Hi all, in particular LaTeX group,

it seems that some old documents (.dtx) don't translate anymore
correctly. That might be simply a bug in the old code, but maybe some
oversight in the LaTeX code.

The attached lgrind.dtx was used and always properly build up to recent
changes (2020 LaTeX), but now fails with
> LaTeX2e <2020-02-02> patch level 5
> L3 programming layer <2020-02-25>
> (/usr/share/texlive/texmf-dist/tex/latex/base/ltxdoc.cls
> Document Class: ltxdoc 2019/09/16 v2.0y Standard LaTeX documentation class
> (/usr/share/texlive/texmf-dist/tex/latex/base/article.cls
> Document Class: article 2019/12/20 v1.4l Standard LaTeX document class
> (/usr/share/texlive/texmf-dist/tex/latex/base/size10.clo))
> (/usr/share/texlive/texmf-dist/tex/latex/base/doc.sty
> (/usr/share/texlive/texmf-dist/tex/latex/tools/multicol.sty)))
> Writing index file lgrind.idx
> Writing glossary file lgrind.glo 
> (./lgrind.dtx
> (/usr/share/texlive/texmf-dist/tex/latex/l3backend/l3backend-dvips.def
> ! LaTeX Error: Missing \begin{document}.
> See the LaTeX manual or LaTeX Companion for explanation.
> Type  H <return>  for immediate help.
>  ...                                              
> l.2 %% T
>         his is file `l3backend-dvips.def',
> ? 
> ! Emergency stop.

Since it contains the
that was recently introduced, I assume there is some breakage with
recent LaTeX.

Any comments?

Please keep Cc, thanks!

All the best


PREINING Norbert                              https://www.preining.info
Accelia Inc. + IFMGA ProGuide + TU Wien + JAIST + TeX Live + Debian Dev
GPG: 0x860CDC13   fp: F7D8 A928 26E3 16A1 9FA0 ACF0 6CAC A448 860C DC13
% \iffalse
% This is LGrind.DTX. This was written by:
% -  Van Jacobson, Lawrence Berkeley Laboratory (based on
%    vgrind by Dave Presotto & William Joy of UC Berkeley),
%    the original written for \TeX.
% -  Jerry Leichter of Yale University, modifications for \LaTeX.
% -  George V. Reilly of Brown University, renamed to lgrind
% -  Michael Piefel, Humboldt-University Berlin, for \LaTeXe
%    and a thousand little changes
% \fi
% \iffalse
%% Based on Van Jacobson's ``tgrindmac'', a macro package for TeX.
%% Modified, 1987 by Jerry Leichter. Put '@' in all internal names.
%% Modified, 1991 by George Reilly. Changed name from tgrind to lgrind.
%% Modified, 1995 by Michael Piefel. Made it work with \LaTeXe.
%%          -1999    Hundreds of bells and whistles. No changelog here.
          [2002/01/28 v3.67 LGrind environment and supporting stuff]
%<package>          [2002/01/28 v3.67 LGrind environment and supporting stuff]
% \fi
% \newcommand{\LG}{\textsf{LGrind}}
% \newcommand{\NFSS}{\textsf{NFSS}}
% \frenchspacing
% \GetFileInfo{lgrind.dtx}
% \begin{document}
% \title{The \LG{} package\thanks{This file
%         has version number \fileversion, last
%         revised \filedate.}}
% \author{Various Artists}
% \date{\filedate}
% \maketitle
% \CheckSum{679}
% \begin{abstract}
% The \LG{} package is a pretty printer for source
% code. It evolved from vgrind, supported \TeX{} (tgrind) and
% \LaTeX{} and now finally \LaTeXe, in particular \NFSS.
% \end{abstract}
% \changes{v1.0}{1985/02/10}{Written}
% \changes{v2.0}{1991/09/06}{Upgrade to \LaTeX2.09}
% \changes{v3.0}{1995/09/24}{Upgrade to \LaTeXe}
% \section{Introduction}
% \subsection{What is it?}
% The \LG{} package consists of three files:
% \begin{itemize}
% \item \texttt{lgrind} or \texttt{lgrind.exe} is the executable. It will
% convert an \LaTeX-File with embedded listings or a source code file
% into a lot of macro-calls.
% \item \texttt{lgrind.sty} provides the environments and macros to make
% the produced mess readable.
% \item \texttt{lgrindef} contains the information
% needed to tell keywords from comments, comments from strings \dots
% \end{itemize}
% \subsection{Who is to blame?}
% \LG{} is not the work of a single one. The program is based on
% \textsf{vgrind} by Dave Presotto \& William Joy of UC Berkeley.
% Van Jacobson wrote \textsf{tgrind} for \TeX. Jerry Leichter
% of Yale University modified it for \LaTeX. George V. Reilly of 
% Brown University changed the name to lgrind and added the
% program-text-within-comments and @-within-\LaTeX{} features,
% and finally
% Michael Piefel of the Humboldt-University Berlin converted it to
% work under \LaTeXe, i.\,e. with \NFSS, and improved the documentation.
% However, there have been many contributors who supported the development; in
% particular the range of supported languages is mainly due to them.
% Unfortunately I do not know all of them, but my thanks go to everybody.
% A special Thank You to
% Torsten Sch\"utze for his OS/2 support and many and various hints.
% \section{\LG{} -- grind nice program listings}
% \begin{tabbing}
% \texttt{lgrind} \= \texttt{[-s] [-e] [-i] [-o \meta{file}] [-n] [-c] [-t 
% \texttt{] [-h }\meta{header}\texttt{]} \\
% \>\texttt{[-v }\meta{varfile}\texttt{] [-d[!] }\meta{description 
% \texttt{[-l}\meta{language}\texttt{] [}\meta{name}\texttt{\textbar-]}
% \end{tabbing}
% \LG{} processes its input file(s) and writes the result to standard
% output. This output can be saved for later editing, inclusion in a
% larger document, etc.
% The options are:
% \makebox[2em][l]{-e} process a \LaTeX-file for embedded text.
% \makebox[2em][l]{-i} process for inclusion in a \LaTeX-document.
% \makebox[2em][l]{-}  take input from standard input.
% \changes{v3.5}{1998/05/23}{Output redirection (-o) implemented.}
% \makebox[2em][l]{-o} redirect output.
% \makebox[2em][l]{-n} don't boldface keywords.
% \makebox[2em][l]{-a} don't treat @, etc. specially in \LaTeX.
% \makebox[2em][l]{-c} don't treat @, etc. specially in comments.
% \makebox[2em][l]{-t} change tab width (default 8).
% \makebox[2em][l]{-h} specifies text to go on the left side of every output 
% page (default is none).
% \makebox[2em][l]{-v} take variable substitution strings from file.
% \makebox[2em][l]{-d} specifies the language definitions file.
% \changes{v3.3}{1996/08/23}{"lgrindef" position can be changed permanently}
% \makebox[2em][l]{-d!} same as -d, except the change is permanent (modifies
% executable)
% \makebox[2em][l]{-l} specifies the language to use.
% \makebox[2em][l]{-s} shows a list of currently known languages.
% The standard for \LG{} is to take its input from the file given on the
% command line and write on standard output. You can change this behaviour with
% the options - and -o, respectively. Please note that as soon as a file is
% detected on the command line (either its name or a -) it is processed with the
% options then in effect, thus
% allowing to give multiple files on one line with possibly multiple targets.
% If neither -e nor -i are specified, a complete \LaTeX-file is produced.  When
% no language is given on the command line, \LG{} tries to figure out the
% language via the extension of the file. A table of extensions and their
% languages is in the definition file. If the extension is unknown, C is chosen
% as a default.
% When \LG{} is started without any parameters, it will show a short help 
% The same happens when the appropriate option is given, but this is
% implementation dependent (usually what is common for the operation system, the
% default for DOS is |-?| and for Unix |--help|).
% The position of the \texttt{lgrindef}-file is determined by giving it on the
% command line (highest priority), by defining an environment variable
% \texttt{LGRINDEF}, and by the position fixed in \LG{}s executable. The latter
% can be changed by using -d! and then using the newly created file as the new
% \LG{}.
% \changes{v3.5}{1998/05/30}{Now testing for \texttt{LGRINDEF} environment 
% The languages which are currently known are stored in the language definition
% file; their number increases more or less rapidly. At the time of writing the
% languages in the table below are part of the distribution.
% \begin{table}\begin{minipage}{0.8\linewidth}
% \let\thefootnote\thempfootnote
% \newcommand{\fnm}[1]{\textsuperscript{\itshape#1}}
% \begin{tabbing}
% \hspace*{2em} \= \hspace{9em} \= \hspace{9em} \= \\
% \> Ada             \> \LaTeX\footnote{John Leis, University of Southern 
Queensland, l...@usq.edu.au.}
%                                       \> RATFOR          \\
% \> Asm             \> LDL             \> RLaB\footnote{Jim Green, National 
Physical Laboratory, j...@cise.npl.co.uk} \\
% \> Asm68           \> Lex             \> Russell         \\
% \> BASIC\fnm{a}    \> Linda           \> SAS\footnote{Michael Friendly, 
frien...@hotspur.psych.yorku.ca} \\
% \> Batch\fnm{b}    \> Lisp (Emacs)    \> Scheme\footnote{Neale Pickett, 
npick...@watchguard.com} \\
% \> C               \> MASM\fnm{a}     \> sh              \\
% \> C++             \> MATLAB\fnm{a}   \> SICStus\fnm{b}  \\
% \> csh             \> ML\fnm{d}       \> src             \\
% \> FORTRAN         \> Mercury         \> SQL             \\
% \> Gnuplot\footnote{Denis Petrovic, denis.petro...@public.srce.hr} 
%                    \> model           \> Tcl/Tk\footnote{Alexander Bednarz, 
Forsch.-zentrum J\"ulich, a.bedn...@kfa-juelich.de} \\
% \> Icon            \> Modula2         \> VisualBasic\fnm{a} \\
% \> IDL\footnote{Diego Berrueta, di...@berrueta.net} 
%                    \> Pascal          \> VMSasm          \\
% \> ISP             \> PERL            \> yacc            \\
% \> Java            \> PostScript      \>                 \\
% \> Kimwitu++       \> PROLOG          \>
% \end{tabbing}\par
% \let\footnoterule\relax
% \end{minipage}\end{table}
% \changes{v3.67}{2002/01/28}{Added Scheme and ML language definitions}
% \subsection{Operation modes}
% There are three modes of operation: stand-alone, include and embedded.
% Use |lgrind -ly bary.y > bary.tex| (or |lgrind -o bary.tex bary.y|) to
% produce a stand-alone \LaTeX-file from, say, a Yacc file. This results in a
% document which is formatted using Piet van Oostrum's \textsf{fancyhdr.sty} to
% make the headers and footers. BTW: You really should have this package. It's
% marvellous. But of course you can change the layout to your likings by
% editing the \texttt{lgrindef}-file.
% To include a C-file named \texttt{foo.c} into your \LaTeX-document, first
% give the command: |lgrind -i -lc foo.c > foo.tex| This will generate
% \texttt{foo.tex}, which will have the pretty-printed version of
% \texttt{foo.c}.  Then include \texttt{lgrind.sty} as you include any other
% package, namely with |\usepackage{lgrind}| at the beginning of your
% \LaTeX-document. Having done this, within the document you can include
% \texttt{foo.tex} using |\lagrind| and |\lgrindfile| described in the next 
% Finally, for the embedded (and probably most powerful) mode, when you have a
% \LaTeX-file with embedded program listings, you can preprocess it with a 
command like:
% |lgrind -e pretty-sources.lg > even-prettier-sources.tex|
% and get a new \LaTeX-file which you then feed into \LaTeX. Commands you can
% use within embedded texts are described below.
% \section{Preparing documents}
% \subsection{Using the \LG.sty-file}
% The \LG{} package is included via the |\usepackage| command. You have to 
include it
% in your document preamble when you want to include listings and when using 
% mode. It is done automatically for stand-alone listings.
% Currently the following options are supported:
% \begin{description}
% \changes{v3.5}{1998/05/23}{Procedure names added to index in addition to
% printing them in the margin.}
% \item[procnames] prints the names of starting and, if nested
% procedures are allowed, continued procedures in the margin. Don't make the
% margin too small, or don't make the names too long \dots
% \item[noprocindex] do not put found procedure beginnings in the index
% \item[noindent] cancels the indentation. Useful for long listings or listings
% within their own sections.
% \item[fussy] lets \LaTeX\ print all overfull hboxes. The default is to 
% this for about a tenth of an inch.
% \item[norules] lets \LG{} suppress the surrounding rules for included material
% (using |\lagrind| and |\lgrindfile|).
% \item[nolineno] doesn't print line numbers.\footnote{To be exact, prints line
% numbers every 50,000 lines. But source code should never get so long in a
% single file~-- that's over 3 MByte! If you really want no numbers,
% set $\backslash$\texttt{LGnuminterval} to zero; then you won't get procedure
% names, either.}
% \item[lineno5] prints line numbers every 5 lines. The default is 10.
% \item[leftno] print line numbers in the left margin. Default is the right.
% \end{description}
% \subsection{Stand-alone and included listings}
% After processing a source code file with \LG{} without the -e or -i
% options you get a \LaTeX-file which can be directly compiled.
% When using -i \LG{} will produce a file which can be included with the
% following macros:
% \DescribeMacro{\lgrindfile}
% The first is |\lgrindfile{|\meta{file}|}|, which will simply include 
% the file \meta{file} at that point of text, and will draw horizontal
% lines before and after the listing.
% \DescribeMacro{\lagrind}
% The macro
% |\lagrind[|\meta{float}|]{|\meta{file}|}{|\meta{caption}|}{|\meta{label}|}|
% will put the listing also within a figure environment, using the \meta{float}
% options (h, t, b or p), \meta{caption} and \meta{label} you gave. The starred
% form of |\lagrind| will also use the starred |figure*|.
% Note that floats cannot be longer than one page, so you should only use
% |\lagrind| for short fragments, longer pieces should use |\lgrindfile| (which
% is non-floating).
% \DeleteShortVerb{\|}\MakeShortVerb{\"}\MakePercentIgnore
% \subsection{Embedded programs within a \LaTeX-file}
% You don't have to process every single source file with \LG{}, only to
% include it in your document. Within the text of your \LaTeX-file, you can
% mark groups of lines as program code, either text- or display-style to
% be specific. You can use several commands for controlling the inclusion
% of source code into your \LaTeX-file.
% Write your text, don't forget to include \LG.sty. Use the following
% commands. You can `debug' your text without including the lengthy
% listings. As a last step (but one), you process your file with \LG{}
% and its option -e, which will provide you with your final \LaTeX{}
% source file.
% The commands are similar to the math environments. \DescribeEnv{\%( \%)}
% With "%(" and "%)" you obtain code in text style, i.\,e. in the same line.
% \DescribeEnv{@ @}Surrounding the text with "@" is a shorthand.
% \begin{verbatim}
% The expression
% %(
% a + 3
% %)
% produces 10.\end{verbatim}
% produces the same as "The expression @a + 3@ produces 10." The output
% will have `a + 3' set as a program.
% \DescribeEnv{\%[ \%]}As with math, the square bracket equivalent produces
% display style listings, i.\,e. indented text on an own line.
% \DescribeEnv{\%*} As long listings tend not to fit on one page, there will
% be page breaks inserted. Since page breaks can considerably affect readability
% there will be none at all unless you insert lines consisting of just "%*".
% Pages will end here and only here, but not necessarily here. (That is, you
% allow (or recommend) a page break. It will be taken if needed.)
% \changes{v3.3}{1996/09/02}{Allow page breaks in embedded listings}
% \DescribeEnv{\%=} You can insert your own code by using a line starting with
% "%=" in the program text. Whatever you enter after that is left in the output,
% exactly as you typed it. It will be executed in a strange environment, so
% doing anything fancy is very tricky. \DescribeMacro{\Line} A macro, "\Line",
% is provided to help you do simple things. For example,
% \begin{verbatim}
% %[
% %=\Line{________\vdots}
%         a = 1;
% %]\end{verbatim}
% produces:
% \hspace*{4em}\vdots
% \hspace*{4em}\textit{a}\textsf{ = 1;}
% (Within the program text, \_ is active and expands to a
% fixed-width space.  A whole bunch of macros are also defined. If you
% understand how \LG{} sets lines up, you can replace the 8 \_'s
% with a call to "\Tab"---but I'll let you hang yourself on that one.)
% \DescribeEnv{\%<}The "%<"\meta{file} command includes \meta{file} as a
% program listing in your document. Before inclusion it will be pretty
% printed. This is the almost the same as \LG ing the \meta{file}
% separately and with -i and including it via "\lgrindfile", only that
% it's simpler for you. \DescribeEnv{\%!} With "%!"\meta{command} the
% input is taken from a shell command.
% While you can specify the language used on the command line, this does
% not suffice for mixed-language programs (or projects). The command
% \DescribeEnv{\%\#}"%#"\meta{language} provides you a means to change
% the language on the fly wherever you want.
% \DescribeEnv{\%@}The shorthand "@" is very useful, and since "@" is not usable
% in normal \LaTeX{} text there is no conflict. If, however, you use "@" in your
% text (after "\makeatletter") the result produced by \LG{} is not satisfactory.
% To disable the shorthand you can use a command line option, or locally "%@-".
% Using "%@+" will switch it on again.
% Important rules:
% \begin{itemize}
% \item "%" and the following character must be the first two characters on
% the line to be recognized.
% \item
% Put \emph{nothing} on the line after the "%" and the key character.
% If you do that, \LG{} will provide a default environment that will
% produce an "\hbox" for "%( )%", and a "\vbox" for "%[ %]".
% If you put stuff on the line, \LG{} assumes you want to control
% the format completely. Doing this requires understanding \emph{exactly}
% what the code \LG{} produces is doing. (Sometimes I'm not sure I do!)
% \item
% "%)" and "%]" are simply ignored outside
% of a code group, but any extra "%(" or "%[" produces a
% warning, so a missing "%)" or "%]" is usually caught.
% \item
% Remember that the code between "%("/"%[" and "%)"/"%]" is put into a
% single box. Expect the usual problems with long boxes! Use "%*" if needed.
% \end{itemize}
% \subsection{Formatting your source code}
% Well, \LG{} uses a different font for comments. This has as a consequence
% that functions you refer to are typeset differently in the program and
% in the comments, which is unsatisfactory. And, wouldn't it be great to
% use \LaTeX{} commands to produce e.\,g. `\copyright'?
% The \texttt{lgrindef}-file defines environments for exactly these
% purposes. They are usually defined as follows, but of course it is possible
% to use other strings if the standard collides with the syntax of the
% language in question.
% \DescribeEnv{\%\% \%\%}Text which is surrounded by "%%" is directly
% passed to \LaTeX, a pair of curled braces around it. So the copyright
% symbol would be produced with "%%\copyright%%".
% \DescribeEnv{\%\$ \$\%}The "%$"\meta{text}"$%" works much the same,
% except that \meta{text} is set in math mode.
% When \LG{} discovers a line that contains \emph{only} a comment beginning
% right at the start of the line and ending at the very end (no spaces), 
% \emph{only} \LaTeX{} text as in the environment described above, the line
% will be copied verbatim into the resulting \LaTeX{} document, with a newline
% appended. This allows (e.\,g., in C):
% \begin{verbatim}
% /*%%\section{Main program}%%*/
% int main()
% {
%     //%%\subsection{Variables}%%
%     int a;\end{verbatim}
% \changes{v3.6}{1999/05/28}{Complete \LaTeX{} lines allowed}
% The underscore which is normally the subscripting operator in math mode is
% used internally in \LG{}. You can still use the command "\sb" instead (and
% "\sp" for superscripts).\label{underscore}
% \DescribeEnv{\%| |\%}In "%|"\meta{text}"|%" a kind of verbatim
% environment is provided. \meta{Text} is typeset in typewriter.
% \DescribeEnv{@ @}Program text within a comment is surrounded by "@". The text
% is processed exactly as if it wasn't a comment. To produce an at-sign
% you have to use "@@".
% \subsection{Greater control\dots}
% Many things are controllable by re-defining various macros.  You can
% change what fonts \LG{} will use for various kinds of things, how
% much it indents the output, whether it adds line numbers, and if so at
% what interval it prints them and whether it sticks them on the left or
% right, and so on. This stuff is all described below in the code section,
% though probably not very well. The default settings produce output
% that looks reasonable to me, though I can't say I'm ecstatic about it.
% Doing a \emph{really} good job would require defining some special fonts.
% Nonetheless as an example my own private font setup. After having defined a
% font family called ttp (for typewriter proportional), using Boton (a
% commercial font which has a nice `code look' to it), I define:
% \begin{verbatim}
% \def\CMfont{\ttpfamily\itshape}
% \def\KWfont{\ttpfamily\bfseries}
% \def\VRfont{\ttpfamily}
% \def\BGfont{\ttpfamily}
% \end{verbatim}
% You can put these redefinitions in the preamble of your \LaTeX-file when
% using embedded and included mode; for stand-alone listings you have to put
% them into the \texttt{lgrindef}-file. This will change fonts for all modes.
% \subsection{Error messages}
% The output of \LG{} always contains exactly one output line for each input
% line.  Hence, you can look up line numbers in \TeX{} error messages in your
% original file, rather than in the \LG ed (LGround?) file.  (Of course, if
% the problem is in the \LG{} output\dots)
% \subsection{Variable substitution}
% \LG{} usually prints variables exactly the way they appear in the source
% code. However, very often one uses names for variables which really denote
% symbols and have special formatting, only that the input alphabet of the
% target language does of course not allow anything fancier than plain ASCII.
% I find myself using greek variables very often, because they are used in the
% problem domain. So there is a `delta' which really should be `$\delta$',
% there is a `gamma\_1' for `$\gamma\sb1$' and so forth. LG{} allows you to
% change those names back to what you desire by use of a variable substution
% file (using option -v).
% This file is very simple, and so is its parser. There is one substitution per
% line, giving the original name, an equality sign, and the text replacing the
% original:
% \begin{verbatim}
% delta=$\delta$
% gamma_1=$\gamma\sb1$
% \end{verbatim}
% You can do everything you want to here. Remember that \emph{usually} variable
% names are set upright and not in math mode. Therefore don't forget the
% dollar-sign, and use "\sb" instead of "_" (see section \ref{underscore}).
% \section{The \texttt{lgrindef}-file}
% The \texttt{lgrindef}-file is \LG's language definition data base.
% It is here where \LG{} learns what are keywords, what comments, where
% are functions, how to distinguish plain comments from \LaTeX-commands etc.
% The first field is just the language name (and any variants
% of it). Thus the C language could be specified to \LG{} as `c' or `C'.
% \subsection{Capabilities}
% Capabilities are of two types: Boolean capabilities which indicate that
% the language has some particular feature and string capabilities which
% give a regular expression or keyword list.
% Entries may continue onto multiple lines by giving a "\" as the last
% character of a line. Lines starting with "#" are comments.
% The following table names and describes each capability.
% \begin{description}
% \item[ab] Regular expression for the start of an alternate form comment
% \item[ae] Regular expression for the end of an alternate form comment
% \item[bb] Regular expression for the start of a block
% \item[be] Regular expression for the end of a lexical block
% \item[cb] Regular expression for the start of a comment
% \item[ce] Regular expression for the end of a comment
% \item[cf] (Boolean) Use specialized routine for detecting C functions
% \changes{v3.1}{1995/11/12}{C functions are now detected much more reliably}
% \item[id] String giving characters other than letters and digits
% that may legally occur in identifiers (default `\_')
% \item[kw] A list of keywords separated by spaces
% \item[lb] Regular expression for the start of a character constant
% \item[le] Regular expression for the end of a character constant
% \item[mb] Regular expression for the start of \TeX{} math within a comment
% \item[me] Regular expression for the end of \TeX{} math within a comment
% \item[np] Regular expression for a line that does \emph{not} contain
% the start of a procedure (e.\,g. prototypes)
% \item[oc] (Boolean) Present means upper and lower case are equivalent
% \item[pb] Regular expression for start of a procedure
% \item[pl] (Boolean) Procedure definitions are constrained to the lexical
% level matched by the `px' capability
% \item[px] A match for this regular expression indicates that procedure
% definitions may occur at the next lexical level. Useful for lisp-like
% languages in which procedure definitions occur as subexpressions of defuns.
% \item[rb] Regular expression for the start of block outside the actual
% code.\footnote{I included this especially for the \texttt{objects} and
% \texttt{records} in Pascal and Modula-2. They \emph{end} (with the "<"be">"
% expression), but shouldn't have any influence on the surrounding procedure.
% When defining \texttt{record} as normal block start, its \texttt{end} ends
% the procedure. Workaround: Make \texttt{record} itself a procedure start.
% But that prints a continuation mark when \textsf{procnames} is on.}
% \changes{v3.1}{1995/11/12}{Pascal objects now treated properly}
% \item[sb] Regular expression for the start of a string
% \item[se] Regular expression for the end of a string
% \item[tb] Regular expression for the start of \TeX{} text within a comment
% \item[tc] (String) Use the named language entry as a continuation of the
% current one
% \item[te] Regular expression for the end of \TeX{} text within a comment
% \item[tl] (Boolean) Present means procedures are only defined at the top
% lexical level
% \item[vb] Regular expression for the start of typewriter text within a
% comment
% \item[ve] Regular expression for the end of typewriter text within a comment
% \item[zb] Regular expression for the start of program text within a comment
% \item[ze] Regular expression for the end of program text within a comment
% \end{description}
% \egroup
% \subsection{Regular Expressions}
% \texttt{lgrindef} uses regular expressions similar to those of
% \textsf{ex} and \textsf{lex}. The characters `"^"', `"$"', `"|"', `:',
%%stopzone   % VIM syncing
% and `"\"' are reserved characters and must be `quoted' with a preceding
% "\" if they are to be included as normal characters.
% The meta-symbols and their meanings are:
% \begin{description}
% \item[\$] The end of a line
% \item[\^] The beginning of a line
% \item[$\backslash$d] A delimiter (space, tab, newline, start of line)
% \item[$\backslash$a] Matches any string of symbols (like `.*' in lex)
% \item[$\backslash$p] Matches any identifier. In a procedure definition
% (the `pb' capability) the string that matches this symbol is used
% as the procedure name.
% \item[( )] Grouping
% \item[$|$] Alternation
% \item[?] Last item is optional
% \item[$\backslash$e] Preceding any string means that the string will not
% match an input string if the input string is preceded by an escape character
% ("\"). This is typically used for languages (like C) that can include the
% string delimiter in a string by escaping it.
% \end{description}
% Unlike other regular expressions in the system, these match words and
% not characters. Hence something like `(tramp"|"steamer)flies?'
% would match `tramp', `steamer', `trampflies', or `steamerflies'.
% Contrary to some forms of regular expressions, \texttt{lgrindef}
% alternation binds very tightly. Grouping parentheses are likely to
% be necessary in expressions involving alternation.
% \subsection{Keyword List}
% The keyword list is just a list of keywords in the language separated
% by spaces. If the `oc' boolean is specified, indicating that upper
% and lower case are equivalent, then all the keywords should be specified
% in lower case.
% \subsection{Configuration options}
% \changes{v3.1}{1995/11/07}{The \LaTeX-text put into the files
% can be configured}
% In addition to the language definitions the \texttt{lgrindef}-file
% contains various configuration data. When the entries do not exist,
% default values are used:
% \begin{description}
% \item[firstpreamble] is the (\LaTeX-)text that comes at the beginning of
% an stand-alone file created by \LG{} from source code (it must contain
% "\begin{document}" somewhere).
% \item[postamble] is the (\LaTeX-)text that comes at the end of
% an stand-alone file (and must contain "\end{document}"). This is the place to
% put a "\printindex" if you wish so (don't forget "\usepackage{makeidx}" and
% "\makeindex" in the preamble).
% \item[filepreamble] is inserted before every processed source file in a
% stand-alone \LaTeX-file. In these two preambles you can use "\f", which
% will be substituted by the current input file (e.\,g. to put it into the
% header).
% \item[configuration] follows the opening of the \texttt{lgrind}-environment.
% This is used for redefining the macros used within it, e.\,g. the fonts
% or the width of a space (the "\@ts" unit).
% \item[chartab] is a list of characters that will be substituted by a
% \LaTeX-string. This is useful when you do (or can) not use any of the fancy
% methods to persuade \LaTeX{} into using your extended ASCII-characters.
% The format is a two digit hex number (the ASCII- (or whatever) value of the
% character), an equal sign, and the according \LaTeX{}-string, ended with a
% colon. You have to escape certain characters (like the backslash). So if
% you, e.\,g., have IBM ASCII code page 437 input and use the
% \textsf{german}-package, you can have your \"a using \verb+84="a+. Note
% that the substituting string must contain more than one character;
% otherwise it will be ignored. To print a `b' instead of an `a' you can
% use \verb+61={b}+.
% \end{description}
% \StopEventually{\PrintChanges\PrintIndex}
% \section{The Implementation of \LG.sty}
%    \begin{macrocode}
%    \end{macrocode}
% \setlength{\parindent}{0pt}
% \begin{macro}{\LGnuminterval}\begin{macro}{\lc@unt}\begin{macro}{\ln@xt}
% The counter "\LGnuminterval" represents the line numbering interval.
% Its default is 10, it is set by two options and can be changed everywhere
% you want to. "\lc@unt" counts the current line, "\ln@xt" contains the next
% line to get numbered.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}
% \begin{macro}{\LGleftnum}
% Line numbers are usually on the right. By setting "LGleftnum" to true
% or false this behaviour can be altered.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\LGindent}
% "\LGindent" is the indentation for all display style listing lines.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \changes{v3.4}{1997/02/05}{Rules around included material can be suppressed.
%                            New option norules.}
% \begin{macro}{\LGnorules}
% Normally \LG{} puts rules around everything that is included (via "\lagrind"
% and "\lgrindfile"), this can be changed with an option.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \changes{v3.4}{1997/01/30}{Prevent slightly overfull hboxes. New option 
% \begin{macro}{\LGsloppy}
% "\LGsloppy" is the amount that a horizontal box may be overfull without 
% a warning from \LaTeX. This is useful since there are often many boxes which 
% overfull by only a few points, and this does not really show since listings 
% very ragged.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\Proc}\begin{macro}{\ProcCont}
% There's a "\Proc{"\meta{ProcName}"}" at the start of each procedure.  If
% the language allows nested procedures (e.\,g. Pascal), there
% will be a "\ProcCont{"\meta{ProcName}"}" at the end of each inner procedure.
% (In this case, \meta{ProcName} is the name of the outer procedure. I.\,e.,
% "\ProcCont" marks the continuation of \meta{ProcName}).
% \begin{macro}{\DefaultProc}
% \changes{v3.0}{1995/09/24}{Reintroduced procedure names in the margins}
% \begin{macro}{\DefaultProcCont}
% Default is not to do anything with the name. Optionally the names are
% printed in the same margin as the line numbers. The name is put into a
% box which will be output whenever it is not empty.
%    \begin{macrocode}
\global\setbox\procbox=\hbox{\PNsize #1}}
\global\setbox\procbox=\hbox{\PNsize\dots #1}}}
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}
% \begin{macro}{\ifLGnoprocindex}
% \changes{v3.6}{1999/05/27}{Added option to suppress indexing of functions}
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% End of initialization, execute any options.
%    \begin{macrocode}
%    \end{macrocode}
% \changes{v3.0}{1995/09/28}{Added package options}
% \begin{macro}{\BGfont}\begin{macro}{\CMfont}\begin{macro}{\NOfont}
% \changes{v3.1}{1995/11/12}{Numbers can now have an own style (i.\,e. font)}
% \begin{macro}{\KWfont}\begin{macro}{\STfont}\begin{macro}{\TTfont}
% \begin{macro}{\VRfont}\begin{macro}{\PNsize}\begin{macro}{\LGsize}
% \begin{macro}{\LGfsize}
% These are the fonts and sizes for background (everything that doesn't fit
% elsewhere), comments, numbers, keywords, strings, verbatim text, variables,
% the procedure names
% in the margins, displayed code ("%[ ]%"), and included code ("\lgrindfile"
% and "\lagrind"), respectively. Note that the suffixes `font' and `size'
% have been chosen solely for the author's intention; you can do anything
% you want, e.\,g. "\tiny" comments. You have to use, however, font changes
% which don't require an argument.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}
% \end{macro}\end{macro}\end{macro}
% \end{macro}\end{macro}\end{macro}
% \end{macro}
% \begin{macro}{\ifLGinline}\begin{macro}{\ifLGd@fault}
% \begin{macro}{\LGbegin}\begin{macro}{\LGend}
% The flag "LGinline" is true for in-line code. "\LGbegin" and "\LGend" are
% default commands to open and close a code example and use it to perform
% certain ops depending whether we're in-line or display style."\LGend" is
% a no-op unless "\LGbegin" (where "LGd@fault" is set true) was executed,
% so you can provide explicit open
% code on the "%[" or "%(" line without providing any special code on the
% matching "%]" or "%)" line.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}
%%stopzone   % VIM syncing
% \begin{macro}{\ifc@omment}\begin{macro}{\ifstr@ng}
% These two conditions indicate if we are setting a comment or maybe
% a string constant, respectively.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}\end{macro}
% \begin{macro}{\ifright@}
% To get decent quotes (opening and closing) within comments, we remember
% whether the next one is going to be `{}``'{} or, if true, `{}'''{}.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ls@far}\begin{macro}{\tb@x}\begin{macro}{\TBw@d}
% These three are all for the sake of tabbing. "\ls@far" stores the
% ``line so far''. The tabwidth goes in "\TBw@d", whilst "\tb@x" is 
% merely a temporary variable for "\Tab" and setting "\@ts".
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}
% The underscore marks a point where the pre-processor wants a fixed-width
% space (of width "\@ts").
%    \begin{macrocode}
{\catcode`\_=\active \gdef\@setunder{\let_=\sp@ce}}
%    \end{macrocode}
% \begin{macro}{\lgrindhead}
% \begin{macro}{\lgrindfilename}\begin{macro}{\lgrindfilesize}
% \begin{macro}{\lgrindmodyear}\begin{macro}{\lgrindmodmonth}
% \begin{macro}{\lgrindmodday}\begin{macro}{\lgrindmodtime}
% We pollute the global namspace once more with these macros, for when they
% are used in the headers or footers, their values must still be known.
% Therefore they cannot be local to the "lgrind" environment.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}
% \end{macro}\end{macro}\end{macro}
% \begin{environment}{lgrind}
% This is the environment that eventually defines all necessary macros for
% formatting. All \LG ed text goes into such an environment, no matter if
% directly so or from within another one. It takes one optional argument,
% the line number.
%    \begin{macrocode}
%    \end{macrocode}
% \begin{macro}{\Line}
% The "\Line" macro is provided for use with "%=" in embedded listings.
% It's just there to hide the actual structure of this, for nobody 
% \emph{really} wants to know anyway.
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\Head}\begin{macro}{\File}
% \changes{v3.3}{1996/09/11}{given a meaningful definition}
% The next are primarily meant for stand-alone listings. "\Head" and
% "\File" are inserted by \LG, they define macros that contain
% a user-specified string (the header option -h), the name, size
% and modification time of the processed file. These can then be
% used e.\,g. in the headers and footers.
%    \begin{macrocode}
\newcommand{\File}[6]{\gdef\lgrindfilename{##1}\message{(LGround: ##1)}%
%    \end{macrocode}
% \end{macro}\end{macro}
% The "\Proc"s now get what
% was specified for them in the options section.
%    \begin{macrocode}
%    \end{macrocode}
% We set a "\hfuzz" to prevent some of the lesser overfull hbox warnings.
%    \begin{macrocode}
%    \end{macrocode}
% \begin{macro}{\NewPage}
% Each formfeed in the input is replaced by a "\NewPage" macro.  If
% you really want a page break here, define this as "\vfill\eject".
%    \begin{macrocode}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\L}
% Each line of displayed program text is enclosed by a "\L{"\dots"}". We 
% turn each line into an hbox. Firstly we look whether we are in-line.
% Every "\LGnuminterval" lines we output a 
% small line number in past the margin.
%    \begin{macrocode}
%    \end{macrocode}
% \begin{macro}{\r@ghtlno}\begin{macro}{\l@ftlno}
% Things get more difficult for display style listings. Here we set
% "\r@ghtlno" and "\l@ftlno" to no-ops, only to redefine them shortly
% after.
%    \begin{macrocode}
%    \end{macrocode}
% If there was a procedure name somewhere, "\procbox" is not empty and
% thus ready to be printed. Otherwise we test "\lc@unt" against "\ln@ext"
% to determine whether or not to print a line number.
%    \begin{macrocode}
     \global\advance\ln@xt by\LGnuminterval%
%    \end{macrocode}
% And once again when the line number is meant to be on the right.
%    \begin{macrocode}
     \global\advance\ln@xt by\LGnuminterval%
%    \end{macrocode}
% \end{macro}\end{macro}
% "\lc@unt" is incremented and everything is squeezed into a "\hbox".
%    \begin{macrocode}
  \global\advance\lc@unt by1%
  \hbox to \linewidth{\hskip\LGindent\l@ftlno ##1\egroup%
%    \end{macrocode}
% \end{macro}
% The initialization of "\lc@unt" and "\ln@xt". Every "lgrind"-environment
% starts over unless given a line number as argument.
%    \begin{macrocode}
\lc@unt=#1\advance\lc@unt by-1%
\ln@xt=\LGnuminterval\advance\ln@xt by-1%
\loop\ifnum\lc@unt>\ln@xt\advance\ln@xt by\LGnuminterval\repeat%
%    \end{macrocode}
% \begin{macro}{\LB}\begin{macro}{\Tab}
% The following weirdness is to deal with tabs. ``Pieces'' of a line
% between tabs are output as "\LB{"\dots"}". E.\,g., a line with a tab at
% column 16 would be output as "\LB{xxx}\Tab{16}\LB{yyy}". (Actually, to
% reduce the number of characters in the ".tex" file the "\Tab" macro
% supplies the 2nd \& subsequent "\LB"s.) We accumulate the "\LB" stuff in 
% an "\hbox". When we see a "\Tab", we grab this hbox (using "\lastbox") 
% and turn it into a box that extends to the tab position.  We stash this 
% box in "\ls@far" \& stick it on in front of the next piece of the line.  
% (There must be a better way of doing tabs but I'm not enough of a \TeX 
% wizard to come up with it. Suggestions would be appreciated. Oh, well,
% this comment's been in here for a decade. I don't believe in Santa Claus.)
%    \begin{macrocode}
 \advance\TBw@d by 1\@ts\ifdim\TBw@d>##1\@ts%
  \setbox\ls@far=\hbox{\box\ls@far \box\tb@x \sp@ce}\else%
  \setbox\ls@far=\hbox to ##1\@ts{\box\ls@far \box\tb@x \hfil}\fi\LB}%
%    \end{macrocode}
% \end{macro}\end{macro}
% A normal space is too thin for code listings.  We make spaces \& tabs
% be in "\@ts" units, which for displays are 80 \% the width of a ``0'' in the
% typewriter font. For inline stuff, on the other hand, we prefer a 
% somewhat smaller space -- actually, the same size as normal inter-word 
% spaces -- to help make the included stuff look like a unit.
%    \begin{macrocode}
\ifLGinline\def\sp@ce{{\hskip .3333em}}%
\else \setbox\tb@x=\hbox{\texttt{0}}%
      \@ts=0.8\wd\tb@x \def\sp@ce{{\hskip 1\@ts}}\fi%
\catcode`\_=\active \@setunder%
%    \end{macrocode}
% \begin{macro}{\CF}\begin{macro}{\N}\begin{macro}{\K}\begin{macro}{\V}
% \begin{macro}{\ic@r}
% \begin{macro}{\C}\begin{macro}{\CE}\begin{macro}{\S}\begin{macro}{\SE}
% Font changing. Since we are usually changing the font inside of a "\LB"
% macro, we remember the current font in "\CF" \& stick a "\CF" at the 
% start of each new box. Also, the characters ``\texttt{\char'042}'' and 
% ``"'"'' behave differently in comments than in code, and others behave 
% differently in strings than in code.
% "\N" is for numbers, "\K" marks keywords, "\V" variables, "\C" and "\CE"
% surround comments, "\S" and "\SE" strings.
% "\ic@r" inserts an optional "\/".
%    \begin{macrocode}
\def\N##1{{\NOfont ##1}\global\futurelet\next\ic@r}%
\def\K##1{{\KWfont ##1}\global\futurelet\next\ic@r}%
\def\V##1{{\VRfont ##1}\global\futurelet\next\ic@r}%
\def\C{\egroup\bgroup\CMfont \global\c@mmenttrue \global\right@false}%
\def\CE{\egroup\bgroup \global\c@mmentfalse}%
\def\S{\egroup\bgroup\STfont \global\str@ngtrue}%
\def\SE{\egroup\bgroup \global\str@ngfalse}%
%    \end{macrocode}
% \end{macro}\end{macro}\end{macro}\end{macro}\end{macro}
% \end{macro}\end{macro}\end{macro}\end{macro}
% \begin{macro}{\,}\begin{macro}{\!}
% We need positive and negative thinspaces in both text and math modes, so
% we re-define "\," and "\!" here.  The definition for "\," isn't really 
% needed for \LaTeX, but we try to be more complete.  Note that in \LaTeX{}
% terms, the new definition isn't robust, like the old~-- but nothing we 
% produce here is likely to be robust --~or \emph{needs} to be!~-- anyway!
%    \begin{macrocode}
\def\,{\relax \ifmmode\mskip\thinmuskip \else\thinspace \fi}%
\def\!{\relax \ifmmode\mskip-\thinmuskip \else\negthinspace \fi}%
%    \end{macrocode}
% \end{macro}\end{macro}
% Special characters. "\CH" chooses its first option alone in math mode;
% its second option in a string; and its third option, enclosed in "$"s,
%%stopzone   % VIM syncing
% otherwise. (At the moment, nothing is ever set in math mode, but you
% never know \dots)
%    \begin{macrocode}
\def\CH##1##2##3{\relax\ifmmode ##1\relax%
\else\ifstr@ng ##2\relax\else$##3$\fi\fi }%
\def\|{\CH|||}%  not necessary for T1
\def\<{\CH<<<}%  dto.
\def\>{\CH>>>}%  dto.
\def\-{\CH---}%  minus sign nicer than hyphen
\def\_{\ifstr@ng {\char'137}\else%
  \leavevmode \kern.06em \vbox{\hrule width.35em}%
  \ifdim\fontdimen\@ne\font=\z@ \kern.06em \fi\fi }%
\def\2{\CH\backslash {\char'134}\backslash }%          % \
\def\3{\ifc@mment\ifright@ ''\global\right@false%
                      \else``\global\right@true \fi%
   \else{\texttt{\char'042}}\fi}%                      % "
\def\5{{\texttt{\char'136}}}%                          % ^ 
%    \end{macrocode}
% Finally we don't want any indentation other than our own. We allow \LaTeX{}
% to stretch our listings a bit. Then we open a group, select the background
% font and (fanfare!) are ready to begin.
%    \begin{macrocode}
\parindent\z@\parskip\z@ plus 1pt%
%    \end{macrocode}
% This is the end of the "lgrind" environment. Rather short (in comparison!)
%    \begin{macrocode}
{\egroup\@@par}           % end of environment lgrind
%    \end{macrocode}
% \end{environment}
% The following are generated as part of opening and closing included
% code sequences. 
%    \begin{macrocode}
%    \end{macrocode}
% \begin{macro}{\lagrind}
% The "lagrind" environment is one of two for including files. It puts
% its argument inside a "figure" environment. It can be used without or
% with a star (first line), and with or without the usual floating
% arguments (second and third).
%    \begin{macrocode}

%    \end{macrocode}
% \begin{macro}{\@@lagrind}
% The unstarred version. Everything is pretty obvious, we open a "figure",
% put in a "minipage", input the file in question, make caption and label
% and that's it.
%    \begin{macrocode}
\vskip .5\baselineskip
\input #2\relax
\vskip .5\baselineskip plus .5\baselineskip
\ifLGnorules\else\hrule\fi\vskip .5\baselineskip
\vskip 2pt
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\@@slagrind}
% D\'ej\`a vu? The starred version got an asterisk attached to "figure".
%    \begin{macrocode}
\vskip .5\baselineskip
\input #2\relax
\vskip .5\baselineskip plus .5\baselineskip
\ifLGnorules\else\hrule\fi\vskip .5\baselineskip
\vskip 2pt
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \begin{macro}{\lgrindfile}
% This is similar. We draw lines above and below, no "figure". But it can
% get longer than one page.
%    \begin{macrocode}
    \vskip .5\baselineskip
    \input #1\relax
    \vskip .5\baselineskip
%    \end{macrocode}
% \end{macro}
% And now \dots
%    \begin{macrocode}
%    \end{macrocode}
% That's it. Thank you for reading up to here.
% Michael Piefel
% \vspace{2ex}
% \hrule
% \Finale
% \iffalse


This is a \LaTeX file with embedded code. Process it with \textsf{LGrind}
and its option -e.

%<example>%< lgrind.c

%<example>%< regexp.h
%<example>%< regexp.c

%<example>%< lgrindef.h
%<example>%< lgrindef.c

% \fi

Reply via email to