All checks were successful
Deploy Trade-In / build-frontend (push) Has been skipped
Deploy Trade-In / test (push) Successful in 1m35s
Deploy Trade-In / changes (push) Successful in 8s
Deploy Trade-In / build-browser (push) Has been skipped
Deploy Trade-In / build-backend (push) Successful in 1m24s
Deploy Trade-In / deploy (push) Successful in 1m0s
383 lines
15 KiB
Python
383 lines
15 KiB
Python
"""Golden-parity: `scraper_kit.orchestration.pipeline` ≡ `app.services.scrape_pipeline`.
|
||
|
||
Strangler-инвариант (#2135): новая scraper_kit-копия оркестрации sweep-pipeline
|
||
должна давать ТОТ ЖЕ наблюдаемый эффект, что и старый боевой orchestrator, на
|
||
одинаковом мокнутом I/O. Развязка `app.*` (config/matcher/enrichment/runs/shutdown
|
||
инжектируются) не должна менять ветвление.
|
||
|
||
Фокус — КРИТИЧНАЯ логика оркестрации (то, ради чего PR):
|
||
- ban/rotation state-machine: SERP-блок на anchor'е → abort sweep;
|
||
- partial-ban intake (#1950): SERP intake сохранён + detail заблокирован →
|
||
mark_done (не mark_banned) под флагом avito_serp_ok_not_banned;
|
||
- detail-фаза: N подряд блоков → propagate → anchor-handler;
|
||
- counters aggregation по anchor'ам + IMV-фаза;
|
||
- последовательность вызовов scrape_runs (heartbeat / mark_done / mark_banned).
|
||
|
||
Метод: гоняем ОБА orchestrator'а (`_drive_old` / `_drive_new`) на идентичном
|
||
описании сценария, перехватывая scrape_runs recording-mock'ом. Сравниваем:
|
||
- итоговый counters.to_dict();
|
||
- последовательность (method, numeric-counters) вызовов scrape_runs.
|
||
|
||
Без сети, без БД.
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import os
|
||
from types import SimpleNamespace
|
||
from typing import Any
|
||
from unittest.mock import AsyncMock, MagicMock, patch
|
||
|
||
import pytest
|
||
|
||
os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
|
||
|
||
from scraper_kit.avito_exceptions import (
|
||
AvitoBlockedError as NewAvitoBlockedError,
|
||
)
|
||
from scraper_kit.orchestration.pipeline import (
|
||
run_avito_city_sweep as new_city_sweep,
|
||
)
|
||
|
||
from app.services.scrape_pipeline import run_avito_city_sweep as old_city_sweep
|
||
from app.services.scrapers.avito_exceptions import (
|
||
AvitoBlockedError as OldAvitoBlockedError,
|
||
)
|
||
|
||
# ── recording scrape_runs ────────────────────────────────────────────────────
|
||
|
||
_NON_NUMERIC_KEYS = {"enrichment_abort_note"}
|
||
|
||
|
||
class _RunsRecorder:
|
||
"""Записывает вызовы scrape_runs-финализаторов в общий список.
|
||
|
||
is_cancelled всегда False (кооп-cancel не тестируем здесь). Каждый терминальный
|
||
вызов пишет (method, counters_dict) — сравниваем последовательности old/new.
|
||
"""
|
||
|
||
def __init__(self) -> None:
|
||
self.calls: list[tuple[str, dict[str, Any]]] = []
|
||
|
||
def is_cancelled(self, db: Any, run_id: int) -> bool:
|
||
return False
|
||
|
||
def update_heartbeat(self, db: Any, run_id: int, counters: dict[str, Any]) -> None:
|
||
self.calls.append(("update_heartbeat", dict(counters)))
|
||
|
||
def mark_done(self, db: Any, run_id: int, counters: dict[str, Any]) -> None:
|
||
self.calls.append(("mark_done", dict(counters)))
|
||
|
||
def mark_banned(self, db: Any, run_id: int, error: str, counters: dict[str, Any]) -> None:
|
||
self.calls.append(("mark_banned", dict(counters)))
|
||
|
||
def mark_failed(self, db: Any, run_id: int, error: str, counters: dict[str, Any]) -> None:
|
||
self.calls.append(("mark_failed", dict(counters)))
|
||
|
||
|
||
_DriveResult = tuple[dict[str, int], list[tuple[str, dict[str, int]]]]
|
||
|
||
|
||
def _normalize(calls: list[tuple[str, dict[str, Any]]]) -> list[tuple[str, dict[str, int]]]:
|
||
"""Оставить только имя метода + числовые counters (убрать note-строки)."""
|
||
out: list[tuple[str, dict[str, int]]] = []
|
||
for method, counters in calls:
|
||
numeric = {k: v for k, v in counters.items() if k not in _NON_NUMERIC_KEYS}
|
||
out.append((method, numeric))
|
||
return out
|
||
|
||
|
||
# ── scenario description ─────────────────────────────────────────────────────
|
||
|
||
|
||
class _Scenario:
|
||
"""Декларативное описание одного прогона city-sweep для обоих orchestrator'ов."""
|
||
|
||
def __init__(
|
||
self,
|
||
*,
|
||
anchors: list[tuple[float, float, str]],
|
||
# per-anchor: ("lots", n_lots, ins, upd) | ("block",)
|
||
per_anchor: list[tuple[Any, ...]],
|
||
enrich_houses: bool = False,
|
||
detail_top_n: int = 0,
|
||
detail_rows: int = 0,
|
||
detail_behavior: str = "ok", # "ok" | "block_all"
|
||
enrich_imv: bool = False,
|
||
imv_result: tuple[int, int, int] | None = None,
|
||
avito_serp_ok_not_banned: bool = True,
|
||
avito_proxy_max_rotations: int = 0,
|
||
lots_have_house_url: bool = False,
|
||
) -> None:
|
||
self.anchors = anchors
|
||
self.per_anchor = per_anchor
|
||
self.enrich_houses = enrich_houses
|
||
self.detail_top_n = detail_top_n
|
||
self.detail_rows = detail_rows
|
||
self.detail_behavior = detail_behavior
|
||
self.enrich_imv = enrich_imv
|
||
self.imv_result = imv_result
|
||
self.avito_serp_ok_not_banned = avito_serp_ok_not_banned
|
||
self.avito_proxy_max_rotations = avito_proxy_max_rotations
|
||
self.lots_have_house_url = lots_have_house_url
|
||
|
||
def _config(self) -> SimpleNamespace:
|
||
return SimpleNamespace(
|
||
scraper_fetch_mode="curl_cffi",
|
||
browser_http_endpoint="http://browser.test/fetch",
|
||
scraper_proxy_url=None,
|
||
avito_proxy_rotate_url=None,
|
||
avito_proxy_max_rotations=self.avito_proxy_max_rotations,
|
||
avito_serp_ok_not_banned=self.avito_serp_ok_not_banned,
|
||
avito_proxy_rotate_settle_s=0.0,
|
||
proxy_rotate_attempts=1,
|
||
proxy_rotate_attempt_timeout_s=1.0,
|
||
cian_proxy_rotate_url=None,
|
||
cian_proxy_max_rotations=0,
|
||
yandex_proxy_rotate_url=None,
|
||
yandex_proxy_max_rotations=0,
|
||
scraper_skip_seen_today=False,
|
||
)
|
||
|
||
def _fetch_around_side_effects(self, blocked_exc: type[Exception]) -> list[Any]:
|
||
effects: list[Any] = []
|
||
for spec in self.per_anchor:
|
||
if spec[0] == "block":
|
||
effects.append(blocked_exc("SERP blocked"))
|
||
else:
|
||
_, n_lots, _ins, _upd = spec
|
||
house_url = "/catalog/houses/ekb/h1/100" if self.lots_have_house_url else None
|
||
effects.append([MagicMock(house_url=house_url) for _ in range(n_lots)])
|
||
return effects
|
||
|
||
def _save_side_effects(self) -> list[tuple[int, int]]:
|
||
return [(spec[2], spec[3]) for spec in self.per_anchor if spec[0] == "lots" and spec[1] > 0]
|
||
|
||
def _detail_rows(self) -> list[dict[str, str]]:
|
||
return [{"source_url": f"https://www.avito.ru/x/{i}"} for i in range(self.detail_rows)]
|
||
|
||
def _fetch_detail_side_effects(self, blocked_exc: type[Exception]) -> Any:
|
||
if self.detail_behavior == "block_all":
|
||
return blocked_exc("detail blocked")
|
||
return MagicMock(house_catalog_url=None)
|
||
|
||
|
||
def _make_db(scenario: _Scenario) -> MagicMock:
|
||
db = MagicMock()
|
||
# detail-фаза: db.execute(...).mappings().all() → priority rows
|
||
db.execute.return_value.mappings.return_value.all.return_value = scenario._detail_rows()
|
||
return db
|
||
|
||
|
||
def _make_scraper(scenario: _Scenario, blocked_exc: type[Exception]) -> MagicMock:
|
||
scraper = MagicMock()
|
||
scraper._cffi = None
|
||
scraper._browser = None
|
||
scraper.fetch_around = AsyncMock(side_effect=scenario._fetch_around_side_effects(blocked_exc))
|
||
return scraper
|
||
|
||
|
||
def _async_session_cm() -> MagicMock:
|
||
sess = MagicMock()
|
||
sess.__aenter__ = AsyncMock(return_value=sess)
|
||
sess.__aexit__ = AsyncMock(return_value=None)
|
||
sess.close = AsyncMock()
|
||
return sess
|
||
|
||
|
||
async def _drive_old(scenario: _Scenario) -> _DriveResult:
|
||
recorder = _RunsRecorder()
|
||
db = _make_db(scenario)
|
||
scraper = _make_scraper(scenario, OldAvitoBlockedError)
|
||
save_mock = MagicMock(side_effect=scenario._save_side_effects())
|
||
|
||
imv_res = None
|
||
if scenario.imv_result is not None:
|
||
checked, saved, errors = scenario.imv_result
|
||
imv_res = SimpleNamespace(checked=checked, saved=saved, errors=errors)
|
||
|
||
with (
|
||
patch("app.services.scrape_pipeline.AvitoScraper", return_value=scraper),
|
||
patch("app.services.scrape_pipeline.save_listings", save_mock),
|
||
patch(
|
||
"app.services.scrape_pipeline.fetch_house_catalog",
|
||
AsyncMock(return_value=MagicMock()),
|
||
),
|
||
patch(
|
||
"app.services.scrape_pipeline.save_house_catalog_enrichment",
|
||
return_value={"house_id": 1},
|
||
),
|
||
patch(
|
||
"app.services.scrape_pipeline.fetch_detail",
|
||
AsyncMock(side_effect=scenario._fetch_detail_side_effects(OldAvitoBlockedError)),
|
||
),
|
||
patch("app.services.scrape_pipeline.save_detail_enrichment", return_value=True),
|
||
patch("app.services.scrape_pipeline.scrape_runs", recorder),
|
||
patch(
|
||
"app.services.house_imv_backfill.process_houses_imv_batch",
|
||
AsyncMock(return_value=imv_res),
|
||
),
|
||
patch("app.services.scrape_pipeline.settings", scenario._config()),
|
||
patch("app.services.scrape_pipeline.shutdown_requested", return_value=False),
|
||
patch("app.services.scrape_pipeline.asyncio.sleep", AsyncMock()),
|
||
patch("app.services.scrape_pipeline.AsyncSession", return_value=_async_session_cm()),
|
||
):
|
||
counters = await old_city_sweep(
|
||
db,
|
||
run_id=1,
|
||
radius_m=1000,
|
||
anchors=scenario.anchors,
|
||
pages_per_anchor=1,
|
||
enrich_houses=scenario.enrich_houses,
|
||
detail_top_n=scenario.detail_top_n,
|
||
request_delay_sec=0.0,
|
||
enrich_imv=scenario.enrich_imv,
|
||
)
|
||
return counters.to_dict(), _normalize(recorder.calls)
|
||
|
||
|
||
async def _drive_new(scenario: _Scenario) -> _DriveResult:
|
||
recorder = _RunsRecorder()
|
||
db = _make_db(scenario)
|
||
scraper = _make_scraper(scenario, NewAvitoBlockedError)
|
||
save_mock = MagicMock(side_effect=scenario._save_side_effects())
|
||
|
||
imv_res = None
|
||
if scenario.imv_result is not None:
|
||
checked, saved, errors = scenario.imv_result
|
||
imv_res = SimpleNamespace(checked=checked, saved=saved, errors=errors)
|
||
|
||
enrichment = MagicMock()
|
||
enrichment.process_houses_imv_batch = AsyncMock(return_value=imv_res)
|
||
|
||
pfx = "scraper_kit.orchestration.pipeline"
|
||
with (
|
||
patch(f"{pfx}.AvitoScraper", return_value=scraper),
|
||
patch(f"{pfx}.save_listings", save_mock),
|
||
patch(f"{pfx}.fetch_house_catalog", AsyncMock(return_value=MagicMock())),
|
||
patch(f"{pfx}.save_house_catalog_enrichment", return_value={"house_id": 1}),
|
||
patch(
|
||
f"{pfx}.fetch_detail",
|
||
AsyncMock(side_effect=scenario._fetch_detail_side_effects(NewAvitoBlockedError)),
|
||
),
|
||
patch(f"{pfx}.save_detail_enrichment", return_value=True),
|
||
patch(f"{pfx}.runs", recorder),
|
||
patch(f"{pfx}.asyncio.sleep", AsyncMock()),
|
||
patch(f"{pfx}.AsyncSession", return_value=_async_session_cm()),
|
||
):
|
||
counters = await new_city_sweep(
|
||
db,
|
||
run_id=1,
|
||
config=scenario._config(),
|
||
matcher=MagicMock(),
|
||
enrichment=enrichment,
|
||
shutdown_requested=lambda: False,
|
||
radius_m=1000,
|
||
anchors=scenario.anchors,
|
||
pages_per_anchor=1,
|
||
enrich_houses=scenario.enrich_houses,
|
||
detail_top_n=scenario.detail_top_n,
|
||
request_delay_sec=0.0,
|
||
enrich_imv=scenario.enrich_imv,
|
||
)
|
||
return counters.to_dict(), _normalize(recorder.calls)
|
||
|
||
|
||
async def _assert_parity(scenario: _Scenario) -> dict[str, int]:
|
||
old_counters, old_calls = await _drive_old(scenario)
|
||
new_counters, new_calls = await _drive_new(scenario)
|
||
assert (
|
||
new_counters == old_counters
|
||
), f"counters diverged:\n old={old_counters}\n new={new_counters}"
|
||
assert (
|
||
new_calls == old_calls
|
||
), f"scrape_runs calls diverged:\n old={old_calls}\n new={new_calls}"
|
||
return new_counters
|
||
|
||
|
||
# ── scenarios ────────────────────────────────────────────────────────────────
|
||
|
||
_ANCHORS_2 = [(56.84, 60.60, "A1"), (56.79, 60.53, "A2")]
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_parity_happy_path_counters_aggregation() -> None:
|
||
"""2 anchor'а, SERP+save, без houses/detail/imv → mark_done + агрегированные counters."""
|
||
scenario = _Scenario(
|
||
anchors=_ANCHORS_2,
|
||
per_anchor=[("lots", 10, 8, 2), ("lots", 6, 5, 1)],
|
||
)
|
||
counters = await _assert_parity(scenario)
|
||
# Убеждаемся что сценарий действительно прогнал агрегацию (не пустой прогон).
|
||
assert counters["lots_fetched"] == 16
|
||
assert counters["lots_inserted"] == 13
|
||
assert counters["lots_updated"] == 3
|
||
assert counters["anchors_done"] == 2
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_parity_full_ban_no_intake_marks_banned() -> None:
|
||
"""Первый же anchor SERP-блок, лоты не сохранены → mark_banned (не done)."""
|
||
scenario = _Scenario(
|
||
anchors=_ANCHORS_2,
|
||
per_anchor=[("block",), ("block",)],
|
||
)
|
||
await _assert_parity(scenario)
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_parity_partial_ban_intake_marks_done() -> None:
|
||
"""Anchor1 intake сохранён, anchor2 SERP-блок → partial-intake → mark_done не banned."""
|
||
scenario = _Scenario(
|
||
anchors=_ANCHORS_2,
|
||
per_anchor=[("lots", 10, 9, 1), ("block",)],
|
||
avito_serp_ok_not_banned=True,
|
||
)
|
||
await _assert_parity(scenario)
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_parity_partial_ban_flag_off_marks_banned() -> None:
|
||
"""avito_serp_ok_not_banned=False: даже при intake SERP-блок → mark_banned."""
|
||
scenario = _Scenario(
|
||
anchors=_ANCHORS_2,
|
||
per_anchor=[("lots", 10, 9, 1), ("block",)],
|
||
avito_serp_ok_not_banned=False,
|
||
)
|
||
await _assert_parity(scenario)
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_parity_detail_consecutive_block_propagates() -> None:
|
||
"""detail-фаза: 3 подряд блока (rotations=0) → propagate → anchor-handler.
|
||
|
||
Лоты сохранены на этом же anchor'е → partial-intake → mark_done.
|
||
"""
|
||
scenario = _Scenario(
|
||
anchors=[(56.84, 60.60, "A1")],
|
||
per_anchor=[("lots", 5, 5, 0)],
|
||
detail_top_n=5,
|
||
detail_rows=3,
|
||
detail_behavior="block_all",
|
||
avito_serp_ok_not_banned=True,
|
||
avito_proxy_max_rotations=0,
|
||
)
|
||
await _assert_parity(scenario)
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_parity_imv_phase_counters() -> None:
|
||
"""IMV-фаза: touched houses → process_houses_imv_batch → imv-counters агрегированы."""
|
||
scenario = _Scenario(
|
||
anchors=[(56.84, 60.60, "A1")],
|
||
per_anchor=[("lots", 4, 4, 0)],
|
||
enrich_houses=True,
|
||
lots_have_house_url=True,
|
||
enrich_imv=True,
|
||
imv_result=(3, 2, 1),
|
||
)
|
||
counters = await _assert_parity(scenario)
|
||
# touched house → IMV-фаза отработала, imv-counters агрегированы одинаково.
|
||
assert counters["imv_attempted"] == 3
|
||
assert counters["imv_enriched"] == 2
|
||
assert counters["imv_failed"] == 1
|