Conveyor

Compute and use Jupyter notebook cell outputs in other notebooks and scripts with just a couple lines of code.

Use Cases

- Split ordered steps across multiple notebooks, picking up the most recent state information with each new notebook - Prototype and organize workflows entirely in Jupyter notebooks without having to manage script exports - Improve performance by reducing overheads from notebook servers and unused code cells

Requirements

Conveyor currently only supports Jupyter notebooks written in Python. It is important to ensure that you have the version of Python (and any dependencies) used in your notebooks also installed locally, or on the machine using this library.

Examples

For quick and simple access to values in a prior notebook:

conveyor.runnotebook("conveyor/examples/Sample Calculations I.ipynb", importglobals=True)
all notebook globals pushed to conveyor.nbglobals with import_globals flag
from conveyor.nbglobals import x, y, z, fig
print(y)

Conveyor can also provide you with any information available in a prior Jupyter workspace.

results = conveyor.run_notebook("conveyor/examples/tests/Sample Calculations I.ipynb")
Cells are zero-indexed, only code cells are counted
codecellidx = 1
Get a cell's source code
code = results[codecellidx]['code']
Get a cell result
cellresult = results[codecell_idx]['result']
Get cell stdout
cellstdout = results[codecell_idx]['stdout']
To get any variable available in notebook
x = results.getvar('x')

There are options for running notebooks that can be used to optimize notebook execution, or stitch notebooks together in pipelines.

from conveyor.multinb import Pipeline
...
data_processing = Pipeline()
The variable 'df' from load_data.ipynb will replace 'df' in 
process_data.ipynb, starting from the third code cell.
dataprocessing.addnotebook(filename="conveyor/examples/tests/loaddata.ipynb", carryvars=['df'])
dataprocessing.addnotebook(filename="conveyor/examples/tests/process_data.ipynb", 
    carryvars=['magicnumber'], startcellidx=3)
Add custom intermediary steps
def transformmagic(fromstate):
    to_state = dict()
    tostate['transformedmagicnumber'] = -1 * fromstate['magic_number']
    return to_state
dataprocessing.addtransform(transform_magic)
Get output of selected variables from each stage
results = data_processing.run()

Installation

This package is available on pypi. Install it using pip with

pip install jupyter-conveyor

Conveyor is only compatible with Python versions > 3.5.

Documentation

See the docs folder, or read them here.