IAAI scraper

tests/test_db.py

@@ -0,0 +1,79 @@
+from __future__ import annotations
+
+import tempfile
+import unittest
+from pathlib import Path
+
+from sqlalchemy import select
+
+from iaai_scraper.core.config import Settings
+from iaai_scraper.storage.db import PersistenceService
+from iaai_scraper.storage.models import Car, Image
+from iaai_scraper.storage.schemas import CarRecord, ImageRecord
+
+
+class TestPersistenceServiceIntegration(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        db_path = Path(self.tmp_dir.name) / "test.sqlite"
+
+        self.settings = Settings()
+        self.settings.database.url = f"sqlite:///{db_path.as_posix()}"
+        self.settings.database.echo = False
+
+        self.persistence = PersistenceService(self.settings)
+        self.persistence.create_tables()
+
+    def tearDown(self) -> None:
+        self.persistence.engine.dispose()
+        self.tmp_dir.cleanup()
+
+    @staticmethod
+    def _record(origin_id: str, *, price: int = 1000, content_hash: str = "hash1") -> CarRecord:
+        return CarRecord(
+            parser_id=f"iaai:{origin_id}",
+            brand="Toyota",
+            model="Camry",
+            year=2014,
+            price=price,
+            origin_url=f"https://www.iaai.com/VehicleDetail/{origin_id}~US",
+            origin_id=origin_id,
+            slug=f"toyota-camry-{origin_id}",
+            content_hash=content_hash,
+            images=[
+                ImageRecord(
+                    fullres_image="https://vis.iaai.com/resizer?imageKeys=1&width=845&height=633",
+                    preview_image="https://vis.iaai.com/resizer?imageKeys=1&width=400&height=300",
+                    order_index=0,
+                )
+            ],
+            raw_attributes={"foo": "bar"},
+        )
+
+    def test_insert_update_and_skip_flow(self) -> None:
+        first = self._record("777", price=1000, content_hash="same")
+        inserted = self.persistence.upsert_car(first)
+        self.assertEqual(inserted["action"], "inserted")
+        self.assertEqual(inserted["images_upserted"], 1)
+
+        same = self._record("777", price=1000, content_hash="same")
+        updated_same = self.persistence.upsert_car(same)
+        self.assertEqual(updated_same["action"], "updated")
+        self.assertEqual(updated_same["images_upserted"], 1)
+
+        changed = self._record("777", price=1500, content_hash="changed")
+        updated = self.persistence.upsert_car(changed)
+        self.assertEqual(updated["action"], "updated")
+        self.assertEqual(updated["images_upserted"], 1)
+
+        with self.persistence.session_scope() as session:
+            cars = session.execute(select(Car)).scalars().all()
+            images = session.execute(select(Image)).scalars().all()
+
+        self.assertEqual(len(cars), 1)
+        self.assertEqual(cars[0].price, 1500)
+        self.assertEqual(len(images), 1)
+
+
+if __name__ == "__main__":
+    unittest.main()