public Document(string html_)
{
HTML = String.Copy(html_);
Root = new Node("DocumentRoot", null);
Root.setInnerHTML(0, HTML.Length, HTML);
Position = 0;
MaxDepth = 0;
NodeCount = 0;
}
public Node(string name_, Node parent_ = null)
{
Name = name_;
Parent = parent_;
Attributes = new Dictionary<string, string>();
Children = new List<Node>();
TrailingSlash = false;
HTMLDoc = Parent != null ? Parent.HTMLDoc : null;
OpenTagStart = -1;
InnerHTMLStart = -1;
InnerHTMLEnd = -1;
CloseTagEnd = -1;
if (Parent != null)
{
Parent.addChild(this);
}
}
public void addChild(Node tag)
{
Children.Add(tag);
}
public bool Parse()
{
int depth = 0;
Node currentTag = Root;
Stack<string> hierarchy = new Stack<string>();
while (SkipToNextTag())
{
Move();
char c = Peek();
if (c == '/')
{
// Found the closing tag, close the node.
currentTag.InnerHTMLEnd = Position - 1;
Move();
string closeName = ParseTagName();
// NOTE:
// Despite the point of this program being the use of a Stack,
// the Stack is actually unneeded. The way the DOM tree is parsed,
// children are effectively "pushed" and "popped" by changing the
// node currentTag is refrencing.
// As a result, the line below could be written:
// (closeName == currentTag.Name)
//
// This completely eliminates the need for a stack.
if (closeName == hierarchy.Peek())
{
Move();
currentTag.CloseTagEnd = Position;
currentTag = currentTag.Parent;
depth--;
hierarchy.Pop();
}
else
{
Console.WriteLine("Error: tried to close {0} before closing {1}!", closeName, currentTag.Name);
return false;
}
}
else
{
NodeCount++;
depth++;
if (depth > MaxDepth) MaxDepth = depth;
currentTag = new Node(ParseTagName(), currentTag);
if (currentTag.OpenTagStart == -1) currentTag.OpenTagStart = Position - currentTag.Name.Length - 1;
hierarchy.Push(currentTag.Name);
SkipWhitespace();
while (Peek() != '>')
{
if (Peek() == '/')
{
// Handle trailing slashes on tags like <br/> and close the node.
Move();
currentTag.TrailingSlash = true;
currentTag.CloseTagEnd = Position;
currentTag = currentTag.Parent;
depth--;
hierarchy.Pop();
SkipWhitespace();
}
else
{
// Parse Attributes
string attrName = ParseAttributeName();
string attrValue = "";
SkipWhitespace();
if (Peek() == '=')
{
Move();
SkipWhitespace();
attrValue = ParseAttributeValue();
SkipWhitespace();
}
currentTag.addAttribute(attrName, attrValue);
}
}
Move();
currentTag.InnerHTMLStart = Position;
}
}
return true;
}