Colour support! Better rule attribute handling. Lots of new hooks.

[mdwtools] / sverb.dtx

% \begin{meta-comment}
%
% $Id: sverb.dtx,v 1.2 2003/09/05 16:09:30 mdw Exp $
%
% Verbatim typesetting done properly (ahem)
%
% (c) 1996 Mark Wooding
%
%----- Revision history -----------------------------------------------------
%
% $Log: sverb.dtx,v $
% Revision 1.2  2003/09/05 16:09:30  mdw
% Make listing formatting more hookable.  Add colour support, and `splitverb'
% environment.
%
% Revision 1.1  2002/02/03 20:49:03  mdw
% Checkin for new build system.
%
% Revision 1.3  1996/11/19 21:01:18  mdw
% Entered into RCS
%
%
% \end{meta-comment}
%
% \begin{meta-comment} <general public licence>
%%
%% sverb package -- handling of verbatim text
%% Copyright (c) 1996 Mark Wooding
%%
%% This program is free software; you can redistribute it and/or modify
%% it under the terms of the GNU General Public License as published by
%% the Free Software Foundation; either version 2 of the License, or
%% (at your option) any later version.
%%
%% This program is distributed in the hope that it will be useful,
%% but WITHOUT ANY WARRANTY; without even the implied warranty of
%% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
%% GNU General Public License for more details.
%%
%% You should have received a copy of the GNU General Public License
%% along with this program; if not, write to the Free Software
%% Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
%%
% \end{meta-comment}
%
% \begin{meta-comment} <Package preamble>
%<+package>\NeedsTeXFormat{LaTeX2e}
%<+package>\ProvidesPackage{sverb}
%<+package>                [2003/09/04 1.4 Verbatim typesetting]
%<+colour>\NeedsTeXFormat{LaTeX2e}
%<+colour>\ProvidesPackage{svcolour}
%<+colour>                [2003/09/04 1.4 Colour support for sverb]
%<+color>\NeedsTeXFormat{LaTeX2e}
%<+color>\ProvidesPackage{svcolor}
%<+color>                [2003/09/04 1.4 Fix for people who can't spell]
%<+split>\NeedsTeXFormat{LaTeX2e}
%<+split>\ProvidesPackage{svsplit}
%<+split>                [2003/09/04 1.4 Verbatim, but with line breaking]
% \end{meta-comment}
%
% \CheckSum{1009}
%% \CharacterTable
%%  {Upper-case    \A\B\C\D\E\F\G\H\I\J\K\L\M\N\O\P\Q\R\S\T\U\V\W\X\Y\Z
%%   Lower-case    \a\b\c\d\e\f\g\h\i\j\k\l\m\n\o\p\q\r\s\t\u\v\w\x\y\z
%%   Digits        \0\1\2\3\4\5\6\7\8\9
%%   Exclamation   \!     Double quote  \"     Hash (number) \#
%%   Dollar        \$     Percent       \%     Ampersand     \&
%%   Acute accent  \'     Left paren    \(     Right paren   \)
%%   Asterisk      \*     Plus          \+     Comma         \,
%%   Minus         \-     Point         \.     Solidus       \/
%%   Colon         \:     Semicolon     \;     Less than     \<
%%   Equals        \=     Greater than  \>     Question mark \?
%%   Commercial at \@     Left bracket  \[     Backslash     \\
%%   Right bracket \]     Circumflex    \^     Underscore    \_
%%   Grave accent  \`     Left brace    \{     Vertical bar  \|
%%   Right brace   \}     Tilde         \~}
%%
%
% \begin{meta-comment}
%
%<*driver>
\input{mdwtools}
\describespackage{sverb}
\describespackage{svcolour}
\describespackage{svsplit}
\mdwdoc
%</driver>
%
% \end{meta-comment}
%
% \section{User guide}
%
% The \package{sverb} package provides some useful commands and environments
% for doing things with verbatim text.  I prefer this code to the standard
% \package{verbatim} package (by Rainer Sch\"opf et al.)\ although I'm
% biased.
%
% The package was written to fulfil a particular purpose: I wanted to be able
% to typeset ARM assembler code, 77~columns wide, on A5~paper, with the
% fields separated by \textit{tab} characters.  It's grown up fairly
% organically from that, and I've tidied it when I've seen the code get too
% ugly.
%
% The current features are:
%
% \begin{itemize}
%
% \item A `listing' environment which typesets verbatim text nicely.
%
% \item A command to read verbatim text from an external file.
%
% \item Support for arbitrary-sized chunks of text without overflowing \TeX's
%       memory.
%
% \item Support for \textit{tab} characters in the verbatim text.
%
% \item An environment for typesetting demonstrations of \LaTeX\ markup.
%
% \item It all works correctly with the \package{doc} system for documenting
%       \LaTeX\ packages.
%
% \item A fairly hairy but quite powerful programmer interface to the yukky
%       bits of the package.
%
% \end{itemize}
%
% The interface is described in its own section, so that more timid readers
% can avoid it.  That said, some of the stuff in this section gets rather
% technical.
%
% Note that this package doesn't even try to do anything with short bits of
% verbatim text (as handled by the |\verb:...:| command).  I have a separate
% package (\package{syntax}) which does all sorts of horrible things along
% those lines.
%
% \subsection{The \env{listing} environment}
%
% \DescribeEnv{listing}
% The main method for typesetting verbatim text is the \env{listing}
% environment.  This works pretty much the same as the standard
% \env{verbatim} environment, with some exceptions, which are described
% below.
%
% So that you know exactly what you're getting, here are the rules by which
% \package{sverb} decides what the verbatim text actually is:
%
% \begin{itemize}
%
% \item If there's any text, other than spaces, on the same line as the
%       `|\begin{listing}|', then the contents of the environment begins
%       immediately after the closing brace (with all leading spaces
%       preserved).  Otherwise, the text begins on the following line.
%
% \item If there is any text, other than spaces, before the
%       `|\end{listing}|', but on the same line, this is considered to be the
%       last line of the text; otherwise the text is presumed to have ended
%       at the end of the previous line.
%
% \item Any text following the |\end{listing}| on the same line is thrown
%       away.  There are good reasons for this, but they're technical.
%       Essentially there's nothing I can do about it.
%
% \end{itemize}
%
% \begin{figure}
% \begin{demo}[w]{The \env{listing} environment}
%\dots in the following code:
%
%\begin{listing}
%init            MOV     R0,#200         ;Version 2.00 please
%                LDR     R1,=&4B534154   ;Magic number (`TASK')
%                ADR     R2,appName      ;Find application name
%                SWI     Wimp_Initialise ;Register as a WIMP task
%\end{listing}
%
%The next step is to \dots
% \end{demo}
% \end{figure}
%
% Tab characters are supported within the environment: tab stops are set
% every eighth column, although this can be modified.
%
% \subsubsection{Configuring the \env{listing} environment}
%
% \DescribeMacro\listingsize
% The text size used in the \env{listing} environment is set by the
% |\listingsize| command.  By default, this is set to |\footnotesize|,
% although you can redefine it in the document preamble, or it can be set in
% the document class.  You can put other declarations (e.g., colours) here if
% you like.
%
% \DescribeMacro\listingindent
% The amount by which the listing text is indented is controlled by the
% |\listingindent| length parameter.  This is a fixed length, whose default
% value is 1\,em.
%
% \DescribeMacro\listinghook
% \DescribeMacro\svafter
% \DescribeMacro\svline
% \DescribeMacro\svdoline
% \DescribeEnv{listinglist}
% The |\listinghook| command is called by the \env{listing} environment (and
% |\verbinput| and \env{demo}) to set up the formatting of the listing.  It
% can do any setting up it likes, and may configure |\svline| and |\svafter|
% as necessary.  The macro |\svline| is run once for each line of verbatim
% text, with the line gathered into a box register, the number of which is
% given as an argument.  The macro |\svafter| is called when processing has
% finished.
%
% The default setting for |\listinghook| is (similar to)
%\begin{listing}
%\newcommand{\listinghook}{%
%  \par%
%  \begin{listinglist}%
%  \listingsize%
%  \renewcommand{\svline}{\listingline}%
%  \renewcommand{\svafter}{\end{listinglist}}%
%}
%\end{listing}
% (see the source for the true definition).  The default |\listingline| macro
% just writes out the line using |\svdoline|, which is a simple no-nonsense
% macro which just writes the text.  As an example, you could say
%\begin{listing}
%\renewcommand{\listingline}{\leavevmode\llap{\strut\vrule\space}\svdoline}
%\end{listing}
% to put a rule down the left-hand side of your listings.
%
% The \env{listinglist} environment is a relatively straightforward
% \env{list}-based environment which sets pu the indentation of a listing.
% Feel free to redefine it.
%
% \subsubsection{Choosing a different end-text}
%
% \DescribeEnv{listing*}
% The \env{listing} environment is terminated by the exact character sequence
% `|\end{listing}|'.  This isn't too much of a problem, unless you want to
% include this string in the text.  This is achieved by the \env{listing$*$}
% environment, which allows you to specify the end-text to find as an
% argument.
%
% For example:
%
% \begin{demo}{The \env{listing$*$} environment}
%Type a listing as follows:
%
%\begin{listing*}{<end-listing*>}
%\begin{listing}
%This is a listing.  Yes.
%\end{listing}
%<end-listing*>
%\end{demo}
%
% Don't include `special' characters in your chosen end-text unless you know
% what you're doing.
%
% \subsection{Writing text to a file}
%
% \DescribeEnv{verbwrite}
% You can write verbatim text to a file using the \env{verbwrite}
% environment.  The syntax is fairly straightforward:
%
% \begin{quote}
% \syntax{"\\begin{verbwrite}{"<file-name>"}" \dots "\\end{verbwrite}"}
% \end{quote}
%
% The text of the environment is written to the named file.  The rules about
% where the text actually starts and ends are the same as for the
% \env{listing} environment.
%
% There is also a $*$-variant, like \env{listing$*$}, which allows you to
% choose the end-text.  The end-text is the first argument, the filename
% comes second.
%
% There is a restriction on the characters you can write to the file: they
% must all be considered `printable' by \TeX; otherwise they will be read
% back in as `\syntax{"^^"<chars>}' which isn't too good.  Unfortunately,
% this includes tab characters, so you can't write them.\footnote{^^A
%   Well, not without doing serious surgery on \TeX\ itself, anyway. }
%
% \iffalse [Example time...  Ho hum.  There is evilness here.] \fi
%\begin{verbwrite*}{<end-write>}{wrdemo1.tmp}
%\begin{verbwrite}{wrdemo.tmp}
%This is some text written to
%a file near the beginning of
%the file.
%\end{verbwrite}
%<end-write>
%
% For example: \verbinput{wrdemo1.tmp}
%
% \input{wrdemo1.tmp} \iffalse [Now build the file ;-) ] \fi
%
% \subsection{The \cmd\verbinput\ command}
%
% \DescribeMacro{\verbinput}
% You can input a pre-prepared text file exactly as it is in the input using
% the |\verbinput| command.  The filename is given as an argument.  For
% example:
%
% \begin{demo}{The \cmd\verbinput\ command}
%\verbinput{wrdemo.tmp}
% \end{demo}
%
% \subsection{The \env{demo} environment}
%
% Package authors need to document their packages, and it's common to want
% to display examples showing the original text and the output side-by-side
% (or, when space doesn't permit this, one above the other).  Both the
% \LaTeX\ book and \textit{The \LaTeX\ Companion} contain such examples.
%
% The \env{demo} environment allows such displays to be created easily.  The
% syntax of the environment is as follows:
%
% \begin{quote}
% \syntax{"\\begin{demo}["<shape>"]{"<title>"}" \dots "\\end{demo}"}
% \end{quote}
%
% The optional \synt{shape} argument can be either `|w|' (wide), or `|n|'
% (narrow).  A `wide' shape places the input and output one above the other,
% while the `narrow' shape puts them side-by-side.  The default shape is
% `narrow'.  An attractive border is drawn around the display to finish it
% off nicely.
%
% An example:
%
%\begin{demo*}{<end-demo>}[w]{The \env{demo} environment}
%\begin{demo}{From the \textit{\TeX book}}
%\[ \sum_{p\;\rm prime}
%   f(p) = \int_{t>1}
%      f(t)\,{\rm d}\pi(t) \]
%\end{demo}
%<end-demo>
%
% \DescribeEnv{demo*}
% As with the other environments created by this package, there's a
% $*$-variant which takes the end-text as an argument.
%
% \DescribeMacro\demohook
% The |\demohook| does the same job for \env{demo} environments as
% |\listinghook| does for \env{listing}s.  The default version just says
%\begin{listing}
%\newcommand{\demohook}{\setlength{\listingindent}{0pt}\listinghook}
%\end{listing}
% (near enough), which turns off the indentation for the listing (which would
% otherwise look rather odd).
%
%
% \section{Programmer interface}
%
% This section describes the publicly available routines provided by the
% \package{sverb} package.  Routines not described here are libable to be
% changed or even removed without warning, so don't use them.
%
% \subsection{Environment hooks}
%
% Each of the environments created here works in the same way.  For each
% environment \env{foo}, there's a main command responsible for doing the
% work, called |\sv@foo|.  This is given all the arguments of the normal
% environment, and two more:
%
% \begin{itemize}
%
% \item The `end-text' to search for, which marks the end of the environment.
%
% \item Some actions to perform after the text has been read and processed.
%       This allows the calling macro to do some extra actions, like closing
%       boxes, etc.
%
% \end{itemize}
%
% All the environments do is call the main command with appropriate
% arguments.
%
% \subsection{Reading the verbatim text}
%
% \DescribeMacro{\sv@read}
% The main scanning routine is |\sv@read|.  It is called with three
% arguments:
%
% \begin{itemize}
%
% \item The end-text marking the end of the environment.
%
% \item The name of a macro (which must be a single token) which is called
%       with a line of text as its single argument.  This is given each
%       line of text which is read from the environment in turn.
%
% \item A macro, or other sort of action, which is to be done when the text
%       has been read and processed.
%
% \end{itemize}
%
% The macro |\sv@read| assumes that the caller has already made some
% provision for removing the category codes of the following text, by either
% calling |\@verbatim| or using the construction
% \begin{listing}
%\let\do=\@makeother
%\dospecials
% \end{listing}
%
% \DescribeMacro{\sv@safespc}
% Note that any space characters you read using |\sv@read| will be catcoded
% as |\active|.  Normally this is OK because |\obeyspaces| (or
% |\@vobeyspaces|) will be in effect.  If you're doing something more exotic,
% like writing text to a file or building a command string, you can call
% |\sv@safespc| which defines the active-space character to be a normal
% whitespace-space when expanded.
%
% \section{Colour support}
%
% There's now a little colour support in \package{sverb}.  To use it, give
% the \textsf{colour} (or \textsf{color}) package option, or load the
% \package{svcolour} package.
%
% \DescribeMacro\svcolourline
% Say \syntax{"\\svcolourline["<model>"]{"<colour>"}{"<box>"}"} to typeset
% \<box> against a background of the given colour.  This is a good thing to
% put in your |\listingline| command.
%\begin{demo}{Coloured listings}
%\renewcommand{\listingline}
%  {\svcolourline[rgb]{1, 0.8, 0.9}}
%Consider, for example, this more
%complicated program.
%\begin{listing}
%#include <stdio.h>
%
%int main(void)
%{
%  puts("Hello, world!");
%  return (0);
%}
%\end{listing}
%\end{demo}
% For coloured text rather than background, put a |\color| command in
% |\listinghook| itself.
%
% \section{The \package{svsplit} package}
%
% A new toy!
%
% \DescribeEnv{splitverb}
% \DescribeEnv{splitverb*}
% \DescribeMacro\svsplitchars
% The \env{splitverb} environment typesets verbatim material very slowly.  On
% the plus side, however, it does know how to do simple line-breaking.  It
% will break lines at spaces or tabs, or after any character listed in
% |\svsplitchars|.  Continuation lines have the same initial intentation as
% the original.  If a line has no `good' breaking point, it's broken as late
% as possible, and a little hyphen is inserted.
%\begin{demo}[w]{The \env{splitverb} environment}
%\begin{multicols}{2}
%\begin{splitverb}
%The \package{url} package is rather fine at splitting up long URLs such as
%  \url{http://www.excessus.demon.co.uk/tex}
%though it can't do its thing in the midst of verbatim text.  It
%also doesn't cope when
%  allthespacesinalongphrasehavemysteriouslydisappeared!
%\end{splitverb}
%\end{multicols}
%\end{demo}
%
% \implementation
%
% \section{Implementation}
%
% This section defines several macros and environments which allow verbatim
% typing, with a high degree of configurability.  OK, so this sort of
% thing's been done so often before that it isn't true, but I don't really
% care.
%
%    \begin{macrocode}
%<*package>
%    \end{macrocode}
%
% \subsection{Options processing}
%
% Notice options, load package.
%
%    \begin{macrocode}
\newif\ifsv@colour\sv@colourfalse
\DeclareOption{colour}{\sv@colourtrue}
\DeclareOption{color}{\sv@colourtrue}
\ProcessOptions
%    \end{macrocode}
%
% \subsection{Simple things}
%
% To help us build funny macros which involve strange and different category
% codes, I'll write some simple macros which I can use while building my
% complicated and clever ones.
%
% \begin{macro}{\@cspecials}
%
% This macro is used to assist the definition of some of the environments.
% It makes `|\|', `|{|' and `|}|' into `other' characters, and replaces them
% with `\verb"|"', `|<|' and `|>|' respectively.  Note that `|[|' and `|]|'
% aren't used, because they make defining commands which take optional
% arguments awkward.  Note that we open a group here.  This should be closed
% using \verb"|endgroup" at the end of the special section.
%
%    \begin{macrocode}
\def\@cspecials{%
  \begingroup%
  \catcode`|0%
  \catcode`<1%
  \catcode`>2%
  \catcode`\{12%
  \catcode`\}12%
  \catcode`\\12%
}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\sv@addtobox}
%
% Add stuff to a horizontal box.
%
%    \begin{macrocode}
\def\sv@addtobox#1#2{\setbox#1\hbox{\unhbox#1\box#2}}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@emptybox}
%
% Clear out a horizontal box.
%
%    \begin{macrocode}
\def\sv@emptybox#1{\setbox#1\hbox{}}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@startlisting}
%
% This macro sets everything up nicely for a \env{listing}-type verbatim
% environment.
%
%    \begin{macrocode}
\def\sv@startlisting{%
  \def\par{\@@par\penalty\interlinepenalty}%
  \@@par%
  \leftskip\@totalleftmargin%
  \obeylines%
  \@noligs%
  \let\do\@makeother\dospecials%
  \verbatim@font%
  \frenchspacing%
  \@vobeyspaces%
  \settabwidth%
  \catcode9\active%
  \lccode`\~9\lowercase{\let~\sv@vtab}%
  \lccode`\~13\lowercase{\let~\vinput@cr}%
  \interlinepenalty500%
}
%    \end{macrocode}
%
% \end{macro}
%
% \subsection{Tab character handling}
%
% One of the things we want to do here is handle tab characters properly.
% (Here, `properly' means `moving to the next column which is a multiple of
% eight', the way these things were always meant to.)
%
% \begin{macro}{\settabwidth}
%
% The tabs used by our tabbed verbatim environments are set up by this
% routine.  It sets the tab width parameter |\svtab| to 8 times the width
% of a |\tt| space.  If you really want, you can redefine this macro.
%
%    \begin{macrocode}
\newdimen\svtab
\def\settabwidth{\setbox\z@\hbox{\texttt{\space}}\svtab8\wd\z@}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@vtab}
%
% Here we handle tabs inside verbatim environments.  We expect to be inside
% |\box|~0.  This is padded to the correct width and contributed to |\box|~2;
% |\box|~0 is then cleared and re-entered.
%
% The idea is that you make tab active, and set it to this macro.  We stop
% the current box, stretch it to the right width, and start another one
% straight after, so nobody knows the difference.  The code here is straight
% from Appendix~D of \textit{The \TeX book}.
%
%    \begin{macrocode}
\def\sv@vtab{%
  \hfill\egroup%
  \@tempdima\wd\z@%
  \divide\@tempdima\svtab%
  \multiply\@tempdima\svtab%
  \advance\@tempdima\svtab%
  \wd\z@\@tempdima%
  \sv@addtobox\tw@\z@%
  \setbox\z@\hbox\bgroup%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\verbinput}
%
% We allow input from a file, by the |\verbinput| command.  We display the
% text pretty much the same as the \env{listing} environment below.
%
% We set tab and return active, and get them to do appropriate things.  This
% isn't actually all that hard.
%
%    \begin{macrocode}
\def\verbinput{\listinghook\@ifstar{\verbinput@\@input}{\verbinput@\input}}
\def\verbinput@#1#2{%  
  \sv@startlisting%
  \setbox\z@\hbox\bgroup%
  #1{#2}%
  \sv@stripspc%
  \egroup%
  \sv@addtobox\tw@\z@%
  \ifdim\wd\tw@=\z@\listingline\tw@\fi%
  \svafter%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\vinput@cr}
%
% This macro handles return characters while inputting text in |\verbinput|.
% We just output our current box, and start another.
%
%    \begin{macrocode}
\def\vinput@cr{%
  \egroup%
  \sv@addtobox\tw@\z@%
  \listingline\tw@%
  \sv@emptybox\tw@%
  \setbox\z@\hbox\bgroup%
}
%    \end{macrocode}
%
% \end{macro}
%
% \subsection{Reading verbatim text}
%
% The traditional way of reading verbatim text is to use a delimited
% argument, as described in the \textit{\TeX book}.  This works well-ish if
% the text isn't very long.  A better solution would be to pick out the text
% line-by-line and process it like that.  So this is what we do.
%
% \begin{macro}{\matcher}
%
% For long verbatim environments, we need to be able to find the end text.
% This is rather tricky.  The solution here is rather horrible.  The
% environment picks out each line of the text at a time, as an argument, and
% tests to see if it contains the text we're after.  We do the test in a
% particularly yukky way: we add the actual target text to the end of the
% line, and inspect the text following the match to see if the match is at
% the end.
%
% The |\matcher| macro creates a `matcher' which will test strings to see if
% they contain something interesting.
%
% To create a matcher, say
% \syntax{"\\matcher{"<cmd-name>"}{"<target>"}{"<process-cmd>"}"}.  The
% command \synt{cmd-name} accepts a line of text as an argument and calls
% the \synt{process-cmd} with the text of the line before the match, or the
% whole lot.  It also sets |\@ifmatched| appropriately.
%
% (Having spent ages coming up with this cruft myself, I found some very
% similar, but slightly better, code in Appendix~D.  So I've changed mine to
% match Donald's.  Anyway, credit where it's due: cheers Don.)
%
%    \begin{macrocode}
\newif\if@matched
\def\matcher#1#2#3{%
  \expandafter\def\csname\string#1$match\endcsname##1#2##2##3\end{%
    \ifx##2\relax%
      \@matchedfalse%
    \else%
      \@matchedtrue%
    \fi%
    #3{##1}%
  }%
  \expandafter\def\expandafter#1\expandafter##\expandafter1\expandafter{%
    \csname\string#1$match\endcsname##1#2\relax\end%
  }%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@stripspc}
%
% This macro strips any trailing glue in the current horizontal list.  This
% is fairly simple, actually: we just loop while glue is the last item.  It's
% slightly complicated by penalties which \TeX\ puts into the list between
% the glue items, but we just remove them too.
%
%    \begin{macrocode}
\def\sv@stripspc{%
  \unpenalty%
  \ifdim\lastskip=\z@\else%
    \unskip\expandafter\sv@stripspc%
  \fi%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@percent}
%
% This macro strips a single leading percent character if there is one, and
% if the \env{doc} package is loaded.  We store the possibly stripped text in
% |\@tempa|.
%
%    \begin{macrocode}
\begingroup
\catcode`\%=12
\gdef\sv@percent#1#2\relax
    {\ifx\check@percent\@@undefined
                     \ifx#1\relax\def\@tempa{}\else
                         \def\@tempa{#1#2}\fi\else
                     \ifx#1\relax\def\@tempa{}\else
                         \ifx#1%\def\@tempa{#2}\else
                             \def\@tempa{#1#2}\fi\fi\fi}
\endgroup
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\@isspaces}
%
% We want to avoid writing the first and last lines of the environment to the
% file if there's nothing in them.  To do this, we need to know whether a
% piece of text contains only space characters.  This macro does this, in a
% rather nasty way.  See the other macros below for details of how this
% works.
%
% We define |\sv@safespc| at the same time: this makes space active and
% expand to a space character which is not active.  Neat, huh?
%
%    \begin{macrocode}
\lccode`\~32
\lccode`\!32
\lowercase{%
  \def\@isspaces#1{%
    \ifx#1\relax%
      \def\@tempb{\@tempswafalse}%
    \else\ifx#1~%
      \let\@tempb\@isspaces%
    \else%
      \def\@tempb##1\relax{}%
    \fi\fi%
    \@tempb%
  }
  \def\sv@safespc{%
    \catcode32\active%
    \def~{ }%
  }
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@read}
%
% This macro does the main job of reading a chunk of verbatim text.  You call
% it like this:
%
% \begin{quote}
% \syntax{"\\sv@read{"<end-text>"}{"<process-line-proc>"}{"<end-proc>"}"}
% \end{quote}
%
% The \synt{end-text} is the text to find at the end of the `environment': we
% stop when we find it.
%
% The \synt{process-line-proc} is a macro which is passed as an argument each
% line which we read from the text.
%
% The \synt{end-proc} is a macro to call once we've finished reading all of
% the text.  This can tidy up an environment or close a file or whatever.
%
% We read the text by picking out newlines using a delimited macro.  We have
% to be a little clever, because newlines are active in verbatim text.
%
% We will also strip `|%|' signs off the beginning if the \package{doc}
% package is here (\package{doc} tries to play with \LaTeX's verbatim stuff,
% and doesn't understand the way we do things).
%
%    \begin{macrocode}
\def\sv@read#1#2#3{%
%    \end{macrocode}
%
% This code does all sorts of evil things, so I'll start by opening a group.
%
%    \begin{macrocode}
  \begingroup%
%    \end{macrocode}
%
% So that I can spot the end-text, I'll create a matcher macro.
%
%    \begin{macrocode}
  \matcher\@match{#1}\sv@read@ii%
%    \end{macrocode}
%
% So that I can identify line ends, I'll make them active.  I'll also make
% spaces active so that they can expand to whatever they ought to expand
% to (spaces in files, or funny \verb*" " characters or whatever.
%
%    \begin{macrocode}
  \catcode13\active%
  \catcode32\active%
%    \end{macrocode}
%
% I'll use the |\if@tempswa| flag to tell me whether I ought to output the
% current line.  This is a little messy, so I'll describe it later.  I'll
% initialise it to false because this is the correct thing to do.
%
%    \begin{macrocode}
  \@tempswafalse%
%    \end{macrocode}
%
% Most of the job is done by two submacros.  I'll define them in terms of
% my current arguments (to save lots of token munging).  The first just
% extracts the next line (which ends at the next newline character) and
% tries to match it.
%
%    \begin{macrocode}
  \lccode`\~13\lowercase{%
    \def\sv@read@i##1~{\@match{##1}}%
  }%
%    \end{macrocode}
%
% The results of the match get passed here, along with the text of the
% line up to the matched text.
%
%    \begin{macrocode}
  \def\sv@read@ii##1{%
%    \end{macrocode}
%
% The first job to do is to maybe strip off percent signs from the beginning,
% to keep \package{doc} happy.
%
%    \begin{macrocode}
    \sv@percent##1\relax\relax%
%    \end{macrocode}
%
% Now I need to decide whether I ought to output this line.  The method goes
% like this: if this is the first line (|\if@tempswa| is false) or the last
% (|\if@matched| is true), \emph{and} the text consists only of spaces, then
% I'll ignore it.
%
% The first thing to do is to notice the last line -- if |\if@matched| is
% true, then I'll make |\if@tempswa| false to make the first-line and
% last-line cases work the same way.
%
%    \begin{macrocode}
    \if@matched\@tempswafalse\fi%
%    \end{macrocode}
%
% Now if this is the first or last line, I'll examine it for spaces.  This
% is done in a separate macro.  It will set |\if@tempswa| false if the
% text contains only spaces.
%
%    \begin{macrocode}
    \if@tempswa\else\@tempswatrue\expandafter\@isspaces\@tempa\relax\fi%
%    \end{macrocode}
%
% Now, if |\if@tempswa| is still true, perform the \<process-line-proc> on
% the line of text.  I'll provide a group, so that it doesn't upset me
% too much.
%
%    \begin{macrocode}
    \if@tempswa%
      \begingroup%
      \expandafter#2\expandafter{\@tempa}%
      \endgroup%
    \fi%
%    \end{macrocode}
%
% The next line won't be the first one, so I'll set the flag true in
% readiness.
%
%    \begin{macrocode}
    \@tempswatrue%
%    \end{macrocode}
%
% Now, if that wasn't the last line, go round again; otherwise end the group
% I started ages ago, and do the user's \<end-proc>.
%
%    \begin{macrocode}
    \if@matched\def\@tempa{\endgroup#3}\else\let\@tempa\sv@read@i\fi%
    \@tempa%
  }%
%    \end{macrocode}
%
% Now to start the thing up.  I'll read the first line.
%
%    \begin{macrocode}
  \sv@read@i%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@readenv}
%
% This macro works out an appropriate end-text for the current environment.
% If you say \syntax{"\\sv@readenv{"<macro-name>"}"}, it will expand do
% \begin{listinglist} \listingsize \synshorts
% <macro-name>"{\\"$_{12}$"end{"$_{12}$<current-env-name>"}"$_{12}$"}"^^A
%               "{\\end{"<current-env-name>"}}"
% \end{listinglist}
% Easy, no?
%
% This is all done with mirrors.  No, err\dots\ it's done with
% |\expandafter|.
%
%    \begin{macrocode}
\begingroup
\lccode`\<=`\{
\lccode`\>=`\}
\lccode`\|=`\\
\lowercase{\endgroup
\def\sv@readenv#1{\expandafter\sv@readenv@i\expandafter{\@currenvir}{#1}}
\def\sv@readenv@i#1#2{#2{|end<#1>}{\end{#1}}}
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@verbline}
%
% This macro typesets a line in a verbatim way, so you can construct a real
% verbatim environment from it.  It's a bit tricky in the way that it catches
% the last line.  Don't worry about this: it's easy really.  Note the
% |\relax| after the |\par| -- this is because \package{doc} tries to do
% clever things with |\par| to strip `|%|' signs out.
%
%    \begin{macrocode}
\def\sv@verbline#1{%
  \sv@emptybox\tw@%
  \setbox\z@\hbox{#1\sv@stripspc}%
  \sv@addtobox\tw@\z@%
  \if1\ifdim\wd\tw@=\z@\if@matched0\else1\fi\else1\fi%
    \svline\tw@\relax%
  \fi%
}
%    \end{macrocode}
%
% \end{macro}
%
% \subsection{Listing environments}
%
% The \env{listing} environment is our equivalent of the standard
% \env{verbatim} environment.  We do some slightly cleverer things, though,
% to make sure (for example) that even text which contains |\end{listing}|
% can be typeset.
%
% \begin{macro}{\listinghook}
%
% Set everything up as required.  This is here for customization.  The
% underlying machinery doesn't mess with this directly, but assumes that
% |\svline| and |\svafter| are set up appropriately.
%
%    \begin{macrocode}
\def\listinghook{%
  \par%
  \begingroup
  \listinglist%
  \listingsize%
  \let\svline\listingline%
  \def\svafter{\endlistinglist\endgroup}%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\listinglist}
% \begin{environment}{listinglist}
%
% This defines the layout for the \env{listing} environment.  It starts a
% list with the appropriate shape.  It's also made into an environment, so
% that the end-paragraph-environment bits work correctly.
%
% The |\listingindent| length parameter sets up the indentation of the
% listings.  If there's a |\parindent| setting, I'll line listings up with
% that; otherwise I'll just choose something which looks right.
%
%    \begin{macrocode}
\newdimen\listingindent
\AtBeginDocument{%
  \ifdim\parindent=\z@\listingindent1em\else\listingindent\parindent\fi%
}
%    \end{macrocode}
%
% Now to define a size hook for the environment.  This is fairly simple
% stuff.
%
%    \begin{macrocode}
\ifx\listingsize\@@undefined
  \let\listingsize\footnotesize
\fi
%    \end{macrocode}
%
% Now to define the environment itself.  Suppress the indentation if we're
% first thing on a new list item, so that the listing lines up with
% everything else.
%
%    \begin{macrocode}
\def\listinglist{%
  \list{}{%
    \if@inlabel%
      \leftmargin\z@%
    \else%
      \leftmargin\listingindent%
    \fi%
    \rightmargin\z@%
    \labelwidth\z@%
    \labelsep\z@%
    \itemindent\z@%
    \listparindent\z@%
    \let\makelabel\relax%
    \parsep\z@skip%
  }%
  \parfillskip\@flushglue%
  \item\relax%
}
\let\endlistinglist\endlist
%    \end{macrocode}
%
% \end{environment}
% \end{macro}
%
% \begin{macro}{\svline}
% \begin{macro}{\svdoline}
% \begin{macro}{\listingline}
%
% The simple spit-out-a-line macro.
%
%    \begin{macrocode}
\def\svdoline#1{\leavevmode\box#1\par}
\let\svline\svdoline
\let\listingline\svline
%    \end{macrocode}
%
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\svafter}
%
% This is called when the machinery finishes.  A default is set for safety's
% sake.
%
%    \begin{macrocode}
\let\svafter\relax
%    \end{macrocode}
%
% \end{macro}
%
% \begin{environment}{listing}
%
% The \env{listing} environment is the only real verbatim-like environment we
% create will all this kit, although it does the job very nicely.
%
% The environment indents its contents slightly, unlike \env{verbatim}, and
% uses a smaller typeface in an attempt to fit 77-column text on an A5~page.
% There is also a $*$-variant, which allows you to specify the terminating
% text.  This enables you to include absolutely any text in the environment,
% including |\end{listing}|.
%
% First, we must define the |\listing| command.
%
%    \begin{macrocode}
\def\listing{\listinghook\sv@readenv\sv@listing}
%    \end{macrocode}
%
% Now we define the |\@listing| command, which does most of the work.  We
% base the \env{listing} environment on a \env{list}.
%
%    \begin{macrocode}
\def\sv@listing#1#2{\sv@startlisting\sv@read{#1}\sv@verbline{\svafter#2}}
%    \end{macrocode}
%
% Now we define the starred version.  The command name needs to include the
% `|*|' character, so we must use |\csname|.  There's some hacking here to
% allow us to read the name using the appropriate catcodes for otherwise
% normal characters: \LaTeX\ activates some characters and makes them typeset
% themselves to suppress some ligaturing.
%
%    \begin{macrocode}
\expandafter\def\csname listing*\endcsname{%
  \listinghook\begingroup\@noligs\listing@star%
}
\def\listing@star#1{\endgroup\sv@listing{#1}{\end{listing*}}}
%    \end{macrocode}
%
% \end{environment}
%
% \begin{environment}{ignore}
%
% The \env{ignore} environment entirely ignores its contents.  Anything at
% all may be put into the environment: it is discarded utterly.
%
% We define some macros for defining ignoring environments, because this can
% be useful for version control, possibly.
%
%    \begin{macrocode}
\def\sv@ignore#1#2{%
  \@bsphack%
  \let\do\@makeother\dospecials%
  \sv@read{#1}\@gobble{\@esphack#2}%
}
\def\ignore{\sv@readenv\sv@ignore}
\def\ignoreenv#1{%
  \expandafter\let\csname #1\endcsname\ignore%
}
\def\unignoreenv#1{%
  \expandafter\def\csname #1\endcsname{\endgroup}%
  \expandafter\def\csname end#1\endcsname%
      {\begingroup\def\@currenvir{#1}}%
}
%    \end{macrocode}
%
% \end{environment}
%
% \subsection{The \env{verbwrite} environment}
%
% The \env{verbwrite} environment allows text to be written to a file in a
% verbatim way.  Note that tab characters don't work, because \TeX\ refuses
% to be nice.
%
% \begin{macro}{\sv@write}
%
% As seems to be traditional now, we first define a general hookable macro
% which allows a caller to specify the end-text and what to do afterwards.
%
%    \begin{macrocode}
\newwrite\sv@writefile
\def\sv@write#1#2{%
  \begingroup%
  \@bsphack%
  \let\do\@makeother\dospecials%
  \sv@safespc%
  \sv@read{#1}\sv@writeline{\sv@endwrite#2}%
}
\def\sv@writeline#1{%
  \immediate\write\sv@writefile{#1}%
}
\def\sv@endwrite{%
  \@esphack%
  \endgroup%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{environment}{verbwrite}
%
% Now we can define the actual environment.  We define a $*$-variant which
% allows the user to specify the end-text, just to make sure.
%
%    \begin{macrocode}
\def\verbwrite#1{%
  \immediate\openout\sv@writefile#1\relax%
  \sv@readenv\sv@write%
}
\def\endverbwrite{\immediate\closeout\sv@writefile}
\expandafter\def\csname verbwrite*\endcsname#1#2{%
  \immediate\openout\sv@writefile#2\relax%
  \sv@write{#1}{\immediate\closeout\sv@writefile\end{verbwrite*}}%
}
%    \end{macrocode}
%
% \end{environment}
%
% \subsection{The \env{demo} environment}
%
% By way of tying all of this together, I present an environment for
% displaying demonstrations of \LaTeX\ markup.  We read the contents of the
% environment, write it to a temporary file, and read it back twice,
% typesetting it the first time and displaying it verbatim the second time.
%
% \begin{macro}{\sv@demoname}
%
% This macro expands to the filename to use for the temporary data.  To
% allow the package documentation to demonstrate the \env{demo} environment
% itself, we need to keep a nesting count.  This avoids too much hackery,
% which unfortunately appears to plague all of my \TeX\ code.
%
%    \begin{macrocode}
\newcount\sv@nestcount
\def\sv@demoname{demo\number\sv@nestcount.tmp}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@demo}
%
% As for listing, we do all the business through a private macro.  This is
% good because it means we can leave the main macro readable.  The argument
% is the end-text to spot.
%
%    \begin{macrocode}
\def\sv@demo#1#2{%
  \@ifnextchar[{\sv@demo@i{#1}{#2}}{\sv@demo@i{#1}{#2}[n]}%
}
\def\sv@demo@i#1#2[#3]#4{%
  \advance\sv@nestcount by\@ne%
  \immediate\openout\sv@writefile\sv@demoname\relax%
  \sv@write{#1}{%
    \immediate\closeout\sv@writefile%
    \sv@dodemo{#2}{#3}{#4}%
  }%
}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{environment}{demo}
%
% This is the real environment.  We provide \env{demo$*$} too, to allow the
% user to choose the end-text.
%
%    \begin{macrocode}
\def\demo{\let\@demohook\demohook\sv@readenv\sv@demo}
\expandafter\def\csname demo*\endcsname#1%
  {\let\@demohook\demohook\sv@demo{#1}{\end{demo*}}}
%    \end{macrocode}
%
% \end{environment}
%
% \begin{macro}{\demohook}
%
% Like |\listinghook|.  So much so that we just call it, but first ensure
% that the indent is zero (otherwise it looks really odd!).
%
%    \begin{macrocode}
\def\demohook{\listingindent\z@\listinghook}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\sv@dodemo}
%
% First, let's define some common bits of code in the stuff below.  The
% minipages used to typeset the material has some clever stuff to avoid
% strange spacing in the output.
%
%    \begin{macrocode}
\def\sv@demosmp{%
  \begin{minipage}[t]{\@tempdima}%
  \vskip8\p@%
  \hrule\@height\z@%
  \raggedright%
  \vbox\bgroup%
}
\def\sv@demoemp{%
    \par\unpenalty\unskip%
  \egroup%
  \vskip8\p@%
  \hrule\@height\z@%
  \end{minipage}%
}
%    \end{macrocode}
%
% This is the macro which actually typesets the demonstration.
%
%    \begin{macrocode}
\def\sv@dodemo#1#2#3{%
%    \end{macrocode}
%
% Now work out some values.  We set |\hsize| to the line width leaving 2\,em
% of space on either side.  The size of the minipages is calculated depending
% on the shape of the demonstration.  This is all fairly simple.
%
%    \begin{macrocode}
  \begingroup%
  \@tempdima\linewidth%
  \advance\@tempdima-2em%
  \hsize\@tempdima%
  \if#2w%
    \advance\@tempdima-2em%
  \else%
    \advance\@tempdima-3em%
    \divide\@tempdima2%
  \fi%
%    \end{macrocode}
%
% Now we open a big vertical box, and put in a header to mark off the
% demonstration.
%
%    \begin{macrocode}
  \par%
  \setbox\z@\hbox{\strut\enspace#3\enspace\strut}%
  \@tempdimb.5\dp\z@%
  \advance\@tempdimb-.5\ht\z@%
  \ht\z@\@tempdimb\dp\z@\@tempdimb%
  \noindent\hskip1em\vtop{%
    \hb@xt@\hsize{%
      \hrulefill%
      \raise\@tempdimb\box\z@%
      \hrulefill%
    }%
    \nointerlineskip%
    \hb@xt@\hsize{\vrule\@height5\p@\hfil\vrule\@height5\p@}%
    \nointerlineskip%
%    \end{macrocode}
%
% Now we insert the output text in the first minipage.  I'll force `|%|'
% to be a comment character, in case something like \package{doc} has had its
% wicked way.
%
%    \begin{macrocode}
    \vskip-\parskip%
    \noindent\hbox{}\hskip1em%
    \sv@demosmp%
    \catcode`\%14\relax%
    \@input{\sv@demoname}%
    \sv@demoemp%
%    \end{macrocode}
%
% Insert some kind of separation between the two.  In `wide' format, we start
% a new line, and put a ruleoff between the two.  In `narrow' format, we just
% leave some space.
%
%    \begin{macrocode}
    \if#2w%
      \vskip8\p@\hrule\vskip8\p@%
      \noindent\hbox{}%
    \fi%
    \hskip1em%
%    \end{macrocode}
%
% Now we put the verbatim copy of the text in the other minipage.
%
%    \begin{macrocode}
    \sv@demosmp%
    \@demohook%
    \verbinput@\@input\sv@demoname%
    \sv@demoemp%
    \par%
    \nointerlineskip%
    \hb@xt@\hsize{\vrule\@height5\p@\hfil\vrule\@height5\p@}%
    \hrule%
  }%
  \endgroup%
  \par%
  \vskip\baselineskip%
  #1%
}
%    \end{macrocode}
%
% \end{macro}
%
% \subsection{Loading the colour package}
%
% If requested, we load the \package{svcolour} package here.  This ensures
% that it can patch this code if it needs to.
%
%    \begin{macrocode}
\ifsv@colour
  \RequirePackage{svcolour}
\fi
%    \end{macrocode}
%
% That's all there is.  Have fun.
%
%    \begin{macrocode}
%</package>
%    \end{macrocode}
%
% \subsection{The \package{svcolour} package}
%
% This is in a separate package to avoid dragging in the \package{color}
% package if it's unwanted.
%
% I prefer English spellings.  Here's a trivial redirection for Americans.
%
%    \begin{macrocode}
%<*color>
\DeclareOption*{\PassOptionsToPackage{\CurrentOption}{svcolour}}
\ProcessOptions
\RequirePackage{svcolour}
%</color>
%    \end{macrocode}
%
% And now we can start the thing properly.
%
%    \begin{macrocode}
%<*colour>
\RequirePackage{color}
%    \end{macrocode}
%
% \begin{macro}{\@snarfcolour}
%
% Reading a colour specification is something we'll need to do a few times,
% so an abstraction is useful.  Its single argument is a continuation to
% which we pass a colour-spec acceptable to the |\color| command.  (This is
% the same code as found in the \package{mdwtab} package.  Remember to keep
% them in step.)
%
%    \begin{macrocode}
\def\@snarfcolour#1{%
  \@ifnextchar[{\@snarfcolour@i{#1}}{\@snarfcolour@ii{#1}{}}%
}
\def\@snarfcolour@i#1[#2]{\@snarfcolour@ii{#1}{[#2]}}
\def\@snarfcolour@ii#1#2#3{#1{#2{#3}}}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\svcolourline}
% \begin{macro}{\svcolorline}
%
% Snarf the option, and plot the coloured bar.  Note the penalties which are
% meant to stick the glue and leaders onto the colour specials.
%
%    \begin{macrocode}
\def\svcolourline{\@snarfcolour\svcl@i}
\def\svcl@i#1#2{%
  \skip@\wd#2%
  \advance\skip@\parfillskip%
  \advance\skip@.2em%
  \strut%
  \kern.2em%
  \begingroup\color#1\nobreak\leaders\vrule\hskip\skip@\endgroup%
  \nobreak\hskip-\skip@%
  \kern.2em%
  \box#2%
  \nobreak\hskip-\rightskip\vadjust{}%
  \par%
}
\let\svcolorline\svcolourline
%    \end{macrocode}
%
% \end{macro}
% \end{macro}
%
% Done!
%
%    \begin{macrocode}
%</colour>
%    \end{macrocode}
%
% \subsection{The \package{svsplit} package}
%
%    \begin{macrocode}
%<*split>
\RequirePackage{sverb}
%    \end{macrocode}
%
% \begin{environment}{splitverb}
% \begin{environment}{splitverb*}
%
% The basic environments are simple enough.
%
%    \begin{macrocode}
\def\splitverb{\listinghook\sv@readenv\splitverb@}
\expandafter\def\csname splitverb*\endcsname%
  {\listinghook\begingroup\@noligs\svsplit@star}
\def\svsplit@star#1{\endgroup\splitverb@{#1}{\end{splitverb*}}}
%    \end{macrocode}
%
% \end{environment}
% \end{environment}
%
% \begin{macro}{\splitverb@}
%
% Even this isn't so bad, really.
%
%    \begin{macrocode}
\def\splitverb@#1#2{\sv@startlisting\sv@read{#1}\svsplit@line{\svafter#2}}
%    \end{macrocode}
%
% \end{macro}
%
% \begin{macro}{\svsplit@line}
%
% For the sake of readability (and maybe saving a few tokens), we define some
% synonyms for \TeX's scratch registers.  |\svsplit@remain| will be a
% |\global| register containing the remaining horizontal space on the line;
% |\svsplit@indent| is a local register containing the amount of initial
% whitespace on the line.
%
%    \begin{macrocode}
\dimendef\svsplit@remain=1 
\dimendef\svsplit@indent=2
%    \end{macrocode}
%
% The switch |\svsplit@| is set if we've found a good place to split the
% current line.
%
%    \begin{macrocode}
\newif\ifsvsplit@
%    \end{macrocode}
%
% And finally a delimiter.  This is the same one I use everywhere else.
%
%    \begin{macrocode}
\def\q@delim{\q@delim}
%    \end{macrocode}
%
%    \begin{macrocode}
\begingroup
\catcode`\~=\active \lccode`\~=32
\catcode`\!=\active \lccode`\!=9
\lowercase{\endgroup
%    \end{macrocode}
%
% So far, so good.  The |\svsplit@line| macro is given a line of text.  We
% initialize |\svtab| to be a \emph{single} space, |\svsplit@remain| to be
% the text width, and |\svsplit@indent| to zero.  Then we embark on the first
% loop, which attempts to find the width of the leading whitespace.
%
%    \begin{macrocode}
\def\svsplit@line#1{%
  \divide\svtab8%
  \global\svsplit@remain\linewidth%
  \svsplit@indent\z@%
  \sv@emptybox\tw@%
  \let\next@\svsplit@findindent%
  \next@#1\q@delim%
}
%    \end{macrocode}
%
% A straightforward tail-recursive loop finds out how much whitespace there
% is at the start of the current line.  Note that |\next@| is already set up
% for the optimized case of continuing the loop.  Also, if we reach the end
% then this is a blank line, so only emit something if we didn't see the
% end-marker.  This is the only place we need to check for this.
%
%    \begin{macrocode}
\def\svsplit@findindent#1{%
  \ifx~#1%
    \advance\svsplit@indent\svtab%
  \else\ifx!#1%
    \dimen@8\svtab%
    \divide\svsplit@indent\dimen@%
    \multiply\svsplit@indent\dimen@%
    \advance\svsplit@indent\dimen@%
  \else\ifx\q@delim#1%
    \if@matched\else\svline\tw@\fi%
    \let\next@\relax%
  \else%
    \def\next@{\svsplit@scanline{#1}}%
  \fi\fi\fi%
  \next@%
}
%    \end{macrocode}
%
% Now we have to actually scan the line to find breakpoints.  We build the
% current unbreakable chunk in |\box|~0.  When we find a breakpoint, we close
% the box, maybe stretch it to take into account trailing space, and attach
% it to |\box|~2, which is gathering the current line.  If |\svsplit@remain|
% hits zero then we flush |\box|~2 to the output and continue on the next
% line with a (more-or-less) clean slate.
%
% If there's no breakpoint then we're hosed.  In that case, we just insert a
% (|\normalfont|) hyphen and eject what we've got.
%
% Note that this assumes that the indentation will fit.  If not, then we're
% deeply stuffed.
%
%    \begin{macrocode}
\def\svsplit@scanline{%
  \svsplit@false%
  \let\next@\svsplit@char%
  \setbox\z@\hbox\bgroup%
    \kern\svsplit@indent%
    \global\advance\svsplit@remain-\svsplit@indent%
    \next@%
}
%    \end{macrocode}
%
% Scanning a character isn't so bad, if we take it a step at a time.
%
%    \begin{macrocode}
\def\svsplit@char#1{%
%    \end{macrocode}
%
% If the character is a space or a tab, then we call |\svsplit@space| which
% knows about adding breakable whitespace.  For tabs, this involves computing
% the correct tab size.
%
%    \begin{macrocode}
    \ifx~#1%
      \svsplit@space\svtab%
    \else\ifx!#1%
      \@tempdima\linewidth%
      \advance\@tempdima-\svsplit@remain%
      \@tempdimb\@tempdima%
      \dimen@8\svtab%
      \divide\@tempdimb\dimen@%
      \multiply\@tempdimb\dimen@%
      \advance\@tempdimb\dimen@%
      \advance\@tempdimb-\@tempdima%
      \svsplit@space\@tempdimb%
%    \end{macrocode}
%
% We might have reached the end of the line.  If so, then we finish off.
%
%    \begin{macrocode}
    \else\ifx\q@delim#1%
      \let\next@\svsplit@done%
%    \end{macrocode}
%
% Otherwise it's a normal character.  If there's not enough space then force
% a break.  
%
%    \begin{macrocode}
    \else%
      \ifdim\svsplit@remain<2\svtab%
        \ifsvsplit@\else\normalfont-\svsplit@break\fi%
        \svsplit@eject%
      \fi%
%    \end{macrocode}
%
% Insert the character and decrement the distance-left register.
%
%    \begin{macrocode}
      #1%
      \global\advance\svsplit@remain-\svtab%
%    \end{macrocode}
%
% Now we see if it's a breakable-after character and if so mark it as being
% breakable.
%
%    \begin{macrocode}
      \def\temp@##1#1##2\q@delim%
        {\ifx\q@delim##2\q@delim\else\svsplit@break\fi}%
      \expandafter\temp@\svsplitchars#1\q@delim%
%    \end{macrocode}
%
% And with that, we're done.
%
%    \begin{macrocode}
    \fi\fi\fi%
    \next@%
}
%    \end{macrocode}
%
% Our next macro is the break-insertion subroutine, which is quite easy.
%
%    \begin{macrocode}
\def\svsplit@break{%
  \egroup%
  \sv@addtobox\tw@\z@%
  \svsplit@true%
  \setbox\z@\hbox\bgroup%
}
%    \end{macrocode}
%
% Now we add space to the current box.  The argument is a dimen register.
%
%    \begin{macrocode}
\def\svsplit@space#1{%
    \ifdim\svsplit@remain>#1\kern#1\global\advance\svsplit@remain-#1\fi%
    \svsplit@break%
    \ifdim\svsplit@remain>#1\else\svsplit@eject\fi%
}
%    \end{macrocode}
%
% We now come to a slightly involved piece of code, which is how to flush out
% a line, and then fix up the registers for the next line correctly.
%
%    \begin{macrocode}
\def\svsplit@eject{%
  \egroup%
  \svline\tw@%
  \sv@emptybox\tw@%
  \svsplit@false%
  \setbox\z@\hbox\bgroup%
    \kern\svsplit@indent%
    \global\svsplit@remain\linewidth%
    \global\advance\svsplit@remain-\svsplit@indent%
    \global\advance\svsplit@remain-\wd\z@%
    \unhbox\z@%
}
%    \end{macrocode}
%
% Finally, how to finish the line and go home.
%
%    \begin{macrocode}
\def\svsplit@done{%
  \egroup%
  \sv@addtobox\tw@\z@%
  \svline\tw@%
}
%    \end{macrocode}
%
% End the |\lowercase| hack.
%
%    \begin{macrocode}
}
%    \end{macrocode}
%
% \end{macro}
%
% Finally, set the breakable characters to something plausible.
%
%    \begin{macrocode}
\def\svsplitchars{:/.}
%    \end{macrocode}
%
% And with that, we're done!
%
%    \begin{macrocode}
%</split>
%    \end{macrocode}
%
% \hfill Mark Wooding, \today
%
% \Finale
%
\endinput