public void ParseHTML(string link)
{
//HtmlWeb web = new HtmlWeb();
var web = new HtmlWeb()
{
PreRequest = request =>
{
// Make any changes to the request object that will be used.
request.AutomaticDecompression = DecompressionMethods.Deflate | DecompressionMethods.GZip;
return(true);
}
};
WebClient w = new WebClient();
WebException except = new WebException();
try
{
string s = w.DownloadString(link);
var htmlDoc = web.Load(link);
var node = htmlDoc.DocumentNode.Descendants("title").FirstOrDefault();
string title = "No Title";
if (node != null)
{
title = node.InnerText;
}
if (title.Equals("Error"))
{
sm.AddErrorToPerformance(link, title);
}
else
{
// Get Date if possible
HtmlNode mdnode = htmlDoc.DocumentNode.SelectSingleNode("//meta[@name='pubdate']");
if (mdnode != null)
{
HtmlAttribute desc;
desc = mdnode.Attributes["content"];
string date = desc.Value;
string[] wordsInTitle = title.Split(' ');
if (date.Length > 8)
{
DateTime d = this.StringToDateTime(date);
// Go word by word and insert into the table
for (int i = 0; i < wordsInTitle.Length; i++)
{
// To LowerCase
string toAdd = wordsInTitle[i].ToLower();
sm.AddLinkToTableStorage(link, toAdd, title, d);
}
}
else
{
// Go word by word and insert into the table
for (int i = 0; i < wordsInTitle.Length; i++)
{
// ToLowerCase
string toAdd = wordsInTitle[i].ToLower();
sm.AddLinkToTableStorage(link, toAdd, title);
}
}
}
sm.AddLinkToPerformance(link, title);
// Extracting Links:
List <string> linksOnPage = this.ExtractAllAHrefTags(htmlDoc);
List <string> addToURLQueue = this.ExamineLinksOnPage(linksOnPage);
this.AddToURLQueue(addToURLQueue);
}
}
catch (WebException e)
{
except = e;
sm.AddErrorToPerformance(e.ToString(), "Catched Error");
}
}
