Python in Detail: __dict__

• Python is flexible when working with object attributes. We can access attributes with some_obj.some_attr, or via the hasattr and getattr built-in functions. We can also set attributes directly on the instance with some_obj.some_attr = value, or with setattr.

• This flexibility is powered by dictionaries, which may be a surprise at first! For regular classes like the ones that we've defined so far, attributes are stored in some_obj.__dict__. The dictionary's keys are the attribute names, and its values are the attribute values. Functions like setattr manipulate that dictionary, but we can also manipulate it directly when needed.

• >

  class User:
    def __init__(self, id, admin):
      self.id = id
      self.admin = admin
  
  amir = User(1, False)
  amir.__dict__

  Result:

  Run Code

• >

  class Cat:
    def __init__(self, name):
      self.name = name
  
  keanu = Cat("Keanu")
  keanu.__dict__

  Result:

  {'name': 'Keanu'}

• The .__dict__ dictionary is not merely another way to look at attributes; it's where the attributes are actually stored! Modifying it changes the object's attributes. In the next example, we manipulate .__dict__ to change cat.name.

• >

  class Cat:
    def __init__(self, name):
      self.name = name
  
  keanu = Cat("Keanu")
  keanu.__dict__["name"] = "Ms. Fluff"
  keanu.name

  Result:

  'Ms. Fluff'

• Because .__dict__ is a regular dictionary, all of the usual dictionary operations work. For example, we can list its keys.

• >

  class Cat:
    def __init__(self, name):
      self.name = name
  
  keanu = Cat("Keanu")
  keanu.vaccinated = True
  
  # Note that the instance's two attribute keys will come out in the order
  # they were defined.
  list(keanu.__dict__.keys())

  Result:

  ['name', 'vaccinated']

• We can also create new attributes by adding new keys, like the vaccinated attribute in the next example.

• >

  class Cat:
    def __init__(self, name):
      self.name = name
  
  keanu = Cat("Keanu")
  keanu.__dict__["vaccinated"] = True
  
  keanu.vaccinated

  Result:

  True

• The attribute dictionary stores attributes assigned to the instance. It doesn't include methods, which are associated with the class rather than a specific instance.

• >

  class Orange:
    def calories(self):
      return 75
  
  orange = Orange()
  "calories" in orange.__dict__

  Result:

  False

• Classes themselves are objects, so they also have a .__dict__! That's where the methods live.

• >

  class Cat:
    family = "felidae"
  
    def __init__(self, name):
      self.name = name
  
    def vaccinated(self):
      return False
• Note: this code example reuses elements (variables, etc.) defined in earlier examples.

  >

  "__init__" in Cat.__dict__

  Result:

  True
• Note: this code example reuses elements (variables, etc.) defined in earlier examples.

  >

  "vaccinated" in Cat.__dict__

  Result:

  True
• Note: this code example reuses elements (variables, etc.) defined in earlier examples.

  >

  "run" in Cat.__dict__

  Result:

  False

• The class dictionary also contains class attributes.

• Note: this code example reuses elements (variables, etc.) defined in earlier examples.

  >

  "name" in Cat.__dict__

  Result:

  False
• Note: this code example reuses elements (variables, etc.) defined in earlier examples.

  >

  "family" in Cat.__dict__

  Result:

  True

• We can even call methods directly from .__dict__.

• >

  class Cat:
    def __init__(self, name):
      self.name = name
  
    def vaccinated(self):
      return False
  
  keanu = Cat("Keanu")
  vaccinated_method = Cat.__dict__["vaccinated"]
  vaccinated_method(keanu)

  Result:

  False

• Most values in Python have a .__dict__, but there are exceptions. One notable exception is dictionaries themselves.

• >

  a_dict = {}
  hasattr(a_dict, "__dict__")

  Result:

  False

• Why don't dictionaries have a .__dict__? There are many reasons, but one simple one is performance. For performance reasons, we can't change built-in types like dicts in the same way that we can modify custom classes like Cat. Integers, floats, strings, lists, and tuples also lack a .__dict__ for the same reason.

• >

  hasattr([], "__dict__")

  Result:

  False

• >

  hasattr(5, "__dict__")

  Result:

  False

• Here's a code problem:

  Write a setattr_via_dict function. It tries to set an attribute by modifying the target object's .__dict__. However, not all Python values have a .__dict__.

  If the object has a .__dict__ attribute, set the attribute value and return True. Otherwise return False.

  9

  1

  2

  3

  4

  5

  6

  def setattr_via_dict(obj, attr, value):

  if hasattr(obj, "__dict__"):

  obj.__dict__[attr] = value

  return True

  else:

  return False

  105

  7

  8

  9

  10

  11

  12

  13

  14

  15

  16

  17

  18

  some_dict = {

  "a": 1

  }

  assert setattr_via_dict(some_dict, "some_attr", 2) == False

  assert some_dict["a"] == 1

  
  

  class C:

  pass

  
  

  c = C()

  assert setattr_via_dict(c, "some_attr", 1)

  assert c.some_attr == 1

  Goal:

  None

  Yours:

  None

  Show Author's Answer