strip_jupyter_notebook.py

#!/usr/bin/env python

"""
This script is used to remove / reset outputs, execution numbers and metadata from .ipynb files for better version control.
I created my own script to do this as I could not find any solution that performs all 3 operations properly.

Usage:
Put this script in the same directory as the jupyter notebook (.ipynb) files and run it.

Note:
Inspired by the jq method here: http://timstaley.co.uk/posts/making-git-and-jupyter-notebooks-play-nice/
Also this gist: https://gist.github.com/pbugnion/ea2797393033b54674af
"""

import sys
import json
import glob

def strip_cell(cell):
    if "outputs" in cell:
        cell["outputs"] = []
    if "execution_count" in cell:
        cell["execution_count"] = None
    if "metadata" in cell:
        cell["metadata"] = {}

ipynb_files = glob.glob('*.ipynb')
for ipynb_file in ipynb_files:
    print('Cleaning file: ', ipynb_file)
    with open(ipynb_file) as f:
        ipynb = json.load(f)
    
    ipy_version = int(ipynb["nbformat"])-1 # nbformat is 1 more than actual version.

    if ipy_version == 2:
        for sheet in ipynb["worksheets"]:
            for cell in sheet["cells"]:
                strip_cell(cell)
    else:
        for cell in ipynb["cells"]:
            strip_cell(cell)
    
    ipynb["metadata"] = {"language_info": {"name":"python", "pygments_lexer": "ipython3"}}
    json.dump(ipynb, open(ipynb_file, 'w'), sort_keys=True, indent=1, separators=(",",": "))