Merge pull request 'fix(tradein): 095 drop v_data_quality before dropping sale_type_text — RED deploy (#731)' (#903) from fix/731-sale-type-text-view-dep into main

Reviewed-on: #903

tradein-mvp/backend/data/sql/095_dead_schema.sql

@@ -5,9 +5,18 @@
 --   sale_type_text: столбец 100% NULL в listings, code refs удалены в том же PR.
 --   audit_log: 0 строк, 0 python refs, триггеры не зависят (верифицировано).
 --
+-- ⚠️ View-зависимость (фикс после падения деплоя 2026-05-31):
+--   v_data_quality (recreated в 094) содержит CTE `SELECT * FROM listings`, что
+--   фиксирует column-level зависимость на ВСЕ колонки listings, включая
+--   sale_type_text — поэтому `DROP COLUMN sale_type_text` падал
+--   ("cannot drop ... other objects depend on it"). Решение: drop view →
+--   drop column → recreate view ТЕМ ЖЕ DDL (после дропа колонки `SELECT *`
+--   её просто не включит). Проверено: только v_data_quality делает SELECT *
+--   из listings; listings_search_mv / v_price_divergence берут явные колонки.
+--
 -- Idempotency:
---   ALTER TABLE ... DROP COLUMN IF EXISTS — безопасен при повторном запуске.
+--   DROP VIEW / DROP COLUMN IF EXISTS / DROP TABLE IF EXISTS / CREATE OR REPLACE —
---   DROP TABLE IF EXISTS — безопасен при повторном запуске.
+--   безопасны при повторном запуске.
 --
 -- Явно НЕ удаляем: houses_price_dynamics, agents, sellers, estimate_photos,
 --                  house_reviews, listings_snapshots.
@@ -15,9 +24,64 @@
 BEGIN;
 
 -- #731: drop dead schema (verified-safe subset of TradeinDb_Audit_May30)
--- sale_type_text: 100% NULL in listings, code refs removed in same PR
+
+-- v_data_quality зависит от listings.sale_type_text через `SELECT * FROM listings`
+-- (см. 094). Дропаем view перед дропом колонки, пересоздаём ниже тем же DDL.
+DROP VIEW IF EXISTS v_data_quality;
+
+-- sale_type_text: 100% NULL in listings, code refs removed in same PR (#731)
 ALTER TABLE IF EXISTS listings DROP COLUMN IF EXISTS sale_type_text;
 
+-- Recreate v_data_quality (DDL идентичен 094 — sale_type_text не в выводе, только
+-- неявно через SELECT * CTE; после дропа колонки * её не включает).
+CREATE OR REPLACE VIEW v_data_quality AS
+WITH active_listings AS (
+    SELECT * FROM listings WHERE is_active = true
+)
+SELECT
+    (SELECT count(*) FROM houses)                                                  AS houses_total,
+    (SELECT count(*) FROM houses h
+       WHERE EXISTS (SELECT 1 FROM house_sources hs WHERE hs.house_id = h.id))      AS houses_with_source,
+    (SELECT count(*) FROM houses h
+       WHERE EXISTS (SELECT 1 FROM house_sources hs
+                      WHERE hs.house_id = h.id AND hs.ext_source = 'avito'))        AS houses_with_avito,
+    (SELECT count(*) FROM houses h
+       WHERE EXISTS (SELECT 1 FROM house_sources hs
+                      WHERE hs.house_id = h.id AND hs.ext_source LIKE 'cian%'))     AS houses_with_cian,
+    (SELECT count(*) FROM houses h
+       WHERE EXISTS (SELECT 1 FROM house_sources hs
+                      WHERE hs.house_id = h.id AND hs.ext_source = 'yandex'))       AS houses_with_yandex,
+    (SELECT count(*) FROM (
+        SELECT house_id FROM house_sources GROUP BY house_id HAVING count(*) >= 2
+    ) sub)                                                                          AS houses_2plus_sources,
+    (SELECT count(*) FROM (
+        SELECT house_id FROM house_sources GROUP BY house_id HAVING count(*) >= 3
+    ) sub)                                                                          AS houses_3plus_sources,
+    (SELECT count(*) FROM active_listings)                                         AS listings_active,
+    (SELECT count(*) FROM (
+        SELECT listing_id FROM listing_sources
+         WHERE listing_id IN (SELECT id FROM active_listings)
+         GROUP BY listing_id HAVING count(*) >= 2
+    ) sub)                                                                          AS listings_dedup_2sources,
+    (SELECT count(*) FROM active_listings WHERE lat IS NOT NULL) * 100.0
+        / NULLIF((SELECT count(*) FROM active_listings), 0)                         AS pct_geocoded,
+    (SELECT count(*) FROM active_listings WHERE cadastral_number IS NOT NULL) * 100.0
+        / NULLIF((SELECT count(*) FROM active_listings), 0)                         AS pct_cadastr,
+    (SELECT count(*) FROM active_listings WHERE description IS NOT NULL) * 100.0
+        / NULLIF((SELECT count(*) FROM active_listings), 0)                         AS pct_description,
+    (SELECT count(*) FROM active_listings l
+       JOIN houses h ON h.id = l.house_id_fk
+       WHERE h.year_built IS NOT NULL) * 100.0
+        / NULLIF((SELECT count(*) FROM active_listings), 0)                         AS pct_year_built,
+    NOW() - (SELECT max(scraped_at) FROM listings WHERE source = 'avito')           AS avito_last_scrape_ago,
+    NOW() - (SELECT max(scraped_at) FROM listings WHERE source = 'cian')            AS cian_last_scrape_ago,
+    NOW() - (SELECT max(scraped_at) FROM listings WHERE source = 'yandex')          AS yandex_last_scrape_ago,
+    (SELECT count(*) FROM v_price_divergence)                                       AS price_disagreements_count,
+    (SELECT count(*) FROM listings WHERE is_outlier = true)                         AS outliers_flagged;
+
+COMMENT ON VIEW v_data_quality IS
+  'KPI snapshot. Refreshed on-demand by /api/v1/admin/data-quality endpoint (Master Plan sec 8.1).';
+
 -- audit_log: 0 rows, 0 python refs, no triggers depend on it (verified)
 DROP TABLE IF EXISTS audit_log;