% $StyleId: rcs.doc,v 2.10 2003/02/02 20:42:38 schrod Exp $
% With modifications by Jeff Goldberg, December 2, 1999
%----------------------------------------------------------------------
% Written by Joachim Schrod <jschrod@acm.org>.
% Minor modification by Jeffrey Goldberg <J.Goldberg@Cranfield.ac.uk>
% Copyright conditions see below.

%
% LaTeX package rcs
% to use RCS tag values in one's document, and typeset revision logs
%
% [LaTeX in MAKEPROG]
% (history at end)

% If you have received this style file without the user manual (in the
% file rcs-user.tex or the respective DVI file), it's incomplete and
% near to useless. If it was given to you as something that you shall
% use as an author -- complain bitterly to your provider. You need the
% documentation and you have a right on it! (Below you can find info
% where to get the reference version.)



%%%%
%%%%
%%%% These TeX macros were documented with the documentation system
%%%% MAKEPROG and automatically converted to the current form.
%%%% If you have MAKEPROG available you may transform it back to
%%%% the original input: Remove every occurence of three percents
%%%% and one optional blank from the beginning of a line and remove
%%%% every line which starts with four percents.  The following Perl
%%%% command will do this:
%%%%
%%%%	perl -pe 's/^%%%%.*\n//; s/^%%%( )?//'
%%%%
%%%%	If you just want to print the documentation you may fetch
%%%% the archive print-makeprog.tar.Z from ftp.th-darmstadt.de (directory
%%%% pub/tex/latex). It contains *all* used styles -- but beware, they
%%%% may not be in a documented form...
%%%%
%%%%
%%% \documentclass{progltx}

%%% \usepackage{rcs-doc}
%%% \usepackage{path}
%%% \usepackage{alltt}
%%% \usepackage{fullpage}


%%% \RCS $StyleDate: 2003/02/02 20:42:38 $


%%% \begin{document}


%%% \title{The \texttt{rcs} Package\\
%%%     {\large (Implementation)}%
%%%     }
%%% \author{%                       % LaTeX does not discard unnecessary glue...
%%%     Joachim Schrod%
%%%     \thanks{%
%%%         \protect\raggedright
%%%         R\"odermark, Germany.
%%%         Email: \texttt{jschrod@acm.org}%
%%%         }%
%%%     }
%%% \date{%
%%%     Revision \RCSStyleRevision\\
%%%     (as of \RCSStyleDate)%
%%%     }

%%% \maketitle



%%% % ------------------------------------------------------------
%%% %
%%% % subdocument: The user interface of rcs.sty
%%% %

%%% \input{rcs-user}

%%% %
%%% % ------------------------------------------------------------
%%% %
%%% % subdocument: The internal interface
%%% %

%%% \input{rcs-conf}

%%% %
%%% % ------------------------------------------------------------



%%% \chap Implementation.

%%% This module reserves the namespace |rcs|.

%%% \beginprog
\ifx \rcs@loaded\undefined
    \def\rcs@loaded{$StyleRevision: 2.10 $}
\else
    \PackageWarningNoLine{rcs}%
	{Some other package already uses namespace `rcs'}
\fi
%%% \endprog


%%% \sect Let's identify this package.

%%% Now I'm caught in a catch-22 situation. The code below---part of my
%%% standard templates for \LaTeX{} packages---uses principles that are
%%% explained later in this document. In particular, the explanation how
%%% the information from the following RCS fields are gathered starts at
%%% chunk~\ref{chunk:value-split}.

%%% There you'll find also the explanation for the funny strings behind
%%% the RCS fields. Actually, I would not need them any more, they make
%%% the initial document (before the first check-in) work. But they won't
%%% be discarded in this package source, as they'll demonstrate another
%%% utilization of RCS at work.

%%% \beginprog
\begingroup
    \def\RCSPackage#1#2 $#3: #4 #5\endRCS $#6: #7 #8\endRCS{%
	\def\date{#4}\def\id{v#7}%
	\ProvidesPackage{#1}[\date\space\id\space #2]%
	}

  \RCSPackage{rcs}{typeset info from RCS fields}
  $StyleDate: 2003/02/02 20:42:38 $: 9999/00/00 \endRCS
  $StyleRevision: 2.10 $: 0.0 \endRCS
\endgroup
%%% \endprog


%%% \sect This module is supported. Send bug reports, comments, and
%%% repairs.

%%% The reference version may be retrieved via anonymous ftp from CTAN,
%%% directory \path|/tex-archive/macros/latex/contrib/supported/rcs|.


%%% \sect This is freely distributable software; you can redistribute it
%%% and/or modify it under the terms of the GNU General Public License
%%% (version~2) as published by the Free Software Foundation.

%%% This software is distributed in the hope that it will be useful, but
%%% \textbf{without any warranty}; without even the implied warranty of
%%% \textbf{merchantability} or \textbf{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 in
%%% the file |License| along with this package; if not, write to the Free
%%% Software Foundation, Inc., 675~Mass Ave, Cambridge, MA~02139,~USA.


%%% \sect Before we start we declare some shorthands for category codes.
%%% By declaring the underscore~`(|_|)' as letter we can use it in our
%%% macros. (I agree with {\sc D.~Knuth} that
%%% |\identifier_several_words_long| is more readable than
%%% |\IdentifierSeveralWordsLong| and in every case better than
%%% |\p@@@s|.) As this is a \LaTeX{} style option the at sign is a letter
%%% anyhow; so we can use the ``private'' Plain and \LaTeX{} macros; and
%%% with the underscore we can make our own macros more readable.  But as
%%% we have to restore this category code at the end of this macro file
%%% we store its former value in the control sequence |\CatUsCode|. This
%%% method is better than to use a group because not all macros have to
%%% be defined global this way.

%%% \beginprog
\ifx \CatEscape\undefined
    \chardef\CatEscape=0
    \chardef\CatOpen=1
    \chardef\CatClose=2
    \chardef\CatIgnore=9
    \chardef\CatLetter=11
    \chardef\CatOther=12
    \chardef\CatActive=13

    \chardef\CatUsCode=\catcode`\_
\fi

\catcode`\_=\CatLetter                  % top level macro file
%%% \endprog




%%% \chap Parsing RCS fields.

%%% Each RCS field consists of a keyword and an optional value. The value
%%% is delimited by `\verb*|: |' from the keyword, this delimiter is
%%% missing if no value does exist. At the end of a value there's always
%%% a space. I.e., the possible input looks like one of the two following
%%% lines:
%%%  %
%%% \begin{quote}
%%%   \itshape
%%%   |\RCS $|Keyword\/|$|\\
%%%   |\RCS $|Keyword\/\verb*|: |value\/\verb*| $|
%%% \end{quote}
%%%  %
%%% where \textit{Keyword} and \textit{value} may be arbitrary strings. In
%%% fact, we assume that \textit{Keyword} is a sequence of letters---only
%%% then the creation of a cseq |\RCS|\textit{Keyword} is sensible. Our
%%% implementation will not produce an error with an obscure keyword, but
%%% the result will not be of use for the author.
%%% We must not depend on a fixed set of keywords, this style shall be
%%% usable with a configurable RCS version. In such a version one can
%%% bind arbitrary keywords to the information available in RCS\@. This
%%% means we will parse all RCS fields with the same macro,
%%% |\rcs_split_field|. This macro will place its results in two macros:
%%% |\RCS_keyword| and |\RCS_value|. Afterwards we can bind the expansion
%%% of |\RCS_value| to the target cseq.


%%% \sect |\RCS_value| will be set to the expansion of |\RcsEmptyValue|
%%% if there was no RCS value available, i.e., if the field was not
%%% processed by RCS yet. This way the values of non-expanded RCS fields
%%% can be configured. The default value is the empty token list.

%%% \beginprog
\def\RcsEmptyValue{}
%%% \endprog


%%% \sect For those fields where a special handling is demanded, a cseq
%%% |\RcsHandle|\textit{Keyword} must exist. This cseq must handle the
%%% parsing of the value. Usually this means that it will create a macro
%%% named |\RCS|\textit{Keyword} with a transformed (or filtered) value.
%%% If you happen to write a handler you might want to look at the
%%% definition of |\RcsHandleDate| starting at section~\ref{sec:date}
%%% If a special handler for a field exists, the original value is
%%% placed automatically in the macro |\RCSRaw|\textit{Keyword}. This is done
%%% before the handler is called.


%%% \sect \label{chunk:value-split} The macro
%%% |\rcs_split_field| will split the field into |\RCS_keyword| and
%%% |\RCS_value|. This split is done by the argument parsing capability
%%% of \TeX{}: We can separate parameters by token lists and \TeX{} will
%%% use a shortest pattern match to determine the arguments.

%%% Let's look what kind of pattern we can use. As an example we use the
%%% RCS field `\verb*|$Revision: 2.11 $|', i.e., the unexpanded field is
%%% `|$Revision: 2.11 $|'. The macro |\RCS| will use dollars to delimit its
%%% argument, we don't need to worry for them. Eventually we'll have to
%%% discard the space of the expanded field's value---after all,
%%% |\RCSRevision| shall expand to `1.1', not to `1.1~'.

%%% First, we can see that the token list $\it (colon, space)$ may be used
%%% as the seperator of keyword and value:
%%% %
%%% \begin{quote}
%%%   \macroCall{\\rcs_split_field \[Revision\(: \)1.1 \]\\end_value}
%%% \end{quote}
%%% %
%%% ($d_i$ is the separating token list between parameter $i$ and~$i+1$.
%%% |#1| denotes the argument of |\RCS|. |\end_value| is a stop token,
%%% used to delimit the last parameter; we have to supply this token at
%%% the call. In fact, all tokens marked by $s$ and the dotted
%%% line---i.e., those after |#1|---are supplied by us.)

%%% When we want this token list as a separator, we have to supply it for
%%% the unexpanded field, since there is no $\it (colon, space)$ in there:
%%% %
%%% \begin{quote}
%%%   \macroCall{\\rcs_split_field \[Revision\]\(: \)\\end_value}
%%% \end{quote}
%%% %
%%% Then the call with the expanded field gets
%%% %
%%% \begin{quote}
%%%   \macroCall{\\rcs_split_field \[Revision\(: \)1.1 \]: \\end_value}
%%% \end{quote}

%%% Well, we want neither the space nor the colon and the other space
%%% after the value---we have to get rid of them. We can utilize the fact
%%% that no RCS field will ever have the token list $\it (space, colon)$
%%% in it. This means we can use it as a terminator for the RCS value:
%%% %
%%% \begin{quote}
%%%   \macroCall{\\rcs_split_field \[Revision\(: \)1.1\( \]:\) \\end_value}
%%% \end{quote}
%%% %
%%% Please note, one token of the delimiter~$d_2$ comes from the
%%% |#1|~argument, the other from the tokens we'll add at the call.

%%% Let's check our unexpanded field again---we don't have $d_2$ there. We
%%% have to add it by the caller:
%%% %
%%% \begin{quote}
%%%   \macroCall{\\rcs_split_field \[Revision\]\(: \)\{\}\( :\)\\end_value}
%%% \end{quote}
%%% %
%%% (Without the empty group the two spaces would be combined into one
%%% token.) Going back to the expanded field, we notice that these
%%% additional tokens do not harm us---we'll ignore everything after $d_2$
%%% anyhow.
%%% %
%%% \begin{quote}
%%%   \macroCall{\\rcs_split_field \[Revision\(: \)1.1\( \]:\) \{\} :\\end_value}
%%% \end{quote}

%%% We can take another look at the whole affair and point out the
%%% parameters we get for the split. Remember, that spaces after a control
%%% word belong to this word. First, the expanded field:
%%% %
%%% \begin{quote}
%%%   \macroCall
%%%       [p]{\\rcs_split_field \[\(Revision\): \(1.1\) \]:\( \{\} :\)\\end_value}
%%% \end{quote}
%%% %
%%% Then the unexpanded one:
%%% %
%%% \begin{quote}
%%%   \macroCall[p]{\\rcs_split_field \[\(Revision\)\]: \(\{\}\) :\(\|\)\\end_value}
%%% \end{quote}
%%% %
%%% ($p_3$, the third parameter will be empty in this case.)

%%% \medskip

%%% \noindent As explained above: if the value does not exist, a default
%%% value is supplied.

%%% \beginprog
\def\rcs_split_field #1: #2 :#3\end_value{%
    \def\RCS_keyword{#1}%
    \def\RCS_value{#2}%
    \ifx \RCS_value\empty
	\let\RCS_value\RcsEmptyValue
    \fi
    }
%%% \endprog


%%% \sect |\RCS| triggers the split. After it has occured, we can bind the
%%% expansion of |\RCS_value| to a new name. This name is constructed from
%%% the prefix |RCS|, the string |Raw| iff a handle for this field exists,
%%% and the keyword itself.

%%% Afterwards we call the handle, on non-existance this will be a no-op.

%%% Gathering the field is a bit more complicated, as we need to pay
%%% attention to underscores in file names.

%%% \beginprog
\def\RCS{%
    \bgroup
        \catcode`\_ \CatActive
	\RCS_get_argument
    }
\def\RCS_get_argument $#1${%
        \gdef\RCS_argument{#1}%
    \egroup
    \RCS_process
    }

\def\RCS_process{%
    \expandafter\rcs_split_field \RCS_argument: {} :\end_value
    \expandafter\let \csname RCS\rcs_raw_if_handle\RCS_keyword\endcsname
		     \RCS_value
    \csname RcsHandle\RCS_keyword\endcsname
    }
\def\rcs_raw_if_handle{%
    \@ifundefined{RcsHandle\RCS_keyword}{}{Raw}%
    }
%%% \endprog




%%% \chap Transforming the value of \Date{} fields.
%%% \label{sec:date}

%%% When we implement the special handling for the \Date{} field, we take
%%% care that we don't rely on the fact that the keyword is |Date|.
%%% Certain RCS versions (one of them created at our site) allow the
%%% configuration of the used keywords. I.e., there might be another
%%% keyword used for this field, but we still want to use the
%%% functionality provided here.
%%% For example, let's assume the keyword of the \Date{} field is
%%% |StyleDate|. Then the assignment
%%% %
%%% \begin{quote}
%%%   |\let\RcsHandleStyleDate=\RcsHandleDate|
%%% \end{quote}
%%% %
%%% is all that's needed to turn on the `date special handling' for this
%%% field. (In fact, that's done in this style file you're reading.)


%%% \sect The special handler for \Date{} fields sets up
%%% |\RCS|\textit{Keyword} (|\RCSDate| usually) to produce a textual
%%% representation of the date part of the \Date{} value in a
%%% |\today|-like format. Thus the constructed tag can be used in many
%%% circumstances, e.g., as the argument of the |\date| tag on the
%%% titlepage.

%%% In addition, |\RCSTime| is defined; it expands to the checkin time.

%%% If the value is empty, i.e., if there's no \Date{} value from RCS
%%% available, we use the current date. Later we might supply the current
%%% time, too---currently |\RCSTime| expands to an empty token list in
%%% such a case.
%%% Of course, this whole special handling is only done if |\today| is
%%% defined in some way; otherwise we just handle the \Date{} field like
%%% every other RCS field: The value is stored unfiltered in
%%% |\RCS|\textit{Keyword}.

%%% \beginprog
\def\RcsHandleDate{%
    \ifx \today\undefined
	\expandafter\let \csname RCS\RCS_keyword\endcsname \RCS_value
    \else
	\ifx \RCS_value\RcsEmptyValue
	    \@namedef{RCS\RCS_keyword}{\today}%
	    \let\RCSTime\empty
	\else
	    \expandafter\rcs_set_date \RCS_value\end_date
	\fi
    \fi
    }
%%% \endprog


%%% \sect If the value is not empty we split it using the argument parsing
%%% mechanism of \TeX{}. Then we'll define the value using the current
%%% definition of |\today|. This way language-specific styles (e.g., from
%%% the \textsf{babel} system) may supply their own way to present a date.

%%% Modified put the group and |\day|, |\month|, and |\year| setting
%%% within the definition of RCSDate to avoid having to do an edef
%%% which would expand |\today| before babel could redefine it.  ---JPG

%%% Further modified to use simple character parsing to allow for both
%%% / delimiter used by original RCS and - delimiter used by modern
%%% CVS, taking care that old 2-digit RCS years are still handled
%%% properly.  ---KHR

%%% \beginprog
\def\rcs_set_date #1 #2\end_date{%
    \expandafter\def \csname RCS\RCS_keyword\endcsname
    {\begingroup\rcs_set_date_really#1\today\endgroup}%
    \def\RCSTime{#2}%
    }

\def\rcs_set_date_really#1#2#3#4#5{%
    \ifx#3/ % Very old RCS file with two digit year...
	\year 19#1#2		% <-- space!
	\def\_{\rcs_set_date_really_ii#4#5}\expandafter\_
    \else
	\year #1#2#3#4		% <-- space!
        \expandafter\rcs_set_date_really_ii
    \fi}

\def\rcs_set_date_really_ii#1#2#3#4#5{%
    \month #1#2		% <-- space!
    \day #4#5		% <-- space!
    }
%%% \endprog



%%% \chap Typesetting revision logs.

%%% If a revision log is to be typeset, we demand a special format.
%%% Namely, it is to be enclosed in the |rcslog| environment. This
%%% environment has an optional parameter, the configuration tokens. The
%%% contents of the environment is the \Log{} field followed by a list of
%%% revision entries. The \Log{} field must be on a line of its own. Each
%%% revision entry is introduced by a line consisting of the tag
%%% |\Revision|, the revision number, the checkin date, the checkin time,
%%% and the author's user id.

%%% I.e., the input looks like the following example sketch, where italic
%%% material is substituted by current values.
%%% %
%%% \begin{quote}
%%%   \chardef\\=`\\                        % just for the example
%%%   \chardef\{=`\{
%%%   \chardef\}=`\}
%%%   \begin{alltt}\frenchspacing
%%% \\begin\{rcslog\}
%%% $Log: rcs.doc,v $
%%% Revision 2.11  2003/02/02 22:30:22  schrod
%%% Removed dependeny on \LaTeX2.09 font selection commands. (Though
%%% they're still used in the package documentation, as convenience.)

%%% \\Revision {\it revision date time uid}
%%% \qquad {\it log text\/}

%%% \quad \([\,\ldots\) {\it perhaps more entries\/} \(\ldots\,]\)

%%% \\end\{rcslog\}
%%%   \end{alltt}
%%% \end{quote}

%%% We typeset the revision log similar to a |description| environment.
%%% Each revision entry is an item.


%%% \sect The tag for the log intro is implemented as a token register. In
%%% fact, |\rcsLogIntro| is a hook.

%%% \beginprog
\newtoks\rcsLogIntro
	\rcsLogIntro={}
%%% \endprog


%%% \sect The author names are stored in a table named |rcs_author|. We
%%% supply the lookup macro |\RCS_get_author|, which expands to the table
%%% value or to the table key itself.

%%% \beginprog
\def\rcsAuthor#1#2{\@namedef{rcs_author:#1}{#2}}
\def\RCS_get_author#1{%
    \@ifundefined{rcs_author:#1}{#1}{\@nameuse{rcs_author:#1}}%
    }
%%% \endprog


%%% \sect Since this style option shall be usable with many styles, we
%%% must provide means to configure its behaviour. In particular, it must
%%% be possible to configure the layout. What remains fixed, is the layout
%%% of the log like a |description| list. I might add this flexibility if
%%% there is enough interest (or if somebody sends me tested changes).

%%% The first thing to parametrize is the overall layout and the header.
%%% This is set up by |\RcsLogStyle|. This cseq \emph{must} define
%%% |\RcsLogHeading|, which is the macro to set the header. E.g., here
%%% one could insert a |\newpage| to start a new page for the revision log.

%%% In addition, we reduce the base size of the log. That's the reason why
%%% these two configurations are combined in one macro---if you change
%%% one, you also have to consider to change the other.

%%% \beginprog
\def\RcsLogStyle{%
    \def\RcsLogHeading{\subsubsection*}%
    \footnotesize
    }
%%% \endprog


%%% \sect The header itself is created by |\rcs_log_hdr|, which is called
%%% with the expanded \Log{} value as its argument, terminated by
%%% |\end_value|.  We use \TeX{}'s parameter parsing mechanism to chop off
%%% the tail of the value---if the value exists, it always ends in~`|,v|'.
%%% Since only an existing value ends in~`|,v|', we have to supply a
%%% fitting empty value. We will use |\rcs_empty_log_value| for that.

%%% The empty log value gets a bit more complicated: If we have an empty
%%% \Log{} value, we will not have any revision entries. (Well, except in
%%% pathological cases.  Then we will produce spurious text.) Missing
%%% revision entries are missing items; since \LaTeX{} will complain
%%% about a missing |\item|, we must provide one to stop it from doing
%%% so. But here is no place for such tokens, the list environment hasn't
%%% started yet. We will set the flag |@rcs_empty_log@| and check this
%%% flag after we invoked the list.

%%% \beginprog
\def\rcs_log_hdr#1,v\end_value{%
    \@rcs_empty_log@false
    \RcsLogHeading{\RcsLogHeadingName #1}%
    \the\rcsLogIntro
    }
\def\rcs_empty_log_value{%
    \global\@rcs_empty_log@true		% that's most probably in a group!
    \RcsUnknownFile			% text for unknown file name
    ,v%					% assert calling env
    }
\newif\if@rcs_empty_log@
%%% \endprog


%%% \sect The user might want to adapt the text output. He can do so by
%%% redefining |\RcsLogHeadingName|, |\RcsUnknownFile|, and |\RcsEmptyLog|.

%%% \beginprog
\def\RcsLogHeadingName{Revision Log for \ttfamily} % file name in monospace
\def\RcsUnknownFile{\rmfamily\mdseries $\langle\,$Unknown file name$\,\rangle$}
\def\RcsEmptyLog{{\itshape No log entries available.}}
%%% \endprog


%%% \sect OK, we have fixed the header, let's address the entries. Each
%%% entry is typeset by |\RcsLogRevision| which is called with four
%%% arguments: (1)~the revision number, (2)~the checkin date, (3)~the
%%% checkin time, and (4)~the uid of the author.

%%% Such an entry is inserted as an item in the list. The uid is
%%% transformed to an author name, if the respective information is
%%% available. The checkin timestamp is mapped to a neater appearance by
%%% |\RcsLogDate| and |\RcsLogTime|.

%%% \beginprog
\def\RcsLogRevision#1#2#3#4{%
    \item [Revision #1]%
      (created at \RcsLogDate #2\end_date\space \RcsLogTime{#3} % <-- space!
      by \RCS_get_author{#4})\\\relax
    }
%%% \endprog


%%% \sect |\RcsLogDate| sets the date. We can use \TeX{}'s parameter parsing
%%% mechanism to separate the date value (passed as the argument) in its
%%% sub-values.  The code is very similar to |\rcs_set_date|.

%%% |\RcsLogTime| sets the time. As explained in the user manual, the time
%%% is ignored usually. (I don't think that this info is relevant in a
%%% printout.) Here we assume something about the context it's expanded
%%% in: Spaces will be before and after the tag. One of them must be
%%% discarded. The user option |\settime| is implemented here as a
%%% private macro, too.

%%% \beginprog
\def\RcsLogDate #1\end_date{%
    \begingroup\rcs_set_date_really#1\today\endgroup
    }

\def\RcsLogTime#1{\ignorespaces}
\def\rcs_settime{%
    \def\RcsLogTime##1{[##1]}%
    }
%%% \endprog


%%% \sect Well, that are all possibilities to configure the revision log.
%%% Let's put together the pieces. Within the |rcslog| environment the
%%% tags |\Revision| and |\settime| are used, they are implemented in
%%% the |rcs| namespace.

%%% At the start, we use the optional argument of the environment to
%%% configure it. |\settime| might be placed here, or font switches, etc.

%%% The next entity shall be the \Log{} field.  In fact, it shall be on
%%% the next line. If an optional environment argument was there, the
%%% newline is still in the input and must be discarded before we can use
%%% |\RCS| to parse the field.

%%% \beginprog
\def\rcslog{%
    \@ifnextchar[%			% ] (Emacs)
	\rcslog_configure
	{\rcslog_configure[]}%
    }
\def\rcslog_configure[#1]{%
    \let\Revision\rcslog_revision
    \let\settime\rcs_settime
    \RcsLogStyle
    #1%
    \afterassignment\rcslog_skipcr
    \let\next
    }
%%% \endprog


%%% \sect There are users---programmer's life would be so nice without
%%% them. An empty line (i.e., a |\par|) might have been inserted between
%%% |\begin{rcslog}| and the \Log{} field. There might be even other
%%% garbage between them. We have to check these possibilities and provide
%%% proper responses.

%%% \beginprog
\def\rcslog_skipcr{%
    \ifx \next $%
	\def\next{\rcslog_field $}%
    \else\ifx \next\par
	\let\next\rcslog_field
    \else
	\PackageError{rcs}{Missing RCS Log field in environment}{%
The RCS Log field must be the very first text in the rcslog\MessageBreak
environment. I will discard the rest of your input line and try to\MessageBreak
recover then. If you think this won't work, exit now.\MessageBreak
Otherwise press <Return> and cross your fingers!}
	\let\next\rcslog_discard_line
    \fi\fi
    \next
    }
%%% \endprog


%%% \sect This is the code for the luser. He provided tokens we didn't
%%% wait for---we discard them and check anew if we've got it right now.
%%% (Remember, we're waiting for the `|$|' of the \Log{} field.) %$ Emacs

%%% \beginprog
\def\rcslog_discard_line{%
    \begingroup
	\obeylines
	\rcslog_gobble_line
    }
\begingroup
\obeylines
\gdef\rcslog_gobble_line #1^^M{%
    \endgroup%
    \afterassignment\rcslog_skipcr%
    \let\next%
    }
\endgroup
%%% \endprog


%%% \sect Yep, we've reached a dollar; let's hope that it's our \Log{}
%%% field. It must be on one line, this line is fetched first and then
%%% analyzed.

%%% \beginprog
\def\rcslog_field{%
    \begingroup
	\obeylines
	\catcode`\_ \CatActive
	\rcslog_get_field
    }

\begingroup
\obeylines
\gdef\rcslog_get_field $#1$^^M{%
        \gdef\RCS_argument{#1}%
    \endgroup%
    \rcslog_set_field%
    }
\endgroup
%%% \endprog


%%% \sect When we analyze the \Log{} field, we must supply the default
%%% value. Within the log list there will be no other RCS fields,
%%% therefore we can redefine |\RcsEmptyValue| in this environment.

%%% The header gets passed the expanded \Log{} value, as explained above.
%%% We use |\RCS_value| instead of |\RCSLog| since we don't know which
%%% RCS keyword is used.

%%% At last, we start the list and check the |@rcs_empty_log@| flag,
%%% perhaps we have to supply a dummy |\item|.

%%% \beginprog
\def\rcslog_set_field{%
    \let\RcsEmptyValue\rcs_empty_log_value
    \RCS_process
    \expandafter\rcs_log_hdr \RCS_value\end_value
    \list{}\RcsLogListStyle		% empty label, layout configurable
    \if@rcs_empty_log@
	\item []\RcsEmptyLog		% supply missing item
    \fi
    }
%%% \endprog


%%% \sect The setup of the log list is done by a macro from the protected
%%% interface, the user shall be able to change it. In the default
%%% supplied, there is no skip between paragraphs, but a quad indentation.
%%% |rcslog| is here no list environment in the sense like |itemize| or
%%% |enumerate|, therefore a usage of these environments within an entry
%%% shall result in top-level item markers.

%%% \textbf{Caveat}:\quad Since we globally set |\@listdepth| at the start
%%% of the environment, we must not execute |\endlist| that decrements it.
%%% Instead we use |\endtrivlist| for the termination of the |rcslog|
%%% environment. It might be, that this must be changed if
%%% |\RcsLogListStyle| is changed.

%%% \beginprog
\def\RcsLogListStyle{%
    \global\@listdepth\z@
    \labelwidth\z@		    % item label w/o fixed size
    \itemindent -\leftmargin	    % move label before start of line
    \advance\itemindent \labelsep   % label was moved too much
    \def\makelabel##1{\bf ##1}%	    % how to typeset labels
    \parsep\z@skip		    % no space between pars in items
    \listparindent 1em		    % <-- space!  indent pars in items
    }%

\let\endrcslog\endtrivlist
%%% \endprog


%%% \sect The arguments of |\Revision| are separated by spaces and
%%% terminated by the end of the line. We have to fetch them, then we can
%%% pass them to |\RcsLogRevision| which does the real work.

%%% \beginprog
\def\rcslog_revision{%
    \begingroup
	\obeylines
	\rcslog_get_revinfo
    }

\begingroup
\obeylines
\gdef\rcslog_get_revinfo #1 #2 #3 #4^^M{%
    \endgroup%
    \RcsLogRevision{#1}{#2}{#3}{#4}%
    }
\endgroup
%%% \endprog




%%% \chap Convenience tags.

%%% As a shortcut we
%%% support the direct supply of the checkin date for the titlepage
%%% material. Here we also take care not to depend on the keyword.

%%% \beginprog
\def\RCSdate $#1${%
    \RCS $#1$%
    \expandafter\date \expandafter{\csname RCS\RCS_keyword\endcsname}%
    }
%%% \endprog


%%% \sect |\RCSID| parses the RCS field and redefines the footline. This
%%% tag is from Nelson's |rcs| style. In his style he defines that the RCS
%%% field is passed as an argument to the macro. To be upward compatible,
%%% we have to support this as well: We look at the next token, if it is
%%% an opening brace we fetch the param and trigger the `real' invocation
%%% by |\rcs_id|. If no brace follows, we call |\rcs_id| immediately.

%%% \beginprog
\def\RCSID{%
    \@ifnextchar\bgroup
	\rcsid_get_field
	\rcs_id
    }
\def\rcsid_get_field#1{\rcs_id #1}
%%% \endprog


%%% \sect As a side effect, we'll produce the normal value tags. When we
%%% define the footline, we must assert that |\RCS_keyword| and
%%% |\RCS_value| are expanded. (They will get new bindings at the next
%%% invocation of |\RCS|.) All other macros must not be expanded, they
%%% shall be used in the context of the footline, not in our definition
%%% context. We utilize |\protect| for that, |\framebox| is the only macro
%%% that must be protected.

%%% \beginprog
\def\rcs_id $#1${%
    \RCS $#1$%
    \begingroup
	\let\protect\noexpand
	\xdef\@oddfoot{%
	    \reset@font
	    \protect\framebox[\textwidth]{\RCS_keyword: \RCS_value}%
	    }%
    \endgroup
    \global\let\@evenfoot\@oddfoot
    }
%%% \endprog


%%% \sect |\RCSdef| just outputs its parameter before it's passed to
%%% |\RCS|. This tag is from Tom Verhoeff's |rcs| style. In fact, we're
%%% not fully compatible: Tom did output `RCS keyword string'. I
%%% substitute this by `RCS field', this is consistent with the rest of
%%% this style's description.

%%% \beginprog
\def\RCSdef $#1${%
    \typeout{RCS field: $#1$}%
    \RCS $#1$%
    }
%%% \endprog




%%% \chap The end.

%%% Well, after all we're finished with this style. We must not forget to
%%% restore the underscore catcode.

%%% \beginprog
\catcode`\_=\CatUsCode

\endinput
%%% \endprog


%%% \sect I would like to thank those who helped me to improve this style.
%%% % In particular, XXX provided XXXsubstantial parts of the code.

%%% \textsc{Marcus Speh} pointed out that Nelson's style option exists.
%%% \textsc{Lee Wittenberg} pointed out that Tom style option exists---Tom
%%% posted the style option on |comp.programming.literate| afterwards. I
%%% borrowed the idea of appending a second colon to the call of
%%% |\rcs_split_field| from Tom's style. \textsc{Harald Fuchs} asked for the
%%% possibility to configure the log list layout.

%%% \textsc{Dave Love} brought the problem with the |-kv| option of |co| to
%%% my attention. It's a pitty that I haven't found a solution yet. (I
%%% mean, one that convinces me.)

%%% \textsc{Christine Detig} told me where my documentation was
%%% uncomprehensible. In particular, you won't have the visual explanation
%%% of |\rcs_split_field| without her. barbara beeton proofread the user
%%% manual. In addition, she pointed out where a non-programmer does not
%%% understand it.

%%% After 7.5 years of quiteness, \textsc{Martin Kneil} emerged and asked
%%% for support of underscores in file names. He got it, and I used the
%%% opportunity to update contact information.



%%% \vskip \PltxPreSectSkip

%%% \rcsLogIntro{Of course, we shouldn't stop without the RCS log of this
%%%     style file.}


%%% \begin{rcslog}
%%% $StyleLog: rcs.doc,v $
%%% \Revision 2.10 2003/02/02 20:42:38 schrod
%%% Support underscores in file names in RCS field values. E.g., in
%%% \rcsField{Id} or \Log{}.

%%% Update my contact information.

%%% \Revision 2.9 1995/08/02 12:09:07 schrod
%%% Transformed this style option into a \LaTeXe{} package.

%%% Updated to \LaTeXe{}.

%%% \Revision 2.8 1993/11/10 12:29:49 schrod
%%% Added acknowledgement of barbara.

%%% \Revision 2.7 1993/11/08 20:15:15 schrod
%%% Added possibility to configure the layout of the log list
%%% (|\RcsLogListStyle|).

%%% Now the log list is on level~0. Thereby item lists or enumerations in
%%% revision entries get first-level labels. This has consequences for
%%% the configuration, this implementation strategy must therefore be
%%% specified in the internal interface description.

%%% \Revision 2.6 1993/11/03 20:04:57 schrod
%%% Cleaned up for distribution: Added email address to each document,
%%% added copyright info to |rcs.doc|, added acknowledgements. Checked my
%%% English and the spacing.

%%% Explained the restriction concerning the |-kv| option of |co|.

%%% Rewrote the explanation to |\rcs_split_field|. Added the visual clues
%%% to explain the (mis?)usage of macro expansion and pattern matching.

%%% \Revision 2.5 1993/11/02 20:02:51 schrod
%%% Implemented |\RCSdef|.

%%% Deleted unnecessary macro that did discard the space of an RCS value.
%%% Moved this in the parameter context of |\rcs_split_field|. Added an
%%% explanation to |\RCS| where it is called.

%%% \Revision 2.4 1993/11/02 18:44:49 schrod
%%% Implemented |\RCSID|.\\
%%% Need |\reset@font| now, for the definition of the footline in |\RCSID|.

%%% Paragraphs in revision logs are not separated by vertical space any
%%% more. They are indented instead. I use the ``classic'' indentation of
%%% one quad.

%%% \Revision 2.3 1993/11/01 19:49:49 schrod
%%% Added subdocument about the internal interface.

%%% Cleaned the cseq names: All public names start with `|RCS|' or
%%% `|rcs|', protected names start with `|Rcs|', both have only letters.
%%% Observer names start with `|RCS|', internal names with `|rcs|', both
%%% feature non-letters in the names.

%%% \Revision 2.2 1993/11/01 19:20:58 schrod
%%% Cleaned up the documentation. The enhancement of the |rcslog|
%%% environment was just appended, now it's integrated in the whole
%%% document. Improved the explanation of the implementation: the valid
%%% input is now explained more clearly.

%%% Use |rcs-doc.sty| for documentation of |rcs| style. It does not only
%%% load style options, but does also define |\RCSStyleRevision| for
%%% access of the style's revision.

%%% Added hint that the user manual is urgently needed.\\
%%%  The load tag is now in the protected interface.\\
%%%  Commented out section with definition of |\reset@font|, that cseq is
%%% not needed in the implementation.\\
%%%  OALD tells me to use ``heading'' instead of ``header.''

%%% \Revision 2.1 1993/10/29 19:07:38 schrod
%%% Reorganized the source structure---the style gets too large to keep
%%% everything in one directory. I use the usual setup for my larger
%%% styles.

%%% Changed the RCS keywords. All keywords in use now start with
%%% `|Style|'. Stuff in the text which looks like RCS fields, but where
%%% the keywords do not start with `|Style|', are examples!

%%% \Revision 1.3 1993/10/29 18:08:19 schrod
%%% Each field value supplied by RCS ends with a space; this space is
%%% discarded.

%%% Supports the typesetting of revision logs. The overall documentation
%%% is not yet fully integrated. Of course, our own revision log is
%%% typeset at the end, too.\\
%%%  But we'll have to use other RCS keywords for this document in the
%%% future. E.g., I can't use original RCS keywords currently in the
%%% examples without much hassle.

%%% An hook for the configuration of this style option is provided.

%%% \Revision 1.2  1993/09/08  15:49:00  schrod
%%% Use canonical names for category codes.

%%% \Revision 1.1  1993/09/03  21:01:29  schrod
%%% Re-implemented |rcs| style option. Made it a documented option.

%%% \end{rcslog}




%%% \end{document}


%%% 
%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Local Variables:
%%% mode: LaTeX
%%% TeX-brace-indent-level: 4
%%% indent-tabs-mode: t
%%% TeX-auto-untabify: nil
%%% TeX-auto-regexp-list: LaTeX-auto-regexp-list
%%% End: