Function process_notebook

src/gitingest/utils/notebook.py:19–74

Process a Jupyter notebook file and return an executable Python script as a string. Parameters ---------- file : Path The path to the Jupyter notebook file. include_output : bool Whether to include cell outputs in the generated script (default: ``True``). Return

(file: Path, *, include_output: bool = True)

def process_notebook(file: Path, *, include_output: bool = True) -> str:
    """Process a Jupyter notebook file and return an executable Python script as a string.

    Parameters
    ----------
    file : Path
        The path to the Jupyter notebook file.
    include_output : bool
        Whether to include cell outputs in the generated script (default: ``True``).

    Returns
    -------
    str
        The executable Python script as a string.

    Raises
    ------
    InvalidNotebookError
        If the notebook file is invalid or cannot be processed.

    """
    try:
        with file.open(encoding="utf-8") as f:
            notebook: dict[str, Any] = json.load(f)
    except json.JSONDecodeError as exc:
        msg = f"Invalid JSON in notebook: {file}"
        raise InvalidNotebookError(msg) from exc

    # Check if the notebook contains worksheets
    worksheets = notebook.get("worksheets")
    if worksheets:
        logger.warning(
            "Worksheets are deprecated as of IPEP-17. Consider updating the notebook. "
            "(See: https://github.com/jupyter/nbformat and "
            "https://github.com/ipython/ipython/wiki/IPEP-17:-Notebook-Format-4#remove-multiple-worksheets "
            "for more information.)",
        )

        if len(worksheets) > 1:
            logger.warning(
                "Multiple worksheets detected. Combining all worksheets into a single script.",
            )

        cells = list(chain.from_iterable(ws["cells"] for ws in worksheets))

    else:
        cells = notebook["cells"]

    result = ["# Jupyter notebook converted to Python script."]

    for cell in cells:
        cell_str = _process_cell(cell, include_output=include_output)
        if cell_str:
            result.append(cell_str)

    return "\n\n".join(result) + "\n"