I want to randomize some of the text contained within some letters. For e.g.:
\random{Dear \name,}{Hello \name,}{Greetings \name,}{Hi \name,}{To \name,}
\random{You are invited to}{You are welcome to}{I would like to invite you to}{This is an invitation to}{You may want to} attend...
When compiled, this would randomly select one of the phrases to use, so that each of the letters is a bit different. How can this be done?
Here an implementation for four arguments based on Marc's answer.
It uses \ifcase instead of \ifodd. Because random(4) returns 1--4, but \ifcase starts from 0, the first case is actually kept empty.
If more than four arguments are wanted simple change [4] and (4) to a higher number and add \or#5 etc. However, more than 9 arguments needs extra work.
\documentclass{article}
\usepackage{pgf}
\newcommand*{\random}[4]{%
\pgfmathparse{random(4)}%
\ifcase\pgfmathresult\relax
\or#1\or#2\or#3\or#4%
\fi%
}
\begin{document}
% Tests:
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\random{1}{2}{3}{4}
\end{document}
Using the very experimental l3rand package (currently in the l3trial directory of the LaTeX3 code repository, which means it is really not stable), you can do the following.
\documentclass{article}
\usepackage{l3rand,xparse}
\ExplSyntaxOn
\rand_seed_from_time:
\cs_generate_variant:Nn \tl_item:nn { nf }
\DeclareDocumentCommand {\random} {m}
{
\tl_item:nf {#1} { \rand_range:nn {0} { \tl_length:n {#1} - 1 } }
\rand_clean:
}
\ExplSyntaxOff
\begin{document}
\newcommand{\name}{Dave}
\random { {Dear \name,} {Hello \name,} {Greetings \name,} {Hi \name,} {To \name,} }
\random { {You are invited to} {You are welcome to} {I would like to invite you to} {This is an invitation to} {You may want to} }
attend...
\end{document}
Currently, \rand_range:nn {<begin>} {<end>} produces a random number between <begin> and <end> inclusive, hence the need to subtract 1. This may very well be changed later. Also, \rand_seed_from_time: might be renamed by the time this solution is used by anyone.
I added braces around the whole argument of \random, otherwise TeX has no easy way of knowing where it ends.
l3rand initialize with a random seed by default, letting the user override it with \rand_seed:n { <int> }, or should it favor reproducibility, with the user typing \rand_seed_from_time: explicitly if he wants random initialization?
Commented
Dec 25, 2011 at 13:13
\tl_length:n should be replaced by \tl_count:n (not doing the change now, since this would bump the answer on the front page, and l3rand is still as experimental as always).
Commented
Jul 18, 2014 at 16:04
The following should do the trick. For some reason I cannot get it to work without loading the entire tikz package (pgfmath should have sufficed).
\usepackage{tikz}
\newcommand*{\random}[2]{%
\pgfmathparse{random(2)}%
\ifodd\pgfmathresult\relax#1\else#2\fi%
}
pgfmath alone. These are fixed in the current develop version. As a workaround loading the pgf package is enough, no need to load the full higher-level tikz layer.
Commented
Dec 25, 2011 at 13:13
\ifodd will expand all tokens behind it until it finds something which is not part of a number. Because \pgfmathresult holds a single digit only the content of #1 is searched for a potential rest of this number. This leads to that e.g. \random{1 always}{never} always picks the first argument, because the 1 is appended to the result, making it always odd. You need to terminate the numerical input of \ifodd explicitly either using a \relax or a \space behind \pgfmathresult.
Commented
Dec 25, 2011 at 13:22