Class 19 - Structured data

Date: October 26, 2026

⚠️Homework

Homework 7 due November 1, 2026

‼️Ending Skills

Last week for skills: 10, 11

🆕New Skills

First week for skills: 5

Skills: 6

Notes (review before class)

Structured Data

• Some data has several attributes that belong together: a song has a name, singer, and year; an email has a sender, subject, and body.
• Structured data: a single datum made of a fixed number of parts, called fields, which can have different types (and can themselves be structured: a date has a day, month, and year).

Defining Structured Data

A dataclass defines a new type, with a type annotation for each field (dataclasses need from dataclasses import dataclass at the top of the file):

from dataclasses import dataclass

@dataclass
class ITunesSong:
    name: str
    singer: str
    year: int

• The names of types we define start with a capital letter.
• ITunesSong is also a constructor: a function that takes a value for each field, in order, and returns a new ITunesSong.
• Always follow a data definition with a few named examples:

lver = ITunesSong("La Vie en Rose", "Édith Piaf", 1945)
so = ITunesSong("Stressed Out", "twenty one pilots", 2015)
wnkkhs = ITunesSong("Waqt Ne Kiya Kya Haseen Sitam", "Geeta Dutt", 1959)

• In the directory, each example is a name for an ITunesSong value, like any other.
• The new type can be used in annotations: lver: ITunesSong = ITunesSong(...). (Try annotating it as str instead, to see the error.)

Field Access

s.year is the year field of s:

def song_age(s: ITunesSong) -> int:
    """returns how many years old the song is"""
    return 2026 - s.year

def test_song_age():
    assert song_age(lver) == 81
    assert song_age(so) == 11
    assert song_age(wnkkhs) == 67

Recap (10 mins)

A library program keeps a record of each book: its title, author, and number of pages. Read through the program:

from dataclasses import dataclass

@dataclass
class BookRecord:
    title: str
    author: str
    pages: int

the_dispossessed = BookRecord("The Dispossessed", "Ursula K. Le Guin", 387)
to_the_lighthouse = BookRecord("To the Lighthouse", "Virginia Woolf", 209)
brave_new_world = BookRecord("Brave New World", "Aldous Huxley", 268)

def pages_left(b: BookRecord, pages_read: int) -> int:
    """how many pages of b are left to read, after reading pages_read of them"""
    return b.pages - pages_read

def test_pages_left():
    assert pages_left(the_dispossessed, 0) == 387
    assert pages_left(to_the_lighthouse, 9) == 200

• What is brave_new_world.author? What is pages_left(brave_new_world, 68)?
• What would BookRecord("Brave New World", 268, "Aldous Huxley") do? Why?
• If the library also recorded the year each book was published, which lines would change?

Class Exercise (45 mins)

1. Design a function that returns a "summary string" for a BookRecord, including the title, author, and pages.

2. Design a function is_long_book that takes in a BookRecord and returns whether or not the book has more than 350 pages.

3. Design a new data type for a Podcast, and based on your own knowledge figure out the fields that make sense to include in it.

4. Design a function podcast_summary that takes in a Podcast and produces a string summarizing a Podcast.

5. Design a Recipe data definition, and then write a row_to_recipe function that consumes a row from the following table (convert numeric columns after loading):

   recipes_raw = load_table("https://raw.githubusercontent.com/neu-pdi/cs2000-public-resources/refs/heads/main/static/support/5-recipes.csv")
   recipes_with_servings = recipes_raw.transform_column("servings", int)
   recipes = recipes_with_servings.transform_column("prep time", int)

6. Test your function using recipes.row(0) or some other row number, and then use it to add a new column with add_column.

Wrap-up (5 mins)

• Structured data groups multiple fields together.
• We define structured types using @dataclass and class, and create instances by calling the constructor.
• Fields are accessed with the dot notation.