"tree" command output with "pure" (7-bit) ASCII output

Question

The "tree" command uses nice box-drawing characters to show the tree but I want to use the output in a "code-page-neutral" context (I know that really there's always a code page, but by restricting it to the lower characters I hope to be free of worries that someone in Ulan Bator sees smiley faces, etc).

For example instead of:

├── include
│  ├── foo
│  └── bar

I'd like something like:

+-- include
| +-- foo
| \-- bar

but none of the "tree" switch combinations I tried gave this (seems more as if they take the box-drawing chars as the baseline and make it yet prettier)

I also looked for box-drawing filters to perform such conversions without finding anything beyond an infinite amount of ASCII art :-). A generic filter smells like something to be cooked-up in 15 mins - plus two more incremental days stumbling into all the amusing corner cases :-)

• Joe: Thanks for the question. I needed this so I could pipe the output of tree into enscript to get contol of the print formatting (using dprint/dprintm from my duplexpr package sourceforge.net/projects/duplexpr/.)
• Sridhar Sarnobat: Thanks for posting. I was a bit stumped trying to copy and paste my plaintext art into html pages with all the garbled fonts.

Answer 1

I'm not sure about this but I think all you need is

tree | sed 's/├/\+/g; s/─/-/g; s/└/\\/g'

For example:

$ tree
.
├── file0
└── foo
 ├── bar
 │  └── file2
 └── file1
2 directories, 3 files
$ tree | sed 's/├/\+/g; s/─/-/g; s/└/\\/g'
.
+-- file0
\-- foo
 +-- bar
 │  \-- file2
 \-- file1
2 directories, 3 files

Alternatively, you can use the --charset option:

$ tree --charset=ascii
.
|-- file0
`-- foo
 |-- bar
 | `-- file2
 `-- file1
2 directories, 3 files

• Stéphane Chazelas: There's no GNU tree, there's no Unix tree, there's no POSIX tree. The only tree implementation I'm aware of is mama.indstate.edu/users/ice/tree.
• terdon: @StephaneChazelas thanks, so the --charset option should always be available then. Answer edited.
• Tom Goodfellow: Splendid! (especially the sed version - for me sed is the wild lands beyond the awk frontier)
• Tom Goodfellow: Now for a homework exercise I could add the other 125 box drawing chars to (try to) make a generic remapper. A crude one would be quite straightforward, eg all 19 cross-overs become just '+' - the two days would be used trying to get something that captures the full strange beauty of box drawings
• rob74: Actually I think the --charset option should be mentioned first - the alternative using sed is instructive, but more complicated...

Answer 2

What about tree --charset unicode ?

|-- boot_print
| |-- config-2.6.32-5-amd64
| |-- grub
| | |-- 915resolution.mod
| | |-- acpi.mod
| | |-- affs.mod
| | |-- afs_be.mod
| | |-- afs.mod
| | |-- aout.mod
| | |-- ata.mod
| | |-- ata_pthru.mod
| | |-- at_keyboard.mod
| | |-- befs_be.mod
| | |-- befs.mod
| | |-- biosdisk.mod
| | |-- bitmap.mod
| | |-- bitmap_scale.mod
| | |-- blocklist.mod
| | |-- boot.img

• Stéphane Chazelas: That's the same as tree --charset nwildner
• user34720: Yup. The other answer was edited so, i didn´t noticed that @terdon gave a more complete answer while i has writing mine ;)

Answer 3

I tried the following to change locale. It also outputs ascii code draw lines same as --charset=ascii.

> LANG=C tree

• Sridhar Sarnobat: Damn, I was hoping this would work for this really nice tool but I guess software is not magic, it needs to be supported: github.com/yzhong52/ascii_tree