Convert jupyter notebook to sphinx gallery notebook styled examples.

ipynb_to_gallery.py

"""Convert jupyter notebook to sphinx gallery notebook styled examples.

Usage: python ipynb_to_gallery.py <notebook.ipynb>

Dependencies:
pypandoc: install using `pip install pypandoc`
"""
import pypandoc as pdoc
import json

def convert_ipynb_to_gallery(file_name):
    python_file = ""

    nb_dict = json.load(open(file_name))
    cells = nb_dict['cells']

    for i, cell in enumerate(cells):
        if i == 0:  
            assert cell['cell_type'] == 'markdown', \
                'First cell has to be markdown'

            md_source = ''.join(cell['source'])
            rst_source = pdoc.convert_text(md_source, 'rst', 'md')
            python_file = '"""\n' + rst_source + '\n"""'
        else:
            if cell['cell_type'] == 'markdown':
                md_source = ''.join(cell['source'])
                rst_source = pdoc.convert_text(md_source, 'rst', 'md')
                commented_source = '\n'.join(['# ' + x for x in
                                              rst_source.split('\n')])
                python_file = python_file + '\n\n\n' + '#' * 70 + '\n' + \
                    commented_source
            elif cell['cell_type'] == 'code':
                source = ''.join(cell['source'])
                python_file = python_file + '\n' * 2 + source

    python_file = python_file.replace("\n%", "\n# %")
    open(file_name.replace('.ipynb', '.py'), 'w').write(python_file)

if __name__ == '__main__':
    import sys
    convert_ipynb_to_gallery(sys.argv[-1])

Hi,

This script is very useful! However lots of people now uses magic command such as %matplotlib inline and so, the resulting file is not executable by regular python.

I suggest adding something like:

    python_file = python_file.replace("\n%", "\n# %")

just before the writing of the file, l. 35.

thanks again :)

Modified!

Really useful script. When using it I found that the pdoc conversion of markdown adds "\r\n" after every new line of the file I tried.
This causes errors in the conversion since the filer writer will add write 2 new lines splitting rst cells.

A quick fix was to add

rst_source = rst_source.replace('\r', '')

after line 23 and 28.

There might be a more clever way with how to write the file in the end.

May be replace

nb_dict = json.load(open(file_name))

with

with open(file_name) as f:
    nb_dict = json.load(f)

in line 14 and replace

open(file_name.replace('.ipynb', '.py'), 'w').write(python_file)

with

with open(file_name.replace('.ipynb', '.py'), 'w') as f:
    f.write(python_file)

in line 38.

I had the opposite issue of @weber-s : my first cell was magic code so as a quick fixed I modified the first conditional block as

if cell["cell_type"] == "markdown":
    md_source = "".join(cell["source"])
    rst_source = pdoc.convert_text(md_source, "rst", "md")
    python_file = '"""\n' + rst_source + '\n"""'
elif cell["cell_type"] == "code":
    source = "".join(cell["source"])
    python_file = python_file + "\n" * 2 + source
else:
    raise ValueError(f"Unsupported starting cell type {cell["cell_type"]}.")