public static List <string> Extract_Tag_with_AngleSharp(string tagName, string source)
{
numberofElement = 0;
//prepare nodename and its attributes
List <string[]> res = TagProcessing.NodenameAndAttributes(tagName);
List <string> list_sonuc = new List <string>();
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
var parser = new HtmlParser();
var document = parser.Parse(source);
stopwatch.Stop();
preProcessTime = stopwatch.Elapsed.TotalMilliseconds;
stopwatch.Restart();
//Do something with LINQ
if (document != null)
{
numberofElement = document.All.Length;
string[] nodename = res[0];
List <AngleSharp.Dom.IElement> temp = document.All.Where(m => m.LocalName == nodename[0]).ToList();
for (int i = 1; i < res.Count; i++)
{
string[] att = res[i];
temp = temp.Where(m => m.Attributes[att[0]] != null && m.Attributes[att[0]].Value == att[1]).ToList();
}
if (temp != null)
{
foreach (AngleSharp.Dom.IElement node in temp)
{
list_sonuc.Add(node.InnerHtml);
}
}
else
{
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(null);
}
}
else
{
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(null);
}
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(list_sonuc);
}
public static List <string> Extract_Tag_with_HAP(string tagname, string source)
{
//prepare xpath
string tagname_xpath = TagProcessing.ToXPath(tagname);
List <string> list_sonuc = new List <string>();
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
HtmlDocument htmlDoc = new HtmlDocument();
// There are various options, set as needed
htmlDoc.OptionFixNestedTags = true;
// filePath is a path to a file containing the html
htmlDoc.LoadHtml(source);
stopwatch.Stop();
preProcessTime = stopwatch.Elapsed.TotalMilliseconds;
stopwatch.Restart();
HtmlNodeCollection _htc = htmlDoc.DocumentNode.SelectNodes(tagname_xpath);
if (_htc != null)
{
foreach (HtmlNode node in _htc)
{
list_sonuc.Add(node.InnerHtml);
}
}
else
{
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(null);
}
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(list_sonuc);
}
public static List <string> Extract_Tag_with_IHTMLDocument(string tagName, string source)
{
//prepare nodename and its attributes
List <string[]> res = TagProcessing.NodenameAndAttributes(tagName);
List <string> list_sonuc = new List <string>();
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
HTMLDocument doc = new HTMLDocument();
IHTMLDocument2 doc2 = (IHTMLDocument2)doc;
doc2.clear();
doc2.designMode = "On";
doc2.write(source);
stopwatch.Stop();
preProcessTime = stopwatch.Elapsed.TotalMilliseconds;
stopwatch.Restart();
if (null != doc)
{
string[] nodename = res[0];
nodename[0] = nodename[0].ToUpper(new CultureInfo("en-US", false));
for (int i = 1; i < res.Count; i++)
{
string[] att = res[i];
if (att[0] == "id")
{
//id means only one record
list_sonuc.Add(doc.getElementById(att[1]).innerHTML);
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(list_sonuc);
}
}
foreach (IHTMLElement element in doc.getElementsByTagName(nodename[0]))
{
bool sonuc = true;
for (int i = 1; i < res.Count; i++)
{
string[] att = res[i];
if (att[0] == "class")
{
if (element.className != att[1])
{
sonuc = false;
break;
}
}
else
{
if (element.innerHTML != null)
{
string tag_temp = element.outerHTML.Substring(0, element.outerHTML.IndexOf(">"));
if (!(tag_temp.Contains(att[0]) && tag_temp.Contains(att[1])))
{
sonuc = false;
break;
}
}
else
{
sonuc = false;
break;
}
}
}
if (sonuc)
{
list_sonuc.Add(element.innerHTML);
}
}
}
else
{
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(null);
}
stopwatch.Stop();
searchTime = stopwatch.Elapsed.TotalMilliseconds;
return(list_sonuc);
}
