Python script showing how html diffs of Jupyter notebooks can be created non-interactively (e.g. from the command-line in CI) using nbdime.

test-nbdime-html-diff.py

"""
Python script showing how html diffs of Jupyter notebooks can be created
non-interactively (e.g. from the command-line in CI) using nbdime
(https://github.com/jupyter/nbdime) in combination with nbdime PR #744:
https://github.com/jupyter/nbdime/pull/744

Usage:

  $ python test-nbdime-html-diff.py NB_PATH1 NB_PATH2

"""

from pathlib import Path
import sys

import jinja2
from jinja2 import Environment, FileSystemLoader

import nbdime.diffing.notebooks as notebooks
import nbdime.utils as nb_utils
import nbdime.webapp as webapp

WEBAPP_DIR = Path(webapp.__file__).parent


def read_notebook(path):
    with path.open() as f:
        data = nb_utils.read_notebook(f, on_null='empty')
    return data


def diff_notebooks(path1, path2, output_path, static_dir):
    nb_data1 = read_notebook(path1)
    nb_data2 = read_notebook(path2)

    # Here is where you can tell nbdime what to include and ignore.
    # For example, tell nbdime to ignore differences in cell metadata.
    notebooks.set_notebook_diff_targets(metadata=False)
    diff_result = notebooks.diff_notebooks(nb_data1, nb_data2)

    template_dir = WEBAPP_DIR / 'templates'
    template_loader = FileSystemLoader(template_dir)
    env = Environment(
        loader=template_loader, autoescape=jinja2.select_autoescape(),
    )
    template = env.get_template('diff.html')

    def static_url(rel_path):
        path = static_dir / rel_path
        return path

    # The showDiff() function in the nbdime package's diff.ts file
    # expects key-values "base" and "diff".
    diff_data = {
        'base': nb_data1,
        'diff': diff_result,
    }
    config_data = {
        'base': str(path1),
        'remote': str(path2),
        'diff_data': diff_data,
    }
    context = {
        # The nbdime package's base nbdimepage.html template calls tojson()
        # on the config_data value.
        'config_data': config_data,
        'static_url': static_url,
    }
    html = template.render(context)
    output_path.write_text(html)


def main():
    args = sys.argv[1:]
    assert len(args) == 2
    path1, path2 = (Path(path) for path in args)
    output_path = Path('test-diff.html')

    # Pass a static_dir that lets us test the locally generated JS.
    static_dir = WEBAPP_DIR.absolute() / 'static'
    diff_notebooks(
        path1, path2=path2, output_path=output_path, static_dir=static_dir,
    )
    print(f'wrote output to: {output_path}')


if __name__ == '__main__':
    main()