2013-02-18 03:48:21 +01:00
|
|
|
<?php
|
|
|
|
|
|
|
|
namespace PicoFeed;
|
|
|
|
|
|
|
|
class Reader
|
|
|
|
{
|
|
|
|
private $url = '';
|
|
|
|
private $content = '';
|
|
|
|
|
|
|
|
|
|
|
|
public function __construct($content = '')
|
|
|
|
{
|
|
|
|
$this->content = $content;
|
|
|
|
|
|
|
|
return $this;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public function download($url)
|
|
|
|
{
|
|
|
|
if (strpos($url, 'http') !== 0) {
|
|
|
|
|
|
|
|
$url = 'http://'.$url;
|
|
|
|
}
|
|
|
|
|
|
|
|
$this->url = $url;
|
|
|
|
$this->content = @file_get_contents($this->url);
|
|
|
|
|
|
|
|
return $this;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public function getContent()
|
|
|
|
{
|
|
|
|
return $this->content;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public function getUrl()
|
|
|
|
{
|
|
|
|
return $this->url;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2013-02-24 20:03:14 +01:00
|
|
|
public function getFirstTag($data)
|
2013-02-18 03:48:21 +01:00
|
|
|
{
|
2013-02-24 20:03:14 +01:00
|
|
|
if (strpos($data, '<?xml') !== false) {
|
2013-02-18 03:48:21 +01:00
|
|
|
|
2013-02-24 20:03:14 +01:00
|
|
|
$data = substr($data, strrpos($data, '?>') + 2);
|
2013-02-18 03:48:21 +01:00
|
|
|
|
2013-02-24 20:03:14 +01:00
|
|
|
$open_tag = strpos($data, '<');
|
|
|
|
$close_tag = strpos($data, '>');
|
2013-02-18 03:48:21 +01:00
|
|
|
|
2013-02-24 20:03:14 +01:00
|
|
|
return substr($data, $open_tag, $close_tag);
|
2013-02-18 03:48:21 +01:00
|
|
|
}
|
|
|
|
|
2013-02-24 20:03:14 +01:00
|
|
|
return $data;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public function getParser($discover = false)
|
|
|
|
{
|
|
|
|
$first_tag = $this->getFirstTag($this->content);
|
|
|
|
|
|
|
|
if (strpos($first_tag, '<feed ') !== false) {
|
2013-02-18 03:48:21 +01:00
|
|
|
|
|
|
|
return new Atom($this->content);
|
|
|
|
}
|
2013-02-24 20:03:14 +01:00
|
|
|
else if (strpos($first_tag, '<rss ') !== false && strpos($first_tag, 'version="2.0"') !== false) {
|
2013-02-18 03:48:21 +01:00
|
|
|
|
|
|
|
return new Rss20($this->content);
|
2013-02-24 20:03:14 +01:00
|
|
|
}
|
|
|
|
else if ($discover === true) {
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
else if ($this->discover()) {
|
2013-02-18 03:48:21 +01:00
|
|
|
|
2013-02-24 20:03:14 +01:00
|
|
|
return $this->getParser(true);
|
|
|
|
}
|
2013-02-18 03:48:21 +01:00
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
public function discover()
|
|
|
|
{
|
2013-02-24 20:03:14 +01:00
|
|
|
if (! $this->content) {
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2013-02-18 03:48:21 +01:00
|
|
|
\libxml_use_internal_errors(true);
|
|
|
|
|
|
|
|
$dom = new \DOMDocument;
|
|
|
|
$dom->loadHTML($this->content);
|
|
|
|
|
|
|
|
$xpath = new \DOMXPath($dom);
|
|
|
|
|
|
|
|
$queries = array(
|
|
|
|
"//link[@type='application/atom+xml']",
|
|
|
|
"//link[@type='application/rss+xml']"
|
|
|
|
);
|
|
|
|
|
|
|
|
foreach ($queries as $query) {
|
|
|
|
|
|
|
|
$nodes = $xpath->query($query);
|
|
|
|
|
|
|
|
if ($nodes->length !== 0) {
|
|
|
|
|
|
|
|
$link = $nodes->item(0)->getAttribute('href');
|
|
|
|
|
|
|
|
// Relative links
|
|
|
|
if (strpos($link, 'http') !== 0) {
|
|
|
|
|
|
|
|
if ($link{0} === '/') $link = substr($link, 1);
|
|
|
|
if ($this->url{strlen($this->url) - 1} !== '/') $this->url .= '/';
|
|
|
|
|
|
|
|
$link = $this->url.$link;
|
|
|
|
}
|
|
|
|
|
|
|
|
$this->download($link);
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|