📦 typepeek

typepeek is a lightweight Python package that infers accurate, human-readable type hints from runtime data — including nested and complex containers like lists, dictionaries, tuples, sets, and even third-party objects like PyTorch tensors.

🚀 Quick Start

Installation

pip install typepeek

Example Usage

from typepeek import infer_type
import torch

data = [torch.tensor(1), torch.tensor(2), 3]
print(infer_type(data))
# Output: List[Union[torch.Tensor, int]]
print(infer_type(data, agnostic=False))
#typing.List[torch.Tensor, torch.Tensor, int]

✨ Features

• ✅ Precise Type Inference — Accurately infers human-readable type hints from runtime values
• 🔁 Deep Nested Structure Support — Handles arbitrarily nested containers (e.g., List[Dict[str, Tuple[int, float]]])
• 🧹 Third-Party Object Compatibility — Understands common libraries like torch.Tensor, np.ndarray, and more
• 🔄 Ordered and Unordered Type Support — Handles both ordered collections (e.g., List[int, float, str, int]) and unordered collections (e.g., List[Union[int, float, str]]).

📚 Examples

infer_type([1, 2, 3])
# typing.List[int]

infer_type(["a", 1, 3.14])
# typing.List[typing.Union[str, int, float]]

infer_type({"name": "Alice", "age": 30})
# typing.Dict[str, Union[str, int]]

infer_type((1, "hello", 3.5), agnostic=False)
#typing.Tuple[int, str, float]

infer_type([[1, 2], [3, 4]])
# typing.List[typing.List[int]]

infer_type([torch.tensor(1), np.array(2)], agnostic=False)
#typing.List[torch.Tensor, numpy.ndarray]

🛠 Use Cases

• 📦 Auto-generate type hints for untyped or runtime-generated data
• 🧪 Write better tests for dynamic outputs
• 🧠 Debug and inspect complex runtime object structures

🙌 Contributing

Contributions are welcome! If you have an idea, bug, or feature request, feel free to open an issue or submit a pull request.

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.

👤 Author

👨‍💻 Le Hoang Viet
🐙 GitHub: Mikyx-1