44

Is there an easy way to create a list of all external files (complete path) which are used by a LaTeX document (and its "sub-documents") by

  • \input
  • \include
  • \includegraphics ?

(may I have forgotten some input sources?)

Stefan Kottwitz
  • 231,401
MostlyHarmless
  • 12,160
  • You can parse the .log file (not very trivial, though). You can also use strace or similar tool (but this one counts as hardcore). – Andrey Vihrov Jul 31 '11 at 13:19
  • 1
    My dateiliste package includes a list of TeX files parsed in the document, but without complete path. (And not files used by includegraphics or similar.) – Paŭlo Ebermann Jul 31 '11 at 14:22

6 Answers6

42

The snapshot package gives you a list of the external dependencies of a LaTeX document. Use it by saying

\RequirePackage{snapshot}

before the \documentclass command (to have the information written to a .dep file), or by saying

\RequirePackage[log]{snapshot}

before the \documentclass command (to have the information written to the .log file).

Gonzalo Medina
  • 505,128
  • 3
    Is there any way to use this information during the LaTeX build? I am trying to include all the distributed .tex files during a build using \usepackage{embedfile} and \embedfile{<...>.tex}... – fgysin Feb 28 '13 at 08:38
20

use the perl script mkjobtexmf available with every TeX distribution and run it like

mkjobtexmf --jobname <latex file> --cmd-tex pdflatex

it creates an file <latex file>.fls which shows all used files, e.g. for a testfile named latex6:

PWD /home/voss/Documents
INPUT /usr/local/texlive/2011/texmf.cnf
INPUT /usr/local/texlive/2011/texmf/web2c/texmf.cnf
INPUT /usr/local/texlive/2011/texmf-var/web2c/pdftex/latex.fmt
INPUT latex6.tex
OUTPUT latex6.log
INPUT /usr/local/texlive/2011/texmf-dist/tex/latex/base/article.cls
INPUT /usr/local/texlive/2011/texmf-dist/tex/latex/base/article.cls
INPUT /usr/local/texlive/2011/texmf-dist/tex/latex/base/size10.clo
INPUT /usr/local/texlive/2011/texmf-dist/tex/latex/base/size10.clo
[ ... ]
  • 9
    Or you can simply run pdflatex -recorder <latex-file> to get the same effect. This is what mkjobtexmf does behind the scenes. – Lev Bishop Jul 31 '11 at 20:42
  • 1
    sure, what else should it do? However, you can create a texmf tree depending to the job which is more what -recorder can do. –  Jul 31 '11 at 21:00
  • 1
    the -recorder option is great ! – Suresh Jul 31 '11 at 21:55
  • 3
    -recorder works better than mkjobtexmf when it comes to pgf plots. With -recorder you also get the input files used by pgf for the plots, whereas you dont with mkjobtexmf – masgo Jun 26 '15 at 12:01
  • 1
    Both ways work for me, but I end up with what seems like every line duplicated, so I used this technique with awk. It helped a lot, but I still have things like "filename" and "./filename". I'm trying to make a list of files to bundle up to send to a publisher (who won't download TeXlive?), and I'd like to reduce the list as much as possible. – Liam Oct 23 '18 at 18:23
  • 1
    @LevBishop Excellent: as far as I'm concerned -recorder is The Answer. If this were an answer I'd up-vote it. – Norman Gray Jun 28 '21 at 11:54
11

This is a modified version of the @Gonzales answer with an additional python code to copy the figures to a new folder.

After using snapshot package to generate the .dep file:

\RequirePackage{snapshot}
\documentclass{article}

use the following python code (say copy_figs.py) to copy the figures to a separate folder (for example, figs_used):

"""Copy figures used by document."""
import os
import shutil


DEP_FILE = 'main.dep'
TARGET_DIR = 'other_img/'
EXTENSIONS = ['pdf', 'pdf_tex', 'png']


def copy_image_files():
    with open(DEP_FILE, 'r') as f:
        for line in f:
            if '*{file}' not in line:
                continue
            value = line.split('{')[2].split('}')
            source = value[0]
            _, e = os.path.splitext(source)
            e = e.lower()[1:]
            if e not in EXTENSIONS:
                continue
            print(source)
            shutil.copy(source, TARGET_DIR)


if __name__ == '__main__':
    copy_image_files()

To run the python code:

c:\Python27\python.exe copy_figs.py

in the folder where the Latex file is placed. It is assumed the original figures are in figs subfolder, and those figures used in the Latex file are copied to figs_used subfolder. The code copies .png and .pdf figure files.

0 _
  • 958
imriss
  • 777
  • Extremely useful piece of code. Easy to filter with if '*{file}' not in line:. – JeT May 01 '21 at 21:36
  • This is piece of gold. For inclusion of tables that are in other files, you could add or re.search(r'\.code\.tex',source) or source=="xkeyval.tex" to the last if statement of the function and texobviously to the extensions. Edited it also for my other needs, full code can be accessed here https://gist.github.com/samuelsaari/99f3ffe35a63482698571d713e9ee594 – Samuel Saari Jul 01 '22 at 11:03
9

If you put \listfiles as the very first line of the master file, the name of every used file is dumped to standard output, including style and font definition files. This is plain old LaTeX.

tadm
  • 91
8

latexmk does a good job with this:

latexmk -deps main.tex

It prints everything: other .tex files, .bib files, .sty files, graphics.

japreiss
  • 425
2

Here is a quick-and-dirty implementation of a different approach, that does not require modifications of the source code, nor even recompilation. It just parses the log file, and needs a Unix-like command-line. Just run the following command from a terminal:

~ egrep -o '\./[^>) ]*' document.log

Caveats:

  • it only detect relative files
  • it will not work with funny file names (spaces or > or parentheses)
Vincent Fourmond
  • 486
  • 2
  • 11