---
id: ast-serialize
version: "0.8.0"
license: MIT
license_treatment: permissive
maintenance: active
---

# ast-serialize — Python bindings for mypy AST serialization

License: permissive · Maintenance: active · Popularity: top 1,000 on PyPI

## Install

pip install ast-serialize
uv add ast-serialize
poetry add ast-serialize

## Description

# ast_serialize - Python Parser and AST Serializer

This is a fast Python extension for parsing Python files and serializing the AST using 
the native binary format used by mypy. This will eventually replace the current mypy parser,
which uses the Python stdlib `ast` module for parsing.

**This is work in progress.**

## Development

Prerequisites:

- Recent Rust toolchain
- Python 3.10+ (3.13t or 3.14t for free-threaded builds)
- [maturin](https://github.com/PyO3/maturin): `pip install maturin`
- Checkout of [mypy repo](https://github.com/python/mypy) for testing

Development build (fast compilation, unoptimized):
```bash
maturin develop
```

Optimized development build:
```bash
maturin develop --release
```

## Testing

**Rust unit tests:**
```bash
cargo test
```

**Python integration tests:** 
Run end-to-end parser and serialization/deserialization tests using mypy's test suite in 
the new-parser branch:
```bash
cd ~/src/mypy  [or wherever you have mypy]
pytest mypy -k NativeParser
```

Add new test cases to `test-data/unit/native-parser.test` (in the mypy repository). See
the main parser test cases in `parse.test` for the expected output format.

**Note:** Run `maturin...

## AI interpretation — verify before relying

A fast Python extension that parses Python source files and serializes their abstract syntax trees using mypy's native binary format, intended to eventually replace mypy's stdlib-based parser.

Verdict: A young but actively maintained Rust-based parser designed for mypy integration, offering performance gains over stdlib ast. Zero known vulnerabilities and permissive licensing make it safe to evaluate, but alpha status and tight coupling to mypy's development cycle mean it is not yet a general-purpose replacement for standard parsing.

[View on SkillFed](https://skillfed.io/packages/ast-serialize) · [View on PyPI](https://pypi.org/project/ast-serialize/)