aaren / Notedown

Python 2/3 and IPython 4 / Jupyter compatible!

Convert IPython Notebooks to markdown (and back)

notedown is a simple tool to create IPython notebooks from markdown (and r-markdown).

notedown separates your markdown into code and not code. Code blocks (fenced or indented) go into input cells, everything else goes into markdown cells.

Usage:

notedown input.md > output.ipynb

Installation:

pip install notedown

or the latest on github:

pip install https://github.com/aaren/notedown/tarball/master

Conversion to markdown

Convert a notebook into markdown, stripping all outputs:

notedown input.ipynb --to markdown --strip > output.md

Convert a notebook into markdown, with output JSON intact:

notedown input.ipynb --to markdown > output_with_outputs.md

The outputs are placed as JSON in a code-block immediately after the corresponding input code-block. notedown understands this convention as well, so it is possible to convert this markdown-with-json back into a notebook.

This means it is possible to edit markdown, convert to notebook, play around a bit and convert back to markdown.

NB: currently, notebook and cell metadata is not preserved in the conversion.

Strip the output cells from markdown:

notedown with_output_cells.md --to markdown --strip > no_output_cells.md

Running an IPython Notebook

notedown notebook.md --run > executed_notebook.ipynb

Editing in the browser (new!)

You can configure IPython / Jupyter to seamlessly use markdown as its storage format. Add the following to your config file:

c.NotebookApp.contents_manager_class = 'notedown.NotedownContentsManager'

Now you can edit your markdown files in the browser, execute code, create plots - all stored in markdown!

For Jupyter, your config file is jupyter_notebook_config.py in ~/.jupyter. For IPython your config is ipython_notebook_config.py in your ipython profile (probably ~/.ipython/profile_default):

Editing in vim

There is a vim plugin that allows editing notebooks (ipynb files) directly in vim. They will be automatically converted to markdown on opening the file, and converted back to the original json format on writing.

R-markdown

You can use notedown to convert r-markdown as well. We just need to tell notedown to use knitr to convert the r-markdown. This requires that you have R installed with knitr.

Convert r-markdown into markdown:

notedown input.Rmd --to markdown --knit > output.md

Convert r-markdown into an IPython notebook:

notedown input.Rmd --knit > output.ipynb

• --rmagic will add %load_ext rpy2.ipython at the start of the notebook, allowing you to execute code cells using the rmagic extension (requires rpy2). notedown does the appropriate %R cell magic automatically.

Magic

Fenced code blocks annotated with a language other than python are read into cells using IPython's %% cell magic.

You can disable this with --nomagic.

• --pre lets you add arbitrary code to the start of the notebook. e.g. notedown file.md --pre '%matplotlib inline' 'import numpy as np'

How do I put a literal code block in my markdown?

By using the --match argument. notedown defaults to converting all code-blocks into code-cells. This behaviour can be changed by giving a different argument to --match:

• --match=all: convert all code blocks (the default)

• --match=fenced: only convert fenced code blocks

• --match=language: only convert fenced code blocks with 'language' as the syntax specifier (or any member of the block attributes)

• --match=strict: only convert code blocks with Pandoc style attributes containing 'python' and 'input' as classes. i.e. code blocks must look like

    ```{.python .input}
    code
    ```
  

This isn't very interactive!

Try editing the markdown in the IPython Notebook using the NotedownContentsManager (see above).

You can get an interactive ipython session in vim by using vim-ipython, which allows you to connect to a running ipython kernel. You can send code from vim to ipython and get code completion from the running kernel. Try it!

Where's my syntax highlighting?!

Try using either vim-markdown or vim-pandoc. Both are clever enough to highlight code in markdown.

Rendering outputs in markdown

This is experimental!

Convert a notebook into markdown, rendering cell outputs as native markdown elements:

notedown input.ipynb --render

This means that e.g. png outputs become ![](data-uri) images and that text is placed in the document.

Of course, you can use this in conjuntion with runipy to produce markdown-with-code-and-figures from markdown-with-code:

notedown input.md --run --render > output.md

Not a notebook in sight!

The --render flag forces the output format to markdown.

TODO

• [x] Python 3 support
• [x] unicode support
• [x] IPython 3 support
• [x] IPython 4 (Jupyter) support
• [ ] Allow kernel specification