feat(ods): реализован ODS слой и DAG загрузки из STG

- Зачем:
  - подготовлена учебная реализация ODS слоя с типизацией, UPSERT и DQ, чтобы продолжить работу от STG к DDS/DM.
- Что:
  - добавлены SQL-скрипты `sql/ods/*_ddl.sql`, `sql/ods/*_load.sql`, `sql/ods/*_dq.sql` для 9 сущностей bookings.
  - добавлены DAG `bookings_ods_ddl` и `bookings_to_gp_ods`, а также smoke-тесты для новых графов.
  - ODS DDL интегрирован в `sql/ddl_gp.sql`; документация и план обновлены под единый запуск через `make ddl-gp`.
- Проверка:
  - `make test`.
  - `make ddl-gp`.
This commit is contained in:
2026-02-28 22:13:11 +03:00
parent 2d629b6e80
commit 85fea19818
37 changed files with 2704 additions and 27 deletions
+92
View File
@@ -0,0 +1,92 @@
-- Загрузка ODS по airplanes: SCD1 (UPDATE изменившихся + INSERT новых).
-- Statement 1: UPDATE существующих строк.
WITH src AS (
SELECT
s.airplane_code,
s.model,
NULLIF(s.range, '')::INTEGER AS range_km,
NULLIF(s.speed, '')::INTEGER AS speed_kmh,
ROW_NUMBER() OVER (
PARTITION BY s.airplane_code
ORDER BY s.load_dttm DESC, s.src_created_at_ts DESC NULLS LAST
) AS rn
FROM stg.airplanes AS s
WHERE s.batch_id = '{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text
)
UPDATE ods.airplanes AS o
SET model = s.model,
range_km = s.range_km,
speed_kmh = s.speed_kmh,
_load_id = '{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text,
_load_ts = now()
FROM src AS s
WHERE s.rn = 1
AND o.airplane_code = s.airplane_code
AND (
o.model IS DISTINCT FROM s.model
OR o.range_km IS DISTINCT FROM s.range_km
OR o.speed_kmh IS DISTINCT FROM s.speed_kmh
);
-- Statement 2: INSERT новых строк.
WITH src AS (
SELECT
s.airplane_code,
s.model,
NULLIF(s.range, '')::INTEGER AS range_km,
NULLIF(s.speed, '')::INTEGER AS speed_kmh,
ROW_NUMBER() OVER (
PARTITION BY s.airplane_code
ORDER BY s.load_dttm DESC, s.src_created_at_ts DESC NULLS LAST
) AS rn
FROM stg.airplanes AS s
WHERE s.batch_id = '{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text
)
INSERT INTO ods.airplanes (
airplane_code,
model,
range_km,
speed_kmh,
_load_id,
_load_ts
)
SELECT
s.airplane_code,
s.model,
s.range_km,
s.speed_kmh,
'{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text,
now()
FROM src AS s
WHERE s.rn = 1
AND NOT EXISTS (
SELECT 1
FROM ods.airplanes AS o
WHERE o.airplane_code = s.airplane_code
);
-- Statement 3: DELETE ключей, которых нет в snapshot текущего батча.
-- Это делает ODS для справочника действительно "current state".
WITH src_keys AS (
SELECT d.airplane_code
FROM (
SELECT
s.airplane_code,
ROW_NUMBER() OVER (
PARTITION BY s.airplane_code
ORDER BY s.load_dttm DESC, s.src_created_at_ts DESC NULLS LAST
) AS rn
FROM stg.airplanes AS s
WHERE s.batch_id = '{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text
) AS d
WHERE d.rn = 1
)
DELETE FROM ods.airplanes AS o
WHERE NOT EXISTS (
SELECT 1
FROM src_keys AS s
WHERE s.airplane_code = o.airplane_code
);
ANALYZE ods.airplanes;