Jupyter Notebook to Python Script Converter

Jupyter notebooks are great for exploration and visualization, but they have limitations in production: they can't be imported as modules, are difficult to version control cleanly (they store outputs in JSON), don't work well with CI/CD pipelines, and can have hidden state from out-of-order cell execution. Converting to a .py script makes code importable, diffable, testable with pytest, runnable from the command line via cron or task schedulers, and compatible with code review workflows.

Magic commands are Jupyter-specific syntax: line magics start with % (e.g., %matplotlib inline, %timeit, %load_ext) and cell magics start with %% (e.g., %%bash, %%sql). Shell commands start with ! (e.g., !pip install, !ls). These are interpreted by the IPython kernel and are not valid Python syntax — they will raise SyntaxError if run as a plain .py file. Remove them when converting to a script, or replace %matplotlib inline with matplotlib.use('Agg') and !pip install X with a requirements.txt entry.

The # %% comment is a special marker recognized by VS Code (with the Jupyter extension) and Spyder IDE as a 'code cell' boundary. This means your converted .py file can still be run interactively cell-by-cell inside VS Code — just like a notebook — without being a .ipynb file. Click 'Run Cell' above any # %% block to execute just that section. This gives you the best of both worlds: a clean .py file for version control and CI/CD, with notebook-like interactivity during development.

After conversion, run the script with: python script.py. For scheduled jobs, use cron (Linux/Mac) or Task Scheduler (Windows). For parameterized runs, consider Papermill (papermill notebook.ipynb output.ipynb -p param value) or replace hardcoded values with argparse. For ML pipelines, tools like Prefect, Airflow, or Dagster can orchestrate .py scripts as pipeline tasks. Always test the script in a clean virtual environment first to catch any missing dependency or implicit global state.