/**
*
* @return \webignition\WebResource\Sitemap\Sitemap
*/
protected function createSitemap()
{
$configuration = new SitemapConfiguration();
$configuration->setTypeToUrlExtractorClassMap(array('sitemaps.org.xml' => 'webignition\\WebResource\\Sitemap\\UrlExtractor\\SitemapsOrgXmlUrlExtractor', 'sitemaps.org.txt' => 'webignition\\WebResource\\Sitemap\\UrlExtractor\\SitemapsOrgTxtUrlExtractor', 'application/atom+xml' => 'webignition\\WebResource\\Sitemap\\UrlExtractor\\NewsFeedUrlExtractor', 'application/rss+xml' => 'webignition\\WebResource\\Sitemap\\UrlExtractor\\NewsFeedUrlExtractor', 'sitemaps.org.xml.index' => 'webignition\\WebResource\\Sitemap\\UrlExtractor\\SitemapsOrgXmlIndexUrlExtractor'));
$sitemap = new Sitemap();
$sitemap->setConfiguration($configuration);
return $sitemap;
}
/**
*
* @return array
*/
public function getUrls()
{
if (is_null($this->urls)) {
$this->urls = array();
$extractorClass = $this->configuration->getExtractorClassForType($this->getType());
if (is_null($extractorClass)) {
return array();
}
$extractor = new $extractorClass();
$urls = $extractor->extract($this->getContent());
foreach ($urls as $url) {
$normalisedUrl = (string) new NormalisedUrl($url);
if (!array_key_exists($normalisedUrl, $this->urls)) {
$this->urls[$normalisedUrl] = true;
}
}
}
return array_keys($this->urls);
}
