private void ParseXml(string xml)
{
XmlDocument doc = new XmlDocument();
doc.LoadXml(DownloadText(xml));
XmlNodeList sitemaps = doc.GetElementsByTagName("sitemap");
if (sitemaps.Count <= 0)
{
// xml doc is a list of urls, not sitemaps
XmlNodeList sites = doc.GetElementsByTagName("url");
foreach (XmlNode node in sites)
{
string urlToAdd = node["loc"].InnerText.Trim();
string lastMod = string.Empty;
if (node["lastmod"] != null)
{
lastMod = node["lastmod"].InnerText.Trim();
}
else if (node["news:news"] != null && node["news:news"]["news:publication_date"] != null)
{
lastMod = node["news:news"]["news:publication_date"].InnerText.Trim();
}
else
{// generic url
AddToUrlQueue(urlToAdd);
}
if (IsRecentUrl(urlToAdd, lastMod))
{
AddToUrlQueue(urlToAdd);
}
}
}
else
{
// xml is sitemaps
foreach (XmlNode node in sitemaps)
{
string xmlToAdd = node["loc"].InnerText.Trim();
string lastMod = node["lastmod"].InnerText.Trim();
if (IsRecentUrl(xmlToAdd, lastMod))
{
connection.AddToXmlQueue(xmlToAdd);
}
}
}
}
public void AddXmlDebug(string url)
{
connection.AddToXmlQueue(url);
connection.ResumeWorkers();
}
