GhidraX — Pure-Python Ghidra Decompiler Engine

A faithful Python port of Ghidra's C++ decompiler core. SLEIGH instruction decoding is provided via a pybind11 native module (sleigh_native.pyd); everything else — IR, data-flow analysis, SSA construction, optimization rules, control-flow structuring, and C output — is pure Python.

Git Repository

• Canonical repository: https://github.com/fjqisba/GhidraX.git
• Documentation: https://github.com/fjqisba/GhidraX/tree/main/docs

git clone https://github.com/fjqisba/GhidraX.git
cd GhidraX

Project Structure

GhidraX/
├── src/                        # Python decompiler package (src-layout)
│   └── ghidra/
│       ├── analysis/           # Heritage (SSA), flow, data-flow
│       ├── arch/               # Architecture abstraction
│       ├── block/              # Control-flow structuring
│       ├── core/               # Address, AddrSpace, opcodes, marshal
│       ├── database/           # Symbol database, variable mapping
│       ├── fspec/              # Function signatures, calling conventions
│       ├── ir/                 # Varnode, PcodeOp, Funcdata
│       ├── output/             # PrintC / PrintLanguage emission
│       ├── sleigh/             # SLEIGH engine + native .pyd modules
│       ├── transform/          # Action/Rule optimization chain
│       └── types/              # Type system, casts
├── native/                     # C++ source (Ghidra decompiler + pybind11)
│   ├── CMakeLists.txt
│   ├── build.bat               # One-click Windows build
│   ├── sleigh_bind.cpp         # sleigh_native.pyd binding
│   └── decompiler_bind.cpp     # decompiler_native.pyd binding
├── specs/                      # Ghidra processor specifications
│   └── Processors/x86/data/languages/
├── tools/                      # Utilities, comparison, and deployment helpers
│   ├── deploy.bat              # IDA plugin deployer
│   ├── console.py              # CLI utility
│   └── action_compare.py       # Python/native staged action comparison
├── docs/                       # Documentation
│   ├── ARCHITECTURE.md         # Porting roadmap & design
│   ├── AUDIT.md                # Code audit notes
│   └── progress.md             # Module porting progress
├── pyproject.toml              # Build config + pytest settings
├── LICENSE
└── README.md

Quick Start

1. Build Native Modules

cd native
build.bat

Auto-detects MSVC 2022, CMake, Ninja, Python, pybind11, and zlib. Outputs are copied to src/ghidra/sleigh/.

Prerequisites

Tool	Install
Visual Studio 2022	"Desktop development with C++" workload
CMake ≥ 3.15	winget install Kitware.CMake
Ninja	winget install Ninja-build.Ninja
Python ≥ 3.10	python.org
pybind11	pip install pybind11
zlib (static)	vcpkg install zlib:x64-windows-static

2. Decompile a Function

from ghidra.sleigh.decompiler_python import DecompilerPython

dp = DecompilerPython()
dp.use_python_heritage = True
dp.use_python_rules = True
dp.use_python_printc = True
dp.initialize()

code = dp.decompile(
    sla_path="specs/Processors/x86/data/languages/x86-64.sla",
    target="x86:LE:64:default",
    image=binary_bytes,
    base_addr=0x140000000,
    entry=0x140001000,
)
print(code)

3. Use the C++ Native Engine

from ghidra.sleigh.decompiler_native import DecompilerNative

dn = DecompilerNative()
dn.add_spec_path("specs/Processors/x86/data/languages")
dn.initialize()

code = dn.decompile(
    "specs/Processors/x86/data/languages/x86-64.sla",
    "x86:LE:64:default",
    binary_bytes, 0x140000000, 0x140001000, 0,
)
print(code)

Architecture

Layer	Implementation	Purpose
SLEIGH Engine	C++ pybind11 .pyd	Instruction decode & P-code lifting
Decompiler Core	Pure Python	IR, SSA, optimization, structuring, C output
Native Baseline	C++ pybind11 .pyd	Full C++ Ghidra decompiler for comparison

The C++ source under native/ is the ground truth — Python must produce semantically equivalent output. See docs/ARCHITECTURE.md for the full porting plan.

SLA Specification Files

Processor specs live under specs/Processors/. The SLA search order:

1. Paths added via arch_map.add_sla_search_dir()
2. PYGHIDRA_SLA_DIR environment variable
3. <project_root>/specs/Processors/<arch>/data/languages/

To add architectures, copy .sla + .pspec + .cspec from your Ghidra install and add an entry in src/ghidra/sleigh/arch_map.py.

Development

pip install -e ".[dev]"
python -m pytest tests -v --timeout=120

License

Apache 2.0