IAAI scraper: Playwright + SQLAlchemy, парсинг авто с аукциона, Docker-ready

iaai_scraper/storage/schemas.py

@@ -0,0 +1,56 @@
+from datetime import datetime, timezone
+from typing import Any
+
+from pydantic import BaseModel, Field
+
+
+class ImageRecord(BaseModel):
+	fullres_image: str
+	preview_image: str
+	order_index: int = 0
+
+
+class CarRecord(BaseModel):
+	parser_id: str
+	brand: str
+	model: str
+	year: int | None = None
+	price: int | None = None
+	currency: str = "USD"
+	mileage: int = 0
+	country: str = "US"
+	is_sold: bool = False
+	color: str = "other"
+	drive: str | None = None
+	gearbox: str | None = None
+	steering_wheel: str | None = None
+	body_type: str = "OTHER"
+	engine_volume: int | None = None
+	selling_type: str = "AUCTION"
+	one_owner: bool = False
+	new_car: bool = False
+	is_hidden: bool = False
+	origin: str = "NA"
+	origin_url: str
+	origin_id: str
+	is_damaged: bool = False
+	evaluation: str | None = None
+	non_smoking: bool = True
+	rental: bool = False
+	repair_history: bool = False
+	slug: str
+	last_seen_at: datetime = Field(default_factory=lambda: datetime.now(timezone.utc))
+	content_hash: str = ""
+	images: list[ImageRecord] = Field(default_factory=list)
+	raw_attributes: dict[str, Any] = Field(default_factory=dict)
+	mapping_notes: list[str] = Field(default_factory=list)
+
+
+class ScrapeExport(BaseModel):
+	source_url: str
+	fetched_at_epoch: int
+	vehicle_summary: dict[str, Any] = Field(default_factory=dict)
+	payload_insights: dict[str, Any] = Field(default_factory=dict)
+	db_record: CarRecord | None = None
+	network: dict[str, Any] = Field(default_factory=dict)
+	access_notes: dict[str, Any] = Field(default_factory=dict)