apply-adapter-pattern

Apply Adapter Pattern

Convert the interface of a class into the interface clients expect, enabling classes with incompatible interfaces to work together.

Why This Is Best Practice

Adopted by: Java's InputStreamReader (adapts byte-stream InputStream to character-stream Reader — in every JVM), Python's io.TextIOWrapper (same adaptation in the standard library), Django REST Framework's Serializer (adapts ORM models to JSON-serializable dicts), and every database driver layer (JDBC, ODBC) which adapts vendor-specific protocols to a standard interface. Impact: GoF documents that the Adapter is the primary mechanism for integrating legacy or third-party code into a new architecture without modifying either side. Every major language's standard library uses it to bridge abstraction layers — evidence of universal adoption across all major ecosystems. Why best: The alternative — modifying the existing class — creates coupling to the consumer and may be impossible (third-party code, compiled libraries). The alternative of modifying the client breaks encapsulation. Adapter wraps without touching either side.

Sources: Gamma et al. (1994) pp. 139–150; Java I/O class hierarchy documentation; Python io module documentation

Steps

Step 1: Identify the target interface the client expects

from abc import ABC, abstractmethod

class JSONLogger(ABC):
    @abstractmethod
    def log(self, event: dict) -> None: ...

Step 2: Identify the adaptee (existing class with incompatible interface)

class LegacyLogger:
    def write_line(self, timestamp: str, level: str, message: str) -> None:
        print(f"[{timestamp}] {level}: {message}")

Step 3: Write the adapter — wraps adaptee, implements target interface

from datetime import datetime

class LegacyLoggerAdapter(JSONLogger):
    def __init__(self, legacy: LegacyLogger):
        self._legacy = legacy

    def log(self, event: dict) -> None:
        self._legacy.write_line(
            timestamp=event.get("ts", datetime.now().isoformat()),
            level=event.get("level", "INFO"),
            message=event.get("message", ""),
        )

Step 4: Wire the adapter at the composition root

legacy = LegacyLogger()
logger: JSONLogger = LegacyLoggerAdapter(legacy)

# Client uses only the target interface — unaware of LegacyLogger
logger.log({"level": "ERROR", "message": "payment failed", "ts": "2026-06-10T12:00:00"})

Step 5: For two-way adaptation, create bidirectional adapters

If both sides need to call each other, write two adapters — one for each direction. A single class implementing both interfaces creates hidden coupling and is hard to test.

When NOT to Use

• When you control both sides — if you own both the client and the class being used, change one interface to match the other directly. Adapter is for cases where modification is impossible or undesirable.
• When the interfaces differ too fundamentally — if the adaptee's semantics are incompatible (not just syntactically different), an adapter produces a leaky abstraction. Consider a apply-facade-pattern instead.

Common Mistakes

Adapting too much. An adapter that translates complex business logic is a Facade or a service, not an adapter. Keep adapters thin — only interface translation, no logic.

Modifying the adaptee to make the adapter simpler. If you can modify the adaptee, there may be no need for an adapter at all. If you can't, don't.

Forgetting to handle adapter errors. When the adaptee throws an exception using its own error types, the adapter should either translate them to the target interface's error types or let them propagate with clear context.