add stiky

This commit is contained in:
Yuriy Yuriev
2026-05-29 21:47:40 +07:00
parent 61ff90e51b
commit b02b7518cf
6 changed files with 109 additions and 220 deletions
+77 -86
View File
@@ -1,5 +1,5 @@
"""
Browser service с поддержкой SOCKS5 через локальный HTTP туннель.
Browser service с поддержкой HTTP и sticky-прокси.
"""
import asyncio
@@ -12,11 +12,9 @@ from dataclasses import dataclass
from camoufox import DefaultAddons
from camoufox.async_api import AsyncCamoufox
from camoufox.exceptions import InvalidIP
from camoufox.exceptions import InvalidIP, InvalidProxy
from config.settings import settings
from managers.proxy_manager import ProxyManager, Proxy, ProxyType
from services.socks5_to_http_proxy import Socks5ProxyPool
logger = logging.getLogger(__name__)
@@ -73,6 +71,44 @@ def _patch_camoufox():
_patch_camoufox()
def _make_sticky_proxy_config() -> tuple:
"""Возвращает (proxy_config, proxy_url_with_auth, proxy_info) для sticky-сессии."""
import re
session_id = random.randint(100000, 999999)
# Заменяем session-XXXXXX в username, если уже есть — иначе добавляем
base = settings.STICKY_PROXY_USER
if re.search(r'session-\d+', base):
username = re.sub(r'session-\d+', f'session-{session_id}', base)
else:
username = f"{base}-session-{session_id}"
config = {
"server": f"http://{settings.STICKY_PROXY_HOST}:{settings.STICKY_PROXY_PORT}",
"username": username,
"password": settings.STICKY_PROXY_PASS,
}
proxy_url = (
f"http://{username}:{settings.STICKY_PROXY_PASS}"
f"@{settings.STICKY_PROXY_HOST}:{settings.STICKY_PROXY_PORT}"
)
return config, proxy_url, f"sticky-{session_id}@{settings.STICKY_PROXY_HOST}"
async def _resolve_sticky_ip(proxy_url: str) -> Optional[str]:
"""Определяет реальный IP sticky-сессии через сам прокси."""
import aiohttp
try:
async with aiohttp.ClientSession() as session:
async with session.get(
"https://api.ipify.org",
proxy=proxy_url,
timeout=aiohttp.ClientTimeout(total=10),
) as resp:
return (await resp.text()).strip()
except Exception as e:
logger.warning(f"Sticky proxy IP resolve failed: {e}")
return None
def _locale_for_ip(ip: str) -> str:
"""Возвращает BCP47 locale без script-тега по реальному IP прокси."""
try:
@@ -104,14 +140,15 @@ class VisitResult:
class BrowserService:
"""
Сервис для посещения страниц через Camoufox.
Поддерживает HTTP, SOCKS5 (через туннель) и прямое соединение.
Поддерживает HTTP и sticky-прокси.
"""
def __init__(self, proxy_manager: ProxyManager):
self.proxy_manager = proxy_manager
self.socks5_pool = Socks5ProxyPool(idle_timeout=300)
_PROXY_FAIL_THRESHOLD = 5 # пауза задачи после N подряд неудачных визитов
def __init__(self):
self.screenshots_dir = Path(settings.SCREENSHOTS_DIR)
self.screenshots_dir.mkdir(exist_ok=True)
self._proxy_fail_streak: int = 0
async def visit_page(
self,
@@ -125,101 +162,61 @@ class BrowserService:
settings.DEFAULT_MAX_READING
)
if not settings.STICKY_PROXY_HOST:
return VisitResult(url=url, success=False, error="No proxy configured")
for attempt in range(3):
proxy = None
use_socks5 = False
try:
if self.proxy_manager.has_proxies():
proxy = await self.proxy_manager.acquire_proxy(timeout=30)
proxy_config, proxy_url, proxy_info = _make_sticky_proxy_config()
real_ip = await _resolve_sticky_ip(proxy_url)
if real_ip is None:
logger.warning(f"Sticky proxy IP resolve failed (attempt {attempt + 1}), retrying with new session")
continue
_set_real_ip(real_ip)
proxy_locale = _locale_for_ip(real_ip)
logger.info(f"Using sticky proxy: {proxy_info} ip={real_ip}")
camoufox_kwargs = dict(
headless=True,
geoip=True,
humanize=True,
exclude_addons=[DefaultAddons.UBO],
proxy=proxy_config,
)
if proxy_locale:
camoufox_kwargs["locale"] = proxy_locale
proxy_locale = None
async with AsyncCamoufox(**camoufox_kwargs) as browser:
result = await self._browse_page(browser, url, proxy_info, reading_time)
try:
if proxy and proxy.proxy_type in [ProxyType.SOCKS5, ProxyType.SOCKS4]:
proxy_config = await self.socks5_pool.get_proxy_config(proxy)
use_socks5 = True
real_ip = proxy.id.split("://")[-1].split(":")[0]
_set_real_ip(real_ip)
proxy_locale = _locale_for_ip(real_ip)
has_auth = bool(proxy.login and proxy.password)
logger.info(f"Using SOCKS5 tunnel for: {proxy.id} auth={'yes' if has_auth else 'NO'} locale={proxy_locale}")
elif proxy:
proxy_config = proxy.proxy_config
real_ip = proxy.id.split("://")[-1].split(":")[0]
_set_real_ip(real_ip)
proxy_locale = _locale_for_ip(real_ip)
logger.info(f"Using HTTP proxy: {proxy.id} locale={proxy_locale}")
else:
proxy_config = None
_set_real_ip(None)
logger.info("Direct connection")
except asyncio.CancelledError:
# Задача отменена — прокси не виноват, освобождаем без штрафа
if proxy:
await self.proxy_manager.release_proxy(proxy, success=True)
raise
if proxy_config:
camoufox_kwargs = dict(
headless=True,
geoip=True,
humanize=True,
exclude_addons=[DefaultAddons.UBO],
proxy=proxy_config,
)
if proxy_locale:
camoufox_kwargs["locale"] = proxy_locale
async with AsyncCamoufox(**camoufox_kwargs) as browser:
result = await self._browse_page(
browser, url,
proxy.id if proxy else "direct",
reading_time
)
else:
return VisitResult(url=url, success=False, error="No proxy available")
# Определяем тип ошибки: ошибка соединения (прокси сломан) vs ошибка загрузки (прокси ok)
is_proxy_dead = result.error and any(
k in result.error for k in ("502", "Bad Gateway", "SOCKS", "NS_ERROR_PROXY")
k in result.error for k in ("502", "Bad Gateway", "NS_ERROR_PROXY")
)
is_load_error = result.error and not is_proxy_dead and any(
k in result.error for k in ("Timeout", "ERR_", "NS_ERROR_", "Connection")
)
if proxy:
# Таймаут/ошибка загрузки — прокси не виноват, освобождаем без штрафа
release_success = True if is_load_error else result.success
await self.proxy_manager.release_proxy(proxy, release_success)
if use_socks5:
await self.socks5_pool.release(proxy)
if is_proxy_dead:
logger.warning(f"Visit attempt {attempt + 1} proxy connection failed, retrying with new proxy")
continue
elif is_load_error:
# Ошибка загрузки — не меняем прокси, просто возвращаем результат
# (сайт мог быть временно недоступен, прокси ок)
logger.warning(f"Visit attempt {attempt + 1} page load error: {result.error or 'unknown'}")
self._proxy_fail_streak = 0
return result
except (ValueError, InvalidIP) as e:
except (ValueError, InvalidIP, InvalidProxy) as e:
logger.warning(f"Visit attempt {attempt + 1} proxy error, retrying: {e}")
if proxy:
await self.proxy_manager.release_proxy(proxy, success=False)
if use_socks5:
await self.socks5_pool.release(proxy)
except Exception as e:
logger.error(f"Visit error: {e}", exc_info=True)
if proxy:
# Ошибка кода/браузера — прокси не виноват
await self.proxy_manager.release_proxy(proxy, success=True)
if use_socks5:
await self.socks5_pool.release(proxy)
return VisitResult(url=url, success=False, error=str(e))
self._proxy_fail_streak += 1
if self._proxy_fail_streak >= self._PROXY_FAIL_THRESHOLD:
logger.error(f"Proxy fail streak {self._proxy_fail_streak} — signalling PROXY_UNAVAILABLE")
self._proxy_fail_streak = 0
return VisitResult(url=url, success=False, error="PROXY_UNAVAILABLE")
return VisitResult(url=url, success=False, error="All attempts failed (fingerprint error)")
async def _browse_page(
@@ -373,13 +370,7 @@ class BrowserService:
return None
async def cleanup(self):
"""Очистка с таймаутом."""
try:
await asyncio.wait_for(self.socks5_pool.stop_all(), timeout=10)
except asyncio.TimeoutError:
logger.warning("SOCKS5 pool cleanup timeout")
except Exception as e:
logger.error(f"Cleanup error: {e}")
pass
async def visit_page_from_twitch(self, url: str, channel: str, reading_time: int = None) -> VisitResult:
return await self.visit_page(url, reading_time)