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`.
2026-02-28 22:13:11 +03:00
parent 2d629b6e80
commit 85fea19818
37 changed files with 2704 additions and 27 deletions
@@ -0,0 +1,104 @@
+-- Загрузка ODS по airports: SCD1 (UPDATE изменившихся + INSERT новых).
+
+-- Statement 1: UPDATE существующих строк.
+WITH src AS (
+    SELECT
+        s.airport_code,
+        s.airport_name,
+        s.city,
+        s.country,
+        s.coordinates,
+        s.timezone,
+        ROW_NUMBER() OVER (
+            PARTITION BY s.airport_code
+            ORDER BY s.load_dttm DESC, s.src_created_at_ts DESC NULLS LAST
+        ) AS rn
+    FROM stg.airports AS s
+    WHERE s.batch_id = '{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text
+)
+UPDATE ods.airports AS o
+SET airport_name = s.airport_name,
+    city         = s.city,
+    country      = s.country,
+    coordinates  = s.coordinates,
+    timezone     = s.timezone,
+    _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.airport_code = s.airport_code
+    AND (
+        o.airport_name IS DISTINCT FROM s.airport_name
+        OR o.city IS DISTINCT FROM s.city
+        OR o.country IS DISTINCT FROM s.country
+        OR o.coordinates IS DISTINCT FROM s.coordinates
+        OR o.timezone IS DISTINCT FROM s.timezone
+    );
+
+-- Statement 2: INSERT новых строк.
+WITH src AS (
+    SELECT
+        s.airport_code,
+        s.airport_name,
+        s.city,
+        s.country,
+        s.coordinates,
+        s.timezone,
+        ROW_NUMBER() OVER (
+            PARTITION BY s.airport_code
+            ORDER BY s.load_dttm DESC, s.src_created_at_ts DESC NULLS LAST
+        ) AS rn
+    FROM stg.airports AS s
+    WHERE s.batch_id = '{{ ti.xcom_pull(task_ids="resolve_stg_batch_id") }}'::text
+)
+INSERT INTO ods.airports (
+    airport_code,
+    airport_name,
+    city,
+    country,
+    coordinates,
+    timezone,
+    _load_id,
+    _load_ts
+)
+SELECT
+    s.airport_code,
+    s.airport_name,
+    s.city,
+    s.country,
+    s.coordinates,
+    s.timezone,
+    '{{ 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.airports AS o
+        WHERE o.airport_code = s.airport_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.airport_code
+    FROM (
+        SELECT
+            s.airport_code,
+            ROW_NUMBER() OVER (
+                PARTITION BY s.airport_code
+                ORDER BY s.load_dttm DESC, s.src_created_at_ts DESC NULLS LAST
+            ) AS rn
+        FROM stg.airports 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.airports AS o
+WHERE NOT EXISTS (
+    SELECT 1
+    FROM src_keys AS s
+    WHERE s.airport_code = o.airport_code
+);
+
+ANALYZE ods.airports;