public void ConsumeLetterThenDigitSequence()
{
CharacterReader r = new CharacterReader("One12 Two &bar; qux");
Assert.AreEqual("One12", r.ConsumeLetterThenDigitSequence());
Assert.AreEqual(' ', r.Consume());
Assert.AreEqual("Two", r.ConsumeLetterThenDigitSequence());
Assert.AreEqual(" &bar; qux", r.ConsumeToEnd());
}
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));
}
}