Python Data vs. Non-data Descriptors
Summary: in this tutorial, you’ll learn the differences between data and non-data descriptors.
Descriptors have two types:
- Data descriptors are objects of a class that implements
__set__method (and/or__delete__method) - Non-data descriptors are objects of a class that have the
__get__method only.
Both descriptor types can optionally implement the __set_name__ method. The __set_name__ method doesn’t affect the classification of the descriptors.
The descriptor types determine how Python resolves object’s attributes lookup.
Non-data descriptor
If a class uses a non-data descriptor, Python will search the attribute in instance attributes first (instance.__dict__). If Python doesn’t find the attribute in the instance attributes, it’ll use the data descriptor.
Let’s take a look at the following example.
First, define a non-data descriptor class FileCount that has the __get__ method which returns the number of files in a folder:
class FileCount:
def __get__(self, instance, owner):
print('The __get__ was called')
return len(os.listdir(instance.path))Code language: Python (python)
Second, define a Folder class that uses the FileCount descriptor:
class Folder:
count = FileCount() def __init__(self, path):
self.path = path
Code language: Python (python)
Third, create an instance of the Folder class and access the count attribute:
folder = Folder('/')
print('file count: ', folder.count)Code language: Python (python)
Python called the __get__ descriptor:
The __get__ was called
file count: 32Code language: Python (python)
After that, set the count attribute of the folder instance to 100 and access the count attribute:
folder.__dict__['count'] = 100
print('file count: ', folder.count)Code language: Python (python)
Output:
file count: 100Code language: Python (python)
In this example, Python can find the count attribute in the instance dictionary __dict__. Therefore, it does not use data descriptors.
Data descriptor
When a class has a data descriptor, Python will look for an instance’s attribute in the data descriptor first. If Python doesn’t find the attribute, it’ll look for the attribute in the instance dictionary (__dict__). For example:
First, define a Coordinate descriptor class:
class Coordinate:
def __get__(self, instance, owner):
print('The __get__ was called') def __set__(self, instance, value):
print('The __set__ was called')
Code language: Python (python)
Second, define a Point class that uses the Coordinate descriptor:
class Point:
x = Coordinate()
y = Coordinate()Code language: Python (python)
Third, create a new instance of the Point class and assign a value to the x attribute of the p instance:
p = Point()
p.x = 10Code language: Python (python)
Output:
The __set__ was calledCode language: Python (python)
Python called the __set__ method of the x descriptor.
Finally, access the x attribute of the p instance:
p.xCode language: Python (python)
Output:
The __get__ was calledCode language: Python (python)
Python called the __get__ method of the x descriptor.
Summary
- Data descriptors are objects of a class that implements
__set__method (and/or__delete__method) - Non-data descriptors are objects of a class that have the
__get__method only. - When accessing object’s attributes, data descriptors override the instance’s attributes and instance’s attributes override non-data descriptors.