Simple Python Scripting for Daily Automation

Most Python tutorials build toy apps. Real daily scripting is different. You need something that runs in ten seconds, reads a file, does one thing, and exits. Here are the patterns I use for the little scripts that save me an hour a day.

The Single-File Script

My automation scripts live in ~/scripts/. Each one is a single .py file. No packages, no setup.py, no virtual environments. Just a shebang line and a function.

#!/usr/bin/env python3
"""Rename all downloads to include their date."""
import os
import sys
from datetime import datetime
 
def main():
    downloads = os.path.expanduser('~/Downloads')
    for name in os.listdir(downloads):
        full = os.path.join(downloads, name)
        if os.path.isfile(full):
            date = datetime.fromtimestamp(os.path.getctime(full))
            prefix = date.strftime('%Y-%m-%d')
            if not name.startswith(prefix):
                os.rename(full, os.path.join(downloads, f'{prefix}_{name}'))
 
if __name__ == '__main__':
    main()

Save, chmod +x, run. Done.

Use the Standard Library

For scripts, I prefer stdlib over third-party packages. Not because pip is bad, but because scripts that use only stdlib work on any Python install without setup.

• json for parsing JSON files
• urllib.request for simple HTTP
• csv for spreadsheets
• pathlib for path manipulation
• subprocess for shelling out

These five modules cover 80% of what I automate.

Keep Arguments Simple

Two options for a single-purpose script:

• Use sys.argv directly for one or two args
• Use argparse if there are more than two

I skip fancy frameworks like Click for throwaway scripts. They add a dependency for features I do not need.

Fail Loudly

Scripts should crash when things go wrong, not silently skip. I do not wrap my scripts in try: except:. If a file is missing, let Python raise. If the API is down, let urllib raise. I read the traceback and fix the issue. Silent failures are how automation lies to you.

Scheduling

macOS: launchd. Linux: cron. Both handle the 'run this every hour' use case without any Python-specific tooling. I write the script to run once and let the OS schedule it.

The Compounding Payoff

Each script saves a few minutes. But I have 30 of them now. That is an hour a day freed up for actual work. The investment was maybe an hour per script, so the break-even was fast.

A Script Index

With 30 scripts, I forget what each one does. I maintain a README.md in ~/scripts/ that lists every script, its purpose, and an example invocation. When I want to clean up my downloads folder, I open the README and find the right script name. Two seconds of lookup beats remembering filenames.

Naming for Discoverability

I use verb-noun names: rename-downloads.py, archive-screenshots.py, sync-notes.py. The verb tells me what it does, the noun tells me what it acts on. Tab completion finds the right script in two keystrokes.

Upgrading Old Scripts

Occasionally a script breaks because an API changed or a file format changed. I fix it when it breaks, not before. Pre-emptively 'modernizing' scripts that still work is a distraction. The goal is never elegant code; it is time saved.

See the Python standard library documentation for the full set of modules available without installing anything.