add helper scripts

tests/test_mappers.py

@@ -0,0 +1,181 @@
+from __future__ import annotations
+
+import unittest
+
+from dubizzle_scraper.parsing.mapper import CarMapper
+
+
+class TestCarMapper(unittest.TestCase):
+    def setUp(self) -> None:
+        self.mapper = CarMapper()
+
+    def test_extracts_make_model_from_algolia_details(self) -> None:
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://dubizzle.com/s/DOBI7J3",
+            vehicle_summary={
+                "id": 16810399,
+                "details_v2": {
+                    "tertiary": [
+                        {"slug": "make", "value": {"en": "Toyota", "ar": "تويوتا"}},
+                        {"slug": "model", "value": {"en": "Land Cruiser", "ar": "لاند كروزر"}},
+                    ]
+                },
+                "category_v2": {
+                    "names_en": ["Motors", "Used Cars", "Toyota", "Land Cruiser"],
+                    "slug_paths": [
+                        "motors",
+                        "motors/used-cars",
+                        "motors/used-cars/toyota",
+                        "motors/used-cars/toyota/land-cruiser",
+                    ],
+                },
+            },
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+
+        self.assertEqual(record.brand, "Toyota")
+        self.assertEqual(record.model, "Land Cruiser")
+
+    def test_extracts_make_model_from_legacy_details(self) -> None:
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://dubizzle.com/s/DOBIlegacy",
+            vehicle_summary={
+                "objectID": "item:legacy:1",
+                "details": {
+                    "Make": {"en": {"label": "Make", "value": "Honda"}},
+                    "Model": {"en": {"label": "Model", "value": "Civic"}},
+                },
+            },
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+
+        self.assertEqual(record.brand, "Honda")
+        self.assertEqual(record.model, "Civic")
+
+    def test_extracts_extended_algolia_fields(self) -> None:
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://dubizzle.com/s/DOBGaGs",
+            vehicle_summary={
+                "id": 16345114,
+                "year": 2012,
+                "price": 59000.0,
+                "kilometers": 91800,
+                "seller_type": "DL",
+                "site": {"en": "Dubai"},
+                "category_v2": {
+                    "names_en": ["Motors", "Used Cars", "Porsche", "Cayenne"],
+                    "slug_paths": [
+                        "motors",
+                        "motors/used-cars",
+                        "motors/used-cars/porsche",
+                        "motors/used-cars/porsche/cayenne",
+                    ],
+                },
+                "details": {
+                    "Body Type": {"en": {"value": "SUV"}},
+                    "Transmission Type": {"en": {"value": "Automatic Transmission"}},
+                    "Exterior Color": {"en": {"value": "Brown"}},
+                    "Steering Side": {"en": {"value": "Left Hand"}},
+                    "Engine Capacity (cc)": {"en": {"value": "4800 cc"}},
+                    "Make": {"en": {"value": "Porsche"}},
+                    "Model": {"en": {"value": "Cayenne"}},
+                },
+                "photo_mains": ["https://dbz-images.dubizzle.com/images/example.jpeg?impolicy=dpv"],
+            },
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+
+        self.assertEqual(record.brand, "Porsche")
+        self.assertEqual(record.model, "Cayenne")
+        self.assertEqual(record.price, 59000)
+        self.assertEqual(record.mileage, 91800)
+        self.assertEqual(record.color, "brown")
+        self.assertEqual(record.body_type, "SUV")
+        self.assertEqual(record.gearbox, "AT")
+        self.assertEqual(record.steering_wheel, "LEFT")
+        self.assertEqual(record.engine_volume, 4800)
+        self.assertEqual(record.country, "AE")
+
+    def test_deduplicates_images_by_image_key(self) -> None:
+        urls = [
+            "https://vis.dubizzle.com/resizer?imageKeys=1&width=200&height=150",
+            "https://vis.dubizzle.com/resizer?imageKeys=1&width=845&height=633",
+            "https://vis.dubizzle.com/resizer?imageKeys=2&width=400&height=300",
+        ]
+
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://www.dubizzle.com/VehicleDetail/123~US",
+            vehicle_summary={"make": "Honda", "model": "Civic", "image_urls": urls},
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+
+        self.assertEqual(len(record.images), 2)
+        self.assertIn("width=845", record.images[0].fullres_image)
+        self.assertIn("height=633", record.images[0].fullres_image)
+
+    def test_no_damage_marker_is_not_damaged(self) -> None:
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://www.dubizzle.com/VehicleDetail/123~US",
+            vehicle_summary={"make": "Ford", "model": "Focus"},
+            payload_insights={
+                "vehicle_core": {},
+                "pricing": {},
+                "damage": {"primary": "normal wear"},
+                "auction": {},
+                "images": {},
+            },
+        )
+
+        self.assertFalse(record.is_damaged)
+
+    def test_unknown_empty_values_and_normalization(self) -> None:
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://www.dubizzle.com/VehicleDetail/999~US",
+            vehicle_summary={"make": "  ", "model": None, "drive": "???", "gearbox": "unknown"},
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+        self.assertEqual(record.brand, "UNKNOWN")
+        self.assertEqual(record.model, "UNKNOWN")
+        self.assertEqual(record.drive, "NA")
+        self.assertEqual(record.gearbox, "NA")
+
+        # Нормализация регистра и пробелов.
+        record2 = self.mapper.map_to_car_record(
+            vehicle_url="https://www.dubizzle.com/VehicleDetail/888~US",
+            vehicle_summary={"make": "Honda", "model": "Civic", "drive": "  Front Wheel Drive  ", "gearbox": " AUTOMATIC "},
+            payload_insights={"vehicle_core": {}, "pricing": {}, "damage": {}, "auction": {}, "images": {}},
+        )
+        self.assertEqual(record2.drive, "FWD")
+        self.assertEqual(record2.gearbox, "AT")
+
+    def test_price_and_currency_parsing(self) -> None:
+        record = self.mapper.map_to_car_record(
+            vehicle_url="https://www.dubizzle.com/VehicleDetail/777~US",
+            vehicle_summary={"make": "Toyota", "model": "Corolla"},
+            payload_insights={
+                "vehicle_core": {},
+                "pricing": {"buy_now": "USD 4,500 - 5,200"},
+                "damage": {},
+                "auction": {},
+                "images": {},
+            },
+        )
+        self.assertEqual(record.price, 5200)
+
+        record2 = self.mapper.map_to_car_record(
+            vehicle_url="https://www.dubizzle.com/VehicleDetail/778~US",
+            vehicle_summary={"make": "Toyota", "model": "Corolla"},
+            payload_insights={
+                "vehicle_core": {},
+                "pricing": {"buy_now": "€4.500,00"},
+                "damage": {},
+                "auction": {},
+                "images": {},
+            },
+        )
+        self.assertEqual(record2.currency, "EUR")
+        self.assertEqual(record2.price, 4500)
+
+
+if __name__ == "__main__":
+    unittest.main()