nbqa 0.1.17

Run any Python code quality tool on a Jupyter Notebook!

nbQA

Adapter to run any code-quality tool on a Jupyter notebook. This is intended to be run as a pre-commit hook and/or during continuous integration.

Documentation is hosted here.

Prerequisites

If you don’t have pip installed, this Python installation guide can guide you through the process.

Installation

Install nbqa with

$ pip install nbqa

There are no dependencies for nbqa so installation should be lightning-fast. Check your installation with

$ nbqa --version
nbqa 0.1.17

Quickstart

The general syntax is

nbqa <command> <notebook or directory> <args>

where command is any standard Python code quality tool.

Examples

Check static type annotations:

$ nbqa mypy tweet-sentiment-roberta-pytorch.ipynb --ignore-missing-imports
tweet-sentiment-roberta-pytorch.ipynb:cell_10:5: error: Argument "batch_size" to "get_test_loader" has incompatible type "str"; expected "int"

Check any examples in your docstrings are correct:

$ nbqa pytest tweet-sentiment-roberta-pytorch.ipynb --doctest-modules
============================= test session starts ==============================
platform linux -- Python 3.8.2, pytest-5.4.3, py-1.9.0, pluggy-0.13.1
rootdir: /home/marco/tweet-sentiment-extraction
plugins: cov-2.10.0
collected 3 items

tweet-sentiment-roberta-pytorch.ipynb .                                  [100%]

============================== 1 passed in 0.03s ===============================

Format your notebooks using black:

$ nbqa black . --line-length=96 --nbqa-mutate
reformatted tweet-sentiment-roberta-pytorch.ipynb
All done! ✨ 🍰 ✨
1 files reformatted.

Note that if, as in this last example, you expect your notebooks to be modified, you will need to pass the --nbqa-mutate flag (alternatively, you could set mutate=1 in your .nbqa.ini file, see “Configuration”).

Empty __init__.py files

Some tools, such as mypy, require (possibly empty) __init__.py files to be in each subdirectory you wish to analyse. To make nbQA aware of this, you need to pass the --nbqa-preserve-init flag, e.g.

nbqa mypy my_dir/my_subdir/my_notebook.ipynb --nbqa-preserve-init

Alternatively, you could set preserve_init=1 in your .nbqa.ini file (see “Configuration”).

Configuration

You can tell nbQA which config file to use either by using the --nbqa-config flag, or by specifying it in a .nbqa.ini file.

So for example, if you wanted to run mypy in such a way that it respects your .mypy.ini file _and_ with the --pretty flag, then you could either run

nbqa mypy my_notebook.ipynb --pretty --nbqa-config .mypy.ini --nbqa-preserve-init

or, you could put the following in your .nbqa.ini file

[mypy]
addopts = --pretty
config = .mypy.ini
preserve_init = 1

and then simply run

nbqa mypy my_notebook.ipynb

You can also tell nbQA to allow mutations, e.g.

[black]
mutate=1

Usage as pre-commit hook

If you want to use nbqa with pre-commit, here’s an example of what you could add to your .pre-commit-config.yaml file:

- repo: https://github.com/nbQA-dev/nbQA
  rev: 0.1.17
  hooks:
    - id: nbqa
      args: ['flake8']
      name: nbqa-flake8
      additional_dependencies: ['flake8']
    - id: nbqa
      args: ['isort', '--nbqa-mutate']
      name: nbqa-isort
      additional_dependencies: ['isort']
    - id: nbqa
      args: ['mypy']
      name: nbqa-mypy
      additional_dependencies: ['mypy']

Supported third party packages

In theory, nbqa can adapt any Python code-quality tool to a Jupyter Notebook.

In practice, here are the tools it’s been tested with:

• flake8

• black

• pytest

• isort

• mypy (you will need to have __init__ files in each subdirectory)

• doctest (as long as you run it via pytest with the –doctest-modules flag)

See Also

Here are some specialised code quality tools for Jupyter Notebooks:

• flake8-nb;

• black-nb.

Project template from cookiecutter.