diff --git a/tradein-mvp/backend/app/services/cian_price_history.py b/tradein-mvp/backend/app/services/cian_price_history.py index 9069cdb8..93d155e5 100644 --- a/tradein-mvp/backend/app/services/cian_price_history.py +++ b/tradein-mvp/backend/app/services/cian_price_history.py @@ -15,9 +15,10 @@ import logging import time from dataclasses import dataclass, field -# #2306: fetch_detail/save_detail_enrichment migrated to scraper_kit (golden-parity -# with legacy app.services.scrapers.cian_detail, см. tests/test_scraper_kit_cian_golden_parity.py -# + tests/test_scraper_kit_pricehistory_session_parity.py). RealScraperConfig — тот же +# #2306: fetch_detail/save_detail_enrichment migrated to scraper_kit (byte-identical +# golden-parity была доказана против legacy app.services.scrapers.cian_detail до его +# удаления, #2397 Part E2 — см. tests/test_scraper_kit_pricehistory_session_parity.py +# для оставшейся extract_state/ScrapedLot parity). RealScraperConfig — тот же # read-only адаптер над settings, что и остальные kit-инжекции (#2131) — сохраняет # proxy-поведение (config.cian_proxy_url) идентичным прежнему прямому импорту settings. from scraper_kit.providers.cian.detail import fetch_detail, save_detail_enrichment diff --git a/tradein-mvp/backend/app/services/scrapers/avito_houses.py b/tradein-mvp/backend/app/services/scrapers/avito_houses.py deleted file mode 100644 index 60c238f9..00000000 --- a/tradein-mvp/backend/app/services/scrapers/avito_houses.py +++ /dev/null @@ -1,1392 +0,0 @@ -"""avito_houses.py — Avito Houses Catalog parser (Stage 2c). - -Парсит страницу /catalog/houses//, извлекает window.__preloadedState__ -и разбирает 10 виджетов: housePage, reviews, miniSerp, housePlacementHistory, -recommendations. - -Avito (≈2026, MFE) перенёс полезную нагрузку виджетов на уровень глубже — -под ключ с именем самого виджета (props['reviews']['entries'], -props['miniSerp']['items'], props['housePlacementHistory']['items'] и т.д.). -Разворачивание делает _widget_payload с fallback на старую форму (props напрямую). - -Flow: - fetch_house_catalog(house_url) - → HTTP GET (curl_cffi chrome120) - → regex extract __preloadedState__ - → JS-unescape + json.loads - → parse_houses_state(state, house_url) - → HouseCatalogEnrichment - -Таблицы: - houses (009_houses.sql + 010_houses_alter.sql) - house_reviews (014_house_reviews.sql) - sellers (012_sellers.sql) - listings.seller_id_fk (013_listings_alter_seller.sql) - house_placement_history (017_house_placement_history.sql) -""" - -from __future__ import annotations - -import json -import logging -import re -import urllib.parse -from dataclasses import dataclass, field -from datetime import date -from typing import TYPE_CHECKING, Any - -from sqlalchemy import text -from sqlalchemy.orm import Session - -from app.services.matching.houses import match_or_create_house -from app.services.scrapers.avito import _is_firewall_page -from app.services.scrapers.avito_exceptions import AvitoBlockedError, AvitoRateLimitedError -from app.services.scrapers.avito_shared import RUS_MONTHS, _unix_to_date - -if TYPE_CHECKING: - from app.services.scrapers.browser_fetcher import BrowserFetcher - -logger = logging.getLogger(__name__) - -AVITO_BASE = "https://www.avito.ru" - -# Regex для извлечения __preloadedState__ из HTML. -# -# Avito (≈2026) сменил формат state-блоба: раньше был -# window.__preloadedState__ = JSON.parse(""); -# теперь — URL-encoded строковый литерал: -# window.__preloadedState__ = "%7B%22abCentral%22%3A..."; (%7B='{', %22='"') -# -# PRELOADED_URLENC_RE — текущая форма (декод через urllib.parse.unquote). -# PRELOADED_JSPARSE_RE — старая форма (оставлена как fallback). -PRELOADED_URLENC_RE = re.compile( - r'window\.__preloadedState__\s*=\s*"((?:[^"\\]|\\.)*)"\s*;', - re.DOTALL, -) -PRELOADED_JSPARSE_RE = re.compile( - r'window\.__preloadedState__\s*=\s*JSON\.parse\("(.*?)"\)\s*;', - re.DOTALL, -) - -# Маппинг expandParams type → (field_name, cast_type) -EXPAND_PARAMS_MAP: dict[str, tuple[str, str]] = { - "Год постройки": ("year_built", "int"), - "Этажей": ("total_floors", "int"), - "Тип дома": ("house_type", "house_type"), - "Горячее водоснабжение": ("hot_water", "str"), - "Пассажирский лифт": ("passenger_elevators", "int"), - "Грузовой лифт": ("cargo_elevators", "int"), - "Консьерж": ("has_concierge", "bool_da"), - "Перекрытия": ("material_floors", "str"), - "Парковка": ("parking_type", "str"), - "Детская площадка": ("has_playground", "bool_da"), - "Закрытый двор": ("closed_yard", "bool_da"), - # Новостройки (developmentPage) — другие подписи параметров. Лифты во - # множественном числе; "Этажность" — диапазон ("от 4 до 24"), int-каст - # вернёт None (хранится только в raw_characteristics). - "Пассажирских лифтов": ("passenger_elevators", "int"), - "Грузовых лифтов": ("cargo_elevators", "int"), - # Класс жилья (#2003) — ТОЛЬКО developmentPage (новостройки); housePage - # (вторичка) этот параметр не отдаёт. Сырое значение "Комфорт"/"Бизнес"/... → - # каноничный enum houses.house_class через _normalize_house_class. - "Класс жилья": ("house_class", "house_class"), -} - -# URL-маркер новостройки: /catalog/novostroyki/... (vs вторичка /catalog/houses/...). -NOVOSTROYKA_URL_MARKER = "/catalog/novostroyki/" - -# Нормализация типа дома из Avito текста в DB enum -HOUSE_TYPE_MAP: dict[str, str] = { - "монолитный": "monolith", - "монолит": "monolith", - "панельный": "panel", - "панель": "panel", - "кирпичный": "brick", - "кирпич": "brick", - "монолитно-кирпичный": "monolith_brick", - "блочный": "block", - "деревянный": "wood", -} - -# Нормализация класса жилья из Avito "Класс жилья" в canonical houses.house_class -# (#2003). Avito сам отдаёт только 4 значения (Комфорт/Бизнес/Премиум/Эконом — -# live-подтверждено), но карта дополнена "Комфорт+"/"Элит" для согласованности с -# вокабуляром yandex_helpers.parse_house_class (тот же столбец пишут оба источника). -HOUSE_CLASS_MAP: dict[str, str] = { - "эконом": "economy", - "комфорт": "comfort", - "комфорт+": "comfort_plus", - "комфорт плюс": "comfort_plus", - "бизнес": "business", - "премиум": "premium", - "элит": "elite", - "элитный": "elite", -} - - -# --------------------------------------------------------------------------- -# Dataclasses -# --------------------------------------------------------------------------- - - -@dataclass -class HouseInfo: - ext_id: int # avitoId (int) - ext_id_hash: str | None = None # base64 internal ID - title: str | None = None - short_address: str | None = None # краткий адрес - full_address: str | None = None # полный адрес - lat: float | None = None - lon: float | None = None - year_built: int | None = None - total_floors: int | None = None - house_type: str | None = None # нормализован: monolith/panel/brick/... - house_class: str | None = None # нормализован: comfort/business/premium/economy/... (#2003) - material_floors: str | None = None - hot_water: str | None = None - passenger_elevators: int | None = None - cargo_elevators: int | None = None - has_concierge: bool | None = None - parking_type: str | None = None - has_playground: bool | None = None - closed_yard: bool | None = None - developer_name: str | None = None - developer_key: str | None = None - infrastructure_summary: str | None = None - infrastructure_walk_distance: str | None = None - rating_score: float | None = None - rating_string: str | None = None - reviews_count: int | None = None - rating_distribution: list[dict[str, Any]] = field(default_factory=list) - map_pins: list[dict[str, Any]] = field(default_factory=list) - raw_characteristics: list[dict[str, Any]] = field(default_factory=list) - - -@dataclass -class HouseReview: - ext_review_id: int - author_name: str | None = None - review_title: str | None = None - score: int | None = None - model_experience: str | None = None - rated_date: date | None = None - text_main: str | None = None - text_pros: str | None = None - text_cons: str | None = None - raw_payload: dict[str, Any] | None = None - - -@dataclass -class PlacementHistoryItem: - ext_item_id: str - title: str | None = None - start_price: int | None = None - start_price_date: date | None = None - last_price: int | None = None - last_price_date: date | None = None - exposure_days: int | None = None - removed_date: date | None = None # ВСЕГДА None для source='avito_widget' - raw_payload: dict[str, Any] | None = None - - -@dataclass -class SellerInfo: - ext_seller_id: str - name: str - seller_type: str | None = None - on_platform_since: str | None = None - profile_url: str | None = None - - -@dataclass -class MiniSerpListing: - ext_item_id: int - title: str | None = None - price_rub: int | None = None - description: str | None = None - url: str = "" - metro_text: str | None = None - metro_color: str | None = None - seller: SellerInfo | None = None - - -@dataclass -class RecommendationItem: - title: str | None = None - price_text: str | None = None - address: str | None = None - url: str = "" - image_url: str | None = None - - -@dataclass -class HouseCatalogEnrichment: - house_url: str - house: HouseInfo - reviews: list[HouseReview] = field(default_factory=list) - placement_history: list[PlacementHistoryItem] = field(default_factory=list) - mini_serp: list[MiniSerpListing] = field(default_factory=list) - recommendations: list[RecommendationItem] = field(default_factory=list) - reviews_next_page_url: str | None = None - - -# --------------------------------------------------------------------------- -# Helper functions -# --------------------------------------------------------------------------- - - -def _parse_ru_date(text_val: str | None) -> date | None: - """Парсит дату вида "9 октября 2025" → date(2025, 10, 9).""" - if not text_val: - return None - m = re.match(r"(\d+)\s+([а-я]+)\s+(\d{4})", text_val.strip(), re.IGNORECASE) - if not m: - return None - day = int(m.group(1)) - month_word = m.group(2).lower() - year = int(m.group(3)) - month = RUS_MONTHS.get(month_word) - return date(year, month, day) if month else None - - -def _strip_price(price_str: str | None) -> int | None: - """Извлекает целое число из строки "11 990 000 ₽" → 11990000.""" - if not price_str: - return None - digits = re.sub(r"[^\d]", "", price_str) - return int(digits) if digits else None - - -def _normalize_house_type(raw: str | None) -> str | None: - """Нормализует тип дома: "Монолитный" → "monolith".""" - if not raw: - return None - return HOUSE_TYPE_MAP.get(raw.lower(), "other") - - -def _normalize_house_class(raw: str | None) -> str | None: - """Нормализует класс жилья: "Комфорт" → "comfort" (#2003). - - В отличие от _normalize_house_type, неизвестное значение → None (НЕ "other"): - house_class используется как точный фильтр (`= ANY(...)` в search_query.py) — - мусорный bucket-токен там бесполезен и только засоряет список значений. - """ - if not raw: - return None - return HOUSE_CLASS_MAP.get(raw.strip().lower()) - - -def _cast_expand_param(value: str, cast_type: str) -> Any: - """Кастует строковое значение expandParams к нужному типу.""" - if cast_type == "int": - try: - return int(value) - except (ValueError, TypeError): - return None - if cast_type == "bool_da": - return value.strip().lower() == "да" - if cast_type == "house_type": - return _normalize_house_type(value) - if cast_type == "house_class": - return _normalize_house_class(value) - # str - return value - - -def _parse_expand_params(items: list[dict[str, Any]]) -> dict[str, Any]: - """Извлекает поля из expandParams['items'] по EXPAND_PARAMS_MAP. - - Возвращает dict с распарсенными полями + raw_characteristics (полный список). - """ - result: dict[str, Any] = {} - for category in items: - for param in category.get("params", []): - param_type = param.get("type", "") - if param_type in EXPAND_PARAMS_MAP: - field_name, cast_type = EXPAND_PARAMS_MAP[param_type] - raw_value = param.get("value", "") - result[field_name] = _cast_expand_param(str(raw_value), cast_type) - return result - - -def _flatten_widgets(placeholders: list[dict[str, Any]]) -> list[dict[str, Any]]: - """Возвращает плоский список виджетов из placeholders. - - Avito (≈2026) обернул 10 виджетов в контейнер: placeholders теперь - [{"id": ..., "widgets": [{"type": "housePage", ...}, ...]}] - раньше виджеты лежали в placeholders напрямую: - [{"type": "housePage", ...}, {"type": "reviews", ...}, ...] - - Хелпер раскрывает любой элемент с ключом "widgets" (список) в его виджеты, - остальные элементы оставляет как есть → поддержка обеих форм. - """ - flat: list[dict[str, Any]] = [] - for entry in placeholders: - if isinstance(entry, dict) and isinstance(entry.get("widgets"), list): - flat.extend(w for w in entry["widgets"] if isinstance(w, dict)) - elif isinstance(entry, dict): - flat.append(entry) - return flat - - -def _get_widget(placeholders: list[dict[str, Any]], widget_type: str) -> dict[str, Any] | None: - """Ищет виджет по type в массиве placeholders (уже flat).""" - for w in placeholders: - if w.get("type") == widget_type: - return w - return None - - -def _widget_payload(widget: dict[str, Any], widget_type: str) -> dict[str, Any]: - """Возвращает полезную нагрузку виджета, разворачивая type-вложенность. - - Avito (≈2026, MFE) переместил данные виджета на уровень глубже — - под ключ с именем самого виджета: - новый: widget['props'][widget_type] = {... entries / items ...} - старый: widget['props'] = {... entries / items ...} - - Хелпер пробует новую форму (props[widget_type]), fallback на старую - (props напрямую) → поддержка обеих форм. - """ - props = widget.get("props", {}) - if not isinstance(props, dict): - return {} - nested = props.get(widget_type) - if isinstance(nested, dict): - return nested - return props - - -# --------------------------------------------------------------------------- -# Widget parsers -# --------------------------------------------------------------------------- - - -def _parse_house_page(widget: dict[str, Any]) -> HouseInfo: - """Парсит виджет housePage → HouseInfo. - - developmentData — основной объект с данными дома. Avito (≈2026) изменил путь: - новый: widget['props']['housePage']['developmentData'] - старый: widget['props']['developmentData'] - Аналогично рейтинг переехал в developmentData['ratingPreview'] - (раньше ratingBadge / ratingSummaryStat лежали прямо в developmentData). - Хелпер пробует новые пути с fallback на старые → поддержка обеих форм. - """ - props = widget.get("props", {}) - dd = props.get("housePage", {}).get("developmentData") or props.get("developmentData", {}) - - # expandParams → характеристики дома - about = dd.get("aboutDevelopment", {}) - expand_params_items: list[dict[str, Any]] = about.get("expandParams", {}).get("items", []) - parsed_params = _parse_expand_params(expand_params_items) - - # Рейтинг: новая форма — внутри ratingPreview, старая — прямо в developmentData - rating_root = dd.get("ratingPreview") or dd - rating_badge = rating_root.get("ratingBadge", {}) - rating_info = rating_badge.get("info", {}) - - rating_summary = rating_root.get("ratingSummaryStat", {}) - rating_stat: list[dict[str, Any]] = rating_summary.get("ratingStat", []) - - # Карта/инфраструктура - map_preview = dd.get("mapPreview", {}) - - # Разработчик - developer = dd.get("developer", {}) - - # Координаты - coords = dd.get("coords", {}) - - return HouseInfo( - ext_id=dd.get("avitoId", 0), - ext_id_hash=dd.get("id"), - title=dd.get("title"), - short_address=dd.get("address"), - full_address=dd.get("fullAddress"), - lat=coords.get("lat"), - lon=coords.get("lng"), - year_built=parsed_params.get("year_built"), - total_floors=parsed_params.get("total_floors"), - house_type=parsed_params.get("house_type"), - house_class=parsed_params.get("house_class"), - material_floors=parsed_params.get("material_floors"), - hot_water=parsed_params.get("hot_water"), - passenger_elevators=parsed_params.get("passenger_elevators"), - cargo_elevators=parsed_params.get("cargo_elevators"), - has_concierge=parsed_params.get("has_concierge"), - parking_type=parsed_params.get("parking_type"), - has_playground=parsed_params.get("has_playground"), - closed_yard=parsed_params.get("closed_yard"), - developer_name=developer.get("name"), - developer_key=developer.get("key"), - infrastructure_summary=map_preview.get("objects"), - infrastructure_walk_distance=map_preview.get("distance"), - rating_score=rating_info.get("score"), - rating_string=rating_info.get("scoreString"), - reviews_count=rating_summary.get("reviewCount"), - rating_distribution=rating_stat, - map_pins=map_preview.get("pins", []), - raw_characteristics=expand_params_items, - ) - - -def _is_novostroyka_url(house_url: str) -> bool: - """True если house_url — страница новостройки (/catalog/novostroyki/...). - - Структура новостройки отличается от вторички: вместо виджета 'housePage' - используется 'developmentPage', нет отзывов/истории размещений. - """ - return NOVOSTROYKA_URL_MARKER in house_url - - -def _parse_development_page(widget: dict[str, Any]) -> HouseInfo: - """Парсит виджет 'developmentPage' (новостройка) → HouseInfo. - - Структура отличается от 'housePage' (вторичка): - - данные под props['developmentPage']['developmentData'] (fallback props['developmentData']) - - характеристики не в developmentData.aboutDevelopment.expandParams.items, - а в aboutDevelopment.tabs[*].expandParams.items[*].params (плоский список params). - - НЕТ рейтинга (ratingBadge/ratingSummaryStat), года постройки (дом строится — - есть "Срок сдачи"), отзывов, истории размещений. - Извлекаем то, что есть: id/адрес/координаты/застройщик/инфраструктура + тип дома, - этажность (диапазон → raw_characteristics), лифты из таблицы "Параметры". - """ - props = widget.get("props", {}) - dd = props.get("developmentPage", {}).get("developmentData") or props.get("developmentData", {}) - - # Характеристики новостройки лежат в tabs[*].expandParams.items[*].params. - # Собираем их в форму, совместимую с _parse_expand_params (category → params). - char_categories: list[dict[str, Any]] = [] - about = dd.get("aboutDevelopment", {}) - for tab in about.get("tabs", []): - if not isinstance(tab, dict): - continue - items = tab.get("expandParams", {}).get("items", []) - for item in items: - if isinstance(item, dict) and isinstance(item.get("params"), list): - char_categories.append(item) - parsed_params = _parse_expand_params(char_categories) - - map_preview = dd.get("mapPreview", {}) - developer = dd.get("developer", {}) - coords = dd.get("coords", {}) - - return HouseInfo( - ext_id=dd.get("avitoId", 0), - ext_id_hash=dd.get("id"), - title=dd.get("title"), - short_address=dd.get("address"), - full_address=dd.get("fullAddress") or None, - lat=coords.get("lat"), - lon=coords.get("lng"), - year_built=parsed_params.get("year_built"), # обычно None — дом строится - total_floors=parsed_params.get("total_floors"), - house_type=parsed_params.get("house_type"), - house_class=parsed_params.get("house_class"), - material_floors=parsed_params.get("material_floors"), - hot_water=parsed_params.get("hot_water"), - passenger_elevators=parsed_params.get("passenger_elevators"), - cargo_elevators=parsed_params.get("cargo_elevators"), - has_concierge=parsed_params.get("has_concierge"), - parking_type=parsed_params.get("parking_type"), - has_playground=parsed_params.get("has_playground"), - closed_yard=parsed_params.get("closed_yard"), - developer_name=developer.get("name"), - developer_key=developer.get("key"), - infrastructure_summary=map_preview.get("objects"), - infrastructure_walk_distance=map_preview.get("distance"), - rating_score=None, - rating_string=None, - reviews_count=None, - rating_distribution=[], - map_pins=map_preview.get("pins", []), - raw_characteristics=char_categories, - ) - - -def _parse_single_review(r: dict[str, Any]) -> HouseReview: - """Парсит один объект отзыва из entries.""" - text_main: str | None = None - text_pros: str | None = None - text_cons: str | None = None - - for section in r.get("textSections", []): - title = section.get("title", "") - section_text = section.get("text") - if title == "": - text_main = section_text - elif title == "Преимущества": - text_pros = section_text - elif title == "Недостатки": - text_cons = section_text - - author = r.get("author", {}) - return HouseReview( - ext_review_id=r["id"], - author_name=author.get("title"), - review_title=r.get("reviewTitle"), - score=r.get("score"), - model_experience=r.get("modelExperience"), - rated_date=_parse_ru_date(r.get("rated")), - text_main=text_main, - text_pros=text_pros, - text_cons=text_cons, - raw_payload=r, - ) - - -def _parse_reviews(widget: dict[str, Any]) -> tuple[list[HouseReview], str | None]: - """Парсит виджет reviews → (список HouseReview, следующая страница URL). - - Entries — смешанный список с type='rating' (отзывы) и type='pages' (пагинация). - """ - reviews: list[HouseReview] = [] - next_page_url: str | None = None - - payload = _widget_payload(widget, "reviews") - entries = payload.get("entries", []) - for entry in entries: - entry_type = entry.get("type") - value = entry.get("value", {}) - - if entry_type == "rating": - try: - reviews.append(_parse_single_review(value)) - except (KeyError, TypeError) as exc: - logger.warning("Не удалось распарсить отзыв: %r — %s", value.get("id"), exc) - elif entry_type == "pages": - next_page_url = value.get("nextPageUrl") - - return reviews, next_page_url - - -def _parse_seller(seller_raw: dict[str, Any]) -> SellerInfo | None: - """Парсит seller из miniSerp item → SellerInfo.""" - if not seller_raw: - return None - name = seller_raw.get("name", "") - if not name: - return None - - profile_url = seller_raw.get("url", "") - # ext_seller_id — последний сегмент URL: "/brands/domrf66" → "domrf66" - ext_seller_id = profile_url.strip("/").split("/")[-1] if profile_url else "" - if not ext_seller_id: - logger.warning("Не удалось извлечь ext_seller_id из seller URL: %r", profile_url) - return None - - return SellerInfo( - ext_seller_id=ext_seller_id, - name=name, - seller_type=seller_raw.get("type"), - on_platform_since=seller_raw.get("from"), - profile_url=profile_url or None, - ) - - -def _parse_mini_serp(widget: dict[str, Any]) -> list[MiniSerpListing]: - """Парсит виджет miniSerp → список активных объявлений в доме.""" - listings: list[MiniSerpListing] = [] - - items = _widget_payload(widget, "miniSerp").get("items", []) - for item in items: - geo = item.get("geo", {}) - colors: list[str] = geo.get("colors", []) - - seller_raw = item.get("seller") - seller = _parse_seller(seller_raw) if seller_raw else None - - price_raw = item.get("price") - price_rub = _strip_price(price_raw) if isinstance(price_raw, str) else price_raw - - listings.append( - MiniSerpListing( - ext_item_id=item["id"], - title=item.get("title"), - price_rub=price_rub, - description=item.get("description"), - url=item.get("url", ""), - metro_text=geo.get("content"), - metro_color=colors[0] if colors else None, - seller=seller, - ) - ) - - return listings - - -def _parse_placement_history(widget: dict[str, Any]) -> list[PlacementHistoryItem]: - """Парсит виджет housePlacementHistory → список PlacementHistoryItem. - - ВАЖНО: removed_date ВСЕГДА None — виджет не отдаёт эту дату. - Точная дата только из IMV API (Stage 2d). - """ - result: list[PlacementHistoryItem] = [] - - items = _widget_payload(widget, "housePlacementHistory").get("items", []) - for item in items: - item_copy = {k: v for k, v in item.items() if k != "itemImage"} - result.append( - PlacementHistoryItem( - ext_item_id=str(item["id"]), - title=item.get("title"), - start_price=item.get("startPrice"), - start_price_date=_unix_to_date(item.get("startPriceDate")), - last_price=item.get("lastPrice"), - last_price_date=_unix_to_date(item.get("lastPriceDate")), - exposure_days=item.get("exposure"), - removed_date=None, # widget не отдаёт! IMV API Stage 2d - raw_payload=item_copy, - ) - ) - - return result - - -def _parse_recommendations(widget: dict[str, Any]) -> list[RecommendationItem]: - """Парсит виджет recommendations → lightweight список похожих объявлений.""" - result: list[RecommendationItem] = [] - - items = _widget_payload(widget, "recommendations").get("items", []) - for item in items: - additional_info: list[str] = item.get("additionalInfo", []) - address = additional_info[0] if additional_info else None - - images: list[dict[str, Any]] = item.get("images", []) - image_url: str | None = None - if images: - image_url = images[0].get("208x156") - - result.append( - RecommendationItem( - title=item.get("title"), - price_text=item.get("priceRange"), - address=address, - url=item.get("url", ""), - image_url=image_url, - ) - ) - - return result - - -# --------------------------------------------------------------------------- -# Core parser (pure, no network) -# --------------------------------------------------------------------------- - - -def _extract_preloaded_state(html: str) -> dict[str, Any] | None: - """Извлекает window.__preloadedState__ из HTML → распарсенный dict. - - Pure функция (нет I/O) — тестируется на фикстуре, переиспользуется из - fetch_house_catalog. - - Две формы (см. PRELOADED_*_RE): - 1. URL-encoded строковый литерал (текущая, ≈2026): - window.__preloadedState__ = "%7B%22...%22%7D"; - → urllib.parse.unquote + json.loads - 2. JSON.parse с js-экранированием (старая, fallback): - window.__preloadedState__ = JSON.parse("{\\"...\\"}"); - → ручной unescape + json.loads - - Возвращает dict или None, если маркер не найден / не распарсился. - """ - # Форма 1 — URL-encoded литерал (приоритет) - m = PRELOADED_URLENC_RE.search(html) - if m is not None: - try: - return json.loads(urllib.parse.unquote(m.group(1))) - except (json.JSONDecodeError, ValueError) as exc: - logger.warning("URL-encoded __preloadedState__ не распарсился: %s", exc) - - # Форма 2 — старый JSON.parse("...") (fallback) - m = PRELOADED_JSPARSE_RE.search(html) - if m is not None: - # JS-стиль unescaping: \\" → " , \\\\ → \ , \\/ → / - # НЕ .decode("unicode_escape") — ломает многобайтный UTF-8 (кириллица). - raw = m.group(1).replace('\\"', '"').replace("\\\\", "\\").replace("\\/", "/") - try: - return json.loads(raw) - except json.JSONDecodeError: - logger.warning("JSON.parse-форма с replace-unescape не удалась, пробуем unicode_escape") - try: - return json.loads(bytes(raw, "utf-8").decode("unicode_escape")) - except (json.JSONDecodeError, UnicodeDecodeError) as exc2: - logger.warning("JSON.parse-форма __preloadedState__ не распарсилась: %s", exc2) - - return None - - -def _parse_development_or_skip( - placeholders: list[dict[str, Any]], house_url: str -) -> HouseCatalogEnrichment: - """Обрабатывает страницу без виджета 'housePage' (новостройка или неизвестная форма). - - - Есть виджет 'developmentPage' → парсим его (partial: дом/застройщик/тип/лифты, - без отзывов/истории/рейтинга). - - Нет developmentPage, но URL — новостройка → graceful skip: warning + минимальное - обогащение (ext_id из URL, без характеристик). НЕ кидаем ValueError, чтобы один - незнакомый шейп новостройки не валил весь enrichment-loop pipeline. - - Иначе (вторичка без housePage) → ValueError (регрессия шейпа вторички). - """ - dev_widget = _get_widget(placeholders, "developmentPage") - if dev_widget is not None: - house_info = _parse_development_page(dev_widget) - logger.info( - "developmentPage (новостройка) распарсен: ext_id=%s type=%s url=%s", - house_info.ext_id, - house_info.house_type, - house_url, - ) - return HouseCatalogEnrichment(house_url=house_url, house=house_info) - - if _is_novostroyka_url(house_url): - logger.warning( - "Новостройка без виджета 'developmentPage'/'housePage' (%s) — " - "graceful skip, обогащение пропущено", - house_url, - ) - ext_id = _ext_id_from_url(house_url) - return HouseCatalogEnrichment(house_url=house_url, house=HouseInfo(ext_id=ext_id)) - - raise ValueError("Виджет 'housePage' не найден в placeholders") - - -def _ext_id_from_url(house_url: str) -> int: - """Извлекает числовой avitoId из хвоста house_url (...//). 0 если нет.""" - m = re.search(r"/(\d+)/?(?:\?|$)", house_url) - return int(m.group(1)) if m else 0 - - -def parse_houses_state(state: dict[str, Any], house_url: str) -> HouseCatalogEnrichment: - """Парсит уже разобранный __preloadedState__ dict → HouseCatalogEnrichment. - - Pure функция: нет HTTP, нет I/O. Принимает state dict и house_url (для поля - HouseCatalogEnrichment.house_url). Используется в тестах и из fetch_house_catalog. - - Новостройки (/catalog/novostroyki/...) не имеют виджета 'housePage' — вместо него - 'developmentPage'. Для них парсим что есть (id/адрес/застройщик/тип дома/лифты) и - возвращаем partial-обогащение без отзывов/истории. Если структура совсем не - распознана — graceful: warning + минимальное обогащение (НЕ ValueError). - - Raises: - ValueError если placeholders не массив, или (только для вторички) housePage - виджет не найден. - """ - try: - raw_placeholders: list[dict[str, Any]] = state["data"]["data"]["page"]["placeholders"] - except (KeyError, TypeError) as exc: - raise ValueError(f"Не удалось найти placeholders в __preloadedState__: {exc}") from exc - - if not isinstance(raw_placeholders, list): - raise ValueError( - f"placeholders должен быть массивом, получен: {type(raw_placeholders).__name__}" - ) - - # Avito (≈2026) оборачивает виджеты в контейнер {id, widgets:[...]} — - # раскрываем в плоский список (поддержка старой и новой формы). - placeholders = _flatten_widgets(raw_placeholders) - - # housePage — обязательный виджет для вторички. Новостройки используют - # 'developmentPage' и не имеют housePage — обрабатываем отдельно (graceful). - house_page_widget = _get_widget(placeholders, "housePage") - if house_page_widget is None: - return _parse_development_or_skip(placeholders, house_url) - - house_info = _parse_house_page(house_page_widget) - - # reviews — опциональный - reviews: list[HouseReview] = [] - next_page_url: str | None = None - reviews_widget = _get_widget(placeholders, "reviews") - if reviews_widget is not None: - reviews, next_page_url = _parse_reviews(reviews_widget) - if not reviews: - logger.warning("Виджет 'reviews' найден, но 0 отзывов распарсено — проверь шейп props") - # #1789: рейтинг переехал из housePage.developmentData.ratingPreview в - # reviews.filterEntries[type=="score"]. Применяем как fallback — только - # если housePage не дал рейтинг (чтобы не затирать уже правильный источник). - if house_info.rating_score is None: - reviews_payload = _widget_payload(reviews_widget, "reviews") - filter_entries: list[dict[str, Any]] = reviews_payload.get("filterEntries", []) - for fe in filter_entries: - if fe.get("type") == "score": - score_value = fe.get("value", {}) - if isinstance(score_value, dict): - score_float = score_value.get("scoreFloat") - if score_float is not None: - house_info.rating_score = float(score_float) - house_info.rating_string = score_value.get("scoreString") - rc = score_value.get("reviewCount") - if rc is not None: - house_info.reviews_count = int(rc) - rating_stat = score_value.get("ratingStat") - if isinstance(rating_stat, list): - house_info.rating_distribution = rating_stat - logger.info( - "parse_houses_state: rating из reviews.filterEntries " - "url=%s score=%.3f string=%s reviews=%s", - house_url, - house_info.rating_score, - house_info.rating_string, - house_info.reviews_count, - ) - break - else: - logger.debug("Виджет 'reviews' не найден — пропускаем") - - # miniSerp — опциональный - mini_serp: list[MiniSerpListing] = [] - mini_serp_widget = _get_widget(placeholders, "miniSerp") - if mini_serp_widget is not None: - mini_serp = _parse_mini_serp(mini_serp_widget) - else: - logger.debug("Виджет 'miniSerp' не найден — пропускаем") - - # housePlacementHistory — опциональный - placement_history: list[PlacementHistoryItem] = [] - history_widget = _get_widget(placeholders, "housePlacementHistory") - if history_widget is not None: - placement_history = _parse_placement_history(history_widget) - if not placement_history: - logger.warning( - "Виджет 'housePlacementHistory' найден, но 0 items распарсено — проверь шейп props" - ) - else: - logger.debug("Виджет 'housePlacementHistory' не найден — пропускаем") - - # recommendations — опциональный - recommendations: list[RecommendationItem] = [] - recs_widget = _get_widget(placeholders, "recommendations") - if recs_widget is not None: - recommendations = _parse_recommendations(recs_widget) - else: - logger.debug("Виджет 'recommendations' не найден — пропускаем") - - return HouseCatalogEnrichment( - house_url=house_url, - house=house_info, - reviews=reviews, - placement_history=placement_history, - mini_serp=mini_serp, - recommendations=recommendations, - reviews_next_page_url=next_page_url, - ) - - -# --------------------------------------------------------------------------- -# HTTP fetch (thin wrapper, requires curl_cffi) -# --------------------------------------------------------------------------- - - -async def fetch_house_catalog( - house_url: str, - *, - cffi_session: Any | None = None, - browser_fetcher: BrowserFetcher | None = None, -) -> HouseCatalogEnrichment: - """GET {AVITO_BASE}{house_url} → extract window.__preloadedState__ → parse 10 виджетов. - - Параметры: - house_url — относительный путь: - /catalog/houses/ekaterinburg/ul_postovskogo/3171365 - cffi_session — опциональная curl_cffi AsyncSession (для переиспользования пула соединений). - Если None — создаётся временная сессия. - browser_fetcher — если передан, используется браузерный fetch вместо curl_cffi. - - Raises: - RuntimeError если curl_cffi не установлен (в curl-режиме). - httpx.HTTPStatusError / curl_cffi аналог если status != 200. - ValueError если __preloadedState__ не найден или placeholders не массив. - """ - url = f"{AVITO_BASE}{house_url}" - logger.info("Houses Catalog fetch: %s", url) - - if browser_fetcher is not None: - html = await browser_fetcher.fetch(url) - if _is_firewall_page(html): - raise AvitoBlockedError(f"Avito houses firewall (browser-mode) for {url}") - state = _extract_preloaded_state(html) - if state is None: - raise ValueError( - f"window.__preloadedState__ не найден / не распарсился в HTML страницы: {url}" - ) - logger.info("Houses Catalog state extracted (browser-mode), parsing widgets") - return parse_houses_state(state, house_url) - - try: - from curl_cffi.requests import AsyncSession as CffiAsyncSession - except ImportError as exc: - raise RuntimeError( - "curl_cffi не установлен. Добавь 'curl-cffi>=0.7.0' в pyproject.toml." - ) from exc - - _own_session = False - if cffi_session is None: - cffi_session = CffiAsyncSession(impersonate="chrome120") - _own_session = True - - try: - resp = await cffi_session.get(url) - if resp.status_code == 403: - raise AvitoBlockedError(f"Avito houses HTTP 403 for {url}") - if resp.status_code == 429: - raise AvitoRateLimitedError(f"Avito houses HTTP 429 for {url}") - resp.raise_for_status() - html = resp.text - - # Извлекаем + декодируем window.__preloadedState__ (URL-encoded или JSON.parse). - state = _extract_preloaded_state(html) - if state is None: - raise ValueError( - f"window.__preloadedState__ не найден / не распарсился в HTML страницы: {url}" - ) - - logger.info("Houses Catalog state extracted, parsing widgets") - return parse_houses_state(state, house_url) - - finally: - if _own_session: - await cffi_session.close() - - -# --------------------------------------------------------------------------- -# Save functions -# --------------------------------------------------------------------------- - - -def upsert_house(db: Session, h: HouseInfo) -> int | None: - """Match-or-create канонический дом через match_or_create_house, затем обновить enrichment. - - Source='avito'. Возвращает houses.id, либо None если matcher отказал (P1: - безномерный адрес без кадастра, method='no_house_number'). - Делегирует в _persist_house (без url — url пуст как раньше для этого пути). - - Маппинг полей HouseInfo → DDL колонки (009 + 010): - ext_id → ext_house_id (text) - ext_id_hash → avito_id_hash - short_address → short_address (010) - full_address → full_address (010) - address (DDL) → используем short_address как fallback - rating_score → rating_score (010, numeric(4,2)) - rating (DDL) → округлённый numeric(2,1) из rating_score - """ - return _persist_house(db, h, "") - - -def _upsert_house_with_url(db: Session, h: HouseInfo, house_url: str) -> int | None: - """Внутренний вариант upsert_house с правильным URL. - - Делегирует в _persist_house — используй его вместо прямого вызова. - Возвращает None если matcher отказал (P1 numberless без кадастра). - """ - return _persist_house(db, h, house_url) - - -def _persist_house(db: Session, h: HouseInfo, house_url: str) -> int | None: - """Канонический путь сохранения дома-avito через match_or_create_house. - - Шаги: - 1. match_or_create_house — находит или создаёт канонический houses-ряд - (3-tier: source_exact → fingerprint → geo → new), защищён advisory lock. - 2. UPDATE houses SET ... — записывает все avito-enrichment поля в найденный/новый дом. - - Avito-авторитетные поля (rating*, developer*, технические хар-ки, url, - avito_id_hash) перезаписываются напрямую. - - address/lat/lon/year_built — через COALESCE чтобы не затирать данные - канонического дома от cian/кадастра. - - house_class (#2003) — COALESCE(new, existing): свежее avito-значение - (топ-приоритет источник) побеждает, но None не затирает значение от - более низкоприоритетного источника (см. HOUSE_FIELD_PRIORITY). - 3. Не вызывает db.commit() — коммитит вызывающий save_house_catalog_enrichment - (advisory lock xact-scoped, должен держаться до commit). - - Returns: - houses.id (int) канонического дома. - """ - house_id, _conf, _method = match_or_create_house( - db, - ext_source="avito", - ext_id=str(h.ext_id), - # short preferred: aliases have no city/region prefix; full is last resort - address=h.short_address or h.full_address, - lat=h.lat, - lon=h.lon, - year_built=h.year_built, - source_url=house_url or None, - ) - - # P1: matcher refuses a numberless address without a cadastral number - # (method 'no_house_number' → house_id None). No canonical row exists to enrich, - # so skip the UPDATE and signal the graceful skip up to save_house_catalog_enrichment. - if house_id is None: - logger.info( - "_persist_house: skip numberless avito house ext_id=%s addr=%r (no cadastral)", - h.ext_id, - h.short_address or h.full_address, - ) - return None - - db.execute( - text(""" - UPDATE houses SET - url = CAST(:url AS text), - short_address = CAST(:short_address AS text), - full_address = CAST(:full_address AS text), - avito_id_hash = CAST(:avito_id_hash AS text), - house_type = CAST(:house_type AS text), - -- house_class (#2003): avito_houses_catalog топ-приоритет источник - -- (conflict_resolution.HOUSE_FIELD_PRIORITY) — свежее avito-значение - -- всегда побеждает, но None (вторичка не отдаёт "Класс жилья") не - -- затирает значение, уже проставленное более низкоприоритетным - -- источником (напр. yandex_realty_nb). - house_class = COALESCE(CAST(:house_class AS text), houses.house_class), - total_floors = CAST(:total_floors AS int), - passenger_elevators = CAST(:passenger_elevators AS int), - cargo_elevators = CAST(:cargo_elevators AS int), - has_concierge = CAST(:has_concierge AS boolean), - closed_yard = CAST(:closed_yard AS boolean), - material_floors = CAST(:material_floors AS text), - hot_water = CAST(:hot_water AS text), - parking_type = CAST(:parking_type AS text), - has_playground = CAST(:has_playground AS boolean), - developer_name = CAST(:developer_name AS text), - developer_key = CAST(:developer_key AS text), - infrastructure_summary = CAST(:infrastructure_summary AS text), - infrastructure_walk_distance = CAST(:infrastructure_walk_distance AS text), - rating = CAST(:rating AS numeric), - rating_score = CAST(:rating_score AS numeric), - rating_string = CAST(:rating_string AS text), - reviews_count = CAST(:reviews_count AS int), - rating_distribution = CAST(:rating_distribution AS jsonb), - map_pins = CAST(:map_pins AS jsonb), - raw_characteristics = CAST(:raw_characteristics AS jsonb), - address = COALESCE(houses.address, CAST(:address AS text)), - lat = COALESCE(houses.lat, CAST(:lat AS double precision)), - lon = COALESCE(houses.lon, CAST(:lon AS double precision)), - year_built = COALESCE(houses.year_built, CAST(:year_built AS int)), - last_scraped_at = NOW(), - avito_validated_at = NOW() - WHERE id = CAST(:house_id AS bigint) - """), - { - "house_id": house_id, - "url": house_url or "", - "short_address": h.short_address, - "full_address": h.full_address, - "avito_id_hash": h.ext_id_hash, - "house_type": h.house_type, - "house_class": h.house_class, - "total_floors": h.total_floors, - "passenger_elevators": h.passenger_elevators, - "cargo_elevators": h.cargo_elevators, - "has_concierge": h.has_concierge, - "closed_yard": h.closed_yard, - "material_floors": h.material_floors, - "hot_water": h.hot_water, - "parking_type": h.parking_type, - "has_playground": h.has_playground, - "developer_name": h.developer_name, - "developer_key": h.developer_key, - "infrastructure_summary": h.infrastructure_summary, - "infrastructure_walk_distance": h.infrastructure_walk_distance, - # rating (numeric(2,1)) — округлённое значение для совместимости с 009 DDL - "rating": round(h.rating_score, 1) if h.rating_score is not None else None, - "rating_score": h.rating_score, - "rating_string": h.rating_string, - "reviews_count": h.reviews_count, - "rating_distribution": json.dumps(h.rating_distribution, ensure_ascii=False), - "map_pins": json.dumps(h.map_pins, ensure_ascii=False), - "raw_characteristics": json.dumps(h.raw_characteristics, ensure_ascii=False), - "address": h.short_address, - "lat": h.lat, - "lon": h.lon, - "year_built": h.year_built, - }, - ) - - logger.info( - "_persist_house: ext_id=%s url=%s → houses.id=%s (method=%s)", - h.ext_id, - house_url, - house_id, - _method, - ) - return int(house_id) - - -def save_house_reviews(db: Session, house_id: int, reviews: list[HouseReview]) -> int: - """INSERT INTO house_reviews ON CONFLICT (source, ext_review_id) DO UPDATE. - - Source='avito'. Возвращает количество вставленных/обновлённых строк. - """ - if not reviews: - return 0 - - count = 0 - for r in reviews: - db.execute( - text(""" - INSERT INTO house_reviews ( - house_id, source, ext_review_id, - author_name, review_title, score, - model_experience, rated_date, - text_main, text_pros, text_cons, - raw_payload, scraped_at - ) VALUES ( - CAST(:house_id AS bigint), 'avito', CAST(:ext_review_id AS bigint), - CAST(:author_name AS text), CAST(:review_title AS text), CAST(:score AS int), - CAST(:model_experience AS text), CAST(:rated_date AS date), - CAST(:text_main AS text), CAST(:text_pros AS text), CAST(:text_cons AS text), - CAST(:raw_payload AS jsonb), NOW() - ) - ON CONFLICT (source, ext_review_id) DO UPDATE SET - author_name = EXCLUDED.author_name, - review_title = EXCLUDED.review_title, - score = EXCLUDED.score, - model_experience = EXCLUDED.model_experience, - rated_date = EXCLUDED.rated_date, - text_main = EXCLUDED.text_main, - text_pros = EXCLUDED.text_pros, - text_cons = EXCLUDED.text_cons, - raw_payload = EXCLUDED.raw_payload, - scraped_at = NOW() - """), - { - "house_id": house_id, - "ext_review_id": r.ext_review_id, - "author_name": r.author_name, - "review_title": r.review_title, - "score": r.score, - "model_experience": r.model_experience, - "rated_date": r.rated_date, - "text_main": r.text_main, - "text_pros": r.text_pros, - "text_cons": r.text_cons, - "raw_payload": ( - json.dumps(r.raw_payload, ensure_ascii=False) if r.raw_payload else None - ), - }, - ) - count += 1 - - logger.info("save_house_reviews: house_id=%s, saved=%d", house_id, count) - return count - - -def save_sellers(db: Session, sellers: list[SellerInfo]) -> dict[str, int]: - """INSERT INTO sellers ON CONFLICT (source, ext_seller_id) DO UPDATE. - - Source='avito'. Возвращает {ext_seller_id: sellers.id}. - """ - result: dict[str, int] = {} - for s in sellers: - row = db.execute( - text(""" - INSERT INTO sellers ( - source, ext_seller_id, - name, seller_type, on_platform_since, profile_url, - last_seen_at - ) VALUES ( - 'avito', CAST(:ext_seller_id AS text), - CAST(:name AS text), CAST(:seller_type AS text), - CAST(:on_platform_since AS text), CAST(:profile_url AS text), - NOW() - ) - ON CONFLICT (source, ext_seller_id) DO UPDATE SET - name = EXCLUDED.name, - seller_type = EXCLUDED.seller_type, - on_platform_since = EXCLUDED.on_platform_since, - profile_url = EXCLUDED.profile_url, - last_seen_at = NOW() - RETURNING id - """), - { - "ext_seller_id": s.ext_seller_id, - "name": s.name, - "seller_type": s.seller_type, - "on_platform_since": s.on_platform_since, - "profile_url": s.profile_url, - }, - ) - seller_db_id: int = row.scalar_one() - result[s.ext_seller_id] = seller_db_id - - logger.info("save_sellers: saved=%d", len(result)) - return result - - -def link_listings_to_seller(db: Session, ext_item_id_to_seller_db_id: dict[int, int]) -> int: - """UPDATE listings SET seller_id_fk = :seller_id WHERE source='avito' AND source_id = :item_id. - - Возвращает количество обновлённых строк. - """ - if not ext_item_id_to_seller_db_id: - return 0 - - total = 0 - for item_id, seller_db_id in ext_item_id_to_seller_db_id.items(): - result = db.execute( - text(""" - UPDATE listings - SET seller_id_fk = CAST(:seller_id AS bigint) - WHERE source = 'avito' - AND source_id = CAST(:item_id AS text) - AND seller_id_fk IS DISTINCT FROM CAST(:seller_id AS bigint) - """), - {"seller_id": seller_db_id, "item_id": str(item_id)}, - ) - total += result.rowcount - - logger.info("link_listings_to_seller: updated=%d listings", total) - return total - - -def save_placement_history(db: Session, items: list[PlacementHistoryItem]) -> int: - """INSERT INTO house_placement_history с source='avito_widget' и removed_date=NULL. - - ON CONFLICT (source, ext_item_id) DO UPDATE last_price/last_price_date/exposure_days. - Возвращает количество строк. - """ - if not items: - return 0 - - count = 0 - for item in items: - db.execute( - text(""" - INSERT INTO house_placement_history ( - source, ext_item_id, - title, - start_price, start_price_date, - last_price, last_price_date, - exposure_days, - removed_date, - raw_payload, scraped_at - ) VALUES ( - 'avito_widget', CAST(:ext_item_id AS text), - CAST(:title AS text), - CAST(:start_price AS bigint), CAST(:start_price_date AS date), - CAST(:last_price AS bigint), CAST(:last_price_date AS date), - CAST(:exposure_days AS int), - NULL, - CAST(:raw_payload AS jsonb), NOW() - ) - ON CONFLICT (source, ext_item_id) DO UPDATE SET - title = EXCLUDED.title, - last_price = EXCLUDED.last_price, - last_price_date = EXCLUDED.last_price_date, - exposure_days = EXCLUDED.exposure_days, - raw_payload = EXCLUDED.raw_payload, - scraped_at = NOW() - """), - { - "ext_item_id": item.ext_item_id, - "title": item.title, - "start_price": item.start_price, - "start_price_date": item.start_price_date, - "last_price": item.last_price, - "last_price_date": item.last_price_date, - "exposure_days": item.exposure_days, - "raw_payload": ( - json.dumps(item.raw_payload, ensure_ascii=False) if item.raw_payload else None - ), - }, - ) - count += 1 - - logger.info("save_placement_history: saved=%d items (source=avito_widget)", count) - return count - - -def save_house_catalog_enrichment( - db: Session, - e: HouseCatalogEnrichment, -) -> dict[str, int]: - """Orchestrator — полный pipeline сохранения одного обогащения. - - Операции (в порядке): - 1. upsert_house → houses.id - 2. save_house_reviews(house_id, reviews) - 3. save_sellers([m.seller for m in mini_serp if m.seller]) → {ext_seller_id: db_id} - 4. link_listings_to_seller({item_id: seller_db_id}) - 5. save_placement_history(placement_history) - - Возвращает: - {'house_id': N, 'reviews': N, 'sellers': N, 'listings_linked': N, 'placement_history': N} - """ - # 1. Upsert дом с правильным URL из enrichment. - # Гард #1820: id-less graceful-skip новостройки (ext_id=0, нет адреса/координат) — - # НЕ персистим canonical house, иначе все безымянные URL склеиваются в один bogus - # avito:0 через source_exact-tier и плодят мусорную house_sources("avito","0"). - if not e.house.ext_id and not (e.house.short_address or e.house.full_address): - logger.warning( - "save_house_catalog_enrichment: пропуск id-less новостройки без адреса" - " (ext_id=%s url=%s) — не создаём bogus avito:0 в house_sources", - e.house.ext_id, - e.house_url, - ) - return { - "house_id": 0, - "reviews": 0, - "sellers": 0, - "listings_linked": 0, - "placement_history": 0, - } - house_id = _upsert_house_with_url(db, e.house, e.house_url) - - # P1: matcher refused to create a canonical house for a numberless address without a - # cadastral number (house_id None). Downstream saves (reviews / placement history) are - # house-scoped and meaningless with a NULL house_id, so graceful-skip the whole - # enrichment — same zero-counter contract as the id-less guard above. - if house_id is None: - logger.warning( - "save_house_catalog_enrichment: пропуск дома с безномерным адресом без кадастра" - " (ext_id=%s addr=%r) — matcher вернул no_house_number", - e.house.ext_id, - e.house.short_address or e.house.full_address, - ) - return { - "house_id": 0, - "reviews": 0, - "sellers": 0, - "listings_linked": 0, - "placement_history": 0, - } - - # 2. Сохранить отзывы - reviews_count = save_house_reviews(db, house_id, e.reviews) - - # 3. Сохранить продавцов (дедуп по ext_seller_id) - sellers_to_save = [m.seller for m in e.mini_serp if m.seller is not None] - seen_seller_ids: set[str] = set() - unique_sellers: list[SellerInfo] = [] - for s in sellers_to_save: - if s.ext_seller_id not in seen_seller_ids: - unique_sellers.append(s) - seen_seller_ids.add(s.ext_seller_id) - - seller_id_map = save_sellers(db, unique_sellers) - - # 4. Слинковать объявления с продавцами - # Маппинг: ext_item_id (int) → seller db_id - item_to_seller: dict[int, int] = {} - for listing in e.mini_serp: - if listing.seller and listing.seller.ext_seller_id in seller_id_map: - item_to_seller[listing.ext_item_id] = seller_id_map[listing.seller.ext_seller_id] - - listings_linked = link_listings_to_seller(db, item_to_seller) - - # 5. Сохранить историю размещений - history_count = save_placement_history(db, e.placement_history) - - db.commit() - - result = { - "house_id": house_id, - "reviews": reviews_count, - "sellers": len(seller_id_map), - "listings_linked": listings_linked, - "placement_history": history_count, - } - logger.info("save_house_catalog_enrichment: %s", result) - return result diff --git a/tradein-mvp/backend/app/services/scrapers/cian.py b/tradein-mvp/backend/app/services/scrapers/cian.py deleted file mode 100644 index 4b514c6a..00000000 --- a/tradein-mvp/backend/app/services/scrapers/cian.py +++ /dev/null @@ -1,1185 +0,0 @@ -"""Cian.ru scraper — state-based SERP parser (137 fields per offer). - -Стратегия (Stage 3 рефактор): -- Cian React micro-frontends хранят state в window._cianConfig['frontend-serp'].push(). -- URL pattern: https://ekb.cian.ru/cat.php?deal_type=sale&offer_type=flat&engine_version=2 - ekb.cian.ru — city-specific subdomain для ЕКБ (per Schema_Cian_SERP_Inventory sec 13). - -ВАЖНО: Циан блокирует httpx/curl по TLS+fingerprint и сыплет Google reCAPTCHA на -датацентр-IP. Транспорт SERP переведён на BrowserFetcher (camoufox real-browser -fingerprint через tradein-browser сервис, #1806 / epic #883 Phase 2). Браузер -отдаёт полный SSR-HTML с Redux initialState — парсинг state не меняется. -Мобильный прокси (ha.mobileproxy.space) применяется server-side в браузер-контейнере -и ротирует IP сам, поэтому code-side _rotate_ip (changeip) больше не нужен. - -НЕ используем anchor jitter: offer.geo.coordinates.{lat,lng} — точные координаты -прямо из SERP state. Jitter запрещён per implementation plan. - -Exhaustive load (fetch_all_secondary): -- Cian SERP — регион-wide (не geo-bbox). Anchor-loop в существующем city-sweep - избыточен для Cian (все anchor'ы дают одну выдачу). Полный сбор = ОДИН проход - с партиционированием по КОМНАТНОСТИ × ЦЕНЕ (адаптивное бинарное деление диапазона). -""" - -from __future__ import annotations - -import asyncio -import hashlib -import inspect -import logging -import math -import re -from collections.abc import Callable -from datetime import UTC, datetime -from typing import Any -from urllib.parse import urlencode - -import sentry_sdk - -from app.core.config import settings -from app.services.scraper_settings import get_scraper_delay -from app.services.scrapers.base import BaseScraper, ScrapedLot -from app.services.scrapers.browser_fetcher import BrowserFetcher -from app.services.scrapers.cian_state_parser import extract_state -from app.services.scrapers.house_type_normalizer import normalize_house_type -from app.services.scrapers.price_brackets import get_price_seed_brackets -from app.services.scrapers.repair_state_normalizer import ( - infer_repair_state_from_text, - normalize_repair_state, -) - -logger = logging.getLogger(__name__) - -# Регион 4743 = Свердловская область (Cian internal region ID) -CIAN_EKB_REGION_ID = 4743 - -# SERP MFE name и state key (per Schema sec 1.2) -_MFE_SERP = "frontend-serp" -_STATE_KEY = "initialState" - -# ── Константы exhaustive-загрузки ────────────────────────────────────────────── - -# Верхняя граница цены при первом рекурсивном делении (нет явного hi). -# 200 млн ₽ заведомо выше ТОПа ЕКБ — дальние бакеты дадут 0 офферов. -_MAX_PRICE = 200_000_000 - -# Минимальный ценовой диапазон для рекурсии; при hi - lo < MIN_BRACKET -# прекращаем деление и пагинируем как есть (принимаем возможный «хвост»). -_MIN_BRACKET = 50_000 - -# Cian SERP показывает ~28 офферов на страницу. -_CIAN_OFFERS_PER_PAGE = 28 - -# Комнатности по умолчанию для exhaustive-загрузки. -# Коды Cian: room1=1к, room2=2к, room3=3к, room4=4к, room5=5к, room6=6+. -# TODO: студии — Cian использует отдельный тип flat_type=studio, не room-param. -# Если добавить room9 — запрос может давать 0 офферов или смешивать типы. -# До верификации кода студии — студии НЕ включаем в exhaustive-загрузку. -_DEFAULT_ROOMS_BUCKETS: list[tuple[int, ...]] = [ - (1,), - (2,), - (3,), - (4,), - (5,), - (6,), -] - - -class CianScraper(BaseScraper): - """Cian SERP scraper. Тянет SSR-HTML через BrowserFetcher (camoufox). - - Извлекает 137 полей на offer из Redux initialState через cian_state_parser. - Координаты точные — НЕТ anchor jitter (offer.geo.coordinates). - - Транспорт: одна BrowserFetcher(source="cian") сессия открывается в __aenter__ - и переиспользуется на всех page-fetch'ах (#1806). camoufox-fingerprint + мобильный - прокси server-side обходят Cian reCAPTCHA; IP ротирует сам прокси. - """ - - name = "cian" - # ekb.cian.ru — city-specific subdomain для ЕКБ (per Schema sec 13, closed Q4) - base_url = "https://ekb.cian.ru" - # Класс-дефолт; реальное значение загружается из scraper_settings при создании экземпляра. - request_delay_sec = 5.0 # консервативно: Cian менее агрессивен чем Avito, но 5s безопасно - - def __init__(self) -> None: - super().__init__() - self.request_delay_sec = get_scraper_delay(self.name) - self._browser: BrowserFetcher | None = None - - async def __aenter__(self) -> CianScraper: - await super().__aenter__() - # Одна BrowserFetcher(source="cian") сессия на весь scrape — camoufox real-browser - # fingerprint через tradein-browser сервис (#1806). Сервер роутит /fetch на - # отдельный браузер+мобильный прокси для cian. Прокси ротирует IP сам, поэтому - # code-side warm-up cookies и changeip-ротация больше не нужны. - self._browser = BrowserFetcher(source="cian") - await self._browser.__aenter__() - return self - - async def __aexit__(self, *args: Any) -> None: - if self._browser is not None: - await self._browser.__aexit__(*args) - self._browser = None - await super().__aexit__(*args) - - # ── Anti-block: IP rotation ─────────────────────────────────────────────── - - async def _rotate_ip(self) -> bool: - """No-op после миграции на BrowserFetcher (#1806). - - Мобильный прокси в браузер-контейнере ротирует IP сам, code-side changeip - больше не нужен. Метод сохранён (returns False) для обратной совместимости - с вызовом в _walk_price_range и тестами, которые его мокают. - """ - return False - - async def fetch_around( - self, - lat: float, - lon: float, - radius_m: int = 1000, - rooms: tuple[int, ...] | None = None, - page: int = 1, - ) -> list[ScrapedLot]: - """Найти объявления Циан по ЕКБ (lat/lon для reference; Cian не поддерживает bbox). - - Cian SERP отдаёт весь ЕКБ — фильтрация по радиусу происходит в postgres - через ST_DWithin после сохранения (точные coords из state). - - rooms — список из 1,2,3,4 (Cian's room codes). Если None — все. - page — страница выдачи (Cian ~28 объявлений на страницу). - """ - url = self._build_url(rooms, page) - try: - assert self._browser is not None - html = await self._browser.fetch(url) - except Exception: - logger.exception("cian BrowserFetcher fetch failed for url=%s", url) - return [] - - lots = self._parse_serp_html(html) - logger.info( - "cian: %d lots fetched rooms=%s page=%d url=%s", - len(lots), - rooms, - page, - url, - ) - await self.sleep_between_requests() - return lots - - async def fetch_around_multi_room( - self, - lat: float, - lon: float, - radius_m: int = 1000, - pages: int = 8, - ) -> list[ScrapedLot]: - """Скрейп Циан по 1к/2к/3к/4+ × N страниц — расширяет выборку. - - 4 комнаты × N страниц × ~28 = до ~330+ лотов до дедупликации. - """ - seen: dict[str, ScrapedLot] = {} - for rooms in ((1,), (2,), (3,), (4,)): - for page in range(1, pages + 1): - try: - lots = await self.fetch_around(lat, lon, radius_m, rooms=rooms, page=page) - except Exception: - logger.exception("cian multi-room fetch failed rooms=%s page=%d", rooms, page) - continue - if not lots: - break # пустая страница → дальше смысла нет - for lot in lots: - key = lot.source_id or lot.source_url - if key and key not in seen: - seen[key] = lot - logger.info( - "cian multi-room: %d unique lots (lat=%.4f lon=%.4f radius=%dm)", - len(seen), - lat, - lon, - radius_m, - ) - return list(seen.values()) - - def _build_url( - self, - rooms: tuple[int, ...] | None = None, - page: int = 1, - min_price: int | None = None, - max_price: int | None = None, - ) -> str: - """URL для Cian каталога вторички ЕКБ. - - Используем ekb.cian.ru (city-specific subdomain). - Регион задаётся через region= param как fallback для reliability. - min_price / max_price — опциональные ценовые границы (руб.) для price-бакетов. - """ - params: list[tuple[str, Any]] = [ - ("deal_type", "sale"), - ("engine_version", "2"), - ("offer_type", "flat"), - ("region", CIAN_EKB_REGION_ID), - ("sort", "creation_date_desc"), - ] - if rooms: - for r in rooms: - params.append((f"room{r}", "1")) - if min_price is not None: - params.append(("minprice", min_price)) - if max_price is not None: - params.append(("maxprice", max_price)) - if page > 1: - params.append(("p", page)) - return f"{self.base_url}/cat.php?{urlencode(params)}" - - def _extract_total_offers(self, html: str) -> int | None: - """Извлечь totalOffers из Redux state Cian SERP. - - Использует тот же extract_state, что и _parse_serp_html. - Возвращает None при captcha/ошибке парсинга. - """ - state = extract_state(html, mfe=_MFE_SERP, key=_STATE_KEY) - if state is None: - return None - total = state.get("results", {}).get("totalOffers") - if total is None: - return None - try: - return int(total) - except (TypeError, ValueError): - return None - - async def _fetch_page_html( - self, - rooms: tuple[int, ...] | None, - page: int, - min_price: int | None, - max_price: int | None, - ) -> str | None: - """GET одной SERP-страницы через BrowserFetcher, возвращает HTML или None при ошибке.""" - assert self._browser is not None - url = self._build_url(rooms, page, min_price, max_price) - try: - return await self._browser.fetch(url) - except Exception: - logger.exception("cian: BrowserFetcher fetch failed url=%s", url) - return None - - async def fetch_all_secondary( - self, - *, - rooms_buckets: list[tuple[int, ...]] | None = None, - price_cap_per_bucket: int = 1400, - max_pages_per_bucket: int = 54, - concurrency: int = 5, - secondary_only: bool = True, - on_bucket: Callable[..., Any] | None = None, - on_progress: Callable[[int], None] | None = None, - skip_buckets: set[str] | None = None, - ) -> list[ScrapedLot]: - """Exhaustive-загрузка Cian ЕКБ вторички через партиционирование КОМНАТНОСТЬ × ЦЕНА. - - Обходит Cian SERP-cap (~54 стр/запрос ≈ 1500 результатов на запрос). Вместо - одной корень-бисекции [0, _MAX_PRICE] на комнатность (холостые probe-спуски - через пустой верх) — стартуем с общей data-driven seed-сетки ценовых брекетов - (get_price_seed_brackets() — shared EKB grid, общая с avito/yandex). Каждый - (комнатность × seed-брекет) дальше дробится _walk_price_range если - totalOffers > price_cap_per_bucket. Последний seed-брекет ОТКРЫТ (hi=None → - maxprice не ставится) — ловит весь хвост люкса без потолка. - Страницы внутри leaf-бакета запрашиваются параллельно (asyncio.gather + Semaphore). - Дедуп по source_id (dict seen), ОБЩИЙ по всем брекетам и комнатностям. - - Параметры: - rooms_buckets: список room-кодов Cian для перебора (default: 1-6). - price_cap_per_bucket: максимум офферов в бакете перед делением (< 1500). - max_pages_per_bucket: Cian hard cap ~54; не превышать. - concurrency: максимум параллельных page-фетчей в leaf-бакете (default=5). - secondary_only: если True (default) — отбрасывает новостройки - (listing_segment=="novostroyki") после парсинга, до save/on_bucket. - object_type=1 SERP-параметр Cian ненадёжен (~5% выдачи), поэтому - фильтруем по authoritative listing_segment из offer.newbuilding.id. - on_bucket: опциональный callback(bucket_key, list[ScrapedLot]) после каждого - leaf-бакета. Может быть async или sync. Если кидает исключение — прерывает прогон. - on_progress: опциональный callback(unique_count) для heartbeat (per room-bucket). - skip_buckets: множество ключей «room_label:lo:hi» уже завершённых бакетов — - пагинация и on_bucket для них пропускаются. Probe-запросы (для split-решения) - всё равно выполняются (их мало по сравнению с пагинацией). - - Возвращает list[ScrapedLot] уникальных лотов (дедуп по source_id/source_url). - """ - _buckets = rooms_buckets if rooms_buckets is not None else _DEFAULT_ROOMS_BUCKETS - seen: dict[str, ScrapedLot] = {} - - for rooms in _buckets: - room_label = f"room{'_'.join(str(r) for r in rooms)}" - logger.info( - "cian exhaustive: starting %s (price_cap=%d concurrency=%d secondary_only=%s)", - room_label, - price_cap_per_bucket, - concurrency, - secondary_only, - ) - before = len(seen) - # Перебираем seed-брекеты: для ЗАКРЫТЫХ передаём hi = br_hi - 1 - # (maxprice эксклюзивный сверху → соседние брекеты не пересекаются), для - # ОТКРЫТОГО (br_hi is None) — hi=None (без потолка). Дедуп по source_id - # в общем `seen` страхует на стыках. - for br_lo, br_hi in get_price_seed_brackets(): - walk_hi = br_hi - 1 if br_hi is not None else None - await self._walk_price_range( - rooms=rooms, - lo=br_lo, - hi=walk_hi, - seen=seen, - price_cap_per_bucket=price_cap_per_bucket, - max_pages_per_bucket=max_pages_per_bucket, - concurrency=concurrency, - secondary_only=secondary_only, - on_bucket=on_bucket, - skip_buckets=skip_buckets, - ) - room_collected = len(seen) - before - logger.info( - "cian exhaustive: %s done — collected %d (total unique=%d)", - room_label, - room_collected, - len(seen), - ) - if on_progress is not None: - on_progress(len(seen)) - - logger.info("cian exhaustive: DONE — total unique=%d lots", len(seen)) - return list(seen.values()) - - async def _walk_price_range( - self, - *, - rooms: tuple[int, ...], - lo: int, - hi: int | None, - seen: dict[str, ScrapedLot], - price_cap_per_bucket: int, - max_pages_per_bucket: int, - concurrency: int = 5, - secondary_only: bool = True, - on_bucket: Callable[..., Any] | None = None, - skip_buckets: set[str] | None = None, - _depth: int = 0, - ) -> None: - """Рекурсивное адаптивное бинарное партиционирование ценового диапазона [lo, hi]. - - Алгоритм для ЗАКРЫТОГО брекета (hi is not None): - 1. Запросить page=1 с min_price=lo, max_price=hi → totalOffers. - 2. Если totalOffers <= cap → пагинировать leaf-бакет параллельно (concurrency). - 3. Если totalOffers > cap → разбить бакет пополам (рекурсия). - Guard: hi - lo < _MIN_BRACKET → пагинировать как есть (логируем WARNING). - - ОТКРЫТЫЙ брекет (hi is None) — верхний seed-брекет без потолка (maxprice не - ставится, Cian отдаёт всё ≥lo). Делить нельзя (mid=(lo+hi)//2 невозможен) — - probe + пагинируем leaf напрямую. Хвост 250М+ крошечный; за cap обычно не - выходит, если вышел — пагинируем до max_pages + WARNING (tail-loss accepted). - - После пагинации leaf-бакета вызывает on_bucket(bucket_key, bucket_lots) если задан. - bucket_key = "room_label:lo:hi" (закрытый) либо "room_label:lo:open" (открытый) — - checkpoint-ключ для resume. - on_bucket может быть async или sync. Исключение в on_bucket прерывает прогон. - skip_buckets: если bucket_key в skip_buckets — пагинацию и on_bucket пропускаем. - secondary_only: если True — новостройки (listing_segment=="novostroyki") отбрасываются - после сбора bucket_lots, до дедупа в seen и вызова on_bucket. - """ - room_label = f"room{'_'.join(str(r) for r in rooms)}" - _lo_param = lo if lo > 0 else None - _hi_repr = "open" if hi is None else str(hi) - - # ── Шаг 1: probe page 1 ──────────────────────────────────────────────── - html = await self._fetch_page_html(rooms, 1, _lo_param, hi) - await self.sleep_between_requests() - - total: int | None = None - if html is not None: - total = self._extract_total_offers(html) - - # Ретрай на captcha/ошибку: 1 повторный fetch. Мобильный прокси в браузер- - # контейнере ротирует IP сам (#1806), поэтому отдельный changeip-вызов - # (_rotate_ip) больше не нужен — просто повторяем запрос на чистом IP. - if total is None: - logger.warning( - "cian: totalOffers=None for %s [%d, %s] depth=%d — retry (proxy auto-rotates IP)", - room_label, - lo, - _hi_repr, - _depth, - ) - await self._rotate_ip() # no-op (back-compat); прокси ротирует IP сам - html = await self._fetch_page_html(rooms, 1, _lo_param, hi) - await self.sleep_between_requests() - if html is not None: - total = self._extract_total_offers(html) - - if total is None: - logger.error( - "cian: skipping bucket %s [%d, %s] — totalOffers unavailable after retry", - room_label, - lo, - _hi_repr, - ) - return - - logger.info( - "cian: %s [%d, %s] totalOffers=%d depth=%d", - room_label, - lo, - _hi_repr, - total, - _depth, - ) - - if total == 0: - return - - # ── ОТКРЫТЫЙ брекет (hi is None): делить нельзя — пагинируем leaf напрямую ─ - # total известен из probe → max_pages = min(ceil(total/per_page), cap). - # Если total > cap (не должно для 250М+, но guard) — пагинируем как есть до - # max_pages_per_bucket с WARNING (хвост люкса крошечный, tail-loss accepted). - if hi is None: - if total > price_cap_per_bucket: - logger.warning( - "cian: OPEN bucket %s [%d, open] totalOffers=%d > cap=%d — paginating " - "without split (tail-loss accepted, lux tail tiny)", - room_label, - lo, - total, - price_cap_per_bucket, - ) - await self._paginate_open_bucket( - rooms=rooms, - room_label=room_label, - lo=lo, - html=html, - total=total, - seen=seen, - max_pages_per_bucket=max_pages_per_bucket, - concurrency=concurrency, - secondary_only=secondary_only, - on_bucket=on_bucket, - skip_buckets=skip_buckets, - ) - return - - # ── Шаг 2: деление или пагинация (только ЗАКРЫТЫЙ брекет, hi: int) ───── - bracket_size = hi - lo - need_split = total > price_cap_per_bucket - too_narrow = bracket_size < _MIN_BRACKET - - if need_split and too_narrow: - logger.warning( - "cian: %s [%d, %d] totalOffers=%d > cap=%d but bracket=%d < MIN_BRACKET=%d " - "— paginating as-is (tail loss ~%d)", - room_label, - lo, - hi, - total, - price_cap_per_bucket, - bracket_size, - _MIN_BRACKET, - max(0, total - price_cap_per_bucket), - ) - need_split = False # принудительно пагинируем - - if need_split: - mid = (lo + hi) // 2 - # [lo, mid] - await self._walk_price_range( - rooms=rooms, - lo=lo, - hi=mid, - seen=seen, - price_cap_per_bucket=price_cap_per_bucket, - max_pages_per_bucket=max_pages_per_bucket, - concurrency=concurrency, - secondary_only=secondary_only, - on_bucket=on_bucket, - skip_buckets=skip_buckets, - _depth=_depth + 1, - ) - # [mid+1, hi] - await self._walk_price_range( - rooms=rooms, - lo=mid + 1, - hi=hi, - seen=seen, - price_cap_per_bucket=price_cap_per_bucket, - max_pages_per_bucket=max_pages_per_bucket, - concurrency=concurrency, - secondary_only=secondary_only, - on_bucket=on_bucket, - skip_buckets=skip_buckets, - _depth=_depth + 1, - ) - return - - # Закрытый leaf-бакет — общая пагинация (см. _paginate_leaf_bucket). - await self._paginate_leaf_bucket( - rooms=rooms, - room_label=room_label, - lo=lo, - hi=hi, - html=html, - total=total, - seen=seen, - max_pages_per_bucket=max_pages_per_bucket, - concurrency=concurrency, - secondary_only=secondary_only, - on_bucket=on_bucket, - skip_buckets=skip_buckets, - ) - - async def _paginate_open_bucket( - self, - *, - rooms: tuple[int, ...], - room_label: str, - lo: int, - html: str | None, - total: int, - seen: dict[str, ScrapedLot], - max_pages_per_bucket: int, - concurrency: int, - secondary_only: bool, - on_bucket: Callable[..., Any] | None, - skip_buckets: set[str] | None, - ) -> None: - """Пагинация ОТКРЫТОГО leaf-бакета (hi=None, без maxprice).""" - await self._paginate_leaf_bucket( - rooms=rooms, - room_label=room_label, - lo=lo, - hi=None, - html=html, - total=total, - seen=seen, - max_pages_per_bucket=max_pages_per_bucket, - concurrency=concurrency, - secondary_only=secondary_only, - on_bucket=on_bucket, - skip_buckets=skip_buckets, - ) - - async def _paginate_leaf_bucket( - self, - *, - rooms: tuple[int, ...], - room_label: str, - lo: int, - hi: int | None, - html: str | None, - total: int, - seen: dict[str, ScrapedLot], - max_pages_per_bucket: int, - concurrency: int, - secondary_only: bool, - on_bucket: Callable[..., Any] | None, - skip_buckets: set[str] | None, - ) -> None: - """Параллельная пагинация leaf-бакета [lo, hi] (закрытого или открытого). - - ОТКРЫТЫЙ (hi is None): maxprice не ставится (_build_url отдаёт всё ≥lo), - bucket_key = "room_label:lo:open". Страница 1 берётся из probe-`html`. - """ - _lo_param = lo if lo > 0 else None - _hi_repr = "open" if hi is None else str(hi) - bucket_key = f"{room_label}:{lo}:{_hi_repr}" - - # Если бакет уже завершён в предыдущем запуске — пропускаем пагинацию и on_bucket. - if skip_buckets and bucket_key in skip_buckets: - logger.info( - "cian: skip bucket %s — already done (resume)", - bucket_key, - ) - return - - pages_needed = math.ceil(total / _CIAN_OFFERS_PER_PAGE) - max_pages = min(pages_needed, max_pages_per_bucket) - - if pages_needed > max_pages_per_bucket: - # Cian hard-cap срабатывает: часть офферов недостижима через пагинацию. - # Это происходит когда bracket < _MIN_BRACKET но totalOffers всё равно > cap. - tail_loss = total - max_pages_per_bucket * _CIAN_OFFERS_PER_PAGE - logger.warning( - "cian: %s [%d, %s] totalOffers=%d exceeds page cap " - "(max_pages=%d × %d=%d offers) — tail loss ~%d offers, " - "consider narrowing price_cap_per_bucket or _MIN_BRACKET", - room_label, - lo, - _hi_repr, - total, - max_pages_per_bucket, - _CIAN_OFFERS_PER_PAGE, - max_pages_per_bucket * _CIAN_OFFERS_PER_PAGE, - tail_loss, - ) - - # Страница 1 уже есть (html из probe выше); остальные — параллельно. - sem = asyncio.Semaphore(concurrency) - - async def _one_page(p: int) -> list[ScrapedLot]: - if p == 1: - # Используем уже полученный HTML от probe - return self._parse_serp_html(html) if html else [] - async with sem: - page_html = await self._fetch_page_html(rooms, p, _lo_param, hi) - await asyncio.sleep(self.request_delay_sec) - if page_html is None: - logger.warning( - "cian: page_html=None %s [%d, %s] page=%d — skipping page", - room_label, - lo, - _hi_repr, - p, - ) - return [] - return self._parse_serp_html(page_html) - - page_results = await asyncio.gather( - *[_one_page(p) for p in range(1, max_pages + 1)], - return_exceptions=True, - ) - - bucket_lots: list[ScrapedLot] = [] - for p_idx, res in enumerate(page_results, start=1): - if isinstance(res, BaseException): - logger.warning( - "cian: page exception %s [%d, %s] page=%d — %r", - room_label, - lo, - _hi_repr, - p_idx, - res, - ) - else: - bucket_lots.extend(res) - - collected_this_bucket = len(bucket_lots) - - # ── Фильтр новостроек (secondary_only) ──────────────────────────────── - dropped_nb = 0 - if secondary_only: - filtered = [lot for lot in bucket_lots if lot.listing_segment != "novostroyki"] - dropped_nb = collected_this_bucket - len(filtered) - bucket_lots = filtered - - # Дедуп в общий seen - for lot in bucket_lots: - key = lot.source_id or lot.source_url - if key: - seen[key] = lot - - logger.info( - "cian: %s [%d, %s] paginated=%d pages collected=%d dropped_nb=%d unique_total=%d", - room_label, - lo, - _hi_repr, - max_pages, - collected_this_bucket, - dropped_nb, - len(seen), - ) - - # ── on_bucket callback: инкрементальный save ────────────────────────── - if on_bucket is not None and bucket_lots: - res_cb = on_bucket(bucket_key, bucket_lots) - if inspect.isawaitable(res_cb): - await res_cb - - def _parse_serp_html(self, html: str) -> list[ScrapedLot]: - """Извлечь offers из Cian Redux state. - - Использует cian_state_parser.extract_state() — общая утилита Stage 2. - Возвращает пустой список если state не найден. - """ - state = extract_state(html, mfe=_MFE_SERP, key=_STATE_KEY) - if state is None: - logger.warning( - "cian SERP state extraction failed (mfe=%s key=%s) — " - "возможно Cian изменил структуру или вернул captcha", - _MFE_SERP, - _STATE_KEY, - ) - return [] - - offers_data: list[dict[str, Any]] = state.get("results", {}).get("offers", []) - if not offers_data: - logger.warning( - "cian state found but results.offers пуст (totalOffers=%s)", - state.get("results", {}).get("totalOffers", "?"), - ) - return [] - - logger.info( - "cian SERP state ok: %d offers (totalOffers=%s)", - len(offers_data), - state.get("results", {}).get("totalOffers", "?"), - ) - - lots: list[ScrapedLot] = [] - for offer in offers_data: - lot = self._offer_to_lot(offer) - if lot is not None: - lots.append(lot) - - raw_count = len(offers_data) - saved_count = len(lots) - # Охрана от silent-failure: если получили offers, но ни один не прошёл парсинг - # — likely schema regression (как descriptionMinhash str→list[int] 2026-05-23). - if raw_count > 0 and saved_count == 0: - logger.error( - "cian SERP: 0/%d offers прошли _offer_to_lot — возможна schema regression", - raw_count, - ) - try: - if settings.glitchtip_dsn: - sentry_sdk.capture_message( - f"cian SERP: {raw_count}/{raw_count} offers failed _offer_to_lot" - " — possible schema regression", - level="error", - ) - except Exception: - pass # sentry_sdk not initialised in dev - - return lots - - def _offer_to_lot(self, offer: dict[str, Any]) -> ScrapedLot | None: - """Парсинг одного Cian offer (137 fields) → ScrapedLot. - - Маппинг полей per Schema_Cian_SERP_Inventory sec 3, sec 8, sec 11. - Без anchor jitter: координаты из offer.geo.coordinates (точные). - """ - try: - # ── Identity ───────────────────────────────────────────────────── - offer_id = offer.get("cianId") or offer.get("id") - if not offer_id: - logger.debug("cian offer пропущен: нет cianId/id") - return None - - source_id = str(offer_id) - url = offer.get("fullUrl") - if not url: - url = f"{self.base_url}/sale/flat/{offer_id}/" - - # ── Цена ───────────────────────────────────────────────────────── - bargain: dict[str, Any] = offer.get("bargainTerms") or {} - price = bargain.get("priceRur") or bargain.get("price") - if not price: - logger.debug("cian offer %s пропущен: нет цены", offer_id) - return None - try: - price_rub = int(price) - except (TypeError, ValueError): - logger.debug("cian offer %s: не удалось распарсить цену %r", offer_id, price) - return None - if price_rub <= 0: - return None - - # ── Параметры квартиры ─────────────────────────────────────────── - rooms: int | None = offer.get("roomsCount") - # Студия: flatType == 'studio' → 0 комнат - if offer.get("flatType") == "studio" and rooms is None: - rooms = 0 - - area_raw = offer.get("totalArea") - area_m2: float | None = None - if area_raw is not None: - try: - area_m2 = float(area_raw) - except (TypeError, ValueError): - pass - - living_area_raw = offer.get("livingArea") - living_area_m2: float | None = None - if living_area_raw is not None: - try: - living_area_m2 = float(living_area_raw) - except (TypeError, ValueError): - pass - - # kitchen_area: сырое (str) в raw_payload + промоутим float в ScrapedLot - # колонку kitchen_area_m2 (#2008, Class B — дашборд/matching, не valuation). - kitchen_area_raw = offer.get("kitchenArea") - kitchen_area_m2: float | None = None - if kitchen_area_raw is not None: - try: - kitchen_area_m2 = float(kitchen_area_raw) - except (TypeError, ValueError): - pass - - floor: int | None = offer.get("floorNumber") - - # ── Здание ─────────────────────────────────────────────────────── - building: dict[str, Any] = offer.get("building") or {} - total_floors: int | None = building.get("floorsCount") - year_built: int | None = building.get("buildYear") - # house_type: cian materialType (camelCase: monolithBrick / gasSilicateBlock - # / stalin / ...) НЕ равен каноничному enum → estimator soft-penalty ложно - # штрафовал аналог. Нормализуем на ингесте (#2008, Class A); сырьё — в - # raw_payload['raw_material_type'] для аудита. - raw_material_type: str | None = building.get("materialType") - house_type: str | None = normalize_house_type(raw_material_type) - - # Newbuilding deadline year — если нет buildYear - if year_built is None: - deadline = building.get("deadline") or {} - year_built = deadline.get("year") - - # ── Кадастр ───────────────────────────────────────────────────── - # cadastralNumber — кадастр КВАРТИРЫ - cadastral_number: str | None = offer.get("cadastralNumber") - # buildingCadastralNumber — кадастр ДОМА - building_cadastral_number: str | None = offer.get("buildingCadastralNumber") - - # ── Геолокация (точные координаты — NO jitter) ─────────────────── - geo: dict[str, Any] = offer.get("geo") or {} - coords: dict[str, Any] = geo.get("coordinates") or {} - lat: float | None = coords.get("lat") - lon: float | None = coords.get("lng") - # Если нет coords → сохраняем без lat/lon (geocode-missing обработает позже) - if lat is None or lon is None: - logger.debug( - "cian offer %s: нет точных координат в state (geo=%s)", - offer_id, - coords, - ) - lat = lon = None - - # ── Адрес из geo.address[] ─────────────────────────────────────── - # newbuilding нужен ДО _format_address: для ЖК geo.address[] часто без - # части {type:"house"} → дом-номер восстанавливаем из building/кадастра/ - # названия ЖК (#1773), иначе все корпуса схлопываются в house-less - # catch-all и same-building anchor рвётся. - address_parts = geo.get("address") or [] - newbuilding: dict[str, Any] = offer.get("newbuilding") or {} - nb_id = newbuilding.get("id") - is_newbuilding = bool(nb_id) and int(nb_id) > 0 - address = _format_address( - address_parts, - building=building, - building_cadastral_number=building_cadastral_number, - newbuilding=newbuilding, - is_newbuilding=is_newbuilding, - ) - - # ── Metro stations ─────────────────────────────────────────────── - undergrounds: list[dict[str, Any]] = geo.get("undergrounds") or [] - metro_stations = [ - { - "name": u.get("name"), - "time": u.get("time"), - "mode": u.get("transportType"), - "line_color": u.get("lineColor"), - "line_id": u.get("lineId"), - "is_default": u.get("isDefault", False), - } - for u in undergrounds - if u.get("name") - ] - - # ── Newbuilding link ───────────────────────────────────────────── - # newbuilding / nb_id / is_newbuilding уже вычислены выше (до адреса). - # id == 0 — sentinel для вторички (per Schema sec 20.5) - house_source: str | None = None - house_ext_id: str | None = None - listing_segment: str | None = None - - if is_newbuilding: - # Предпочитаем кадастр ДОМА как same-building anchor (#1773): nb_id - # группирует по ЖК целиком, а один ЖК — это несколько корпусов - # (6679/11110/7019). buildingCadastralNumber стабильно идентифицирует - # ФИЗИЧЕСКИЙ дом → Tier 0 cadastr_exact в match_or_create_house - # схлопывает раздробленные корпуса. nb_id оставляем fallback'ом. - if building_cadastral_number: - house_source = "cian_building_cad" - house_ext_id = building_cadastral_number - else: - house_source = "cian_newbuilding" - house_ext_id = str(nb_id) - listing_segment = "novostroyki" - else: - house_source = "cian" - listing_segment = "vtorichka" - - # ── Дополнительные характеристики ──────────────────────────────── - balconies_count: int | None = offer.get("balconiesCount") - loggias_count: int | None = offer.get("loggiasCount") - bedrooms_count: int | None = offer.get("bedroomsCount") - - has_balcony: bool | None = None - if balconies_count is not None: - has_balcony = balconies_count > 0 - elif loggias_count is not None: - has_balcony = loggias_count > 0 - - # Мебель и repair_state - # hasFurniture (мебель вообще) и isSoldFurnished (продаётся с мебелью) — - # разная семантика, нельзя OR-склеивать: явный False у hasFurniture - # затирался бы коалесингом. Берём hasFurniture; только если он None, - # падаем на isSoldFurnished как fallback (#1388). - _has_furniture = offer.get("hasFurniture") - has_furniture: bool | None = ( - _has_furniture if _has_furniture is not None else offer.get("isSoldFurnished") - ) - # decoration — для новостроек (отделка); repair_state — для вторички - # Cian SERP: offer.decoration содержит raw значения (without/cosmetic/euro/design/...) - # Нормализуем до enum needs_repair/standard/good/excellent при инgesте. - repair_state: str | None = normalize_repair_state(offer.get("decoration")) - - # ── Продавец ───────────────────────────────────────────────────── - # PII purge (#2217): телефоны продавцов больше не сохраняем. - # ScrapedLot.phones остаётся в модели (см. base.py), но на вход - # всегда передаём пустой список — write-path на listings.phones - # выключен здесь, а не на уровне upsert. - phones: list[dict[str, Any]] = [] - is_homeowner: bool | None = offer.get("isByHomeowner") - is_pro_seller: bool | None = offer.get("isPro") - - # ── Сделка ─────────────────────────────────────────────────────── - sale_type: str | None = bargain.get("saleType") - bargain_allowed: bool | None = bargain.get("bargainAllowed") - - # ── Class B promote (#2008): ипотека / апартаменты / проверка ЕГРН ── - # Раньше лежали только в raw_payload; промоутим в колонки listings для - # дашборда покрытия и matching (НЕ для valuation — estimator их не читает). - mortgage_available: bool | None = bargain.get("mortgageAllowed") - is_apartments: bool | None = offer.get("isApartments") - is_rosreestr_checked: bool | None = offer.get("isRosreestrChecked") - - # ── Description + minhash ───────────────────────────────────────── - description: str | None = offer.get("description") - # Cian предоставляет готовый minhash для dedup и cross-source matching. - # С ~2026-05-23 Cian изменил тип descriptionMinhash: str → list[int]. - # Нормализуем оба варианта; НЕ sorted() — порядок band-ов значим для LSH. - _raw_minhash = offer.get("descriptionMinhash") - if isinstance(_raw_minhash, list): - description_minhash: str | None = ",".join(str(x) for x in _raw_minhash) or None - elif isinstance(_raw_minhash, str): - description_minhash = _raw_minhash or None - else: - description_minhash = None - # Если Cian не дал minhash — вычисляем простой SHA1 из description - if not description_minhash and description: - description_minhash = hashlib.sha1( - description.lower().encode("utf-8", errors="replace") - ).hexdigest()[:32] - - # Fallback: repair_state из описания, если decoration отсутствует (#622) - if repair_state is None and description: - repair_state = infer_repair_state_from_text(description) - - # ── Фото ───────────────────────────────────────────────────────── - photo_urls: list[str] = [ - p["fullUrl"] - for p in (offer.get("photos") or []) - if isinstance(p, dict) and p.get("fullUrl") - ] - - # ── Дата публикации ─────────────────────────────────────────────── - added_ts = offer.get("addedTimestamp") - listing_date = None - if added_ts: - try: - listing_date = datetime.fromtimestamp(int(added_ts), tz=UTC).date() - except (TypeError, ValueError, OSError): - pass - - # ── Raw payload (полный offer для enrichment в будущем) ─────────── - raw_payload: dict[str, Any] = { - "cian_id": offer_id, - "flat_type": offer.get("flatType"), - "is_apartments": offer.get("isApartments"), - "offer_type": offer.get("offerType"), - "category": offer.get("category"), - "has_furniture": has_furniture, - # сырой kitchenArea (str) — колонка kitchen_area_m2 (float) промоучена выше - "kitchen_area_m2": kitchen_area_raw, - # сырой materialType (camelCase) — house_type нормализован выше (#2008) - "raw_material_type": raw_material_type, - "mortgage_allowed": bargain.get("mortgageAllowed"), - "sale_type": sale_type, - "newbuilding_name": newbuilding.get("name"), - "is_from_developer": ( - offer.get("fromDeveloper") or newbuilding.get("isFromDeveloper") - ), - "builders_ids": offer.get("buildersIds"), - "is_rosreestr_checked": offer.get("isRosreestrChecked"), - "is_layout_approved": offer.get("isLayoutApproved"), - "user_id": offer.get("userId"), - "published_user_id": offer.get("publishedUserId"), - "is_by_commercial_owner": offer.get("isByCommercialOwner"), - "is_cian_partner": offer.get("isCianPartner"), - "description_words_highlighted": offer.get("descriptionWordsHighlighted"), - "building_parking": building.get("parking"), - "building_total_area": building.get("totalArea"), - } - - return ScrapedLot( - source="cian", - source_url=url, - source_id=source_id, - address=address, - lat=lat, - lon=lon, - rooms=rooms, - area_m2=area_m2, - floor=floor, - total_floors=total_floors, - year_built=year_built, - house_type=house_type, - repair_state=repair_state, - has_balcony=has_balcony, - kitchen_area_m2=kitchen_area_m2, - mortgage_available=mortgage_available, - is_apartments=is_apartments, - is_rosreestr_checked=is_rosreestr_checked, - kadastr_num=cadastral_number, - house_source=house_source, - house_ext_id=house_ext_id, - listing_segment=listing_segment, - price_rub=price_rub, - price_per_m2=None, # compute_price_per_m2() вычислит - listing_date=listing_date, - photo_urls=photo_urls, - raw_payload=raw_payload, - # Cian-specific (Stage 2 fields) - living_area_m2=living_area_m2, - bedrooms_count=bedrooms_count, - balconies_count=balconies_count, - loggias_count=loggias_count, - description_minhash=description_minhash, - cadastral_number=cadastral_number, - building_cadastral_number=building_cadastral_number, - phones=phones, - is_homeowner=is_homeowner, - is_pro_seller=is_pro_seller, - bargain_allowed=bargain_allowed, - sale_type=sale_type, - metro_stations=metro_stations, - ) - - except Exception: - _oid = offer.get("cianId") or offer.get("id") - logger.exception("cian _offer_to_lot failed for offer_id=%s", _oid) - return None - - -# ── Address formatter ──────────────────────────────────────────────────────── - -# Дом-токен в хвосте строки: «29», «29А», «12 к1», «5/2», «10 стр.3», «1С7». -# Якорим в конец строки (\Z) — берём именно завершающий номер, не цифры из -# названия ЖК («Суходольский квартал»). Не матчим «р-н», «мкр» и пр. слова. -_HOUSE_TOKEN_RE = re.compile( - r"(\d+[А-Яа-яA-Za-z]?" # базовый номер: 29 / 29А / 1С - r"(?:\s*(?:к|корп\.?|стр\.?|с)\s*\d+[А-Яа-яA-Za-z]?)?" # корпус/строение - r"(?:\s*/\s*\d+[А-Яа-яA-Za-z]?)?)" # дробь 5/2 - r"\s*\Z" -) - - -def _recover_newbuilding_house( - *, - building: dict[str, Any] | None, - building_cadastral_number: str | None, - newbuilding: dict[str, Any] | None, -) -> str | None: - """Восстановить дом-токен для новостройки, когда geo.address[] без {type:"house"}. - - Приоритет (#1773): - (a) building.houseNumber / building.address — если SERP-state их отдал; - (b) building_cadastral_number — стабильный same-building anchor: если ни - номер, ни название не дали дом, добавляем «кад. 66:41:...:11396» как - ключ-якорь, чтобы раздробленные корпуса схлопывались по кадастру; - (c) хвостовой дом-токен из newbuilding.name («… 29» / «… 29А к1»). - - НЕ выдумываем catch-all номер: если ничего из (a)/(b)/(c) не нашли — None. - Кадастр (b) предпочтительнее как house anchor (см. _offer_to_lot). - """ - building = building or {} - newbuilding = newbuilding or {} - - # (a) building.houseNumber / building.address — прямые поля SERP-state - house_number = building.get("houseNumber") - if house_number is not None: - token = str(house_number).strip() - if token: - return token - b_addr = (building.get("address") or "").strip() - if b_addr: - m = _HOUSE_TOKEN_RE.search(b_addr) - if m: - return m.group(1).strip() - - # (c) хвостовой дом-токен из названия ЖК (до кадастра — он читабельнее) - nb_name = (newbuilding.get("name") or "").strip() - if nb_name: - m = _HOUSE_TOKEN_RE.search(nb_name) - if m: - return m.group(1).strip() - - # (b) кадастр дома как anchor-ключ (последний приоритет в строке адреса, - # но самый надёжный для матчинга — Tier 0 cadastr_exact) - if building_cadastral_number: - cad = building_cadastral_number.strip() - if cad: - return f"кад. {cad}" - - return None - - -def _format_address( - address_parts: list[dict[str, Any]], - *, - building: dict[str, Any] | None = None, - building_cadastral_number: str | None = None, - newbuilding: dict[str, Any] | None = None, - is_newbuilding: bool = False, -) -> str: - """Сформировать читаемый адрес из geo.address[] Cian. - - Пропускаем location (страна/регион) и metro — берём раион/улицу/дом. - Пример: [location=Москва, raion=Пресненский, street=..., house=1С7, metro=Москва-Сити] - → 'Пресненский, улица ..., 1С7' - - #1773: для новостроек (is_newbuilding) geo.address[] часто без {type:"house"}. - Если части house нет — восстанавливаем номер дома через _recover_newbuilding_house - (building → название ЖК → кадастр) и дописываем его в конец строки, чтобы - same-building anchor (fingerprint/кадастр) не рвался по house-less улице. - """ - if not address_parts: - return "Екатеринбург (Cian)" - - skip_types = {"location", "metro"} - parts: list[str] = [] - has_house = False - for part in address_parts: - if not isinstance(part, dict): - continue - ptype = part.get("type", "") - if ptype == "house": - has_house = True - if ptype in skip_types: - continue - name = (part.get("fullName") or part.get("name") or "").strip() - if name: - parts.append(name) - - # Восстановление дома только для новостроек без части house — вторичка с - # house-less адресом не трогается (не выдумываем номер). - if is_newbuilding and not has_house: - recovered = _recover_newbuilding_house( - building=building, - building_cadastral_number=building_cadastral_number, - newbuilding=newbuilding, - ) - if recovered: - parts.append(recovered) - - return ", ".join(parts) if parts else "Екатеринбург (Cian)" diff --git a/tradein-mvp/backend/app/services/scrapers/cian_detail.py b/tradein-mvp/backend/app/services/scrapers/cian_detail.py deleted file mode 100644 index d3b03742..00000000 --- a/tradein-mvp/backend/app/services/scrapers/cian_detail.py +++ /dev/null @@ -1,464 +0,0 @@ -"""Cian.ru detail page scraper — single offer enrichment. - -Different from SERP (cian.py): -- URL: https://ekb.cian.ru/sale/flat// or https://www.cian.ru/sale/flat// -- MFE: 'frontend-offer-card' -- State KEY: 'defaultState' (NOT 'initialState' — SERP uses initialState) -- Sister containers in same _cianConfig: bti, priceChanges, stats, agent, newObject - -Parses ~88 offer fields + sister data → DetailEnrichment dataclass. -Stage 5 of CianScraper v1. -""" - -from __future__ import annotations - -import json -import logging -from dataclasses import dataclass, field -from typing import TYPE_CHECKING, Any - -from curl_cffi.requests import AsyncSession -from sqlalchemy import text -from sqlalchemy.orm import Session - -from app.core.config import settings -from app.services.scrapers.cian_state_parser import extract_all_states, extract_state -from app.services.scrapers.repair_state_normalizer import ( - infer_repair_state_from_text, - normalize_repair_state, -) -from app.services.scrapers.snapshot_writer import upsert_listing_snapshot - -if TYPE_CHECKING: - from app.services.scrapers.browser_fetcher import BrowserFetcher - -logger = logging.getLogger(__name__) - -# offer_price_history.diff_percent — NUMERIC(8,2) (миграция -# 131_fix_diff_percent_overflow.sql). Прямой INSERT ниже читает diff_percent из -# скрапленного Cian JSON и не проходит через БД-триггер record_listing_price_change() -# (тот срабатывает только на UPDATE listings) — гарбадж/экстремальное значение -# с редизайна source-страницы иначе переполняет колонку. -_DIFF_PERCENT_BOUND = 999999.99 - - -def _clamp_diff_percent(value: float | int | None) -> float | None: - """Клампит diff_percent в границы NUMERIC(8,2) (зеркало migration 131 trigger-clamp).""" - if value is None or isinstance(value, bool): - return None - try: - numeric = float(value) - except (TypeError, ValueError): - return None - return min(max(numeric, -_DIFF_PERCENT_BOUND), _DIFF_PERCENT_BOUND) - - -@dataclass -class DetailEnrichment: - """Result of cian_detail.fetch_detail() — fields для UPDATE listings + sister rows.""" - - # Core listing enrichment (extend existing listings row) - cian_id: int | None = None - windows_view_type: str | None = None # 'yard' / 'street' / 'yardAndStreet' - separate_wcs_count: int | None = None - combined_wcs_count: int | None = None - ceiling_height: float | None = None # meters — offer.building.ceilingHeight (#1791) - repair_type: str | None = None # raw Cian: offer.repairType OR offer.decoration (#1791) - repair_state: str | None = None # enum: needs_repair/standard/good/excellent - kitchen_area_m2: float | None = None # offer.kitchenArea (м²) - description: str | None = None # offer.description (текст объявления) - views_total: int | None = None # Cian stats.totalViewsFormattedString → int - views_today: int | None = None - - # BTI (вторичка only — primary buildings don't have БТИ) - bti_data: dict[str, Any] | None = None # raw bti.houseData snapshot - - # Price history (offer.priceChanges) - price_changes: list[dict[str, Any]] = field(default_factory=list) - - # Agent profile (offer.agent.* expanded) - agent_profile: dict[str, Any] | None = None - - # Newbuilding link (if offer is from a newbuilding) - newbuilding_data: dict[str, Any] | None = None - - # Raw payload для debugging / future-extraction - raw_offer: dict[str, Any] | None = None - raw_sister_states: dict[str, Any] = field(default_factory=dict) - - -async def fetch_detail( - offer_url: str, - *, - session: AsyncSession | None = None, - browser_fetcher: BrowserFetcher | None = None, -) -> DetailEnrichment | None: - """Fetch detail page and extract all containers. - - Args: - offer_url: e.g. 'https://ekb.cian.ru/sale/flat/327830237/' - session: optional shared curl_cffi session (для batched scrapes). - Ignored when browser_fetcher is provided. - browser_fetcher: optional BrowserFetcher instance (camoufox HTTP service). - When provided, fetches JS-rendered HTML via the browser service instead of - curl_cffi — enables priceChanges extraction which requires JS rendering. - Caller is responsible for the context-manager lifecycle of the fetcher. - - Returns: DetailEnrichment, or None если fetch / parse failed. - """ - if browser_fetcher is not None: - # Browser path: get fully JS-rendered HTML; same parse path follows. - try: - html = await browser_fetcher.fetch(offer_url) - except Exception as exc: - logger.warning("Cian detail browser fetch failed %s: %s", offer_url, exc) - return None - else: - # curl_cffi path (legacy, back-compat). - close_session = False - if session is None: - # proxies: mobile-proxy egress (#806) — без прокси datacenter-IP блокируется Cian. - # Пусто (env не задан) → прямое подключение (dev/no-op). - _proxy_url = settings.cian_proxy_url - _proxies = {"http": _proxy_url, "https": _proxy_url} if _proxy_url else None - session = AsyncSession( - impersonate="chrome120", - timeout=25.0, - proxies=_proxies, - headers={ - "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8", - "Accept-Language": "ru-RU,ru;q=0.9,en;q=0.8", - }, - ) - close_session = True - - try: - resp = await session.get(offer_url, allow_redirects=True) - if resp.status_code != 200: - logger.warning("Cian detail fetch %s → HTTP %d", offer_url, resp.status_code) - return None - html = resp.text - finally: - if close_session: - await session.close() - - # ── Shared parse path (identical regardless of HTML source) ────────────── - - # Primary state container — defaultState in frontend-offer-card MFE - # NOTE: detail pages use 'defaultState', SERP uses 'initialState' - offer_state = extract_state(html, mfe="frontend-offer-card", key="defaultState") - if offer_state is None: - logger.warning("Cian detail %s: defaultState extraction failed", offer_url) - return None - - offer_data = offer_state.get("offerData", {}) - offer = offer_data.get("offer", {}) - - result = DetailEnrichment( - cian_id=offer.get("cianId") or offer.get("id"), - windows_view_type=_extract_windows_view(offer), - separate_wcs_count=offer.get("separateWcsCount"), - combined_wcs_count=offer.get("combinedWcsCount"), - ceiling_height=_extract_ceiling_height(offer), - repair_type=_extract_repair_type(offer), - repair_state=_extract_repair_state(offer), - kitchen_area_m2=_parse_float(offer.get("kitchenArea")), - description=offer.get("description") or None, - raw_offer=offer, - ) - - # Stats (views) — from stats key in offerData or top-level state - stats = offer_data.get("stats") or offer_state.get("stats") or {} - result.views_total = _parse_views(stats.get("totalViewsFormattedString")) - result.views_today = _parse_views(stats.get("todayViewsFormattedString")) - - # Price changes — browser HTML: offerData.priceChanges (priceData.price format) - # curl_cffi HTML: offer.priceChanges or state.priceChanges (flat price field) - result.price_changes = _extract_price_changes(offer, offer_data, offer_state) - - # Agent profile expansion - result.agent_profile = _extract_agent_profile(offer) - - # Sister state containers: bti, newObject, plus raw stash of everything else - all_states = extract_all_states(html) - offer_card_states = all_states.get("frontend-offer-card", {}) - - bti_state = offer_card_states.get("bti") - if bti_state: - result.bti_data = bti_state.get("houseData") - result.raw_sister_states["bti"] = bti_state - - # Newbuilding link (if applicable) - newbuilding = offer.get("newbuilding") - if newbuilding and newbuilding.get("id"): - result.newbuilding_data = newbuilding - - # Stash all non-defaultState sister containers for debugging / future-extraction - for k, v in offer_card_states.items(): - if k != "defaultState": - result.raw_sister_states[k] = v - - return result - - -# ── Field extractors ────────────────────────────────────────────────────────── - - -def _extract_windows_view(offer: dict[str, Any]) -> str | None: - return offer.get("windowsViewType") - - -def _extract_ceiling_height(offer: dict[str, Any]) -> float | None: - """Высота потолков (м). - - На detail-странице (defaultState) значение лежит в ``offer.building.ceilingHeight``, - а НЕ в ``offer.ceilingHeight`` (последнее всегда null) — #1791. Top-level оставлен - как fallback на случай иной структуры карточки. - """ - building = offer.get("building") or {} - return _parse_float(building.get("ceilingHeight") or offer.get("ceilingHeight")) - - -def _extract_repair_type(offer: dict[str, Any]) -> str | None: - # Cian: 'cosmetic' / 'design' / 'no' / 'euro' / ... - # На detail-странице структурное поле чаще приходит как `decoration` - # ('preFine'/'fine'/...), а `repairType` пустое (#1791) — читаем оба. - return offer.get("repairType") or offer.get("decoration") - - -def _extract_repair_state(offer: dict[str, Any]) -> str | None: - """Нормализованный enum из repairType/decoration → needs_repair/standard/good/excellent. - - Если структурное поле пустое — fallback на инференс из описания (#622). - Структурное значение всегда в приоритете. - """ - state = normalize_repair_state(offer.get("repairType") or offer.get("decoration")) - if state is None: - state = infer_repair_state_from_text(offer.get("description")) - return state - - -def _parse_float(value: Any) -> float | None: - if value is None: - return None - try: - return float(value) - except (ValueError, TypeError): - return None - - -def _parse_views(formatted_str: str | None) -> int | None: - """Cian's stats.totalViewsFormattedString — '1 234' → 1234.""" - if not formatted_str: - return None - try: - return int(str(formatted_str).replace(" ", "").replace(",", "")) - except (ValueError, TypeError): - return None - - -def _extract_price_changes( - offer: dict[str, Any], - offer_data: dict[str, Any], - state: dict[str, Any], -) -> list[dict[str, Any]]: - """Normalize priceChanges from any of the known Cian locations/formats. - - Lookup priority: - 1. offer.priceChanges — curl_cffi flat format: {changeTime, price, diffPercent} - 2. offerData.priceChanges — browser-rendered format: {changeTime, priceData:{price}} - 3. state.priceChanges — top-level fallback (rare) - - Normalizes to: {change_time, price_rub, diff_percent}. - """ - raw = ( - offer.get("priceChanges") - or offer_data.get("priceChanges") - or state.get("priceChanges") - or [] - ) - if not isinstance(raw, list): - return [] - result = [] - for entry in raw: - if not isinstance(entry, dict): - continue - # price_rub: flat field (curl_cffi) OR nested priceData.price (browser) - price_data = entry.get("priceData") or {} - price_rub = entry.get("price") or entry.get("priceRub") or price_data.get("price") - result.append( - { - "change_time": entry.get("changeTime") or entry.get("date"), - "price_rub": price_rub, - "diff_percent": entry.get("diffPercent") or entry.get("diff_percent"), - } - ) - return result - - -def _extract_agent_profile(offer: dict[str, Any]) -> dict[str, Any] | None: - """offer.agent or offer.agency or offer.user — extract canonical profile shape.""" - agent = offer.get("agent") or {} - user = offer.get("user") or {} - if not agent and not user: - return None - return { - "ext_agent_id": agent.get("id") or user.get("cianUserId"), - "company_name": agent.get("companyName") or user.get("companyName"), - "is_pro": agent.get("isPro") or user.get("isPro"), - "skills": agent.get("skills") or [], - "account_type": agent.get("accountType") or user.get("accountType"), - } - - -# ── Save helpers ────────────────────────────────────────────────────────────── - - -def save_detail_enrichment(db: Session, listing_id: int, enrichment: DetailEnrichment) -> None: - """Persist DetailEnrichment to DB: - - - UPDATE listings SET windows_view_type, ceiling_height, ... WHERE id = listing_id - - INSERT INTO offer_price_history rows (if price_changes non-empty) - - INSERT INTO agents (if agent_profile) and link via agent_id_fk - - bti_data не сохраняется здесь — это задача Stage 6 (houses), где есть house_id - - Idempotency: re-running безопасно (UPSERT / ON CONFLICT DO NOTHING semantics). - """ - # Extend listings row with detail-enriched fields - db.execute( - text(""" - UPDATE listings SET - windows_view_type = COALESCE(:wvt, windows_view_type), - separate_wcs_count = COALESCE(:swc, separate_wcs_count), - combined_wcs_count = COALESCE(:cwc, combined_wcs_count), - ceiling_height = COALESCE(:ch, ceiling_height), - repair_type = COALESCE(:rt, repair_type), - repair_state = COALESCE(:rs, repair_state), - kitchen_area_m2 = COALESCE(CAST(:ka AS double precision), kitchen_area_m2), - description = COALESCE(:descr, description), - views_total = COALESCE(:vt, views_total), - views_today = COALESCE(:vd, views_today), - detail_enriched_at = NOW() - WHERE id = CAST(:lid AS bigint) - """), - { - "lid": listing_id, - "wvt": enrichment.windows_view_type, - "swc": enrichment.separate_wcs_count, - "cwc": enrichment.combined_wcs_count, - "ch": enrichment.ceiling_height, - "rt": enrichment.repair_type, - "rs": enrichment.repair_state, - "ka": enrichment.kitchen_area_m2, - "descr": enrichment.description, - "vt": enrichment.views_total, - "vd": enrichment.views_today, - }, - ) - - # Snapshot: point-in-time observation в listings_snapshots. - # Берём price_rub / price_per_m2 из listings строки (только что обновлённой). - # Используем COALESCE — на случай если listings.price_rub NULL (не должно быть, но safe). - _snap_row = db.execute( - text(""" - SELECT price_rub, price_per_m2 - FROM listings - WHERE id = CAST(:lid AS bigint) - """), - {"lid": listing_id}, - ).fetchone() - _snap_written = False - if _snap_row is not None and _snap_row.price_rub is not None: - try: - with db.begin_nested(): - upsert_listing_snapshot( - db, - listing_id=listing_id, - price_rub=int(_snap_row.price_rub), - price_per_m2=int(_snap_row.price_per_m2) if _snap_row.price_per_m2 else None, - run_id=None, - status="active", - ) - _snap_written = True - except Exception as exc: - logger.warning( - "save_detail_enrichment: snapshot write failed listing_id=%s: %s", - listing_id, - exc, - ) - - # Price changes — INSERT new rows, de-dup by listing_id + change_time - for change in enrichment.price_changes: - if not change.get("change_time") or not change.get("price_rub"): - continue - db.execute( - text(""" - INSERT INTO offer_price_history - (listing_id, change_time, price_rub, diff_percent, source) - VALUES ( - CAST(:lid AS bigint), - CAST(:ct AS timestamptz), - CAST(:price AS numeric), - CAST(:diff AS numeric), - 'cian' - ) - ON CONFLICT ON CONSTRAINT offer_price_history_listing_change_uq DO NOTHING - """), - { - "lid": listing_id, - "ct": change["change_time"], - "price": change["price_rub"], - "diff": _clamp_diff_percent(change.get("diff_percent")), - }, - ) - - # Agent profile UPSERT — store agent row and link to listing - if enrichment.agent_profile and enrichment.agent_profile.get("ext_agent_id"): - ap = enrichment.agent_profile - row = db.execute( - text(""" - INSERT INTO agents - (ext_source, ext_agent_id, company_name, is_pro, skills, account_type) - VALUES ( - 'cian', - :eai, - :cn, - :ip, - CAST(:sk AS jsonb), - :at - ) - ON CONFLICT (ext_source, ext_agent_id) DO UPDATE SET - company_name = COALESCE(EXCLUDED.company_name, agents.company_name), - is_pro = COALESCE(EXCLUDED.is_pro, agents.is_pro), - skills = COALESCE(EXCLUDED.skills, agents.skills), - account_type = COALESCE(EXCLUDED.account_type, agents.account_type) - RETURNING id - """), - { - "eai": str(ap["ext_agent_id"]), - "cn": ap.get("company_name"), - "ip": ap.get("is_pro"), - "sk": json.dumps(ap.get("skills") or []), - "at": ap.get("account_type"), - }, - ) - agent_db_id = row.scalar_one_or_none() - if agent_db_id is not None: - db.execute( - text(""" - UPDATE listings - SET agent_id_fk = CAST(:aid AS bigint) - WHERE id = CAST(:lid AS bigint) - AND agent_id_fk IS NULL - """), - {"aid": agent_db_id, "lid": listing_id}, - ) - - db.commit() - logger.info( - "Cian detail enrichment saved for listing_id=%s (price_changes=%d, agent=%s, snapshot=%s)", - listing_id, - len(enrichment.price_changes), - bool(enrichment.agent_profile), - _snap_written, - ) diff --git a/tradein-mvp/backend/app/services/scrapers/domclick.py b/tradein-mvp/backend/app/services/scrapers/domclick.py deleted file mode 100644 index 4ed48e0b..00000000 --- a/tradein-mvp/backend/app/services/scrapers/domclick.py +++ /dev/null @@ -1,667 +0,0 @@ -"""DomClick.ru scraper — вторичка через JSON BFF API (#1968). - -Стратегия: GET https://bff-search-web.domclick.ru/api/offers/v1?... через -BrowserFetcher(source="domclick") (generic provider → shared mobile proxy). -QRATOR банит прямые datacenter-запросы, но пропускает через mobile proxy. - -Ответ BrowserFetcher содержит JSON, обёрнутый в HTML (
 или bare body).
-Парсинг: _extract_json() вытаскивает первый {...} из ответа.
-
-Стратегия нумерации комнат (ROOM_BUCKETS):
-  "st"  — студии (force rooms=0 в ScrapedLot)
-  "1"–"4" — соответственно
-  "5+"  — 5 и более комнат
-
-Пагинация: offset 0, 20, 40, …, cap=2000. Если snippetsCount > 2000
-для бакета, рекурсивно делим по цене (binary split) до тех пор пока
-каждая ветка укладывается в cap.
-
-# TODO Layer B (#1846 follow-up): backfill renovation/wallType/priceHistory
-через detail-эндпоинт после первоначального сбора.
-"""
-
-from __future__ import annotations
-
-import hashlib
-import json
-import logging
-from datetime import date
-from typing import Any
-from urllib.parse import urlencode
-
-from app.services.scraper_settings import get_scraper_delay
-from app.services.scrapers.base import BaseScraper, ScrapedLot
-from app.services.scrapers.domclick_exceptions import DomClickBlockedError
-from app.services.scrapers.repair_state_normalizer import infer_repair_state_from_text
-
-logger = logging.getLogger(__name__)
-
-# ── API constants ─────────────────────────────────────────────────────────────
-
-_BFF_BASE = "https://bff-search-web.domclick.ru"
-_EKB_ADDRESS_GUID = "0d475b79-88de-4054-818c-37d8f9d0d440"
-_EKB_AREA_ID = "20561"
-
-# Buckets to sweep — порядок влияет на логи.
-ROOM_BUCKETS: tuple[str, ...] = ("st", "1", "2", "3", "4", "5+")
-
-OFFSET_CAP: int = 2000  # max offset принятый BFF API
-PAGE_SIZE: int = 20  # items per page (жёстко задан API)
-# Потолок цены для binary split. 1 млрд ₽ — безопасно выше любой ЕКБ-квартиры;
-# выбран так чтобы НЕ отбрасывать листинги дороже потолка при сплите unbounded-бакета
-# (родительский count не ограничен сверху → clamped _lte должен покрывать весь хвост).
-LTE_MAX: int = 1_000_000_000
-MIN_PRICE_SPAN: int = 100_000  # ниже этого span прекращаем делить
-
-# ── QRATOR block detection ────────────────────────────────────────────────────
-
-_QRATOR_MARKERS: tuple[str, ...] = (
-    "qrator",
-    "bot_mitigation",
-    "система защиты",
-    "403 | домклик",
-    "captcha",
-    "access denied",
-)
-
-# ── EKB geo guard ─────────────────────────────────────────────────────────────
-
-_EKB_LAT_MIN: float = 56.6
-_EKB_LAT_MAX: float = 57.0
-_EKB_LON_MIN: float = 60.2
-_EKB_LON_MAX: float = 60.9
-_EKB_REGION_NAME: str = "Екатеринбург"
-
-
-# ── JSON extraction ───────────────────────────────────────────────────────────
-
-
-def _extract_json(html: str) -> dict[str, Any]:
-    """Извлекает JSON-объект из ответа BrowserFetcher.
-
-    BrowserFetcher оборачивает JSON в HTML-страницу двумя способами:
-    1. Bare body: ``{"result": ...}`` напрямую в .
-    2. 
-wrapped: ``
{"result": ...}
``.
-
-    Стратегия: найти первый '{' и последний '}' — работает для обоих вариантов.
-
-    Raises:
-        DomClickBlockedError: если ответ содержит QRATOR/captcha маркеры.
-        ValueError: если JSON не найден или не является dict.
-    """
-    # Сканируем ВЕСЬ ответ (а не только первые 4096B): block-маркер может
-    # стоять за пределами head в крупных challenge-страницах.
-    html_lower = html.lower()
-    if any(m in html_lower for m in _QRATOR_MARKERS):
-        raise DomClickBlockedError(
-            f"DomClick BFF: QRATOR block page detected (markers checked: {_QRATOR_MARKERS[:2]})"
-        )
-
-    start = html.find("{")
-    end = html.rfind("}")
-    if start == -1 or end == -1 or end <= start:
-        raise ValueError(f"No JSON object found in BFF response (len={len(html)})")
-
-    data = json.loads(html[start : end + 1])
-    if not isinstance(data, dict):
-        raise ValueError(f"BFF response JSON is not a dict: {type(data)}")
-    return data  # type: ignore[return-value]
-
-
-# ── URL builders ──────────────────────────────────────────────────────────────
-
-
-def _build_offers_url(
-    rooms: str,
-    price_gte: int | None,
-    price_lte: int | None,
-    offset: int,
-) -> str:
-    """Строит URL для GET /api/offers/v1 с пагинацией.
-
-    urlencode кодирует "5+" → "5%2B" (literal '+' в query string = space → reject).
-    """
-    params: list[tuple[str, str]] = [
-        ("address", _EKB_ADDRESS_GUID),
-        ("aids", _EKB_AREA_ID),
-        ("deal_type", "sale"),
-        ("category", "living"),
-        ("offer_type", "flat"),
-        ("rooms", rooms),
-        ("sort", "qi"),
-        ("sort_dir", "desc"),
-        ("offset", str(offset)),
-        ("limit", str(PAGE_SIZE)),
-    ]
-    if price_gte is not None:
-        params.append(("sale_price__gte", str(price_gte)))
-    if price_lte is not None:
-        params.append(("sale_price__lte", str(price_lte)))
-    return f"{_BFF_BASE}/api/offers/v1?{urlencode(params)}"
-
-
-def _build_count_url(
-    rooms: str,
-    price_gte: int | None,
-    price_lte: int | None,
-) -> str:
-    """Строит URL для GET /api/offers/count/v1 (без offset/limit)."""
-    params: list[tuple[str, str]] = [
-        ("address", _EKB_ADDRESS_GUID),
-        ("aids", _EKB_AREA_ID),
-        ("deal_type", "sale"),
-        ("category", "living"),
-        ("offer_type", "flat"),
-        ("rooms", rooms),
-        ("sort", "qi"),
-        ("sort_dir", "desc"),
-    ]
-    if price_gte is not None:
-        params.append(("sale_price__gte", str(price_gte)))
-    if price_lte is not None:
-        params.append(("sale_price__lte", str(price_lte)))
-    return f"{_BFF_BASE}/api/offers/count/v1?{urlencode(params)}"
-
-
-def _parse_publish_date(iso: str | None) -> date | None:
-    """Парсит ISO8601 дату публикации → date. None при ошибке."""
-    if not iso:
-        return None
-    try:
-        return date.fromisoformat(iso[:10])
-    except (ValueError, TypeError):
-        return None
-
-
-def _extract_agency_name(seller: dict[str, Any]) -> str | None:
-    """Извлечь читаемое имя агентства/агента из seller-блока BFF offer.
-
-    seller.company / seller.agent могут быть строкой ИЛИ dict (defensive —
-    реальная форма поля варьируется). Приоритет: company → agent. Для dict
-    пробуем типичные name-поля. Возвращает None если ничего читаемого нет.
-    """
-    for key in ("company", "agent"):
-        val = seller.get(key)
-        if isinstance(val, str):
-            name = val.strip()
-            if name:
-                return name
-        elif isinstance(val, dict):
-            for name_key in ("name", "title", "fullName"):
-                nm = val.get(name_key)
-                if isinstance(nm, str) and nm.strip():
-                    return nm.strip()
-    return None
-
-
-# ── DomClickScraper ──────────────────────────────────────────────────────────
-
-
-class DomClickScraper(BaseScraper):
-    """DomClick вторичка через JSON BFF API. Источник = 'domklik'.
-
-    Использует BrowserFetcher(source="domclick") (generic provider → shared mobile
-    proxy), который обходит QRATOR. Прямые curl/httpx-запросы с datacenter-IP
-    блокируются QRATOR.
-
-    Основной метод: fetch_city(city_id, rooms, pages).
-    fetch_around() не реализован: DomClick не поддерживает geo-radius в URL.
-
-    Counters (публичные после fetch_city):
-        parse_failures  — офферы с ошибкой маппинга
-        geo_filtered    — офферы вне ЕКБ bbox или с неверным offerRegionName
-        blocked         — True если sweep был прерван QRATOR-блоком
-        fetch_errors    — не-block ошибки извлечения JSON (truncated/garbled/bad shape)
-    """
-
-    name = "domklik"
-    source = "domklik"
-    base_url = "https://domclick.ru"
-    # Консервативная задержка между страницами (fallback до init)
-    request_delay_sec = 8.0
-
-    def __init__(self) -> None:
-        super().__init__()
-        self.request_delay_sec = get_scraper_delay(self.name)
-        self.parse_failures: int = 0
-        self.geo_filtered: int = 0
-        self.blocked: bool = False
-        # Не-block ошибки извлечения JSON (truncated/garbled response, неверная
-        # структура). В отличие от parse_failures (per-item), это per-fetch ошибки,
-        # которые ограничивают сбор бакета. Учитываются в honest-status pipeline.
-        self.fetch_errors: int = 0
-
-    async def __aenter__(self) -> DomClickScraper:
-        await super().__aenter__()
-        return self
-
-    # ── fetch_around — stub ───────────────────────────────────────────────────
-
-    async def fetch_around(self, lat: float, lon: float, radius_m: int = 1000) -> list[ScrapedLot]:
-        """DomClick не поддерживает geo-radius. Используй fetch_city()."""
-        raise NotImplementedError(
-            "DomClick не поддерживает geo-radius; используй fetch_city(city_id=...)"
-        )
-
-    # ── Основной метод ────────────────────────────────────────────────────────
-
-    async def fetch_city(
-        self,
-        city_id: int,
-        rooms: list[int] | None = None,
-        pages: int = 100,
-    ) -> list[ScrapedLot]:
-        """Citywide sweep через BFF JSON API.
-
-        Аргументы city_id и rooms принимаются для совместимости сигнатуры с
-        вызывающим кодом (run_domclick_city_sweep), но:
-        - city_id — vestigial (EКБ захардкожен через GUID).
-        - rooms — игнорируется; всегда обходятся все ROOM_BUCKETS внутри.
-
-        Args:
-            city_id: игнорируется (EKB захардкожен).
-            rooms: игнорируется (ROOM_BUCKETS перебирается всегда).
-            pages: максимальное число страниц на бакет (safety cap).
-
-        Returns:
-            Дедуплицированный по source_id список ScrapedLot.
-        """
-        # TODO Layer B (#1846 follow-up): backfill renovation/wallType/priceHistory
-
-        from app.services.scrapers.browser_fetcher import BrowserFetcher
-
-        out_lots: list[ScrapedLot] = []
-        seen_ids: set[str] = set()
-
-        async with BrowserFetcher(source="domclick") as fetcher:
-            for bucket in ROOM_BUCKETS:
-                logger.info(
-                    "domklik: BFF sweep rooms=%r city_id=%d pages_cap=%d",
-                    bucket,
-                    city_id,
-                    pages,
-                )
-                try:
-                    await self._sweep_bucket(
-                        fetcher=fetcher,
-                        rooms=bucket,
-                        price_gte=None,
-                        price_lte=None,
-                        seen_ids=seen_ids,
-                        out_lots=out_lots,
-                        pages=pages,
-                    )
-                except DomClickBlockedError:
-                    self.blocked = True
-                    logger.error(
-                        "domklik: QRATOR block during rooms=%r — aborting all buckets",
-                        bucket,
-                    )
-                    break
-                except (ValueError, TypeError) as exc:
-                    # Defensive: bucket-level ошибка не должна убивать весь sweep.
-                    # _count/_paginate уже глотают эти ошибки per-fetch (fetch_errors++),
-                    # но если что-то всё же всплыло — переходим к следующему бакету.
-                    self.fetch_errors += 1
-                    logger.warning(
-                        "domklik: bucket rooms=%r failed (%s) — skipping to next bucket",
-                        bucket,
-                        exc,
-                        exc_info=True,
-                    )
-                    continue
-
-        logger.info(
-            "domklik: fetch_city done city_id=%d total=%d "
-            "parse_failures=%d geo_filtered=%d fetch_errors=%d blocked=%s",
-            city_id,
-            len(out_lots),
-            self.parse_failures,
-            self.geo_filtered,
-            self.fetch_errors,
-            self.blocked,
-        )
-        return out_lots
-
-    # ── Internal sweep helpers ────────────────────────────────────────────────
-
-    async def _count(
-        self,
-        fetcher: Any,
-        rooms: str,
-        price_gte: int | None,
-        price_lte: int | None,
-    ) -> int:
-        """Запрашивает snippetsCount для данного бакета (rooms + price range).
-
-        DomClickBlockedError пробрасывается наверх (abort sweep). Прочие ошибки
-        извлечения JSON (truncated/garbled/bad shape) → fetch_errors++ и return 0
-        (бакет пропускается, sweep продолжается).
-        """
-        url = _build_count_url(rooms, price_gte, price_lte)
-        logger.debug("domklik: count url=%s", url)
-        html = await fetcher.fetch(url)
-        try:
-            data = _extract_json(html)
-        except DomClickBlockedError:
-            raise
-        except (ValueError, TypeError, AttributeError):
-            self.fetch_errors += 1
-            logger.warning(
-                "domklik: _count JSON extract failed rooms=%r price=[%s,%s] — skip bucket",
-                rooms,
-                price_gte,
-                price_lte,
-                exc_info=True,
-            )
-            return 0
-        # {"result": null} → None, {"snippetsCount": null} → None: harden обе.
-        res = data.get("result") or {}
-        raw = res.get("snippetsCount")
-        return int(raw) if raw else 0
-
-    async def _sweep_bucket(
-        self,
-        fetcher: Any,
-        rooms: str,
-        price_gte: int | None,
-        price_lte: int | None,
-        seen_ids: set[str],
-        out_lots: list[ScrapedLot],
-        pages: int,
-    ) -> None:
-        """Рекурсивный sweep бакета (rooms, price_gte, price_lte).
-
-        Если snippetsCount > OFFSET_CAP — делим диапазон цен пополам и
-        рекурсируем в каждую половину. Остановка рекурсии:
-        - span <= MIN_PRICE_SPAN → пагинируем как есть (с предупреждением о truncation)
-        - snippetsCount == 0 → пропускаем
-        - snippetsCount <= OFFSET_CAP → пагинируем
-
-        Raises:
-            DomClickBlockedError: если QRATOR-блок — propagate наверх.
-        """
-        count = await self._count(fetcher, rooms, price_gte, price_lte)
-        if count == 0:
-            return
-
-        if count > OFFSET_CAP:
-            _gte = price_gte if price_gte is not None else 0
-            _lte = price_lte if price_lte is not None else LTE_MAX
-            span = _lte - _gte
-            if span <= MIN_PRICE_SPAN:
-                logger.warning(
-                    "domklik: rooms=%r price=[%s,%s] count=%d > cap=%d "
-                    "but span=%d <= min=%d — paginating as-is (bucket truncated at %d)",
-                    rooms,
-                    price_gte,
-                    price_lte,
-                    count,
-                    OFFSET_CAP,
-                    span,
-                    MIN_PRICE_SPAN,
-                    OFFSET_CAP,
-                )
-                await self._paginate(
-                    fetcher, rooms, price_gte, price_lte, seen_ids, out_lots, pages
-                )
-                return
-            mid = (_gte + _lte) // 2
-            logger.debug(
-                "domklik: rooms=%r count=%d > cap=%d — price-split [%d,%d] → [%d,%d]+[%d,%d]",
-                rooms,
-                count,
-                OFFSET_CAP,
-                _gte,
-                _lte,
-                _gte,
-                mid,
-                mid + 1,
-                _lte,
-            )
-            await self._sweep_bucket(fetcher, rooms, _gte, mid, seen_ids, out_lots, pages)
-            await self._sweep_bucket(fetcher, rooms, mid + 1, _lte, seen_ids, out_lots, pages)
-        else:
-            await self._paginate(fetcher, rooms, price_gte, price_lte, seen_ids, out_lots, pages)
-
-    async def _paginate(
-        self,
-        fetcher: Any,
-        rooms: str,
-        price_gte: int | None,
-        price_lte: int | None,
-        seen_ids: set[str],
-        out_lots: list[ScrapedLot],
-        pages: int,
-    ) -> None:
-        """Пагинирует один бакет (rooms, price range) до исчерпания или cap.
-
-        DomClickBlockedError пробрасывается наверх (abort sweep). Прочие ошибки
-        извлечения JSON (truncated/garbled/bad shape) → fetch_errors++ и break
-        (трактуем как конец бакета — уже собранные lots сохраняются).
-
-        Raises:
-            DomClickBlockedError: propagate из _extract_json при QRATOR-блоке.
-        """
-        force_rooms = 0 if rooms == "st" else None
-        page_idx = 0
-        while page_idx < pages:
-            offset = page_idx * PAGE_SIZE
-            if offset >= OFFSET_CAP:
-                break
-            url = _build_offers_url(rooms, price_gte, price_lte, offset)
-            logger.debug("domklik: offers url=%s", url)
-            html = await fetcher.fetch(url)
-            try:
-                data = _extract_json(html)
-            except DomClickBlockedError:
-                raise
-            except (ValueError, TypeError, AttributeError):
-                self.fetch_errors += 1
-                logger.warning(
-                    "domklik: _paginate JSON extract failed rooms=%r offset=%d "
-                    "— ending bucket (lots so far preserved)",
-                    rooms,
-                    offset,
-                    exc_info=True,
-                )
-                break
-
-            # {"result": null} → None; harden до .get("items").
-            items: list[dict[str, Any]] = (data.get("result") or {}).get("items") or []
-            if not items:
-                break
-
-            for item in items:
-                lot = self._map_item(item, force_rooms=force_rooms)
-                if lot is None:
-                    continue
-                if not self._is_geo_ok(item):
-                    self.geo_filtered += 1
-                    continue
-                key = lot.source_id or lot.source_url
-                if key in seen_ids:
-                    continue
-                seen_ids.add(key)
-                out_lots.append(lot)
-
-            page_idx += 1
-            if page_idx < pages and len(items) == PAGE_SIZE:
-                await self.sleep_between_requests()
-            elif not items or len(items) < PAGE_SIZE:
-                break
-
-    def _is_geo_ok(self, item: dict[str, Any]) -> bool:
-        """Гео-гард: пропускает только листинги ЕКБ в bbox.
-
-        aids=20561 даёт чистый ЕКБ, но гард оставляем как defensive проверку.
-        """
-        region = item.get("offerRegionName", "")
-        if region != _EKB_REGION_NAME:
-            return False
-        loc = item.get("location") or {}
-        lat = loc.get("lat")
-        lon = loc.get("lon")
-        if lat is None or lon is None:
-            return False
-        return (
-            _EKB_LAT_MIN <= float(lat) <= _EKB_LAT_MAX
-            and _EKB_LON_MIN <= float(lon) <= _EKB_LON_MAX
-        )
-
-    def _map_item(
-        self, item: dict[str, Any], *, force_rooms: int | None = None
-    ) -> ScrapedLot | None:
-        """Маппинг offer-item из BFF API → ScrapedLot.
-
-        Возвращает None если price <= 0 или при ошибке парсинга.
-        """
-        try:
-            price = item.get("price", 0) or 0
-            if price <= 0:
-                return None
-
-            item_id = item.get("id")
-            source_id = str(item_id) if item_id is not None else None
-            source_url: str = item.get("path", "") or ""
-
-            loc = item.get("location") or {}
-            lat: float | None = loc.get("lat")
-            lon: float | None = loc.get("lon")
-
-            address_obj = item.get("address") or {}
-            address: str | None = address_obj.get("displayName")
-
-            # 0 — sentinel «неизвестно» для area/floor/total_floors/buildYear
-            # (DomClick эмитит 0 для under-construction / unknown) → None.
-            obj_info = item.get("objectInfo") or {}
-            area_raw = obj_info.get("area")
-            area_m2: float | None = float(area_raw) if area_raw else None
-            floor_raw = obj_info.get("floor")
-            floor: int | None = int(floor_raw) if floor_raw else None
-
-            house = item.get("house") or {}
-            floors_raw = house.get("floors")
-            total_floors: int | None = int(floors_raw) if floors_raw else None
-            year_raw = house.get("buildYear")
-            year_built: int | None = int(year_raw) if year_raw else None
-
-            # rooms: 0 — ВАЛИДНОЕ значение (студия), поэтому проверка is not None,
-            # а не truthy. force_rooms=0 (st-бакет) перебивает objectInfo.
-            if force_rooms is not None:
-                rooms: int | None = force_rooms
-            else:
-                rooms_raw = obj_info.get("rooms")
-                rooms = int(rooms_raw) if rooms_raw is not None else None
-
-            square_price_raw = item.get("squarePrice")
-            price_per_m2: int | None = int(square_price_raw) if square_price_raw else None
-            if price_per_m2 is None and area_m2 and area_m2 > 0:
-                price_per_m2 = int(price / area_m2)
-
-            # ── Описание + repair-инференс + minhash (Layer A enrichment) ─────
-            # DomClick search не отдаёт renovation-enum: repair_state выводим из
-            # текста описания (как cian #622 fallback — «евроремонт»/«без отделки»).
-            # description_minhash — sha1 от описания (mirror cian) для cross-source
-            # дедупа.
-            description_raw = item.get("description")
-            description: str | None = None
-            if isinstance(description_raw, str):
-                description = description_raw.strip() or None
-
-            repair_state: str | None = None
-            description_minhash: str | None = None
-            if description:
-                repair_state = infer_repair_state_from_text(description)
-                description_minhash = hashlib.sha1(
-                    description.lower().encode("utf-8", errors="replace")
-                ).hexdigest()[:32]
-
-            # ── Продавец / агентство ──────────────────────────────────────────
-            seller = item.get("seller")
-            if not isinstance(seller, dict):
-                seller = {}
-            agency_name = _extract_agency_name(seller)
-
-            flat_complex = item.get("flatComplex") or {}
-            raw_payload: dict[str, Any] = {
-                "isRosreestrApproved": item.get("isRosreestrApproved"),
-                "squarePrice": square_price_raw,
-                "lastPriceHistoryState": item.get("lastPriceHistoryState"),
-                "flatComplex": {
-                    "name": flat_complex.get("name"),
-                    "id": flat_complex.get("id"),
-                    "slug": flat_complex.get("slug"),
-                }
-                if flat_complex
-                else None,
-                "updatedDate": item.get("updatedDate"),
-            }
-
-            # ── Доп. SERP-поля (Layer A enrichment) — мержим в raw_payload, ────
-            # сбрасывая None-значения чтобы payload оставался компактным; ничего
-            # из существующих ключей не теряем.
-            seller_compact = {
-                k: seller.get(k) for k in ("agent", "company") if seller.get(k) is not None
-            }
-            _enrich: dict[str, Any] = {
-                "is_sber_collateral": item.get("isSberCollateral"),
-                "has_discount": item.get("hasDiscount"),
-                "discount_value": item.get("discountValue"),
-                "duplicates_offer_count": item.get("duplicatesOfferCount"),
-                "seller": seller_compact or None,
-            }
-            raw_payload.update({k: v for k, v in _enrich.items() if v is not None})
-
-            publish_date = _parse_publish_date(item.get("publishedDate"))
-
-            return ScrapedLot(
-                source="domklik",
-                source_url=source_url,
-                source_id=source_id,
-                address=address,
-                lat=lat,
-                lon=lon,
-                rooms=rooms,
-                area_m2=area_m2,
-                floor=floor,
-                total_floors=total_floors,
-                year_built=year_built,
-                price_rub=int(price),
-                price_per_m2=price_per_m2,
-                listing_segment="vtorichka",
-                publish_date=publish_date,
-                description=description,
-                repair_state=repair_state,
-                description_minhash=description_minhash,
-                agency_name=agency_name,
-                raw_payload=raw_payload,
-            )
-        except Exception:
-            self.parse_failures += 1
-            logger.warning(
-                "domklik: _map_item failed for item id=%r (parse_failures=%d)",
-                item.get("id"),
-                self.parse_failures,
-                exc_info=True,
-            )
-            return None
-
-
-# ── Convenience runner (для Celery tasks) ────────────────────────────────────
-
-
-async def scrape_domclick_city(
-    city_id: int,
-    rooms: list[int] | None = None,
-    pages: int = 100,
-) -> list[ScrapedLot]:
-    """Удобная точка входа для вызова из Celery tasks.
-
-    Пример::
-
-        import asyncio
-        lots = asyncio.run(scrape_domclick_city(city_id=4, rooms=None))
-    """
-    async with DomClickScraper() as scraper:
-        return await scraper.fetch_city(city_id=city_id, rooms=rooms, pages=pages)
diff --git a/tradein-mvp/backend/app/services/scrapers/domclick_exceptions.py b/tradein-mvp/backend/app/services/scrapers/domclick_exceptions.py
deleted file mode 100644
index 247de7fe..00000000
--- a/tradein-mvp/backend/app/services/scrapers/domclick_exceptions.py
+++ /dev/null
@@ -1,27 +0,0 @@
-"""DomClick-specific exceptions для anti-bot detection."""
-
-
-class DomClickBlockedError(Exception):
-    """DomClick BFF вернул QRATOR block-страницу (HTTP 200 + block HTML).
-
-    QRATOR (qrator.net) — WAF/DDoS-защита domclick.ru. Блокирует datacenter-IP
-    и возвращает HTML с маркерами: "qrator", "bot_mitigation", "система защиты",
-    "403 | домклик", "captcha", "access denied".
-
-    Обходится через shared mobile proxy (BrowserFetcher(source="domclick") →
-    generic provider → мобильный egress).
-    """
-
-
-class DomClickParseError(Exception):
-    """DomClick detail-карточка получена (HTTP 200, не block), но SSR-стейт
-    не извлекается / не парсится.
-
-    Возникает в Layer B (domclick_detail.py) когда:
-    - `window.__SSR_STATE__` не найден в HTML (но это и не challenge-страница);
-    - object-литерал не сбалансирован по скобкам;
-    - json.loads падает после санитайза bare `undefined` → `null`.
-
-    В отличие от DomClickBlockedError (anti-bot), это сигнал дрейфа схемы карточки
-    или повреждённого ответа — оркестратор считает его parse-failure, а не блоком.
-    """
diff --git a/tradein-mvp/backend/app/services/scrapers/yandex_realty.py b/tradein-mvp/backend/app/services/scrapers/yandex_realty.py
deleted file mode 100644
index 640a97f2..00000000
--- a/tradein-mvp/backend/app/services/scrapers/yandex_realty.py
+++ /dev/null
@@ -1,1107 +0,0 @@
-"""Yandex.Nedvizhimost scraper (realty.yandex.ru) -- gate-API JSON parser.
-
-Transport: BrowserFetcher (camoufox real-browser fingerprint via tradein-browser service).
-camoufox's fingerprint is not tarpitted by yandex and does not burn the mobile-proxy IP;
-raw system-curl over SOCKS5 gets the proxy IP flagged. One BrowserFetcher context is opened
-per scraper session (__aenter__) and reused across all page fetches.
-
-camoufox returns the gate-API JSON wrapped in HTML (verified prod 2026-06-17):
-    ...
{"response":...}

-JSON is extracted via _extract_json_from_content() (handles the 
 wrapper).
-
-API path: GET https://realty.yandex.ru/gate/react-page/get/
-  _pageType=search&_providers=react-search-results-data
-  type=SELL&category=APARTMENT&newFlat=NO&rgid=559132 (EKB)
-  roomsTotal=&priceMin=...&priceMax=...&page=<1-based>
-
-Response shape: response.search.offers.{entities[], pager{page,pageSize,totalItems,totalPages}}
-
-Page param is 1-based: page=1 -> first page (pager.page=0 in response).
-page=0 -> API returns error.  Paginate from page=1 to totalPages (inclusive).
-
-Studio roomsTotal:
-  Request: roomsTotal=STUDIO
-  Response entities: roomsTotal=None -> mapped to rooms=0 (existing convention).
-"""
-
-from __future__ import annotations
-
-import asyncio
-import json
-import logging
-import math
-import re
-from dataclasses import dataclass
-from datetime import UTC, date, datetime
-from typing import Any
-from urllib.parse import urlencode
-
-from curl_cffi.requests import AsyncSession as _CurlCffiSession
-
-from app.services.scraper_settings import get_scraper_delay
-from app.services.scrapers.base import BaseScraper, ScrapedLot
-from app.services.scrapers.browser_fetcher import BrowserFetcher
-from app.services.scrapers.house_type_normalizer import normalize_house_type
-from app.services.scrapers.price_brackets import get_price_seed_brackets
-
-logger = logging.getLogger(__name__)
-
-# camoufox renders JSON endpoints as 
{"response":...}
 inside an HTML wrapper.
-_RE_PRE_JSON = re.compile("]*>(\\{.*?)
", re.DOTALL | re.IGNORECASE)
-
-DEFAULT_CITY = "ekaterinburg"
-_GATE_MAX_PAGES_CAP = 50
-_EKB_RGID = 559132
-_GATE_URL = "https://realty.yandex.ru/gate/react-page/get/"
-
-ROOM_GATE_PARAM: dict[str, str | int] = {
-    "studio": "STUDIO",
-    "1": 1,
-    "2": 2,
-    "3": 3,
-    "4+": 4,
-}
-
-ROOM_PATH: dict[str, str] = {
-    "studio": "studiya",
-    "1": "odnokomnatnaya",
-    "2": "dvuhkomnatnaya",
-    "3": "trehkomnatnaya",
-    "4+": "4-i-bolee",
-}
-
-DEFAULT_PRICE_RANGES: list[tuple[int | None, int | None]] = [
-    (None, 5_000_000),
-    (5_000_000, 7_000_000),
-    (7_000_000, 10_000_000),
-    (10_000_000, 15_000_000),
-    (15_000_000, 25_000_000),
-    (25_000_000, None),
-]
-
-_YANDEX_MAX_PRICE = 200_000_000
-_YANDEX_MIN_BRACKET = 50_000
-_GATE_PAGE_SIZE = 20
-_YANDEX_TARPIT_MAX_RETRIES: int = 2
-
-# Таймаут httpx-клиента BrowserFetcher для yandex-пути: 30s вместо глобального 120s.
-# Один проблемный combo занимает ≤30s×(1+_YANDEX_TARPIT_MAX_RETRIES)=90s вместо 360s.
-_YANDEX_BROWSER_FETCH_TIMEOUT_S: float = 30.0
-
-
-@dataclass
-class _CurlResponse:
-    status_code: int
-    text: str
-
-
-def _is_gate_error(payload: dict[str, Any]) -> bool:
-    """Return True when gate-API returned an error object instead of search results."""
-    return "error" in payload or "response" not in payload
-
-
-def _extract_gate_data(
-    payload: dict[str, Any],
-) -> tuple[list[dict[str, Any]], dict[str, Any]] | None:
-    """Extract (entities, pager) from valid gate-API response. None on missing path."""
-    try:
-        offers = payload["response"]["search"]["offers"]
-        entities: list[dict[str, Any]] = offers["entities"]
-        pager: dict[str, Any] = offers["pager"]
-        return entities, pager
-    except (KeyError, TypeError):
-        return None
-
-
-def _extract_json_from_content(content: str) -> str | None:
-    """Extract raw JSON text from camoufox HTML-wrapped content.
-
-    camoufox renders a JSON endpoint as:
-        ...
{"response":...}

-
-    Strategy:
-    1. Try 
...
 regex (primary -- matches prod output).
-    2. Fallback: slice from the first '{' (handles raw-JSON response, no wrapper).
-
-    Returns the JSON string, or None if no '{' is found.
-    """
-    m = _RE_PRE_JSON.search(content)
-    if m:
-        return m.group(1)
-
-    idx = content.find("{")
-    if idx != -1:
-        return content[idx:]
-
-    return None
-
-
-def _segment_for(new_flat: str) -> str:
-    """Map a newFlat gate-API value to the listings.listing_segment value."""
-    return "novostroyki" if new_flat == "YES" else "vtorichka"
-
-
-def _parse_creation_date(raw: Any) -> tuple[date | None, int | None]:
-    """Parse gate-API creationDate ('2026-05-24T15:10:27Z') → (date, days_on_market).
-
-    days_on_market = floor((now - creationDate).days), clamped to >= 0.
-    Returns (None, None) when raw is absent or unparseable (never raises).
-    """
-    if not raw or not isinstance(raw, str):
-        return None, None
-    txt = raw.strip()
-    # datetime.fromisoformat handles 'Z' only from py3.11+, but normalise defensively.
-    if txt.endswith("Z"):
-        txt = txt[:-1] + "+00:00"
-    try:
-        dt = datetime.fromisoformat(txt)
-    except ValueError:
-        return None, None
-    if dt.tzinfo is None:
-        dt = dt.replace(tzinfo=UTC)
-    listing_date = dt.date()
-    delta_days = (datetime.now(UTC) - dt).days
-    days_on_market = max(delta_days, 0)
-    return listing_date, days_on_market
-
-
-def _author_fields(author: dict[str, Any]) -> tuple[bool | None, bool | None, str | None]:
-    """Map entity.author → (is_homeowner, is_pro_seller, agency_name).
-
-    author.category ∈ {OWNER, AGENCY, AGENT, ...}.
-      OWNER          → is_homeowner=True
-      AGENCY / AGENT → is_pro_seller=True, agency_name = organization or agentName
-    Returns (None, None, None) when author/category absent.
-    """
-    category = (author.get("category") or "").upper()
-    if not category:
-        return None, None, None
-    if category == "OWNER":
-        return True, None, None
-    if category in {"AGENCY", "AGENT"}:
-        agency_name = author.get("organization") or author.get("agentName") or None
-        return None, True, agency_name
-    return None, None, None
-
-
-def _to_bigint(raw: Any) -> int | None:
-    """Coerce a gate-API numeric value (str or number) to int. None when unparseable."""
-    if raw is None:
-        return None
-    try:
-        return int(float(raw))
-    except (TypeError, ValueError):
-        return None
-
-
-def _predicted_prices(entity: dict[str, Any]) -> tuple[int | None, int | None, int | None]:
-    """Extract (value, min, max) from entity.predictions.predictedPrice.
-
-    Yandex's own per-offer valuation. The gate-API nests it under
-    ``predictions.predictedPrice`` (NOT ``predictedPrice.predictedPrice`` —
-    that key does not exist; verified live 2026-06-18 against the gate-API).
-    Fields are strings (e.g. "7554000"). Returns (None, None, None) when the
-    nested path is absent (e.g. novostroyki offers carry no valuation).
-    """
-    outer = entity.get("predictions") or {}
-    inner = outer.get("predictedPrice") or {}
-    if not isinstance(inner, dict):
-        return None, None, None
-    return (
-        _to_bigint(inner.get("value")),
-        _to_bigint(inner.get("min")),
-        _to_bigint(inner.get("max")),
-    )
-
-
-def _metro_stations(location: dict[str, Any]) -> list[dict[str, Any]]:
-    """Build metro_stations [{name, min}] from location.metroList[] (+ location.metro)."""
-    stations: list[dict[str, Any]] = []
-    raw_list = location.get("metroList")
-    if isinstance(raw_list, list):
-        for m in raw_list:
-            if isinstance(m, dict) and m.get("name"):
-                stations.append({"name": m.get("name"), "min": m.get("timeToMetro")})
-    if not stations:
-        single = location.get("metro")
-        if isinstance(single, dict) and single.get("name"):
-            stations.append({"name": single.get("name"), "min": single.get("timeToMetro")})
-    return stations
-
-
-def _parse_gate_json(
-    payload: dict[str, Any], page_param: int = 1, new_flat: str = "NO"
-) -> list[ScrapedLot]:
-    """Parse gate-API JSON payload into a list of ScrapedLot.
-
-    Verified field mapping (from .issdump/yandex_gate_api_sample.json + live prod 2026-06-17):
-      offerId                         -> source_id
-      url (//realty.yandex.ru/offer/) -> source_url (canonical per-offer, #2235:
-                                         partner/redirect url -> offer_id-based)
-      price.value                     -> price_rub
-      price.valuePerPart              -> price_per_m2
-      area.value                      -> area_m2
-      livingSpace.value               -> living_area_m2
-      kitchenSpace.value              -> kitchen_area_m2 (+ raw_payload["kitchen_area_m2"])
-      roomsTotal (int | None)         -> rooms (None -> 0 for studio)
-      floorsOffered[0]                -> floor
-      floorsTotal                     -> total_floors
-      building.builtYear              -> year_built
-      building.buildingType           -> house_type (normalize_house_type: MONOLIT->monolith)
-      ceilingHeight                   -> ceiling_height_m (+ raw_payload["ceiling_height"])
-      location.point.latitude/longitude -> lat / lon
-      location.geocoderAddress        -> address (full, with house number)
-      mainImages[]                    -> photo_urls (prepend "https:", up to 5)
-      building.siteId                 -> house_ext_id (JK linkage)
-      listing_segment                 -> "novostroyki" (newFlat=YES) | "vtorichka" (newFlat=NO)
-      creationDate (ISO Z)            -> listing_date / publish_date + days_on_market
-      description                     -> description
-      author.category/organization   -> is_homeowner / is_pro_seller / agency_name
-      predictedPrice.predictedPrice.* -> predicted_price_rub / _min / _max
-      price.trend / price.previous    -> price_trend / price_previous_rub
-      location.metroList[]            -> metro_stations [{name, min}]
-      offerId                         -> yandex_offer_id
-    """
-    result = _extract_gate_data(payload)
-    if result is None:
-        logger.warning("yandex gate: response missing response.search.offers path")
-        return []
-
-    entities, _pager = result
-    lots: list[ScrapedLot] = []
-    for entity in entities:
-        lot = _entity_to_lot(entity, page_param=page_param, new_flat=new_flat)
-        if lot is not None:
-            lots.append(lot)
-    return lots
-
-
-def _canonical_source_url(url_raw: str, offer_id: str) -> str:
-    """Каноничный per-offer source_url для yandex-оффера (#2235).
-
-    gate-API `url` для вторички = `//realty.yandex.ru/offer/` (уникален на
-    оффер). Для новостроек/partner-офферов там НЕканоническая ссылка на сайт
-    застройщика или рекламный редирект (na100.pro/go.php, atlasdevelopment.su, …),
-    которую делят МНОГО разных квартир одного ЖК → source_url перестаёт быть
-    уникальным (204 «дубля» среди is_active). Возвращаем стабильный уникальный
-    URL из offer_id, если исходный не ведёт на realty.yandex; иначе — исходный.
-    """
-    source_url = f"https:{url_raw}" if url_raw.startswith("//") else url_raw
-    is_yandex_realty = "realty.yandex" in source_url or "realty.ya.ru" in source_url
-    if not source_url or not is_yandex_realty:
-        return f"https://realty.yandex.ru/offer/{offer_id}/"
-    return source_url
-
-
-def _entity_to_lot(
-    entity: dict[str, Any], page_param: int = 1, new_flat: str = "NO"
-) -> ScrapedLot | None:
-    """Convert one gate-API entity dict to ScrapedLot. None on missing required fields."""
-    try:
-        offer_id = str(entity.get("offerId") or "")
-        if not offer_id:
-            return None
-
-        # #2235: для новостроек gate-API кладёт в `url` НЕканоническую ссылку —
-        # сайт застройщика / рекламный редирект (na100.pro/go.php, atlasdevelopment.su,
-        # unistroyrf.ru, …). Много разных офферов одного ЖК делят ОДНУ такую страницу
-        # → 204 «дубля» source_url среди is_active (разные source_id, разные квартиры).
-        # Канонизируем: если `url` не ведёт на realty.yandex — строим стабильный
-        # per-offer URL из offer_id (он всегда есть, см. проверку выше). Вторичный
-        # //realty.yandex.ru/offer/ оставляем как есть.
-        source_url = _canonical_source_url(entity.get("url") or "", offer_id)
-
-        price_dict = entity.get("price") or {}
-        price_val = price_dict.get("value")
-        if not price_val or price_val <= 0:
-            return None
-        price_rub = int(price_val)
-
-        price_per_m2_raw = price_dict.get("valuePerPart")
-        price_per_m2 = int(price_per_m2_raw) if price_per_m2_raw else None
-
-        area_dict = entity.get("area") or {}
-        area_m2_raw = area_dict.get("value")
-        area_m2 = float(area_m2_raw) if area_m2_raw is not None else None
-
-        living_dict = entity.get("livingSpace") or {}
-        living_area_raw = living_dict.get("value")
-        living_area_m2 = float(living_area_raw) if living_area_raw is not None else None
-
-        kitchen_dict = entity.get("kitchenSpace") or {}
-        kitchen_area_raw = kitchen_dict.get("value")
-        kitchen_area_m2 = float(kitchen_area_raw) if kitchen_area_raw is not None else None
-
-        rooms_raw = entity.get("roomsTotal")
-        if rooms_raw is None:
-            rooms = 0  # studio
-        else:
-            try:
-                rooms = int(rooms_raw)
-            except (TypeError, ValueError):
-                rooms = None
-
-        floors_offered = entity.get("floorsOffered") or []
-        floor = int(floors_offered[0]) if floors_offered else None
-        floors_total_raw = entity.get("floorsTotal")
-        total_floors = int(floors_total_raw) if floors_total_raw is not None else None
-
-        ceiling_height = entity.get("ceilingHeight")
-        # Колонка ceiling_height — numeric(3,2), max 9.99. Yandex SERP отдаёт мусор
-        # (видели 18 м) → запись out-of-range уронила бы весь батч DataError'ом
-        # (per-lot SAVEPOINT ловит только IntegrityError). Берём только правдоподобный
-        # диапазон 2.0–6.0 м, иначе None; сырое значение остаётся в raw_payload. (#2007)
-        _ceiling_raw = float(ceiling_height) if ceiling_height is not None else None
-        ceiling_height_m = (
-            _ceiling_raw if _ceiling_raw is not None and 2.0 <= _ceiling_raw <= 6.0 else None
-        )
-
-        building = entity.get("building") or {}
-        year_built_raw = building.get("builtYear")
-        year_built = int(year_built_raw) if year_built_raw is not None else None
-        # buildingType — SCREAMING (MONOLIT/BRICK/...). Нормализуем в канон, иначе
-        # estimator soft-penalty штрафует yandex-аналоги (#2007).
-        raw_building_type = building.get("buildingType")
-        house_type = normalize_house_type(raw_building_type)
-
-        site_id = building.get("siteId")
-        site_name = building.get("siteName")
-        house_source: str | None = None
-        house_ext_id: str | None = None
-        if site_id:
-            house_source = "yandex_realty_nb"
-            house_ext_id = str(site_id)
-
-        location = entity.get("location") or {}
-        point = location.get("point") or {}
-        lat_raw = point.get("latitude")
-        lon_raw = point.get("longitude")
-        lat = float(lat_raw) if lat_raw is not None else None
-        lon = float(lon_raw) if lon_raw is not None else None
-
-        address = location.get("geocoderAddress") or location.get("streetAddress") or None
-
-        # ── Rich gate-API fields ───────────────────────────────────────────
-        listing_date, days_on_market = _parse_creation_date(entity.get("creationDate"))
-        description = entity.get("description") or None
-
-        author = entity.get("author") or {}
-        is_homeowner, is_pro_seller, agency_name = _author_fields(author)
-
-        predicted_value, predicted_min, predicted_max = _predicted_prices(entity)
-
-        price_trend = price_dict.get("trend") or None
-        price_previous_rub = _to_bigint(price_dict.get("previous"))
-
-        metro_stations = _metro_stations(location)
-
-        photo_urls: list[str] = []
-        for img in entity.get("mainImages") or []:
-            if img:
-                photo_urls.append(f"https:{img}" if img.startswith("//") else img)
-            if len(photo_urls) >= 5:
-                break
-
-        return ScrapedLot(
-            source="yandex",
-            source_url=source_url,
-            source_id=offer_id,
-            address=address,
-            lat=lat,
-            lon=lon,
-            rooms=rooms,
-            area_m2=area_m2,
-            living_area_m2=living_area_m2,
-            floor=floor,
-            total_floors=total_floors,
-            year_built=year_built,
-            house_type=house_type,
-            kitchen_area_m2=kitchen_area_m2,
-            ceiling_height_m=ceiling_height_m,
-            price_rub=price_rub,
-            price_per_m2=price_per_m2,
-            photo_urls=photo_urls,
-            house_source=house_source,
-            house_ext_id=house_ext_id,
-            house_url=None,
-            listing_segment=_segment_for(new_flat),
-            listing_date=listing_date,
-            publish_date=listing_date,
-            days_on_market=days_on_market,
-            description=description,
-            is_homeowner=is_homeowner,
-            is_pro_seller=is_pro_seller,
-            agency_name=agency_name,
-            metro_stations=metro_stations,
-            yandex_offer_id=offer_id,
-            predicted_price_rub=predicted_value,
-            predicted_price_min=predicted_min,
-            predicted_price_max=predicted_max,
-            price_trend=price_trend,
-            price_previous_rub=price_previous_rub,
-            raw_payload={
-                "page_param": page_param,
-                "ceiling_height": ceiling_height,
-                "kitchen_area_m2": kitchen_area_raw,
-                "raw_building_type": raw_building_type,  # сырой SCREAMING — для отладки
-                "site_name": site_name,
-                "offer_id": offer_id,
-            },
-        )
-    except Exception:
-        logger.exception("yandex gate: entity parse failed offer_id=%s", entity.get("offerId"))
-        return None
-
-
-class YandexRealtyScraper(BaseScraper):
-    name = "yandex"
-    base_url = "https://realty.yandex.ru"
-    request_delay_sec = 5.0
-
-    def __init__(self, city: str = DEFAULT_CITY) -> None:
-        super().__init__()
-        self.city = city
-        self.request_delay_sec = get_scraper_delay(self.name)
-        self._browser: BrowserFetcher | None = None
-        # _cffi_session retained only for _rotate_ip (changeip call).
-        self._cffi_session: _CurlCffiSession | None = None
-        self._cookies: dict[str, str] = {}
-
-    async def __aenter__(self) -> YandexRealtyScraper:  # type: ignore[override]
-        """Open ONE BrowserFetcher (camoufox) session, reused across all page fetches.
-
-        camoufox's real-browser fingerprint avoids the mobile-proxy IP tarpit
-        escalation that raw system-curl over SOCKS5 triggers. Opening per-page is
-        expensive and unnecessary.
-
-        fetch_timeout_s=_YANDEX_BROWSER_FETCH_TIMEOUT_S (30s) ограничивает потери
-        на один тайм-аутовый запрос: worst-case 30s×(1+retries)=90s на combo вместо 360s.
-        """
-        self._browser = BrowserFetcher(
-            source="yandex", fetch_timeout_s=_YANDEX_BROWSER_FETCH_TIMEOUT_S
-        )
-        await self._browser.__aenter__()
-        return self
-
-    async def __aexit__(self, *args: object) -> None:  # type: ignore[override]
-        if self._browser is not None:
-            await self._browser.__aexit__(*args)
-            self._browser = None
-        if self._cffi_session is not None:
-            await self._cffi_session.close()
-            self._cffi_session = None
-
-    async def _http_get(self, url: str, **kwargs: object) -> _CurlResponse:  # type: ignore[override]
-        """Fetch gate-API URL via the shared BrowserFetcher (camoufox).
-
-        camoufox returns HTML-wrapped JSON; the JSON is extracted and returned as
-        the body of a _CurlResponse so existing callers (json.loads(resp.text),
-        resp.status_code checks) keep working unchanged.
-
-        Maps outcomes onto the curl-style status contract:
-          - JSON extracted  -> status_code=200, text=
-          - fetch raised / no JSON found -> status_code=0 (treated as transient
-            failure by the caller retry logic)
-        """
-        if self._browser is None:
-            raise RuntimeError("YandexRealtyScraper must be used as async context manager")
-
-        try:
-            content = await self._browser.fetch(url)
-        except Exception:
-            logger.warning("yandex gate: BrowserFetcher.fetch failed url=%s", url, exc_info=True)
-            return _CurlResponse(status_code=0, text="")
-
-        json_text = _extract_json_from_content(content)
-        if not json_text:
-            logger.warning(
-                "yandex gate: could not extract JSON from content url=%s first200=%r",
-                url,
-                content[:200],
-            )
-            return _CurlResponse(status_code=0, text="")
-
-        return _CurlResponse(status_code=200, text=json_text)
-
-    async def _rotate_ip(self) -> bool:
-        """Rotate mobile proxy IP via changeip URL. Returns True on success."""
-        from app.core.config import settings as _settings
-
-        rotate_url = _settings.yandex_proxy_rotate_url or _settings.avito_proxy_rotate_url
-        if not rotate_url:
-            return False
-        sep = "&" if "?" in rotate_url else "?"
-        try:
-            async with _CurlCffiSession(timeout=30) as rot:
-                await rot.get(f"{rotate_url}{sep}format=json")
-            await asyncio.sleep(9)
-            logger.info("yandex proxy: IP rotated via changeip")
-            return True
-        except Exception:
-            logger.warning("yandex proxy: IP rotation failed", exc_info=True)
-            return False
-
-    def _build_url(
-        self,
-        page: int = 1,
-        rooms: str | None = None,
-        price_min: int | None = None,
-        price_max: int | None = None,
-        new_flat: str = "NO",
-    ) -> str:
-        """Build gate-API URL. page: 1-based (page=0 -> API error).
-
-        new_flat: "NO" -> vtorichka (default, preserves legacy behavior),
-                  "YES" -> novostroyki (primary-sale / reassignment segment).
-        """
-        params: dict[str, str | int] = {
-            "rgid": _EKB_RGID,
-            "type": "SELL",
-            "category": "APARTMENT",
-            "newFlat": new_flat,
-            "_pageType": "search",
-            "_providers": "react-search-results-data",
-            "page": page,
-        }
-        if rooms and rooms in ROOM_GATE_PARAM:
-            params["roomsTotal"] = ROOM_GATE_PARAM[rooms]
-        if price_min is not None:
-            params["priceMin"] = price_min
-        if price_max is not None:
-            params["priceMax"] = price_max
-        return f"{_GATE_URL}?{urlencode(params)}"
-
-    async def _fetch_page_json(
-        self,
-        rooms: str | None,
-        page: int,
-        price_min: int | None,
-        price_max: int | None,
-        new_flat: str = "NO",
-    ) -> dict[str, Any] | None:
-        """Fetch one gate-API page. Returns parsed JSON dict or None on error/invalid."""
-        url = self._build_url(
-            page=page, rooms=rooms, price_min=price_min, price_max=price_max, new_flat=new_flat
-        )
-        try:
-            resp = await self._http_get(url, timeout=60)
-        except Exception:
-            logger.exception("yandex gate: GET failed url=%s", url)
-            return None
-        if resp.status_code != 200:  # type: ignore[union-attr]
-            logger.warning("yandex gate: HTTP %d url=%s", resp.status_code, url)  # type: ignore[union-attr]
-            return None
-        try:
-            payload: dict[str, Any] = json.loads(resp.text)  # type: ignore[union-attr]
-        except (json.JSONDecodeError, ValueError):
-            logger.warning("yandex gate: JSON parse failed url=%s", url)
-            return None
-        if _is_gate_error(payload):
-            logger.warning("yandex gate: error response url=%s keys=%s", url, list(payload.keys()))
-            return None
-        return payload
-
-    async def fetch_around(
-        self,
-        lat: float,
-        lon: float,
-        radius_m: int = 1000,
-        page: int = 1,
-        rooms: str | None = None,
-        price_min: int | None = None,
-        price_max: int | None = None,
-        new_flat: str = "NO",
-    ) -> list[ScrapedLot]:
-        """Fetch ONE page of gate-API results. lat/lon/radius_m ignored (uses rgid).
-
-        page: 1-based. Tarpit resilience: status_code==0 or JSON error
-        -> rotate IP + retry up to _YANDEX_TARPIT_MAX_RETRIES times.
-        Non-tarpit failures (4xx/5xx) are not retried.
-        new_flat: "NO" -> vtorichka, "YES" -> novostroyki.
-        """
-        url = self._build_url(
-            page=page, rooms=rooms, price_min=price_min, price_max=price_max, new_flat=new_flat
-        )
-
-        payload: dict[str, Any] | None = None
-        for attempt in range(1 + _YANDEX_TARPIT_MAX_RETRIES):
-            try:
-                response = await self._http_get(url, timeout=60)
-            except Exception:
-                logger.exception("yandex gate fetch failed: %s", url)
-                return []
-
-            status = response.status_code  # type: ignore[union-attr]
-
-            if status == 0:
-                logger.warning(
-                    "yandex gate: status=0 (tarpit?) rooms=%s page=%d"
-                    " -- rotating IP + retry attempt %d",
-                    rooms,
-                    page,
-                    attempt + 1,
-                )
-                await self._rotate_ip()
-                await asyncio.sleep(2)
-                continue
-
-            if status != 200:
-                logger.warning(
-                    "yandex gate: HTTP %d rooms=%s page=%d url=%s", status, rooms, page, url
-                )
-                return []
-
-            try:
-                payload = json.loads(response.text)  # type: ignore[union-attr]
-            except (json.JSONDecodeError, ValueError):
-                logger.warning(
-                    "yandex gate: JSON parse failed rooms=%s page=%d -- rotating + retry %d",
-                    rooms,
-                    page,
-                    attempt + 1,
-                )
-                await self._rotate_ip()
-                await asyncio.sleep(2)
-                continue
-
-            if _is_gate_error(payload):
-                logger.warning(
-                    "yandex gate: error payload rooms=%s page=%d keys=%s -- rotating + retry %d",
-                    rooms,
-                    page,
-                    list(payload.keys()),
-                    attempt + 1,
-                )
-                await self._rotate_ip()
-                await asyncio.sleep(2)
-                payload = None
-                continue
-
-            break
-
-        if payload is None:
-            logger.warning(
-                "yandex gate: all retries exhausted rooms=%s page=%d -- returning empty",
-                rooms,
-                page,
-            )
-            return []
-
-        lots = _parse_gate_json(payload, page_param=page, new_flat=new_flat)
-        logger.info(
-            "yandex gate page=%d rooms=%s price=[%s-%s] newFlat=%s: %d cards",
-            page,
-            rooms or "all",
-            price_min,
-            price_max,
-            new_flat,
-            len(lots),
-        )
-        await self.sleep_between_requests()
-        return lots
-
-    async def fetch_around_multi_room(
-        self,
-        lat: float,
-        lon: float,
-        radius_m: int = 1000,
-        max_pages: int = _GATE_MAX_PAGES_CAP,
-        rooms_list: list[str] | None = None,
-        price_ranges: list[tuple[int | None, int | None]] | None = None,
-        segments: list[str] | None = None,
-        on_combo: Any = None,
-        **_legacy_kwargs: Any,
-    ) -> list[ScrapedLot]:
-        """Fetch via segment x rooms x price-range combos; paginate each combo to totalPages.
-
-        Pagination driven by pager.totalPages from first page response.
-        Each combo iterates page=1 .. min(totalPages, max_pages, _GATE_MAX_PAGES_CAP).
-        Deduplicates by source_id/source_url across ALL combos AND segments (one `seen`).
-        Legacy mode (rooms_list=None, price_ranges=None): single citywide sweep.
-        segments: list of newFlat values, default ["NO"] (vtorichka only). Pass
-            ["NO", "YES"] to sweep both vtorichka and novostroyki in one call.
-
-        on_combo: опциональный callback(combo_label: str, new_lots: list[ScrapedLot]) -> None.
-            Вызывается после каждого combo с новыми (не повторными) лотами из него.
-            Позволяет инкрементальный save: вызывающий код сохраняет new_lots сразу,
-            не дожидаясь конца sweep'а. Дубли между combo НЕ передаются повторно.
-        """
-        seen: dict[str, ScrapedLot] = {}
-        _segments = segments or ["NO"]
-        combos_skipped = 0
-
-        if rooms_list is None and price_ranges is None:
-            combos: list[tuple[str | None, int | None, int | None]] = [(None, None, None)]
-        else:
-            r_list = rooms_list or list(ROOM_PATH.keys())
-            p_ranges = price_ranges or DEFAULT_PRICE_RANGES
-            combos = [(r, lo, hi) for r in r_list for lo, hi in p_ranges]
-
-        for new_flat in _segments:
-            _seg = _segment_for(new_flat)
-            for rooms, price_min, price_max in combos:
-                combo_label = f"{_seg}/{_combo_label(rooms, price_min, price_max)}"
-                total_pages: int | None = None
-                combo_new_lots: list[ScrapedLot] = []
-                combo_skipped = False
-
-                for page in range(1, max_pages + 1):
-                    if page == 1:
-                        payload_p1: dict[str, Any] | None = None
-                        p1_url = self._build_url(
-                            page=1,
-                            rooms=rooms,
-                            price_min=price_min,
-                            price_max=price_max,
-                            new_flat=new_flat,
-                        )
-                        for _attempt in range(1 + _YANDEX_TARPIT_MAX_RETRIES):
-                            try:
-                                resp = await self._http_get(p1_url, timeout=60)
-                            except Exception:
-                                logger.exception(
-                                    "yandex gate combo fetch failed combo=%s", combo_label
-                                )
-                                break
-                            if resp.status_code == 0:  # type: ignore[union-attr]
-                                logger.warning(
-                                    "yandex gate: tarpit combo=%s page=1 attempt=%d -- rotating",
-                                    combo_label,
-                                    _attempt + 1,
-                                )
-                                await self._rotate_ip()
-                                await asyncio.sleep(2)
-                                continue
-                            if resp.status_code != 200:  # type: ignore[union-attr]
-                                logger.warning(
-                                    "yandex gate: HTTP %d combo=%s page=1",
-                                    resp.status_code,
-                                    combo_label,  # type: ignore[union-attr]
-                                )
-                                break
-                            try:
-                                payload_p1 = json.loads(resp.text)  # type: ignore[union-attr]
-                            except (json.JSONDecodeError, ValueError):
-                                logger.warning(
-                                    "yandex gate: JSON parse error combo=%s page=1", combo_label
-                                )
-                                payload_p1 = None
-                                await self._rotate_ip()
-                                await asyncio.sleep(2)
-                                continue
-                            if _is_gate_error(payload_p1):
-                                logger.warning(
-                                    "yandex gate: error payload combo=%s page=1", combo_label
-                                )
-                                payload_p1 = None
-                                break
-                            break
-
-                        if payload_p1 is None:
-                            logger.warning(
-                                "yandex gate combo [%s] page=1: all retries exhausted -- skipping",
-                                combo_label,
-                            )
-                            combo_skipped = True
-                            combos_skipped += 1
-                            break
-
-                        result = _extract_gate_data(payload_p1)
-                        if result is None:
-                            combo_skipped = True
-                            combos_skipped += 1
-                            break
-                        _entities_p1, pager_p1 = result
-                        total_pages = min(
-                            pager_p1.get("totalPages", 1),
-                            max_pages,
-                            _GATE_MAX_PAGES_CAP,
-                        )
-                        lots_p1 = _parse_gate_json(payload_p1, page_param=1, new_flat=new_flat)
-                        if not lots_p1:
-                            logger.debug(
-                                "yandex gate combo [%s] page=1: empty -- stopping", combo_label
-                            )
-                            break
-                        for lot in lots_p1:
-                            key = lot.source_id or lot.source_url
-                            if key and key not in seen:
-                                seen[key] = lot
-                                combo_new_lots.append(lot)
-                        await self.sleep_between_requests()
-                        if total_pages <= 1:
-                            break
-                        continue
-
-                    if total_pages is not None and page > total_pages:
-                        break
-
-                    lots = await self.fetch_around(
-                        lat,
-                        lon,
-                        radius_m,
-                        page=page,
-                        rooms=rooms,
-                        price_min=price_min,
-                        price_max=price_max,
-                        new_flat=new_flat,
-                    )
-                    if not lots:
-                        logger.debug(
-                            "yandex gate combo [%s] page=%d: empty -- stopping",
-                            combo_label,
-                            page,
-                        )
-                        break
-                    for lot in lots:
-                        key = lot.source_id or lot.source_url
-                        if key and key not in seen:
-                            seen[key] = lot
-                            combo_new_lots.append(lot)
-
-                # Инкрементальный save: вызываем on_combo если есть новые лоты.
-                # Скипнутые combo (combo_skipped=True) не вызывают on_combo.
-                if on_combo is not None and combo_new_lots and not combo_skipped:
-                    try:
-                        on_combo(combo_label, combo_new_lots)
-                    except Exception:
-                        logger.exception(
-                            "yandex gate: on_combo callback failed combo=%s", combo_label
-                        )
-
-        logger.info(
-            "yandex gate aggregate: %d unique lots (%d combos x %d segments=%s, skipped=%d)",
-            len(seen),
-            len(combos),
-            len(_segments),
-            _segments,
-            combos_skipped,
-        )
-        return list(seen.values())
-
-    async def fetch_all_secondary(
-        self,
-        *,
-        rooms_buckets: list[str] | None = None,
-        price_cap_per_bucket: int = 500,
-        max_pages_per_bucket: int = _GATE_MAX_PAGES_CAP,
-        concurrency: int = 4,
-        on_bucket: Any = None,
-        on_progress: Any = None,
-        skip_buckets: set[str] | None = None,
-    ) -> list[ScrapedLot]:
-        """Exhaustive EKB vtorichka load via KOMNATNOST x TSENA partitioning.
-
-        Uses gate-API JSON (totalItems from pager) instead of DOM state extraction.
-        Стартует с общей data-driven seed-сетки ценовых брекетов
-        (get_price_seed_brackets() — shared EKB grid, общая с avito/cian) вместо
-        одной корень-бисекции [0, _YANDEX_MAX_PRICE]. Закрытые seed-брекеты передают
-        hi = br_hi - 1 (priceMax эксклюзивный сверху → нет пересечений), открытый
-        (br_hi is None) — hi=None (без priceMax, весь хвост люкса). Дедуп по
-        source_id в общем seen страхует на стыках.
-        """
-        _buckets = rooms_buckets if rooms_buckets is not None else list(ROOM_PATH.keys())
-        seen: dict[str, ScrapedLot] = {}
-
-        for rooms_key in _buckets:
-            logger.info(
-                "yandex gate exhaustive: starting rooms=%s (price_cap=%d concurrency=%d)",
-                rooms_key,
-                price_cap_per_bucket,
-                concurrency,
-            )
-            before = len(seen)
-            for br_lo, br_hi in get_price_seed_brackets():
-                walk_hi = br_hi - 1 if br_hi is not None else None
-                await self._walk_price_range(
-                    rooms=rooms_key,
-                    lo=br_lo,
-                    hi=walk_hi,
-                    seen=seen,
-                    price_cap_per_bucket=price_cap_per_bucket,
-                    max_pages_per_bucket=max_pages_per_bucket,
-                    concurrency=concurrency,
-                    on_bucket=on_bucket,
-                    skip_buckets=skip_buckets,
-                )
-            room_collected = len(seen) - before
-            logger.info(
-                "yandex gate exhaustive: rooms=%s done -- collected %d (total unique=%d)",
-                rooms_key,
-                room_collected,
-                len(seen),
-            )
-            if on_progress is not None:
-                on_progress(len(seen))
-
-        logger.info("yandex gate exhaustive: DONE -- total unique=%d lots", len(seen))
-        return list(seen.values())
-
-    async def _walk_price_range(
-        self,
-        *,
-        rooms: str | None,
-        lo: int,
-        hi: int | None,
-        seen: dict[str, ScrapedLot],
-        price_cap_per_bucket: int,
-        max_pages_per_bucket: int,
-        concurrency: int = 4,
-        on_bucket: Any = None,
-        skip_buckets: set[str] | None = None,
-        _depth: int = 0,
-    ) -> None:
-        """Recursive binary price-range partitioning.
-
-        Probe page=1 -> read totalItems from pager.
-        Closed bracket (hi is not None): totalItems > cap and bracket >= MIN_BRACKET
-        -> split; else paginate leaf.
-        Open bracket (hi is None): top seed-bracket without priceMax ceiling. Cannot
-        split (no mid) -> paginate leaf directly (lux tail tiny, tail-loss accepted).
-        Fallback: totalItems=None -> paginate-until-empty (safe degradation).
-        skip_buckets: resume checkpoint -- skip leaf buckets already done.
-        """
-        _lo_param = lo if lo > 0 else None
-        _hi_repr = "open" if hi is None else str(hi)
-
-        probe_payload = await self._fetch_page_json(rooms, 1, _lo_param, hi)
-        await asyncio.sleep(self.request_delay_sec)
-
-        total: int | None = None
-        probe_lots: list[ScrapedLot] = []
-
-        if probe_payload is None:
-            logger.warning(
-                "yandex gate: probe failed rooms=%s [%d, %s] depth=%d -- rotating + retry",
-                rooms,
-                lo,
-                _hi_repr,
-                _depth,
-            )
-            rotated = await self._rotate_ip()
-            if rotated:
-                probe_payload = await self._fetch_page_json(rooms, 1, _lo_param, hi)
-                await asyncio.sleep(self.request_delay_sec)
-
-        if probe_payload is not None:
-            result = _extract_gate_data(probe_payload)
-            if result is not None:
-                _, pager = result
-                total = pager.get("totalItems")
-                probe_lots = _parse_gate_json(probe_payload, page_param=1)
-            else:
-                logger.warning(
-                    "yandex gate: probe data extract failed rooms=%s [%d, %s]", rooms, lo, _hi_repr
-                )
-
-        if total is None:
-            # Degraded path: paginate-until-empty without knowing total
-            logger.warning(
-                "yandex gate: totalItems unknown rooms=%s [%d, %s] -- paginate until empty",
-                rooms,
-                lo,
-                _hi_repr,
-            )
-            page = 1
-            bucket_key = _combo_label(rooms, lo, hi)
-            if skip_buckets and bucket_key in skip_buckets:
-                return
-            pages_fetched = 0
-            while pages_fetched < max_pages_per_bucket:
-                payload = await self._fetch_page_json(rooms, page, _lo_param, hi)
-                await asyncio.sleep(self.request_delay_sec)
-                if payload is None:
-                    break
-                lots = _parse_gate_json(payload, page_param=page)
-                if not lots:
-                    break
-                for lot in lots:
-                    if lot.source_id and lot.source_id not in seen:
-                        seen[lot.source_id] = lot
-                page += 1
-                pages_fetched += 1
-            if on_bucket is not None:
-                on_bucket(bucket_key, len(seen))
-            return
-
-        # Split or paginate. ОТКРЫТЫЙ брекет (hi is None) делить нельзя (нет mid) —
-        # сразу пагинируем leaf напрямую (хвост люкса крошечный, tail-loss accepted).
-        cap = price_cap_per_bucket
-        _min_bracket = 500_000
-
-        if hi is not None and total > cap and (hi - lo) >= _min_bracket and _depth < 8:
-            mid = (lo + hi) // 2
-            await self._walk_price_range(
-                rooms=rooms,
-                lo=lo,
-                hi=mid,
-                seen=seen,
-                price_cap_per_bucket=cap,
-                max_pages_per_bucket=max_pages_per_bucket,
-                concurrency=concurrency,
-                on_bucket=on_bucket,
-                skip_buckets=skip_buckets,
-                _depth=_depth + 1,
-            )
-            await self._walk_price_range(
-                rooms=rooms,
-                lo=mid + 1,
-                hi=hi,
-                seen=seen,
-                price_cap_per_bucket=cap,
-                max_pages_per_bucket=max_pages_per_bucket,
-                concurrency=concurrency,
-                on_bucket=on_bucket,
-                skip_buckets=skip_buckets,
-                _depth=_depth + 1,
-            )
-            return
-
-        # Leaf bucket: paginate
-        bucket_key = _combo_label(rooms, lo, hi)
-        if skip_buckets and bucket_key in skip_buckets:
-            logger.debug("yandex gate: skip bucket %s (checkpoint)", bucket_key)
-            return
-
-        total_pages = min(
-            math.ceil(total / 20),
-            _GATE_MAX_PAGES_CAP,
-            max_pages_per_bucket,
-        )
-        logger.info("yandex gate: leaf bucket %s total=%d pages=%d", bucket_key, total, total_pages)
-
-        # Add probe lots (page 1 already fetched)
-        for lot in probe_lots:
-            if lot.source_id and lot.source_id not in seen:
-                seen[lot.source_id] = lot
-
-        if total_pages <= 1:
-            if on_bucket is not None:
-                on_bucket(bucket_key, len(seen))
-            return
-
-        # Paginate pages 2..total_pages with concurrency
-        sem = asyncio.Semaphore(concurrency)
-
-        async def _fetch_leaf_page(pg: int) -> list[ScrapedLot]:
-            async with sem:
-                payload = await self._fetch_page_json(rooms, pg, _lo_param, hi)
-                await asyncio.sleep(self.request_delay_sec)
-                if payload is None:
-                    return []
-                return _parse_gate_json(payload, page_param=pg)
-
-        tasks = [_fetch_leaf_page(pg) for pg in range(2, total_pages + 1)]
-        results = await asyncio.gather(*tasks)
-        for page_lots in results:
-            for lot in page_lots:
-                if lot.source_id and lot.source_id not in seen:
-                    seen[lot.source_id] = lot
-
-        if on_bucket is not None:
-            on_bucket(bucket_key, len(seen))
-
-
-def _combo_label(rooms: str | None, lo: int | None, hi: int | None) -> str:
-    """Уникальная метка бакета для checkpoint-логики."""
-    return f"{rooms or 'any'}:{lo}-{hi}"
diff --git a/tradein-mvp/backend/tests/scrapers/test_admin_avito_kit_parity.py b/tradein-mvp/backend/tests/scrapers/test_admin_avito_kit_parity.py
deleted file mode 100644
index 540d4766..00000000
--- a/tradein-mvp/backend/tests/scrapers/test_admin_avito_kit_parity.py
+++ /dev/null
@@ -1,112 +0,0 @@
-"""Parity-proof: admin.py Avito debug-роуты (#2305, Group A) — legacy vs scraper_kit.
-
-`app/api/v1/admin.py::scrape_avito_house` теперь зовёт `scraper_kit.providers.avito.houses`
-вместо `app.services.scrapers.avito_houses` (тот же `/scrape` root-роут —
-`scraper_kit.providers.avito.serp.AvitoScraper`). Каждая пара функций ниже — чистое,
-offline-тестируемое parsing-ядро, которое реально используется соответствующим
-debug-эндпоинтом (не сам HTTP-fetch — см. `tests/support/README.md`: "Live network/DB
-в parity-тестах избегайте").
-
-- `_parse_house_page` — `scrape_avito_house` → `fetch_house_catalog` парсит виджет
-  `housePage` этой функцией.
-- `_extract_rooms_from_title` / `_extract_area_from_title` — использует
-  `AvitoScraper` (SERP-парсинг `/scrape` root-роута) для title-fallback полей.
-
-Легаси `app.services.scrapers.avito_imv` (`scrape_avito_imv` → `evaluate_via_imv` →
-`_parse_price`) удалён (#2277 финальный шаг scraper_kit-миграции, 0 runtime
-importers) — parity-проверка `_parse_price` убрана вместе с ним; kit-сторона
-`_parse_price` покрыта отдельно (`tests/scrapers/test_avito_imv_kit_parity.py` /
-`tests/test_avito_imv_parse.py`).
-"""
-
-from __future__ import annotations
-
-import dataclasses
-import os
-
-# Соседние модули app.services.scrapers.* импортируют app.core.config при импорте
-# пакета — фиктивный DSN мостит офлайн-запуск (mirror test_avito_detail_kit_parity.py).
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
-
-from scraper_kit.providers.avito.houses import _parse_house_page as kit_parse_house_page
-from scraper_kit.providers.avito.serp import (
-    _extract_area_from_title as kit_extract_area_from_title,
-)
-from scraper_kit.providers.avito.serp import (
-    _extract_rooms_from_title as kit_extract_rooms_from_title,
-)
-
-from app.services.scrapers.avito import _extract_area_from_title as legacy_extract_area_from_title
-from app.services.scrapers.avito import (
-    _extract_rooms_from_title as legacy_extract_rooms_from_title,
-)
-from app.services.scrapers.avito_houses import _parse_house_page as legacy_parse_house_page
-from tests.support.parity import assert_parity
-
-_HOUSE_WIDGET = {
-    "props": {
-        "developmentData": {
-            "avitoId": 3171365,
-            "id": "abc123hash",
-            "title": "ЖК Тестовый",
-            "address": "ул. Тестовая, 1",
-            "fullAddress": "Екатеринбург, ул. Тестовая, 1",
-            "coords": {"lat": 56.8, "lng": 60.6},
-            "aboutDevelopment": {
-                "expandParams": {
-                    "items": [
-                        {
-                            "params": [
-                                {"type": "Год постройки", "value": "2020"},
-                                {"type": "Этажей", "value": "16"},
-                                {"type": "Пассажирский лифт", "value": "2"},
-                            ]
-                        }
-                    ]
-                }
-            },
-            "ratingBadge": {"info": {"score": 4.5, "scoreString": "4.5"}},
-            "ratingSummaryStat": {"reviewCount": 10, "ratingStat": []},
-            "mapPreview": {"objects": "школа рядом", "distance": "5 мин", "pins": []},
-            "developer": {"name": "ООО Застройщик", "key": "dev1"},
-        }
-    }
-}
-
-
-def test_avito_house_page_parse_are_class_distinct() -> None:
-    """Sanity: legacy/kit HouseInfo — разные классы, обычный `==` даёт False."""
-    legacy_result = legacy_parse_house_page(_HOUSE_WIDGET)
-    kit_result = kit_parse_house_page(_HOUSE_WIDGET)
-
-    assert type(legacy_result) is not type(kit_result)
-    assert legacy_result != kit_result
-    assert dataclasses.asdict(legacy_result) == dataclasses.asdict(kit_result)
-
-
-def test_parity_avito_house_page() -> None:
-    """`scrape_avito_house` (admin.py) → fetch_house_catalog → _parse_house_page."""
-    assert_parity(
-        legacy_fn=legacy_parse_house_page,
-        kit_fn=kit_parse_house_page,
-        fixtures=[(_HOUSE_WIDGET,)],
-    )
-
-
-def test_parity_avito_serp_title_extract() -> None:
-    """`/scrape` root-роут (admin.py) → AvitoScraper SERP title-fallback fields."""
-    titles = [
-        "2-к. квартира, 54.3 м², 3/9 эт.",
-        "Студия, 25 м², 1/5 эт.",
-        "3-к. кв., 75,5 м², 12/16 эт.",
-    ]
-    assert_parity(
-        legacy_fn=legacy_extract_rooms_from_title,
-        kit_fn=kit_extract_rooms_from_title,
-        fixtures=titles,
-    )
-    assert_parity(
-        legacy_fn=legacy_extract_area_from_title,
-        kit_fn=kit_extract_area_from_title,
-        fixtures=titles,
-    )
diff --git a/tradein-mvp/backend/tests/scrapers/test_admin_cian_kit_parity.py b/tradein-mvp/backend/tests/scrapers/test_admin_cian_kit_parity.py
deleted file mode 100644
index 39fdc616..00000000
--- a/tradein-mvp/backend/tests/scrapers/test_admin_cian_kit_parity.py
+++ /dev/null
@@ -1,51 +0,0 @@
-"""Parity-proof: admin.py `scrape_cian_detail` (#2305, Group A) — legacy vs scraper_kit.
-
-`app/api/v1/admin.py::scrape_cian_detail` теперь зовёт
-`scraper_kit.providers.cian.detail.fetch_detail/save_detail_enrichment` вместо
-`app.services.scrapers.cian_detail`. `_extract_price_changes` — чистая,
-offline-тестируемая normalize-функция, которую `fetch_detail` вызывает при
-сборке `DetailEnrichment.price_changes` (см. `tests/support/README.md`:
-"Live network/DB в parity-тестах избегайте").
-"""
-
-from __future__ import annotations
-
-import os
-
-# cian_detail.py импортирует app.core.config при module-level import — фиктивный
-# DSN мостит офлайн-запуск (mirror test_avito_detail_kit_parity.py).
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
-
-from scraper_kit.providers.cian.detail import _extract_price_changes as kit_extract_price_changes
-
-from app.services.scrapers.cian_detail import (
-    _extract_price_changes as legacy_extract_price_changes,
-)
-from tests.support.parity import assert_parity
-
-# Три реальных формы priceChanges: curl_cffi flat, browser-rendered nested, оба-пусто.
-_OFFER_FLAT = {
-    "priceChanges": [
-        {"changeTime": "2026-01-15T00:00:00", "price": 9000000, "diffPercent": -2.5},
-        {"changeTime": "2026-02-01T00:00:00", "priceRub": 8800000, "diff_percent": -2.2},
-    ]
-}
-_OFFER_DATA_NESTED = {
-    "priceChanges": [
-        {"changeTime": "2026-01-15T00:00:00", "priceData": {"price": 9000000}},
-    ]
-}
-_STATE_EMPTY: dict[str, object] = {}
-
-
-def test_parity_cian_detail_price_changes() -> None:
-    """`scrape_cian_detail` (admin.py) → fetch_detail → _extract_price_changes."""
-    assert_parity(
-        legacy_fn=legacy_extract_price_changes,
-        kit_fn=kit_extract_price_changes,
-        fixtures=[
-            (_OFFER_FLAT, {}, _STATE_EMPTY),
-            ({}, _OFFER_DATA_NESTED, _STATE_EMPTY),
-            ({}, {}, _STATE_EMPTY),
-        ],
-    )
diff --git a/tradein-mvp/backend/tests/scrapers/test_admin_yandex_kit_parity.py b/tradein-mvp/backend/tests/scrapers/test_admin_yandex_kit_parity.py
deleted file mode 100644
index 67cd2bce..00000000
--- a/tradein-mvp/backend/tests/scrapers/test_admin_yandex_kit_parity.py
+++ /dev/null
@@ -1,99 +0,0 @@
-"""Parity-proof: admin.py Yandex debug-роуты (#2305, Group A) — legacy vs scraper_kit.
-
-`app/api/v1/admin.py::scrape_yandex_detail` / `scrape_yandex_valuation` теперь
-зовут `scraper_kit.providers.yandex.{detail,valuation}` вместо
-`app.services.scrapers.yandex_{detail,valuation}` (тот же `/scrape` root-роут —
-`scraper_kit.providers.yandex.serp.YandexRealtyScraper`). Каждая функция ниже —
-чистое, offline-тестируемое parsing-ядро (см. `tests/support/README.md`:
-"Live network/DB в parity-тестах избегайте").
-
-- `_entity_to_lot` — `/scrape` root-роут → `YandexRealtyScraper.fetch_around`
-  конвертирует gate-API entity → ScrapedLot этой функцией (также доказывает
-  ScrapedLot-контракт, общий с `save_listings`/`scripts/ingest_domclick_jsonl.py`).
-
-Легаси `app.services.scrapers.yandex_detail` / `yandex_valuation` удалены (#2277
-финальный шаг scraper_kit-миграции, 0 runtime importers) — `_parse_card_fields` /
-`_parse_house_meta` parity-проверки убраны отсюда вместе с ними; kit-side coverage
-для обеих функций сохранена standalone в `tests/test_yandex_detail_structural.py` /
-`tests/test_yandex_valuation.py`. `yandex_realty` остаётся легаси (не мигрирован) —
-его `_entity_to_lot` parity-проверка не тронута.
-"""
-
-from __future__ import annotations
-
-import os
-
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
-
-from scraper_kit.providers.yandex.serp import _entity_to_lot as kit_entity_to_lot
-
-from app.services.scrapers.yandex_realty import _entity_to_lot as legacy_entity_to_lot
-from tests.support.parity import assert_parity
-
-# gate-API entity (realty.yandex.ru), без creationDate/author — исключает
-# time-dependent поля (days_on_market) из фикстуры (deterministic offline).
-_YANDEX_ENTITY = {
-    "offerId": "12345",
-    "url": "//realty.yandex.ru/offer/12345",
-    "price": {"value": 8500000, "valuePerPart": 157000},
-    "area": {"value": 54.3},
-    "livingSpace": {"value": 32.1},
-    "kitchenSpace": {"value": 9.5},
-    "roomsTotal": 2,
-    "floorsOffered": [5],
-    "floorsTotal": 9,
-    "ceilingHeight": 2.7,
-    "building": {"builtYear": 2015, "buildingType": "MONOLIT"},
-    "location": {
-        "point": {"latitude": 56.83, "longitude": 60.6},
-        "geocoderAddress": "Екатеринбург, ул. Тестовая, 1",
-    },
-    "mainImages": ["//avatars.mds.yandex.net/img1.jpg"],
-}
-
-
-def test_yandex_entity_to_lot_are_class_distinct() -> None:
-    """Sanity: legacy/kit ScrapedLot — разные классы (pydantic), обычный `==` даёт False."""
-    legacy_result = legacy_entity_to_lot(_YANDEX_ENTITY)
-    kit_result = kit_entity_to_lot(_YANDEX_ENTITY)
-
-    assert legacy_result is not None
-    assert kit_result is not None
-    assert type(legacy_result) is not type(kit_result)
-    assert legacy_result.model_dump() == kit_result.model_dump()
-
-
-def test_parity_yandex_serp_entity_to_lot() -> None:
-    """`/scrape` root-роут (admin.py) → YandexRealtyScraper.fetch_around → _entity_to_lot.
-
-    Также доказывает ScrapedLot-контракт, общий с `save_listings`
-    (используется и `/scrape`, и `scripts/ingest_domclick_jsonl.py`).
-    """
-    assert_parity(
-        legacy_fn=legacy_entity_to_lot,
-        kit_fn=kit_entity_to_lot,
-        fixtures=[(_YANDEX_ENTITY,)],
-    )
-
-
-def test_parity_harness_catches_yandex_entity_divergence() -> None:
-    """Harness реально ЛОВИТ расхождение — мутируем price_rub в kit-результате."""
-    import pytest
-
-    from tests.support.parity import ParityMismatchError
-
-    legacy_result = legacy_entity_to_lot(_YANDEX_ENTITY)
-    kit_result = kit_entity_to_lot(_YANDEX_ENTITY)
-    assert legacy_result is not None
-    assert kit_result is not None
-
-    # ScrapedLot — pydantic BaseModel (НЕ dataclass) → model_copy, не dataclasses.replace.
-    mutated_kit_result = kit_result.model_copy(update={"price_rub": kit_result.price_rub + 1})
-
-    with pytest.raises(ParityMismatchError) as exc_info:
-        assert_parity(
-            legacy_fn=lambda: legacy_result,
-            kit_fn=lambda: mutated_kit_result,
-            fixtures=[()],
-        )
-    assert "price_rub" in str(exc_info.value)
diff --git a/tradein-mvp/backend/tests/scrapers/test_avito_unix_date_tz_consistency.py b/tradein-mvp/backend/tests/scrapers/test_avito_unix_date_tz_consistency.py
index 683339be..4ee0ab61 100644
--- a/tradein-mvp/backend/tests/scrapers/test_avito_unix_date_tz_consistency.py
+++ b/tradein-mvp/backend/tests/scrapers/test_avito_unix_date_tz_consistency.py
@@ -3,7 +3,7 @@
 
 До фикса было 3 независимые копии с разными timezone:
   - avito.py (inline в `_build_sort_timestamp_map`) — MSK (документировано #726)
-  - avito_houses.py — UTC
+  - avito_houses.py (легаси, удалён #2397 Part E2) — UTC
   - avito_imv.py (легаси, удалён #2277) — naive/system-local
     (`datetime.fromtimestamp(ts)` без tz)
 
@@ -17,6 +17,11 @@ avito_imv_getdata.json: removedDate=1756077885 → UTC 2025-08-24, MSK 2025-08-2
 kit-эквивалент `scraper_kit.providers.avito.imv`, который использует ТОТ ЖЕ паттерн
 (`from scraper_kit.providers.avito.shared import _unix_to_date`) — регрессия
 "3 независимые копии с разными TZ" остаётся под наблюдением и на kit-стороне.
+
+Легаси `app.services.scrapers.avito_houses` удалён (#2397 Part E2, 0 runtime
+importers) — houses-ветка consolidation-проверки переведена на kit-эквивалент
+`scraper_kit.providers.avito.houses`, который использует ТОТ ЖЕ паттерн
+(`from scraper_kit.providers.avito.shared import RUS_MONTHS, _unix_to_date`).
 """
 
 from __future__ import annotations
@@ -26,10 +31,11 @@ from datetime import date
 
 os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
 
+from scraper_kit.providers.avito import houses as kit_avito_houses
 from scraper_kit.providers.avito import imv as kit_avito_imv
 from scraper_kit.providers.avito import shared as kit_avito_shared
 
-from app.services.scrapers import avito, avito_houses, avito_shared
+from app.services.scrapers import avito, avito_shared
 
 # Живой пример из fixtures/avito_imv_getdata.json (placementHistory.items[0].removedDate) —
 # именно тот timestamp, на котором UTC и MSK расходились в date() ДО фикса
@@ -44,14 +50,15 @@ def test_shared_unix_to_date_uses_msk() -> None:
 
 
 def test_avito_houses_reexports_shared_unix_to_date() -> None:
-    """avito_houses.py больше не держит свою копию — импортирует общую функцию."""
-    assert avito_houses._unix_to_date is avito_shared._unix_to_date
-    assert avito_houses._unix_to_date(_EPOCH_SECONDS) == _EXPECTED_MSK_DATE
+    """kit houses.py не держит свою копию — импортирует общую функцию (тот же паттерн,
+    что и легаси avito_houses.py до удаления, #2397 Part E2)."""
+    assert kit_avito_houses._unix_to_date is kit_avito_shared._unix_to_date
+    assert kit_avito_houses._unix_to_date(_EPOCH_SECONDS) == _EXPECTED_MSK_DATE
 
 
 def test_kit_avito_imv_reexports_shared_unix_to_date() -> None:
     """kit imv.py не держит свою копию — импортирует общую функцию (тот же паттерн,
-    что и легаси avito_houses.py/avito_shared.py после фикса #2129)."""
+    что и kit houses.py/avito_shared.py после фикса #2129)."""
     assert kit_avito_imv._unix_to_date is kit_avito_shared._unix_to_date
     assert kit_avito_imv._unix_to_date(_EPOCH_SECONDS) == _EXPECTED_MSK_DATE
 
@@ -66,11 +73,11 @@ def test_avito_serp_sort_timestamp_matches_shared_date() -> None:
     )
     ts_map = avito._build_sort_timestamp_map(html)
     assert ts_map["8043936560"] == _EXPECTED_MSK_DATE
-    # Легаси avito/avito_houses/avito_shared + kit avito_imv все согласованы на одном
+    # Легаси avito/avito_shared + kit avito_houses/avito_imv все согласованы на одном
     # моменте времени:
     assert (
         ts_map["8043936560"]
-        == avito_houses._unix_to_date(_EPOCH_SECONDS)
+        == kit_avito_houses._unix_to_date(_EPOCH_SECONDS)
         == kit_avito_imv._unix_to_date(_EPOCH_SECONDS)
         == avito_shared._unix_to_date(_EPOCH_SECONDS)
     )
@@ -81,7 +88,7 @@ def test_rus_months_single_source() -> None:
     from app.services.scrapers import avito_detail
 
     assert avito_detail.RUS_MONTHS is avito_shared.RUS_MONTHS
-    assert avito_houses.RUS_MONTHS is avito_shared.RUS_MONTHS
+    assert kit_avito_houses.RUS_MONTHS is kit_avito_shared.RUS_MONTHS
     assert avito_shared.RUS_MONTHS["декабря"] == 12
 
 
diff --git a/tradein-mvp/backend/tests/scrapers/test_cian_exhaustive.py b/tradein-mvp/backend/tests/scrapers/test_cian_exhaustive.py
deleted file mode 100644
index 888918ce..00000000
--- a/tradein-mvp/backend/tests/scrapers/test_cian_exhaustive.py
+++ /dev/null
@@ -1,792 +0,0 @@
-"""Тесты exhaustive-загрузки Cian: fetch_all_secondary + _walk_price_range.
-
-Мокаем сетевые вызовы (_fetch_page_html, _extract_total_offers, sleep_between_requests)
-для детерминированного контроля поведения без реального HTTP.
-"""
-
-from __future__ import annotations
-
-import os
-from unittest.mock import AsyncMock, MagicMock, patch
-
-# Settings requires DATABASE_URL at import time — set dummy DSN before any app import.
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")
-
-import logging
-
-import pytest
-
-from app.services.scrapers.base import ScrapedLot
-from app.services.scrapers.cian import (
-    _CIAN_OFFERS_PER_PAGE,
-    _MAX_PRICE,
-    _MIN_BRACKET,
-    CianScraper,
-)
-from app.services.scrapers.price_brackets import get_price_seed_brackets
-
-_SEED = get_price_seed_brackets()
-
-
-def _make_lot(source_id: str, price: int = 3_000_000) -> ScrapedLot:
-    """Минимальный ScrapedLot для теста."""
-    return ScrapedLot(
-        source="cian",
-        source_url=f"https://ekb.cian.ru/sale/flat/{source_id}/",
-        source_id=source_id,
-        address="Екатеринбург",
-        price_rub=price,
-    )
-
-
-def _make_html_stub(total_offers: int, lot_ids: list[str]) -> str:
-    """Заглушка HTML — не парсится напрямую; используется через мокнутые методы."""
-    return f"total={total_offers} ids={lot_ids}"
-
-
-@pytest.fixture
-def scraper() -> CianScraper:
-    """CianScraper без реального __aenter__ (без HTTP warm-up)."""
-    s = CianScraper.__new__(CianScraper)
-    s.name = "cian"
-    s.base_url = "https://ekb.cian.ru"
-    s.request_delay_sec = 0.0
-    s._browser = MagicMock()  # не None → assert self._browser is not None пройдёт
-    return s
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_splits_on_cap(scraper: CianScraper) -> None:
-    """totalOffers > cap → бакет разбивается; в итоге оба суб-бакета пагинируются."""
-    # Настройка: rooms=(1,), один бакет [0, None]
-    # Probe [0, None]: totalOffers=3000 > cap=1400 → split на [0, MAX_PRICE]
-    # Probe [0, MAX_PRICE]: totalOffers=3000 > 1400 → split на [0, mid] и [mid+1, MAX_PRICE]
-    # Probe [0, mid] (mid = MAX_PRICE//2 = 100_000_000): totalOffers=800 <= cap → пагинировать
-    # Probe [mid+1, MAX_PRICE]: totalOffers=700 <= cap → пагинировать
-
-    mid = _MAX_PRICE // 2
-
-    call_log: list[tuple] = []
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        call_log.append((rooms, page, min_price, max_price))
-        return f"page={page} min={min_price} max={max_price}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        # Парсим min/max из fake html
-        import re
-
-        min_m = re.search(r"min=(\w+)", html)
-        max_m = re.search(r"max=(\w+)", html)
-        min_p = int(min_m.group(1)) if min_m and min_m.group(1) != "None" else 0
-        max_p = int(max_m.group(1)) if max_m and max_m.group(1) != "None" else _MAX_PRICE
-
-        # Имитируем разделение: весь диапазон > 1400, левая и правая половины <= 1400
-        if max_p is None or max_p >= _MAX_PRICE:
-            if min_p == 0:
-                return 3000  # весь диапазон — нужно split
-        if max_p == mid:
-            return 800  # левая половина — пагинировать
-        if min_p == mid + 1:
-            return 700  # правая половина — пагинировать
-        return 3000  # всё остальное — split
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        min_m = re.search(r"min=(\w+)", html)
-        max_m = re.search(r"max=(\w+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        min_p = min_m.group(1) if min_m else "0"
-        max_p = max_m.group(1) if max_m else "MAX"
-        # Возвращаем по 2 лота на первой странице каждого пагинируемого бакета
-        if page == 1:
-            return [
-                _make_lot(f"lot_{min_p}_{max_p}_p{page}_1"),
-                _make_lot(f"lot_{min_p}_{max_p}_p{page}_2"),
-            ]
-        return []  # страница 2+ — пусто → early stop
-
-    async def fake_sleep(*_: object, **__: object) -> None:
-        pass
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        lots = await scraper.fetch_all_secondary(
-            rooms_buckets=[(1,)],
-            price_cap_per_bucket=1400,
-        )
-
-    # Должны получить лоты из ОБОИХ суб-бакетов после split
-    assert len(lots) > 0, "Должны получить лоты после разделения бакета"
-
-    # Проверяем что был probe с большим диапазоном И с sub-бакетами
-    probed_ranges = [(min_p, max_p) for (_, page, min_p, max_p) in call_log if page == 1]
-    # Должен быть probe [0, None] или [0, MAX] — начальный probe всего диапазона
-    initial_probes = [r for r in probed_ranges if r[0] is None or r[0] == 0]
-    assert (
-        len(initial_probes) >= 1
-    ), f"Ожидался probe всего диапазона, probed_ranges={probed_ranges}"
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_paginates_when_under_cap(scraper: CianScraper) -> None:
-    """totalOffers <= cap → пагинирует бакет полностью (страница за страницей)."""
-    pages_fetched: list[int] = []
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        pages_fetched.append(page)
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 56  # 56 офферов → ceil(56/28) = 2 страницы
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        if page <= 2:
-            return [_make_lot(f"lot_p{page}_{i}") for i in range(28)]
-        return []
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        lots = await scraper.fetch_all_secondary(
-            rooms_buckets=[(2,)],
-            price_cap_per_bucket=1400,
-        )
-
-    # 28 лотов на стр 1 + 28 на стр 2 = 56, дедуп по source_id
-    assert len(lots) == 56, f"Ожидалось 56 уникальных лотов, получено {len(lots)}"
-    # Должно быть 2 страницы пагинации (page=1 probe + page=2)
-    assert 2 in pages_fetched, "Должна быть запрошена страница 2"
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_min_bracket_guard(scraper: CianScraper) -> None:
-    """Если hi - lo < MIN_BRACKET → пагинируем как есть (не делим бесконечно)."""
-    call_count = 0
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        nonlocal call_count
-        call_count += 1
-        if call_count > 100:
-            raise RuntimeError("Бесконечная рекурсия — MIN_BRACKET guard не сработал")
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        # Всегда > cap → без guard делилось бы бесконечно
-        return 5000
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        return []  # пустой → early stop в пагинации
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        # Запускаем с очень узким диапазоном — guard должен остановить деление
-        await scraper._walk_price_range(
-            rooms=(1,),
-            lo=1_000_000,
-            hi=1_000_000 + _MIN_BRACKET - 1,  # bracket < MIN_BRACKET
-            seen={},
-            price_cap_per_bucket=1400,
-            max_pages_per_bucket=54,
-        )
-
-    # Guard сработал: нет RuntimeError (бесконечная рекурсия не возникла).
-    # Параллельная пагинация запускает до max_pages_per_bucket=54 страниц единовременно
-    # (все страницы бакета параллельны), поэтому call_count может быть до 54.
-    # Важно что нет рекурсии (bracket < MIN_BRACKET → один проход пагинации, не деление).
-    assert call_count <= 55, f"Слишком много вызовов ({call_count}) — guard не сработал"
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_on_bucket_called_per_leaf(scraper: CianScraper) -> None:
-    """on_bucket вызывается после каждого leaf-бакета с лотами бакета."""
-    # totalOffers=56 ≤ cap → leaf-бакет, пагинируется параллельно 2 страницы
-    bucket_calls: list[tuple[str, list[ScrapedLot]]] = []
-
-    def fake_on_bucket(bucket_key: str, lots: list[ScrapedLot]) -> None:
-        bucket_calls.append((bucket_key, list(lots)))
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 56  # ceil(56/28) = 2 страницы
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return [_make_lot(f"lot_p{page}_{i}") for i in range(10)]
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        await scraper.fetch_all_secondary(
-            rooms_buckets=[(1,)],
-            price_cap_per_bucket=1400,
-            concurrency=3,
-            on_bucket=fake_on_bucket,
-        )
-
-    # on_bucket вызван по разу на каждый seed-брекет (все непустые leaf-бакеты).
-    assert len(bucket_calls) == len(
-        _SEED
-    ), f"Ожидалось {len(_SEED)} вызовов on_bucket, получено {len(bucket_calls)}"
-    key, lots = bucket_calls[0]
-    # Ключ бакета содержит room_label:lo:hi
-    assert "room1" in key, f"bucket_key должен содержать room1, got {key!r}"
-    # Лоты из обеих страниц переданы в on_bucket
-    assert len(lots) == 20, f"Ожидалось 20 лотов в on_bucket (2 стр × 10), получено {len(lots)}"
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_on_bucket_cancel_stops_run(scraper: CianScraper) -> None:
-    """Если on_bucket кидает RuntimeError('cancelled') — прогон прерывается."""
-
-    def cancel_on_bucket(bucket_key: str, lots: list[ScrapedLot]) -> None:
-        raise RuntimeError("cancelled")
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 28  # один лист → 1 страница
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        return [_make_lot("lot_1")]
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        with pytest.raises(RuntimeError, match="cancelled"):
-            await scraper.fetch_all_secondary(
-                rooms_buckets=[(1,), (2,)],
-                price_cap_per_bucket=1400,
-                on_bucket=cancel_on_bucket,
-            )
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_concurrent_pages_deduped(scraper: CianScraper) -> None:
-    """Параллельная пагинация не ломает дедупликацию по source_id."""
-    fetch_pages: list[int] = []
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        fetch_pages.append(page)
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 84  # ceil(84/28) = 3 страницы
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        # Страница 1 и 2 дают уникальные лоты; страница 3 дублирует страницу 1
-        if page == 3:
-            return [_make_lot(f"lot_p1_{i}") for i in range(10)]  # дубликаты
-        return [_make_lot(f"lot_p{page}_{i}") for i in range(10)]
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        lots = await scraper.fetch_all_secondary(
-            rooms_buckets=[(1,)],
-            price_cap_per_bucket=1400,
-            concurrency=5,
-        )
-
-    # Страницы 1+2 = 20 уникальных, страница 3 дублирует → итого 20
-    assert len(lots) == 20, f"Ожидалось 20 уникальных лотов (дедуп), получено {len(lots)}"
-    # Все 3 страницы запрошены (параллельно или нет — нам важен факт)
-    assert 3 in fetch_pages or 3 in [
-        p for p in fetch_pages
-    ], f"Страница 3 не была запрошена, fetch_pages={fetch_pages}"
-
-
-@pytest.mark.asyncio
-async def test_skip_buckets_skips_pagination_and_on_bucket(scraper: CianScraper) -> None:
-    """skip_buckets: для скипнутого бакета пагинация не вызывается, on_bucket не вызывается."""
-    fetch_calls: list[tuple] = []
-    on_bucket_calls: list[str] = []
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        fetch_calls.append((rooms, page, min_price, max_price))
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 28  # один leaf-бакет, 1 страница
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        return [_make_lot("lot_1")]
-
-    def fake_on_bucket(bucket_key: str, lots: list[ScrapedLot]) -> None:
-        on_bucket_calls.append(bucket_key)
-
-    # Скипаем ВСЕ seed-бакеты комнатности room1. Закрытые → ':', открытый → ':open'.
-    skip_set: set[str] = set()
-    for br_lo, br_hi in _SEED:
-        hi_repr = "open" if br_hi is None else str(br_hi - 1)
-        skip_set.add(f"room1:{br_lo}:{hi_repr}")
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        lots = await scraper.fetch_all_secondary(
-            rooms_buckets=[(1,)],
-            price_cap_per_bucket=1400,
-            on_bucket=fake_on_bucket,
-            skip_buckets=skip_set,
-        )
-
-    # Probe (page=1) ВСЁ РАВНО вызывается на каждый seed-брекет (нужен для split-решения),
-    # но НЕ должно быть страниц пагинации (все бакеты в skip_buckets).
-    # Probe = page=1; пагинация пропущена ДО fetch остальных страниц → ровно len(_SEED) probe.
-    assert len(fetch_calls) == len(
-        _SEED
-    ), f"Ожидалось {len(_SEED)} probe-вызовов, получено {len(fetch_calls)}: {fetch_calls}"
-    assert all(c[1] == 1 for c in fetch_calls), "Все вызовы должны быть probe page=1"
-    # on_bucket не вызывается для skip-бакетов
-    assert (
-        len(on_bucket_calls) == 0
-    ), f"on_bucket не должен вызываться для skip-бакетов, got: {on_bucket_calls}"
-    # lots пустой — ничего не собрали (пагинация пропущена)
-    assert len(lots) == 0, f"Ожидалось 0 лотов для skip-бакетов, got {len(lots)}"
-
-
-@pytest.mark.asyncio
-async def test_on_bucket_receives_key_and_lots(scraper: CianScraper) -> None:
-    """on_bucket получает (bucket_key, lots) — bucket_key в формате room_label:lo:hi."""
-    received: list[tuple[str, int]] = []  # (key, len(lots))
-
-    def fake_on_bucket(bucket_key: str, lots: list[ScrapedLot]) -> None:
-        received.append((bucket_key, len(lots)))
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 28  # 1 страница
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return [_make_lot(f"lot_p{page}_{i}") for i in range(5)]
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        await scraper.fetch_all_secondary(
-            rooms_buckets=[(2,)],
-            price_cap_per_bucket=1400,
-            on_bucket=fake_on_bucket,
-        )
-
-    assert len(received) == len(
-        _SEED
-    ), f"Ожидалось {len(_SEED)} вызовов on_bucket, получено {len(received)}"
-    keys = [k for (k, _c) in received]
-    # Закрытые брекеты: hi = br_hi - 1. Открытый (последний): ':open'.
-    first_lo, first_hi = _SEED[0]
-    assert first_hi is not None
-    assert keys[0] == f"room2:{first_lo}:{first_hi - 1}", f"Неверный ключ: {keys[0]!r}"
-    last_lo, last_hi = _SEED[-1]
-    assert last_hi is None
-    assert keys[-1] == f"room2:{last_lo}:open", f"Открытый ключ неверен: {keys[-1]!r}"
-    assert all(count == 5 for (_k, count) in received)
-
-
-# ── secondary_only фильтр ─────────────────────────────────────────────────────
-
-
-@pytest.mark.asyncio
-async def test_secondary_only_drops_novostroyki(scraper: CianScraper) -> None:
-    """secondary_only=True: новостройки (listing_segment=='novostroyki') не попадают
-    в on_bucket и в итоговый список; вторичка и None-сегмент — пропускаются.
-    """
-    on_bucket_calls: list[list[ScrapedLot]] = []
-
-    def fake_on_bucket(bucket_key: str, lots: list[ScrapedLot]) -> None:
-        on_bucket_calls.append(list(lots))
-
-    # Микс лотов: 2 новостройки + 2 вторички + 1 None-сегмент
-    mixed_lots = [
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/nb_1/",
-            source_id="nb_1",
-            address="ЖК Новый",
-            price_rub=5_000_000,
-            listing_segment="novostroyki",
-        ),
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/nb_2/",
-            source_id="nb_2",
-            address="ЖК Новый 2",
-            price_rub=6_000_000,
-            listing_segment="novostroyki",
-        ),
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/vt_1/",
-            source_id="vt_1",
-            address="ул. Ленина",
-            price_rub=3_500_000,
-            listing_segment="vtorichka",
-        ),
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/vt_2/",
-            source_id="vt_2",
-            address="ул. Мира",
-            price_rub=4_000_000,
-            listing_segment="vtorichka",
-        ),
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/none_seg/",
-            source_id="none_seg",
-            address="ул. Пушкина",
-            price_rub=3_000_000,
-            listing_segment=None,
-        ),
-    ]
-
-    page_call_count = 0
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        nonlocal page_call_count
-        page_call_count += 1
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 5  # <= cap → leaf-бакет
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        # Первая страница отдаёт весь микс; вторая — пусто
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return mixed_lots if page == 1 else []
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        lots = await scraper.fetch_all_secondary(
-            rooms_buckets=[(1,)],
-            price_cap_per_bucket=1400,
-            secondary_only=True,
-            on_bucket=fake_on_bucket,
-        )
-
-    # Итоговый список: только vtorichka + None-сегмент (3 лота), без новостроек
-    assert len(lots) == 3, f"Ожидалось 3 лота (без novostroyki), получено {len(lots)}"
-    lot_ids = {lot.source_id for lot in lots}
-    assert "nb_1" not in lot_ids, "nb_1 (novostroyki) не должен присутствовать"
-    assert "nb_2" not in lot_ids, "nb_2 (novostroyki) не должен присутствовать"
-    assert "vt_1" in lot_ids, "vt_1 (vtorichka) должен присутствовать"
-    assert "vt_2" in lot_ids, "vt_2 (vtorichka) должен присутствовать"
-    assert "none_seg" in lot_ids, "none_seg (listing_segment=None) должен присутствовать"
-
-    # on_bucket вызывается на каждый seed-брекет (тот же микс на page=1 каждого),
-    # но новостройки отфильтрованы в каждом вызове.
-    assert len(on_bucket_calls) == len(
-        _SEED
-    ), f"Ожидалось {len(_SEED)} вызовов on_bucket, получено {len(on_bucket_calls)}"
-    for call_lots in on_bucket_calls:
-        bucket_ids = {lot.source_id for lot in call_lots}
-        assert "nb_1" not in bucket_ids
-        assert "nb_2" not in bucket_ids
-        assert len(call_lots) == 3
-
-
-@pytest.mark.asyncio
-async def test_secondary_only_false_passes_all(scraper: CianScraper) -> None:
-    """secondary_only=False: новостройки НЕ отфильтровываются — все лоты проходят."""
-    mixed_lots = [
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/nb_1/",
-            source_id="nb_1",
-            address="ЖК Новый",
-            price_rub=5_000_000,
-            listing_segment="novostroyki",
-        ),
-        ScrapedLot(
-            source="cian",
-            source_url="https://ekb.cian.ru/sale/flat/vt_1/",
-            source_id="vt_1",
-            address="ул. Ленина",
-            price_rub=3_500_000,
-            listing_segment="vtorichka",
-        ),
-    ]
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 2
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return mixed_lots if page == 1 else []
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        lots = await scraper.fetch_all_secondary(
-            rooms_buckets=[(1,)],
-            price_cap_per_bucket=1400,
-            secondary_only=False,
-        )
-
-    assert len(lots) == 2, f"Ожидалось 2 лота (фильтр выключен), получено {len(lots)}"
-    lot_ids = {lot.source_id for lot in lots}
-    assert "nb_1" in lot_ids, "nb_1 должен присутствовать при secondary_only=False"
-    assert "vt_1" in lot_ids
-
-
-# ── #1393: tail-loss warning при превышении page cap ─────────────────────────
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_warns_on_page_cap_tail_loss(
-    scraper: CianScraper, caplog: pytest.LogCaptureFixture
-) -> None:
-    """#1393: когда totalOffers > max_pages_per_bucket × offers_per_page,
-    _walk_price_range логирует WARNING с числом потерянных офферов.
-
-    Сценарий: bracket < _MIN_BRACKET (guard срабатывает → force-paginate),
-    totalOffers = max_pages * offers_per_page + 50 → tail_loss = 50.
-    """
-    max_pages_cap = 5  # нарочно мало, чтобы быстро дойти до cap
-    offers_per_page = _CIAN_OFFERS_PER_PAGE  # 28
-    # total > cap (2000 > любой price_cap_per_bucket=1400) И bracket < _MIN_BRACKET
-    # → force-paginate path; pages_needed = ceil(2000/28) = 72 > max_pages_cap=5
-    total_offers = max_pages_cap * offers_per_page + 50  # 190; tail_loss=50
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return total_offers
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return [_make_lot(f"lot_p{page}_{i}") for i in range(offers_per_page)]
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-        caplog.at_level(logging.WARNING, logger="app.services.scrapers.cian"),
-    ):
-        await scraper._walk_price_range(
-            rooms=(1,),
-            # Узкий диапазон: bracket < _MIN_BRACKET → force-paginate guard
-            lo=1_000_000,
-            hi=1_000_000 + _MIN_BRACKET - 1,
-            seen={},
-            price_cap_per_bucket=1400,  # total(190) < 1400 → нет split, только leaf
-            max_pages_per_bucket=max_pages_cap,
-        )
-
-    # Должно быть предупреждение о tail-loss
-    tail_warnings = [
-        r for r in caplog.records if "tail loss" in r.message and r.levelno == logging.WARNING
-    ]
-    assert tail_warnings, (
-        f"Ожидалось WARNING о tail loss, но записей нет. "
-        f"Все warnings: {[r.message for r in caplog.records if r.levelno == logging.WARNING]}"
-    )
-    # Сообщение должно содержать число потерянных офферов (50)
-    assert any(
-        "50" in r.message for r in tail_warnings
-    ), f"WARNING должен содержать tail_loss=50, got: {[r.message for r in tail_warnings]}"
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_no_warning_when_within_page_cap(
-    scraper: CianScraper, caplog: pytest.LogCaptureFixture
-) -> None:
-    """#1393: когда totalOffers <= max_pages_per_bucket × offers_per_page,
-    tail-loss WARNING не должен эмитироваться (данные собраны полностью).
-    """
-    max_pages_cap = 54
-    total_offers = 56  # ceil(56/28) = 2 страницы < 54 cap → нет потерь
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return total_offers
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return [_make_lot(f"lot_p{page}_{i}") for i in range(10)]
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-        caplog.at_level(logging.WARNING, logger="app.services.scrapers.cian"),
-    ):
-        await scraper._walk_price_range(
-            rooms=(1,),
-            lo=0,
-            hi=_MAX_PRICE,
-            seen={},
-            price_cap_per_bucket=1400,
-            max_pages_per_bucket=max_pages_cap,
-        )
-
-    tail_warnings = [
-        r for r in caplog.records if "tail loss" in r.message and r.levelno == logging.WARNING
-    ]
-    assert not tail_warnings, (
-        f"Не ожидалось WARNING о tail loss при total={total_offers} <= page cap, "
-        f"но получено: {[r.message for r in tail_warnings]}"
-    )
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_open_bracket_no_split(scraper: CianScraper) -> None:
-    """Открытый брекет (hi=None) не делится даже при total > cap — leaf без maxprice.
-
-    bucket_key оканчивается на ':open'; maxprice в запросах всегда None (без потолка).
-    """
-    on_bucket_keys: list[str] = []
-    seen_max: list[int | None] = []
-
-    async def fake_fetch_page_html(
-        rooms: tuple, page: int, min_price: int | None, max_price: int | None
-    ) -> str:
-        seen_max.append(max_price)
-        return f"page={page}"
-
-    def fake_extract_total_offers(html: str) -> int | None:
-        return 5000  # > cap, но открытый брекет делить нельзя → leaf
-
-    def fake_parse_serp_html(html: str) -> list[ScrapedLot]:
-        import re
-
-        page_m = re.search(r"page=(\d+)", html)
-        page = int(page_m.group(1)) if page_m else 1
-        return [_make_lot(f"lot_p{page}")] if page == 1 else []
-
-    with (
-        patch.object(scraper, "_fetch_page_html", side_effect=fake_fetch_page_html),
-        patch.object(scraper, "_extract_total_offers", side_effect=fake_extract_total_offers),
-        patch.object(scraper, "_parse_serp_html", side_effect=fake_parse_serp_html),
-        patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock),
-        patch.object(scraper, "_rotate_ip", return_value=False),
-    ):
-        await scraper._walk_price_range(
-            rooms=(1,),
-            lo=250_000_000,
-            hi=None,
-            seen={},
-            price_cap_per_bucket=1400,
-            max_pages_per_bucket=54,
-            on_bucket=lambda k, lots: on_bucket_keys.append(k),
-        )
-
-    # Открытый брекет: ровно один leaf-бакет с ключом ':open', maxprice всегда None.
-    assert on_bucket_keys == [
-        "room1:250000000:open"
-    ], f"Открытый брекет должен дать один ':open'-бакет, got {on_bucket_keys}"
-    assert all(
-        m is None for m in seen_max
-    ), f"Открытый брекет: maxprice всегда None (без потолка), got {seen_max}"
diff --git a/tradein-mvp/backend/tests/scrapers/test_domclick.py b/tradein-mvp/backend/tests/scrapers/test_domclick.py
deleted file mode 100644
index dc613020..00000000
--- a/tradein-mvp/backend/tests/scrapers/test_domclick.py
+++ /dev/null
@@ -1,630 +0,0 @@
-"""Unit-тесты для DomClick BFF scraper (#1968).
-
-Тестируем:
-- _extract_json: bare + 
-wrapped → dict; QRATOR-маркер → DomClickBlockedError
-- _build_offers_url / _build_count_url: корректные параметры, "5+" → "5%2B",
-  offers имеет offset/limit, count — нет
-- _map_item: маппинг realistic offer dict → ScrapedLot; price=0 → None; studio
-- _is_geo_ok: гео-гард по bbox + offerRegionName
-- _sweep_bucket: price-split при count > OFFSET_CAP
-- DomClickScraper: source/name, fetch_around raises NotImplementedError
-"""
-
-from __future__ import annotations
-
-import asyncio
-from typing import Any
-
-import pytest
-
-from app.services.scrapers.domclick import (
-    LTE_MAX,
-    OFFSET_CAP,
-    DomClickScraper,
-    _build_count_url,
-    _build_offers_url,
-    _extract_json,
-)
-from app.services.scrapers.domclick_exceptions import DomClickBlockedError
-
-# ── _extract_json ─────────────────────────────────────────────────────────────
-
-
-def test_extract_json_bare_body() -> None:
-    """Bare JSON body (без HTML-обёртки) корректно парсится."""
-    html = '{"result": {"items": [], "snippetsCount": 42}}'
-    data = _extract_json(html)
-    assert data["result"]["snippetsCount"] == 42
-
-
-def test_extract_json_pre_wrapped() -> None:
-    """JSON в 
-теге корректно парсится."""
-    html = (
-        "DomClick"
-        '
{"result": {"items": [{"id": 1}]}}
'
-    )
-    data = _extract_json(html)
-    assert data["result"]["items"][0]["id"] == 1
-
-
-def test_extract_json_qrator_marker_raises() -> None:
-    """HTML с QRATOR-маркером вызывает DomClickBlockedError."""
-    blocked_html = (
-        "QRATOR protection"
-        "Access denied by bot_mitigation system"
-    )
-    with pytest.raises(DomClickBlockedError):
-        _extract_json(blocked_html)
-
-
-def test_extract_json_captcha_raises() -> None:
-    """HTML с captcha-маркером вызывает DomClickBlockedError."""
-    html = "Please solve the captcha to continue"
-    with pytest.raises(DomClickBlockedError):
-        _extract_json(html)
-
-
-def test_extract_json_no_json_raises() -> None:
-    """HTML без JSON-объекта вызывает ValueError."""
-    with pytest.raises(ValueError, match="No JSON"):
-        _extract_json("plain text")
-
-
-# ── _build_offers_url ────────────────────────────────────────────────────────
-
-
-def test_build_offers_url_basic_params() -> None:
-    url = _build_offers_url("1", None, None, 0)
-    assert "deal_type=sale" in url
-    assert "category=living" in url
-    assert "offer_type=flat" in url
-    assert "rooms=1" in url
-    assert "offset=0" in url
-    assert "limit=20" in url
-    assert "sale_price__gte" not in url
-    assert "sale_price__lte" not in url
-
-
-def test_build_offers_url_5plus_encoded() -> None:
-    """rooms='5+' должен кодироваться как '5%2B' в URL."""
-    url = _build_offers_url("5+", None, None, 0)
-    assert "5%2B" in url
-    # literal '+' в URL не должно быть для значения rooms
-    assert "rooms=5+" not in url
-
-
-def test_build_offers_url_with_price_range() -> None:
-    url = _build_offers_url("2", 3_000_000, 6_000_000, 40)
-    assert "sale_price__gte=3000000" in url
-    assert "sale_price__lte=6000000" in url
-    assert "offset=40" in url
-
-
-def test_build_offers_url_ekb_guid_present() -> None:
-    url = _build_offers_url("3", None, None, 0)
-    assert "0d475b79-88de-4054-818c-37d8f9d0d440" in url
-    assert "aids=20561" in url
-
-
-def test_build_offers_url_bff_host() -> None:
-    url = _build_offers_url("1", None, None, 0)
-    assert url.startswith("https://bff-search-web.domclick.ru/api/offers/v1?")
-
-
-# ── _build_count_url ─────────────────────────────────────────────────────────
-
-
-def test_build_count_url_no_offset_limit() -> None:
-    """count url не должен содержать offset и limit."""
-    url = _build_count_url("2", None, None)
-    assert "offset" not in url
-    assert "limit" not in url
-
-
-def test_build_count_url_5plus_encoded() -> None:
-    url = _build_count_url("5+", None, None)
-    assert "5%2B" in url
-    assert "rooms=5+" not in url
-
-
-def test_build_count_url_with_price() -> None:
-    url = _build_count_url("1", 1_000_000, 4_000_000)
-    assert "sale_price__gte=1000000" in url
-    assert "sale_price__lte=4000000" in url
-
-
-def test_build_count_url_bff_host() -> None:
-    url = _build_count_url("1", None, None)
-    assert url.startswith("https://bff-search-web.domclick.ru/api/offers/count/v1?")
-
-
-# ── _map_item → ScrapedLot ────────────────────────────────────────────────────
-
-_REALISTIC_ITEM: dict[str, Any] = {
-    "id": 2075671636,
-    "path": "https://domclick.ru/card/sale__flat__2075671636",
-    "location": {"lat": 56.838, "lon": 60.612},
-    "address": {"displayName": "Екатеринбург, улица Малышева, 1"},
-    "objectInfo": {"area": 53.5, "rooms": 2, "floor": 5, "isApartment": False},
-    "house": {"floors": 15, "buildYear": 2003},
-    "price": 5_200_000,
-    "squarePrice": 97196,
-    "flatComplex": {"id": 12345, "name": "ЖК Тест", "slug": "zhk-test"},
-    "isRosreestrApproved": True,
-    "publishedDate": "2026-06-01T10:00:00+05:00",
-    "updatedDate": "2026-06-15T12:00:00+05:00",
-    "lastPriceHistoryState": {"direction": "DECREASED"},
-    "offerRegionName": "Екатеринбург",
-}
-
-
-def test_map_item_basic_fields() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    lot = scraper._map_item(_REALISTIC_ITEM)
-    assert lot is not None
-    assert lot.source == "domklik"
-    assert lot.source_id == "2075671636"
-    assert lot.source_url == "https://domclick.ru/card/sale__flat__2075671636"
-    assert lot.lat == pytest.approx(56.838)
-    assert lot.lon == pytest.approx(60.612)
-    assert lot.address == "Екатеринбург, улица Малышева, 1"
-    assert lot.rooms == 2
-    assert lot.area_m2 == pytest.approx(53.5)
-    assert lot.floor == 5
-    assert lot.total_floors == 15
-    assert lot.year_built == 2003
-    assert lot.price_rub == 5_200_000
-    assert lot.price_per_m2 == 97196
-    assert lot.listing_segment == "vtorichka"
-
-
-def test_map_item_publish_date() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    from datetime import date
-
-    lot = scraper._map_item(_REALISTIC_ITEM)
-    assert lot is not None
-    assert lot.publish_date == date(2026, 6, 1)
-
-
-def test_map_item_raw_payload() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    lot = scraper._map_item(_REALISTIC_ITEM)
-    assert lot is not None
-    assert lot.raw_payload is not None
-    assert lot.raw_payload["isRosreestrApproved"] is True
-    assert lot.raw_payload["flatComplex"]["name"] == "ЖК Тест"
-    assert lot.raw_payload["lastPriceHistoryState"]["direction"] == "DECREASED"
-
-
-def test_map_item_studio_force_rooms_zero() -> None:
-    """force_rooms=0 (студия-бакет) перебивает objectInfo.rooms."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {**_REALISTIC_ITEM, "objectInfo": {"area": 28.0, "rooms": 1, "floor": 3}}
-    lot = scraper._map_item(item, force_rooms=0)
-    assert lot is not None
-    assert lot.rooms == 0
-
-
-def test_map_item_zero_price_returns_none() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {**_REALISTIC_ITEM, "price": 0}
-    assert scraper._map_item(item) is None
-
-
-def test_map_item_zero_sentinels_become_none() -> None:
-    """buildYear/area/floor/total_floors == 0 (sentinel «неизвестно») → None.
-
-    rooms=0 НЕ затрагивается (валидная студия), но здесь force_rooms не задан и
-    objectInfo.rooms=2, так что rooms остаётся 2.
-    """
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {
-        **_REALISTIC_ITEM,
-        "objectInfo": {"area": 0, "rooms": 2, "floor": 0},
-        "house": {"floors": 0, "buildYear": 0},
-    }
-    lot = scraper._map_item(item)
-    assert lot is not None
-    assert lot.area_m2 is None
-    assert lot.floor is None
-    assert lot.total_floors is None
-    assert lot.year_built is None
-    assert lot.rooms == 2  # rooms 0-sentinel НЕ применяется
-
-
-def test_map_item_rooms_zero_studio_preserved() -> None:
-    """force_rooms=0 даёт rooms=0 (студия) — 0 здесь валиден, не sentinel."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {**_REALISTIC_ITEM, "objectInfo": {"area": 24.0, "rooms": 0, "floor": 2}}
-    lot = scraper._map_item(item, force_rooms=0)
-    assert lot is not None
-    assert lot.rooms == 0
-
-
-def test_map_item_price_per_m2_fallback() -> None:
-    """Если squarePrice отсутствует, price_per_m2 вычисляется из price / area_m2."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {**_REALISTIC_ITEM, "squarePrice": None}
-    lot = scraper._map_item(item)
-    assert lot is not None
-    assert lot.price_per_m2 == int(5_200_000 / 53.5)
-
-
-# ── _map_item Layer A enrichment (description / repair / minhash / seller) ─────
-
-_VALID_REPAIR_ENUM = {"needs_repair", "standard", "good", "excellent"}
-
-
-def test_map_item_description_repair_minhash() -> None:
-    """Offer с описанием → description set, repair_state выведен, minhash len 32."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {
-        **_REALISTIC_ITEM,
-        "description": "Продаётся светлая квартира с евроремонтом, рядом метро.",
-    }
-    lot = scraper._map_item(item)
-    assert lot is not None
-    assert lot.description == "Продаётся светлая квартира с евроремонтом, рядом метро."
-    # «евроремонт» → good (см. repair_state_normalizer._TEXT_PATTERNS)
-    assert lot.repair_state == "good"
-    assert lot.repair_state in _VALID_REPAIR_ENUM
-    assert lot.description_minhash is not None
-    assert len(lot.description_minhash) == 32
-
-
-def test_map_item_description_stripped() -> None:
-    """description триммится; пустой/whitespace → None."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {**_REALISTIC_ITEM, "description": "   Без отделки, требуется ремонт.  "}
-    lot = scraper._map_item(item)
-    assert lot is not None
-    assert lot.description == "Без отделки, требуется ремонт."
-    # «без отделк» → needs_repair
-    assert lot.repair_state == "needs_repair"
-
-
-def test_map_item_enrichment_raw_payload() -> None:
-    """Новые SERP-поля попадают в raw_payload (is_sber_collateral и т.д.)."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {
-        **_REALISTIC_ITEM,
-        "isSberCollateral": True,
-        "hasDiscount": True,
-        "discountValue": 2.3,
-        "duplicatesOfferCount": 3,
-        "seller": {"company": "Этажи", "agent": "Иван Петров"},
-    }
-    lot = scraper._map_item(item)
-    assert lot is not None
-    assert lot.raw_payload is not None
-    assert lot.raw_payload["is_sber_collateral"] is True
-    assert lot.raw_payload["has_discount"] is True
-    assert lot.raw_payload["discount_value"] == 2.3
-    assert lot.raw_payload["duplicates_offer_count"] == 3
-    assert lot.raw_payload["seller"] == {"company": "Этажи", "agent": "Иван Петров"}
-    # agency_name извлекается из seller.company (приоритет company → agent)
-    assert lot.agency_name == "Этажи"
-
-
-def test_map_item_agency_name_from_agent_dict() -> None:
-    """seller.company отсутствует, agent — dict с name → agency_name = name."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    item = {**_REALISTIC_ITEM, "seller": {"agent": {"name": "Мария С."}}}
-    lot = scraper._map_item(item)
-    assert lot is not None
-    assert lot.agency_name == "Мария С."
-    assert lot.raw_payload is not None
-    assert lot.raw_payload["seller"] == {"agent": {"name": "Мария С."}}
-
-
-def test_map_item_no_description_no_crash() -> None:
-    """Offer без description: не падает, description/repair_state/minhash = None."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    # _REALISTIC_ITEM не содержит description/seller/enrichment-полей
-    lot = scraper._map_item(_REALISTIC_ITEM)
-    assert lot is not None
-    assert lot.description is None
-    assert lot.repair_state is None
-    assert lot.description_minhash is None
-    assert lot.agency_name is None
-
-
-def test_map_item_enrichment_preserves_existing_raw_payload_and_drops_none() -> None:
-    """Существующие raw_payload-ключи не теряются; None-enrich-ключи отброшены."""
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    lot = scraper._map_item(_REALISTIC_ITEM)
-    assert lot is not None
-    assert lot.raw_payload is not None
-    # Существующие ключи на месте
-    assert lot.raw_payload["isRosreestrApproved"] is True
-    assert lot.raw_payload["flatComplex"]["name"] == "ЖК Тест"
-    assert lot.raw_payload["lastPriceHistoryState"]["direction"] == "DECREASED"
-    assert "updatedDate" in lot.raw_payload
-    # None-значные enrich-ключи отсутствуют (компактный payload)
-    assert "is_sber_collateral" not in lot.raw_payload
-    assert "has_discount" not in lot.raw_payload
-    assert "discount_value" not in lot.raw_payload
-    assert "duplicates_offer_count" not in lot.raw_payload
-    assert "seller" not in lot.raw_payload
-
-
-# ── _is_geo_ok — гео-гард ────────────────────────────────────────────────────
-
-
-def test_is_geo_ok_valid_ekb_item() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    assert scraper._is_geo_ok(_REALISTIC_ITEM) is True
-
-
-def test_is_geo_ok_wrong_region_name() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    item = {**_REALISTIC_ITEM, "offerRegionName": "Москва"}
-    assert scraper._is_geo_ok(item) is False
-
-
-def test_is_geo_ok_out_of_bbox_lat() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    item = {**_REALISTIC_ITEM, "location": {"lat": 55.5, "lon": 60.6}}
-    assert scraper._is_geo_ok(item) is False
-
-
-def test_is_geo_ok_out_of_bbox_lon() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    item = {**_REALISTIC_ITEM, "location": {"lat": 56.8, "lon": 61.5}}
-    assert scraper._is_geo_ok(item) is False
-
-
-def test_is_geo_ok_missing_coords() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    item = {**_REALISTIC_ITEM, "location": {"lat": None, "lon": None}}
-    assert scraper._is_geo_ok(item) is False
-
-
-# ── _sweep_bucket price-split ────────────────────────────────────────────────
-
-
-async def test_sweep_bucket_price_split_boundaries(monkeypatch: pytest.MonkeyPatch) -> None:
-    """_sweep_bucket: top-level unbounded split даёт contiguous non-overlapping ranges.
-
-    rooms="2" count>cap один раз, потом ≤cap → ровно 2 leaf-бакета, покрывающих
-    [0, LTE_MAX] без перекрытия и без зазора: [(0, mid), (mid+1, LTE_MAX)].
-    """
-    call_count: dict[str, int] = {"n": 0}
-    paginate_calls: list[tuple[str, int | None, int | None]] = []
-
-    async def fake_count(
-        self: DomClickScraper,
-        fetcher: object,
-        rooms: str,
-        price_gte: int | None,
-        price_lte: int | None,
-    ) -> int:
-        call_count["n"] += 1
-        # Первый вызов (unbounded) → over cap → price-split; остальные → под cap
-        return OFFSET_CAP + 1 if call_count["n"] == 1 else 1
-
-    async def fake_paginate(
-        self: DomClickScraper,
-        fetcher: object,
-        rooms: str,
-        price_gte: int | None,
-        price_lte: int | None,
-        seen_ids: set[str],
-        out_lots: list,
-        pages: int,
-    ) -> None:
-        paginate_calls.append((rooms, price_gte, price_lte))
-
-    monkeypatch.setattr(DomClickScraper, "_count", fake_count)
-    monkeypatch.setattr(DomClickScraper, "_paginate", fake_paginate)
-
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    scraper.geo_filtered = 0
-    scraper.blocked = False
-    scraper.fetch_errors = 0
-    scraper.request_delay_sec = 0.0
-
-    await scraper._sweep_bucket(
-        fetcher=object(),
-        rooms="2",
-        price_gte=None,
-        price_lte=None,
-        seen_ids=set(),
-        out_lots=[],
-        pages=10,
-    )
-
-    # 1 count call (over cap) + 2 recursive count calls (≤ cap) = 3 total
-    assert call_count["n"] == 3
-    # Both sub-buckets paginated, с точными границами split'а unbounded-диапазона.
-    mid = (0 + LTE_MAX) // 2
-    assert paginate_calls == [("2", 0, mid), ("2", mid + 1, LTE_MAX)]
-    # Sanity: contiguous (нет зазора) + non-overlapping + покрывают весь хвост.
-    assert paginate_calls[0][2] + 1 == paginate_calls[1][1]  # mid + 1
-    assert paginate_calls[0][1] == 0
-    assert paginate_calls[1][2] == LTE_MAX
-
-
-async def test_sweep_bucket_zero_count_no_paginate(monkeypatch: pytest.MonkeyPatch) -> None:
-    """_sweep_bucket при count==0 не вызывает _paginate."""
-    paginate_called = False
-
-    async def fake_count(self: DomClickScraper, *_args: object, **_kwargs: object) -> int:
-        return 0
-
-    async def fake_paginate(self: DomClickScraper, *_args: object, **_kwargs: object) -> None:
-        nonlocal paginate_called
-        paginate_called = True
-
-    monkeypatch.setattr(DomClickScraper, "_count", fake_count)
-    monkeypatch.setattr(DomClickScraper, "_paginate", fake_paginate)
-
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    scraper.geo_filtered = 0
-    scraper.blocked = False
-    scraper.fetch_errors = 0
-    scraper.request_delay_sec = 0.0
-
-    await scraper._sweep_bucket(
-        fetcher=object(),
-        rooms="3",
-        price_gte=None,
-        price_lte=None,
-        seen_ids=set(),
-        out_lots=[],
-        pages=5,
-    )
-    assert paginate_called is False
-
-
-# ── _paginate resilience — mid-bucket parse error ────────────────────────────
-
-
-class _FakeFetcher:
-    """Стаб BrowserFetcher: возвращает заранее заданные ответы по порядку вызовов."""
-
-    def __init__(self, responses: list[str]) -> None:
-        self._responses = responses
-        self._idx = 0
-
-    async def fetch(self, _url: str) -> str:
-        resp = self._responses[min(self._idx, len(self._responses) - 1)]
-        self._idx += 1
-        return resp
-
-
-async def test_paginate_mid_bucket_parse_error_preserves_lots() -> None:
-    """Garbled JSON на page 2 → bucket breaks, page-1 lots survive, fetch_errors++.
-
-    Page 1 — валидный JSON с 20 items (PAGE_SIZE) → продолжаем; page 2 — мусор
-    (ValueError из _extract_json) → fetch_errors++ и break, уже собранные lots целы.
-    """
-    import json
-
-    page1_items = [
-        {
-            "id": 1000 + i,
-            "path": f"https://domclick.ru/card/sale__flat__{1000 + i}",
-            "location": {"lat": 56.84, "lon": 60.6},
-            "address": {"displayName": "Екатеринбург, тест"},
-            "objectInfo": {"area": 40.0, "rooms": 1, "floor": 3},
-            "house": {"floors": 9, "buildYear": 2005},
-            "price": 4_000_000 + i,
-            "squarePrice": 100000,
-            "offerRegionName": "Екатеринбург",
-        }
-        for i in range(20)
-    ]
-    page1 = json.dumps({"result": {"items": page1_items}})
-    garbled = "
{not valid json at all
"
-    fetcher = _FakeFetcher([page1, garbled])
-
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    scraper.geo_filtered = 0
-    scraper.blocked = False
-    scraper.fetch_errors = 0
-    scraper.request_delay_sec = 0.0
-
-    out_lots: list = []
-    await scraper._paginate(
-        fetcher=fetcher,
-        rooms="1",
-        price_gte=None,
-        price_lte=None,
-        seen_ids=set(),
-        out_lots=out_lots,
-        pages=10,
-    )
-
-    assert len(out_lots) == 20  # page-1 lots сохранены
-    assert scraper.fetch_errors == 1  # page-2 garbled посчитан
-
-
-async def test_fetch_city_bucket_error_continues_to_next_bucket(
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    """Ошибка в одном бакете НЕ убивает весь sweep — остальные бакеты отрабатывают."""
-    swept: list[str] = []
-
-    async def fake_sweep_bucket(
-        self: DomClickScraper,
-        fetcher: object,
-        rooms: str,
-        price_gte: int | None,
-        price_lte: int | None,
-        seen_ids: set[str],
-        out_lots: list,
-        pages: int,
-    ) -> None:
-        swept.append(rooms)
-        if rooms == "2":
-            raise ValueError("garbled bucket")
-
-    # Заглушка BrowserFetcher через async context manager
-    class _BFStub:
-        async def __aenter__(self) -> _BFStub:
-            return self
-
-        async def __aexit__(self, *_a: object) -> None:
-            return None
-
-        async def fetch(self, _url: str) -> str:
-            return "{}"
-
-    import app.services.scrapers.browser_fetcher as bf_mod
-
-    monkeypatch.setattr(DomClickScraper, "_sweep_bucket", fake_sweep_bucket)
-    monkeypatch.setattr(bf_mod, "BrowserFetcher", lambda **_k: _BFStub())
-
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    scraper.geo_filtered = 0
-    scraper.blocked = False
-    scraper.fetch_errors = 0
-    scraper.request_delay_sec = 0.0
-
-    lots = await scraper.fetch_city(city_id=4, rooms=None, pages=5)
-
-    # Все 6 бакетов посещены несмотря на ошибку в "2"
-    assert swept == ["st", "1", "2", "3", "4", "5+"]
-    assert scraper.fetch_errors == 1
-    assert scraper.blocked is False
-    assert lots == []
-
-
-# ── DomClickScraper — class attrs ─────────────────────────────────────────────
-
-
-def test_scraper_source_name() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    scraper.parse_failures = 0
-    scraper.geo_filtered = 0
-    scraper.blocked = False
-    scraper.request_delay_sec = 0.0
-    assert scraper.name == "domklik"
-    assert scraper.source == "domklik"
-    assert scraper.base_url == "https://domclick.ru"
-
-
-def test_fetch_around_raises() -> None:
-    scraper = DomClickScraper.__new__(DomClickScraper)
-    with pytest.raises(NotImplementedError, match="geo-radius"):
-        asyncio.run(scraper.fetch_around(56.8, 60.6))
diff --git a/tradein-mvp/backend/tests/scrapers/test_domclick_detail.py b/tradein-mvp/backend/tests/scrapers/test_domclick_detail.py
index cd629987..ff80ca8e 100644
--- a/tradein-mvp/backend/tests/scrapers/test_domclick_detail.py
+++ b/tradein-mvp/backend/tests/scrapers/test_domclick_detail.py
@@ -24,9 +24,10 @@ test_cian_detail.py — в проекте нет live-DB фикстуры, ид
 scraper_kit-миграции, 0 runtime importers) — тесты переведены на kit-эквивалент
 `scraper_kit.providers.domclick.detail` без изменения покрытия. Exceptions тоже
 переведены на `scraper_kit.domclick_exceptions` (kit держит свою независимую копию,
-`parse_detail_html`(kit) поднимает именно её — legacy `app.services.scrapers.
-domclick_exceptions` остаётся живым модулем для `domclick.py`, но здесь больше не
-подходящий класс для `pytest.raises`).
+`parse_detail_html`(kit) поднимает именно её). Legacy `app.services.scrapers.
+domclick`/`domclick_exceptions` удалены целиком (#2397 Part E2, 0 runtime
+importers) — `scraper_kit.domclick_exceptions` остаётся единственным живым классом
+для `pytest.raises`.
 """
 
 from __future__ import annotations
diff --git a/tradein-mvp/backend/tests/test_avito_houses_dedup.py b/tradein-mvp/backend/tests/test_avito_houses_dedup.py
deleted file mode 100644
index d32ad4b9..00000000
--- a/tradein-mvp/backend/tests/test_avito_houses_dedup.py
+++ /dev/null
@@ -1,381 +0,0 @@
-"""Tests for avito_houses dedup fix — _persist_house делегирует в match_or_create_house.
-
-Проверяет:
-  1. Два HouseInfo с разными ext_id но одинаковым адресом и координатами
-     → _persist_house возвращает ОДИН house_id (не плодит дубль).
-  2. Enrichment-поля (rating_score, developer_name) реально записываются в UPDATE.
-  3. _persist_house не вызывает db.commit().
-  4. upsert_house и _upsert_house_with_url делегируют в _persist_house.
-  5. #1820: два id-less URL (ext_id=0, без адреса) НЕ создают house_sources(avito,"0").
-
-Использует тот же паттерн mock-DB что в test_matching.py и test_same_building_match.py.
-"""
-
-import json
-import os
-
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
-
-from unittest.mock import MagicMock, patch
-
-from app.services.scrapers.avito_houses import (
-    HouseCatalogEnrichment,
-    HouseInfo,
-    _persist_house,
-    _upsert_house_with_url,
-    save_house_catalog_enrichment,
-    upsert_house,
-)
-
-# ---------------------------------------------------------------------------
-# Helpers
-# ---------------------------------------------------------------------------
-
-
-def _make_house(ext_id: int = 111, address: str = "ул. Ленина, 5") -> HouseInfo:
-    return HouseInfo(
-        ext_id=ext_id,
-        ext_id_hash="hash_" + str(ext_id),
-        short_address=address,
-        full_address="Свердловская обл., Екатеринбург, " + address,
-        lat=56.8378,
-        lon=60.5946,
-        year_built=2010,
-        house_type="monolith",
-        total_floors=16,
-        passenger_elevators=2,
-        cargo_elevators=1,
-        has_concierge=True,
-        closed_yard=False,
-        material_floors="concrete",
-        hot_water="central",
-        parking_type="underground",
-        has_playground=True,
-        developer_name="ПИК",
-        developer_key="pik",
-        infrastructure_summary="Школа, детсад",
-        infrastructure_walk_distance="5 мин",
-        rating_score=4.5,
-        rating_string="4.5 из 5",
-        reviews_count=120,
-        rating_distribution=[{"score": 5, "count": 80}],
-        map_pins=[{"type": "metro", "name": "Уральская"}],
-        raw_characteristics=[{"title": "Год постройки", "value": "2010"}],
-    )
-
-
-def _make_db_for_persist(canonical_id: int = 42) -> MagicMock:
-    """Mock DB для _persist_house: match_or_create_house ходит через db.execute,
-    поэтому патчим match_or_create_house целиком и проверяем вызов UPDATE.
-    """
-    db = MagicMock()
-    # db.execute для UPDATE houses SET ... — просто возвращаем None (rowcount не важен)
-    db.execute.return_value = MagicMock()
-    return db
-
-
-# ---------------------------------------------------------------------------
-# Test 1: два HouseInfo с разными ext_id, одинаковый адрес → один house_id
-# ---------------------------------------------------------------------------
-
-
-def test_persist_house_same_address_returns_same_id():
-    """Два HouseInfo с разными ext_id но одинаковым адресом → один canonical house_id.
-
-    Моделируем случай: первый вызов (ext_id=111) → match_or_create_house вернул 42 (new).
-    Второй вызов (ext_id=222, тот же адрес) → match_or_create_house вернул 42 (fingerprint).
-    Оба _persist_house должны вернуть 42.
-    """
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h1 = _make_house(ext_id=111)
-    h2 = _make_house(ext_id=222)  # другой ext_id, тот же адрес/координаты
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        side_effect=[(42, 1.0, "new"), (42, 0.9, "fingerprint")],
-    ) as mock_match:
-        id1 = _persist_house(db, h1, "https://avito.ru/house/111")
-        id2 = _persist_house(db, h2, "https://avito.ru/house/222")
-
-    assert id1 == 42, f"Первый вызов должен вернуть 42, получили {id1}"
-    assert id2 == 42, f"Второй вызов должен вернуть тот же 42, получили {id2}"
-    assert id1 == id2, "Оба вызова должны вернуть один house_id (нет дубля)"
-    assert mock_match.call_count == 2
-
-
-# ---------------------------------------------------------------------------
-# Test 2: enrichment-поля записываются в UPDATE
-# ---------------------------------------------------------------------------
-
-
-def test_persist_house_enrichment_fields_written():
-    """Enrichment-поля (rating_score, developer_name) передаются в UPDATE-запрос."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h = _make_house(ext_id=111)
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(42, 1.0, "source_exact"),
-    ):
-        _persist_house(db, h, "https://avito.ru/house/111")
-
-    # db.execute должен быть вызван 1 раз (UPDATE houses SET ...)
-    assert (
-        db.execute.call_count == 1
-    ), f"Ожидался 1 вызов db.execute (UPDATE), получили {db.execute.call_count}"
-
-    # Проверяем параметры bind
-    call_args = db.execute.call_args
-    sql_obj = call_args[0][0]
-    bind = call_args[0][1]
-
-    assert "UPDATE houses" in str(sql_obj), "SQL должен содержать UPDATE houses"
-    assert bind["rating_score"] == 4.5, f"rating_score={bind.get('rating_score')!r}"
-    assert bind["developer_name"] == "ПИК", f"developer_name={bind.get('developer_name')!r}"
-    assert bind["house_id"] == 42, f"house_id={bind.get('house_id')!r}"
-
-    # avito_id_hash должен быть передан
-    assert bind["avito_id_hash"] == "hash_111", f"avito_id_hash={bind.get('avito_id_hash')!r}"
-
-    # rating_distribution должен быть JSON-строкой
-    rd = bind.get("rating_distribution")
-    assert isinstance(rd, str), f"rating_distribution должен быть str (JSON), получили {type(rd)}"
-    assert json.loads(rd) == [{"score": 5, "count": 80}]
-
-
-# ---------------------------------------------------------------------------
-# Test 3: _persist_house не вызывает db.commit()
-# ---------------------------------------------------------------------------
-
-
-def test_persist_house_no_commit():
-    """_persist_house не должен вызывать db.commit() — коммитит вызывающий."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h = _make_house()
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(7, 1.0, "new"),
-    ):
-        _persist_house(db, h, "")
-
-    db.commit.assert_not_called()
-
-
-# ---------------------------------------------------------------------------
-# Test 4: upsert_house делегирует в _persist_house (с пустым url)
-# ---------------------------------------------------------------------------
-
-
-def test_upsert_house_delegates_to_persist_house():
-    """upsert_house(db, h) → _persist_house(db, h, "") → возвращает house_id."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-    h = _make_house()
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(55, 1.0, "new"),
-    ):
-        result = upsert_house(db, h)
-
-    assert result == 55
-
-    # Проверяем что UPDATE был вызван с url="" (пустая строка — нет url в этом пути)
-    bind = db.execute.call_args[0][1]
-    assert bind["url"] == "", f"url должен быть пустой строкой, получили {bind.get('url')!r}"
-
-
-# ---------------------------------------------------------------------------
-# Test 5: _upsert_house_with_url делегирует в _persist_house (с реальным url)
-# ---------------------------------------------------------------------------
-
-
-def test_upsert_house_with_url_delegates_to_persist_house():
-    """_upsert_house_with_url(db, h, url) → _persist_house(db, h, url) → house_id."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-    h = _make_house()
-    url = "https://www.avito.ru/catalog/houses/ekb/dom-1234"
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(99, 1.0, "new"),
-    ):
-        result = _upsert_house_with_url(db, h, url)
-
-    assert result == 99
-
-    # Проверяем что правильный url передан в UPDATE
-    bind = db.execute.call_args[0][1]
-    assert bind["url"] == url, f"url должен быть {url!r}, получили {bind.get('url')!r}"
-
-
-# ---------------------------------------------------------------------------
-# Test 6: match_or_create_house вызывается с правильными аргументами
-# ---------------------------------------------------------------------------
-
-
-def test_persist_house_calls_match_with_correct_args():
-    """_persist_house передаёт правильные аргументы в match_or_create_house."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h = _make_house(ext_id=333)
-    url = "https://avito.ru/333"
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(10, 1.0, "new"),
-    ) as mock_match:
-        _persist_house(db, h, url)
-
-    mock_match.assert_called_once_with(
-        db,
-        ext_source="avito",
-        ext_id="333",
-        address=h.short_address or h.full_address,
-        lat=h.lat,
-        lon=h.lon,
-        year_built=h.year_built,
-        source_url=url,
-    )
-
-
-# ---------------------------------------------------------------------------
-# Test 7: COALESCE — адрес/координаты не затираются через UPDATE параметры
-#   (проверяем что bind содержит address/lat/lon для COALESCE в SQL)
-# ---------------------------------------------------------------------------
-
-
-def test_persist_house_coalesce_fields_in_bind():
-    """UPDATE передаёт address/lat/lon в bind для COALESCE(houses., CAST(:x AS ...))."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h = _make_house()
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(1, 1.0, "new"),
-    ):
-        _persist_house(db, h, "")
-
-    bind = db.execute.call_args[0][1]
-    assert bind["address"] == h.short_address
-    assert bind["lat"] == h.lat
-    assert bind["lon"] == h.lon
-    assert bind["year_built"] == h.year_built
-
-
-# ---------------------------------------------------------------------------
-# Test 7b: #2003 — house_class передаётся в UPDATE bind (COALESCE fill-only)
-# ---------------------------------------------------------------------------
-
-
-def test_persist_house_house_class_in_bind():
-    """UPDATE передаёт house_class в bind для COALESCE(CAST(:house_class AS text), houses.*)."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h = _make_house()
-    h.house_class = "comfort"
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(1, 1.0, "new"),
-    ):
-        _persist_house(db, h, "")
-
-    bind = db.execute.call_args[0][1]
-    assert bind["house_class"] == "comfort"
-
-    sql_obj = db.execute.call_args[0][0]
-    assert "house_class" in str(sql_obj)
-
-
-def test_persist_house_house_class_none_does_not_null_via_coalesce():
-    """house_class=None в bind — SQL использует COALESCE(new, existing), не голый overwrite."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    h = _make_house()  # house_class остаётся дефолтным None
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(1, 1.0, "new"),
-    ):
-        _persist_house(db, h, "")
-
-    bind = db.execute.call_args[0][1]
-    assert bind["house_class"] is None
-
-    sql_obj = str(db.execute.call_args[0][0])
-    # SQL должен явно защищать существующее значение через COALESCE, а не
-    # безусловный overwrite (который занулил бы значение от другого источника).
-    assert "COALESCE(CAST(:house_class AS text), houses.house_class)" in sql_obj
-
-
-# ---------------------------------------------------------------------------
-# Test 8: #1820 — ext_id=0 graceful-skip гард
-#   Два id-less URL (ext_id=0, без адреса) → match_or_create_house НЕ вызывается,
-#   house_sources("avito","0") НЕ создаётся, возвращается sentinel dict с house_id=0.
-# ---------------------------------------------------------------------------
-
-
-def test_save_house_catalog_enrichment_skips_zero_ext_id_no_address():
-    """#1820: ext_id=0 + нет адреса → гард пропускает персистирование canonical house."""
-    db = MagicMock()
-
-    # Создаём два id-less graceful-skip enrichment (как _parse_development_or_skip возвращает)
-    enrichment_a = HouseCatalogEnrichment(
-        house_url="https://www.avito.ru/catalog/novostroyki/dom-A/",
-        house=HouseInfo(ext_id=0),
-    )
-    enrichment_b = HouseCatalogEnrichment(
-        house_url="https://www.avito.ru/catalog/novostroyki/dom-B/",
-        house=HouseInfo(ext_id=0),
-    )
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-    ) as mock_match:
-        result_a = save_house_catalog_enrichment(db, enrichment_a)
-        result_b = save_house_catalog_enrichment(db, enrichment_b)
-
-    # match_or_create_house НЕ должен вызываться — гард сработал до него
-    mock_match.assert_not_called()
-
-    # Возвращается sentinel dict с house_id=0, не плодятся дубли
-    assert result_a["house_id"] == 0, f"ожидали house_id=0, получили {result_a['house_id']}"
-    assert result_b["house_id"] == 0, f"ожидали house_id=0, получили {result_b['house_id']}"
-
-    # db.commit() НЕ вызывается (нет реальной транзакции)
-    db.commit.assert_not_called()
-
-
-def test_save_house_catalog_enrichment_with_address_persists_despite_zero_ext_id():
-    """#1820: ext_id=0 НО есть адрес → НЕ пропускаем (developmentPage с адресом, без avitoId)."""
-    db = MagicMock()
-    db.execute.return_value = MagicMock()
-
-    enrichment = HouseCatalogEnrichment(
-        house_url="https://www.avito.ru/catalog/novostroyki/dom-C/",
-        house=HouseInfo(ext_id=0, short_address="ул. Ленина, 1"),
-    )
-
-    with patch(
-        "app.services.scrapers.avito_houses.match_or_create_house",
-        return_value=(77, 0.9, "fingerprint"),
-    ) as mock_match:
-        result = save_house_catalog_enrichment(db, enrichment)
-
-    # Адрес есть → персистим через fingerprint tier
-    mock_match.assert_called_once()
-    assert result["house_id"] == 77
diff --git a/tradein-mvp/backend/tests/test_avito_houses_parse.py b/tradein-mvp/backend/tests/test_avito_houses_parse.py
deleted file mode 100644
index 69323072..00000000
--- a/tradein-mvp/backend/tests/test_avito_houses_parse.py
+++ /dev/null
@@ -1,790 +0,0 @@
-"""Offline smoke for avito_houses parser.
-
-Uses minimal JSON fixture mimicking Avito __preloadedState__ structure.
-Не требует сети, БД или curl_cffi — тестирует только чистый парсер.
-"""
-
-from datetime import date
-from pathlib import Path
-
-import pytest
-
-from app.services.scrapers.avito_houses import (
-    HouseCatalogEnrichment,
-    HouseInfo,
-    HouseReview,
-    MiniSerpListing,
-    PlacementHistoryItem,
-    RecommendationItem,
-    SellerInfo,
-    _extract_preloaded_state,
-    _parse_ru_date,
-    parse_houses_state,
-)
-
-FIXTURES_DIR = Path(__file__).parent / "fixtures"
-
-
-# ---------------------------------------------------------------------------
-# Minimal __preloadedState__ fixture (Python dict — не raw HTML)
-# ---------------------------------------------------------------------------
-
-MINIMAL_STATE: dict = {
-    "data": {
-        "data": {
-            "page": {
-                "placeholders": [
-                    {"type": "breadcrumbs"},
-                    {"type": "title"},
-                    {"type": "gallery", "props": {}},
-                    {
-                        "type": "housePage",
-                        "props": {
-                            "developmentData": {
-                                "avitoId": 3171365,
-                                "id": "MTE3LjM3MDg5MA",
-                                "title": "Akademika Postovskogo 17a",
-                                "address": "ул. Постовского, 17а",
-                                "fullAddress": (
-                                    "Свердловская обл., Екатеринбург, ул. Постовского, 17а"
-                                ),
-                                "coords": {"lat": 56.790699, "lng": 60.580191},
-                                "aboutDevelopment": {
-                                    "expandParams": {
-                                        "items": [
-                                            {
-                                                "title": "Параметры дома",
-                                                "params": [
-                                                    {"type": "Год постройки", "value": "2019"},
-                                                    {"type": "Этажей", "value": "25"},
-                                                    {"type": "Тип дома", "value": "Монолитный"},
-                                                    {"type": "Пассажирский лифт", "value": "2"},
-                                                    {"type": "Консьерж", "value": "Да"},
-                                                    {"type": "Закрытый двор", "value": "Да"},
-                                                    {"type": "Детская площадка", "value": "Да"},
-                                                ],
-                                            },
-                                        ]
-                                    }
-                                },
-                                "developer": {"name": "Атомстройкомплекс", "key": "atomstroy"},
-                                "mapPreview": {
-                                    "distance": "в 5 минутах ходьбы",
-                                    "objects": "школа, парк, аптека",
-                                    "pins": [{"lat": 56.79, "lng": 60.58, "type": "school"}],
-                                },
-                                "ratingBadge": {"info": {"score": 4.666, "scoreString": "4,7"}},
-                                "ratingSummaryStat": {
-                                    "ratingStat": [
-                                        {"score": 5, "count": 4},
-                                        {"score": 4, "count": 2},
-                                    ],
-                                    "reviewCount": 6,
-                                },
-                            }
-                        },
-                    },
-                    {
-                        "type": "reviews",
-                        "props": {
-                            "entries": [
-                                {
-                                    "type": "rating",
-                                    "value": {
-                                        "id": 12345,
-                                        "author": {"title": "Иван"},
-                                        "reviewTitle": "Хорошо",
-                                        "score": 5,
-                                        "modelExperience": "Живу в своей квартире",
-                                        "rated": "9 октября 2025",
-                                        "textSections": [
-                                            {"title": "", "text": "Общий отзыв"},
-                                            {"title": "Преимущества", "text": "Парк рядом"},
-                                            {"title": "Недостатки", "text": "Парковки мало"},
-                                        ],
-                                    },
-                                },
-                                {"type": "pages", "value": {"nextPageUrl": "/reviews?page=2"}},
-                            ]
-                        },
-                    },
-                    {
-                        "type": "miniSerp",
-                        "props": {
-                            "items": [
-                                {
-                                    "id": 7986882804,
-                                    "title": "3-к. квартира",
-                                    "description": "Описание",
-                                    "url": "/ekaterinburg/kvartiry/test_7986882804",
-                                    "price": "11 990 000 ₽",
-                                    "geo": {"content": "Чкаловская, 2,7 км", "colors": ["#cf2734"]},
-                                    "seller": {
-                                        "name": "DOMRF66",
-                                        "type": "Компания",
-                                        "from": "На Авито с июля 2012",
-                                        "url": "/brands/domrf66",
-                                        "logos": [],
-                                    },
-                                },
-                            ]
-                        },
-                    },
-                    {
-                        "type": "housePlacementHistory",
-                        "props": {
-                            "items": [
-                                {
-                                    "id": 7000000001,
-                                    "title": "1-к. квартира",
-                                    "startPrice": 5000000,
-                                    "startPriceDate": 1697000000,
-                                    "lastPrice": 4900000,
-                                    "lastPriceDate": 1715000000,
-                                    "exposure": 210,
-                                    "itemImage": {},
-                                },
-                            ]
-                        },
-                    },
-                    {"type": "uxFeedback"},
-                    {
-                        "type": "recommendations",
-                        "props": {
-                            "items": [
-                                {
-                                    "title": "Похожая",
-                                    "priceRange": "от 6 млн",
-                                    "additionalInfo": ["Адрес 1"],
-                                    "url": "/rec1",
-                                    "images": [{"208x156": "https://avito.ru/img1.jpg"}],
-                                },
-                            ]
-                        },
-                    },
-                    {"type": "padding"},
-                ]
-            }
-        }
-    }
-}
-
-
-# ---------------------------------------------------------------------------
-# Helpers
-# ---------------------------------------------------------------------------
-
-
-def test_ru_date_parse() -> None:
-    assert _parse_ru_date("9 октября 2025") == date(2025, 10, 9)
-    assert _parse_ru_date("31 декабря 2020") == date(2020, 12, 31)
-    assert _parse_ru_date("1 января 2000") == date(2000, 1, 1)
-    assert _parse_ru_date(None) is None
-    assert _parse_ru_date("bogus") is None
-    assert _parse_ru_date("") is None
-
-
-# ---------------------------------------------------------------------------
-# Main parse test
-# ---------------------------------------------------------------------------
-
-
-def test_parse_minimal_state() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-
-    assert isinstance(e, HouseCatalogEnrichment)
-    assert e.house_url == "/catalog/houses/ekaterinburg/test/3171365"
-
-
-def test_house_info() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    h = e.house
-
-    assert isinstance(h, HouseInfo)
-    assert h.ext_id == 3171365
-    assert h.ext_id_hash == "MTE3LjM3MDg5MA"
-    assert h.title == "Akademika Postovskogo 17a"
-    assert h.short_address == "ул. Постовского, 17а"
-    assert h.full_address == "Свердловская обл., Екатеринбург, ул. Постовского, 17а"
-    assert h.lat == 56.790699
-    assert h.lon == 60.580191
-
-
-def test_house_expand_params() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    h = e.house
-
-    assert h.year_built == 2019
-    assert h.total_floors == 25
-    assert h.house_type == "monolith"
-    assert h.passenger_elevators == 2
-    assert h.has_concierge is True
-    assert h.closed_yard is True
-    assert h.has_playground is True
-
-
-def test_house_developer_and_rating() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    h = e.house
-
-    assert h.developer_name == "Атомстройкомплекс"
-    assert h.developer_key == "atomstroy"
-    assert h.rating_score == 4.666
-    assert h.rating_string == "4,7"
-    assert h.reviews_count == 6
-    assert len(h.rating_distribution) == 2
-
-
-def test_house_map_and_infra() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    h = e.house
-
-    assert h.infrastructure_walk_distance == "в 5 минутах ходьбы"
-    assert h.infrastructure_summary == "школа, парк, аптека"
-    assert len(h.map_pins) == 1
-    assert h.map_pins[0]["type"] == "school"
-
-
-def test_house_raw_characteristics() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    # raw_characteristics = полный expandParams items (1 категория)
-    assert len(e.house.raw_characteristics) == 1
-    assert e.house.raw_characteristics[0]["title"] == "Параметры дома"
-
-
-def test_reviews() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-
-    assert len(e.reviews) == 1
-    assert e.reviews_next_page_url == "/reviews?page=2"
-
-    r = e.reviews[0]
-    assert isinstance(r, HouseReview)
-    assert r.ext_review_id == 12345
-    assert r.author_name == "Иван"
-    assert r.review_title == "Хорошо"
-    assert r.score == 5
-    assert r.model_experience == "Живу в своей квартире"
-    assert r.rated_date == date(2025, 10, 9)
-    assert r.text_main == "Общий отзыв"
-    assert r.text_pros == "Парк рядом"
-    assert r.text_cons == "Парковки мало"
-    assert r.raw_payload is not None
-
-
-def test_mini_serp() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-
-    assert len(e.mini_serp) == 1
-    m = e.mini_serp[0]
-    assert isinstance(m, MiniSerpListing)
-    assert m.ext_item_id == 7986882804
-    assert m.price_rub == 11990000
-    assert m.metro_text == "Чкаловская, 2,7 км"
-    assert m.metro_color == "#cf2734"
-    assert m.description == "Описание"
-
-
-def test_seller_extraction() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    m = e.mini_serp[0]
-
-    assert m.seller is not None
-    s = m.seller
-    assert isinstance(s, SellerInfo)
-    assert s.ext_seller_id == "domrf66"
-    assert s.name == "DOMRF66"
-    assert s.seller_type == "Компания"
-    assert s.on_platform_since == "На Авито с июля 2012"
-    assert s.profile_url == "/brands/domrf66"
-
-
-def test_placement_history() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-
-    assert len(e.placement_history) == 1
-    p = e.placement_history[0]
-    assert isinstance(p, PlacementHistoryItem)
-    assert p.ext_item_id == "7000000001"
-    assert p.title == "1-к. квартира"
-    assert p.start_price == 5000000
-    assert p.last_price == 4900000
-    assert p.exposure_days == 210
-    # removed_date ВСЕГДА None — виджет не отдаёт!
-    assert p.removed_date is None
-    # raw_payload не содержит itemImage (скипается)
-    assert p.raw_payload is not None
-    assert "itemImage" not in p.raw_payload
-    # start_price_date и last_price_date — из unix timestamp
-    assert isinstance(p.start_price_date, date)
-    assert isinstance(p.last_price_date, date)
-
-
-def test_recommendations() -> None:
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-
-    assert len(e.recommendations) == 1
-    rec = e.recommendations[0]
-    assert isinstance(rec, RecommendationItem)
-    assert rec.title == "Похожая"
-    assert rec.price_text == "от 6 млн"
-    assert rec.address == "Адрес 1"
-    assert rec.url == "/rec1"
-    assert rec.image_url == "https://avito.ru/img1.jpg"
-
-
-def test_missing_optional_widgets() -> None:
-    """Парсер не падает если опциональные виджеты отсутствуют."""
-    state_minimal: dict = {
-        "data": {
-            "data": {
-                "page": {
-                    "placeholders": [
-                        {
-                            "type": "housePage",
-                            "props": {
-                                "developmentData": {
-                                    "avitoId": 9999,
-                                    "coords": {"lat": 56.0, "lng": 60.0},
-                                    "aboutDevelopment": {"expandParams": {"items": []}},
-                                    "developer": {},
-                                    "mapPreview": {},
-                                    "ratingBadge": {"info": {}},
-                                    "ratingSummaryStat": {},
-                                }
-                            },
-                        },
-                    ]
-                }
-            }
-        }
-    }
-    e = parse_houses_state(state_minimal, "/catalog/houses/test/9999")
-    assert e.house.ext_id == 9999
-    assert e.reviews == []
-    assert e.mini_serp == []
-    assert e.placement_history == []
-    assert e.recommendations == []
-    assert e.reviews_next_page_url is None
-
-
-def test_invalid_state_raises() -> None:
-    """ValueError если placeholders не найден."""
-    with pytest.raises(ValueError, match="placeholders"):
-        parse_houses_state({}, "/catalog/houses/test/1")
-
-
-def test_missing_house_page_raises() -> None:
-    """ValueError если housePage виджет отсутствует (вторичка — /catalog/houses/)."""
-    state: dict = {
-        "data": {
-            "data": {
-                "page": {
-                    "placeholders": [
-                        {"type": "breadcrumbs"},
-                    ]
-                }
-            }
-        }
-    }
-    with pytest.raises(ValueError, match="housePage"):
-        parse_houses_state(state, "/catalog/houses/test/1")
-
-
-def test_novostroyka_minimal_development_page() -> None:
-    """Новостройка: виджет 'developmentPage' парсится в partial HouseInfo без ValueError."""
-    state: dict = {
-        "data": {
-            "data": {
-                "page": {
-                    "placeholders": [
-                        {"type": "reNavigation"},
-                        {
-                            "type": "developmentPage",
-                            "props": {
-                                "developmentPage": {
-                                    "developmentData": {
-                                        "avitoId": 12879380,
-                                        "id": "MGDKHN70PQ",
-                                        "title": "ЖК Береговой",
-                                        "address": "ул. Щербакова, жилые дома",
-                                        "fullAddress": "",
-                                        "coords": {"lat": 56.77005, "lng": 60.67244},
-                                        "developer": {"name": "Талан", "key": "talan-1"},
-                                        "mapPreview": {
-                                            "distance": "в 15 минутах ходьбы",
-                                            "objects": "школа, детский сад",
-                                            "pins": [{"lat": 56.77, "lng": 60.67}],
-                                        },
-                                        "aboutDevelopment": {
-                                            "tabs": [
-                                                {
-                                                    "title": "Параметры",
-                                                    "expandParams": {
-                                                        "items": [
-                                                            {
-                                                                "params": [
-                                                                    {
-                                                                        "type": "Тип дома",
-                                                                        "value": "Монолит",
-                                                                    },
-                                                                    {
-                                                                        "type": "Этажность",
-                                                                        "value": "от 4 до 24",
-                                                                    },
-                                                                    {
-                                                                        "type": (
-                                                                            "Пассажирских лифтов"
-                                                                        ),
-                                                                        "value": "42",
-                                                                    },
-                                                                    {
-                                                                        "type": "Грузовых лифтов",
-                                                                        "value": "60",
-                                                                    },
-                                                                ]
-                                                            }
-                                                        ]
-                                                    },
-                                                }
-                                            ]
-                                        },
-                                    }
-                                }
-                            },
-                        },
-                        {"type": "uxFeedback"},
-                    ]
-                }
-            }
-        }
-    }
-    url = "/catalog/novostroyki/ekaterinburg/zhk_beregovoy/12879380"
-    e = parse_houses_state(state, url)
-
-    assert isinstance(e, HouseCatalogEnrichment)
-    assert e.house_url == url
-    h = e.house
-    assert h.ext_id == 12879380
-    assert h.ext_id_hash == "MGDKHN70PQ"
-    assert h.title == "ЖК Береговой"
-    assert h.short_address == "ул. Щербакова, жилые дома"
-    assert h.lat == 56.77005
-    assert h.lon == 60.67244
-    assert h.house_type == "monolith"  # "Монолит" → monolith
-    assert h.passenger_elevators == 42
-    assert h.cargo_elevators == 60
-    # Этажность = диапазон "от 4 до 24" → int-каст None (только в raw_characteristics)
-    assert h.total_floors is None
-    assert h.year_built is None  # дом строится, года постройки нет
-    assert h.developer_name == "Талан"
-    assert h.developer_key == "talan-1"
-    assert h.infrastructure_summary == "школа, детский сад"
-    assert h.infrastructure_walk_distance == "в 15 минутах ходьбы"
-    assert len(h.map_pins) == 1
-    # Нет рейтинга/отзывов/истории у новостройки
-    assert h.rating_score is None
-    assert e.reviews == []
-    assert e.placement_history == []
-    assert e.recommendations == []
-
-
-def test_novostroyka_no_development_page_graceful_skip() -> None:
-    """Новостройка без developmentPage/housePage → graceful skip (НЕ ValueError)."""
-    state: dict = {
-        "data": {
-            "data": {
-                "page": {
-                    "placeholders": [
-                        {"type": "reNavigation"},
-                        {"type": "uxFeedback"},
-                    ]
-                }
-            }
-        }
-    }
-    url = "/catalog/novostroyki/ekaterinburg/zhk_unknown/55667788"
-    e = parse_houses_state(state, url)
-
-    assert isinstance(e, HouseCatalogEnrichment)
-    assert e.house_url == url
-    # ext_id извлечён из хвоста URL
-    assert e.house.ext_id == 55667788
-    assert e.reviews == []
-    assert e.placement_history == []
-
-
-def test_novostroyka_real_fixture_end_to_end() -> None:
-    """Реальная фикстура новостройки: parse не кидает ValueError, отдаёт partial-обогащение."""
-    html = (FIXTURES_DIR / "avito_novostroyka_12879380.html").read_text(encoding="utf-8")
-    state = _extract_preloaded_state(html)
-    assert state is not None, "URL-encoded __preloadedState__ должен распарситься"
-
-    url = "/catalog/novostroyki/ekaterinburg/semeynyy_kvartal_beregovoy/12879380"
-    e = parse_houses_state(state, url)
-
-    assert isinstance(e, HouseCatalogEnrichment)
-    h = e.house
-    assert h.ext_id == 12879380
-    assert h.ext_id_hash  # base64 internal id
-    assert h.house_type == "monolith"
-    assert h.passenger_elevators is not None and h.passenger_elevators > 0
-    assert h.lat is not None and h.lon is not None
-    assert h.developer_name  # застройщик присутствует
-    # Новостройка: нет отзывов/истории/года постройки
-    assert h.year_built is None
-    assert e.reviews == []
-    assert e.placement_history == []
-
-
-def test_house_type_normalization() -> None:
-    """Все варианты типа дома нормализуются корректно."""
-    from app.services.scrapers.avito_houses import _normalize_house_type
-
-    assert _normalize_house_type("Монолитный") == "monolith"
-    assert _normalize_house_type("Панельный") == "panel"
-    assert _normalize_house_type("Кирпичный") == "brick"
-    assert _normalize_house_type("Монолитно-кирпичный") == "monolith_brick"
-    assert _normalize_house_type("Блочный") == "block"
-    assert _normalize_house_type("Деревянный") == "wood"
-    assert _normalize_house_type("Неизвестный") == "other"
-    assert _normalize_house_type(None) is None
-
-
-def test_house_class_normalization() -> None:
-    """Все варианты класса жилья нормализуются корректно (#2003)."""
-    from app.services.scrapers.avito_houses import _normalize_house_class
-
-    assert _normalize_house_class("Комфорт") == "comfort"
-    assert _normalize_house_class("Бизнес") == "business"
-    assert _normalize_house_class("Премиум") == "premium"
-    assert _normalize_house_class("Эконом") == "economy"
-    assert _normalize_house_class("Комфорт+") == "comfort_plus"
-    assert _normalize_house_class("Элит") == "elite"
-    # Регистронезависимость + пробелы по краям
-    assert _normalize_house_class("  комфорт  ") == "comfort"
-    # Неизвестное значение → None (НЕ "other" — house_class точный ANY(...) фильтр)
-    assert _normalize_house_class("Неизвестный класс") is None
-    assert _normalize_house_class(None) is None
-    assert _normalize_house_class("") is None
-
-
-def test_house_class_parsed_from_novostroyka_expand_params() -> None:
-    """ "Класс жилья" из developmentPage expandParams → HouseInfo.house_class (#2003)."""
-    state: dict = {
-        "data": {
-            "data": {
-                "page": {
-                    "placeholders": [
-                        {
-                            "type": "developmentPage",
-                            "props": {
-                                "developmentPage": {
-                                    "developmentData": {
-                                        "avitoId": 42,
-                                        "coords": {"lat": 56.8, "lng": 60.6},
-                                        "developer": {},
-                                        "mapPreview": {},
-                                        "aboutDevelopment": {
-                                            "tabs": [
-                                                {
-                                                    "expandParams": {
-                                                        "items": [
-                                                            {
-                                                                "params": [
-                                                                    {
-                                                                        "type": "Класс жилья",
-                                                                        "value": "Комфорт",
-                                                                    },
-                                                                ]
-                                                            }
-                                                        ]
-                                                    }
-                                                }
-                                            ]
-                                        },
-                                    }
-                                }
-                            },
-                        },
-                    ]
-                }
-            }
-        }
-    }
-    e = parse_houses_state(state, "/catalog/novostroyki/ekaterinburg/test/42")
-    assert e.house.house_class == "comfort"
-
-
-def test_house_class_absent_on_secondary_listing() -> None:
-    """Вторичка (housePage) не отдаёт "Класс жилья" → house_class остаётся None."""
-    e = parse_houses_state(MINIMAL_STATE, "/catalog/houses/ekaterinburg/test/3171365")
-    assert e.house.house_class is None
-
-
-def test_strip_price() -> None:
-    """_strip_price корректно извлекает числа из строк Avito."""
-    from app.services.scrapers.avito_houses import _strip_price
-
-    assert _strip_price("11 990 000 ₽") == 11990000
-    assert _strip_price("5 000 000 ₽") == 5000000
-    assert _strip_price(None) is None
-    assert _strip_price("") is None
-
-
-def test_mini_serp_no_seller() -> None:
-    """Объявление без продавца не вызывает ошибку."""
-    state: dict = {
-        "data": {
-            "data": {
-                "page": {
-                    "placeholders": [
-                        {
-                            "type": "housePage",
-                            "props": {
-                                "developmentData": {
-                                    "avitoId": 111,
-                                    "coords": {"lat": 56.0, "lng": 60.0},
-                                    "aboutDevelopment": {"expandParams": {"items": []}},
-                                    "developer": {},
-                                    "mapPreview": {},
-                                    "ratingBadge": {"info": {}},
-                                    "ratingSummaryStat": {},
-                                }
-                            },
-                        },
-                        {
-                            "type": "miniSerp",
-                            "props": {
-                                "items": [
-                                    {
-                                        "id": 12345,
-                                        "title": "1-к.",
-                                        "url": "/test",
-                                        "price": "3 000 000 ₽",
-                                        "geo": {"content": "Центр", "colors": []},
-                                    },
-                                ]
-                            },
-                        },
-                    ]
-                }
-            }
-        }
-    }
-    e = parse_houses_state(state, "/catalog/houses/test/111")
-    assert len(e.mini_serp) == 1
-    assert e.mini_serp[0].seller is None
-    assert e.mini_serp[0].price_rub == 3000000
-
-
-# ---------------------------------------------------------------------------
-# Real-HTML fixture (regression for issue #566 — state-blob encoding/shape change)
-# ---------------------------------------------------------------------------
-#
-# Fixture: реальная страница /catalog/houses/ekaterinburg/ul_yumasheva_6/115188,
-# захвачена curl_cffi chrome120 (2026-05-28). Avito сменил формат __preloadedState__
-# на URL-encoded литерал + обернул виджеты в placeholders[0]['widgets'] +
-# вложил developmentData под props['housePage']. Этот тест ловит регрессию декода/shape.
-
-
-def test_extract_preloaded_state_from_real_html() -> None:
-    """_extract_preloaded_state декодирует текущий URL-encoded __preloadedState__."""
-    html = (FIXTURES_DIR / "avito_houses_sample.html").read_text(encoding="utf-8")
-
-    state = _extract_preloaded_state(html)
-
-    assert state is not None, "URL-encoded __preloadedState__ должен распарситься"
-    # Avito SSR обёртка — успешный ответ
-    assert state["data"]["status"] == 200
-    # Inner path до placeholders сохранился
-    placeholders = state["data"]["data"]["page"]["placeholders"]
-    assert isinstance(placeholders, list) and placeholders, "placeholders непустой массив"
-
-
-def test_parse_real_fixture_end_to_end() -> None:
-    """Фикстура HTML → _extract_preloaded_state → parse_houses_state → HouseInfo.ext_id."""
-    html = (FIXTURES_DIR / "avito_houses_sample.html").read_text(encoding="utf-8")
-    state = _extract_preloaded_state(html)
-    assert state is not None
-
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/ul_yumasheva_6/115188")
-
-    assert isinstance(e, HouseCatalogEnrichment)
-    # housePage распарсился — ext_id (avitoId) присутствует и совпадает с URL
-    assert e.house.ext_id == 115188
-    assert e.house.ext_id_hash  # base64 internal id присутствует
-    # expandParams распарсились (год постройки реальной фикстуры — 2016)
-    assert e.house.year_built == 2016
-    assert e.house.total_floors == 17
-    assert e.house.house_type == "monolith"
-    assert e.house.lat is not None and e.house.lon is not None
-
-
-def test_extract_preloaded_state_jsparse_fallback() -> None:
-    """Старая форма JSON.parse("...") всё ещё декодируется (fallback-ветка)."""
-    inner = '{\\"data\\":{\\"data\\":{\\"page\\":{\\"placeholders\\":[]}}}}'
-    html = f''
-
-    state = _extract_preloaded_state(html)
-
-    assert state is not None
-    assert state["data"]["data"]["page"]["placeholders"] == []
-
-
-def test_extract_preloaded_state_missing_returns_none() -> None:
-    """Нет маркера → None (а не исключение)."""
-    assert _extract_preloaded_state("no state here") is None
-
-
-# ---------------------------------------------------------------------------
-# Real-HTML fixture (regression for issue #1789 — MFE props nesting)
-# ---------------------------------------------------------------------------
-#
-# Fixture: реальная страница /catalog/houses/ekaterinburg/ul_soni_morozovoy_190/572538,
-# захвачена браузер-сервисом (2026-06-19). Avito перенёс данные виджетов на уровень
-# глубже — под ключ с именем виджета:
-#   reviews:               props['reviews']['entries']   (было props['entries'])
-#   miniSerp:              props['miniSerp']['items']    (было props['items'])
-#   housePlacementHistory: props['housePlacementHistory']['items'] (было props['items'])
-#   recommendations:       props['recommendations']['items']       (было props['items'])
-# Из-за этого reviews/placement/seller-обогащение НЕ долетало в БД (0 строк).
-# Этот тест ловит регрессию шейпа props.
-
-
-def test_parse_mfe_fixture_widgets_reach_db() -> None:
-    """Фикстура MFE: reviews/placement/miniSerp/house_type/year_built распарсены непусто."""
-    html = (FIXTURES_DIR / "avito_house_572538.html").read_text(encoding="utf-8")
-    state = _extract_preloaded_state(html)
-    assert state is not None, "URL-encoded __preloadedState__ должен распарситься"
-
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/ul_soni_morozovoy_190/572538")
-
-    # housePage — тип дома и год постройки долетают
-    assert e.house.ext_id == 572538
-    assert e.house.house_type is not None, "house_type должен парситься из expandParams"
-    assert e.house.house_type == "brick"
-    assert e.house.year_built == 1994
-    assert e.house.reviews_count == 3
-
-    # reviews — >=1 отзыв с текстом, автором, оценкой
-    assert len(e.reviews) >= 1, "reviews должны распарситься из props['reviews']['entries']"
-    r = e.reviews[0]
-    assert r.ext_review_id, "у отзыва должен быть id"
-    assert r.author_name, "у отзыва должен быть автор"
-    assert r.score is not None, "у отзыва должна быть оценка"
-    assert r.text_main, "у отзыва должен быть основной текст (textSections)"
-
-    # placement — >=1 item из props['housePlacementHistory']['items']
-    assert len(e.placement_history) >= 1, "placement должен распарситься"
-    p = e.placement_history[0]
-    assert p.ext_item_id, "у placement item должен быть id"
-    assert p.start_price is not None
-    assert p.last_price is not None
-    assert isinstance(p.last_price_date, date)
-
-    # miniSerp — >=1 объявление из props['miniSerp']['items']
-    assert len(e.mini_serp) >= 1, "miniSerp должен распарситься"
-    m = e.mini_serp[0]
-    assert m.ext_item_id
-    assert m.price_rub is not None
diff --git a/tradein-mvp/backend/tests/test_backfill_wave2.py b/tradein-mvp/backend/tests/test_backfill_wave2.py
index 158aeb8b..8aaefd7b 100644
--- a/tradein-mvp/backend/tests/test_backfill_wave2.py
+++ b/tradein-mvp/backend/tests/test_backfill_wave2.py
@@ -234,9 +234,10 @@ async def test_backfill_yandex_addresses_error_on_non_200():
 @pytest.mark.asyncio
 async def test_cian_price_history_saves_changes():
     """Happy path: fetch returns DetailEnrichment with price_changes → saved."""
+    from scraper_kit.providers.cian.detail import DetailEnrichment
+
     from app.services.cian_price_history import backfill_cian_price_history
     from app.services.scraper_adapters import RealScraperConfig
-    from app.services.scrapers.cian_detail import DetailEnrichment
 
     enrichment = DetailEnrichment(
         cian_id=123456,
@@ -290,9 +291,10 @@ async def test_cian_price_history_saves_changes():
 @pytest.mark.asyncio
 async def test_cian_price_history_skips_no_changes():
     """Fetch succeeds but no price_changes → skipped (not error)."""
+    from scraper_kit.providers.cian.detail import DetailEnrichment
+
     from app.services.cian_price_history import backfill_cian_price_history
     from app.services.scraper_adapters import RealScraperConfig
-    from app.services.scrapers.cian_detail import DetailEnrichment
 
     enrichment = DetailEnrichment(cian_id=111, price_changes=[])
 
diff --git a/tradein-mvp/backend/tests/test_cian_detail.py b/tradein-mvp/backend/tests/test_cian_detail.py
deleted file mode 100644
index e23323f5..00000000
--- a/tradein-mvp/backend/tests/test_cian_detail.py
+++ /dev/null
@@ -1,830 +0,0 @@
-"""Tests for cian_detail.py — Stage 5 of CianScraper v1."""
-
-from __future__ import annotations
-
-from pathlib import Path
-from unittest.mock import AsyncMock, MagicMock
-
-import pytest
-
-from app.services.scrapers.cian_detail import (
-    DetailEnrichment,
-    _extract_agent_profile,
-    _extract_ceiling_height,
-    _extract_price_changes,
-    _extract_repair_state,
-    _extract_repair_type,
-    _parse_views,
-    fetch_detail,
-    save_detail_enrichment,
-)
-
-_FIXTURE_DIR = Path(__file__).parent / "fixtures"
-
-# ── _parse_views ──────────────────────────────────────────────────────────────
-
-
-def test_parse_views_space_formatted():
-    assert _parse_views("1 234") == 1234
-
-
-def test_parse_views_plain():
-    assert _parse_views("567") == 567
-
-
-def test_parse_views_none():
-    assert _parse_views(None) is None
-
-
-def test_parse_views_invalid():
-    assert _parse_views("invalid") is None
-
-
-def test_parse_views_comma_formatted():
-    assert _parse_views("1,234") == 1234
-
-
-# ── _extract_repair_state — структурное поле в приоритете, fallback на текст (#622) ──
-
-
-def test_repair_state_structured_wins_over_description():
-    """repairType присутствует → используется он, описание игнорируется."""
-    offer = {
-        "repairType": "cosmetic",  # → standard
-        "description": "Шикарный дизайнерский ремонт",  # инференс дал бы excellent
-    }
-    assert _extract_repair_state(offer) == "standard"
-
-
-def test_repair_state_falls_back_to_description_when_structured_missing():
-    """repairType отсутствует → инференс из описания."""
-    offer = {
-        "repairType": None,
-        "description": "Сделан свежий евроремонт, заезжай и живи",
-    }
-    assert _extract_repair_state(offer) == "good"
-
-
-def test_repair_state_none_when_both_missing():
-    """Нет structured и нет сигнала в тексте → None (не фабрикуем)."""
-    offer = {"description": "Светлая квартира рядом с парком"}
-    assert _extract_repair_state(offer) is None
-
-
-# ── _extract_ceiling_height — offer.building.ceilingHeight, не offer.ceilingHeight (#1791) ──
-
-
-def test_ceiling_height_from_building_node():
-    """На detail-странице высота лежит в offer.building.ceilingHeight."""
-    offer = {"ceilingHeight": None, "building": {"ceilingHeight": "3.0"}}
-    assert _extract_ceiling_height(offer) == 3.0
-
-
-def test_ceiling_height_falls_back_to_top_level():
-    """Если building.ceilingHeight отсутствует — fallback на offer.ceilingHeight."""
-    offer = {"ceilingHeight": "2.75", "building": {}}
-    assert _extract_ceiling_height(offer) == 2.75
-
-
-def test_ceiling_height_none_when_absent_everywhere():
-    assert _extract_ceiling_height({"building": {}}) is None
-    assert _extract_ceiling_height({}) is None
-
-
-# ── _extract_repair_type / state — offer.decoration fallback (#1791) ──────────
-
-
-def test_repair_type_prefers_repair_type_then_decoration():
-    assert _extract_repair_type({"repairType": "euro", "decoration": "cosmetic"}) == "euro"
-    assert _extract_repair_type({"repairType": None, "decoration": "preFine"}) == "preFine"
-    assert _extract_repair_type({}) is None
-
-
-def test_repair_state_from_decoration_when_repair_type_missing():
-    """offer.decoration='preFine' → needs_repair (через нормализатор)."""
-    offer = {"repairType": None, "decoration": "preFine", "description": ""}
-    assert _extract_repair_state(offer) == "needs_repair"
-
-
-# ── _extract_price_changes ────────────────────────────────────────────────────
-
-
-def test_extract_price_changes_normalizes():
-    """curl_cffi flat format: {changeTime, price, diffPercent}."""
-    offer = {
-        "priceChanges": [
-            {"changeTime": "2026-04-01T00:00:00Z", "price": 5000000, "diffPercent": -2.5},
-            {"changeTime": "2026-04-10T00:00:00Z", "price": 4900000, "diffPercent": -2.0},
-        ]
-    }
-    result = _extract_price_changes(offer, {}, {})
-    assert len(result) == 2
-    assert result[0]["change_time"] == "2026-04-01T00:00:00Z"
-    assert result[0]["price_rub"] == 5000000
-    assert result[0]["diff_percent"] == -2.5
-
-
-def test_extract_price_changes_handles_empty_offer():
-    assert _extract_price_changes({}, {}, {}) == []
-
-
-def test_extract_price_changes_handles_none_value():
-    assert _extract_price_changes({"priceChanges": None}, {}, {}) == []
-
-
-def test_extract_price_changes_falls_back_to_state():
-    state = {
-        "priceChanges": [
-            {"changeTime": "2026-03-01T00:00:00Z", "price": 6000000, "diffPercent": 0},
-        ]
-    }
-    result = _extract_price_changes({}, {}, state)
-    assert len(result) == 1
-    assert result[0]["price_rub"] == 6000000
-
-
-def test_extract_price_changes_skips_non_dict_entries():
-    offer = {"priceChanges": [{"changeTime": "2026-04-01T00:00:00Z", "price": 5000000}, "bad"]}
-    result = _extract_price_changes(offer, {}, {})
-    assert len(result) == 1
-
-
-def test_extract_price_changes_offer_data_browser_format():
-    """Browser-rendered HTML: offerData.priceChanges with nested priceData.price."""
-    offer_data = {
-        "priceChanges": [
-            {
-                "changeTime": "2026-06-05T15:32:41Z",
-                "priceData": {"currency": "rur", "price": 13300000},
-            },
-            {
-                "changeTime": "2026-06-03T08:05:04Z",
-                "priceData": {"currency": "rur", "price": 13600000},
-            },
-        ]
-    }
-    result = _extract_price_changes({}, offer_data, {})
-    assert len(result) == 2
-    assert result[0]["change_time"] == "2026-06-05T15:32:41Z"
-    assert result[0]["price_rub"] == 13300000
-    assert result[1]["price_rub"] == 13600000
-
-
-def test_extract_price_changes_offer_wins_over_offer_data():
-    """offer.priceChanges takes priority over offerData.priceChanges (flat format wins)."""
-    offer = {
-        "priceChanges": [
-            {"changeTime": "2026-04-01T00:00:00Z", "price": 5000000},
-        ]
-    }
-    offer_data = {
-        "priceChanges": [
-            {"changeTime": "2026-06-01T00:00:00Z", "priceData": {"price": 9000000}},
-        ]
-    }
-    result = _extract_price_changes(offer, offer_data, {})
-    assert len(result) == 1
-    assert result[0]["price_rub"] == 5000000
-
-
-# ── _extract_agent_profile ────────────────────────────────────────────────────
-
-
-def test_extract_agent_profile_from_agent_key():
-    offer = {
-        "agent": {
-            "id": 12345,
-            "companyName": "Новосёл",
-            "isPro": True,
-            "skills": [{"id": 1, "name": "Продажа"}],
-            "accountType": "specialist",
-        }
-    }
-    profile = _extract_agent_profile(offer)
-    assert profile is not None
-    assert profile["ext_agent_id"] == 12345
-    assert profile["company_name"] == "Новосёл"
-    assert profile["is_pro"] is True
-    assert profile["account_type"] == "specialist"
-
-
-def test_extract_agent_profile_from_user_key():
-    offer = {
-        "user": {
-            "cianUserId": 99999,
-            "companyName": "Агентство Икс",
-            "isPro": False,
-            "accountType": "agency",
-        }
-    }
-    profile = _extract_agent_profile(offer)
-    assert profile is not None
-    assert profile["ext_agent_id"] == 99999
-    assert profile["company_name"] == "Агентство Икс"
-
-
-def test_extract_agent_profile_returns_none_when_no_agent():
-    assert _extract_agent_profile({}) is None
-    assert _extract_agent_profile({"agent": {}, "user": {}}) is None
-
-
-# ── DetailEnrichment defaults ─────────────────────────────────────────────────
-
-
-def test_dataclass_defaults():
-    enrich = DetailEnrichment()
-    assert enrich.cian_id is None
-    assert enrich.price_changes == []
-    assert enrich.raw_sister_states == {}
-    assert enrich.bti_data is None
-    assert enrich.agent_profile is None
-    assert enrich.newbuilding_data is None
-
-
-# ── fetch_detail integration (mocked) ────────────────────────────────────────
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_returns_none_on_no_state(monkeypatch):
-    """If state extraction returns None → fetch_detail returns None gracefully."""
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: None,
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/123/", session=session)
-    assert result is None
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_returns_none_on_non_200(monkeypatch):
-    """Non-200 HTTP response → returns None, no exception raised."""
-    response = MagicMock()
-    response.status_code = 403
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/456/", session=session)
-    assert result is None
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_extracts_core_fields(monkeypatch):
-    """Verify core fields are extracted correctly from mocked state."""
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 327830237,
-                "windowsViewType": "yardAndStreet",
-                "separateWcsCount": 1,
-                "combinedWcsCount": 0,
-                "ceilingHeight": 2.65,
-                "repairType": "cosmetic",
-                "priceChanges": [
-                    {
-                        "changeTime": "2026-04-01T00:00:00Z",
-                        "price": 5500000,
-                        "diffPercent": -1.8,
-                    }
-                ],
-                "agent": {
-                    "id": 7654321,
-                    "companyName": "Первичка Плюс",
-                    "isPro": True,
-                    "skills": [],
-                    "accountType": "specialist",
-                },
-            },
-            "stats": {
-                "totalViewsFormattedString": "1 042",
-                "todayViewsFormattedString": "7",
-            },
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/327830237/", session=session)
-    assert result is not None
-    assert result.cian_id == 327830237
-    assert result.windows_view_type == "yardAndStreet"
-    assert result.separate_wcs_count == 1
-    assert result.combined_wcs_count == 0
-    assert result.ceiling_height == 2.65
-    assert result.repair_type == "cosmetic"
-    assert result.views_total == 1042
-    assert result.views_today == 7
-    assert len(result.price_changes) == 1
-    assert result.price_changes[0]["price_rub"] == 5500000
-    assert result.agent_profile is not None
-    assert result.agent_profile["ext_agent_id"] == 7654321
-    assert result.agent_profile["company_name"] == "Первичка Плюс"
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_extracts_bti_from_sister_states(monkeypatch):
-    """BTI sister state is captured into bti_data."""
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 111111,
-            }
-        }
-    }
-    fake_all_states = {
-        "frontend-offer-card": {
-            "bti": {
-                "houseData": {
-                    "seriesName": "П-44Т",
-                    "floorsCount": 17,
-                    "flatCount": 120,
-                }
-            }
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: fake_all_states,
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/111111/", session=session)
-    assert result is not None
-    assert result.bti_data is not None
-    assert result.bti_data["seriesName"] == "П-44Т"
-    assert "bti" in result.raw_sister_states
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_extracts_newbuilding_link(monkeypatch):
-    """Newbuilding reference is captured from offer.newbuilding."""
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 222222,
-                "newbuilding": {
-                    "id": 54016,
-                    "name": "Екатерининский парк",
-                    "url": "https://zhk-ekaterininskiy-park-ekb-i.cian.ru/",
-                },
-            }
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/222222/", session=session)
-    assert result is not None
-    assert result.newbuilding_data is not None
-    assert result.newbuilding_data["id"] == 54016
-
-
-# ── kitchen_area_m2 extraction ────────────────────────────────────────────────
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_extracts_kitchen_area(monkeypatch):
-    """kitchenArea из offer → kitchen_area_m2 в DetailEnrichment."""
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 333333,
-                "kitchenArea": 9.5,
-            }
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/333333/", session=session)
-    assert result is not None
-    assert result.kitchen_area_m2 == 9.5
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_kitchen_area_none_when_missing(monkeypatch):
-    """Если kitchenArea отсутствует в offer — kitchen_area_m2 остаётся None."""
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 444444,
-            }
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/444444/", session=session)
-    assert result is not None
-    assert result.kitchen_area_m2 is None
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_extracts_description(monkeypatch):
-    """offer.description → DetailEnrichment.description (для persist в listings)."""
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 555555,
-                "description": "Продаётся уютная квартира с видом на парк.",
-            }
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = ""
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://ekb.cian.ru/sale/flat/555555/", session=session)
-    assert result is not None
-    assert result.description == "Продаётся уютная квартира с видом на парк."
-
-
-# ── save_detail_enrichment — kitchen_area_m2 persisted ───────────────────────
-
-
-def test_save_detail_enrichment_writes_kitchen_area():
-    """save_detail_enrichment передаёт kitchen_area_m2 в UPDATE listings."""
-    db = MagicMock()
-
-    # Мокаем fetchone для snapshot-запроса — price_rub=None → snapshot не пишется
-    db.execute.return_value.fetchone.return_value = None
-
-    enrichment = DetailEnrichment(
-        windows_view_type=None,
-        separate_wcs_count=None,
-        combined_wcs_count=None,
-        ceiling_height=None,
-        repair_type=None,
-        repair_state=None,
-        kitchen_area_m2=8.3,
-        views_total=None,
-        views_today=None,
-    )
-
-    save_detail_enrichment(db, listing_id=999, enrichment=enrichment)
-
-    # Проверяем что UPDATE listings вызван с параметром ka=8.3
-    update_call_params = db.execute.call_args_list[0][0][1]
-    assert update_call_params["ka"] == 8.3
-
-
-def test_save_detail_enrichment_kitchen_area_none_passthrough():
-    """Если kitchen_area_m2 = None — параметр ka=None, COALESCE сохраняет существующее."""
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-
-    enrichment = DetailEnrichment(kitchen_area_m2=None)
-
-    save_detail_enrichment(db, listing_id=1001, enrichment=enrichment)
-
-    update_call_params = db.execute.call_args_list[0][0][1]
-    assert update_call_params["ka"] is None
-
-
-# ── save_detail_enrichment — description + detail_enriched_at stamp ───────────
-
-
-def test_save_detail_enrichment_writes_description():
-    """save_detail_enrichment передаёт description в UPDATE listings."""
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-
-    enrichment = DetailEnrichment(description="Светлая квартира рядом с парком")
-
-    save_detail_enrichment(db, listing_id=1002, enrichment=enrichment)
-
-    update_call_params = db.execute.call_args_list[0][0][1]
-    assert update_call_params["descr"] == "Светлая квартира рядом с парком"
-
-
-def test_save_detail_enrichment_stamps_detail_enriched_at():
-    """UPDATE listings содержит detail_enriched_at = NOW() — иначе строка
-    переобогащается каждый run (detail_enriched_at IS NULL filter)."""
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-
-    enrichment = DetailEnrichment(kitchen_area_m2=8.3)
-
-    save_detail_enrichment(db, listing_id=1003, enrichment=enrichment)
-
-    update_sql = str(db.execute.call_args_list[0][0][0])
-    assert "detail_enriched_at = NOW()" in update_sql
-
-
-# ── save_detail_enrichment — diff_percent клампится перед INSERT (#2277) ──────
-# offer_price_history.diff_percent — NUMERIC(8,2) (миграция 131). Этот прямой
-# INSERT не проходит через БД-триггер record_listing_price_change() (тот
-# срабатывает только на UPDATE listings) — garbage/экстремальный diff_percent
-# из скрапленного Cian JSON иначе переполняет колонку.
-
-
-def _insert_diff_param(db: MagicMock) -> float | None:
-    insert_calls = [c for c in db.execute.call_args_list if "offer_price_history" in str(c[0][0])]
-    assert len(insert_calls) == 1
-    return insert_calls[0][0][1]["diff"]
-
-
-def test_save_detail_enrichment_clamps_extreme_positive_diff():
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-    enrichment = DetailEnrichment(
-        price_changes=[
-            {"change_time": "2026-04-01T00:00:00Z", "price_rub": 5000000, "diff_percent": 5000000}
-        ]
-    )
-    save_detail_enrichment(db, listing_id=2001, enrichment=enrichment)
-    assert _insert_diff_param(db) == 999999.99
-
-
-def test_save_detail_enrichment_clamps_extreme_negative_diff():
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-    enrichment = DetailEnrichment(
-        price_changes=[
-            {
-                "change_time": "2026-04-01T00:00:00Z",
-                "price_rub": 5000000,
-                "diff_percent": -5000000,
-            }
-        ]
-    )
-    save_detail_enrichment(db, listing_id=2002, enrichment=enrichment)
-    assert _insert_diff_param(db) == -999999.99
-
-
-def test_save_detail_enrichment_normal_diff_unchanged():
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-    enrichment = DetailEnrichment(
-        price_changes=[
-            {"change_time": "2026-04-01T00:00:00Z", "price_rub": 5000000, "diff_percent": -2.5}
-        ]
-    )
-    save_detail_enrichment(db, listing_id=2003, enrichment=enrichment)
-    assert _insert_diff_param(db) == -2.5
-
-
-def test_save_detail_enrichment_none_diff_unchanged():
-    db = MagicMock()
-    db.execute.return_value.fetchone.return_value = None
-    enrichment = DetailEnrichment(
-        price_changes=[
-            {"change_time": "2026-04-01T00:00:00Z", "price_rub": 5000000, "diff_percent": None}
-        ]
-    )
-    save_detail_enrichment(db, listing_id=2004, enrichment=enrichment)
-    assert _insert_diff_param(db) is None
-
-
-# ── fetch_detail browser_fetcher param ───────────────────────────────────────
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_browser_fetcher_uses_browser_html(monkeypatch):
-    """browser_fetcher provided → HTML comes from BrowserFetcher.fetch, not curl_cffi.
-
-    Simulates the browser-rendered format where priceChanges is at offerData level
-    with nested priceData.price (verified on prod 2026-06-16).
-    """
-    fake_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 328647045,
-            },
-            # Browser-rendered: priceChanges at offerData level, priceData.price format
-            "priceChanges": [
-                {
-                    "changeTime": "2026-06-05T15:32:41Z",
-                    "priceData": {"currency": "rur", "price": 13300000},
-                },
-                {
-                    "changeTime": "2026-06-03T08:05:04Z",
-                    "priceData": {"currency": "rur", "price": 13600000},
-                },
-            ],
-        }
-    }
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    # Mock BrowserFetcher: fetch() returns HTML tagged so we can verify it was called.
-    mock_bf = MagicMock()
-    mock_bf.fetch = AsyncMock(return_value="browser-rendered")
-
-    result = await fetch_detail(
-        "https://ekb.cian.ru/sale/flat/328647045/",
-        browser_fetcher=mock_bf,
-    )
-
-    assert result is not None
-    assert result.cian_id == 328647045
-    # price_changes populated from browser-rendered HTML via offerData.priceChanges
-    assert len(result.price_changes) == 2
-    assert result.price_changes[0]["price_rub"] == 13300000
-    assert result.price_changes[0]["change_time"] == "2026-06-05T15:32:41Z"
-    assert result.price_changes[1]["price_rub"] == 13600000
-    # BrowserFetcher.fetch was called with the offer URL
-    mock_bf.fetch.assert_called_once_with("https://ekb.cian.ru/sale/flat/328647045/")
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_browser_fetcher_no_curl_session_used(monkeypatch):
-    """When browser_fetcher is given, curl_cffi AsyncSession is NOT instantiated."""
-    fake_state = {"offerData": {"offer": {"cianId": 555555}}}
-
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: fake_state if key == "defaultState" else None,
-    )
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_all_states",
-        lambda html: {},
-    )
-
-    # Patch AsyncSession so we can assert it was NOT called.
-    mock_session_cls = MagicMock()
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.AsyncSession",
-        mock_session_cls,
-    )
-
-    mock_bf = MagicMock()
-    mock_bf.fetch = AsyncMock(return_value="")
-
-    result = await fetch_detail(
-        "https://ekb.cian.ru/sale/flat/555555/",
-        browser_fetcher=mock_bf,
-    )
-
-    assert result is not None
-    mock_session_cls.assert_not_called()
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_browser_fetcher_returns_none_on_exception(monkeypatch):
-    """If BrowserFetcher.fetch raises, fetch_detail returns None (no crash)."""
-    mock_bf = MagicMock()
-    mock_bf.fetch = AsyncMock(side_effect=RuntimeError("browser service unavailable"))
-
-    result = await fetch_detail(
-        "https://ekb.cian.ru/sale/flat/999999/",
-        browser_fetcher=mock_bf,
-    )
-
-    assert result is None
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_browser_fetcher_state_missing_returns_none(monkeypatch):
-    """browser HTML parsed but no defaultState → fetch_detail returns None."""
-    monkeypatch.setattr(
-        "app.services.scrapers.cian_detail.extract_state",
-        lambda html, mfe, key: None,
-    )
-
-    mock_bf = MagicMock()
-    mock_bf.fetch = AsyncMock(return_value="no state here")
-
-    result = await fetch_detail(
-        "https://ekb.cian.ru/sale/flat/777777/",
-        browser_fetcher=mock_bf,
-    )
-
-    assert result is None
-
-
-# ── Real-card fixture (#1791) — снята с www.cian.ru/sale/flat/330982715/ ──────
-# Регрессия на field-drift: ceiling/kitchen/description должны извлекаться из
-# актуальной structure detail-страницы (defaultState, concat-формат).
-
-
-@pytest.mark.asyncio
-async def test_fetch_detail_real_fixture_extracts_drifted_fields():
-    """На реальной карточке ceiling берётся из offer.building.ceilingHeight,
-    kitchen/description — из offer top-level. До #1791 ceiling был 0%."""
-    html = (_FIXTURE_DIR / "cian_flat_330982715.html").read_text(encoding="utf-8")
-
-    response = MagicMock()
-    response.status_code = 200
-    response.text = html
-
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-
-    result = await fetch_detail("https://www.cian.ru/sale/flat/330982715/", session=session)
-
-    assert result is not None
-    # ceiling_height — РАНЬШЕ был None (читали offer.ceilingHeight=null);
-    # теперь из offer.building.ceilingHeight=3.0
-    assert result.ceiling_height == 3.0
-    # kitchen_area_m2 — offer.kitchenArea='14.5' (строка → float)
-    assert result.kitchen_area_m2 == 14.5
-    # description — непустой текст объявления
-    assert result.description is not None
-    assert len(result.description) > 50
-    assert result.cian_id is not None
diff --git a/tradein-mvp/backend/tests/test_cian_serp_scraper.py b/tradein-mvp/backend/tests/test_cian_serp_scraper.py
deleted file mode 100644
index 0bb42868..00000000
--- a/tradein-mvp/backend/tests/test_cian_serp_scraper.py
+++ /dev/null
@@ -1,792 +0,0 @@
-"""Smoke тесты для cian.py SERP scraper (Stage 3 рефактор).
-
-Проверяют парсинг 137 полей из синтетического Redux state без сетевых запросов.
-Fixture формирует минимальный JS push() conforming to _cianConfig pattern.
-"""
-
-import json
-from unittest.mock import AsyncMock, MagicMock, patch
-
-import pytest
-
-from app.services.scrapers.cian import CianScraper, _format_address
-
-# ── Fixture helpers ──────────────────────────────────────────────────────────
-
-
-def _make_cian_serp_html(offers: list[dict]) -> str:
-    """Обернуть offers в валидный _cianConfig['frontend-serp'].push() HTML."""
-    state = {
-        "results": {
-            "totalOffers": len(offers),
-            "offers": offers,
-        }
-    }
-    state_json = json.dumps(state, ensure_ascii=False)
-    return f"""
-
-"""
-
-
-def _make_full_offer(
-    cian_id: int = 328739308,
-    price_rur: int = 6500000,
-    rooms: int = 2,
-    total_area: str = "52.0",
-    living_area: str = "35.0",
-    kitchen_area: str = "11.5",
-    floor: int = 7,
-    floors_count: int = 16,
-    year_built: int = 2010,
-    material_type: str = "panel",
-    lat: float = 56.8385,
-    lng: float = 60.5940,
-    cadastral_number: str = "66:41:0204016:1234",
-    building_cadastral_number: str = "66:41:0204016:100",
-    description: str = "Продаётся отличная квартира",
-    description_minhash: str | list[int] = "abc123def456abc1",
-    newbuilding_id: int = 0,
-    has_phones: bool = True,
-    is_homeowner: bool = True,
-    is_pro: bool = False,
-    bargain_allowed: bool = True,
-    sale_type: str = "free",
-) -> dict:
-    """Синтетический Cian offer с полным набором полей (137 полей subset)."""
-    offer: dict = {
-        "cianId": cian_id,
-        "id": cian_id,
-        "fullUrl": f"https://ekb.cian.ru/sale/flat/{cian_id}/",
-        "category": "flatSale",
-        "dealType": "sale",
-        "offerType": "flat",
-        "flatType": "rooms",
-        "isApartments": False,
-        "roomsCount": rooms,
-        "status": "published",
-        # Area
-        "totalArea": total_area,
-        "livingArea": living_area,
-        "kitchenArea": kitchen_area,
-        "balconiesCount": 1,
-        "loggiasCount": 0,
-        "bedroomsCount": None,
-        "hasFurniture": False,
-        "isSoldFurnished": False,
-        "decoration": None,
-        # Floor / building
-        "floorNumber": floor,
-        "building": {
-            "floorsCount": floors_count,
-            "materialType": material_type,
-            "buildYear": year_built,
-            "deadline": None,
-            "parking": None,
-        },
-        # Cadastral
-        "cadastralNumber": cadastral_number,
-        "buildingCadastralNumber": building_cadastral_number,
-        # Price
-        "bargainTerms": {
-            "priceRur": price_rur,
-            "price": price_rur,
-            "currency": "rur",
-            "saleType": sale_type,
-            "mortgageAllowed": True,
-            "bargainAllowed": bargain_allowed,
-            "priceType": "all",
-        },
-        "formattedShortPrice": f"{price_rur // 1000000} млн ₽",
-        # Geo
-        "geo": {
-            "coordinates": {"lat": lat, "lng": lng},
-            "address": [
-                {"type": "location", "name": "Свердловская область"},
-                {"type": "city", "name": "Екатеринбург"},
-                {"type": "street", "name": "улица Постовского"},
-                {"type": "house", "name": "17А"},
-            ],
-            "undergrounds": [
-                {
-                    "name": "Геологическая",
-                    "time": 10,
-                    "transportType": "walk",
-                    "lineId": 10,
-                    "lineColor": "FF0000",
-                    "isDefault": True,
-                    "id": 100,
-                }
-            ],
-            "districts": [],
-        },
-        # Newbuilding
-        "newbuilding": {"id": newbuilding_id, "name": ""},
-        "buildersIds": [],
-        # Seller
-        "phones": [{"number": "9012345678", "countryCode": "7"}] if has_phones else [],
-        "userId": 13332612,
-        "cianUserId": 13332612,
-        "publishedUserId": 13332612,
-        "isByHomeowner": is_homeowner,
-        "isByCommercialOwner": False,
-        "isCianPartner": False,
-        "isPro": is_pro,
-        "isCalltrackingEnabled": False,
-        # Description
-        "description": description,
-        "descriptionMinhash": description_minhash,
-        "isDuplicatedDescription": False,
-        # Time
-        "addedTimestamp": 1716451200,  # 2024-05-23 00:00:00 UTC
-        "added": "вчера",
-        # Photos
-        "photos": [
-            {
-                "id": 1001,
-                "fullUrl": f"https://images.cdn-cian.ru/images/{cian_id}-1.jpg",
-                "thumbnailUrl": f"https://images.cdn-cian.ru/images/{cian_id}-2.jpg",
-            }
-        ],
-        # Verification
-        "isRosreestrChecked": True,
-        "isLayoutApproved": None,
-        # Misc
-        "isNew": False,
-        "isTop3": False,
-        "isPremium": False,
-    }
-    return offer
-
-
-# ── Tests ────────────────────────────────────────────────────────────────────
-
-
-def test_cian_serp_parses_full_offer():
-    """Базовый тест: полный offer → ScrapedLot с ключевыми полями."""
-    offer = _make_full_offer()
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    assert len(lots) == 1
-    lot = lots[0]
-
-    assert lot.source == "cian"
-    assert lot.source_id == "328739308"
-    assert lot.source_url == "https://ekb.cian.ru/sale/flat/328739308/"
-
-
-def test_cian_serp_coords_no_jitter():
-    """Координаты берутся точно из state — NO anchor jitter."""
-    offer = _make_full_offer(lat=56.8385, lng=60.5940)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    assert len(lots) == 1
-    lot = lots[0]
-    assert lot.lat == pytest.approx(56.8385, abs=1e-9)
-    assert lot.lon == pytest.approx(60.5940, abs=1e-9)
-
-
-def test_cian_serp_kitchen_area_in_raw_payload():
-    """Сырой kitchenArea (str) остаётся в raw_payload (#2008 промоутит float
-    в колонку ScrapedLot.kitchen_area_m2 отдельно — см. соответств. тест)."""
-    offer = _make_full_offer(kitchen_area="11.5")
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.raw_payload is not None
-    assert lot.raw_payload["kitchen_area_m2"] == "11.5"
-
-
-def test_cian_serp_description_minhash():
-    """description_minhash берётся из Cian state, если есть."""
-    offer = _make_full_offer(
-        description_minhash="cian_minhash_abc123",
-        description="Тест описания",
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.description_minhash == "cian_minhash_abc123"
-
-
-def test_cian_serp_description_minhash_fallback():
-    """Если Cian не дал minhash — вычисляем SHA1 из description."""
-    offer = _make_full_offer(description_minhash="", description="Уникальное описание квартиры")
-    offer["descriptionMinhash"] = ""  # явно пусто
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.description_minhash is not None
-    assert len(lot.description_minhash) == 32  # SHA1 hex [:32]
-
-
-def test_cian_serp_description_minhash_list_int():
-    """descriptionMinhash как list[int] (новый формат Cian ~2026-05-23) → строка через запятую.
-
-    Порядок элементов сохраняется как-есть — sorted() НЕ применяется,
-    т.к. порядок LSH-band'ов значим для dedup/cross-source matching.
-    Offer должен быть принят (не отброшен) — возвращает ScrapedLot, не None.
-    """
-    minhash_list = [8439545, 5781654, 555882, 407717]
-    offer = _make_full_offer(
-        description_minhash=minhash_list,  # type: ignore[arg-type]
-        description="Продаётся квартира с видом на парк",
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    assert len(lots) == 1, "offer с list[int] minhash должен быть принят, не отброшен"
-    lot = lots[0]
-    assert lot.description_minhash == "8439545,5781654,555882,407717"
-
-
-def test_cian_serp_phones():
-    """PII purge (#2217): даже если raw offer содержит phones, lot.phones пуст —
-    скрейпер больше не читает offer.get("phones")."""
-    offer = _make_full_offer(has_phones=True)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert isinstance(lot.phones, list)
-    assert lot.phones == []
-
-
-def test_cian_serp_house_ext_id_newbuilding():
-    """Новостройка с кадастром дома → anchor = building cad (#1773), не nb_id.
-
-    _make_full_offer несёт building_cadastral_number, поэтому same-building anchor
-    предпочитает кадастр дома (надёжнее, чем nb_id, который группирует весь ЖК)."""
-    offer = _make_full_offer(newbuilding_id=54016)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.house_source == "cian_building_cad"
-    assert lot.house_ext_id == "66:41:0204016:100"
-    assert lot.listing_segment == "novostroyki"
-
-
-def test_cian_serp_house_ext_id_newbuilding_nb_id_fallback():
-    """Без кадастра дома anchor падает на nb_id (ЖК) — back-compat fallback."""
-    offer = _make_full_offer(newbuilding_id=54016, building_cadastral_number="")
-    offer["buildingCadastralNumber"] = None
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.house_source == "cian_newbuilding"
-    assert lot.house_ext_id == "54016"
-    assert lot.listing_segment == "novostroyki"
-
-
-def test_cian_serp_vtorichka_no_house_ext_id():
-    """Вторичка (newbuilding.id == 0) — house_ext_id None."""
-    offer = _make_full_offer(newbuilding_id=0)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.house_source == "cian"
-    assert lot.house_ext_id is None
-    assert lot.listing_segment == "vtorichka"
-
-
-def test_cian_serp_metro_stations():
-    """metro_stations — список dicts с name/time/mode/line_color."""
-    offer = _make_full_offer()
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert isinstance(lot.metro_stations, list)
-    assert len(lot.metro_stations) == 1
-    station = lot.metro_stations[0]
-    assert station["name"] == "Геологическая"
-    assert station["time"] == 10
-    assert station["mode"] == "walk"
-    assert station["line_color"] == "FF0000"
-
-
-def test_cian_serp_cadastral_numbers():
-    """Кадастровые номера квартиры и здания."""
-    offer = _make_full_offer(
-        cadastral_number="66:41:0204016:1234",
-        building_cadastral_number="66:41:0204016:100",
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.cadastral_number == "66:41:0204016:1234"
-    assert lot.building_cadastral_number == "66:41:0204016:100"
-    # kadastr_num (base field) тоже заполнен
-    assert lot.kadastr_num == "66:41:0204016:1234"
-
-
-def test_cian_serp_is_homeowner_seller():
-    """is_homeowner и is_pro_seller из offer."""
-    offer = _make_full_offer(is_homeowner=True, is_pro=False)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.is_homeowner is True
-    assert lot.is_pro_seller is False
-
-
-def test_cian_serp_bargain_and_sale_type():
-    """bargain_allowed и sale_type из bargainTerms."""
-    offer = _make_full_offer(bargain_allowed=True, sale_type="alternative")
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.bargain_allowed is True
-    assert lot.sale_type == "alternative"
-
-
-def test_cian_serp_building_fields():
-    """Поля здания: year_built, house_type, total_floors."""
-    offer = _make_full_offer(year_built=1988, material_type="brick", floors_count=9)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.year_built == 1988
-    assert lot.house_type == "brick"
-    assert lot.total_floors == 9
-
-
-def test_cian_serp_promotes_class_b_fields():
-    """#2008: kitchen_area_m2 / mortgage_available / is_apartments /
-    is_rosreestr_checked промоутятся в колонки ScrapedLot (Class B —
-    дашборд/matching, НЕ valuation)."""
-    # _make_full_offer: mortgageAllowed=True, isApartments=False, isRosreestrChecked=True
-    offer = _make_full_offer(kitchen_area="11.5")
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.kitchen_area_m2 == pytest.approx(11.5)
-    assert lot.mortgage_available is True
-    assert lot.is_apartments is False
-    assert lot.is_rosreestr_checked is True
-
-
-def test_cian_serp_house_type_normalized_monolith_brick():
-    """#2008 (Class A): cian materialType 'monolithBrick' → канон 'monolith_brick';
-    сырьё остаётся в raw_payload['raw_material_type']."""
-    offer = _make_full_offer(material_type="monolithBrick")
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.house_type == "monolith_brick"
-    assert lot.raw_payload is not None
-    assert lot.raw_payload["raw_material_type"] == "monolithBrick"
-
-
-def test_cian_serp_house_type_block_synonyms_normalized():
-    """#2008: gasSilicate/aerocrete/foamConcrete Block + stalin → канон."""
-    scraper = CianScraper()
-    for raw, canon in (
-        ("gasSilicateBlock", "block"),
-        ("aerocreteBlock", "block"),
-        ("foamConcreteBlock", "block"),
-        ("stalin", "brick"),  # «сталинка» = кирпич
-    ):
-        offer = _make_full_offer(material_type=raw)
-        lots = scraper._parse_serp_html(_make_cian_serp_html([offer]))
-        assert lots[0].house_type == canon, f"{raw} → {lots[0].house_type} != {canon}"
-
-
-def test_cian_serp_house_type_canonical_passthrough():
-    """#2008: уже-каноничные monolith/panel остаются неизменными (idempotency)."""
-    scraper = CianScraper()
-    for raw in ("monolith", "panel", "brick", "block", "wood"):
-        offer = _make_full_offer(material_type=raw)
-        lots = scraper._parse_serp_html(_make_cian_serp_html([offer]))
-        assert lots[0].house_type == raw
-
-
-def test_cian_serp_house_type_unknown_becomes_none():
-    """#2008: неизвестный materialType → None (нейтрально для estimator soft-penalty)."""
-    offer = _make_full_offer(material_type="someUnknownMaterial")
-    lots = CianScraper()._parse_serp_html(_make_cian_serp_html([offer]))
-    assert lots[0].house_type is None
-
-
-def test_cian_serp_living_area():
-    """living_area_m2 из offer.livingArea."""
-    offer = _make_full_offer(living_area="35.0")
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.living_area_m2 == pytest.approx(35.0)
-
-
-def test_cian_serp_state_missing_returns_empty():
-    """Если state нет в HTML — возвращаем пустой список."""
-    html = "
no cian state here
"
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-    assert lots == []
-
-
-def test_cian_serp_skips_offer_without_price():
-    """Offer без цены — пропускаем."""
-    offer = _make_full_offer()
-    del offer["bargainTerms"]["priceRur"]
-    del offer["bargainTerms"]["price"]
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-    assert lots == []
-
-
-def test_cian_serp_multiple_offers():
-    """Несколько offers → список ScrapedLot."""
-    offers = [_make_full_offer(cian_id=100 + i) for i in range(5)]
-    html = _make_cian_serp_html(offers)
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-    assert len(lots) == 5
-    ids = [lot.source_id for lot in lots]
-    assert sorted(ids) == ["100", "101", "102", "103", "104"]
-
-
-def test_cian_serp_photo_urls():
-    """photo_urls — полные URL из offer.photos[].fullUrl."""
-    offer = _make_full_offer(cian_id=999)
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert len(lot.photo_urls) == 1
-    assert "cdn-cian.ru" in lot.photo_urls[0]
-    assert "999-1.jpg" in lot.photo_urls[0]
-
-
-def test_cian_serp_no_coords_does_not_crash():
-    """Если координаты отсутствуют — lot.lat/lon = None, не краш."""
-    offer = _make_full_offer()
-    del offer["geo"]["coordinates"]
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    assert len(lots) == 1
-    assert lots[0].lat is None
-    assert lots[0].lon is None
-
-
-# ── Transport: BrowserFetcher (camoufox) миграция (#1806) ────────────────────
-
-
-@pytest.mark.asyncio
-async def test_cian_aenter_opens_browser_fetcher() -> None:
-    """__aenter__ открывает ОДНУ BrowserFetcher(source="cian") сессию (camoufox)."""
-    mock_browser = MagicMock()
-    mock_browser.__aenter__ = AsyncMock(return_value=mock_browser)
-    mock_browser.__aexit__ = AsyncMock(return_value=None)
-
-    captured: dict = {}
-
-    def _fake_browser(*args, **kwargs):
-        captured["source"] = kwargs.get("source") if kwargs else (args[0] if args else None)
-        return mock_browser
-
-    scraper = CianScraper()
-    with patch("app.services.scrapers.cian.BrowserFetcher", _fake_browser):
-        await scraper.__aenter__()
-
-    mock_browser.__aenter__.assert_awaited_once()
-    assert scraper._browser is mock_browser
-    assert captured["source"] == "cian"
-
-    await scraper.__aexit__(None, None, None)
-    mock_browser.__aexit__.assert_awaited_once()
-    assert scraper._browser is None
-
-
-@pytest.mark.asyncio
-async def test_cian_fetch_page_html_uses_browser_fetcher() -> None:
-    """_fetch_page_html тянет HTML через BrowserFetcher.fetch, не через curl_cffi."""
-    scraper = CianScraper.__new__(CianScraper)
-    scraper.name = "cian"
-    scraper.base_url = "https://ekb.cian.ru"
-    scraper.request_delay_sec = 0.0
-
-    fetched_urls: list[str] = []
-
-    async def fake_fetch(url: str) -> str:
-        fetched_urls.append(url)
-        return "serp-page"
-
-    scraper._browser = MagicMock()
-    scraper._browser.fetch = AsyncMock(side_effect=fake_fetch)
-
-    html = await scraper._fetch_page_html((2,), 1, None, None)
-
-    assert html == "serp-page"
-    assert len(fetched_urls) == 1
-    assert "cat.php" in fetched_urls[0]
-
-
-@pytest.mark.asyncio
-async def test_cian_fetch_page_html_returns_none_on_browser_error() -> None:
-    """Если BrowserFetcher.fetch падает — _fetch_page_html возвращает None (не краш)."""
-    scraper = CianScraper.__new__(CianScraper)
-    scraper.name = "cian"
-    scraper.base_url = "https://ekb.cian.ru"
-    scraper.request_delay_sec = 0.0
-
-    scraper._browser = MagicMock()
-    scraper._browser.fetch = AsyncMock(side_effect=RuntimeError("browser down"))
-
-    html = await scraper._fetch_page_html((1,), 2, 1_000_000, 5_000_000)
-    assert html is None
-
-
-@pytest.mark.asyncio
-async def test_cian_fetch_around_parses_browser_html() -> None:
-    """fetch_around парсит лоты из SSR-HTML, отданного BrowserFetcher."""
-    offer = _make_full_offer(cian_id=777)
-    serp_html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper.__new__(CianScraper)
-    scraper.name = "cian"
-    scraper.base_url = "https://ekb.cian.ru"
-    scraper.request_delay_sec = 0.0
-    scraper._browser = MagicMock()
-    scraper._browser.fetch = AsyncMock(return_value=serp_html)
-
-    with patch.object(scraper, "sleep_between_requests", new_callable=AsyncMock):
-        lots = await scraper.fetch_around(56.84, 60.60, rooms=(2,), page=1)
-
-    assert len(lots) == 1
-    assert lots[0].source_id == "777"
-    scraper._browser.fetch.assert_awaited_once()
-
-
-@pytest.mark.asyncio
-async def test_cian_rotate_ip_is_noop() -> None:
-    """_rotate_ip после миграции — no-op, всегда False (прокси ротирует IP сам)."""
-    scraper = CianScraper.__new__(CianScraper)
-    result = await scraper._rotate_ip()
-    assert result is False
-
-
-# ── Address formatter tests ──────────────────────────────────────────────────
-
-
-def test_format_address_standard():
-    """Стандартный geo.address[]."""
-    parts = [
-        {"type": "location", "name": "Свердловская область"},
-        {"type": "city", "name": "Екатеринбург"},
-        {"type": "district", "name": "Чкаловский"},
-        {"type": "street", "name": "улица Постовского"},
-        {"type": "house", "name": "17А"},
-        {"type": "metro", "name": "Геологическая"},
-    ]
-    result = _format_address(parts)
-    # location и metro пропускаем
-    assert "Свердловская" not in result
-    assert "Геологическая" not in result
-    assert "Екатеринбург" in result
-    assert "17А" in result
-
-
-def test_format_address_empty():
-    """Пустой address[] → fallback string."""
-    assert _format_address([]) == "Екатеринбург (Cian)"
-
-
-# ── #1773: новостройка без part{type:"house"} → восстановление номера дома ─────
-
-
-def _make_newbuilding_houseless_offer(
-    cian_id: int = 330047129,
-    newbuilding_id: int = 54016,
-    newbuilding_name: str = "Суходольский квартал",
-    building_cadastral_number: str | None = "66:41:0313010:11396",
-    building_house_number: str | None = None,
-) -> dict:
-    """Cian-оффер новостройки: geo.address[] БЕЗ {type:"house"} (ЖК Суходольский,
-    ул. Евгения Савкова 29, cian 330047129 — реальный кейс из #1773)."""
-    offer = _make_full_offer(
-        cian_id=cian_id,
-        newbuilding_id=newbuilding_id,
-        building_cadastral_number=building_cadastral_number or "",
-    )
-    # geo.address[] без части house — корень бага: «…улица Евгения Савкова» без «29»
-    offer["geo"]["address"] = [
-        {"type": "location", "name": "Свердловская область"},
-        {"type": "city", "name": "Екатеринбург"},
-        {"type": "district", "name": "Академический"},
-        {"type": "street", "name": "улица Евгения Савкова"},
-    ]
-    offer["newbuilding"] = {"id": newbuilding_id, "name": newbuilding_name}
-    building = offer.setdefault("building", {})
-    if building_house_number is not None:
-        building["houseNumber"] = building_house_number
-    if building_cadastral_number is None:
-        offer["buildingCadastralNumber"] = None
-    return offer
-
-
-def test_cian_newbuilding_cad_number_anchor():
-    """Новостройка с кадастром дома → house_source/house_ext_id = building cad
-    (same-building anchor #1773), адрес дополнен кадастром."""
-    offer = _make_newbuilding_houseless_offer(
-        building_cadastral_number="66:41:0313010:11396",
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    assert len(lots) == 1
-    lot = lots[0]
-    # Кадастр дома предпочтён как anchor-ключ вместо nb_id (ЖК целиком)
-    assert lot.house_source == "cian_building_cad"
-    assert lot.house_ext_id == "66:41:0313010:11396"
-    assert lot.building_cadastral_number == "66:41:0313010:11396"
-    assert lot.listing_segment == "novostroyki"
-    # geo.address[] был без house → кадастр дописан в адрес как якорь
-    assert "Евгения Савкова" in (lot.address or "")
-    assert "66:41:0313010:11396" in (lot.address or "")
-
-
-def test_cian_newbuilding_recovers_house_from_building_number():
-    """building.houseNumber имеет приоритет (a): «29» восстановлен в адрес,
-    кадастр остаётся anchor-ключом."""
-    offer = _make_newbuilding_houseless_offer(
-        building_cadastral_number="66:41:0313010:11396",
-        building_house_number="29",
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    addr = lot.address or ""
-    assert "Евгения Савкова" in addr
-    assert "29" in addr  # номер дома восстановлен из building.houseNumber
-    # анкор по-прежнему кадастр дома (надёжнее nb_id)
-    assert lot.house_ext_id == "66:41:0313010:11396"
-
-
-def test_cian_newbuilding_recovers_house_from_name_when_no_cad():
-    """Без кадастра дома: дом-токен «29» вытягивается из названия ЖК (приоритет c),
-    anchor падает на nb_id."""
-    offer = _make_newbuilding_houseless_offer(
-        newbuilding_name="ЖК на Савкова 29",
-        building_cadastral_number=None,
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    assert lot.building_cadastral_number is None
-    # нет кадастра → anchor = nb_id (ЖК)
-    assert lot.house_source == "cian_newbuilding"
-    assert lot.house_ext_id == "54016"
-    addr = lot.address or ""
-    assert "29" in addr  # дом восстановлен из названия ЖК
-
-
-def test_cian_newbuilding_no_house_no_fabrication():
-    """Ни building, ни кадастра, ни номера в названии → номер НЕ выдумываем."""
-    offer = _make_newbuilding_houseless_offer(
-        newbuilding_name="Суходольский квартал",  # без числа-хвоста
-        building_cadastral_number=None,
-    )
-    html = _make_cian_serp_html([offer])
-
-    scraper = CianScraper()
-    lots = scraper._parse_serp_html(html)
-
-    lot = lots[0]
-    addr = lot.address or ""
-    assert "Евгения Савкова" in addr
-    # никаких фейковых catch-all номеров (например 7019) в адресе
-    assert not addr.rstrip().endswith("7019")
-
-
-def test_format_address_vtorichka_houseless_untouched():
-    """Вторичка с house-less адресом НЕ трогается — номер не выдумываем."""
-    parts = [
-        {"type": "city", "name": "Екатеринбург"},
-        {"type": "street", "name": "улица Евгения Савкова"},
-    ]
-    result = _format_address(
-        parts,
-        building={"houseNumber": "29"},
-        building_cadastral_number="66:41:0313010:11396",
-        newbuilding={"id": 0, "name": ""},
-        is_newbuilding=False,
-    )
-    # вторичка: is_newbuilding=False → ни «29», ни кадастр не дописываются
-    assert "29" not in result
-    assert "66:41" not in result
diff --git a/tradein-mvp/backend/tests/test_domclick_sweep.py b/tradein-mvp/backend/tests/test_domclick_sweep.py
index e7b6908c..0476100c 100644
--- a/tradein-mvp/backend/tests/test_domclick_sweep.py
+++ b/tradein-mvp/backend/tests/test_domclick_sweep.py
@@ -6,21 +6,25 @@ cancel, honest mark_done/mark_failed statuses) удалён вместе с
 теперь в `test_scraper_kit_pipeline_parity2.py::test_domclick_city_sweep_{done,
 blocked_failed}` (kit `run_domclick_city_sweep`).
 
+`app.services.scrapers.domclick` удалён (#2397 Part E2, 0 runtime-импортёров) —
+`_map_item` ретаргетирован на kit `scraper_kit.providers.domclick.serp.DomClickScraper`
+(единственный живой путь).
+
 Остались:
   - DomClickCitySweepCounters (dataclass, из scraper_kit) — defaults + to_dict
-  - DomClickScraper._map_item — BFF offer-item → ScrapedLot маппинг (не зависит
-    от оркестрации, `app.services.scrapers.domclick` живёт до Part E2)
+  - DomClickScraper._map_item (kit) — BFF offer-item → ScrapedLot маппинг (не
+    зависит от оркестрации)
 """
 
 from __future__ import annotations
 
 import os
+from types import SimpleNamespace
 
 os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")
 
 import pytest
-
-from app.services.scrapers.domclick import DomClickScraper
+from scraper_kit.providers.domclick.serp import DomClickScraper
 
 # ── DomClickCitySweepCounters ───────────────────────────────────────────────
 
@@ -54,7 +58,8 @@ def test_domclick_sweep_counters_to_dict_all_keys() -> None:
 
 def test_map_item_basic_mapping() -> None:
     """DomClickScraper._map_item маппит BFF offer-item → ScrapedLot корректно."""
-    scraper = DomClickScraper()
+    config = SimpleNamespace(browser_http_endpoint="http://tradein-browser:9000")
+    scraper = DomClickScraper(config, delay_provider=lambda _name: 8.0)
     item = {
         "id": 111222333,
         "path": "https://domclick.ru/card/sale__flat__111222333",
diff --git a/tradein-mvp/backend/tests/test_houses_rating_filter_entries_1789.py b/tradein-mvp/backend/tests/test_houses_rating_filter_entries_1789.py
deleted file mode 100644
index 428c28b7..00000000
--- a/tradein-mvp/backend/tests/test_houses_rating_filter_entries_1789.py
+++ /dev/null
@@ -1,189 +0,0 @@
-"""#1789 — rating_score из reviews.filterEntries[type==score] (fallback).
-
-Avito перенёс сводный рейтинг из housePage.developmentData.ratingPreview в виджет
-reviews.filterEntries[{type: 'score', value: {scoreFloat, scoreString, reviewCount, ratingStat}}].
-
-Тест 1: когда housePage НЕ даёт рейтинг (ratingBadge пуст), rating берётся из
-        reviews.filterEntries — проверяем rating_score/rating_string/reviews_count/
-        rating_distribution.
-
-Тест 2: housePage даёт рейтинг (старый путь ratingBadge) → filterEntries НЕ
-        перезаписывает (приоритет housePage).
-
-Тест 3: новостройка (developmentPage) без reviews-виджета → graceful, нет краша.
-
-Без сети, без БД.
-"""
-
-from __future__ import annotations
-
-import os
-
-import pytest
-
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
-
-from app.services.scrapers.avito_houses import parse_houses_state
-
-
-def _make_state(
-    *,
-    house_page_rating: dict | None = None,
-    reviews_filter_entries: list | None = None,
-) -> dict:
-    """Минимальный __preloadedState__ с housePage и опциональными reviews.
-
-    house_page_rating — если задан, проставляем в developmentData (старый путь).
-    reviews_filter_entries — если задан, проставляем в reviews.props.reviews.filterEntries
-    """
-    dev_data: dict = {
-        "avitoId": 555666,
-        "id": "aGFzaA==",
-        "title": "Test House",
-        "address": "ул. Тестовая, 1",
-        "coords": {"lat": 56.8, "lng": 60.6},
-        "aboutDevelopment": {"expandParams": {"items": []}},
-        "developer": {},
-        "mapPreview": {},
-    }
-    if house_page_rating is not None:
-        dev_data.update(house_page_rating)
-
-    placeholders: list = [
-        {
-            "type": "housePage",
-            "props": {"developmentData": dev_data},
-        },
-    ]
-
-    if reviews_filter_entries is not None:
-        placeholders.append(
-            {
-                "type": "reviews",
-                "props": {
-                    # MFE-форма: данные под ключом "reviews" внутри props
-                    "reviews": {
-                        "entries": [],
-                        "filterEntries": reviews_filter_entries,
-                    }
-                },
-            }
-        )
-
-    return {"data": {"data": {"page": {"placeholders": placeholders}}}}
-
-
-# ---------------------------------------------------------------------------
-# Тест 1: рейтинг из reviews.filterEntries (housePage пуст)
-# ---------------------------------------------------------------------------
-
-
-def test_rating_from_filter_entries_when_housepage_empty() -> None:
-    """Когда housePage не даёт рейтинг — берём из reviews.filterEntries."""
-    state = _make_state(
-        reviews_filter_entries=[
-            {
-                "type": "score",
-                "value": {
-                    "scoreFloat": 4.666,
-                    "scoreString": "4,7",
-                    "reviewCount": 3,
-                    "ratingStat": [
-                        {"score": 5, "count": 2},
-                        {"score": 4, "count": 1},
-                    ],
-                },
-            }
-        ]
-    )
-
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/test/555666")
-    h = e.house
-
-    assert h.rating_score == pytest.approx(4.666, rel=1e-3)
-    assert h.rating_string == "4,7"
-    assert h.reviews_count == 3
-    assert len(h.rating_distribution) == 2
-    assert h.rating_distribution[0]["score"] == 5
-
-
-def test_rating_filter_entries_float_type() -> None:
-    """scoreFloat из filterEntries конвертируется в float (не остаётся int/str)."""
-    state = _make_state(
-        reviews_filter_entries=[
-            {
-                "type": "score",
-                "value": {"scoreFloat": 5, "scoreString": "5,0", "reviewCount": 1},
-            }
-        ]
-    )
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/test/555666")
-    assert isinstance(e.house.rating_score, float)
-    assert e.house.rating_score == 5.0
-
-
-# ---------------------------------------------------------------------------
-# Тест 2: housePage даёт рейтинг → filterEntries НЕ перезаписывает
-# ---------------------------------------------------------------------------
-
-
-def test_housepage_rating_takes_priority_over_filter_entries() -> None:
-    """Если housePage.ratingBadge уже заполнен — filterEntries не перезаписывает."""
-    state = _make_state(
-        house_page_rating={
-            "ratingBadge": {"info": {"score": 3.9, "scoreString": "3,9"}},
-            "ratingSummaryStat": {"reviewCount": 10, "ratingStat": []},
-        },
-        reviews_filter_entries=[
-            {
-                "type": "score",
-                "value": {
-                    "scoreFloat": 4.9,
-                    "scoreString": "4,9",
-                    "reviewCount": 99,
-                    "ratingStat": [],
-                },
-            }
-        ],
-    )
-
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/test/555666")
-    h = e.house
-
-    # Значение из housePage должно сохраниться
-    assert h.rating_score == pytest.approx(3.9, rel=1e-3)
-    assert h.rating_string == "3,9"
-    assert h.reviews_count == 10
-
-
-# ---------------------------------------------------------------------------
-# Тест 3: нет reviews-виджета вообще → None, не падаем
-# ---------------------------------------------------------------------------
-
-
-def test_no_reviews_widget_rating_is_none() -> None:
-    """Без виджета reviews и без ratingBadge → rating_score=None, graceful."""
-    state = _make_state()  # нет reviews_filter_entries
-
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/test/555666")
-    h = e.house
-
-    assert h.rating_score is None
-    assert h.rating_string is None
-    assert h.reviews_count is None
-
-
-# ---------------------------------------------------------------------------
-# Тест 4: filterEntries без entry type==score → rating остаётся None
-# ---------------------------------------------------------------------------
-
-
-def test_filter_entries_without_score_type() -> None:
-    """filterEntries без type==score → рейтинг не заполняется."""
-    state = _make_state(
-        reviews_filter_entries=[
-            {"type": "something_else", "value": {"scoreFloat": 9.9}},
-        ]
-    )
-    e = parse_houses_state(state, "/catalog/houses/ekaterinburg/test/555666")
-    assert e.house.rating_score is None
diff --git a/tradein-mvp/backend/tests/test_pipeline_browser_routing.py b/tradein-mvp/backend/tests/test_pipeline_browser_routing.py
index 9d86d8e9..48530e6b 100644
--- a/tradein-mvp/backend/tests/test_pipeline_browser_routing.py
+++ b/tradein-mvp/backend/tests/test_pipeline_browser_routing.py
@@ -12,6 +12,10 @@ Legacy `run_avito_pipeline` browser/curl-mode integration tests (было ран
 удалены вместе с `app.services.scrape_pipeline` (#2397 Part E1) — browser-routing
 покрытие на уровне `fetch_detail`/`fetch_house_catalog`/`AvitoScraper._fetch_serp_html`
 ниже остаётся нетронутым.
+
+`fetch_house_catalog` ретаргетирован на `scraper_kit.providers.avito.houses` (#2397
+Part E2) — legacy `app.services.scrapers.avito_houses` удалён (0 runtime-импортёров,
+kit — единственный живой путь).
 """
 
 from __future__ import annotations
@@ -26,9 +30,11 @@ import pytest
 
 os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
 
+from scraper_kit.avito_exceptions import AvitoBlockedError as KitAvitoBlockedError
+from scraper_kit.providers.avito.houses import fetch_house_catalog
+
 from app.services.scrapers.avito_detail import fetch_detail
 from app.services.scrapers.avito_exceptions import AvitoBlockedError
-from app.services.scrapers.avito_houses import fetch_house_catalog
 
 FIXTURES = Path(__file__).parent / "fixtures"
 
@@ -205,10 +211,10 @@ async def test_fetch_house_catalog_browser_no_curl_used() -> None:
 
 @pytest.mark.asyncio
 async def test_fetch_house_catalog_browser_firewall_raises() -> None:
-    """Firewall HTML через browser path → AvitoBlockedError."""
+    """Firewall HTML через browser path → AvitoBlockedError (kit)."""
     mock_fetcher = make_mock_fetcher(FIREWALL_HTML)
 
-    with pytest.raises(AvitoBlockedError, match="browser-mode"):
+    with pytest.raises(KitAvitoBlockedError, match="browser-mode"):
         await fetch_house_catalog(
             "/catalog/houses/ekaterinburg/ul_test/9999",
             browser_fetcher=mock_fetcher,
diff --git a/tradein-mvp/backend/tests/test_scraper_kit_bisection_parity.py b/tradein-mvp/backend/tests/test_scraper_kit_bisection_parity.py
index 011c8d70..44e1508c 100644
--- a/tradein-mvp/backend/tests/test_scraper_kit_bisection_parity.py
+++ b/tradein-mvp/backend/tests/test_scraper_kit_bisection_parity.py
@@ -1,14 +1,15 @@
-"""Golden-parity бисекции: старый боевой обход `app.services.scrapers.`
+"""Golden-parity бисекции: старый боевой обход `app.services.scrapers.avito`
 ≡ новый единый движок `scraper_kit.pricing.walk_price_range` (#2134).
 
 Четыре независимые копии адаптивной бисекции «комнатность × цена» были
 консолидированы в один движок (`scraper_kit/pricing/bisection.py`). Провайдеры
-(avito/cian/yandex/domclick) теперь передают в движок свои probe/paginate + свой
-`BisectionConfig`. Этот тест доказывает, что unification НЕ изменил ОБХОД: для
-каждого провайдера при одинаковом входе (детерминированный мок счётчика по
-диапазону) старая боевая реализация и новая (движок) посещают ИДЕНТИЧНУЮ
-последовательность бакетов — probe-запросы, leaf-пагинации, деления, ранние
-возвраты на нуле, probe-fail-ветки.
+cian/domclick/yandex_realty удалены (#2397 Part E2, 0 runtime-импортёров) —
+их legacy-vs-kit секции обхода убраны отсюда вместе с ними (легаси-эталон
+исчез). `avito` остаётся легаси (не мигрирован) — его parity-проверка ниже
+доказывает, что unification НЕ изменил ОБХОД: при одинаковом входе
+(детерминированный мок счётчика по диапазону) старая боевая реализация и
+новая (движок) посещают ИДЕНТИЧНУЮ последовательность бакетов — probe-запросы,
+leaf-пагинации, деления, ранние возвраты на нуле, probe-fail-ветки.
 
 Механика: на ОБЕ реализации (старую и новую) ставятся ОДИНАКОВЫЕ моки на
 низкоуровневые методы (probe-fetch + leaf-пагинатор), которые (а) возвращают
@@ -22,7 +23,7 @@ import os
 from collections.abc import Callable
 from types import SimpleNamespace
 from typing import Any
-from unittest.mock import AsyncMock, MagicMock, patch
+from unittest.mock import AsyncMock, patch
 
 import pytest
 
@@ -30,14 +31,8 @@ os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/
 
 from scraper_kit.price_brackets import get_price_seed_brackets
 from scraper_kit.providers.avito.serp import AvitoScraper as NewAvito
-from scraper_kit.providers.cian.serp import CianScraper as NewCian
-from scraper_kit.providers.domclick.serp import DomClickScraper as NewDomClick
-from scraper_kit.providers.yandex.serp import YandexRealtyScraper as NewYandex
 
 from app.services.scrapers.avito import AvitoScraper as OldAvito
-from app.services.scrapers.cian import CianScraper as OldCian
-from app.services.scrapers.domclick import DomClickScraper as OldDomClick
-from app.services.scrapers.yandex_realty import YandexRealtyScraper as OldYandex
 
 # ── Оракул счётчиков ─────────────────────────────────────────────────────────
 
@@ -55,7 +50,7 @@ def make_oracle(
 
     span считается по клампнутым границам (None-lo → 0, None-hi → open_ceiling),
     чтобы широкие брекеты делились, а узкие уходили в leaf. `zeros` форсит 0 (ранний
-    возврат / leaf-на-нуле у yandex), `nones` форсит None (probe-fail).
+    возврат), `nones` форсит None (probe-fail).
     """
 
     def oracle(lo: Any, hi: Any) -> int | None:
@@ -71,10 +66,10 @@ def make_oracle(
     return oracle
 
 
-# ── Провайдер-специфичные «сборщики событий» ─────────────────────────────────
+# ── avito «сборщик событий» ──────────────────────────────────────────────────
 #
-# Каждый ставит одинаковые моки на старый/новый экземпляр и возвращает список
-# событий обхода. `("probe", lo, hi)` — запрос счётчика бакета; `("leaf", ...)` —
+# Ставит одинаковые моки на старый/новый экземпляр и возвращает список событий
+# обхода. `("probe", lo, hi)` — запрос счётчика бакета; `("leaf", ...)` —
 # пагинация leaf-бакета.
 
 
@@ -91,81 +86,6 @@ def _patch_avito(scraper: Any, oracle: Oracle, events: list[Any]) -> None:
     scraper._rotate_ip = AsyncMock(return_value=False)
 
 
-def _patch_cian(scraper: Any, oracle: Oracle, events: list[Any]) -> None:
-    holder: dict[str, tuple[Any, Any]] = {}
-
-    def fph(rooms: Any, page: int, min_price: Any, max_price: Any) -> str:
-        events.append(("probe", min_price, max_price))
-        holder["b"] = (min_price, max_price)
-        return "H"
-
-    def eto(html: str) -> int | None:
-        return oracle(*holder["b"])
-
-    def leaf(**kw: Any) -> None:
-        events.append(("leaf", kw["lo"], kw["hi"], kw["total"]))
-
-    scraper._fetch_page_html = AsyncMock(side_effect=fph)
-    scraper._extract_total_offers = MagicMock(side_effect=eto)
-    scraper._paginate_leaf_bucket = AsyncMock(side_effect=leaf)
-    scraper._rotate_ip = AsyncMock(return_value=False)
-    scraper.sleep_between_requests = AsyncMock()
-
-
-def _patch_domclick(scraper: Any, oracle: Oracle, events: list[Any]) -> None:
-    def count(fetcher: Any, rooms: Any, gte: Any, lte: Any) -> int:
-        events.append(("probe", gte, lte))
-        val = oracle(gte, lte)
-        return val if val is not None else 0
-
-    def paginate(
-        fetcher: Any, rooms: Any, gte: Any, lte: Any, seen_ids: Any, out_lots: Any, pages: Any
-    ) -> None:
-        events.append(("leaf", gte, lte))
-
-    scraper._count = AsyncMock(side_effect=count)
-    scraper._paginate = AsyncMock(side_effect=paginate)
-
-
-def _patch_yandex(
-    scraper: Any, oracle: Oracle, events: list[Any], *, record_pages: bool = False
-) -> None:
-    """Yandex: leaf-пагинация встроена в walk (нет отдельного метода-хука).
-
-    Стабим `_fetch_page_json` (page=1 отдаёт payload с totalItems из оракула;
-    page>1 — пустой payload) и передаём `on_bucket`, который пишет завершение
-    leaf/degraded бакета. probe = page-1-fetch; leaf-страницы 2..N идут через
-    asyncio.gather (порядок недетерминирован) — их НЕ пишем в норм-режиме; факт
-    leaf-обхода фиксируется через on_bucket. Для degraded-теста (пагинация
-    последовательная) `record_pages=True` пишет все страницы.
-    """
-
-    def gate(total: int | None, entities: list[Any] | None = None) -> dict[str, Any]:
-        pager: dict[str, Any] = {}
-        if total is not None:
-            pager["totalItems"] = total
-        return {"response": {"search": {"offers": {"entities": entities or [], "pager": pager}}}}
-
-    def fpj(rooms: Any, page: int, price_min: Any, price_max: Any, new_flat: str = "NO") -> Any:
-        if page == 1:
-            events.append(("probe", price_min, price_max))
-            return gate(oracle(price_min, price_max))
-        if record_pages:
-            events.append(("page", price_min, price_max, page))
-        return gate(0, entities=[])
-
-    scraper._fetch_page_json = AsyncMock(side_effect=fpj)
-    scraper._rotate_ip = AsyncMock(return_value=False)
-    scraper.request_delay_sec = 0.0
-
-
-def _yandex_on_bucket(events: list[Any]) -> Callable[[str, int], None]:
-    def on_bucket(bucket_key: str, count: int) -> None:
-        events.append(("bucket", bucket_key))
-
-    return on_bucket
-
-
 # ── Конструкторы старых/новых экземпляров (offline, без DB) ───────────────────
 
 _KIT_CONFIG = SimpleNamespace(
@@ -185,42 +105,14 @@ def _old_avito() -> Any:
         return OldAvito()
 
 
-def _old_cian() -> Any:
-    with patch("app.services.scrapers.cian.get_scraper_delay", return_value=5.0):
-        return OldCian()
+# ── avito: probe → split → leaf ───────────────────────────────────────────────
 
-
-def _old_yandex() -> Any:
-    with patch("app.services.scrapers.yandex_realty.get_scraper_delay", return_value=5.0):
-        return OldYandex()
-
-
-def _old_domclick() -> Any:
-    with patch("app.services.scrapers.domclick.get_scraper_delay", return_value=8.0):
-        return OldDomClick()
-
-
-# ── avito / cian: одинаковый скелет (probe → split → leaf) ────────────────────
-
-
-# Провайдер-специфичные kwargs для _walk_price_range (avito идентифицирует бакет
-# room_slug/room_label; cian — кортежем rooms).
 _AVITO_WALK_EXTRA = {"room_slug": "room1", "room_label": "1к"}
-_CIAN_WALK_EXTRA = {"rooms": (1,)}
 
 
 @pytest.mark.asyncio
-@pytest.mark.parametrize(
-    "old_factory, new_factory, patch_fn, walk_extra",
-    [
-        (_old_avito, lambda: NewAvito(config=_KIT_CONFIG), _patch_avito, _AVITO_WALK_EXTRA),
-        (_old_cian, lambda: NewCian(config=_KIT_CONFIG), _patch_cian, _CIAN_WALK_EXTRA),
-    ],
-)
-async def test_bisection_parity_avito_cian(
-    old_factory: Any, new_factory: Any, patch_fn: Any, walk_extra: dict[str, Any]
-) -> None:
-    """avito/cian: split-дерево + leaf + ранний возврат на нуле — обход идентичен.
+async def test_bisection_parity_avito() -> None:
+    """avito: split-дерево + leaf + ранний возврат на нуле — обход идентичен.
 
     Оракул: divisor=1000, cap=1400 → брекет 8М делится 3 уровня (8М→4М→2М→1М-leaf);
     один под-брекет форсим в 0 (ранний возврат без leaf).
@@ -237,7 +129,7 @@ async def test_bisection_parity_avito_cian(
 
     async def run(scraper: Any) -> list[Any]:
         events: list[Any] = []
-        patch_fn(scraper, oracle, events)
+        _patch_avito(scraper, oracle, events)
         for lo, hi in brackets:
             await scraper._walk_price_range(
                 lo=lo,
@@ -249,12 +141,12 @@ async def test_bisection_parity_avito_cian(
                 secondary_only=True,
                 on_bucket=None,
                 skip_buckets=None,
-                **walk_extra,
+                **_AVITO_WALK_EXTRA,
             )
         return events
 
-    old_events = await run(old_factory())
-    new_events = await run(new_factory())
+    old_events = await run(_old_avito())
+    new_events = await run(NewAvito(config=_KIT_CONFIG))
     assert old_events == new_events
     # Санити: обход реально нетривиален (несколько probe + несколько leaf).
     assert sum(1 for e in old_events if e[0] == "leaf") >= 4
@@ -300,140 +192,18 @@ async def test_bisection_parity_avito_probe_fail() -> None:
     assert ("leaf", 25_000_000, None, 100) in old_events
 
 
-# ── domclick: count-endpoint + None-both-ends + split открытого корня ─────────
-
-
-@pytest.mark.asyncio
-async def test_bisection_parity_domclick() -> None:
-    """domclick: unbounded-корень (None,None) клампится к [0,1e9] и делится;
-    count-endpoint; span<=MIN_PRICE_SPAN строго → leaf. Обход идентичен старому."""
-    oracle = make_oracle(divisor=100_000, open_ceiling=1_000_000_000)
-
-    async def run(scraper: Any) -> list[Any]:
-        events: list[Any] = []
-        _patch_domclick(scraper, oracle, events)
-        await scraper._sweep_bucket(
-            fetcher=None,
-            rooms="1",
-            price_gte=None,
-            price_lte=None,
-            seen_ids=set(),
-            out_lots=[],
-            pages=100,
-        )
-        return events
-
-    old_events = await run(_old_domclick())
-    new_events = await run(NewDomClick(_make_domclick_config()))
-    assert old_events == new_events
-    assert sum(1 for e in old_events if e[0] == "leaf") >= 4
-    # Корень (None,None) обязательно пробится первым.
-    assert old_events[0] == ("probe", None, None)
-
-
-def _make_domclick_config() -> Any:
-    return SimpleNamespace(browser_http_endpoint="http://tradein-browser:9000")
-
-
-# ── yandex: max_depth=8, min_bracket=500k, count-based on_bucket ──────────────
-
-
-@pytest.mark.asyncio
-async def test_bisection_parity_yandex_normal() -> None:
-    """yandex: split-дерево (порог 500k, cap 500) + leaf-on-bucket + zero-as-leaf.
-
-    Оракул divisor=1000, cap=500 → 4М→2М→1М→500k-leaf. Один под-брекет=0
-    (yandex вызывает on_bucket даже на нуле — treat_zero_as_leaf).
-    """
-    oracle = make_oracle(
-        divisor=1000,
-        open_ceiling=30_000_000,
-        zeros=frozenset({(2_000_000, 2_999_999)}),
-    )
-    brackets: list[tuple[int, int | None]] = [(0, 3_999_999), (25_000_000, None)]
-
-    async def run(scraper: Any) -> list[Any]:
-        events: list[Any] = []
-        _patch_yandex(scraper, oracle, events)
-        for lo, hi in brackets:
-            await scraper._walk_price_range(
-                rooms="1",
-                lo=lo,
-                hi=hi,
-                seen={},
-                price_cap_per_bucket=500,
-                max_pages_per_bucket=50,
-                concurrency=4,
-                on_bucket=_yandex_on_bucket(events),
-                skip_buckets=None,
-            )
-        return events
-
-    old_events = await run(_old_yandex())
-    new_events = await run(NewYandex(config=_KIT_CONFIG))
-    assert old_events == new_events
-    assert sum(1 for e in old_events if e[0] == "bucket") >= 4
-    assert sum(1 for e in old_events if e[0] == "probe") >= 6
-
-
-@pytest.mark.asyncio
-async def test_bisection_parity_yandex_degraded() -> None:
-    """yandex probe-fail (totalItems неизвестен) → degraded paginate-until-empty.
-
-    Оракул форсит None на бакете → probe page1 payload без totalItems → total=None
-    → degraded последовательно пагинирует. record_pages=True фиксирует все страницы.
-    """
-    oracle = make_oracle(
-        divisor=1000,
-        open_ceiling=30_000_000,
-        nones=frozenset({(None, 3_999_999)}),
-    )
-
-    async def run(scraper: Any) -> list[Any]:
-        events: list[Any] = []
-        _patch_yandex(scraper, oracle, events, record_pages=True)
-        await scraper._walk_price_range(
-            rooms="1",
-            lo=0,
-            hi=3_999_999,
-            seen={},
-            price_cap_per_bucket=500,
-            max_pages_per_bucket=50,
-            concurrency=4,
-            on_bucket=_yandex_on_bucket(events),
-            skip_buckets=None,
-        )
-        return events
-
-    old_events = await run(_old_yandex())
-    new_events = await run(NewYandex(config=_KIT_CONFIG))
-    assert old_events == new_events
-    # degraded: probe page1 + затем последовательная пагинация страниц 1..K до пустой.
-    assert ("probe", None, 3_999_999) in old_events
-    assert any(e[0] == "bucket" for e in old_events)
-
-
 # ── Full seed-grid обход (все seed-брекеты одной комнатности) ─────────────────
 
 
 @pytest.mark.asyncio
-@pytest.mark.parametrize(
-    "old_factory, new_factory, patch_fn, walk_extra",
-    [
-        (_old_avito, lambda: NewAvito(config=_KIT_CONFIG), _patch_avito, _AVITO_WALK_EXTRA),
-        (_old_cian, lambda: NewCian(config=_KIT_CONFIG), _patch_cian, _CIAN_WALK_EXTRA),
-    ],
-)
-async def test_bisection_parity_full_seed_grid(
-    old_factory: Any, new_factory: Any, patch_fn: Any, walk_extra: dict[str, Any]
-) -> None:
+async def test_bisection_parity_full_seed_grid() -> None:
     """Полный обход всех seed-брекетов (как в fetch_all_secondary) — включая
-    открытый catch-all — идентичен между старым и новым для avito/cian."""
+    открытый catch-all — идентичен между старым и новым для avito."""
     oracle = make_oracle(divisor=1500, open_ceiling=300_000_000)
 
     async def run(scraper: Any) -> list[Any]:
         events: list[Any] = []
-        patch_fn(scraper, oracle, events)
+        _patch_avito(scraper, oracle, events)
         for br_lo, br_hi in get_price_seed_brackets():
             walk_hi = br_hi - 1 if br_hi is not None else None
             await scraper._walk_price_range(
@@ -446,10 +216,10 @@ async def test_bisection_parity_full_seed_grid(
                 secondary_only=True,
                 on_bucket=None,
                 skip_buckets=None,
-                **walk_extra,
+                **_AVITO_WALK_EXTRA,
             )
         return events
 
-    old_events = await run(old_factory())
-    new_events = await run(new_factory())
+    old_events = await run(_old_avito())
+    new_events = await run(NewAvito(config=_KIT_CONFIG))
     assert old_events == new_events
diff --git a/tradein-mvp/backend/tests/test_scraper_kit_cian_golden_parity.py b/tradein-mvp/backend/tests/test_scraper_kit_cian_golden_parity.py
index 9c36f8f5..15378f5c 100644
--- a/tradein-mvp/backend/tests/test_scraper_kit_cian_golden_parity.py
+++ b/tradein-mvp/backend/tests/test_scraper_kit_cian_golden_parity.py
@@ -1,20 +1,22 @@
-"""Golden-parity: `scraper_kit.providers.cian.*` парсинг ≡ `app.services.scrapers.cian*`.
+"""Golden-parity: `scraper_kit.providers.cian.*` парсинг ≡ ещё живым `app.services.scrapers.cian*`.
 
 Strangler-инвариант (#2133 cian): новая scraper_kit-копия cian-провайдера должна давать
 БАЙТ-ИДЕНТИЧНЫЙ результат парсинга старому боевому коду на одинаковом входе. Развязка
 (settings→ScraperConfig, get_scraper_delay→delay_provider, cian_session→локальная копия)
 НЕ меняет распарсенные данные.
 
-Гоняем ОБА модуля на одних фикстурах и сравниваем результат:
-  - SERP `_parse_serp_html` (Redux state → list[ScrapedLot]) на синтетическом _cianConfig
-    push-HTML (вторичка + новостройка + студия), сравнение model_dump();
-  - SERP `_offer_to_lot` пооферно + `_format_address` (pure, восстановление дома ЖК);
-  - `_extract_total_offers` (totalOffers из state);
-  - `extract_state` — app-версия ≡ kit-версия на одном HTML (общий парсер state);
-  - detail `fetch_detail` (defaultState + bti sister → DetailEnrichment, dataclasses.asdict);
-  - valuation `_parse_valuation_state` (estimation/chart/houseInfo → CianValuationResult);
-  - newbuilding pure-экстракторы (_extract_chart / _extract_reliability_checks /
-    _extract_nested_offers / _extract_zhk_url_from_serp).
+`app.services.scrapers.cian` / `cian_detail` удалены (#2397 Part E2, 0 runtime-
+импортёров — admin.py/cian_history_backfill.py уже на kit). Parity-тесты SERP
+`_parse_serp_html`/`_offer_to_lot`/`_extract_total_offers`/`_format_address` и
+detail `fetch_detail` удалены вместе с ними (легаси-эталон исчез).
+
+Оставшееся здесь сравнивает kit ≡ ЕЩЁ ЖИВЫЕ легаси-модули:
+  - `extract_state` — `app.services.scrapers.cian_state_parser` (живой, общий парсер state);
+  - valuation `_parse_valuation_state` / `estimate_via_cian_valuation` —
+    `app.services.scrapers.cian_valuation` (живой);
+  - newbuilding pure-экстракторы — `app.services.scrapers.cian_newbuilding` (живой);
+  - kit mandatory-config guards (`estimate_via_cian_valuation`, `cian.session`) — kit-only;
+  - strangler-guard: kit-провайдер не импортирует `app.*`.
 
 Идентичность результата = kit-парсер верно скопирован (развязка не задела логику).
 """
@@ -27,17 +29,17 @@ import json
 import os
 from types import SimpleNamespace
 from typing import Any
-from unittest.mock import AsyncMock, MagicMock, patch
+from unittest.mock import MagicMock
 
 import pytest
 
-# Старый app.services.scrapers.cian* импортирует app.core.config.settings=Settings(),
-# которому нужен DATABASE_URL. Офлайн-парсинг БД не трогает — фиктивный DSN достаточен.
+# Старый app.services.scrapers.cian_state_parser/cian_valuation/cian_newbuilding
+# импортируют app.core.config.settings=Settings(), которому нужен DATABASE_URL.
+# Офлайн-парсинг БД не трогает — фиктивный DSN достаточен.
 os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
 
 # НОВЫЙ (scraper_kit) cian-провайдер.
 from scraper_kit.cian_state_parser import extract_state as new_extract_state
-from scraper_kit.providers.cian.detail import fetch_detail as new_fetch_detail
 from scraper_kit.providers.cian.newbuilding import _extract_chart as new_extract_chart
 from scraper_kit.providers.cian.newbuilding import (
     _extract_nested_offers as new_extract_nested_offers,
@@ -48,8 +50,6 @@ from scraper_kit.providers.cian.newbuilding import (
 from scraper_kit.providers.cian.newbuilding import (
     _extract_zhk_url_from_serp as new_extract_zhk_url,
 )
-from scraper_kit.providers.cian.serp import CianScraper as NewCianScraper
-from scraper_kit.providers.cian.serp import _format_address as new_format_address
 from scraper_kit.providers.cian.valuation import (
     _parse_valuation_state as new_parse_valuation,
 )
@@ -62,10 +62,7 @@ from scraper_kit.providers.cian.valuation import (
 # #2306's cian_price_history.py (см. app/services/scraper_adapters.py).
 from app.services.scraper_adapters import RealScraperConfig
 
-# СТАРЫЙ (app.services.scrapers) cian-провайдер.
-from app.services.scrapers.cian import CianScraper as OldCianScraper
-from app.services.scrapers.cian import _format_address as old_format_address
-from app.services.scrapers.cian_detail import fetch_detail as old_fetch_detail
+# СТАРЫЙ (app.services.scrapers) — ещё живые cian-модули (cian.py/cian_detail.py удалены).
 from app.services.scrapers.cian_newbuilding import _extract_chart as old_extract_chart
 from app.services.scrapers.cian_newbuilding import (
     _extract_nested_offers as old_extract_nested_offers,
@@ -85,14 +82,11 @@ from app.services.scrapers.cian_valuation import (
 )
 from tests.support.parity import assert_parity
 
-# ── Fixtures: SERP offers → _cianConfig push HTML ────────────────────────────
-
-_CONFIG = SimpleNamespace(glitchtip_dsn=None)
+# ── extract_state parity (state-парсер, живой в cian_state_parser.py) ────────
 
 
-def _serp_html(offers: list[dict[str, Any]]) -> str:
-    """Обернуть offers в валидный _cianConfig['frontend-serp'].push() HTML."""
-    state = {"results": {"totalOffers": len(offers), "offers": offers}}
+def _serp_html_for_state(state: dict[str, Any]) -> str:
+    """Обернуть произвольный state в валидный _cianConfig['frontend-serp'].push() HTML."""
     state_json = json.dumps(state, ensure_ascii=False)
     return (
         ""
-    )
-
-
-def _mock_session(html: str) -> MagicMock:
-    response = MagicMock()
-    response.status_code = 200
-    response.text = html
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-    return session
-
-
-@pytest.mark.asyncio
-async def test_detail_fetch_parity() -> None:
-    """fetch_detail (defaultState + bti) → DetailEnrichment идентичен (все поля)."""
-    html = _detail_html()
-    url = "https://ekb.cian.ru/sale/flat/327830237/"
-
-    old_res = await old_fetch_detail(url, session=_mock_session(html))
-    new_res = await new_fetch_detail(url, session=_mock_session(html))
-
-    assert old_res is not None
-    assert new_res is not None
-    assert dataclasses.asdict(old_res) == dataclasses.asdict(new_res)
-
-
-# ── valuation _parse_valuation_state parity ──────────────────────────────────
+# ── valuation _parse_valuation_state parity (живой cian_valuation.py) ────────
 
 
 def _valuation_state() -> dict[str, Any]:
@@ -595,7 +312,7 @@ def test_kit_cian_session_config_requirement_asymmetry() -> None:
     db2.execute.assert_called_once()
 
 
-# ── newbuilding pure-экстракторы parity ──────────────────────────────────────
+# ── newbuilding pure-экстракторы parity (живой cian_newbuilding.py) ──────────
 
 
 def test_newbuilding_extractors_parity() -> None:
diff --git a/tradein-mvp/backend/tests/test_scraper_kit_domclick_golden_parity.py b/tradein-mvp/backend/tests/test_scraper_kit_domclick_golden_parity.py
deleted file mode 100644
index 312fc866..00000000
--- a/tradein-mvp/backend/tests/test_scraper_kit_domclick_golden_parity.py
+++ /dev/null
@@ -1,183 +0,0 @@
-"""Golden-parity: `scraper_kit.providers.domclick.serp` ≡ `app.services.scrapers.domclick`.
-
-Strangler-инвариант (#2133): новая scraper_kit-копия domclick-провайдера должна
-давать БАЙТ-ИДЕНТИЧНЫЙ результат парсинга старому боевому коду на одинаковом входе.
-Развязка (get_scraper_delay→delay_provider, base/exceptions/repair-normalizer→
-scraper_kit.*) НЕ меняет распарсенные данные.
-
-Гоняем ОБА модуля на одной BFF-JSON-фикстуре и сравниваем:
-  - `_extract_json` (JSON из HTML-обёртки + QRATOR-block detection);
-  - `_map_item` (offer → ScrapedLot): студия rooms=0, sentinel 0→None, agency;
-  - `_is_geo_ok` (bbox + offerRegionName guard);
-  - url-билдеры и чистые хелперы (_build_offers_url/_build_count_url/
-    _parse_publish_date/_extract_agency_name).
-
-Идентичность результата = kit-парсер верно скопирован (развязка не задела логику).
-"""
-
-from __future__ import annotations
-
-import json
-import os
-from pathlib import Path
-from types import SimpleNamespace
-from typing import Any
-from unittest.mock import patch
-
-# Старый app.services.scrapers.domclick тянет app.core.config.settings=Settings(),
-# которому нужен DATABASE_URL. Офлайн-парсинг БД не трогает — фиктивный DSN достаточен.
-os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
-
-# НОВЫЙ (scraper_kit) — ruff сортирует scraper_kit (first-party) выше app-импортов.
-from scraper_kit.providers.domclick.serp import DomClickScraper as NewDomClickScraper
-from scraper_kit.providers.domclick.serp import _build_count_url as new_build_count_url
-from scraper_kit.providers.domclick.serp import _build_offers_url as new_build_offers_url
-from scraper_kit.providers.domclick.serp import _extract_agency_name as new_extract_agency
-from scraper_kit.providers.domclick.serp import _extract_json as new_extract_json
-from scraper_kit.providers.domclick.serp import _parse_publish_date as new_parse_publish_date
-
-from app.services.scrapers.domclick import DomClickScraper as OldDomClickScraper
-from app.services.scrapers.domclick import _build_count_url as old_build_count_url
-from app.services.scrapers.domclick import _build_offers_url as old_build_offers_url
-from app.services.scrapers.domclick import _extract_agency_name as old_extract_agency
-from app.services.scrapers.domclick import _extract_json as old_extract_json
-from app.services.scrapers.domclick import _parse_publish_date as old_parse_publish_date
-
-_FIXTURES = Path(__file__).parent / "fixtures"
-_BFF_JSON = (_FIXTURES / "domclick_bff_offers_sample.json").read_text(encoding="utf-8")
-_BFF_ITEMS: list[dict[str, Any]] = json.loads(_BFF_JSON)["result"]["items"]
-
-
-def _make_old() -> OldDomClickScraper:
-    """Старый DomClickScraper без DB: patch get_scraper_delay."""
-    with patch("app.services.scrapers.domclick.get_scraper_delay", return_value=8.0):
-        return OldDomClickScraper()
-
-
-def _make_new() -> NewDomClickScraper:
-    """Новый DomClickScraper с инжектируемым ScraperConfig + delay_provider.
-
-    Парсинг-тесты не дёргают fetch_city → browser_http_endpoint не используется,
-    но конструктор требует config (duck-typed namespace достаточен).
-    """
-    config = SimpleNamespace(browser_http_endpoint="http://tradein-browser:9000")
-    return NewDomClickScraper(config, delay_provider=lambda _name: 8.0)
-
-
-# ── _extract_json parity ──────────────────────────────────────────────────────
-
-
-def test_extract_json_parity_bare() -> None:
-    """_extract_json на bare JSON-теле идентичен."""
-    assert old_extract_json(_BFF_JSON) == new_extract_json(_BFF_JSON)
-
-
-def test_extract_json_parity_pre_wrapped() -> None:
-    """_extract_json на 
-обёрнутом JSON идентичен (реальная форма ответа)."""
-    wrapped = f"
{_BFF_JSON}
"
-    assert old_extract_json(wrapped) == new_extract_json(wrapped)
-
-
-def test_extract_json_parity_qrator_block() -> None:
-    """QRATOR block-страница → DomClickBlockedError в обоих (идентичное поведение)."""
-    from scraper_kit.domclick_exceptions import DomClickBlockedError as NewBlocked
-
-    from app.services.scrapers.domclick_exceptions import DomClickBlockedError as OldBlocked
-
-    block_html = "Система защиты QRATOR: captcha"
-    old_raised = new_raised = False
-    try:
-        old_extract_json(block_html)
-    except OldBlocked:
-        old_raised = True
-    try:
-        new_extract_json(block_html)
-    except NewBlocked:
-        new_raised = True
-    assert old_raised and new_raised
-
-
-# ── _map_item parity (offer → ScrapedLot) ─────────────────────────────────────
-
-
-def test_map_item_parity_all_items() -> None:
-    """Каждый offer-item → ScrapedLot идентичен (студия rooms=0, sentinel 0→None)."""
-    old = _make_old()
-    new = _make_new()
-    for item in _BFF_ITEMS:
-        # force_rooms=0 для студий (st-бакет) — проверяем обе ветки маппинга.
-        for force_rooms in (None, 0):
-            old_lot = old._map_item(item, force_rooms=force_rooms)
-            new_lot = new._map_item(item, force_rooms=force_rooms)
-            if old_lot is None or new_lot is None:
-                assert old_lot is None and new_lot is None
-                continue
-            assert old_lot.model_dump() == new_lot.model_dump()
-    # Счётчик parse-ошибок эволюционирует одинаково.
-    assert old.parse_failures == new.parse_failures
-
-
-# ── _is_geo_ok parity (bbox + region guard) ───────────────────────────────────
-
-
-def test_is_geo_ok_parity_all_items() -> None:
-    """Гео-гард (offerRegionName + bbox) идентичен на всех items."""
-    old = _make_old()
-    new = _make_new()
-    for item in _BFF_ITEMS:
-        assert old._is_geo_ok(item) == new._is_geo_ok(item)
-
-
-# ── url-билдеры + чистые хелперы ──────────────────────────────────────────────
-
-
-def test_url_builders_parity() -> None:
-    """_build_offers_url / _build_count_url идентичны (вкл. '5+' → '5%2B')."""
-    cases: list[tuple[str, int | None, int | None]] = [
-        ("st", None, None),
-        ("2", 3_000_000, 8_000_000),
-        ("5+", 15_000_000, None),
-        ("1", None, 5_000_000),
-    ]
-    for rooms, gte, lte in cases:
-        assert old_build_count_url(rooms, gte, lte) == new_build_count_url(rooms, gte, lte)
-        for offset in (0, 20, 1980):
-            assert old_build_offers_url(rooms, gte, lte, offset) == new_build_offers_url(
-                rooms, gte, lte, offset
-            )
-
-
-def test_parse_publish_date_parity() -> None:
-    """_parse_publish_date идентичен на валидных/битых ISO-строках."""
-    for iso in [None, "", "2026-06-28T09:15:00+05:00", "2026-06-27", "not-a-date", "2026-13-40"]:
-        assert old_parse_publish_date(iso) == new_parse_publish_date(iso)
-
-
-def test_extract_agency_name_parity() -> None:
-    """_extract_agency_name идентичен на строковых/dict/пустых seller-блоках."""
-    sellers: list[dict[str, Any]] = [
-        {},
-        {"company": "АН Этажи", "agent": "Иванов"},
-        {"company": "   ", "agent": {"name": "Резерв"}},
-        {"agent": {"fullName": "Петров П."}},
-        {"company": {"title": "СберРиелт"}},
-        {"agent": 12345},
-    ]
-    for seller in sellers:
-        assert old_extract_agency(seller) == new_extract_agency(seller)
-
-
-# ── strangler-guard: kit-провайдер не импортирует app.* ───────────────────────
-
-
-def test_kit_domclick_has_no_app_imports() -> None:
-    """Ни один модуль scraper_kit.providers.domclick.* не импортирует app.* (развязка)."""
-    import inspect
-
-    from scraper_kit.providers.domclick import serp
-
-    source = inspect.getsource(serp)
-    for line in source.splitlines():
-        stripped = line.strip()
-        assert not stripped.startswith("from app"), f"serp: {line!r}"
-        assert not stripped.startswith("import app"), f"serp: {line!r}"
diff --git a/tradein-mvp/backend/tests/test_scraper_kit_group_c_backfill_kit_parity.py b/tradein-mvp/backend/tests/test_scraper_kit_group_c_backfill_kit_parity.py
index 26edff4b..6fd9acf0 100644
--- a/tradein-mvp/backend/tests/test_scraper_kit_group_c_backfill_kit_parity.py
+++ b/tradein-mvp/backend/tests/test_scraper_kit_group_c_backfill_kit_parity.py
@@ -32,16 +32,21 @@ Migrates legacy scraper imports of these files to their `scraper_kit` equivalent
 
 This file proves the DELTA introduced by the #2310 migration specifically —
 general avito/cian-provider parsing golden-parity is already covered by
-`tests/test_scraper_kit_avito_golden_parity.py` and
+`tests/test_scraper_kit_avito_golden_parity.py` and (while it existed)
 `tests/test_scraper_kit_cian_golden_parity.py`. Uses `tests/support/parity.assert_parity`
 per the epic's parity-gate convention (issue #2304).
+
+`test_cian_fetch_detail_via_browser_fetcher_parity` (legacy `app.services.scrapers.
+cian_detail.fetch_detail` vs kit, browser_fetcher call shape) removed (#2397 Part E2)
+— `app.services.scrapers.cian_detail` deleted, 0 runtime importers (cian_history_backfill.py
+already calls the kit `fetch_detail` exclusively, proven by
+`test_cian_history_backfill_browser_fetcher_uses_settings_endpoint` below).
 """
 
 from __future__ import annotations
 
 import asyncio
 import dataclasses
-import json
 import os
 from typing import Any
 from unittest.mock import AsyncMock, MagicMock, patch
@@ -53,12 +58,10 @@ os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/
 from scraper_kit.avito_exceptions import AvitoBlockedError as KitAvitoBlockedError
 from scraper_kit.providers.avito.detail import fetch_detail as kit_avito_fetch_detail
 from scraper_kit.providers.avito.serp import AvitoScraper as KitAvitoScraper
-from scraper_kit.providers.cian.detail import fetch_detail as kit_cian_fetch_detail
 
 from app.core.config import settings
 from app.services.scraper_adapters import RealScraperConfig
 from app.services.scrapers.avito_detail import fetch_detail as legacy_avito_fetch_detail
-from app.services.scrapers.cian_detail import fetch_detail as legacy_cian_fetch_detail
 from tests.support.parity import assert_parity
 
 # ── avito fetch_detail: backconnect config-gate (avito_detail_backfill.py delta) ──
@@ -144,30 +147,7 @@ def test_avito_scraper_rotate_ip_reads_injected_config() -> None:
     assert scraper._config.avito_proxy_rotate_url == settings.avito_proxy_rotate_url
 
 
-# ── cian fetch_detail via browser_fetcher (cian_history_backfill.py call shape) ──
-
-
-def _cian_detail_html() -> str:
-    default_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 777,
-                "priceChanges": [
-                    {"changeTime": "2026-06-01T00:00:00Z", "price": 5900000, "diffPercent": -0.5}
-                ],
-            },
-            "stats": {},
-        }
-    }
-    ds_json = json.dumps(default_state, ensure_ascii=False)
-    return (
-        ""
-    )
+# ── avito fetch_detail via browser_fetcher (avito_detail_backfill.py browser_mode) ──
 
 
 def _fake_browser_fetcher(html: str) -> Any:
@@ -176,30 +156,6 @@ def _fake_browser_fetcher(html: str) -> Any:
     return bf
 
 
-def test_cian_fetch_detail_via_browser_fetcher_parity() -> None:
-    """cian_history_backfill.py's listings block ALWAYS calls fetch_detail(url,
-    browser_fetcher=bf) — never the curl_cffi/config own-session path exercised by
-    tests/test_scraper_kit_pricehistory_session_parity.py (cian_price_history.py's
-    call shape). Proves parity holds for THIS call shape too."""
-    html = _cian_detail_html()
-    url = "https://ekb.cian.ru/sale/flat/777/"
-
-    def _run_legacy() -> Any:
-        return dataclasses.asdict(
-            asyncio.run(legacy_cian_fetch_detail(url, browser_fetcher=_fake_browser_fetcher(html)))
-        )
-
-    def _run_kit() -> Any:
-        return dataclasses.asdict(
-            asyncio.run(kit_cian_fetch_detail(url, browser_fetcher=_fake_browser_fetcher(html)))
-        )
-
-    assert_parity(legacy_fn=_run_legacy, kit_fn=_run_kit, fixtures=[()])
-
-
-# ── avito fetch_detail via browser_fetcher (avito_detail_backfill.py browser_mode) ──
-
-
 def test_avito_fetch_detail_via_browser_fetcher_parity() -> None:
     """browser_mode branch of avito_detail_backfill.py calls fetch_detail(url,
     browser_fetcher=browser_fetcher) — config is never read on this path (early
diff --git a/tradein-mvp/backend/tests/test_scraper_kit_pricehistory_session_parity.py b/tradein-mvp/backend/tests/test_scraper_kit_pricehistory_session_parity.py
index 8ce767e5..6daf481e 100644
--- a/tradein-mvp/backend/tests/test_scraper_kit_pricehistory_session_parity.py
+++ b/tradein-mvp/backend/tests/test_scraper_kit_pricehistory_session_parity.py
@@ -17,32 +17,31 @@ cian-provider / ScrapedLot golden-parity is already covered by
 `tests/test_scraper_kit_cian_golden_parity.py` and
 `tests/test_scraper_kit_base_golden_parity.py`. Uses `tests/support/parity.assert_parity`
 per the epic's parity-gate convention (issue #2304).
+
+`test_fetch_detail_with_config_injection_parity` / `test_fetch_detail_own_session_proxy_
+wiring_parity` (legacy `app.services.scrapers.cian_detail.fetch_detail` vs kit) removed
+(#2397 Part E2) — `cian_detail` deleted, 0 runtime importers. `extract_state`
+(`cian_state_parser`, still legacy-live) and `ScrapedLot` (`base`, still legacy-live)
+parity below is unaffected.
 """
 
 from __future__ import annotations
 
-import asyncio
-import dataclasses
 import json
 import os
 from typing import Any
-from unittest.mock import AsyncMock, MagicMock, patch
 
-# Старый app.services.scrapers.cian* / app.core.config.settings требует DATABASE_URL
-# при импорте (тот же паттерн, что в test_scraper_kit_cian_golden_parity.py).
+# Старый app.services.scrapers.cian_state_parser/base / app.core.config.settings
+# требует DATABASE_URL при импорте (тот же паттерн, что в
+# test_scraper_kit_cian_golden_parity.py).
 os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
 
 # Kit — то, на что теперь ссылаются мигрированные файлы (#2306).
 from scraper_kit.base import ScrapedLot as KitScrapedLot
 from scraper_kit.cian_state_parser import extract_state as kit_extract_state
-from scraper_kit.providers.cian.detail import fetch_detail as kit_fetch_detail
 
-from app.core.config import settings
-from app.services.scraper_adapters import RealScraperConfig
-
-# Legacy — эталон сравнения (эти модули НЕ меняются в #2306).
+# Legacy — эталон сравнения (эти модули НЕ меняются в #2306, остаются live).
 from app.services.scrapers.base import ScrapedLot as LegacyScrapedLot
-from app.services.scrapers.cian_detail import fetch_detail as legacy_fetch_detail
 from app.services.scrapers.cian_state_parser import extract_state as legacy_extract_state
 from tests.support.parity import assert_parity
 
@@ -108,113 +107,3 @@ def test_scraped_lot_model_dump_parity() -> None:
         kit_fn=lambda kw: KitScrapedLot(**kw).model_dump(),
         fixtures=[kwargs],
     )
-
-
-# ── fetch_detail / DetailEnrichment parity (cian_price_history.py dependency) ──
-
-
-def _detail_html() -> str:
-    default_state = {
-        "offerData": {
-            "offer": {
-                "cianId": 555111,
-                "priceChanges": [
-                    {"changeTime": "2026-05-01T00:00:00Z", "price": 6100000, "diffPercent": -1.2}
-                ],
-            },
-            "stats": {},
-        }
-    }
-    ds_json = json.dumps(default_state, ensure_ascii=False)
-    return (
-        ""
-    )
-
-
-def _mock_session(html: str) -> MagicMock:
-    response = MagicMock()
-    response.status_code = 200
-    response.text = html
-    session = MagicMock()
-    session.get = AsyncMock(return_value=response)
-    session.close = AsyncMock()
-    return session
-
-
-def _run_legacy_fetch(url: str, html: str) -> Any:
-    return dataclasses.asdict(asyncio.run(legacy_fetch_detail(url, session=_mock_session(html))))
-
-
-def _run_kit_fetch(url: str, html: str) -> Any:
-    # Точно тот вызов, что теперь делает cian_price_history.backfill_cian_price_history:
-    # config=RealScraperConfig() добавлен по миграции (#2306); session= общий mock здесь
-    # только чтобы избежать сетевого I/O — сам config не участвует в session-пути, так
-    # что этот тест доказывает: инжекция config= не меняет результат парсинга.
-    result = asyncio.run(
-        kit_fetch_detail(url, config=RealScraperConfig(), session=_mock_session(html))
-    )
-    return dataclasses.asdict(result)
-
-
-def test_fetch_detail_with_config_injection_parity() -> None:
-    html = _detail_html()
-    url = "https://ekb.cian.ru/sale/flat/555111/"
-    assert_parity(
-        legacy_fn=_run_legacy_fetch,
-        kit_fn=_run_kit_fetch,
-        fixtures=[(url, html)],
-    )
-
-
-def test_fetch_detail_own_session_proxy_wiring_parity(monkeypatch: Any) -> None:
-    """cian_price_history calls fetch_detail WITHOUT session= (own-session curl_cffi path).
-
-    Kit fetch_detail without config= silently drops the proxy (direct connection) — a
-    real behaviour regression risk (#806: datacenter-IP gets banned by Cian without a
-    proxy). This proves config=RealScraperConfig() (our actual call site) constructs
-    AsyncSession(proxies=...) identically to the legacy direct-settings path.
-
-    #2362 (F4b): kit's own-session branch now builds the session via
-    `scraper_kit.providers._base.build_curl_cffi_session()` instead of constructing
-    `AsyncSession` directly in `detail.py` — so the spy patches `AsyncSession` where
-    the constructor call now actually happens (`_base`), not in `detail` anymore.
-    The asserted kwargs (proxies=...) are unchanged.
-    """
-    monkeypatch.setattr(settings, "cian_proxy_url_env", "http://test-proxy.local:8080")
-
-    captured: dict[str, dict[str, Any]] = {}
-
-    def _make_fake_session(name: str) -> Any:
-        class _FakeSession:
-            def __init__(self, **kwargs: Any) -> None:
-                captured[name] = kwargs
-
-            async def get(self, *_args: Any, **_kwargs: Any) -> Any:
-                response = MagicMock()
-                response.status_code = 200
-                response.text = _detail_html()
-                return response
-
-            async def close(self) -> None:
-                pass
-
-        return _FakeSession
-
-    url = "https://ekb.cian.ru/sale/flat/1/"
-
-    with patch("app.services.scrapers.cian_detail.AsyncSession", _make_fake_session("legacy")):
-        asyncio.run(legacy_fetch_detail(url))
-
-    with patch("scraper_kit.providers._base.AsyncSession", _make_fake_session("kit")):
-        asyncio.run(kit_fetch_detail(url, config=RealScraperConfig()))
-
-    assert captured["legacy"].get("proxies") == captured["kit"].get("proxies")
-    assert captured["kit"]["proxies"] == {
-        "http": "http://test-proxy.local:8080",
-        "https": "http://test-proxy.local:8080",
-    }
diff --git a/tradein-mvp/backend/tests/test_scraper_kit_yandex_golden_parity.py b/tradein-mvp/backend/tests/test_scraper_kit_yandex_golden_parity.py
index 85bfb7b2..eec7fd3f 100644
--- a/tradein-mvp/backend/tests/test_scraper_kit_yandex_golden_parity.py
+++ b/tradein-mvp/backend/tests/test_scraper_kit_yandex_golden_parity.py
@@ -1,4 +1,5 @@
-"""Golden-parity: `scraper_kit.providers.yandex.*` парсинг ≡ `app.services.scrapers.yandex_*`.
+"""Golden-parity: `scraper_kit.providers.yandex.*` парсинг ≡ ещё живым
+`app.services.scrapers.yandex_*`.
 
 Strangler-инвариант (#2133): новая scraper_kit-копия yandex-провайдера должна давать
 БАЙТ-ИДЕНТИЧНЫЙ результат парсинга старому боевому коду на одинаковом входе.
@@ -6,34 +7,34 @@ Strangler-инвариант (#2133): новая scraper_kit-копия yandex-
 house_type_normalizer→providers.yandex.shared, base/browser_fetcher/helpers→scraper_kit.*)
 НЕ меняет распарсенные данные.
 
-Гоняем ОБА модуля на одних фикстурах и сравниваем результат:
-  - SERP `_entity_to_lot` (gate-API entity → ScrapedLot) на репрезентативных entity;
-  - SERP `_parse_gate_json` (payload → list[ScrapedLot]) с новостройкой/вторичкой;
-  - `normalize_house_type` (SCREAMING/camelCase → канон);
-  - newbuilding `parse` (ЖК-лендинг → YandexNewbuildingInfo) на синтетике;
-  - `_build_url` (gate-API URL builder) — детерминированность.
+`app.services.scrapers.yandex_realty` удалён (#2397 Part E2, 0 runtime-импортёров).
+SERP `_entity_to_lot`/`_parse_gate_json`/`_build_url` parity-тесты (сравнение с
+`YandexRealtyScraper`) удалены вместе с ним (легаси-эталон исчез).
 
-Идентичность результата = kit-парсер верно скопирован (развязка не задела логику).
+Оставшееся сравнивает kit ≡ ЕЩЁ ЖИВЫЕ легаси-модули:
+  - `normalize_house_type` — `app.services.scrapers.house_type_normalizer` (живой);
+  - newbuilding `parse` (ЖК-лендинг → YandexNewbuildingInfo) —
+    `app.services.scrapers.yandex_newbuilding` (живой);
+  - strangler-guard: kit-провайдер не импортирует `app.*`.
 
 Легаси `app.services.scrapers.yandex_detail` / `yandex_valuation` удалены (#2277
 финальный шаг scraper_kit-миграции, 0 runtime importers) — detail/valuation
-parity-проверки убраны отсюда вместе с ними. Kit-side coverage для detail/valuation
+parity-проверки убраны отсюда ранее. Kit-side coverage для detail/valuation
 `.parse()` сохранена без изменений в `tests/test_yandex_detail.py` /
 `tests/test_yandex_detail_structural.py` / `tests/test_yandex_valuation.py` /
 `tests/test_yandex_valuation_save.py` (переведены на kit-эквиваленты тем же PR).
-`yandex_realty` / `yandex_newbuilding` / `house_type_normalizer` остаются легаси
-(не мигрированы) — их parity-проверки не тронуты.
+
+Идентичность результата = kit-парсер верно скопирован (развязка не задела логику).
 """
 
 from __future__ import annotations
 
 import inspect
 import os
-from types import SimpleNamespace
-from typing import Any
 
-# Старый app.services.scrapers.yandex_* импортирует app.core.config.settings=Settings(),
-# которому нужен DATABASE_URL. Офлайн-парсинг БД не трогает — фиктивный DSN достаточен.
+# Старый app.services.scrapers.house_type_normalizer/yandex_newbuilding импортирует
+# app.core.config.settings=Settings(), которому нужен DATABASE_URL. Офлайн-парсинг
+# БД не трогает — фиктивный DSN достаточен.
 os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/test_db")
 
 # НОВЫЙ (scraper_kit) и СТАРЫЙ (app.services.scrapers) yandex-парсеры — оба в одном
@@ -41,9 +42,6 @@ os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost/
 from scraper_kit.providers.yandex.newbuilding import (
     YandexNewbuildingScraper as NewNewbuildingScraper,
 )
-from scraper_kit.providers.yandex.serp import YandexRealtyScraper as NewRealtyScraper
-from scraper_kit.providers.yandex.serp import _entity_to_lot as new_entity_to_lot
-from scraper_kit.providers.yandex.serp import _parse_gate_json as new_parse_gate_json
 from scraper_kit.providers.yandex.shared import normalize_house_type as new_normalize_house_type
 
 from app.services.scrapers.house_type_normalizer import (
@@ -52,192 +50,6 @@ from app.services.scrapers.house_type_normalizer import (
 from app.services.scrapers.yandex_newbuilding import (
     YandexNewbuildingScraper as OldNewbuildingScraper,
 )
-from app.services.scrapers.yandex_realty import YandexRealtyScraper as OldRealtyScraper
-from app.services.scrapers.yandex_realty import _entity_to_lot as old_entity_to_lot
-from app.services.scrapers.yandex_realty import _parse_gate_json as old_parse_gate_json
-
-# Инжектируемый конфиг для kit-скрапперов (duck-typed namespace, structurally
-# удовлетворяет scraper_kit.contracts.ScraperConfig в объёме читаемых полей).
-_KIT_CONFIG = SimpleNamespace(
-    scraper_proxy_url=None,
-    yandex_proxy_rotate_url=None,
-    avito_proxy_rotate_url=None,
-)
-
-
-# ── Fixtures: gate-API entities (mirror test_yandex_realty_serp) ─────────────
-
-_ENTITY_FULL: dict = {
-    "offerId": 4740475460451078271,
-    "url": "//realty.yandex.ru/offer/4740475460451078271",
-    "price": {"currency": "RUR", "value": 4500000, "valuePerPart": 119048},
-    "area": {"value": 37.8, "unit": "SQUARE_METER"},
-    "livingSpace": {"value": 20, "unit": "SQUARE_METER"},
-    "kitchenSpace": {"value": 12.7, "unit": "SQUARE_METER"},
-    "roomsTotal": 1,
-    "floorsOffered": [8],
-    "floorsTotal": 25,
-    "ceilingHeight": 2.7,
-    "building": {
-        "builtYear": 2016,
-        "buildingType": "MONOLIT",
-        "siteId": 280938,
-        "siteName": "Peremen",
-    },
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Evgenia Savkova, 8",
-        "point": {"latitude": 56.79512, "longitude": 60.49186},
-    },
-    "mainImages": [
-        "//avatars.mds.yandex.net/get-realty-offers/11904018/abc/main",
-        "//avatars.mds.yandex.net/get-realty-offers/14717586/def/main",
-    ],
-}
-
-_ENTITY_STUDIO: dict = {
-    "offerId": 9999999,
-    "url": "//realty.yandex.ru/offer/9999999",
-    "price": {"value": 3200000, "valuePerPart": 114285},
-    "area": {"value": 28.0},
-    "livingSpace": None,
-    "kitchenSpace": None,
-    "roomsTotal": None,  # studio -> rooms=0
-    "floorsOffered": [2],
-    "floorsTotal": 9,
-    "ceilingHeight": None,
-    "building": {"builtYear": 2010, "buildingType": "PANEL", "siteId": None, "siteName": None},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Lenina, 5",
-        "point": {"latitude": 56.838, "longitude": 60.597},
-    },
-    "mainImages": [],
-}
-
-_ENTITY_RICH_OWNER: dict = {
-    "offerId": 5500000000000000001,
-    "url": "//realty.yandex.ru/offer/5500000000000000001",
-    "creationDate": "2026-05-24T15:10:27Z",
-    "description": "Продаётся светлая квартира с видом на парк.",
-    "author": {"category": "OWNER", "agentName": "Иван", "organization": None},
-    "price": {"value": 7200000, "valuePerPart": 150000, "trend": "DECREASED", "previous": 7500000},
-    "predictions": {"predictedPrice": {"value": "7554000", "min": "7100000", "max": "8000000"}},
-    "area": {"value": 48.0},
-    "roomsTotal": 2,
-    "floorsOffered": [5],
-    "floorsTotal": 12,
-    "building": {"builtYear": 2018, "buildingType": "MONOLIT"},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Mira, 10",
-        "point": {"latitude": 56.84, "longitude": 60.61},
-        "metroList": [
-            {"name": "Ploshchad 1905 goda", "timeToMetro": 7},
-            {"name": "Geologicheskaya", "timeToMetro": 12},
-        ],
-    },
-    "mainImages": [],
-}
-
-_ENTITY_RICH_AGENCY: dict = {
-    "offerId": 5500000000000000002,
-    "url": "//realty.yandex.ru/offer/5500000000000000002",
-    "creationDate": "2026-06-01T09:00:00Z",
-    "description": "Агентская продажа.",
-    "author": {"category": "AGENCY", "agentName": "Петров", "organization": "Агентство «Диал»"},
-    "price": {"value": 9900000, "trend": "UNCHANGED"},
-    "area": {"value": 60.0},
-    "roomsTotal": 3,
-    "floorsOffered": [3],
-    "floorsTotal": 9,
-    "building": {"builtYear": 2012, "buildingType": "BRICK"},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Lenina, 1",
-        "point": {"latitude": 56.83, "longitude": 60.6},
-        "metro": {"name": "Dinamo", "timeToMetro": 4},
-    },
-    "mainImages": [],
-}
-
-# #2235: novostroyka offer whose gate-API `url` is a developer/redirect landing
-# (shared by many flats). Both parsers must canonicalise it to offer_id-based url.
-_ENTITY_PARTNER_URL: dict = {
-    "offerId": 7416316697684447350,
-    "url": "http://na100.pro/go.php?link=LtWgA6kWd3xjlrUBCLl6VGjo",
-    "price": {"value": 7790000, "valuePerPart": 130000},
-    "area": {"value": 60.0},
-    "roomsTotal": 2,
-    "floorsOffered": [5],
-    "floorsTotal": 16,
-    "building": {"builtYear": 2024, "buildingType": "MONOLIT", "siteId": 999001},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Titova, 1",
-        "point": {"latitude": 56.8, "longitude": 60.6},
-    },
-    "mainImages": [],
-}
-
-_ALL_ENTITIES = [
-    _ENTITY_FULL,
-    _ENTITY_STUDIO,
-    _ENTITY_RICH_OWNER,
-    _ENTITY_RICH_AGENCY,
-    _ENTITY_PARTNER_URL,
-]
-
-
-def _make_gate_payload(entities: list, pager: dict | None = None) -> dict:
-    if pager is None:
-        pager = {"page": 0, "pageSize": 20, "totalItems": len(entities), "totalPages": 1}
-    return {"response": {"search": {"offers": {"entities": entities, "pager": pager}}}}
-
-
-def _dump_lots(lots: list[Any]) -> list[dict[str, Any]]:
-    return [lot.model_dump() for lot in lots]
-
-
-# ── SERP _entity_to_lot parity ───────────────────────────────────────────────
-
-
-def test_entity_to_lot_parity_all_entities() -> None:
-    """gate-API entity → ScrapedLot идентичен на всех репрезентативных entity."""
-    for entity in _ALL_ENTITIES:
-        old_lot = old_entity_to_lot(entity)
-        new_lot = new_entity_to_lot(entity)
-        assert (old_lot is None) == (new_lot is None)
-        assert old_lot is not None and new_lot is not None
-        assert old_lot.model_dump() == new_lot.model_dump()
-
-
-def test_entity_to_lot_parity_new_flat_segments() -> None:
-    """new_flat=YES/NO → listing_segment идентичен (новостройка/вторичка)."""
-    for new_flat in ("NO", "YES"):
-        old_lot = old_entity_to_lot(_ENTITY_FULL, new_flat=new_flat)
-        new_lot = new_entity_to_lot(_ENTITY_FULL, new_flat=new_flat)
-        assert old_lot is not None and new_lot is not None
-        assert old_lot.model_dump() == new_lot.model_dump()
-
-
-def test_entity_to_lot_parity_ceiling_out_of_range() -> None:
-    """ceilingHeight мусор (18 / 1.6) → одинаково дропается в None (#2007)."""
-    for bad_ceiling in (18, 1.6):
-        entity = {**_ENTITY_FULL, "ceilingHeight": bad_ceiling}
-        old_lot = old_entity_to_lot(entity)
-        new_lot = new_entity_to_lot(entity)
-        assert old_lot is not None and new_lot is not None
-        assert old_lot.model_dump() == new_lot.model_dump()
-
-
-# ── SERP _parse_gate_json parity ─────────────────────────────────────────────
-
-
-def test_parse_gate_json_parity() -> None:
-    """payload → list[ScrapedLot] идентичен (вторичка и новостройка)."""
-    payload = _make_gate_payload(_ALL_ENTITIES)
-    for new_flat in ("NO", "YES"):
-        old_lots = old_parse_gate_json(payload, page_param=2, new_flat=new_flat)
-        new_lots = new_parse_gate_json(payload, page_param=2, new_flat=new_flat)
-        assert len(old_lots) == len(new_lots) > 0
-        assert _dump_lots(old_lots) == _dump_lots(new_lots)
-
 
 # ── normalize_house_type parity ──────────────────────────────────────────────
 
@@ -296,23 +108,6 @@ def test_newbuilding_parse_parity() -> None:
     assert old_res.model_dump() == new_res.model_dump()
 
 
-# ── _build_url parity ────────────────────────────────────────────────────────
-
-
-def test_build_url_parity() -> None:
-    """gate-API URL builder идентичен на репрезентативных комбо."""
-    old_scraper = OldRealtyScraper()
-    new_scraper = NewRealtyScraper(config=_KIT_CONFIG)
-    combos = [
-        dict(page=1),
-        dict(page=3, rooms="2", price_min=5_000_000, price_max=7_000_000),
-        dict(page=1, rooms="studio", new_flat="YES"),
-        dict(page=2, rooms="4+", price_min=25_000_000),
-    ]
-    for combo in combos:
-        assert old_scraper._build_url(**combo) == new_scraper._build_url(**combo)
-
-
 # ── strangler-guard: kit-провайдер не импортирует app.* ──────────────────────
 
 
diff --git a/tradein-mvp/backend/tests/test_scraper_proxy.py b/tradein-mvp/backend/tests/test_scraper_proxy.py
index 98bd116e..bb52143f 100644
--- a/tradein-mvp/backend/tests/test_scraper_proxy.py
+++ b/tradein-mvp/backend/tests/test_scraper_proxy.py
@@ -177,30 +177,13 @@ _EXPECTED_PROXIES = {"http": _PROXY_URL, "https": _PROXY_URL}
 
 
 # ── YandexRealtyScraper ──────────────────────────────────────────────────────
-
-
-@pytest.mark.asyncio
-async def test_yandex_realty_opens_browser_fetcher(monkeypatch):
-    """YandexRealtyScraper.__aenter__ opens ONE BrowserFetcher (camoufox) session.
-
-    Transport moved off the proxied curl_cffi session to BrowserFetcher (camoufox);
-    the mobile proxy is applied server-side by the tradein-browser service, not via a
-    proxies= kwarg here.
-    """
-    from app.services.scrapers.yandex_realty import YandexRealtyScraper
-
-    mock_browser = MagicMock()
-    mock_browser.__aenter__ = AsyncMock(return_value=mock_browser)
-    mock_browser.__aexit__ = AsyncMock(return_value=None)
-
-    monkeypatch.setattr(
-        "app.services.scrapers.yandex_realty.BrowserFetcher", lambda *a, **k: mock_browser
-    )
-    scraper = YandexRealtyScraper()
-    await scraper.__aenter__()
-
-    mock_browser.__aenter__.assert_awaited_once()
-    assert scraper._browser is mock_browser
+#
+# `test_yandex_realty_opens_browser_fetcher` (legacy `app.services.scrapers.
+# yandex_realty.YandexRealtyScraper.__aenter__` opens ONE BrowserFetcher session)
+# removed (#2397 Part E2) — `yandex_realty` deleted, 0 runtime importers. Kit
+# equivalent already covered (opens exactly one BrowserFetcher session via
+# `build_browser_fetcher`) by `test_kit_serp_proxy_pool.py::
+# test_yandex_serp_no_pool_by_default` (and the pool-wiring variants alongside it).
 
 
 # ── YandexValuationScraper ───────────────────────────────────────────────────
diff --git a/tradein-mvp/backend/tests/test_snapshot_writer.py b/tradein-mvp/backend/tests/test_snapshot_writer.py
index b259a78d..8c023c64 100644
--- a/tradein-mvp/backend/tests/test_snapshot_writer.py
+++ b/tradein-mvp/backend/tests/test_snapshot_writer.py
@@ -9,14 +9,23 @@
 5. Ошибка snapshot не прерывает основной INSERT в save_listings.
 
 Используем unittest.mock — без реальной БД.
+
+`app.services.scrapers.cian_detail` удалён (#2397 Part E2, 0 runtime-импортёров) —
+`DetailEnrichment`/`save_detail_enrichment` ретаргетированы на
+`scraper_kit.providers.cian.detail` (byte-identical strangler-копия; `snapshot_writer`
+тестируется по-прежнему через живой `app.services.scrapers.snapshot_writer` —
+kit и app-версии этого модуля идентичны, kit's `save_detail_enrichment` импортирует
+`upsert_listing_snapshot` из `scraper_kit.snapshot_writer`).
 """
+
 from __future__ import annotations
 
 from contextlib import contextmanager
 from datetime import date
 from unittest.mock import MagicMock, patch
 
-from app.services.scrapers.cian_detail import DetailEnrichment, save_detail_enrichment
+from scraper_kit.providers.cian.detail import DetailEnrichment, save_detail_enrichment
+
 from app.services.scrapers.snapshot_writer import upsert_listing_snapshot
 
 # ── Helpers ───────────────────────────────────────────────────────────────────
@@ -220,7 +229,7 @@ def test_save_detail_enrichment_snapshot_failure_does_not_abort():
     db = _mock_db_detail(price_rub=4_000_000)
 
     with patch(
-        "app.services.scrapers.cian_detail.upsert_listing_snapshot",
+        "scraper_kit.providers.cian.detail.upsert_listing_snapshot",
         side_effect=RuntimeError("db snapshot blip"),
     ):
         enrichment = DetailEnrichment(ceiling_height=2.5)
@@ -261,9 +270,9 @@ def test_save_detail_enrichment_oph_on_conflict_constraint():
     sqls = _get_all_sqls(db)
     oph_sqls = [s for s in sqls if "offer_price_history" in s]
     assert oph_sqls, "INSERT offer_price_history не найден"
-    assert "offer_price_history_listing_change_uq" in oph_sqls[0], (
-        "ON CONFLICT должен ссылаться на UNIQUE constraint"
-    )
+    assert (
+        "offer_price_history_listing_change_uq" in oph_sqls[0]
+    ), "ON CONFLICT должен ссылаться на UNIQUE constraint"
 
 
 def test_save_detail_enrichment_skips_price_change_without_change_time():
@@ -271,7 +280,7 @@ def test_save_detail_enrichment_skips_price_change_without_change_time():
     db = _mock_db_detail(price_rub=5_000_000)
     enrichment = DetailEnrichment(
         price_changes=[
-            {"change_time": None, "price_rub": 5_000_000},       # пропустить
+            {"change_time": None, "price_rub": 5_000_000},  # пропустить
             {"change_time": "2026-04-01T00:00:00Z", "price_rub": None},  # пропустить
             {"change_time": "2026-04-01T00:00:00Z", "price_rub": 5_000_000},  # записать
         ]
diff --git a/tradein-mvp/backend/tests/test_yandex_city_sweep.py b/tradein-mvp/backend/tests/test_yandex_city_sweep.py
index df2fbccc..b50729a9 100644
--- a/tradein-mvp/backend/tests/test_yandex_city_sweep.py
+++ b/tradein-mvp/backend/tests/test_yandex_city_sweep.py
@@ -11,7 +11,9 @@ Part E1) — эквивалентная regression-coverage теперь в
 
 Coverage:
   - YandexCitySweepCounters корректно заполняются (dataclass shape)
-  - Combos watchdog timeout существенно больше ANCHOR_TIMEOUT_SEC (arithmetic-only)
+  - Combos watchdog timeout существенно больше ANCHOR_TIMEOUT_SEC (arithmetic-only;
+    `DEFAULT_PRICE_RANGES`/`ROOM_PATH` — kit `scraper_kit.providers.yandex.serp`,
+    ретаргетировано с удалённого `app.services.scrapers.yandex_realty`, #2397 Part E2)
   - Admin endpoints (start/cancel/list-runs) — kit sweep вызван с DI
     (config/matcher/enrichment)
 """
@@ -194,8 +196,7 @@ def test_combos_sweep_timeout_substantially_larger_than_anchor_timeout() -> None
         _YANDEX_COMBOS_PER_FETCH_S,
         ANCHOR_TIMEOUT_SEC,
     )
-
-    from app.services.scrapers.yandex_realty import DEFAULT_PRICE_RANGES, ROOM_PATH
+    from scraper_kit.providers.yandex.serp import DEFAULT_PRICE_RANGES, ROOM_PATH
 
     # Параметры prod-режима
     rooms_list = list(ROOM_PATH.keys())  # 5 комнатностей
diff --git a/tradein-mvp/backend/tests/test_yandex_realty_serp.py b/tradein-mvp/backend/tests/test_yandex_realty_serp.py
deleted file mode 100644
index 1fdf0d40..00000000
--- a/tradein-mvp/backend/tests/test_yandex_realty_serp.py
+++ /dev/null
@@ -1,1363 +0,0 @@
-"""Unit tests for YandexRealtyScraper gate-API JSON parser.
-
-Replaced DOM/HTML SERP tests (2026-06-17): new transport uses
-GET realty.yandex.ru/gate/react-page/get/ (JSON gate-API) instead of HTML SERP.
-"""
-
-import json
-from unittest.mock import AsyncMock, MagicMock, patch
-
-import pytest
-
-from app.services.scrapers.base import ScrapedLot
-from app.services.scrapers.yandex_realty import (
-    _GATE_MAX_PAGES_CAP,
-    DEFAULT_CITY,
-    DEFAULT_PRICE_RANGES,
-    ROOM_PATH,
-    YandexRealtyScraper,
-    _canonical_source_url,
-    _combo_label,
-    _CurlResponse,
-    _entity_to_lot,
-    _extract_gate_data,
-    _extract_json_from_content,
-    _is_gate_error,
-    _parse_gate_json,
-)
-
-# ---------------------------------------------------------------------------
-# Minimal gate-API JSON fixtures (derived from real captured sample 2026-06-17)
-# pager shape: {page, pageSize, totalItems, totalPages}
-# ---------------------------------------------------------------------------
-
-_ENTITY_FULL: dict = {
-    "offerId": 4740475460451078271,
-    "url": "//realty.yandex.ru/offer/4740475460451078271",
-    "price": {
-        "currency": "RUR",
-        "value": 4500000,
-        "valuePerPart": 119048,
-    },
-    "area": {"value": 37.8, "unit": "SQUARE_METER"},
-    "livingSpace": {"value": 20, "unit": "SQUARE_METER"},
-    "kitchenSpace": {"value": 12.7, "unit": "SQUARE_METER"},
-    "roomsTotal": 1,
-    "floorsOffered": [8],
-    "floorsTotal": 25,
-    "ceilingHeight": 2.7,
-    "building": {
-        "builtYear": 2016,
-        "buildingType": "MONOLIT",
-        "siteId": 280938,
-        "siteName": "Peremen",
-    },
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Evgenia Savkova, 8",
-        "point": {"latitude": 56.79512, "longitude": 60.49186},
-    },
-    "mainImages": [
-        "//avatars.mds.yandex.net/get-realty-offers/11904018/abc/main",
-        "//avatars.mds.yandex.net/get-realty-offers/14717586/def/main",
-    ],
-}
-
-_ENTITY_STUDIO: dict = {
-    "offerId": 9999999,
-    "url": "//realty.yandex.ru/offer/9999999",
-    "price": {"value": 3200000, "valuePerPart": 114285},
-    "area": {"value": 28.0},
-    "livingSpace": None,
-    "kitchenSpace": None,
-    "roomsTotal": None,  # studio -> rooms=0
-    "floorsOffered": [2],
-    "floorsTotal": 9,
-    "ceilingHeight": None,
-    "building": {"builtYear": 2010, "buildingType": "PANEL", "siteId": None, "siteName": None},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Lenina, 5",
-        "point": {"latitude": 56.838, "longitude": 60.597},
-    },
-    "mainImages": [],
-}
-
-_ENTITY_NO_PRICE: dict = {
-    "offerId": 1111,
-    "url": "//realty.yandex.ru/offer/1111",
-    "price": {"value": 0},
-    "area": {"value": 40.0},
-    "roomsTotal": 1,
-    "floorsOffered": [3],
-    "floorsTotal": 5,
-    "building": {},
-    "location": {"point": {"latitude": 56.8, "longitude": 60.6}},
-    "mainImages": [],
-}
-
-_ENTITY_NO_OFFER_ID: dict = {
-    "offerId": None,
-    "price": {"value": 5000000},
-    "area": {"value": 50.0},
-}
-
-
-# Entity carrying the rich gate-API fields (creationDate, description, author,
-# predictedPrice, price.trend/previous, location.metroList) — verified live 2026-06-17.
-_ENTITY_RICH_OWNER: dict = {
-    "offerId": 5500000000000000001,
-    "url": "//realty.yandex.ru/offer/5500000000000000001",
-    "creationDate": "2026-05-24T15:10:27Z",
-    "description": "Продаётся светлая квартира с видом на парк.",
-    "author": {"category": "OWNER", "agentName": "Иван", "organization": None},
-    "price": {
-        "value": 7200000,
-        "valuePerPart": 150000,
-        "trend": "DECREASED",
-        "previous": 7500000,
-    },
-    # Real gate-API nests valuation under predictions.predictedPrice (verified
-    # live 2026-06-18) — NOT predictedPrice.predictedPrice.
-    "predictions": {"predictedPrice": {"value": "7554000", "min": "7100000", "max": "8000000"}},
-    "area": {"value": 48.0},
-    "roomsTotal": 2,
-    "floorsOffered": [5],
-    "floorsTotal": 12,
-    "building": {"builtYear": 2018, "buildingType": "MONOLIT"},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Mira, 10",
-        "point": {"latitude": 56.84, "longitude": 60.61},
-        "metroList": [
-            {"name": "Ploshchad 1905 goda", "timeToMetro": 7},
-            {"name": "Geologicheskaya", "timeToMetro": 12},
-        ],
-    },
-    "mainImages": [],
-}
-
-_ENTITY_RICH_AGENCY: dict = {
-    "offerId": 5500000000000000002,
-    "url": "//realty.yandex.ru/offer/5500000000000000002",
-    "creationDate": "2026-06-01T09:00:00Z",
-    "description": "Агентская продажа.",
-    "author": {"category": "AGENCY", "agentName": "Петров", "organization": "Агентство «Диал»"},
-    "price": {"value": 9900000, "trend": "UNCHANGED"},
-    "area": {"value": 60.0},
-    "roomsTotal": 3,
-    "floorsOffered": [3],
-    "floorsTotal": 9,
-    "building": {"builtYear": 2012, "buildingType": "BRICK"},
-    "location": {
-        "geocoderAddress": "Ekaterinburg, ulitsa Lenina, 1",
-        "point": {"latitude": 56.83, "longitude": 60.6},
-        "metro": {"name": "Dinamo", "timeToMetro": 4},
-    },
-    "mainImages": [],
-}
-
-
-def _make_gate_payload(entities: list, pager: dict | None = None) -> dict:
-    """Wrap entities + pager into gate-API response shape."""
-    if pager is None:
-        pager = {"page": 0, "pageSize": 20, "totalItems": len(entities), "totalPages": 1}
-    return {
-        "response": {
-            "search": {
-                "offers": {
-                    "entities": entities,
-                    "pager": pager,
-                }
-            }
-        }
-    }
-
-
-# ---------------------------------------------------------------------------
-# PART A: pure helper functions
-# ---------------------------------------------------------------------------
-
-
-def test_is_gate_error_on_error_key():
-    assert _is_gate_error({"error": "bad request"}) is True
-
-
-def test_is_gate_error_on_missing_response():
-    assert _is_gate_error({"something": "else"}) is True
-
-
-def test_is_gate_error_on_valid_payload():
-    payload = _make_gate_payload([_ENTITY_FULL])
-    assert _is_gate_error(payload) is False
-
-
-def test_extract_gate_data_returns_entities_and_pager():
-    payload = _make_gate_payload([_ENTITY_FULL, _ENTITY_STUDIO])
-    result = _extract_gate_data(payload)
-    assert result is not None
-    entities, pager = result
-    assert len(entities) == 2
-    assert "totalItems" in pager
-
-
-def test_extract_gate_data_returns_none_on_bad_shape():
-    assert _extract_gate_data({}) is None
-    assert _extract_gate_data({"response": {}}) is None
-    assert _extract_gate_data({"response": {"search": {}}}) is None
-
-
-# ---------------------------------------------------------------------------
-# PART B: _entity_to_lot field mapping
-# ---------------------------------------------------------------------------
-
-
-def test_entity_to_lot_full_entity():
-    lot = _entity_to_lot(_ENTITY_FULL)
-    assert lot is not None
-    assert lot.source == "yandex"
-    assert lot.source_id == "4740475460451078271"
-    assert lot.source_url == "https://realty.yandex.ru/offer/4740475460451078271"
-    assert lot.price_rub == 4_500_000
-    assert lot.price_per_m2 == 119_048
-    assert lot.area_m2 == pytest.approx(37.8)
-    assert lot.living_area_m2 == pytest.approx(20.0)
-    assert lot.rooms == 1
-    assert lot.floor == 8
-    assert lot.total_floors == 25
-    assert lot.year_built == 2016
-    assert lot.house_type == "monolith"  # normalize_house_type(MONOLIT) — #2007
-    assert lot.kitchen_area_m2 == pytest.approx(12.7)  # promoted to column — #2007
-    assert lot.ceiling_height_m == pytest.approx(2.7)  # promoted to column — #2007
-    assert lot.lat == pytest.approx(56.79512)
-    assert lot.lon == pytest.approx(60.49186)
-    assert lot.address == "Ekaterinburg, ulitsa Evgenia Savkova, 8"
-    assert lot.house_source == "yandex_realty_nb"
-    assert lot.house_ext_id == "280938"
-    assert lot.listing_segment == "vtorichka"
-    assert len(lot.photo_urls) == 2
-    assert all(u.startswith("https://avatars.mds.yandex.net") for u in lot.photo_urls)
-    assert lot.raw_payload is not None
-    assert lot.raw_payload["ceiling_height"] == pytest.approx(2.7)
-    assert lot.raw_payload["kitchen_area_m2"] == pytest.approx(12.7)
-    assert lot.raw_payload["raw_building_type"] == "MONOLIT"  # сырой SCREAMING сохранён
-
-
-def test_entity_to_lot_ceiling_out_of_range_dropped():
-    """ceiling_height — numeric(3,2) (max 9.99); неправдоподобный yandex SERP мусор
-    (видели 18 м) отсекаем в None, иначе DataError уронил бы весь батч. Сырое
-    значение остаётся в raw_payload. (#2007)"""
-    lot_hi = _entity_to_lot({**_ENTITY_FULL, "ceilingHeight": 18})
-    assert lot_hi is not None
-    assert lot_hi.ceiling_height_m is None
-    assert lot_hi.raw_payload["ceiling_height"] == 18  # сырое сохранено
-    lot_lo = _entity_to_lot({**_ENTITY_FULL, "ceilingHeight": 1.6})
-    assert lot_lo is not None
-    assert lot_lo.ceiling_height_m is None
-
-
-def test_entity_to_lot_segment_vtorichka_default():
-    """Default new_flat='NO' -> listing_segment='vtorichka'."""
-    lot = _entity_to_lot(_ENTITY_FULL)
-    assert lot is not None
-    assert lot.listing_segment == "vtorichka"
-
-
-def test_entity_to_lot_segment_novostroyki_on_new_flat_yes():
-    """new_flat='YES' -> listing_segment='novostroyki'."""
-    lot = _entity_to_lot(_ENTITY_FULL, new_flat="YES")
-    assert lot is not None
-    assert lot.listing_segment == "novostroyki"
-
-
-def test_parse_gate_json_propagates_segment():
-    """_parse_gate_json threads new_flat down to each lot's listing_segment."""
-    payload = _make_gate_payload([_ENTITY_FULL, _ENTITY_STUDIO])
-    lots_nb = _parse_gate_json(payload, new_flat="YES")
-    assert lots_nb and all(lot.listing_segment == "novostroyki" for lot in lots_nb)
-    lots_vt = _parse_gate_json(payload, new_flat="NO")
-    assert lots_vt and all(lot.listing_segment == "vtorichka" for lot in lots_vt)
-
-
-def test_entity_to_lot_studio_rooms_zero():
-    lot = _entity_to_lot(_ENTITY_STUDIO)
-    assert lot is not None
-    assert lot.rooms == 0
-    assert lot.area_m2 == pytest.approx(28.0)
-    assert lot.floor == 2
-    assert lot.total_floors == 9
-    assert lot.house_source is None
-    assert lot.house_ext_id is None
-    assert lot.photo_urls == []
-
-
-def test_entity_to_lot_no_price_returns_none():
-    lot = _entity_to_lot(_ENTITY_NO_PRICE)
-    assert lot is None
-
-
-def test_entity_to_lot_no_offer_id_returns_none():
-    lot = _entity_to_lot(_ENTITY_NO_OFFER_ID)
-    assert lot is None
-
-
-def test_entity_to_lot_photo_url_prefix():
-    """mainImages starting with // must get https: prepended."""
-    lot = _entity_to_lot(_ENTITY_FULL)
-    assert lot is not None
-    # Both URLs must start with https:// (not protocol-relative //)
-    assert all(u.startswith("https://") for u in lot.photo_urls)
-    assert not any(u.startswith("//") for u in lot.photo_urls)
-
-
-def test_entity_to_lot_photo_capped_at_5():
-    entity = dict(_ENTITY_FULL)
-    entity["mainImages"] = [f"//avatars.mds.yandex.net/img{i}/main" for i in range(10)]
-    lot = _entity_to_lot(entity)
-    assert lot is not None
-    assert len(lot.photo_urls) == 5
-
-
-def test_entity_to_lot_url_fallback_when_missing():
-    """When url field absent, fall back to constructed /offer// URL."""
-    entity = dict(_ENTITY_FULL)
-    entity["url"] = None
-    lot = _entity_to_lot(entity)
-    assert lot is not None
-    expected_id = _ENTITY_FULL["offerId"]
-    assert lot.source_url == f"https://realty.yandex.ru/offer/{expected_id}/"
-
-
-# ---------------------------------------------------------------------------
-# #2235: source_url canonicalisation — partner/redirect urls of novostroyki
-# offers must NOT be stored verbatim (many flats of one ЖК share one url →
-# duplicate active source_url). They must become offer_id-based canonical urls.
-# ---------------------------------------------------------------------------
-
-
-def test_canonical_source_url_keeps_yandex_realty_url():
-    """A real //realty.yandex.ru/offer/ url is preserved verbatim (vtorichka)."""
-    assert (
-        _canonical_source_url("//realty.yandex.ru/offer/123", "123")
-        == "https://realty.yandex.ru/offer/123"
-    )
-
-
-@pytest.mark.parametrize(
-    "partner_url",
-    [
-        "http://na100.pro/go.php?link=LtWgA6kWd3xjlrUBCLl6VGjo",  # ad redirect
-        "https://atlasdevelopment.su/",  # ЖК landing shared by many flats
-        "https://unistroyrf.ru/",
-        "https://vtmn.ru/complexes/vitamin-kvartal-na-titova",
-        "",  # missing url
-    ],
-)
-def test_canonical_source_url_rewrites_partner_urls(partner_url):
-    """Non-realty.yandex partner/redirect urls become offer_id-based canonical urls."""
-    assert (
-        _canonical_source_url(partner_url, "777888999")
-        == "https://realty.yandex.ru/offer/777888999/"
-    )
-
-
-def test_entity_to_lot_novostroyka_partner_url_is_canonicalised():
-    """Two distinct offers sharing one developer url get DISTINCT canonical source_urls.
-
-    Reproduces the #2235 root cause: the gate-API `url` for novostroyki points to
-    the developer site (shared by many flats). Verbatim storage produced 204
-    active source_url duplicates. After canonicalisation each offer_id yields its
-    own unique url, so the duplication disappears at the source.
-    """
-    shared = "http://na100.pro/go.php?link=LtWgA6kWd3xjlrUBCLl6VGjo"
-    ent_a = {**_ENTITY_FULL, "offerId": 1111111111111111111, "url": shared}
-    ent_b = {**_ENTITY_FULL, "offerId": 2222222222222222222, "url": shared}
-    lot_a = _entity_to_lot(ent_a, new_flat="YES")
-    lot_b = _entity_to_lot(ent_b, new_flat="YES")
-    assert lot_a is not None and lot_b is not None
-    assert lot_a.source_url == "https://realty.yandex.ru/offer/1111111111111111111/"
-    assert lot_b.source_url == "https://realty.yandex.ru/offer/2222222222222222222/"
-    assert lot_a.source_url != lot_b.source_url
-
-
-def test_entity_to_lot_no_house_when_site_id_absent():
-    """Entity without siteId produces house_source=None, house_ext_id=None."""
-    entity = dict(_ENTITY_FULL)
-    entity["building"] = {"builtYear": 2015, "buildingType": "BRICK", "siteId": None}
-    lot = _entity_to_lot(entity)
-    assert lot is not None
-    assert lot.house_source is None
-    assert lot.house_ext_id is None
-
-
-def test_entity_to_lot_page_param_in_raw():
-    """page_param passed to _entity_to_lot appears in raw_payload."""
-    lot = _entity_to_lot(_ENTITY_FULL, page_param=3)
-    assert lot is not None
-    assert lot.raw_payload["page_param"] == 3
-
-
-# ---------------------------------------------------------------------------
-# PART B2: rich gate-API fields (predictedPrice, seller, dates, trend, metro)
-# ---------------------------------------------------------------------------
-
-
-def test_entity_to_lot_rich_owner_fields():
-    lot = _entity_to_lot(_ENTITY_RICH_OWNER)
-    assert lot is not None
-    # creationDate → listing_date / publish_date / days_on_market
-    assert lot.listing_date is not None
-    assert lot.listing_date.isoformat() == "2026-05-24"
-    assert lot.publish_date == lot.listing_date
-    assert lot.days_on_market is not None
-    assert lot.days_on_market >= 0
-    # description
-    assert lot.description == "Продаётся светлая квартира с видом на парк."
-    # author OWNER
-    assert lot.is_homeowner is True
-    assert lot.is_pro_seller is None
-    assert lot.agency_name is None
-    # predictedPrice
-    assert lot.predicted_price_rub == 7_554_000
-    assert lot.predicted_price_min == 7_100_000
-    assert lot.predicted_price_max == 8_000_000
-    # price trend / previous
-    assert lot.price_trend == "DECREASED"
-    assert lot.price_previous_rub == 7_500_000
-    # metro_stations from metroList
-    assert lot.metro_stations == [
-        {"name": "Ploshchad 1905 goda", "min": 7},
-        {"name": "Geologicheskaya", "min": 12},
-    ]
-    # yandex_offer_id
-    assert lot.yandex_offer_id == "5500000000000000001"
-
-
-def test_entity_to_lot_rich_agency_fields():
-    lot = _entity_to_lot(_ENTITY_RICH_AGENCY)
-    assert lot is not None
-    # AGENCY → pro seller + agency_name from organization
-    assert lot.is_pro_seller is True
-    assert lot.is_homeowner is None
-    assert lot.agency_name == "Агентство «Диал»"
-    # price.trend present, no previous → price_previous_rub None
-    assert lot.price_trend == "UNCHANGED"
-    assert lot.price_previous_rub is None
-    # no predictedPrice path → all None
-    assert lot.predicted_price_rub is None
-    assert lot.predicted_price_min is None
-    assert lot.predicted_price_max is None
-    # single location.metro fallback
-    assert lot.metro_stations == [{"name": "Dinamo", "min": 4}]
-
-
-def test_entity_to_lot_no_rich_fields_are_none():
-    """Base fixture lacks rich fields → all new fields default to None/[] (no crash)."""
-    lot = _entity_to_lot(_ENTITY_FULL)
-    assert lot is not None
-    assert lot.listing_date is None
-    assert lot.days_on_market is None
-    assert lot.description is None
-    assert lot.is_homeowner is None
-    assert lot.is_pro_seller is None
-    assert lot.agency_name is None
-    assert lot.predicted_price_rub is None
-    assert lot.price_trend is None
-    assert lot.price_previous_rub is None
-    assert lot.metro_stations == []
-    # yandex_offer_id is always populated from offerId
-    assert lot.yandex_offer_id == "4740475460451078271"
-
-
-def test_entity_to_lot_bad_creation_date_is_safe():
-    """Unparseable creationDate → listing_date/days_on_market None, no exception."""
-    entity = dict(_ENTITY_RICH_OWNER)
-    entity["creationDate"] = "not-a-date"
-    lot = _entity_to_lot(entity)
-    assert lot is not None
-    assert lot.listing_date is None
-    assert lot.days_on_market is None
-
-
-# ---------------------------------------------------------------------------
-# PART C: _parse_gate_json (list-level)
-# ---------------------------------------------------------------------------
-
-
-def test_parse_gate_json_returns_list():
-    payload = _make_gate_payload([_ENTITY_FULL, _ENTITY_STUDIO])
-    lots = _parse_gate_json(payload)
-    assert len(lots) == 2
-    assert all(isinstance(lot, ScrapedLot) for lot in lots)
-
-
-def test_parse_gate_json_skips_invalid():
-    """Entities that fail validation (no price, no id) are silently skipped."""
-    payload = _make_gate_payload([_ENTITY_FULL, _ENTITY_NO_PRICE, _ENTITY_NO_OFFER_ID])
-    lots = _parse_gate_json(payload)
-    assert len(lots) == 1
-    assert lots[0].source_id == str(_ENTITY_FULL["offerId"])
-
-
-def test_parse_gate_json_empty_entities():
-    payload = _make_gate_payload([])
-    lots = _parse_gate_json(payload)
-    assert lots == []
-
-
-def test_parse_gate_json_bad_payload():
-    """Error/missing-path payload returns empty list without raising."""
-    assert _parse_gate_json({}) == []
-    assert _parse_gate_json({"error": "bad"}) == []
-
-
-# ---------------------------------------------------------------------------
-# PART D: _build_url gate-API URL shape
-# ---------------------------------------------------------------------------
-
-
-def test_build_url_contains_required_params():
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1)
-    assert "realty.yandex.ru/gate/react-page/get/" in url
-    assert "rgid=559132" in url
-    assert "type=SELL" in url
-    assert "category=APARTMENT" in url
-    assert "newFlat=NO" in url
-    assert "_pageType=search" in url
-    assert "_providers=react-search-results-data" in url
-    assert "page=1" in url
-
-
-def test_build_url_page_1_based():
-    """page param is 1-based; page=1 is first page."""
-    s = YandexRealtyScraper()
-    url1 = s._build_url(page=1)
-    url2 = s._build_url(page=2)
-    assert "page=1" in url1
-    assert "page=2" in url2
-
-
-def test_build_url_rooms_studio():
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1, rooms="studio")
-    assert "roomsTotal=STUDIO" in url
-
-
-def test_build_url_rooms_1():
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1, rooms="1")
-    assert "roomsTotal=1" in url
-
-
-def test_build_url_rooms_4plus():
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1, rooms="4+")
-    assert "roomsTotal=4" in url
-
-
-def test_build_url_price_range():
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1, price_min=5_000_000, price_max=7_000_000)
-    assert "priceMin=5000000" in url
-    assert "priceMax=7000000" in url
-
-
-def test_build_url_unknown_rooms_no_rooms_param():
-    """Unknown rooms key must not produce a roomsTotal param."""
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1, rooms="99-room-penthouse")
-    assert "roomsTotal" not in url
-
-
-def test_build_url_combos_are_unique():
-    """All rooms x price-range combos produce distinct URLs."""
-    s = YandexRealtyScraper()
-    urls = set()
-    for rooms in ROOM_PATH.keys():
-        for lo, hi in DEFAULT_PRICE_RANGES:
-            urls.add(s._build_url(page=1, rooms=rooms, price_min=lo, price_max=hi))
-    expected = len(ROOM_PATH) * len(DEFAULT_PRICE_RANGES)
-    assert len(urls) == expected, f"URL collisions: {expected} combos but {len(urls)} unique"
-
-
-def test_build_url_new_flat_default_no():
-    """Default new_flat must keep newFlat=NO (vtorichka) — backward compat."""
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1)
-    assert "newFlat=NO" in url
-    assert "newFlat=YES" not in url
-
-
-def test_build_url_new_flat_yes():
-    """new_flat='YES' must produce newFlat=YES (novostroyki segment)."""
-    s = YandexRealtyScraper()
-    url = s._build_url(page=1, new_flat="YES")
-    assert "newFlat=YES" in url
-
-
-def test_default_city():
-    assert DEFAULT_CITY == "ekaterinburg"
-
-
-def test_gate_max_pages_cap():
-    assert _GATE_MAX_PAGES_CAP == 50
-
-
-# ---------------------------------------------------------------------------
-# PART E: _combo_label
-# ---------------------------------------------------------------------------
-
-
-def test_combo_label_none_none():
-    result = _combo_label(None, 0, 200_000_000)
-    assert isinstance(result, str)
-    assert "any" in result
-
-
-def test_combo_label_room_and_price():
-    result = _combo_label("1", 5_000_000, 7_000_000)
-    assert "1" in result
-    assert "5000000" in result
-    assert "7000000" in result
-
-
-def test_combo_label_different_ranges_differ():
-    k1 = _combo_label("1", 5_000_000, 7_000_000)
-    k2 = _combo_label("1", 7_000_000, 10_000_000)
-    k3 = _combo_label("2", 5_000_000, 7_000_000)
-    assert k1 != k2
-    assert k1 != k3
-
-
-# ---------------------------------------------------------------------------
-# PART F: transport layer
-# ---------------------------------------------------------------------------
-
-
-# camoufox renders the gate-API JSON wrapped in an HTML 
 (verified prod 2026-06-17):
-_CAMOUFOX_HTML_PREFIX = (
-    '
'
-)
-_CAMOUFOX_HTML_SUFFIX = "
"
-
-
-def _wrap_camoufox(json_text: str) -> str:
-    """Wrap a JSON string the way camoufox renders a JSON endpoint."""
-    return f"{_CAMOUFOX_HTML_PREFIX}{json_text}{_CAMOUFOX_HTML_SUFFIX}"
-
-
-def test_extract_json_from_content_unwraps_camoufox_pre():
-    """_extract_json_from_content must unwrap camoufox's 
{json}
 HTML.
-
-    Feeds the real prod-observed wrapper and asserts the extracted text parses
-    back to the original gate-API payload.
-    """
-    payload = _make_gate_payload([_ENTITY_FULL])
-    wrapped = _wrap_camoufox(json.dumps(payload))
-
-    extracted = _extract_json_from_content(wrapped)
-    assert extracted is not None
-    assert extracted.startswith('{"response"')
-    assert json.loads(extracted) == payload
-
-
-def test_extract_json_from_content_raw_json_fallback():
-    """Raw JSON (no HTML wrapper) is returned via the first-'{' fallback."""
-    raw = json.dumps(_make_gate_payload([]))
-    assert _extract_json_from_content(raw) == raw
-
-
-def test_extract_json_from_content_returns_none_when_no_json():
-    assert _extract_json_from_content("nope") is None
-
-
-@pytest.mark.asyncio
-async def test_aexit_closes_browser_and_session():
-    s = YandexRealtyScraper()
-    mock_browser = AsyncMock()
-    mock_session = AsyncMock()
-    s._browser = mock_browser
-    s._cffi_session = mock_session
-    await s.__aexit__(None, None, None)
-    mock_browser.__aexit__.assert_awaited_once()
-    mock_session.close.assert_awaited_once()
-    assert s._browser is None
-    assert s._cffi_session is None
-
-
-@pytest.mark.asyncio
-async def test_http_get_raises_when_no_browser():
-    """_http_get must raise RuntimeError outside async context manager (no browser)."""
-    s = YandexRealtyScraper()
-    with pytest.raises(RuntimeError, match="async context manager"):
-        await s._http_get("https://realty.yandex.ru/test/")
-
-
-@pytest.mark.asyncio
-async def test_http_get_unwraps_camoufox_html_to_json_body():
-    """_http_get fetches via BrowserFetcher and returns extracted JSON with status 200."""
-    s = YandexRealtyScraper()
-    payload = _make_gate_payload([_ENTITY_FULL])
-    wrapped = _wrap_camoufox(json.dumps(payload))
-
-    mock_browser = AsyncMock()
-    mock_browser.fetch = AsyncMock(return_value=wrapped)
-    s._browser = mock_browser
-
-    resp = await s._http_get("https://realty.yandex.ru/test/", timeout=30)
-
-    assert isinstance(resp, _CurlResponse)
-    assert resp.status_code == 200
-    assert json.loads(resp.text) == payload
-    mock_browser.fetch.assert_awaited_once_with("https://realty.yandex.ru/test/")
-
-
-@pytest.mark.asyncio
-async def test_http_get_status_0_on_fetch_exception():
-    """BrowserFetcher.fetch raising -> _CurlResponse(status_code=0) (transient failure)."""
-    s = YandexRealtyScraper()
-    mock_browser = AsyncMock()
-    mock_browser.fetch = AsyncMock(side_effect=RuntimeError("browser down"))
-    s._browser = mock_browser
-
-    resp = await s._http_get("https://realty.yandex.ru/test/")
-
-    assert resp.status_code == 0
-    assert resp.text == ""
-
-
-@pytest.mark.asyncio
-async def test_http_get_status_0_when_no_json_in_content():
-    """Content with no JSON blob -> _CurlResponse(status_code=0)."""
-    s = YandexRealtyScraper()
-    mock_browser = AsyncMock()
-    mock_browser.fetch = AsyncMock(return_value="blocked")
-    s._browser = mock_browser
-
-    resp = await s._http_get("https://realty.yandex.ru/test/")
-
-    assert resp.status_code == 0
-
-
-# ---------------------------------------------------------------------------
-# PART G: fetch_around — tarpit resilience
-# ---------------------------------------------------------------------------
-
-
-def _valid_gate_text(entity: dict | None = None) -> str:
-    ent = entity or _ENTITY_FULL
-    pager = {"page": 0, "pageSize": 20, "totalItems": 1, "totalPages": 1}
-    return json.dumps(_make_gate_payload([ent], pager))
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_returns_lots_on_200_json():
-    s = YandexRealtyScraper()
-    body = _valid_gate_text()
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        return _CurlResponse(status_code=200, text=body)
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-            lots = await s.fetch_around(lat=56.8, lon=60.6, rooms="1")
-
-    assert len(lots) == 1
-    assert lots[0].source_id == str(_ENTITY_FULL["offerId"])
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_rotates_and_retries_on_status_0():
-    """fetch_around rotates IP and retries on status_code==0 (tarpit)."""
-    s = YandexRealtyScraper()
-    call_count = 0
-    body = _valid_gate_text()
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        nonlocal call_count
-        call_count += 1
-        if call_count == 1:
-            return _CurlResponse(status_code=0, text="")
-        return _CurlResponse(status_code=200, text=body)
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "_rotate_ip", new_callable=AsyncMock) as mock_rotate:
-            with patch("asyncio.sleep", new_callable=AsyncMock):
-                with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-                    lots = await s.fetch_around(lat=56.84, lon=60.60, rooms="1")
-
-    assert mock_rotate.call_count == 1
-    assert call_count == 2
-    assert len(lots) >= 1
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_rotates_on_json_error():
-    """fetch_around rotates when response body is not valid JSON."""
-    s = YandexRealtyScraper()
-    call_count = 0
-    body = _valid_gate_text()
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        nonlocal call_count
-        call_count += 1
-        if call_count == 1:
-            return _CurlResponse(status_code=200, text="not json")
-        return _CurlResponse(status_code=200, text=body)
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "_rotate_ip", new_callable=AsyncMock) as mock_rotate:
-            with patch("asyncio.sleep", new_callable=AsyncMock):
-                with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-                    lots = await s.fetch_around(lat=56.84, lon=60.60, rooms="2")
-
-    assert mock_rotate.call_count == 1
-    assert len(lots) >= 1
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_exhausts_retries_returns_empty():
-    """fetch_around returns [] when all retries exhausted."""
-    s = YandexRealtyScraper()
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        return _CurlResponse(status_code=0, text="")
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "_rotate_ip", new_callable=AsyncMock) as mock_rotate:
-            with patch("asyncio.sleep", new_callable=AsyncMock):
-                lots = await s.fetch_around(lat=56.84, lon=60.60, rooms="1")
-
-    assert mock_rotate.call_count >= 1
-    assert lots == []
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_no_rotate_on_404():
-    """fetch_around does NOT rotate on 404 -- only on tarpit (status==0) / bad JSON."""
-    s = YandexRealtyScraper()
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        return _CurlResponse(status_code=404, text="Not Found")
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "_rotate_ip", new_callable=AsyncMock) as mock_rotate:
-            lots = await s.fetch_around(lat=56.84, lon=60.60)
-
-    assert mock_rotate.call_count == 0
-    assert lots == []
-
-
-# ---------------------------------------------------------------------------
-# PART H: fetch_around_multi_room
-# ---------------------------------------------------------------------------
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_multi_room_dedup():
-    """Multi-room fetch deduplicates offers seen in multiple combos.
-
-    fetch_around_multi_room fetches page=1 inline via _http_get, not via
-    fetch_around. We mock _http_get to return JSON payloads with shared/unique
-    offer IDs to verify deduplication.
-    """
-    s = YandexRealtyScraper()
-
-    # Combo 0: shared_offer + unique_0
-    # Combo 1: shared_offer + unique_1
-    entity_shared = dict(
-        _ENTITY_FULL, offerId="shared_offer_1234", url="//realty.yandex.ru/offer/shared_offer_1234"
-    )
-    entity_unique_0 = dict(
-        _ENTITY_FULL, offerId="unique_offer_0", url="//realty.yandex.ru/offer/unique_offer_0"
-    )
-    entity_unique_1 = dict(
-        _ENTITY_FULL, offerId="unique_offer_1", url="//realty.yandex.ru/offer/unique_offer_1"
-    )
-
-    pager = {"page": 0, "pageSize": 20, "totalItems": 2, "totalPages": 1}
-    payload_combo0 = _make_gate_payload([entity_shared, entity_unique_0], pager)
-    payload_combo1 = _make_gate_payload([entity_shared, entity_unique_1], pager)
-
-    combo_payloads = [payload_combo0, payload_combo1]
-    call_idx = 0
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        nonlocal call_idx
-        idx = call_idx
-        call_idx += 1
-        if idx < len(combo_payloads):
-            return _CurlResponse(status_code=200, text=json.dumps(combo_payloads[idx]))
-        return _CurlResponse(status_code=200, text=json.dumps(_make_gate_payload([])))
-
-    mock_settings = MagicMock()
-    mock_settings.yandex_proxy_url = None
-
-    with patch.dict("sys.modules", {"app.core.config": MagicMock(settings=mock_settings)}):
-        with patch.object(s, "_http_get", side_effect=fake_http_get):
-            with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-                lots = await s.fetch_around_multi_room(
-                    lat=0.0,
-                    lon=0.0,
-                    rooms_list=["1"],
-                    price_ranges=[(None, 5_000_000), (5_000_000, 7_000_000)],
-                    max_pages=1,
-                )
-
-    ids = {lot.source_id for lot in lots}
-    assert "shared_offer_1234" in ids
-    assert len(ids) == 3
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_multi_room_runs_both_segments_with_shared_dedup():
-    """segments=['NO','YES'] sweeps both vtorichka+novostroyki; dedup spans both passes.
-
-    Mocks _http_get and inspects requested URLs: both newFlat=NO and newFlat=YES
-    must be requested. The same offerId returned in both passes must be deduped to
-    one lot (single shared `seen` across passes), tagged by the FIRST pass's segment.
-    """
-    s = YandexRealtyScraper()
-
-    requested_new_flat: list[str] = []
-
-    entity_shared = dict(
-        _ENTITY_FULL, offerId="shared_xseg", url="//realty.yandex.ru/offer/shared_xseg"
-    )
-    entity_vt = dict(_ENTITY_FULL, offerId="vt_only", url="//realty.yandex.ru/offer/vt_only")
-    entity_nb = dict(_ENTITY_FULL, offerId="nb_only", url="//realty.yandex.ru/offer/nb_only")
-    pager = {"page": 0, "pageSize": 20, "totalItems": 2, "totalPages": 1}
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        if "newFlat=YES" in url:
-            requested_new_flat.append("YES")
-            payload = _make_gate_payload([entity_shared, entity_nb], pager)
-        else:
-            requested_new_flat.append("NO")
-            payload = _make_gate_payload([entity_shared, entity_vt], pager)
-        return _CurlResponse(status_code=200, text=json.dumps(payload))
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-            lots = await s.fetch_around_multi_room(
-                lat=0.0,
-                lon=0.0,
-                rooms_list=["1"],
-                price_ranges=[(None, 5_000_000)],
-                segments=["NO", "YES"],
-                max_pages=1,
-            )
-
-    assert "NO" in requested_new_flat
-    assert "YES" in requested_new_flat
-
-    by_id = {lot.source_id: lot for lot in lots}
-    assert set(by_id) == {"shared_xseg", "vt_only", "nb_only"}
-    # Shared offer deduped to ONE lot, tagged by the first pass (NO -> vtorichka).
-    assert by_id["shared_xseg"].listing_segment == "vtorichka"
-    assert by_id["vt_only"].listing_segment == "vtorichka"
-    assert by_id["nb_only"].listing_segment == "novostroyki"
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_multi_room_default_segment_no_only():
-    """Without segments arg, only newFlat=NO (vtorichka) is requested (legacy default)."""
-    s = YandexRealtyScraper()
-    requested_new_flat: list[str] = []
-    pager = {"page": 0, "pageSize": 20, "totalItems": 1, "totalPages": 1}
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        requested_new_flat.append("YES" if "newFlat=YES" in url else "NO")
-        body = json.dumps(_make_gate_payload([_ENTITY_FULL], pager))
-        return _CurlResponse(status_code=200, text=body)
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-            await s.fetch_around_multi_room(
-                lat=0.0,
-                lon=0.0,
-                rooms_list=["1"],
-                price_ranges=[(None, 5_000_000)],
-                max_pages=1,
-            )
-
-    assert set(requested_new_flat) == {"NO"}
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_multi_room_legacy_single_sweep():
-    """Legacy mode (no rooms_list/price_ranges) uses single citywide sweep."""
-    s = YandexRealtyScraper()
-    calls: list[dict] = []
-
-    async def fake_fetch_around(
-        lat, lon, radius_m=1000, page=1, rooms=None, price_min=None, price_max=None
-    ):
-        calls.append({"page": page, "rooms": rooms, "price_min": price_min})
-        return []
-
-    with patch.object(s, "fetch_around", side_effect=fake_fetch_around):
-        await s.fetch_around_multi_room(lat=0.0, lon=0.0, max_pages=3)
-
-    assert all(c["rooms"] is None for c in calls)
-    assert all(c["price_min"] is None for c in calls)
-
-
-# ---------------------------------------------------------------------------
-# PART I: exhaustive load -- fetch_all_secondary + _walk_price_range
-# ---------------------------------------------------------------------------
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_splits_when_total_exceeds_cap():
-    """When totalItems > cap, _walk_price_range splits into sub-ranges recursively."""
-    s = YandexRealtyScraper()
-
-    call_ranges: list[tuple] = []
-    call_count = 0
-
-    async def fake_fetch_page_json(rooms, page, price_min, price_max):
-        nonlocal call_count
-        call_count += 1
-        call_ranges.append((price_min, price_max))
-        # First call (wide range): 600 items -> triggers split
-        total = 600 if call_count == 1 else 10
-        entities = [dict(_ENTITY_FULL, offerId=f"offer_{call_count}")]
-        pager = {
-            "page": 0,
-            "pageSize": 20,
-            "totalItems": total,
-            "totalPages": max(1, total // 20),
-        }
-        return _make_gate_payload(entities, pager)
-
-    seen: dict = {}
-    with patch.object(s, "_fetch_page_json", side_effect=fake_fetch_page_json):
-        with patch("asyncio.sleep", new_callable=AsyncMock):
-            await s._walk_price_range(
-                rooms="1",
-                lo=0,
-                hi=10_000_000,
-                seen=seen,
-                price_cap_per_bucket=500,
-                max_pages_per_bucket=50,
-                concurrency=2,
-            )
-
-    assert len(call_ranges) >= 3, f"Expected >=3 probes, got {call_ranges}"
-    assert call_ranges[0] == (None, 10_000_000)  # lo=0 -> _lo_param=None
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_fallback_when_total_none():
-    """totalItems absent in pager triggers paginate-until-empty fallback."""
-    s = YandexRealtyScraper()
-    call_count = 0
-
-    async def fake_fetch_page_json(rooms, page, price_min, price_max):
-        nonlocal call_count
-        call_count += 1
-        if call_count == 1:
-            # Probe: pager without totalItems
-            pager = {"page": 0, "pageSize": 20}
-            return _make_gate_payload([_ENTITY_FULL], pager)
-        if call_count == 2:
-            pager = {"page": 0, "pageSize": 20}
-            return _make_gate_payload([dict(_ENTITY_FULL, offerId=f"offer_{call_count}")], pager)
-        return _make_gate_payload([])
-
-    seen: dict = {}
-    bucket_calls: list = []
-
-    with patch.object(s, "_fetch_page_json", side_effect=fake_fetch_page_json):
-        with patch("asyncio.sleep", new_callable=AsyncMock):
-            await s._walk_price_range(
-                rooms="2",
-                lo=0,
-                hi=15_000_000,
-                seen=seen,
-                price_cap_per_bucket=500,
-                max_pages_per_bucket=10,
-                concurrency=2,
-                on_bucket=lambda k, n: bucket_calls.append((k, n)),
-            )
-
-    assert len(bucket_calls) >= 1, "on_bucket must fire in fallback path"
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_fires_on_bucket_for_leaf():
-    """on_bucket callback fires once for a leaf bucket with small totalItems."""
-    s = YandexRealtyScraper()
-    bucket_calls: list = []
-
-    async def fake_fetch_page_json(rooms, page, price_min, price_max):
-        pager = {"page": 0, "pageSize": 20, "totalItems": 5, "totalPages": 1}
-        return _make_gate_payload([_ENTITY_FULL], pager)
-
-    seen: dict = {}
-    with patch.object(s, "_fetch_page_json", side_effect=fake_fetch_page_json):
-        with patch("asyncio.sleep", new_callable=AsyncMock):
-            await s._walk_price_range(
-                rooms="1",
-                lo=0,
-                hi=20_000_000,
-                seen=seen,
-                price_cap_per_bucket=500,
-                max_pages_per_bucket=5,
-                concurrency=2,
-                on_bucket=lambda k, n: bucket_calls.append(k),
-            )
-
-    assert len(bucket_calls) >= 1
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_skip_buckets():
-    """skip_buckets prevents on_bucket from firing for already-done bucket."""
-    s = YandexRealtyScraper()
-    on_bucket_calls: list = []
-
-    async def fake_fetch_page_json(rooms, page, price_min, price_max):
-        pager = {"page": 0, "pageSize": 20, "totalItems": 10, "totalPages": 1}
-        return _make_gate_payload([_ENTITY_FULL], pager)
-
-    # lo=0 -> _lo_param=None -> bucket key uses None
-    skip_set = {_combo_label("1", 0, 10_000_000)}
-    seen: dict = {}
-    with patch.object(s, "_fetch_page_json", side_effect=fake_fetch_page_json):
-        with patch("asyncio.sleep", new_callable=AsyncMock):
-            await s._walk_price_range(
-                rooms="1",
-                lo=0,
-                hi=10_000_000,
-                seen=seen,
-                price_cap_per_bucket=500,
-                max_pages_per_bucket=5,
-                concurrency=2,
-                on_bucket=lambda k, n: on_bucket_calls.append(k),
-                skip_buckets=skip_set,
-            )
-
-    assert skip_set.pop() not in on_bucket_calls
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_deduplicates_across_rooms():
-    """fetch_all_secondary deduplicates offers across room buckets."""
-    s = YandexRealtyScraper()
-
-    shared_lot = MagicMock(spec=ScrapedLot)
-    shared_lot.source_id = "shared_999"
-    shared_lot.source_url = "https://realty.yandex.ru/offer/shared_999/"
-
-    unique_lot_1 = MagicMock(spec=ScrapedLot)
-    unique_lot_1.source_id = "unique_001"
-    unique_lot_1.source_url = "https://realty.yandex.ru/offer/unique_001/"
-
-    unique_lot_2 = MagicMock(spec=ScrapedLot)
-    unique_lot_2.source_id = "unique_002"
-    unique_lot_2.source_url = "https://realty.yandex.ru/offer/unique_002/"
-
-    call_count = 0
-
-    async def fake_walk(*, rooms, lo, hi, seen, **kwargs):
-        nonlocal call_count
-        seen["shared_999"] = shared_lot
-        if call_count == 0:
-            seen["unique_001"] = unique_lot_1
-        else:
-            seen["unique_002"] = unique_lot_2
-        call_count += 1
-
-    with patch.object(s, "_walk_price_range", side_effect=fake_walk):
-        result = await s.fetch_all_secondary(rooms_buckets=["1", "2"])
-
-    ids = {lot.source_id for lot in result}
-    assert "shared_999" in ids
-    assert "unique_001" in ids
-    assert "unique_002" in ids
-    assert len(ids) == 3
-
-
-@pytest.mark.asyncio
-async def test_fetch_all_secondary_walks_shared_seed_grid():
-    """fetch_all_secondary итерирует общую seed-сетку: len(_SEED) walk-вызовов/комнатность.
-
-    Закрытые брекеты передают hi = br_hi - 1; открытый (последний) — hi=None.
-    """
-    from app.services.scrapers.price_brackets import get_price_seed_brackets
-
-    seed = get_price_seed_brackets()
-    s = YandexRealtyScraper()
-    walk_args: list[tuple[int, int | None]] = []
-
-    async def fake_walk(*, rooms, lo, hi, seen, **kwargs):
-        walk_args.append((lo, hi))
-
-    with patch.object(s, "_walk_price_range", side_effect=fake_walk):
-        await s.fetch_all_secondary(rooms_buckets=["1"])
-
-    assert len(walk_args) == len(
-        seed
-    ), f"Ожидалось {len(seed)} walk-вызовов (seed-брекеты), получено {len(walk_args)}"
-    # Первый брекет закрыт: hi = seed[0].hi - 1.
-    first_lo, first_hi = seed[0]
-    assert first_hi is not None
-    assert walk_args[0] == (first_lo, first_hi - 1)
-    # Последний брекет открыт: hi=None (без priceMax-потолка).
-    last_lo, last_hi = seed[-1]
-    assert last_hi is None
-    assert walk_args[-1] == (last_lo, None)
-
-
-@pytest.mark.asyncio
-async def test_walk_price_range_open_bracket_no_split():
-    """Открытый брекет (hi=None) не делится — пагинируется leaf напрямую (one bucket).
-
-    totalItems > cap, но hi=None → split невозможен → ровно один leaf-бакет.
-    """
-    s = YandexRealtyScraper()
-    bucket_calls: list = []
-    seen_prices: list[int | None] = []
-
-    async def fake_fetch_page_json(rooms, page, price_min, price_max):
-        seen_prices.append(price_max)
-        # total > cap, но 1 страница лотов → leaf пагинируется без split
-        pager = {"page": 0, "pageSize": 20, "totalItems": 9999, "totalPages": 1}
-        return _make_gate_payload([_ENTITY_FULL], pager)
-
-    seen: dict = {}
-    with patch.object(s, "_fetch_page_json", side_effect=fake_fetch_page_json):
-        with patch("asyncio.sleep", new_callable=AsyncMock):
-            await s._walk_price_range(
-                rooms="1",
-                lo=250_000_000,
-                hi=None,
-                seen=seen,
-                price_cap_per_bucket=500,
-                max_pages_per_bucket=5,
-                concurrency=2,
-                on_bucket=lambda k, n: bucket_calls.append(k),
-            )
-
-    # Открытый брекет: НЕ делится (один on_bucket), priceMax всегда None (без потолка).
-    assert len(bucket_calls) == 1, f"Открытый брекет не должен делиться, got {bucket_calls}"
-    assert all(
-        pm is None for pm in seen_prices
-    ), f"Открытый брекет: priceMax всегда None, got {seen_prices}"
-
-
-# ---------------------------------------------------------------------------
-# PART J: fetch_around_multi_room on_combo callback (FIX A — incremental save)
-# ---------------------------------------------------------------------------
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_multi_room_on_combo_called_per_combo():
-    """on_combo вызывается после каждого combo с новыми (деduped) лотами.
-
-    2 combo (rooms_list=['1'], 2 price_ranges): каждый имеет уникальный лот.
-    on_combo должен быть вызван дважды — по одному разу на combo.
-    """
-    s = YandexRealtyScraper()
-
-    entity_0 = dict(_ENTITY_FULL, offerId="combo0_lot", url="//realty.yandex.ru/offer/combo0_lot")
-    entity_1 = dict(_ENTITY_FULL, offerId="combo1_lot", url="//realty.yandex.ru/offer/combo1_lot")
-    pager = {"page": 0, "pageSize": 20, "totalItems": 1, "totalPages": 1}
-
-    call_idx = 0
-    payloads = [
-        _make_gate_payload([entity_0], pager),
-        _make_gate_payload([entity_1], pager),
-    ]
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        nonlocal call_idx
-        idx = call_idx
-        call_idx += 1
-        if idx < len(payloads):
-            return _CurlResponse(status_code=200, text=json.dumps(payloads[idx]))
-        return _CurlResponse(status_code=200, text=json.dumps(_make_gate_payload([])))
-
-    on_combo_calls: list[tuple[str, list]] = []
-
-    def on_combo(label: str, new_lots: list) -> None:
-        on_combo_calls.append((label, list(new_lots)))
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-            lots = await s.fetch_around_multi_room(
-                lat=0.0,
-                lon=0.0,
-                rooms_list=["1"],
-                price_ranges=[(None, 5_000_000), (5_000_000, 10_000_000)],
-                max_pages=1,
-                on_combo=on_combo,
-            )
-
-    # Два уникальных лота, два combo.
-    assert len(lots) == 2
-    # on_combo вызван ровно дважды — по одному на combo.
-    assert len(on_combo_calls) == 2, f"Ожидалось 2 on_combo вызова, получено {len(on_combo_calls)}"
-    # Каждый on_combo получил ровно один новый лот.
-    for _label, combo_lots in on_combo_calls:
-        assert len(combo_lots) == 1
-    # Конкретные source_id в правильных batch'ах.
-    all_ids = {lot.source_id for _, combo_lots in on_combo_calls for lot in combo_lots}
-    assert all_ids == {"combo0_lot", "combo1_lot"}
-
-
-@pytest.mark.asyncio
-async def test_fetch_around_multi_room_on_combo_not_called_for_failed_combo():
-    """on_combo НЕ вызывается для combo, у которого все page-1 retries исчерпаны.
-
-    FIX B: bounded retry — если status_code=0 на всех попытках page=1,
-    combo пропускается. on_combo не должен быть вызван для пропущенного combo.
-    Успешный combo (второй) — on_combo должен сработать.
-    """
-    s = YandexRealtyScraper()
-
-    entity_ok = dict(_ENTITY_FULL, offerId="ok_lot", url="//realty.yandex.ru/offer/ok_lot")
-    pager = {"page": 0, "pageSize": 20, "totalItems": 1, "totalPages": 1}
-
-    call_idx = 0
-
-    async def fake_http_get(url: str, **kwargs: object) -> _CurlResponse:
-        nonlocal call_idx
-        call_idx += 1
-        # Первые (1 + _YANDEX_TARPIT_MAX_RETRIES) запросы — это retries для combo[0] → fail.
-        from app.services.scrapers.yandex_realty import _YANDEX_TARPIT_MAX_RETRIES
-
-        if call_idx <= 1 + _YANDEX_TARPIT_MAX_RETRIES:
-            return _CurlResponse(status_code=0, text="")
-        # Следующий запрос — combo[1] page=1 → success.
-        return _CurlResponse(
-            status_code=200, text=json.dumps(_make_gate_payload([entity_ok], pager))
-        )
-
-    on_combo_calls: list[str] = []
-
-    def on_combo(label: str, new_lots: list) -> None:
-        on_combo_calls.append(label)
-
-    with patch.object(s, "_http_get", side_effect=fake_http_get):
-        with patch.object(s, "_rotate_ip", new_callable=AsyncMock):
-            with patch("asyncio.sleep", new_callable=AsyncMock):
-                with patch.object(s, "sleep_between_requests", new_callable=AsyncMock):
-                    lots = await s.fetch_around_multi_room(
-                        lat=0.0,
-                        lon=0.0,
-                        rooms_list=["1"],
-                        price_ranges=[(None, 5_000_000), (5_000_000, 10_000_000)],
-                        max_pages=1,
-                        on_combo=on_combo,
-                    )
-
-    # combo[0] скипнут, combo[1] успешен → 1 лот в итоге.
-    assert len(lots) == 1
-    assert lots[0].source_id == "ok_lot"
-    # on_combo вызван только для успешного combo, не для скипнутого.
-    assert len(on_combo_calls) == 1, (
-        f"on_combo должен быть вызван только 1 раз (успешный combo), "
-        f"получено {len(on_combo_calls)}: {on_combo_calls}"
-    )
diff --git a/tradein-mvp/backend/tests/test_yandex_scrapers_delay_wiring.py b/tradein-mvp/backend/tests/test_yandex_scrapers_delay_wiring.py
index ea6b61ef..43d7f54e 100644
--- a/tradein-mvp/backend/tests/test_yandex_scrapers_delay_wiring.py
+++ b/tradein-mvp/backend/tests/test_yandex_scrapers_delay_wiring.py
@@ -1,8 +1,10 @@
 """Smoke: each Yandex scraper instance reads request_delay_sec from scraper_settings.
 
 Легаси `app.services.scrapers.yandex_detail` / `yandex_valuation` удалены (#2277
-финальный шаг scraper_kit-миграции). `yandex_realty` / `yandex_newbuilding` пока
-остаются легаси (не мигрированы) — их delay-wiring тесты не тронуты.
+финальный шаг scraper_kit-миграции). `yandex_newbuilding` пока остаётся легаси (не
+мигрирован) — его delay-wiring тест не тронут. `yandex_realty` удалён (#2397 Part
+E2, 0 runtime-импортёров) — его delay-wiring тест убран (kit `YandexRealtyScraper`
+delay-provider wiring покрыт `test_kit_serp_proxy_pool.py`).
 `yandex_detail`/`yandex_valuation` переведены на kit-эквиваленты, которые используют
 инжектируемый `delay_provider=` (strangler DI, #2133) вместо прямого импорта
 `get_scraper_delay` — тест адаптирован под реальный контракт kit-класса.
@@ -15,7 +17,6 @@ from scraper_kit.providers.yandex.detail import YandexDetailScraper
 from scraper_kit.providers.yandex.valuation import YandexValuationScraper
 
 from app.services.scrapers.yandex_newbuilding import YandexNewbuildingScraper
-from app.services.scrapers.yandex_realty import YandexRealtyScraper
 
 _KIT_CONFIG = SimpleNamespace(
     scraper_proxy_url=None,
@@ -24,12 +25,6 @@ _KIT_CONFIG = SimpleNamespace(
 )
 
 
-def test_yandex_realty_picks_up_delay():
-    with patch("app.services.scrapers.yandex_realty.get_scraper_delay", return_value=9.5):
-        s = YandexRealtyScraper()
-        assert s.request_delay_sec == 9.5
-
-
 def test_yandex_detail_picks_up_delay():
     """Kit YandexDetailScraper: delay_provider callable вызывается с self.name."""
     s = YandexDetailScraper(delay_provider=lambda _name: 9.5)
diff --git a/tradein-mvp/scripts/local-sweep-ekb-cian.py b/tradein-mvp/scripts/local-sweep-ekb-cian.py
deleted file mode 100644
index 6250e8ec..00000000
--- a/tradein-mvp/scripts/local-sweep-ekb-cian.py
+++ /dev/null
@@ -1,359 +0,0 @@
-"""Cian ЕКБ vtorichka sweep — local residential IP → prod tradein DB через SSH tunnel.
-
-Why: ekb.cian.ru SERP отдаёт ВЕСЬ город (без geo-фильтра по lat/lon), поэтому
-"10 anchors" из Avito-эталона тут бесполезны — одинаковый ответ × 10.
-Вместо anchors — proper room-segmented citywide pagination
-(rooms 1/2/3/4 × N pages × ~28 listings), per-page save (visible progress +
-не теряем данные при крэше).
-
-Cian реально менее агрессивен чем Avito, но **residential IP всё равно может попасть
-в blacklist при слишком частых запросах** — default delay 15s.
-
-Setup (один раз):
-  ssh -f -N -L 15433:172.20.0.2:5432 gendesign
-  cd tradein-mvp/backend
-  $env:DATABASE_URL='postgresql+psycopg://tradein:@localhost:15433/tradein'
-  $env:ENVIRONMENT='local'; $env:GLITCHTIP_DSN=''
-
-Usage:
-  uv run python ../scripts/local-sweep-ekb-cian.py                 # default: 4 rooms × 8 pages
-  uv run python ../scripts/local-sweep-ekb-cian.py --pages 20      # больше глубины
-  uv run python ../scripts/local-sweep-ekb-cian.py --delay 10      # быстрее (risk)
-  uv run python ../scripts/local-sweep-ekb-cian.py --rooms 0       # всё (включая студии — no room param)
-"""
-from __future__ import annotations
-
-# stdlib
-import argparse
-import asyncio
-import json
-import sys
-import time
-import types
-from pathlib import Path
-
-# Auto-inject backend/ в sys.path чтобы найти app.* package
-_BACKEND_DIR = Path(__file__).resolve().parent.parent / "backend"
-if str(_BACKEND_DIR) not in sys.path:
-    sys.path.insert(0, str(_BACKEND_DIR))
-
-# Stub weasyprint (Windows без GTK ломает import chain через trade_in.py)
-for _m in (
-    "weasyprint",
-    "weasyprint.text",
-    "weasyprint.text.ffi",
-    "weasyprint.css",
-    "weasyprint.css.targets",
-):
-    sys.modules[_m] = types.ModuleType(_m)
-sys.modules["weasyprint"].HTML = type("HTML", (), {"__init__": lambda self, *a, **kw: None})
-sys.modules["weasyprint"].CSS = type("CSS", (), {"__init__": lambda self, *a, **kw: None})
-
-# app imports после stubs
-import logging  # noqa: E402
-import re  # noqa: E402
-
-from app.core.db import SessionLocal  # noqa: E402
-from app.services.scrapers.base import ScrapedLot, save_listings  # noqa: E402
-from app.services.scrapers.cian import CianScraper  # noqa: E402
-
-# Шумный WARNING из upstream cian_state_parser ('state extraction failed') — у нас
-# fallback на новый .concat() markup отрабатывает, warning нерелевантен.
-logging.getLogger("app.services.scrapers.cian").setLevel(logging.ERROR)
-
-# ── New-markup fallback parser ───────────────────────────────────────────────
-# Cian переключил MFE bootstrap с .push({...}) на .concat([{...},{...}]):
-#   window._cianConfig['frontend-serp'] = (window._cianConfig['frontend-serp'] || []).concat([…])
-# Существующий cian_state_parser.py не ловит этот шаблон → 0 lots на свежей выдаче.
-# Чтобы не трогать shared scraper module (правилами запрещено), парсим тут.
-_RE_CONCAT_START = re.compile(
-    r"window\._cianConfig\['(?P[\w-]+)'\]\s*=\s*"
-    r"\(window\._cianConfig\['[\w-]+'\]\s*\|\|\s*\[\]\)\.concat\(\["
-)
-
-
-def _extract_balanced_array(text: str, start: int) -> str | None:
-    """Из позиции `start` (где '[') вернуть текст массива включая внешние '[' и ']'.
-
-    JSON-aware: учитывает strings и escapes, чтобы не сбить счётчик '[' ']' на
-    скобках внутри string literals.
-    """
-    if text[start] != "[":
-        return None
-    depth = 0
-    i = start
-    in_str = False
-    esc = False
-    while i < len(text):
-        ch = text[i]
-        if in_str:
-            if esc:
-                esc = False
-            elif ch == "\\":
-                esc = True
-            elif ch == '"':
-                in_str = False
-        else:
-            if ch == '"':
-                in_str = True
-            elif ch == "[":
-                depth += 1
-            elif ch == "]":
-                depth -= 1
-                if depth == 0:
-                    return text[start : i + 1]
-        i += 1
-    return None
-
-
-def _parse_serp_v2(scraper: CianScraper, html: str) -> list[ScrapedLot]:
-    """Fallback парсер для нового шаблона .concat([...]) — ekb.cian.ru/cat.php c 2026-05."""
-    state: dict | None = None
-    for m in _RE_CONCAT_START.finditer(html):
-        if m.group("mfe") != "frontend-serp":
-            continue
-        # Позиция '[' — это последний символ матча минус 1 (regex кончается на '[')
-        arr_start = m.end() - 1
-        arr_str = _extract_balanced_array(html, arr_start)
-        if arr_str is None:
-            continue
-        try:
-            entries = json.loads(arr_str)
-        except json.JSONDecodeError:
-            continue
-        for entry in entries:
-            if (
-                isinstance(entry, dict)
-                and entry.get("key") == "initialState"
-                and isinstance(entry.get("value"), dict)
-            ):
-                state = entry["value"]
-                break
-        if state is not None:
-            break
-    if state is None:
-        return []
-    offers = (state.get("results") or {}).get("offers") or []
-    lots: list[ScrapedLot] = []
-    for offer in offers:
-        # Cian свежий markup: descriptionMinhash = list[int] (multi-hash signatures).
-        # ScrapedLot ожидает str → сбрасываем чтобы _offer_to_lot вычислил SHA1 от description.
-        if isinstance(offer.get("descriptionMinhash"), list):
-            offer["descriptionMinhash"] = None
-        lot = scraper._offer_to_lot(offer)
-        if lot is not None:
-            lots.append(lot)
-    return lots
-
-
-def _load_cookies_file(path: Path) -> dict[str, str]:
-    """Загрузить cookies из JSON-файла.
-
-    Поддерживает два формата:
-    - browser export (list[{name,value,...}]) — DevTools / EditThisCookie / Cookie-Editor
-    - plain dict {name: value}
-    """
-    raw = json.loads(path.read_text(encoding="utf-8"))
-    if isinstance(raw, list):
-        return {c["name"]: c["value"] for c in raw if c.get("name") and c.get("value")}
-    if isinstance(raw, dict):
-        return {str(k): str(v) for k, v in raw.items()}
-    raise SystemExit(f"--cookies-file: unknown JSON shape (got {type(raw).__name__})")
-
-
-async def phase_citywide(
-    pages: int,
-    delay: float,
-    room_segments: tuple[tuple[int, ...] | None, ...],
-    cookies_file: Path | None = None,
-    start_page: int = 1,
-) -> int:
-    """Room-segmented citywide pagination — per-page save, incremental.
-
-    room_segments — кортеж из tuple (1,) / (2,) / ... / None (= no room filter).
-    None segment даёт смешанную выдачу (студии + всё подряд) — Cian
-    не различает их без room param.
-    """
-    seg_labels = [
-        ("r" + str(seg[0]) if seg else "all")
-        for seg in room_segments
-    ]
-    print(
-        f"\n=== PHASE CITYWIDE: Cian ЕКБ vtorichka "
-        f"({len(room_segments)} segments={seg_labels} × {pages} pages, delay {delay}s) ===\n",
-        flush=True,
-    )
-    total = 0
-    cum_ins = 0
-    cum_upd = 0
-    async with CianScraper() as s:
-        # Override per-request delay (in addition to between-anchors delay)
-        s.request_delay_sec = delay
-        # Inject cookies (если есть): прокидываем в curl_cffi session перед warm-up
-        if cookies_file is not None:
-            cookies = _load_cookies_file(cookies_file)
-            assert s._cffi is not None
-            s._cffi.cookies.update(cookies)
-            print(f"  loaded {len(cookies)} cookies from {cookies_file.name}", flush=True)
-        # Warm-up: посетить homepage чтобы получить _CIAN_GK / _yasc + не выглядеть как наглый бот
-        try:
-            assert s._cffi is not None
-            r_warm = await s._cffi.get(f"{s.base_url}/")
-            print(
-                f"  warmup GET / → HTTP {r_warm.status_code} "
-                f"(cookies={len(list(s._cffi.cookies.jar))})",
-                flush=True,
-            )
-            # Subsequent SERP requests should look like in-site navigation
-            s._cffi.headers.update(
-                {
-                    "Referer": f"{s.base_url}/",
-                    "Sec-Fetch-Site": "same-origin",
-                    "Sec-Fetch-Mode": "navigate",
-                    "Sec-Fetch-User": "?1",
-                    "Sec-Fetch-Dest": "document",
-                }
-            )
-            await asyncio.sleep(min(5.0, delay))
-        except Exception as e:  # noqa: BLE001
-            print(f"  warmup FAIL: {type(e).__name__}: {e}", flush=True)
-        for seg_idx, seg in enumerate(room_segments):
-            seg_label = "r" + str(seg[0]) if seg else "all"
-            # --start-page применяется только к первому segment'у — продолжение прерванного sweep
-            page_from = start_page if seg_idx == 0 else 1
-            for page in range(page_from, pages + 1):
-                # Manual fetch+parse чтобы можно было дампнуть HTML при empty results
-                try:
-                    url_ = s._build_url(rooms=seg, page=page)
-                    assert s._cffi is not None
-                    resp = await s._cffi.get(url_)
-                    if resp.status_code != 200:
-                        print(
-                            f"  seg={seg_label} page={page} HTTP {resp.status_code} → skip",
-                            flush=True,
-                        )
-                        await asyncio.sleep(delay)
-                        continue
-                    lots = s._parse_serp_html(resp.text)
-                    if not lots:
-                        # Fallback: новый markup .concat([...])
-                        lots = _parse_serp_v2(s, resp.text)
-                        if lots:
-                            print(
-                                f"    (parsed via v2 .concat fallback)",
-                                flush=True,
-                            )
-                    if not lots:
-                        # Дамп для разбора: captcha vs пустая страница vs изменение markup
-                        dump_path = Path(f"cian-empty-seg-{seg_label}-p{page}.html")
-                        dump_path.write_text(resp.text, encoding="utf-8")
-                        print(f"    DUMP: {dump_path} ({len(resp.text)} bytes)", flush=True)
-                    await s.sleep_between_requests()
-                except Exception as e:  # noqa: BLE001
-                    print(
-                        f"  seg={seg_label} page={page} FAIL fetch: "
-                        f"{type(e).__name__}: {e}",
-                        flush=True,
-                    )
-                    await asyncio.sleep(delay)
-                    continue
-                if not lots:
-                    print(
-                        f"  seg={seg_label} page={page}: 0 lots — end of segment "
-                        "(либо captcha)",
-                        flush=True,
-                    )
-                    break
-                db = SessionLocal()
-                try:
-                    ins, upd = save_listings(db, lots)
-                    cum_ins += ins
-                    cum_upd += upd
-                    total += len(lots)
-                    print(
-                        f"  seg={seg_label} page={page}/{pages}: {len(lots)} lots → "
-                        f"ins={ins} upd={upd} "
-                        f"[cum: total={total} new={cum_ins} upd={cum_upd}]",
-                        flush=True,
-                    )
-                except Exception as e:  # noqa: BLE001
-                    print(
-                        f"  seg={seg_label} page={page} FAIL save: "
-                        f"{type(e).__name__}: {e}",
-                        flush=True,
-                    )
-                finally:
-                    db.close()
-                # scraper.fetch_around() уже спит request_delay_sec * jitter — extra sleep не нужен
-    print(
-        f"  PHASE CITYWIDE TOTAL: fetched={total} ins={cum_ins} upd={cum_upd}",
-        flush=True,
-    )
-    return total
-
-
-async def main() -> None:
-    p = argparse.ArgumentParser()
-    p.add_argument(
-        "--pages",
-        type=int,
-        default=8,
-        help="страниц на room-segment (default 8 → ~28×4×8 = ~900 до дедупа)",
-    )
-    p.add_argument(
-        "--delay",
-        type=float,
-        default=15,
-        help="sec между HTTP requests (residential-safe ≥10, default 15)",
-    )
-    p.add_argument(
-        "--rooms",
-        type=str,
-        default="1,2,3,4",
-        help="room segments через запятую (1,2,3,4) или 'all' для одного citywide "
-        "запроса без room param (default '1,2,3,4')",
-    )
-    p.add_argument(
-        "--cookies-file",
-        type=Path,
-        default=None,
-        help="JSON-файл с Cian cookies (browser-export или {name:value}). "
-        "Опционально — для обхода captcha. См. scripts/.cian-cookies.json",
-    )
-    p.add_argument(
-        "--start-page",
-        type=int,
-        default=1,
-        help="С какой страницы начать ПЕРВЫЙ сегмент (для resume прерванного sweep). "
-        "Остальные segments идут с p=1. Default 1.",
-    )
-    args = p.parse_args()
-
-    # Parse rooms argument
-    if args.rooms.strip().lower() == "all":
-        segments: tuple[tuple[int, ...] | None, ...] = (None,)
-    else:
-        try:
-            nums = [int(x.strip()) for x in args.rooms.split(",") if x.strip()]
-        except ValueError as e:
-            raise SystemExit(f"--rooms must be comma-separated ints or 'all': {e}") from e
-        segments = tuple((n,) for n in nums)
-
-    cookies_file: Path | None = args.cookies_file
-    if cookies_file is not None and not cookies_file.exists():
-        raise SystemExit(f"--cookies-file not found: {cookies_file}")
-
-    t0 = time.perf_counter()
-    await phase_citywide(
-        args.pages,
-        args.delay,
-        segments,
-        cookies_file=cookies_file,
-        start_page=args.start_page,
-    )
-    elapsed = time.perf_counter() - t0
-    print(f"\n=== ALL DONE in {elapsed / 60:.1f} min ===", flush=True)
-
-
-if __name__ == "__main__":
-    asyncio.run(main())
diff --git a/tradein-mvp/scripts/local-sweep-ekb-yandex.py b/tradein-mvp/scripts/local-sweep-ekb-yandex.py
deleted file mode 100644
index 9fc87cf9..00000000
--- a/tradein-mvp/scripts/local-sweep-ekb-yandex.py
+++ /dev/null
@@ -1,376 +0,0 @@
-"""Yandex.Недвижимость ЕКБ vtorichka sweep — runs locally через residential IP.
-
-Why local: Yandex anti-bot (captcha redirect) включается быстрее на datacenter
-IP + httpx fingerprint. Запуск с residential IP + Chrome120 TLS impersonation
-(curl_cffi) + browser session cookies — проходит. Скрипт пишет в prod tradein
-DB через SSH tunnel (localhost:15433 → tradein-postgres docker container).
-
-Modes:
-  --mode citywide  (default): один общий /vtorichniy-rynok/ sweep, ~25 страниц
-                              × 23 cards = ~575 уникальных. Yandex SERP лимит.
-  --mode combos:    итерация 5 rooms × 6 price ranges = 30 combos × до 25 pages.
-                    DB-level дедуп → coverage близко к total (~4000+ vtorichka
-                    ЕКБ). ~75 min полный прогон.
-
-Setup:
-  ssh -f -N -L 15433:172.20.0.2:5432 gendesign   # tradein-postgres tunnel
-  cd tradein-mvp/backend
-  export DATABASE_URL='postgresql+psycopg://tradein:@localhost:15433/tradein'
-  export ENVIRONMENT='local'
-  export GLITCHTIP_DSN=''
-
-Usage:
-  uv run python ../scripts/local-sweep-ekb-yandex.py                       # citywide
-  uv run python ../scripts/local-sweep-ekb-yandex.py --mode combos          # full ~4000
-  uv run python ../scripts/local-sweep-ekb-yandex.py --mode combos \\
-      --rooms 1,2 --price-ranges 5-7,7-10                                   # subset
-  uv run python ../scripts/local-sweep-ekb-yandex.py --pages 1              # smoke
-"""
-from __future__ import annotations
-
-# stdlib
-import argparse
-import asyncio
-import json
-import random
-import sys
-import time
-import types
-from pathlib import Path
-from urllib.parse import urlencode
-
-# Auto-inject backend/ в sys.path чтобы найти app.* package
-_BACKEND_DIR = Path(__file__).resolve().parent.parent / "backend"
-if str(_BACKEND_DIR) not in sys.path:
-    sys.path.insert(0, str(_BACKEND_DIR))
-
-# Stub weasyprint (Windows без GTK ломает import chain через trade_in.py)
-for _m in (
-    "weasyprint",
-    "weasyprint.text",
-    "weasyprint.text.ffi",
-    "weasyprint.css",
-    "weasyprint.css.targets",
-):
-    sys.modules[_m] = types.ModuleType(_m)
-sys.modules["weasyprint"].HTML = type("HTML", (), {"__init__": lambda self, *a, **kw: None})
-sys.modules["weasyprint"].CSS = type("CSS", (), {"__init__": lambda self, *a, **kw: None})
-
-# app imports после stubs
-from curl_cffi.requests import AsyncSession  # noqa: E402
-
-from app.core.db import SessionLocal  # noqa: E402
-from app.services.scrapers.base import save_listings  # noqa: E402
-from app.services.scrapers.yandex_realty import YandexRealtyScraper  # noqa: E402
-
-DEFAULT_CITY = "ekaterinburg"
-DEFAULT_COOKIES_FILE = Path(__file__).resolve().parent / ".yandex-cookies.json"
-
-# Path-сегменты Yandex для room buckets (после канонизации). Ключ — slug для
-# --rooms CLI; значение — segment в URL.
-ROOM_PATH: dict[str, str] = {
-    "studio": "studiya",
-    "1": "odnokomnatnaya",
-    "2": "dvuhkomnatnaya",
-    "3": "trehkomnatnaya",
-    "4+": "4-i-bolee",
-}
-
-# Price ranges (price_min, price_max) в рублях. None — open-ended boundary.
-PRICE_RANGES: list[tuple[int | None, int | None]] = [
-    (None, 5_000_000),
-    (5_000_000, 7_000_000),
-    (7_000_000, 10_000_000),
-    (10_000_000, 15_000_000),
-    (15_000_000, 25_000_000),
-    (25_000_000, None),
-]
-
-_CAPTCHA_MARKERS = ("/showcaptcha", "smartcaptcha", "checkboxcaptcha")
-
-
-def _load_cookies(path: Path) -> dict[str, str]:
-    if not path.exists():
-        return {}
-    raw = json.loads(path.read_text(encoding="utf-8"))
-    return {item["name"]: item["value"] for item in raw if "name" in item and "value" in item}
-
-
-def _is_captcha(html: str) -> bool:
-    head = html[:5000].lower()
-    return any(m in head for m in _CAPTCHA_MARKERS) or "докажите, что вы не робот" in head
-
-
-def _build_url(
-    city: str,
-    page: int = 0,
-    rooms: str | None = None,
-    price_min: int | None = None,
-    price_max: int | None = None,
-) -> str:
-    """Yandex SERP URL. Если rooms задан — path-based бакет; иначе общий vtorichka.
-
-    newFlat=NO_DEAL принудительно отфильтровывает новостройки (нужно при
-    combo, т.к. Yandex canonicalize иногда дропает /vtorichniy-rynok/).
-    """
-    if rooms:
-        path = f"/{city}/kupit/kvartira/{ROOM_PATH[rooms]}/vtorichniy-rynok/"
-    else:
-        path = f"/{city}/kupit/kvartira/vtorichniy-rynok/"
-    base = f"https://realty.yandex.ru{path}"
-
-    params: dict[str, str | int] = {}
-    if price_min is not None:
-        params["priceMin"] = price_min
-    if price_max is not None:
-        params["priceMax"] = price_max
-    if price_min is not None or price_max is not None:
-        # Гарантия vtorichka — Yandex иногда дропает /vtorichniy-rynok/ при canonicalize
-        params["newFlat"] = "NO_DEAL"
-    if page > 0:
-        params["page"] = page
-
-    if params:
-        return f"{base}?{urlencode(params)}"
-    return base
-
-
-def _combo_label(rooms: str | None, price_min: int | None, price_max: int | None) -> str:
-    r = rooms or "all-rooms"
-    if price_min is None and price_max is None:
-        return f"{r}/any-price"
-    lo = f"{price_min // 1_000_000}M" if price_min else "0"
-    hi = f"{price_max // 1_000_000}M" if price_max else "inf"
-    return f"{r}/{lo}-{hi}"
-
-
-async def sweep_one(
-    session: AsyncSession,
-    parser: YandexRealtyScraper,
-    city: str,
-    pages: int,
-    delay: float,
-    rooms: str | None = None,
-    price_min: int | None = None,
-    price_max: int | None = None,
-) -> tuple[int, int, int]:
-    """Один scan по фиксированному (rooms, price) — пагинация до pages или empty.
-
-    Returns: (fetched, inserted, updated)
-    """
-    label = _combo_label(rooms, price_min, price_max)
-    total = cum_ins = cum_upd = 0
-    for page in range(pages):
-        url = _build_url(city, page=page, rooms=rooms, price_min=price_min, price_max=price_max)
-        try:
-            resp = await session.get(url, timeout=20)
-        except Exception as e:  # noqa: BLE001
-            print(f"    [{label}] page={page} FAIL fetch: {type(e).__name__}: {e}", flush=True)
-            await asyncio.sleep(delay)
-            continue
-        if resp.status_code != 200:
-            print(f"    [{label}] page={page} HTTP {resp.status_code} — stop combo", flush=True)
-            break
-        if _is_captcha(resp.text):
-            print(
-                f"    [{label}] page={page} ⚠️ CAPTCHA — stop combo. "
-                f"Подожди 30-60min или refresh .yandex-cookies.json",
-                flush=True,
-            )
-            raise RuntimeError("captcha")
-        # NBSP → space: yandex_helpers.parse_rub() багует на \xa0 в group(1) → price None
-        html = resp.text.replace("\xa0", " ")
-        lots = parser._parse_html(html, page=page)
-        if not lots:
-            print(f"    [{label}] page={page}: 0 lots — end of pagination, stop", flush=True)
-            break
-        db = SessionLocal()
-        try:
-            ins, upd = save_listings(db, lots)
-            cum_ins += ins
-            cum_upd += upd
-            total += len(lots)
-            print(
-                f"    [{label}] page={page + 1}/{pages}: {len(lots)} lots → "
-                f"ins={ins} upd={upd} [combo: {total}/{cum_ins}/{cum_upd}]",
-                flush=True,
-            )
-        except Exception as e:  # noqa: BLE001
-            print(f"    [{label}] page={page} FAIL save: {type(e).__name__}: {e}", flush=True)
-        finally:
-            db.close()
-        if page + 1 < pages:
-            await asyncio.sleep(delay * random.uniform(0.85, 1.3))
-    return total, cum_ins, cum_upd
-
-
-async def phase_citywide(pages: int, delay: float, city: str, cookies: dict[str, str]) -> None:
-    print(
-        f"\n=== PHASE CITYWIDE: Yandex vtorichka (city={city}, {pages} pages, "
-        f"delay {delay}s) ===\n",
-        flush=True,
-    )
-    parser = YandexRealtyScraper(city=city)
-    async with AsyncSession(impersonate="chrome120", cookies=cookies or {}) as session:
-        total, ins, upd = await sweep_one(session, parser, city, pages, delay)
-    print(f"  CITYWIDE TOTAL: fetched={total} ins={ins} upd={upd}", flush=True)
-
-
-async def phase_combos(
-    pages: int,
-    delay: float,
-    city: str,
-    cookies: dict[str, str],
-    rooms_list: list[str],
-    price_ranges: list[tuple[int | None, int | None]],
-    combo_delay: float,
-) -> None:
-    combos = [(r, lo, hi) for r in rooms_list for lo, hi in price_ranges]
-    print(
-        f"\n=== PHASE COMBOS: {len(combos)} combos × ≤{pages} pages "
-        f"(city={city}, delay {delay}s, combo-delay {combo_delay}s) ===",
-        flush=True,
-    )
-    parser = YandexRealtyScraper(city=city)
-    grand_fetched = grand_ins = grand_upd = 0
-    async with AsyncSession(impersonate="chrome120", cookies=cookies or {}) as session:
-        for i, (rooms, lo, hi) in enumerate(combos, 1):
-            label = _combo_label(rooms, lo, hi)
-            print(f"\n  [{i}/{len(combos)}] combo={label}", flush=True)
-            try:
-                f, ins, upd = await sweep_one(
-                    session, parser, city, pages, delay,
-                    rooms=rooms, price_min=lo, price_max=hi,
-                )
-            except RuntimeError as e:
-                if str(e) == "captcha":
-                    print("  ⚠️ stopping combo run — captcha", flush=True)
-                    break
-                raise
-            grand_fetched += f
-            grand_ins += ins
-            grand_upd += upd
-            print(
-                f"  [{i}/{len(combos)}] {label} done: f={f} i={ins} u={upd} "
-                f"[grand: f={grand_fetched} i={grand_ins} u={grand_upd}]",
-                flush=True,
-            )
-            if i < len(combos):
-                await asyncio.sleep(combo_delay * random.uniform(0.85, 1.3))
-    print(
-        f"\n  COMBOS GRAND TOTAL: fetched={grand_fetched} "
-        f"ins={grand_ins} upd={grand_upd}",
-        flush=True,
-    )
-
-
-def _parse_rooms_arg(s: str) -> list[str]:
-    if s == "all":
-        return list(ROOM_PATH)
-    out: list[str] = []
-    for tok in s.split(","):
-        tok = tok.strip()
-        if tok not in ROOM_PATH:
-            raise argparse.ArgumentTypeError(
-                f"unknown room '{tok}'. Allowed: {sorted(ROOM_PATH)} or 'all'"
-            )
-        out.append(tok)
-    return out
-
-
-def _parse_price_ranges_arg(s: str) -> list[tuple[int | None, int | None]]:
-    if s == "all":
-        return list(PRICE_RANGES)
-    out: list[tuple[int | None, int | None]] = []
-    for tok in s.split(","):
-        tok = tok.strip()
-        if "-" not in tok:
-            raise argparse.ArgumentTypeError(f"range '{tok}' missing '-'")
-        lo_s, hi_s = tok.split("-", 1)
-        lo = int(float(lo_s) * 1_000_000) if lo_s else None
-        hi = int(float(hi_s) * 1_000_000) if hi_s else None
-        out.append((lo, hi))
-    return out
-
-
-async def main() -> None:
-    p = argparse.ArgumentParser(description="Yandex.Недвижимость ЕКБ sweep")
-    p.add_argument(
-        "--mode",
-        choices=["citywide", "combos"],
-        default="citywide",
-        help="citywide = один общий vtorichka sweep (~575 уникальных). "
-        "combos = rooms × price = ~4000+",
-    )
-    p.add_argument(
-        "--pages",
-        type=int,
-        default=25,
-        help="max pages per sweep (Yandex SERP cap ≈ 25; stops earlier on empty)",
-    )
-    p.add_argument(
-        "--delay",
-        type=float,
-        default=15,
-        help="sec между HTTP requests внутри combo (default 15, residential-safe)",
-    )
-    p.add_argument(
-        "--combo-delay",
-        type=float,
-        default=20,
-        help="sec между combos (default 20, дополнительный jitter combos mode)",
-    )
-    p.add_argument(
-        "--rooms",
-        type=_parse_rooms_arg,
-        default="all",
-        help=f"combos: comma-list из {sorted(ROOM_PATH)} или 'all' (default)",
-    )
-    p.add_argument(
-        "--price-ranges",
-        type=_parse_price_ranges_arg,
-        default="all",
-        help="combos: comma-list 'lo-hi' в млн руб ('-5,5-7,7-10,10-15,15-25,25-') "
-        "или 'all' (default). Empty side = open-ended.",
-    )
-    p.add_argument(
-        "--city",
-        type=str,
-        default=DEFAULT_CITY,
-        help=f"Yandex city slug (default '{DEFAULT_CITY}')",
-    )
-    p.add_argument(
-        "--cookies-file",
-        type=Path,
-        default=DEFAULT_COOKIES_FILE,
-        help=f"Browser cookies JSON (default '{DEFAULT_COOKIES_FILE.name}'). "
-        f"Empty/missing → start без cookies (риск captcha).",
-    )
-    args = p.parse_args()
-
-    cookies = _load_cookies(args.cookies_file)
-    if cookies:
-        print(f"loaded {len(cookies)} cookies from {args.cookies_file}", flush=True)
-    else:
-        print(
-            f"⚠️ no cookies loaded ({args.cookies_file}, exists={args.cookies_file.exists()}). "
-            f"Yandex anti-bot risk.",
-            flush=True,
-        )
-
-    t0 = time.perf_counter()
-    if args.mode == "citywide":
-        await phase_citywide(args.pages, args.delay, args.city, cookies)
-    else:
-        await phase_combos(
-            args.pages, args.delay, args.city, cookies,
-            rooms_list=args.rooms,
-            price_ranges=args.price_ranges,
-            combo_delay=args.combo_delay,
-        )
-    elapsed = time.perf_counter() - t0
-    print(f"\n=== ALL DONE in {elapsed / 60:.1f} min ===", flush=True)
-
-
-if __name__ == "__main__":
-    asyncio.run(main())
diff --git a/tradein-mvp/scripts/local-sweep-ekb.py b/tradein-mvp/scripts/local-sweep-ekb.py
deleted file mode 100644
index 35c2dd98..00000000
--- a/tradein-mvp/scripts/local-sweep-ekb.py
+++ /dev/null
@@ -1,443 +0,0 @@
-"""Full ЕКБ sweep + houses enrichment — runs locally через residential IP.
-
-Why: prod VPS Beget IP заблокирован Avito (datacenter fingerprint → 429).
-Local IP (домашний ISP) = residential → проходит. Скрипт пишет в prod DB
-через SSH tunnel (localhost:15433 → tradein-postgres docker container).
-
-Setup:
-  ssh -f -N -L 15433:172.20.0.2:5432 gendesign   # tradein-postgres tunnel
-  cd tradein-mvp/backend
-  export DATABASE_URL='postgresql+psycopg://tradein:@localhost:15433/tradein'
-
-Usage:
-  uv run python ../scripts/local-sweep-ekb.py                     # defaults
-  uv run python ../scripts/local-sweep-ekb.py --pages 3 --detail-top 20 --delay 30
-  uv run python ../scripts/local-sweep-ekb.py --skip-search       # только detail + houses
-  uv run python ../scripts/local-sweep-ekb.py --skip-houses       # без houses
-"""
-from __future__ import annotations
-
-# stdlib
-import argparse
-import asyncio
-import sys
-import time
-import types
-from pathlib import Path
-from urllib.parse import urlparse
-
-# Auto-inject backend/ в sys.path чтобы найти app.* package
-# (скрипт лежит в tradein-mvp/scripts/, app — в tradein-mvp/backend/)
-_BACKEND_DIR = Path(__file__).resolve().parent.parent / "backend"
-if str(_BACKEND_DIR) not in sys.path:
-    sys.path.insert(0, str(_BACKEND_DIR))
-
-# Stub weasyprint (Windows без GTK ломает import chain через trade_in.py)
-for _m in (
-    "weasyprint",
-    "weasyprint.text",
-    "weasyprint.text.ffi",
-    "weasyprint.css",
-    "weasyprint.css.targets",
-):
-    sys.modules[_m] = types.ModuleType(_m)
-sys.modules["weasyprint"].HTML = type("HTML", (), {"__init__": lambda self, *a, **kw: None})
-sys.modules["weasyprint"].CSS = type("CSS", (), {"__init__": lambda self, *a, **kw: None})
-
-# app imports после stubs
-from sqlalchemy import text  # noqa: E402
-
-from app.core.db import SessionLocal  # noqa: E402
-from app.services.scrapers.avito import AvitoScraper  # noqa: E402
-
-
-class CityWideAvitoScraper(AvitoScraper):
-    """Subclass без geo-фильтра — search всего города (ЕКБ) через pagination.
-
-    Override `_build_web_url` чтобы НЕ добавлять `geoCoords`/`radius` — Avito
-    отдаёт ВСЕ объявления в городе sorted by date, до 100 pages (cap ~5000).
-    """
-
-    def _build_web_url(self, lat: float, lon: float, radius_km: int, page: int = 1) -> str:  # noqa: ARG002
-        from urllib.parse import urlencode
-        params = {"s": 104, "p": page}  # s=104 sort by date
-        return f"{self.base_url}/ekaterinburg/kvartiry/prodam-ASgBAgICAUSSA8YQ?{urlencode(params)}"
-
-
-# Avito URL slugs для room-filter (extracted из real search page HTML 2026-05-23)
-ROOM_SLUGS: list[tuple[str, str]] = [
-    ("студии",       "studii-ASgBAgICAkSSA8YQygj~WA"),
-    ("1-комн.",      "1-komnatnye-ASgBAgICAkSSA8YQygiAWQ"),
-    ("2-комн.",      "2-komnatnye-ASgBAgICAkSSA8YQygiCWQ"),
-    ("3-комн.",      "3-komnatnye-ASgBAgICAkSSA8YQygiEWQ"),
-    ("4-комн.",      "4-komnatnye-ASgBAgICAkSSA8YQygiGWQ"),
-    ("5+комн.",      "5-komnatnye-ASgBAgICAkSSA8YQygiIWQ"),
-    ("своб.планир.", "svobodnaya_planirovka-ASgBAgICAkSSA8YQygj8zzI"),
-]
-
-
-class CityRoomsAvitoScraper(AvitoScraper):
-    """Subclass с URL slug filter по комнатности — для splitting когда total > 5000.
-
-    URL pattern: /ekaterinburg/kvartiry/prodam/?s=104&p=N
-    """
-
-    def __init__(self, slug: str) -> None:
-        super().__init__()
-        self.room_slug = slug
-
-    def _build_web_url(self, lat: float, lon: float, radius_km: int, page: int = 1) -> str:  # noqa: ARG002
-        from urllib.parse import urlencode
-        params = {"s": 104, "p": page}
-        return f"{self.base_url}/ekaterinburg/kvartiry/prodam/{self.room_slug}?{urlencode(params)}"
-from app.services.scrapers.avito_detail import (  # noqa: E402
-    fetch_detail,
-    save_detail_enrichment,
-)
-from app.services.scrapers.avito_houses import (  # noqa: E402
-    fetch_house_catalog,
-    save_house_catalog_enrichment,
-)
-from app.services.scrapers.base import save_listings  # noqa: E402
-
-# 10 anchors покрывающих центральные/северные/южные/восточные/западные районы ЕКБ
-ANCHORS: list[tuple[str, float, float]] = [
-    ("Центр",         56.8400, 60.6050),
-    ("ЮЗ",            56.7950, 60.5300),
-    ("Уралмаш",       56.8970, 60.6100),
-    ("Академический", 56.7700, 60.5500),
-    ("Пионерский",    56.8650, 60.6200),
-    ("Эльмаш",        56.8567, 60.6700),
-    ("Втузгородок",   56.8400, 60.6650),
-    ("ВИЗ",           56.8500, 60.5650),
-    ("Юго-Запад",     56.7770, 60.5800),
-    ("Сортировка",    56.8950, 60.5600),
-]
-
-
-async def phase_citywide(pages: int, delay: float) -> int:
-    """City-wide pagination — ВСЕ объявления ЕКБ (no geo filter), per-page save.
-
-    Per-page save важно: visible прогресс + не теряем данные если crash mid-way.
-    """
-    print(
-        f"\n=== PHASE CITYWIDE: full ЕКБ pagination ({pages} pages, delay {delay}s) ===\n",
-        flush=True,
-    )
-    total = 0
-    cum_ins = 0
-    cum_upd = 0
-    async with CityWideAvitoScraper() as s:
-        for page in range(1, pages + 1):
-            url = s._build_web_url(0.0, 0.0, 1, page=page)
-            try:
-                assert s._cffi is not None
-                response = await s._cffi.get(url)
-                if response.status_code != 200:
-                    print(f"  ⚠️ page={page} HTTP {response.status_code} — stop", flush=True)
-                    break
-                lots = s._parse_html(response.text, source_url_base=url)
-            except Exception as e:  # noqa: BLE001
-                print(f"  page={page} FAIL fetch: {type(e).__name__}: {e}", flush=True)
-                await asyncio.sleep(delay)
-                continue
-            if not lots:
-                print(f"  page={page}: 0 lots — end of pagination", flush=True)
-                break
-            db = SessionLocal()
-            try:
-                ins, upd = save_listings(db, lots)
-                cum_ins += ins
-                cum_upd += upd
-                total += len(lots)
-                print(
-                    f"  page={page}/{pages}: {len(lots)} lots → ins={ins} upd={upd} "
-                    f"[cum: total={total} new={cum_ins} upd={cum_upd}]",
-                    flush=True,
-                )
-            except Exception as e:  # noqa: BLE001
-                print(f"  page={page} FAIL save: {type(e).__name__}: {e}", flush=True)
-            finally:
-                db.close()
-            if page < pages:
-                await asyncio.sleep(delay)
-    print(f"  PHASE CITYWIDE TOTAL: fetched={total} ins={cum_ins} upd={cum_upd}", flush=True)
-    return total
-
-
-async def phase_byrooms(pages: int, delay: float) -> int:
-    """Rooms-split sweep — 7 categories × pages, per-page save.
-
-    Для total > 5000 (Avito search cap) split по комнатности позволяет получить
-    все listings (каждый segment <5000).
-    """
-    print(
-        f"\n=== PHASE BYROOMS: 7 категорий × {pages} pages, delay {delay}s ===\n",
-        flush=True,
-    )
-    grand_total = 0
-    grand_ins = 0
-    grand_upd = 0
-    import random
-    rate_limit_hits = 0
-    for ci, (name, slug) in enumerate(ROOM_SLUGS, 1):
-        print(f"\n--- [{ci}/{len(ROOM_SLUGS)}] Категория: {name} (slug={slug[:30]}...) ---", flush=True)
-        cat_total = 0
-        cat_ins = 0
-        cat_upd = 0
-        async with CityRoomsAvitoScraper(slug) as s:
-            for page in range(1, pages + 1):
-                url = s._build_web_url(0.0, 0.0, 1, page=page)
-                try:
-                    assert s._cffi is not None
-                    response = await s._cffi.get(url)
-                    if response.status_code == 429:
-                        # Exponential backoff на rate-limit: 60s × 2^N (max 30 min)
-                        rate_limit_hits += 1
-                        backoff = min(60 * (2 ** (rate_limit_hits - 1)), 1800)
-                        print(f"  {name} p={page} HTTP 429 (hit #{rate_limit_hits}) → backoff {backoff}s", flush=True)
-                        await asyncio.sleep(backoff)
-                        # retry same page after backoff
-                        try:
-                            response = await s._cffi.get(url)
-                            if response.status_code != 200:
-                                print(f"  {name} p={page} retry HTTP {response.status_code} — skip page", flush=True)
-                                await asyncio.sleep(delay * 2)
-                                continue
-                        except Exception as e:  # noqa: BLE001
-                            print(f"  {name} p={page} retry FAIL: {type(e).__name__}: {e}", flush=True)
-                            continue
-                    elif response.status_code != 200:
-                        print(f"  {name} p={page} HTTP {response.status_code} — skip page", flush=True)
-                        await asyncio.sleep(delay)
-                        continue
-                    lots = s._parse_html(response.text, source_url_base=url)
-                except Exception as e:  # noqa: BLE001
-                    print(f"  {name} p={page} FAIL fetch: {type(e).__name__}: {e}", flush=True)
-                    await asyncio.sleep(delay)
-                    continue
-                if not lots:
-                    print(f"  {name} page={page}: 0 lots — end of category", flush=True)
-                    break
-                db = SessionLocal()
-                try:
-                    ins, upd = save_listings(db, lots)
-                    cat_ins += ins
-                    cat_upd += upd
-                    cat_total += len(lots)
-                    print(
-                        f"  {name} p={page}/{pages}: {len(lots)} lots → ins={ins} upd={upd} "
-                        f"[cat: tot={cat_total} new={cat_ins}]",
-                        flush=True,
-                    )
-                except Exception as e:  # noqa: BLE001
-                    print(f"  {name} page={page} FAIL save: {type(e).__name__}: {e}", flush=True)
-                finally:
-                    db.close()
-                if page < pages:
-                    # Jitter: ±30% от delay чтобы выглядеть менее robotic
-                    jittered = delay * random.uniform(0.7, 1.3)
-                    await asyncio.sleep(jittered)
-        print(
-            f"  --- {name} TOTAL: fetched={cat_total} ins={cat_ins} upd={cat_upd} ---",
-            flush=True,
-        )
-        grand_total += cat_total
-        grand_ins += cat_ins
-        grand_upd += cat_upd
-    print(
-        f"\n  PHASE BYROOMS GRAND TOTAL: fetched={grand_total} new={grand_ins} updated={grand_upd}",
-        flush=True,
-    )
-    return grand_total
-
-
-async def phase_a_search(pages: int, delay: float) -> int:
-    """Search всех anchors × pages, save listings."""
-    print(
-        f"\n=== PHASE A: search ({len(ANCHORS)} anchors × {pages} pages, "
-        f"delay {delay}s) ===\n",
-        flush=True,
-    )
-    total = 0
-    for i, (name, lat, lon) in enumerate(ANCHORS, 1):
-        print(f"[{i}/{len(ANCHORS)}] {name} ({lat},{lon})", flush=True)
-        try:
-            async with AvitoScraper() as s:
-                lots = await s.fetch_around(
-                    lat, lon, 1500, pages=pages, delay_override_sec=delay
-                )
-        except Exception as e:  # noqa: BLE001
-            print(f"  FAIL fetch: {type(e).__name__}: {e}", flush=True)
-            await asyncio.sleep(delay)
-            continue
-        print(f"  fetched {len(lots)} lots", flush=True)
-        if not lots:
-            print("  ⚠️ 0 lots — possible 429, stop phase A", flush=True)
-            break
-        db = SessionLocal()
-        try:
-            ins, upd = save_listings(db, lots)
-            print(f"  saved: ins={ins} upd={upd}", flush=True)
-            total += len(lots)
-        except Exception as e:  # noqa: BLE001
-            print(f"  FAIL save: {type(e).__name__}: {e}", flush=True)
-        finally:
-            db.close()
-        if i < len(ANCHORS):
-            await asyncio.sleep(delay)
-    return total
-
-
-async def phase_b_detail(top_n: int, delay: float) -> tuple[int, set[str]]:
-    """Detail enrichment для top-N recent listings без detail.
-
-    Returns: (enriched_count, set of unique house_catalog_url paths extracted)
-    """
-    print(
-        f"\n=== PHASE B: detail enrichment (top {top_n}, delay {delay}s) ===\n",
-        flush=True,
-    )
-    db = SessionLocal()
-    try:
-        rows = db.execute(
-            text(
-                """
-                SELECT source_url, source_id FROM listings
-                WHERE source='avito' AND detail_enriched_at IS NULL
-                  AND source_url IS NOT NULL AND price_rub > 0
-                ORDER BY scraped_at DESC NULLS LAST
-                LIMIT :limit
-                """
-            ),
-            {"limit": top_n},
-        ).mappings().all()
-    finally:
-        db.close()
-
-    print(f"  queue: {len(rows)} listings to enrich", flush=True)
-    enriched = 0
-    failed = 0
-    house_urls: set[str] = set()
-    for i, row in enumerate(rows, 1):
-        url = row["source_url"]
-        path = urlparse(url).path if url.startswith("http") else url
-        print(f"[{i}/{len(rows)}] detail {row['source_id']}", flush=True)
-        try:
-            enr = await fetch_detail(path)
-            db = SessionLocal()
-            try:
-                ok = save_detail_enrichment(db, enr)
-                print(
-                    f"  enriched: ok={ok} "
-                    f"rooms={enr.rooms} area={enr.area_m2} "
-                    f"kitchen={enr.kitchen_area_m2} owners={enr.owners_count} "
-                    f"house_url={enr.house_catalog_url}",
-                    flush=True,
-                )
-                if ok:
-                    enriched += 1
-                # Извлекаем house path (strip query string + filter только /catalog/houses/)
-                if enr.house_catalog_url:
-                    p = urlparse(enr.house_catalog_url).path
-                    if p.startswith("/catalog/houses/"):
-                        house_urls.add(p)
-            finally:
-                db.close()
-        except Exception as e:  # noqa: BLE001
-            print(f"  FAIL: {type(e).__name__}: {e}", flush=True)
-            failed += 1
-        if i < len(rows):
-            await asyncio.sleep(delay)
-    print(f"  total: enriched={enriched} failed={failed} unique_houses_extracted={len(house_urls)}", flush=True)
-    return enriched, house_urls
-
-
-async def phase_c_houses(delay: float, extra_urls: set[str] | None = None) -> int:
-    """Houses enrichment — unique house_url из listings (DB) + extra_urls (из Phase B in-memory).
-
-    save_detail_enrichment не пишет house_url в listings (баг Stage 2b),
-    поэтому Phase C принимает extra_urls собранные Phase B вручную.
-    """
-    print("\n=== PHASE C: houses enrichment ===\n", flush=True)
-    db = SessionLocal()
-    try:
-        rows = db.execute(
-            text(
-                """
-                SELECT DISTINCT house_url FROM listings
-                WHERE source='avito' AND house_url IS NOT NULL
-                  AND house_id_fk IS NULL
-                ORDER BY house_url
-                """
-            )
-        ).fetchall()
-    finally:
-        db.close()
-
-    urls_from_db = {r[0] for r in rows if r[0]}
-    urls = urls_from_db | (extra_urls or set())
-    print(f"  queue: {len(urls)} unique houses (db={len(urls_from_db)} +extras={len(extra_urls or set())})", flush=True)
-    if not urls:
-        print("  no houses to enrich (skip)", flush=True)
-        return 0
-
-    enriched = 0
-    failed = 0
-    for i, url in enumerate(sorted(urls), 1):
-        path = urlparse(url).path if url.startswith("http") else url
-        print(f"[{i}/{len(urls)}] {path}", flush=True)
-        try:
-            enr = await fetch_house_catalog(path)
-            db = SessionLocal()
-            try:
-                counters = save_house_catalog_enrichment(db, enr)
-                print(f"  saved: {counters}", flush=True)
-                enriched += 1
-            finally:
-                db.close()
-        except Exception as e:  # noqa: BLE001
-            print(f"  FAIL: {type(e).__name__}: {e}", flush=True)
-            failed += 1
-        if i < len(urls):
-            await asyncio.sleep(delay)
-    print(f"  total: enriched={enriched} failed={failed}", flush=True)
-    return enriched
-
-
-async def main() -> None:
-    p = argparse.ArgumentParser()
-    p.add_argument(
-        "--mode",
-        choices=["anchors", "citywide", "byrooms", "both"],
-        default="anchors",
-        help="anchors=10 anchors; citywide=full pagination без geo; byrooms=7 категорий комнат (для total>5000); both=anchors+citywide",
-    )
-    p.add_argument("--pages", type=int, default=5, help="pages per anchor (anchors mode) или total pages (citywide)")
-    p.add_argument("--detail-top", type=int, default=50, help="top-N для detail enrichment")
-    p.add_argument("--delay", type=float, default=30, help="sec между HTTP requests")
-    p.add_argument("--skip-search", action="store_true")
-    p.add_argument("--skip-detail", action="store_true")
-    p.add_argument("--skip-houses", action="store_true")
-    args = p.parse_args()
-
-    t0 = time.perf_counter()
-    if not args.skip_search:
-        if args.mode in ("anchors", "both"):
-            await phase_a_search(args.pages, args.delay)
-        if args.mode in ("citywide", "both"):
-            cw_pages = args.pages if args.mode == "citywide" else 100
-            await phase_citywide(cw_pages, args.delay)
-        if args.mode == "byrooms":
-            await phase_byrooms(args.pages, args.delay)
-    extracted_house_urls: set[str] = set()
-    if not args.skip_detail:
-        _, extracted_house_urls = await phase_b_detail(args.detail_top, args.delay)
-    if not args.skip_houses:
-        await phase_c_houses(args.delay, extra_urls=extracted_house_urls)
-    elapsed = time.perf_counter() - t0
-    print(f"\n=== ALL DONE in {elapsed / 60:.1f} min ===", flush=True)
-
-
-if __name__ == "__main__":
-    asyncio.run(main())