e.g. the first code block in the indigenous post has an empty line at the top and several empty lines at the bottom:

And add error/warning if the output of the code cell is unknown

Error when running plotly-converter on https://blog.ubicenter.org/20201214/canada-homelessness-experiment.html

Error message:
Traceback (most recent call last):
File "C:\Users\yuchi\src\UBI-Center\blog\jb\20201214\make_graphs.py", line 14, in
io.write_html(fig, "../../plotly-converter/assets/graphs/2020-12-14-canada-homelessness-experiment-graph1.html", full_html = False, include_plotlyjs = False)
NameError: name 'fig' is not defined
File "C:\Users\yuchi\src\UBI-Center\blog\jb\20201214\make_graphs.py", line 70
io.write_html(fig, "../../plotly-converter/assets/graphs/2020-12-14-canada-homelessness-experiment-graph2.html", full_html = False, include_plotlyjs = False)
^
SyntaxError: invalid syntax
Traceback (most recent call last):
File "C:\Users\yuchi\src\UBI-Center\blog\plotly-converter\2converter.py", line 128, in
for line in file:
File "C:\Program Files\Python39\lib\encodings\cp1252.py", line 23, in decode
return codecs.charmap_decode(input,self.errors,decoding_table)[0]
UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 1588: character maps to

Instead of with the post name in the filename. I think this will improve organization of graphs in the main ubicenter.org repo.

Is it possible to make this button more subtle?

We've moved to using fig.show(config={"displayModeBar": False}) to avoid the clutter in the upper right of the plotly charts, would be good to have that in the new site too.

There are currently only a few extra cases besides poorly graphs, so the code is currently sufficient. Right now I check if "DataFrame" or "Stargazer" appears in the code block to try to determine what the output is supposed to be. But if these functions or objects are used in a later block I won't know what they refer to there. A potential solution is to include a series of try-catch lines to attempt every "to_html" function in every code block. Although it's not very elegant I think it would work. Something to consider in the future to make this more robust. The alternative is to mandate one specific name for the object displayed (eg, star for stargazer, df for data frame, fig for plotly, etc) or a keyword in the block

To avoid issues with functions that show figures.

For example, maybe the https://github.com/executablebooks/jupyter-book code would be relevant, since they create html (including plotly charts) from notebooks.

I think the tool could save some copy/pasting by taking the following two arguments:

1. nb_path: Path to the notebook. The program could break this apart into nb_folder and nb_filename, and change the working directory to nb_folder, so that if the notebook uses files based on relative paths, they execute properly.
2. site_path: Path to the UBICenter/ubicenter.org repo. The folder structure is consistent there, so the program should then write the markdown file and place it into _posts, create the assets/graphs/[nb_filename] folder, and place the resulting plotly html files in that folder.

@yuchida522

Still says "Click to show code" when toggled:

From https://ubicenter.github.io/ubicenter.org/indigenous-ubi

Jupyter notebooks have to escape $ to avoid creating LaTeX, so converting to markdown requires removing the slashes. Example from https://ubicenter.github.io/ubicenter.org/indigenous-ubi:

Low priority but could be nice