nbconvert jupyter notebook to latex format

[DancingQuanta]

I have a jupyter notebook I wanted to convert to latex with nbconvert. I have set up my notebook with following settings

hv.extension('matplotlib')
hv.notebook_extension(display_formats=['html', 'svg'])
hv.output(fig='svg')

I am using holoviews master 91259b96.

I used the following nbconvert command.

jupyter nbconvert --no-prompt --TagRemovePreprocessor.enabled=True --TagRemovePreprocessor.remove_input_tags="{'remove_input'}" --TagRemovePreprocessor.remove_cell_tags="{'remove_cell', 'injected-parameters'}" --to=latex --LatexExporter.template_file="revtex.tplx" --output="analysis.tex" "analysis.ipynb"

A tex document was created with text repr of the holoviews object. The nbconvert also appears to create a pdf of each plot. I am expecting a latex enviroment with graphic rendering tex code in place.

I may have been missing this in the docs or this latex rendering is not yet fully supported.

[DancingQuanta]

This issue was caused by my revtex.tplx in the ways I did not understand. Originally I have modifed someone's else template for revtex document class to have some cell tags code to manipulate contents passed to nbconvert. Since nbconvert have added more control over the cell tags I no longer needed my modifications any more and forgot about them. Since them this tags code have been mashed things up for holoviews objects!

So this issue is reduced to lack of documenataion on hotebook_extension and use of nbconvert.

[DancingQuanta]

I am writing my observations here for using nbconvert to generate markdown export from a jupyter notebook (especially from Download as interface).

I am using the following imports and hv setup

import numpy as np
import holoviews as hv
import hvplot.xarray
import xarray as xr
hv.extension('matplotlib')
hv.notebook_extension(display_formats=['html', 'svg'])
hv.output(fig='svg')

• case 1

hv.notebook_extension(display_formats=['html', 'svg'])

This sets all the rendered plots to be a SVG images which you can easily right click to get context menu and save as. However, when exporting a markdown version of a notebook the images was extracted as normal and the new markdown file contain reference to SVG images as an HTML img src tag not markdown image link as I expected.

• case 2

hv.notebook_extension(display_formats=['svg'])

This case is missing the html item from the display format and the rendered images are not SVG images as far I can tell because I cannot save as them. Interestlngly enough, when converting to markdown the new markdown file contain references to images as proper markdown image link which is what I wanted all along.

I am using Chrome with a useful extension called Dark Reader which transform websites to white texxt on black. Using this extension I have observed visual differences between above cases. For the case 1, the figures are unaffected by my extension modifying the notebook to black so they show up white. In case 2, the figures are affected and is showing up black background and plot elements are inverted colours. This made me suspect that addition of html to the display formats wraps the SVG in HTML tags which allows it to be saved through context menu.

As far as I can tell, it is nice to see flexibility in exposing SVG images through an HTML tags though unfortunately do not convert to markdown image links upon export to markdown format.

Summary: hv.notebook_extension(display_formats=['svg']) is great for clean markdown export from a notebook with SVG images.

Edit; the notebook I am using is here