feat(scrapers): yandex detail-enrichment — save + ночной backfill через прокси (part of #1553) #1554
2 changed files with 230 additions and 77 deletions
|
|
@ -8,12 +8,19 @@ Yandex city sweep does not call YandexDetailScraper — SERP data only.
|
|||
area_m2 coverage 23%, living/kitchen 0%, repair_state 1% on prod.
|
||||
|
||||
Solution: single snapshot SELECT at start (guarantees termination), fetch each
|
||||
offer detail page via YandexDetailScraper (httpx + BaseScraper retry), persist
|
||||
with save_detail_enrichment. No explicit block-exception from YandexDetailScraper —
|
||||
fetch_detail returns None on HTTP error or parse failure. Track consecutive None
|
||||
results; abort after max_consecutive_blocks (mark_done, not mark_failed — retry next
|
||||
night via NULL detail_enriched_at). YandexDetailScraper has no _rotate_ip: plain
|
||||
httpx with UA rotation; no proxy layer required.
|
||||
offer detail page via curl_cffi AsyncSession (chrome120 + proxy) — mirrors
|
||||
yandex_address_backfill.py which already gets full HTML from Yandex on prod.
|
||||
Parse HTML via YandexDetailScraper.parse (pure, no network). Persist via
|
||||
save_detail_enrichment. Track consecutive parse→None results; abort after
|
||||
max_consecutive_blocks (mark_done, not mark_failed — retry next night via
|
||||
NULL detail_enriched_at).
|
||||
|
||||
Why curl_cffi and not YandexDetailScraper.fetch_detail:
|
||||
fetch_detail uses BaseScraper._http_get (plain httpx, no proxy, no TLS
|
||||
fingerprinting). On datacenter IPs Yandex returns captcha / shell-HTML
|
||||
→ parse always returns None → backfill would be 0% effective. The
|
||||
curl_cffi path (chrome120 impersonation + mobile proxy) is already proven
|
||||
by yandex_address_backfill, which fetches identical offer detail pages.
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
|
@ -23,9 +30,11 @@ import logging
|
|||
import time
|
||||
from dataclasses import dataclass, field
|
||||
|
||||
from curl_cffi.requests import AsyncSession
|
||||
from sqlalchemy import text
|
||||
from sqlalchemy.orm import Session
|
||||
|
||||
from app.core.config import settings
|
||||
from app.services import scrape_runs as runs_mod
|
||||
from app.services.scrapers.yandex_detail import YandexDetailScraper, save_detail_enrichment
|
||||
|
||||
|
|
@ -61,21 +70,22 @@ async def run_yandex_detail_backfill(
|
|||
run_id: int,
|
||||
params: dict,
|
||||
) -> YandexDetailBackfillResult:
|
||||
"""Backfill detail_enriched_at for legacy yandex listings via YandexDetailScraper.
|
||||
"""Backfill detail_enriched_at for legacy yandex listings via curl_cffi + parse.
|
||||
|
||||
Params (from default_params jsonb in scrape_schedules):
|
||||
batch_size: int -- snapshot size (SELECT LIMIT), default 800.
|
||||
budget_sec: float -- wall-clock budget per run, default 3600s.
|
||||
request_delay_sec: float -- delay between listings (overrides scraper default), default 5s.
|
||||
max_consecutive_blocks: int -- consecutive None results before abort, default 5.
|
||||
request_delay_sec: float -- delay between listings, default 5s.
|
||||
max_consecutive_blocks: int -- consecutive parse→None before abort, default 5.
|
||||
|
||||
Fetch mechanism:
|
||||
curl_cffi AsyncSession(impersonate="chrome120") + scraper_proxy_url — mirrors
|
||||
yandex_address_backfill. On HTTP 200: pass resp.text to
|
||||
YandexDetailScraper().parse(html, offer_url). parse→None counts as a fail
|
||||
(possible captcha wall); consecutive None → abort after max_consecutive_blocks.
|
||||
|
||||
Lifecycle: update_heartbeat -> snapshot -> loop with budget guard ->
|
||||
mark_done (incl. partial / consecutive-None abort) / mark_failed (exception only).
|
||||
|
||||
YandexDetailScraper.fetch_detail returns None on HTTP error or parse failure
|
||||
(no explicit block exception). Consecutive None results signal a possible
|
||||
temporary ban or network issue; we abort after max_consecutive_blocks and let
|
||||
the scheduler retry the next night (NULL detail_enriched_at rows reappear).
|
||||
"""
|
||||
batch_size = int(params.get("batch_size", 800))
|
||||
budget_sec = float(params.get("budget_sec", 3600))
|
||||
|
|
@ -130,10 +140,22 @@ async def run_yandex_detail_backfill(
|
|||
max_consecutive_blocks,
|
||||
)
|
||||
|
||||
# Build proxies dict once — mirrors yandex_address_backfill.py
|
||||
_proxy = settings.scraper_proxy_url
|
||||
_proxies = {"http": _proxy, "https": _proxy} if _proxy else None
|
||||
|
||||
consecutive_none = 0
|
||||
do_sleep = False
|
||||
scraper = YandexDetailScraper()
|
||||
|
||||
async with YandexDetailScraper() as scraper:
|
||||
async with AsyncSession(
|
||||
impersonate="chrome120",
|
||||
timeout=30.0,
|
||||
proxies=_proxies,
|
||||
headers={
|
||||
"Accept-Language": "ru-RU,ru;q=0.9,en;q=0.8",
|
||||
},
|
||||
) as session:
|
||||
for idx, row in enumerate(snapshot):
|
||||
# Budget guard
|
||||
elapsed = time.monotonic() - start
|
||||
|
|
@ -159,16 +181,64 @@ async def run_yandex_detail_backfill(
|
|||
counters.attempted += 1
|
||||
|
||||
try:
|
||||
enrichment = await scraper.fetch_detail(source_url)
|
||||
try:
|
||||
resp = await session.get(source_url, allow_redirects=True)
|
||||
except Exception as fetch_exc:
|
||||
consecutive_none += 1
|
||||
counters.failed += 1
|
||||
logger.warning(
|
||||
"yandex_detail_backfill: run_id=%d listing_id=%d "
|
||||
"fetch error (consecutive=%d): %s",
|
||||
run_id,
|
||||
listing_id,
|
||||
consecutive_none,
|
||||
fetch_exc,
|
||||
)
|
||||
if consecutive_none >= max_consecutive_blocks:
|
||||
logger.error(
|
||||
"yandex_detail_backfill: run_id=%d ABORT -- %d consecutive "
|
||||
"errors. enriched=%d attempted=%d",
|
||||
run_id,
|
||||
consecutive_none,
|
||||
counters.enriched,
|
||||
counters.attempted,
|
||||
)
|
||||
break
|
||||
continue
|
||||
|
||||
if resp.status_code != 200:
|
||||
consecutive_none += 1
|
||||
counters.failed += 1
|
||||
logger.warning(
|
||||
"yandex_detail_backfill: run_id=%d listing_id=%d "
|
||||
"HTTP %d (consecutive=%d)",
|
||||
run_id,
|
||||
listing_id,
|
||||
resp.status_code,
|
||||
consecutive_none,
|
||||
)
|
||||
if consecutive_none >= max_consecutive_blocks:
|
||||
logger.error(
|
||||
"yandex_detail_backfill: run_id=%d ABORT -- %d consecutive "
|
||||
"non-200 responses. enriched=%d attempted=%d",
|
||||
run_id,
|
||||
consecutive_none,
|
||||
counters.enriched,
|
||||
counters.attempted,
|
||||
)
|
||||
break
|
||||
continue
|
||||
|
||||
enrichment = scraper.parse(resp.text, offer_url=source_url)
|
||||
|
||||
if enrichment is None:
|
||||
# fetch_detail returns None on block / parse failure.
|
||||
# parse→None: captcha wall / shell-HTML / no JSON-LD.
|
||||
# Do not mark listing as done — retry next night.
|
||||
consecutive_none += 1
|
||||
counters.failed += 1
|
||||
logger.warning(
|
||||
"yandex_detail_backfill: run_id=%d listing_id=%d source_url=%s "
|
||||
"-> None (consecutive=%d)",
|
||||
"-> parse None (consecutive=%d)",
|
||||
run_id,
|
||||
listing_id,
|
||||
source_url,
|
||||
|
|
@ -177,7 +247,7 @@ async def run_yandex_detail_backfill(
|
|||
if consecutive_none >= max_consecutive_blocks:
|
||||
logger.error(
|
||||
"yandex_detail_backfill: run_id=%d ABORT -- %d consecutive "
|
||||
"None results (possible ban). enriched=%d attempted=%d",
|
||||
"parse-None results (captcha wall?). enriched=%d attempted=%d",
|
||||
run_id,
|
||||
consecutive_none,
|
||||
counters.enriched,
|
||||
|
|
|
|||
|
|
@ -1,4 +1,8 @@
|
|||
"""Tests for app.tasks.yandex_detail_backfill and save_detail_enrichment."""
|
||||
"""Tests for app.tasks.yandex_detail_backfill.
|
||||
|
||||
Fetch mechanism changed: curl_cffi AsyncSession(chrome120+proxy) + YandexDetailScraper.parse
|
||||
(instead of old fetch_detail path). Mocks target session.get and scraper.parse.
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
|
|
@ -23,10 +27,12 @@ from app.tasks.yandex_detail_backfill import ( # noqa: E402
|
|||
# Path constants for patching
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
_FETCH = "app.tasks.yandex_detail_backfill.YandexDetailScraper"
|
||||
_ASYNC_SESSION = "app.tasks.yandex_detail_backfill.AsyncSession"
|
||||
_PARSE = "app.tasks.yandex_detail_backfill.YandexDetailScraper.parse"
|
||||
_SAVE = "app.tasks.yandex_detail_backfill.save_detail_enrichment"
|
||||
_RUNS = "app.tasks.yandex_detail_backfill.runs_mod"
|
||||
_SLEEP = "app.tasks.yandex_detail_backfill.asyncio.sleep"
|
||||
_SETTINGS = "app.tasks.yandex_detail_backfill.settings"
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Helpers
|
||||
|
|
@ -48,6 +54,33 @@ def _mock_db(snapshot: list[dict]) -> MagicMock:
|
|||
return db
|
||||
|
||||
|
||||
def _make_resp(status: int = 200, text: str = "<html>ok</html>") -> MagicMock:
|
||||
"""Fake curl_cffi response."""
|
||||
resp = MagicMock()
|
||||
resp.status_code = status
|
||||
resp.text = text
|
||||
return resp
|
||||
|
||||
|
||||
def _make_session_ctx(get_side_effect) -> MagicMock:
|
||||
"""Build AsyncSession context-manager mock with session.get side_effect."""
|
||||
session = AsyncMock()
|
||||
session.get = AsyncMock(side_effect=get_side_effect)
|
||||
|
||||
ctx = MagicMock()
|
||||
ctx.__aenter__ = AsyncMock(return_value=session)
|
||||
ctx.__aexit__ = AsyncMock(return_value=None)
|
||||
|
||||
session_cls = MagicMock(return_value=ctx)
|
||||
return session_cls, session
|
||||
|
||||
|
||||
def _mock_settings(proxy: str | None = "http://proxy:3128") -> MagicMock:
|
||||
s = MagicMock()
|
||||
s.scraper_proxy_url = proxy
|
||||
return s
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Tests: run_yandex_detail_backfill
|
||||
# ---------------------------------------------------------------------------
|
||||
|
|
@ -55,17 +88,15 @@ def _mock_db(snapshot: list[dict]) -> MagicMock:
|
|||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_empty_snapshot_marks_done() -> None:
|
||||
"""Empty snapshot -> mark_done immediately, no fetch calls."""
|
||||
"""Empty snapshot -> mark_done immediately, no session.get calls."""
|
||||
db = _mock_db([])
|
||||
runs = MagicMock()
|
||||
mock_scraper_cls = MagicMock()
|
||||
mock_scraper_inst = AsyncMock()
|
||||
mock_scraper_cls.return_value.__aenter__ = AsyncMock(return_value=mock_scraper_inst)
|
||||
mock_scraper_cls.return_value.__aexit__ = AsyncMock(return_value=None)
|
||||
session_cls, session = _make_session_ctx([])
|
||||
|
||||
with (
|
||||
patch(_FETCH, mock_scraper_cls),
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db, run_id=1, params={"batch_size": 10, "budget_sec": 60}
|
||||
|
|
@ -74,30 +105,29 @@ async def test_backfill_empty_snapshot_marks_done() -> None:
|
|||
assert isinstance(result, YandexDetailBackfillResult)
|
||||
assert result.attempted == 0
|
||||
assert result.enriched == 0
|
||||
mock_scraper_inst.fetch_detail.assert_not_called()
|
||||
session.get.assert_not_called()
|
||||
runs.mark_done.assert_called_once()
|
||||
runs.mark_failed.assert_not_called()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_processes_snapshot_to_completion() -> None:
|
||||
"""3 listings -> all fetched and enriched, mark_done called."""
|
||||
"""3 listings -> all fetched (200) + parse success -> enriched=3, mark_done."""
|
||||
snapshot = _make_snapshot(3)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
mock_enrichment = MagicMock()
|
||||
|
||||
mock_scraper_cls = MagicMock()
|
||||
mock_scraper_inst = AsyncMock()
|
||||
mock_scraper_inst.fetch_detail = AsyncMock(return_value=mock_enrichment)
|
||||
mock_scraper_cls.return_value.__aenter__ = AsyncMock(return_value=mock_scraper_inst)
|
||||
mock_scraper_cls.return_value.__aexit__ = AsyncMock(return_value=None)
|
||||
responses = [_make_resp(200), _make_resp(200), _make_resp(200)]
|
||||
session_cls, session = _make_session_ctx(responses)
|
||||
|
||||
with (
|
||||
patch(_FETCH, mock_scraper_cls),
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_PARSE, return_value=mock_enrichment),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SAVE, return_value=True),
|
||||
patch(_SLEEP, new_callable=AsyncMock),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db, run_id=2, params={"batch_size": 10, "budget_sec": 3600}
|
||||
|
|
@ -106,28 +136,27 @@ async def test_backfill_processes_snapshot_to_completion() -> None:
|
|||
assert result.attempted == 3
|
||||
assert result.enriched == 3
|
||||
assert result.failed == 0
|
||||
assert mock_scraper_inst.fetch_detail.call_count == 3
|
||||
assert session.get.call_count == 3
|
||||
runs.mark_done.assert_called_once()
|
||||
runs.mark_failed.assert_not_called()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_none_abort_after_max_consecutive() -> None:
|
||||
"""5 consecutive None results -> abort, mark_done (NOT mark_failed)."""
|
||||
async def test_backfill_parse_none_abort_after_max_consecutive() -> None:
|
||||
"""5 consecutive parse→None results -> abort, mark_done (NOT mark_failed)."""
|
||||
snapshot = _make_snapshot(10)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
|
||||
mock_scraper_cls = MagicMock()
|
||||
mock_scraper_inst = AsyncMock()
|
||||
mock_scraper_inst.fetch_detail = AsyncMock(return_value=None)
|
||||
mock_scraper_cls.return_value.__aenter__ = AsyncMock(return_value=mock_scraper_inst)
|
||||
mock_scraper_cls.return_value.__aexit__ = AsyncMock(return_value=None)
|
||||
responses = [_make_resp(200)] * 10
|
||||
session_cls, _session = _make_session_ctx(responses)
|
||||
|
||||
with (
|
||||
patch(_FETCH, mock_scraper_cls),
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_PARSE, return_value=None),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SLEEP, new_callable=AsyncMock),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db,
|
||||
|
|
@ -143,24 +172,24 @@ async def test_backfill_none_abort_after_max_consecutive() -> None:
|
|||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_none_resets_on_success() -> None:
|
||||
"""1 None + 1 success + 1 None -> consecutive resets, no abort after 1 success."""
|
||||
async def test_backfill_parse_none_resets_on_success() -> None:
|
||||
"""1 parse→None + 1 success + 1 parse→None -> consecutive resets, no abort."""
|
||||
snapshot = _make_snapshot(3)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
mock_enrichment = MagicMock()
|
||||
|
||||
mock_scraper_cls = MagicMock()
|
||||
mock_scraper_inst = AsyncMock()
|
||||
mock_scraper_inst.fetch_detail = AsyncMock(side_effect=[None, mock_enrichment, None])
|
||||
mock_scraper_cls.return_value.__aenter__ = AsyncMock(return_value=mock_scraper_inst)
|
||||
mock_scraper_cls.return_value.__aexit__ = AsyncMock(return_value=None)
|
||||
responses = [_make_resp(200)] * 3
|
||||
session_cls, _session = _make_session_ctx(responses)
|
||||
parse_results = [None, mock_enrichment, None]
|
||||
|
||||
with (
|
||||
patch(_FETCH, mock_scraper_cls),
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_PARSE, side_effect=parse_results),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SAVE, return_value=True),
|
||||
patch(_SLEEP, new_callable=AsyncMock),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db,
|
||||
|
|
@ -174,27 +203,55 @@ async def test_backfill_none_resets_on_success() -> None:
|
|||
runs.mark_done.assert_called_once()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_non200_counts_as_fail_and_aborts() -> None:
|
||||
"""5 consecutive HTTP 403 responses -> abort after max_consecutive_blocks."""
|
||||
snapshot = _make_snapshot(10)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
|
||||
responses = [_make_resp(403)] * 10
|
||||
session_cls, _session = _make_session_ctx(responses)
|
||||
|
||||
with (
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_PARSE, return_value=MagicMock()),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SLEEP, new_callable=AsyncMock),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db,
|
||||
run_id=5,
|
||||
params={"batch_size": 10, "budget_sec": 3600, "max_consecutive_blocks": 5},
|
||||
)
|
||||
|
||||
assert result.failed == 5
|
||||
assert result.enriched == 0
|
||||
runs.mark_done.assert_called_once()
|
||||
runs.mark_failed.assert_not_called()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_budget_guard_stops_loop() -> None:
|
||||
"""Budget expired before first listing -> fetch_detail not called."""
|
||||
"""Budget expired before first listing -> session.get not called."""
|
||||
snapshot = _make_snapshot(5)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
|
||||
mock_scraper_cls = MagicMock()
|
||||
mock_scraper_inst = AsyncMock()
|
||||
mock_scraper_cls.return_value.__aenter__ = AsyncMock(return_value=mock_scraper_inst)
|
||||
mock_scraper_cls.return_value.__aexit__ = AsyncMock(return_value=None)
|
||||
responses = [_make_resp(200)] * 5
|
||||
session_cls, session = _make_session_ctx(responses)
|
||||
|
||||
mono_values = iter([0.0, 999.0, 999.0])
|
||||
with (
|
||||
patch(_FETCH, mock_scraper_cls),
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_RUNS, runs),
|
||||
patch("app.tasks.yandex_detail_backfill.time.monotonic", side_effect=mono_values),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
await run_yandex_detail_backfill(db, run_id=5, params={"batch_size": 5, "budget_sec": 1})
|
||||
await run_yandex_detail_backfill(db, run_id=6, params={"batch_size": 5, "budget_sec": 1})
|
||||
|
||||
mock_scraper_inst.fetch_detail.assert_not_called()
|
||||
session.get.assert_not_called()
|
||||
runs.mark_done.assert_called_once()
|
||||
|
||||
|
||||
|
|
@ -207,10 +264,11 @@ async def test_backfill_top_level_exception_marks_failed() -> None:
|
|||
|
||||
with (
|
||||
patch(_RUNS, runs),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
with pytest.raises(RuntimeError, match="DB connection lost"):
|
||||
await run_yandex_detail_backfill(
|
||||
db, run_id=6, params={"batch_size": 5, "budget_sec": 60}
|
||||
db, run_id=7, params={"batch_size": 5, "budget_sec": 60}
|
||||
)
|
||||
|
||||
runs.mark_failed.assert_called_once()
|
||||
|
|
@ -219,39 +277,67 @@ async def test_backfill_top_level_exception_marks_failed() -> None:
|
|||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_fetch_exception_continues() -> None:
|
||||
"""RuntimeError on one listing -> failed++, db.rollback(), loop continues for next."""
|
||||
"""session.get raises on first listing -> failed++, loop continues for second."""
|
||||
snapshot = _make_snapshot(2)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
mock_enrichment = MagicMock()
|
||||
|
||||
mock_scraper_cls = MagicMock()
|
||||
mock_scraper_inst = AsyncMock()
|
||||
mock_scraper_inst.fetch_detail = AsyncMock(
|
||||
side_effect=[RuntimeError("parse error"), mock_enrichment]
|
||||
)
|
||||
mock_scraper_cls.return_value.__aenter__ = AsyncMock(return_value=mock_scraper_inst)
|
||||
mock_scraper_cls.return_value.__aexit__ = AsyncMock(return_value=None)
|
||||
# First call raises, second returns 200 OK
|
||||
responses = [RuntimeError("connection reset"), _make_resp(200)]
|
||||
session_cls, _session = _make_session_ctx(responses)
|
||||
|
||||
with (
|
||||
patch(_FETCH, mock_scraper_cls),
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_PARSE, return_value=mock_enrichment),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SAVE, return_value=True),
|
||||
patch(_SLEEP, new_callable=AsyncMock),
|
||||
patch(_SETTINGS, _mock_settings()),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db, run_id=7, params={"batch_size": 10, "budget_sec": 3600}
|
||||
db, run_id=8, params={"batch_size": 10, "budget_sec": 3600}
|
||||
)
|
||||
|
||||
assert result.failed == 1
|
||||
assert result.enriched == 1
|
||||
assert result.attempted == 2
|
||||
db.rollback.assert_called()
|
||||
# fetch exception is caught by inner try/except (not DB) — no rollback needed
|
||||
db.rollback.assert_not_called()
|
||||
runs.mark_done.assert_called_once()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_backfill_no_proxy_when_settings_none() -> None:
|
||||
"""scraper_proxy_url=None -> AsyncSession called with proxies=None."""
|
||||
snapshot = _make_snapshot(1)
|
||||
db = _mock_db(snapshot)
|
||||
runs = MagicMock()
|
||||
mock_enrichment = MagicMock()
|
||||
|
||||
responses = [_make_resp(200)]
|
||||
session_cls, _session = _make_session_ctx(responses)
|
||||
|
||||
with (
|
||||
patch(_ASYNC_SESSION, session_cls),
|
||||
patch(_PARSE, return_value=mock_enrichment),
|
||||
patch(_RUNS, runs),
|
||||
patch(_SAVE, return_value=True),
|
||||
patch(_SLEEP, new_callable=AsyncMock),
|
||||
patch(_SETTINGS, _mock_settings(proxy=None)),
|
||||
):
|
||||
result = await run_yandex_detail_backfill(
|
||||
db, run_id=9, params={"batch_size": 10, "budget_sec": 3600}
|
||||
)
|
||||
|
||||
assert result.enriched == 1
|
||||
# Verify AsyncSession was constructed with proxies=None
|
||||
call_kwargs = session_cls.call_args[1]
|
||||
assert call_kwargs.get("proxies") is None
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Tests: save_detail_enrichment
|
||||
# Tests: save_detail_enrichment (unchanged — kept for coverage)
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
|
||||
|
|
@ -296,21 +382,18 @@ def test_save_detail_enrichment_maps_fields_to_update() -> None:
|
|||
db.execute.assert_called_once()
|
||||
db.commit.assert_called_once()
|
||||
|
||||
# Verify the SQL call contains metro_stations and photo_urls as JSON
|
||||
call_args = db.execute.call_args
|
||||
params = call_args[0][1] # positional second arg to execute()
|
||||
params = call_args[0][1]
|
||||
assert params["rooms"] == 2
|
||||
assert params["area_m2"] == 54.5
|
||||
assert params["agency_name"] == "Агентство «Тест»"
|
||||
|
||||
# metro_stations must be JSON-serialised
|
||||
metro_val = params["metro_stations"]
|
||||
assert metro_val is not None
|
||||
parsed_metro = json.loads(metro_val)
|
||||
assert parsed_metro[0]["name"] == "Чкаловская"
|
||||
assert parsed_metro[0]["walk_min"] == 11
|
||||
|
||||
# photo_urls must be JSON-serialised
|
||||
photo_val = params["photo_urls"]
|
||||
assert photo_val is not None
|
||||
parsed_photos = json.loads(photo_val)
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue