infer-types

A CLI tool to automatically add type annotations into Python code.

The main scenario for using the tool is to help you with annotating a big and old codebase. It won't solve the task for you 100% but will definitely help you tremendously, because many of the functions in the real world have quite simple return types that are easy to infer automatically.

Features:

• 100% automated, get a bunch of type annotations with no effort.
• 100% static, all types are inferred without running the code.
• A lot of heuristics and smart inference.
• Actively uses typeshed to find annotations for unannotated dependencies.

Example

Let's say, you have the following method:

class Database:
    def users_count(self):
        return len(self.users)

Since len always returns int, infer-types is able to infer the return type of the method. So, after running the tool, the code will look like this:

class Database:
    def users_count(self) -> int:
        return len(self.users)

Installation

python3 -m pip install infer-types

Usage

python3 -m infer_types ./example/

The tool will add new import statements that can be duplicated and are located not at the top of the file. To fix it, run isort:

python3 -m isort ./example/

The infer-types tool uses the new fancy syntax for type annotations introduced in Python 3.10. So, instead of Optional[str] it will emit str | None. If your code is supposed to run on an older version of Python, add from __future__ import annotations at the beginning of each file. It will solve the issue and also make startup of your app faster. You can also do that with isort:

python3 -m isort --add-import 'from __future__ import annotations' ./example/

See awesome-python-typing for more tools to help you with annotating your code.

How it works

• Most of heuristics live in astypes package. Check it out learn more about the main inference logic.
• If the same method is defined in a base class, copy the type annotations from there.
• If there are no return statements returning a value, the return type is None.
• If there is a yield statement, the return type is typing.Iterator.
• In some cases, the return type can be guessed from the function name. For example, is_open function is assumed to return bool because it starts with is_.

You can run only a specific heuristic using the --only flag.