rspyai

A Textual app to explore your Rust codebase.

Note

This is a personal project. It might break for some reason.

overview

.
├── Cargo.toml
├── pyproject.toml
├── python
│   └── rspyai
│       ├── __init__.py
│       ├── __pycache__
│       ├── models
│       │   ├── __pycache__
│       │   └── function_summary.py
│       ├── rspyai.pyi
│       ├── settings.py
│       ├── tui.py
│       └── widgets
│           ├── __pycache__
│           ├── function_details.py
│           ├── function_summary.py
│           └── function_tree.py
├── src
│   ├── function.rs
│   ├── lib.rs
│   ├── sample.rs
│   └── scanner.rs
└── uv.lock

flowchart TB
    subgraph Rust["Rust Backend (via PyO3)"]
        Scanner["ProjectScanner"]
        RustFunc["RustFunction"]
        Scanner -->|extracts| RustFunc
        Debug["Debug Utils"]
        
        subgraph RustAPI["Python-exposed API"]
            scan["scan_rust_project()"]
            meta["get_function_metadata()"]
        end
        
        Scanner --> scan
        Scanner --> meta
        RustFunc --> scan
        RustFunc --> meta
    end

    subgraph Python["Python Frontend"]
        subgraph TUI["Textual UI Components"]
            FuncBrowser["FunctionBrowser (Main App)"]
            FuncTree["FunctionTree Widget"]
            FuncDetails["FunctionDetails Widget"]
            FuncSummary["FunctionSummary Widget"]
            
            FuncBrowser -->|contains| FuncTree
            FuncBrowser -->|contains| FuncDetails
            FuncDetails -->|contains| FuncSummary
        end
        
        subgraph Settings["Configuration"]
            Config["Settings (pydantic)"]
            AIConfig["AI Model Settings"]
            Config --> AIConfig
        end
        
        subgraph AI["AI Integration"]
            Agent["pydantic-ai Agent"]
            Summary["Function Summaries"]
            Agent -->|generates| Summary
        end
    end

    RustAPI -->|exposes functions to| Python
    FuncTree -->|calls| scan
    FuncDetails -->|calls| meta
    FuncSummary -->|uses| Agent
    AIConfig -->|configures| Agent

    style Rust fill:#deb887
    style Python fill:#4682b4
    style TUI fill:#2f4f4f
    style AI fill:#556b2f

Loading

This project combines Rust and Python via pyo3 and maturin to provide a Textual-based TUI that explores Rust codebases, highlighting public functions for potential Python exposure.

1. Rust Backend:

   • src/scanner.rs: Recursively scans a Rust project directory, identifying .rs files and extracting public functions using syn and walkdir.
   • src/function.rs: Defines RustFunction and logic for parsing signatures, documentation, and source code.
   • src/lib.rs: Exposes scan_rust_project and get_function_metadata as Python-callable functions via PyO3.

2. Python Frontend:

   • rspyai Python package:
     • rspyai/__init__.py: Imports Rust-implemented functions as scan_rust_project and get_function_metadata.
     • rspyai/tui.py: Implements the FunctionBrowser TUI using Textual. This app:
       • Scans and displays a tree of Rust functions.
       • Shows detailed metadata (signature, doc, source) in a split-pane layout.
       • Integrates with pydantic-ai to generate AI-based summaries of selected functions.
     • rspyai/widgets/*: Modular widgets for tree navigation, details display, and AI-generated summaries.
   • pyproject.toml and Cargo.toml: Configure Python (with maturin) and Rust builds, specifying dependencies and linking the Rust cdylib as a Python extension.

3. AI Integration:

   • pydantic-ai used to streamline calling LLMs.
   • Queries about functions are sent to a chosen AI model to stream summaries of the selected function.

4. Settings:

   • rspyai/settings.py: Defines settings for the TUI and AI that allow:
     • setting the ai model
     • setting the system prompt used by the ai model
     • TODO (add more settings)

requirements

• uv to bootstrap a python environment
• an api key for an LLM provider supported by pydantic-ai that supports streaming responses

Important

by default, rspyai uses openai:gpt-4o and requires an OPENAI_API_KEY set in your environment.

usage

# start the function browser 
uvx rspyai@latest

# start the function scanner at a specific path
uvx rspyai@latest [path_to_rust_project]

# start the function browser using a different ai model
RSPYAI_AI_MODEL=ollama:qwen2.5 uvx rspyai@latest

interactive TUI

the TUI provides:

• function tree browser
• information on each function: signature, docstring, parent file, etc.
• ai-generated summaries with AI agent summary (pydantic-ai)

rspyai-pydantic-core.mov

development

# Clone the repository
git clone https://github.com/zzstoatzz/rspyai.git
cd rspyai

# Install development dependencies
uv sync --dev --all-extras

# Rebuild the pyo3 extension
maturin develop

# Run your local copy of the app
rspyai