private XmlToken TokenizeSimple(string text, XmlTokenKind kind, XmlTokenizerMode nextMode)
{
var token = new XmlToken(kind, text.Length);
position += text.Length;
mode = nextMode;
return(token);
}
private XmlToken TokenizeEntity()
{
Debug.Assert(mode == XmlTokenizerMode.OutsideElement);
Debug.Assert(input[position] == '&');
var token = new XmlToken(XmlTokenKind.Entity, input.IndexOf(';', position) - position);
position += token.Length;
return(token);
}
private XmlToken TokenizeAttributeValue()
{
Debug.Assert(mode == XmlTokenizerMode.AfterAttributeEquals);
var closePosition = input.IndexOf(input[position], position + 1);
var token = new XmlToken(XmlTokenKind.AttributeValue, closePosition + 1 - position);
position = closePosition + 1;
mode = XmlTokenizerMode.InsideElement;
return(token);
}
private XmlToken NextToken()
{
if (position >= input.Length)
{
return(new XmlToken(XmlTokenKind.EOF, 0));
}
XmlToken token;
switch (mode)
{
case XmlTokenizerMode.AfterAttributeEquals:
token = TokenizeAttributeValue();
break;
case XmlTokenizerMode.AfterAttributeName:
token = TokenizeSimple("=", XmlTokenKind.Equals, XmlTokenizerMode.AfterAttributeEquals);
break;
case XmlTokenizerMode.AfterOpen:
token = TokenizeName(XmlTokenKind.ElementName, XmlTokenizerMode.InsideElement);
break;
case XmlTokenizerMode.InsideCData:
token = TokenizeInsideCData();
break;
case XmlTokenizerMode.InsideComment:
token = TokenizeInsideComment();
break;
case XmlTokenizerMode.InsideElement:
token = TokenizeInsideElement();
break;
case XmlTokenizerMode.InsideProcessingInstruction:
token = TokenizeInsideProcessingInstruction();
break;
case XmlTokenizerMode.OutsideElement:
token = TokenizeOutsideElement();
break;
default:
token = new XmlToken(XmlTokenKind.EOF, 0);
throw new Exception("missing case");
}
return(token);
}
private XmlToken TokenizeInsideComment()
{
Debug.Assert(mode == XmlTokenizerMode.InsideComment);
var tokenend = input.IndexOf("-->", position);
if (position == tokenend)
{
position += "-->".Length;
mode = XmlTokenizerMode.OutsideElement;
return(new XmlToken(XmlTokenKind.CommentEnd, "-->".Length));
}
var token = new XmlToken(XmlTokenKind.CommentText, tokenend - position);
position = tokenend;
return(token);
}
private XmlToken TokenizeInsideCData()
{
Debug.Assert(mode == XmlTokenizerMode.InsideCData);
var tokenend = input.IndexOf("]]>", position);
if (position == tokenend)
{
position += "]]>".Length;
mode = XmlTokenizerMode.OutsideElement;
return(new XmlToken(XmlTokenKind.CDataEnd, "]]>".Length));
}
var token = new XmlToken(XmlTokenKind.TextContent, tokenend - position);
position = tokenend;
return(token);
}
private XmlToken TokenizeInsideProcessingInstruction()
{
Debug.Assert(mode == XmlTokenizerMode.InsideProcessingInstruction);
var tokenend = input.IndexOf("?>", position);
if (position == tokenend)
{
position += "?>".Length;
mode = XmlTokenizerMode.OutsideElement;
return(new XmlToken(XmlTokenKind.CloseProcessingInstruction, "?>".Length));
}
var token = new XmlToken(XmlTokenKind.TextContent, tokenend - position);
position = tokenend;
return(token);
}
private XmlToken TokenizeElementWhitespace()
{
int i;
for (i = position; i < input.Length; i++)
{
if (!char.IsWhiteSpace(input[i]))
{
break;
}
}
var token = new XmlToken(XmlTokenKind.ElementWhitespace, i - position);
position = i;
return(token);
}
private XmlToken TokenizeName(XmlTokenKind kind, XmlTokenizerMode nextMode)
{
Debug.Assert(mode == XmlTokenizerMode.AfterOpen || mode == XmlTokenizerMode.InsideElement);
int i;
for (i = position; i < input.Length; i++)
{
if (!IsNameCharacter(input[i]))
{
break;
}
}
var token = new XmlToken(kind, i - position);
mode = nextMode;
position = i;
return(token);
}
//// We break on newlines because that makes it easier for us
//// to ignore the space after comments
//private Token TokenizeWhitespaceContent()
//{
// Debug.Assert(char.IsWhiteSpace(input[position]));
// bool sawNewline = false;
// int i;
// for (i = position; i < input.Length; i++) {
// if (!char.IsWhiteSpace(input[i])) {
// break;
// } else if (input[i] == '\n' || input[i] == '\r') {
// sawNewline = true;
// } else if (sawNewline) {
// break;
// }
// }
// Token token = new Token(TokenKind.WhitespaceContent, i - position);
// position = i;
// return token;
//}
private XmlToken TokenizeText()
{
Debug.Assert(input[position] != '<');
Debug.Assert(input[position] != '&');
Debug.Assert(mode == XmlTokenizerMode.OutsideElement);
int i;
for (i = position; i < input.Length; i++)
{
if (input[i] == '<' || input[i] == '&')
{
break;
}
}
var token = new XmlToken(XmlTokenKind.TextContent, i - position);
position = i;
return(token);
}
