feat: harden sync flow, proxy bridge, and data normalization
This commit is contained in:
@@ -4,7 +4,7 @@ from contextlib import contextmanager
|
||||
from datetime import datetime, timezone
|
||||
from typing import Iterator
|
||||
|
||||
from sqlalchemy import create_engine, select
|
||||
from sqlalchemy import create_engine, or_, select
|
||||
from sqlalchemy.orm import Session, sessionmaker
|
||||
|
||||
from ..core.config import Settings
|
||||
@@ -76,6 +76,16 @@ class PersistenceService:
|
||||
def start_sync_run(self, lane: str) -> int:
|
||||
# Создаём запись о запуске синхронизации.
|
||||
with self.session_scope() as session:
|
||||
# Если предыдущий процесс умер, оставив run в `running`,
|
||||
# помечаем его как failed перед новым запуском.
|
||||
now = datetime.now(timezone.utc)
|
||||
stale_runs = session.execute(select(SyncRun).where(SyncRun.status == "running")).scalars().all()
|
||||
for stale in stale_runs:
|
||||
stale.status = "failed"
|
||||
stale.finished_at = now
|
||||
if not stale.error_summary:
|
||||
stale.error_summary = "Recovered stale running sync run before starting a new run"
|
||||
|
||||
run = SyncRun(status="running", lane=lane, ids_fetched=0, cars_upserted=0, cars_failed=0, images_upserted=0)
|
||||
session.add(run)
|
||||
session.flush()
|
||||
@@ -95,6 +105,22 @@ class PersistenceService:
|
||||
run.images_upserted = images_upserted
|
||||
run.error_summary = error_summary
|
||||
|
||||
def get_existing_origin_urls(self, origin_urls: list[str]) -> set[str]:
|
||||
# Возвращает уже существующие в БД origin_url для фильтрации only-new запусков.
|
||||
if not origin_urls:
|
||||
return set()
|
||||
with self.session_scope() as session:
|
||||
rows = session.execute(select(Car.origin_url).where(Car.origin_url.in_(origin_urls))).all()
|
||||
return {str(row[0]) for row in rows if row and row[0]}
|
||||
|
||||
def get_existing_origin_ids(self, origin_ids: list[str]) -> set[str]:
|
||||
# Возвращает уже существующие в БД origin_id для фильтрации только новых авто.
|
||||
if not origin_ids:
|
||||
return set()
|
||||
with self.session_scope() as session:
|
||||
rows = session.execute(select(Car.origin_id).where(Car.origin_id.in_(origin_ids))).all()
|
||||
return {str(row[0]) for row in rows if row and row[0]}
|
||||
|
||||
@staticmethod
|
||||
def _add_images(session: Session, car_id: int, images: list[dict[str, object]]) -> None:
|
||||
for image_payload in images:
|
||||
@@ -104,7 +130,7 @@ class PersistenceService:
|
||||
def _car_payload(record: CarRecord) -> dict[str, object]:
|
||||
payload = record.model_dump(mode="python")
|
||||
result = {key: value for key, value in payload.items() if key in CAR_DB_FIELDS}
|
||||
# Serialize raw_attributes dict to JSON string for Text column.
|
||||
|
||||
if "raw_attributes" in result and isinstance(result["raw_attributes"], dict):
|
||||
result["raw_attributes"] = json.dumps(result["raw_attributes"], ensure_ascii=False, default=str)
|
||||
return result
|
||||
@@ -116,8 +142,11 @@ class PersistenceService:
|
||||
images = [image.model_dump(mode="python") for image in record.images]
|
||||
content_hash = str(payload.get("content_hash") or "")
|
||||
with self.session_scope() as session:
|
||||
# поиск по origin_id
|
||||
car = session.execute(select(Car).where(Car.origin_id == record.origin_id)).scalar_one_or_none()
|
||||
# Сначала пытаемся найти по origin_id, а если ранее origin_id был неполный,
|
||||
# подхватываем существующую запись по origin_url, чтобы не плодить дубли.
|
||||
car = session.execute(
|
||||
select(Car).where(or_(Car.origin_id == record.origin_id, Car.origin_url == record.origin_url))
|
||||
).scalar_one_or_none()
|
||||
action = "inserted"
|
||||
if car is None:
|
||||
car = Car(**payload)
|
||||
|
||||
@@ -5,4 +5,4 @@ STEERING_WHEEL_ENUM_VALUES = ("LEFT", "RIGHT", "left", "right", "NA")
|
||||
BODY_TYPE_ENUM_VALUES = ("COUPE", "SUV", "HATCHBACK", "MINIVAN", "SEDAN", "NA", "Station Wagon", "Pickup", "Truck", "Open", "RV", "Other", "STATION_WAGON", "PICKUP", "TRUCK", "OPEN", "OTHER")
|
||||
COUNTRY_ENUM_VALUES = ("JP", "KR", "US", "CA", "NA")
|
||||
ORIGIN_ENUM_VALUES = ("TAU", "CARSENSOR", "HANAMARU", "ENCAR", "KURUMA_TRADER", "carsensor", "encar", "kuruma_trader", "asnet", "kababa", "ACV", "COPART", "copart", "NA", "ASNET", "KABABA", "IAAI")
|
||||
SELLING_TYPE_ENUM_VALUES = ("STOCK", "AUCTION", "TENDER", "stock", "auction", "tender", "NA")
|
||||
SELLING_TYPE_ENUM_VALUES = ("STOCK", "AUCTION", "TENDER", "stock", "auction", "tender", "NA")
|
||||
|
||||
Reference in New Issue
Block a user