improve scraper runtime

This commit is contained in:
qananasikq
2026-04-08 13:30:45 +03:00
parent 21bdf17f35
commit f96e5ca5f8
9 changed files with 207 additions and 21 deletions

View File

@@ -11,6 +11,7 @@ logger = logging.getLogger("iaai_scraper.browser")
def _build_init_script() -> str:
"""JS-патч признаков автоматизации."""
# Подмешиваем базовые browser fingerprint overrides.
hardware_concurrency = random.choice([4, 8, 12, 16])
device_memory = random.choice([4, 8, 16])
languages = ["en-US", "en"]
@@ -61,11 +62,12 @@ def _build_init_script() -> str:
class BrowserFactory:
def __init__(self, settings: Settings) -> None:
# Фабрика браузера и контекстов на основе env-настроек.
self.settings = settings
def create_browser(self, playwright: Playwright) -> Browser:
# пробуем Chrome, если нет — Chromium
launch_kwargs = {
launch_kwargs: dict = {
"headless": self.settings.headless,
"args": [
"--disable-blink-features=AutomationControlled",
@@ -74,6 +76,10 @@ class BrowserFactory:
"--disable-features=IsolateOrigins,site-per-process",
],
}
proxy_dict = self.settings.proxy.to_playwright_dict()
if proxy_dict:
launch_kwargs["proxy"] = proxy_dict
logger.info("Using proxy: %s", self.settings.proxy.server)
try:
logger.info("Trying to launch real Chrome channel")
return playwright.chromium.launch(channel="chrome", **launch_kwargs)
@@ -83,6 +89,7 @@ class BrowserFactory:
def create_context(self, browser: Browser, storage_state: str | None = None) -> BrowserContext:
# рандом viewport/timezone под каждый контекст
# Контекст изолирует cookies, headers и fingerprint-параметры.
viewport = random.choice(self.settings.fingerprint.viewport_presets)
timezone_id = random.choice(self.settings.fingerprint.timezone_candidates)
color_scheme = random.choice(["light", "dark"])