学ぶ Runtime Checkable Protocols

メニューを表示するにはスワイプしてください

By default, a Protocol is a static typing construct – it exists for type checkers like mypy but has no runtime presence. @runtime_checkable changes this, enabling isinstance() checks at runtime. Understanding exactly what this check does and does not verify is essential for using it safely.

What `@runtime_checkable` Enables

Without the decorator, calling isinstance() on a Protocol raises TypeError. With it, the check passes if the object has all the required methods as attributes:


              12345678910111213141516171819202122232425
            
from typing import Protocol, runtime_checkable

@runtime_checkable
class Cacheable(Protocol):
    def cache_key(self) -> str:
        ...

    def invalidate(self) -> None:
        ...

class UserSession:
    def __init__(self, session_id):
        self.session_id = session_id

    def cache_key(self):
        return f"session:{self.session_id}"

    def invalidate(self):
        print(f"Invalidating {self.session_id}")

class EmptyClass:
    pass

print(isinstance(UserSession("s-001"), Cacheable))  # True – both methods present
print(isinstance(EmptyClass(), Cacheable))           # False – methods missing

The Limitation: Signatures Are Not Checked

@runtime_checkable only checks that the names exist as attributes – it does not verify argument counts, return types, or whether the attribute is actually callable:


              123456789101112131415161718
            
from typing import Protocol, runtime_checkable

@runtime_checkable
class Processable(Protocol):
    def process(self, data: list) -> dict:
        ...

# Wrong signature – but isinstance() still passes
class BrokenProcessor:
    def process(self):  # Missing the data argument
        return "not a dict"

# Not even callable – but isinstance() still passes
class FakeProcessor:
    process = "not a function"  # A string, not a method

print(isinstance(BrokenProcessor(), Processable))  # True – name exists
print(isinstance(FakeProcessor(), Processable))    # True – name exists

This is why @runtime_checkable is a structural check by name, not a full contract verification. Use it for quick guards, not for guaranteeing correct behavior.

Protocol Attributes

Protocols can also require data attributes, not just methods. @runtime_checkable checks for their presence the same way:


              1234567891011121314151617181920
            
from typing import Protocol, runtime_checkable

@runtime_checkable
class HasMetadata(Protocol):
    version: str
    author: str

class ReportTemplate:
    version = "2.0"
    author = "analytics-team"

    def render(self):
        return f"Report v{self.version} by {self.author}"

class IncompleteTemplate:
    version = "1.0"
    # Missing author

print(isinstance(ReportTemplate(), HasMetadata))     # True – both attributes present
print(isinstance(IncompleteTemplate(), HasMetadata)) # False – author missing

Protocol Inheritance

Protocols can inherit from other Protocols to compose larger interfaces:


              1234567891011121314151617181920212223242526
            
from typing import Protocol, runtime_checkable

@runtime_checkable
class Readable(Protocol):
    def read(self) -> str:
        ...

@runtime_checkable
class Writable(Protocol):
    def write(self, data: str) -> None:
        ...

@runtime_checkable
class ReadWritable(Readable, Writable, Protocol):
    pass  # Combines both interfaces

class FileBuffer:
    def read(self):
        return "buffered content"

    def write(self, data):
        print(f"Writing: {data}")

print(isinstance(FileBuffer(), Readable))     # True
print(isinstance(FileBuffer(), Writable))     # True
print(isinstance(FileBuffer(), ReadWritable)) # True – satisfies both

When to Use `@runtime_checkable`

Use @runtime_checkable when:

You need a lightweight guard at a function entry point to give a clear error message;
You are working with dynamically loaded plugins and need to verify basic interface presence;
You want to filter a list of objects by capability.

Do not rely on it as a complete correctness check – call the methods and let AttributeError or TypeError surface any mismatches naturally.


              12345678910111213
            
from typing import Protocol, runtime_checkable

@runtime_checkable
class Exporter(Protocol):
    def export(self, data: list) -> str:
        ...

def run_export(exporter, data):
    if not isinstance(exporter, Exporter):
        raise TypeError(
            f"Expected an Exporter, got {type(exporter).__name__}"
        )
    return exporter.export(data)

すべて明確でしたか？

フィードバックありがとうございます！

セクション 3. 章 2

AIに質問する

何でも質問するか、提案された質問の1つを試してチャットを始めてください

セクション 3. 章 2

Runtime Checkable Protocols

What @runtime_checkable Enables

The Limitation: Signatures Are Not Checked

Protocol Attributes

Protocol Inheritance

When to Use @runtime_checkable

What `@runtime_checkable` Enables

When to Use `@runtime_checkable`