245 lines
9 KiB
Python
245 lines
9 KiB
Python
"""Avito newbuilding (novostroyka) citywide sweep.
|
||
|
||
Тесты (без сети, без реального DB):
|
||
1. _build_newbuilding_url — novostroyka-slug в пути, s=104, page param, без geo.
|
||
2. _parse_html на novostroyka-SERP fixture → listing_segment='novostroyki' +
|
||
newbuilding_id заполнен (DOM-маркер застройщика + zhk-якорь).
|
||
3. fetch_newbuildings break-on-empty + dedup (mirror fetch_city_wide).
|
||
4. NewbuildingSweepCounters — defaults + to_dict.
|
||
5. run_avito_newbuilding_sweep wiring — мокаем fetch_newbuildings + save_listings,
|
||
assert save вызван и SERP-счётчики записаны (mark_done).
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import os
|
||
from pathlib import Path
|
||
from typing import Any
|
||
from unittest.mock import AsyncMock, MagicMock, patch
|
||
|
||
import pytest
|
||
|
||
os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")
|
||
|
||
from app.services import scrape_pipeline
|
||
from app.services.scrapers.avito import NOVOSTROYKA_SLUG, AvitoScraper
|
||
from app.services.scrapers.base import ScrapedLot
|
||
|
||
FIXTURE = Path(__file__).parent / "fixtures" / "avito_serp_novostroyka.html"
|
||
|
||
|
||
def _make_lot(source_id: str) -> ScrapedLot:
|
||
return ScrapedLot(
|
||
source="avito",
|
||
source_url=f"https://www.avito.ru/ekaterinburg/kvartiry/{source_id}",
|
||
source_id=source_id,
|
||
price_rub=6_000_000,
|
||
listing_segment="novostroyki",
|
||
)
|
||
|
||
|
||
# ── 1. _build_newbuilding_url ───────────────────────────────────────────────
|
||
|
||
|
||
def test_build_newbuilding_url_contains_slug_and_sort() -> None:
|
||
s = AvitoScraper()
|
||
url = s._build_newbuilding_url(page=1)
|
||
assert NOVOSTROYKA_SLUG in url
|
||
assert "novostroyka-" in url
|
||
assert "geoCoords" not in url
|
||
assert "radius" not in url
|
||
assert "s=104" in url
|
||
assert "p=1" in url
|
||
assert "ekaterinburg/kvartiry/prodam/" in url
|
||
|
||
|
||
def test_build_newbuilding_url_page_param() -> None:
|
||
s = AvitoScraper()
|
||
url = s._build_newbuilding_url(page=2)
|
||
assert "novostroyka-" in url
|
||
assert "p=2" in url
|
||
|
||
|
||
def test_build_newbuilding_url_differs_from_citywide() -> None:
|
||
s = AvitoScraper()
|
||
assert s._build_newbuilding_url(page=1) != s._build_citywide_url(page=1)
|
||
|
||
|
||
# ── 2. _parse_html classifies novostroyka cards ─────────────────────────────
|
||
|
||
|
||
def test_parse_novostroyka_fixture_segments_and_newbuilding_id() -> None:
|
||
html = FIXTURE.read_text(encoding="utf-8")
|
||
s = AvitoScraper()
|
||
lots = s._parse_html(html, source_url_base=s._build_newbuilding_url(page=1))
|
||
|
||
assert len(lots) == 2
|
||
for lot in lots:
|
||
assert lot.listing_segment == "novostroyki"
|
||
assert lot.newbuilding_id is not None
|
||
assert lot.newbuilding_url is not None
|
||
|
||
ids = {lot.newbuilding_id for lot in lots}
|
||
assert "meridian-ekaterinburg" in ids
|
||
assert "severnyy-kvartal-ekaterinburg" in ids
|
||
|
||
|
||
# ── 3. fetch_newbuildings break-on-empty + dedup ────────────────────────────
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_fetch_newbuildings_break_on_empty() -> None:
|
||
s = AvitoScraper()
|
||
pages_data: list[list[ScrapedLot]] = [[_make_lot("A"), _make_lot("B")], []]
|
||
call_n = 0
|
||
|
||
async def mock_html(url: str, page: int) -> str:
|
||
assert "novostroyka-" in url
|
||
return f"<html>{page}</html>"
|
||
|
||
def mock_parse(html: str, source_url_base: str) -> list[ScrapedLot]:
|
||
nonlocal call_n
|
||
idx = call_n
|
||
call_n += 1
|
||
return pages_data[idx] if idx < len(pages_data) else []
|
||
|
||
with patch.object(s, "_fetch_serp_html", side_effect=mock_html):
|
||
with patch.object(s, "_parse_html", side_effect=mock_parse):
|
||
with patch.object(s, "sleep_between_requests", new=AsyncMock()):
|
||
result = await s.fetch_newbuildings(pages=10, delay_override_sec=0)
|
||
|
||
assert len(result) == 2
|
||
assert call_n == 2 # page1 + page2(empty) → stop
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_fetch_newbuildings_dedup_by_source_id() -> None:
|
||
s = AvitoScraper()
|
||
dup = _make_lot("SAME")
|
||
pages_data: list[list[ScrapedLot]] = [[dup], [dup], []]
|
||
call_n = 0
|
||
|
||
async def mock_html(url: str, page: int) -> str:
|
||
return f"<html>{page}</html>"
|
||
|
||
def mock_parse(html: str, source_url_base: str) -> list[ScrapedLot]:
|
||
nonlocal call_n
|
||
idx = call_n
|
||
call_n += 1
|
||
return pages_data[idx] if idx < len(pages_data) else []
|
||
|
||
with patch.object(s, "_fetch_serp_html", side_effect=mock_html):
|
||
with patch.object(s, "_parse_html", side_effect=mock_parse):
|
||
with patch.object(s, "sleep_between_requests", new=AsyncMock()):
|
||
result = await s.fetch_newbuildings(pages=10, delay_override_sec=0)
|
||
|
||
assert len(result) == 1
|
||
assert result[0].source_id == "SAME"
|
||
|
||
|
||
def test_fetch_city_wide_url_unchanged() -> None:
|
||
"""fetch_city_wide остаётся citywide-URL — рефактор не сменил его поведение."""
|
||
s = AvitoScraper()
|
||
assert "novostroyka-" not in s._build_citywide_url(page=1)
|
||
assert "prodam-ASgBAgICAUSSA8YQ" in s._build_citywide_url(page=1)
|
||
|
||
|
||
# ── 4. NewbuildingSweepCounters ─────────────────────────────────────────────
|
||
|
||
|
||
def test_newbuilding_counters_defaults_and_to_dict() -> None:
|
||
from app.services.scrape_pipeline import NewbuildingSweepCounters
|
||
|
||
c = NewbuildingSweepCounters()
|
||
assert c.lots_fetched == 0
|
||
assert c.lots_inserted == 0
|
||
assert c.lots_updated == 0
|
||
assert c.errors_count == 0
|
||
d = c.to_dict()
|
||
assert set(d.keys()) == {"lots_fetched", "lots_inserted", "lots_updated", "errors_count"}
|
||
|
||
|
||
# ── 5. run_avito_newbuilding_sweep wiring ───────────────────────────────────
|
||
|
||
|
||
class _FakeRuns:
|
||
def __init__(self) -> None:
|
||
self.heartbeats: list[dict[str, int]] = []
|
||
self.done: dict[str, int] | None = None
|
||
self.failed: tuple[str, dict[str, int]] | None = None
|
||
self.banned: tuple[str, dict[str, int]] | None = None
|
||
|
||
def is_cancelled(self, _db: Any, _run_id: int) -> bool:
|
||
return False
|
||
|
||
def update_heartbeat(self, _db: Any, _run_id: int, counters: dict[str, int]) -> None:
|
||
self.heartbeats.append(dict(counters))
|
||
|
||
def mark_done(self, _db: Any, _run_id: int, counters: dict[str, int]) -> None:
|
||
self.done = dict(counters)
|
||
|
||
def mark_failed(self, _db: Any, _run_id: int, error: str, counters: dict[str, int]) -> None:
|
||
self.failed = (error, dict(counters))
|
||
|
||
def mark_banned(self, _db: Any, _run_id: int, error: str, counters: dict[str, int]) -> None:
|
||
self.banned = (error, dict(counters))
|
||
|
||
|
||
def _install_runs(monkeypatch: pytest.MonkeyPatch, fake: _FakeRuns) -> None:
|
||
monkeypatch.setattr(scrape_pipeline.scrape_runs, "is_cancelled", fake.is_cancelled)
|
||
monkeypatch.setattr(scrape_pipeline.scrape_runs, "update_heartbeat", fake.update_heartbeat)
|
||
monkeypatch.setattr(scrape_pipeline.scrape_runs, "mark_done", fake.mark_done)
|
||
monkeypatch.setattr(scrape_pipeline.scrape_runs, "mark_failed", fake.mark_failed)
|
||
monkeypatch.setattr(scrape_pipeline.scrape_runs, "mark_banned", fake.mark_banned)
|
||
|
||
|
||
@pytest.mark.asyncio
|
||
async def test_run_avito_newbuilding_sweep_saves_and_counts(
|
||
monkeypatch: pytest.MonkeyPatch,
|
||
) -> None:
|
||
fake_runs = _FakeRuns()
|
||
_install_runs(monkeypatch, fake_runs)
|
||
|
||
fake_lots = [_make_lot("N1"), _make_lot("N2"), _make_lot("N3")]
|
||
|
||
async def fake_fetch_newbuildings(
|
||
self: AvitoScraper,
|
||
pages: int = 30,
|
||
*,
|
||
delay_override_sec: float | None = None,
|
||
) -> list[ScrapedLot]:
|
||
return list(fake_lots)
|
||
|
||
monkeypatch.setattr(AvitoScraper, "fetch_newbuildings", fake_fetch_newbuildings)
|
||
|
||
save_calls: list[int] = []
|
||
|
||
def fake_save_listings(_db: Any, lots: Any, **_kw: Any) -> tuple[int, int]:
|
||
save_calls.append(len(lots))
|
||
return (len(lots), 0)
|
||
|
||
monkeypatch.setattr(scrape_pipeline, "save_listings", fake_save_listings)
|
||
# Force cffi-mode (no browser) so we don't spin up BrowserFetcher.
|
||
monkeypatch.setattr(scrape_pipeline.settings, "scraper_fetch_mode", "cffi")
|
||
|
||
# Stub shared AsyncSession (used by the sweep in cffi-mode) — no real session.
|
||
fake_session = AsyncMock()
|
||
fake_session.__aenter__ = AsyncMock(return_value=fake_session)
|
||
fake_session.__aexit__ = AsyncMock(return_value=None)
|
||
monkeypatch.setattr(scrape_pipeline, "AsyncSession", lambda **_kw: fake_session)
|
||
|
||
counters = await scrape_pipeline.run_avito_newbuilding_sweep(
|
||
MagicMock(),
|
||
run_id=42,
|
||
pages=2,
|
||
request_delay_sec=0.0,
|
||
)
|
||
|
||
assert save_calls == [3]
|
||
assert counters.lots_fetched == 3
|
||
assert counters.lots_inserted == 3
|
||
assert counters.lots_updated == 0
|
||
assert fake_runs.done is not None
|
||
assert fake_runs.done["lots_fetched"] == 3
|
||
assert fake_runs.failed is None
|
||
assert fake_runs.banned is None
|