public function testToStringWithConvertIdnToUtf8Enabled()
{
$source = 'http://artesan.xn--a-iga.com/';
$expectedUrl = 'http://artesan.ía.com/';
$url = new NormalisedUrl($source);
$url->getConfiguration()->enableConvertIdnToUtf8();
$this->assertEquals($expectedUrl, (string) $url);
}
public function matches($content = null)
{
if (trim($content) == '') {
return false;
}
if ($content != strip_tags($content)) {
return false;
}
$contentLines = explode("\n", $content);
foreach ($contentLines as $contentLine) {
$url = new NormalisedUrl($contentLine);
if (!$url->hasScheme()) {
return false;
}
}
return true;
}
private function derivePass()
{
if (!$this->absoluteUrl->hasPass() && $this->sourceUrl->hasPass()) {
$this->absoluteUrl->setPass($this->sourceUrl->getPass());
}
}
/**
*
* @param array $refDomains
* @return array
*/
private function getExpectedErrorValuesByRefDomain($refDomains = array())
{
$expectedErrorValues = array();
foreach ($this->expectedErrorValues as $errorValueSet) {
$errorValueSetRefUrl = new NormalisedUrl($errorValueSet['ref']);
if (in_array((string) $errorValueSetRefUrl->getHost(), $refDomains)) {
$expectedErrorValues[] = $errorValueSet;
}
}
return $expectedErrorValues;
}
/**
*
* @param NormalisedUrl $url
* @return boolean
*/
private function isUrlDomainToBeIncluded(NormalisedUrl $url)
{
return !in_array($url->getHost(), $this->getConfiguration()->getDomainsToExclude());
}
/**
* Get the URL where we expect to find the robots.txt file
*
* @return string
*/
public function getExpectedRobotsTxtFileUrl()
{
if ($this->getConfiguration()->getRootUrl()->getRoot() == '') {
return (string) $this->rootUrl . self::DEFAULT_SITEMAP_TXT_FILE_NAME;
}
$rootUrl = new NormalisedUrl($this->getConfiguration()->getRootUrl()->getRoot());
$rootUrl->setPath('/' . self::ROBOTS_TXT_FILE_NAME);
return (string) $rootUrl;
}