Signal ReadString(ref Readable <char> @in, out Tokenized @out)
{
int start = @in.Offset;
for (; [email protected]; @in.Move())
{
switch (@in.Peek)
{
//below should do another move(), but checking to see if empty...
case '\\': //BUT! what about "\\", eh??? need to lookahead to know what to do
@in.Move(); //which we can't do, as we might be at end of buffer
break; //if we're at end, then, what? it's like we want to look enter a special mode when we get here
//or - maybe we have to emit a StringPart and start again
case '"':
int length = @in.Offset - start;
@in.Move();
Pop();
return(Emit(out @out, start, length, Token.String));
}
}
//x.Emit(Token.StringPart, 0, i); //but only if i > 1...!
return(Underrun(out @out));
}
Signal ReadNumber(ref Readable <char> @in, out Tokenized @out)
{
int start = @in.Offset;
@in.Move();
for (; [email protected]; @in.Move())
{
if (!IsNumeric(@in.Peek))
{
Pop();
int len = @in.Offset - start;
return(Emit(out @out, start, len, Token.Number));
}
}
return(Underrun(out @out));
}
public Signal Next(ref Readable <char> @in, out Tokenized @out)
{
start:
if (@in.IsEmpty) //should just try reading, surely...
{
return(Underrun(out @out));
}
char current = @in.Peek;
if (_mode != Mode.String && IsWhitespace(current))
{
SkipWhitespace(ref @in);
goto start;
}
//stupid edge case
//of line feed without the carriage return
//requires look-ahead
switch (_mode)
{
case Mode.Line:
if (current == 0) //shouldn't we just, you know, return the signal here...
{
@in.Move();
return(End(out @out));
}
else
{
Push(Mode.LineEnd);
Switch(Mode.Value);
goto start;
}
case Mode.LineEnd:
if (current == 0)
{
@in.Move();
return(End(out @out));
}
throw new NotImplementedException("Handle line break?");
case Mode.Value:
switch (current)
{
case '"':
@in.Move();
Switch(Mode.String);
goto start;
case '{':
@in.Move();
Switch(Mode.Object1);
return(Emit(out @out, 1, 0, Token.Object));
case '[':
@in.Move();
Switch(Mode.Array1);
return(Emit(out @out, 1, 0, Token.Array));
case char c when IsNumeric(c):
return(ReadNumber(ref @in, out @out));
}
break;
case Mode.Object1:
switch (current)
{
case '}':
@in.Move();
Pop();
return(Emit(out @out, 1, 0, Token.ObjectEnd));
case '"':
@in.Move();
Push(Mode.Object2);
Switch(Mode.String);
goto start;
}
break;
case Mode.Object2:
switch (current)
{
case ':':
@in.Move();
Push(Mode.Object3);
Switch(Mode.Value);
goto start;
}
break;
case Mode.Object3:
switch (current)
{
case ',':
@in.Move();
Switch(Mode.Object1);
goto start;
case '}':
@in.Move();
Pop();
return(Emit(out @out, 1, 0, Token.ObjectEnd));
}
break;
case Mode.Array1:
switch (current)
{
case ']':
@in.Move();
Pop();
return(Emit(out @out, 1, 0, Token.ArrayEnd));
default:
Push(Mode.Array2);
Switch(Mode.Value);
goto start;
}
case Mode.Array2:
switch (current)
{
case ']':
@in.Move();
Pop();
return(Emit(out @out, 1, 0, Token.ArrayEnd));
case ',':
@in.Move();
Switch(Mode.Array1);
goto start;
}
break;
case Mode.String:
return(ReadString(ref @in, out @out));
}
return(BadInput(out @out));
}
static Signal Return(out Tokenized @out, Signal status)
{
@out = default;
return(status);
}
static Signal BadInput(out Tokenized @out) => Return(out @out, Signal.BadInput);
static Signal End(out Tokenized @out) => Return(out @out, Signal.End);
static Signal Underrun(out Tokenized @out) => Return(out @out, Signal.Underrun);
=> c == ' '; //more to add!
// Result Ok(out Tokenized @out, int chars = 0) {
// _offset += chars;
// @out = default; //but there needs to be some way of signalling that there's no token returned... BUT why are we even yielding, if there's no token, and no signal?
// return (Status.Ok, chars);
// }
Signal Emit(out Tokenized @out, int before, int length, Token token)
{
@out = new Tokenized(before, length, token);
return(Signal.Ok);
}