Passing command-line arguments to LaTeX document

Question

Sometimes, I define new commands such as the following.

\newcommand{\comment}[1]{\textbf{#1}}
%\necommand{\comment}[1]{\emph{#1}} 

The above commands enable me to change the style of parts of my code all at once. If I want to generate both of the possible styles, I have to compile my LaTeX document two times each time modifying the source code to enable the desired style.

Is there a way to avoid the source code modification in such cases? That is, can I pass latex some command-line arguments so that I can choose which style to use based on that argument?

Answer 1

That is, can I pass latex some command-line arguments so that I can choose which style to use based on that argument?

Yes. Three options:

One

In your source file, write

\providecommand{\comment}[1]{\emph{#1}}% fallback definition

and then compile the LaTeX document ("myfile.tex") as

pdflatex (whatever options you need) "\newcommand\comment[1]{\textbf{#1}}\input{myfile}"

Two

Alternatively,

pdflatex "\let\ifmyflag\iftrue\input{myfile}"

and then have in the source

\ifcsname ifmyflag\endcsname\else
  \expandafter\let\csname ifmyflag\expandafter\endcsname
                  \csname iffalse\endcsname
\fi
...
\ifmyflag
  \newcommand\comment[1]{\emph{#1}}
\else
  \newcommand\comment[1]{\textbf{#1}}
\fi

Three

Or even

pdflatex "\def\myflag{}\input{myfile}"

with

\ifdefined\myflag
  \newcommand\comment[1]{\emph{#1}}
\else
  \newcommand\comment[1]{\textbf{#1}}
\fi

which is probably the shortest, albeit slightly fragile because you never know when a package might define \myflag behind your back.

Answer 2

You should use Will's approaches when you need fairly flexible one-off options, like say changing the position line on your resume. If otoh you are producing the same selection of options over & over, then you should consider avoiding command line arguments, or working them into a build script or makefile.

I'll give two techniques for avoiding command line arguments :

Trick 1: If you're producing a fixed array of documents that must remain accessible, like your two styles example, then I'd recommend simply implementing Will's latex code inside another tex file, i.e. thesis.tex contains a \providecommand\comment[1]{\emph{#1}} and thesis-ugly.tex consists of \newcommand\comment[1]{\textbf{#1}} \input thesis.tex.

You must of course rerun tools like bibtex when using this technique, unless you symlink the intermediary files, ala ln -s thesis.aux thesis-ugly.aux and ln -s thesis.bbl thesis-ugly.bbl.

Trick 2: I found trick 1 awkward for changing document papersizes, so I wrote the following perl script, called simply papersize. The command papersize A4 teaching.tex modifies teaching.tex in place, and symlinks teaching.pdf to teaching-A4.pdf, so that running pdflatex teaching creates teaching-A4.pdf, but does not disturb the pre-existing teaching-letter.pdf and does not require rerunning bibtex teaching. It does obviously require rerunning pdflatex twice for documents with internal references.

#!/usr/bin/perl -i~ -n

BEGIN {
die "Usage: papersize letter/A4/etc. [filename]\n" if ($#ARGV < 0);
$SIZE = shift @ARGV;  @files=@ARGV;
$FLAG = "% paper size :: ";
}

if (/$FLAG(\w+)/) {
    if ($1 eq $SIZE) {
        s/^\% //;
    } else {
        s/^([^\%])/\% \1/;
    }
}
print $_;

END {
foreach (@files) {
    if (s/\.tex//) {
    $l = "$_-$SIZE.pdf";  $_ .= ".pdf";
    unlink($_) if (-l $_);
    symlink($l,$_) if (! -e $_);
} }
}

You must add the special comments % paper size :: ... to every file line that should be changed when you change the paper size.

\documentclass[letterpaper,11pt]{article}  % paper size :: letter
% \documentclass[a4paper,11pt]{article}  % paper size :: A4
\usepackage[text={6.5in,8.8in}]{geometry}  % paper size :: letter
% \usepackage[text={16.4cm,24.5cm}]{geometry}  % paper size :: A4

You could obviously work papersize into a build script or makefile too or modify the above script for .dvi files.. or generalize the script to other modifications.

Answer 3

This may not be the optimal solution for the situation in question (also a very old post), but I thought I'd present an alternative solution which was useful for me. It is simply to define the two commands in two different files, link a third file to the desired choice, and \input that third file. For example, define:

% in main.tex
\input{options}
% ...

.

% in option1.tex
\newcommand{\comment}[1]{\textbf{#1}}

.

% in option2.tex
\newcommand{\comment}[1]{\emph{#1}} 

Then to compile with option1, simply run:

> ln -s option1.tex options.tex && pdflatex main.tex

I prefer this because I am more comfortable with bash than obscure latex, have no problem with a proliferation of files (these can easily be organized in subdirectories), and have scripts compile my document anyway so this additional step comes at little cost.

Answer 4

To provide my dissertation in both the required, ugly, tree wasting format, and a compact prettier version, I used ifthen an a kludge of make and sed that rewrote a bit of the header.

I think Will's approaches are all nicer.