private void ResolveWrongClosingTag(DothtmlElementNode element)
{
Debug.Assert(element.IsClosingTag);
var startElement = ElementHierarchy.Peek() as DothtmlElementNode;
Debug.Assert(startElement != null);
Debug.Assert(startElement.FullTagName != element.FullTagName);
while (startElement != null && !startElement.FullTagName.Equals(element.FullTagName, StringComparison.OrdinalIgnoreCase))
{
ElementHierarchy.Pop();
if (HtmlWriter.SelfClosingTags.Contains(startElement.FullTagName))
{
// automatic immediate close of the tag (for <img src="">)
ElementHierarchy.Peek().Content.AddRange(startElement.Content);
startElement.Content.Clear();
}
else if (AutomaticClosingTags.Contains(startElement.FullTagName))
{
// elements than can contain itself like <p> are closed on the first occurance of element with the same name
var sameElementIndex = startElement.Content.FindIndex(a => (a as DothtmlElementNode)?.FullTagName == startElement.FullTagName);
if (sameElementIndex >= 0)
{
var count = startElement.Content.Count - sameElementIndex;
ElementHierarchy.Peek().Content.AddRange(startElement.Content.Skip(sameElementIndex));
startElement.Content.RemoveRange(sameElementIndex, count);
}
}
// otherwise just pop the element
startElement = ElementHierarchy.Peek() as DothtmlElementNode;
}
}
private void ResolveWrongClosingTag(DothtmlElementNode element)
{
Debug.Assert(element.IsClosingTag);
var startElement = ElementHierarchy.Peek() as DothtmlElementNode;
Debug.Assert(startElement != null);
Debug.Assert(startElement.FullTagName != element.FullTagName);
while (startElement != null && !startElement.FullTagName.Equals(element.FullTagName, StringComparison.OrdinalIgnoreCase))
{
ElementHierarchy.Pop();
if (HtmlWriter.IsSelfClosing(startElement.FullTagName))
{
// automatic immediate close of the tag (for <img src="">)
ElementHierarchy.Peek().Content.AddRange(startElement.Content);
startElement.Content.Clear();
startElement.AddWarning("End tag is missing, the element is implicitly self-closed.");
}
else if (AutomaticClosingTags.Contains(startElement.FullTagName))
{
// elements than can contain itself like <p> are closed on the first occurance of element with the same name
var sameElementIndex = startElement.Content.FindIndex(a => (a as DothtmlElementNode)?.FullTagName == startElement.FullTagName);
if (sameElementIndex < 0)
{
startElement.AddWarning($"End tag is missing, the element is implicitly closed with its parent tag or by the end of file.");
}
else if (sameElementIndex >= 0)
{
startElement.AddWarning($"End tag is missing, the element is implicitly closed by following <{startElement.Content[sameElementIndex].As<DothtmlElementNode>()?.FullTagName}> tag.");
startElement.Content[sameElementIndex].AddWarning($"Previous <{startElement.FullTagName}> is implicitly closed here.");
var count = startElement.Content.Count - sameElementIndex;
ElementHierarchy.Peek().Content.AddRange(startElement.Content.Skip(sameElementIndex));
startElement.Content.RemoveRange(sameElementIndex, count);
}
}
else
{
startElement.AddWarning($"End tag is missing, the element is implicitly closed by </{element.FullTagName}>.");
element.AddWarning($"Element <{startElement.FullTagName}> is implicitly closed here.");
}
// otherwise just pop the element
startElement = ElementHierarchy.Peek() as DothtmlElementNode;
}
}
public DothtmlRootNode Parse(List <DothtmlToken> tokens)
{
Root = null;
Tokens = tokens;
CurrentIndex = 0;
ElementHierarchy.Clear();
// read file
var root = new DothtmlRootNode();
root.Tokens.Add(Tokens);
ElementHierarchy.Push(root);
// read content
var doNotAppend = false;
while (Peek() is DothtmlToken token)
{
if (token.Type == DothtmlTokenType.DirectiveStart)
{
// directive
root.Directives.Add(ReadDirective());
doNotAppend = true;
}
else if (token.Type == DothtmlTokenType.OpenTag)
{
// element - check element hierarchy
var element = ReadElement();
if (ElementHierarchy.Any())
{
element.ParentNode = ElementHierarchy.Peek() as DothtmlElementNode;
}
if (!element.IsSelfClosingTag)
{
if (!element.IsClosingTag)
{
// open tag
CurrentElementContent.Add(element);
ElementHierarchy.Push(element);
}
else
{
// close tag
if (ElementHierarchy.Count <= 1)
{
element.AddWarning($"The closing tag '</{element.FullTagName}>' doesn't have a matching opening tag!");
CurrentElementContent.Add(element);
}
else
{
var beginTag = (DothtmlElementNode)ElementHierarchy.Peek();
var beginTagName = beginTag.FullTagName;
if (!beginTagName.Equals(element.FullTagName, StringComparison.OrdinalIgnoreCase))
{
element.AddWarning($"The closing tag '</{element.FullTagName}>' doesn't have a matching opening tag!");
ResolveWrongClosingTag(element);
if (ElementHierarchy.Peek() is DothtmlElementNode newBeginTag && beginTagName != newBeginTag.FullTagName)
{
newBeginTag.CorrespondingEndTag = element;
ElementHierarchy.Pop();
}
else
{
CurrentElementContent.Add(element);
}
}
else
{
ElementHierarchy.Pop();
beginTag.CorrespondingEndTag = element;
}
}
}
}
