浏览器重启逻辑

app/core/config.py

@@ -53,6 +53,8 @@ class Settings(BaseSettings):
     browser_channel: str | None = "chrome"
     browser_pool_size: int = 3
     max_site_concurrency: int = 5  # 最大并发页面数
+    browser_recycle_after_contexts: int = 200  # 浏览器累计创建上下文达到该值后主动回收，0 关闭
+    browser_recycle_after_seconds: float = 3600.0  # 浏览器运行超过该秒数后主动回收，0 关闭
     page_acquire_timeout_seconds: float = 60.0  # 获取页面槽位超时
     request_timeout_seconds: float = 120.0  # 页面请求超时
     cloudflare_wait_timeout_seconds: float = 60.0  # 等待 Cloudflare 挑战超时

app/services/browser_pool.py

@@ -32,6 +32,10 @@ class BrowserPool:
         self._startup_error: str | None = None
         self._retained_contexts: list[Any] = []
         self._retained_pages: list[Any] = []
+        # 主动回收相关计数：累计创建的上下文数、本次浏览器启动时刻、当前在用上下文数
+        self._contexts_since_start = 0
+        self._browser_started_at: float | None = None
+        self._active_contexts = 0
 
     @property
     def enabled(self) -> bool:
@@ -95,6 +99,8 @@ class BrowserPool:
             )
             self._started = True
             self._startup_error = None
+            self._contexts_since_start = 0
+            self._browser_started_at = asyncio.get_running_loop().time()
             logger.info("浏览器启动完成")
         except Exception as exc:
             self._playwright = None
@@ -163,6 +169,51 @@ class BrowserPool:
 
         await self._restart_browser("browser is disconnected before acquiring page session")
 
+    def _recycle_due(self) -> bool:
+        """判断浏览器是否达到主动回收阈值（累计上下文数或运行时长任一超限）。"""
+        max_contexts = self._settings.browser_recycle_after_contexts
+        if max_contexts > 0 and self._contexts_since_start >= max_contexts:
+            return True
+
+        max_age = self._settings.browser_recycle_after_seconds
+        if max_age > 0 and self._browser_started_at is not None:
+            elapsed = asyncio.get_running_loop().time() - self._browser_started_at
+            if elapsed >= max_age:
+                return True
+
+        return False
+
+    async def _maybe_recycle(self) -> None:
+        """达到回收阈值且当前无在用上下文时，主动重建浏览器以释放累积内存。
+
+        采用「空闲时回收」策略：若仍有其他会话在使用上下文，则跳过本次，
+        交由后续空闲时机处理，避免关闭浏览器时打断进行中的请求。
+        """
+        if not self._recycle_due():
+            return
+
+        async with self._restart_lock:
+            # 持锁后二次确认，避免与并发回收/重建重复执行
+            if not self._recycle_due():
+                return
+            if self._active_contexts > 0:
+                logger.debug(
+                    "浏览器达到回收阈值但仍有 %s 个上下文在用，延迟回收",
+                    self._active_contexts,
+                )
+                return
+
+            logger.info(
+                "浏览器达到回收阈值，主动重建：contexts=%s elapsed=%.0fs",
+                self._contexts_since_start,
+                0.0 if self._browser_started_at is None
+                else asyncio.get_running_loop().time() - self._browser_started_at,
+            )
+            await self.close()
+            await self.start()
+            if not self._is_browser_connected():
+                raise BrowserUnavailableError(self._startup_error or "Browser recycle failed")
+
     @staticmethod
     def _should_retry_context_creation(exc: Exception) -> bool:
         """识别 Playwright driver 断链类错误，允许执行一次重建后重试。"""
@@ -240,7 +291,10 @@ class BrowserPool:
 
         context = None
         page = None
+        context_counted = False
         try:
+            # 创建上下文前检查是否需要主动回收（仅在空闲时执行，不打断进行中的会话）
+            await self._maybe_recycle()
             for attempt in range(2):
                 try:
                     context = await self._browser.new_context(
@@ -258,6 +312,11 @@ class BrowserPool:
                         continue
                     raise
 
+            # 上下文创建成功，计入回收阈值统计并标记为在用
+            self._contexts_since_start += 1
+            self._active_contexts += 1
+            context_counted = True
+
             if self._settings.browser_stealth_enabled:
                 try:
                     from playwright_stealth import Stealth
@@ -296,4 +355,6 @@ class BrowserPool:
                 await page.close()
             if context is not None:
                 await context.close()
+            if context_counted:
+                self._active_contexts -= 1
             self._semaphore.release()