fixing jupytext markdown inconsistencies #288
Conversation
| else: | ||
| # Otherwise, create an empty notebook and add all of the file contents as a markdown file | ||
| ntbk = nbf.v4.new_notebook() | ||
| ntbk['cells'].append(nbf.v4.new_markdown_cell(source=''.join(content))) |
There was a problem hiding this comment.
Hi Chris, yes sure, at least this way we're going to fix the rendering of the Jupyter Book book itself. I think this is good enough for a quick fix, and as you suggest we will follow-up on this.
| language: python | ||
| name: python3 | ||
| --- | ||
|
|
||
| # Creating book content |
There was a problem hiding this comment.
Maybe changing this file is not necessary.
And the changes I see below (with Markdown comments surrounding the python code cells) raise an interesting question: should Python code blocks in the Markdown file always appear as executable code cells? As you noticed, these Markdown comments allow to preserve Jupyter's Markdown cells whatever there content is.
There was a problem hiding this comment.
In this case, I needed to change the file so that I could use the right sidebar popout, which Jupyter Book can only use with notebooks instead of markdown files, this is why I added the jupytext header. I'll put my thoughts about choosing markdown syntax in the issue so that we don't lose the conversation when this is merged!
|
OK, merging this one since the current master branch is wonky due to markdown processing inconsistencies! It's a bit hacky but it'll get the job done :-) and perhaps @mwouts and I can figure out a more graceful solution to some of these syntax questions! (thanks for you help figuring this out @mwouts!!) |
This changes the logic for reading in text files in order to be more consistent about how markdown files are handled.
The current situation:
Any content file will be read in with jupytext. This is awesome, with the exception that some people's code blocks may be incorrectly parsed as code cells in the markdown files. In the short term, it'd be tricky to avoid that behavior from within juyptext
This PR's approach:
If a
.markdownor.mdfile is found, read it in to memory. First see if it has any YAML header.yaml.get('jupyter').get('jupytext')returns None.yaml.jupyter.jupytext) then usejupytext.readsto load in the notebook.The main assumption this is making is that jupytext files will always have a YAML header with something in the
jupyter.jupytextfield, while non-jupytext markdown files will not. That seems like a reasonable assumption to make.@mwouts what do you think about this implementation? It's a bit hacky, but at least in the short-term it solves the markdown parsing differences problem. Perhaps longer-term we can figure out a way to solve this more gracefully!