public function testStripXmlTag()
{
$data = file_get_contents('tests/fixtures/jeux-linux.fr.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/ezrss.it');
$this->assertEquals('<!DOC', substr(Filter::stripXmlTag($data), 0, 5));
$data = file_get_contents('tests/fixtures/fulltextrss.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/sametmax.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/ibash.ru.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/pcinpact.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/resorts.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/rue89.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/cercle.psy.xml');
$this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
$data = file_get_contents('tests/fixtures/lagrange.xml');
$this->assertEquals('<feed', substr(Filter::stripXmlTag($data), 0, 5));
$data = file_get_contents('tests/fixtures/googleblog.xml');
$this->assertEquals('<feed', substr(trim(Filter::stripXmlTag($data)), 0, 5));
$data = file_get_contents('tests/fixtures/atomsample.xml');
$this->assertEquals('<feed', substr(trim(Filter::stripXmlTag($data)), 0, 5));
$data = file_get_contents('tests/fixtures/planete-jquery.xml');
$this->assertEquals('<rdf:RDF', trim(substr(trim(Filter::stripXmlTag($data)), 0, 8)));
}
/**
* Constructor.
*
* @param string $content Feed content
* @param string $http_encoding HTTP encoding (headers)
* @param string $fallback_url Fallback url when the feed provide relative or broken url
*/
public function __construct($content, $http_encoding = '', $fallback_url = '')
{
$this->fallback_url = $fallback_url;
$xml_encoding = XmlParser::getEncodingFromXmlTag($content);
// Strip XML tag to avoid multiple encoding/decoding in the next XML processing
$this->content = Filter::stripXmlTag($content);
// Encode everything in UTF-8
Logger::setMessage(get_called_class() . ': HTTP Encoding "' . $http_encoding . '" ; XML Encoding "' . $xml_encoding . '"');
$this->content = Encoding::convert($this->content, $xml_encoding ?: $http_encoding);
$this->itemPostProcessor = new ItemPostProcessor($this->config);
$this->itemPostProcessor->register(new ContentGeneratorProcessor($this->config));
$this->itemPostProcessor->register(new ContentFilterProcessor($this->config));
}
/**
* Constructor
*
* @access public
* @param string $content Feed content
* @param string $http_encoding HTTP encoding (headers)
* @param string $fallback_url Fallback url when the feed provide relative or broken url
*/
public function __construct($content, $http_encoding = '', $fallback_url = '')
{
$this->date = new DateParser();
$this->fallback_url = $fallback_url;
$xml_encoding = XmlParser::getEncodingFromXmlTag($content);
// Strip XML tag to avoid multiple encoding/decoding in the next XML processing
$this->content = Filter::stripXmlTag($content);
// Encode everything in UTF-8
Logger::setMessage(get_called_class() . ': HTTP Encoding "' . $http_encoding . '" ; XML Encoding "' . $xml_encoding . '"');
$this->content = Encoding::convert($this->content, $xml_encoding ?: $http_encoding);
// Workarounds
$this->content = Filter::normalizeData($this->content);
}
