summaryrefslogtreecommitdiff
path: root/Master/texmf-dist/source/latex/dialogl/dialog.dtx
diff options
context:
space:
mode:
Diffstat (limited to 'Master/texmf-dist/source/latex/dialogl/dialog.dtx')
-rw-r--r--Master/texmf-dist/source/latex/dialogl/dialog.dtx1436
1 files changed, 0 insertions, 1436 deletions
diff --git a/Master/texmf-dist/source/latex/dialogl/dialog.dtx b/Master/texmf-dist/source/latex/dialogl/dialog.dtx
deleted file mode 100644
index 3a645e8b5fc..00000000000
--- a/Master/texmf-dist/source/latex/dialogl/dialog.dtx
+++ /dev/null
@@ -1,1436 +0,0 @@
-%%% ====================================================================
-%%% @LaTeX-style-file{
-%%% filename = "dialog.sty",
-%%% version = "0.9y",
-%%% date = "1994/11/08",
-%%% time = "13:31:30 EST",
-%%% checksum = "20801 1436 7278 54732",
-%%% author = "Michael Downes",
-%%% email = "mjd@math.ams.org (Internet)",
-%%% codetable = "ISO/ASCII",
-%%% keywords = "TeX, dialog",
-%%% supported = "yes",
-%%% abstract = "This file provides functions for writing
-%%% messages and menus on screen, and reading user responses. It
-%%% can be used with LaTeX as a documentstyle option, or in
-%%% other forms of TeX by a standard \input statement.",
-%%% docstring = "The checksum field above contains a CRC-16
-%%% checksum as the first value, followed by the equivalent of
-%%% the standard UNIX wc (word count) utility output of lines,
-%%% words, and characters. This is produced by Robert Solovay's
-%%% checksum utility.",
-%%% }
-%%% ====================================================================
-%
-% \iffalse
-%<*driver>
-\input{dia-driv.tex}
-%</driver>
-% \fi
-%
-% \section{History}
-% This file, \fn{dialog.sty}, was born out of a utility called
-% \fn{listout.tex} that I wrote for my personal use. The purpose of
-% \fn{listout.tex} was to facilitate printing out plain text
-% files\Dash electronic mail, program source files in various
-% programming languages, and, foremost, \tex/ macro files and log
-% files. An important part of my \tex/ programming practice is to
-% print out a macro file on paper and read it through, marking
-% corrections along the way, then use the marked copy as a script
-% for editing the file. (For one thing, this allows me to analyze
-% and mark corrections while riding the bus to work, or
-% sitting out in the back yard to supervise the kids.) The output I
-% normally desired was two `pages' per sheet of U.S. letter-size
-% paper printed landscape, in order to conserve paper.
-%
-% Once created, \fn{listout.tex} quickly became my favorite means
-% of printing out plain text files, not to mention an indispensable
-% tool in my debugging toolbox: I turn on \cw{tracingmacros} and
-% \cw{tracingcommands}, then print out the resulting log file so
-% that I can see several hundred lines of the log at once (by
-% spreading out two or three pages on my desk with 100+ lines per
-% page); then I trace through, cross things out, label other
-% things, draw arrows, and so forth.
-%
-% I soon added a filename prompting loop to make it convenient to
-% print multiple files in a single run. In the process of
-% perfecting this simple prompting routine\Dash over two or three
-% years\Dash and adding the ability to optionally specify things like
-% number of columns at run time, eventually I wrote so much
-% dialog-related macro code that it became clear this code should
-% be moved out of \fn{listout.tex} into its own module. The result
-% was \fn{dialog.sty}.
-%
-% Before getting into the macro definitions and technical
-% commentary, here are descriptions from the user's
-% perspective of the functions defined in this file.
-%
-% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \section{Message-sending functions}
-%
-%\begin{usage}
-%\mesj{<text>}
-%\end{usage}
-% Sends the message verbatim: category 12 for all special
-% characters except braces, tab characters, and carriage returns:
-%
-%^^V { } ^^I ^^M
-%
-% Naturally, the catcode changes are effective only if \cw{mesj} is
-% used directly, not inside a macro argument or definition
-% replacement text.
-%
-% Multiple spaces in the argument of \cw{mesj} print as multiple
-% spaces on screen. A tab character produces always eight spaces;
-% `smart' handling of tabs is more complicated than I would care to
-% attempt.
-%
-% Line breaks in the argument of \cw{mesj} will produce line breaks
-% on screen. That is, you don't need to enter a special sequence
-% such as \ctrl{J}\qc{\%} to get line breaks. See the technical
-% commentary for \cw{mesjsetup} for details. Even though curly
-% braces are left with their normal catcodes, they can be printed
-% in a message without any problem, if they occur in balanced
-% pairs. If not, the message should be sent using \cw{xmesj}
-% instead of \cw{mesj}.
-%
-% Because of its careful handling of the message text, \cw{mesj} is
-% extremely easy to use. The only thing you have to worry about is
-% having properly matched braces. Beyond that, you simply type
-% everything exactly as you want it to appear on screen.
-%
-%\begin{usage}
-%\xmesj{<text>}
-%\end{usage}
-% This is like \cw{mesj} but expands embedded control sequences
-% instead of printing them verbatim. All special characters have
-% category 12 except backslash, percent, braces, tab, return, and
-% ampersand:
-%
-%^^V \ % { } ^^I ^^M &
-%
-% The first four have normal \tex/ catcodes to make it possible to
-% use most normal \tex/ commands, and comments, in the message
-% text. \ctrl{I} and \ctrl{M} are catcode 13 and behave as
-% described for \cw{mesj}. The \qc{\&} is a special convenience, an
-% abbreviation for \cw{noexpand}, to use for controlling expansion
-% inside the message text.
-%
-% Doubled backslash \cs{\\} in the argument will produce a single
-% category 12 backslash character\Dash thus, \verb'\\xxx' can be used
-% instead of \cw{string}\cw{xxx} or \cw{noexpand}\cw{xxx} (notice
-% that this works even for outer things like \cw{bye} or
-% \cw{newif}). Similarly \cs{\%}, \cs{\{}, \cs{\}} and \cs{\&}
-% produce the corresponding single characters.
-%
-% Category 12 space means that you cannot write something like
-%
-%^^V \ifvmode h\else v\fi rule
-%
-% in the argument of \cw{xmesj} without getting a space after the
-% \cw{ifvmode}, \cw{else}, and \cw{fi}.%^^A
-%^^A
-%\footnote{Well, actually, you could replace each space by
-%\qc{\%}\meta{newline} to get rid of it. But that makes the message
-%text harder to read for the programmer.}
-%^^A
-% Since occasionally this may be troublesome, \cs{\.} is defined
-% inside the argument of \cw{xmesj} to be a `control word
-% terminator': If the expansion of \cw{foo} is \verb'abc', then
-% \verb'\foo\.xyz' produces \verb'abcxyz' on screen (as opposed to
-% \verb'\foo xyz' which would produce \verb'abc xyz'). Thus the
-% above conditional could be written as
-%
-%^^V \ifvmode\.h\else\.v\fi\.rule
-%
-% Even though the catcode changes done by \cw{xmesj} setup have no
-% effect if \cw{xmesj} is used inside an argument or definition
-% replacement text, I find it convenient occasionally to use
-% \cw{xmesj} in those contexts, in order to get other aspects of the
-% \cw{xmesj} setup. For instance, if you need to embed a message
-% that contains a percent sign inside a definition, you can write
-%
-%^^V \def\foo{...
-%^^V \xmesj{... this is a percent
-%^^V sign: \% (sans backslash) ...}
-%^^V ...}
-%
-% To further support such uses of \cw{xmesj}, the following changes
-% are also done by \cw{xmesj} setup: the backslash-space control
-% symbol {\tt\bslash\char32} is made equivalent to \cw{space};
-% \cs{\^^J} and \cs{\^^M} are defined to produce a \cw{newlinechar};
-% and active tilde \qc{\~} will produce a category-12 tilde.
-%
-% Among other things, this setup makes it easier to obtain
-% newlines and multiple spaces in an embedded message. For example,
-% in the following definition the message will have a line break on
-% screen for each backslash at the end of a line, and the third
-% line will be indented four spaces.
-%
-%^^V \def\bar{...
-%^^V \xmesj{First line\
-%^^V Second line\
-%^^V \ \ \ \ Indented line\
-%^^V Last line}%
-%^^V ...}
-%
-% The alternative of defining a separate message function
-% \cw{barfoo} with \cw{f[x]mesj} and calling \cw{barfoo} inside of
-% \cw{bar} would allow more natural entry of the newlines and the
-% multiple spaces, but would be slightly more expensive in string
-% pool and hash table usage.
-%
-%\begin{usage}
-%\promptmesj{<text>}
-%\promptxmesj{<text>}
-%\end{usage}
-% These are like \cw{mesj}, \cw{xmesj} but use \cw{message} rather
-% than \cw{immediate}\cw{write}\verb'16' internally, thus if the
-% following operation is a \cw{read}, the user will see the cursor
-% on screen at the end of the last line, as may be desired when
-% prompting for a short reply, rather than at the beginning of the
-% next line. The character \qc{\!} is preempted internally for
-% newlinechar, for these two functions only,
-% which means that it cannot be actually printed in
-% the message text. Use of a visible character such as \qc{\!},
-% rather than the normal \cw{newlinechar} \ctrl{J}, is necessary
-% for robustness because of the fact that the \cw{message}
-% primitive was unable to use an `invisible' character (outside the
-% range 32--126) for newlines up until \tex/ version 3.1415, which
-% some users do not yet have (at the time of this writing\Dash July
-% 1994).
-%
-%\begin{usage}
-%\storemesj\foo{<text>}
-%\storexmesj\foo{<text>}
-%\end{usage}
-% These functions are similar to \cw{mesj}, \cw{xmesj} but store the
-% given text in the control sequence \cw{foo} instead of immediately
-% sending the message. Standard \tex/ parameter syntax can be used
-% to make \cw{foo} a function with arguments, e.g. after
-%
-%^^V \storemesj\foo#1{...#1...}
-%
-% then you can later write
-%
-%^^V \message{\foo{\the\hsize}}
-%
-% and get the current value of \cw{hsize} into the middle of the
-% message text. Consequently also in the x-version \cw{storexmesj} a
-% category-12 \qc{\#} character can be obtained with \cs{\#}.
-%
-%\begin{usage}
-%\fmesj\foobar#1#2...{...#1...#2...}
-%\end{usage}
-% Defines \cw{foobar} as a function that will take the given
-% arguments, sow them into the message text \verb"{...}", and send
-% the message. In the message text all special characters are
-% category 12 except for braces, \qc{\#}, tab, and carriage return.
-%
-% If an unmatched brace or a \qc{\#} must be printed in the message
-% text \cw{fxmesj} must be used instead. (\arg{#} could be used to
-% insert a single category-6 \qc{\#} token into the message text, and
-% \tex/ would print it without an error, but both \cw{message} and
-% \cw{write} would print it as two \arg{#} characters, even though
-% it's only a single token internally.)
-%
-%\begin{usage}
-%\fxmesj\foobar#1#2...{...#1...#2...}
-%\end{usage}
-% Combination of \cw{xmesj} and \cw{fmesj}. Defines \cw{foobar} like
-% \cw{fmesj}, but with full expansion of the replacement text and
-% with normal category codes for backslash, percent, braces, and hash
-% \qc{\#}. The control symbols \cs{\\} \cs{\%} \cs{\{} \cs{\}}
-% \cs{\&} and \cs{\.} can be used as in \cw{xmesj}, with also \cs{\#}
-% for printing a \qc{\#} character of category 12.
-%
-% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \section{Reading functions}
-%
-%\begin{usage}
-%\readline{<default>}\answer
-%\end{usage}
-% This reads a line of input from the user into the macro
-% \cw{answer}. (The macro name can be anything chosen by the
-% programmer, not just \cw{answer}.) Before reading, all special
-% characters are deactivated, so that the primitive \cw{read} will
-% not choke if the user happens to enter something like \cw{newif}
-% or {\sc control-l} or \qc{\}}. Depending on the operating
-% system, certain characters\Dash e.g., {\sc control-c}, {\sc
-% control-z}, {\sc control-d}, {\sc control-h}\Dash might have
-% special effects instead of being entered into the replacement
-% text of \cw{answer}, regardless of the catcode changes. To
-% take the most obvious example, under most
-% operating systems, typing {\sc control-h} (the Rubout or
-% Backward-Delete key) will delete the previous character from the
-% user's response, instead of entering an \ascii/ character 8
-% into \cw{answer}.
-%
-% There is one significant exception from the catcode changes that
-% are done for \cw{readline}: spaces and tabs retain their normal
-% catcode of 10, so that multiple spaces in an answer will be
-% reduced to a single space, and macros with normal space-delimited
-% arguments will work when applied to the answer. (I can't think of
-% any likely scenario where category 12 for spaces would be
-% useful.) Also, the catcode of \ctrl{M} is set to 9 (ignore) so
-% that an empty line\Dash meaning that the user just pressed the
-% carriage return/enter key\Dash will result in an empty \cw{answer}.
-% If the answer is empty, the given default string will be
-% substituted. The default string can be empty.
-%
-%\begin{usage}
-%\xreadline{<default>}\answer
-%\end{usage}
-% Like \cw{readline} but the answer is read as executable tokens;
-% the usual catcodes of the \tex/ special characters remain in
-% effect while reading the answer. A few common outer things
-% (\cw{bye}, \cs{\+}, \cw{newif}, \ctrl{L}, among others) are
-% neutralized before the \cw{read} is done, but the user can still
-% cause problems by entering some other outer control sequence or
-% unbalanced braces. I doubt there's any bulletproof solution, if
-% the tokens are to remain executable, short of the usual last
-% resort: reading the answer using \cw{readline}, writing it to a
-% file, then inputting the file.
-%
-%\begin{usage}
-%\readchar{<default>}\answer
-%\end{usage}
-% This is like \cw{readline} but it reduces the answer to its first
-% character. \meta{default} is either a single character or empty.
-%
-%\begin{usage}
-%\readChar{<default>}\answer
-%\end{usage}
-% This is like \cw{readchar} and also uppercases the answer.
-%
-%\begin{usage}
-%\changecase\uppercase\answer
-%\end{usage}
-% The function \cw{changecase} redefines its second argument, which
-% must be a macro, to contain the same text as before, but
-% uppercased or lowercased according to the first argument. Thus
-% \cw{readChar}\verb'{Q}'\cw{answer} is equivalent to
-%
-%^^V \readchar{q}\answer
-%^^V \changecase\uppercase\answer
-%
-% It might sometimes be desirable to force lower case before using
-% a file name given by the user, for example.
-%
-% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \section{Checking functions}
-%
-%\begin{usage}
-%\checkinteger\reply\tempcount
-%\end{usage}
-% To read in and check an answer that is supposed to be an integer,
-% use \cw{readline}{}\cw{reply} and then apply \cw{checkinteger} to
-% the \cw{reply}, supplying a count register, not necessarily named
-% \cw{tempcount}, wherein \cw{checkinteger} will leave the validated
-% integer. If \cw{reply} does not contain a valid integer the
-% returned value will be \verb'-'\cw{maxdimen}.
-%
-% At the present time only decimal digits are handled;
-% some valid \tex/ numbers such as \verb'"AB', \verb'`\@',
-% \cw{number}\cw{prevgraf}, or a count register name, will not be
-% recognized by \cw{checkinteger}. There seems to be no bulletproof
-% way to allow these possibilities.
-%
-% Tests that hide \cw{checkinteger} under the hood, such as a
-% \cw{nonnegativeinteger} test, are not provided because as often as
-% not the number being prompted for will have to be tested to see if
-% it falls inside a more specific range, such as 0\dash 255 for an
-% 8-bit number or 1\dash 31 for a date, and it seems common sense to
-% omit overhead if it would usually be redundant. It's easy enough to
-% define such a test for yourself, if you want one.
-%
-%\begin{usage}
-%\checkdimen\reply\tempdim
-%\end{usage}
-% Analog of \cw{checkinteger} for dimension values. If \cw{reply}
-% does not contain a valid dimension the value returned in
-% \cw{tempdim} will be \verb'-'\cw{maxdimen}.
-%
-% Only explicit dimensions with decimal digits, optional decimal
-% point and more decimal digits, followed by explicit units
-% \verb'pt' \verb'cm' \verb'in' or whatever are checked for; some
-% valid \tex/ dimensions such as \cw{parindent},
-% \verb'.3'\cw{baselineskip}, or \cw{fontdimen}\verb'5'\cw{font}
-% will not be recognized by \cw{checkdimen}.
-%
-% \subsection*{What good is all this?}
-% What good is all this stuff, practically speaking?\Dash you may
-% ask. Well, a typical application might be something like: At the
-% beginning of a document, prompt interactively to find out if the
-% user wants to print on A4 or US letter-size paper, or change the
-% top or left margin. Such a query could be done like this:
-%
-%^^V \promptxmesj{
-%^^V Do you want to print on A4 or US letter paper?
-%^^V Enter u or U for US letter, anything else for A4: }
-%^^V \readChar{A}\reply % default = A4
-%^^V \if U\reply \textheight=11in \textwidth=8.5in
-%^^V \else \textheight=297mm \textwidth=210mm \fi
-%^^V % Subtract space for 1-inch margins
-%^^V \addtolength{\textheight}{-2in}
-%^^V \addtolength{\textwidth}{-2in}
-%^^V
-%^^V \promptxmesj{
-%^^V Left margin setting? [Return = keep current value,
-%^^V \the\oddsidemargin]: }
-%^^V \readline{\the\oddsidemargin}\reply
-%^^V \checkdimen\reply{\dimen0}
-%^^V \ifdim\dimen0>-\maxdimen
-%^^V \setlength\oddsidemargin{\dimen0}%
-%^^V \xmesj{OK, using new left margin of %
-%^^V \the\oddsidemargin.}
-%^^V \else
-%^^V \xmesj{Sorry, I don't understand %
-%^^V that reply: `\reply'.\
-%^^V Using default value: \the\oddsidemargin.}
-%^^V \fi
-%
-% Although \latex/'s \cw{typeout} and \cw{typein} functions can
-% be used for this same task, they are rather more awkward, and
-% checking the margin value for validity would be quite difficult.
-%
-% \StopEventually{}
-%
-% \section{Implementation}
-% Standard package identification:
-% \begin{macrocode}
-%<*2e>
-\NeedsTeXFormat{LaTeX2e}
-\ProvidesPackage{dialog}[1994/11/08 v0.9y]
-%</2e>
-% \end{macrocode}
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \section{Preliminaries}
-%
-% \begin{macrocode}
-%<*2e>
-\RequirePackage{grabhedr}
-%</2e>
-% \end{macrocode}
-%
-% If \fn{grabhedr.sty} is not already loaded, load it now. The
-% \cw{trap.input} function is explained in \fn{grabhedr.doc}.
-% \begin{macrocode}
-%<*209>
-\csname trap.input\endcsname
-\input grabhedr.sty \relax
-\fileversiondate{dialog.sty}{0.9y}{1994/11/08}%
-%</209>
-% \end{macrocode}
-%
-% The functions \cw{localcatcodes} and \cw{restore\-catcodes} are
-% defined in \fn{grabhedr.sty}. We use them to save and restore
-% catcodes of any special characters needed in this file whose
-% current catcodes might not be what we want them to be. Saving and
-% restoring catcode of at-sign \qc{\@} makes this file work equally
-% well as a \latex/ documentstyle option or as a simple input file in
-% other contexts. The double quote character \qc{\"} might be active
-% for German and other languages. Saving and restoring tilde \qc{\~},
-% hash \qc{\#}, caret \qc{\^}, and left quote \qc{\`} catcodes is
-% normally redundant but reduces the number of assumptions we must
-% rely on. (The following catcodes are assumed: \qc{\\} 0, \qc{\{} 1,
-% \qc{\}} 2, \qc{\%} 14, {\tt a}\dash {\tt z} {\tt A}\dash {\tt Z}
-% 11, {\tt 0}\dash {\tt 9} \qc{\.} \qc{\-} 12. Also note that
-% \cw{endlinechar} is assumed to have a non-null value.)
-% \begin{macrocode}
-%% The line break is significant here:
-\localcatcodes{\@{11}\ {10}\
-{5}\~{13}\"{12}\#{6}\^{7}\`{12}}
-% \end{macrocode}
-%
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \section{Definitions}
-%
-% \begin{macro}{\otherchars}
-%
-% For deactivating characters with special catcodes during
-% \cw{readline} we use, instead of \cw{dospecials}, a more
-% bulletproof (albeit slower) combination of \cw{otherchars},
-% \cw{controlchars}, and \cw{highchars} that covers all characters in
-% the range 0\dash 255 except letters and digits. Handling the
-% characters above 127 triples the overhead done for each read
-% operation or message definition but seems mandatory for maximum
-% robustness.^^A
-%^^A%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \footnote{If you are using \fn{dialog.sty} functions on a slow
-% computer, you might want to try setting \cw{highchars} = empty to
-% see if that helps the speed.}
-%^^A%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%
-% \cw{otherchars} includes the thirty-three nonalphanumeric visible
-% characters (counting space as visible). It is intended as an
-% executable list like \cw{dospecials} but, as an exercise in
-% memory conservation, it is constructed without the \cw{do}s. For
-% the usual application of changing catcodes, the list can still be
-% executed nicely as shown below. Also, if we arrange to
-% make sure that each character token gets category 12, it's not
-% necessary to use control symbols such as \cs{\%} in place of
-% those few special characters that would otherwise be difficult to
-% place inside of a definition. This avoids a problem that would
-% otherwise arise if we included \cs{\+} in the list and tried to
-% process the list with a typical definition of \cs{do}: \cs{\+} is
-% `outer' in plain \tex/ and would cause an error message when
-% \cw{do} attempted to read it as an argument. (As a matter of fact
-% the catcode changes below show a different way around that
-% problem, but a list of category-12 character tokens is a fun
-% thing to have around anyway.)
-%
-% \begin{macrocode}
-\begingroup
-% \end{macrocode}
-% First we start a group to localize \cw{catcode} changes. Then we
-% change all relevant catcodes to 12 except for backslash, open
-% brace, and close brace, which can be handled by judicious
-% application of \cw{escapechar}, \cw{string}, \cw{edef}, and
-% \cw{xdef}. By defining \cw{do} in a slightly backward way, so that
-% it doesn't take an argument, we don't need to worry about the
-% presence of \cs{\+} in the list of control symbols. Notice the
-% absence of \cs{\`} from the list of control symbols; it was already
-% catcoded to 12 in the \cw{localcatcodes} declaration at the
-% beginning of this file\Dash otherwise it would be troublesome to
-% make the definition of \cw{do} bulletproof (consider the
-% possibilities that \qc{\`} might have catcode 0, 5, 9, or 14).
-%
-% \begin{macrocode}
-\def\do{12 \catcode`}
-\catcode`\~\do\!\do\@\do\#\do\$\do\^\do\&
-\do\*\do\(\do\)\do\-\do\_\do\=\do\[\do\]
-\do\;\do\:\do\'\do\"\do\<\do\>\do\,\do\.
-\do\/\do\?\do\|12\relax
-% \end{macrocode}
-% To handle backslash and braces, we define \cs{\\},
-% \cs{\{}, and \cs{\}} to produce the corresponding category-12
-% character tokens. Setting \cw{escapechar} to $-1$ means that
-% \cw{string} will omit the leading backslash that it would
-% otherwise produce when applied to a control sequence.
-% \begin{macrocode}
-\escapechar -1
-\edef\\{\string\\}
-\edef\{{\string\{}\edef\}{\string\}}
-% \end{macrocode}
-% Space and percent are done last. Then, with almost all the
-% special characters now category 12, it's rather easy to define
-% \cw{otherchars}.
-% \begin{macrocode}
-\catcode`\ =12\catcode`\%=12
-\xdef\otherchars
-{ !"#$%&'()*+,-./:;<=>?[\\]^_`\{|\}~}
-\endgroup % ^ ^ ^
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\controlchars}
-%
-% \cw{controlchars} is another list for the control characters
-% \ascii/ 0\dash 31 and 127. The construction of this list is similar
-% to the construction of \cw{otherchars}. We need to turn off
-% \cw{endlinechar} because the catcode of \ctrl{M} is going to be
-% changed. The \ctrl{L} inside the \cw{gdef} is not a problem (as it
-% might have been, due to the usual outerness of \ctrl{L}) because
-% the catcode is changed from 13 to 12 before that point.
-%
-% \begin{macrocode}
-\begingroup
-\endlinechar = -1
-\def\do{12 \catcode`}
-\catcode`\^^@\do\^^A\do\^^B\do\^^C
-\do\^^D\do\^^E\do\^^F\do\^^G\do\^^H\do\^^I
-\do\^^J\do\^^K\do\^^L\do\^^M\do\^^N\do\^^O
-\do\^^P\do\^^Q\do\^^R\do\^^S\do\^^T\do\^^U
-\do\^^V\do\^^W\do\^^X\do\^^Y\do\^^Z\do\^^[
-\do\^^\\do\^^]\do\^^^\do\^^_\do\^^? 12\relax
-%
-\gdef\controlchars{^^@^^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^^[^^\^^]^^^^^_^^?}
-\endgroup
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\highchars}
-% And finally, the list \cw{highchars} contains characters
-% 128\dash 255, the ones that have the eighth bit set.
-%
-% \begin{macrocode}
-\begingroup
-\def\do{12 \catcode`}
-\catcode`\^^80\do\^^81\do\^^82\do\^^83\do\^^84
-\do\^^85\do\^^86\do\^^87\do\^^88\do\^^89\do\^^8a
-\do\^^8b\do\^^8c\do\^^8d\do\^^8e\do\^^8f
-\do\^^90\do\^^91\do\^^92\do\^^93\do\^^94\do\^^95
-% \end{macrocode}
-%\verbdots\iffalse
-\do\^^96\do\^^97\do\^^98\do\^^99\do\^^9a\do\^^9b
-\do\^^9c\do\^^9d\do\^^9e\do\^^9f
-\do\^^a0\do\^^a1\do\^^a2\do\^^a3\do\^^a4\do\^^a5
-\do\^^a6\do\^^a7\do\^^a8\do\^^a9\do\^^aa\do\^^ab
-\do\^^ac\do\^^ad\do\^^ae\do\^^af
-\do\^^b0\do\^^b1\do\^^b2\do\^^b3\do\^^b4\do\^^b5
-\do\^^b6\do\^^b7\do\^^b8\do\^^b9\do\^^ba\do\^^bb
-\do\^^bc\do\^^bd\do\^^be\do\^^bf
-\do\^^c0\do\^^c1\do\^^c2\do\^^c3\do\^^c4\do\^^c5
-\do\^^c6\do\^^c7\do\^^c8\do\^^c9\do\^^ca\do\^^cb
-\do\^^cc\do\^^cd\do\^^ce\do\^^cf
-\do\^^d0\do\^^d1\do\^^d2\do\^^d3\do\^^d4\do\^^d5
-\do\^^d6\do\^^d7\do\^^d8\do\^^d9\do\^^da\do\^^db
-\do\^^dc\do\^^dd\do\^^de\do\^^df
-\do\^^e0\do\^^e1\do\^^e2\do\^^e3\do\^^e4\do\^^e5
-\do\^^e6\do\^^e7\do\^^e8\do\^^e9\do\^^ea\do\^^eb
-\do\^^ec\do\^^ed\do\^^ee\do\^^ef
-\do\^^f0\do\^^f1\do\^^f2\do\^^f3\do\^^f4\do\^^f5
-\do\^^f6\do\^^f7\do\^^f8\do\^^f9\do\^^fa\do\^^fb
-%\fi
-% \begin{macrocode}
-\do\^^fc\do\^^fd\do\^^fe\do\^^ff 12\relax
-%
-\gdef\highchars{%
-^^80^^81^^82^^83^^84^^85^^86^^87^^88%
-^^89^^8a^^8b^^8c^^8d^^8e^^8f%
-^^90^^91^^92^^93^^94^^95^^96^^97^^98%
-% \end{macrocode}
-%\verbdots\iffalse
-^^99^^9a^^9b^^9c^^9d^^9e^^9f%
-^^a0^^a1^^a2^^a3^^a4^^a5^^a6^^a7^^a8%
-^^a9^^aa^^ab^^ac^^ad^^ae^^af%
-^^b0^^b1^^b2^^b3^^b4^^b5^^b6^^b7^^b8%
-^^b9^^ba^^bb^^bc^^bd^^be^^bf%
-^^c0^^c1^^c2^^c3^^c4^^c5^^c6^^c7^^c8%
-^^c9^^ca^^cb^^cc^^cd^^ce^^cf%
-^^d0^^d1^^d2^^d3^^d4^^d5^^d6^^d7^^d8%
-^^d9^^da^^db^^dc^^dd^^de^^df%
-^^e0^^e1^^e2^^e3^^e4^^e5^^e6^^e7^^e8%
-^^e9^^ea^^eb^^ec^^ed^^ee^^ef%
-^^f0^^f1^^f2^^f3^^f4^^f5^^f6^^f7^^f8%
-%\fi
-% \begin{macrocode}
-^^f9^^fa^^fb^^fc^^fd^^fe^^ff}
-\endgroup
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\actively}
-%
-% The function \cw{actively} makes a given character active and
-% carries out the assignment given as the first argument. The
-% assignment can be embedded in the replacement text of a macro
-% definition without requiring any special setup to produce an active
-% character in the replacement text. The argument should be a control
-% symbol, e.g. \cw{@} or \cs{\#} or \cs{\^^M}, rather than a single
-% character. (Except that \qc{\+} is safer than \cs{\+} in
-% \plaintex/.) If the assignment is a definition (\cw{def},
-% \cw{edef}, \cw{gdef}, \cw{xdef}) it is allowed to take arguments in
-% the normal \tex/ way. Prefixes such as \cw{global}, \cw{long}, or
-% \cw{outer} must go inside the first argument rather than before
-% \cw{actively}.
-%
-% Usage:
-%\begin{usage}
-%\actively\def\?{<replacement text>}
-%\actively\def\%#1#2{<replacement text>}
-%\actively{\global\let}\^^@=\space
-%\end{usage}
-% One place where this function can be put to good use is in
-% making \ctrl{M} active in order to get special action at the end
-% of each line of input. The usual way of going about this would be
-% to write
-%
-%^^V \def\par{something}\obeylines
-%
-% \noindent which is a puzzling construction to the \tex/ novice
-% who doesn't know what \cw{obeylines} does with \cw{par}. The same
-% effect could be gotten a little more transparently with
-%
-%^^V \actively\def\^^M{something}
-%
-% In the definition of \cw{actively} we use the unique properties
-% of \cw{lowercase} to create an active character with the right
-% character code, overlapping with a \cw{begingroup} \cw{endgroup}
-% structure that localizes the necessary lc-code change.
-% \begin{macrocode}
-\def\actively#1#2{\catcode`#2\active
- \begingroup \lccode`\~=`#2\relax
- \lowercase{\endgroup#1~}}
-% \end{macrocode}
-% \end{macro}
-%
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-%
-% \begin{macro}{\mesjsetup}
-%
-% The \cw{mesjsetup} function starts a group to localize catcode
-% changes. The group will be closed eventually by a separate
-% function that does the actual sending or stores the message text
-% for later retrieval.
-%
-% We want to change the catcode of each character in the three
-% lists \cw{otherchars}, \cw{controlchars}, and \cw{highchars} to
-% 12. After giving \cw{do} a recursive definition, we apply it to
-% each of the three lists, adding a suitable element at the end of
-% the list to make the recursion stop there. This allows
-% leaving out the \cw{do} tokens from the character lists, without
-% incurring the cost of an if test at each recursion step.
-% \begin{macrocode}
-\def\mesjsetup{\begingroup \count@=12
- \def\do##1{\catcode`##1\count@ \do}%
-% \end{macrocode}
-% The abbreviation \cw{xp@} = \cw{expandafter} is from
-% \fn{grabhedr.sty}.
-% \begin{macrocode}
- \xp@\do\otherchars{a11 \@gobbletwo}%
- \xp@\do\controlchars{a11 \@gobbletwo}%
- \xp@\do\highchars{a11 \@gobbletwo}%
-% \end{macrocode}
-% Make the tab character produce eight spaces:
-% \begin{macrocode}
- \actively\edef\^^I{ \space\space\space
- \space\space\space\space}%
-% \end{macrocode}
-% The convenient treatment of newlines in the argument of \cw{mesj}
-% (every line break produces a line break on screen) is achieved by
-% making the \ctrl{M} character active and defining it to produce a
-% category-12 \ctrl{J} character. Although for \cw{mesj} it would
-% have sufficed to make \ctrl{M} category 12 and locally set
-% \cw{newlinechar} = \ctrl{M} while sending the message, it turns out
-% to be useful for other functions to have the \ctrl{M} character
-% active, so that it can be remapped to an arbitrary function for
-% handling new lines (e.g., perhaps adding extra spaces at the
-% beginning of each line). And if \cw{mesj} treats \ctrl{M} the same,
-% we can arrange for it to share the setup routines needed for the
-% other functions.
-% \begin{macrocode}
- \endlinechar=`\^^M\actively\let\^^M=\relax
- \catcode`\{=1 \catcode`\}=2 }
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\sendmesj}
-%
-% In \cw{sendmesj} we go to a little extra trouble to make sure
-% \ctrl{M} produces a newline character, no matter what the value of
-% \cw{newlinechar} might be in the surrounding environment. The
-% impending \cw{endgroup} will restore \cw{newlinechar} to its
-% previous value. One reason for using \ctrl{J} (instead of, say,
-% \ctrl{M} directly) is to allow e.g. \cw{mesj}\verb'{xxx^^Jxxxx}' to
-% be written inside a definition, as is sometimes convenient. This
-% would be difficult with \ctrl{M} instead of \ctrl{J} because of
-% catcodes.
-%
-% \begin{macrocode}
-\def\sendmesj{\newlinechar`\^^J%
- \actively\def\^^M{^^J}%
- \immediate\write\sixt@@n{\mesjtext}\endgroup}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\mesj}
-%
-% Given the support functions defined above, the definition of
-% \cw{mesj} is easy: Use \cw{mesjsetup} to clear all special
-% catcodes, then set up \cw{sendmesj} to be triggered by the next
-% assignment, then read the following balanced-braces group into
-% \cw{mesjtext}. As soon as the definition is completed, \tex/ will
-% execute \cw{sendmesj}, which will send the text and close the
-% group that was started in \cw{mesjsetup} to localize the catcode
-% changes.
-%
-% \begin{macrocode}
-\def\mesj{\mesjsetup \afterassignment\sendmesj
- \def\mesjtext}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\sendprompt}
-%
-% The \cw{sendprompt} function is just like \cw{sendmesj} except
-% that it uses \cw{message} instead of \cw{write}, as might be
-% desired when prompting for user input, so that the on-screen
-% cursor stays on the same line as the prompt instead of hopping
-% down to the beginning of the next line. In order for newlines to
-% work with \cw{message} we must use a visible character instead of
-% \ctrl{J}. When everyone has \tex/ version 3.1415 or later this
-% will no longer be true. The choice of \qc{\!} might be construed
-% (if you wish) as editorial comment that \qc{\!} should not be
-% shouted at the user in a prompt.
-%
-% \begin{macrocode}
-\def\sendprompt{%
- \newlinechar`\!\relax \actively\def\^^M{!}%
- \message{\mesjtext}\endgroup}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\promptmesj}
-%
-% This function is like \cw{mesj} but uses \cw{sendprompt} instead of
-% \cw{sendmesj}.
-%
-% \begin{macrocode}
-\def\promptmesj{\mesjsetup
- \afterassignment\sendprompt \def\mesjtext}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\storemesj}
-%
-% Arg \arg{1} of \cw{storemesj} is the control sequence under which
-% the message text is to be stored.
-% \begin{macrocode}
-\def\storemesj#1{\mesjsetup
- \catcode`\#=6 % to allow arguments if needed
- \afterassignment\endgroup
- \long\gdef#1}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\fmesj}
-%
-% While \cw{storemesj}\cw{foo}\verb"{...}" is more or less the
-% same as \cw{def}\cw{foo}\verb"{...}" with special catcode
-% changes, \cw{fmesj}\cw{foo}\verb"{...}" corresponds to
-% \cw{def}\cw{foo}\qc{\{}\cw{mesj}\verb"{...}}", that is, after
-% \cw{fmesj}\cw{foo} the function \cw{foo} can be executed directly
-% to send the message. Thus \cw{storemesj} is typically used for
-% storing pieces of messages, while \cw{fmesj} is used for storing
-% entire messages.
-%
-% To read the parameter text \arg{2}, we use the peculiar
-% \verb'#{' feature of \tex/ to read everything up to the opening
-% brace.
-% \begin{macrocode}
-\def\fmesj#1#2#{\mesjsetup
- \catcode`\#=6 % restore to normal
-% \end{macrocode}
-% The parameter text \arg{2} must be stored in a token register
-% rather than a macro to avoid problems with \qc{\#} characters.
-% The \cw{long} prefix is just to admit the (unlikely) possibility
-% of using \cw{fmesj} to define something such as an error message
-% saying `You can't use \arg{1} here' where one of the
-% possibilities for \arg{1} is \cw{string}\cw{par}.
-% \begin{macrocode}
- \toks@{\long\gdef#1#2}%
-% \end{macrocode}
-% Define \cw{@tempa} to put together the first two arguments and
-% [pseudo]argument \arg{3} and make the definition of \arg{1}.
-% \begin{macrocode}
- \def\@tempa{%
- \edef\@tempa{%
- \the\toks@{%
-% \end{macrocode}
-% The abbreviation \cw{nx@} = \cw{noexpand} is from
-% \fn{grabhedr.sty}.
-% \begin{macrocode}
- \begingroup\def\nx@\mesjtext{\the\toks2 }%
- \nx@\sendmesj}%
- }%
- \@tempa
- \endgroup % Turn off the \mesjsetup catcodes
- }%
- \afterassignment\@tempa
- \toks2=}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\xmesjsetup}
-%
-% \cw{xmesjsetup} is like \cw{mesjsetup} except it prepares to allow
-% control sequence tokens and normal comments in the message text.
-% For \tex/nicians' convenience certain other features are
-% thrown in.
-%
-% Here, unlike the setup for \cw{xreadline}, I don't bother to
-% remove the outerness of \cw{bye}, \cw{newif}, etc., because I
-% presume the arguments of \cw{xmesj}, \cw{fxmesj},
-% \cw{storexmesj}, \cw{fxmenu}, etc.\ are more likely to be written
-% by a \tex/nician than by an average end user, whereas
-% \cw{xreadline} is designed to handle arbitrary input from
-% arbitrary users.
-%
-% \begin{macrocode}
-\def\xmesjsetup{\mesjsetup
-% \end{macrocode}
-% Throw in pseudo braces just in case we are inside an \cw{halign}
-% with \cs{\\} let equal to \cw{cr} at the time when \cw{xmesjsetup}
-% is called. (As might happen in \amstex/.)
-% \begin{macrocode}
- \iffalse{\fi
- \catcode`\\=0 \catcode`\%=14
-% \end{macrocode}
-% Define {\def~{\spacefactor1500 \space}%
-% \cs{\%}~\cs{\\}~\cs{\{}~\cs{\}}~\cs{\&}} to produce the
-% corresponding single characters, category 12. The \cw{lowercase}
-% trick here allows these definitions to be nonglobal.
-% \begin{macrocode}
- \begingroup \lccode`\0=`\\\lccode`\1=`\{%
- \lccode`\2=`\}\lccode`\3=`\%%
- \lowercase{\endgroup \def\\{0}\def\{{1}%
- \def\}{2}\def\%{3}}%
- \iffalse}\fi
- \edef\&{\string &}%
-% \end{macrocode}
-% Let \qc{\&} \qc{\=} \cw{noexpand} for expansion control inside the
-% argument text; let active \ctrl{M} \qc{\=} \cw{relax} so that
-% newlines will remain inert during the expansion.
-% \begin{macrocode}
- \actively\let\&=\noexpand
- \actively\let\^^M=\relax
-% \end{macrocode}
-% Define \cs{\.} to be a noop, for terminating a control word when
-% it is followed by letters and no space is wanted.
-% \begin{macrocode}
- \def\.{}%
-% \end{macrocode}
-% Support for use of \cw{xmesj} inside a definition replacement
-% text or macro argument: control-space
-% \verb*"\ " = \cw{space}, tilde \qc{\~} prints as itself,
-% \cs{\^^M} (i.e., a lone backslash at the end of a line) will
-% produce a newline, also \cs{\^^J}, while finally \cw{par} \qc{\=}
-% blank line translates to two newlines.
-% \begin{macrocode}
- \def\ { }\edef~{\string ~}%
-% \end{macrocode}
-% Define \cs{\^^M} to produce an active \ctrl{M} character, which (we
-% hope) will be suitably defined to produce a newline or whatever.
-% \begin{macrocode}
- \begingroup \lccode`\~=`\^^M%
- \lowercase{\endgroup \def\^^M{~}}%
- \let\^^J\^^M \def\par{\^^M\^^M}%
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\xmesj}
-%
-% \cw{xmesj} uses \cw{xmesjsetup} and \cw{edef}.
-% \begin{macrocode}
-\def\xmesj{\xmesjsetup \afterassignment\sendmesj
- \edef\mesjtext}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\promptxmesj}
-%
-% \cw{promptxmesj} is analogous to \cw{promptmesj}, but with
-% expansion.
-% \begin{macrocode}
-\def\promptxmesj{\xmesjsetup
- \afterassignment\sendprompt \edef\mesjtext}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\storexmesj}
-%
-% And \cw{storexmesj} is like \cw{storemesj}, with expansion. Since
-% we allow arguments for the function being defined, we also must
-% define \cs{\#} to produce a single category-12 \qc{\#} character so
-% that there will be a way to print \qc{\#} in the message text.
-% \begin{macrocode}
-\def\storexmesj#1#2#{\xmesjsetup
- \catcode`\#=6 % to allow arguments if needed
- \edef\#{\string##}%
- \afterassignment\endgroup
- \long\xdef#1#2}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\fxmesj}
-%
-% And \cw{fxmesj} is the expansive analog of \cw{fmesj}.
-% \begin{macrocode}
-\def\fxmesj#1#2#{\xmesjsetup
- \catcode`\#=6 % restore to normal
- \edef\#{\string##}%
- \toks@{\long\xdef#1#2}%
- \def\@tempa{%
- \edef\@tempa{%
- \the\toks@{\begingroup
- \def\nx@\nx@\nx@\mesjtext{\the\toks\tw@}%
- \nx@\nx@\nx@\sendmesj}}%
- \@tempa % execute the constructed xdef
- \endgroup % restore normal catcodes
- }%
- \afterassignment\@tempa
- \toks\tw@=}
-% \end{macrocode}
-% \end{macro}
-%
-%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% \section{Reading functions}
-%
-% \begin{macro}{\readline}
-% The \cw{readline} function gets one line of input from the user.
-% Arguments are: \arg{1} default to be used if the user response is
-% empty (i.e., if the user just pressed the return/enter key),
-% \arg{2} macro to receive the input.
-%
-% \begin{macrocode}
-\def\readline#1#2{%
- \begingroup \count@ 12 %
- \def\do##1{\catcode`##1\count@ \do}%
- \xp@\do\otherchars{a11 \@gobbletwo}%
- \xp@\do\controlchars{a11 \@gobbletwo}%
- \xp@\do\highchars{a11 \@gobbletwo}%
-% \end{macrocode}
-% Make spaces and tabs normal instead of category 12.
-% \begin{macrocode}
- \catcode`\ =10 \catcode`\^^I=10 %
- \catcode`\^^M=9 % ignore
-% \end{macrocode}
-% Reset end-of-line char to normal, just in case.
-% \begin{macrocode}
- \endlinechar`\^^M
-% \end{macrocode}
-% We go to a little trouble to avoid \cw{gdef}-ing \arg{2}, in order
-% to prevent save stack buildup if the user of \cw{readline} carries
-% on unaware doing local redefinitions of \arg{2} after the initial
-% read.
-% \begin{macrocode}
- \read\m@ne to#2%
- \edef#2{\def\nx@#2{#2}}%
- \xp@\endgroup #2%
- \ifx\@empty#2\def#2{#1}\fi
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\xreadline}
-%
-% \cw{xreadline} is like \cw{readline} except that it leaves almost
-% all catcodes unchanged so that the return value is executable
-% tokens instead of strictly character tokens of category 11 or 12.
-% \begin{macrocode}
-\def\xreadline#1#2{%
- \begingroup
-% \end{macrocode}
-% Render some outer control sequences innocuous.
-% \begin{macrocode}
- \xp@\let\csname bye\endcsname\relax
- \xp@\let\csname newif\endcsname\relax
- \xp@\let\csname newcount\endcsname\relax
- \xp@\let\csname newdimen\endcsname\relax
- \xp@\let\csname newskip\endcsname\relax
- \xp@\let\csname newmuskip\endcsname\relax
- \xp@\let\csname newtoks\endcsname\relax
- \xp@\let\csname newbox\endcsname\relax
- \xp@\let\csname newinsert\endcsname\relax
- \xp@\let\csname +\endcsname\relax
- \actively\let\^^L\relax
- \catcode`\^^M=9 % ignore
- \endlinechar`\^^M% reset to normal
- \read\m@ne to#2%
- \toks@\xp@{#2}%
- \edef\@tempa{\def\nx@#2{\the\toks@}}%
- \xp@\endgroup \@tempa
- \ifx\@empty#2\def#2{#1}\fi
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\readchar}
-%
-% \cw{readchar} reduces the user response to a single character.
-% \begin{macrocode}
-\def\readchar#1#2{%
- \readline{#1}#2%
-% \end{macrocode}
-% If the user's response and the default response are both empty,
-% we need something after \arg{1} to keep \cw{@car} from running
-% away, so we add an empty pair of braces.
-% \begin{macrocode}
- \edef#2{\xp@\@car#2#1{}\@nil}%
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\readChar}
-%
-% \cw{readChar} reduces the user response to a single uppercase
-% character. (This is useful to simplify testing the response
-% later with \cw{if}.)
-%
-% \begin{macrocode}
-\def\readChar#1#2{%
- \readline{#1}#2%
- \changecase\uppercase#2%
-% \end{macrocode}
-% Reduce \arg{2} to its first character, or the first character of
-% \arg{1}, if \arg{2} is empty. The extra braces \verb'{}' are to
-% prevent a runaway argument error from \cw{@car} if \arg{2} and
-% \arg{1} are both empty.
-% \begin{macrocode}
- \edef#2{\xp@\@car #2#1{}\@nil}%
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\changecase}
-% The function \cw{changecase} uppercases or lowercases the
-% replacement text of its second argument, which
-% must be a macro. The first argument should be \cw{uppercase} or
-% \cw{lowercase}.
-% \begin{macrocode}
-\def\changecase#1#2{\@casetoks\xp@{#2}%
- \edef#2{#1{\def\nx@#2{\the\@casetoks}}}#2}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\@casetoks}
-% We allocate a token register just for the use of \cw{changecase}
-% because it might be used at a low level internally where we don't
-% want to interfere with other uses of the scratch token registers
-% 0\dash 9.
-% \begin{macrocode}
-\newtoks\@casetoks
-% \end{macrocode}
-% \end{macro}
-%
-% A common task in reading user input is to verify, when an answer
-% of a certain kind was requested, that the response has indeed the
-% desired form\Dash for example, if a nonnegative integer is required
-% for subsequent processing, it behooves us to verify that we have
-% a nonnegative integer in hand before doing anything that might
-% lead to inconvenient error messages. However, it's not easy to
-% decide how best to handle such verification. One possibility
-% might be to have a function
-%
-%^^V \readnonnegativeinteger\foo
-%
-% to do all the work of going out and fetching a number from the
-% user and leaving it in the macro \cw{foo}. Another possibility
-% would be to read the response using \cw{readline} and then apply
-% a separate function that can be used in combination with \cw{if},
-% for example
-%
-%^^V \readline{}\reply
-%^^V \if\validnumber\reply ... \else ... \fi
-%
-% For maximum flexibility, a slightly lower-level approach is
-% chosen here. The target syntax is
-%
-%^^V \readline{}\reply
-%^^V \checkinteger\reply\tempcount
-%
-% where \cw{tempcount} will be set to \verb'-'\cw{maxdimen} if
-% \cw{reply} does {\em not\/} contain a valid integer. (Negative
-% integers are allowed, as long as they are greater than
-% \verb'-'\cw{maxdimen}.) Then the function that calls
-% \cw{checkinteger} is free to make additional checks on the range
-% of the reply and give error messages tailored to the
-% circumstances. And the handling of an empty \cw{reply} can be
-% arbitrarily customized, something that would tend to be
-% inconvenient for the first method mentioned.
-%
-% The first and second approaches can be built on top of the third
-% if desired, e.g.\ (for the second approach)
-%
-%^^V \def\validnumber#1{TT\fi
-%^^V \checkinteger#1\tempcount%
-%^^V \ifnum\tempcount>-\maxdimen }
-%
-% The curious \verb'TT\fi'\verbdots\cw{ifnum} construction is from
-% \TeXhax{} 1989, no.~20 and no.~38 (a suggestion of D. E. Knuth in
-% reply to a query by S. von Bechtolsheim).
-%
-% \begin{macro}{\checkinteger}
-%
-% The arguments of \cw{checkinteger}'s are: \arg{2}, a count register
-% to hold the result; \arg{1}, a macro holding zero or more arbitrary
-% characters of category 11 or 12.
-%
-% \begin{macrocode}
-\def\checkinteger#1#2{\let\scansign@\@empty
- \def\scanresult@{#2}%
- \xp@\scanint#1x\endscan}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\scanint}
-%
-% To validate a number, the function \cw{scanint} must first scan
-% away leading \qc{\+} or \qc{\-} signs (keeping track in
-% \cw{scansign@}), then look at the first token after that: if it's
-% a digit, fine, scan that digit and any succeeding digits into the
-% given count register (\cw{scanresult@}), ending with \cw{endscan}
-% to get rid of any following garbage tokens that might just
-% possibly show up.
-%
-% Typical usage of \cw{scanint} includes initializing \cw{scansign@}
-% to empty, as in the definition of \cw{checkinteger}.
-%\begin{usage}
-%\let\scansign@\@empty
-%\def\scanresult@{\tempcount}%
-%\xp@\scanint\reply x\endscan
-%\end{usage}
-% Assumption: \cw{reply} is either empty or contains only category
-% 11 or 12 characters (which it will if you used \cw{readline}!). If
-% a separate check is done earlier to trap the case where \cw{reply}
-% is empty\Dash for example, by using a nonempty default for
-% \cw{readline}\Dash then the \verb'x' before \cw{endscan} is
-% superfluous.
-%
-% Arg \arg{1} = next character from the string being tested.
-% The test whether \arg{1} is a decimal digit is similar in spirit
-% to the test \verb'\if!#1!' to see if an argument is empty
-% (\TB, Appendix~D, p.~376).
-% \begin{macrocode}
-\def\scanint#1{%
- \ifodd 0#11 %
-% \end{macrocode}
-% Is \arg{1} a decimal digit? If so read all digits into
-% \cw{scanresult@} with the sign prefix.
-% \begin{macrocode}
- \def\@tempa{\afterassignment\endscan
- \scanresult@=\scansign@#1}%
- \else
- \if -#1\relax
-% \end{macrocode}
-% Here we flipflop the sign; watch closely.
-% \begin{macrocode}
- \edef\scansign@{%
- \ifx\@empty\scansign@ -\fi}%
- \def\@tempa{\scanint}%
- \else
-% \end{macrocode}
-% A plus sign can just be ignored.
-% \begin{macrocode}
- \if +#1\relax
- \def\@tempa{\scanint}%
- \else % not a valid number
- \def\@tempa{%
- \scanresult@=-\maxdimen\endscan}%
- \fi\fi\fi
- \@tempa
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\endscan}
-% The \cw{endscan} function just gobbles any remaining garbage. It
-% uses its own name as the argument delimiter.
-% \begin{macrocode}
-\def\endscan#1\endscan{}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\dimenfirstpart}
-% \begin{macro}{\dimentoks}
-% \cw{dimenfirstpart}, a count register, receives the digits, if
-% any, preceding the decimal point. \cw{dimentoks}, a token
-% register, receives any digits after the decimal point.
-% \begin{macrocode}
-\newcount\dimenfirstpart
-\newtoks\dimentoks
-% \end{macrocode}
-% \end{macro}
-% \end{macro}
-%
-% \begin{macro}{\scandimen}
-%
-% \cw{scandimen} is similar to \cw{scanint} but has to call some
-% auxiliary functions to scan the various subcomponents of a
-% dimension (leading digits, decimal point, fractional part, and
-% units, with optional {\tt true}, in addition to the sign). The
-% minimum requirements of \tex/'s syntax for dimensions are a digit
-% or decimal point \qc{\+} the units; all the other components are
-% optional (\TB, Exercise~10.3, p.~58).
-%
-% When scanning for the digits of a fractional part, we can't throw
-% away leading zeros; therefore we don't read the fractional part
-% into a count register as we did for the digits before the decimal
-% point; instead we read the digits one by one and store them in
-% \cw{dimentoks}.
-%
-% The function that calls \cw{scandimen} should initialize
-% \cw{scansign@} to \cw{@empty}, \cw{dimenfirstpart} to \cw{z@},
-% \cw{dimentoks} to empty \verb'{}', and \cw{dimentrue@} to
-% \cw{@empty}.
-%
-% Test values: {\tt 0pt}, {\tt 1.1in}, {\tt -2cm}, {\tt .3mm}, {\tt
-% 0.4dd}, {\tt 5.cc}, {\tt.10000000009pc}, \cw{hsize}, {\tt em}.
-%
-% \begin{macrocode}
-\def\scandimen#1{%
- \ifodd 0#11
- \def\@tempa{\def\@tempa{\scandimenb}%
- \afterassignment\@tempa
- \dimenfirstpart#1}%
- \else
-% \end{macrocode}
-%
-% The following test resolves to true if \arg{1} is either a period
-% or a comma (both recognized by \tex/ as decimal point
-% characters).
-%
-% \begin{macrocode}
- \if \if,#1.\else#1\fi.%
- \def\@tempa{\scandimenc}%
- \else
- \if -#1% then flipflop the sign
- \edef\scansign@{%
- \ifx\@empty\scansign@ -\fi}%
- \def\@tempa{\scandimen}%
- \else
- \if +#1% then ignore it
- \def\@tempa{\scandimen}%
- \else % not a valid dimen
- \def\@tempa{%
- \scanresult@=-\maxdimen\endscan}%
- \fi\fi\fi\fi
- \@tempa
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\scandimenb}
-%
-% Scan for an optional decimal point.
-%
-% \begin{macrocode}
-\def\scandimenb#1{%
- \if \if,#1.\else#1\fi.%
- \def\@tempa{\scandimenc}%
- \else
-% \end{macrocode}
-% If the decimal point is absent, we need to put back \arg{2} and
-% rescan it to see if it is the first letter of the units.
-% \begin{macrocode}
- \def\@tempa{\scanunitsa#1}%
- \fi
- \@tempa
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\scandimenc}
-%
-% Scan for the fractional part: digits after the decimal point.
-%
-% \begin{macrocode}
-\def\scandimenc#1{%
-% \end{macrocode}
-% If \arg{1} is a digit, add it to \cw{dimentoks}.
-% \begin{macrocode}
- \ifodd 0#11 \dimentoks\xp@{%
- \the\dimentoks#1}%
- \def\@tempa{\scandimenc}%
- \else
-% \end{macrocode}
-% Otherwise rescan \arg{1}, presumably the first letter of the
-% units.
-% \begin{macrocode}
- \def\@tempa{\scanunitsa#1}%
- \fi
- \@tempa
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\scanunitsa}
-%
-% \begin{macrocode}
-\def\scanunitsa#1\endscan{%
-% \end{macrocode}
-% Check for \verb'true' qualifier.
-% \begin{macrocode}
- \def\@tempa##1true##2##3\@tempa{##2}%
-% \end{macrocode}
-%
-% The peculiar nature of \cw{lowercase} is evident here as we are
-% able to apply it to only the test part of the conditional without
-% running into brace-matching problems. (Compare the braces in this
-% example to something like \cw{message}\qc{\{}\cw{iffalse}
-% \verb"A}"\cw{else} \verb"B}"\cw{fi}.)
-%
-% \begin{macrocode}
- \lowercase{%
- \xp@\ifx\xp@\end
- \@tempa#1true\end\@tempa
- }%
-% \end{macrocode}
-% No \verb'true' was found:
-% \begin{macrocode}
- \let\dimentrue@\@empty
- \def\@tempa{\scanunitsb#1\endscan}%
- \else
- \def\dimentrue@{true}%
- \def\@tempa##1true##2\@tempa{%
- \def\@tempa{##1}%
- \ifx\@tempa\@empty
- \def\@tempa{\scanunitsb##2\endscan}%
- \else
- \def\@tempa{\scanunitsb xx\endscan}%
- \fi}%
- \@tempa#1\@tempa
- \fi
- \@tempa
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\scanunitsb}
-%
-% Scan for the name of the units and complete the assignment of the
-% scanned value to \cw{scanresult@}. Notice that, because of the way
-% \cw{scanunitsb} picks up \arg{1} and \arg{2} as macro arguments,
-% \verb"p t" is allowed as a variation of \verb"pt". Eliminating
-% this permissiveness doesn't seem worth the speed penalty
-% that would be incurred in \cw{scanunitsb}.
-%
-% The method for detecting a valid units string is to define the
-% scratch function \cw{@tempa} to apply \tex/'s parameter-matching
-% abilities to a special string that will yield a boolean value
-% of true if and only if the given string is a valid \tex/ unit.
-% \begin{macrocode}
-\def\scanunitsb#1#2{%
- \def\@tempa##1#1#2##2##3\@nil{##2}%
- \def\@tempb##1{T\@tempa
- pcTptTcmTccTemTexTinTmmTddTspT##1F\@nil}%
-% \end{macrocode}
-% Force lowercase just in case the units were entered with
-% uppercase letters (accepted by \tex/, so we had better accept
-% uppercase also).
-% \begin{macrocode}
- \lowercase{%
- \if\@tempb{#1#2}%
- }%
- \scanresult@=\scansign@
- \number\dimenfirstpart.\the\dimentoks
- \dimentrue@#1#2\relax
- \else
- \scanresult@=-\maxdimen
- \fi
-% \end{macrocode}
-% Call \cw{endscan} to gobble garbage tokens, if any.
-% \begin{macrocode}
- \endscan
-}
-% \end{macrocode}
-% \end{macro}
-%
-% \begin{macro}{\checkdimen}
-%
-% Argument \arg{2} must be a dimen register; \arg{1} is expected to
-% be a macro holding zero or more arbitrary characters of category 11
-% or 12.
-% \begin{macrocode}
-\def\checkdimen#1#2{%
- \let\scansign@\@empty \def\scanresult@{#2}%
- \let\dimentrue@\@empty
- \dimenfirstpart\z@ \dimentoks{}%
- \xp@\scandimen#1xx\endscan
-}
-% \end{macrocode}
-% \end{macro}
-%
-% Finish up.
-% \begin{macrocode}
-\restorecatcodes
-\endinput
-% \end{macrocode}
-%
-% \CheckSum{858}
-% \Finale