persist raw attributes and tighten sync assertions

tests/test_db.py

@@ -104,6 +104,18 @@ class TestPersistenceServiceIntegration(unittest.TestCase):
         self.assertEqual(result["action"], "skipped")
         self.assertEqual(result["images_upserted"], 0)
 
+    def test_persistence_ignores_non_db_fields(self) -> None:
+        record = self._record("1000", content_hash="schema-test")
+        record.raw_attributes = {"vin": "123"}
+        record.mapping_notes = ["note"]
+
+        result = self.persistence.upsert_car(record)
+
+        self.assertEqual(result["action"], "inserted")
+        with self.persistence.session_scope() as session:
+            car = session.execute(select(Car).where(Car.origin_id == "1000")).scalar_one()
+        self.assertEqual(car.origin_id, "1000")
+
 
 if __name__ == "__main__":
     unittest.main()

tests/test_mappers.py

@@ -25,6 +25,30 @@ class TestCarMapper(unittest.TestCase):
         # Длина hex-представления SHA-256
         self.assertEqual(len(record.content_hash), 64)
 
+    def test_content_hash_changes_when_image_set_changes(self) -> None:
+        record_one = self.mapper.map_to_car_record(
+            vehicle_url="https://www.iaai.com/VehicleDetail/45089484~US",
+            vehicle_summary={
+                "make": "Toyota",
+                "model": "Camry",
+                "year": "2014",
+                "image_urls": ["https://vis.iaai.com/resizer?imageKeys=1&width=845&height=633"],
+            },
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+        record_two = self.mapper.map_to_car_record(
+            vehicle_url="https://www.iaai.com/VehicleDetail/45089484~US",
+            vehicle_summary={
+                "make": "Toyota",
+                "model": "Camry",
+                "year": "2014",
+                "image_urls": ["https://vis.iaai.com/resizer?imageKeys=2&width=845&height=633"],
+            },
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+
+        self.assertNotEqual(record_one.content_hash, record_two.content_hash)
+
     def test_deduplicates_images_by_image_key(self) -> None:
         urls = [
             "https://vis.iaai.com/resizer?imageKeys=1&width=200&height=150",

tests/test_scraper.py

@@ -52,8 +52,9 @@ class TestScraperSync(unittest.TestCase):
         scraper.persistence.upsert_car = MagicMock(return_value={"action": "inserted", "images_upserted": 1})
 
         scraper.collect_listing = MagicMock(return_value={"vehicle_urls": ["https://www.iaai.com/VehicleDetail/222~US"]})
+        page = MagicMock()
         scraper._scrape_on_page = MagicMock(return_value={"db_record": make_db_record("222")})
-        scraper._get_page = MagicMock(return_value=MagicMock())
+        scraper._get_page = MagicMock(return_value=page)
         scraper.car_mapper.map_to_car_record = MagicMock(side_effect=AssertionError("should not be called"))
 
         result = scraper.sync_listing()
@@ -63,6 +64,7 @@ class TestScraperSync(unittest.TestCase):
         self.assertIn("trace_id", result)
         self.assertIn("elapsed_seconds", result)
         self.assertEqual(scraper.persistence.upsert_car.call_count, 1)
+        page.close.assert_called_once()
 
     def test_sync_listing_respects_limit(self) -> None:
         scraper = self._make_scraper()
@@ -84,6 +86,21 @@ class TestScraperSync(unittest.TestCase):
 
         self.assertEqual(scraper._scrape_on_page.call_count, 1)
 
+    def test_sync_listing_does_not_count_skipped_as_upserted(self) -> None:
+        scraper = self._make_scraper()
+
+        scraper.persistence.create_tables = MagicMock()
+        scraper.persistence.start_sync_run = MagicMock(return_value=4)
+        scraper.persistence.finish_sync_run = MagicMock()
+        scraper.persistence.upsert_car = MagicMock(return_value={"action": "skipped", "images_upserted": 0})
+        scraper.collect_listing = MagicMock(return_value={"vehicle_urls": ["https://www.iaai.com/VehicleDetail/444~US"]})
+        scraper._scrape_on_page = MagicMock(return_value={"db_record": make_db_record("444")})
+        scraper._get_page = MagicMock(return_value=MagicMock())
+
+        result = scraper.sync_listing()
+
+        self.assertEqual(result["cars_upserted"], 0)
+
     def test_close_resets_browser_state(self) -> None:
         scraper = self._make_scraper()
         scraper.context = MagicMock()