gendesign/tradein-mvp/backend/tests/test_city_sweep.py
bot-backend 0a4e126b30
All checks were successful
CI Trade-In / changes (pull_request) Successful in 9s
CI / changes (pull_request) Successful in 9s
CI Trade-In / frontend-checks (pull_request) Has been skipped
CI / backend-tests (pull_request) Has been skipped
CI / frontend-tests (pull_request) Has been skipped
CI / openapi-codegen-check (pull_request) Has been skipped
CI Trade-In / backend-tests (pull_request) Successful in 2m28s
fix(tradein/scraper): проставлять город объявления из контекста развёртки (#2594)
Скрапер знает город в момент сбора (city_slug из CITY_LOCATIONS/CITY_ANCHORS,
scraper_kit.orchestration.pipeline), но раньше нигде его не записывал. Провайдеры
(avito/cian) часто отдают адрес БЕЗ города в тексте ("ул. Победы, 30" вместо
"Нижний Тагил, ул. Победы, 30" — cian даже явно вырезает location-часть перед
записью, providers/cian/serp.py _format_address skip_types={"location",...}).
Без города такой адрес при геокодинге считался "город не назван" и коллизировал
с одноимённой екатеринбургской улицей (Ленина/Победы/Тенистая — сотни совпадений
в ЕКБ-реестрах) → объявление получало координаты Екатеринбурга.

Fix: отдельная колонка listings.city (196_listings_city.sql), проставляется из
sweep-контекста через save_listings(..., city=...) — НЕ парсингом/дописыванием
в address. Раздельная колонка не портит исходный текст адреса: downstream
text-парсеры (geocoder._parse_street_house/_names_non_ekb_city, estimator
house-matching) продолжают работать на исходном сыром тексте неизменёнными —
дописывание города в address ломало бы bare-form адреса без street-маркера
("Дружинина, 33" без "ул.") в этих же парсерах.

Симметрия: EKB-варианты city-sweep функций (city_slug=None) тоже получают
city="Екатеринбург" — resolve_city_name(None) даёт тот же ЕКБ-дефолт, что и
get_city_location/get_city_anchors. Проставлено во всех продовых write-путях:
run_avito_city_sweep/run_yandex_city_sweep/run_cian_city_sweep (city_slug-aware),
run_avito_newbuilding_sweep/run_cian_full_load/run_yandex_full_load/
run_avito_full_load (подтверждённо EKB-only по докстрингам), run_domclick_city_sweep
(EKB city_id, oblast B2 ещё не wired — честный None для неизвестного city_id).

Scope: только write-path для НОВЫХ листингов. Бэкфилл накопленных строк и
консультация city в geocode_missing_listings/backfill_coords_from_geoportal
(gate там пока text-only, _names_non_ekb_city) — geocoder.py намеренно не
тронут (#2582/#2580) — отдельные follow-up задачи.
2026-07-31 22:14:47 +03:00

426 lines
16 KiB
Python
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

"""Offline smoke tests for city sweep utilities + endpoints.
Тесты НЕ требуют реального DB — мокируем SessionLocal и scrape_runs модуль.
"""
from __future__ import annotations
import os
os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")
from datetime import UTC
from unittest.mock import AsyncMock, MagicMock, patch
import pytest
# ── EKB_ANCHORS ────────────────────────────────────────────────────────────
def test_ekb_anchors_count() -> None:
from scraper_kit.orchestration.pipeline import EKB_ANCHORS
assert len(EKB_ANCHORS) == 5
for lat, lon, name in EKB_ANCHORS:
assert 56.5 < lat < 57.0, f"lat={lat} out of EKB range for anchor {name}"
assert 60.3 < lon < 60.8, f"lon={lon} out of EKB range for anchor {name}"
assert isinstance(name, str) and name
# ── resolve_city_name (#2594) ────────────────────────────────────────────────
def test_resolve_city_name_known_oblast_slugs() -> None:
"""Каждый city_slug из CITY_LOCATIONS резолвится в человекочитаемое имя."""
from scraper_kit.orchestration.pipeline import CITY_LOCATIONS, resolve_city_name
expected = {
"nizhniy_tagil": "Нижний Тагил",
"kamensk_uralskiy": "Каменск-Уральский",
"pervouralsk": "Первоуральск",
"verkhnyaya_pyshma": "Верхняя Пышма",
"serov": "Серов",
}
# CITY_DISPLAY_NAMES обязан покрывать ровно те же slug'и, что CITY_LOCATIONS
# (иначе oblast-город бы тихо получил ЕКБ-дефолт вместо своего имени).
assert set(expected) == set(CITY_LOCATIONS)
for slug, name in expected.items():
assert resolve_city_name(slug) == name
def test_resolve_city_name_none_defaults_to_ekaterinburg() -> None:
"""city_slug=None — ЕКБ-развёртка той же функции, НЕ «город неизвестен» (#2594 симметрия)."""
from scraper_kit.orchestration.pipeline import EKATERINBURG_CITY_NAME, resolve_city_name
assert resolve_city_name(None) == EKATERINBURG_CITY_NAME == "Екатеринбург"
def test_resolve_city_name_unknown_slug_defaults_to_ekaterinburg() -> None:
"""Неизвестный slug — тот же ЕКБ-дефолт, что и get_city_location/get_city_anchors."""
from scraper_kit.orchestration.pipeline import resolve_city_name
assert resolve_city_name("nonexistent_city") == "Екатеринбург"
# ── CitySweepCounters ───────────────────────────────────────────────────────
def test_city_sweep_counters_defaults() -> None:
from scraper_kit.orchestration.pipeline import CitySweepCounters
c = CitySweepCounters()
assert c.anchors_total == 0
assert c.lots_fetched == 0
assert c.errors_count == 0
def test_city_sweep_counters_to_dict() -> None:
from scraper_kit.orchestration.pipeline import CitySweepCounters
c = CitySweepCounters(anchors_total=5, anchors_done=2, lots_fetched=150)
d = c.to_dict()
assert d["anchors_total"] == 5
assert d["anchors_done"] == 2
assert d["lots_fetched"] == 150
assert "errors_count" in d
assert "houses_enriched" in d
assert "detail_enriched" in d
def test_city_sweep_counters_to_dict_all_keys() -> None:
from dataclasses import fields
from scraper_kit.orchestration.pipeline import CitySweepCounters
c = CitySweepCounters()
d = c.to_dict()
expected_keys = {f.name for f in fields(c)}
assert set(d.keys()) == expected_keys
# ── scrape_runs utilities (unit, no DB) ────────────────────────────────────
def test_scrape_runs_create_run_returns_id() -> None:
from app.services.scrape_runs import create_run
mock_db = MagicMock()
mock_row = MagicMock()
mock_row.id = 42
mock_db.execute.return_value.fetchone.return_value = mock_row
run_id = create_run(mock_db, source="avito_city_sweep", params={"pages_per_anchor": 3})
assert run_id == 42
mock_db.commit.assert_called()
def test_scrape_runs_is_cancelled_true() -> None:
from app.services.scrape_runs import is_cancelled
mock_db = MagicMock()
mock_row = MagicMock()
mock_row.status = "cancelled"
mock_db.execute.return_value.fetchone.return_value = mock_row
assert is_cancelled(mock_db, 99) is True
def test_scrape_runs_is_cancelled_false_when_running() -> None:
from app.services.scrape_runs import is_cancelled
mock_db = MagicMock()
mock_row = MagicMock()
mock_row.status = "running"
mock_db.execute.return_value.fetchone.return_value = mock_row
assert is_cancelled(mock_db, 99) is False
def test_scrape_runs_is_cancelled_false_when_not_found() -> None:
from app.services.scrape_runs import is_cancelled
mock_db = MagicMock()
mock_db.execute.return_value.fetchone.return_value = None
assert is_cancelled(mock_db, 999) is False
def test_scrape_runs_mark_cancelled_returns_bool() -> None:
from app.services.scrape_runs import mark_cancelled
mock_db = MagicMock()
mock_db.execute.return_value.fetchone.return_value = MagicMock() # row found
result = mark_cancelled(mock_db, 10)
assert result is True
mock_db.commit.assert_called()
def test_scrape_runs_mark_cancelled_returns_false_when_not_running() -> None:
from app.services.scrape_runs import mark_cancelled
mock_db = MagicMock()
mock_db.execute.return_value.fetchone.return_value = None # no row (not running)
result = mark_cancelled(mock_db, 10)
assert result is False
# ── total_seen / new_count column population (audit #1926) ──────────────────
def _captured_params(mock_db: MagicMock) -> dict:
"""Извлечь dict bind-параметров из последнего db.execute(text(...), params)."""
assert mock_db.execute.call_args is not None, "db.execute was not called"
return mock_db.execute.call_args.args[1]
def test_column_counts_maps_lots_fetched_inserted() -> None:
"""_column_counts: lots_fetched→total_seen, lots_inserted→new_count."""
from app.services.scrape_runs import _column_counts
total_seen, new_count = _column_counts(
{"lots_fetched": 150, "lots_inserted": 12, "anchors_done": 5}
)
assert total_seen == 150
assert new_count == 12
def test_column_counts_prefers_explicit_total_seen_new_count() -> None:
"""Если в counters уже есть total_seen/new_count — они приоритетнее lots_*."""
from app.services.scrape_runs import _column_counts
total_seen, new_count = _column_counts(
{"total_seen": 99, "new_count": 7, "lots_fetched": 1, "lots_inserted": 1}
)
assert total_seen == 99
assert new_count == 7
def test_column_counts_returns_none_when_absent() -> None:
"""Нет ни одного из ключей → None (COALESCE сохранит старое значение колонки)."""
from app.services.scrape_runs import _column_counts
total_seen, new_count = _column_counts({"anchors_done": 3})
assert total_seen is None
assert new_count is None
def test_mark_done_persists_total_seen_new_count_columns() -> None:
"""mark_done пишет total_seen/new_count из lots_fetched/lots_inserted (audit #1926)."""
from app.services.scrape_runs import mark_done
mock_db = MagicMock()
mock_db.execute.return_value.first.return_value = MagicMock(id=1) # row updated
mark_done(mock_db, 1, {"lots_fetched": 200, "lots_inserted": 18, "anchors_done": 5})
params = _captured_params(mock_db)
assert params["total_seen"] == 200
assert params["new_count"] == 18
# SQL must SET the dedicated columns, not only the jsonb blob
sql = str(mock_db.execute.call_args.args[0])
assert "total_seen" in sql
assert "new_count" in sql
mock_db.commit.assert_called()
def test_update_heartbeat_persists_total_seen_new_count_columns() -> None:
"""update_heartbeat также обновляет колонки (live progress, не только финал)."""
from app.services.scrape_runs import update_heartbeat
mock_db = MagicMock()
update_heartbeat(mock_db, 1, {"lots_fetched": 75, "lots_inserted": 4})
params = _captured_params(mock_db)
assert params["total_seen"] == 75
assert params["new_count"] == 4
mock_db.commit.assert_called()
def test_mark_done_yandex_counters_populate_columns() -> None:
"""End-to-end shape: YandexCitySweepCounters.to_dict() → mark_done → columns set."""
from scraper_kit.orchestration.pipeline import YandexCitySweepCounters
from app.services.scrape_runs import mark_done
counters = YandexCitySweepCounters(
anchors_total=1, anchors_done=1, lots_fetched=540, lots_inserted=42
)
mock_db = MagicMock()
mock_db.execute.return_value.first.return_value = MagicMock(id=1)
mark_done(mock_db, 1, counters.to_dict())
params = _captured_params(mock_db)
assert params["total_seen"] == 540
assert params["new_count"] == 42
def test_mark_done_avito_cian_counters_populate_columns() -> None:
"""End-to-end shape: CitySweepCounters.to_dict() (avito/cian) → mark_done → columns set.
CitySweepCounters обслуживает avito + cian city-sweep'ы и проходит тот же путь
counters.to_dict() → mark_done, что и Yandex — фикс #1926 должен заполнять
total_seen/new_count и для них (issue: «то же касается CitySweepCounters»).
"""
from scraper_kit.orchestration.pipeline import CitySweepCounters
from app.services.scrape_runs import mark_done
counters = CitySweepCounters(
anchors_total=4, anchors_done=4, lots_fetched=1234, lots_inserted=56
)
mock_db = MagicMock()
mock_db.execute.return_value.first.return_value = MagicMock(id=1)
mark_done(mock_db, 1, counters.to_dict())
params = _captured_params(mock_db)
assert params["total_seen"] == 1234
assert params["new_count"] == 56
def test_mark_failed_persists_total_seen_new_count_columns() -> None:
"""Даже при mark_failed колонки заполняются из частично собранных counters (#1926).
Sweep может упасть на середине, успев собрать тысячи лотов — observability должна
видеть total_seen>0, а не 0 для failed-прогона.
"""
from app.services.scrape_runs import mark_failed
mock_db = MagicMock()
mock_db.execute.return_value.first.return_value = MagicMock(id=1)
mark_failed(mock_db, 1, "boom", {"lots_fetched": 300, "lots_inserted": 9})
# mark_failed после UPDATE дёргает _alert_on_run_id (SELECT source ...), поэтому
# последний execute — это alert-SELECT, а не наш UPDATE. Ищем UPDATE-вызов явно.
update_call = next(
c for c in mock_db.execute.call_args_list if "UPDATE scrape_runs" in str(c.args[0])
)
params = update_call.args[1]
assert params["total_seen"] == 300
assert params["new_count"] == 9
sql = str(update_call.args[0])
assert "total_seen" in sql
assert "new_count" in sql
# ── FastAPI endpoints (offline) ─────────────────────────────────────────────
@pytest.fixture
def app_with_admin():
from fastapi import FastAPI
from fastapi.testclient import TestClient
# Lazy import to avoid side effects before env is set
from app.api.v1 import admin as admin_module
from app.core.db import get_db
app = FastAPI()
app.include_router(admin_module.router, prefix="/api/v1/admin")
def fake_db():
yield MagicMock()
app.dependency_overrides[get_db] = fake_db
return TestClient(app)
def test_start_endpoint_validates_pages_per_anchor(app_with_admin) -> None:
"""pages_per_anchor=99 превышает max=10 → 422."""
r = app_with_admin.post(
"/api/v1/admin/scrape/avito-city-sweep",
json={"pages_per_anchor": 99},
)
assert r.status_code == 422
def test_start_endpoint_validates_request_delay_too_low(app_with_admin) -> None:
"""request_delay_sec=1.0 меньше min=3.0 → 422."""
r = app_with_admin.post(
"/api/v1/admin/scrape/avito-city-sweep",
json={"request_delay_sec": 1.0},
)
assert r.status_code == 422
def test_start_endpoint_ok(app_with_admin) -> None:
"""Valid request → 200, run_id в ответе; kit sweep вызван с DI (config/matcher/enrichment)."""
from app.services.scraper_adapters import (
RealEnrichmentJobs,
RealMatcherAdapter,
RealScraperConfig,
)
with (
patch("app.api.v1.admin.has_running_run", return_value=False),
patch("app.services.scrape_runs.create_run", return_value=7),
patch("app.api.v1.admin.run_avito_city_sweep", new_callable=AsyncMock) as mock_sweep,
):
r = app_with_admin.post(
"/api/v1/admin/scrape/avito-city-sweep",
json={"pages_per_anchor": 2, "detail_top_n": 10},
)
assert r.status_code == 200
body = r.json()
assert body["run_id"] == 7
assert body["status"] == "running"
assert body["pages_per_anchor"] == 2
mock_sweep.assert_called_once()
kw = mock_sweep.call_args.kwargs
assert isinstance(kw["config"], RealScraperConfig)
assert isinstance(kw["matcher"], RealMatcherAdapter)
assert isinstance(kw["enrichment"], RealEnrichmentJobs)
assert "proxy_provider" in kw
def test_cancel_endpoint_exists(app_with_admin) -> None:
"""Cancel endpoint returns run_id + cancelled flag."""
with patch("app.services.scrape_runs.mark_cancelled", return_value=True):
r = app_with_admin.post("/api/v1/admin/scrape/avito-city-sweep/123/cancel")
assert r.status_code == 200
body = r.json()
assert body["run_id"] == 123
assert body["cancelled"] is True
assert body["ok"] is True
def test_cancel_endpoint_not_running(app_with_admin) -> None:
"""Cancel на уже завершённый run → cancelled=False."""
with patch("app.services.scrape_runs.mark_cancelled", return_value=False):
r = app_with_admin.post("/api/v1/admin/scrape/avito-city-sweep/456/cancel")
assert r.status_code == 200
assert r.json()["cancelled"] is False
def test_list_runs_endpoint(app_with_admin) -> None:
"""List runs endpoint возвращает список."""
from datetime import datetime
fake_rows = [
{
"run_id": 1,
"source": "avito_city_sweep",
"status": "done",
"params": {"pages_per_anchor": 3},
"counters": {"lots_fetched": 300},
"error": None,
"started_at": datetime(2025, 1, 1, tzinfo=UTC),
"finished_at": datetime(2025, 1, 1, 1, tzinfo=UTC),
"heartbeat_at": datetime(2025, 1, 1, 1, tzinfo=UTC),
}
]
with patch("app.services.scrape_runs.list_recent", return_value=fake_rows):
r = app_with_admin.get("/api/v1/admin/scrape/avito-city-sweep/runs")
assert r.status_code == 200
body = r.json()
assert len(body) == 1
assert body[0]["run_id"] == 1
assert body[0]["status"] == "done"
assert "2025-01-01" in body[0]["started_at"]