\documentclass[a4paper]{article}

\usepackage{skak}

\def\package#1{``\textsf{#1}''}
\def\command#1{``\texttt{\symbol{92}#1}''}
\def\commandwarg#1#2{\texttt{\symbol{92}#1}\{\textit{#2}\}}
\def\base#1{$_{\scriptscriptstyle #1}$}
\def\abbrev#1{\textsf{#1}}
\def\bnf#1{\textit{#1}}
\def\bnfterm#1{\textbf{#1}}
\def\bnfprod{$\rightarrow$}
\def\bnfempty{$\epsilon$}
\def\bnfor{$|$}
\def\bnfjoin{$\bowtie$}

\setlength{\parindent}{0pt}

\title{The Syntax of Chess Moves in the \LaTeX{} Package \package{skak} (Draft)}
\author{Dirk B\"achle}

\begin{document}

\maketitle

\begin{abstract}
This short document contains some thoughts and ideas about the ``to be supported''
syntax of chess moves, including SAN (Short Algebraic Notation) as well as LAN
(Long Algebraic Notation). 

The main purpose of this draft is to specify a concrete set of allowed chess moves.
In the next step, extensive test routines for this syntax should be generated
that can be used to verify the final implementation\ldots
\end{abstract}

\section*{Stage 1: Parsing \command{mainline}/\command{variation}}

The commands \command{mainline} and \command{variation} should accept
a nonemtpy list of space separated ``move tokens'' (\abbrev{MT}):

\medskip
\hfil\commandwarg{mainline}{MT\_list}\hfil or \hfil\commandwarg{variation}{MT\_list}\hfil

where

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MT\_list} \bnfprod\={} \bnf{MT} \bnfterm{space} \bnf{MT\_list}\\
 \>\bnfor{} \bnf{MT}
\end{tabbing}
\end{minipage}
\end{center}

using BNF (Backus-Naur Form) notation with `\bnfterm{space}' as the
terminal symbol for the character '\ ' (ASCII code 
32\base{10}).

\section*{Stage 2: Splitting off move numbers}

Each \abbrev{MT} may be either a chess move (\abbrev{CM}), a move number (\abbrev{MN})
or a token like ``\texttt{2.Kg1}'' that combines both:

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MT} \bnfprod\={} \bnf{MN} \bnf{CM}\\
 \>\bnfor{} \bnf{CM}\\
 \>\bnfor{} \bnf{MN}
\end{tabbing}
\end{minipage}
\end{center}

The only task of this stage is to separate ``combined tokens'' (first rule in the
production above) and, therefore, supply the next stage 3 with a steady stream
of single move tokens (\abbrev{M}). 

\begin{quote}
Remark: This ``separation'' probably has to be done by inspecting the whole
token character for character. Thus, it appears to be efficient to already
collect information about the move number \abbrev{MN} (White/Black to move,
getting the number itself, suppressing leading zeros) in this early stage\ldots
\end{quote}

No semantic checking is done here, i.e.~this stage doesn't
detect errors like \mbox{``\texttt{2.~Kg1 3.~Bf4}''}.

\section*{Stage 3: Parsing single move tokens}

At this point, we can be sure to get either a chess move \abbrev{CM}
or a move number \abbrev{MN}:

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{M} \bnfprod\={} \bnf{CM}\\
 \>\bnfor{} \bnf{MN}
\end{tabbing}
\end{minipage}
\end{center}

Both can be distinguished well by inspecting the first character of the token, since
only move numbers \abbrev{MN} may start with a digit (\abbrev{D}).

This stage would be the right place to check for the correct order
of chess moves and move numbers, i.e.~if a \abbrev{MN} is encountered
the right side (White or Black) should be to move.

\section*{Parsing move numbers}

Move numbers \abbrev{MN} consist of an integer value (\abbrev{N}), immediately
followed by one or three dots as terminal symbols, signalling whether it's
Whites or Blacks move.

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MN} \bnfprod\={} \bnf{N} \bnfterm{...}\\
 \>\bnfor{} \bnf{N} \bnfterm{.}
\end{tabbing}
\end{minipage}
\end{center}

An integer value \abbrev{N} is a nonempty list of single digits 
(\abbrev{D})\ldots

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{N} \bnfprod\={} \bnf{D} \bnf{N}\\
 \>\bnfor{} \bnf{D}
\end{tabbing}
\end{minipage}
\end{center}

\ldots where each digit \abbrev{D} should be contained in the set of terminal
symbols 0--9.

\begin{center}
\begin{minipage}{5cm}
\begin{tabbing}
\bnf{D} \bnfprod\={} \bnfterm{0}\bnfor\bnfterm{1}\bnfor\bnfterm{2}%
\bnfor\bnfterm{3}\bnfor\bnfterm{4}\bnfor\bnfterm{5}\bnfor\bnfterm{6}%
\bnfor\bnfterm{7}\bnfor\bnfterm{8}\bnfor\bnfterm{9}
\end{tabbing}
\end{minipage}
\end{center}

Here, the move number can be checked against the internal move counter
if necessary.

\section*{Parsing chess moves}

A chess move \abbrev{CM} starts with the ``move specification'' (\abbrev{MS}),
giving information about the piece that is to move and the source and
destination squares. Additionally, an arbitrary number of character tokens may follow
as a move comment (\abbrev{MC}). So, after enough ``hints'' for executing the move
are collected, i.e.~\abbrev{MS} was ``matched'', the rest of the token is regarded
as comment \abbrev{MC} and is output unchanged.

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{CM} \bnfprod\={} \bnf{MS} \bnf{MC}\\
 \>\bnfor{} \bnf{MS}
\end{tabbing}
\end{minipage}
\end{center}

Before defining the move specification \abbrev{MS} itself a few helping nonterminal
symbols are introduced. Please, remember that bold characters within
the ``rules'' denote terminal symbols\ldots

\begin{enumerate}
\item A ``piece'' character (\abbrev{P})

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{P} \bnfprod\={} \bnfterm{K}\bnfor\bnfterm{Q}\bnfor\bnfterm{B}%
\bnfor\bnfterm{N}\bnfor\bnfterm{R}
\end{tabbing}
\end{minipage}
\end{center}

Remark: This definition uses the english letters for the single pieces. The
real implementation should be independent of the used language.

\item A ``file'' character (\abbrev{f})

\begin{center}
\begin{minipage}{5cm}
\begin{tabbing}
\bnf{f} \bnfprod\={} \bnfterm{a}\bnfor\bnfterm{b}\bnfor\bnfterm{c}%
\bnfor\bnfterm{d}\bnfor\bnfterm{e}\bnfor\bnfterm{f}\bnfor\bnfterm{g}%
\bnfor\bnfterm{h}
\end{tabbing}
\end{minipage}
\end{center}

\item A ``rank'' character (\abbrev{r})

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{r} \bnfprod\={} \bnfterm{1}\bnfor\bnfterm{2}\bnfor\bnfterm{3}%
\bnfor\bnfterm{4}\bnfor\bnfterm{5}\bnfor\bnfterm{6}\bnfor\bnfterm{7}%
\bnfor\bnfterm{8}
\end{tabbing}
\end{minipage}
\end{center}

\item A ``capture'' character (\bnfjoin)

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnfjoin{} \bnfprod\={} \bnfterm{-}\\
 \>\bnfor{} \bnfterm{x}\\
 \>\bnfor{} \bnfempty
\end{tabbing}
\end{minipage}
\end{center}

where '\bnfempty' denotes the ``empty symbol''.

\end{enumerate}


Move specifications (\abbrev{MS}) can be subdivided into the following groups:

\begin{itemize}
\item Starting with a ``piece'' character

\begin{itemize}
\item With source square
\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MS} \bnfprod\={} \bnf{P} \bnf{f} \bnf{r} \bnfjoin{} \bnf{f} \bnf{r}\\
 \>\bnfor{} \bnf{P} \bnf{f} \bnfjoin{} \bnf{f} \bnf{r}\\
 \>\bnfor{} \bnf{P} \bnf{r} \bnfjoin{} \bnf{f} \bnf{r}
\end{tabbing}
\end{minipage}
\end{center}

Examples: ``\texttt{Qf3-f4}'', ``\texttt{Ree6}'', ``\texttt{N1xc3}''
\item Only a destination square present

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MS} \bnfprod\={} \bnf{P} \bnfjoin{} \bnf{f} \bnf{r}
\end{tabbing}
\end{minipage}
\end{center}

Example: ``\texttt{Nxc4}''
\end{itemize}

\item Without leading ``piece'' character

\begin{itemize}
\item With source square

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MS} \bnfprod\={} \bnf{f} \bnf{r} \bnfjoin{} \bnf{f} \bnf{r} \bnf{P}\\
 \>\bnfor{} \bnf{f} \bnfjoin{} \bnf{f} \bnf{r} \bnf{P}\\
 \>\bnfor{} \bnf{f} \bnf{r} \bnfjoin{} \bnf{f} \bnf{r}\\
 \>\bnfor{} \bnf{f} \bnfjoin{} \bnf{f} \bnf{r}
\end{tabbing}
\end{minipage}
\end{center}

Examples: ``\texttt{f7-f8R}'', ``\texttt{dxe8B}'', ``\texttt{g2-g4}'', 
``\texttt{fxe6}''
\item Only a destination square present

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MS} \bnfprod\={} \bnf{f} \bnf{r} \bnf{P}\\
 \>\bnfor{} \bnf{f} \bnf{r}
\end{tabbing}
\end{minipage}
\end{center}

Examples: ``\texttt{e8R}'', ``\texttt{a6}''
\end{itemize}


\item Castlings

\begin{center}
\begin{minipage}{3cm}
\begin{tabbing}
\bnf{MS} \bnfprod\={} \bnfterm{O-O-O}\\
 \>\bnfor{} \bnfterm{O-O}
\end{tabbing}
\end{minipage}
\end{center}

Both of these terminals use the letter 'O' (ASCII code 
79\base{10}) and not the digit '0'
(ASCII code 48\base{10})!
\end{itemize}

\section*{Final remark}

This text tries to provide only the syntax for the frontend of the new ``move
machine''. All the things that happen after the input was ``matched'' and
information was drawn out of the given ``chess moves'' to the largest extent,
i.e.~the semantic actions, are beyond the scope of this document.

So, at the moment it is perfectly legal and syntactically correct to say
``\texttt{4. gxf8K}''\ldots

\end{document}