JuBox 0.4.0

Object oriented interface to Jupyter Notebook

JuBox

Programmatical Jupyter Notebook API for humans

Jupyter Notebook extension, Python 3, Easy to use

---

What?

• Pythonic approach to load, modify, run and save Jupyter Notebooks in Python code
• This enables to quickly and easily create:
  • Parametrized notebooks
  • Version controlable notebooks directly from inserted code snippets, Python files or even straight from Python functions and classes.
  • Notebooks with code generated markdown

Installation

• Pip install from pip

pip install jubox

• Clone the source code and pip install:

git clone https://github.com/Miksus/jubox.git
cd jubox
pip install -e .

Example

See demo or cook book for detailed examples.

from jubox import JupyterNotebook, CodeCell

# Load jupyter Notebook
nb = JupyterNotebook("my_file.ipynb")
nb.load()

# Run the notebook
# by making a copy (output is lost if raised exception)
nb_runned = nb()

# or in place (maintains exceptions in the notebook)
nb(inplace=True)


# save the file to the original file
nb.save()
# or to new file
nb.to_ipynb("run_notebook.ipynb")


# Get cells with specific tags
param_cells = nb.cells.get(tags=["parametrized"])

# Rewrite the first tagged cell with dict of parameters
new_cell = CodeCell.from_variable_dict(run_date="2020-01-01", category="blue")
param_cells[0].overwrite(new_cell)

---

Features

• Convenient object oriented API for Jupyter Notebooks.
• API includes:
  • Load, save and run notebooks in Python
  • Conveniently select specific cells using tags, source, regex or the type
  • Conveniently select specific outputs of a cell and render it to HTML or plain text
  • Render Jupyter Notebooks in a Jupyter Notebook!

Test

Pytest was chosen as testing suites. Tests are found in test directory inside the source.

---

Author

• Mikael Koli - Miksus - [e-mail]

---

License

• MIT license