Constraints on dictionary keys

Key Uniqueness and Updating Key-Value Pairs

While dictionaries are very flexible and useful tools, there are some things we must keep in mind when working with them. The first is key uniqueness. Simply put, all keys in a dictionary must be unique. You cannot have two identical keys in a dictionary. Let’s take a look at an example for a different class.

class_dict = {'instructors': 2, 'TAs': 10, 'tutors': 35, 'students': 600, 'instructors': 5}

What happens when we inspect the content of this dictionary with a call to print(class_dict)? We get:

{'instructors': 5, 'TAs': 10, 'tutors': 35, 'students': 600}

As you can see, only the second key-value assignment of 'instructors' : 5 remains in the dictionary. Similarly, if we were to run the line class_dict['instructors'] = 7 after, the dictionary would then be:

{'instructors': 7, 'TAs': 10, 'tutors': 35, 'students': 600}

Any value assignment to an existing key will overwrite the existing value associated with said key, not create a new key-value pair.

This process is referred to as updating a key-value pair.

Updating Key-Value Pairs

We can update an existing pair with the same syntax as adding a pair, using dictionary[key] = value. For example, after changing the value associated with key 'instructors', if we want to change the value associated with the key 'TAs' to 5, we can run the line class_dict['TAs'] = 5. Then, calling print(class_dict) would display:

{'instructors': 7, 'TAs': 5, 'tutors': 35, 'students': 600}

Value Uniqueness?

While it is impossible to have duplicate keys, there is no restriction on duplicate values. You may have as many duplicate values as you please. For example,

class_dict = {'instructors': 1, 'TAs': 1, 'tutors': 1, 'students': 1}

is a perfectly valid dictionary (but a pretty weird class).

Key Immutability

Keys need not be strings, but they should be some immutable datatype. This means keys can be integers, strings, or even tuples, but not lists or other dictionaries. Keep in mind though, values can be anything (including lists, dicts, and other mutable datatypes).