Bulk Convert Python files to IPython Notebook Files (py to ipynb conversion)

I had a bunch of Python files that I needed to convert in bulk to Jupyter Notebook files. Jupyter Notebook files are simple JSON files with a cells array containing one or more unnamed cell nodes.

For example, here’s the code for the simple Jupyter notebook shown in the image above. Note the cells array and, in particular, the source key.

{
    "cells": [
        {
            "cell_type": "code",
            "execution_count": null,
            "metadata": {
                "collapsed": false
            },
            "outputs": [],
            "source": [
                "try:\n",
                "    1/0\n",
                "except:\n",
                "    print('You cannot divide by zero!')"
            ]
        }
    ],
    "metadata": {
        "kernelspec": {
            "display_name": "Python 3",
            "language": "python",
            "name": "python3"
        },
        "language_info": {
            "codemirror_mode": {
                "name": "ipython",
                "version": 3
            },
            "file_extension": ".py",
            "mimetype": "text/x-python",
            "name": "python",
            "nbconvert_exporter": "python",
            "pygments_lexer": "ipython3",
            "version": "3.4.3"
        }
    },
    "nbformat": 4,
    "nbformat_minor": 0
}

The only piece we need to change to convert Python files to Jupyter Notebook files is the highlighted source value, which contains the Python code that goes in the cell. Everything else we can copy verbatim.

My py-to-ipynb.py script looks like this:

import os
from json.encoder import JSONEncoder
nb_start = '''{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": ['''
nb_end =''']
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.4.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}'''
def main():
    for dirpath, dirnames, filenames in os.walk('.'):
        for fname in filenames:
            path = dirpath+'/'+fname
            if fname[-3:] == '.py' and not os.path.samefile(path,'.\\py-to-ipynb.py'):
                path = dirpath+'/'+fname
                nb_path = path[:-3] + '.ipynb'
                with open(path,'r') as f_in:
                    f_in_content = JSONEncoder().encode(f_in.read())
                    nb_content = nb_start + f_in_content + nb_end
                    
                    with open(nb_path,'w') as f_out:
                        f_out.write(nb_content)
                        
                print("Created",nb_path)
main()

Here’s what it does:

1. Walks through all the files within the current directory and its subdirectories.
2. Reads in the contents of every file (except itself) ending with .py, encodes it as JSON, and stores it as f_in_content.
3. Creates a nb_content variable by concatenating…
   1. nb_start – the JSON text that goes before the source value in the Jupyter Notebook JSON.
   2. f_in_content
   3. nb_end – the JSON text that goes after the source value in the Jupyter Notebook JSON.
4. Writes nb_content to a new file with the same name, but with a .ipynb extension.

I know it’s a hack, but I couldn’t find a built-in way of doing this. I’ve only done some minimal testing, so use at your own risk. 🙂