% \def\fileversion{2.3}\def\filedate{2001/05/06}\def\docdate{2013/11/26}
%\iffalse These lines are necessary for the doc and docstrip utilities
%<*asciidocu>

%%#
%%# Author:
%%#
%%# M. V\"ath      martin@mvath.de
%%#
%%# This package bases on ideas of indextwo.sty which was written jointly with
%%#
%%# O. Karch       karch@informatik.uni-wuerzburg.de
%%# H.-C. Wirth    wirth@informatik.uni-wuerzburg.de

%%# The package may be distributed and/or modified under the conditions of
%%# the LaTeX Project Public License (version 1.3c or later).

%%#############################################################################
%%#
%%#  This LaTeX package provides a luxury front-end for the \index command.
%%#  For example, it allows to generate multiple \index entries in
%%#  almost any form by a single command. Extremely customizable.
%%#  Works with all versions of LaTeX and probably most other TeX formats, too.
%%#  This package requires the toolbox package at least in version 3.1.
%%#
%%#############################################################################

%%#
%%# This package was tested with LaTeX 2.09 and LaTeX 2e and should work
%%# with all other (future) versions of LaTeX and friends, too.
%%# It should even run with other (non-LaTeX) formats if an \index command
%%# with LaTeX-Syntax is provided.
%%#

% If you have access to (reasonable new) docstrip TeXfiles, you should run
% tex/latex on the file
%   `varindex.ins'
% to produce the files `varindex.sty', `varindex.txt' and `varindex.tex'.
% (In case `varindex.ins' is lost: You can find its content in the last
%  few lines of the file you are currently reading between the
%  lines %<*insfile> and %</insfile>;
%  BTW: You should name this file here `varindex.dtx' before you run tex
%  on varindex.ins).
%
% The file `varindex.txt' contains a `brief' (but complete) ascii
% documentation. If you have additionally access to LaTeX2e, you may run
% latex on the generated file `varindex.tex' to produce `varindex.dvi' which
% contains a more luxury documentation. (If you want an index or
% implementation details in the documentation read the comments in
% `varindex.tex' before running it through latex).
%
% If you do not have access to the docstrip TeXfiles you may simply
% copy/rename this file here into
%   `varindex.sty'
% to use the package and read the ascii documentation in this file.
% The ascii documentation (the content of `varindex.txt') consists of all
% lines in this file which start with %%# (this is the above block and another
% (very large) block which follows soon).
%
% But I do not recommended this, because TeX will need more time to read this
% longer file, and because there might be a (very slight) incompatibility if
% you use it (namely if you load `varindex.sty' with some *extremely* strange
% catcodes).
% I recommend in this case instead to generate the file `varindex.sty' as
% anything between the lines %<*package> and %</package> by hand.
% (This is not precisely the file which you get with the docstrip TeXfiles,
% but a reasonable good substitute. To speed up loading, you may also
% eliminate all lines starting with `% ' in `varindex.sty').
% For your information: The file `varindex.tex' produced by the docstrip
% TeXfiles is anything between the lines %<*driver> and %</driver>
% (with a line containing \endinput and two comment lines appended)
%
%
% In any case, you end up with the file `varindex.sty'.
%
% If you can generate `varindex.tex', you will probably want to stop
% reading now and instead read the `luxury' documentation in `varindex.dvi'.
%
% Otherwise, you have to make due with the following ascii documentation.
%
%
% \fi\catcode\#14\relax\iffalse  (This line is needed for the doc utility)
%

%%# To use varindex, you have to put `varindex.sty' in a path where TeX looks
%%# for its input files. You must also have the file `toolbox.sty' of the
%%# toolbox package in your path. The TeX documents using varindex need the
%%# following modifications in their header:
%%#
%%# a) If you use LaTeX2.09, use varindex as a style option, e.g.
%%#      \documentstyle[varindex]{article}
%%#    or
%%#      \documentstyle[varindex,12pt]{article}
%%# b) If you use LaTeX2e, put in the preamble the command
%%#      \usepackage{varindex}
%%# c) If you use some other (non-LaTeX) format, you will probably have to
%%#    insert a line like
%%#          \catcode`\@=11\input varindex.sty\catcode`\@=12\relax
%%#
%%# For TeX insiders: LaTeX-specific commands used in `varindex.sty' are only:
%%#
%%#   \makeatother
%%#   \makeatletter
%%#   \typeout
%%#   \RequirePackage
%%#   \newcommand (only in the form \newcommand{\command}{} to ensure that
%%#                 \command was not defined before)
%%#   \ProvidesPackage
%%#
%%#   The above commands are used only if they are defined (otherwise,
%%#   natural substitutes are used (\newcommand and \Provides package are
%%#   then not used at all))


%%#
%%# I strongly recommend to have additionally a look at the documentation
%%# of the "toolbox" package, in particular the \toolboxMakeDef command
%%# (provided since toolbox v3.1) which supplements \varindex.
%%#

%%#
%%# The \varindex command is highly customizable.
%%# So the following examples can only give you a rough impression of what
%%# you can actually do with it. You do not need to understand them now:
%%# The general usage is explained later. (Maybe you want to look up the
%%# examples afterwards).
%%#
%%#
%%# 1. Typical example of usage (using the \varindextwo macros):
%%#
%%#
%%#
%%# Suppose the following customization of the \varindex command was defined:
%%#     \def\Index{\varindex(){\varindextwoScan}{\varindextwo}[\emph]}
%%# Since we used the \varindextwo macros here, by default
%%# TWO index entries are generated with a single command:
%%#
%%# \Index{ring with $1$@one}
%%#
%%# ring
%%#    ~ with $1$            [sorted like: "with one"]
%%#
%%# $1$                      [sorted like: "one" ]
%%#    ring with ~           [sorted like: "ring with" (without " ~") ]
%%#
%%# DIFFERENT entries may be generated by your own rules. E.g.:
%%#
%%# \Index{Hausdorff measure of noncompactness}[4,23 2!~34!1= ]
%%#
%%# generates THREE index entries:
%%#
%%# noncompactness, measure of ~ (sorted like "noncompactness, measure of")
%%#
%%# measure
%%#   ~ of noncompactness
%%#         Hausdorff =    (here, = stands for the $\approx$ symbol)
%%#
%%# Hausdorff measure of noncompactness
%%#
%%# The last of these entries occurs only, because the last character in
%%# the format argument [...] was a space.
%%#
%%# If you want only slight modification(s) of the default, you need also just
%%# slight modifications of the command, e.g.
%%#
%%# \Index*,{ring with $1$@one} is as above, just that the second entry reads
%%#
%%# $1$, ring with ~  [sorted like "one, ring with" ]
%%#
%%# Moreover, since the `*' occured in the previous command, additionally
%%#
%%# \emph{ring with $1$}
%%#
%%# appears in the running text. Praxis shows that changes in the entries as
%%# above have to be made frequently in the `fine tuning' of the index of a
%%# book. Note that with the original \index the command would have to be
%%# changed completely for the above entries, because we have a "sort" entry
%%# for which a different rule applies if we use a subitem instead of a ",".
%%# (BTW: \varindex produces sort entries only if they are necessary).
%%#
%%# Of course, you may combine the default with your own format:
%%#
%%# \Index{internal integrable function}[23,1~ +]  generates
%%#
%%# integrable function, internal ~
%%#
%%# internal
%%#    ~ integrable function
%%#
%%# function
%%#    internal integrable ~
%%#
%%#
%%# 2. A simple example without the \varindextwo macros:
%%#
%%# \varindex*{topology of a normed space}{5!4~!1= 45,12}[|see{norm}]
%%#
%%# produces the two index entries:
%%#
%%#  space
%%#     normed ~
%%#         topology =
%%#
%%#  normed space, topology of
%%#
%%# In both entries not the pagenumber is cited, but the text "see also norm".
%%# Additionally, the tokens {topology of a normed space} are
%%# inserted into the running text (this would not happen if
%%# \varindex is used in place of \varindex*).
%%# Only in the first of the above entries, an `@' part will be added to the
%%# *.idx file: This entry is sorted as if " ~" resp. " =" would not exist.
%%#
%%#
%%# 3. Another example without the \varindextwo macros:
%%#
%%# \varindex[\emph][|textbf]%
%%#     {$\sigma$@sigma!$(\sigma-)$ finite {measure space}}%
%%#     {*1.23 23 23,1- 1-23}
%%#
%%# generates the three index entries:
%%#
%%#  finite measure space
%%#
%%#  finite measure space, $\sigma$- (sorted as "finite measure space, sigma-")
%%#
%%#  $\sigma$-finite measure space   (sorted as "sigma-finite measure space")
%%#
%%# In all three entries the pages are printed using \textbf.
%%#
%%# Moreover, the tokens
%%#   \emph{($\sigma$-)finite measure space}
%%# are put into the running text.
%%#
%%#
%%# 4. A simple example using a flag (without the \varindextwo macros):
%%#
%%# \varindex1{$L$ and $M$@L and M}{}
%%#          ^
%%#
%%# generates the index entry "$L$ and $M$" which is sorted like "L and M".
%%# In the form
%%#   \varindex1*{$L$ and $M$@L and M}{}
%%# or
%%#   \varindex*1{$L$ and $M$@L and M}{}
%%# additionally, {$L$ and $M$} is output into the running text.
%%#
%%# Note that without the `1', the index generated were
%%# "$L$ and $M$ and M" which were sorted like "$L$ and L and M".
%%#
%%#
%%# 5. An example for very primitive customized index commands (without the
%%#    \varindextwo macros):
%%#
%%# \def\myindexA{\varindex[\emph][|textbf]*}
%%# \def\myindexB{\varindex(){}{}[\emph][|textbf]}
%%#
%%# \myindexA{Foo1}{}
%%# \myindexB{Foo2}[][]
%%# \myindexB*{Foo3!{Foo 3}}
%%#
%%# Generates the index entries Foo1, Foo2, and Foo3. Moreover, \emph{Foo1}
%%# and \emph{Foo 3} is put into the running text. The pagenumber of Foo 2 in
%%# the index is printed normal, the others with \textbf.
%%#
%%#
%%# 6. An example of a primitive customized index command (without the
%%#    \varindextwo macros):
%%#
%%# \def\myindex{\varindex(\ifnum\varindexCount=2 1!2 2,1\fi){}{}[\emph]}
%%#
%%# \myindex{Foo Bar}
%%#    generates the index entries
%%#     Foo
%%#        Bar
%%#
%%#     Bar, Foo
%%#
%%# \myindex*{Foo Bar}
%%#    generates the same index entries as above and additionally outputs
%%#    \emph{Foo Bar} into the running text.
%%#
%%# \myindex*{other format}[21]
%%#   generates the index entry
%%#     format other
%%#   and outputs \emph{other format} into the running text.
%%#
%%# \myindex[|textbf]*{BoBo}
%%#   generates the index entry
%%#      BoBo
%%#   where the pagenumber is printed with \textbf.
%%#   Moreover, \emph{BoBo} is output into the running text.
%%#
%%#
%%# 7. A primitive varying customized index command (without the \varindextwo
%%#    macros):
%%#
%%# \def\myindex{\varindex(\ifnum\varindexCount=2 1!2 2,1\fi)%
%%#     (\ifnum\varindexCount=2*2,1 1!2 2,1\fi){}{}[\emph]}
%%#
%%# With the same commands as in 6. the same index entries are generated,
%%# but the tokens \emph{Far, Boo}, \emph{other format}, and \emph{Bobo}
%%# are output into the running text.
%%#
%%#
%%# 8. Examples with the \varindextwo macros:
%%#    We use a similar definition for \Index as in 1., and define an \iemph
%%#    macro which has `*' as a default and outputs with \emph.
%%#    Both sort of definitions are highly recommended for you private style.
%%#
%%# \def\Index{\varindex(){\varindextwoScan}{\varindextwo}[]}
%%# \def\iemph{\varindex(){\varindextwoScan\def\varindexStarPos{}}%
%%#              {\varindextwo}[\emph]}
%%# (note that \def\iemph{\varindex(){\varindextwoScan}{\varindextwo}[\emph]*}
%%#  is not so good as the above definition of \iemph as is explained below).
%%#
%%# \Index*{measure of $\gamma$-noncompactness@gamma-noncompactness!FOO}
%%#
%%#   generates the index entries
%%#
%%#   measure
%%#      ~ of $\gamma$-noncompactness   (sorted as "of gamma-noncompactness")
%%#
%%#   $\gamma$-noncompacntess
%%#      measure of ~
%%#
%%#   and outputs the tokens {measure of FOO} into the running text.
%%#
%%# \Index*,?_'{flic of flac}  generates
%%#
%%#   flic of flac
%%#
%%#   flac, flic
%%#
%%#   and outputs {flic of flac}.
%%#
%%# \Index^={$\sigma$@sigma algebra} generates
%%#
%%#   $\sigma$ (sorted as "sigma")
%%#      ~-algebra   (sorted as "algebra" resp. "-algebra", see below)
%%#
%%#   algebra
%%#      $\sigma$-~  (sorted as "sigma-")
%%#
%%# Here, the sorting "-algebra" is chosen when you used
%%#    \let\varindexOutSortDashBeg\varindexOutSortDash
%%# as described later.
%%#
%%# \Index-^>{$\sigma$@sigma algebra} generates
%%#
%%#   $\sigma$- (sorted as "sigma-")
%%#      ~ algebra   (sorted as "algebra")
%%#
%%# \Index^.,:{ultra filter} generates
%%#
%%#   ultra
%%#      ~filter        (no space following ~)
%%#   filter, ultra~    (no space in front of ~)
%%#
%%# \iemph[|see{recursive recursion}],_{recursive recursion}  generates
%%#
%%#  recursive recursion, \see{recursive recursion}
%%#
%%#  recursion, recursive ~, \see{recursive recursion}
%%#
%%#  and outputs \emph{recursive recursion}. If we would have used instead the
%%#  definition \def\iemph{\varindex(){\varindextwoScan}{\varindextwo}[]*},
%%#  the above call were not possible, since the optional arguments [|see...]
%%#  must occur in front of the flag `*'.
%%#
%%# \Index-;*{Flip Flop}   generates
%%#
%%#    Flip-Flop
%%#
%%#    FlopFlip
%%#
%%#    and outputs {Flip-Flop}  (!!!!!!!)
%%#
%%# \Index>{Flippy Flop!GO}[*2-1-2 2!1-1 1-12 +]
%%#
%%#   generates
%%#
%%#     Flop
%%#        Flippy-Flippy (no tilde!)
%%#
%%#     Flippy-Flippy Flop
%%#
%%#     Flippy
%%#        ~ Flop
%%#
%%#   and outputs {GO-Flippy-GO} (!!!)
%%#
%%# \Index*_,{shape {of a} of star}  generates
%%#
%%#    shape of a star
%%#
%%#    star, shape of ~ (sorted like "star, shape of")
%%#
%%# and outputs {shape of a star}.
%%#
%%# \Index1*{$L$ feature@L feature}
%%# \Index,{No $L$@L feature}
%%# generate
%%#
%%# $L$ feature (sorted like "L feature")
%%#
%%# No
%%#   ~ $L$ feature   (sorted like "L feature")
%%#
%%# feature, No $L$ ~ (sorted like "feature, No L")
%%#
%%# and outputs {$L$ feature}.
%%#
%%#
%%# 9. Example of a simple modification of the \varindextwo macros:
%%#
%%# \def\myindextwoScan{%
%%#      \varindextwoScan
%%#      \varindexNewFlag ~\vxtSpaceA
%%#      \varindexNewFlag 1\vxtSpaceB
%%#      \varindexNewFlag !\varindexOneEntry
%%#      \varindexMakeSplitExpand{/}{vxtSplitAtMagic}%
%%#      \def\varindexStarPos{}%
%%# }
%%# \def\myIndex{\varindex(){\myindextwoScan}{\varindextwo}[]}
%%#
%%# \myIndex behaves similar to the previous example with
%%# the following differences:
%%# 1. The flag `1' now has the previous meaning of "~".
%%# 2. One may now use "~" and "_" equivalently.
%%# 3. The flag `!' now has the originial meaning of the flag `1'.
%%# 4. Instead of `+' in the format string, the character `/' has to be used.

%%#
%%#
%%# Description:
%%#
%%# There are two possible calls:
%%#
%%# a) \varindex[textstyle][pagestyleA]flags{main entries}{format}[pagestyleB]
%%#
%%#    Here, [textstyle], [pagestyleA], [pagestyleB], and flags are optional.
%%#    (If [textstyle] is omitted, also [pagestyleA] must be omitted).
%%#
%%#    textstyle: describes the textstyle used for the output into the
%%#               running text (typically, textstyle is \emph or empty).
%%#    pagestyle: describes the pagestyle used for the output of the pagenumber
%%#               (you may also use a construct like |see{...} here).
%%#               If [pagestyleB] is present, this is the value used.
%%#               Otherwise, the value of [pagestyleA] is used (resp. nothing).
%%#    flags can be one (or both) of the following tokens:
%%#      `*': "Output into running text":
%%#           If no * is used in the format string (see below),
%%#           then the format is automatically prepended by the tokens `* '
%%#           resp. `*' (depending whether format starts with ` ' or not)
%%#      `1': The content of {main entries} is considered as a single entry
%%#           (see below).
%%#    {main entries} and
%%#    {format} are explained later.
%%#
%%# b) \varindex(formatA)(format*A){scanprogram}{mainprogram}
%%#         [textstyle][pagestyleA]flags{main entries}[formatB][pagestyleB]
%%#
%%#    Concerning [textstyle], [pagestyleA], flags, [pagestyleB], the same
%%#    remarks as in a) hold. Also (format*A) and (formatA) are optional.
%%#    (if (formatA) is omitted, also (format*A) must be omitted).
%%#
%%#    The "default" format is formatA resp. format*A (depending whether the
%%#    flag "*" has been used or not). If [formatB] is given, then this is
%%#    used as the format (i.e. it overrides the default).
%%#
%%#    scanprogram is executed immediately when \varindex reads it.
%%#    This can be used to introduce FURTHER flags (other than '*' and `1'),
%%#    see below.
%%#    mainprogram is executed AFTER the format string(s) has been expanded.
%%#
%%#
%%#
%%# It is recommended to define personal macros which contain all parameters up
%%# to the place where the flags occur (either with or without some flags,
%%# depending on the intention). See the examples above.
%%#
%%#
%%#
%%#
%%# The main argument (called {main entries} above) has the form
%%#
%%# entry1[`space' entry2][`space' entry3] ...
%%#
%%# where each of entry1 entry2 entry3 ...  in turn has the form
%%#   indextext@sorttext!outputtext or
%%#   indextext!outputtext@sorttext
%%# In each of these forms, @sorttext and !outputtext are optional.
%%# Here,
%%#   indextext  is the text which is put into the index for the entry
%%#   sorttext   is what is used as a sort criterion for the entry
%%#   outputtext is what is output into the running text for the entry
%%#
%%# Note that the symbol @ has in a sense the opposite meaning as in the
%%# original \index command:
%%# BEFORE that symbol, the desired text appears, and
%%# AFTER the symbol the place where it has to be sorted.
%%# However, we chose the symbols @ and !, because these two
%%# symbols are forbidden anyway (otherwise, makeindex will become confused).
%%#
%%# As usual, multiple spaces are considered as ONE space and do not generate
%%# empty "ghost" entries. Moreover, a space after a TeX-command like \LaTeX
%%# is eliminated by TeX and thus also not considered as a separator.
%%# You may use braces {...} either around a whole entry or
%%# around indextext resp. sorttext resp. outputtext to allow spaces
%%# which do not act as "separators" in the corresponding part. The braces
%%# around these parts will vanish. In particular, you may generate an
%%# empty entry with {} or ! or @ (bordered by spaces).
%%# If you want that braces occur in the OUTPUT, you have to add an
%%# additional pair of braces. Be aware that you write sufficiently
%%# many braces, if you really should need them: For example,
%%# the command \varindex{{{{A B}}}}{} produces the index entry {A B}:
%%# The outermost pair of braces is eliminated immediately by TeX. The second
%%# pair is eliminated, because this is a pair around a complete entry.
%%# Finally, another pair is eliminated, because it is a brace around
%%# indextext.
%%# With the flag `1', {main entries} is considered as one single entry.
%%# Nevertheless, also with this flag, an additional pair of braces around
%%# main entry is eliminated.
%%#
%%#
%%# The format string  has one of the following three forms
%%#   indexa[`space' indexb][`space' indexc] ... [*outputa][*outputb] ...
%%#   [*outputa][`space' indexa][`space' indexb][*outputc] ...
%%#   *
%%# where the order of arguments can be arbitrarily mixed (except for the first
%%# indexa in the first form). indexa indexb ... outputa ... describe the
%%# format of the index entries resp. of the output into the running text.
%%# The output is generated in the given order (this might be important, if
%%# a pagebreak occurs during the output into the running text).
%%# The last case is exceptional and equivalent to `*space'.
%%# The following characters/strings describe the format:
%%#
%%#  1-9: entry1-9
%%#  0:   entry 10
%%#  ~:   a placeholder sign
%%#  =:   alternative placeholder sign
%%#  _:   a space character
%%#  s:   The token \space (but _ is used for the sort entry)
%%#  .:   No space will follow
%%#  ,:   The character , (space will follow)
%%#  -:   The character - (no space will follow)
%%#  !:   Create a new sublevel
%%# All other tokens in this argument are forbidden.
%%# The token ! makes no sense for the output in the running text and thus
%%# is forbidden behind *.
%%#
%%# By a heuristic rule, a space is automatically inserted between two entries
%%# which generate non-empty text.  If the heuristic rule fails, you may always
%%# force a space by either "_" or "s", or forbid a space by ".".
%%#
%%# If a format is empty, all entries are taken in the order of input.
%%# Note that TeX eliminates concatenated spaces, and so you are able to
%%# produce an empty format entry only at the end or in front of a `*'.
%%# If you want to force an empty OUTPUT (is this ever useful?), you may use
%%# just "." as the entry.
%%#
%%# A sort entry is only written to the *.idx file if it differs from the
%%# text entry.

%%#
%%#
%%# Description of the \varindextwo macros:
%%#
%%# The macros
%%#
%%#   \varindextwoScan \varindextwo
%%#
%%# can be used together as parameters {scanprogram} resp. {mainprogram}
%%# for the \varindex command.
%%# If \varindex is called with these macros, and no
%%# explicit format argument is given, these macros generate a format
%%# depending on the number of entries in {main entries}:
%%#
%%# a) If there is only one entry in {main entries},
%%#    then the format "1" resp. "*1 1" is generated
%%#    (depending whether the * flag was used or not), i.e. the
%%#    entry is put into the index, and printed if \varindex was
%%#    called with the * flag.
%%#
%%# b) If there are two main entries like {graulty bazola},
%%#    then the format becomes { 1!~2 1!2~}, i.e. the index entries
%%#
%%#    graulty
%%#      ~ bazola
%%#
%%#    bazola
%%#       graulty ~
%%#
%%#    are produced.
%%#
%%#    You can modify the first format entry with the following flags:
%%#     flags  format used:           If used with `*' additionally:
%%#      _     12                     *12
%%#      -     1-2                    *1-2
%%#      ^     1!~-2 resp.  1!-2      *1-2
%%#      -^    1-!~.2 resp. 1-!~2     *1-2
%%#      .     1.2                    *1.2
%%#      .^    1!~.2                  *1.2
%%#      /     1!2   (without ~)      *12
%%#      <     no entry               *12
%%#
%%#    Here, the combinations -^ and .^ mean that both flags are used
%%#    (the order plays no role).
%%#    The output for ^ respectively -^ is determined by the switches
%%#        \varindextwoDashTildetrue (default)
%%#        \varindextwoDashTildefalse
%%#        \varindextwoDashSpacetrue
%%#        \varindextwoDashSpacefalse   (default)
%%#
%%#    You can modify the second format entry with the following flags:
%%#     flags  format used:
%%#      ,     2,1~ resp. 2,1
%%#      ~     21
%%#      =     2!1-~
%%#      ,=    2,1-~ resp. 2,1-
%%#      +     2-1
%%#      :     2!1.~
%%#      ,:    2,1.~ resp. 2,1
%%#      ;     2.1
%%#      >     no entry
%%#      !     Append ~ (if not already there)
%%#      ?     Without ~
%%#
%%#    Whether the first or the second alternatives in the above forms is used
%%#    depends on the status of the switch
%%#          \varindextwoCommaTildetrue   (default)
%%#          \varindextwoCommaTildefalse
%%#    We point out that "\varindextwoCommaTildefalse" was in earlier versions
%%#    of "varindex" the default (and could not be changed).
%%#    Note that this switch has no effect if the ! or ? flags are used.
%%#
%%#    HINT for remembering the symbols: The symbols _ . - for the first entry
%%#    are the same flags as for the output and the same flags which are used
%%#    in the format string. The corresponding symbols ~ : (and ;) = (and +)
%%#    for the second entry "look similar". The "," is only useful in the
%%#    second entry (and is the same symbol as in the format string).
%%#    The two exceptional symbols ">" and "<" can be read as "generate only
%%#    the forward-directed resp. backward-directed entry".
%%#
%%# c) If there are three main entries like {graulty of bazola}, then the
%%#    format becomes { 1!~23 3!12~}, i.e. the output becomes
%%#
%%#    graulty
%%#      ~ of bazola
%%#
%%#    bazola
%%#      graulty of ~
%%#
%%#    The flags have an analogous effect to b).
%%#    In addition, if the flags  ` resp. ' are used, the second word
%%#    (in our example: `of') is omitted from the output in the first resp.
%%#    in the second index entry. If the * flag and ` are used together, the
%%#    second entry is also omitted from the output into the running text.
%%#
%%# d) If there are four main entries like {graulty of OF bazola}, then the
%%#    format becomes { 1!~24 4!31~}, i.e. the output becomes
%%#
%%#    graulty
%%#      ~ of bazola
%%#
%%#    bazola
%%#      graulty OF ~
%%#
%%#    i.e. we have a similar effect as in c) except that for the first entry
%%#    the third word is skipped, and for the second entry the second word is
%%#    skipped. All flags as in c) can be used with an analogous meaning.
%%#    Also if the * flag is used, the output into the running text is
%%#    analogous to c) (the third word is skipped).
%%#
%%# e) If there are more than four main entries, then
%%#    (formatA) resp. (format*A) is used.
%%#
%%# If an explicit format argument is provided, then this is used except
%%# when it contains the symbol +. In this case, this symbol is replaced by the
%%# output which would be generated "normally". If additionally the `*'-flag
%%# is used, there is a special rule:
%%# If the explicit format contains a `*', then no additional `*'-output is
%%# produced. Otherwise, the `*'-output from a)-e) is prepended to the given
%%# format string (with a space at the end, unless the format string is empty).
%%# This means that `normally' you have the output from a)-e),
%%# unless you write an own `*'-part in the argument.
%%#
%%# If you do not like the tokens chosen for the default symbols, you can
%%# easily choose your own tokens by just replacing \varindextwoScan with your
%%# own macro (or defining your own ``magic'' tokens after \varindextwoScan,
%%# see Example 9. above).

%%#
%%#
%%# Primitive customization:
%%#
%%#
%%# You may use the command
%%#     \varindexUsePlaceholderAfalse
%%#     \varindexUsePlaceholderBfalse
%%# to ignore the sign ~ resp. = in the format entry.
%%# You can easily restore the default by
%%#     \varindexUsePlaceholderAtrue
%%#     \varindexUsePlaceholderBtrue
%%#
%%# If you want to switch off the placeholder only at the beginning of
%%# a new entry (resp. of a new subentry), you can use the commands
%%#     \varindexLeadingPlaceholderAfalse
%%#     \varindexLeadingPlaceholderBfalse
%%# The default is restored with
%%#     \varindexLeadingPlaceholderAtrue
%%#     \varindexLeadingPlaceholderBtrue
%%#
%%# By default, the "," in the format string is in this connection
%%# treated as a symbol generating a new "subentry". You can change this
%%# default with the command
%%#  \varindexCommaLeadsfalse
%%# You may switch back with
%%#  \varindexCommaLeadstrue
%%#
%%#
%%# The content of the macro
%%#   \varindexPlaceholderAText
%%#   \varindexPlaceholderASort
%%#   \varindexPlaceholderBText
%%#   \varindexPlaceholderBSort
%%# is used as the placeholder ~ resp. =  in the index resp. sort entry.
%%# Note that if one of these entries expands empty, different rules for
%%# the automatic insertion of spaces apply (for the index and sort entry
%%# separately).
%%#
%%# You may prefer that the placeholder text resp. sort content changes at
%%# run time to the context of the previous ! entry. For example, you may
%%# want that the command \varindex{Gibble Gobble}{2!~2} produces the index
%%# entry
%%#   Gibble
%%#     Gibble Gobble
%%# (or is at least SORTED as such an entry).
%%#
%%# You can achieve this with the commands
%%#  \varindexPreviousATexttrue
%%#  \varindexPreviousASorttrue
%%#  \varindexPreviousBTexttrue
%%#  \varindexPreviousBSorttrue
%%# for the text and sort entry and the placeholders ~ and =, respectively.
%%# With these commands, the content of \varindexPlaceholderAText etc
%%# is only used as the default if no previous ! entry was given.
%%# You can switch back to the default mode with the respective commands
%%#  \varindexPreviousATextfalse
%%#  \varindexPreviousASortfalse
%%#  \varindexPreviousBTextfalse
%%#  \varindexPreviousBSortfalse
%%#
%%# By default, the ',' in the format entry is considered in this connection
%%# similar to '!'. If you want to change this, use the command
%%#  \varindexCommaExpandfalse
%%# You may switch back with
%%#  \varindexCommaExpandtrue
%%#
%%# Since version 2.2, the dash "-" occurring at the beginning of entry
%%# (or after "!" or even after "," if \varindexCommaExpandtrue is in effect)
%%# is ignored for sorting. You can switch off this feature with the command
%%#
%%#   \let\varindexOutSortDashBeg\varindexOutSortDash
%%#
%%# To restore the new default, use
%%#
%%#   \let\varindexOutSortDashBeg\toolboxEmpty
%%#
%%#
%%# The commands
%%#
%%#   \varindexSetWordString{|}     % Use | as separator for entries instead
%%#                                 % of spaces.
%%#   \varindexSetSortString{>}     % default: @
%%#   \varindexSetPlainString{<}    % default: !
%%#
%%# set the default "magic" strings used for the main argument.
%%# The argument of the above macros is intentionally NOT expanded (so that
%%# you do not have troubles with active characters like ~).
%%# To force expansion, use e.g.
%%#   \expandafter\varindexSetWordString\expandafter{\MacroToExpand}
%%# or some similar construct.
%%#
%%# If you use a different separator than `space' for the entries, you may want
%%# to change the rule how braces are eliminated. With the commands
%%#   \varindexEliminateOuterBracetrue
%%#   \varindexEliminateInnerBracetrue
%%#   \varindexEliminateOuterBracefalse
%%#   \varindexEliminateInnerBracefalse
%%# you may allow resp. forbid the elimination of braces around the entries
%%# resp. around {indextext} {sorttext} and {outputtext}
%%# With the flag `1', {main entries} is considered as one single entry,
%%# but if \varindexEliminateOuterBracetrue is set (which is the default)
%%# also in this case an additional pair of braces around main entry is
%%# eliminated.
%%#
%%# Similarly as above,
%%#
%%#   \varindexSetIndexString{|}    % default: space
%%#   \varindexSetOutputString{<}   % default: *
%%#   \varindexSetTildeAString{~}   % (is default)
%%#   \varindexSetTildeBString{=}   % (is default)
%%#   \varindexSetSpaceString{_}    % (is default)
%%#   \varindexSetSpaceTokString{s} % (is default)
%%#   \varindexSetOmitString{.}     % (is default)
%%#   \varindexSetCommaString{,}    % (is default)
%%#   \varindexSetDashString{-}     % (is default)
%%#   \varindexSetExclamString{!}   % (is default)
%%#   \varindexSetStringForNr 1{a}  % default: 1
%%#   \varindexSetStringForNr{11}{b}% No default!
%%#
%%# sets the default "magic" strings used for the format argument.
%%# In contrast to before, the arguments are fully expanded (with
%%# \varindexedef, see below).
%%# Note that the last command provides a way to access more than 10 entries!
%%#
%%# If you use \varindexSetIndexString{|} (with some non-space token),
%%# you may still use spaces in the format which then are just ignored.
%%#
%%#
%%# Avoid that one string is a prefix of another string: In this case,
%%# strange effects may happen, and this behavior may even change in
%%# future releases of this package.
%%# Note that the above effect may arise unintentionally if you use active
%%# chars. For this reason, `~' is defined to expand to the letter `~',
%%# before the expansion is executed. Maybe in later releases of this package
%%# there will be further such redefinitions.
%%# You can explicitly use this feature in your own macros by using
%%# \varindexedef instead of the usual \edef, see below.
%%#
%%#
%%# You can change the output for the text resp sort entry of the various
%%# symbols used in the format string. More precisely, you can redefine
%%#
%%#     \varindexOutExclam     % Text output for !
%%#     \varindexFollowsExclam % Decide whether magic space after ! is inserted
%%#     \varindexOutTextComma  % Text output for ,
%%#     \varindexOutSortComma  % Sort output for ,
%%#     \varindexFollowsComma  % Decide whether magic space after , is inserted
%%#     \varindexPreceedsComma % Decide whether magic space before , is erased
%%#     \varindexOutTextSpace  % Text output for _
%%#     \varindexOutSortSpace  % Sort output for _
%%#     \varindexFollowsSpace  % Decide whether magic space after _ is inserted
%%#     \varindexPreceedsSpace % Decide whether magic space before _ is erase
%%#     \varindexOutTextSpTok  % Text output for s
%%#     \varindexOutSortSpTok  % Sort output for s
%%#     \varindexFollowsSpTok  % Decide whether magic space after s is inserted
%%#     \varindexPreceedsSpTok % Decide whether magic space before s is erased
%%#     \varindexOutTextDash   % Text output for -
%%#     \varindexOutSortDash   % Sort output for -
%%#     \varindexOutSortDashBeg% Sort output for - if nothing preceeds
%%#     \varindexFollowsDash   % Decide whether magic space after - is inserted
%%#     \varindexPreceedsDash  % Decide whether magic space before - is erased
%%#     \varindexOutTextOmit   % Text output for .
%%#     \varindexOutSortOmit   % Sort output for .
%%#     \varindexFollowsOmit   % Decide whether magic space after . is inserted
%%#     \varindexPreceedsOmit  % Decide whether magic space before . is erased
%%#
%%# The meaning is as follows: \varindexOut... should just expand to the
%%# tokens which should be written into the text resp. sort output.
%%# The corresponding macro \varindexFollows... is typically defined with
%%# one of the following commands:
%%#     \let\varindexFollows...\varindexNextSpace
%%#     \let\varindexFollows...\varindexNoNextSpace
%%# In the first case, a space is `magically' inserted in front of a
%%# subsequent token (unless this token erases the magic space).
%%# In the second case, no space follows, of course.
%%# There is an alternative call:
%%#     \def\varindexFollows...{\varindexSpaceTexttrue\varindexSpaceSortfalse}
%%# This definition achieves that for the text output a space should follow,
%%# but not for the sort output. Of course, you can also use similarly the
%%# commands \varindexSpaceTextfalse and/or \varindexSpaceSorttrue
%%# in the above definition (the effect should be obvious).
%%# In particular, \varindexNextSpace is equivalent to
%%# \varindexSpaceTexttrue\varindexSorttrue.
%%#
%%# The macro \varindexPreceeds... is similarly as \varindexFollows... with
%%# the difference that it is executed BEFORE the token in question is
%%# output. In particular, you can ignore a previous `magic space'.
%%# All of the `magic space' commands are defined by default with
%%#      \let\varindexFollows...\toolboxEmpty
%%#      \let\varindexPreceeds...\varindexNoNextSpace
%%# with the following two exceptions:
%%#      \let\varindexFollowsExclam\varindexNoNextSpace
%%#      \let\varindexFollowsComma\varindexNextSpace
%%#
%%# Documented Features (in later versions, these will still be available,
%%#  and if they contain a bug now, they will be repaired to work this way):
%%#
%%# (In a future release, varindex will cook your coffee, too)
%%#
%%#
%%#
%%# The main argument is NOT expanded, so you can actually write macros
%%# into the *.idx file. The command actually used to write the index is
%%#
%%#      \varindexIndex
%%#
%%# (which by default is of course defined as \index).
%%# Since some implementations of the \index command still (partially) expand
%%# their argument (which might be considered as a bug), the argument of
%%# \varindexIndex is translated before the actual call with the aid of
%%# \toolboxMakeHarmless.
%%# If you want to redefine \varindexIndex to a personal \index-output
%%# function, you might want to skip the \toolboxMakeHarmless step.
%%# This is arranged with the command
%%#
%%#    \varindexNoConvert
%%#
%%# You can cancel the effect of this command with
%%#
%%#    \varindexDoConvert
%%#
%%# Even if \varindexNoConvert is not in effect, the main argument is not
%%# expanded (and in particular, the argument of \varindexIndex consists of
%%# the corresponding entries in an unexpanded form).
%%#
%%# The whole \varindex... procedure takes place in a group, so all your
%%# variables are local to a single call (unless you use \gdef...
%%# (and friends) of course).
%%#
%%# There are no \global directives in any macro related with this package.
%%# In particular, if you call the above macros in a group (or redefine
%%# \varindexIndex), the effect holds only until the end of the group.
%%#
%%# In contrast to the main argument, the format argument is expanded
%%# (with \varindexedef, see below)
%%# (and also formatA resp. format*A is expanded before),
%%# so you can define abbreviations (even simple conditional abbreviations)
%%# for the format. Note, however, that the expansion is essentially only
%%# evaluated with \edef. So, you will probably not want to use e.g. \relax,
%%# since this command expands to itself (and  not into nothing - use
%%# \toolboxEmpty if you want the latter).
%%# If you want more complex abbreviations, use {mainprogram} instead.
%%#
%%# In order to help you write conditional abbreviations, the
%%# following variables are defined when your macro is expanded (and in
%%# particular also in {mainprogram}). You may of course redefine them
%%# appropriately:
%%#
%%#    \varindexAppend     The argument pagestyleA resp. pagestyleB.
%%#    \varindexCount      a TeX counter containing the number of entries in
%%#                        the main argument.
%%#    \varindexLastEntry  This is a macro (NOT a counter) which is usually
%%#                        the value of \the\varindexCount. See below.
%%#    \varindexEntry1-... The (index) text occuring in entry 1-...
%%#    \varindexSort1-...  The corresponding sort entry. If no sort entry was
%%#                        given, this is the same as the corresponding
%%#                        \varindexEntry1-...
%%#    \varindexPlain1-... The text which should be output in the text. If no
%%#                        such entry was given, this is the same as the
%%#                        corresponding \varindexEntry1-...
%%#
%%# \varindexCount may be larger than 10, and correspondingly there may be
%%# also more than 10 different \varindexEntry/Sort/Plain macros.
%%# If you ADD entries, you have to increase \varindexCount correspondingly,
%%# otherwise an error is produced if the format string contains an entry
%%# larger then \varindexCount. However, your additional entries are NOT
%%# output for EMPTY formats: For empty formats the number used is
%%# \varindexLastEntry. So if you want to output also your additional entries
%%# in empty formats, you have to set \varindexLastEntry to your modified
%%# value of \varindexCount in mainprogram. You may e.g. achieve this with
%%# the following lines:
%%#   \expandafter
%%#   \def\expandafter\varindexLastEntry\expandafter{\the\varindexCount}%
%%# It is admissible that \varindexLastEntry is larger than \varindexCount:
%%# In this case all entries until \varindexLastEntry are
%%# output in empty formats without producing an error (provided, of course,
%%# that the corresponding variables \varindexEntry.../Sort.../Plain... are
%%# defined).
%%#
%%# Note that numbers in TeX command names require special handling,
%%# i.e., you have to use something like 
%%#
%%# \csname varindexPlain2\endcsname
%%#
%%# to access variables. To avoid this, you may want to use the macros
%%# \toolboxLet and \toolboxWithNr of the toolbox package. Examples are
%%#   \toolboxWithNr 1\let{varindexEntry}\toolboxEmpty
%%#   \toolboxWithNr {10}\def{varindexSort}{Foo}
%%#   \toolboxLet    \mymacro{varindexPlain\the\mycount}
%%# These commands set \varindexEntry1 to \toolboxEmpty, \varindexSort10 to
%%# Foo, and \mymacro to the content of \varindexPlain5 (if \mycount=5).
%%#
%%# All these variables are also available when {mainprogram} is expanded.
%%# In {mainprogram} also the following functions are useful:
%%#
%%# \varindexFormat
%%#    This macro expands to the actual format string which is used.
%%#    The main purpose of {mainprogram} will usually be to (re)define
%%#    this macro. After {mainprogram} has been called, this macro is
%%#    modified in several ways:
%%#    1. \varindexFormat is expanded (with \varindexedef, see below).
%%#       (thus, \varindexFormat is expanded BEFORE AND AFTER the call
%%#        of {mainprogram}).
%%#    2. \ifvarindexStar is true (see below) a `*' resp. `* ' is prepended.
%%#    3. If it is only "*", it is replaced by "* ".
%%#    Note that before execution of {mainprogram}, no tests on the
%%#    validity of the format are done: You may use your own symbols to
%%#    `communicate' with {mainprogram} (if {mainprogram} eliminates these
%%#    symbols afterwards).
%%#
%%# \varindexFormatDefault
%%#    This macro expands to formatA resp format*A (in the expanded form)
%%#    depending whether the * flag has been used in the call.
%%#    Note that this expansion was executed BEFORE the optional format
%%#    argument is expanded for the first time.
%%#
%%# \ifvarindexStar ifpart [ \else elsepart] \fi
%%#    If the optional * was given, the ifpart is executed, otherwise the
%%#    else-part. This is a TeX-if-command. In particular, by
%%#    \varindexStarfalse resp. \varindexStartrue
%%#    you may change the behavior for future if's. This can be used
%%#    to prevent/force that a `*' resp. `* ' is prepended to the format
%%#    after the execution of {mainprogram}.
%%#    Setting of this variable has no effect in {scanprogram}
%%#
%%# \ifvarindexExplicitFormat
%%#    This is a TeX-if-command (see above) which is true if the optional
%%#    format argument [formatB] was given.
%%#
%%#
%%# To "compose" the format, you may want to use the macros
%%#
%%# \toolboxDef\macrotodefine{argumentlist}
%%# \toolboxAppend\macrotoappend{argumentlist}
%%# \varindexedef\macrotodefine{argumentlist}
%%#
%%# All commands work similar to \def resp. \edef with two differences:
%%#
%%# For \toolboxDef\macro the argumentlist is expanded precisely by one level
%%# (for details, see the documentation of the toolbox package).
%%# Of course, for \toolboxAppend, the new definition is appended to the old
%%# definition of \macro.
%%#
%%# \varindexedef fully expands argumentlist. However, some active symbols
%%# (currently only ~, but additional symbols might follow in a future release)
%%# are deactivated before expansion, so that ~ actually expands to ~ and not
%%# to a strange command sequence.
%%#
%%#
%%# To "decompose" the format, you may want to use one of the macros
%%#
%%# \toolboxSplitAt{arg}{critical}{\before}{\after}
%%# \toolboxMakeSplit{critical}{command} resp.
%%# \varindexMakeSplitExpand{\MacrosWhichExpandToCritical}{command}
%%# \varindexMakeVarSplit\variable{\MacrosWhichExpandToCritical}{command}
%%#
%%# The first two macros are described in the toolbox package.
%%# The remaining two macros are similar to \varindexMakeSplit with the
%%# difference that the argument {critical} is obtained by expanding
%%# \MacrosWhichExpandToCritical with the aid of \varindexedef.
%%# In the last form, additionally \variable is (re)defined to expand to
%%# the value of critical.
%%#
%%# The following instances of a command created by \toolboxMakeSplit
%%# exist (the content of "critical" should be obvious):
%%#
%%# \varindexSplitAtIndex   (generated by \varindexSetIndexString)
%%# \varindexSplitAtOutput
%%# \varindexSplitAtTildeA
%%# \varindexSplitAtTildeB
%%# \varindexSplitAtSpace   <- Here the default content of "critical" is "_"
%%# \varindexSplitAtSpaceTok<- Here the default content of "critical" is "s"
%%# \varindexSplitAtOmit
%%# \varindexSplitAtComma
%%# \varindexSplitAtDash
%%# \varindexSplitAtExclam
%%# \varindexSplitAtNr?? (here, ?? is the number as a string; use
%%#    \csname ....\endcsname to call this macro. Only those numbers are
%%#    admissible which have been introduced with \varindexSetStringForNr
%%#    (this is the case for 1-10)).
%%#
%%# There is also the instance
%%# \varindexSplitSpace (to be distinguished from \varindexSplitAtSpace)
%%# where "critical" is the space symbol.
%%#
%%#
%%# In addition, you may use the variables
%%#   \varindexIndexString
%%#   \varindexOutputString
%%#   \varindexTildeAString
%%#   ....
%%#   \varindexNr??String
%%# which expand to the corresponding strings.
%%#
%%# All the previous macros should not be redefined `by hand'.
%%# They are implicitly redefined by the \varindexSet... commands.
%%#
%%#
%%#
%%#
%%# To drop possible braces, use the command
%%#
%%#   \toolboxDropBrace\variable
%%#
%%# of the toolbox package.
%%#
%%#
%%# In {scanprogram} you may already want to call \toolboxMakeSplit:
%%# In this way many choices for the tokens are put into {scanprogram},
%%# and in this sense, you can consider {mainprogram} as `customizable'.
%%# Anyway, the main task of {scanprogram} is to introduce new flags with
%%#
%%#   \varindexNewFlag\token\position[program\varindexEndOfFlag]
%%# or
%%#   \varindexNewFlag\token\position[\programmacro]
%%#
%%# Here, \position is an (undefined) macro, and \token an arbitrary token.
%%# The [program(macro)] part is optional and is explained later.
%%# If token is entered in the current call of \varindex instead (or in
%%# addition to) the star, then \position is defined to expand to a
%%# (literally) number, namely the (last) position where token was given
%%# inside the "*-argument-list" (counting from 0). For example, if
%%#   \varindexNewFlag ,\commapos
%%#   \varindexNewFlag -\minuspos
%%#   \varindexNewFlag .\pointpos
%%# are used in \programA, then in the call
%%# \varindex(){\programA}{\programB},-**-{}
%%# the variable \commapos will expand in \programB to `0', while
%%# \minuspos will expand to `4' (the last position is taken). Finally,
%%# \pointpos is \undefined (unless you defined \pointpos differently AFTER
%%# the call of \varindexNewFlag; in this case, this is the default).
%%# If \varindexNewFlag is called multiple times with the same token,
%%# only the LAST call with this token takes effect.
%%#
%%# The flags `*' and `1' are introduced in this way with
%%#
%%# \varindexNewFlag *\varindexStarPos
%%# \varindexNewFlag 1\varindexOneEntry
%%#
%%# BEFORE {scanprogram} is executed. This means:
%%# 1. Usually, \varindexStarPos contains the (last) position of `*'
%%#    (resp. is undefined). Moreover, if you define \varindexStarPos in
%%#    {scanprogram} or in some flag, you get the same effect as if `*'
%%#    had been used.
%%#    An analogous remark holds for \varindexOneEntry.
%%# 2. If you introduce `*' with \varindexNewFlag, the `*' looses its
%%#    original meaning. The same holds for `1'.
%%#
%%# If you have added a [program(macro)] part, this part is expanded
%%# whenever the flag introduced by \token is used in the call of \varindex
%%# (note that it is not excluded that this happens several times within one
%%# call). More precisely, program is expanded AFTER the variable \position has
%%# been set to the corresponding value, so you may already use \position in
%%# the program part.
%%#
%%# IMPORTANT: The last token expanded in program MUST be \varindexEndOfFlag.
%%# This is not nice but has to do with the way TeX parses its arguments.
%%# Also if you use the form \programmacro, the VERY LAST token expanded
%%# must be \varindexEndOfFlag. Even a construction like
%%# \def\myprogrammacro{\ifx....
%%#   \CallSomeMacroWithvarindexEndOfFlagAtTheEnd
%%# \fi
%%# \varindexEndOfFlag}
%%# is forbidden: In \CallSomeMacroWithvarindexEndOfFlagAtTheEnd an error
%%# would occur at the end, since still the tokens \fi\varindexEndOfFlag
%%# are in the tokenlist when EndOfFlag is reached there.
%%# As a workaround, you may e.g. use
%%# \def\myprogrammacro{\ifx...
%%#    \def\execute{....\varindexEndOfFlag}%
%%# \else
%%#    \def\execute{\varindexEndOfFlag}%
%%# \execute}
%%#
%%# If you use the form [\programmacro], your macro may even read additional
%%# parameters. These parameters are expected in the call of \varindex BEHIND
%%# the flag you have introduced. So you may actually use flags with
%%# parameters. For example, if \scanprogram contains a macro like
%%#    \varindexNewFlag -\minuspos[\readpara]
%%# and you have defined
%%#    \def\readpara#1{\def\merk{#1}\varindexEndOfFlag}
%%# then the call \varindex(){\scanprogram}{\mainprg}*-{Foo}-{Foo 2}*{Entries}
%%# is admissible, and in mainprogram, \merk will have the value Foo 2.
%%#
%%# If you are more familiar with TeX, you can even allow OPTIONAL arguments
%%# following your flag: The value of the next (non-space) token is at the
%%# call of \programmacro already saved into the variable
%%#     \varindexNextToken,
%%# so you can just use it to test for e.g. [. In this connection, you may
%%# also want to use the commands
%%#     \varindexTestAndExec
%%#     \varindexSkipAndExec
%%# Please, see the program text how these commands are applied.

%%# Since version 2.1 you can also hack in personal extensions of the format
%%# string. To do this, use the command
%%#
%%#    \varindexNewFormat\splitcommand{action}
%%#
%%# Here, \splitcommand is a command generated by \toolboxMakeSplit or
%%# friends (preferably by \varindexMakeSplitExpand, because the command should
%%# act on the format string which is expand with \varindexedef).
%%# The string where it splits is the new string you can use
%%# in the format argument after this call. For each
%%# occurrence of the corresponding string in the format argument,
%%# action will be executed. If \splitcommand splits at a string which already
%%# had a previous meaning in the format string (or which is a prefix to
%%# such a string), the old meaning of these tokens in the format string is
%%# overridden.
%%#
%%# Typically, action will contain the following commands:
%%# One action will probably be to output a desired token (sequence) via
%%#
%%#    \varindexTokensOut{TextToken}{SortToken}
%%# or
%%#    \varindexTokensOutExpand\MacroForTextToken\MacroForSortToken
%%#
%%# In the first form, TextToken resp. SortToken is the token sequence put
%%# into the index or running text respectively into the sort entry of the
%%# index for the corresponding format entry. The second form is similar with
%%# the only difference that the arguments must be single macros which expand
%%# to TextToken and SortToken, respectively.
%%#
%%# The variable \ifvarindexIndexMode can be used to test whether the
%%# output goes into the running text or into the index (i.e. whether a `*'
%%# preceeded the current entry). For output into the text, SortToken is
%%# ignored, of course).
%%#
%%# Another action in \varindexNewFormat will probably be to take care of the
%%# magic space flags. This is achieved by a call to \varindexNextSpace or
%%# \varindexNoNextSpace (or separately via \varindexSpaceTexttrue/false resp.
%%# \varindexSpaceTexttrue/false); see the earlier description.
%%# The magic space flags are taken into account in "\varindexTokensOut".
%%# Thus, if you want to ignore the previous flag for some reason you should
%%# set them correspondingly before this call.
%%# However, after the call you should also set them correspondingly for
%%# further processing.
%%#
%%# Example:
%%#   \varindexMakeSplitExpand{:}{splitAtColon}
%%#   \varindexNewFormat\splitAtColon{%
%%#      \varindexNoNextSpace
%%#      \ifvarindexIndexMode
%%#         \varindexTokensOut{:}{}\varindexSpaceTexttrue
%%#      \else
%%#         \varindexTokensOut{---}{}\varindexNoNextSpace
%%#      \fi}
%%#
%%# defines a new format entry `:' which has the meaning that a colon
%%# (automagically followed by a space) is put into the index but not into
%%# the sort entry. Moreover, in the running text, the colon appears as a
%%# long dash with no space followed. In any case, there is no magic space
%%# output in front of the colon.
%%# 

%%# As an alternative action in \varindexNewFormat, you can also call the
%%# default commands for the format entries. The corresponding macros are
%%#
%%#        \varindexAddSpace           _
%%#        \varindexAddSpTok           s
%%#        \varindexAddOmit            .
%%#        \varindexAddDash            -
%%#        \varindexAddComma           ,
%%#        \varindexAddExclam          !
%%#        \varindexAddTildeA          ~
%%#        \varindexAddTildeB          =
%%#        \varindexAddNumber{number}  0-9 (the argument is 1-...)
%%#
%%# The precise meaning of these macros is intentionally NOT documented,
%%# because some details or side effects might change in a future release of
%%# "varindex". But just for this reason, it might be advantageous to use the
%%# above macros instead of writing personal substitutes which may fail to
%%# have such side effects.
%%#
%%# Example:
%%#    \varindexMakeSplitExpand{~-}{splitPhrase}% It is important to expand
%%#                                             % (to have correct ~ catcode)
%%#    \varindexNewFormat\splitPhrase{%
%%#       \varindexAddTildeA
%%#       \let\remember\varindexPreceedsDash
%%#       \let\varindexPreceedsDash\toolboxEmpty
%%#       \varindexAddDash
%%#       \let\varindexPreceedsDash\remember}
%%#
%%# After the above command, "~" and "-" have their usual meaning in
%%# the format string except when they follow immediately in the form "~-".
%%# In this case, the behavior of "-" changes as if \varindexPreceedsDash
%%# were empty (which has the effect that the output looks like "~ -" instead
%%# of "~-".
%%# Note that although "~" is a prefix to "~-", the converse is not true:
%%# Thus, the above command does not change the previous meaning of "~"
%%# (and of course also not of "-").

%%#
%%# Some hack:
%%# Tex ignores leading spaces in the argument list of a `normal' macro.
%%# This has the effect that you should be able to insert spaces BETWEEN any
%%# of your arguments without any trouble.
%%# If this does not work the expected way, you can use the command
%%# 
%%#    \let\varindexArgumentSpace\varindexSkipTricky
%%#
%%# which implements an own macro which does this task. Sadly, this can only
%%# be done with some catcode trickery which in turn might bring you in trouble
%%# with some (VERY) exotic packages. You can restore the default with
%%#
%%#    \let\varindexArgumentSpace\varindexSkipDefault
%%#
%%#
%%# All above custamization commands/variables may be used anytime between
%%# two \varindex calls. They take effect with the next \varindex call.



%%# The macros \varindextwoScan and \varindextwo are considered as a
%%# (very useful) EXAMPLE for the customization of the \varindex command.
%%# If you need further customization, have a look at their definition first.

%  This is the end of the ascii documentation
%</asciidocu>
%\fi\catcode\#6\relax
%
% \iffalse
%<*metainfo>
% \fi

% \def\basename{varindex}

%% \CharacterTable
%%  {Upper-case    \A\B\C\D\E\F\G\H\I\J\K\L\M\N\O\P\Q\R\S\T\U\V\W\X\Y\Z
%%   Lower-case    \a\b\c\d\e\f\g\h\i\j\k\l\m\n\o\p\q\r\s\t\u\v\w\x\y\z
%%   Digits        \0\1\2\3\4\5\6\7\8\9
%%   Exclamation   \!     Double quote  \"     Hash (number) \#
%%   Dollar        \$     Percent       \%     Ampersand     \&
%%   Acute accent  \'     Left paren    \(     Right paren   \)
%%   Asterisk      \*     Plus          \+     Comma         \,
%%   Minus         \-     Point         \.     Solidus       \/
%%   Colon         \:     Semicolon     \;     Less than     \<
%%   Equals        \=     Greater than  \>     Question mark \?
%%   Commercial at \@     Left bracket  \[     Backslash     \\
%%   Right bracket \]     Circumflex    \^     Underscore    \_
%%   Grave accent  \`     Left brace    \{     Vertical bar  \|
%%   Right brace   \}     Tilde         \~}

% \CheckSum{2042}

% \iffalse
%   \RecordChanges
% \fi
% \def\mypar{}
% \def\ChangesTab{}\let\orichanges\changes
%   \def\changes#1#2#3{\orichanges{#1}{#2}{#3}%
%   \expandafter\def\expandafter\ChangesTab\expandafter{\ChangesTab
%   \item[v#1 {\mdseries(\emph{#2})}] {#3.}}}
% \newcommand{\cmdcite}[1]{\texttt{\textbackslash#1}}

% \changes{2.3}{2013/11/26}{Update email, add license. Date/version remains the
%              same since only commments and documentation are modified}
% \changes{2.3}{2001/05/06}{Introduced new macros \cmdcite{varindexIndex},
%              \cmdcite{varindexDoConvert}, \cmdcite{varindexNoConvert}.
%              The layout of the documentation was slightly improved.
%              Moreover, a completely new section was added (on
%              \cmdcite{toolboxMakeDef}).\mypar
%              Use \cmdcite{typeout} instead of \cmdcite{message},
%              but provide a workaround for plain \TeX. Avoid error with
%              plain \TeX\ and \LaTeX 2.09 (because \cmdcite{newcount} is
%              defined there as \cmdcite{outer} which is rather annoying).
%              Since \texttt{toolbox} works around this problem, finally also
%              the local \cmdcite{if}'s now actually \emph{are} local}
% \changes{2.2}{2001/05/01}{Introduced \texttt{=:\^} flags into
%              \cmdcite{varindextwo}. Changed sorting of entries
%              starting with ``\texttt{-}''
%              (see \cmdcite{varindexOutSortDashBeg}).\mypar
%              Improved documentation, in particular of \cmdcite{varindextwo}
%              and \cmdcite{varindexNewFormat}.
%              Also new examples were added for both commands}
% \changes{2.1}{2001/04/30}{Finally implemented placeholder replacement
%              (\cmdcite{varindexPreviousATexttrue} and friends) such that
%              it works also in the expected way with commas.\mypar
%              Introduced separated command \cmdcite{varindexCommaExpandtrue}
%              which was in earlier versions automatically coupled with
%              \cmdcite{varindexCommaLeadstrue}. Changed the default of both
%              values to `true'.\mypar
%              Introduced \cmdcite{varindexOutText...} and friends and enabled
%              a hook into the format string processing via
%              \cmdcite{varindexNewFormat}.\mypar
%              Reordered documentation such that the more interesting
%              customization parts are now mentioned earlier}
% \changes{1.21}{2001/03/29}{Divided package into two: Many macros which are
%              of independent interest have been put into the \texttt{toolbox}
%              package. For the user visible is that the macros are now called
%              e.\,g.\ \cmdcite{toolboxMakeSplit} instead of
%              \cmdcite{varindexMakeSplit}. For backward compatibility,
%              the old (now obsolete) names are still available, although not
%              documented anymore}
% \changes{1.20}{2001/01/13}{Changed \cmdcite{varindextwo}:
%              Added the \cmdcite{varindextwoCommaTilde} functionality and the
%              corresponding three flags \texttt{?!/}. Moreover, changed the
%              default to \cmdcite{varindextwoCommaTildetrue}. Note that due to
%              this change the generated index entries may differ from those
%              of earlier versions. To get a backward compatible output,
%              you have to use \cmdcite{varindextwoCommaTildefalse}.\mypar
%              Also use now the symbols \texttt{`} and \texttt{'}
%              instead of the (now obsolete) \texttt{P} and \texttt{p} flags.
%              The reason for the latter is that \texttt{P} and \texttt{p}
%              could not be used immediately after a command without a space}
% \changes{1.19}{2000/12/07}{Eliminated a bug in the \texttt{P} flag}
% \changes{1.18}{2000/11/30}{Eliminated a bug in \cmdcite{SkipTricky}}
% \changes{1.17}{2000/11/21}{\cmdcite{varindexNr??String} is now defined
%              as documented}
% \changes{1.16}{2000/11/19}{Documentation rewritten
%              for doc and docstrip utilities (required files are now
%              \texttt{varindex.dtx} and \texttt{varindex.ins})}
% \changes{1.15}{2000/10/10}{Fixed harmless bug which always caused a warning}
% \changes{1.14}{2000/06/01}{First version released into public
%              (as \texttt{varindex.sty})}
%
% \iffalse
%
% The following part is the code needed to compile the documentation.
% We do some trickery here, because we want that this file here could be used
% for *both*: Compiling the documentation *and* as `varindex.sty'
% (if really required).
% Everything between the lines %<*driver> and %</driver>
% goes to the file `varindex.tex'.
%
% \fi
\ifcat @a% If this test is positive, then `@' has letter catcode which means
%          that this file is loaded as a style file (or as described in the
%          install section of the documentation).
\def\TemporaryDummy{}
\else\long\def\TemporaryDummy{%
%
%
% \iffalse
%</metainfo>
%<*driver>
\documentclass{ltxdoc}
\OnlyDescription%% Comment this out, or
%%\AlsoImplementation%% uncomment this line, if you want to see the source.

%% If you want an index, uncomment one of the following lines.
%% After running latex a first time, you have to use
%%      makeindex -s gind.ist varindex
%% Then run latex again.
%%\CodelineIndex% comment this out if you want an index referring to codelines
%%\PageIndex% comment this out if you want an index referring to pages
%%\OldMakeindex% uncomment this line, if your MakeIndex is pre-v2.9

\setlength\hfuzz{15pt}
\hbadness=7000
\begin{document}
\DocInput{varindex.dtx}
\end{document}
%</driver>
%<*metainfo>
% \fi
}
\fi\TemporaryDummy\let\TemporaryDummy\undefined
%
% \iffalse
%          Personal macros used for the doc utility:
% \fi
% \MakeShortVerb{"}
% \def\bracemeta#1{\texttt{\{}\meta{#1}\texttt{\}}}
% \def\nocomment{}
% \def\myspace{\texttt{\textvisiblespace}}
% \parindent0pt

%
% \newenvironment{myquote}{\begin{quote}\begin{raggedright}\relax
%}{\end{raggedright}\end{quote}}
% \newenvironment{myindex}{\begin{myquote}}{\end{myquote}}
% \def\myi{\par\hangindent 40pt\relax}
% \def\mys{\myi\hspace*{20pt}}
% \def\myss{\myi\hspace*{30pt}}
% \def\pnoa#1{, #1}
% \def\pno{\pnoa{17}}
% \def\mysee#1{\pnoa{\emph{see also} #1}}
% \def\sortedlike#1{\hspace*{10pt}(\emph{sorted like} #1)}
% \def\sortedlikeq#1{\sortedlike{``#1''}}
% \def\proz{\texttt{\%}}
% \def\ppari{\proz\\\hspace*{20pt}}
% \def\pari{\\\hspace*{20pt}}
% \def\myquad{\mbox{$\qquad$}}
% \def\macrott#1{{\mdseries\cmdcite{#1}}}

% \sloppy
% \title{The \texttt{varindex} package\thanks
%   {This package has version number \fileversion{}, last revised \filedate.
%   The package may be distributed and/or modified under the conditions of the
%   LaTeX Project Public License, either version 1.3c of this license or (at
%   your option) any later version. The latest version of this license is in
%   \texttt{http://www.latex-project.org/lppl.txt}, and version 1.3c or later
%   is part of all distributions of LaTeX version 2005/12/01 or later.}}
% \author{Martin V\"ath\thanks{This package bases on ideas of
%   \texttt{indextwo.sty} which was written jointly with
%   O.~Karch \texttt{karch@informatik.uni-wuerzburg.de} and
%   H.-C.~Wirth \texttt{wirth@informatik.uni-wuerzburg.de}}\\
%       {\texttt{martin@mvath.de}}}
% \date{\docdate}
% \maketitle          

% \begin{abstract}
% This \LaTeX\ package provides a luxury front-end for the "\index" command.
% For example, it allows to generate multiple \index entries in
% almost any form by a single command. It is extremely customizable, and works
% with all versions of \LaTeX\ and probably most other \TeX\ formats, too.
% \end{abstract}

% You may copy this package freely, as long as you distribute only unmodified
% and complete versions.

% \tableofcontents

% \section{Changes}
%
% \begin{description}
%    \def\mypar{\par}\ChangesTab
% \end{description}
%

% \section{Introduction}
%
% If you have to write an index for a larger scientific text, you will probably
% sooner or later have the problem that you want to put certain texts into the
% index which consist of more than one word like \emph{Theorem of Picard}.
% \par
% There are several reasonable ways how one might do this: One might put the
% text "Theorem of Picard" in the index, or "Picard theorem".
% An alternative way is "Picard, Theorem", or "Picard's Theorem"
% or "Theorem!Picard"
% (here, "!" is the symbol used by "makeindex" to separate a
% subindex, i.\,e.\ the output will actually look like
% \begin{myindex}
% \myi Theorem
% \mys Picard\pno
% \end{myindex}
% This is useful if one also has other entries starting with \emph{Theorem}).
% Several other natural alternatives are also possible.
% \par
% One may choose some or all of the above alternatives and then code them into
% \TeX. However, if more than one of these alternatives is desired in the
% index, this requires several "\index" commands (which are similar but not
% identical). Moreover, if later another choice of the above alternatives
% is required, the index entries must be completely rewritten.
% This rewriting may be very cumbersome if one makes use of subindices
% for two reasons.
% \par
% First, it may happen that one has to change even entries which are
% apparently not affected. For example, assume there were just two subindices
% for the index entry \emph{Theorem}, namely "Theorem!Picard" and
% "Theorem!Riemann". If the theorem of Picard is eliminated from the main
% text, one not only has to delete the entry "Theorem!Picard": The single
% subindex "Theorem!Riemann" does not make much sense -- one should replace
% it by "Theorem, Riemann". One could write a program to do this task
% automatically, but my experience shows that it is better to do such things
% by hand, because there are some situations where exceptions are desirable.
% \par
% However, things become really cumbersome, if one also has to sort the
% entries in a different way than they are written. For example, suppose
% that instead of \emph{Theorem} in the above example, one has the text
% \emph{$\zeta$-function} which one wants to have sorted like
% \emph{zeta-function}. In this case, the "\index"
% command must be changed from
% \begin{verbatim}
%   \index{zeta-function@$\zeta$-function!Riemann}\end{verbatim}
%
% into
% \begin{verbatim}
%   \index{zeta-function, Riemann@$\zeta$-function, Riemann}\end{verbatim}
%
% which is rather different!
% \par
% Moreover, one may also want to have a placeholder. For example, instead of
% generating the entries
% \begin{myindex}
% \myi Theorem
% \mys Liouville\pno
% \mys Open Mapping\pno
% \mys Picard\pno
% \end{myindex}
% One might like to have entries like
% \begin{myindex}
% \myi Theorem
% \mys $\sim$ of Liouville\pno
% \mys Open Mapping $\sim$\pno
% \mys $\sim$ of Picard\pno
% \end{myindex}
% One might use the symbol "$\sim$" for the placeholder, but this needs of
% course its own sort entry in the "\index" command (in the above example,
% we sorted by \emph{Liouville}, \emph{Open Mapping}, and \emph{Picard},
% respectively, which is perhaps the most natural choice).
% \par
% With the package "varindex" it is rather simple to generate any of the
% above "\index" entries, and it is easy to modify them, e.\,g.\ the above
% placeholders can be introduced easily and then activated or deactivated by
% a switch. Moreover, several "\index" entries like "Picard, Theorem of"
% and "Theorem of Picard" can be generated simultaneously with a
% \emph{single} command (i.\,e.\ you have to write the phrase
% "Theorem of Picard" only once). In addition, it is also possible to produce
% not only corresponding "\index" commands but also to output the content
% into the running text.
% \par
% Since my experience shows that in each new book there arise new special cases
% for the "\index" command, I was trying to provide a
% \emph{highly customizable} solution which can be modified for all
% needs (I hope).

% \section{Installation}
%
% This package was tested with \LaTeX 2.09 and \LaTeXe\ and should work
% with all other (future) versions of \LaTeX\ and friends, too.
% It should even run with other (non-\LaTeX) formats if an "\index" command
% (with \LaTeX-Syntax) is provided.
%
% To use ``varindex'', you have to put the file "varindex.sty" in a path where
% \TeX\ looks for its input files. You must also have the file "toolbox.sty"
% of the ``toolbox'' package in your path (toolbox~v3.1 or newer is required).
% The \TeX\ documents using  varindex need the
% following modifications in their header:
%
% \begin{itemize}
% \item If you use \LaTeXe, put in the preamble the command
%   \begin{verbatim}
%     \usepackage{varindex}\end{verbatim}
%
% \item If you use \LaTeX 2.09, use "varindex" as a style option, e.\,g.\
%   \begin{verbatim}
%     \documentstyle[varindex]{article}\end{verbatim}
%    or
%   \begin{verbatim}
%     \documentstyle[varindex,12pt]{article}\end{verbatim}
%
% \item If you use some other (non-\LaTeX) format, you will probably have to
%    insert a line like
% \begin{verbatim}
% \catcode`\@=11\relax\input varindex.sty\catcode`\@=12\relax\end{verbatim}
%
% \end{itemize}
%
% For \TeX\ insiders: \LaTeX-specific commands used in "varindex.sty" are only:
%
% \begin{itemize}
% \item "\makeatother"
% \item "\makeatletter"
% \item "\typeout"
% \item "\RequirePackage"
% \item "\newcommand" (used only in the form "\newcommand{"\meta{command}"}{}"
%       to ensure that \meta{command} was not defined before)
% \item "\ProvidesPackage"
% \end{itemize}
% The above commands are used only if they are defined (otherwise,
% natural substitutes are used ("\newcommand" and "\ProvidesPackage" are
% then not used at all)).\bigskip

% \section{Additional hint}
%
% Although this package provides a convenient way to write "\index" commands,
% this may not be sufficient: Also with this package, the "\index" commands
% may still be rather complex. For some books, it might be necessary to write
% the same "\index" entries several times (to get various page numbers).
% The first idea that one might have in this case is to define a list of the
% used "\index" commands near the beginning of the document e.\,g.\ as:
% \begin{verbatim}
% \newcommand{\Index}{\varindex(){\varindextwoScan}{\varindextwo}[]}
% \newcommand{\Riemann}{\Index{Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}
% \newcommand{\sigmaa}{\Index-={$\sigma$@sigma algebra}}\end{verbatim}
% (the first line and the usage of "\Index" is explained in later sections;
% here, it is sufficient to know that "\Index" produces in the above form
% several "\index" entries).
% To produce the corresponding "\index" entries in the
% running text, one then just needs to use "\Riemann" resp.\ "\sigmaa".
% However, this method has some disadvantages:
% \begin{enumerate}
% \item If one has many "\index" entries, it is easy to forget that e.\,g.\
%       "\sigmaa" is a command which should produce an index entry.
%       Then "\sigmaa" in the main text might be rather confusing.
% \item One has to take care of macros already provided by \TeX, \LaTeX\ or
%       some packages. For example, it is not possible to use the name
%       "\sigma" for the above purpose.
% \end{enumerate}
% To avoid these problems, one may be very disciplinary and call the
% involved macros systematically e.\,g.\ "\IndexRiemann" "\Indexsigmaa" etc.
% However, this produces terrible long and unreadable macro names in the main
% text.\par
% The "toolbox" package (version 3.1 or newer) provides a more convenient
% solution: You can use the command
% \begin{verbatim}
%   \newcommand{\Index}{\varindex(){\varindextwoScan}{\varindextwo}[]}
%   \toolboxMakeDef{Idx}{#1}\end{verbatim}
% and afterwards
% \begin{verbatim}
%   \NewIdx{R-S}{\Index{Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}
%   \NewIdx{sigma-algebra}{\Index-={$\sigma$@sigma algebra}}\end{verbatim}
% After these definitions, you can use the intuitive commands
% "\Idx{R-S}" and "\Idx{sigma-algebra}" to produce the corresponding
% index entries.\par
% Note also the symbol ``-'' in the above names which is usually not allowed
% in \TeX\ macros.\par
%
% If several page numbers occur for one index entry, it may be convenient for
% the reader if the ``main page number(s)'' (e.\,g.\ the place(s)
% where the corresponding notion is defined) is written in a different style.
% Of course, this is supported by "\varindex" (as explained later). To use this
% feature in connection with "\toolboxMakeDef", there are several
% possibilities. A straightforward way is to add in addition to the above
% commands also
% \begin{verbatim}
%   \NewIdx{R-S*}{\Index[|textbf]%
%     {Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}
%   \NewIdx{sigma-algebra*}{\Index[|textbf]%
%     -={$\sigma$@sigma algebra}}\end{verbatim}
% Then "\Idx{R-S*}" and "\Idx{sigma-algebra*}" produce index entries
% with boldface page numbers. Of course, you can automatize the task of
% generating macros which produce normal and boldface page numbers
% e.\,g.\ as follows:
% \begin{verbatim}
%   \newcommand{\Index}{\varindex(){\varindextwoScan}{\varindextwo}[]}
%   \toolboxMakeDef{Idx}{#1}
%   \newcommand{\MakeIdx}[2]{%
%     \NewIdx{#1}{\Index#2}%
%     \NewIdx{#1*}{\Index[|textbf]#2}}
%
%   \MakeIdx{R-S}{{Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}
%   \MakeIdx{sigma-algebra}{-={$\sigma$@sigma algebra}}\end{verbatim}
% This provides the commands "\Idx{R-S}" "\Idx{R-S*}"
% "\Idx{sigma-algebra}" and "\Idx{sigma-algebra*}" with the same functionality
% as explained before.\par
%
% The above approach has the disadvantage that the names ``$\ldots$"*"'' have
% automatically a fixed meaning. Alternatively, you can also generate different
% main commands. This is immediately supported by "\toolboxMakDef":
% \begin{verbatim}
%   \newcommand{\Index}{\varindex(){\varindextwoScan}{\varindextwo}[]}
%   \toolboxMakeDef{Idx}{\Index#1}
%   \toolboxMakeDef{IdxMain}{\Index[|textbf]#1}
%
%   \NewIdx{R-S}{{Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}
%   \NewIdxMain{R-S}{%
%     {Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}\end{verbatim}
% Now you can use "\Idx{R-S}" and "\IdxMain{R-S}" to produce the
% normal respectively boldface page numbers. Of course, you can also automatize
% in this case the task of generating these commands:
% \begin{verbatim}
%   \newcommand{\Index}{\varindex(){\varindextwoScan}{\varindextwo}[]}
%   \toolboxMakeDef{Idx}{\Index#1}
%   \toolboxMakeDef{IdxMain}{\Index[|textbf]#1}
%   \newcommand{\NewIdxBoth}[2]{%
%     \NewIdx{#1}{#2}%
%     \NewIdxMain{#1}{#2}}
%
%   \NewIdxBoth{R-S}{{Riemann Stieltjes integral}[1-23 2,1-~3 3!1-2~]}
%   \NewIdxBoth{sigma-algebra}{-={$\sigma$@sigma algebra}}\end{verbatim}
% After the above commands, you can use "\Idx{R-S}", "\IdxMain{R-S}",
% "\Idx{sigma-algebra}" and "\IdxMain{sigma-algebra}" with their obvious
% meaning.\par\bigskip
%
% \textbf{Summarizing}: The last of the above solutions is the one which
% I recommend.\par\bigskip
%
% For further possibilities of the "\toolboxMakeDef" command (like making
% copies of entries, redefining entries, etc.), please read the
% original documentation of the "toolbox" package.

% \section{Examples}
%
% Probably you are very impatient and want to see some examples.
% So, here you are -- but be warned that the examples are not explained in
% detail. You have to read the later sections to understand more precisely
% why the examples work the way they do.
% \par
% Since the "\varindex" command is highly customizable,
% the following examples can only give you a rough impression of what
% you can actually do with it.
%
% \subsection{Typical example of usage (using the \texttt{varindextwo} macros)}
% \label{s:firstex}
%
% Suppose the following customization of the "\varindex" command was defined:
% \begin{verbatim}
% \def\Index{\varindex(){\varindextwoScan}{\varindextwo}[\emph]}\end{verbatim}
%
% Since we used the "varindextwo" macros here, by default
% \emph{two} "\index" entries are generated with a single command.
% After the above definition, the command
% \begin{verbatim}
%   \Index{ring with $1$@one}\end{verbatim}
%
% will produces the "\index" entries for:
%
% \begin{myindex}
% \myi ring
% \mys $\sim$ with $1$\pno\sortedlikeq{with one}
% \myi $1$\sortedlikeq{one}
% \mys ring with $\sim$\pno\sortedlike
%     {``ring with'' (without \mbox{``$\sim$''!})}
% \end{myindex}
%
% One of the examples of the introduction can be generated with
% \begin{verbatim}
%   \Index*{Theorem of@ Picard}\end{verbatim}
% which produces the two "\index" entries for
%
% \begin{myindex}
% \myi Theorem
% \mys $\sim$ of Picard\pno\sortedlikeq{Picard}
% \myi Picard
% \mys Theorem of $\sim$\pno\sortedlikeq{Theorem}
% \end{myindex}
% In this example, the word "of" is ignored for the sorting, because we put
% the symbol "@"\meta{nothing} behind it which means that it should be sorted
% as the empty string (in cases like this, also the space in front of "Picard"
% is automatically ignored for the sort entry).
%
% Since we used in the above example the form "\Index*" instead of "\Index",
% additionally the text "\emph{Theorem of Picard}" is output into the running
% text (the string "\emph" stems from our customization of "\Index" defined
% in the beginning).
% \bigskip
%
% \emph{Completely} different entries may be generated by your own rules.
% To this end, an optional parameter "["\meta{format}"]" can be appended.
% Internally, the main argument is splitted into words which are numbered
% $1,2,\ldots$; in \meta{format}, you can refer to these words simply
% by using the corresponding numbers. An ``empty'' entry in \meta{format}
% means all words in the original order (i.\,e.\ it is essentially equivalent
% to $12345\ldots$). Some other symbols besides numbers are also allowed
% which are listed later.
%
% \begin{verbatim*}
%\Index{Hausdorff measure of noncompactness}[4,23 2!~34!1= ]\end{verbatim*}
% generates \emph{three} index entries:
%
% \begin{myindex}
% \myi noncompactness, measure of\pno
% \myi measure
% \mys $\sim$ of noncompactness
% \myss Hausdorff $\approx$\pno
% \myi Hausdorff measure of noncompactness\pno
% \end{myindex}
%
% The last of these entries occurs only, because the last character in
% the \meta{format} argument "[4,23"\myspace"2!~34!1="\myspace"]" is a space
% (the space following "=" tells "varindex" that another (empty) entry is
% desired).
% \par
%
% If you want only slight modification(s) of the default, you need also just
% slight modification(s) of the command, e.\,g.
% \begin{verbatim}
% \Index*,{ring with $1$@one}\end{verbatim}
%
% has the same effect as in our earlier example,
% just that the second entry reads
%
% \begin{myindex}
% \myi $1$, ring with $\sim$\pno\sortedlikeq{one, ring with}
% \end{myindex}
% Moreover, since the `"*"' occurred in the previous command, additionally
%
% \begin{verbatim}
%   \emph{ring with $1$}\end{verbatim}
%
% appears in the running text. Praxis shows that changes in the entries as
% above have to be made frequently in the `fine tuning' of the index of a
% book. Note that with the original "\index", the command would have to be
% changed completely for the above entries, because we have a ``sort'' entry
% for which a different rule applies if we use a subitem (``"!"'')
% instead of a ``","\myspace''.
% (BTW: "\varindex" produces sort entries only if they are necessary).
% \par
%
% Of course, you may combine the default with your own format:
% \begin{verbatim*}\Index{internal integrable function}[23,1~ +]\end{verbatim*}
%
% generates
% \begin{myindex}
% \myi internal function, integrable $\sim$\pno
% \myi internal
% \mys $\sim$ integrable function\pno
% \myi function
% \mys internal integrable $\sim$\pno
% \end{myindex}
% In this example, the first entry is generated by the symbols "23,1~" in the
% \meta{format} argument, and the last two entries are those entries which
% would have been generated by the "varindextwo" macros if no \meta{format}
% argument had been given -- the magic symbol "+" instructs the "varindextwo"
% macros to not suppress these entries.
%
%
% \subsection{A simple example without the \texttt{varindextwo} macros}
%
% As known from "\index", a "|see{...}" can be used to produce a reference to
% another index entry instead of a page number. Such a command may optionally
% be appended.
% \begin{verbatim*}
%\varindex*{topology of a normed space}{5!4~!1= 45,12}[|see{norm}]
%\end{verbatim*}
% \begin{myindex}
% \myi space
% \mys normed $\sim$
% \myss topology $\approx$\mysee{norm}
% \myi normed space, topology of\mysee{norm}
% \end{myindex}
% (the precise appearance of ``\emph{see also} '' depends on how the "\see"
% macro is defined in your style). The above command additionally inserts the
% tokens "{topology of a normed space}" into the running text (this would not
% happen if "\varindex" is used in place of "\varindex*").
% Only in the first of the above entries, an `"@"' part will be added to the
% "*.idx" file: This entry is sorted as if ``\myspace$\sim$'' resp.\
% ``\myspace$\approx$'' would not exist.
%
%
% \subsection{Another example without the \texttt{varindextwo} macros}
%
% The appearance of the cited page number and the appearance in the
% running text (only available with ``"*"'') can be customized easily.
%
% \begin{verbatim}
% \varindex[\emph][|textbf]%
%     {$\sigma$@sigma!$(\sigma-)$ finite {measure space}}%
%     {*1.23 23 23,1- 1-23}\end{verbatim}
%
% generates the three index entries
% \begin{myindex}
% \myi finite measure space\pnoa{\textbf{42}}
% \myi finite measure space, $\sigma$-\pnoa{\textbf{42}}\sortedlikeq
%   {finite measure space, \mbox{sigma-}}
% \myi $\sigma$-finite measure space\pnoa{\textbf{42}}\sortedlikeq
%   {sigma-finite measure space}
% \end{myindex}
% In all three entries the pages are printed using "\textbf".
% Moreover, the tokens
% \begin{verbatim}
%   \emph{($\sigma$-)finite measure space}\end{verbatim}
%
% are put into the running text.
%
% \subsection{A simple example using a flag
%    (without the \texttt{varindextwo} macros)}
%
% Note the token "1" in the following command:
% \begin{verbatim}
%  \varindex1{$L$ and $M$@L and M}{}\end{verbatim}
%
% This command generates the index entry
% \begin{myindex}
%   \myi $L$ and $M$\pno\sortedlikeq{L and M}
% \end{myindex}
% With the command
% \begin{verbatim}
%   \varindex1*{$L$ and $M$@L and M}{}\end{verbatim}
% or
% \begin{verbatim}
%   \varindex*1{$L$ and $M$@L and M}{}\end{verbatim}
%
% additionally, "{$L$ and $M$}" is output into the running text.
%
% Note that without the `"1"', the index generated were
% \begin{myindex}
%   \myi "$L$ and $M$ and M"\pno\sortedlikeq{"\$L\$ and L and M"}
% \end{myindex}
%
% \subsection{An example for very primitive customized index commands
%   (without the \texttt{varindextwo} macros)}
%
% If you defined your ``default'' shortcuts appropriately, you can still
% override some of your own defaults.
%
% \begin{verbatim}
% \def\myindexA{\varindex[\emph][|textbf]*}
% \def\myindexB{\varindex(){}{}[\emph][|textbf]}
%
% \myindexA{Foo1}{}
% \myindexB{Foo2}[][]
% \myindexB*{Foo3!{Foo 3}}\end{verbatim}
%
% Generates the index entries "Foo1", "Foo2", and "Foo3". Moreover,
% "\emph{Foo1}" and "\emph{Foo 3}" is put into the running text.
% The page number of "Foo 2" in the index is printed normal (i.\,e.\ "|textbf"
% is overridden), the others with "\textbf". Note that by using braces,
% it was possible to include a space into the text "Foo 3".
%
%
% \subsection{An example of a primitive customized index command
%   (without the \texttt{varindextwo} macros)}\label{s:myindex}
%
% \TeX\ code can be inserted to customize the default formatting string.
% \begin{verbatim}
% \def\myindex{\varindex(\ifnum\varindexCount=2 1!2 2,1\fi){}{}[\emph]}
%
% \myindex{Foo Bar}\end{verbatim}
%
%    generates the index entries
% \begin{myindex}
% \myi Foo
% \mys Bar\pno
% \myi Bar, Foo\pno
% \end{myindex}
%
% \begin{verbatim}
% \myindex*{Foo Bar}\end{verbatim}
%
% generates the same index entries as above and additionally outputs
% "\emph{Foo Bar}" into the running text.
%
% \begin{verbatim}
% \myindex*{other format}[21]\end{verbatim}
%
% generates the index entry
% \begin{myindex}
% \myi format other\pno
% \end{myindex}
% and outputs "\emph{other format}" into the running text.
%
% \begin{verbatim}
% \myindex[|textbf]*{BoBo}\end{verbatim}
%
% generates the index entry
% \begin{myindex}
% \myi BoBo\pnoa{\textbf{42}}
% \end{myindex}
%  where the page number is printed with "\textbf".
%  Moreover, "\emph{BoBo}" is output into the running text.
%
% \subsection{A primitive varying customized index command
%     (without the \texttt{varindextwo} macros)}
%
% \begin{verbatim}
% \def\myindex{\varindex(\ifnum\varindexCount=2 1!2 2,1\fi)%
%     (\ifnum\varindexCount=2*2,1 1!2 2,1\fi){}{}[\emph]}\end{verbatim}
%
% With the same commands as in Section~\ref{s:myindex}, the same index entries
% are generated, but the tokens "\emph{Far, Boo}", "\emph{other format}",
% and "\emph{Bobo}" are output into the running text.
%
%
% \subsection{Examples with the \texttt{varindextwo} macros}
%
% As mentioned earlier, if the "varindextwo" macros are used (and if
% the main arguments contains at least two entries), then two
% "\index" entries are generated by default. Roughly speaking, one "\index"
% entry is the argument in its given order and the other is the argument in
% reverse order (this is not precise, but explains the idea).\par
%
% For the following examples,
% we use a similar definition for "\Index" as in Section~\ref{s:firstex},
% and define an "\iemph" macro which has `"*"' as a default and outputs with
% "\emph". I recommend to put both sort of definitions into the preamble of
% all of your files which should use "varindex" (the names are of course
% subject to your personal taste).
%
% \begin{verbatim}
% \def\Index{\varindex(){\varindextwoScan}{\varindextwo}[]}
% \def\iemph{\varindex(){\varindextwoScan\def\varindexStarPos{}}%
%              {\varindextwo}[\emph]}\end{verbatim}
%
% (note that the apparently simpler definition
% \begin{verbatim}
% \def\iemph{\varindex(){\varindextwoScan}{\varindextwo}[\emph]*}\end{verbatim}
%  is not as good as the above definition of "\iemph" as is explained below).
%
% After the above customization, you may use the following commands.
% \begin{verbatim}
% \Index*{measure of $\gamma$-noncompactness@gamma-noncompactness!FOO}
% \end{verbatim}
% generates the index entries
% \begin{myindex}
% \myi measure
% \mys $\sim$ of $\gamma$-noncompactness\pno\sortedlikeq
%        {of gamma-noncompactness}
% \myi $\gamma$-noncompactness\sortedlike{gamma-noncompactness}
% \mys measure of $\sim$\pno
% \end{myindex}
% and outputs the tokens "{measure of FOO}" into the running text.
% Note the order of the words!
%
% \begin{verbatim}
%   \Index*,?_'{flic of flac}\end{verbatim}
%
% generates
% \begin{myindex}
% \myi flic of flac\pno
% \myi flac, flic\pno
% \end{myindex}
% and outputs "{flic of flac}". Here, the two tokens "_" and "," had the
% effect that a space resp.\ ``","\myspace'' was used instead of a subentry
% in the first resp.\ second "\index" entry; the token "'" caused the
% preposition "of" to disappear in the second entry, and the token "?"
% suppressed a trailing $\sim$ in the second entry.\par
%
% A more realistic example is
% \begin{verbatim}
%   \Index^={$\sigma$@sigma algebra}\end{verbatim}
%
% which in both entries inserts a ``-'':
%
% \begin{myindex}
% \myi $\sigma$\sortedlikeq{sigma}
% \mys $\sim$-algebra\sortedlike{``algebra'' or ``-algebra'', see below}
% \myi algebra
% \mys $\sigma$-$\sim$\pno\sortedlikeq{sigma-}
% \end{myindex}
% Here, the sorting ``-algebra'' is chosen when you used
% \begin{verbatim}
%   \let\varindexOutSortDashBeg\varindexOutSortDash\end{verbatim}
% as described later.\par
%
% The flag "^" in conjunction with "-" has a special meaning:
% \begin{verbatim}
%   \Index^->{$\sigma$@sigma algebra}\end{verbatim}
% Then the first entry reads instead (the second is canceled because of ">"):
%
% \begin{myindex}
% \myi $\sigma$-\sortedlikeq{sigma-}
% \mys $\sim$ algebra\pno\sortedlikeq{algebra}
% \end{myindex}
%
% Similarly, the conjunction "^" and "." has a special meaning:
% \begin{verbatim}
%   \Index^.,:{ultra filter}\end{verbatim}
%
% generates entries without a space following $\sim$ (and in view of the flag
% "," no new subentry is used for the second entry but instead a ``,''):
%
% \begin{myindex}
% \myi ultra
% \mys $\sim$filter\pno\sortedlike{filter}
% \myi filter, ultra$\sim$\pno\sortedlike{filter, ultra}
% \end{myindex}
%
% For crossreferences, one may use something like
% \begin{verbatim}
%   \iemph[|see{recursive recursion}],_{recursive recursion}\end{verbatim}
%
% which generates
% \begin{myindex}
% \myi recursive recursion\mysee{recursive recursion}
% \myi recursion, recursive $\sim$\mysee{recursive recursion}
% \end{myindex}
% and outputs "\emph{recursive recursion}". If we would have used instead the
% definition
% \begin{verbatim}
%   \def\iemph{\varindex(){\varindextwoScan}{\varindextwo}[]*}\end{verbatim}
%
% the above call were not possible, since the optional arguments "[|see...]"
%  must occur in front of the flag `"*"'.
%
% \begin{verbatim}
%   \Index-;*{Flip Flop}\end{verbatim}
%
% generates
% \begin{myindex}
% \myi Flip-Flop\pno
% \myi FlopFlip\pno
% \end{myindex}
% and outputs "{Flip-Flop}" (note that the symbol "-" which is caused by the
% flag "-" occurs also in the output).
% 
% One may freely create other entries:
% \begin{verbatim}
%   \Index>{Flippy Flop!GO}[*2-1-2 2!1-1 1-12 +]\end{verbatim}
%
% generates
% \begin{myindex}
% \myi Flop
% \mys Flippy-Flippy\pno\qquad\emph{(no ``$\sim$''!)}
% \myi Flippy-Flippy Flop\pno
% \myi Flippy
% \mys $\sim$ Flop\pno
% \end{myindex}
% and outputs "{GO-Flippy-GO}" (note that even without an explicit "*" flag,
% we were able to generate this output, just by using the "*" in
% \meta{format}).
%
% \begin{verbatim}
%   \Index*_,{shape {of a} of star}\end{verbatim}
%
% generates
% \begin{myindex}
% \myi shape of a star\pno
% \myi star, shape of $\sim$\pno\sortedlike{star, shape of}
% \end{myindex}
% and outputs "{shape of a star}". Note that for the first entry the
% proposition ``"of a"'' was chosen while for the second entry the
% alternative proposition "of" was chosen.
%
% Also with the "varindextwo" macros the other flags keep their old meaning.
% For example, the flag "1" still means that the main argument is
% considered as a single entry.
% \begin{verbatim}
%   \Index1*{$L$ feature@L feature}
%   \Index,{No $L$@L feature}\end{verbatim}
%
% generate
% \begin{myindex}
% \myi $L$ feature\pno\sortedlikeq{"L feature"}
% \myi No
% \mys $\sim$ $L$ feature\pno\sortedlikeq{"L feature"}
% \myi feature, No $L$ $\sim$\pno\sortedlikeq{"feature, No L"}
% \end{myindex}
%  and outputs "{$L$ feature}".
%
%
% \subsection{Example of a simple modification of the
%     \texttt{varindextwo} macros}\label{ss:expatch}
%
% All flags and magic tokens in "varindextwo" can be customized to fit your
% own taste, just by modifying "\varindextwoScan".
% 
% \begin{verbatim}
% \def\myindextwoScan{%
%      \varindextwoScan
%      \varindexNewFlag ~\vxtSpaceA
%      \varindexNewFlag 1\vxtSpaceB
%      \varindexNewFlag !\varindexOneEntry
%      \varindexMakeSplitExpand{/}{vxtSplitAtMagic}%
%      \def\varindexStarPos{}%
% }
% \def\myIndex{\varindex(){\myindextwoScan}{\varindextwo}[]}\end{verbatim}
%
%
% "\myIndex" behaves similar to the previous example with
% the following differences:
% \begin{enumerate}
% \item The flag "1" now has the previous meaning of "~".
% \item One may now use "~" and "_" equivalently.
% \item The flag "!" now has the original meaning of the flag "1".
% \item Instead of "+" in the format string, the character "/" has to be used.
% \end{enumerate}

%
% \section{Main Description}
%
% Without additional customization, there are two possible calls for the
% \DescribeMacro{\varindex}"\varindex" command:
%
% \begin{enumerate}
% \item[a)] The first call has the form
% \begin{myquote}
% "\varindex["\meta{textstyle}"]["\meta{pagestyleA}"]"\meta{flags}\ppari
% \bracemeta{main entries}\bracemeta{format}"["\meta{pagestyleB}"]"
% \end{myquote}
% Here, "["\meta{textstyle}"]", "["\meta{pagestyleA}"]",
% "["\meta{pagestyleB}"]", and \meta{flags} are optional.
% (But if "["\meta{textstyle}"]" is omitted, also "["\meta{pagestyleA}"]"
% must be omitted).
%
% The meaning of the arguments is as follows:
% \begin{description}
% \item[\mdseries\meta{textstyle}] describes the textstyle used for the output
%     into the running text (typically, textstyle is "\emph" or empty).
% \item[\mdseries\meta{pagestyle}] describes the pagestyle used for the output
%     of the page number (you may also use a construct like "|see{...}" here).
%     If "["\meta{pagestyleB}"]" is present, this is the value used.
%     Otherwise, the value of "["\meta{pagestyleA}"]" is used
%     (resp.\ nothing).
% \item[\mdseries\meta{flags}] can be one (or both) of the following tokens:
%   \begin{description}
%   \item[\mdseries "*"] If this token appears in \meta{flag}, it has roughly
%              the meaning ``Output into running text''. More precisely, if no
%              "*" is occurs in \meta{format} (see below), then \meta{format}
%              is automatically prepended by the tokens ``"*"\myspace'' resp.\
%              ``"*"'' (depending whether \meta{format} starts with a space
%              or not).
%   \item[\mdseries "1"] With this flag, the content of \meta{main entries} is
%              considered as a single entry (see below).
%   \end{description}
% \item[\mdseries\meta{main entries}] and
% \item[\mdseries\meta{format}] are explained later.
% \end{description}
%
% \item[b)] The alternative form to call "\varindex" is
% \begin{myquote}
%   "\varindex("\meta{formatA}")("\meta{format*A}")"\relax
%   \bracemeta{scan program}\ppari
%   \bracemeta{main program}"["\meta{textstyle}"]["\meta{pagestyleA}"]"\relax
%\meta{flags}\ppari
%   \bracemeta{main entries}"["\meta{formatB}"]["\meta{pagestyleB}"]"
% \end{myquote}
% The arguments "["\meta{textstyle}"]", "["\meta{pagestyleA}"]", \meta{flags},
% and "["\meta{pagestyleB}"]" have the same meaning as in a) (in particular,
% they are optional, but if "["\meta{textstyle}"]" is omitted, also
% "["\meta{pagestyleA}"]" must be omitted).
%
% Also "("\meta{format*A}")" and "("\meta{formatA}")" are optional,
% but if "("\meta{formatA}")" is omitted, also "("\meta{format*A}")"
% must be omitted.
%
% The ``default'' value for \meta{format} is \meta{formatA} resp.\
% \meta{format*A} (depending whether the flag "*" has been used or not).
% If "["\meta{formatB}"]" is given, then this is used as the format (i.\,e.\
% the optional argument \meta{formatB} can be used to override the default).
%
% The other values are:
% \begin{description}
% \item[\mdseries\meta{scan program}] is executed (i.\,e.\ expanded)
%    immediately when "\varindex" reads it. This parameter can be used
%    to introduce \emph{further} flags (other than "*" and "1"), see below.
% \item[\mdseries\meta{main program}] is executed \emph{after} the
%    \meta{format} string(s)  has been expanded.
% \end{description}
% \end{enumerate}
%
% It is recommended to define personal macros which contain all parameters up
% to the place where \meta{flags} occurs (either with or without some flags,
% depending on the intention). See the examples section.
% \bigskip\par
%
% \meta{main entries} is the main argument which has the following form
%
% \begin{myquote}
% \meta{entry1}[\myspace\meta{entry2}][\myspace\meta{entry3}]$\ldots$
% \end{myquote}
%
% where each of \meta{entry1} \meta{entry2} \meta{entry3} $\ldots$
% in turn has the form
% \begin{myquote}
%   \meta{indextext}"@"\meta{sorttext}"!"\meta{outputtext}
% \end{myquote}
% or
% \begin{myquote}
%   \meta{indextext}"!"\meta{outputtext}"@"\meta{sorttext}
% \end{myquote}
% In each of these forms, "@"\meta{sorttext} and "!"\meta{outputtext}
% are optional. Here,
% \begin{description}
% \item[\mdseries\meta{indextext}]
%     is the text which is put into the index for the entry
% \item[\mdseries\meta{sorttext}]
%     is what is used as a sort criterion for the entry
% \item[\mdseries\meta{outputtext}]
%     is what is output into the running text for the entry
% \end{description}
%
% Note that the symbol "@" has in a sense the opposite meaning as in the
% original "\index" command:
% \emph{Before} that symbol, the desired text appears, and
% \emph{after} the symbol the place where it has to be sorted.
% However, we chose the symbols "@" and "!", because these two
% symbols are forbidden anyway (otherwise, "makeindex" will become confused).
%
% As usual, multiple spaces are considered as \emph{one} space and do not
% generate empty ``ghost'' entries. Moreover, a space after a \TeX-command
% like "\LaTeX" is eliminated by \TeX\ and thus also not considered as a
% separator. You may use braces"{"$\ldots$"}" either around a whole entry or
% around \meta{indextext} or \meta{sorttext} or \meta{outputtext} to allow
% spaces which do not act as ``separators'' in the corresponding part.
% The braces around these parts will vanish. In particular, you may generate an
% empty entry with "{}" or "!" or "@" (bordered by spaces).
% If you want that braces occur in the \emph{output}, you have to add an
% additional pair of braces. Be aware that you write sufficiently
% many braces, if you really should need them: For example,
% the command "\varindex{{{{A"\myspace"B}}}}{}" produces the index entry
% "{A"\myspace"B}": The outermost pair of braces is eliminated immediately
% by \TeX. The second pair is eliminated, because this is a pair around a
% complete entry in \meta{main entries}. Finally, another pair is eliminated,
% because it is a brace around \meta{indextext}.
% With the flag `"1"', \meta{main entries} is considered as one single entry.
% Nevertheless, also with this flag, an additional pair of braces around
% \meta{main entry} is eliminated.\bigskip
%
%
% The \meta{format} has one of the following three forms:
% \begin{myquote}
% \meta{IndexA}[\myspace\meta{IndexB}][\myspace\meta{IndexC}]$\ldots$\relax
%["*"\meta{OutputA}]["*"\meta{OutputB}]$\ldots$\\\relax
%["*"\meta{OutputA}][\myspace\meta{IndexA}][\myspace
%\meta{IndexB}]["*"\meta{OutputC}]$\ldots$\\
%  "*"
% \end{myquote}
% where the order of arguments can be arbitrarily mixed (except for the first
% \meta{IndexA} in the first form). \meta{IndexA} \meta{IndexB} $\ldots$
% \meta{OutputA} $\ldots$ describe the format of the index entries resp.\
% of the output into the running text.
% The output is generated in the given order (this might be important, if
% a pagebreak occurs during the output into the running text).
% The last case is exceptional and equivalent to ``"*"\myspace''.
% The following characters/strings are allowed in the format entries:
% \par\bigskip
%
% \begin{tabular}[t]{ll}
% token   &meaning\\\hline
% "1"--"9"&\meta{entry1}--\meta{entry9}\\
% "0"     &\meta{entry10}\\
% "~"     &a placeholder sign (default is $\sim$)\\
% "="     &a secondary placeholder sign (default is $\approx$)\\
% "_"     &the space character ``\myspace''\\
% "s"     &The token ``"\space"''
%          (but ``\myspace'' is used for the sort entry)\\
% "."     &No space will follow (see below)\\
% ","     &The character ``","'' (space will follow)\\
% "-"     &The character ``"-"'' (no space will follow)\\
% "!"     &Create a new sublevel (subitem).
% \end{tabular}\par\bigskip
%
% All other tokens in this argument are forbidden. (Note that the magic
% symbol "+" in \meta{format} is handled by the "varindextwo" macros, and
% not by "varindex").\par
% The token ``"!"'' above makes no sense for the output in the running text
% and thus is forbidden behind "*".\par
%
% By a heuristic rule, a space is automatically inserted between two entries
% which generate non-empty text. If the heuristic rule fails, you may always
% force a space by either ``"_"'' or ``"s"'', or forbid a space by ``"."''.
%
% If a format is empty, all entries are taken in the order of input.
% Note that \TeX\ eliminates concatenated spaces, and so you are able to
% produce an empty format entry only at the end of \meta{format} or in front
% of a ``"*"''. If you want to force an empty \emph{output}
% (is this ever useful?), you may use just ``"."'' as the entry.\\
%
% A sort entry is only written to the "*.idx" file if it differs from the
% text entry.

%
% \section{Description of the \texttt{varindextwo} macros}
%
% The macros
% \DescribeMacro{\varindextwoScan}
% \DescribeMacro{\varindextwo}
% \begin{verbatim}
%   \varindextwoScan \varindextwo\end{verbatim}
%
% can be used together as parameters \meta{scan program} resp.\
% \meta{main program} for the "\varindex" command.
% If "\varindex" is called with these macros, and no
% explicit format argument is given, these macros generate a format
% depending on the number of entries in \meta{main entries}:
%
% \begin{enumerate}
% \item[a)] If there is only one entry in \meta{main entries}, then the format
%    ``"1"'' resp.\ ``"*1"\myspace"1"'' is
%    generated (depending whether the "*" flag was used or not), i.\,e.\ the
%    entry is put into the index, and printed if "\varindex" was
%    called with the "*" flag.
%
% \item[b)] If there are two entries in \meta{main entries}, then
%    \meta{format} becomes ``\myspace"1!~2"\myspace"1!2~"''. For example, if
%    \meta{main entries} is ``"graulty bazola"'', then the "\index" entries
%
%   \begin{myindex}
%   \myi graulty
%   \mys $\sim$ bazola\pno
%   \myi bazola
%   \mys graulty $\sim$\pno
%   \end{myindex}
%
%   are produced.\par
%   You can modify the first format entry with the following flags:\par
%   \bigskip
%
%   \begin{tabular}[t]{lll}
%   flags&format used&with ``"*"'' additionally\\\hline
%   "_" &"12"                   &"*12"\\
%   "-" &"1-2"                  &"*1-2"\\
%   "^" &"1!~-2"  resp.\ "1!-2" &"*1-2"\\
%   "-^"&"1-!~.2" resp.\ "1-!~2"&"*1-2"\\
%   "." &"1.2"                  &"*1.2"\\
%   ".^"&"1!~.2"                &"*1.2"\\
%   "/" &"1!2" (without "~")    &"*12"\\
%   "<" &no entry               &"*12"
%   \end{tabular}\par\bigskip
%
%   Here, the combinations "-^" and ".^" mean that both flags are used
%   (the order plays no role).
%   The output for "^" respectively "-^" is determined by the switches
%   \begin{myquote}
%   \DescribeMacro{\varindextwoDashTildetrue}"\varindextwoDashTildetrue"
%   (default)\par
%   \DescribeMacro{\varindextwoDashTildefalse}"\varindextwoDashTildefalse"
%   \end{myquote}
%   and
%   \begin{myquote}
%   \DescribeMacro{\varindextwoDashSpacetrue}"\varindextwoDashSpacetrue"\\
%   \DescribeMacro{\varindextwoDashSpacefalse}"\varindextwoDashSpacefalse"
%   (default)
%   \end{myquote}
%   respectively. The last entry in the above table is an additional format
%   entry which is generated if also the ``"*"'' flag is used.\par
%
%   You can modify the second format entry with the following flags:\par
%   \bigskip
%
%   \begin{tabular}[t]{ll}
%   flags&format used\\\hline
%   "," &"2,1~" resp.\ "2,1"\\
%   "~" &"21"\\
%   "=" &"2!1-~"\\
%   ",="&"2,1-~" resp.\ "2,1-"\\
%   "+" &"2-1"\\
%   ":" &"2!1.~"\\
%   ",:"&"2,1.~" resp.\ "2,1"\\
%   ";" &"2.1"\\
%   ">" &no entry\\
%   "!" &Append "~" (if not already there)\\
%   "?" &Without trailing "~"
%   \end{tabular}\par\bigskip
%
%   Whether the first or the second alternatives in the above forms is used
%   depends on the status of the switch
%   \begin{myquote}
%   \DescribeMacro{\varindextwoCommaTildetrue}"\varindextwoCommaTildetrue"
%   (default)\\
%   \DescribeMacro{\varindextwoCommaTildefalse}"\varindextwoCommaTildefalse"
%   \end{myquote}
%   We point out that "\varindextwoCommaTildefalse" was in earlier versions of
%   "varindex" the default (and could not be changed).
%   Note that this switch has no effect if the ! or ? flags are used.\par
%
%   \textbf{Hint} for remembering the symbols: The symbols "_" "." "-"
%   for the first entry are the same flags as for the output and the same
%   flags which are used in the \meta{format} argument.
%   The corresponding symbols "~" ":" (and ";") "=" (and "+") for the second
%   entry ``look similar''. The "," flag is only useful in the second entry
%   (and is the same symbol as in the \meta{format} argument).
%   The two exceptional symbols ">" and "<" can be read as ``generate only the
%   `forward-directed' resp.\ `backward-directed entry'\,''.\par
%
%
% \item[c)] If there are three entries in \meta{main entries}, then
%   \meta{format} becomes ``\myspace"1!~23"\myspace"3!12~"''.
%   For example, if \meta{main entries} is ``"graulty of bazola"'', then
%   the following two "\index" entries are produced.
%
%   \begin{myindex}
%   \myi graulty
%   \mys $\sim$ of bazola\pno
%   \myi bazola
%   \mys graulty of $\sim$\pno
%   \end{myindex}
%
%    The flags have an analogous effect to~b).
%    In addition, if the flags "`" resp.\ "'" are used, the second entry
%    (in our example: ``of'') is omitted from the output in the first
%    resp.\ in the second index entry. If the flags "*" and "`"
%    are used together, the second entry is also omitted from the output
%    into the running text.
%
% \item[d)] If there are four entries in \meta{main entries}, then
%    \meta{format} becomes ``\myspace"1!~24"\myspace"4!31~"''. For example, if
%    \meta{main entries} is ``"graulty of@ OF bazola"'', then the following
%    two "\index" entries are produced.
%
%   \begin{myindex}
%   \myi graulty
%   \mys $\sim$ of bazola\pno\sortedlikeq{bazola}
%   \myi bazola
%   \mys graulty OF $\sim$\pno\sortedlikeq{graulty OF}
%   \end{myindex}
%    In other words, we have a similar effect as in~c) except that for the
%    first entry the third word is skipped, and for the second entry the
%    second word is skipped. All flags as in~c) can be used with an
%    analogous meaning.
%    Also, if the "*" flag is used, the output into the running text is
%    analogous to~c) (the third word is skipped).
%
% \item[e)] If there are more than four entries in \meta{main entries}, then
%    \meta{formatA} resp.\ \meta{format*A} is used.
% \end{enumerate}
%
% If an explicit \meta{format} argument is given to "\varindex" (together with
% the "\varindextwo" macro as \meta{main program}), then this format is used,
% except when it contains the symbol "+". In this case, this symbol "+" is
% replaced by the format which would be generated by the rules~a)--e).
% If additionally the "*" flag is used, there is a special rule:
% If the explicit format contains a "*", then no additional "*"-output is
% produced. Otherwise, the "*"-output from~a)--e) is prepended to the given
% \meta{format} (with a space at the end, unless the format string is empty).
% This means that ``normally'' you have the output from a)--e),
% unless you write an own explicit "*"-part in the \meta{format}.
%
% If you do not like the tokens chosen for the default symbols, you can
% easily choose your own tokens by just replacing "\varindextwoScan" with your
% own macro (or defining your own ``magic'' tokens after "\varindextwoScan",
% see the example in Section~\ref{ss:expatch}).

%
% \section{Primitive customization}
%
%
% You may use the command
% \DescribeMacro{\varindexUsePlaceholderAfalse}
% \DescribeMacro{\varindexUsePlaceholderBfalse}
% \begin{verbatim}
%     \varindexUsePlaceholderAfalse
%     \varindexUsePlaceholderBfalse\end{verbatim}
% to ignore the symbol "~" resp.\ "=" in the format entry.
% You can easily restore the default by
% \DescribeMacro{\varindexUsePlaceholderAtrue}
% \DescribeMacro{\varindexUsePlaceholderBtrue}
% \begin{verbatim}
%     \varindexUsePlaceholderAtrue
%     \varindexUsePlaceholderBtrue\end{verbatim}
%
% If you want to switch off the placeholder only at the beginning of
% a new entry (resp.\ of a new subentry), you can use the commands
% \DescribeMacro{\varindexLeadingPlaceholderAfalse}
% \DescribeMacro{\varindexLeadingPlaceholderBfalse}
% \begin{verbatim}
%     \varindexLeadingPlaceholderAfalse
%     \varindexLeadingPlaceholderBfalse\end{verbatim}
% The default is restored with
% \DescribeMacro{\varindexLeadingPlaceholderAtrue}
% \DescribeMacro{\varindexLeadingPlaceholderBtrue}
% \begin{verbatim}
%     \varindexLeadingPlaceholderAtrue
%     \varindexLeadingPlaceholderBtrue\end{verbatim}
%
% By default, the ``","'' in the format string is in this connection
% treated as a symbol generating a new ``subentry''. You can change this
% default with the command
% \DescribeMacro{\varindexCommaLeadsfalse}
% \begin{verbatim}
%  \varindexCommaLeadsfalse\end{verbatim}
% You may switch back with
% \DescribeMacro{\varindexCommaLeadstrue}
% \begin{verbatim}
%  \varindexCommaLeadstrue\end{verbatim}
%
%
% The content of the macros
% \DescribeMacro{\varindexPlaceholderAText}
% \DescribeMacro{\varindexPlaceholderASort}
% \DescribeMacro{\varindexPlaceholderBText}
% \DescribeMacro{\varindexPlaceholderBSort}
% \begin{verbatim}
%   \varindexPlaceholderAText
%   \varindexPlaceholderASort
%   \varindexPlaceholderBText
%   \varindexPlaceholderBSort\end{verbatim}
% is used as the placeholder "~" resp.\ "="  in the index resp.\ sort entry.
% Note that if one of these entries expands empty, different rules for
% the automatic insertion of spaces apply (for the index and sort entry
% separately).
%
% You may prefer that the placeholder text resp.\ sort content changes at
% run time to the context of the previous "!" entry. For example, you may
% want that the command "\varindex{Gibble Gobble}{2!~2}" produces the index
% entry
%  \begin{myindex}
%  \myi Gibble
%  \mys Gibble Gobble\pno
%  \end{myindex}
% (or is at least \emph{sorted} as such an entry).
%
% You can achieve this with the commands
% \DescribeMacro{\varindexPreviousATexttrue}
% \DescribeMacro{\varindexPreviousASorttrue}
% \DescribeMacro{\varindexPreviousBTexttrue}
% \DescribeMacro{\varindexPreviousBSorttrue}
% \begin{verbatim}
%  \varindexPreviousATexttrue
%  \varindexPreviousASorttrue
%  \varindexPreviousBTexttrue
%  \varindexPreviousBSorttrue\end{verbatim}
% for the text and sort entry and the placeholders "~" and "=", respectively.
% With these commands, the content of "\varindexPlaceholderAText" etc
% is only used as the default if no previous "!" entry was given.
% You can switch back to the default mode with the respective commands
% \DescribeMacro{\varindexPreviousATextfalse}
% \DescribeMacro{\varindexPreviousASortfalse}
% \DescribeMacro{\varindexPreviousBTextfalse}
% \DescribeMacro{\varindexPreviousBSortfalse}
% \begin{verbatim}
%  \varindexPreviousATextfalse
%  \varindexPreviousASortfalse
%  \varindexPreviousBTextfalse
%  \varindexPreviousBSortfalse\end{verbatim}
% By default, the ``","'' in the format entry is considered in this connection
% similar to ``"!"''. If you want to change this, use the command
% \DescribeMacro{\varindexCommaExpandfalse}
% \begin{verbatim}
%   \varindexCommaExpandfalse\end{verbatim}
% You may switch back with
% \DescribeMacro{\varindexCommaExpandtrue}
% \begin{verbatim}
%  \varindexCommaExpandtrue\end{verbatim}
%
% Since version 2.2, the dash ``"-"'' occurring at the beginning of entry
% (or after ``"!"'' or even after ``","'' if "\varindexCommaExpandtrue" is in
% effect) is ignored for sorting.
% You can switch off this feature with the command
% \DescribeMacro{\varindexOutSortDashBeg}
% \begin{verbatim}
%   \let\varindexOutSortDashBeg\varindexOutSortDash\end{verbatim}
% To restore the new default, use
% \begin{verbatim}
%   \let\varindexOutSortDashBeg\toolboxEmpty\end{verbatim}
% \bigskip
% The commands
% \begin{myquote}
% \DescribeMacro{\varindexSetWordString}
% "\varindexSetWordString{|}"\myquad
% (Use "|" as separator for entries instead of spaces)\par
% \DescribeMacro{\varindexSetSortString}
% "\varindexSetSortString{>}"\myquad(default: "@")\par
% \DescribeMacro{\varindexSetPlainString}
% "\varindexSetPlainString{<}"\myquad(default: "!")
% \end{myquote}
% set the default ``magic'' strings used for \meta{main entries}.
% The argument of the above macros is intentionally \emph{not} expanded (so
% that you do not have troubles with active characters like "~").
% To force expansion, use e.\,g.
% \begin{verbatim}
% \expandafter\varindexSetWordString\expandafter{\MacroToExpand}\end{verbatim}
% or some similar construct.
%
% If you use a different separator than ``space'' for the entries, you may
% want to change the rule how braces are eliminated. With the commands
% \DescribeMacro{\varindexEliminateOuterBracetrue}
% \DescribeMacro{\varindexEliminateInnerBracetrue}
% \DescribeMacro{\varindexEliminateOuterBracefalse}
% \DescribeMacro{\varindexEliminateInnerBracefalse}
% \begin{verbatim}
%   \varindexEliminateOuterBracetrue
%   \varindexEliminateInnerBracetrue
%   \varindexEliminateOuterBracefalse
%   \varindexEliminateInnerBracefalse\end{verbatim}
% you may allow resp.\ forbid the elimination of braces around the entries
% resp.\ around \meta{indextext} \meta{sorttext} and \meta{outputtext}
% With the flag ``"1"'', \meta{main entries} is considered as one single entry,
% but if "\varindexEliminateOuterBracetrue" is set (which is the default)
% also in this case an additional pair of braces around main entry is
% eliminated.
%
% Similarly as above,
% \begin{myquote}
% \DescribeMacro{\varindexSetIndexString}
%   "\varindexSetIndexString{|}"   \myquad default: space\par
% \DescribeMacro{\varindexSetOutputString}
%   "\varindexSetOutputString{<}"  \myquad default: "*"\par
% \DescribeMacro{\varindexSetTildeAString}
%   "\varindexSetTildeAString{~}"  \myquad (is default)\par
% \DescribeMacro{\varindexSetTildeBString}
%   "\varindexSetTildeBString{=}"  \myquad (is default)\par
% \DescribeMacro{\varindexSetSpaceString}
%   "\varindexSetSpaceString{_}"   \myquad (is default)\par
% \DescribeMacro{\varindexSetSpaceTokString}
%   "\varindexSetSpaceTokString{s}"\myquad (is default)\par
% \DescribeMacro{\varindexSetOmitString}
%   "\varindexSetOmitString{.}"    \myquad (is default)\par
% \DescribeMacro{\varindexSetCommaString}
%   "\varindexSetCommaString{,}"   \myquad (is default)\par
% \DescribeMacro{\varindexSetDashString}
%   "\varindexSetDashString{-}"    \myquad (is default)\par
% \DescribeMacro{\varindexSetExclamString}
%   "\varindexSetExclamString{!}"  \myquad (is default)\par
% \DescribeMacro{\varindexSetStringForNr}
%   "\varindexSetStringForNr 1{a}" \myquad default: "1"\par
%   "\varindexSetStringForNr{11}{b}"\myquad No default!
% \end{myquote}
% set the default ``magic'' strings used for \meta{format}.
% In contrast to before, the arguments are fully expanded (with
% "\varindexedef", see Section~\ref{s:docufeature}).
% Note that the last command provides a way to access more than $10$ entries!
%
% If you use "\varindexSetIndexString{|}" (with some non-space token),
% you may still use spaces in the format which then are just ignored.
%
%
% Avoid that one string is a prefix of another string: In this case,
% strange effects may happen, and this behavior may even change in
% future releases of this package.
% Note that the above effect may arise unintentionally if you use active
% chars. For this reason, ``"~"'' is defined to expand to the letter ``"~"'',
% before the expansion is executed. Maybe in later releases of this package
% there will be further such redefinitions.
% You can explicitly use this feature in your own macros by using
% "\varindexedef" instead of the usual "\edef",
% see Section~\ref{s:docufeature}.
%
% You can change the output for the text resp sort entry of the various
% symbols used in the format string. More precisely, you can redefine
%
% \begin{myquote}
% \DescribeMacro{\varindexOutExclam}
%    "\varindexOutExclam"      \myquad Text output for "!"\par
% \DescribeMacro{\varindexFollowsExclam}
%    "\varindexFollowsExclam"  \myquad
%Decide whether magic space after "!" is inserted\par
% \DescribeMacro{\varindexOutTextComma}
%    "\varindexOutTextComma"   \myquad Text output for ","\par
% \DescribeMacro{\varindexOutSortComma}
%    "\varindexOutSortComma"   \myquad Sort output for ","\par
% \DescribeMacro{\varindexFollowsComma}
%    "\varindexFollowsComma"   \myquad
%Decide whether magic space after "," is inserted\par
% \DescribeMacro{\varindexPreceedsComma}
%    "\varindexPreceedsComma"   \myquad
%Decide whether magic space before "," is erased\par
% \DescribeMacro{\varindexOutTextSpace}
%    "\varindexOutTextSpace"   \myquad Text output for "_"\par
% \DescribeMacro{\varindexOutSortSpace}
%    "\varindexOutSortSpace"   \myquad Sort output for "_"\par
% \DescribeMacro{\varindexFollowsSpace}
%    "\varindexFollowsSpace"   \myquad
%Decide whether magic space after "_" is inserted\par
% \DescribeMacro{\varindexPreceedsSpace}
%    "\varindexPreceedsSpace"   \myquad
%Decide whether magic space before "_" is erased\par
% \DescribeMacro{\varindexOutTextSpTok}
%    "\varindexOutTextSpTok"   \myquad Text output for "s"\par
% \DescribeMacro{\varindexOutSortSpTok}
%   "\varindexOutSortSpTok"   \myquad Sort output for "s"\par
% \DescribeMacro{\varindexFollowsSpTok}
%    "\varindexFollowsSpTok"   \myquad
%Decide whether magic space after "s" is inserted\par
% \DescribeMacro{\varindexPreceedsSpTok}
%    "\varindexPreceedsSpTok"   \myquad
%Decide whether magic space before "s" is erased\par
% \DescribeMacro{\varindexOutTextDash}
%    "\varindexOutTextDash"    \myquad Text output for "-"\par
% \DescribeMacro{\varindexOutSortDash}
%    "\varindexOutSortDash"    \myquad Sort output for "-"\par
% \DescribeMacro{\varindexOutSortDashBeg}
%    "\varindexOutSortDashBeg"    \myquad Sort output for "-"
%if nothing preceeds.\par
% \DescribeMacro{\varindexFollowsDash}
%    "\varindexFollowsDash"    \myquad
%Decide whether magic space after "-" is inserted\par
% \DescribeMacro{\varindexPreceedsDash}
%    "\varindexPreceedsDash"   \myquad
%Decide whether magic space before "-" is erased\par
% \DescribeMacro{\varindexOutTextOmit}
%    "\varindexOutTextOmit"    \myquad Text output for "."\par
% \DescribeMacro{\varindexOutSortOmit}
%    "\varindexOutSortOmit"    \myquad Sort output for "."\par
% \DescribeMacro{\varindexFollowsOmit}
%    "\varindexFollowsOmit"    \myquad
%Decide whether magic space after "." is inserted\par
% \DescribeMacro{\varindexPreceedsOmit}
%    "\varindexPreceedsOmit"   \myquad
%Decide whether magic space before "." is erased
% \end{myquote}
%
% The meaning is as follows: "\varindexOut..." should just expand to the
% tokens which should be written into the text resp.\ sort output.
% The corresponding macro "\varindexFollows..." is typically defined with
% one of the following commands:
% \begin{verbatim}
%     \let\varindexFollows...\varindexNextSpace
%     \let\varindexFollows...\varindexNoNextSpace\end{verbatim}
% In the first case, a space is `magically' inserted in front of a
% subsequent token (unless this token erases the magic space).
% In the second case, no space follows, of course.
% There is an alternative call:
% \begin{myquote}
% \DescribeMacro{\varindexSpaceTexttrue}
% \DescribeMacro{\varindexSpaceTextfalse}
% \DescribeMacro{\varindexSpaceSorttrue}
% \DescribeMacro{\varindexSpaceSortfalse}
% "\def\varindexFollows...{\varindexSpaceTexttrue\varindexSpaceSortfalse}"
% \end{myquote}
% This definition achieves that for the text output a space should follow,
% but not for the sort output. Of course, you can also use similarly the
% commands "\varindexSpaceTextfalse" and/or "\varindexSpaceSorttrue" in the
% above definition (the effect should be obvious).
% In particular, "\varindexNextSpace" is equivalent to
% "\varindexSpaceTexttrue\varindexSorttrue".\par
%
% The macro "\varindexPreceeds..." is similarly as "\varindexFollows..." with
% the difference that it is executed \emph{before} the token in question is
% output. In particular, you can ignore a previous `magic space'.
% All of the `magic space' commands are defined by default with
% \begin{verbatim}
%     \let\varindexFollows...\toolboxEmpty
%     \let\varindexPreceeds...\varindexNoNextSpace\end{verbatim}
% with the following two exceptions:
% \begin{verbatim}
%     \let\varindexFollowsExclam\varindexNextSpace
%     \let\varindexFollowsComma\varindexNextSpace\end{verbatim}
%
% \section{Documented Features}\label{s:docufeature}
%
% The title ``documented features'' means that these are hacks which allow
% further customization but which are ``documented'' in the sense that these
% hacks will also be available in later versions. (You use an ``undocumented''
% feature if you must use a macro name of the "varindex" package which
% contains the symbol "@"). If a feature described in this section does not
% work in the documented way, this is a bug and might be repaired in a
% later version of "\varindex".
%
% (In a future release, "varindex" will cook your coffee, too).\bigskip
%
%
% The argument \meta{main entries} is \emph{never} expanded, so you can
% actually write macros into the "*.idx" file.
% The command actually used to write the index is
% \begin{myquote}
% \DescribeMacro{\varindexIndex}"\varindexIndex"
% \end{myquote}
% (which by default is of course defined as "\index").
% Since some implementations of the "\index" command still (partially) expand
% their argument (which might be considered as a bug), the argument of
% "\varindexIndex" is translated before the actual call with the aid of
% "\toolboxMakeHarmless".
% If you want to redefine "\varindexIndex" to a personal "\index"-output
% function, you might want to skip the "\toolboxMakeHarmless" step.
% This is arranged with the command
% \begin{myquote}
% \DescribeMacro{\varindexNoConvert}"\varindexNoConvert"
% \end{myquote}
% You can cancel the effect of this command with
% \begin{myquote}
% \DescribeMacro{\varindexNoConvert}"\varindexNoConvert"
% \end{myquote}
% Even if "\varindexNoConvert" is not in effect, \meta{main entries} is not
% expanded (and in particular, the argument of "\varindexIndex" consists of
% the corresponding entries in an unexpanded form).\par\bigskip
%
% The whole "\varindex" expansion takes place in a group, so all your
% variables are local to a single call (unless you use "\gdef"
% (and friends) of course).\par
%
% There are no "\global" directives used in any macro of this package.
% In particular, if you call the above macros in a group (or redefine
% "\varindexIndex"), the effect holds only until the end of the group.
% \par\bigskip
%
% In contrast to \meta{main entries}, the \meta{format} argument is expanded
% (with "\varindexedef", see below) (and also \meta{formatA} resp.\
% \meta{format*A} is expanded before), so you can define abbreviations
% (even simple conditional % abbreviations) for \meta{format}.
% Note, however, that the expansion is essentially only evaluated with "\edef".
% So, you will probably not want to use e.\,g.\ "\relax", since this command
% expands to itself (and not into nothing -- use "\toolboxEmpty" if you want
% the latter).
% If you want more complex abbreviations, use \meta{main program} instead.
%
% In order to help you write conditional abbreviations, the
% following variables are defined when your macro is expanded (and in
% particular also in \meta{main program}). You may of course redefine them
% appropriately:
%
% \begin{description}
% \item[\macrott{varindexAppend}]\DescribeMacro{\varindexAppend}
%    The argument \meta{pagestyleA} resp.\ \meta{pagestyleB}.
% \item[\macrott{varindexCount}]\DescribeMacro{\varindexCount}
%    A \TeX\ counter containing the number of entries in \meta{main entries}.
% \item[\macrott{varindexLastEntry}]\DescribeMacro{\varindexLastEntry}
%   This is a macro (\emph{not} a counter) which is usually the value of
%   "\the\varindexCount". See below.
% \item[\macrott{varindexEntry1} \macrott{varindexEntry2} \mdseries$\ldots$]
%\DescribeMacro{\varindexEntry1-...}
%    The (index) text occurring in \meta{Entry1} \meta{Entry2} $\ldots$
% \item[\macrott{varindexSort1} \macrott{varindexSort2} \mdseries$\ldots$]
%\DescribeMacro{\varindexSort1-...}
%    The corresponding sort entry. If no sort entry was given, this is
%    the same as the corresponding "\varindexEntry1-..."
% \item[\macrott{varindexPlain1} \macrott{varindexPlain2} \mdseries$\ldots$]
%\DescribeMacro{\varindexPlain1-...}
%    The text which should be output in the text. If no such entry was given,
%    this is the same as the corresponding "\varindexEntry1-..."
% \end{description}
% "\varindexCount" may be larger than $10$, and correspondingly there may be
% also more than $10$ different "\varindexEntry"/"Sort"/"Plain" macros.
% If you \emph{add} entries, you have to increase "\varindexCount"
% correspondingly, otherwise an error is produced if the format string
% contains an entry larger then "\varindexCount". However, your additional
% entries are \emph{not} output for \emph{empty} formats:
% For empty formats the entries output are $1$--"\varindexLastEntry".
% So if you want to output also your additional entries
% in empty formats, you have to set "\varindexLastEntry" to your modified
% value of "\varindexCount" in \meta{main program}. You may e.\,g.\ achieve
% this with the following lines:
% \begin{myquote}
%  "\expandafter"\\
%  "\def\expandafter\varindexLastEntry\expandafter{\the\varindexCount}"\proz
% \end{myquote}
% It is admissible that "\varindexLastEntry" is larger than "\varindexCount":
% In this case all entries until "\varindexLastEntry" are
% output in empty formats without producing an error (provided, of course,
% that the corresponding variables "\varindexEntry..."/"Sort..."/"Plain..." are
% defined).
%
% Note that numbers in \TeX\ command names require special handling,
% i.\,e., you have to use something like 
%
% \begin{verbatim}
% \csname varindexPlain2\endcsname\end{verbatim}
%
% to access variables. To avoid this, you may want to use the macros
% \DescribeMacro{\toolboxLet}"\toolboxLet" and
% \DescribeMacro{\toolboxWithNr}"\toolboxWithNr" of the "toolbox" package.
% Examples are
% \begin{verbatim}
%   \toolboxWithNr 1\let{varindexEntry}\toolboxEmpty
%   \toolboxWithNr {10}\def{varindexSort}{Foo}
%   \toolboxLet    \mymacro{varindexPlain\the\mycount}\end{verbatim}
% These commands set "\varindexEntry1" to "\toolboxEmpty", "\varindexSort10"
% to Foo, and "\mymacro" to the content of "\varindexPlain5" (if "\mycount=5").
%
% All these variables are also available when \meta{main program} is expanded.
% In \meta{main program} also the following functions are useful:
%
% \begin{description}
% \item[\macrott{varindexFormat}]\DescribeMacro{\varindexFormat}
%    This macro expands to the actual \meta{format} which is used.
%    The main purpose of \meta{main program} will usually be to (re)define
%    this macro. After \meta{main program} has been called, this macro is
%    modified in several ways:
%    \begin{enumerate}
%    \item "\varindexFormat" is expanded (with "\varindexedef", see below).
%       (thus, "\varindexFormat" is expanded \emph{before and after} the call
%        of \meta{main program}).
%    \item "\ifvarindexStar" is true (see below)
%        a ``"*"'' resp.\ ``"*"\myspace'' is prepended.
%    \item If it is only ``"*"'', it is replaced by ``"*"\myspace''.
%    \end{enumerate}
%    Note that before execution of \meta{main program}, no tests on the
%    validity of the format are done: You may use your own symbols to
%    `communicate' with \meta{main program} (provided that \meta{main program}
%    eliminates these symbols afterwards).
% \item[\macrott{varindexFormatDefault}]\DescribeMacro{\varindexFormatDefault}
%    This macro expands to \meta{formatA} resp.\ \meta{format*A}
%    (in the expanded form) depending whether the "*" flag has been used in
%    the call. Note that this expansion was executed \emph{before}
%    the optional format argument is expanded for the first time.
% \item[\macrott{ifvarindexstar}]\DescribeMacro{\ifvarindexstar}
%    \meta{ifpart} [ "\else" \meta{elsepart} ] "\fi"\par
%    If the optional "*" was given, \meta{ifpart} is executed, otherwise
%    \meta{elsepart}. This is a \TeX\ "if" command. In particular, by
%    \DescribeMacro{\varindexStarfalse}\DescribeMacro{\varindexStartrue}
%    "\varindexStarfalse" resp.\ "\varindexStartrue"
%    you may change the behavior for future "if"'s. This can be used
%    to prevent/force that a ``"*"'' resp.\ ``"*"\myspace'' is prepended to
%    the format after the execution of \meta{main program}.
%    Setting of this variable has no effect in \meta{scan program}.
% \item[\macrott{ifvarindexExplicitFormat}]
%    \DescribeMacro{\ifvarindexExplicitFormat}
%    This is a \TeX\ "if" command (see above) which is true if the optional
%    format argument \meta{formatB} was given.
% \end{description}
%
% To ``compose'' the format, you may want to use the macros
% \begin{myquote}
% \DescribeMacro{\toolboxDef}
% "\toolboxDef\macrotodefine"\bracemeta{argumentlist}\par
% \DescribeMacro{\toolboxAppend}
% "\toolboxAppend\macrotoappend"\bracemeta{argumentlist}\par
% \DescribeMacro{\varindexedef}
% "\varindexedef\macrotodefine"\bracemeta{argumentlist}
% \end{myquote}
% All commands work similar to "\def" resp.\ "\edef" with two differences:
%
% For "\toolboxDef\macro", \meta{argumentlist} is expanded precisely by one
% level (for details, see the documentation of the "toolbox" package).
% Of course, for "\toolboxAppend", the new definition is appended to the old
% definition of "\macro".
%
% "\varindexedef" fully expands \meta{argumentlist}. However, some active
% symbols (currently only "~", but additional symbols might follow in a
% future release) are deactivated before expansion, so that "~" actually
% expands to "~" and not to a strange command sequence.
%
%
% To ``decompose'' the format, you may want to use one of the macros
%
% \begin{myquote}
% \DescribeMacro{\toolboxSplitAt}
% "\toolboxSplitAt"\bracemeta{arg}\bracemeta{critical}\bracemeta{command}\par
% \DescribeMacro{\toolboxMakeSplit}
% "\toolboxMakeSplit"\bracemeta{critical}\bracemeta{command}\par
% \DescribeMacro{\varindexMakeSplitExpand}
% "\varindexMakeSplitExpand"\relax
%\bracemeta{macros expanding to critical}\bracemeta{command}\par
% \DescribeMacro{\varindexMakeVarSplit}
% "\varindexMakeVarSplit\variable"\relax
%\bracemeta{macros expanding to critical}\bracemeta{command}
% \end{myquote}
% The first two macros are described in the "toolbox" package.
% The remaining two macros are similar to "\varindexMakeSplit" with the
% difference that the argument \meta{critical} is obtained by expanding
% \meta{macros expanding to critical} with the aid of "\varindexedef".
% In the last form, additionally "\variable" is (re)defined to expand to
% \meta{critical} ("\variable" is any free name).
%
% The following instances of a command created by "\toolboxMakeSplit"
% exist (the content of \meta{critical} should be obvious):
%
% \begin{myquote}
% \DescribeMacro{\varindexSplitAtIndex}
% "\varindexSplitAtIndex"\myquad(generated by "\varindexSetIndexString")\par
% \DescribeMacro{\varindexSplitAtOutput}
% "\varindexSplitAtOutput"\myquad(generated by "\varindexSetOutputString")\par
% \DescribeMacro{\varindexSplitAtTildeA}
% "\varindexSplitAtTildeA"\myquad(generated by "\varindexSetTildeAString")\par
% \DescribeMacro{\varindexSplitAtTildeB}
% "\varindexSplitAtTildeB"\myquad(generated by "\varindexSetTildeBString")\par
% \DescribeMacro{\varindexSplitAtSpace}
% "\varindexSplitAtSpace"\myquad(generated by "\varindexSetSpaceString";
% by default, \meta{critical} is "_")\par
% \DescribeMacro{\varindexSplitAtSpaceTok}
% "\varindexSplitAtSpaceTok"\myquad
%  (generated by "\varindexSetSpaceTokString";
%  by default, \meta{critical} is "s")\par
% \DescribeMacro{\varindexSplitAtOmit}
% "\varindexSplitAtOmit"\myquad(generated by "\varindexSetOmitString")\par
% \DescribeMacro{\varindexSplitAtComma}
% "\varindexSplitAtComma"\myquad(generated by "\varindexSetCommaString")\par
% \DescribeMacro{\varindexSplitAtDash}
% "\varindexSplitAtDash"\myquad(generated by "\varindexSetDashString")\par
% \DescribeMacro{\varindexSplitAtExclam}
% "\varindexSplitAtExclam"\myquad(generated by "\varindexSetExclamString")\par
% \DescribeMacro{\varindexSplitAtNr??}
% "\varindexSplitAtNr1"\myquad(generated by "\varindexSetStringForNr")\par
% "\varindexSplitAtNr2"\par
% $\ldots$
% \end{myquote}
% Note that you must use "\csname ....\endcsname" to call e.\,g.\ the macro
% "\varindexSplitAtNr1". Only those numbers are admissible for
% "\varindexSplitAtNr??" which have been introduced with
% "\varindexSetStringForNr" (by default, this is the case for $1$--$10$).
%
% There is also the instance
% \DescribeMacro{\varindexSplitSpace}
% \begin{myquote}
% "\varindexSplitSpace"\myquad
% (to be distinguished from "\varindexSplitAtSpace")
% \end{myquote}
% where \meta{critical} is the space symbol.
%
%
% In addition, you may use the variables
% \begin{myquote}
% \DescribeMacro{\varindexIndexString}"\varindexIndexString"\par
% \DescribeMacro{\varindexOutputString}"\varindexOutputString"\par
% \DescribeMacro{\varindexTildeAString}"\varindexTildeAString"\par
% \DescribeMacro{\varindexTildeBString}"\varindexTildeBString"\par
% \DescribeMacro{\varindexSpaceString}"\varindexSpaceString"\par
% \DescribeMacro{\varindexSpaceTokString}"\varindexSpaceTokString"\par
% \DescribeMacro{\varindexOmitString}"\varindexOmitString"\par
% \DescribeMacro{\varindexCommaString}"\varindexCommaString"\par
% \DescribeMacro{\varindexDashString}"\varindexDashString"\par
% \DescribeMacro{\varindexExclamString}"\varindexExclamString"\par
% \DescribeMacro{\varindexNr??String}\relax
%"\varindexNr1String \varindexNr2String ..."
% \end{myquote}
% which expand to the corresponding strings.\par
%
% All the previous macros should not be redefined ``by hand''.
% They are implicitly redefined by the "\varindexSet..." commands
% (which partially also do other tasks).\par
%
%
%
% To drop possible braces, use the command
% \DescribeMacro{\toolboxDropBrace}
% \begin{verbatim}
%   \toolboxDropBrace\variable\end{verbatim}
% of the "toolbox" package.\par
%
% In \meta{scan program} you may already want to call "\toolboxMakeSplit":
% In this way, the choices for the ``magic'' tokens are made in the
% (usually small) argument \meta{scan program}, and in this sense, you can
% keep your (possibly rather complex) macro \meta{main program}
% ``customizable'' (i.\,e.\ you can use the same macro also with a different
% set of ``magic'' strings).
% However, the main task of \meta{scan program} is to introduce
% new flags with
%
% \begin{myquote}
% \DescribeMacro{\varindexNewFlag}
%"\varindexNewFlag"\meta{token}"\position["\meta{program}"\varindexEndOfFlag]"
% \end{myquote}
% or
% \begin{myquote}
%   "\varindexNewFlag"\meta{token}"\position[\programmacro]"
% \end{myquote}
% Here, "\position" is an (undefined) macro, and \meta{token} an arbitrary
% token. The "[program(macro)]" part is optional and is explained later.
% If \meta{token} appears in \meta{flags}, then "\position" is defined to
% expand to a (literally) number, namely the (last) position where token
% was given inside the \meta{flags} list (counting from $0$). For example, if
% \begin{verbatim}
%   \varindexNewFlag ,\commapos
%   \varindexNewFlag -\minuspos
%   \varindexNewFlag .\pointpos\end{verbatim}
% are used in "\programA", then in the call
% \begin{verbatim}
% \varindex(){\programA}{\programB},-**-{}\end{verbatim}
% the variable "\commapos" will expand in "\programB" to "0", while
% "\minuspos" will expand to "4" (the last position is taken). Finally,
% "\pointpos" is "\undefined" (unless you defined "\pointpos" differently
% \emph{after} the call of "\varindexNewFlag"; in this case, this is the
% default).
% If "\varindexNewFlag" is called multiple times with the same token,
% only the \emph{last} call with this token takes effect.
%
% The flags "*" and "1" are introduced in this way with
% \DescribeMacro{\varindexStarPos}
% \DescribeMacro{\varindexOneEntry}
% \begin{verbatim}
% \varindexNewFlag *\varindexStarPos
% \varindexNewFlag 1\varindexOneEntry\end{verbatim}
% \emph{before} \meta{scan program} is executed. This means:
% \begin{enumerate}
% \item Usually, "\varindexStarPos" contains the (last) position of "*"
%    (resp.\ is "\undefined"). Moreover, if you define "\varindexStarPos" in
%    \meta{scan program} or in some flag, you get the same effect as if "*"
%    had been used.
%    An analogous remark holds for "\varindexOneEntry".
% \item If you introduce "*" with "\varindexNewFlag", the "*" looses its
%    original meaning. The same holds for "1".
% \end{enumerate}
%
% If you have added a "[program(macro)]" part in the call of
% "\varindexNewFlag", this part is expanded whenever the flag introduced
% by "\token" is used in the call of "\varindex"
% (note that it is not excluded that this happens several times within one
% call). More precisely, program is expanded \emph{after} the variable
% "\position" has been set to the corresponding value, so you may already
% use "\position" in the program part.
%
% \textbf{Important:} The last token expanded in program \emph{must} be
% \DescribeMacro{\varindexEndOfFlag}"\varindexEndOfFlag".
% This is not nice but has to do with the way \TeX\ parses its arguments.
% Also if you use the form "\programmacro", the \emph{very last} token expanded
% must be "\varindexEndOfFlag". Even a construction like
% \begin{verbatim}
% \def\myprogrammacro{\ifx....
%   \CallSomeMacroWithvarindexEndOfFlagAtTheEnd
% \fi
% \varindexEndOfFlag}\end{verbatim}
% is forbidden: In "\CallSomeMacroWithvarindexEndOfFlagAtTheEnd" an error
% would occur at the end, since still the tokens "\fi\varindexEndOfFlag"
% are in the tokenlist when EndOfFlag is reached there.
% As a workaround, you may e.\,g.\ use
% \begin{myquote}
% "\def\myprogrammacro{\ifx..."\pari
%    "\def\execute{....\varindexEndOfFlag}"\ppari
% "\else"\pari
%    "\def\execute{\varindexEndOfFlag}"\ppari
% "\execute}"
% \end{myquote}
%
% If you use the form "[\programmacro]", your macro may even read additional
% parameters. These parameters are expected in the call of "\varindex"
% \emph{behind} the flag you have introduced. So you may actually use flags
% with parameters. For example, if "\scanprogram" contains a macro like
% \begin{verbatim}
%    \varindexNewFlag -\minuspos[\readpara]\end{verbatim}
% and you have defined
% \begin{verbatim}
%    \def\readpara#1{\def\merk{#1}\varindexEndOfFlag}\end{verbatim}
% then the call
% \begin{verbatim}
%   \varindex(){\scanprogram}{\mainprg}*-{Foo}-{Foo 2}*{Entries}\end{verbatim}
% is admissible, and in \meta{main program},
% "\merk" will have the value ``Foo 2''.
%
% If you are more familiar with \TeX, you can even allow \emph{optional}
% arguments following your flag: The value of the next (non-space) token is
% at the call of "\programmacro" already saved into the macro
% \DescribeMacro{\varindexNextToken}
% \begin{verbatim}
%     \varindexNextToken\end{verbatim}
% so you can just use it to test for e.\,g.\ ``"["''. In this connection,
% you may also want to use the commands
% \DescribeMacro{\varindexTestAndExec}
% \DescribeMacro{\varindexSkipAndExec}
% \begin{verbatim}
%     \varindexTestAndExec
%     \varindexSkipAndExec\end{verbatim}
% Please, see the program text how these commands are applied.\bigskip
%
% Since version 2.1 you can also hack in personal extensions of the format
% string. To do this, use the command
% \begin{myquote}
% \DescribeMacro{\varindexNewFormat}
% "\varindexNewFormat\"\meta{splitcommand}\bracemeta{action}
% \end{myquote}
% Here, "\"\meta{splitcommand} is a command generated by "\toolboxMakeSplit" or
% friends (preferably by "\varindexMakeSplitExpand", because the command should
% act on the format string which is expand with "\varindexedef").
% The string where it splits is the new string you can use
% in the format argument after this call. For each
% occurrence of the corresponding string in the format argument,
% \meta{action} will be executed. If "\"\meta{splitcommand} splits at a string
% which already had a previous meaning in the format string (or which is a
% prefix to such a string) the old meaning of this string in the format string
% is overridden.
%
% Typically, action will contain the following commands:
% One action will probably be to output a desired token (sequence) via
% \begin{myquote}
% \DescribeMacro{\varindexTokensOut}
% "\varindexTokensOut"\bracemeta{text token}\bracemeta{sort token}
% \end{myquote}
% or
% \begin{myquote}
% \DescribeMacro{\varindexTokensOutExpand}
% "\varindexTokensOutExpand"\meta{macro for text token}\relax
%\meta{macro for sort token}
% \end{myquote}
% In the first form, \meta{text token} resp.\ \meta{sort token} is the token
% sequence put into the index or running text respectively into the sort entry
% of the index for the corresponding format entry. The second form is similar
% with the only difference that the arguments must be single macros which
% expand to \meta{text token} and \meta{sort token}, respectively.\par
%
% The variable \DescribeMacro{\ifvarindexIndexMode}
% \begin{verbatim}
%   \ifvarindexIndexMode\end{verbatim}
% can be used to test whether the output goes into the running text or into
% the index (i.\,e.\ whether a ``"*"'' preceeded the current entry).
% For output into the text, \meta{sort token} is ignored, of course).
%
% Another action in "\varindexNewFormat" will probably be to take care of
% the magic space flags. This is achieved by a call of "\varindexNextSpace" or
% "\varindexNoNextSpace" (or separately via "\varindexSpaceTexttrue"/"false"
% resp.\ "\varindexSpaceTexttrue"/"false"); see the earlier description.
% The magic space flags are taken into account in "\varindexTokensOut".
% Thus, if you want to ignore the previous flag for some reason you should set
% them correspondingly before this call.
% However, after the call you should also set them correspondingly for
% further processing.\par
%
% Example:
% \begin{verbatim}
%   \toolboxMakeSplit{:}{splitAtColon}
%     \varindexNewFormat\splitAtColon{%
%        \varindexNoNextSpace
%        \ifvarindexIndexMode
%           \varindexTokensOut{:}{}\varindexSpaceTexttrue
%        \else
%           \varindexTokensOut{---}{}\varindexNoNextSpace
%        \fi}\end{verbatim}
% defines a new format entry ``":"'' which has the meaning that a colon
% (automagically followed by a space) is put into the index but not into
% the sort entry. Moreover, in the running text, the colon appears as a
% long dash with no space followed. In any case, there is no magic space output
% in front of the colon.\bigskip
%
% As an alternative action in "\varindexNewFormat", you can also call the
% default commands for the format entries. The corresponding macros are
% \begin{myquote}
% \DescribeMacro{\varindexAddSpace}"\varindexAddSpace"\myquad"_"\par
% \DescribeMacro{\varindexAddSpTok}"\varindexAddSpTok"\myquad"s"\par
% \DescribeMacro{\varindexAddOmit}"\varindexAddOmit"\myquad"."\par
% \DescribeMacro{\varindexAddDash}"\varindexAddDash"\myquad"-"\par
% \DescribeMacro{\varindexAddComma}"\varindexAddComma"\myquad","\par
% \DescribeMacro{\varindexAddExclam}"\varindexAddExclam"\myquad"!"\par
% \DescribeMacro{\varindexAddTildeA}"\varindexAddTildeA"\myquad"~"\par
% \DescribeMacro{\varindexAddTildeB}"\varindexAddTildeB"\myquad"="\par
% \DescribeMacro{\varindexAddNumber}"\varindexAddNumber"\bracemeta{number}
%\myquad"0-9" (\meta{number} is "1", "2", $\ldots$)
% \end{myquote}
% The precise meaning of these macros is intentionally \emph{not} documented,
% because some details or side effects might change in a future release of
% "varindex". But just for this reason, it might be advantageous to use the
% above macros instead of writing personal substitutes which may fail to
% have such side effects.\par
%
% Example:
% \begin{verbatim}
%    \varindexMakeSplitExpand{~-}{splitPhrase}% It is important to expand
%                                             % (to have correct ~ catcode)
%    \varindexNewFormat\splitPhrase{%
%       \varindexAddTildeA
%       \let\remember\varindexPreceedsDash
%       \let\varindexPreceedsDash\toolboxEmpty
%       \varindexAddDash
%       \let\varindexPreceedsDash\remember}\end{verbatim}
% After the above command, ``"~"'' and ``"-"'' have their usual meaning in
% the format string except when they follow immediately in the form ``"~-"''.
% In this case, the behavior of "-" changes as if "\varindexPreceedsDash"
% were empty (which has the effect that the output looks like~``$\sim$~-''
% instead of~``$\sim$-'').\par
% Note that although "~" is a prefix to "~-", the converse is not true:
% Thus, the above command does not change the previous meaning of~"~"
% (and of course also not of~"-").\par
% \bigskip
%
% \textbf{Some hack:}
% \TeX\ ignores leading spaces in the argument list of a ``normal'' macro.
% This has the effect that you should be able to insert spaces \emph{between}
% any of your arguments without any trouble.
% If this does not work the expected way, you can use the command
% \DescribeMacro{\varindexArgumentSpace}
% \DescribeMacro{\varindexSkipTricky}
% \begin{verbatim}
%    \let\varindexArgumentSpace\varindexSkipTricky\end{verbatim}
% which implements an own macro which does this task. Sadly, this can only
% be done with some catcode trickery which in turn might bring you in trouble
% with some (\emph{very}) exotic packages. You can restore the default with
% \DescribeMacro{\varindexSkipDefault}
% \begin{verbatim}
%    \let\varindexArgumentSpace\varindexSkipDefault\end{verbatim}
%
% All above customization commands/variables may be used anytime between
% two "\varindex" calls. They take effect with the next "\varindex" call.

%
% The macros "\varindextwoScan" and "\varindextwo" are considered as a
% (very useful) \emph{example} for the customization of the "\varindex"
% command. If you need further customization, have a look at their
% definition first.


% \iffalse
%</metainfo>
% \fi
% \iffalse--------------------------------------------------------------\fi
% \StopEventually{\PrintChanges}
%                     \section{Implementation}
%
% \iffalse--------------------------------------------------------------
%
%
%<*package>
% \fi
%    \begin{macrocode}
%%
%% This file was generated by the docstrip utility from varindex.dtx.
%% You may distribute this file only together with the source varindex.dtx
%% (and if possible with varindex.ins) and only if these files are unmodified.
%%
%% Author:
%%
%% M. V\"ath      martin@mvath.de
%%

\ifx\typeout\undefined
  \def\vx@tmp{\immediate\write17}
\else
  \let\vx@tmp\typeout
\fi\vx@tmp{varindex.sty by M. Vaeth: Release 2.3}
\ifx\ProvidesPackage\undefined\else
  \ProvidesPackage{varindex}[2001/05/06 v2.3]
\fi
\ifx\RequirePackage\undefined
  \ifx\toolboxDef\undefined
    \input toolbox.sty
  \fi
  \ifx\toolboxNewifFalse\undefined
    \errmessage{varindex.sty requires toolbox.sty version 3.1 or higher}
  \fi
\else
  \RequirePackage{toolbox}[2001/05/06]
\fi

%    \end{macrocode}
%
% \subsection{The \texttt{varindextwo} (sample) macros and their ``submacros''}
% \begin{macro}{\varindextwo...}\nocomment
%
%    \begin{macrocode}
%%
%% The file starts with the documented (sample) macros with their "sub"macros:
%%

\ifx\newcommand\undefined\else
  \newcommand{\varindextwoScan}{}
  \newcommand{\varindextwo}{}
  \newcommand{\vxtAddmyStar}{}
  \newcommand{\vxtMakemyFormat}{}
\fi

%    \end{macrocode}
%
%  \begin{macro}{\varindextwoScan}
% The demo ``scan'' macro. It provides some new flags.
%    \begin{macrocode}
\def\varindextwoScan{%
  \varindexNewFlag _\vxtSpaceA
  \varindexNewFlag -\vxtDashA
  \varindexNewFlag .\vxtDotA
  \varindexNewFlag <\vxtOmitA
  \varindexNewFlag /\vxtNoTildeA
  \varindexNewFlag `\vxtNopropA
  \varindexNewFlag P\vxtNopropA% Now obsolete; only for backward compatibility
  \varindexNewFlag ^\vxtBreakifyA
  \varindexNewFlag ,\vxtCommaB
  \varindexNewFlag ~\vxtSpaceB
  \varindexNewFlag +\vxtDashB
  \varindexNewFlag ;\vxtDotB
  \varindexNewFlag >\vxtOmitB
  \varindexNewFlag !\vxtTildeB
  \varindexNewFlag ?\vxtNoTildeB
  \varindexNewFlag p\vxtNopropB% Now obsolete; only for backward compatibility
  \varindexNewFlag '\vxtNopropB
  \varindexNewFlag =\vxtDashifyB
  \varindexNewFlag :\vxtOmitifyB
  \varindexMakeSplitExpand{+}{vxtSplitAtMagic}% + is the magic symbol
}

%    \end{macrocode}
%  \end{macro}
%
%  \begin{macro}{\varindextwo}
% The ``main'' demo macro which generates two "\index" commands
% whose format depends on the number of entries in \meta{main entries}
%    \begin{macrocode}
\def\varindextwo{%
  \let\vxtStarFormat\undefined
  \def\vxtexecute{\let\vxtmyFormat\varindexFormatDefault}%
  \ifcase\varindexCount
    % We have 0 entries (this cannot happen: {} generates one (empty) entry)
  \or% 1 entry:
    \def\vxtexecute{%
      \toolboxDef\vxtmyFormat{1}%
      \toolboxDef\vxtStarFormat{\varindexOutputString 1}}%
  \or% 2 entries:
    \def\vxtentryAa{1}%
    \def\vxtentryAb{2}%
    \def\vxtentryBa{2}%
    \def\vxtentryBb{1}%
    \let\vxtexecute\vxtMakemyFormat
  \or% 3 entries:
    \def\vxtentryAa{1}%
    \ifx\vxtNopropA\undefined
      \def\vxtentryAb{23}%
    \else
      \def\vxtentryAb{3}%
    \fi
    \def\vxtentryBa{3}%
    \ifx\vxtNopropB\undefined
      \def\vxtentryBb{12}%
    \else
      \def\vxtentryBb{1}%
    \fi
    \let\vxtexecute\vxtMakemyFormat
  \or% 4 entries:
    \def\vxtentryAa{1}%
    \ifx\vxtNopropA\undefined
      \def\vxtentryAb{24}%
    \else
      \def\vxtentryAb{4}%
    \fi
    \def\vxtentryBa{4}%
    \ifx\vxtNopropB\undefined
      \def\vxtentryBb{13}%
    \else
      \def\vxtentryBb{1}%
    \fi
    \let\vxtexecute\vxtMakemyFormat
  \else% more than 4 entries:
    \def\vxtexecute{\let\vxtmyFormat\varindexFormatDefault}%
  \fi
  \ifvarindexExplicitFormat% the optional format string was given:
    \expandafter\vxtSplitAtMagic\expandafter{\varindexFormat}\vxttmp\vxttmpa
    \ifx\vxttmpa\undefined
      % If the Magic symbol does not occur, we do nothing
    \else% Otherwise:
      % We generate our format and insert it at the place of the Magic symbol:
      \vxtexecute
      \toolboxDef\varindexFormat{\vxttmp\vxtmyFormat\vxttmpa}%
      \vxtAddmyStar
    \fi
  \else% without an optional format string, we generate our own format
    \vxtexecute
    \let\varindexFormat\vxtmyFormat
    \vxtAddmyStar
  \fi
}

%    \end{macrocode}
%  \end{macro}
%  Some flags used for the "varindextwo" macros.
%  \begin{macro}{\ifvarindextwoCommaTilde}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindextwoCommaTilde}
%    \end{macrocode}
%  \end{macro}
%  \begin{macro}{\ifvarindextwoDashTilde}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindextwoDashTilde}
%    \end{macrocode}
%  \end{macro}
%  \begin{macro}{\ifvarindextwoDashSpace}\nocomment
%    \begin{macrocode}
\toolboxNewiffalse{varindextwoDashSpace}
%    \end{macrocode}
%  \end{macro}
%
%  \begin{macro}{\vxtAddmyStar}
% A submacro needed by "\varindextwo"
%    \begin{macrocode}
\def\vxtAddmyStar{%
  \ifvarindexStar% With the * flag, and
    \ifx\vxtStarFormat\undefined\else% if we have our own StarFormat:
      \varindexStarfalse% <- No further prepending of `*' desired afterwards
      \expandafter\varindexSplitAtOutput\expandafter{\varindexFormat}%
        \vxttmp\vxttmp
      \ifx\vxttmp\undefined% Did no `*' occur in the format string yet?
        \ifx\varindexFormat\toolboxEmpty% Empty format means: Only the * part:
          \let\varindexFormat\vxtStarFormat
        \else%Normally, we prepend our *-format and a space
          \toolboxDef\varindexFormat{%
            \vxtStarFormat\varindexIndexString\varindexFormat}%
        \fi
      \fi
    \fi
  \fi
}

%    \end{macrocode}
%  \end{macro}
%
%  \begin{macro}{\vxtAddmyStar}
% A submacro needed by "\varindextwo"
%    \begin{macrocode}
\def\vxtMakemyFormat{%
  \let\vxtstarseparator\toolboxEmpty
  \ifx\vxtBreakifyA\undefined
    \ifx\vxtSpaceA\undefined
      \ifx\vxtDashA\undefined
        \ifx\vxtDotA\undefined
          \vxtExclamSeparator
        \else
          \let\vxtseparator\varindexOmitString
          \let\vxtstarseparator\vxtseparator
        \fi
      \else
        \let\vxtseparator\varindexDashString
        \let\vxtstarseparator\vxtseparator
      \fi
    \else
      \let\vxtseparator\toolboxEmpty
    \fi
  \else% With ^
    \ifx\vxtDotA\undefined
      \let\vxtstarseparator\varindexDashString
      \ifx\vxtDashA\undefined% -^
        \ifvarindextwoDashTilde
          \vxtExclamSeparator
          \toolboxAppend\vxtseparator\vxtstarseparator
        \else
           \toolboxDef\vxtseparator{\varindexExclamString\vxtstarseparator}%
        \fi
      \else% Plain ^
        \vxtExclamSeparator
        \ifvarindextwoDashSpace
          \toolboxDef\vxtseparator{\vxtstarseparator\vxtseparator}%
        \else
          \toolboxDef\vxtseparator{%
            \vxtstarseparator\vxtseparator\varindexOmitString}%
        \fi
      \fi
    \else% .^
      \let\vxtstarseparator\varindexOmitString
      \vxtExclamSeparator
      \toolboxAppend\vxtseparator\vxtstarseparator
    \fi
  \fi
  \ifx\vxtOmitA\undefined
    % We define now the string "\vxtentryAa?\vxtentryAb" where
    % ? is determined by the flags.
    \toolboxDef\vxtmyFormat{\vxtentryAa\vxtseparator\vxtentryAb}%
  \else
    \let\vxtmyFormat\toolboxEmpty
  \fi
  % The *-format is calculated similarly:
  \toolboxDef\vxtStarFormat{\varindexOutputString
    \vxtentryAa\vxtstarseparator\vxtentryAb}%
  \ifx\vxtOmitB\undefined
    \let\vxtappend\toolboxEmpty
    \ifx\vxtSpaceB\undefined
      \ifx\vxtDashB\undefined
        \ifx\vxtDotB\undefined
          \ifx\vxtCommaB\undefined
            \let\vxtseparator\varindexExclamString
            \let\vxtappend\varindexTildeAString
          \else
            \let\vxtseparator\varindexCommaString
            \ifvarindextwoCommaTilde
              \let\vxtappend\varindexTildeAString
            \fi
          \fi
          \ifx\vxtDashifyB\undefined
            \ifx\vxtOmitifyB\undefined\else
              \toolboxDef\vxtappend{\varindexOmitString\vxtappend}%
            \fi
          \else
            \toolboxDef\vxtappend{\varindexDashString\vxtappend}%
          \fi
        \else
          \let\vxtseparator\varindexOmitString
        \fi
      \else
        \let\vxtseparator\varindexDashString
      \fi
    \else
      \let\vxtseparator\toolboxEmpty
    \fi
    \ifx\vxtTildeB\undefined\else
      \let\vxtappend\varindexTildeAString
    \fi
    \ifx\vxtNoTildeB\undefined\else
      \let\vxtappend\toolboxEmpty
    \fi
    % In all cases, we prepend now the string "\vxtentryBa?\vxtentryBb??"
    % where ? and ?? are determined by the flags
    \ifx\vxtmyFormat\toolboxEmpty% If the previous string was empty: no space
      \toolboxDef\vxtmyFormat{\vxtentryBa\vxtseparator\vxtentryBb\vxtappend}%
    \else
      \toolboxAppend\vxtmyFormat{%
        \varindexIndexString\vxtentryBa\vxtseparator\vxtentryBb\vxtappend}%
    \fi
  \fi
}
%    \end{macrocode}
%  \end{macro}
%  \begin{macro}{\vxtAddmyStar}
% A submacro needed by "\vxtMakemyFormat"
%    \begin{macrocode}
\def\vxtExclamSeparator{%
  \ifx\vxtNoTildeA\undefined
    \toolboxDef\vxtseparator{\varindexExclamString\varindexTildeAString}%
  \else
    \let\vxtseparator\varindexExclamString
  \fi}

%    \end{macrocode}
%  \end{macro}
% \end{macro}
%
%    \begin{macrocode}
%% --------------------------------------------------------------------
%    \end{macrocode}
% \subsection{The main implementation of \texttt{varindex}}
%    \begin{macrocode}
%%
%% Now the implementation of \varindex.
%%
%    \end{macrocode}
%
%
% Counters and default variables:\par
%
% We cannot use "\toolboxIf" for "\newcount", because in plain \TeX\ and
% \LaTeX 2.09 the command "\newcount" is defined with the annoying "\outer"
% flag.
%    \begin{macrocode}

\ifx\varindexCount\undefined
  \newcount\varindexCount
\fi

%    \end{macrocode}
% \begin{macro}{\varindexPlaceholderAText}\nocomment
%    \begin{macrocode}
\toolboxIf\undefined\def\varindexPlaceholderAText{$\sim$}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexPlaceholderASort}\nocomment
%    \begin{macrocode}
\toolboxIf\undefined\let\varindexPlaceholderASort\toolboxEmpty
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexPlaceholderBText}\nocomment
%    \begin{macrocode}
\toolboxIf\undefined\def\varindexPlaceholderBText{$\approx$}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexPlaceholderBSort}\nocomment
%    \begin{macrocode}
\toolboxIf\undefined\let\varindexPlaceholderBSort\toolboxEmpty

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexUsePlaceholderA}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexUsePlaceholderA}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexUsePlaceholderB}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexUsePlaceholderB}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexLeadingPlaceholderA}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexLeadingPlaceholderA}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexLeadingPlaceholderB}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexLeadingPlaceholderB}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexCommaLeads}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexCommaLeads}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexCommaExpand}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexCommaExpand}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexPreviousAText}\nocomment
%    \begin{macrocode}
\toolboxNewiffalse{varindexPreviousAText}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexPreviousASort}\nocomment
%    \begin{macrocode}
\toolboxNewiffalse{varindexPreviousASort}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexPreviousBText}\nocomment
%    \begin{macrocode}
\toolboxNewiffalse{varindexPreviousBText}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexPreviousBSort}\nocomment
%    \begin{macrocode}
\toolboxNewiffalse{varindexPreviousBSort}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexEliminateInnerBrace}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexEliminateInnerBrace}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexEliminateOuterBrace}\nocomment
%    \begin{macrocode}
\toolboxNewiftrue{varindexEliminateOuterBrace}

%    \end{macrocode}
% \end{macro}
%
%
% The non-local macros (they should not exist already):
%
%    \begin{macrocode}
\ifx\newcommand\undefined\else
  \newcommand{\varindex}{}
  \newcommand{\varindexNewFlag}{}
  \newcommand{\varindexTestAndExec}{}
  \newcommand{\varindexSkipAndExec}{}
  \newcommand{\varindexSkipDefault}{}
  \newcommand{\varindexSkipTricky}{}
  \newcommand{\varindexMakeVarSplit}{}
  \newcommand{\varindexMakeSplitExpand}{}
  \newcommand{\varindexSetWordString}{}
  \newcommand{\varindexSetSortString}{}
  \newcommand{\varindexSetPlainString}{}
  \newcommand{\varindexSetIndexString}{}
  \newcommand{\varindexSetOutputString}{}
  \newcommand{\varindexSetTildeAString}{}
  \newcommand{\varindexSetTildeBString}{}
  \newcommand{\varindexSetSpaceString}{}
  \newcommand{\varindexSetSpaceTokString}{}
  \newcommand{\varindexSetOmitString}{}
  \newcommand{\varindexSetCommaString}{}
  \newcommand{\varindexSetDashString}{}
  \newcommand{\varindexSetExclamString}{}
  \newcommand{\varindexSetStringForNr}{}
  \newcommand{\varindexedef}{}
  \newcommand{\varindexArgumentSpace}{}
  \newcommand{\varindexNewFormat}{}
  \newcommand{\varindexTokensOut}{}
  \newcommand{\varindexTokensOutExpand}{}
  \newcommand{\varindexAddNumber}{}
  \newcommand{\varindexAddComma}{}
  \newcommand{\varindexAddTildeA}{}
  \newcommand{\varindexAddTildeB}{}
  \newcommand{\varindexAddExclam}{}
  \newcommand{\varindexAddSpace}{}
  \newcommand{\varindexAddSpTok}{}
  \newcommand{\varindexAddOmit}{}
  \newcommand{\varindexAddDash}{}
  \newcommand{\varindexDoConvert}{}
  \newcommand{\varindexNoConvert}{}
%% Obsolete names (only provided for backward compatibility):
  \newcommand{\varindexdef}{}
  \newcommand{\varindexspace}{}
  \newcommand{\varindexDropBrace}{}
  \newcommand{\varindexMakeSplit}{}
  \newcommand{\varindexLet}{}
  \newcommand{\varindexWithNr}{}
\fi
\let\varindexdef\toolboxDef
\let\varindexspace\toolboxSpace
\let\varindexDropBrace\toolboxDropBrace
\let\varindexMakeSplit\toolboxMakeSplit
\let\varindexLet\toolboxLet
\let\varindexWithNr\toolboxWithNr

%    \end{macrocode}
%
% \begin{macro}{\varindex}
%
% The essential ``flow'' is easy:
%
% "\varindex" reads the arguments (using lots of auxiliary macros)
% and then calls "\vx@Main".
%
%    \begin{macrocode}
\def\varindex{\begingroup
%    \end{macrocode}
% Before "toolbox" 3.1 the following "\if"'s could not be made local.
% The reason is that in plain \TeX\ and \LaTeX 2.09 the command "\newif" is
% "\outer". Due to this stupid restriction, we had to define them before.
% "toolbox" 3.1 has now implemented its own "\newif" command for this
% reason.\par\bigskip
%
% "\if"'s for scanning of the parameters:
%
% \begin{macro}{\ifvarindexStar}\nocomment
%    \begin{macrocode}
  \toolboxNewifFalse{varindexStar}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexExplicitFormat}\nocomment
%    \begin{macrocode}
  \toolboxNewifFalse{varindexExplicitFormat}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvx@atespace}
% is true if a space was ate during skipping
%    \begin{macrocode}
  \toolboxNewifFalse{vx@atespace}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvx@savespace}
% is true if we must ``put back'' an eaten space
%    \begin{macrocode}
  \toolboxNewifFalse{vx@savespace}%
%    \end{macrocode}
% \end{macro}
% "\if"'s for "\vx@main" and its submacros:
% \begin{macro}{\ifvx@ffound}\nocomment
%    \begin{macrocode}
  \toolboxNewifFalse{vx@ffound}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexIndexMode}\nocomment
% "\ifvarindexIndexMode" is true if the currently parsed format corresponds
% to an index entry (in contrast to the output into the running text).
%    \begin{macrocode}
  \toolboxNewifTrue{varindexIndexMode}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexSpaceText}\nocomment
%    \begin{macrocode}
  \toolboxNewifFalse{varindexSpaceText}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvarindexSpaceSort}\nocomment
%    \begin{macrocode}
  \toolboxNewifFalse{varindexSpaceSort}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvx@firstiter}\nocomment
%    \begin{macrocode}
  \toolboxNewifTrue{vx@firstiter}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvx@tildeLead}
% Decides whether we have a leading tilde
%    \begin{macrocode}
  \toolboxNewifFalse{vx@tildeLead}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvx@commaTextLead}
% True if "\ifvarindexCommaExpand" is true and we have no token after comma.
%    \begin{macrocode}
  \toolboxNewifFalse{vx@commaTextLead}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\ifvx@commaSortLead}\nocomment
% True if "\ifvarindexCommaExpand" is true and we have no token after comma.
%    \begin{macrocode}
  \toolboxNewifFalse{vx@commaSortLead}%
%    \end{macrocode}
% \end{macro}
% We use "\varindexCount" also for generic flags to save counters.
%    \begin{macrocode}
  \varindexCount=0\relax
  \let\varindexStyle\toolboxEmpty
  \let\varindexAppend\toolboxEmpty
  \let\vx@text\toolboxEmpty
  \let\vx@mainprogram\undefined
  \let\vx@formatA\undefined
  \let\vx@formatAstar\undefined
  \let\varindexFormat\undefined
  \let\varindexFormatDefault\undefined
  \varindexNewFlag 1\varindexOneEntry
  \varindexNewFlag *\varindexStarPos
  \varindexTestAndExec(%optional formatA)
    \vx@ReadFormatA%else
    \vx@TestTextStyle}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadFormatA}\nocomment
%    \begin{macrocode}
\def\vx@ReadFormatA(#1){%
  \def\vx@formatA{#1}%
  \varindexTestAndExec(%optional formatAstar)
    \vx@ReadFormatAstar%else
    \vx@SkipAndReadScan}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadFormatAstar}\nocomment
%    \begin{macrocode}
\def\vx@ReadFormatAstar(#1){%
  \def\vx@formatAstar{#1}\vx@SkipAndReadScan}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@SkipAndReadScan}\nocomment
%    \begin{macrocode}
\def\vx@SkipAndReadScan{\varindexSkipAndExec\vx@ReadScan}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadScan}\nocomment
%    \begin{macrocode}
\def\vx@ReadScan#1{#1\toolboxEmpty\varindexSkipAndExec\vx@ReadProgram}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadProgram}\nocomment
%    \begin{macrocode}
\def\vx@ReadProgram#1{\def\vx@mainprogram{#1}\vx@TestTextStyle}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@TestTextStyle}\nocomment
%    \begin{macrocode}
\def\vx@TestTextStyle{%
  \varindexTestAndExec[%textstyle]
    \vx@ReadTextStyle%else
    \vx@TestFlags}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadTextStyle}\nocomment
%    \begin{macrocode}
\def\vx@ReadTextStyle[#1]{%
  \def\varindexStyle{#1}%
  \varindexTestAndExec[%pagestyleA]
    \vx@ReadPageStyleA%else
    \vx@TestFlags}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadTextStyleA}\nocomment
%    \begin{macrocode}
\def\vx@ReadPageStyleA[#1]{\def\varindexAppend{#1}\vx@TestFlags}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@TestFlags}
% This macro starts the flag read loop.
%    \begin{macrocode}
\def\vx@TestFlags{%
%    \end{macrocode}
% In this loop, we use "\varindexCount" for the flag position.
%    \begin{macrocode}
  \varindexCount=0\relax
  \varindexEndOfFlag}

%    \end{macrocode}
% "\varindexEndOfFlag" is the value of the `first to call' TestFlag...
% (which is actually the last TestFlag... defined)
% Let us first initialize it to TestFlag0:
%    \begin{macrocode}
\varindexCount=0\relax
\expandafter
\def\expandafter\varindexEndOfFlag\expandafter{%
  \csname vx@TestFlag\the\varindexCount\endcsname}
%    \end{macrocode}
%Initialize the `last' "\vx@TestFlag0" as a macro with the meaning
%``start to read \meta{main entries}''.
%    \begin{macrocode}
\expandafter
\def\varindexEndOfFlag{\varindexSkipAndExec\vx@ReadMain}%
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@TestFlag0}
% "\vx@TestFlag0" is the first macro generated by "\varindexNewFlag".
%
%  \begin{macro}{\varindexNewFlag}
%  "\varindexNewFlag" calls "\vx@MakeNewFlag" with appropriate arguments:
%
%    \begin{macrocode}
\def\varindexNewFlag#1#2{%
  \let#2\undefined
  \expandafter
  \def\expandafter\vx@tmp\expandafter{%
    \csname vx@TestFlag\the\varindexCount\endcsname #1#2}%
  \advance\varindexCount by 1\relax
%    \end{macrocode}
% A bug in \TeX? Without this "\relax", the next "\expandafter"
% is executed \emph{before} the counter is actually increased.
%    \begin{macrocode}
  \expandafter
    \toolboxSurround\expandafter{%
      \csname vx@TestFlag\the\varindexCount\expandafter\endcsname
      \csname vx@ReadFlag\the\varindexCount\endcsname}{}\vx@tmp
  \varindexTestAndExec[%program]
    \vx@ReadFlagProgram%else
    {\vx@MakeFlagProgram\varindexEndOfFlag}%
}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\vx@ReadFlagProgram}\nocomment
%    \begin{macrocode}
\def\vx@ReadFlagProgram[#1]{\vx@MakeFlagProgram{#1}}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@MakeFlagProgram}\nocomment
%    \begin{macrocode}
\def\vx@MakeFlagProgram#1{\expandafter\vx@MakeNewFlag\vx@tmp{#1}}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@MakeNewFlag}
% Usage:
% "\vx@MakeNewFlag\TestMacro\ReadMacro\PrevMacro\token\pos"\bracemeta{program}
%    \begin{macrocode}
\def\vx@MakeNewFlag#1#2#3#4#5#6{%
  \def\varindexEndOfFlag{#1}%
  \def#1{\varindexTestAndExec#4{#2}{#3}}%
  \def#2#4{\expandafter
    \def\expandafter#5\expandafter{\the\varindexCount}%
    \advance\varindexCount by 1\relax
    \varindexSkipAndExec{#6}}%
}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadMain}\nocomment
%    \begin{macrocode}
\long\def\vx@ReadMain#1{\def\vx@text{#1}%
  \ifx\vx@formatA\undefined
    \def\vx@exec{\varindexSkipAndExec\vx@ReadFormatB}%
  \else
    \def\vx@exec{%
      \varindexTestAndExec[%FormatB]
        \vx@ReadOptFormatB%else
        {\ifvx@atespace\vx@savespacetrue\fi\vx@TestAppend}}%
  \fi
  \vx@exec}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadFormatB}\nocomment
%    \begin{macrocode}
\def\vx@ReadFormatB#1{\def\varindexFormat{#1}\vx@TestAppend}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadOptFormatB}\nocomment
%    \begin{macrocode}
\def\vx@ReadOptFormatB[#1]{\def\varindexFormat{#1}\vx@TestAppend}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@TestAppend}\nocomment
%    \begin{macrocode}
\def\vx@TestAppend{%
  \varindexTestAndExec[%Append]
    {\vx@savespacefalse\vx@ReadAppend}%else
    {\ifvx@atespace\vx@savespacetrue\fi\vx@Main}}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ReadAppend}\nocomment
%    \begin{macrocode}
\def\vx@ReadAppend[#1]{\def\varindexAppend{#1}\vx@Main}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\varindexTestAndExec}
% "\varindexTestAndExec\Token"\bracemeta{ifpart}\bracemeta{elsepart}
% at the end of a macro:
% Steps forward in the token list as long as there are spaces.
% If a space is eaten, "\vx@atespace" is set true.
% The next (non-space) token is put into "\varindexNextToken".
% If this token is "\Token", then \meta{ifpart} is executed, otherwise
% \meta{elsepart}.
%    \begin{macrocode}
\def\varindexTestAndExec#1#2#3{%
  \varindexSkipAndExec{%
    \ifx\varindexNextToken#1%
      \def\vx@exec{#2}%
    \else
      \def\vx@exec{#3}%
    \fi
    \vx@exec}}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexSkipAndExec}
% "\varindexSkipAndExec"\bracemeta{commands}  at the end of a macro:
% Steps forward in the token list as long as there are spaces.
% The next (non-space) token is put into "\varindexNextToken".
% Then commands are executed.
%    \begin{macrocode}
\def\varindexSkipAndExec#1{%
  \vx@atespacefalse
  \def\vx@skipexec{#1}%
  \varindexArgumentSpace}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexSkipDefault}\nocomment
%    \begin{macrocode}
\def\varindexSkipDefault{\futurelet\varindexNextToken\vx@skipexec}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexSkipTrickery}\nocomment
%    \begin{macrocode}
\def\varindexSkipTricky{%
  \vx@MakeSpaceOther\futurelet\varindexNextToken\vx@SkipLoopTest}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexSkipLoopTest}\nocomment
%    \begin{macrocode}
\def\vx@SkipLoopTest{%
  \vx@MakeSpaceSpace
  \ifx\varindexNextToken\vx@spaceothertok
    \def\vx@skipnext{\vx@gobblespace a}%
  \else
    \let\vx@skipnext\vx@skipexec
  \fi
  \vx@skipnext}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@gobblespace}\nocomment
%    \begin{macrocode}
\def\vx@gobblespace a#1{\vx@atespacetrue\varindexSkipTricky}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@Main}
% This macro is called after all arguments are read into appropriate macros.
% Now the macros will be parsed (possibly using the argument
% \meta{mainprogram}), and then the output is produced.
% Finally, the group is closed and a possible ``eaten space'' is put back.
%    \begin{macrocode}
\def\vx@Main{%
  \ifx\varindexStarPos\undefined
    \varindexStarfalse
  \else
    \varindexStartrue
  \fi
  \ifx\vx@formatA\undefined
%    \end{macrocode}
% The default format is empty if no default format is given
%    \begin{macrocode}
    \let\varindexFormatDefault\toolboxEmpty
  \else
    \ifvarindexStar
      \ifx\vx@formatAstar\undefined
        \let\varindexFormatDefault\vx@formatA
      \else
        \let\varindexFormatDefault\vx@formatAstar
      \fi
    \else
      \let\varindexFormatDefault\vx@formatA
    \fi
  \fi
%    \end{macrocode}
% Now we parse \meta{main entries}:
%    \begin{macrocode}
  \toolboxDef\vx@toparse{\vx@text\vx@word\vx@endparse}%
  \ifx\varindexOneEntry\undefined
    \varindexCount=0\relax
    \let\vx@NextLoop\vx@TextLoop
    \vx@TextLoop
  \else
    \varindexCount=1\relax
    \ifvarindexEliminateOuterBrace
      \toolboxDropBrace\vx@text
    \fi
    \expandafter\vx@TextEntry\expandafter{\vx@text}%
  \fi
  \expandafter
  \def\expandafter\varindexLastEntry\expandafter{\the\varindexCount}%
%    \end{macrocode}
% Now we expand the format(s):
%    \begin{macrocode}
  \varindexedef\varindexFormatDefault{\varindexFormatDefault}%
  \ifx\varindexFormat\undefined
    \varindexExplicitFormatfalse
    \let\varindexFormat\varindexFormatDefault
  \else
    \varindexExplicitFormattrue
    \varindexedef\varindexFormat{\varindexFormat}%
  \fi
%    \end{macrocode}
% Now we execute \meta{mainprogram}, and expand "\varindexFormat" again:
%    \begin{macrocode}
  \ifx\vx@mainprogram\undefined\else
    \vx@mainprogram
    \varindexedef\varindexFormat{\varindexFormat}%
  \fi
%    \end{macrocode}
% Still modify "\varindexFormat":
%    \begin{macrocode}
  \ifvarindexStar
%    \end{macrocode}
% Prepend `"*"' resp. `"*"\myspace'
%    \begin{macrocode}
    \expandafter\varindexSplitAtOutput\expandafter{\varindexFormat}\vx@fa\vx@fb
    \ifx\vx@fb\undefined
%    \end{macrocode}
% The `"*"' does not occur in "\varindexFormat"
%    \begin{macrocode}
      \expandafter\varindexSplitAtIndex\expandafter{\varindexFormat}%
        \vx@fa\vx@fb
      \ifx\vx@fb\undefined
%    \end{macrocode}
% We did not find a `space'
%    \begin{macrocode}
        \let\vx@fa\undefined
      \fi
      \ifx\vx@fa\toolboxEmpty
%    \end{macrocode}
% Did we find a `space' immediately at the start?
% Then prepend only `"*"'
%    \begin{macrocode}
        \toolboxDef\varindexFormat{\varindexOutputString\varindexFormat}%
      \else
%    \end{macrocode}
%         prepend `"*"\myspace'
%    \begin{macrocode}
        \toolboxDef\varindexFormat{%
          \varindexOutputString\varindexIndexString\varindexFormat}%
      \fi
    \fi
  \fi
  \ifx\varindexFormat\varindexOutputString
%    \end{macrocode}
% Replace `"*"' by `"*"\myspace'
%    \begin{macrocode}
    \toolboxDef\varindexFormat{\varindexOutputString\varindexIndexString}%
  \fi
%    \end{macrocode}
% Now we interpret "\varindexFormat" and produce the output:
%    \begin{macrocode}
  \ifx\varindexFormat\toolboxEmpty
    \vx@FormatAll
  \else
    \toolboxDef\vx@toparse{\varindexFormat\varindexIndexString\vx@endparse}%
    \let\vx@NextLoop\vx@FormLoop
    \vx@FormLoop
  \fi
  \ifvx@savespace\aftergroup\space\fi
\endgroup}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@endparse}
% "\vx@endparse" is the end of the string to parse:
%    \begin{macrocode}
\def\vx@endparse{\vx@endparse}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@TextLoop}
% This is the (recursive) macro which parses \meta{main entries}.
% For each entry (separated by "\vx@word"), possibly with braces
% eliminated, the macro "\vx@TextEntry" is called.
%
%    \begin{macrocode}
\def\vx@TextLoop{%
  \ifx\vx@toparse\vx@endparse
    \let\vx@NextLoop\toolboxEmpty
  \else
    \advance\varindexCount by 1\relax
    \expandafter\vx@SplitAtWord\expandafter{\vx@toparse}%
      \vx@textentry\vx@toparse
    \ifvarindexEliminateOuterBrace
      \toolboxDropBrace\vx@textentry
    \fi
    \expandafter\vx@TextEntry\expandafter{\vx@textentry}%
  \fi
  \vx@NextLoop}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@FormLoop}
% This is the (recursive) macro which parses the \meta{format}.
% For each entry (separated by space or by "*") the function
% "\vx@Format" is called with "\varindexIndexMode" set correspondingly.
%    \begin{macrocode}
\def\vx@FormLoop{%
  \ifx\vx@toparse\vx@endparse
    \let\vx@NextLoop\toolboxEmpty
  \else
    \expandafter\varindexSplitAtIndex\expandafter{\vx@toparse}\vx@fa\vx@fb
    \expandafter\varindexSplitAtOutput\expandafter{\vx@fa}\vx@fa\vx@fc
    \ifx\vx@fa\toolboxEmpty
%    \end{macrocode}
% Have we reached an empty entry in the format string?
%    \begin{macrocode}
      \ifvx@firstiter%%       In the first iteration,
%%      \ifx\vx@fc\undefined% if 'space' is the very first sign,
%%        \vx@FormatAll%      we output everything
%%      \fi
      \else
%    \end{macrocode}
% Later, we always output everything.
%    \begin{macrocode}
        \vx@FormatAll
      \fi
    \else
%    \end{macrocode}
% Usually, we just work with our current entry
%    \begin{macrocode}
      \expandafter\vx@Format\expandafter{\vx@fa}%
    \fi
    \vx@firstiterfalse
    \ifx\vx@fc\undefined
%    \end{macrocode}
% If the first separating sign is 'space':
%    \begin{macrocode}
      \varindexIndexModetrue
      \let\vx@toparse\vx@fb
    \else
%    \end{macrocode}
% If the first separating sign is "*":
%    \begin{macrocode}
      \varindexIndexModefalse
      \expandafter
      \varindexSplitAtOutput\expandafter{\vx@toparse}\vx@fa\vx@toparse
    \fi
  \fi
  \vx@NextLoop}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@FormatAll}
% This macro is as "\vx@Format", just that everything is output:
%    \begin{macrocode}
\def\vx@FormatAll{%
  \vx@FormatInit
  \expandafter
  \def\expandafter\vx@Count\expandafter{\the\varindexCount}\relax
  \varindexCount=0\relax
  \let\vx@NextFormatAllLoop\vx@FormatAllLoop
  \vx@FormatAllLoop
  \expandafter\varindexCount\expandafter=\vx@Count\relax
  \vx@Write}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@FormatAllLoop}\nocomment
%    \begin{macrocode}
\def\vx@FormatAllLoop{%
  \advance\varindexCount by 1\relax
  \expandafter
  \varindexAddNumber\expandafter{\the\varindexCount}%
  \expandafter
  \ifnum\varindexLastEntry=\varindexCount
    \let\vx@NextFormatAllLoop\toolboxEmpty
  \fi
  \vx@NextFormatAllLoop}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@AppendwithCount}
% The content of the variable "#2\the\varindexCount" is appended to "#1".
% If both, "#1" and "#2\the\varindexCount" are nonempty, a space
% is inserted between.
%    \begin{macrocode}
\def\vx@AppendWithCount#1#2{%
  \toolboxLet\vx@tmp{#2\the\varindexCount}%
  \ifx#1\toolboxEmpty
    \let#1\vx@tmp
  \else
    \ifx\vx@tmp\toolboxEmpty\else\toolboxAppend#1{\toolboxSpace\vx@tmp}\fi
  \fi}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@Format}
% This is the macro which parses the format string "#1" for one single
% output, either into index or into running text, depending on
% "\ifvarindexIndexMode".
%
%    \begin{macrocode}
\def\vx@Format#1{%
  \vx@FormatInit
  \def\vx@fparse{#1}%
  \toolboxAppend\vx@fparse\vx@endparse
  \let\vx@FormatNextLoop\vx@FormatLoop
  \vx@FormatLoop
  \vx@Write
}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@FormatInit}\nocomment
%    \begin{macrocode}
\def\vx@FormatInit{%
  \let\vx@outputlist\toolboxEmpty
  \let\vx@textlist\toolboxEmpty
  \let\vx@sortlist\toolboxEmpty
  \let\vx@textlistBefore\toolboxEmpty
  \let\vx@sortlistBefore\toolboxEmpty
  \let\vx@tildeAText\varindexPlaceholderAText
  \let\vx@tildeASort\varindexPlaceholderASort
  \let\vx@tildeBText\varindexPlaceholderBText
  \let\vx@tildeBSort\varindexPlaceholderBSort
  \vx@tildeLeadtrue
  \ifvarindexCommaExpand
    \vx@commaTextLeadtrue
    \vx@commaSortLeadtrue
  \else
    \vx@commaTextLeadfalse
    \vx@commaSortLeadfalse
  \fi
  \varindexNoNextSpace
}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@FormatLoop}
% "\vx@FormatLoop" is the main loop which parses one (space-separated) entry
% in the format string. The essential part here is the command
% "\vx@HandleFormat" which is described below.
% Essentially, "\vx@HandleFormat" is a list of subsequent calls to
% "\vx@condexec"
%    \begin{macrocode}
\def\vx@FormatLoop{%
  \vx@ffoundfalse
  \ifx\vx@fparse\vx@endparse
    \let\vx@FormatNextLoop\toolboxEmpty
  \else
    \vx@HandleFormat
    \ifvx@ffound\else
      \expandafter\vx@ErrorFormat\vx@fparse
      \let\vx@FormatNextLoop\toolboxEmpty
    \fi
  \fi
  \vx@FormatNextLoop
}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@condexec}\nocomment
%    \begin{macrocode}
\def\vx@condexec#1#2{\ifvx@ffound\else
  \expandafter #1\expandafter{\vx@fparse}\vx@fparseA\vx@fparseB
  \ifx\vx@fparseB\undefined\else
    \ifx\vx@fparseA\toolboxEmpty
      \let\vx@fparse\vx@fparseB
      \vx@ffoundtrue
      #2%
    \fi
  \fi
\fi}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@HandleFormat}
%  "\vx@HandleFormat" is a list of commands of the form
%  "\vx@condexec"\bracemeta{splitcommand}\bracemeta{action}.
%  The idea is that if \meta{splitcommand} splits at the `current' token
%  of the format string then \meta{action} is executed.
%  For details, see "\vx@condexec".
%  The list "\vx@Handle" format is generated by "\varindexNewFormat".
%    \begin{macrocode}
\let\vx@HandleFormat\toolboxEmpty

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexNewFormat}
%    \begin{macrocode}
\def\varindexNewFormat#1#2{%
  \toolboxSurround{\vx@condexec{#1}{#2}}{}\vx@HandleFormat}

%    \end{macrocode}
% \end{macro}
% Now we initialize the default actions for the format string. However,
% it is necessary to postpone the expansion of the "\vx@NewStdFormat" commands
% (which call "\varindexNewFormat") until "\varindexOutExclam" etc.\ have
% been initialized: In fact, the macro "\vx@NewStdFormat" uses "\csname"
% and thus implicitly sets those variables to "\relax" which conflicts with
% our initialization, because we define "\varindexOutExclam" and friends
% only if they are "\undefined" (so that we do not override previous user
% definitions of these macros).\par
%    \begin{macrocode}
%% This macro is executed near the end of this file.
\def\varindexAddOmit{% The name is a fake: The macro redefines this macro
%    \end{macrocode}
% \begin{macro}{\varindexAddSpace}
% "\varindexAddSpace" is made by a scheme from "\vx@NewStdFormat"
%    \begin{macrocode}
  \vx@NewStdFormat\varindexSplitAtSpace{Space}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddSpTok}
% "\varindexAddSpTok" is made by a scheme from "\vx@NewStdFormat"
%    \begin{macrocode}
  \vx@NewStdFormat\varindexSplitAtSpaceTok{SpTok}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddOmit}
% "\varindexAddOmit" is made by a scheme from "\vx@NewStdFormat"
%    \begin{macrocode}
  \vx@NewStdFormat\varindexSplitAtOmit{Omit}%
%    \end{macrocode}
% \end{macro}
%    \begin{macrocode}
  \varindexNewFormat\varindexSplitSpace{}% Spaces in inner format are ignored
  \varindexNewFormat\varindexSplitAtDash\varindexAddDash
  \varindexNewFormat\varindexSplitAtTildeB\varindexAddTildeB
  \varindexNewFormat\varindexSplitAtTildeA\varindexAddTildeA
  \varindexNewFormat\varindexSplitAtComma\varindexAddComma
  \varindexNewFormat\varindexSplitAtExclam\varindexAddExclam
  \let\vx@NewStdFormat\undefined% We do not need this generic macro anymore
}

%    \end{macrocode}
% \begin{macro}{\vx@NewStdFormat}
% \begin{myquote}
% "\vx@NewStdFormat\"\meta{splitcmd}\bracemeta{name}
% \end{myquote}
% calls
% \begin{myquote}
% "\varindexNewFormat\"\meta{splitcmd}"\varindexAdd"\meta{name}
% \end{myquote}
% and defines the corresponding macro "\varindexAdd"\meta{name} by
% \begin{myquote}
% "\def\varindexAdd"\meta{name}"{\varindexPreceeds"\meta{name}\\
% "  \varindexTokensOutExpand"\\
% "    \varindexOutText"\meta{name}\\
% "    \varindexOutSort"\meta{name}\\
% "  \varindexFollows"\meta{name}"}"
% \end{myquote}
%    \begin{macrocode}
\def\vx@NewStdFormat#1#2{%
  \expandafter
  \varindexNewFormat\expandafter#1\csname varindexAdd#2\endcsname
  \expandafter
  \def\expandafter\vx@tmp\expandafter{%
    \csname varindexPreceeds#2\expandafter\endcsname
    \expandafter
    \varindexTokensOutExpand
      \csname varindexOutText#2\expandafter\endcsname
      \csname varindexOutSort#2\expandafter\endcsname
    \csname varindexFollows#2\endcsname}
  \expandafter\let\csname varindexAdd#2\endcsname\vx@tmp}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddNumber}\nocomment
%    \begin{macrocode}
\def\varindexAddNumber#1{%
  \ifnum\varindexCount<#1%
    \expandafter\vx@ErrorNumber\expandafter{\the\varindexCount}{#1}%
  \fi
  \ifvarindexIndexMode
    \toolboxLet\vx@adda{varindexEntry#1}%
    \toolboxLet\vx@addb{varindexSort#1}%
  \else
    \toolboxLet\vx@adda{varindexPlain#1}%
  \fi
  \vx@Addit\varindexSpaceTexttrue\varindexSpaceSorttrue}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddDash}
%    \begin{macrocode}
\def\varindexAddDash{%
  \varindexPreceedsDash
  \ifx\vx@sortlist\toolboxEmpty
    \varindexTokensOutExpand\varindexOutTextDash\varindexOutSortDashBeg
  \else
    \varindexTokensOutExpand\varindexOutTextDash\varindexOutSortDash
  \fi
  \varindexFollowsDash}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddComma}\nocomment
%    \begin{macrocode}
\def\varindexAddComma{%
  \ifvarindexCommaLeads
    \vx@tildeLeadtrue
  \fi
  \ifvarindexCommaExpand
    \let\vx@tildeAText\vx@textlist
    \let\vx@tildeBText\vx@textlist
    \ifvarindexIndexMode
      \let\vx@tildeASort\vx@sortlist
      \let\vx@tildeBSort\vx@sortlist
    \fi
  \fi
  \varindexPreceedsComma
  \varindexTokensOutExpand\varindexOutTextComma\varindexOutSortComma
  \vx@flushToBefore
  \ifvarindexCommaExpand
    \vx@commaTextLeadtrue\vx@commaSortLeadtrue
  \fi
  \varindexFollowsComma}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@flushToBefore}\nocomment
%    \begin{macrocode}
\def\vx@flushToBefore{%
  \ifvarindexCommaExpand
    \toolboxAppend\vx@textlistBefore\vx@textlist
    \let\vx@textlist\toolboxEmpty
    \ifvarindexIndexMode
      \toolboxAppend\vx@sortlistBefore\vx@sortlist
      \let\vx@sortlist\toolboxEmpty
    \fi
  \fi}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddTildeA}\nocomment
%    \begin{macrocode}
\def\varindexAddTildeA{%
  \ifvarindexUsePlaceholderA
    \ifvx@tildeLead
      \ifvarindexLeadingPlaceholderA
        \vx@AddingTildeA
      \fi
    \else
      \vx@AddingTildeA
    \fi
  \fi}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@AddingTildeA}\nocomment
%    \begin{macrocode}
\def\vx@AddingTildeA{%
  \ifvarindexPreviousAText
    \let\vx@adda\vx@tildeAText
  \else
    \let\vx@adda\varindexPlaceholderAText
  \fi
  \ifvarindexPreviousASort
    \let\vx@addb\vx@tildeASort
  \else
    \let\vx@addb\varindexPlaceholderASort
  \fi
  \vx@Addit\varindexSpaceTexttrue\varindexSpaceSorttrue}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@AddTildeB}\nocomment
%    \begin{macrocode}
\def\varindexAddTildeB{%
  \ifvarindexUsePlaceholderB
    \ifvx@tildeLead
      \ifvarindexLeadingPlaceholderB
        \vx@AddingTildeB
      \fi
    \else
      \vx@AddingTildeB
    \fi
  \fi}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@AddingTildeB}\nocomment
%    \begin{macrocode}
\def\vx@AddingTildeB{%
  \ifvarindexPreviousBText
    \let\vx@adda\vx@tildeBText
  \else
    \let\vx@adda\varindexPlaceholderBText
  \fi
  \ifvarindexPreviousBSort
    \let\vx@addb\vx@tildeBSort
  \else
    \let\vx@addb\varindexPlaceholderBSort
  \fi
  \vx@Addit\varindexSpaceTexttrue\varindexSpaceSorttrue}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexNextSpace}\nocomment
%    \begin{macrocode}
\def\varindexNextSpace{\varindexSpaceTexttrue\varindexSpaceSorttrue}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexNoNextSpace}\nocomment
%    \begin{macrocode}
\def\varindexNoNextSpace{\varindexSpaceTextfalse\varindexSpaceSortfalse}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexTokensOut}\nocomment
%    \begin{macrocode}
\long\def\varindexTokensOut#1#2{\def\vx@adda{#1}\def\vx@addb{#2}\vx@Addit{}{}}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexTokensOutExpand}\nocomment
%    \begin{macrocode}
\def\varindexTokensOutExpand#1#2{\let\vx@adda#1\let\vx@addb#2\vx@Addit{}{}}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@Addit}
% \begin{myquote}
% "\vx@Addit"\bracemeta{command A}\bracemeta{command B}
% \end{myquote}
% Does the following if "\vx@adda"/"\vx@addb" is nonempty:
% \begin{myquote}
% Add (space +) "\vx@adda"/"\vx@addb" to "\vx@textlist"/"\vx@sortlist"\\
% (Shift space to "\vx@...listBefore" if "\ifvx@comma...Lead" is true).\\
% "\vx@tildeLeadfalse" (only if "\vx@adda" is nonempty)\\
% "\varindexSpace...false"\\
% "\vx@comma...Leadfalse"\\
% Execute \meta{command A/B}
% \end{myquote}
%    \begin{macrocode}
\def\vx@Addit#1#2{%
  \ifx\vx@adda\toolboxEmpty\else
    \vx@tildeLeadfalse
    \ifvarindexSpaceText
      \varindexSpaceTextfalse
      \ifvx@commaTextLead
        \toolboxAppend\vx@textlistBefore\toolboxSpace
        \let\vx@textlist\vx@adda
      \else
        \toolboxAppend\vx@textlist{\toolboxSpace\vx@adda}%
      \fi
    \else
      \toolboxAppend\vx@textlist\vx@adda
    \fi
    \vx@commaTextLeadfalse
    #1%
  \fi
  \ifvarindexIndexMode
    \ifx\vx@addb\toolboxEmpty\else
      \ifvarindexSpaceSort
        \varindexSpaceSortfalse
        \ifvx@commaSortLead
          \toolboxAppend\vx@sortlistBefore\toolboxSpace
          \let\vx@sortlist\vx@addb
        \else
          \toolboxAppend\vx@sortlist{\toolboxSpace\vx@addb}%
        \fi
      \else
        \toolboxAppend\vx@sortlist\vx@addb
      \fi
      \vx@commaSortLeadfalse
      #2%
    \fi
  \fi}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexAddExclam}\nocomment
%    \begin{macrocode}
\def\varindexAddExclam{%
  \ifvarindexIndexMode
    \vx@Flush
    \toolboxAppend\vx@outputlist\varindexOutExclam
  \else
    \errmessage{\string\varindex: %
      Subitem can only be used for index, not for running text}%
  \fi
  \varindexFollowsExclam
}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ErrorFormat}\nocomment
% "\vx@ErrorFormat" outputs the format until the content of "\vx@endparse" is
% reached.
%    \begin{macrocode}
\expandafter
\def\expandafter\vx@ErrorFormat\expandafter#\expandafter 1\vx@endparse
  {\errmessage{\string\varindex: Error in format `#1'}}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@ErrorNumber}\nocomment
%    \begin{macrocode}
\def\vx@ErrorNumber#1#2{\errmessage{\string\varindex: %
  Entry `#2' does not exist (only #1 entries)}}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@Write}
% The main output macro (either into "\index" or into running text):
%
%    \begin{macrocode}
\def\vx@Write{%
  \vx@Flush
  \ifvarindexIndexMode
    \toolboxAppend\vx@outputlist\varindexAppend
    \vx@MakeHarmless\vx@outputlist
    \expandafter
    \varindexIndex\expandafter{\vx@outputlist}%
  \else
    \expandafter
    \varindexStyle\expandafter{\vx@outputlist}%
  \fi}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@Flush}
% Put the text buffered in "\vx@textlistBefore\vx@textlist" and
% "\vx@sortlistBefore\vx@sortlist" to "\vx@outputlist" in the way appropriate
% fot the output, and delete the buffers.
% Moreover, prepare "\vx@tilde?Text" and friends correspondingly.
%
%    \begin{macrocode}
\def\vx@Flush{%
  \vx@tildeLeadtrue
  \vx@commaTextLeadtrue
  \vx@commaSortLeadtrue
  \let\vx@tildeAText\vx@textlist
  \let\vx@tildeBText\vx@textlist
  \ifvarindexIndexMode
    \let\vx@tildeASort\vx@sortlist
    \let\vx@tildeBSort\vx@sortlist
    \toolboxDef\vx@textlist{\vx@textlistBefore\vx@textlist}%
    \toolboxDef\vx@sortlist{\vx@sortlistBefore\vx@sortlist}%
    \ifx\vx@textlist\vx@sortlist
      \toolboxAppend\vx@outputlist\vx@textlist
    \else
      \toolboxAppend\vx@outputlist{\vx@sortlist\vx@otherat\vx@textlist}%
    \fi
    \let\vx@sortlistBefore\toolboxEmpty
    \let\vx@sortlist\toolboxEmpty
  \else
    \toolboxAppend\vx@outputlist{\vx@textlistBefore\vx@textlist}%
  \fi
  \let\vx@textlistBefore\toolboxEmpty
  \let\vx@textlist\toolboxEmpty}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\vx@TextEntry}
% "\vx@TextEntry"\bracemeta{\texttt{\#1}} defines
% "\varindexEntry"/"Sort"/"Plain1-..."
% with the values corresponding to entry "#1". Here, "1-..." is
% "\the\varindexCount". If necessary, inner braces are eliminated
%
%    \begin{macrocode}
\def\vx@TextEntry#1{%
  \vx@SplitAtSort{#1}\vx@textEntryA\vx@textEntryB
  \expandafter\vx@SplitAtPlain\expandafter{\vx@textEntryA}%
    \vx@textEntryA\vx@textEntryC
  \vx@LetEliminated{varindexEntry}\vx@textEntryA
  \ifx\vx@textEntryB\undefined
%    \end{macrocode}
% No sort entry:
%    \begin{macrocode}
    \vx@LetEliminated{varindexSort}\vx@textEntryA
    \ifx\vx@textEntryC\undefined
      \vx@LetEliminated{varindexPlain}\vx@textEntryA
    \else
      \vx@LetEliminated{varindexPlain}\vx@textEntryC
    \fi
  \else
%    \end{macrocode}
% Sort entry:
%    \begin{macrocode}
    \ifx\vx@textEntryC\undefined
      \expandafter\vx@SplitAtPlain\expandafter{\vx@textEntryB}%
        \vx@textEntryB\vx@textEntryC
    \fi
    \vx@LetEliminated{varindexSort}\vx@textEntryB
    \ifx\vx@textEntryC\undefined
      \vx@LetEliminated{varindexPlain}\vx@textEntryA
    \else
      \vx@LetEliminated{varindexPlain}\vx@textEntryC
    \fi
  \fi
}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@LetEliminated}\nocomment
%    \begin{macrocode}
\def\vx@LetEliminated#1#2{%
  \let\vx@tmp#2%
  \ifvarindexEliminateInnerBrace
    \toolboxDropBrace\vx@tmp
  \fi
  \toolboxWithNr{\the\varindexCount}\let{#1}\vx@tmp}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexMakeVarSplit}\nocomment
%    \begin{macrocode}
\def\varindexMakeVarSplit#1#2#3{\varindexedef#1{#2}\expandafter
  \toolboxMakeSplit\expandafter{#1}{#3}}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexMakeSplitExpand}\nocomment
%    \begin{macrocode}
\def\varindexMakeSplitExpand#1#2{\varindexMakeVarSplit\vx@tmp{#1}{#2}}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexSetWordString}
%  \begin{macro}{\vx@SplitAtWord}\nocomment
%    \begin{macrocode}
\def\varindexSetWordString#1{\def\vx@word{#1}%
  \toolboxMakeSplit{#1}{vx@SplitAtWord}}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetSortString}
%  \begin{macro}{\vx@SplitAtSort}\nocomment
%    \begin{macrocode}
\def\varindexSetSortString#1{%
  \toolboxMakeSplit{#1}{vx@SplitAtSort}}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetPlainString}
%  \begin{macro}{\vx@SplitAtPlain}\nocomment
%    \begin{macrocode}
\def\varindexSetPlainString#1{%
  \toolboxMakeSplit{#1}{vx@SplitAtPlain}}

%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\vx@tmp}
% Temporarily, we define a generic command:
% \begin{myquote}
% "\vx@tmp"\bracemeta{name}
% \end{myquote}
% is (temporarily) a call which is equivalent to:
% \begin{myquote}
% "\def\varindexSet"\meta{name}"String#1{%"\\
% "  \varindexMakeVarSplit"\relax
%"\varindex"\meta{name}"String{#1}{varindexSplitAt"\meta{name}"}}"
% \end{myquote}
%    \begin{macrocode}
\def\vx@tmp#1{%
  \expandafter
  \def\expandafter\vx@adda\expandafter##\expandafter 1\expandafter{%
    \expandafter
    \varindexMakeVarSplit\csname varindex#1String\endcsname
    {##1}{varindexSplitAt#1}}%
  \expandafter
  \let\csname varindexSet#1String\endcsname\vx@adda}
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexSetIndexString}
%  \begin{macro}{\varindexSplitAtIndex}\nocomment
%    \begin{macrocode}
\vx@tmp{Index}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetOutputString}
%  \begin{macro}{\varindexSplitAtOutput}\nocomment
%    \begin{macrocode}
\vx@tmp{Output}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetTildeAString}
%  \begin{macro}{\varindexSplitAtTildeA}\nocomment
%    \begin{macrocode}
\vx@tmp{TildeA}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetTildeBString}
%  \begin{macro}{\varindexSplitAtTildeB}\nocomment
%    \begin{macrocode}
\vx@tmp{TildeB}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetSpaceString}
%  \begin{macro}{\varindexSplitAtSpace}\nocomment
%    \begin{macrocode}
\vx@tmp{Space}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetSpaceTokString}
%  \begin{macro}{\varindexSplitAtSpaceTok}\nocomment
%    \begin{macrocode}
\vx@tmp{SpaceTok}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetOmitString}
%  \begin{macro}{\varindexSplitAtOmit}\nocomment
%    \begin{macrocode}
\vx@tmp{Omit}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetCommaString}
%  \begin{macro}{\varindexSplitAtComma}\nocomment
%    \begin{macrocode}
\vx@tmp{Comma}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetDashString}
%  \begin{macro}{\varindexSplitAtDash}\nocomment
%    \begin{macrocode}
\vx@tmp{Dash}
%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetExclamString}
%  \begin{macro}{\varindexSplitAtExclam}\nocomment
%    \begin{macrocode}
\vx@tmp{Exclam}

%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\varindexSetStringForNr}
%  \begin{macro}{\varindexSplitAtNr}\nocomment
%    \begin{macrocode}
\def\varindexSetStringForNr#1#2{%
  \edef\vx@adda{#1}%
  \expandafter
    \vx@AddToHandleNrs\csname varindexSplitAtNr\vx@adda\endcsname
%    \end{macrocode}
% It is important to define the instance only \emph{after} the call of
% "\vx@AddToHandleNrs" (because this macro tests the existence).
%    \begin{macrocode}
  \expandafter
    \varindexMakeVarSplit\csname varindexNr\vx@adda String\endcsname
      {#2}{varindexSplitAtNr\vx@adda}}

%    \end{macrocode}
%  \end{macro}
% \end{macro}
%
% \begin{macro}{\vx@AddToHandleNrs}
% "\vx@AddToHandleNrs{\vx@HandleNr1}"
% tests whether "\vx@HandleNr1" is already defined:
% If not (i.\,e.\ "\relax" because we use it with "\csname") the call
% \begin{myquote}
%  "\varindexNewFormat\varindexSplitAtNr1{\varindexAddNumber"\relax
%\bracemeta{content of \cmdcite{vx@adda}}"}"
% \end{myquote}
%  is executed.
%    \begin{macrocode}
\def\vx@AddToHandleNrs#1{%
  \ifx#1\relax
    \expandafter
      \varindexNewFormat\expandafter#1\expandafter{\expandafter
        \varindexAddNumber\expandafter{\vx@adda}}%
  \fi}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\varindexedef}
% This macro is described in the main documentation.
%    \begin{macrocode}
\def\varindexedef#1#2{%
  \let\vx@tildemerk~%
  \let~\vx@nicetilde
  \edef#1{#2}%
  \let~\vx@tildemerk}

%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\varindexDoConvert}\nocomment
%    \begin{macrocode}
\def\varindexDoConvert{\let\vx@MakeHarmless\toolboxMakeHarmless}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexNoConvert}\nocomment
%    \begin{macrocode}
\def\varindexNoConvert{\def\vx@MakeHarmless##1{}}

%    \end{macrocode}
% \end{macro}
%
%
% \par\hrule\par\null\par
% Finally, some definitions which require catcode-trickery:
%
%
% \begin{macro}{\vx@nicetilde}
% "\vx@nicetilde" expands to a tilde with a nonactive catcode.
%
%    \begin{macrocode}
\def\vx@tmp{\catcode`\~=13\relax}
\catcode`\~=11\relax\def\vx@nicetilde{~}\vx@tmp

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@spaceothertok}
%  \begin{macro}{\vx@MakeSpaceOther}\nocomment
%    \begin{macrocode}
\def\vx@MakeSpaceOther{\catcode`\ =13\relax}
%    \end{macrocode}
%  \end{macro}
%  \begin{macro}{\vx@MakeSpaceSpace}\nocomment
%    \begin{macrocode}
\def\vx@MakeSpaceSpace{\catcode`\ =10\relax}
\vx@MakeSpaceOther%<- This % is important
\let\vx@spaceothertok= %<- This ` %' is important
\vx@MakeSpaceSpace

%    \end{macrocode}
%  \end{macro}
% \end{macro}
% \begin{macro}{\vx@otherat}
% "\def\vx@otherat{@}" is a bad idea, since then
% "@" has the wrong catcode. So we use a temporary macro to define it:
% We misuse the name "\varindexArgumentSpace" to this purpose.
% Of course, we must redefine the latter afterwards.
%    \begin{macrocode}
\def\varindexArgumentSpace#1{\def\vx@otherat{#1}}
\ifx\makeatother\undefined\catcode`\@=12\relax\else\makeatother\fi
\varindexArgumentSpace{@}

%    \end{macrocode}
% \end{macro}
%
% Initialize default values (you should have the `usual' catcodes here.
% In particular, "\makeatother" must be called before if `"@"'
% is used as argument).
% \begin{macro}{\varindexArgumentSpace}\nocomment
%    \begin{macrocode}
\let\varindexArgumentSpace\varindexSkipDefault

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\varindexIndex}
% We use "\def" for "\varindexIndex" so that a redefinition of "\index" also
% redefines our command.
%    \begin{macrocode}
\toolboxIf\undefined\def\varindexIndex{\index}

%    \end{macrocode}
% \end{macro}
% \begin{macro}{\vx@MakeHarmless}\nocomment
%    \begin{macrocode}
\varindexDoConvert

%    \end{macrocode}
% \end{macro}
%    \begin{macrocode}
\toolboxIf\undefined\def\varindexOutExclam{!}
\toolboxIf\undefined\let\varindexFollowsExclam\varindexNoNextSpace
\toolboxIf\undefined\def\varindexOutTextComma{,}
\toolboxIf\undefined\def\varindexOutSortComma{,}
\toolboxIf\undefined\let\varindexFollowsComma\varindexNextSpace
\toolboxIf\undefined\let\varindexPreceedsComma\varindexNoNextSpace
\toolboxIf\undefined\def\varindexOutTextSpace{ }
\toolboxIf\undefined\def\varindexOutSortSpace{ }
\toolboxIf\undefined\let\varindexFollowsSpace\toolboxEmpty
\toolboxIf\undefined\let\varindexPreceedsSpace\varindexNoNextSpace
\toolboxIf\undefined\def\varindexOutTextSpTok{\space}
\toolboxIf\undefined\def\varindexOutSortSpTok{ }
\toolboxIf\undefined\let\varindexFollowsSpTok\toolboxEmpty
\toolboxIf\undefined\let\varindexPreceedsSpTok\varindexNoNextSpace
\toolboxIf\undefined\def\varindexOutTextDash{-}
\toolboxIf\undefined\def\varindexOutSortDash{-}
\toolboxIf\undefined\let\varindexOutSortDashBeg\toolboxEmpty
\toolboxIf\undefined\let\varindexFollowsDash\toolboxEmpty
\toolboxIf\undefined\let\varindexPreceedsDash\varindexNoNextSpace
\toolboxIf\undefined\let\varindexOutTextOmit\toolboxEmpty
\toolboxIf\undefined\let\varindexOutSortOmit\toolboxEmpty
\toolboxIf\undefined\let\varindexFollowsOmit\toolboxEmpty
\toolboxIf\undefined\let\varindexPreceedsOmit\varindexNoNextSpace
%    \end{macrocode}
% For the following macro, see the remarks in the definition of
% the fake name "\varindexAddOmit" (after definition of "\varindexNewFormat")
%    \begin{macrocode}
%% It is important that the following is after all \toolboxIf\undefined...
\varindexAddOmit% Execute the fake macro and redefine \varindexAddOmit

%    \end{macrocode}
%
% \begin{macro}{\varindexSplitSpace}\nocomment
%    \begin{macrocode}
\toolboxMakeSplit{ }{varindexSplitSpace}

%    \end{macrocode}
% \end{macro}
%
%    \begin{macrocode}
\varindexSetWordString{ }
\varindexSetSortString{@}
\varindexSetPlainString{!}

\varindexSetIndexString{ }
\varindexSetOutputString{*}

\varindexSetTildeAString{~}
\varindexSetTildeBString{=}
\varindexSetSpaceString{_}
\varindexSetSpaceTokString{s}
\varindexSetOmitString{.}
\varindexSetCommaString{,}
\varindexSetDashString{-}
\varindexSetExclamString{!}
\varindexSetStringForNr{10}{0}
\varindexSetStringForNr 9{9}
\varindexSetStringForNr 8{8}
\varindexSetStringForNr 7{7}
\varindexSetStringForNr 6{6}
\varindexSetStringForNr 5{5}
\varindexSetStringForNr 4{4}
\varindexSetStringForNr 3{3}
\varindexSetStringForNr 2{2}
\varindexSetStringForNr 1{1}

\ifx\makeatletter\undefined\catcode`\@=11\relax\else\makeatletter\fi

%    \end{macrocode}
% \iffalse
%</package>
% \fi
% \Finale\PrintIndex
%
%
%
% \iffalse
%<*metainfo>
% \fi
\iffalse -------------------------------------------------------------
%
% Just in case that for some reason the file `varindex.ins' got lost:
% The content of this file is between the lines %<*insfile> and %</insfile>:
%
%\iffalse
%</metainfo>
%<*insfile>
\def\batchfile{varindex.ins}
\input docstrip.tex
\keepsilent
\nopreamble
\generateFile{varindex.sty}{f}{\from{varindex.dtx}{package}}
\generateFile{varindex.txt}{f}{\from{varindex.dtx}{asciidocu}}
\generateFile{varindex.tex}{f}{\from{varindex.dtx}{driver}}
%%\generateFile{varindex.ins.dup}{f}{\from{varindex.dtx}{insfile}}

\Msg{***************************************************************}
\Msg{*}
\Msg{* To finish the installation you have to move the}
\Msg{* file 'varindex.sty' into a directory searched by TeX.}
\Msg{*}
\Msg{*}
\Msg{* Run the file varindex.tex through LaTeX to produce the}
\Msg{* documentation or read the ascii documentation in varindex.txt}
\Msg{*}
\Msg{*}
\Msg{***************************************************************}
\endbatchfile
%</insfile>
%<*metainfo>
% \fi
\fi
% \iffalse
%</metainfo>
% \fi
\endinput

%%% Local Variables: 
%%% mode: latex
%%% TeX-master: "varindex.tex"
%%% End: