ghidra-sleigh

Pre-compiled Sleigh runtime data from Ghidra, packaged as a Python library.

Ghidra's Sleigh language definitions (.sla files) are large binary artifacts compiled from human-readable .slaspec sources. This package compiles them once at build time and ships the results as ordinary Python package data, so that downstream tools — disassemblers, decompilers, binary analysis frameworks — can load them immediately. No Ghidra installation required, no build step at runtime, no Java.

The package version exactly tracks the upstream Ghidra release tag it was compiled from: 12.0.3 here means the files were compiled from Ghidra_12.0.3_build.

Why this exists

Projects that need Sleigh data — such as flatline, a binary lifter and decompiler — typically have to either vendor the entire Ghidra tree or ask users to install Ghidra manually. Both options are heavy — the raw processor definitions alone are hundreds of megabytes when compiled.

ghidra-sleigh solves this by:

• Compiling all 60+ Ghidra processor families by default (~26 MB), covering every ISA Ghidra supports. A lighter build (~10 MB) with only the major ISAs is available via a build option.
• Exposing a simple Python API — one function call returns the path to the data directory.
• Matching the directory layout expected by SleighArchitecture::scanForSleighDirectories(), so it plugs into any Ghidra-compatible consumer without adaptation.

Supported versions

Bundled processors

By default all 60+ Ghidra processor families are compiled and included (~26 MB installed). The most commonly used processors are:

A lighter build (~10 MB) containing only these major ISAs can be produced by setting all_processors to false — see Building from source.

Installation

python -m venv venv
. venv/bin/activate
pip install ghidra-sleigh

Building from source requires a C++ compiler and zlib. If a pre-built wheel is available for your platform it will be installed directly with no build step.

Usage

import ghidra_sleigh

# Absolute path to the bundled runtime data directory.
data_dir = ghidra_sleigh.get_runtime_data_dir()
# → /path/to/site-packages/ghidra_sleigh/data

# Upstream Ghidra tag this package was compiled from.
print(ghidra_sleigh.GHIDRA_TAG)     # "Ghidra_12.0.3_build"
print(ghidra_sleigh.GHIDRA_COMMIT)  # "09f14c92d3da6e5d5f6b7dea115409719db3cce1"

# Tuple of processor names whose .sla files are included.
print(ghidra_sleigh.PROCESSORS)     # ("DATA", "x86", "AARCH64", "ARM", "RISCV", "MIPS")

The data_dir is structured exactly as SleighArchitecture::scanForSleighDirectories() expects:

data/
└── Ghidra/
    └── Processors/
        ├── DATA/data/languages/    ← .cspec / .pspec
        ├── x86/data/languages/     ← .sla / .ldefs / .pspec / .cspec
        ├── AARCH64/data/languages/
        ├── ARM/data/languages/
        ├── RISCV/data/languages/
        └── MIPS/data/languages/

Pass data_dir as the Sleigh root to any Ghidra-compatible library and it will discover all bundled language definitions automatically.

Building from source

The build requires:

• Minimum supported Python version: 3.13+
• Meson ≥ 1.6
• A C++11 compiler.
  • GCC ≥ 4.8, Clang ≥ 3.3, or MSVC ≥ 19.0 (Visual Studio 2015).
• zlib development headers

# Clone with the Ghidra submodule (required for the C++ sources)
git clone --recurse-submodules https://github.com/patacca/ghidra-sleigh
cd ghidra-sleigh

python -m venv venv
. venv/bin/activate
pip install .

Build options:

# Point to an existing Ghidra checkout
pip install --config-settings=setup-args="-Dghidra_root=/path/to/ghidra" .

# Build with only the major ISAs (lighter, ~10 MB)
pip install --config-settings=setup-args="-Dall_processors=false" .

Development

Install development tooling:

python -m venv venv
. venv/bin/activate
pip install -e ".[dev]"

Run checks via tox:

# Run all environments
tox

# Run one environment
tox -e lint
tox -e format
tox -e py

The configured tox environments are:

• lint: lint checks
• format: formatting checks
• py: tests (active interpreter)

Versioning

Each release of ghidra-sleigh is pinned to exactly one Ghidra upstream revision. The package version mirrors the Ghidra release version directly (e.g. 12.0.3 ↔ Ghidra_12.0.3_build), making the relationship immediately obvious.

Relation to upstream Ghidra

The C++ sleighc compiler and all .slaspec source files are taken verbatim from the Ghidra repository at the pinned tag. No Ghidra code is modified; this project only automates the compilation step and wraps the outputs in a Python package.

Future plans

• Automated release pipeline — GitHub Actions workflow that watches for new Ghidra release tags and publishes a matching ghidra-sleigh version automatically, keeping the package in sync with upstream without manual intervention.

• Pre-built wheels — Provide binary wheels for Linux (x86-64, aarch64), macOS (x86-64, arm64), and Windows so that pip install ghidra-sleigh never needs a C++ compiler.

• Custom processor support — API for loading user-supplied .slaspec definitions alongside the bundled ones, enabling experimental or proprietary ISA support without forking the package.

License

Apache-2.0 — see LICENSE.

The bundled compiled data is derived from the Ghidra project, also Apache-2.0.