-- 139_premium_houses.sql
-- Purpose: premium_houses materialized view — premium / elite EKB buildings
--          identified by listing concentration (#2002, premium-building identification).
--
-- Rationale (#2002):
--   We flag a house as "premium" by the concentration of expensive listings on it.
--   Validated query returns ~297 houses that together cover ~65% of all >=20M RUB
--   elite listings. A house qualifies when it has a meaningful sample (n >= 4 active
--   listings with sane price_per_m2) AND either:
--     - a high median price_per_m2 (>= 220000 RUB/m2), OR
--     - at least 3 listings priced >= 20M RUB (n_elite >= 3).
--   price_per_m2 is clamped to [100000, 800000] to drop scrape garbage / mis-parsed
--   areas before aggregating, so the median is robust.
--
-- house_id dupes: house_dedup is imperfect, so the same physical building can appear
--   under several house_id values. That is FINE here — we deliberately keep every
--   variant in the set; dropping dupes would only shrink coverage, never improve it.
--   The unique index is on house_id (each MV row is one house_id), not on a physical
--   building, so distinct house_id rows never collide.
--
-- Refresh: refreshed by the existing MV-refresh task (app/tasks/refresh_search_matview.py,
--   scheduled as `refresh_search_matview` in scrape_schedules) right AFTER
--   listings_search_mv — premium_houses derives from `listings`, so refreshing it once
--   the search MV is fresh keeps both in step. REFRESH ... CONCURRENTLY requires the
--   UNIQUE index below.
--
-- Dependencies: 002_core_tables.sql (listings: house_id_fk, price_rub, price_per_m2,
--               area_m2, is_active).
-- Deploy order: after 002_core_tables.sql. No backend schema coupling.
-- Re-run safe: CREATE MATERIALIZED VIEW IF NOT EXISTS + CREATE UNIQUE INDEX IF NOT EXISTS.
--   No DROP — dropping would break in-flight REFRESH ... CONCURRENTLY and momentarily
--   delete the object during deploy.

BEGIN;

CREATE MATERIALIZED VIEW IF NOT EXISTS premium_houses AS
WITH lh AS (
    SELECT l.house_id_fk AS house_id,
        count(*) AS n,
        count(*) FILTER (WHERE l.price_rub >= 20000000) AS n_elite,
        percentile_cont(0.5) WITHIN GROUP (ORDER BY l.price_per_m2)
            FILTER (WHERE l.price_per_m2 BETWEEN 100000 AND 800000) AS med_ppm2
    FROM listings l
    WHERE l.is_active AND l.house_id_fk IS NOT NULL AND l.area_m2 > 0
          AND l.price_per_m2 BETWEEN 100000 AND 800000
    GROUP BY l.house_id_fk
)
SELECT house_id, n, n_elite, round(med_ppm2)::int AS med_ppm2
FROM lh
WHERE n >= 4 AND (med_ppm2 >= 220000 OR n_elite >= 3);

-- UNIQUE index is mandatory for REFRESH MATERIALIZED VIEW CONCURRENTLY.
CREATE UNIQUE INDEX IF NOT EXISTS premium_houses_house_id_uidx
    ON premium_houses (house_id);

COMMIT;