chore: настройка определения что мы на странице cf_turnstile

src/antibot/known_providers/cloudflare.py

@@ -84,7 +84,7 @@ class CloudflareDetector(Detector):
 
 	async def detect(self, page: Page, info: ChallengeInfo):
 		log.debug("CloudflareDetector run")
-		await wait_for_networkidle(page)
+		await wait_for_page_stability(page, True, False)
 
 		log.debug("Страница загружена полностью")
 		info.meta = {}
@@ -94,7 +94,9 @@ class CloudflareDetector(Detector):
 
 		log.debug(f"ищем {marker_selector} успех если есть {info.success_locator}")
 
-		if not await page.query_selector(marker_selector):
+		cf_type = await self._detect_cloudflare(page)
+
+		if not cf_type:
 			log.debug("на странице нет маркера преграды")
 			if info.success_locator:
 				info.success = bool(await page.query_selector(info.success_locator))
@@ -108,8 +110,6 @@ class CloudflareDetector(Detector):
 			log.debug("Решение не требуется")
 			return
 
-		content = await page.content()
-		cf_type = next((t for t in self._CTYPES if f"cType: '{t}'" in content), None)
 		info.meta["cf_type"] = cf_type
 		log.debug(f"cType со страницы: {cf_type!r}")
 
@@ -134,6 +134,41 @@ class CloudflareDetector(Detector):
 		else:
 			info.stage = Stage.INTERSTITIAL
 
+	@staticmethod
+	async def _detect_cloudflare(page: Page) -> str | None:
+		"""
+		Detect the type of Cloudflare challenge present in the provided page content.
+
+		This function analyzes the given page content to identify whether a specific
+		type of Cloudflare challenge is present. It checks for three predefined
+		challenge types: non-interactive, managed, and interactive. If a challenge
+		type is detected, it returns the corresponding type as a string. If no
+		challenge type is detected, it returns None.
+
+		Args:
+			page_content (str): The content of the page to analyze for Cloudflare
+				challenge types.
+
+		Returns:
+			str: A string representing the detected Cloudflare challenge type, if
+				found. Returns None if no challenge matches.
+		"""
+		challenge_types = (
+			"non-interactive",
+			"managed",
+			"interactive",
+		)
+		page_content = await page.content()
+		for ctype in challenge_types:
+			if f"cType: '{ctype}'" in page_content:
+				return ctype
+
+		# Check if turnstile captcha is embedded inside the page (Usually inside a closed Shadow iframe)
+		if await page.query_selector('script[src*="challenges.cloudflare.com/turnstile/v"]'):
+			return "embedded"
+
+		return None
+
 
 __CF_PATTERN__ = re.compile(r"^https://challenges\.cloudflare\.com/cdn-cgi/challenge-platform/.*")
 
@@ -170,6 +205,7 @@ class CloudflareSolver(Solver):
 				while time.time() < deadline:
 					if not await page.query_selector(info.detect_locator):
 						return
+					await page.wait_for_timeout(1)
 			return
 		elif info.stage == Stage.INTERSTITIAL:
 			if info.success_locator:
@@ -184,6 +220,7 @@ class CloudflareSolver(Solver):
 				while time.time() < deadline:
 					if not await page.query_selector(info.detect_locator):
 						return
+					await page.wait_for_timeout(1)
 				return
 		else:
 
@@ -193,6 +230,20 @@ class CloudflareSolver(Solver):
 			await captcha_solver.solve(info.meta["container_selector"], page)
 			# TODO: возбудить js
 
+	@classmethod
+	def _challenge_cleared(cls, page_content: str, challenge_type: str) -> bool:
+		"""
+		Check whether the Cloudflare challenge is no longer present in the page content.
+
+		Args:
+			page_content (str): The content of the page to analyze.
+			challenge_type (str): The challenge type returned by `_detect_cloudflare`.
+
+		Returns:
+			bool: True if the challenge is gone from the page, False otherwise.
+		"""
+		return challenge_type == "embedded" or cls._detect_cloudflare(page_content) is None
+
 
 @register
 class ReCaptchaProvider(AntiBotProvider):