Thanks again for your fixes and quick response times.

I found another issue, this time related to multiple outputs, especially outputs of different mime types per cell.

The following could should display 5 headings and 5 tables. The xls sheet however only shows the latest table, and no headings:

import pandas as pd
from IPython.display import *

for i in range(5):
    display(Markdown(f'# Markdown Loop {i}'))
    display(pd.DataFrame(pd.np.random.random((5,5))))

Cool extension. You might consider adding this and the other container DS Jupyter projects to awesome-jupyter, awesome-jupyterlab?

https://github.com/mauhai/awesome-jupyterlab

https://github.com/adebar/awesome-jupyter

https://github.com/quobit/awesome-python-in-education#jupyter

Is there any way to export the charts in a larger size in excel. I am able to use the package on all my notebooks but chart export size seems too small and I have to manually increase each chart size in excel. Please let me know if I missed anything?
Thanks for your work on this, its great!

Issue Description

The export function fails when there are pandas dataframes with NaN values.

The following message is displayed:

nbconvert failed: NAN/INF not supported in write_number() without 'nan_inf_to_errors' Workbook() option

Reproducing

To reproduce the issue, simply create a notebook with the following cell and try to export:

pd.DataFrame([[pd.np.nan]])

Found another one 🐞

from IPython.display import Markdown, display

display(Markdown('''
# Hi Dan
1. Thanks for fixing the last issue so quickly
1. This library is really cool
1. I'll probably start using it for work soon
'''))

results in:

<IPython.core.display.Markdown object>

The same issue occurs without the display. Similarly, IPython.display.HTML and Latex fail. I suspect you might need an explicit resolution of _repr_*_() functions, as it's done in IPython itself: IPython docs and the corresponding code (check the display function)

Note: Somewhat surprisingly (to me, anyway), the following contrived example of a custom object displayed correctly:

from random import choice
from string import ascii_lowercase

def _random_str(length: int):
    return ''.join(choice(ascii_lowercase) for _ in range(length))

class RandomUser:
    def __init__(self):
        self.user = _random_str(10)
        self.password = _random_str(20)
        
    def _repr_html_(self):
        return f'''
        <table>
        <tr><th>User: </th><td>{self.user}</td>
        <tr><th>Password: </th><td>{self.password}</td>
        </table>
        '''
    
RandomUser()

On the other hand, this code:

import json

class DisplayDict(dict):
    def _repr_json_(self):
        return self
    
DisplayDict({'hello': 'world'})

results in another error:

nbconvert failed: split

e.g.:
pd.DataFrame([['a',1,123], ['a',2,345], ['b',1,678], ['a',1,345]]).set_index([0,1])