protected XmlParser(XmlRootState rootState)
{
RootState = rootState;
Context = new XmlParserContext {
CurrentStateLength = 0,
CurrentState = rootState,
PreviousState = rootState,
KeywordBuilder = new StringBuilder(),
Nodes = new NodeStack()
};
Context.Nodes.Push(RootState.CreateDocument());
}
public override XmlParserState PushChar(char c, XmlParserContext context, ref string rollback)
{
System.Diagnostics.Debug.Assert(((XAttribute)context.Nodes.Peek()).Value == null);
if (c == '<')
{
//the parent state should report the error
rollback = string.Empty;
return(Parent);
}
if (context.CurrentStateLength == 0)
{
if (c == '"')
{
context.StateTag = DOUBLEQUOTE;
return(null);
}
if (c == '\'')
{
context.StateTag = SINGLEQUOTE;
return(null);
}
context.StateTag = UNQUOTED;
}
int maskedTag = context.StateTag & TagMask;
if (maskedTag == UNQUOTED)
{
return(BuildUnquotedValue(c, context, ref rollback));
}
if ((c == '"' && maskedTag == DOUBLEQUOTE) || c == '\'' && maskedTag == SINGLEQUOTE)
{
//ending the value
var att = (XAttribute)context.Nodes.Peek();
att.Value = context.KeywordBuilder.ToString();
return(Parent);
}
context.KeywordBuilder.Append(c);
return(null);
}
XmlParserState BuildUnquotedValue(char c, XmlParserContext context, ref string rollback)
{
if (char.IsLetterOrDigit(c) || c == '_' || c == '.')
{
context.KeywordBuilder.Append(c);
return(null);
}
if (context.KeywordBuilder.Length == 0)
{
context.Diagnostics?.LogError($"The value of attribute '{(string)((XAttribute)context.Nodes.Peek ()).Name.FullName}' ended unexpectedly.", context.Position);
rollback = string.Empty;
return(Parent);
}
var att = (XAttribute)context.Nodes.Peek();
att.Value = context.KeywordBuilder.ToString();
rollback = string.Empty;
return(Parent);
}
public override XmlParserState PushChar(char c, XmlParserContext context, ref string rollback)
{
if (context.CurrentStateLength == 0)
{
context.Nodes.Push(new XCData(context.Position - STARTOFFSET));
}
if (c == ']')
{
//make sure we know when there are two ']' chars together
if (context.StateTag == NOMATCH)
{
context.StateTag = SINGLE_BRACKET;
}
else
{
context.StateTag = DOUBLE_BRACKET;
}
}
else if (c == '>' && context.StateTag == DOUBLE_BRACKET)
{
// if the ']]' is followed by a '>', the state has ended
// so attach a node to the DOM and end the state
var cdata = (XCData)context.Nodes.Pop();
cdata.End(context.Position + 1);
if (context.BuildTree)
{
((XContainer)context.Nodes.Peek()).AddChildNode(cdata);
}
return(Parent);
}
else
{
// not any part of a ']]>', so make sure matching is reset
context.StateTag = NOMATCH;
}
return(null);
}
public override XmlParserState PushChar(char c, XmlParserContext context, ref string rollback)
{
if (context.CurrentStateLength == 0)
{
context.Nodes.Push(new XProcessingInstruction(context.Position - QUESTION));
}
if (c == '?')
{
if (context.StateTag == NOMATCH)
{
context.StateTag = QUESTION;
return(null);
}
}
else if (c == '>' && context.StateTag == QUESTION)
{
// if the '?' is followed by a '>', the state has ended
// so attach a node to the DOM and end the state
var xpi = (XProcessingInstruction)context.Nodes.Pop();
// at this point the position isn't incremented to include the '>' yet
// so make sure to include the closing '>' in the span
xpi.End(context.Position + 1);
if (context.BuildTree)
{
((XContainer)context.Nodes.Peek()).AddChildNode(xpi);
}
return(Parent);
}
else
{
context.StateTag = NOMATCH;
}
return(null);
}
public static char?GetDelimiterChar(XmlParserContext context)
=> context.CurrentState is XmlAttributeValueState
? (context.StateTag & TagMask) switch
public override XmlParserState PushChar(char c, XmlParserContext context, ref string rollback)
{
var doc = context.Nodes.Peek() as XDocType;
if (doc == null)
{
doc = new XDocType(context.Position - STARTOFFSET);
context.Nodes.Push(doc);
}
if (!doc.RootElement.IsValid)
{
if (XmlChar.IsWhitespace(c))
{
return(null);
}
else if (XmlChar.IsFirstNameChar(c))
{
rollback = "";
return(nameState);
}
}
else if (doc.PublicFpi == null)
{
if (context.StateTag == 0)
{
if (c == 's' || c == 'S')
{
context.StateTag = 1;
return(null);
}
else if (c == 'p' || c == 'P')
{
context.StateTag = -1;
return(null);
}
if (XmlChar.IsWhitespace(c))
{
return(null);
}
}
else if (Math.Abs(context.StateTag) < 6)
{
if (context.StateTag > 0)
{
if ("YSTEM"[context.StateTag - 1] == c || "ystem"[context.StateTag - 1] == c)
{
context.StateTag++;
if (context.StateTag == 6)
{
context.StateTag = 0;
doc.PublicFpi = "";
}
return(null);
}
}
else
{
int absState = Math.Abs(context.StateTag) - 1;
if ("UBLIC"[absState] == c || "ublic"[absState] == c)
{
context.StateTag--;
return(null);
}
}
}
else
{
if (context.KeywordBuilder.Length == 0)
{
if (XmlChar.IsWhitespace(c))
{
return(null);
}
else if (c == '"')
{
context.KeywordBuilder.Append(c);
return(null);
}
}
else
{
if (c == '"')
{
context.KeywordBuilder.Remove(0, 1);
doc.PublicFpi = context.KeywordBuilder.ToString();
context.KeywordBuilder.Length = 0;
context.StateTag = 0;
}
else
{
context.KeywordBuilder.Append(c);
}
return(null);
}
}
}
else if (doc.Uri == null)
{
if (context.KeywordBuilder.Length == 0)
{
if (XmlChar.IsWhitespace(c))
{
return(null);
}
else if (c == '"')
{
context.KeywordBuilder.Append(c);
return(null);
}
}
else
{
if (c == '"')
{
context.KeywordBuilder.Remove(0, 1);
doc.Uri = context.KeywordBuilder.ToString();
context.KeywordBuilder.Length = 0;
}
else
{
context.KeywordBuilder.Append(c);
}
return(null);
}
}
else if (doc.InternalDeclarationRegion.Length == 0)
{
if (XmlChar.IsWhitespace(c))
{
return(null);
}
switch (context.StateTag)
{
case 0:
if (c == '[')
{
doc.InternalDeclarationRegion = new TextSpan(context.Position + 1, 0);
context.StateTag = 1;
return(null);
}
break;
case 1:
if (c == '<')
{
context.StateTag = 2;
return(null);
}
else if (c == ']')
{
context.StateTag = 0;
doc.InternalDeclarationRegion = TextSpan.FromBounds(doc.InternalDeclarationRegion.Start, context.Position);
return(null);
}
break;
case 2:
if (c == '>')
{
context.StateTag = 1;
}
return(null);
default:
throw new InvalidOperationException();
}
}
doc = (XDocType)context.Nodes.Pop();
if (c == '<')
{
rollback = string.Empty;
context.Diagnostics?.LogError("Doctype ended prematurely.", context.Position);
}
else if (c != '>')
{
context.Diagnostics?.LogError("Unexpected character '" + c + "' in doctype.", context.Position);
}
doc.End(context.Position);
if (context.BuildTree)
{
((XContainer)context.Nodes.Peek()).AddChildNode(doc);
}
return(Parent);
}
protected XmlParser(XmlParserContext context, XmlRootState rootState)
{
Context = context;
RootState = rootState;
}
public override XmlParserState PushChar(char c, XmlParserContext context, ref string rollback)
{
var element = context.Nodes.Peek() as XElement;
if (element == null || element.IsEnded)
{
var parent = element;
element = new XElement(context.Position - STARTOFFSET)
{
Parent = parent
};
context.Nodes.Push(element);
if (context.BuildTree)
{
var parentContainer = (XContainer)context.Nodes.Peek(element.IsClosed ? 0 : 1);
parentContainer.AddChildNode(element);
}
}
if (c == '<')
{
if (element.IsNamed)
{
context.Diagnostics?.LogError("Unexpected '<' in tag '" + element.Name.FullName + "'.", context.Position);
Close(element, context, context.Position);
}
else
{
context.Diagnostics?.LogError("Tag has no name.", element.Span.Start);
}
rollback = string.Empty;
return(Parent);
}
Debug.Assert(!element.IsEnded);
if (element.IsClosed && c != '>')
{
if (char.IsWhiteSpace(c))
{
context.Diagnostics?.LogWarning("Unexpected whitespace after '/' in self-closing tag.", context.Position);
return(null);
}
context.Diagnostics?.LogError("Unexpected character '" + c + "' after '/' in self-closing tag.", context.Position);
context.Nodes.Pop();
return(Parent);
}
//if tag closed
if (c == '>')
{
element.HasEndBracket = true;
if (context.StateTag == MAYBE_SELF_CLOSING)
{
element.Close(element);
}
if (!element.IsNamed)
{
context.Diagnostics?.LogError("Tag closed prematurely.", context.Position);
element.End(context.Position + 1);
}
else
{
Close(element, context, context.Position + 1);
}
return(Parent);
}
if (c == '/')
{
context.StateTag = MAYBE_SELF_CLOSING;
return(null);
}
if (context.StateTag == ATTEMPT_RECOVERY)
{
if (XmlChar.IsWhitespace(c))
{
context.StateTag = RECOVERY_FOUND_WHITESPACE;
}
return(null);
}
if (context.StateTag == RECOVERY_FOUND_WHITESPACE)
{
if (!XmlChar.IsFirstNameChar(c))
{
return(null);
}
}
context.StateTag = FREE;
if (context.CurrentStateLength > 0 && XmlChar.IsFirstNameChar(c))
{
rollback = string.Empty;
return(AttributeState);
}
if (!element.IsNamed && XmlChar.IsFirstNameChar(c))
{
rollback = string.Empty;
return(NameState);
}
if (XmlChar.IsWhitespace(c))
{
return(null);
}
context.Diagnostics?.LogError($"Unexpected character '{c}' in tag.", context.Position - 1);
context.StateTag = ATTEMPT_RECOVERY;
return(null);
}
public static bool IsFree(XmlParserContext context) => context.CurrentState is XmlTagState && context.StateTag == FREE;
