private void MyXmlLexer_LexStateChanged(XmlLexerEvent lexEvent, int startIndex, int len)
{
switch (lexEvent)
{
default:
{
throw new NotSupportedException();
}
case XmlLexerEvent.VisitOpenAngle:
{
//enter new context
}
break;
case XmlLexerEvent.CommentContent:
{
}
break;
case XmlLexerEvent.NamePrefix:
{
//name prefix of
#if DEBUG
string testStr = _textSnapshot.Substring(startIndex, len);
#endif
switch (_parseState)
{
default:
throw new NotSupportedException();
case 0:
_nodeNamePrefix = new TextSpan(startIndex, len);
_hasNodeNamePrefix = true;
break;
case 1:
//attribute part
_attrPrefix = new TextSpan(startIndex, len);
_hasAttrPrefix = true;
break;
case 2: // </a
_nodeNamePrefix = new TextSpan(startIndex, len);
_hasNodeNamePrefix = true;
break;
}
}
break;
case XmlLexerEvent.FromContentPart:
{
//text content of the element
OnTextNode(new TextSpan(startIndex, len));
}
break;
case XmlLexerEvent.AttributeValueAsLiteralString:
{
//assign value and add to parent
//string attrValue = textSnapshot.Substring(startIndex, len);
if (_parseState == 11)
{
//doctype node
//add to its parameter
}
else
{
//add value to current attribute node
_parseState = 1;
OnAttribute(_attrName, new TextSpan(startIndex, len));
}
}
break;
case XmlLexerEvent.Attribute:
{
//create attribute node and wait for its value
_attrName = new TextSpan(startIndex, len);
//string attrName = textSnapshot.Substring(startIndex, len);
}
break;
case XmlLexerEvent.NodeNameOrAttribute:
{
//the lexer dose not store state of element name or attribute name
//so we use parseState to decide here
string name = _textSnapshot.Substring(startIndex, len);
switch (_parseState)
{
case 0:
{
//element name=> create element
if (_currentNodeName != null)
{
OnEnteringElementBody();
_openEltStack.Push(_currentNodeName);
}
_currentNodeName = name;
//enter new node
OnVisitNewElement(new TextSpan(startIndex, len));
_parseState = 1; //enter attribute
_waitingAttrName = null;
}
break;
case 1:
{
//wait for attr value
if (_waitingAttrName != null)
{
//push waiting attr
//create new attribute
//eg. in html
//but this is not valid in Xml
throw new NotSupportedException();
}
_waitingAttrName = name;
}
break;
case 2:
{
//****
//node name after open slash </
//TODO: review here,avoid direct string comparison
if (_currentNodeName == name)
{
OnExitingElementBody();
if (_openEltStack.Count > 0)
{
_waitingAttrName = null;
_currentNodeName = _openEltStack.Pop();
}
_parseState = 3;
}
else
{
//eg. in html
//but this is not valid in Xml
//not match open-close tag
throw new NotSupportedException();
}
}
break;
case 4:
{
//attribute value as id ***
//eg. in Html, but not for general Xml
throw new NotSupportedException();
}
case 10:
{
//eg <!
_parseState = 11;
}
break;
case 11:
{
//comment node
}
break;
default:
{
}
break;
}
}
break;
case XmlLexerEvent.VisitCloseAngle:
{
//close angle of current new node
//enter into its content
if (_parseState == 11)
{
//add doctype to html
}
else
{
}
_waitingAttrName = null;
_parseState = 0;
}
break;
case XmlLexerEvent.VisitAttrAssign:
{
_parseState = 4;
}
break;
case XmlLexerEvent.VisitOpenSlashAngle:
{
_parseState = 2;
}
break;
case XmlLexerEvent.VisitCloseSlashAngle:
{
// />
if (_openEltStack.Count > 0)
{
OnExitingElementBody();
//curTextNode = null;
//curAttr = null;
_waitingAttrName = null;
_currentNodeName = _openEltStack.Pop();
}
_parseState = 0;
}
break;
case XmlLexerEvent.VisitOpenAngleExclimation:
{
_parseState = 10;
}
break;
}
}
void LexStateChanged(HtmlLexerEvent lexEvent, int startIndex, int len)
{
switch (lexEvent)
{
case HtmlLexerEvent.CommentContent:
{
//var commentContent = this.textSnapshot.Copy(startIndex, len);
}
break;
case HtmlLexerEvent.FromContentPart:
{
if (curTextNode == null)
{
curTextNode = _resultHtmlDoc.CreateTextNode(
HtmlDecodeHelper.DecodeHtml(this.textSnapshot, startIndex, len));
if (curHtmlNode != null)
{
curHtmlNode.AddChild(curTextNode);
}
}
else
{
curTextNode.AppendTextContent(HtmlDecodeHelper.DecodeHtml(this.textSnapshot, startIndex, len));
}
}
break;
case HtmlLexerEvent.AttributeValueAsLiteralString:
{
//assign value and add to parent
if (parseState == 11)
{
//document node
//doc
domDocNode.AddParameter(textSnapshot.Substring(startIndex, len));
}
else
{
curAttr.Value = textSnapshot.Substring(startIndex, len);
curHtmlNode.AddAttribute(curAttr);
}
}
break;
case HtmlLexerEvent.Attribute:
{
//create attribute node and wait for its value
string nodename = textSnapshot.Substring(startIndex, len);
curAttr = this._resultHtmlDoc.CreateAttribute(null, nodename);
}
break;
case HtmlLexerEvent.NodeNameOrAttribute:
{
//the lexer dose not store state of element name or attribute name
//so we use parseState to decide here
string name = textSnapshot.Substring(startIndex, len);
switch (parseState)
{
case 0:
{
//create element
DomElement elem = this._resultHtmlDoc.CreateElement(null, name);
if (curHtmlNode != null)
{
curHtmlNode.AddChild(elem);
openEltStack.Push(curHtmlNode);
}
curHtmlNode = elem;
parseState = 1; //attribute
curTextNode = null;
curAttr = null;
waitingAttrName = null;
}
break;
case 1:
{
//wait for attr value
if (waitingAttrName != null)
{
//push waiting attr
curAttr = this._resultHtmlDoc.CreateAttribute(null, waitingAttrName);
curAttr.Value = "";
curHtmlNode.AddAttribute(curAttr);
curAttr = null;
}
waitingAttrName = name;
}
break;
case 2:
{
//****
//node name after open slash
//TODO: review here,avoid direct string comparison
if (curHtmlNode.LocalName == name)
{
if (openEltStack.Count > 0)
{
waitingAttrName = null;
curTextNode = null;
curAttr = null;
curHtmlNode = openEltStack.Pop();
}
parseState = 3;
}
else
{
//if not equal then check if current node need close tag or not
if (HtmlTagMatching.IsSingleTag(curHtmlNode.LocalNameIndex))
{
if (openEltStack.Count > 0)
{
waitingAttrName = null;
curHtmlNode = openEltStack.Pop();
curAttr = null;
curTextNode = null;
}
if (curHtmlNode.LocalName == name)
{
if (openEltStack.Count > 0)
{
curTextNode = null;
curAttr = null;
curHtmlNode = openEltStack.Pop();
waitingAttrName = null;
}
parseState = 3;
}
else
{
//implement err handling here!
throw new NotSupportedException();
}
}
else
{
//implement err handling here!
throw new NotSupportedException();
}
}
}
break;
case 4:
{
//attribute value as id
if (curAttr != null)
{
curAttr.Value = name;
curAttr = null;
parseState = 0;
waitingAttrName = null;
}
else
{
}
}
break;
case 10:
{
//document node
parseState = 11;
//after docnodename , this may be attr of the document node
this.domDocNode = (DomDocumentNode)this._resultHtmlDoc.CreateDocumentNodeElement();
domDocNode.DocNodeName = name;
}
break;
case 11:
{
//doc
domDocNode.AddParameter(name);
}
break;
default:
{
}
break;
}
}
break;
case HtmlLexerEvent.VisitCloseAngle:
{
//close angle of current new node
//enter into its content
if (parseState == 11)
{
//add doctype to html
this._resultHtmlDoc.RootNode.AddChild(this.domDocNode);
domDocNode = null;
}
if (waitingAttrName != null)
{
curAttr = this._resultHtmlDoc.CreateAttribute(null, waitingAttrName);
curAttr.Value = "";
curHtmlNode.AddAttribute(curAttr);
curAttr = null;
}
waitingAttrName = null;
parseState = 0;
curTextNode = null;
curAttr = null;
}
break;
case HtmlLexerEvent.VisitAttrAssign:
{
parseState = 4;
}
break;
case HtmlLexerEvent.VisitOpenSlashAngle:
{
parseState = 2;
}
break;
case HtmlLexerEvent.VisitCloseSlashAngle:
{
if (openEltStack.Count > 0)
{
curTextNode = null;
curAttr = null;
waitingAttrName = null;
curHtmlNode = openEltStack.Pop();
}
parseState = 0;
}
break;
case HtmlLexerEvent.VisitOpenAngleExclimation:
{
//eg. doctype
parseState = 10;
}
break;
default:
{
//1. visit open angle
}
break;
}
}