初始化提交

backend/app/ai_provider/core/resilience.py

@@ -0,0 +1,133 @@
+from __future__ import annotations
+
+import time
+from enum import Enum
+
+
+class CircuitBreakerState(Enum):
+    CLOSED = "closed"
+    OPEN = "open"
+    HALF_OPEN = "half_open"
+
+
+class CircuitBreaker:
+    """标准断路器，冷却后自动半开。"""
+
+    def __init__(
+        self,
+        name: str,
+        failure_threshold: int = 5,
+        cooldown_seconds: float = 30.0,
+    ) -> None:
+        self.name = name
+        self.failure_threshold = failure_threshold
+        self.cooldown_seconds = cooldown_seconds
+        self._state = CircuitBreakerState.CLOSED
+        self._failure_count = 0
+        self._last_failure_time: float = 0.0
+
+    @property
+    def state(self) -> CircuitBreakerState:
+        if (
+            self._state == CircuitBreakerState.OPEN
+            and time.monotonic() - self._last_failure_time >= self.cooldown_seconds
+        ):
+            self._state = CircuitBreakerState.HALF_OPEN
+        return self._state
+
+    def record_success(self) -> None:
+        self._failure_count = 0
+        self._state = CircuitBreakerState.CLOSED
+
+    def record_failure(self) -> None:
+        self._failure_count += 1
+        self._last_failure_time = time.monotonic()
+        if self._failure_count >= self.failure_threshold:
+            self._state = CircuitBreakerState.OPEN
+
+    def allow_request(self) -> bool:
+        return self.state != CircuitBreakerState.OPEN
+
+
+class RetryStrategy:
+    """指数退避重试。"""
+
+    def __init__(
+        self,
+        max_retries: int = 3,
+        base_delay: float = 1.0,
+        max_delay: float = 60.0,
+        multiplier: float = 2.0,
+    ) -> None:
+        self.max_retries = max_retries
+        self.base_delay = base_delay
+        self.max_delay = max_delay
+        self.multiplier = multiplier
+
+    def should_retry(self, attempt_index: int) -> bool:
+        return attempt_index < self.max_retries
+
+    def get_delay(self, attempt_index: int) -> float:
+        delay = self.base_delay * (self.multiplier**attempt_index)
+        return min(delay, self.max_delay)
+
+
+class DowngradeAction:
+    """降级链中的一个步骤。"""
+
+    def __init__(self, type: str, target: str) -> None:  # noqa: A002
+        self.type = type  # "fallback_endpoint", "fallback_model", "cache", "degrade_quality"
+        self.target = target
+
+
+class DowngradeChain:
+    """当模型/端点失败时的降级动作链。"""
+
+    def __init__(self, actions: list[DowngradeAction]) -> None:
+        self._actions = actions
+        self._index = 0
+
+    def next_action(self) -> tuple[str, str] | None:
+        if self._index >= len(self._actions):
+            return None
+        action = self._actions[self._index]
+        return (action.type, action.target)
+
+    def advance(self) -> None:
+        self._index += 1
+
+    def is_exhausted(self) -> bool:
+        return self._index >= len(self._actions)
+
+    def reset(self) -> None:
+        self._index = 0
+
+
+class ResilienceFactory:
+    """弹性组件工厂。"""
+
+    @staticmethod
+    def create_circuit_breaker(
+        name: str,
+        failure_threshold: int = 5,
+        cooldown_seconds: float = 30.0,
+    ) -> CircuitBreaker:
+        return CircuitBreaker(
+            name=name,
+            failure_threshold=failure_threshold,
+            cooldown_seconds=cooldown_seconds,
+        )
+
+    @staticmethod
+    def create_retry_strategy(
+        max_retries: int = 3,
+        base_delay: float = 1.0,
+        max_delay: float = 60.0,
+        multiplier: float = 2.0,
+    ) -> RetryStrategy:
+        return RetryStrategy(
+            max_retries=max_retries,
+            base_delay=base_delay,
+            max_delay=max_delay,
+            multiplier=multiplier,
+        )