Bookbook converts a set of notebooks in a directory to HTML or PDF, preserving cross references within and between notebooks.

This code is in early development, so use it at your own risk.

Bookbook requires Python 3.5.

pip install bookbook

To install locally as an editable install, run:

pip install flit
git clone https://github.com/takluyver/bookbook.git
cd bookbook
flit install --symlink

bookbook expects a directory of notebooks whose names indicate their order. Specifically, the file names must have the form ``x-y.ipynb``, where typically x is a number indicating the order and y is a chapter title; e.g.: 01-introduction.ipynb.

To run bookbook:

python3 -m bookbook.html           # HTML output under html/
python3 -m bookbook.latex [--pdf]  # Latex/PDF output as combined.(tex|pdf)

Add --help to either command for more options.

Each top-level header (# xyz in markdown) will be converted to a top-level latex section (a chapter if using the book or report document class). Lower-level headers (##, ###, etc.) are converted to subsections, subsubsections, etc. A latex label will also be inserted for each. The first cell of each notebook must start with a top-level header.

Markdown references will be converted automatically to latex references. For instance, if the markdown contains the hyperlink [02-foo](02-foo.ipynb) and 02-foo.ipynb is a notebook in the same directory, the link will appear as Chapter \ref{sec:02-foo}. The label \label{sec:02-foo} will be inserted at the start of that notebook, so when the latex is compiled to PDF it will appear as Chapter 2.

References to sections within a notebook work similarly. If a notebook contains (in markdown) the section heading ## bar within the notebook starting with top-level header # foo, then the markdown hyperlink [foo](foo.ipynb#bar) will be converted to the latex reference Section \ref{bar} and when compiled to PDF it will be rendered as something like Section 2.1.

bookbook uses nbconvert under the hood. Custom formatting of latex output can be accomplished by using a template, in the same way as would be done using nbconvert by itself. See the nbconvert documentation for more details.

• Book on Riemann solvers (in development) by David Ketcheson, Mauricio del Razo, and Randall LeVeque. This example uses a custom nbconvert template and shows how to store your notebooks with no output (for version control) while automatically executing them before running bookbook, so that PDF and HTML versions include the output.

If you are writing a book in Jupyter notebooks, you may also find these to be useful:

• nbopen: open notebooks from the command line without launching a new notebook server. We find it useful to launch a single server in your home directory; then nbopen will use that to open each notebook.
• nbdime: diff/merge for notebooks; includes terminal or graphical output.
• nbstripout: remove output from notebooks before committing them.