Skip to content

Lab: M3: functions, files, libraries & errors

You'll need: your own laptop and its terminal. This is your first local install: set aside time and go slowly. Time: ~60 min • Work in your breakout pair.

Heads up: installing software is where everyone hits a snag, it is not a sign you're behind. We go one tiny step at a time, separately for each operating system. Nothing here can harm your computer. If a step doesn't match your screen, check the troubleshooting table in resources/install-guides/python-venv.md.

This lab has two parts: - Part A: set up Python, a virtual environment, and your first library. - Part B: write reusable functions that read a file, save JSON, and survive a bad line.

flowchart LR
  F["expenses.txt<br/>(a file you read)"] --> P["functions transform it<br/>(categorize each)"]
  P --> J["expenses_categorized.json<br/>(a file you write)"]
  P -.bad line?.-> E["try / except<br/>skips it safely"]

Part A: set up Python on your machine

Use python on Windows and python3 on macOS/Linux. The commands below show macOS/Linux; Windows differences are noted. Full per-OS detail: the install guide.

Step 1: Open your terminal

Open Terminal (macOS/Linux) or PowerShell (Windows). New to it? See terminal.md.

You should now see: a window with a prompt ending in $ (macOS/Linux) or > (Windows) and a blinking cursor.

Step 2: Install Python and check the version

Install Python 3.10+ following the install guide, Step 1 for your OS. Then, in a new terminal: 

python3 --version      # Windows: python --version
You should now see: Python 3.12.x (any 3.10 or newer). If not, see the guide's Troubleshooting table, don't push on until this works.

Step 3: Make a project folder

mkdir ai-course && cd ai-course      # Windows: mkdir ai-course; cd ai-course
You should now see: your prompt now ends in ai-course. You're inside the project folder.

Step 4: Create a virtual environment

A virtual environment is a private box for this project's libraries. 

python3 -m venv .venv      # Windows: python -m venv .venv
You should now see: no message, but a new .venv folder now exists in ai-course. (You create this once.)

Step 5: Activate it

source .venv/bin/activate      # Windows PowerShell: .\.venv\Scripts\Activate.ps1
You should now see: your prompt now starts with (.venv). That means the box is active. (Re-run this whenever you open a new terminal. Windows "scripts disabled" error? See the guide.)

Step 6: Install your first library

pip install rich
You should now see: lines ending in Successfully installed rich-15.x …. You just installed someone else's code to reuse, that's pip.

Step 7: Confirm it's there

pip list
You should now see: a list of packages that includes rich (e.g. rich 15.0.0). Setup done. Take a breath, the hard part is behind you.

Part B: reusable code that survives the real world

Step 8: Get the starter files

Put budget_pro_starter.py and expenses.txt (from this module's starters/ folder, your instructor will share them) into your ai-course folder, next to .venv.

You should now see: running ls (Windows: dir) lists budget_pro_starter.py, expenses.txt, and .venv.

Step 9: Run it once and watch errors get handled

With your venv active: 

python budget_pro_starter.py      # macOS/Linux may need python3
You should now see: a line Skipping bad line: 'this line is broken on purpose', then each expense printed with -> None. Two things just happened: it read the file, and the deliberately broken line didn't crash the program: try/except caught it and skipped it. (The sizes are None because we haven't filled in the function yet, next step.)

Step 10: Finish the categorize() function

Open budget_pro_starter.py in any editor. Find categorize() and replace the two "____" blanks with "medium" and "big". Save, then run it again. 

python budget_pro_starter.py
You should now see: each expense now ends in a real size, e.g. Coffee: $4.50 -> small, Groceries: $62.00 -> big. A function is a named, reusable block, you call categorize(...) once per expense and it hands back an answer.

Step 11: Save the results as JSON

In the same file, find the # TODO: save the list ... comment near the bottom and replace the print("Done ...") placeholder with the two lines shown in the hint: 

with open("expenses_categorized.json", "w") as f:
    json.dump(expenses, f, indent=2)
Save and run once more.

You should now see: a new file expenses_categorized.json in your folder. Open it, it's your list of dictionaries written as JSON text. This is the exact format an AI API speaks (M4). You just wrote data to disk.

Step 12: See the library pay off (the full solution)

Now run the finished version, which uses rich to draw a table: 

python ../path/to/solution/budget_pro.py    # or copy solution/budget_pro.py into your folder and run it
(Simplest: copy solution/budget_pro.py into ai-course next to expenses.txt, then python budget_pro.py.)

You should now see: a neat bordered table of your expenses with colours, a bold total, and a green Saved expenses_categorized.json. That polish is the rich library doing the work, code you installed, not code you wrote.

Step 13: Make it yours

Edit expenses.txt to your real spending (keep the item, amount shape), then re-run.

You should now see: your own expenses categorized, totalled, and saved. You didn't touch the functions, they handle whatever the file contains. That's reusable code.

Stuck on the code? The finished, commented version is in ../solution/budget_pro.py. Peek only after you've tried.

Your win

You set up Python locally, installed and used a library, organized your code into functions, read a file, saved JSON, and handled a bad line without crashing, the exact toolkit Part B's AI API needs.

Post it to the chat wins board: a screenshot of your rich table, or "Python's running on my machine, I read a file, saved JSON, and a broken line didn't even slow me down "

Take-home (optional)

Add a new function count_big(expenses) that returns how many expenses are "big", and print it. One small function, reusing categorize(). (Hint: loop and count, like M2's big_count.)