kernel_name argument of papermill.execute seems to have no effect. · nteract/papermill#856

(0 留言) (0 反應) (0 負責人)Python (5,381 star) (402 fork)batch import

bughelp wanted

描述

🐛 Bug

The kernel_name argument of papermill.execute seems to have no effect. The executing notebook uses the Jupyter kernel specified in the original notebook's metatdata. Using papermill v2.6.0. Or at least the display_name in the metadata is unchanged, this is what jupyter hub uses to display the name of the kernal being used.

Something along the lines of this function should fix this:

import nbformat as nbf

def change_notebook_kernel(input_path, output_path, kernel):
    """
    Change a Jupyter Notebook's kernel.

    Parameters
    ----------
    input_path: str
        Path to the input notebook.
    output_path: str
        Path to the output notebook.
    kernel: str
        Name of the kernel_name to change to.
    """
    notebook = nbf.read(input_path, as_version=4)
    notebook['metadata']['kernelspec']['name'] = kernel
    notebook['metadata']['kernelspec']['display_name'] = kernel
    nbf.write(notebook, output_path)

貢獻者指南

技術棧: python
領域: data
議題類型: bug
難度: 2
預計時間: under 1 hour
活動狀態: fresh
清晰度: clear
前置要求: PythonpapermillJupyter notebooks
新手友善度: 75
研究方向: Investigate the execute function in papermill's core module (likely papermill/execute.py) to understand how kernel name is passed to the notebook execution. The fix should update the notebook's metadata kernelspec name and display name after execution, similar to the provided snippet. Also check if there are existing tests for kernel specification.