/// <summary>
/// Utility method to consume reader and unescape entities found within.
/// </summary>
/// <param name="inAttribute"></param>
/// <returns>Unescaped string from reader</returns>
public string UnescapeEntities(bool inAttribute)
{
StringBuilder builder = new StringBuilder();
while (!_reader.IsEmpty())
{
builder.Append(_reader.ConsumeTo('&'));
if (_reader.Matches('&'))
{
_reader.Consume();
char?c = ConsumeCharacterReference(null, inAttribute);
if (c == null)
{
builder.Append('&');
}
else
{
builder.Append(c);
}
}
}
return(builder.ToString());
}
public void consumeLetterSequence()
{
CharacterReader r = new CharacterReader("One &bar; qux");
Assert.AreEqual("One", r.ConsumeLetterSequence());
Assert.AreEqual(" &", r.ConsumeTo("bar;"));
Assert.AreEqual("bar", r.ConsumeLetterSequence());
Assert.AreEqual("; qux", r.ConsumeToEnd());
}
public void consumeToString()
{
CharacterReader r = new CharacterReader("One Two Two Four");
Assert.AreEqual("One ", r.ConsumeTo("Two"));
Assert.AreEqual('T', r.Consume());
Assert.AreEqual("wo ", r.ConsumeTo("Two"));
Assert.AreEqual('T', r.Consume());
Assert.AreEqual("wo Four", r.ConsumeTo("Qux"));
}
public void consumeToChar()
{
CharacterReader r = new CharacterReader("One Two Three");
Assert.AreEqual("One ", r.ConsumeTo('T'));
Assert.AreEqual("", r.ConsumeTo('T')); // on Two
Assert.AreEqual('T', r.Consume());
Assert.AreEqual("wo ", r.ConsumeTo('T'));
Assert.AreEqual('T', r.Consume());
Assert.AreEqual("hree", r.ConsumeTo('T')); // consume to end
}
public void nextIndexOfString()
{
string input = "One Two something Two Three Four";
CharacterReader r = new CharacterReader(input);
Assert.AreEqual(-1, r.NextIndexOf("Foo"));
Assert.AreEqual(4, r.NextIndexOf("Two"));
Assert.AreEqual("One Two ", r.ConsumeTo("something"));
Assert.AreEqual(10, r.NextIndexOf("Two"));
Assert.AreEqual("something Two Three Four", r.ConsumeToEnd());
Assert.AreEqual(-1, r.NextIndexOf("Two"));
}
public void nextIndexOfChar()
{
string input = "blah blah";
CharacterReader r = new CharacterReader(input);
Assert.AreEqual(-1, r.NextIndexOf('x'));
Assert.AreEqual(3, r.NextIndexOf('h'));
String pull = r.ConsumeTo('h');
Assert.AreEqual("bla", pull);
r.Consume();
Assert.AreEqual(2, r.NextIndexOf('l'));
Assert.AreEqual(" blah", r.ConsumeToEnd());
Assert.AreEqual(-1, r.NextIndexOf('x'));
}
public override void Read(Tokeniser t, CharacterReader r)
{
string data = r.ConsumeTo("]]>");
t.Emit(data);
r.MatchConsume("]]>");
t.Transition(Data);
}
public override void Read(Tokeniser t, CharacterReader r)
{
switch (r.Current())
{
case _nullChar:
t.Error(this);
r.Advance();
t.Emit(_replacementChar);
break;
case _eof:
t.Emit(new Token.EOF());
break;
default:
string data = r.ConsumeTo(_nullChar);
t.Emit(data);
break;
}
}
public override void Read(Tokeniser t, CharacterReader r)
{
// todo: handle bogus comment starting from eof. when does that trigger?
// rewind to capture character that lead us here
r.Unconsume();
Token.Comment comment = new Token.Comment();
comment.Data.Append(r.ConsumeTo('>'));
// todo: replace nullChar with replaceChar
t.Emit(comment);
t.AdvanceTransition(Data);
}