32 lines
930 B
Python
32 lines
930 B
Python
from __future__ import annotations
|
|
|
|
import time
|
|
|
|
from scrapling import PlayWrightFetcher
|
|
|
|
from ..models.request import ScrapeRequest
|
|
from ..models.response import ScrapeResponse
|
|
from .base import BaseScraper
|
|
|
|
|
|
class DynamicScraper(BaseScraper):
|
|
"""Wraps Scrapling's PlayWrightFetcher — full browser via Playwright."""
|
|
|
|
async def scrape(self, req: ScrapeRequest) -> ScrapeResponse:
|
|
start = time.perf_counter()
|
|
|
|
kwargs: dict = {
|
|
"url": req.url,
|
|
"headless": req.headless,
|
|
"timeout": req.timeout,
|
|
"network_idle": req.network_idle,
|
|
}
|
|
if req.wait_selector:
|
|
kwargs["wait_selector"] = req.wait_selector
|
|
if req.proxy:
|
|
kwargs["proxy"] = req.proxy
|
|
|
|
fetcher = PlayWrightFetcher(auto_match=False)
|
|
page = await fetcher.async_fetch(**kwargs)
|
|
return self._build_response(req, page, "dynamic", start)
|