\section{Lexical Structure}

\subsection{Constants}

Constants available in \Kaya{} include integers, floating point
values, strings, characters and booleans, as shown in Figure \ref{consts}. 

\FFIG{
\begin{tabular}{rcl}
\MC{constant} & ::= & \MC{int} \\
           & $\mid$ & \MC{float} \\
           & $\mid$ & \MC{string} \\
           & $\mid$ & \MC{bool} \\
           & $\mid$ & \MC{char} \\
\\
\MC{int} & ::= & \oneplus{\MC{digit}} \\
\MC{float} & ::= & \oneplus{\MC{digit}} . \oneplus{\MC{digit}} \maybe{\MC{exp}}\\
\MC{string} & ::= & \CD{"}\zeroplus{\MC{ascii}}\CD{"} \\
\MC{char} & ::= & \CD{'}\MC{ascii}\CD{'} \\
\MC{boolean} & ::= & \CD{true} $\mid$ \CD{false} \\
\\
\MC{digit} & ::= & 0 $\mid$ 1 $\mid$ \ldots $\mid$ 9 \\
\MC{exp} & ::= & \CD{e+} \oneplus{\MC{digit}} \\
         & $\mid$ & \CD{e-} \oneplus{\MC{digit}} \\
         & $\mid$ & \CD{e} \oneplus{\MC{digit}} \\
\MC{ascii} & ::= & $\langle$any printable ASCII character$\rangle$ \\
           & $\mid$ & $\langle$an escape sequence$\rangle$ \\
\end{tabular}
}
{Constants}
{consts}

\subsection{Variables}

A variable name is any sequence of characters (other than a reserved
word, see section \ref{reserved}) beginning with a letter or
underscore, and containing only letters, underscores, digits or
apostrophes.

In the BNF, we refer to variable names as \MC{identifier}. Names which
must begin with an upper case letter are referred to as
\MC{uc\_identifier}. Names which must begin with a lower case letter
are referred to as \MC{lc\_identifier}.

\subsection{Operators}

The following tokens are operators:

\begin{tabular}{ccccccc}
\CD{+} &
\CD{-} &
\CD{*} &
\CD{/} &
\CD{\%} &
\CD{**} &
\CD{!} \\
\CD{++} &
\CD{--} &
\CD{+=} &
\CD{-=} &
\CD{*=} &
\CD{/=} \\
\CD{==} &
\CD{!=} &
\CD{<} &
\CD{>} &
\CD{<=} &
\CD{>=} \\
\CD{\&} &
\CD{|} &
\CD{\^} &
\CD{\&\&} &
\CD{||} &
\CD{<<} &
\CD{>>} \\

\end{tabular}

\subsection{Reserved Words}
\label{reserved}

The lexical analyser recognises the following as reserved words, and
hence they cannot be used as type or variable names:

\CD{Bool}, 
\CD{Char}, 
\CD{Exception}, 
\CD{File}, 
\CD{Float}, 
\CD{Int}, 
\CD{Ptr}, 
\CD{String}, 
\CD{Void}, 
\CD{abstract}, 
\CD{break}, 
\CD{case}, 
\CD{catch}, 
\CD{data}, 
\CD{default}, 
\CD{do},
\CD{else}, 
\CD{end}, 
\CD{false}, 
\CD{finally},
\CD{fnid}, 
\CD{foreign}, 
\CD{for}, 
\CD{globals}, 
\CD{if}, 
\CD{import}, 
\CD{include}, 
\CD{in}, 
\CD{lambda},
\CD{module}, 
\CD{of}, 
\CD{pass}, 
\CD{pure}, 
\CD{private},
\CD{public}, 
\CD{repeat}, 
\CD{return}, 
\CD{size}, 
\CD{throw}, 
\CD{trace}, 
\CD{true}, 
\CD{try}, 
\CD{type}, 
\CD{var}, 
\CD{while}.

\subsection{Delimiters}

The following tokens are used as delimiters:

\begin{tabular}{ccccccc}
\CD{\{} &
\CD{\}} &
\CD{(} &
\CD{)} &
\CD{[} &
\CD{]} \\
\CD{.} &
\CD{=} &
\CD{,} &
\CD{|} &
\CD{;} &
\CD{`} &
\CD{->} \\

\end{tabular}

\subsection{Comments}

A single line comment is introduced with \CD{//}, and continues to the
end of a line, e.g.:

\begin{verbatim}
pi = 3.0; // unreasonable approximation
\end{verbatim}

A multi line comment is introduced with \CD{/*} and continues until
\CD{*/}, e.g.:

\begin{verbatim}
/* We should probably use something more accurate than this,
   but I can't remember any more decimal places */
pi = 3.14;
\end{verbatim}

\subsection{Directives}

Directives (special instructions to the compiler) are prefixed with a
\CD{\%} sign. These may be implementation dependent; those required
by all implementations are:

\CD{imported},
\CD{include},
\CD{link},
\CD{VM}.

\noindent
An implementation is free to introduce others as required.