Skip to content

Latest commit

 

History

History
75 lines (53 loc) · 2.47 KB

README.md

File metadata and controls

75 lines (53 loc) · 2.47 KB
Raw

nbclick - Turn Jupyter notebooks into command line applications

License: MIT PyPI version

nbclick allows you to modify and run Jupyter notebooks from the commandline. It builds on top of nbparameterise which allows programmatic extraction and modification of parameters of Jupyter notebooks.

Installation

nbclick can be installed using pip:

python -m pip install nbclick

It is also possible to run nbclick without prior installation using pipx:

pipx run --system-site-packages nbclick

Note that the --system-site-packages flag is absolutely necessary if your notebook depends on any non-standard library Python package.

Running nbclick

After installation, you can run nbclick using the commandline:

nbclick

The most important argument is the NOTEBOOK parameter. For a given notebook, you can again use --help to display the configuration options:

nbclick mynotebook.ipynb --help

Preparing a notebook for execution with nbclick

nbclick relies on nbparameterise to extract command line options from your Jupyter notebook. The best way to specify customizable parameters is to place them into the first code cell of the notebook as simple assignments:

num_samples = 1000      # The number of samples to draw
outfile = "output.csv"  # The filename to store the results

For above case, the output of nbclick notebook.ipynb --help will be:

Usage: nbclick notebook.ipynb [OPTIONS]

Options:
  --num_samples INTEGER  The number of samples to draw  [default: 1000]
  --outfile TEXT         The filename to store the results  [default:
                         output.csv]
  --help                 Show this message and exit.

Limitations

There are a few known limitations that result from upstream projects that I currently do not plan to fix for nbclick:

  • The number of parameter types recognized is quite small. nbclick is known to work with int, float, bool, str, list (of both homogeneous and heterogeneous type). Most notably, nbparameterise does not support tuples.
  • List parameters are restricted to fixed length (defined by their default). This results from click voluntarily chosing not to provide variable length list parameters, as it introduces ambiguity of the parser.