A markdown kernel for Jupyter

Installation

If you are using a conda environment, you can install markdown-kernel with command

conda install -c conda-forge markdown-kernel

Otherwise, with a working Jupyter installation, you can install markdown-kernel with commands

pip install markdown-kernel
python -m markdown_kernel.install

You can verify if a markdown kernel is properly installed with command

jupyter kernelspec list

Use a markdown kernel by itself

markdown-kernel is a simple Jupyter kernel that displays cell content as markdown. To use this kernel, you can start a Jupyter server with command jupyter notebook or jupyter lab, create a notebook with this kernel, enter and render markdown texts.

But wait, this is stupid because you can do the same thing in Jupyter, only easier. What is the point of a markdown kernel by itself?

Use a Markdown kernel in SoS Notebook

If you are familiar with R Markdown, you might know and like its inline code that allows results to be inserted directly into the text of a .Rmd file by enclosing the code with `r `. This is really convenient for writing Rmarkdown report but cannot be done in Jupyter because Jupyter's markdown cells are rendered at the frontend and do not interact with Jupyter kernels (See ipython/ipython#2592, jupyter/help#41,and jupyter/notebook#3463 for related discussions).

SoS Notebook is a Jupyter kernel that supports the use of multiple kernels in one Jupyter notebook. Although a markdown kernel by itself cannot interpolate and evaluate expressions either, you can the %expand magic of SoS to enable inline expressions in Jupyter, for Python, R, and potentially many other languages. Here is how it works:

Expand expressions with SoS (Python) variables

The SoS kernel is a super kernel that starts and communicates with all other Jupyter kernels. Its syntax is a super set of Python 3.6+ so you can execute any Python code in SoS kernels. The %expand magic treats the content of a markdown cell as a Python f-string and expands expressions enclosed in { }, or other delimiters as option of magic %expand.

NOTE: Starting from jupyterlab-sos version 0.6.3, the input of codecell with Markdown kernel are folded automatically after cell execution.

Expand expressions in subkernels such as R

The --in option of magic %expand allows you to expand the cell content in specified subkernel, if its language module supports the expand protocol. This allows the content in markdown cells to be expanded in other languages such as R. Most notably, you can use

%expand `r ` --in R

to expand inline expressions in R Markdown texts.

This technique is used by sos-rmarkdown, a RMarkdown to SoS Notebook converter to convert markdown text with inline expressions.