优化浏览器context复用
This commit is contained in:
@@ -120,28 +120,42 @@ class CloudflareSessionManager:
|
||||
session_name = self._session_name_for_url(target_url)
|
||||
|
||||
# 1. 优先复用已有保活上下文
|
||||
warm = self._warm_contexts.get(session_name)
|
||||
if warm is not None and not self._warm_expired(warm):
|
||||
html = await self._fetch_via_warm(warm, target_url)
|
||||
if html is not None:
|
||||
reused = await self._reuse_warm(session_name, target_url)
|
||||
if reused is not None:
|
||||
html, warm = reused
|
||||
return html, self._warm_to_session(warm)
|
||||
await self._retire_warm(session_name, warm)
|
||||
|
||||
# 2. 在构建锁内创建/复用保活上下文,避免并发重建风暴
|
||||
lock = self._build_locks.setdefault(session_name, asyncio.Lock())
|
||||
async with lock:
|
||||
# 双重检查:可能已被其他协程在等待锁期间建好
|
||||
warm = self._warm_contexts.get(session_name)
|
||||
if warm is not None and not self._warm_expired(warm):
|
||||
html = await self._fetch_via_warm(warm, target_url)
|
||||
if html is not None:
|
||||
reused = await self._reuse_warm(session_name, target_url)
|
||||
if reused is not None:
|
||||
html, warm = reused
|
||||
return html, self._warm_to_session(warm)
|
||||
await self._retire_warm(session_name, warm)
|
||||
|
||||
logger.info("创建保活上下文:session_name=%s", session_name)
|
||||
warm, html = await self._build_warm_context(session_name, target_url)
|
||||
return html, self._warm_to_session(warm)
|
||||
|
||||
async def _reuse_warm(self, session_name: str, target_url: str) -> tuple[str, WarmContext] | None:
|
||||
"""尝试复用已有保活上下文。
|
||||
|
||||
过期或复用失败时退役(关闭)该上下文并返回 None;
|
||||
成功时返回 (html, warm)。
|
||||
"""
|
||||
warm = self._warm_contexts.get(session_name)
|
||||
if warm is None:
|
||||
return None
|
||||
if self._warm_expired(warm):
|
||||
await self._retire_warm(session_name, warm)
|
||||
return None
|
||||
html = await self._fetch_via_warm(warm, target_url)
|
||||
if html is None:
|
||||
await self._retire_warm(session_name, warm)
|
||||
return None
|
||||
return html, warm
|
||||
|
||||
def _warm_expired(self, warm: WarmContext) -> bool:
|
||||
"""保活上下文是否已退役或超过最大存活时长(沿用 session_ttl)。"""
|
||||
if warm.retired:
|
||||
@@ -158,10 +172,13 @@ class CloudflareSessionManager:
|
||||
warm.in_use += 1
|
||||
try:
|
||||
async with self._browser_pool.page_in_context(warm.context) as page:
|
||||
# 携带同站 Referer,使复用请求看起来像站内跳转而非「凭空直达」,
|
||||
# 规避 CF 对「持有 clearance 却无浏览上下文」请求的 403 判定。
|
||||
response = await page.goto(
|
||||
target_url,
|
||||
wait_until="domcontentloaded",
|
||||
timeout=int(self._settings.request_timeout_seconds * 1000),
|
||||
referer=self._settings.base_url,
|
||||
)
|
||||
status = response.status if response is not None else None
|
||||
if status != 200:
|
||||
@@ -190,6 +207,11 @@ class CloudflareSessionManager:
|
||||
|
||||
async def _build_warm_context(self, session_name: str, target_url: str) -> tuple[WarmContext, str]:
|
||||
"""创建一个通过 Cloudflare 挑战的保活上下文,并返回本次访问的页面 HTML。"""
|
||||
# 退役并关闭可能残留的旧上下文(如已过期未复用的),避免覆盖注册表导致泄漏
|
||||
stale = self._warm_contexts.get(session_name)
|
||||
if stale is not None:
|
||||
await self._retire_warm(session_name, stale)
|
||||
|
||||
context = await self._browser_pool.create_persistent_context()
|
||||
try:
|
||||
async with self._browser_pool.page_in_context(context) as page:
|
||||
|
||||
@@ -111,6 +111,23 @@ async def test_warm_context_rebuilds_and_closes_on_failure():
|
||||
assert sess2.session_id != sess1.session_id # 换了新会话
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_expired_warm_context_is_closed_on_rebuild():
|
||||
pool = _FakeBrowserPool()
|
||||
# 存活时长设为 0:任何已存在的保活上下文都立即判定过期
|
||||
settings = Settings(session_ttl_seconds=0, cloudflare_reuse_wait_timeout_seconds=8.0)
|
||||
mgr = CloudflareSessionManager(settings, pool, SessionStore(settings))
|
||||
|
||||
await mgr.fetch_html(_DETAIL_URL.format(1))
|
||||
assert pool.create_calls == 1
|
||||
|
||||
# 第二次请求:旧上下文过期 -> 退役关闭 -> 重建,旧上下文不应泄漏
|
||||
await mgr.fetch_html(_DETAIL_URL.format(2))
|
||||
assert pool.create_calls == 2
|
||||
assert pool.contexts[0].closed is True
|
||||
assert len(mgr._warm_contexts) == 1 # 注册表中只保留新的那个
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_close_releases_all_warm_contexts():
|
||||
pool = _FakeBrowserPool()
|
||||
|
||||
Reference in New Issue
Block a user