pydolce 0.1.2

The sweet way to catch outdated docstrings

Dolce

Because broken docs leave a bitter taste.

Dolce is a tool designed to help you maintain high-quality docstrings in your Python code. It leverages Large Language Models (LLMs) to ensure that your docstrings are consistent with your code.

Installation

pip install pydolce

Usage

dolce check # Check docstrings in all python files in the current directory and subdirectories
dolce check src # Check in specific directory
dolce check src/myproject/main.py # Check in specific file

Example

dolce check tests/samples

outputs:

[ ERROR ] tests/samples/wrong_descr.py:1 add
  ![#f03c15]- DOC300: Docstring states the function does something that the code does not do. (The docstring claims the function multiplies integers, but the code performs addition.)[/#f03c15]
[ ERROR ] tests/samples/behavior.py:4 post_multiplication
  - DOC300: Docstring states the function does something that the code does not do. (The docstring summary 'Add two integers' does not match the code which performs multiplication and an HTTP POST request.)
  - DOC301: Docstring omits a critical behavior that the code performs. (The code performs a critical behavior (HTTP POST request) but the docstring does not mention this behavior.)
[ ERROR ] tests/samples/typos.py:1 add
  - DOC200: Docstring description contains spelling errors. (The docstring DESCRIPTION contains typo: 'intgers' instead of 'integers'.)
  - DOC201: Docstring parameter description contains spelling errors. (The parameter 'a' description contains typo: 'Te' instead of 'The'.)
[ ERROR ] tests/samples/simple.py:1 fibonacci
  - SIG203: parameters missing
  - SIG503: return missing from docstring
[  OK   ] tests/samples/simple.py:6 subtract

Summary:
✓ Correct: 1
✗ Incorrect: 4

Configure

Right now dolce can be configured via pyproject.toml file. You can specify which rules to check and which to ignore. By default it will check all rules.

[tool.dolce]
target = [
  # Set of rules to check
  "DOC300",
]
disable = [
  # Set of rules to ignore
  "SIG201",
]

Rules checked with LLM

Some rules require an LLM to check docstrings (e.g., docstring spelling, docstring description vs code behavior, etc.). By default dolce will try to run locally qwen3:8b model via ollama provider. You can visit the Ollama site for installation instructions.

qwen3:8b has relatively good performance while fitting in an RTX 4060 GPU (8GB VRAM). However, if you want to use a different model or provider you can configure the default options in the pyproject.toml of your project like this:

[tool.dolce]
url = "http://localhost:11434"
model = "codestral"
provider = "ollama"
api_key = "YOUR_API_KEY_ENVIROMENT_VAR"

Signature check

Signature check is done vía docsig. If you add a [tool.docsig] config section in your pyproject.toml file, dolce will load it to configure the signature check.

# Example from docsign documentation 
# https://docsig.readthedocs.io/en/latest/usage/configuration.html
[tool.docsig]
check-dunders = false
check-overridden = false
check-protected = false
disable = [
    "SIG101",
    "SIG102",
    "SIG402",
]
target = [
    "SIG202",
    "SIG203",
    "SIG201",
]

To be implemented

• Add cache system to avoid re-checking unchanged code
• Support for ignoring specific code segments, files, directories, etc
• Support parallel requests ... much more!

---

📦 For Developers

Make sure you have the following tools installed before working with the project:

• uv → Python project and environment management
• make → run common project tasks via the Makefile

Getting Started

Install dependencies into a local virtual environment:

uv sync --all-groups

This will create a .venv folder and install everything declared in pyproject.toml.

Then, you can activate the environment manually depending on your shell/OS:

• Linux / macOS (bash/zsh):

  source .venv/bin/activate
  

• Windows (PowerShell):

  .venv\Scripts\Activate.ps1
  

• Windows (cmd.exe):

  .venv\Scripts\activate.bat
  

Running

uv run dolce check path/to/your/code

Linting, Formatting, and Type Checking

make qa

Runs Ruff for linting and formatting, and Mypy for type checking.

Running Unit Tests

Before running tests, override any required environment variables in the .env.test file.

make test

Executes the test suite using Pytest.

Building the Project

make build

Generates a distribution package inside the dist/ directory.

Cleaning Up

make clean

Removes build artifacts, caches, and temporary files to keep your project directory clean.

Building docs

make docs

Generates the project documentation inside the dist/docs folder.

When building the project (make build) the docs will also be generated automatically and included in the distribution package.

🤝 Contributing

Contributions are welcome! Please ensure all QA checks and tests pass before opening a pull request.

---

_{🚀 Project starter provided by Cookie Pyrate}