Automating Your Life with Python Scripts

The useful threshold is roughly this: if a task takes five minutes and you do it weekly, that is four hours a year. An hour spent scripting it pays back by spring, and the script never gets bored and skips a step.

pathlib makes filesystem work readable and cross-platform. Handle name collisions from the start, since that is where naive scripts destroy data.

```python
from pathlib import Path
import shutil

GROUPS = { 'images': {'.jpg', '.jpeg', '.png', '.webp', '.heic'}, 'docs': {'.pdf', '.docx', '.xlsx', '.csv'}, 'archives': {'.zip', '.tar', '.gz', '.7z'}, }

def tidy(folder: Path, dry_run: bool = True) -> None: for item in folder.iterdir(): if not item.is_file(): continue group = next((g for g, exts in GROUPS.items() if item.suffix.lower() in exts), 'other') target_dir = folder / group target = target_dir / item.name

n = 1 while target.exists(): target = target_dir / f'{item.stem} ({n}){item.suffix}' n += 1

print(f'{item.name} -> {group}/{target.name}') if not dry_run: target_dir.mkdir(exist_ok=True) shutil.move(item, target) ```

Defaulting dry_run to True is the habit that saves you. Watch it print the plan, then pass False.

Check for an API before parsing HTML, since a documented JSON endpoint is stable and a page layout is not. When you must scrape, identify yourself, rate limit, and cache responses locally so development iterations do not hammer someone's server.

```python
import httpx, time
from selectolax.parser import HTMLParser

headers = {'User-Agent': 'personal-price-tracker (contact: me@example.com)'}

with httpx.Client(headers=headers, timeout=15, follow_redirects=True) as client: for url in urls: r = client.get(url) r.raise_for_status() tree = HTMLParser(r.text) node = tree.css_first('[data-testid=price]') print(url, node.text(strip=True) if node else 'not found') time.sleep(2) ```

A script that runs unattended needs different qualities from one you babysit.

• Log to a file, not stdout. When it fails at 3am you will want the traceback, not a memory.
• Make it idempotent. Running twice should not double-process anything. Keep a record of what you have already handled.
• Keep secrets in environment variables and never in the file you eventually push to GitHub.
• Add argparse flags for dry runs, paths, and verbosity. Editing constants at the top gets old fast.
• Exit non-zero on failure so your scheduler notices something went wrong.

cron on Linux and macOS, Task Scheduler on Windows, GitHub Actions on a cron trigger when the job needs no local machine. For anything with dependencies between steps, reach for a proper task runner rather than chaining scripts with sleep calls.

Start with the annoyance you complained about most recently. That is the one worth twenty lines.