fix docker scraping

improve batch sync

add postgres upsert

fix sync locking

improve listing sync

speed up scraper

clean up project

prepare for github

update docker setup
This commit is contained in:
qananasikq
2026-04-13 16:35:08 +03:00
parent b9557922ed
commit a4c93a1df1
30 changed files with 2239 additions and 624 deletions

View File

@@ -13,7 +13,7 @@ logger = logging.getLogger("iaai_scraper.network")
@dataclass
class NetworkCapture:
# XHR/fetch перехватчик.
# Перехватчик сетевых запросов.
settings: Settings
requests: list[dict[str, Any]] = field(default_factory=list)
@@ -21,13 +21,21 @@ class NetworkCapture:
_seen_req: set[str] = field(default_factory=set)
_seen_resp: set[str] = field(default_factory=set)
_origin: str | None = None
_page: Any = field(default=None)
def attach(self, page: Page, origin_url: str | None = None) -> None:
# Снимаем старые подписки перед повторным attach.
try:
page.remove_listener("request", self._on_request)
page.remove_listener("response", self._on_response)
except Exception:
pass
# Подписка на сетевые события
try:
self._origin = urlparse(origin_url or page.url).netloc.lower() or None
except Exception:
self._origin = None
self._page = page
page.on("request", self._on_request)
page.on("response", self._on_response)