/// <summary>
/// Gets corresponding opening tag to the closing tag name.
/// </summary>
/// <param name="tagName"></param>
/// <param name="element"></param>
/// <returns>The opening DOMElement</returns>
private static DOMElement GetOpeningTag(string tagName, DOMElement element)
{
if (element != null)
{
if (element.TagName == tagName)
{
return(element);
}
else
{
return(GetOpeningTag(tagName, element.ParentNode));
}
}
return(null);
}
/// <summary>
/// Builds DOM tree using tokens generated from tokenizer.
/// </summary>
/// <param name="tokens"></param>
/// <returns>List of parent elements</returns>
public static List <DOMElement> BuildTree(List <string> tokens)
{
List <DOMElement> elements = new List <DOMElement>();
List <string> openedTags = new List <string>();
DOMElement parent = null;
for (int i = 0; i < tokens.Count; i++)
{
string token = tokens[i];
string tagName = GetTagName(token);
TagType tagType = GetTagType(token, tagName);
NodeType nodeType = GetNodeType(token, tagName);
if (tagType == TagType.Closing)
{
if (parent != null)
{
// Check if there's a similar tag previously opened with the same name.
var openedTagIndex = openedTags.LastIndexOf(tagName);
if (openedTagIndex != -1)
{
if (parent.TagName == tagName)
{
// When the current parent element has the same tag name, just go level up.
parent = parent.ParentNode;
// Example:
// <a>
// <b>
// </b>
// parent = <a>
// </a>
// parent = null
}
else
{
// Now, we have to find the corresponding opening tag to the current closing tag.
var el = GetOpeningTag(tagName, parent);
if (el != null)
{
// Set current parent to parent of the found element.
parent = el.ParentNode;
}
// Here's an example where it might occur:
// parent = null
// <a>
// <b>
// <c>
// </b>
// parent = <a>
}
// Remove the opening tag from the collection.
openedTags.RemoveAt(openedTagIndex);
}
}
}
else
{
DOMElement element = new DOMElement()
{
NodeType = nodeType,
};
if (parent != null)
{
element.ParentNode = parent;
parent.Children.Add(element);
}
else
{
elements.Add(element);
}
if (nodeType == NodeType.Element)
{
element.TagName = tagName;
// Extracts all attributes from token.
element.Attributes = GetAttributes(token, tagName);
// Collect opened tags, to correctly close other tags.
openedTags.Add(tagName);
if (tagType == TagType.Opening)
{
// Set current parent to currently processed element.
parent = element;
}
}
else if (nodeType == NodeType.Text)
{
element.NodeValue = token;
}
else if (nodeType == NodeType.Comment)
{
element.NodeValue = GetCommentText(token);
}
}
}
return(elements);
}
