docker/jupyter/README.md

64 lines
2.0 KiB
Markdown

# Jupyter Docker Image
This directory contains build files for a complete jupyter environment. It is based off the following repositories:
- https://github.com/jupyter/docker-stacks
- https://github.com/rgriffogoes/scraper-notebook
- https://github.com/sharpTrick/sage-notebook
The images these files produce contain everything you could possibly want in a jupyter notebook. Additional packages can be installed in notebooks with `%%bash` magic.
## Building the Image
To build this image, run `build/build.fish.` Edit the script to customize the image.
## Configuration
Change the permissions on the configuration volume (see `docker-compose.yml`) if you get errors.
You may want to add the following lines to `jupyter_notebook_config.py`:
```python
c.ServerApp.local_hostnames = ["localhost", "0.0.0.0"]
c.ServerApp.token = "a_long_random_string"
c.TagRemovePreprocessor.enabled = True
c.TagRemovePreprocessor.remove_cell_tags = ("hide_cell",)
c.TagRemovePreprocessor.remove_all_outputs_tags = ("hide_output",)
c.TagRemovePreprocessor.remove_input_tags = ("hide_input",)
```
The first sets `0.0.0.0` to the allowed local hostname list. This prevents a possible "connection reset" when connecting to the docker container. Opens jupyter up to the outside internet, but you should have your inbound ports firewalled off anyway.
The second line sets a persistent token. If this config value is unset, jupyter will generate a new one on each run, making it hard to make a one-click bookmark.
## Image contents
- jupyterlab, jupyterhub, notebook
- latex + pandoc for exporting
- mamba for packages
- Python
- Scrape tools (bs4, requests)
- scipy, scikit-learn, scikit-image
- tensorflow
- sympy
- numpy
- matplotlib
- pandas
- manim
- Octave
- R
- Julia
## TODO
- Auto dark theme
- Fix sage
- GPU support ([source](https://github.com/iot-salzburg/gpu-jupyter))
- Haskell ([source](https://github.com/IHaskell/ihaskell-notebook))
- Rust Kernel
- C++ Kernel
- Perl Kernel
- Maxima Kernel