// Opens structurig element such as Div or Table etc.
private void OpenStructuringElement(XmlElement htmlElement)
{
// Close all pending inline elements
// All block elements are considered as delimiters for inline elements
// which forces all inline elements to be closed and re-opened in the following
// structural element (if any).
// By doing that we guarantee that all inline elements appear only within most nested blocks
if (HtmlSchema.IsBlockElement(htmlElement.LocalName))
{
while (_openedElements.Count > 0 && HtmlSchema.IsInlineElement
(
_openedElements.Peek()
.LocalName))
{
XmlElement htmlInlineElement = _openedElements.Pop();
InvariantAssert
(
_openedElements.Count > 0,
"OpenStructuringElement: stack of opened elements cannot become empty here");
_pendingInlineElements.Push(CreateElementCopy(htmlInlineElement));
}
}
// Add this block element to its parent
if (_openedElements.Count > 0)
{
XmlElement htmlParent = _openedElements.Peek();
// Check some known block elements for auto-closing (LI and P)
if (HtmlSchema.ClosesOnNextElementStart
(
htmlParent.LocalName,
htmlElement.LocalName))
{
_openedElements.Pop();
htmlParent = _openedElements.Count > 0
? _openedElements.Peek()
: null;
}
if (htmlParent != null)
{
// NOTE:
// Actually we never expect null - it would mean two top-level P or LI (without a parent).
// In such weird case we will loose all paragraphs except the first one...
htmlParent.AppendChild(htmlElement);
}
}
// Push it onto a stack
_openedElements.Push(htmlElement);
}
/// <summary>
/// Parses the stream of html tokens starting
/// from the name of top-level element.
/// Returns XmlElement representing the top-level
/// html element
/// </summary>
private XmlElement ParseHtmlContent()
{
// Create artificial root elelemt to be able to group multiple top-level elements
// We create "html" element which may be a duplicate of real HTML element, which is ok, as HtmlConverter will swallow it painlessly..
XmlElement htmlRootElement = _document.CreateElement
(
"html",
XhtmlNamespace);
OpenStructuringElement(htmlRootElement);
while (_htmlLexicalAnalyzer.NextTokenType != HtmlTokenType.Eof)
{
switch (_htmlLexicalAnalyzer.NextTokenType)
{
case HtmlTokenType.OpeningTagStart:
_htmlLexicalAnalyzer.GetNextTagToken();
if (_htmlLexicalAnalyzer.NextTokenType == HtmlTokenType.Name)
{
string htmlElementName = _htmlLexicalAnalyzer.NextToken.ToLower();
_htmlLexicalAnalyzer.GetNextTagToken();
// Create an element
XmlElement htmlElement = _document.CreateElement
(
htmlElementName,
XhtmlNamespace);
// Parse element attributes
ParseAttributes(htmlElement);
if (_htmlLexicalAnalyzer.NextTokenType == HtmlTokenType.EmptyTagEnd ||
HtmlSchema.IsEmptyElement(htmlElementName))
{
// It is an element without content (because of explicit slash or based on implicit knowledge aboout html)
AddEmptyElement(htmlElement);
}
else if (HtmlSchema.IsInlineElement(htmlElementName))
{
// Elements known as formatting are pushed to some special
// pending stack, which allows them to be transferred
// over block tags - by doing this we convert
// overlapping tags into normal heirarchical element structure.
OpenInlineElement(htmlElement);
}
else if (HtmlSchema.IsBlockElement(htmlElementName) ||
HtmlSchema.IsKnownOpenableElement(htmlElementName))
{
// This includes no-scope elements
OpenStructuringElement(htmlElement);
}
}
break;
case HtmlTokenType.ClosingTagStart:
_htmlLexicalAnalyzer.GetNextTagToken();
if (_htmlLexicalAnalyzer.NextTokenType == HtmlTokenType.Name)
{
string htmlElementName = _htmlLexicalAnalyzer.NextToken.ToLower();
// Skip the name token. Assume that the following token is end of tag,
// but do not check this. If it is not true, we simply ignore one token
// - this is our recovery from bad xml in this case.
_htmlLexicalAnalyzer.GetNextTagToken();
CloseElement(htmlElementName);
}
break;
case HtmlTokenType.Text:
AddTextContent(_htmlLexicalAnalyzer.NextToken);
break;
case HtmlTokenType.Comment:
AddComment(_htmlLexicalAnalyzer.NextToken);
break;
}
_htmlLexicalAnalyzer.GetNextContentToken();
}
// Get rid of the artificial root element
if (htmlRootElement.FirstChild is XmlElement &&
htmlRootElement.FirstChild == htmlRootElement.LastChild &&
htmlRootElement.FirstChild.LocalName.ToLower() == "html")
{
htmlRootElement = (XmlElement)htmlRootElement.FirstChild;
}
return(htmlRootElement);
}