gistfile1.cpp

template <typename T>
struct Lexer : boost::spirit::lex::lexer<T>
{
    Lexer(char const* start) : row(1), rowStart(start)
    {
        using boost::spirit::lex::token_def;
        using boost::phoenix::ref;
        using boost::spirit::lex::_end;

        this->self
            // Whitespace
            // How do you save current line iterator for column?
            = token_def<>("[\n]", Whitespace)[++ref(row), ref(rowStart) = _end]
            | token_def<>("[ \r\t]", Whitespace)

            // Keywords must appear before identifier, or else they get tagged as identifiers
            | token_def<>("var", Var)
            | token_def<>("if", If)
            | token_def<>("else", Else)
            | token_def<>("for", For)
            | token_def<>("while", While)
            | token_def<>("return", Return)

            // Types like keywords must appear before identifier
            | token_def<>("bool", Bool)
            | token_def<>("int8", Int8)
            | token_def<>("int16", Int16)
            | token_def<>("int32", Int32)
            | token_def<>("int64", Int64)
            | token_def<>("string", String)
            | token_def<>("uint8", UInt8)
            | token_def<>("uint16", UInt16)
            | token_def<>("uint32", UInt32)
            | token_def<>("uint64", UInt64)
            | token_def<>("void", Void)

            | token_def<>("[a-zA-Z][a-zA-Z0-9]*", Identifier)

            | token_def<>("0b[0-1]+", BinLiteral)
            | token_def<>("0o[0-7]+", OctLiteral)
            | token_def<>("0x[0-9a-fA-F]+", HexLiteral)
            | token_def<>("(0d)?[0-9]+", DecLiteral)

            | token_def<>("\\\".*\\\"", StringLiteral)

            | token_def<>("\\+", Add)
            | token_def<>("\\-", Subtract)
            | token_def<>("\\*", Multiply)
            | token_def<>("\\/", Divide)
            | token_def<>("<=", LowerEqual)
            | token_def<>("<", Lower)
            | token_def<>(">=", GreaterEqual)
            | token_def<>(">", Greater)
            | token_def<>("==", Equal)
            | token_def<>("!=", NotEqual)
            | token_def<>("=", Assign)

            | token_def<>("\\(", ParenthesisOpen)
            | token_def<>("\\)", ParenthesisClose)
            | token_def<>("\\{", CurlyBraceOpen)
            | token_def<>("\\}", CurlyBraceClose)

            | token_def<>(";", Semicolon)
            | token_def<>(",", Comma);
    }

    size_t row;
    char const* rowStart;
};