Let's continue our exploration of Python's magic methods in this second part of the series. This part will focus on numbers and containers, i.e., collections. You can read the first part here.
Container-related methods
Python provides the usual containers, e.g., lists, sets, and dictionaries. You can use the following methods when you want to implement your own.
Common methods
Containers have a size. Python defines two methods to implement to return the number of items in a container: object.__len__(self) for the exact size and object.__length_hint__(self) for an approximation. You should use the latter when getting the exact size is computationally expensive.
Item-related methods
Containers contain objects. Some containers offer index-based access, e.g., list(1), while others offer key-based access, e.g., dict('mykey'). In both cases, here are the methods to implement:
| Method | Functionality |
|---|---|
object.__getitem__(self, key) |
Get the object |
object.__setitem__(self, key, value) |
Set the object |
object.__delitem__(self, key) |
Remove the object |
object.__missing__(self, key) |
Called when the key is not found by the default get(key) implementation |
object.__iter__(self) |
Return an iterator over items (or keys) in the container |
object.__reversed__(self) |
Reverse the objects in the container |
object.__contains__(self, item) |
Check whether an item is part of the container |
Let's create a simple hash-map-like container for illustration purposes:
class Container:
def __init__(self):
self.items = {}
def __getattribute__(self, name):
raise AttributeError()
def __len__(self):
return len(self.items) #1
def __setitem__(self, key, value):
self.items[key] = value #1
def __getitem__(self, key):
return self.items[key] #1
def __delitem__(self, key):
return self.items.pop(key) #1
def __contains__(self, key):
return key in self.items #2
def __iter__(self):
return iter(self.items.keys()) #3
def __reversed__(self):
return iter(reversed(self.items.keys())) #4
container = Container()
container['foo'] = 'foo'
container['bar'] = 'bar'
print(len(container)) #5
for x in container: #6
print(f'{x}: {container[x]}')
print('---')
for x in reversed(container): #7
print(f'{x}: {container[x]}')
print('---')
del container['foo']
for x in container: #8
print(f'{x}: {container[x]}')
print('---')
print('foo' in container) #9
- Delegate on the
itemsdictionary - Check if the key belongs to
items - Get the keys' iterator
- Get the reversed key's iterator
- Print 2 as the container has two items at this point
- Implicitly calls the
__iter__()method - Implicitly calls the
__reversed__()method - Print
bar: barsince thefookey has been deleted - Implicitly calls the
__contains__()method
Number-related methods
Just as we can emulate containers, we can emulate numbers as well.
Arithmetic methods
Arithmetic methods abound; it's easier to summarize them in a table:
| Kind | Method | Operator/function | Comment |
|---|---|---|---|
| All | |||
object.__add__(self, other)
|
+
|
||
object.__sub__(self, other)
|
-
|
||
object.__mul__(self, other)
|
*
|
||
object.__matmul__(self, other)
|
@
|
Matrix multiplication | |
object.__truediv__(self, other)
|
/
|
Regular division | |
object.__floordiv__(self, other)
|
//
|
Division without the reminder | |
object.__mod__(self, other)
|
%
|
Reminder of the division | |
object.__divmod__(self, other)
|
divmod()
|
||
object.__pow__(self, other[, modulo])
|
pow()
|
||
object.__lshift__(self, other)
|
<<
|
||
object.__rshift__(self, other)
|
>>
|
||
object.__and__(self, other)
|
&
|
||
object.__xor__(self, other)
|
^
|
Exclusive OR
|
|
object.__or__(self, other)
|
|
|
Inclusive OR
|
|
| Binary | |||
object.__radd__(self, other)
|
+
|
||
object.__rsub__(self, other)
|
-
|
||
object.__rmul__(self, other)
|
*
|
||
object.__rmatmul__(self, other)
|
@
|
||
object.__rtruediv__(self, other)
|
/
|
||
object.__rfloordiv__(self, other)
|
//
|
||
object.__rmod__(self, other)
|
%
|
||
object.__rdivmod__(self, other)
|
divmod()
|
||
object.__rpow__(self, other[, modulo])
|
pow()
|
||
object.__rlshift__(self, other)
|
<<
|
||
object.__rrshift__(self, other)
|
>>
|
||
object.__rand__(self, other)
|
&
|
||
object.__rxor__(self, other)
|
^
|
||
object.__ror__(self, other)
|
|
|
||
| Assignement | |||
object.__iadd__(self, other)
|
+=
|
||
object.__isub__(self, other)
|
-=
|
||
object.__imul__(self, other)
|
*=
|
||
object.__imatmul__(self, other)
|
@=
|
||
object.__itruediv__(self, other)
|
/=
|
||
object.__ifloordiv__(self, other)
|
//=
|
||
object.__imod__(self, other)
|
%=
|
||
object.__ipow__(self, other[, modulo])
|
pow()=
|
||
object.__ilshift__(self, other)
|
<<=
|
||
object.__irshift__(self, other)
|
>>=
|
||
object.__iand__(self, other)
|
&=
|
||
object.__ixor__(self, other)
|
^=
|
||
object.__ior__(self, other)
|
|=
|
||
| Unary | |||
object.__neg__(self)
|
-
|
||
object.__pos__(self)
|
+
|
||
object.__abs__(self)
|
abs()
|
Absolute value | |
object.__invert__(self)
|
~
|
Bitwise NOT
|
Imagine an e-commerce site with products and stocks of them dispatched in warehouses. We need to subtract stock levels when someone orders and add stock levels when the stock is replenished. Let's implement the latter with some of the methods we've seen so far:
class Warehouse: #1
def __init__(self, id):
self.id = id
def __eq__(self, other): #2
if not isinstance(other, Warehouse):
return False
return self.id == other.id
def __repr__(self): #3
return f'Warehouse(id={self.id})'
class Product: #1
def __init__(self, id):
self.id = id
def __eq__(self, other): #2
if not isinstance(other, Product):
return False
return self.id == other.id
def __repr__(self): #3
return f'Product(id={self.id})'
class StockLevel:
def __init__(self, product, warehouse, quantity):
self.product = product
self.warehouse = warehouse
self.quantity = quantity
def __add__(self, other): #4
if not isinstance(other, StockLevel):
raise Exception(f'{other} is not a StockLevel')
if self.warehouse != other.warehouse:
raise Exception(f'Warehouse are not the same {other.warehouse}')
if self.product != other.product:
raise Exception(f'Product are not the same {other.product}')
return StockLevel(self.product, self.warehouse,\
self.quantity + other.quantity) #5
def __repr__(self):
return f'StockLevel(warehouse={self.warehouse},\
product={self.product},quantity={self.quantity})'
warehouse1 = Warehouse(1)
warehouse2 = Warehouse(2)
product = Product(1) #6
product1 = Product(1) #6
stocklevel111 = StockLevel(product, warehouse1, 1) #7
stocklevel112 = StockLevel(product, warehouse1, 2) #7
stocklevel121 = StockLevel(product1, warehouse2, 1) #7
print(stocklevel111 + stocklevel112) #8
stocklevel111 + stocklevel121 #9
- Define necessary classes
- Override equality to compare ids
- Override representation
- Implement addition. If the warehouse and product don't match, raise an exception.
- Create a new
StockLevelwith the same product and warehouse and the quantity as the sum of both quantities - Define two products that point to the same id; it's the same product for equality purposes
- Create new stock-level objects
- Print
StockLevel(warehouse=Warehouse(id=1),product=Product(id=1),quantity=3) - Raise an exception as warehouses are different, though products are the same
Conversion methods
Conversion methods allow changing an instance to a numeric type, i.e., int, float, or complex.
| Method | Built-in function |
|---|---|
object.__complex__(self) |
complex() |
object.__int__(self) |
int() |
object.__float__(self) |
float() |
If no such method is implemented, Python falls back to the object.__index__(self), for example, when using the instance as an index.
The following sample, however irrelevant it is, highlights the above:
class Foo:
def __init__(self, id):
self.id = id
def __index__(self): #1
return self.id
foo = Foo(1)
array = ['a', 'b', 'c']
what = array[foo] #2
print(what) #3
- Define the fallback method
- Coerce
foointo anint. We didn't implement any conversion method; Python falls back toindex() - Print
b
Other methods
Finally, Python delegates to a magic method when your code calls a specific number-related function.
| Method | Built-in function |
|---|---|
object.__round__(self[, ndigits]) |
round() |
object.__trunc__(self) |
trunc() |
object.__floor__(self) |
floor() |
object.__ceil__(self) |
ceil() |
Context managers' methods
Python's context managers allow fine-grained control over resources that must be acquired and released. It works with the with keyword. For example, here's how you open a file to write to:
with open('file', 'w') as f: #1
f.write('Hello world!')
#2
- Open the file
- At this point, Python has closed the file
A context manager is syntactic sugar. The following code is equivalent to the one from above:
f = open('file', 'w')
try:
f.write('Hello world!')
finally:
f.close()
To write your context manager requires to implement two methods: one for opening the context and one for closing it, respectively, object.__enter__(self) and object.__exit__(self, exc_type, exc_value, traceback).
Let's write a context manager to manage a pseudo-connection.
import traceback
class Connection:
def __enter__(self):
self.connection = Connection()
return self.connection
def __exit__(self, exc_type, exc_value, exc_traceback):
self.connection = None
if exc_type is not None:
print('An exception happened')
print(traceback.format_exception(exc_type, exc_value, exc_traceback))
return True
def do_something(self):
pass
with Connection() as connection:
connection.do_something()
Callable objects
I was first exposed to callable objects in Kotlin. A callable object looks like a function but is an object:
hello = Hello()
hello('world')
The method to implement to make the above code run is object.__call__(self[, args...]).
class Hello:
def __call__(self, who):
print(f'Hello {who}!')
Conclusion
The post concludes our 2-part series on Python "magic" methods. I didn't mention some of them, though, as they are so many. However, they cover the majority of them.
Happy Python!
To go further:
Originally published at A Java Geek on October 22nd, 2023
Top comments (0)