Skip to content

Latest commit

 

History

History
235 lines (181 loc) · 5.51 KB

ex4_3.md

File metadata and controls

235 lines (181 loc) · 5.51 KB

[ Index | Exercise 4.2 | Exercise 4.4 ]

Exercise 4.3

Objectives:

  • Learn about descriptors

Files Created: descrip.py

Files Modified: validate.py

(a) Descriptors in action

Earlier, you created a class Stock that made use of slots, properties, and other features. All of these features are implemented using the descriptor protocol. See it in action by trying this simple experiment.

First, create a stock object, and try looking up a few attributes:

>>> s = Stock('GOOG', 100, 490.10)
>>> s.name
'GOOG'
>>> s.shares
100
>>>

Now, notice that these attributes are in the class dictionary.

>>> Stock.__dict__.keys()
['sell', '__module__', '__weakref__', 'price', '_price', 'shares', '_shares',
'__slots__', 'cost', '__repr__', '__doc__', '__init__']
>>> 

Try these steps which illustrate how descriptors get and set values on an instance:

>>> q = Stock.__dict__['shares']
>>> q.__get__(s)
100
>>> q.__set__(s,75)
>>> s.shares
75
>>> q.__set__(s, '75')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "stock.py", line 23, in shares
    raise TypeError('Expected an integer')
TypeError: Expected an integer
>>>

The execution of __get__() and __set__() occurs automatically whenever you access instances.

(b) Make your own descriptor

Define the descriptor class from the notes:

# descrip.py

class Descriptor:
    def __init__(self, name):
        self.name = name
    def __get__(self, instance, cls):
        print('%s:__get__' % self.name)
    def __set__(self, instance, value):
        print('%s:__set__ %s' % (self.name, value))
    def __delete__(self, instance):
        print('%s:__delete__' % self.name)

Now, try defining a simple class that uses this descriptor:

>>> class Foo:
        a = Descriptor('a')
        b = Descriptor('b')
        c = Descriptor('c')

>>> f = Foo()
>>> f
<__main__.Foo object at 0x38e130> <class __main__.Foo>
>>> f.a
a:__get__
>>> f.b
b:__get__
>>> f.a = 23
a:__set__ 23
>>> del f.a
a:__delete__
>>>

Ponder the fact that you have captured the dot-operator for a specific attribute.

(c) From Validators to Descriptors

In the previous exercise, you wrote a series of classes that could perform checking. For example:

>>> PositiveInteger.check(10)
10
>>> PositiveInteger.check('10')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
    raise TypeError('Expected %s' % cls.expected_type)
TypeError: expected <class 'int'>
>>> PositiveInteger.check(-10)

You can extend this to descriptors by making a simple change to the Validator base class. Change it to the following:

# validate.py

class Validator:
    def __init__(self, name):
        self.name = name

    @classmethod
    def check(cls, value):
        return value

    def __set__(self, instance,	value):
        instance.__dict__[self.name] = self.check(value)

Note: The lack of the __get__() method in the descriptor means that Python will use its default implementation of attribute lookup. This requires that the supplied name matches the name used in the instance dictionary.

No other changes should be necessary. Now, try modifying the Stock class to use the validators as descriptors like this:

class Stock:
    name   = String('name')
    shares = PositiveInteger('shares')
    price  = PositiveFloat('price')

    def __init__(self,name,shares,price):
        self.name = name
        self.shares = shares
        self.price = price

You'll find that your class works the same way as before, involves much less code, and gives you all of the desired checking:

>>> s = Stock('GOOG', 100, 490.10)
>>> s.name
'GOOG'
>>> s.shares
100
>>> s.shares = 75
>>> s.shares = '75'
... TypeError ...
>>> s.shares = -50
... ValueError ...
>>>

This is pretty cool. Descriptors have allowed you to greatly simplify the implementation of the Stock class. This is the real power of descriptors--you get low level control over the dot and can use it to do amazing things.

(d) Fixing the Names

One annoying thing about descriptors is the redundant name specification. For example:

class Stock:
    ...
    shares = PositiveInteger('shares')
    ...

We can fix that. Change the top-level Validator class to include a __set_name__() method like this:

# validate.py

class Validator:
    def __init__(self, name=None):
        self.name = name

    def __set_name__(self, cls, name):
        self.name = name

    @classmethod
    def check(cls, value):
        return value

    def __set__(self, instance,	value):
        instance.__dict__[self.name] = self.check(value)

Now, try rewriting your Stock class so that it looks like this:

class Stock:
    name   = String()
    shares = PositiveInteger()
    price  = PositiveFloat()

    def __init__(self,name,shares,price):
        self.name = name
        self.shares = shares
        self.price = price

Ah, much nicer. Be aware that this ability to set the name is a Python 3.6 feature however. It won't work on older versions.

[ Solution | Index | Exercise 4.2 | Exercise 4.4 ]


>>> Advanced Python Mastery
... A course by dabeaz
... Copyright 2007-2023

. This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License