% \iffalse meta-comment % Copyright (C) 2015--2019 by Edgar Olthof % % This file may be distributed and/or modified under the % conditions of the LaTeX Project Public License, either % version 1.3 of this license or (at your option) any later % version. The latest version of this license is in % % http://www.latex-project.org/lppl.txt % % and version 1.3 or later is part of all distributions of % LaTeX version 1999/12/01 or later. % % \fi % \iffalse %\ProvidesPackage{fcolumn} % [2019/04/24 v1.2 Automatic Currency Package (EHTO)] %\NeedsTeXFormat{LaTeX2e} %\RequirePackage{array}[v2.4k] %<*driver> \documentclass{ltxdoc} \usepackage{fcolumn} \usepackage{booktabs} \usepackage[tableposition=top,aboveskip=0pt]{caption} \DeclareCaptionStyle{normal}{labelfont={rm,bf,small},textfont={rm,small},% indention=0pt,labelsep=colon,justification=raggedright} \captionsetup[table]{style=normal} \captionsetup{singlelinecheck=false} \usepackage{paralist} \usepackage{url} \usepackage[hyperfootnotes=false,hidelinks]{hyperref} \usepackage{eurosym} \EnableCrossrefs \CodelineIndex \RecordChanges \begin{document} \DocInput{fcolumn.dtx} \end{document} % % \fi % \CheckSum{1078} % \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 \~} % % \changes{v0.1}{2015/02/06}{First working version.} % \changes{v1.0}{2015/03/07}{Three-argument version is working properly.} % \changes{v1.1}{2015/09/14}{Automatic checking of column balance performed when number of % F-columns is even (behaviour can be overridden). Empty entries are now recognised and % correctly treated as such, except for the one ended by the double backslash. Not serious; % workaround possible. Furthermore optimisation of code: minimised the number of private % counts and resetting of column counter done in a nicer way.} % \changes{v1.1.1}{2015/09/27}{Installation procedure changed from .ins-in-.dtx to separate % .ins and .dtx after discussion with Karl Berry as well as some minor code improvements.} % \changes{v1.1.2}{2015/09/29}{Some inconsistencies between explanatory text % and actual code removed.} % \changes{v1.2}{2019/03/06}{Input parsing changed after comment from Frank % Mittelbach. He (Frank) also gave various suggestions for improving robustness % or user friendlyness of this package. This version is only backwards compatible % when zero decimal digits were and are specified as modifier.} % \GetFileInfo{fcolumn.sty} % \DoNotIndex{\@arstrut,\@arstrutbox,\@depth,\@firstampfalse,\@firstamptrue,\@gobble,\@halignto} % \DoNotIndex{\@height,\@mkpream,\@ne,\@nextchar,\@preamble,\@sharp,\@tempdima,\@tempswafalse} % \DoNotIndex{\@tempswatrue,\@\temptokena,\@tfor,\@whilesw,\@width,\advance,\arraystretch,\baselineskip} % \DoNotIndex{\begingroup,\bgroup,\count,\count@,\cr,\csname,\def,\divide,\do,\dp,\edef,\egroup} % \DoNotIndex{\else,\endcsname,\endgroup,\expandafter,\extrarowheight,\fi,\gdef,\halignto,\hbox} % \DoNotIndex{\hfil,\ht,\if,\if@firstamp,\if@tempswa,\ifcase,\lineskip,\loop,\m@ne,\m@th,\message} % \DoNotIndex{\multiply,\or,\par,\prepnext@tok,\protect,\repeat,\setbox,\space,\string,\strutbox} % \DoNotIndex{\tabskip,\tabularnewline,\the,\the@toks,\thr@@,\tw@,\unhbox,\vbox,\vcenter,\vrule} % \DoNotIndex{\vskip,\vtop,\wd,\xdef,\z@,\z@skip} % \title{The \textsf{fcolumn} package\thanks{This file has version % number \fileversion, last revised \filedate.}} % \author{Edgar Olthof\\ % \texttt{edgar olthof inter nl net}} % \date{Printed \today} % \maketitle % \begin{abstract} % In financial reports, text and currency amounts are regularly put in % one table, e.g., a year balance or a profit-and-loss overview. This % package provides the settings for automatically typesetting and checking % such columns, including the sum line (preceded by a rule of the correct % width), using the specifier |f|. % \end{abstract} % % \section{Introduction}\label{intro} % The package |fcolumn| provides the macros for an extra tabular specifier % that makes creating financial tables easy. The column specifier |f| % itself is rather simple; it is the predefined version of a generic % column |F|. The generic version expects three arguments: 1)~grouping % character of the integer part on output, 2)~decimal mark used on output, % and 3)~compact additional information on input/output characteristics, % see below. % % The f-column in the current version of the package is defined for the % continental European standard: |\newcolumntype{f}{F{.}{,}{3,2}}|. This % means that a number like 12345,67 will be typeset as $12{.}345{,}67$. % People in the Anglo-saxon world would rather code % |\newcolumntype{f}{F{,}{.}{3,2}}| for the same input, yielding % $12{,}345{.}67$ as output for the number given above. The default % value for |#3| is |3,2|, indicating % that grouping of the integer part is by three digits, that a comma is % used in the \TeX-source to indicate the decimal separator, and that % the decimal part consists of two digits. % If however, in your country or company grouping is done with a thinspace every % four digits, that the separator in the source should be the character |p|, % and there are three digits after the decimal mark---that happens to be a % |\cdot|---, then simply specify |\newcolumntype{f}{F{\,}{\cdot}{4p3}}| in % that case. The input could be 12345p678 then, yielding $1{\,}2345{\cdot}678$ % as output. % % By default two digits are used for the decimal part, so if you really % want no decimal digits (in that case of course also skipping the decimal % mark) you have to explicitly specify |x,0|. If you want no grouping % character, specify |0,x|. % % This package requires and loads the |array| package~\cite{array}. To % show where and how the |F|-column is used, let's look at some typical % financial information as shown in Table~\ref{tab:ex1} % \begin{table}[htb] % \caption{Example Table.} % \label{tab:ex1} % \begin{tabular}{@{}lflf@{}} % \multicolumn4c{\bfseries Balance sheet}\\ % \toprule % properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\ % \midrule % house & 200000 & equity capital& 50000 \\ % bank account & -603,23 & mortgage & 150000 \\ % savings & 28000 & & \\ % cash & 145,85 & profit & 27542,62 \\ % \sumline % \bottomrule % \end{tabular} % \end{table} % and how this is entered in \LaTeX\ (Table~\ref{tab:ex2}). % \begin{table}[htb] % \caption{Verbatim version of Example Table.} % \label{tab:ex2} % \begin{verbatim} %\begin{table}[htb] %\caption{Example Table.} %\label{tab:ex1} %\begin{tabular}{@{}lflf@{}} %\multicolumn4c{\bfseries Balance sheet}\\ %\toprule %properties & \leeg{31 dec 2014} & debts & \leeg{31 dec 2014}\\ %\midrule %house & 200000 & equity capital& 50000 \\ %bank account & -603,23 & mortgage & 150000 \\ %savings & 28000 & & \\ %cash & 145,85 & profit & 27542,62 \\ %\sumline %\bottomrule %\end{tabular} %\end{table} % \end{verbatim} % \end{table} % All the work was done by the column specifier ``f'' (for ``finance''). % In this case it constructs the |\sumline|, typesets the numbers, % calculates the totals, determines the widths of the sumrules, and checks % whether the two columns are in balance; if not, the user is warned via a % |\PackageWarning|. Of course for nice settings the |booktabs| % package~\cite{booktabs} was used, but that is not the point here. % % This package is heavily inspired by the |dcolumn| package by % David Carlisle~\cite{dcolumn}, some constructions are more or less % copied from that package. Version~1.2 (this version) uses a few ideas % from, and incorporates quite some suggestions by, Frank Mittelbach. % % \section{Commands} % The user only needs to know six commands or constructions. These six are given here. % \begin{macro}{F} % In the tabular the column specifier |F| can be given with arguments, or % the predefined version~|f|, where the three arguments of |F| are |{.}|, % |{,}|, and |{3,2}|. If you want |g| to be your own definition like the % curious one given in Section~\ref{intro}, then specify % |\newcolumntype{g}{F{\,}{\cdot}{4p3}}| prior to using |g| in a tabular. % % Entries in an F-column are, from that moment on, treated as numbers unless % explicitly escaped by |\leeg|, see below. The numbers are typeset % according to the template the user gives with his/her |F|-column. % The ``middle'' character of |#3| is an important switch: it does more % than just setting the input decimal mark. By default the input grouping % character is the dot, except when the dot is specified as input decimal % mark; in that case the comma is acting as input grouping character. With % this convention the continental Europe and Anglo-saxon part of the world % is served. % \end{macro} % \begin{macro}{\sumline} % The numbers in an |F|-column are typeset as a financial amount, but the real % benefit comes with the |\sumline|. It does three things: % \begin{compactenum}[1)] % \item It calculates the total of the column so far and the maximum width % encountered so far, including the width of the total; % \item It generates a rule with width calculated in the first item; % \item It checks the columns that are supposed to balance whether or not they % actually do. If so, nothing happens. If not, a |\PackageWarning| is given % that column $i$ and $j$ do not balance, where $i$ and $j$ are the relevant % columns. This is only done if the total number of |F|-columns is even, % e.g., if there are six |F|-columns, then~1 is checked against~4, 2 against~5, % and 3 against~6. If the number of |F|-columns is odd then anything could be % possible in that table and nothing is assumed about structure within the % table. This behaviour can be overridden, see below. % \end{compactenum} % By default the vertical separation between the rule and the total is % 2\thinspace pt, but this can be changed by the optional argument to % |\sumline|. Give, e.g., |\sumline[10pt]|, in case you want this spacing % to be 10\thinspace pt. % \end{macro} % \begin{macro}{\resetsumline} % Suppose you want to typeset one tabular with the profit-and-loss of many % projects individually. The layout of those tabulars is the same and it % were nice if all columns were aligned. This can be done by making it one % big tabular with a fresh start for each project. The macro % |\resetsumlines| is used for that: it resets all totals and all column % widths, see for example Table~\ref{tab:ex3}. % \begin{table}[htb] % \caption{Example: multiple projects.} % \label{tab:ex3} % \begin{tabular}{@{}lfflff@{}} % \multicolumn6c{\bfseries Project~1}\\ % \toprule % expense & \leeg{actual} & \leeg{budget} & % income & \leeg{actual} & \leeg{budget} \\ % \midrule % food & 450,20 & 500 & tickets & 1200 & 1000 \\ % drinks & 547,5 & 400 \\ % music & 180 & 100 \\ % profit & 22,3 \\ % \sumline % \resetsumline % \multicolumn6c{\bfseries Project~2}\\ % \toprule % expense & \leeg{actual} & \leeg{budget} & % income & \leeg{actual} & \leeg{budget} \\ % \midrule % food & 250 & 300 & tickets & 400 & 450 \\ % drinks & 100 & 80 \\ % music & 80 & 70 & loss & 30 \\ % \sumline % \bottomrule % \end{tabular} % \end{table} % Note that the rules in the first and third |F|-columns of project~1 cover % $1{.}200{,}00$ whereas in project~2 those rules are narrower since they % only cover $430{,}00$; still the columns are aligned. The verbatim way % of setting up Table~\ref{tab:ex3} is given in Table~\ref{tab:ex4}. % \begin{table}[htb] % \caption{Verbatim version of Table~\ref{tab:ex3}.} % \label{tab:ex4} % \begin{verbatim} %\begin{table}[htb] %\caption{Example: multiple projects.} %\label{tab:ex3} %\begin{tabular}{@{}lfflff@{}} %\multicolumn6c{\bfseries Project~1}\\ %\toprule %expense & \leeg{actual} & \leeg{budget} & %income & \leeg{actual} & \leeg{budget} \\ %\midrule %food & 450,2 & 500 & tickets & 1200 & 1000 \\ %drinks & 547,5 & 400 \\ %music & 180 & 100 \\ %profit & 22,3 \\ %\sumline %\resetsumline %\multicolumn6c{\bfseries Project~2}\\ %\toprule %expense & \leeg{actual} & \leeg{budget} & %income & \leeg{actual} & \leeg{budget} \\ %\midrule %food & 250 & 300 & tickets & 400 & 450 \\ %drinks & 100 & 80 \\ %music & 80 & 70 & loss & 30 \\ %\sumline %\bottomrule %\end{tabular} %\end{table} % \end{verbatim} % \end{table} % \end{macro} % \begin{macro}{\leeg} % If an |F|-column should be empty then simply leave it empty. If however % it should not be empty but the entry should be treated as text---even % if it is a number---, this can be done with |\leeg|. It expects an % argument and this argument is typeset in the column. The common case is % where |p.m.| (\textit{pro memoria}) is entered. In contrast to v1.1.2 % of this package, now even an empty |F|-column followed by |\\| is allowed. % \end{macro} % \begin{macro}{\checkfcolumns} % The automatic column balance check can also be done manually. If % |F|-columns~1 and~4 should balance and you want them to be checked, % then simply say |\checkfcolumns14|. With more than nine |F|-columns % you may be forced to say something like |\checkfcolumns{10}{12}|. If % |\checkfcolumns| is used, the automatic check is disabled. Multiple % |\checkfcolumns|s are supported; if |F|-columns~1, 2, and~3 should % balance, you specify |\checkfcolumns12| and |\checkfcolumns23|. % There is no explicit command to disable all checking, but % |\checkfcolumns11| obviously also serves that purpose. % \end{macro} % \begin{macro}{\ifstrict@ccounting} % In the rare occasion that a negative number occurs in a financial table, % the sign of that number can be an explicit minus sign ($-$) or the number % is coloured red, or it is typeset between parentheses, and there may be % even other ways. By default (for aesthetic reasons) |fcolumn| typesets % it with a minus~sign, but strict accounting prescibes that the number % should be put between parentheses. The latter can be accomplished by % setting |\strict@ccountingtrue|. % \end{macro} % \StopEventually{\PrintChanges\PrintIndex} % \section{The macros} % Here follows the actual code. % \subsection{Option} % \begin{macro}{option strict} % There is one option. If set, strict accounting rules are used in display. % \begin{macrocode} \newif\ifstrict@ccounting \strict@ccountingfalse \DeclareOption{strict}{\strict@ccountingtrue} \ProcessOptions % \end{macrocode} % \end{macro} %\subsection{Definitions} % \begin{macro}{column F} % \begin{macro}{column f} % The column specifier |F| is the generic one, and |f| is the default % (continental European) one for easy use. Note that the definition of the % column type |f| does not use private macros (no~|@|), so overriding its % definition is easy for a user. % \begin{macrocode} \newcolumntype{F}[3]{>{\b@fi{#1}{#2}{#3}}r<{\e@fi}} \newcolumntype{f}{F{.}{,}{3,2}} % \end{macrocode} % \end{macro} % \end{macro} % \begin{macro}{\FCsc@l} % \begin{macro}{\FCtc@l} % Two \meta{count}s are defined, that both start at zero: the \meta{count} % |\FCsc@l|, that keeps track at which F-column the tabular is working on % and the \meta{count} |\FCtc@l|, that records the number of F-columns % that were encountered so far. Later in the package the code can be % found for generating a new \meta{count} and a new \meta{dimen} if the % number of requested |F|-columns is larger than currently available. % This is of course the case when an |F|-column is used for the first time. % \begin{macrocode} \newcount\FCsc@l \FCsc@l=0 \newcount\FCtc@l \FCtc@l=0 % \end{macrocode} % \end{macro} % \end{macro} % \begin{macro}{\geldm@cro} % The macro |\geldm@cro| takes a number and by default interprets this as % an amount expressed in cents (dollar cents, euro cents, centen, Pfennige, % centimes, kopecks, groszy) and typesets it as the amount in entire % currency units (dollars, euros, guldens, Marke, francs, rubles, z\l oty) % with comma as decimal separator and the dot as grouping character % (thousand separator if the first part of |#1| is~3). % As explained, this can be changed. It uses two private booleans: % |\withs@p| and |\strict@ccounting|. The latter is used to typeset % negative numbers between parentheses. By default it doesn't do this: % a minus sign is used. % \begin{macrocode} \newif\ifwiths@p % \end{macrocode} % Actually |\geldm@cro| is only a wrapper around |\g@ldm@cro|. % \begin{macrocode} \def\geldm@cro#1#2{\withs@pfalse \afterassignment\g@ldm@cro\count@#1\relax{#2}} % \end{macrocode} % \end{macro} % \begin{macro}{\g@ldm@cro} % This macro starts by looking at the sign of |#2|: if it is negative, it prints % the correct indicator (a parenthesis or a minus sign), assigns the absolute value % of~|#2| to |\count2| and goes on. Note that |\geldm@cro| and therefore |\g@ldm@cro| % are always used within |$|s, so it is really a minus sign that is printed, not a hyphen. % All calculations are done with |\count0|, |\count1|, etc.\ i.e., without |F|-column-specific % \meta{count}s because it is all done locally. Leaving the tabular environment will % restore their values. % \begin{macrocode} \def\g@ldm@cro#1\relax#2{\ifnum#2<0 \ifstrict@ccounting (\else -\fi \count2=-#2 \else\count2=#2 \fi % \end{macrocode} % Calculate the entire currency units: this is the result of $x/a$ as integer % division, with $a=10^n$ and $n$ the part of |#1| after the separator (if any). % Here the first character of |#1| is discarded, so the separator in |#1| is not strict: % you could also specify |3.2| instead of |3,2| (or even |3p2|). % \begin{macrocode} \count4=\ifx\relax#1\relax 2 \else \@gobble#1\relax\fi \count3=0 \loop\ifnum\count3<\count4 \divide\count2 by 10 \advance\count3 by \@ne \repeat % \end{macrocode} % Note that |\count3| now equals |\count4|: this going up-and-down will be used more often, % it saves several assignments. The value in |\count2| is then output by |\g@ldens| using % the separation given (and stored in |\count@|). % \begin{macrocode} \g@ldens{\the\count@}% % \end{macrocode} % If there is a decimal part\dots % \begin{macrocode} \ifnum\count3>0\decim@lmark % \end{macrocode} % Next the decimal part is dealt with. Now $x\bmod a$ is calculated in the % usual way: $x-(x/a)*a$ with integer division. The minus sign necessary % for this calculation is introduced in the next line by changing the % comparison from |<| to |>|. % \begin{macrocode} \ifnum#2>0 \count2=-#2\else\count2=#2 \fi \loop\ifnum\count3>0 \divide\count2 by 10 \advance\count3 by \m@ne \repeat % \end{macrocode} % The value of |\count3| is now~0, so counting up again. % \begin{macrocode} \loop\ifnum\count3<\count4 \multiply\count2 by 10 \advance\count3 by \@ne \repeat \ifnum#2>0 \advance\count2 by #2 \else \advance\count2 by -#2 \fi \zerop@d{\number\count3}{\number\count2}% \fi % \end{macrocode} % If the negative number is indicated by putting it between parentheses, % then the closing parenthesis should stick out of the column, otherwise % the alignment of this entry in the column is wrong. This is done by an % |\rlap| and therefore does not influence the column width. For the last % column this means that this parenthesis may even stick out of the table. % I don't like this, therefore I chose to put |\strict@ccountingfalse|. % Change if you like, by setting the option |strict|. % % If overflow was detected, an exclamation mark is output to the right of % the value that caused this. This of course ruins the appearance of the % table, but in this case that serves a clear goal: there's something wrong % and you should know. % \begin{macrocode} \ifx\FCs@gn\m@ne \ifnum#2<0 \ifstrict@ccounting \rlap{)~!}\else\rlap{~!}\fi\else\ifstrict@ccounting \rlap{\phantom{)}~!}\else\rlap{~!}\fi\fi \else \ifnum#2<0 \ifstrict@ccounting\rlap{)}\fi\fi \fi} % \end{macrocode} % \end{macro} % \begin{macro}{\g@ldens} % Here the whole currency units are dealt with. The macro |\g@ldens| is % used recursively, therefore the double braces; this allows to use % |\count0| locally. This also implies that tail recursion is not % possible here, but that is not very important, as the largest number % (which is $2^{31}-1$) will only cause a threefold recursion using the % default |3,2| (ninefold when using |1,0|, but who does that?). The % largest amount this package can deal with is therefore 2.147.483.647 % (using |3,0|). For most people this is probably more than enough if % the currency is euros or dollars. And otherwise make clear that you % use a currency unit of~k\EUR{} (or even M\EUR{} for the very rich). % The author is thinking of ways to use two counters for each number. % The maximum then becomes $2^{63}-1$. Even expressed in cents this % would lead to a maximum of slightly more than 92.2\thinspace P\EUR{}; % about 100 times the current world economy~\cite{world_economy}. Yet % another method is to used Heiko Oberdiek's package |bigintcalc|: % then only memory restrictions apply. This, however, requires a major % rewrite of |fcolumn|. For now, version~1.2 sticks to the moderate amounts. % % There is no straightforward interpretation of |#1| being zero or % negative, therefore this is used as an indicator that no grouping % character should be used. % \begin{macrocode} \def\g@ldens#1{{\count3=\count2 \count0=#1 % \end{macrocode} % First divide by $10^n$, where $n$ is |#1|. % \begin{macrocode} \ifnum\count0<1 \count0=3 \fi \loop \ifnum\count0>0 \divide\count2 by 10 \advance\count0 by \m@ne \repeat % \end{macrocode} % Here is the recursive part, % \begin{macrocode} \ifnum\count2>0 \g@ldens{#1}\fi % \end{macrocode} % and then reconstruct the rest of the number. % \begin{macrocode} \count0=#1 \ifnum\count0<1 \count0=3 \fi \loop \ifnum\count0>0 \multiply\count2 by 10 \advance\count0 by \m@ne \repeat \count2=-\count2 \advance\count2 by \count3 \du@zendprint{#1}}} % \end{macrocode} % \end{macro} % \begin{macro}{\du@zendprint} % The macro |\du@zendprint| takes care for correctly printing the separator % and possible trailing zeros. The former, however, is only done if |#1| % is larger than zero. % \begin{macrocode} \def\du@zendprint#1{\ifwiths@p\ifnum#1>0 \sep@rator\fi \zerop@d{#1}{\number\count2}% \else\zerop@d1{\number\count2}\fi\global\withs@ptrue} % \end{macrocode} % \end{macro} % \begin{macro}{\zerop@d} % The macro |\zerop@d| uses at least |#1| digits for printing the % number |#2|, padding with zeros when necessary. Note: |#1| being % zero or negative is a flag that it should be interpreted as 3. % A bit ugly, but it works, since the related code knows about this. % % It is done within an extra pair of braces, so that |\count0| % and |\count1| can be used without disturbing their values in other macros. % \begin{macrocode} \def\zerop@d#1#2{{\count0=1 \count1=#2 % \end{macrocode} % First determine the number of digits of |#2| (expressed in the decimal % system). This number is in |\count0| and is at least~1. % \begin{macrocode} \loop \divide \count1 by 10 \ifnum\count1>0 \advance\count0 by \@ne \repeat % \end{macrocode} % If |#1| is positive, the number of zeros to be padded is % $\max(0,|#1-\count0|)$ (the second argument can be negative), so a % simple loop suffices. If it is zero or negative, this is a signal % that it should be interpreted as 3 (and no separator will be output). % \begin{macrocode} \ifnum#1>0 \loop \ifnum\count0<#1\relax 0\advance\count0 by \@ne \repeat \else \advance\count0 by -3 \loop \ifnum\count0<0 0\advance\count0 by \@ne \repeat \fi\number#2}} % \end{macrocode} % \end{macro} % \begin{macro}{\zetg@ld} % This macro takes care for several things: it increases the subtotal for a % given F-column, it checks whether or not that subtotal has overflown, it % records the largest width of the entries in that column and it typesets % |#1| via |\geldm@cro|. % \begin{macrocode} \def\zetg@ld#1#2{\count0=#2\relax \let\FCs@gn=\@ne % \end{macrocode} % First it checks whether there is a risk of overflow in this step. If % $A$ and $B$ are two \TeX-registers and $B$ is to be added to $A$, % overflow will not occur if one is (or both are) zero or if $A$ and~$B$ % have different signs. Otherwise, be careful. Note that \TeX\ does not % check for overflow when performing an |\advance| (done in section~1238 % of Ref.~\cite{CandTB}), in contrast to |\multiply|, see section~105. % \begin{macrocode} \ifnum\count0<0 \ifnum\csname FCtot@\romannumeral\FCsc@l\endcsname<0 \let\FCs@gn=\m@ne \fi \fi \ifnum\count0>0 \ifnum\csname FCtot@\romannumeral\FCsc@l\endcsname>0 \let\FCs@gn=\m@ne \fi \fi \global\advance\csname FCtot@\romannumeral\FCsc@l\endcsname by \count0 \ifx\FCs@gn\m@ne % \end{macrocode} % They had the same sign: risk of overflow. Record the sign % of |\count0| (and of the original total of this column; they were % the same) in |\FCs@gn|. Table~\ref{tab:ex5} % \begin{table}[htb] % \def\PackageError#1#2#3{\PackageWarning{#1}{#2}} % \caption{Examples on overflow.} % \label{tab:ex5} % \begin{tabular}{@{}lfff@{}} % \multicolumn4c{\bfseries Projects}\\ % \toprule % income & \leeg{31 dec 2014} & \leeg{31 dec 2015} & \leeg{31 dec 2016} \\ % \midrule % item~1 & 20000000 & 20000000 &20000000 \\ % item~2 & 10000000 & 2000000 &-2000000 \\ % item~3 & 5000000 & -2000000 & 2000000 \\ % \checkfcolumns23 % \sumline % \bottomrule % \end{tabular} % \end{table} % shows what can go wrong if the numbers are too large: in the left % |F|-column the sumline is incorrect and the number that caused % the overflow is indicated by an exclamation mark. In the middle % |F|-column, overflow occurs twice and because this is once positive, % once negative here, cancellation of errors occurs and the sumline % is correct in the end. Nevertheless, it is advised to swap the two % items that caused the overflow, as shown in the right |F|-column. % % Since the absolute value of |\FCs@gn| is unity, no overflow will % occur in the multiplication step below. % \begin{macrocode} \ifnum\count0>0 \let\FCs@gn\@ne \fi \count0=\csname FCtot@\romannumeral\FCsc@l\endcsname \multiply\count0 by \FCs@gn \ifnum\count0<0 \let\FCs@gn=\m@ne \PackageError{fcolumn}{Register overflow}{Overflow occurred in fcolumn \number\FCsc@l. Check your table.}% \else\let\FCs@gn=\@ne \fi \fi % \end{macrocode} % The value of |\FCs@gn| is used in |\geldm@cro| below. % \begin{macrocode} \setbox0=\hbox{$\geldm@cro{#1}{#2}$}% \ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0 \fi\unhbox0} % \end{macrocode} % \end{macro} % \def\<#1>{\leavevmode\hbox{$\langle$#1\/$\rangle$}} % \begin{macro}{\b@fi} % The macro |\b@fi| provides the beginning of the financial column. It % will be inserted in the column to capture the number entered by the user % The separator and decimal mark are within a math environment, so you % can indeed specify |\,| instead of |\thinspace|, but there is an extra % brace around, so it doesn't affect the spacing between the digits % (trick copied from |dcolumn|, Ref.~\cite{dcolumn}). % % The \meta{count}s |\FC@l| captures the part to the left of the decimal % marker, |\FC@r| that to the right. % \begin{macrocode} \newcount\FC@l \newcount\FC@r \def\FC@chklist{} \def\setucc@de#1#2\relax{\uccode`\~=`#1 } \def\b@fi#1#2#3{% % \end{macrocode} % An intermediate macro |\sep@xt| to extract the first character % of |#1|, which in most cases will be the only character. % \begin{macrocode} \def\sep@xt##1##2\end{\def\sep@rator{{##1}}}% \sep@xt#1\end\def\decim@lmark{{#2}}% \def\sp@l{#3}\global\advance\FCsc@l by \@ne \global\FC@l=0 \global\FC@r=1 % \end{macrocode} % The value specified by the user is then captured by |\FC@l| and % this is done in a special way: |\FC@l| is assigned globally within % |\box0|. Why? To use it as scribbling paper to examine what the % user entered, without dumping it into the horizontal list. % % There are four parts to an |F|-column entry, all parts optional, % making 16 combinations. The sequence is % (in the Backus--Naur notation of Ref.~\cite{CandTA}): % \ \ \ % \. Here \ is a plus or minus character with % category code~12, \ is a sequence of zero % or more (decimal) \s, and \ is the middel % part of |#3|, i.e., the comma in |3,2| or the period in |3.2|. % If the \ is absent with no space characters between % the two \ terms, these merge, making four redundant % entries. One of the % combinations is \, a sequence of exactly zero non-space tokens: % this is the only combination that doesn't put anything in an |F|-column. % % The minus sign must be captured separately, because in an entry like % |-0,07| the 7~cents are negative, but this cannot be seen from the % integer part, since $-0$ is $0$ in \TeX\ (in fact in most computer % languages, but not in~MIX~\cite{TAOCP1}), so |\ifnum-0<0| yields false. % |\FCs@gn| is a general purpose flag. Its first use is to capture % the sign. % \begin{macrocode} \let\FCs@gn=\@ne\relax \setbox0\hbox\bgroup$% % \end{macrocode} % Do the scan inside a box and inside math mode. Start with % defining all characters that may appear as the first one in an |F|-column % as active. % \begin{macrocode} \uccode`\~=`0\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=0} \uccode`\~=`1\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=1} \uccode`\~=`2\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=2} \uccode`\~=`3\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=3} \uccode`\~=`4\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=4} \uccode`\~=`5\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=5} \uccode`\~=`6\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=6} \uccode`\~=`7\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=7} \uccode`\~=`8\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=8} \uccode`\~=`9\relax \uppercase{\def~}{\restorem@thcodes \global\FC@l=9} % \end{macrocode} % For the input decimal mark something extra is needed: if it is the % first character in an |F|-column (like in |,07|), it should also % restore the |\mathcode|s of the digits. Checking whether or not it % is the first is easy, since in that case the |\mathcode|s of the % decimal digits is still |"8000|. The assignment to |\FC@r| starts % with~1, so that appended digits get captured correctly, even if they % start with~0. Postprocessing of |\FC@r| is done in |\e@fi|. % The input decimal mark switches itself off as active character, % so at most one input decimal separator is allowed (N.B.: % this makes sense). % \begin{macrocode} \def\deactdecm@rk##1##2\relax{\mathcode`##1=0 }% \afterassignment\setucc@de\count@#3\relax \uppercase{\def~}{\ifnum\mathcode`\0=\mathcode`- \restorem@thcodes\fi \afterassignment\deactdecm@rk\count@#3\relax \global\FC@r=1}% % \end{macrocode} % The input grouping character effectively expands to ``nothing, i.e., % ignore'' in a complicated way: it ignores the character and resumes % scanning the number. The test prior to that action is needed if % the grouping character is the first character encountered in the % |F|-column. Which part to continue with depends on whether or not an % input decimal mark was encountered; that can be checked by looking % at its |\mathcode|. % % The input grouping character is the dot |.|, except when that character % was already chosen as input decimal mark. In that case, the grouping % character will be the comma. This is easy to check because the |\uccode| % of |`\~| is still preserved. % \begin{macrocode} \def\d@cm##1##2{\count@=\mathcode`##1 } \ifnum\uccode`\~=`. \uccode`\~=`,\relax\else \uccode`\~=`.\relax\fi \uppercase{\def~}{\ifnum\mathcode`\0=\mathcode`- \restorem@thcodes\fi \afterassignment\d@cm\count@#3\relax % \end{macrocode} % The |\expandafter| below is necessary because the global assignment % should act after the |\fi|. % \begin{macrocode} \ifnum\count@=\mathcode`- \expandafter\global\FC@l=\the\FC@l \else \expandafter\global\FC@r=\the\FC@r\fi}% % \end{macrocode} % The signs are relatively simply: record the sign, restore |\mathcode|s if % needed (it should be: a minus sign between digits screws up everything), % and start scanning the number. % \begin{macrocode} \uccode`\~=`+\relax \uppercase{\def~}{\ifnum\mathcode`\0=\mathcode`- \restorem@thcodes\fi\global\FC@l=0}% \uccode`\~=`-\relax \uppercase{\def~}{\ifnum\mathcode`\0=\mathcode`- \restorem@thcodes\fi\global\let\FCs@gn\m@ne \global\FC@l=0}% % \end{macrocode} % Now actually activate all these codes. The first is simple, but % after that, one can't say |"8000| anymore because |0| acts as active. % But copying |\mathcode|s still works. % \begin{macrocode} \mathcode`-="8000 \mathcode`+=\mathcode`- \mathcode`.=\mathcode`- % \end{macrocode} % These three remain active until the |$| in |\e@fi| is encountered. % The following ones will, in the general case, have their activeness % turned off at some time. % \begin{macrocode} \def\actdecm@rk##1##2\relax{\ifx##1.\relax \mathcode`,=\mathcode`- \else \mathcode`##1=\mathcode`- \fi}% \afterassignment\actdecm@rk\count@#3\relax \mathcode`\0=\mathcode`- \mathcode`\1=\mathcode`- \mathcode`\2=\mathcode`- \mathcode`\3=\mathcode`- \mathcode`\4=\mathcode`- \mathcode`\5=\mathcode`- \mathcode`\6=\mathcode`- \mathcode`\7=\mathcode`- \mathcode`\8=\mathcode`- \mathcode`\9=\mathcode`- } % \end{macrocode} % \end{macro} % \begin{macro}{\e@fi} % If the digits are still active then either nothing was entered or only % characters that did not deactivate the digits were entered. In % either case the output should be \. To flag this outside % the group that started with the opening |$| of |\b@fi|, |\FC@r| is set % globally to a negative value. This doesn't harm, because it didn't % contain relevant information anyway. Outside the group, the sign % of |\FC@r| can then be tested. This is a slight misuse of this % |\count|, but now it's documented. % % If there was no decimal sign or if there was a decimal sign but no % decimal part, |\FC@r| will still be~1, which doesn't parse well with % |\secd@xt|, so a zero is appended. % \begin{macrocode} \def\e@fi{\ifnum\mathcode`\0=\mathcode`- \global\FC@r=\m@ne\fi$\egroup \ifnum\FC@r>0 \ifnum\FC@r<10 \multiply\FC@r by 10 \fi % \end{macrocode} % Next is a loop for bringing the decimal part in the correct way to the % the integer part. The loop is performed the number of decimal % digits to be printed (the~2 in |3,2| of the default setting). This also % means that if you specified more decimal digits than this, the excess % digit(s) will not be handled and a |\PackageWarning| will be given. % \begin{macrocode} \def\i@ts##1##2{\count0=##2} \afterassignment\i@ts\count@\sp@l \loop\ifnum\count0>0 \multiply\FC@l by 10 \expandafter\secd@xt\number\FC@r\end \advance\count0 by \m@ne \repeat \ifnum\FC@r>10 \def\tw@l##1##2\relax{##2} \PackageWarning{fcolumn}{Excess digit\ifnum\FC@r>100 s\fi\space ``\expandafter\tw@l\number\FC@r\relax'' in decimal part \MessageBreak ignored} \fi % \end{macrocode} % Don't forget to correct for the sign (once this is done, |\FCs@gn| % is free again and can and will be used for other purposes). % Then output the result. % \begin{macrocode} \ifx\FCs@gn\m@ne\relax \FC@l=-\FC@l \fi \zetg@ld{\sp@l}{\FC@l}% \fi} % \end{macrocode} % \end{macro} % \begin{macro}{\secd@xt} % The second digit from the left is needed from a decimal number. The % macro |\secd@xt| extracts that digit, provided that the number has at % least two digits, but that is guaranteed by |\e@fi|. That second digit % is then added to |\FC@l|. A new number is assigned to |\FC@r|, that % consists of the digits of |#1#3|. If |#3| was empty, a zero is appended. % In this way |\FC@r| is prepared for insertion in the next invocation of % |\secd@xt|. In iterating: 1234 yields~134, yields~14, yields~10, % stays~10, etc. % \begin{macrocode} \def\secd@xt#1#2#3\end{\advance\FC@l by #2 \FC@r=#1#3 \ifnum\FC@r<10 \multiply\FC@r by 10 \fi} % \end{macrocode} % \end{macro} % \begin{macro}{\restorem@thcodes} % As shown above, once the first digit, or sign, or decimal separator, or % grouping character is scanned, the decimal digits should loose their % activeness. That is done here rather blunt, since the actual % |\mathcode| is not important---as long as it is not |"8000|---because % the digits are not used for typesetting (and even if they were; it's % inside |\box0|, whose contents will be discarded). When the |$| % in |\e@fi| is encountered, the digits get back their original % |\mathcode|s so that the actual typesetting in |\zetg@ld| is % correct again. % \begin{macrocode} \def\restorem@thcodes{\mathcode`\0=0 \mathcode`\1=0 \mathcode`\2=0 \mathcode`\3=0 \mathcode`\4=0 \mathcode`\5=0 \mathcode`\6=0 \mathcode`\7=0 \mathcode`\8=0 \mathcode`\9=0 } % \end{macrocode} % \end{macro} % \subsection{Adaptations to existing macros} % \begin{macro}{\@array} % The definition of |\@array| had to be extended slightly because it should % also include |\@mksumline| (acting on the same |#2| as |\@mkpream| gets). % This change is transparant: it only adds functionality and if you don't % use that, you won't notice the difference. It starts by just copying the % original definition from v2.4k of the |array| package~\cite{array}. % \begin{macrocode} \def\@array[#1]#2{% \@tempdima \ht \strutbox \advance \@tempdima by\extrarowheight \setbox \@arstrutbox \hbox{\vrule \@height \arraystretch \@tempdima \@depth \arraystretch \dp \strutbox \@width \z@}% % \end{macrocode} % Here comes the first change: after each |\\| (or |\cr| for that matter) % the \meta{count} |\FCsc@l| should be reset. This is easiest done with % |\everycr|, but |\everycr| is put to |{}| by |\ialign|, so that definition % should change. The resetting should be done globally. % \begin{macrocode} \def\ialign{\everycr{\noalign{\global\FCsc@l=0 }}% \tabskip\z@skip\halign} % \end{macrocode} % Then the definition is picked up again. % \begin{macrocode} \begingroup \@mkpream{#2}% \xdef\@preamble{\noexpand \ialign \@halignto \bgroup\@arstrut\@preamble\tabskip\z@\cr}% \endgroup % \end{macrocode} % The combination |\endgroup| followed by |\begingroup| seems redundant, % but that is not the case: the |\endgroup| restores everything that was % not |\global|. With the following |\begingroup| it is ensured that % |\@mksumline| experiences the same settings as |\@mkpream| did. % \begin{macrocode} \begingroup \@mksumline{#2}% \endgroup % \end{macrocode} % As a side product of |\@mksumline| also the \meta{count}s for the totals % and \meta{dimen}s for the widths of the colums are created. The columns % should start fresh, i.e., totals are~0 and widths are 0\thinspace pt. % \begin{macrocode} \res@tsumline % \end{macrocode} % From here on it is just the old definition of |array.sty|. % \begin{macrocode} \@arrayleft \if #1t\vtop \else \if#1b\vbox \else \vcenter \fi \fi \bgroup \let \@sharp ##\let \protect \relax \lineskip \z@ \baselineskip \z@ \m@th \let\\\@arraycr \let\tabularnewline\\\let\par\@empty \@preamble} % \end{macrocode} % Because |\@array| was changed here and it is this version that should % be used, |\@@array| should be |\let| equal to |\@array| again. % \begin{macrocode} \let\@@array=\@array % \end{macrocode} % \end{macro} % \subsection{The sumline, close to a postamble} % \begin{macro}{\@mksumline} % The construction of the sumline is much easier than that of the preamble % for several reasons. It may be safely assumed that the preamble specifier % is grammatically correct because it has already been screened by |\@mkpream|. % Furthermore most entries will simply add nothing to |\s@ml@ne|, e.g., |@|, |!|, % and \verb+|+ can be fully ignored. Ampersands are only inserted by |c|, |l|, |r|, % |p|, |m|, and |b|. So a specifier like |@{}lflf@{}| will yield the sumline % |&\a&&\a\\|, (where |\a| is a macro that prints the desired result of the column, % see later). Had the specifier been \verb+l|f||@{ }l|f+, then the same sumline must % be constructed: all difficulties are already picked up and solved in the creation % of the preamble. % % In reality the sumline must be constructed from the expanded form of the % specifier, so |@{}lf@{}| will expand as % |@{}l>{\b@fi{.}{,}{3,2}}r<{\e@fi}@{}|. The rules for constructing the % sumline are now very simple: % \begin{compactitem}[$\bullet$] % \item add an ampersand when |c|, |l|, |r|, |p|, |m|, or |b| is found, % unless it is the first one (this is the same as in the preamble); % \item add a |\a| when |<{\e@fi}| is found; % \item ignore everything else; % \item close with a |\\|. % \end{compactitem} % (In reality also the column check is inserted just before the |\\|, see % |\aut@check|.) To discriminate, a special version of |\@testpach| could % be written, but that is not necessary: |\@testpach| can do all the work, % although much of it will be discarded. Here speed is sacrificed for space % and this can be afforded because the creation of the sumline is done only % once per |\tabular|. % % The start is copied from |\@mkpream|. % \begin{macrocode} \def\@mksumline#1{\gdef\s@ml@ne{}\@lastchclass 4 \@firstamptrue % \end{macrocode} % At first the column number is reset and the actual code for what was % called |\a| above is made inactive. % \begin{macrocode} \global\FCsc@l=0 \let\prr@sult=\relax % \end{macrocode} % Then |\@mkpream| is picked up again. % \begin{macrocode} \@temptokena{#1} \@tempswatrue \@whilesw\if@tempswa\fi{\@tempswafalse\the\NC@list}% \count0\m@ne \let\the@toks\relax \prepnext@tok % \end{macrocode} % Next is the loop over all tokens in the expanded form of the specifier. The change % with respect to |\@mkpream| is that the body of the loop is now only dealing % with F-classes~0, 2, and 10. What to do in those cases is of course different from % what to do when constructing the preamble, so special definitions are created, see % below. % \begin{macrocode} \expandafter \@tfor \expandafter \@nextchar \expandafter :\expandafter =\the\@temptokena \do {\@testpach \ifcase \@chclass \@classfz \or \or \@classfii \or \or \or \or \or \or \or \or \@classfx \fi \@lastchclass\@chclass}% % \end{macrocode} % And the macro is finished by applying the |\aut@check| and appending the % |\\| to the sumline. Note that the |\aut@check| is performed {\itshape % in\/} the last column, but since it does not put anything in the % horizontal list---it only writes to screen and transcript file---, this % is harmless. % \begin{macrocode} \xdef\s@ml@ne{\s@ml@ne\noexpand\aut@check\noexpand\\}} % \end{macrocode} % \end{macro} % \begin{macro}{\@addtosumline} % Macro |\@addtosumline|, as its name already suggests, adds something to the % sumline, like its counterpart |\@addtopreamble| did to the preamble. % \begin{macrocode} \def\@addtosumline#1{\xdef\s@ml@ne{\s@ml@ne #1}} % \end{macrocode} % \end{macro} % \begin{macro}{\@classfx} % Class~f10 for the sumline creation is a stripped down version of |\@classx|: add an % ampersand unless it is the first. It deals with the specifiers |b|, |m|, |p|, |c|, % |l|, and |r|. % \begin{macrocode} \def\@classfx{\if@firstamp \@firstampfalse \else \@addtosumline &\fi} % \end{macrocode} % \end{macro} % \begin{macro}{\@classfz} % Class~f0 is applicable for specifiers |c|, |l|, and |r|, and if the arguments of |p|, % |m|, or |b| are given. The latter three cases, with |\@chnum| is 0, 1, or~2 should be % ignored and the first three cases are now similar to class~f10. % \begin{macrocode} \def\@classfz{\ifnum\@chnum<\thr@@ \@classfx\fi} % \end{macrocode} % \end{macro} % \begin{macro}{\@classfii} % Here comes the nice and nasty part. Class~f2 is applicable if a~|<| is % specified. This is tested by checking |\@lastchclass|, which should be % equal to~8. Then it is checked that the argument to |<| is indeed % |\e@fi|. This check is rather clumsy but this was the first way, after % many attempts, that worked. It is necessary because the usage of |<| % is not restricted to |\e@fi|: the user may have specified other % \LaTeX-code using~|<|. % \begin{macrocode} \def\@classfii{\ifnum\@lastchclass=8 \edef\t@stm{\expandafter\string\@nextchar} \edef\t@stn{\string\e@fi} \ifx\t@stm\t@stn % \end{macrocode} % If both tests yield |true| then add the macro to typeset everything. % \begin{macrocode} \@addtosumline{\prr@sult} % \end{macrocode} % But we're not done yet: in the following lines of code the appropriate \meta{count}s and % \meta{dimen}s are created, if necessary. Note that |\FCsc@l| was set to~0 in the % beginning of |\@mksumline|, so it is well-defined when |\@classfii| is used. % \begin{macrocode} \global\advance\FCsc@l by \@ne \ifnum\FCsc@l>\FCtc@l % \end{macrocode} % Apparently the number of requested columns is larger than the currently % available number of relevant \meta{count}s and \meta{dimen}s, so new ones % should be created. What is checked here is merely the existence of % |\FCtot@|. If it already exists---although it may not % even be a \meta{count}; that cannot be checked---it is not created by % |fcolumn| and an error is given. In case it is a \meta{count} you're just % lucky, and you could ignore that error, although any change to this % \meta{count} is global anyway, so things will be overwritten. In the % case it is not a \meta{count}, things will go haywire and you'll soon % find out. The remedy then is to rename your \meta{count} prior to % |fcolumn| to avoid this name clash. % \begin{macrocode} \expandafter\ifx\csname FCtot@\romannumeral\FCsc@l\endcsname\relax \expandafter\newcount\csname FCtot@\romannumeral\FCsc@l\endcsname \else \PackageError{fcolumn}{Name clash for }{\expandafter\csname FCtot@\romannumeral\FCsc@l\endcsname is already defined and it may not even be a . If you're\MessageBreak sure it is a , you can press now and I'll proceed, but things\MessageBreak will get overwritten.}% \fi % \end{macrocode} % And the same is applicable for the \meta{dimen}: in case of a name % clash you have to rename your \meta{dimen} prior to |fcolumn|. % \begin{macrocode} \expandafter\ifx\csname FCwd@\romannumeral\FCsc@l\endcsname\relax \expandafter\newdimen\csname FCwd@\romannumeral\FCsc@l\endcsname % \end{macrocode} % If the creation was successful, the \meta{count} |\FCtc@l| should be % increased. % \begin{macrocode} \global\FCtc@l=\FCsc@l \else \PackageError{fcolumn}{Name clash for }{\expandafter\csname FCwd@\romannumeral\FCsc@l\endcsname is already defined and it may not even be a . If you're\MessageBreak sure it is a , you can press now and I'll proceed, but things\MessageBreak will get overwritten.}% \fi \fi \fi \fi} % \end{macrocode} % Once created, it is not necessary to initialise them here because that % is done later in one go. % \end{macro} % \begin{macro}{\sumline} % The command for the sumline has one optional argument: the separation % between the rule and the total. By default this is 2\thinspace pt, but % the user may specify |\sumline[10pt]| if that separation needs to be % 10\thinspace pt. The assignment needs to be global, because it is done % in the first column of the tabular, but is valid for the whole line. % \begin{macrocode} \newdimen\s@mlinesep \def\sumline{\@ifnextchar[\s@mline{\s@mline[2pt]}} \def\s@mline[#1]{\global\s@mlinesep=#1 \s@ml@ne} % \end{macrocode} % \end{macro} % \begin{macro}{\prr@sult} % The macro |\prr@sult| actually puts the information together. It starts % like |\leeg|. % \begin{macrocode} \def\prr@sult{$\egroup \let\e@fi=\relax % \end{macrocode} % Then the information for the last line is computed. It is not sufficient % to calculate the width of the result (in points) to use that as the width % of the rule separating the individual entries and the result. It may be % that the sum is larger (in points) than any of the entries, e.g., when the % result of $6+6$ (using specifier~|3,2|) is typeset. The width of the rule % should be equal to the width of |\hbox{$12{,}00$}| then. On the other % hand the width of the rule when summing $24$ and $-24$ should be that of % |\hbox{$-24{,}00$}| (or |\hbox{$(24{,}00$}|, see above), not the width of % the result |\hbox{$0{,}00$}|. Therefore the maximum of all entry widths, % including the result, was calculated. % \begin{macrocode} \setbox0=\hbox{$\geldm@cro{\sp@l}{\number\csname FCtot@\romannumeral\FCsc@l\endcsname}$}% \ifdim\wd0>\csname FCwd@\romannumeral\FCsc@l\endcsname \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\wd0 \fi \vbox{\hrule width \csname FCwd@\romannumeral\FCsc@l\endcsname \vskip\s@mlinesep \hbox to \csname FCwd@\romannumeral\FCsc@l\endcsname{\hfil\unhbox0}}} % \end{macrocode} % \end{macro} % \subsection{Other checks} % \begin{macro}{\leeg} % This macro is used to overrule the default behaviour of the pair |\b@fi| % and |\e@fi|. It starts with ending the groups in the % same way that |\e@fi| would normally do. Then the effect of |\e@fi| % (that is still in the preamble) is annihilated by |\let|ting it to % be |\relax|. This |\let| is only local to the current column. Then % the argument to |\leeg| is treated in a similar way as |\e@fi| would % do with a typeset number. % % Since the user may from time to time also need a column entry other than % a number in the table, e.g., |\leeg{p.m.}|, this definition is without % at-sign. By defining |\leeg| in this way, instead of |\multicolumn1r{}| % (which contains |\omit|), the default spacing in the column is retained. % It has its normal effect on the column width, but doesn't alter the % width of the sumrule. % \begin{macrocode} \def\leeg#1{$\egroup \let\e@fi=\relax #1} % \end{macrocode} % Note that anything may be given as argument to |\leeg|, so in principle % it can also be used to cheat: |\leeg{0,03}| will insert |0,03| in the % table but it doesn't increase the totals of that column by~3 (assuming % |3,2| coding for the separations). But you won't cheat, won't you? % It may affect the width, so be careful: don't insert the unabridged % version of Romeo and Julia here. % \end{macro} % \begin{macro}{\res@tsumline} % Since all changes to the totals and widths of the columns are global, % they have to be reset actively at the start of a tabular or array. % That is an action by itself, but it may occur more often, on request % of the user, therefore a special macro is defined. A side effect of % this macro is that |\FCsc@l| is reset to~0. This is an advantage: % it should be zero at the beginning of a line in the table (for other % lines this is done by the |\\|). % \begin{macrocode} \def\res@tsumline{\FCsc@l=\FCtc@l\loop\ifnum\FCsc@l>0 \global\csname FCtot@\romannumeral\FCsc@l\endcsname=0 \global\csname FCwd@\romannumeral\FCsc@l\endcsname=\z@ \advance\FCsc@l by \m@ne\repeat} % \end{macrocode} % \end{macro} % \begin{macro}{\resetsumline} % To reset a sumline within a table, it should be done within a |\noalign|. % \begin{macrocode} \def\resetsumline{\noalign{\res@tsumline}} % \end{macrocode} % \end{macro} % \begin{macro}{\aut@check} % If the number of F-columns is even, it is assumed that they are part of % two sets of columns of which each column of the first set should balance % the appropriate column of the second set. If on the other hand the number % of columns is odd, then at least one column has nothing to balance against % and no checking occurs. It is correct to check for oddness of |\FCsc@l| % since this |\aut@check| is only performed in the last column of the % tabular: the value of |\FCsc@l| now equals the number of columns used % in the current tabular (and may differ from |\FCtc@l|). % % The output is only to screen and the transcript file; it doesn't change % the appearance of your document, so in case the assumption is wrong you % can safely ignore the result and go on. The \meta{count}s 0 and~1 are % used here and this can be done because any content of those \meta{count}s % from previous calculations has become irrelevant at this moment. % % If the list |\FC@chklist| is empty, the list for the automatic check is % generated (which will remain empty if |\FCsc@l| is odd). % \begin{macrocode} \def\aut@check{\ifx\@empty\FC@chklist\relax \ifodd\FCsc@l\else \count0=\@ne \count1=\FCsc@l \divide\count1 by \tw@ \loop\ifnum\count1<\FCsc@l \advance\count1 by \@ne \xdef\FC@chklist{\FC@chklist\number\count0,\number\count1;}% \advance\count0 by\@ne \repeat \fi \fi % \end{macrocode} % Then this list is peeled off and processed. % \begin{macrocode} \loop \ifx\FC@chklist\@empty\let\FCs@gn=\@ne\else\let\FCs@gn=\m@ne\fi \ifx\FCs@gn\m@ne \expandafter\fre@t\FC@chklist\end \ifnum\csname FCtot@\romannumeral\count0\endcsname= \csname FCtot@\romannumeral\count1\endcsname\else \PackageWarningNoLine{fcolumn}{F-columns \number\count0 \space and \number\count1 \space do not balance}% \fi \repeat} % \end{macrocode} % When |\aut@check| is finished, |\FC@chklist| is empty again, i.e., well % prepared for the next time it is used. This also means that the default % behaviour kicks in again: if that's not what you want, you should % specify the appropriate |\checkfcolumns| lines again. % \end{macro} % \begin{macro}{\fre@t} % This function eats the first two numbers off |\FC@chklist|. % \begin{macrocode} \def\fre@t#1,#2;#3\end{\count0=#1 \count1=#2 \xdef\FC@chklist{#3}} % \end{macrocode} % \end{macro} % \begin{macro}{\checkfcolumns} % But the assumptions for |\aut@check| may be wrong, therefore manual % control on this checking is also made possible here. The macro % |\checkfcolumns| provides a way to the user to check that the appropriate % columns are balanced (as it should in a balance). Arguments |#1| and % |#2| are the |F|-column numbers to compare. It is the responsibility of % the user to provide the correct numbers here, otherwise bogus output % is generated. If this manual check is inserted, the automatic check % will not be performed. % \begin{macrocode} \def\checkfcolumns#1#2{\noalign{\xdef\FC@chklist{\FC@chklist #1,#2;}}} % \end{macrocode} % \end{macro} % That's it! % \section*{Acknowledgement} % Thanks to Karl Berry for valuable comments regarding the consistency of % the installation procedure of this version. Frank Mittelbach gave % various useful suggestions for improving the input parsing as well as % hints to make the package more \LaTeX-like. % \bibliographystyle{plain} % \begin{thebibliography}{9} % \itemsep0pt % \parsep0pt % \parskip0pt % \bibitem{array} Frank Mittelbach and David Carlisle. % A new implementation of \LaTeX's |tabular| and |array| environment. % \bibitem{booktabs} Simon Fear. Publication quality tables in \LaTeX. % \bibitem{dcolumn} David Carlisle. The |dcolumn| package. % \bibitem{world_economy} According to the IMF \url{www.imf.org}. % \bibitem{CandTB} Donald Knuth, Computers \& Typesetting/B, \TeX: the program. % \bibitem{CandTA} Donald Knuth, Computers \& Typesetting/A, The \TeX book. % \bibitem{TAOCP1} Donald Knuth, The Art of Computers Programming, volume~1. % \end{thebibliography} % \Finale \endinput