Skip to content

api_reference.md

Latest commit

 

History

History
295 lines (193 loc) · 6.21 KB

api_reference.md

File metadata and controls

295 lines (193 loc) · 6.21 KB

RE-Architect API Reference

This document provides detailed reference documentation for the RE-Architect Python API.

Table of Contents

  1. Core Pipeline
  2. Configuration
  3. Decompiler Integration
  4. Binary Loading
  5. Analysis Components
  6. LLM Integration
  7. Test Generation
  8. Visualization
  9. Data Models
  10. Error Handling
  11. Examples
  12. Version Compatibility
  13. Support

Core Pipeline

ReversePipeline Class

Main entry point for binary analysis.

from src.core.pipeline import ReversePipeline
from src.core.config import Config

config = Config.from_file("config.yaml")
pipeline = ReversePipeline(config)

results = pipeline.analyze(
    binary_path="path/to/binary.exe",
    output_dir="./output",
    decompiler="ghidra",
    generate_tests=True
)

Constructor

__init__(config: Config) Initialize the pipeline with a configuration.

  • Parameters:

    • config (Config): Analysis configuration

Methods

analyze(binary_path, output_dir=None, decompiler="auto", generate_tests=False) Perform complete analysis of a binary.

  • Parameters:

    • binary_path (str | Path): Path to binary
    • output_dir (str | Path, optional): Output directory
    • decompiler (str): "ghidra" | "ida" | "binja" | "auto"
    • generate_tests (bool): Whether to generate test harnesses

  • Returns:

    • dict: Results with functions, data structures, and metadata

  • Raises:

    • DecompilerError, AnalysisError, FileNotFoundError

Configuration

Config Class

Manages configuration from YAML or dictionaries.

from src.core.config import Config

# Load from file
config = Config.from_file("config.yaml")

# Load from dict
config = Config({
    "decompiler": {"default": "ghidra"},
    "llm": {"enable": True, "provider": "openai"}
})

Methods

  • from_file(path: str) -> Config — load from YAML
  • get(key: str, default=None) — retrieve using dot notation
  • set(key: str, value) — update value
  • disable_llm() / enable_llm() — toggle LLM features

Decompiler Integration

DecompilerFactory

Factory for decompiler instances.

from src.decompilers.decompiler_factory import DecompilerFactory

factory = DecompilerFactory()
decompiler = factory.create("ghidra")
if decompiler.is_available():
    results = decompiler.decompile(binary_info)
  • create(name: str) -> BaseDecompiler
  • Returns a decompiler instance (ghidra, ida, binja, or auto)

BaseDecompiler

Abstract base class for all decompilers.

  • is_available() -> bool
  • decompile(binary_info: BinaryInfo) -> DecompiledCode
  • get_decompiler_info() -> dict

Binary Loading

BinaryLoader

Loads and parses binaries.

from src.core.binary_loader import BinaryLoader

loader = BinaryLoader()
binary_info = loader.load("binary.exe", auto_unpack=True)

print(binary_info.format, binary_info.architecture)

BinaryInfo

Container with attributes:

  • path, format, architecture, bit_width, endianness
  • entry_point, sections, symbols, imports, exports
  • compiler, stripped

Analysis Components

  • StaticAnalyzer — static analysis of code
  • DataStructureAnalyzer — recover structs
  • EnhancedStaticAnalyzer — deeper function analysis
from src.analysis.static_analyzer import StaticAnalyzer

analyzer = StaticAnalyzer(config)
results = analyzer.analyze(decompiled_code)

LLM Integration

FunctionSummarizer

Generates LLM-based summaries.

from src.llm.function_summarizer import FunctionSummarizer

summarizer = FunctionSummarizer(config)
summary = summarizer.analyze_function_enhanced(func_info, context)
batch = summarizer.analyze_batch_enhanced(functions, context)

Outputs: FunctionSummary objects with attributes:

  • name, purpose, behavior, complexity_analysis
  • arguments, return_value, side_effects, security_notes
  • optimization_suggestions, confidence_score

Test Generation

TestGenerator

Generates per-function harnesses.

from src.test_generation.test_generator import TestGenerator

generator = TestGenerator(config)
tests = generator.generate(functions, data_structures)

Visualization

Start Web Server

python -m src.visualization.server --host 0.0.0.0 --port 8080

From Code

from src.visualization.server import app
app.run(host="localhost", port=5000, debug=True)

Data Models

  • DecompiledCode: container for functions, strings, types
  • FunctionInfo: function attributes (name, address, size, calls, etc.)
  • Instruction: assembly instruction details

Error Handling

Custom exceptions:

  • DecompilerError — decompilation failure
  • AnalysisError — analysis failure
from src.core.exceptions import DecompilerError, AnalysisError

Examples

Full Pipeline

from src.core.pipeline import ReversePipeline
from src.core.config import Config

config = Config.from_file("config.yaml")
pipeline = ReversePipeline(config)

results = pipeline.analyze("sample_binary.exe", "./analysis", "ghidra", True)

print(f"Functions: {len(results['functions'])}")
print(f"Data structures: {len(results['data_structures'])}")

Custom Decompiler

from src.decompilers.base_decompiler import BaseDecompiler, DecompiledCode

class CustomDecompiler(BaseDecompiler):
    def is_available(self): return True
    def decompile(self, binary_info):
        dc = DecompiledCode(binary_info)
        dc.add_function(address=0x401000, code="int main(){return 0;}", name="main")
        return dc

Version Compatibility

  • v1.0+: Static analysis, LLM integration, web UI
  • v1.1+: Dynamic analysis, batch mode
  • v1.2+: Custom decompiler plugins, advanced test generation

Support