Filter an IPython notebook so that lines of console output that were written over by carriage return characters (\\r) are removed.

np_filter.py

"""
Filter an IPython notebook so that lines of console output that were
written over by carriage return characters (\\r) are removed.

This is a UNIX filter, so use as::

   cat my_notebook.ipynb | nb_filter.py > my_notebook.filtered.ipynb
"""

import json
import re
import sys


def filter(s):
    while True:
        match = re.search(ur'(?:^|\r|\n)(.*?\r(?!\n))', s)
        if match is not None:
            s = s[:match.start()] + s[match.end():]
        else:
            break
    return s


content = json.load(sys.stdin)

for worksheet in content['worksheets']:
    for cell in worksheet['cells']:
        texts = {}
        for output in cell['outputs']:
            if output['output_type'] == 'stream':
                for line in output['text']:
                    texts.setdefault(output['stream'], []).append(line)

        cell['outputs'] = []
        for stream, text in texts.items():
            text = ''.join(text)
            text = filter(text)

            cell['outputs'].append({
                'output_type': 'stream',
                'stream': stream,
                'text': [text]})

json.dump(content, sys.stdout, indent=1)