2013-04-05 05:34:07 +02:00
|
|
|
<?php
|
|
|
|
|
2013-06-29 03:50:15 +02:00
|
|
|
namespace PicoFeed\Parsers;
|
2013-04-05 05:34:07 +02:00
|
|
|
|
2014-03-17 02:35:57 +01:00
|
|
|
/**
|
|
|
|
* Atom parser
|
|
|
|
*
|
|
|
|
* @author Frederic Guillot
|
|
|
|
* @package parser
|
|
|
|
*/
|
2013-06-29 03:50:15 +02:00
|
|
|
class Atom extends \PicoFeed\Parser
|
2013-04-05 05:34:07 +02:00
|
|
|
{
|
2014-03-17 02:35:57 +01:00
|
|
|
/**
|
|
|
|
* Parse the document
|
|
|
|
*
|
|
|
|
* @access public
|
|
|
|
* @return mixed Atom instance or false
|
|
|
|
*/
|
2013-04-05 05:34:07 +02:00
|
|
|
public function execute()
|
|
|
|
{
|
2013-08-30 01:34:11 +02:00
|
|
|
\PicoFeed\Logging::log(\get_called_class().': begin parsing');
|
|
|
|
|
2013-04-05 05:34:07 +02:00
|
|
|
\libxml_use_internal_errors(true);
|
|
|
|
$xml = \simplexml_load_string($this->content);
|
|
|
|
|
|
|
|
if ($xml === false) {
|
2013-08-30 01:34:11 +02:00
|
|
|
\PicoFeed\Logging::log(\get_called_class().': XML parsing error');
|
2013-07-06 20:29:45 +02:00
|
|
|
\PicoFeed\Logging::log($this->getXmlErrors());
|
2013-04-05 05:34:07 +02:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2014-03-17 02:35:57 +01:00
|
|
|
$this->language = $this->getXmlLang($this->content);
|
2013-04-05 05:34:07 +02:00
|
|
|
$this->url = $this->getUrl($xml);
|
2014-01-12 01:46:30 +01:00
|
|
|
$this->title = $this->stripWhiteSpace((string) $xml->title) ?: $this->url;
|
2013-04-05 05:34:07 +02:00
|
|
|
$this->id = (string) $xml->id;
|
2013-10-24 00:39:21 +02:00
|
|
|
$this->updated = $this->parseDate((string) $xml->updated);
|
2013-04-05 05:34:07 +02:00
|
|
|
$author = (string) $xml->author->name;
|
|
|
|
|
2014-01-12 01:46:30 +01:00
|
|
|
\PicoFeed\Logging::log(\get_called_class().': Title => '.$this->title);
|
|
|
|
\PicoFeed\Logging::log(\get_called_class().': Url => '.$this->url);
|
|
|
|
|
2013-04-05 05:34:07 +02:00
|
|
|
foreach ($xml->entry as $entry) {
|
|
|
|
|
|
|
|
if (isset($entry->author->name)) {
|
2013-07-27 03:00:39 +02:00
|
|
|
$author = (string) $entry->author->name;
|
2013-04-05 05:34:07 +02:00
|
|
|
}
|
|
|
|
|
2013-07-27 03:00:39 +02:00
|
|
|
$id = (string) $entry->id;
|
|
|
|
|
2013-04-05 05:34:07 +02:00
|
|
|
$item = new \StdClass;
|
|
|
|
$item->url = $this->getUrl($entry);
|
2013-12-18 02:56:53 +01:00
|
|
|
$item->id = $this->generateId($id !== $item->url ? $id : $item->url, $this->isExcludedFromId($this->url) ? '' : $this->url);
|
2013-07-27 03:00:39 +02:00
|
|
|
$item->title = $this->stripWhiteSpace((string) $entry->title);
|
2013-10-24 00:39:21 +02:00
|
|
|
$item->updated = $this->parseDate((string) $entry->updated);
|
2013-04-05 05:34:07 +02:00
|
|
|
$item->author = $author;
|
|
|
|
$item->content = $this->filterHtml($this->getContent($entry), $item->url);
|
2014-03-17 02:35:57 +01:00
|
|
|
$item->language = $this->language;
|
2013-04-05 05:34:07 +02:00
|
|
|
|
2013-05-06 03:44:03 +02:00
|
|
|
if (empty($item->title)) $item->title = $item->url;
|
|
|
|
|
2014-02-18 04:04:49 +01:00
|
|
|
// Try to find an enclosure
|
|
|
|
foreach ($entry->link as $link) {
|
|
|
|
if ((string) $link['rel'] === 'enclosure') {
|
|
|
|
$item->enclosure = (string) $link['href'];
|
|
|
|
$item->enclosure_type = (string) $link['type'];
|
|
|
|
|
|
|
|
if (\PicoFeed\Filter::isRelativePath($item->enclosure)) {
|
|
|
|
$item->enclosure = \PicoFeed\Filter::getAbsoluteUrl($item->enclosure, $this->url);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-04-05 05:34:07 +02:00
|
|
|
$this->items[] = $item;
|
|
|
|
}
|
|
|
|
|
2013-08-30 01:34:11 +02:00
|
|
|
\PicoFeed\Logging::log(\get_called_class().': parsing finished ('.count($this->items).' items)');
|
|
|
|
|
2013-04-05 05:34:07 +02:00
|
|
|
return $this;
|
|
|
|
}
|
|
|
|
|
2014-03-17 02:35:57 +01:00
|
|
|
/**
|
|
|
|
* Get the entry content
|
|
|
|
*
|
|
|
|
* @access public
|
|
|
|
* @param SimpleXMLElement $entry XML Entry
|
|
|
|
* @return string
|
|
|
|
*/
|
2013-04-05 05:34:07 +02:00
|
|
|
public function getContent($entry)
|
|
|
|
{
|
|
|
|
if (isset($entry->content) && ! empty($entry->content)) {
|
|
|
|
|
|
|
|
if (count($entry->content->children())) {
|
|
|
|
return (string) $entry->content->asXML();
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
return (string) $entry->content;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else if (isset($entry->summary) && ! empty($entry->summary)) {
|
|
|
|
return (string) $entry->summary;
|
|
|
|
}
|
|
|
|
|
|
|
|
return '';
|
|
|
|
}
|
|
|
|
|
2014-03-17 02:35:57 +01:00
|
|
|
/**
|
|
|
|
* Get the URL from a link tag
|
|
|
|
*
|
|
|
|
* @access public
|
|
|
|
* @param SimpleXMLElement $xml XML tag
|
|
|
|
* @return string
|
|
|
|
*/
|
2013-04-05 05:34:07 +02:00
|
|
|
public function getUrl($xml)
|
|
|
|
{
|
|
|
|
foreach ($xml->link as $link) {
|
2013-04-13 03:09:05 +02:00
|
|
|
if ((string) $link['type'] === 'text/html' || (string) $link['type'] === 'application/xhtml+xml') {
|
2013-04-05 05:34:07 +02:00
|
|
|
return (string) $link['href'];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return (string) $xml->link['href'];
|
|
|
|
}
|
|
|
|
}
|