// 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>
/// returns true if the current element closes when the new element, whose name has just been read, starts
/// </summary>
/// <param name="currentElementName">
/// string representing current element name
/// </param>
/// <param name="elementName"></param>
/// string representing name of the next element that will start
internal static bool ClosesOnNextElementStart(string currentElementName, string nextElementName)
{
Debug.Assert(currentElementName == currentElementName.ToLower());
switch (currentElementName)
{
case "colgroup":
return(_htmlElementsClosingColgroup.Contains(nextElementName) && HtmlSchema.IsBlockElement(nextElementName));
case "dd":
return(_htmlElementsClosingDD.Contains(nextElementName) && HtmlSchema.IsBlockElement(nextElementName));
case "dt":
return(_htmlElementsClosingDT.Contains(nextElementName) && HtmlSchema.IsBlockElement(nextElementName));
case "li":
return(_htmlElementsClosingLI.Contains(nextElementName));
case "p":
return(HtmlSchema.IsBlockElement(nextElementName));
case "tbody":
return(_htmlElementsClosingTbody.Contains(nextElementName));
case "tfoot":
return(_htmlElementsClosingTfoot.Contains(nextElementName));
case "thead":
return(_htmlElementsClosingThead.Contains(nextElementName));
case "tr":
return(_htmlElementsClosingTR.Contains(nextElementName));
case "td":
return(_htmlElementsClosingTD.Contains(nextElementName));
case "th":
return(_htmlElementsClosingTH.Contains(nextElementName));
}
return(false);
}
/// <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 the real HTML element (a duplicate HTML element
// is ok since HtmlConverter will process it without a problem).
XmlElement htmlRootElement = _document.CreateElement("html", XhtmlNamespace);
OpenStructuringElement(htmlRootElement);
while (_htmlLexicalAnalyzer.NextTokenType != HtmlTokenType.EOF)
{
if (_htmlLexicalAnalyzer.NextTokenType == 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);
}
else
{
// Do nothing. Skip the whole opening tag.
// Ignoring all unknown elements on their start tags.
// Thus we will ignore them on closinng tag as well.
// Anyway we don't know what to do withthem on conversion to Xaml.
}
}
else
{
// Note that the token following opening angle bracket must be a name - lexical analyzer must guarantee that.
// Otherwise - we skip the angle bracket and continue parsing the content as if it is just text.
// TODO: Add the following asserion here, right? or output "<" as a text run instead?:
// InvariantAssert(false, "Angle bracket without a following name is not expected");
}
}
else if (_htmlLexicalAnalyzer.NextTokenType == 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);
}
}
else if (_htmlLexicalAnalyzer.NextTokenType == HtmlTokenType.Text)
{
AddTextContent(_htmlLexicalAnalyzer.NextToken);
}
else if (_htmlLexicalAnalyzer.NextTokenType == HtmlTokenType.Comment)
{
AddComment(_htmlLexicalAnalyzer.NextToken);
}
_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);
}