How to Use Lambda With the Sort Method

At 1:25 in this video, what is the [-1] doing here?

The [-1] gives the last element of a list (in this case the surnames as the names were split with .split(), so that the last element of that list would be the surname). You could use [1], but that would only work if each name given had only two elements - firstname, lastname. if anyone had a middle name, then it wouldn’t work correctly.

When you have the key be lambda x, the function list.sort(key) automatically feeds each element in the list to the lambda function rather than the list itself?

@Zach Thomas Not exactly. Both the sorted() function and the .sort() method accept an optional key parameter, which lets you define the order of your elements. By default, they’re sorted by value, which produces different results depending on their type:

>>> sorted([42, 177, 13])
[13, 42, 177]

>>> sorted(["42", "177", "13"])
['13', '177', '42']

For example, numbers are sorted using total order, while strings are sorted in lexicographic order. You can override this default behavior by providing a custom callable, whether it’s a lambda expression or a fully-fledged function, to define your key. This callable will receive a single element from the list:

>>> fruits = ["watermelon", "fig", "plum", "apple", "orange"]

>>> sorted(fruits, key=len)  # Sort by string length
['fig', 'plum', 'apple', 'orange', 'watermelon']

Python uses the key to know how to compare two elements and decide which one should go first. That is especially helpful with custom data types, which might contain more than one attribute to sort by:

>>> from collections import namedtuple
>>> Person = namedtuple("Person", "first_name last_name age")
>>> people = [
...     Person("Joe", "Doe", 42),
...     Person("John", "Smith", 27),
...     Person("Anna", "Smith", 31),
... ]
>>> sorted(people, key=lambda x: x.age)
[
    Person(first_name='John', last_name='Smith', age=27),
    Person(first_name='Anna', last_name='Smith', age=31),
    Person(first_name='Joe', last_name='Doe', age=42)
]

You can use attrgetter to conveniently grab that attribute:

>>> from operator import attrgetter
>>> sorted(people, key=attrgetter("age"))
[
    Person(first_name='John', last_name='Smith', age=27),
    Person(first_name='Anna', last_name='Smith', age=31),
    Person(first_name='Joe', last_name='Doe', age=42)
]

It looks like you’re incorrectly using __repr__ when you should be using __str__

@P Moz There’s no one “correct” way to implement str()/repr() in your classes. The reason why Darren has implemented .__repr__() instead of .__str__() is because of how Python lists represent their items by default. In particuar, when you print a Python list to the console, it’ll print each item using the .__repr__() method of each item:

>>> class Person:
...     def __init__(self, name):
...         self.name = name
...     
...     def __str__(self):
...         return self.name
... 
>>> people = [
...     Person("Alice"),
...     Person("Bob"),
...     Person("John"),
... ]
>>> 
>>> people
[
    <__main__.Person object at 0x7f8aecc4f910>,
    <__main__.Person object at 0x7f8aecc4dc10>,
    <__main__.Person object at 0x7f8aecc4f650>
]

Even though the Person class provides a .__str__() method, the people list still prints the default .__repr__() representation of each item.

I see, thanks for the information. I understand what you mean, but just in case anyone is browsing this discussion and wanted to learn a bit more, according to official documentation, __str__ should be used to create the easy-to-read string representation of an object (source), while __repr__ should be used to create an unambiguous or “official” string of the object that’s a valid Python expression and could be used to recreate the object with the same value, if possible (source).