% $Id: note.tex,v 1.2 1999/06/10 05:51:53 loreti Exp $

\documentclass[a4paper]{article}
\usepackage{verbatim}

\newdimen\boxfigwidth
\newenvironment{bigbox}
{\begingroup\boxfigwidth=\hsize%
  \advance\boxfigwidth by -2\fboxrule%
  \advance\boxfigwidth by -2\fboxsep%
  \setbox4=\vbox\bgroup\hsize\boxfigwidth%
  \hrule height0pt width\boxfigwidth\par\bigskip%
  \linewidth=\boxfigwidth%
}{\smallskip\egroup\fbox{\box4}\endgroup}

\makeatletter
\renewcommand{\fnum@figure}{\textsc{\figurename~\thefigure}}
\renewcommand{\section}{\@startsection {section}{1}{\z@}%
  {-3.5ex \@plus -1ex \@minus -.2ex}%
  {2.3ex \@plus.2ex}%
  {\normalfont\Large\sffamily\bfseries}}
\makeatother

\begin{document}
\title{\scshape How to use ``Old-Style Digits''\\
  with the Lucida fonts}
\author{\textsc{Maurizio Loreti}\thanks{
  EMail: \texttt{loreti@pd.infn.it}}}
\date{Rev.\ 1.2\quad--\quad 1999-06-10}
\maketitle
\noindent As a system administrator and local \TeX pert, I
have been asked by an user to build a style file to
automagically switch all the occurrences of a digit in the
printed text to the so-called \emph{old-style digits}.

Although I do not share the aesthetic preferences of this
user, I have solved the problem satisfactorily; the main
difficulty has been to understand how to define a
\emph{virtual font}, because guidelines and examples were
difficult to find.
Here follows a short description of what I have done, hoping
that other people wishing to define a virtual font will find
it useful; I want to excuse myself for my bad english, and
to thank Mr.\ Allin Cottrell, whose examples for Computer
Modern have been an important starting point\/\footnote{They
  may be found under his web page in
  \texttt{http://www.wfu.edu/\textasciitilde cottrell}.}.

A couple of notes: the old-style digits are in the
LucidaNewMath-AltItalic font (normal weight) and in the
LucidaNewMath-AltDemiItalic font (DemiBold weight); the
latter is in the ``Lucida Expert'' font package, that is
\emph{required} in order to obtain \emph{bold} old-style
digits.

All the \verb|.fd|, \verb|.tfm| and \verb|.vf| auxilliary
files are built with Unix command procedures for the
\verb|bash| shell, freely available from the Free Software
Foundation\/\footnote{By anonymous ftp from
  \texttt{prep.ai.mit.edu} (and its mirrors) in
  \texttt{/pub/gnu}; download \texttt{/pub/gnu/GNUinfo/FTP}
  for a list of the mirror sites.}; the only used features
missing from the usual Bourne shell environment
(\verb|/bin/sh|) are:
\begin{itemize}
\item The parameter expansion commands like
  \texttt{\$\{par\%word\}} and \texttt{\$\{par\#word\}}:
  they first expand ``word'' to a pattern, and then remove
  from the end (the former) or from the beginning (the
  latter) of the expansion of the environment variable
  ``par'' the substring matching that pattern (in a Bourne
  shell environment they may be easily simulated with
  \verb|sed| pipes).
\item The command substitution syntax: \verb|$(command)| is
  the \verb|bash| equivalent of the Bourne shell construct
  \verb|`command`|.
\end{itemize}

All the file manipulation is done with \verb|awk| programs;
in our installation, the \verb|GNU| version of \verb|awk|
comes first in the path: but the \verb|GNU| extensions to
the language are not used, and any \verb|awk| flavour should
understand these programs.

\section{The style file}
\verb|lucold.sty| is the style file; it has two options,
\verb|text| (or \verb|notext|) and \verb|math| (or
\verb|nomath|), to select (or, respectively, deselect)
old-style digits in the text mode and in the math mode.
The default options are \verb|[text,nomath]|; when
\verb|notext| and \verb|math| are both explicitely selected,
a warning is printed and \verb|notext| changed to
\verb|text|: since having old-style digits in the
mathematics and not in the text has poor meaning (in my
opinion, of course).

\verb|lucold.sty| is straightforward, being just a wrapper
around the single command that remaps the text default font
family to \verb|hlos| (in the current encoding), and the
additional commands that remap the math-mode digits to the
text font; the style file for Lucida, \verb|lucidabr.sty|,
normally uses \verb|hlh| as default text font family.

\section{The font definition files}
The second step is the creation of the \verb|.fd| files for
\verb|hlos| in the needed encodings; the \verb|OT1| (or
Knuth) encoding is the default, but here in Italy we mainly
use the \verb|T1| (or Cork) encoding.

\begin{figure}[htbp]
  \begin{bigbox}
\begin{verbatim}
  #! /bin/bash

  cp $TETEXDIR/texmf/tex/latex/lucidabr/*t1hlh.fd .

  for infile in *.fd
    do outfile="${infile%h.fd}os.fd"
    awk >$outfile -f makefd.awk -v infile=$infile
  done

  rm *t1hlh.fd
\end{verbatim}
  \end{bigbox}
  \caption{the \texttt{makefd} shell procedure.}
  \label{fig:fd}
\end{figure}

The two font definition files \verb|ot1hlos.fd| and
\verb|t1hlos.fd| are easily obtained using
\verb|[o]t1hlh.fd| as a template; the shell procedure in
figure \ref{fig:fd} does exactly that.
The \verb|awk| program \verb|makefd.awk| remaps all the
upright fonts of the \verb|hlos| family to a new set of
(virtual) fonts, called
\begin{center}
  \begin{tabular}{|c|c||c|c|}
    \hline
    & & New & Original \\
    Encoding & Weight & Font & Font \\
    \hline \hline
    \verb|OT1| & Normal & \verb|hlor7t| & \verb|hlhr7t| \\
    \verb|OT1| & Bold   & \verb|hlob7t| & \verb|hlhb7t| \\
    \verb|T1|  & Normal & \verb|hlor8t| & \verb|hlhr8t| \\
    \verb|T1|  & Bold   & \verb|hlob8t| & \verb|hlhb8t| \\
    \hline
  \end{tabular}
\end{center}
using as a template the input file given as a parameter in
the \verb|-v| command option.
All the other variants (italic, slanted, \ldots) of the
\verb|hlos| family are left mapped to the same fonts used in
the original \verb|.fd| files.

\section{The \TeX\ font metric files}
The third step is the creation of the \verb|.tfm| files:
essentially, the shell procedure of figure \ref{fig:tfm}
copies some of the metric files of the original fonts of the
\verb|hlh| family to the current directory; then converts
them to an human-readable form (\verb|.pl| files) using
\verb|tftopl|; then, using \verb|maketfm.awk|, substitutes
the section describing the digits with the corresponding
section taken from the \verb|.tfm| files of the fonts with
the old-style digits, i.e.\ \verb|hlcrima| (normal weight)
and \verb|hlcdima| (boldface).

\begin{figure}[htbp]
  \begin{bigbox}
\begin{verbatim}
  #! /bin/bash

  cp $TETEXDIR/texmf/fonts/tfm/bh/lubright/hlh??t.tfm .
  cp $TETEXDIR/texmf/fonts/tfm/bh/lumath/hlc?im*.tfm .

  for file in *.tfm
    do tftopl $file ${file%tfm}pl
  done

  awk >hlor7t.pl -f maketfm.awk \
      -v infile1=hlhr7t.pl -v infile2=hlcrima.pl
  awk >hlob7t.pl -f maketfm.awk \
      -v infile1=hlhb7t.pl -v infile2=hlcdima.pl
  awk >hlor8t.pl -f maketfm.awk \
      -v infile1=hlhr8t.pl -v infile2=hlcrima.pl
  awk >hlob8t.pl -f maketfm.awk \
      -v infile1=hlhb8t.pl -v infile2=hlcdima.pl

  pltotf hlor7t.pl hlor7t.tfm
  pltotf hlob7t.pl hlob7t.tfm
  pltotf hlor8t.pl hlor8t.tfm
  pltotf hlob8t.pl hlob8t.tfm

  rm hl???t.pl hlc?im*.tfm
\end{verbatim}
  \end{bigbox}
  \caption{the \texttt{maketfm} shell procedure.}
  \label{fig:tfm}
\end{figure}

Next, the merged \verb|.pl| files are converted to binary
\verb|.tfm| files with \verb|pltotf|; and, finally, the few
files that are no more needed are removed.

\section{The virtual font files}
\begin{figure}[htbp]
  \begin{bigbox}
\begin{verbatim}
  #! /bin/bash

  cp $TETEXDIR/texmf/fonts/vf/bh/lubright/hlh??t.vf .

  for file in hlh??t.vf
    do base="${file%vf}"
    vftovp ${base}vf ${base}tfm ${base}vpl
  done

  to="hlor7t"; from="hlhr7t.vpl"; mft="hlcrima"
  ck="$(grep CHECKSUM ${mft}.pl | cut -d" " -f3 | sed 's/)//')"
  awk >tmp -f maketfm.awk -v infile1=$from -v infile2=${mft}.pl
  awk >${to}.vpl -f makevf.awk -v infile=tmp \
      -v mft=$mft -v ck=$ck
  vptovf ${to}.vpl ${to}.vf ${to}.tfm
  rm tmp

  to="hlob7t"; from="hlhb7t.vpl"; mft="hlcdima"
  ck="$(grep CHECKSUM ${mft}.pl | cut -d" " -f3 | sed 's/)//')"
  awk >tmp -f maketfm.awk -v infile1=$from -v infile2=${mft}.pl
  awk >${to}.vpl -f makevf.awk -v infile=tmp \
      -v mft=$mft -v ck=$ck
  vptovf ${to}.vpl ${to}.vf ${to}.tfm
  rm tmp

  to="hlor8t"; from="hlhr8t.vpl"; mft="hlcrima"
  ck="$(grep CHECKSUM ${mft}.pl | cut -d" " -f3 | sed 's/)//')"
  awk >tmp -f maketfm.awk -v infile1=$from -v infile2=${mft}.pl
  awk >${to}.vpl -f makevf.awk -v infile=tmp \
      -v mft=$mft -v ck=$ck
  vptovf ${to}.vpl ${to}.vf ${to}.tfm
  rm tmp

  to="hlob8t"; from="hlhb8t.vpl"; mft="hlcdima"
  ck="$(grep CHECKSUM ${mft}.pl | cut -d" " -f3 | sed 's/)//')"
  awk >tmp -f maketfm.awk -v infile1=$from -v infile2=${mft}.pl
  awk >${to}.vpl -f makevf.awk -v infile=tmp \
      -v mft=$mft -v ck=$ck
  vptovf ${to}.vpl ${to}.vf ${to}.tfm
  rm tmp *.pl *.vpl hlh*.tfm hlh*.vf
\end{verbatim}
  \end{bigbox}
  \caption{the \texttt{makevf} shell procedure.}
  \label{fig:vf}
\end{figure}

The last step is the generation of the \verb|.vf| files;
their human-readable form (the \verb|.vpl| files) are
essentially \verb|.pl| files, with the addition of some
sections: a \verb|VTITLE|, one or more \verb|MAPFONT| records
describing the real fonts, and a \verb|MAP| entry in the
record of every character describing how to obtain it from
a real font.
In the figure \ref{fig:vf}, the shell procedure that
builds the \verb|.vf| files is listed; it performs the
following steps:
\begin{itemize}
\item The \verb|.vf| files for the original fonts are copied
  to the current directory, and converted to \verb|.vpl|
  files using \verb|vftovp|;
\item The checksum of the \verb|.tfm| file from which the
  old-style digits are taken is extracted from the
  corresponding \verb|.pl| file, and stored in the \verb|ck|
  environment variable;
\item The program \verb|maketfm.awk| is executed, to
  substitute the section of every \verb|.vpl| file
  describing digits with the corresponding section for the
  old-style digits (taken from the relevant \verb|.pl|
  file): note that in that step the \verb|MAP| entry for
  every digit is lost;
\item The program \verb|makevf.awk|, that adds
  to the \verb|.vpl| file a valid \verb|MAPFONT| record and
  valid \verb|MAP| entries for every digit, is executed;
\item The \verb|.vpl| files are converted to binary using
  \verb|vptovf|, and a final cleanup is performed.
\end{itemize}

What is left is simply to move the \verb|*.sty|,
\verb|*.fd|, \verb|*.tfm| and \verb|*.vf| files to the right
directory of the \TeX\ tree.

\section{History and Copyright}
The first version of this package (1.0) has been updated
because the font names in the default Lucida files
distributed with the \LaTeX\ PSNFSS package have changed;
the second version (1.1) stated in an improper way the
license conditions.

This package is free software; you can redistribute it
and/or modify it under the terms of the GNU General Public
License (GPL) as published by the Free Software Foundation;
either version 2, or (at your option) any later version. The
text of the GNU GPL\footnote{Version 2, June 1991.
  Copyright \copyright\ 1989--1991 Free Software Foundation,
  Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307,
  USA.} is included in the file README of this distribution.
\end{document}