jupyterbook to pdf using latex - possible to specify longtable?

[jgunstone]

Hi there,

I'm converting markdown documentation to pdf using jupyterbook and it works great - but i have a long multipage table that isn't wrapping across the pages correctly. Currently latex seems to be using "tabulary" format and i'd like to specify "longtable". Is this possible?

My environment

• running ubuntu20.04
• jupyterbook version

(mfcode_docs) jovyan@PORT4028:/mnt/c/engDev/git_mf/mfcode_docs$ jb --version
Jupyter Book      : 0.11.3
External ToC      : 0.2.3
MyST-Parser       : 0.13.7
MyST-NB           : 0.12.3
Sphinx Book Theme : 0.1.5
Jupyter-Cache     : 0.4.3
NbClient          : 0.5.4

• used the following commands to set up latex from the jupyterbook docs

sudo apt-get install texlive-latex-recommended texlive-latex-extra \
                     texlive-fonts-recommended texlive-fonts-extra \
                     texlive-xetex latexmk

(i also converted a simple longtable example and it worked so to verify i had the longtable plugin installed)

The Issue

here is the markdown table:

• index.md

this is what the output pdf looks like:

and here is the latex:

\begin{savenotes}\sphinxattablestart
\centering
\begin{tabulary}{\linewidth}[t]{|T|T|T|}
\hline

\sphinxAtStartPar

&\sphinxstyletheadfamily 
\sphinxAtStartPar
technology
&\sphinxstyletheadfamily 
\sphinxAtStartPar
description
\\
\hline
\sphinxAtStartPar
0
&
\sphinxAtStartPar
\sphinxhref{https://conda-forge.org/}{\sphinxincludegraphics{{conda-forge-icon}.png}  conda\sphinxhyphen{}forge}
&
\sphinxAtStartPar
community maintained collection of opensource packages and template specifcation for new packages.
\\
\hline
\sphinxAtStartPar
1
&
\sphinxAtStartPar
\sphinxhref{https://docs.conda.io/en/latest/conda-build.html}{\sphinxincludegraphics{{conda-icon}.png}  conda}
&
\sphinxAtStartPar
tool for building conda packages, maintained by Anaconda.
\\
\hline
\sphinxAtStartPar
2
&
\sphinxAtStartPar
\sphinxhref{https://github.com/mamba-org/mamba}{\sphinxincludegraphics{{mamba-icon}.png}  mamba}
&
\sphinxAtStartPar
Mamba is a fast, robust, and cross\sphinxhyphen{}platform package manager; it is a reimplementation of the conda package manager in C++.
\\
\hline
\sphinxAtStartPar
3
&
\sphinxAtStartPar
\sphinxhref{https://docs.ansible.com/}{\sphinxincludegraphics{{ansible-icon}.png}  ansible}
&
\sphinxAtStartPar
Ansible playbooks are used to automate the setup, configuration and update of remote servers from a local machine. This is used to deploy our JupyterHub server.
\\
\hline
\sphinxAtStartPar
4
&
\sphinxAtStartPar
\sphinxhref{https://jupyterhub.readthedocs.io/en/stable/}{\sphinxincludegraphics{{jupyterhub-icon}.png}  jupyterhub}
&
\sphinxAtStartPar
JupyterHub is the best way to serve Jupyter notebook for multiple users. It can be used in a class of students, a corporate data science group or scientific research group.
\\
\hline
\sphinxAtStartPar
5
&
\sphinxAtStartPar
\sphinxhref{https://repo2docker.readthedocs.io/en/latest/index.html}{\sphinxincludegraphics{{repo2docker-icon}.png}  repo2docker}
&
\sphinxAtStartPar
repo2docker fetches a repository (from GitHub, GitLab, Zenodo, Figshare, Dataverse installations, a Git repository or a local directory) and builds a container image in which the code can be executed. The image build process is based on the configuration files found in the repository.
\\
\hline
\sphinxAtStartPar
6
&
\sphinxAtStartPar
\sphinxhref{https://jupyter.org/}{\sphinxincludegraphics{{jupyter-icon}.png}  jupyter}
&
\sphinxAtStartPar
interactive coding and in\sphinxhyphen{}line report production
\\
\hline
\sphinxAtStartPar
7
&
\sphinxAtStartPar
\sphinxhref{https://code.visualstudio.com/}{\sphinxincludegraphics{{vscode-icon}.png}  vscode}
&
\sphinxAtStartPar
state of the art IDE for software development
\\
\hline
\sphinxAtStartPar
8
&
\sphinxAtStartPar
\sphinxhref{https://www.notion.so/pyrevitlabs/pyRevit-bd907d6292ed4ce997c46e84b6ef67a0}{\sphinxincludegraphics{{pyrevit-icon}.png}  pyrevit}
&
\sphinxAtStartPar
pyRevit (with lowercase py) is a Rapid Application Prototyping (RAD) environment for Autodesk Revit. It helps you quickly sketch out your automation and add\sphinxhyphen{}on ideas, in whichever language that you are most comfortable with, inside the Revit environment and using its APIs.
\\
\hline
\sphinxAtStartPar
9
&
\sphinxAtStartPar
\sphinxhref{https://www.food4rhino.com/en/app/ladybug-tools}{\sphinxincludegraphics{{grasshopper+ladybug-tools-icon}.png}  grasshopper+ladybug\sphinxhyphen{}tools}
&
\sphinxAtStartPar
visual coding and python scripting environment inside McNeel Rhino. good for early\sphinxhyphen{}stage concept design development and analysis.
\\
\hline
\sphinxAtStartPar
10
&
\sphinxAtStartPar
\sphinxhref{https://www.iesve.com/software/python-scripting}{\sphinxincludegraphics{{IES-vescripts-icon}.png}  IES\sphinxhyphen{}vescripts}
&
\sphinxAtStartPar
You can use the VE Python API to create your own customised scripts to: automate reports, data input, parametric simulations and many other tasks.
\\
\hline
\sphinxAtStartPar
11
&
\sphinxAtStartPar
\sphinxhref{https://github.com/voila-dashboards/voila}{\sphinxincludegraphics{{voila-icon}.png}  voila}
&
\sphinxAtStartPar
Voilà turns Jupyter notebooks into standalone web applications.
\\
\hline
\sphinxAtStartPar
12
&
\sphinxAtStartPar
\sphinxhref{https://plotly.com/python/getting-started/}{\sphinxincludegraphics{{plotly-icon}.png}  plotly}
&
\sphinxAtStartPar
simple but powerful interactive web visualisation
\\
\hline
\sphinxAtStartPar
13
&
\sphinxAtStartPar
\sphinxhref{https://altair-viz.github.io/}{\sphinxincludegraphics{{altair-icon}.png}  altair}
&
\sphinxAtStartPar
declarative 2D plotting that wraps around vega d3 schemas
\\
\hline
\sphinxAtStartPar
14
&
\sphinxAtStartPar
\sphinxhref{https://github.com/bqplot/bqplot}{\sphinxincludegraphics{{bqplot-icon}.png}  bqplot}
&
\sphinxAtStartPar
plotting library build with jupyter widgets. allows for complex interactions with ipywidgets.
\\
\hline
\sphinxAtStartPar
15
&
\sphinxAtStartPar
\sphinxhref{https://www.postgresql.org/}{\sphinxincludegraphics{{postgresql-icon}.png}  postgresql}
&
\sphinxAtStartPar
PostgreSQL is a powerful, open source object\sphinxhyphen{}relational database system with over 30 years of active development that has earned it a strong reputation for reliability, feature robustness, and performance. It is used for production\sphinxhyphen{}grade tools.
\\
\hline
\sphinxAtStartPar
16
&
\sphinxAtStartPar
\sphinxhref{https://www.sqlite.org/index.html}{\sphinxincludegraphics{{sqlite-icon}.png}  sqlite}
&
\sphinxAtStartPar
SQLite is a C\sphinxhyphen{}language library that implements a small, fast, self\sphinxhyphen{}contained, high\sphinxhyphen{}reliability, full\sphinxhyphen{}featured, SQL database engine. It is used for some standalone apps and prototyping.
\\
\hline
\sphinxAtStartPar
17
&
\sphinxAtStartPar
\sphinxhref{https://www.sqlalchemy.org/}{\sphinxincludegraphics{{sqlalchemy-icon}.png}  sqlalchemy}
&
\sphinxAtStartPar
SQLAlchemy is the Python SQL toolkit and Object Relational Mapper that gives application developers the full power and flexibility of SQL. Table definitions and relations are developed in SQL alchemy, as well as data\sphinxhyphen{}serialisation with pydantic.
\\
\hline
\sphinxAtStartPar
18
&
\sphinxAtStartPar
\sphinxhref{https://pydantic-docs.helpmanual.io/}{\sphinxincludegraphics{{pydantic-icon}.png}  pydantic}
&
\sphinxAtStartPar
Define how data should be in pure, canonical python; validate it with pydantic.Used to create object definitions that feed: databases, calculation tools and documentation.
\\
\hline
\sphinxAtStartPar
19
&
\sphinxAtStartPar
\sphinxhref{https://fastapi.tiangolo.com/}{\sphinxincludegraphics{{FastAPI-icon}.png}  FastAPI}
&
\sphinxAtStartPar
FastAPI framework, high performance, easy to learn, fast to code, ready for \sphinxhref{http://production.It}{production.It} makes maintainence, upgrade and integration of production packages with software scripts simple asthe packages and environment is installed on the server.
\\
\hline
\sphinxAtStartPar
20
&
\sphinxAtStartPar
\sphinxhref{https://quantstack.net/}{\sphinxincludegraphics{{quantstack-icon}.png}  quantstack}
&
\sphinxAtStartPar
for help with the heavy lifting and deployment we call in the pros.Quantstack are one of the core Jupyter developers and are active members of the opensource community.Working with them gives us confidence in our approach to software development,whilst concentrating on what we’re best at: Engineering.
\\
\hline
\sphinxAtStartPar
21
&
\sphinxAtStartPar
\sphinxhref{https://jupyterbook.org/intro.html}{\sphinxincludegraphics{{jupyterbook-icon}.png}  jupyterbook}
&
\sphinxAtStartPar
Jupyter Book is an open source project for building beautiful, publication\sphinxhyphen{}quality books and documents from computational material.
\\
\hline
\sphinxAtStartPar
22
&
\sphinxAtStartPar
\sphinxhref{https://www.sphinx-doc.org/en/master/}{\sphinxincludegraphics{{sphinx-icon}.png}  sphinx}
&
\sphinxAtStartPar
Jupyter book utilises sphinx for many of its functions. Sphinx can also be used to auto\sphinxhyphen{}doc docstrings written in the Google Style Guide (MF standard).
\\
\hline
\end{tabulary}
\par
\sphinxattableend\end{savenotes}

[AakashGfude]

hi @gunstonej , thanks for notifying us about this. You can add packages, like "longtable", using the preamble key in _config.yml :

latex:
   latex_elements:
       preamble:

give this a try and let us know if it works.

[jgunstone]

thanks for the response!

I just added the following to the _config.yml

latex:
  latex_elements:
    preamble: longtable

i also tried:

latex:
  latex_elements:
    preamble: 
      - longtable

but neither of them made a difference.

this reminded me i also tried this yesterday to see if it was possible to specify it through Sphinx but this wouldn't run.

sphinx:
  extra_extensions:
   - longtable

[jgunstone]

sorry - just realised i think i misunderstood the use of "preamble".

is this used to inject latex into the beginning of the document?

i just tried this instead:

latex:
  latex_elements:
    preamble: |
        \usepackage{longtable}

also this (following the syntax here:

sphinx:
  config:
    latex_elements:
        preamble: |
          \usepackage{longtable}

but neither did anything to the output and the latex is still using tabulary

[jgunstone]

to note -

I think another (better) solution might be to use tabularray (see example ), this would be able to deal with tables spanning multiple columns which isn't supported by longtable.

Notes

• tabularray requires TexLive 2021 which isn't available via conda or apt-get.
• pandoc uses longtable by default