/// <summary>
///
/// </summary>
/// <param name="fs"></param>
private void ParseLinks(FileStream fs, Response response)
{
try
{
if (File.Exists(response.TempFile) == false)
{
return;
}
// Lets ensure that the first non-blank line has a HTML header!
using (FileStream temp = System.IO.File.OpenRead(response.TempFile))
{
using (LineReader lr = new LineReader(temp, 4096, Encoding.Default))
{
bool process = true;
while (process == true)
{
string line = lr.ReadLine();
if (line == null)
{
return;
}
line = line.Trim();
if (line.Length == 0)
{
continue;
}
bool validHtml = true;
if (line.IndexOf("<html>", StringComparison.InvariantCultureIgnoreCase) == -1)
{
if (line.IndexOf("<!doctype html", StringComparison.InvariantCultureIgnoreCase) == -1)
{
validHtml = false;
}
}
if (validHtml == true)
{
break;
}
else
{
return;
}
}
}
}
using (TextReader tr = File.OpenText(response.TempFile))
{
SimpleHtmlParser parser = new SimpleHtmlParser();
var doc = parser.Parse(tr);
List<string> links = new List<string>();
foreach (System.Xml.XmlElement link in doc.GetElementsByTagName("a"))
{
if (link.Attributes == null)
{
continue;
}
if (link.Attributes["href"] == null)
{
continue;
}
var href = link.Attributes["href"].Value.Trim();
string md5 = Text.ConvertByteArrayToHexString(Security.GenerateMd5Hash(href));
if (md5.ToLower() == "6666cd76f96956469e7be39d750cc7d9")
{
// Ignore "/"
continue;
}
if (links.Contains(md5) == false)
{
links.Add(md5);
woanware.IO.WriteToFileStream(fs, "LINK: " + href + Environment.NewLine);
}
}
}
}
catch (Exception) { }
}
