public List<HtmlElement> GetElementsByTag(string tag, HtmlElement element, List<HtmlElement> list)
{
if (list == null) list = new List<HtmlElement>();
if (element.GetTagName() == tag) list.Add(element);
if (element.ChildCount > 0)
foreach (HtmlElement child in element.Childs)
GetElementsByTag(tag, child, list);
return list;
}
public List<HtmlElement> GetElementsByTag(string tag, string attribute, string value, HtmlElement element)
{
List<HtmlElement> allTags = GetElementsByTag(tag, element, null);
List<HtmlElement> result = new List<HtmlElement>();
foreach (HtmlElement e in allTags)
{
if (e.GetAttribute(attribute) == value) result.Add(e);
}
return result;
}
public HtmlElement GetFirst(string tag, HtmlElement element)
{
if (element.GetTagName() == tag) return element;
if (element.ChildCount > 0)
foreach (HtmlElement child in element.Childs)
{
HtmlElement e = GetFirst(tag, child);
if (e != null) return e;
}
return null; ;
}
public void ParseOld(string html)
{
if (string.IsNullOrEmpty(html)) return;
HtmlElement currentElement = Root;
bool inTag = false;
string buffer = null;
for (int i = 0; i < html.Length; i++)
{
switch (currentElement.GetTagName())
{
case "script":
case "style":
switch (html[i])
{
case '<':
if (html.Substring(i + 1, 7) == "/script" || html.Substring(i + 1, 6) == "/style")
{
if (!string.IsNullOrEmpty(buffer))
currentElement.InnerText += buffer;
buffer = null;
}
else
buffer += html[i];
break;
case '>':
if (buffer.StartsWith("/script") || buffer.StartsWith("/style"))
{
if (buffer.StartsWith("/"))
{
if (currentElement.Parent != null) currentElement = currentElement.Parent;
}
else
{
HtmlElement newElement = new HtmlElement()
{
Tag = buffer
};
newElement.Parent = currentElement;
currentElement = newElement;
}
buffer = null;
}
else
buffer += html[i];
break;
default:
buffer += html[i];
break;
}
break;
default:
switch (html[i])
{
case '<':
if (!string.IsNullOrEmpty(buffer))
{
currentElement.Childs.Add(
new HtmlElement() { InnerText = buffer });
}
buffer = null;
break;
case '>':
if (buffer != null && buffer.StartsWith("!--"))
{
buffer = null;
continue;
}
if (buffer.StartsWith("/") || buffer.EndsWith("/"))
{
if (buffer.EndsWith("/") && buffer.Contains(" ")) //self closing tag
{
HtmlElement newElement = new HtmlElement()
{
Tag = buffer
};
newElement.Parent = currentElement;
currentElement.Childs.Add(newElement);
}
else
{
string tempTag = buffer.Replace("/", "").ToLowerInvariant();
if (currentElement.Parent != null)
{
if (currentElement.GetTagName() == tempTag)
{
currentElement = currentElement.Parent;
}
else if (currentElement.Parent.Parent != null
&& currentElement.Parent.Parent.GetTagName() == tempTag)
{
currentElement = currentElement.Parent.Parent;
}
}
}
}
else
{
HtmlElement newElement = new HtmlElement()
{
Tag = buffer
};
newElement.Parent = currentElement;
currentElement.Childs.Add(newElement);
if (newElement.GetTagName() != "br"
&& newElement.GetTagName() != "meta"
&& newElement.GetTagName() != "img"
&& newElement.GetTagName() != "!doctype")
{
currentElement = newElement;
}
}
buffer = null;
break;
default:
buffer += html[i];
break;
}
break;
}
}
}
//public void Parse2(string html)
//{
// this.html = html;
// state = 0;
// index = 0;
// while (index < html.Length)
// {
// char c = getChar();
// switch (state)
// {
// case 0:
// while (getChar() != '<') { }
// currentTag = new HtmlElement();
// Root.Childs.Add(currentTag);
// state = 1;
// break;
// case 1: //tag start
// if (c == ' ')
// state= 2;
// else if (c == '/')
// state = 3;
// else if (c == '>')
// state = 3;
// else
// currentTag.Tag += c;
// break;
// case 2: //attributes
// default:
// break;
// }
// }
//}
public void Parse(string html)
{
OriginalText = html;
if (string.IsNullOrEmpty(html)) return;
HtmlElement currentElement = Root;
bool inTag = false;
string buffer = null;
for (int i = 0; i < html.Length; i++)
{
switch (html[i])
{
case '<':
//script içinde direkt > karakteri kullanılmasını destekliyoruz.
if ((i + 1) < html.Length && html[i + 1] != '/' && currentElement.GetTagName() == "script")
{
buffer += html[i];
continue;
}
if (!inTag)
{
if (!string.IsNullOrEmpty(buffer))
{
HtmlElement element = new HtmlElement() { InnerText = buffer };
element.Parent = currentElement;
currentElement.Childs.Add(element);
}
inTag = true;
buffer = null;
}
else
buffer += html[i];
break;
case '>':
if (!inTag) //html içinde direkt > karakteri kullanılmasını destekliyoruz. Hatalı kullanım olmasına rağmen browserlarda çalışıyor.
{
buffer += html[i];
continue;
}
HtmlElement element2 = new HtmlElement() { Tag = buffer };
if (element2.Tag.StartsWith("/"))
{
string temp = element2.Tag.Substring(1).ToLowerInvariant();
if (currentElement.GetTagName() == temp)
{
if (currentElement.Parent != null)
currentElement = currentElement.Parent;
}
else if (currentElement.Parent != null
&& currentElement.Parent.GetTagName() == temp)
{
if (currentElement.Parent != null)
currentElement = currentElement.Parent.Parent;
}
else
{
while (currentElement.Parent != null && inlineTags.ContainsKey(currentElement.GetTagName()))
{
currentElement = currentElement.Parent;
}
}
}
else if (element2.GetTagName() != "br"
&& element2.GetTagName() != "meta"
&& element2.GetTagName() != "img"
&& element2.GetTagName() != "hr"
&& element2.GetTagName() != "link"
&& !element2.GetTagName().StartsWith("!--")
&& element2.GetTagName() != "!doctype"
&& !element2.GetTagName().EndsWith("/"))
{
element2.Parent = currentElement;
currentElement.Childs.Add(element2);
currentElement = element2;
}
else
{
element2.Parent = currentElement;
currentElement.Childs.Add(element2);
}
inTag = false;
buffer = null;
break;
default:
buffer += html[i];
break;
}
}
if (!string.IsNullOrEmpty(buffer))
{
HtmlElement lastElement = new HtmlElement();
lastElement.InnerText = buffer;
Root.Childs.Add(lastElement);
}
}