cleanup and fix runtime bugs

iaai_scraper/api/app.py

@@ -1,4 +1,4 @@
-"""FastAPI app."""
+# Создание FastAPI-приложения и настройка его жизненного цикла.
 
 from contextlib import asynccontextmanager
 
@@ -11,7 +11,7 @@ from .routes import cars, health, tasks
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    """Жизненный цикл."""
+    # Жизненный цикл.
     persistence: PersistenceService = app.state.persistence
     persistence.create_tables()
     yield
@@ -37,5 +37,5 @@ def create_app(settings: Settings | None = None) -> FastAPI:
     return app
 
 
-# Для запуска через uvicorn.
+# Экземпляр приложения для запуска через uvicorn.
 app = create_app()

iaai_scraper/api/deps.py

@@ -1,4 +1,4 @@
-"""FastAPI deps."""
+# Dependency helpers для FastAPI-роутов.
 
 from fastapi import Request
 

iaai_scraper/api/routes/cars.py

@@ -1,4 +1,4 @@
-"""Cars endpoints."""
+# Роуты для просмотра автомобилей и агрегированной статистики.
 
 from fastapi import APIRouter, Depends, HTTPException, Query
 from sqlalchemy import func, select
@@ -6,6 +6,7 @@ from sqlalchemy import func, select
 from ..deps import get_persistence
 from ...storage.db import PersistenceService
 from ...storage.models import Car, Image
+from ...storage.schemas import CarRead
 
 router = APIRouter()
 
@@ -21,7 +22,7 @@ def list_cars(
     is_sold: bool | None = None,
     persistence: PersistenceService = Depends(get_persistence),
 ):
-    """Список автомобилей с пагинацией и фильтрами."""
+    # Список автомобилей с пагинацией и фильтрами
     with persistence.session_scope() as session:
         query = select(Car)
 
@@ -36,11 +37,9 @@ def list_cars(
         if is_sold is not None:
             query = query.where(Car.is_sold == is_sold)
 
-        # Общее количество.
         count_query = select(func.count()).select_from(query.subquery())
         total = session.execute(count_query).scalar() or 0
 
-        # Пагинация.
         offset = (page - 1) * per_page
         cars = session.execute(
             query.order_by(Car.last_seen_at.desc()).offset(offset).limit(per_page)
@@ -51,7 +50,7 @@ def list_cars(
             "page": page,
             "per_page": per_page,
             "pages": (total + per_page - 1) // per_page if per_page else 0,
-            "items": [_car_to_dict(car) for car in cars],
+            "items": [CarRead.model_validate(car).model_dump(mode="json") for car in cars],
         }
 
 
@@ -60,12 +59,12 @@ def get_car(
     car_id: int,
     persistence: PersistenceService = Depends(get_persistence),
 ):
-    """Детальная информация об автомобиле с изображениями."""
+    # Детальная информация об автомобиле с изображениями
     with persistence.session_scope() as session:
         car = session.get(Car, car_id)
         if car is None:
             raise HTTPException(status_code=404, detail="Car not found")
-        return _car_to_dict(car, include_images=True)
+        return CarRead.model_validate(car).model_dump(mode="json")
 
 
 @router.get("/cars/by-origin/{origin_id}")
@@ -73,19 +72,19 @@ def get_car_by_origin(
     origin_id: str,
     persistence: PersistenceService = Depends(get_persistence),
 ):
-    """Поиск автомобиля по origin_id."""
+    # Поиск автомобиля по origin_id
     with persistence.session_scope() as session:
         car = session.execute(
             select(Car).where(Car.origin_id == origin_id)
         ).scalars().first()
         if car is None:
             raise HTTPException(status_code=404, detail="Car not found")
-        return _car_to_dict(car, include_images=True)
+        return CarRead.model_validate(car).model_dump(mode="json")
 
 
 @router.get("/stats")
 def get_stats(persistence: PersistenceService = Depends(get_persistence)):
-    """Общая статистика по БД."""
+    # Общая статистика по БД
     with persistence.session_scope() as session:
         total_cars = session.execute(select(func.count(Car.id))).scalar() or 0
         total_images = session.execute(select(func.count(Image.id))).scalar() or 0
@@ -102,43 +101,3 @@ def get_stats(persistence: PersistenceService = Depends(get_persistence)):
             "top_brands": [{"brand": b, "count": c} for b, c in brands],
         }
 
-
-def _car_to_dict(car: Car, include_images: bool = False) -> dict:
-    """Сериализация Car в dict."""
-    result = {
-        "id": car.id,
-        "parser_id": car.parser_id,
-        "brand": car.brand,
-        "model": car.model,
-        "year": car.year,
-        "price": car.price,
-        "currency": car.currency,
-        "mileage": car.mileage,
-        "country": car.country,
-        "is_sold": car.is_sold,
-        "color": car.color,
-        "drive": car.drive,
-        "gearbox": car.gearbox,
-        "steering_wheel": car.steering_wheel,
-        "body_type": car.body_type,
-        "engine_volume": car.engine_volume,
-        "selling_type": car.selling_type,
-        "origin": car.origin,
-        "origin_url": car.origin_url,
-        "origin_id": car.origin_id,
-        "is_damaged": car.is_damaged,
-        "slug": car.slug,
-        "last_seen_at": car.last_seen_at.isoformat() if car.last_seen_at else None,
-        "content_hash": car.content_hash,
-    }
-    if include_images:
-        result["images"] = [
-            {
-                "id": img.id,
-                "fullres_image": img.fullres_image,
-                "preview_image": img.preview_image,
-                "order_index": img.order_index,
-            }
-            for img in sorted(car.images, key=lambda i: i.order_index)
-        ]
-    return result

iaai_scraper/api/routes/health.py

@@ -1,4 +1,4 @@
-"""Health endpoint."""
+# Роут проверки доступности сервиса и соединения с БД.
 
 from fastapi import APIRouter, Depends
 from sqlalchemy import text
@@ -11,7 +11,7 @@ router = APIRouter()
 
 @router.get("/health")
 def health_check(persistence: PersistenceService = Depends(get_persistence)):
-    """Проверка API и БД."""
+    # Проверка API и БД
     db_ok = False
     try:
         with persistence.session_scope() as session:

iaai_scraper/api/routes/tasks.py

@@ -1,21 +1,17 @@
-"""Task endpoints."""
+# Роуты запуска задач синхронизации и просмотра истории sync-runs.
 
-from datetime import datetime, timezone
-
-from fastapi import APIRouter, Depends, HTTPException, Query
+from fastapi import APIRouter, Depends, Query
 from pydantic import BaseModel
 from sqlalchemy import select, func
 
 from ..deps import get_persistence
 from ...storage.db import PersistenceService
-from ...storage.models import ScrapeTask, SyncRun
+from ...storage.models import SyncRun
 from ...worker.tasks import sync_vehicle_task, sync_listing_task
 
 router = APIRouter()
 
 
-# --- Схемы.
-
 class SyncVehicleRequest(BaseModel):
     vehicle_url: str
     lane: str = "iaai"
@@ -29,26 +25,16 @@ class SyncListingRequest(BaseModel):
     only_new: bool | None = None
 
 
-# --- Эндпоинты.
-
 @router.post("/tasks/sync-vehicle")
 def start_sync_vehicle(
     body: SyncVehicleRequest,
-    persistence: PersistenceService = Depends(get_persistence),
 ):
-    """Запустить задачу скрапинга одного автомобиля через Celery."""
+    # Запустить задачу скрапинга одного автомобиля через Celery
     result = sync_vehicle_task.apply_async(
         kwargs={"vehicle_url": body.vehicle_url, "lane": body.lane},
         queue="scraping",
     )
 
-    # Регистрируем задачу в БД.
-    persistence.create_scrape_task(
-        celery_task_id=result.id,
-        task_type="sync_vehicle",
-        vehicle_url=body.vehicle_url,
-    )
-
     return {
         "task_id": result.id,
         "status": "queued",
@@ -59,9 +45,8 @@ def start_sync_vehicle(
 @router.post("/tasks/sync-listing")
 def start_sync_listing(
     body: SyncListingRequest,
-    persistence: PersistenceService = Depends(get_persistence),
 ):
-    """Запустить задачу полного цикла листинга через Celery."""
+    # Запустить задачу полного цикла листинга через Celery
     result = sync_listing_task.apply_async(
         kwargs={
             "make": body.make,
@@ -73,86 +58,21 @@ def start_sync_listing(
         queue="scraping",
     )
 
-    persistence.create_scrape_task(
-        celery_task_id=result.id,
-        task_type="sync_listing",
-    )
-
     return {
         "task_id": result.id,
         "status": "queued",
     }
 
 
-@router.get("/tasks/{task_id}")
-def get_task_status(
-    task_id: str,
-    persistence: PersistenceService = Depends(get_persistence),
-):
-    """Статус Celery-задачи."""
-    task_info = persistence.get_scrape_task(task_id)
-    if task_info is None:
-        raise HTTPException(status_code=404, detail="Task not found")
-    return task_info
-
-
-@router.get("/tasks")
-def list_tasks(
-    page: int = Query(1, ge=1),
-    per_page: int = Query(20, ge=1, le=100),
-    status: str | None = None,
-    task_type: str | None = None,
-    persistence: PersistenceService = Depends(get_persistence),
-):
-    """Список задач с пагинацией."""
-    with persistence.session_scope() as session:
-        query = select(ScrapeTask)
-        if status:
-            query = query.where(ScrapeTask.status == status)
-        if task_type:
-            query = query.where(ScrapeTask.task_type == task_type)
-
-        total = session.execute(
-            select(func.count()).select_from(query.subquery())
-        ).scalar() or 0
-
-        offset = (page - 1) * per_page
-        tasks = session.execute(
-            query.order_by(ScrapeTask.created_at.desc()).offset(offset).limit(per_page)
-        ).scalars().all()
-
-        return {
-            "total": total,
-            "page": page,
-            "per_page": per_page,
-            "items": [
-                {
-                    "id": t.id,
-                    "celery_task_id": t.celery_task_id,
-                    "task_type": t.task_type,
-                    "status": t.status,
-                    "vehicle_url": t.vehicle_url,
-                    "created_at": t.created_at.isoformat() if t.created_at else None,
-                    "started_at": t.started_at.isoformat() if t.started_at else None,
-                    "finished_at": t.finished_at.isoformat() if t.finished_at else None,
-                    "result_summary": t.result_summary,
-                    "error_message": t.error_message,
-                }
-                for t in tasks
-            ],
-        }
-
-
 @router.get("/sync-runs")
 def list_sync_runs(
     page: int = Query(1, ge=1),
     per_page: int = Query(20, ge=1, le=100),
     persistence: PersistenceService = Depends(get_persistence),
 ):
-    """История запусков синхронизации."""
+    # История запусков синхронизации
     with persistence.session_scope() as session:
         total = session.execute(select(func.count(SyncRun.id))).scalar() or 0
-
         offset = (page - 1) * per_page
         runs = session.execute(
             select(SyncRun).order_by(SyncRun.started_at.desc()).offset(offset).limit(per_page)