refactor(tradein/admin): debug sweep endpoints на kit orchestration (#2397 slice A) #2400

Merged
lekss361 merged 2 commits from refactor/tradein-admin-debug-sweeps-to-kit into main 2026-07-04 09:16:58 +00:00
Showing only changes of commit 42b1ca6d7f - Show all commits

View file

@ -28,9 +28,27 @@ from pydantic import BaseModel, Field, field_validator
# (endpoint не прокинут через config, в отличие от SERP-классов и BaseScraper.__aenter__
# соседних провайдеров) — вне scope этой задачи (см. "только consume, не трогать
# scraper_kit provider-логику"), фиксируем как follow-up.
#
# scraper_kit-миграция (#2397 slice A, эпик #2277): 5 debug city-sweep/full-load
# роутов (avito-city-sweep, cian-city-sweep, cian-full-load, yandex-city-sweep,
# yandex-full-load) переключены с app.services.scrape_pipeline (legacy) на
# scraper_kit.orchestration.pipeline — тот же DI-паттерн, что и app.scheduler_main
# ._run_kit_scheduler / scraper_kit.orchestration.scheduler._job_*: config/matcher/
# enrichment/proxy_provider инжектируются явно вместо module-level импортов app.*
# внутри scrape_pipeline.py. shutdown_requested НЕ прокидывается (нет SIGTERM-drain
# семантики у admin BackgroundTasks — эквивалент дефолту lambda: False, поведение не
# меняется). Боевой scheduler.py (production sweep-cron) остаётся на legacy — это
# отдельный, ещё больший шаг (см. vault Event_Legacy_Scrapers_Partial_Deletion_Jul04).
from scraper_kit.base import save_listings
from scraper_kit.browser_fetcher import BrowserFetcher
from scraper_kit.orchestration.pipeline import DEFAULT_REGION_CODE
from scraper_kit.orchestration.pipeline import (
DEFAULT_REGION_CODE,
run_avito_city_sweep,
run_cian_city_sweep,
run_cian_full_load,
run_yandex_city_sweep,
run_yandex_full_load,
)
from scraper_kit.providers.avito.detail import fetch_detail, save_detail_enrichment
from scraper_kit.providers.avito.houses import fetch_house_catalog, save_house_catalog_enrichment
from scraper_kit.providers.avito.imv import (
@ -54,14 +72,12 @@ from app.services import cian_session as cian_session_svc
from app.services import scrape_runs as runs_mod
from app.services.geocoder import geocode
from app.services.scheduler import has_running_run
from app.services.scrape_pipeline import (
run_avito_city_sweep,
run_cian_city_sweep,
run_cian_full_load,
run_yandex_city_sweep,
run_yandex_full_load,
from app.services.scraper_adapters import (
RealEnrichmentJobs,
RealMatcherAdapter,
RealProxyProvider,
RealScraperConfig,
)
from app.services.scraper_adapters import RealMatcherAdapter, RealProxyProvider, RealScraperConfig
from app.services.scraper_settings import get_scraper_delay, invalidate_cache
from app.services.scrapers.yandex_newbuilding import YandexNewbuildingScraper
from app.tasks.avito_detail_backfill import run_avito_detail_backfill
@ -840,6 +856,10 @@ async def start_avito_city_sweep(
await run_avito_city_sweep(
sweep_db,
run_id=run_id,
config=RealScraperConfig(),
matcher=RealMatcherAdapter(),
enrichment=RealEnrichmentJobs(),
proxy_provider=_kit_proxy_provider(),
radius_m=payload.radius_m,
pages_per_anchor=payload.pages_per_anchor,
enrich_houses=payload.enrich_houses,
@ -930,6 +950,9 @@ async def start_cian_city_sweep(
await run_cian_city_sweep(
sweep_db,
run_id=run_id,
config=RealScraperConfig(),
matcher=RealMatcherAdapter(),
proxy_provider=_kit_proxy_provider(),
radius_m=payload.radius_m,
pages_per_anchor=payload.pages_per_anchor,
request_delay_sec=payload.request_delay_sec,
@ -1059,6 +1082,9 @@ async def start_cian_full_load(
await run_cian_full_load(
task_db,
run_id=run_id,
config=RealScraperConfig(),
matcher=RealMatcherAdapter(),
proxy_provider=_kit_proxy_provider(),
price_cap_per_bucket=payload.price_cap_per_bucket,
request_delay_sec=payload.request_delay_sec,
concurrency=payload.concurrency,
@ -1160,6 +1186,10 @@ async def start_yandex_full_load(
await run_yandex_full_load(
task_db,
run_id=run_id,
config=RealScraperConfig(),
matcher=RealMatcherAdapter(),
enrichment=RealEnrichmentJobs(),
proxy_provider=_kit_proxy_provider(),
price_cap_per_bucket=payload.price_cap_per_bucket,
request_delay_sec=payload.request_delay_sec,
concurrency=payload.concurrency,
@ -1224,6 +1254,10 @@ async def start_yandex_city_sweep(
await run_yandex_city_sweep(
sweep_db,
run_id=run_id,
config=RealScraperConfig(),
matcher=RealMatcherAdapter(),
enrichment=RealEnrichmentJobs(),
proxy_provider=_kit_proxy_provider(),
radius_m=payload.radius_m,
pages_per_anchor=payload.pages_per_anchor,
request_delay_sec=payload.request_delay_sec,