feat(main): заглушки студенческих файлов, ослабление DQ, адаптация тестов

- fact_flight_sales_dq: student SK (passenger/route/airplane) → NOTICE, airport_sk → порог 1% (эталон через ods.routes) - routes_dq: закомментирован RI airplane_code → ods.airplanes - ODS DAG: убрана зависимость dq_ods_airplanes → load_ods_routes - 18 студенческих файлов заменены заглушками (ODS/DDS/DM load+dq) - test_ods_sql_contract: SNAPSHOT_ENTITIES без airplanes/seats - test_dags_smoke: убран assert dq_ods_airplanes → dq_ods_routes Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 23:07:14 +03:00
parent a8148c8afa
commit 5ca4e6b2ca
23 changed files with 120 additions and 1703 deletions
@@ -1,96 +1,4 @@
-- Загрузка DDS dim_airplanes: SCD1 UPSERT (UPDATE изменившихся + INSERT новых).
-
-- Statement 1: UPDATE существующих записей (если атрибуты изменились).
-WITH seats_agg AS (
-    SELECT
-        s.airplane_code,
-        COUNT(*)::INTEGER AS total_seats
-    FROM ods.seats AS s
-    GROUP BY s.airplane_code
-),
-src AS (
-    SELECT
-        a.airplane_code,
-        a.model,
-        a.range_km,
-        a.speed_kmh,
-        COALESCE(sa.total_seats, 0) AS total_seats
-    FROM ods.airplanes AS a
-    LEFT JOIN seats_agg AS sa
-        ON sa.airplane_code = a.airplane_code
-)
-UPDATE dds.dim_airplanes AS d
-SET model       = s.model,
-    range_km    = s.range_km,
-    speed_kmh   = s.speed_kmh,
-    total_seats = s.total_seats,
-    updated_at  = now(),
-    _load_id    = '{{ run_id }}',
-    _load_ts    = now()
-FROM src AS s
-WHERE d.airplane_bk = s.airplane_code
-    AND (
-        d.model IS DISTINCT FROM s.model
-        OR d.range_km IS DISTINCT FROM s.range_km
-        OR d.speed_kmh IS DISTINCT FROM s.speed_kmh
-        OR d.total_seats IS DISTINCT FROM s.total_seats
-    );
-
-- Statement 2: INSERT новых записей (MAX(sk) + ROW_NUMBER()).
-WITH seats_agg AS (
-    SELECT
-        s.airplane_code,
-        COUNT(*)::INTEGER AS total_seats
-    FROM ods.seats AS s
-    GROUP BY s.airplane_code
-),
-src AS (
-    SELECT
-        a.airplane_code,
-        a.model,
-        a.range_km,
-        a.speed_kmh,
-        COALESCE(sa.total_seats, 0) AS total_seats
-    FROM ods.airplanes AS a
-    LEFT JOIN seats_agg AS sa
-        ON sa.airplane_code = a.airplane_code
-),
-- Учебный комментарий: Генерация SK через MAX() + ROW_NUMBER()
-- Этот подход работает безопасно только потому, что Airflow запускает
-- джобы загрузки для одной таблицы строго последовательно (concurrency=1).
-- При параллельной загрузке возникнет состояние гонки (race condition) и возможны дубли SK.
-max_sk AS (
-    SELECT COALESCE(MAX(airplane_sk), 0) AS v
-    FROM dds.dim_airplanes
-)
-INSERT INTO dds.dim_airplanes (
-    airplane_sk,
-    airplane_bk,
-    model,
-    range_km,
-    speed_kmh,
-    total_seats,
-    created_at,
-    updated_at,
-    _load_id,
-    _load_ts
-)
-SELECT
-    (SELECT v FROM max_sk) + ROW_NUMBER() OVER (ORDER BY s.airplane_code)::INTEGER,
-    s.airplane_code,
-    s.model,
-    s.range_km,
-    s.speed_kmh,
-    s.total_seats,
-    now(),
-    now(),
-    '{{ run_id }}',
-    now()
-FROM src AS s
-WHERE NOT EXISTS (
-    SELECT 1
-    FROM dds.dim_airplanes AS d
-    WHERE d.airplane_bk = s.airplane_code
-);
-
-ANALYZE dds.dim_airplanes;
+-- TODO: реализуйте загрузку dds.dim_airplanes (см. ТЗ в docs/assignment/analyst_spec.md)
+-- Паттерн: SCD1 UPSERT (UPDATE изменившихся + INSERT новых).
+-- Эталонную реализацию можно найти в ветке solution.
+SELECT 1;