pdflatex builder failed to display styled dataframes

Describe the problem

When creating a jupyter book with a styled dataframe as described in the Jupyter Book docs here (https://jupyterbook.org/content/code-outputs.html#library-output-formatting), the dataframe is not displayed with the pdf latex builder and instead we are presented with

<pandas.io.formats.style.Styler at 0x1de96f473d0>

Link to your repository or website

No response

Steps to reproduce

1. Create a new Jupyter book with a ipynb containing the following code referenced from (https://jupyterbook.org/content/code-outputs.html#library-output-formatting)

import numpy as np
import pandas as pd

np.random.seed(24)
df = pd.DataFrame({'A': np.linspace(1, 10, 10)})
df = pd.concat([df, pd.DataFrame(np.random.randn(10, 4), columns=list('BCDE'))],
               axis=1)
df.iloc[3, 3] = np.nan
df.iloc[0, 2] = np.nan

def color_negative_red(val):
    """
    Takes a scalar and returns a string with
    the css property `'color: red'` for negative
    strings, black otherwise.
    """
    color = 'red' if val < 0 else 'black'
    return 'color: %s' % color

def highlight_max(s):
    '''
    highlight the maximum in a Series yellow.
    '''
    is_max = s == s.max()
    return ['background-color: yellow' if v else '' for v in is_max]

df.style.\
    applymap(color_negative_red).\
    apply(highlight_max).\
    set_table_attributes('style="font-size: 10px"')

1. Build the book as a latexpdf using

jupyter-book build testreport --builder=pdflatex

1. Open the PDF and see that where the dataframe should have been, there is instead

<pandas.io.formats.style.Styler at 0x1de96f473d0>

The version of Python you’re using

python 3.8.5

Your operating system

Windows 10

Versions of your packages

Jupyter Book      : 0.12.1
External ToC      : 0.2.3
MyST-Parser       : 0.15.2
MyST-NB           : 0.13.1
Sphinx Book Theme : 0.1.7
Jupyter-Cache     : 0.4.3
NbClient          : 0.5.3

Additional context

No response

[andrewellis55]

Quick update on this, I’ve found both the seeming cause and a workaround. Once we call df.style, we no longer have a pd.Dataframe but rather a pd.io.formats.style.Styler. The Styler object does not have _repr_latex_ defined that is required to render it in latex but it does have a to_latex() method. Which means we can just add the following lines to our code to get it to work

    def rep_latex():
        return df.to_latex(convert_css=True)
    df._repr_latex_ = rep_latex

Now once we have this, we will need to make sure a few more packages are added to our latex build. This can be done by adding the following lines to our _conf.yml

sphinx:
  config:
    latex_elements:
      # Note this line cheats the table parameter into the documentclass[] parameter
      papersize: letterpaper,table
      # These lines add the desired packages to the latex build
      preamble: |+
        \usepackage{xcolor}
        \usepackage{booktabs}
        \usepackage{multirow}

[franzhaas]

I think this is the same issue.

https://github.com/executablebooks/jupyter-book/issues/1512

According to pandas, this should be improvded by the 1.4 release, which is currently on the way.