add antibot guard and retry

iaai_scraper/core/exceptions.py

@@ -0,0 +1,10 @@
+class ScraperError(Exception):
+    """Base scraper exception."""
+
+
+class AntiBotDetectedError(ScraperError):
+    """Raised when the target website appears to block automation."""
+
+
+class SiteStructureChangedError(ScraperError):
+    """Raised when the page shape changed and required data is missing."""

iaai_scraper/core/retry.py

@@ -7,6 +7,8 @@ from typing import Any
 
 from playwright.sync_api import Error, TimeoutError as PlaywrightTimeoutError
 
+from .exceptions import AntiBotDetectedError
+
 logger = logging.getLogger("iaai_scraper.retry")
 
 # Типы исключений, при которых retry имеет смысл.
@@ -16,6 +18,7 @@ RETRYABLE_EXCEPTIONS = (
     ConnectionError,
     OSError,
     TimeoutError,
+    AntiBotDetectedError,
 )
 
 

iaai_scraper/parsing/parser.py

@@ -236,12 +236,13 @@ class VehicleParser:
 
 	def _build_access_notes(self, summary: dict[str, Any], responses: list[dict[str, Any]]) -> dict[str, Any]:
 		endpoints = [item.get("url", "") for item in responses]
+		dom_hints = self._dom_hints(" ".join(str(value) for value in summary.values() if value is not None))
 		return {
 			"vin_visible": bool(summary.get("vin")),
 			"images_visible": bool(summary.get("image_urls")),
 			"network_json_count": len(responses),
-			"possible_captcha": False,
-			"possible_antibot": False,
+			"possible_captcha": bool(dom_hints.get("has_captcha_text")),
+			"possible_antibot": bool(dom_hints.get("has_antibot_text")),
 			"observed_endpoints": endpoints[:20],
 		}