule) {
if (preg_match($pattern, $sub_url)) {
foreach ($rule as $search => $replace) {
$content = preg_replace($search, $replace, $content);
}
}
}
}
return $content;
}