public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
string name = r.ConsumeLetterSequence();
t.DataBuffer.Append(name.ToLowerInvariant());
t.Emit(name);
return;
}
char c = r.Consume();
switch (c)
{
case '\t':
case '\n':
case '\r':
case '\f':
case ' ':
case '/':
case '>':
if (t.DataBuffer.ToString().Equals("script"))
{
t.Transition(ScriptDataEscaped);
}
else
{
t.Transition(ScriptDataDoubleEscaped);
}
t.Emit(c);
break;
default:
r.Unconsume();
t.Transition(ScriptDataDoubleEscaped);
break;
}
}
public char?ConsumeCharacterReference(char?additionalAllowedCharacter, bool inAttribute)
{
if (_reader.IsEmpty())
{
return(null);
}
if (additionalAllowedCharacter != null && additionalAllowedCharacter == _reader.Current())
{
return(null);
}
if (_reader.MatchesAny('\t', '\n', '\r', '\f', ' ', '<', '&'))
{
return(null);
}
_reader.Mark();
if (_reader.MatchConsume("#"))
{ // numbered
bool isHexMode = _reader.MatchConsumeIgnoreCase("X");
string numRef = isHexMode ? _reader.ConsumeHexSequence() : _reader.ConsumeDigitSequence();
if (numRef.Length == 0)
{ // didn't match anything
CharacterReferenceError("Numeric reference with no numerals");
_reader.RewindToMark();
return(null);
}
if (!_reader.MatchConsume(";"))
{
CharacterReferenceError("Missing semicolon"); // missing semi
}
int charval = -1;
try
{
int numbase = isHexMode ? 16 : 10;
charval = Convert.ToInt32(numRef, numbase);
}
catch (FormatException)
{
} // skip
if (charval == -1 || (charval >= 0xD800 && charval <= 0xDFFF) || charval > 0x10FFFF)
{
CharacterReferenceError("Character outside of valid range");
return(ReplacementChar);
}
else
{
// todo: implement number replacement table
// todo: check for extra illegal unicode points as parse errors
return((char)charval);
}
}
else
{ // named
// get as many letters as possible, and look for matching entities. unconsume backwards till a match is found
string nameRef = _reader.ConsumeLetterThenDigitSequence();
bool looksLegit = _reader.Matches(';');
// found if a base named entity without a ;, or an extended entity with the ;.
bool found = (Entities.IsBaseNamedEntity(nameRef) || (Entities.IsNamedEntity(nameRef) && looksLegit));
if (!found)
{
_reader.RewindToMark();
if (looksLegit)
{
CharacterReferenceError(string.Format("Invalid named referenece '{0}'", nameRef));
}
return(null);
}
if (inAttribute && (_reader.MatchesLetter() || _reader.MatchesDigit() || _reader.MatchesAny('=', '-', '_')))
{
// don't want that to match
_reader.RewindToMark();
return(null);
}
if (!_reader.MatchConsume(";"))
{
CharacterReferenceError("Missing semicolon"); // missing semi
}
return(Entities.GetCharacterByName(nameRef));
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
t.CreateTagPending(false);
t.TagPending.AppendTagName(char.ToLowerInvariant(r.Current()));
t.DataBuffer.Append(r.Current());
t.AdvanceTransition(ScriptDataEscapedEndTagName);
}
else
{
t.Emit("</");
t.Transition(ScriptDataEscaped);
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
string name = r.ConsumeLetterSequence();
t.TagPending.AppendTagName(name.ToLowerInvariant());
t.DataBuffer.Append(name);
return;
}
if (t.IsAppropriateEndTagToken() && !r.IsEmpty())
{
char c = r.Consume();
switch (c)
{
case '\t':
case '\n':
case '\r':
case '\f':
case ' ':
t.Transition(BeforeAttributeName);
break;
case '/':
t.Transition(SelfClosingStartTag);
break;
case '>':
t.EmitTagPending();
t.Transition(Data);
break;
default:
t.DataBuffer.Append(c);
AnythingElse(t, r);
break;
}
}
else
{
AnythingElse(t, r);
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
t.CreateTagPending(false);
t.Transition(ScriptDataEndTagName);
}
else
{
t.Emit("</");
t.Transition(ScriptData);
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
t.CreateTempBuffer();
t.DataBuffer.Append(char.ToLowerInvariant(r.Current()));
t.Emit("<" + r.Current());
t.AdvanceTransition(ScriptDataDoubleEscapeStart);
}
else if (r.Matches('/'))
{
t.CreateTempBuffer();
t.AdvanceTransition(ScriptDataEscapedEndTagOpen);
}
else
{
t.Emit('<');
t.Transition(ScriptDataEscaped);
}
}
// from < in rcdata
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.Matches('/'))
{
t.CreateTempBuffer();
t.AdvanceTransition(RcDataEndTagOpen);
}
else if (r.MatchesLetter() && !r.ContainsIgnoreCase("</" + t.AppropriateEndTagName()))
{
// diverge from spec: got a start tag, but there's no appropriate end tag (</title>), so rather than
// consuming to EOF; break out here
t.TagPending = new Token.EndTag(t.AppropriateEndTagName());
t.EmitTagPending();
r.Unconsume(); // undo "<"
t.Transition(Data);
}
else
{
t.Emit("<");
t.Transition(RcData);
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.IsEmpty())
{
t.EofError(this);
t.Emit("</");
t.Transition(Data);
}
else if (r.MatchesLetter())
{
t.CreateTagPending(false);
t.Transition(TagName);
}
else if (r.Matches('>'))
{
t.Error(this);
t.AdvanceTransition(Data);
}
else
{
t.Error(this);
t.AdvanceTransition(BogusComment);
}
}
// from < in data
public override void Read(Tokeniser t, CharacterReader r)
{
switch (r.Current())
{
case '!':
t.AdvanceTransition(MarkupDeclarationOpen);
break;
case '/':
t.AdvanceTransition(EndTagOpen);
break;
case '?':
t.AdvanceTransition(BogusComment);
break;
default:
if (r.MatchesLetter())
{
t.CreateTagPending(true);
t.Transition(TagName);
}
else
{
t.Error(this);
t.Emit('<'); // char that got us here
t.Transition(Data);
}
break;
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
string name = r.ConsumeLetterSequence();
t.DoctypePending.Name.Append(name.ToLowerInvariant());
return;
}
char c = r.Consume();
switch (c)
{
case '>':
t.EmitDoctypePending();
t.Transition(Data);
break;
case '\t':
case '\n':
case '\r':
case '\f':
case ' ':
t.Transition(AfterDoctypeName);
break;
case _nullChar:
t.Error(this);
t.DoctypePending.Name.Append(_replacementChar);
break;
case _eof:
t.EofError(this);
t.DoctypePending.ForceQuirks = true;
t.EmitDoctypePending();
t.Transition(Data);
break;
default:
t.DoctypePending.Name.Append(c);
break;
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
if (r.MatchesLetter())
{
t.CreateDoctypePending();
t.Transition(DoctypeName);
return;
}
char c = r.Consume();
switch (c)
{
case '\t':
case '\n':
case '\r':
case '\f':
case ' ':
break; // ignore whitespace
case _nullChar:
t.Error(this);
t.DoctypePending.Name.Append(_replacementChar);
t.Transition(DoctypeName);
break;
case _eof:
t.EofError(this);
t.CreateDoctypePending();
t.DoctypePending.ForceQuirks = true;
t.EmitDoctypePending();
t.Transition(Data);
break;
default:
t.CreateDoctypePending();
t.DoctypePending.Name.Append(c);
t.Transition(DoctypeName);
break;
}
}