Nesting: A List of Dictionaries

Real data is rarely a single list or a single dictionary. It is a list of records, where each record is a dictionary with the same keys: a row of a spreadsheet, a contact in an address book, an order in a shop. This shape is so common that once you can read it, you can read most of what files and the internet will hand you.

Think of a stack of identical index cards. Every card has the same printed headings, name, city and age, and each card is filled in differently. The stack is the list; each card is a dictionary.

Records

A record describes one thing with named fields. In Python, a dictionary does this naturally: {"name": "Anna", "city": "Lisbon", "age": 34}. The keys are the headings on the card, and the values are what is written under them.

A List of Records

Put several records in a list and you have a small table. Walk it with a for loop and read one field from each record: here, every person's name and city.

people = [
    {"name": "Anna", "city": "Lisbon", "age": 34},
    {"name": "Omar", "city": "Leeds", "age": 29},
    {"name": "Lena", "city": "Lisbon", "age": 41},
]

for person in people:
    print(person["name"], "lives in", person["city"])

Anna lives in Lisbon
Omar lives in Leeds
Lena lives in Lisbon

Reaching Inside

To reach one value directly, chain the brackets and read them left to right. people[1] is the second record, Omar's card; people[1]["city"] is the city on that card, Leeds. The common mistake is to skip the first step: people["city"] asks the list for a key, and lists do not have keys, so Python raises a TypeError saying list indices must be integers.

>>> people = [{"city": "Lisbon"}, {"city": "Leeds"}]
>>> people[1]["city"]
'Leeds'
>>> people["city"]
Traceback (most recent call last):
  File "<python-input-2>", line 1, in <module>
    people["city"]
    ~~~~~~^^^^^^^^
TypeError: list indices must be integers or slices, not str

Filtering and Summarising

The patterns you already know combine without anything new. To find everyone in one city, loop and keep the records that match; a comprehension does it in one line. To average an age, total it with sum over the ages and divide by the count.

people = [
    {"name": "Anna", "city": "Lisbon", "age": 34},
    {"name": "Omar", "city": "Leeds", "age": 29},
    {"name": "Lena", "city": "Lisbon", "age": 41},
]

in_lisbon = [p["name"] for p in people if p["city"] == "Lisbon"]
print(in_lisbon)

average_age = sum(p["age"] for p in people) / len(people)
print(f"Average age: {average_age:.1f}")

['Anna', 'Lena']
Average age: 34.7

The second line inside sum has no square brackets: Python lets you feed a comprehension straight into a function like this, without building a list first. Read it the same way: each person's age, for every person.

A First Look at JSON

The same data, written out as text for another program to read, looks almost identical. The json module turns a list of dictionaries into JSON, the text format most of the internet uses to exchange data. Chapter 10 uses it properly; for now, notice how little changes: the same brackets, braces, keys and values.

import json

people = [{"name": "Anna", "city": "Lisbon"}, {"name": "Omar", "city": "Leeds"}]
print(json.dumps(people, indent=2))

[
  {
    "name": "Anna",
    "city": "Lisbon"
  },
  {
    "name": "Omar",
    "city": "Leeds"
  }
]