private SitemapFile ParseSitemapIndex(XmlNode sitemapNode)
{
var result = new SitemapFile();
foreach (XmlNode urlDetail in sitemapNode.ChildNodes)
{
var nodeName = urlDetail.Name.ToLower();
var nodeValue = urlDetail.InnerText;
if (nodeName == "loc")
{
Uri tmpUri;
if (Uri.TryCreate(nodeValue, UriKind.Absolute, out tmpUri))
{
result.Location = tmpUri;
}
}
else if (nodeName == "lastmod")
{
DateTime tmpLastModified;
if (DateTime.TryParse(nodeValue, out tmpLastModified))
{
result.LastModified = tmpLastModified;
}
}
}
return(result);
}
#pragma warning disable CS1998 // Async method lacks 'await' operators and will run synchronously
public async Task <SitemapFile> ParseSitemapAsync(TextReader reader, CancellationToken cancellationToken = default)
#pragma warning restore CS1998 // Async method lacks 'await' operators and will run synchronously
{
var result = new SitemapFile();
XDocument document;
try
{
#if NETSTANDARD2_1
document = await XDocument.LoadAsync(reader, LoadOptions.None, cancellationToken);
#else
document = XDocument.Load(reader, LoadOptions.None);
cancellationToken.ThrowIfCancellationRequested();
#endif
}
catch (XmlException)
{
return(null);
}
foreach (var topNode in document.Elements())
{
var nodeName = topNode.Name.LocalName;
if (nodeName.Equals("urlset", StringComparison.InvariantCultureIgnoreCase))
{
var urls = new List <SitemapEntry>();
foreach (var urlNode in topNode.Elements())
{
cancellationToken.ThrowIfCancellationRequested();
var sitemapEntry = ParseSitemapEntry(urlNode);
urls.Add(sitemapEntry);
}
result.Urls = urls;
}
else if (nodeName.Equals("sitemapindex", StringComparison.InvariantCultureIgnoreCase))
{
var indexedSitemaps = new List <SitemapIndexEntry>();
foreach (var sitemapNode in topNode.Elements())
{
cancellationToken.ThrowIfCancellationRequested();
var indexedSitemap = ParseSitemapIndex(sitemapNode);
indexedSitemaps.Add(indexedSitemap);
}
result.Sitemaps = indexedSitemaps;
}
}
return(result);
}
public SitemapFile ParseSitemap(string rawSitemap)
{
var result = new SitemapFile();
var document = new XmlDocument();
try
{
document.LoadXml(rawSitemap);
}
catch (XmlException)
{
return(null);
}
foreach (XmlNode topNode in document.ChildNodes)
{
if (topNode.Name.ToLower() == "urlset")
{
var urls = new List <SitemapEntry>();
foreach (XmlNode urlNode in topNode.ChildNodes)
{
var sitemapEntry = ParseSitemapEntry(urlNode);
urls.Add(sitemapEntry);
}
result.Urls = urls;
}
else if (topNode.Name.ToLower() == "sitemapindex")
{
var indexedSitemaps = new List <SitemapFile>();
foreach (XmlNode sitemapNode in topNode.ChildNodes)
{
var indexedSitemap = ParseSitemapIndex(sitemapNode);
indexedSitemaps.Add(indexedSitemap);
}
result.Sitemaps = indexedSitemaps;
}
}
return(result);
}