Home / ubuntu / Questions / 660912
In Process
muk.li
Asked: 2015-08-14 03:05:12 +0800 CST

How to grep for all packages used in a tex file

  • 772

Assuming I want to grep all packages used in a tex file:

$ grep usepackage my.tex

will do, with (as far as I can see) one exception: If a package gets loaded with that many options that the author of the file decided to insert line breaks, as in

\usepackage[option1,
option2,
option3]{thepackage}

Would there be a way to tell grep to continue parsing the file until it hits the }?

Ideally I would only get the thepackage as output, so that I could possibly iterate over it, such as:

$ for i in `locate `my grep command`.sty`
do
grep \\\\newcommand{\\\\createstrouble} $i
done

in order to quickly find a package in which a command \createstrouble is defined?

EDIT: One further complication: I have a \usepackage command as follows:

\usepackage[pdftitle={My Title},
 pdfauthor={My Author}]{hyperref}

EDIT: I understood that this approach doesn't work for my purpose, which was to find out conflicting packages. Primarily so, because a package loaded by \usepackage{something} pulls in more packages through RequirePackage which one will not have on ones radar. Better to work with the logfile, possibly using \listfiles in ones preamble, producing a list of files loaded, and grepping through ones texmf directory for the actual command name used by different packages.

latex

2 Answers

  1. You can do multiline matches in regular grep using a combination of the -P (PCRE mode) and -z flags e.g.

    grep -zP '(?s)\\usepackage.+?}' file

    or with pcregrep

    pcregrep -M '(?s)\\usepackage.+?}' file

    The (?s) modifier allows . to match newlines.

    To get just the package name, you could try either

    grep -zPo '(?s)\\usepackage.*?{\K.+?(?=})' file

    or

    pcregrep -Mo '(?s)\\usepackage.*?{\K.+?(?=})' file


    Testing the plain grep expression on a TeX manuscript of my own I get the following package list:

    ~$ grep -zPo '(?s)\\usepackage.*?{\K.+?(?=})' myfile.tex
geometry
changepage
inputenc
textcomp,marvosym
fixltx2e
amsmath,amssymb
cite
nameref,hyperref
lineno
microtype
rotating
setspace
caption
lastpage,fancyhdr,graphicx
epstopdf
    • 2

  2. It's not simple to write a TeX parser. Therefore the following solution can only be try.

    Using awk twice and uniq

    awk '/usepackage/,/}$/ {gsub(/\\/,"\n",$0);printf "%s",$0}' my.tex |\
awk -F{ '{gsub(/}/,"",$NF); if ($NF != "") {print $NF}}' |\
sort -u

    Example

    $ cat foo.tex
\usepackage[pdftitle={My Title},
 pdfauthor={My Author}]{hyperref}
\usepackage{ngerman}
\usepackage[scale=0.85]{geometry}

$ awk '/usepackage/,/}$/ {gsub(/\\/,"\n",$0);printf "%s",$0}' foo.tex |\
  awk -F{ '{gsub(/}/,"",$NF); if ($NF != "") {print $NF}}' |\
  sort -u
geometry
hyperref
ngerman
    • 2