automating recursive file hierarchy (directory tree) diagrams with tikz

Question

So, I've been trying to write a python script that will produce the TikZ/LaTeX source code for one of these filesystem tree / directory structure diagrams:

Preview

---

Code

LaTeX: TikZ

\documentclass[tikz,border=10]{standalone}
\usetikzlibrary{trees}
\tikzstyle{every node}=[draw=black,thick,anchor=west]

\begin{document}

    \begin{tikzpicture}[ %
    grow via three points={one child at (0.5,-0.7) and two children at (0.5,-0.7) and (0.5,-1.4)},
    edge from parent path={(\tikzparentnode.south)|-(\tikzchildnode.west)}]

        \node {/tmp/test}
        %
        child{node{a}
            child{node{1}}
            child{node{2}}
            child{node{3}}}
        child [missing] {} %
        child [missing] {} %
        child [missing] {} %
        %
        child{node{b}
            child{node{1}}
            child{node{2}}}
        child [missing] {} %
        child [missing] {} %
        %
        child{node{c}
            child{node{1}}}
        child [missing] {} %
        %
        child{node{d}};
        %
    \end{tikzpicture}

\end{document}

---

It was supposed to be just a quick little thing, but it turns out programming programs to.. program programs can be bloody confusing. I'd like a way to do it all from within the LaTeX environment (perhaps LuaLaTeX can help?) if possible. Or alternatively some kind of macro or function that takes input in a practical format that can be generated externally and copy/pasted in. Here are a couple of examples of possible ways to generate input that might be suitable (I'm just spitballing here, they won't be perfect, but I can tune them to suit if need be):

---

Additional Info:

Don't be confused by this section, the question isn't about bash or python or whatever, this is just to demonstrate a few ways to generate input, and also to show it can easily be manipulated to the appropriate format. What I need help with is writing a suitable LaTeX macro that can typeset this information properly.

bash: find

in

find .

out

.
./a
./a/1
./a/2
./a/3
./b
./b/1
./b/2
./c
./c/1
./d

---

in

printf '{/tmp/test'; 
find /tmp/test -printf '%P, ';
printf '}\n';

out

{tmp/test, a, a/1, a/2, a/3, b, b/1, b/2, c, c/1, d, }

---

python: os.walk()

in

import os
for (root,dirs,files) in os.walk('.'):
    print(f'{root}\n{dirs}\n{files}')

out

.
['a', 'b', 'c', 'd']
[]
./a
[]
['1', '2', '3']
./b
[]
['1', '2']
./c
[]
['1']
./d
[]
[]

---

in

print('{/tmp/test',end='')
for (root,dirs,files) in os.walk('.'):
    print(f"{root[2:]},[{','.join(dirs+files)}],")
print('}')

out

{/tmp/test,[a,b,c,d],
a,[1,2,3],
b,[1,2],
c,[1],
d,[],
}

Answer 1

It is not an answer as such (but it is too long to put in a commentary). I think it's possible using -shell-escape, the tree command and a few regex (I'm more familiar with perl than python). The first thing is to be able to find a pattern to be able to build the code in tikz. Using :

[pablo@worktex ~] $ pwd
/home/pablo
[pablo@worktex ~] $ tree ltxgit/
ltxgit/
├── build
│   ├── distrib
│   ├── doc
│   ├── local
│   ├── test
│   └── unpacked
├── build.lua
├── mybuild.lua
├── mygitcmd.txt
├── mytree.txt
├── README.md
├── sources
│   ├── README.md
│   └── testfiles
│       ├── test-nospace.tex
│       └── test-pkg.tex
└── texlive
    ├── texlive_install.sh
    ├── texlive_packages
    └── texlive.profile

9 directories, 11 files

The point is to distinguish which are directories and which are just files, then we switch to:

[pablo@worktex ~] $ tree -p ltxgit/
ltxgit/
├── [drwxrwxr-x]  build
│   ├── [drwxrwxr-x]  distrib
│   ├── [drwxrwxr-x]  doc
│   ├── [drwxrwxr-x]  local
│   ├── [drwxrwxr-x]  test
│   └── [drwxrwxr-x]  unpacked
├── [-rw-rw-r--]  build.lua
├── [-rw-rw-r--]  mybuild.lua
├── [-rw-rw-r--]  mygitcmd.txt
├── [-rw-rw-r--]  mytree.txt
├── [-rw-rw-r--]  README.md
├── [drwxrwxr-x]  sources
│   ├── [-rw-rw-r--]  README.md
│   └── [drwxrwxr-x]  testfiles
│       ├── [-rw-rw-r--]  test-nospace.tex
│       └── [-rw-rw-r--]  test-pkg.tex
└── [drwx------]  texlive
    ├── [-rw-rw-r--]  texlive_install.sh
    ├── [-rw-rw-r--]  texlive_packages
    └── [-rw-rw-r--]  texlive.profile

9 directories, 11 files

Differentiate directories from files using -p. Now comes the hard part, finding regularity to pass the code to tikz. We know that the first line corresponds to the root directory and the others follow a regularity based on the beginning of the line |, the spaces and their identification [d.+?]. We can generate a routine and convert it to this:

ltxgit/
├── build/
│   ├── distrib/
│   ├── doc/
│   ├── local/
│   ├── test/
│   └── unpacked/
├── build.lua
├── mybuild.lua
├── mygitcmd.txt
├── mytree.txt
├── README.md
├── sources/
│   ├── README.md
│   └── testfiles/
│       ├── test-nospace.tex
│       └── test-pkg.tex
└── texlive/
    ├── texlive_install.sh
    ├── texlive_packages
    └── texlive.profile

in terms of tikz. I think (I'm not sure about this) that it would be better to use Lua and to do it multiplatform and not depend on tree, but, if you are using python maybe you can play for a while using pythontex until you come up with the solution.

Good luck with this, it would be great to see it in action.

Answer 2

You could do something like the following. This relies on the tree command being present on the system, as well as bash and sed --- so this is useless on Windows, for example --- but the command is run externally, as part of compilation, provided shell escape is enabled.

\documentclass{article}
\usepackage{bashful}
\usepackage[edges]{forest}
\begin{document}
\bash[stdoutFile=mydirtree.tex]
tree -X --noreport /opt/stapler/ | sed -e "s/<tree>/\\\begin{forest}for tree={folder,grow\'=0}/" -e 's/<\/tree>/\\end{forest}/' -e 's/\(<directory name=\)\|\(<file name=\)/[/' -e's/\"\([^\"]*\)\">/\1/' -e 's/<\/directory>/]/' -e 's/<\/file>/]/' -e '/^<?xml version/d' -e 's/_/\\_/g'
\END
\input{mydirtree}

\end{document}

Note that you may need to adjust the sed depending on which characters occur in file names on your system (including directories, which are also files, of course).

I used pdflatex -shell-escape <filename> to compile.