Skip to main content

Clean Jupyter notebooks for versioning

Project description

Licence GitHub release PyPI version Python versions CI status Coverage

nb-clean cleans Jupyter notebooks of cell execution counts, metadata, outputs, and (optionally) empty cells, preparing them for committing to version control. It provides both a Git filter and pre-commit hook to automatically clean notebooks before they're staged, and can also be used with other version control systems, as a command line tool, and as a Python library. It can determine if a notebook is clean or not, which can be used as a check in your continuous integration pipelines.

:warning: nb-clean 2.0.0 introduced a new command line interface to make cleaning notebooks in place easier. If you upgrade from a previous release, you'll need to migrate to the new interface as described under Migrating to nb-clean 2.

Installation

To install the latest release from PyPI, use pip:

python3 -m pip install nb-clean

nb-clean can also be installed with Conda:

conda install -c conda-forge nb-clean

In Python projects using Poetry or Pipenv for dependency management, add nb-clean as a development dependency with poetry add --dev nb-clean or pipenv install --dev nb-clean. nb-clean requires Python 3.7 or later.

Usage

Checking

You can check if a notebook is clean with:

nb-clean check notebook.ipynb

or by passing the notebook contents on standard input:

nb-clean check < notebook.ipynb

To also check for empty cells, add the -e/--remove-empty-cells flag. To ignore cell metadata, add the -m/--preserve-cell-metadata flag. To ignore cell outputs, add the -o/--preserve-cell-outputs flag.

nb-clean will exit with status code 0 if the notebook is clean, and status code 1 if it is not. nb-clean will also print details of cell execution counts, metadata, outputs, and empty cells it finds.

Cleaning (interactive)

You can clean a Jupyter notebook with:

nb-clean clean notebook.ipynb

This cleans the notebook in place. You can also pass the notebook content on standard input, in which case the cleaned notebook is written to standard output:

nb-clean clean < original.ipynb > cleaned.ipynb

To also remove empty cells, add the -e/--remove-empty-cells flag. To preserve cell metadata, add the -m/--preserve-cell-metadata flag. To preserve cell outputs, add the -o/--preserve-cell-outputs flag.

Cleaning (Git filter)

To add a filter to an existing Git repository to automatically clean notebooks when they're staged, run the following from the working tree:

nb-clean add-filter

This will configure a filter to remove cell execution counts, metadata, and outputs. To also remove empty cells, use:

nb-clean add-filter --remove-empty-cells

To preserve cell metadata, such as that required by tools such as papermill, use:

nb-clean add-filter --preserve-cell-metadata

To preserve cell outputs, use:

nb-clean add-filter --preserve-cell-outputs

nb-clean will configure a filter in the Git repository in which it is run, and won't mutate your global or system Git configuration. To remove the filter, run:

nb-clean remove-filter

Cleaning (pre-commit hook)

nb-clean can also be used as a pre-commit hook. You may prefer this to the Git filter if your project already uses the pre-commit framework.

Note that the Git filter and pre-commit hook work differently, with different effects on your working directory. The pre-commit hook operates on the notebook on disk, cleaning the copy in your working directory. The Git filter cleans notebooks as they are added to the index, leaving the copy in your working directory dirty. This means cell outputs are still visible to you in your local Jupyter instance when using the Git filter, but not when using the pre-commit hook.

After installing pre-commit, add the nb-clean hook by adding the following snippet to .pre-commit-config.yaml in the root of your repository:

repos:
  - repo: https://github.com/srstevenson/nb-clean
    rev: "2.3.0"
    hooks:
      - id: nb-clean

You can pass additional arguments to nb-clean such as --remove-empty-cells with an args array as follows:

repos:
  - repo: https://github.com/srstevenson/nb-clean
    rev: "2.3.0"
    hooks:
      - id: nb-clean
        args:
          - --remove-empty-cells

Run pre-commit install to ensure the hook is installed, and pre-commit autoupdate to update the hook to the latest release of nb-clean.

Migrating to nb-clean 2

The following table maps from the command line interface of nb-clean 1.6.0 to that of nb-clean 2.0.0.

Description nb-clean 1.6.0 nb-clean 2.0.0
Clean notebook nb-clean clean -i/--input notebook.ipynb | sponge notebook.ipynb nb-clean clean notebook.ipynb
Clean notebook (remove empty cells) nb-clean clean -i/--input notebook.ipynb -e/--remove-empty nb-clean clean -e/--remove-empty-cells notebook.ipynb
Clean notebook (preserve cell metadata) nb-clean clean -i/--input notebook.ipynb -m/--preserve-metadata nb-clean clean -m/--preserve-cell-metadata notebook.ipynb
Clean notebook (preserve cell outputs) nb-clean clean -o/--preserve-cell-outputs notebook.ipynb
Check notebook nb-clean check -i/--input notebook.ipynb nb-clean check notebook.ipynb
Check notebook (remove empty cells) nb-clean check -i/--input notebook.ipynb -e/--remove-empty nb-clean check -e/--remove-empty-cells notebook.ipynb
Check notebook (preserve cell metadata) nb-clean check -i/--input notebook.ipynb -m/--preserve-metadata nb-clean check -m/--preserve-cell-metadata notebook.ipynb
Check notebook (preserve cell outputs) nb-clean check -o/--preserve-cell-outputs notebook.ipynb
Add Git filter to clean notebooks nb-clean configure-git nb-clean add-filter
Remove Git filter nb-clean unconfigure-git nb-clean remove-filter

Copyright

Copyright © 2017-2022 Scott Stevenson.

nb-clean is distributed under the terms of the ISC licence.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

nb-clean-2.3.0.tar.gz (22.4 kB view hashes)

Uploaded source

Built Distribution

nb_clean-2.3.0-py3-none-any.whl (19.6 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page