"""Unit tests for YandexRealtyScraper DOM-based SERP parser.""" from datetime import date import pytest from app.services.scrapers.yandex_realty import ( DEFAULT_CITY, MAX_PAGES, YandexRealtyScraper, ) # Realistic single-card fixture (trimmed but selector-faithful). # Note: selectolax.text(strip=True) concatenates all text nodes without # added separators — so floor/price data lives in a single line div with # middot (·) separators, matching real Yandex SERP card layout. # # The address container mirrors the live DOM shape (2026-05): #
# улица Малышева, 125 #
# selectolax sees the trailing text node ", 125" as a sibling of , so # text(strip=False) yields "улица Малышева, 125" while text(strip=True) # collapses whitespace to "улица Малышева,125". SINGLE_CARD_HTML = """
Открыть
улица Фрезеровщиков, 84
ЖК Татлин
36,8 м² · 1-комнатная квартира · 8 этаж из 9 · 4 399 000 ₽ · 119 566 ₽ за м²
торг возможен · опубликовано 17 февраля 2026
""" STUDIO_CARD_HTML = """
x
Студия · 28 м² · 2 этаж из 9 · 3 200 000 ₽
""" NO_PRICE_CARD_HTML = """
x
40 м² · 1-комнатная · 3 этаж из 5
""" # House number with letter suffix (e.g. "2В") — real-world variant LETTER_HOUSE_CARD_HTML = """
x
улица Пехотинцев, 2В
40 м² · 1-комнатная · 3 этаж из 5 · 5 000 000 ₽
""" # House number with slash (e.g. "14/3") — corner building SLASH_HOUSE_CARD_HTML = """
x
улица Академика Ландау, 14/3
50 м² · 2-комнатная · 4 этаж из 9 · 7 000 000 ₽
""" # District / city prefix before the street link (cards in suburbs or # named micro-districts) — must keep the full prefixed address. DISTRICT_PREFIX_CARD_HTML = """
x
Берёзовский, Александровский проспект, 5А
35 м² · 1-комнатная · 2 этаж из 12 · 4 500 000 ₽
""" # Legacy fallback case: no AddressWithGeoLinks container — must still # capture street name via the aggregator link (street-only) to avoid NULL. NO_ADDR_CONTAINER_CARD_HTML = """
x улица Малышева
30 м² · 1-комнатная · 1 этаж из 5 · 3 500 000 ₽
""" EMPTY_PAGE_HTML = "
Нет объявлений
" def test_max_pages_default(): assert MAX_PAGES == 3 def test_default_city(): assert DEFAULT_CITY == "ekaterinburg" def test_build_url_first_page(): s = YandexRealtyScraper() assert s._build_url(page=0) == ( "https://realty.yandex.ru/ekaterinburg/kupit/kvartira/vtorichniy-rynok/" ) def test_build_url_paginated(): s = YandexRealtyScraper() assert s._build_url(page=2) == ( "https://realty.yandex.ru/ekaterinburg/kupit/kvartira/vtorichniy-rynok/?page=2" ) def test_build_url_other_city(): s = YandexRealtyScraper(city="moscow") assert s._build_url(page=0) == ( "https://realty.yandex.ru/moscow/kupit/kvartira/vtorichniy-rynok/" ) def test_parse_html_extracts_card(): s = YandexRealtyScraper() lots = s._parse_html(SINGLE_CARD_HTML, page=0) assert len(lots) == 1 lot = lots[0] assert lot.source == "yandex" assert lot.source_id == "7567094292504417257" assert lot.source_url == "https://realty.yandex.ru/offer/7567094292504417257/" # Address must include the house number (was "улица Фрезеровщиков" only — buggy) assert lot.address == "улица Фрезеровщиков, 84" assert any(ch.isdigit() for ch in lot.address) assert lot.area_m2 == pytest.approx(36.8) assert lot.rooms == 1 assert lot.floor == 8 assert lot.total_floors == 9 assert lot.price_rub == 4_399_000 assert lot.price_per_m2 == 119_566 assert lot.bargain_allowed is True assert lot.listing_date == date(2026, 2, 17) assert lot.house_source == "yandex_realty_nb" assert lot.house_ext_id == "1592987" assert lot.house_url == ( "https://realty.yandex.ru/ekaterinburg/kupit/novostrojka/tatlin-1592987/" ) assert lot.listing_segment == "vtorichka" assert lot.lat is None and lot.lon is None # SERP has no coords # Photos: 2 from yandex CDN, the third (other-domain) skipped assert len(lot.photo_urls) == 2 assert all("avatars.mds.yandex" in u for u in lot.photo_urls) # Size upgrade applied: app_snippet_small → main assert all("app_snippet_small" not in u for u in lot.photo_urls) assert all("main" in u for u in lot.photo_urls) def test_parse_html_studio_rooms_zero(): s = YandexRealtyScraper() lots = s._parse_html(STUDIO_CARD_HTML, page=0) assert len(lots) == 1 assert lots[0].rooms == 0 # studio convention assert lots[0].area_m2 == pytest.approx(28.0) assert lots[0].floor == 2 and lots[0].total_floors == 9 def test_parse_html_no_price_skipped(): s = YandexRealtyScraper() lots = s._parse_html(NO_PRICE_CARD_HTML, page=0) assert lots == [] def test_parse_html_empty_page(): s = YandexRealtyScraper() lots = s._parse_html(EMPTY_PAGE_HTML, page=0) assert lots == [] def test_parse_html_multi_card(): multi = SINGLE_CARD_HTML + STUDIO_CARD_HTML s = YandexRealtyScraper() lots = s._parse_html(multi, page=1) assert len(lots) == 2 assert {lot.source_id for lot in lots} == {"7567094292504417257", "9999999"} assert all(lot.raw_payload["page"] == 1 for lot in lots) def test_address_with_letter_house_number(): """House numbers with letter suffix (e.g. '2В') must round-trip intact.""" s = YandexRealtyScraper() lots = s._parse_html(LETTER_HOUSE_CARD_HTML, page=0) assert len(lots) == 1 assert lots[0].address == "улица Пехотинцев, 2В" def test_address_with_slash_house_number(): """Corner-building numbers like '14/3' must be preserved verbatim.""" s = YandexRealtyScraper() lots = s._parse_html(SLASH_HOUSE_CARD_HTML, page=0) assert len(lots) == 1 assert lots[0].address == "улица Академика Ландау, 14/3" def test_address_keeps_district_prefix(): """Cards with district / city prefix must keep the full prefixed address — the geocoder benefits from disambiguation.""" s = YandexRealtyScraper() lots = s._parse_html(DISTRICT_PREFIX_CARD_HTML, page=0) assert len(lots) == 1 assert lots[0].address == "Берёзовский, Александровский проспект, 5А" def test_address_fallback_to_street_link(): """When the AddressWithGeoLinks container is missing, fall back to the street aggregator link (street-only) — better than NULL.""" s = YandexRealtyScraper() lots = s._parse_html(NO_ADDR_CONTAINER_CARD_HTML, page=0) assert len(lots) == 1 assert lots[0].address == "улица Малышева"