Inheritance: pull up duplicated behavior
Write shared behavior once in a superclass, so every subclass gets each change.
Watch the narrated video
Every step
Problem
Steps 1–3Refactor
Steps 4–6Payoff
Steps 7–8Recap
Steps 9–10The code: before and after
Before
class Shipment(ABC):
@abstractmethod
def price(self) -> int: ...
class DomesticShipment(Shipment):
def __init__(self, tracking: str, weight_kg: int):
self.tracking = tracking
self.weight_kg = weight_kg
def price(self) -> int:
return 5 + self.weight_kg
def label(self) -> str:
parts = [self.tracking, f"{self.weight_kg} kg", f"${self.price()}"]
return " | ".join(parts)
class InternationalShipment(Shipment):
def __init__(self, tracking: str, weight_kg: int):
self.tracking = tracking
self.weight_kg = weight_kg
def price(self) -> int:
return 20 + self.weight_kg * 4
def label(self) -> str:
parts = [self.tracking, f"{self.weight_kg} kg", f"${self.price()}"]
return " | ".join(parts)
def main() -> None:
home = DomesticShipment("DM-1", 2)
abroad = InternationalShipment("IN-7", 2)
print(home.label()) # DM-1 | 2 kg | $7
print(abroad.label()) # IN-7 | 2 kg | $28
if __name__ == "__main__":
main()After
class Shipment(ABC):
def __init__(self, tracking: str, weight_kg: int):
self.tracking = tracking
self.weight_kg = weight_kg
self._fragile = False
@abstractmethod
def price(self) -> int: ...
def mark_fragile(self) -> None:
self._fragile = True
def label(self) -> str:
parts = [self.tracking, f"{self.weight_kg} kg", f"${self.price()}"]
if self._fragile:
parts.append("FRAGILE")
return " | ".join(parts)
class DomesticShipment(Shipment):
def price(self) -> int:
return 5 + self.weight_kg
class InternationalShipment(Shipment):
def price(self) -> int:
return 20 + self.weight_kg * 4
def main() -> None:
home = DomesticShipment("DM-1", 2)
abroad = InternationalShipment("IN-7", 2)
abroad.mark_fragile()
print(home.label()) # DM-1 | 2 kg | $7
print(abroad.label()) # IN-7 | 2 kg | $28 | FRAGILE
if __name__ == "__main__":
main()Transcript, step by step
Problem
Step 1: A courier's two shipments Before
A parcel courier's dispatch desk works with two kinds of shipment. Shipment is the superclass, the class the others extend, and each subclass, a class that extends it, says what its parcel costs. Both subclasses also set a tracking code and a weight in their own init, and both build a label from them. Main prints one label for a domestic parcel and one for an international one.
Step 2: Fragile parcels Code smell
The courier now needs the word fragile on the label of fragile parcels. The quick way adds a fragile attribute, a mark fragile method and a label check to Domestic Shipment, and then the same three again to International Shipment. Main marks the parcel going abroad, so only the second label ends in fragile.
Step 3: Two copies of everything Code smell
Duplicated code across subclasses is when two classes that share a superclass each write the same fields and methods on their own. You can see the fields here and here, and the label here and here. So every change to the label is made twice, and the day someone edits only one copy, a fragile parcel ships without the word on it. A third shipment type would copy all of it a third time.
Refactor
Step 4: Refactor first Back it out
Let's take the fragile flag out again. We'll refactor first, so that the shared parts move up into Shipment, and then we'll add the flag in one place.
Step 5: One __init__ Pull Up Constructor Body
In Python a field comes into being when init assigns it, so the fields and the constructor body move up together. Shipment's init now sets the tracking code and the weight, once. The two subclass constructors only repeated those lines, so both are deleted, and each subclass inherits the init from Shipment. Main still builds both parcels with a tracking code and a weight, and prints the same two labels.
Step 6: One label Pull Up Method
Now the two label bodies are identical, so one copy moves up into Shipment, and both subclass copies are deleted. Label still works because it calls price, and each subclass still answers price its own way. This is inheritance, where a subclass gets the fields and methods of its superclass, so the shared parts are written once, in Shipment. What's left in each subclass is its price. A superclass method that calls a step each subclass fills in, like label calling price, is the Template Method pattern.
Payoff
Step 7: Fragile, once Add the flag
Now the fragile flag comes back, and this time it goes into Shipment only. Shipment gains one fragile attribute and one mark fragile method, and label adds the word once. Domestic and International Shipment didn't change, and main only gains one call that marks the parcel going abroad.
Step 8: Shared code lives once Payoff
The fields now exist once, here, and the label exists once, here. The fragile rule is written once too, so the two labels can't drift apart, and the run prints the word only on the parcel going abroad. It fits because both classes are shipments, each with a tracking code, a weight and a label, and only the price differs. Abstraction hid how a class stores its data, and inheritance shares what two classes both do.
Recap
Step 9: Where we started, where we ended Before / after
On the left, each subclass keeps its own init and its own label. On the right, Shipment holds the init, the fragile flag and the label, each written once. A subclass like Domestic Shipment keeps only its own price.
Step 10: When to reach for it Takeaways
Reach for it when two or more subclasses of one superclass repeat the same fields or the same method body. Skip it when two classes look alike today but will change for different reasons, because a superclass ties them together, so use Extract Class and delegate to it. You'll spot the need by the same method copied into sibling subclasses, and the same init lines in each. Python's Mapping base class works this way, because a subclass writes three methods, get item, len and iter, and inherits get, keys and items.