name: refactor_nb description: Jupyter Notebook (.ipynb) 内のコード要素を一括置換・リファクタリングする

refactor_notebooks Skill

This skill is for programmatically analyzing Jupyter Notebook (.ipynb) files and performing batch replacement or modification of specific import patterns or code blocks.

Instructions

1. Analyze Notebook Structure:

   • Load .ipynb files as JSON using Python and iterate through the cells list.
   • Verify if the cell_type of each cell is code.

2. Filter and Match:

   • Join the source field (list format) into a string and use regular expressions or keyword matching to identify target cells.
   • Target specific import statements (e.g., from gwexpy.noise import asd), function calls, or targeted comments.

3. Implement Transformation:

   • Create a transformation script to rewrite the source list of the cells in memory.
   • The updated source must be in list format (each element as a string ending with a newline).

4. Write and Verify:

   • Save the file using json.dump, maintaining an indent of 1 (a gwexpy convention) and specifying ensure_ascii=False.
   • Verify that the resulting notebook is valid JSON and that the intended changes have been applied using view_file or similar.

Usage Guidelines

• When modifying multiple notebooks across a directory, create a loop script that collects .ipynb files using glob or similar.
• For complex refactoring, adopt a workflow of saving the logic as a temporary .py script, executing it via run_command, and then deleting the script.