Automating simple tasks on Google Drive and Google Storage for the Gutenberg Changelog podcast

This week, I succeeded again using AI to automate a few interconnected tasks. Gemini is a great buddy for that. Now I have command line tool, written in Python that does this:

Set-up a new podcast episode on Google Drive:

Create a new folder for the next episode
Find the Run Down tempate in the podcast top folder
Drop a copy of the Run Down template into the next episode’s folder.

Publish the mp3 file on Google Storage

After the producer drops the final episode file in the episode’s folder this command line script does the following

Find the file
Download the file
Upload it to Google Cloud bucket
Make it public
Display the URL of the public file.

Key Technologies and Features

When I look through the final list of the various touchpoints, I probably wouldn’t have succeeded in creating this in a million years all by myself. With Gemini, I now have super powers ????.

Python: The programming language used for both scripts.
Google Drive API & Google Cloud Storage API: These are the services the scripts interact with.
OAuth 2.0: Used for secure, user-based authentication for Google Drive operations.
Service Accounts: Implemented for highly reliable and automated authentication for Google Cloud Storage uploads.
config.ini: A separate configuration file where you store important IDs (like your main “Gutenberg Changelog” folder ID, template name, and GCS bucket name) and other settings, making the scripts easy to customize without editing code.
Command-Line Arguments (argparse): Allows you to pass episode details directly when running the scripts, making them efficient for automation.
Flexible File Finding: The publish_episode.py script includes smart search capabilities to find folders and MP3s based on partial names and modified dates.
Temporary Files & Cleanup: MP3s are downloaded to temporary locations and automatically cleaned up after upload.
Confirmation Prompt: A safety step in publish_episode.py to confirm the selected MP3 before public upload.

It doesn’t sound like much, but I wanted to do this a long time. However the time it would have taken myself to figure out all the pieces, I probably could have handled 300 episodes manually. So only useing an AI made this part feasible for me.

Gemini helped me get all the google credentials right for Google Drive, and Google Storage. It’s not an easy thing.

source venv/bin/activate
    python3 publish_episode.py --episode 120

While going through the processes, and debugging all the permission errors one at a time, I realized something: Who ever wrecked the IAM service at Google and brought down half of the Internet, in the Great Outage of the Internet of June 12, 2025, might not have worked with an AI. The AI was spot on with my problems. ????? <joking/>

I also played with Sora again to make a couple of Gibli Podcast Studo graphics.