sheet

Tools to extract content from ipython (jupyter) notebooks

To install: pip install sheet

Examples

Getting the filepath of the current notebook

filepath = get_path_of_current_notebook()
# Example: '~/my_notebook_folder/some_notebook.ipynb'

Access to cells of a notebook given its filepath

from sheet.contents import get_ipynb_cells, get_ipynb_cells_source
filepath = '~/my_notebook_folder/some_notebook.ipynb'

cells = get_ipynb_cells(filepath)
assert type(cells), type(cells[0]) == (list, dict)

cells = get_ipynb_cells_source(filepath)
assert type(cells), type(cells[0]) == (list, str)

from sheet.contents import get_ipynb_cells_full_text
notebook_text = get_ipynb_cells_full_text(filepath)
print(notebook_text)

Search the cells of a single notebook

Index and search the cells of a notebook

from sheet import SingleNotebookSearch

search = SingleNotebookSearch(src=None)  # if no filepath (src) to a notebook is given, will use the "current notebook"

result_indices = search('lines iterize')
print(result_indices)
print("\n---- Contents of first hit ----")
print(search[result_indices[0]])  # print the contents of the first result

[70 225 226 198 199 196 200 201 193 197]

---- Contents of first hit ----
process_wf = Line(
    partial(fixed_step_chunker, chk_size=DFLT_CHK_SIZE),
    iterize(process_chk)
)

Search the contents of the notebooks under a directory

from sheet.contents import SearchNotebooks

search = SearchNotebooks('~/my_notebooks_folder', max_levels=0)  # enter max_levels=None for full recursive
search('bayesian')

array(['Spyn 01 - Potentials.ipynb',
       'Bayes 01 - Potentials-Only explanation.ipynb', 'taped.ipynb',
       'separation of concerns - how py2store does it.ipynb',
       'equate.ipynb', 'peruse.ipynb',
       'hum, taped, lined -- feeding audio to a pipeline.ipynb',
       'owner.ipynb', 'best of 2020.ipynb',
       'Bayes 02 - Potentials - And drug data example.ipynb'],
      dtype=object)

Okay, we have a list of notebooks that match our query (i.e. the highest average alignment to our query -- not just keyword matching!), but what cells in particular have the highest relevance?

Well, we can now peruse our notebook at that level, with a notebook cells searcher. (Note: You can combine both to make a cell-level searcher from the folder level.)

ss = search.search_notebook('Spyn 01 - Potentials.ipynb')
ss('bayesian')

array([['Spyn 01 - Potentials.ipynb', 6],
       ['Spyn 01 - Potentials.ipynb', 2],
       ['Spyn 01 - Potentials.ipynb', 71],
       ['Spyn 01 - Potentials.ipynb', 88],
       ['Spyn 01 - Potentials.ipynb', 91],
       ['Spyn 01 - Potentials.ipynb', 84],
       ['Spyn 01 - Potentials.ipynb', 85],
       ['Spyn 01 - Potentials.ipynb', 86],
       ['Spyn 01 - Potentials.ipynb', 87],
       ['Spyn 01 - Potentials.ipynb', 82]], dtype=object)

ss['Spyn 01 - Potentials.ipynb', 6]

'# Potentials - A key data structure to Discrete Bayesian Inference'

ss['Spyn 01 - Potentials.ipynb', 87]