How to use matplotlib config for generated pages ?

[ksachdeva]

Hi,

I would prefer to keep the quality of images low in the jupyter notebooks while writing it as well as when storing it in git however when jupyter-book generates the pages I would prefer that the images are of high quality (retina).

I generally have this at the top of my notebooks

%config InlineBackend.figure_format = "retina"
plt.rcParams["axes.spines.right"] = False
plt.rcParams["axes.spines.top"] = False

Is there a way to configure these types settings outside the notebook and provide a file (rc files, an additional section in _config.yml etc) to jupyter-book to use?

Regards & thanks
Kapil

[chrisjsewell]

Heya, IIRC no not currently I'm afraid.

I'm a little hesitant of modifying the actual contents of notebooks within jupyter-book.
But perhaps it would be interesting to see if environmental variables can be injected into the execution process, something like

import os

if "JUPYTER_BOOK" in os.environ:
  %config InlineBackend.figure_format = "retina"
  plt.rcParams["axes.spines.right"] = False
  plt.rcParams["axes.spines.top"] = False

jupyter/nbclient#79 may also be relevance

[choldgraf]

oh interesting - got it

[mgeier]

Just for reference, in nbsphinx kernel arguments can be set in conf.py see https://nbsphinx.readthedocs.io/en/0.7.1/configuring-kernels.html#Kernel-Arguments

This can be used for Matplotlib settings, see https://nbsphinx.readthedocs.io/en/0.7.1/code-cells.html#Plots

The ExecutePreprocessor from nbconvert has the extra_arguments parameter for this.

[mgeier]

PS: I've asked a related question at jupyter/nbconvert#1344, which I think would be a great feature, but this would not help in @ksachdeva's case (because those setting would also affect JupyterLab).

[chrisjsewell]

Cheers @mgeier, yeh we should most probably allow to parse such arguments also.
Although the environmental variable approach feels like it would be more generally applicable to all kernels/packages

[mgeier]

@chrisjsewell AFAICT, the extra_arguments parameter is generally applicable to all kernels/packages, isn't it?
The config option nbsphinx_execute_arguments is just a list of strings, kernels can interpret this however they see fit.

Nothing against environment variables though, those can of course also be used, see https://nbsphinx.readthedocs.io/en/0.7.1/configuring-kernels.html#Environment-Variables, but this creates some noise in the notebook.

[ksachdeva]

Thanks, guys; I think the usage of conditioning based on environment variable is an acceptable way to handle it.

That said, I see the value of the external configuration as well.

Instead of injecting code cells is there is a way to specify the directory/configuration/rc file for visualization libraries. Most of the visualization libraries have this notion of themes/config and incorporating these themes using the configuration options in jupyter-book may be another way to help customize the generation of final document format.

[loftusa]

Bumping on this. I'm writing a book at the moment and I'd like to have custom global matplotlib rcParams. Don't want to have to make a new cell at the top of every file...