Skip to content

M3: Functions, files, libraries & errors

Until now your code lived in a browser and ran top to bottom. Today you move to your own machine and learn the four habits that turn scripts into real software: wrap code in functions you can reuse, pull in libraries other people wrote, read and write files (including JSON: the text form of M2's dictionaries), and catch errors so a typo doesn't crash everything. This is the exact toolkit the AI API needs in Part B.

Today's win: set up Python on your own computer, then run a program that reads a file, uses a library you installed, organizes its work into functions, saves JSON, and survives a bad line of data, and you understand every piece.

Today you will

  • Install Python + a virtual environment + pip on your own machine (your first local install)
  • Organize code into functions and pull in an installed library
  • Read a file → transform it → save it as JSON, and add try/except so mistakes don't crash you

Run of show (~75 min)

Time What we do
0:00 Hook + the win we're chasing
0:05 The one idea: reusable code that survives the real world (full read in notes.md)
0:10 Lab Part A: install Python + venv + a library (go slow, pairs)
0:40 Lab Part B: functions, read a file, save JSON, handle errors
1:10 Show: post your rich table + saved .json to the wins board
1:15 Wrap + take-home

If you get stuck

  • Installing is the hard part, and getting stuck is normal: that's exactly why we go one tiny step at a time, per OS. Nothing here can harm your computer.
  • Re-read the You should now see line. Most setup problems are: the (.venv) prefix is missing (re-activate), or it's python vs python3 (use whichever your version check accepted).
  • The full per-OS steps and a troubleshooting table live in resources/install-guides/python-venv.md. Lean on it.

Instructor note: for a group new to the terminal, run Part A (setup) as its own session and Part B (the code) as the next. The lab is cut at that seam. This is the most setup-heavy module in the course, budget extra time.

Optional challenge

Only if you'll run AI models on your own machine later: skim the PyTorch tooling box in notes.md and install the CPU-only build using the official selector. You do not need it for this course, we use hosted APIs, so skip it with a clear conscience otherwise.