tqdm failure with Jupyter Notebook

Question

tqdm is one of my favorite Python packages, but I have an annoying issue with it and trying to figure out if this is my fault or not.

While running a loop with tqdm using Jupyter Notebook and encoutering a RunTime Error the recovery is really hard - fixing the bug and reruning the loop ofter result in multiline prints, instead of tqdm original single line. The only way to get back is to restart the Kernel which is not very useful.

Are the any other solutions?

I can't reproduce the issue intentionally but here is an example code that might cause it:

from tqdm import trange
s=0
for i in trange(100):
    s+=i
    if i==10:
        raise ValueError

and then reruning the cell.

Answer 1

To overcome this issue, I use the following conditional import which works in code that might run in Jupyter Notebook or in the console (and do nothing if not run interactively):

# progress.py
import sys


def tqdm(iterable, **kwargs):
    """Fake progress function."""

    return iterable


# check if running notebook and use notebook backend for tqdm progress bar
if 'IPython' in sys.modules:

    from IPython import get_ipython
    from IPython.display import display_javascript, Javascript

    from tqdm import tqdm

    ip = get_ipython()

    if 'IPKernelApp' in ip.config:
        monkey_patch = f"""
            from tqdm import tqdm_notebook as tqdm
            import {__name__} as module
            module.tqdm = tqdm
        """
        monkey_patch = ';'.join(x.strip() for x in monkey_patch.strip().split('\n'))
        display_javascript(Javascript(f"""IPython.notebook.kernel.execute("{monkey_patch}");"""))

Then use it:

$ jupyter console  # or ipython3
In [1]: import progress

In [2]: list(progress.tqdm(range(10)))
100%|██████████| 10/10 [00:00<00:00, 120873.31it/s]
Out[2]: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]