KumiScript/parser/Lexer.cs

using KumiScript.Interpreter;

namespace KumiScript.Reader
{
    public class Lexer
    {
        Stack<Token> _tokenStack;
        readonly StreamReader _streamReader;

        public Lexer(Stream stream)
        {
            _tokenStack = new Stack<Token>(16);
            _streamReader = new StreamReader(stream);
        }

        public Token NextToken()
        {
            if (_tokenStack.IsEmpty())
                return LexFromStream();

            return _tokenStack.Pop();
        }

        public Token PeekToken()
        {
            Token t = NextToken();
            PushToken(t);
            return t;
        }

        public void PushToken(Token token)
        {
            _tokenStack.Push(token);
            return;
        }

        private Token LexFromStream()
        {
            int c = GetChar();

            while (char.IsWhiteSpace((char) c))
                c = GetChar();
            
            if (c == ';')
            {
                while ((c = GetChar()) != '\n');
                c = GetChar();
            }

            if (c == -1)
                return new EndOfFileToken();
            if (c == '(')
                return new ParenthesisToken(true);
            if (c == ')')
                return new ParenthesisToken(false);
            if (c == '"')
                return LexString();
            if (isspec(c))
                return new SpecialToken((char) c);

            return LexGeneric(c);
        }

        private Token LexString()
        {
            List<char> chars = new List<char>();
            int c = GetChar();

            while (c != '"')
            {
                if (c == '\\')
                    c = GetChar();

                chars.Add((char) c);
                c = GetChar();
                if (c == -1)
                    throw new Exception("Unexpected EOF!");
            }
                        
            return new StringToken(new string(chars.ToArray()));
        }
        
        private Token LexGeneric(int c)
        {
            List<char> chars = new List<char>();
            chars.Add((char) c); //NOTE: The language server is WRONG!  List<char>((char) c) BREAKS! **** you m$

            int p = PeekChar();
            while (!char.IsWhiteSpace((char) p) && !isspecparen(p) && p != -1)
            {
                c = GetChar();
                chars.Add((char) c);
                p = PeekChar();
            }

            return new AtomToken(new string(chars.ToArray()));
        }

        private static bool isspec(int c)
        {
            switch (c)
            {
                case '#':
                    return true;
                case '\'':
                    return true;
                case '\\':
                    return true;
                case '`':
                    return true;
                case ',':
                    return true;
            }

            return false;
        }

        private static bool isspecparen(int c)
        {
            switch (c)
            {
                case '(':
                    return true;
                case ')':
                    return true;
                case '#':
                    return true;
                case '\'':
                    return true;
                case '\\':
                    return true;
                case '`':
                    return true;
                case ',':
                    return true;
            }

            return false;
        }

        private int GetChar()
        {
            return _streamReader.Read();
        }

        private int PeekChar()
        {
            return _streamReader.Peek();
        }
    }
}