Python Evaluates Assignments From Right To Left

Charlton Asino — Wed, 27 Sep 2023 14:00:46 +0000

Why does it matter?
Assigning variables in Python may range from simple statements such as count=5 to others that may require evaluation such as str_len=len("eat one piece of the fruit and you drown")

When assigning variables in Python the evaluation
is done from right to left.Meaning the length of the string str_len will be evaluated first and then assigned.

Tuples are immutable
When working with more complex objects the evaluation order may affect the behaviour of your objects.An example being tuples.Tuples are immutable in Python in that they cannot be modified once created.

>>> a=(1, [2,3,4])
>>> type(a)
<class 'tuple'>
>>> a[0]=8
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'tuple' object does not support item assignment
>>> a[1]=[5,6,7,8]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'tuple' object does not support item assignment
>>> a
(1, [2, 3, 4])

Modifying tuple a raises an exception and inspecting it after shows that the tuple has not been modified.
Murky waters
Depending on how a tuple is operated on and its composition, you can get some unexpected behaviour(they become mutable). This can occur when using augmented assignment operators in particular __iadd__ (in-place addition) which is commonly used as +=.

>>> a
(1, [2, 3, 4])
>>> a[1]+=[100,101,700]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'tuple' object does not support item assignment

An attempt to modify the tuple raises an exception which conforms to the expected behaviour that tuples are immutable.

>>> a
(1, [2, 3, 4])
>>> a[1]+=[100,101,700]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'tuple' object does not support item assignment
>>> a
(1, [2, 3, 4, 100, 101, 700])

But on inspection as per the above example.The tuple has been modified ,the list in the tuple in particular even though we got an exception.

This occurred due to two aspects of the code:

In-place addition which is supported in lists.
Evaluation of assignments from right to left.

a[1] is a list [2, 3, 4] which supports inplace addition .The statement a[1]+=[100,101,700] is evaluated from right to left meaning the list at a[1] is modified first then a assigned into the tuple and this raises an exception.

Take away
Understading how python evaluates statements is important, especially when performing operations on objects which contain types with different behaviours.

Python Objects As Dictionary Keys

Charlton Asino — Thu, 07 Sep 2023 13:34:20 +0000

When can it happen?
You have a custom class(object) and you want to store related metadata in memory . You decide to use Python dict and leverage the faster look ups O(1) .
In the case of a user object:

Python dicts ,eq and the hash function
For the user object , tracking the number of clicks made the user is required for later computations.The object will be the key for our dict and the number of clicks will be the value .In this case the user will be instatiated more than once.



>>> u1=User(1,"Jill")
>>> clicks=dict()
>>> clicks[u1]=1
>>> clicks
{<User(1 Jill)>: 1}
>>> u2=User(2,"Lock")
>>> clicks[u2]=2
>>> clicks
{<User(1 Jill)>: 1, <User(2 Lock)>: 2}
>>> u3=User(1,"Jill")
>>> clicks[u3]=3
>>> clicks
{<User(1 Jill)>: 1, <User(2 Lock)>: 2, <User(1 Jill)>: 3}

Both u1 andu3 are representations of the same user ,Jill with the same user_id 1 but when updating the value for u3a new key is created.

This happens because Python uses the object id to generate a hash which is used as a key in the dict.Both u1and u3 have different object ids hence will have different dictionary keys despite representing the same user.



>>> id(u1)
140030203095248
>>> id(u3)
140030203095440
>>>

<User(1 Jill)>value needs to be updated for this case since it already exists in the dictionary .To a achieve this a __hash__ and__eq__ dunder methods should be implented for the user object where if the objects are equal they will have the same hash.

Creating a dict with the user updated object gives the required output.



>>> u1=User(1,"Jill")
>>> clicks=dict()
>>> clicks[u1]=1
>>> clicks
{<User(1,Jill)>: 1}
>>> u2=User(2,"Lock")
>>> clicks[u2]=1
>>> clicks
{<User(1,Jill)>: 1, <User(2,Lock)>: 1}
>>> u3=User(1,"Jill")
>>> clicks[u3]=3
>>> clicks
{<User(1,Jill)>: 3, <User(2,Lock)>: 1}
>>>

Upating theclicksdict with the object u3 which is represents the user Jill with user_id 1 ,updates the existing object in the dict u1 which is respresents the same user.

DEV Community: Charlton Asino

Python Evaluates Assignments From Right To Left

Python Objects As Dictionary Keys