Skip to main content

Stage 8: It Understands

Course progressStage 8 of 10
~35 min
Build

keyword matching so messy sentences still work

Learn

how string methods clean text and how in finds a word

Ship

an assistant that answers 'can you tell me a JOKE please?'

The big idea​

Until now your assistant has been picky. It only reacts if you type the command exactly — joke, all lowercase, nothing extra. Type JOKE or tell me a joke and it shrugs. Real assistants are more forgiving. This stage is the single biggest jump in how smart yours feels.

Two ideas do the work. First, string methods — little skills attached to text. .lower() makes everything lowercase so JOKE and joke match. .strip() removes accidental spaces at the ends. You chain them right onto the input.

Second, the in operator. "joke" in command asks "does the word joke appear anywhere in what they typed?" That one change means "tell me a joke please" triggers the joke skill, because the word joke is in there somewhere.

New words
string method
A skill attached to text with a dot, like command.lower() or command.strip().
.lower()
Makes text all lowercase, so JOKE and joke count as the same.
in
Checks whether one piece of text appears inside another: "joke" in command.
Before you start

Make sure you finished Stage 7: It Knows Things — your assistant should tell the real time and look colorful.

Build it​

Step 1 — Clean the input​

People type messily — capital letters, extra spaces. We fix that the moment we read the command by chaining two string methods onto input().

Think first

Read the chain

After command = input("> ").lower().strip(), if someone types ' JOKE ' with capitals and spaces, what is stored in command?

Check your thinking

"joke" — .lower() makes it lowercase and .strip() trims the spaces off both ends. We clean it once, up front, so the rest of the loop can stay simple.

Python code task
Update your file

assistant.py

Where it goes: Change the first line inside your while loop to add .lower().strip().

Clean the text once, right where you read it.

    command = input("> ").lower().strip()

Step 2 — Match on keywords instead of exact words​

Now switch every == test to an in test. Instead of "is the command exactly joke?" we ask "is the word joke somewhere in the command?"

Trace it

Trace a messy command

  1. The user types "Can you tell me a JOKE please".
  2. .lower().strip() turns it into "can you tell me a joke please".
  3. "joke" in command is True, because joke really is in there.
  4. tell_joke() runs. It understood you.
Python code task
Update your file

assistant.py

Where it goes: Replace your whole if/elif chain inside the loop with this keyword-matching version.

Every check is now 'in' instead of '=='. The quit check still comes first.

    if "quit" in command or "exit" in command:
        speak("Goodbye for now!")
        break
    elif "hello" in command:
        greet(name)
    elif "bye" in command:
        say_bye()
    elif "joke" in command:
        tell_joke()
    elif "time" in command:
        tell_time()
    elif "date" in command:
        tell_date()
    elif "roll" in command or "dice" in command:
        roll_dice()
    elif "math" in command:
        do_math()
    elif "help" in command:
        show_help()
    else:
        speak("I'm not sure what you mean. Type help.")

Run it and talk to your assistant like a person: type tell me a joke, what is the TIME?, please roll the dice. It understands all of them now. This is the moment it stops feeling like a vending machine and starts feeling like it gets you.

Understand it​

Nothing about the meaning changed — the assistant still has no idea what a joke is. But by cleaning the text and searching for a keyword, you made its rules far more forgiving, and forgiving rules feel like understanding. That's a real insight into how a lot of "smart" software works: clever, flexible matching on top of simple logic you wrote.

This approach has an honest weakness, and it's worth knowing: keyword matching can be fooled. Type "what time is it, no jokes" and it might tell a joke, because the word joke is in there. The order of your elif branches decides which keyword wins when two match. There's no perfect rule here — just tradeoffs you choose. Noticing that tradeoff is exactly the kind of thinking that separates someone who copies code from someone who understands it.

Try this​

Learning beat

Try this

Three short experiments. Predict before you run, then test your guess.

Predict first

Type a sentence that contains two keywords, like "tell me the time and a joke". Predict which one your assistant does first. (Hint: look at the order of your elif branches.)

Compare

Try the same command with and without .lower(). Type JOKE after temporarily removing .lower(). Compare what happens. Now you can feel exactly what that one method does.

Connect

Your assistant can understand commands, but it forgets everything the moment you quit. What if you wanted it to remember a fact you told it — like your favorite color? (Stage 9 gives it a memory.)

Test your stage​

Stuck? Compare carefully
Answer check
Debug compare only

assistant.py

Where it goes: Compare this against your whole file. Use it only if your program won't run. Only the loop changed this stage.

This is the full file at the end of Stage 8.

import subprocess
import random
import datetime
import math
from rich.console import Console


console = Console()


def speak(text):
    print(text)
    subprocess.run(["say", text])


def greet(person):
    speak(f"Hello again, {person}!")


def say_bye():
    speak("See you later!")


def show_help():
    console.print("[bold cyan]I can do:[/bold cyan] hello, bye, joke, time, date, roll, math, help, quit")
    speak("Here are my commands. They're on the screen.")


jokes = [
    "Why did the computer go to the doctor? It caught a virus!",
    "Why was the math book sad? It had too many problems.",
    "What do you call a bear with no teeth? A gummy bear!",
    "Why do programmers prefer dark mode? Because light attracts bugs!",
]


def tell_joke():
    joke = random.choice(jokes)
    speak(joke)


def tell_time():
    now = datetime.datetime.now()
    speak("The time is " + now.strftime("%I:%M %p") + ".")


def tell_date():
    today = datetime.date.today()
    speak("Today is " + today.strftime("%A, %B %d") + ".")


def roll_dice():
    number = random.randint(1, 6)
    speak(f"You rolled a {number}.")


def do_math():
    number = input("Square root of what number? ")
    answer = math.sqrt(float(number))
    speak(f"The square root of {number} is {answer}.")


console.print("[bold green]==== PIXEL ====[/bold green]")
console.print("[cyan]Your Python assistant is online.[/cyan]")
speak("Hi! I'm Pixel, your assistant.")

name = input("What is your name? ")
speak(f"Nice to meet you, {name}!")
speak(f"Type commands, {name}. Type help to hear what I can do.")

while True:
    command = input("> ").lower().strip()

    if "quit" in command or "exit" in command:
        speak("Goodbye for now!")
        break
    elif "hello" in command:
        greet(name)
    elif "bye" in command:
        say_bye()
    elif "joke" in command:
        tell_joke()
    elif "time" in command:
        tell_time()
    elif "date" in command:
        tell_date()
    elif "roll" in command or "dice" in command:
        roll_dice()
    elif "math" in command:
        do_math()
    elif "help" in command:
        show_help()
    else:
        speak("I'm not sure what you mean. Type help.")
  • Typing JOKE in capitals now works.
  • Typing a full sentence like tell me a joke triggers the joke.
  • Extra spaces before or after a command don't break it.
  • Design check. Try to fool your assistant with a tricky sentence. Did you find a case where keyword matching guesses wrong? That's a real limitation, not your mistake.
  • From memory. Without looking, write the line that reads a command and cleans it. Did you chain .lower().strip()?

If it breaks​

  • AttributeError: 'str' object has no attribute 'Lower'. Method names are lowercase: .lower(), not .Lower(). Python is picky about this.
  • Everything triggers the same skill. Check your branch order and keywords. A very common word like a single letter will match too much — use distinctive keywords like joke and time.
  • A sentence does the wrong skill. That's keyword matching being fooled, not a bug. The earliest matching elif wins — reorder your branches if a different one should take priority.
  • hello stopped working. Make sure you switched to "hello" in command, with quotes around the keyword and in between.