dirplot 0.2.0

Static treemap bitmaps for directory trees, displayed as inline terminal images

dirplot

dirplot creates static nested treemap images for directory trees. It can display them in the system image viewer (default, works everywhere) or inline inside the terminal using the iTerm2 inline image protocol or the Kitty graphics protocol — detected automatically at runtime.

Features

• Directories are shown as labelled, nested containers, mirroring the actual hierarchy.
• Squarified treemap layout — rectangles are as square as possible for easy reading.
• File area proportional to file size; colour determined by file extension.
• ~500 known extensions mapped to colours from the GitHub Linguist table; unknown extensions get an MD5-stable colour from any Matplotlib colormap.
• Label colour (black/white) chosen automatically based on background luminance.
• Output resolution matches the current terminal window pixel size (via TIOCGWINSZ), or a custom WIDTHxHEIGHT.
• Van Wijk cushion shading gives tiles a raised 3-D appearance (optional).
• Display via system image viewer or inline in the terminal (iTerm2 and Kitty protocols, auto-detected).
• Save output to a PNG file with --output.
• Exclude paths with --exclude (repeatable).
• Works on macOS, Linux, and Windows; WSL2 fully supported.

How It Works

1. Scans the directory tree, collecting each file's path, extension, and size in bytes.
2. Computes a squarified dirplot layout recursively — directories allocate space for their children.
3. Renders to a PNG via Pillow (PIL) at the exact pixel dimensions of the current terminal window (detected via TIOCGWINSZ), or at a custom size when --size is given.
4. Displays via the system image viewer (open / xdg-open) or inline via an auto-detected terminal graphics protocol (iTerm2 or Kitty).

Extension colours come from the GitHub Linguist language colour table (~500 known extensions). Unknown extensions fall back to an MD5-stable colour derived from the chosen --colormap. File label text is automatically black or white depending on the background luminance.

Installation

Note: The recommended install methods below use uv. See the uv installation guide if you don't have it yet. Alternatively, use pip install dirplot to install without uv.

# As a standalone tool (recommended)
uv tool install dirplot

# Into the current environment
pip install dirplot

From GitHub

# As a standalone tool
uv tool install git+https://github.com/deeplook/dirplot

# Into the current environment
pip install git+https://github.com/deeplook/dirplot

Platform Support

This tool has been developed and tested on macOS and is CI-tested on Linux and Windows (Python 3.10+). The default --show mode works on all platforms. The --inline mode requires a terminal emulator that supports the iTerm2 or Kitty graphics protocol; on Windows, only WezTerm currently qualifies. WSL2 is treated as Linux and has full support. Feedback and bug reports are welcome — please open an issue on GitHub.

Usage

# Use it before installing it
uvx dirplot --help

# Show dirplot for the current directory (opens image in system viewer)
dirplot .

# Save to a file without displaying
dirplot . --output dirplot.png --no-show

# Display inline (protocol auto-detected: iTerm2, Kitty, Ghostty)
dirplot . --inline

# Exclude directories
dirplot . --exclude .venv --exclude .git

# Use a different colormap and larger directory labels
dirplot . --colormap Set2 --font-size 18

# Render at a fixed resolution instead of terminal size
dirplot . --size 1920x1080 --output dirplot.png --no-show

# Don't apply cushion shading — makes tiles look flat
dirplot . --no-cushion

Options

Flag	Short	Default	Description
--output	-o	—	Save PNG to this path
--show/--no-show		--show	Display the image after rendering
--inline		off	Display in terminal (protocol auto-detected)
--legend/--no-legend		--no-legend	Show file-extension colour legend
--font-size	-s	12	Directory label font size in pixels
--colormap	-c	tab20	Matplotlib colormap for unknown extensions
--exclude	-e	—	Path to exclude (repeatable)
--size		terminal size	Output dimensions as WIDTHxHEIGHT (e.g. 1920x1080)
--header/--no-header		--header	Print info lines before rendering
--cushion/--no-cushion		--cushion	Apply van Wijk cushion shading for a raised 3-D look
--log/--no-log		--no-log	Use log of file sizes for layout, making small files more visible

Inline Display

The --inline flag renders the image directly in the terminal. The protocol is auto-detected at runtime: terminals that support the Kitty graphics protocol use APC chunks (ESC_G…); all others fall back to the iTerm2 inline image protocol (ESC]1337;File=…).

Terminal	Platform	Protocol
iTerm2	macOS	iTerm2
WezTerm	macOS, Linux, Windows	Kitty & iTerm2
Warp	macOS, Linux	iTerm2
Hyper	macOS, Linux, Windows	iTerm2
Kitty	macOS, Linux	Kitty
Ghostty	macOS, Linux	Kitty

The default mode (--show, no --inline) opens the PNG in the system viewer (open on macOS, xdg-open on Linux, system default on Windows) and works in any terminal.

Windows note: Common Windows shells (PowerShell, cmd, Git Bash) and terminal emulators (Windows Terminal, ConEmu) do not support any inline image protocol. --inline will silently produce no output in these environments. WezTerm is currently the only mainstream Windows terminal emulator that supports inline image rendering (via the Kitty graphics protocol). WSL2 is treated as Linux and has full support.

Note: --inline does not work in AI coding assistants such as Claude Code, Cursor, or GitHub Copilot Chat. These tools intercept terminal output as plain text and do not implement any graphics protocol, so the escape sequences are either stripped or displayed as garbage. Use the default --show mode (system viewer) or --output to save the PNG to a file instead. Or use Pi where it is easily added as an extension.

Python API

The public API is small — build_tree, create_treemap, and the display helpers:

from pathlib import Path
from dirplot.scanner import build_tree
from dirplot.render import create_treemap

# Build the tree and render to a PNG in memory
root = build_tree(Path("/path/to/project"))
buf = create_treemap(root, width_px=1920, height_px=1080, colormap="tab20", cushion=True)

# Save to disk
Path("treemap.png").write_bytes(buf.read())

# In a Jupyter notebook: display inline as a cell output (no save needed)
from PIL import Image
Image.open(buf)  # Jupyter renders PIL images automatically via _repr_png_()

To open the result in the system image viewer or display it inline:

from dirplot.display import display_window, display_inline

buf.seek(0)
display_window(buf)   # system viewer (works everywhere)

buf.seek(0)
display_inline(buf)   # inline in terminal (iTerm2 / Kitty / WezTerm)

Development

git clone https://github.com/deeplook/dirplot
cd dirplot
make test

See CONTRIBUTING.md for full details.

License

MIT