feat(dds): реализован слой dds для bookings
- Зачем: - подготовлен учебный Star Schema слой для перехода от ODS к аналитике и витринам. - Что: - добавлены 21 SQL-файл для DDS (DDL/LOAD/DQ) с SCD1/SCD2 и фактом `fact_flight_sales`. - добавлены DAG `bookings_dds_ddl` и `bookings_to_gp_dds`, а также smoke-тесты структуры DAG. - обновлены `sql/ddl_gp.sql` и документация (`README`, `docs/*`, `db_schema`) под поток `stg -> ods -> dds`. - Проверка: - make test.
This commit is contained in:
@@ -0,0 +1,17 @@
|
||||
-- DDL для DDS-слоя по таблице dim_airplanes (SCD1-измерение).
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.dim_airplanes (
|
||||
airplane_sk INTEGER NOT NULL,
|
||||
airplane_bk TEXT NOT NULL,
|
||||
model TEXT NOT NULL,
|
||||
range_km INTEGER,
|
||||
speed_kmh INTEGER,
|
||||
total_seats INTEGER,
|
||||
created_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
_load_id TEXT NOT NULL,
|
||||
_load_ts TIMESTAMP NOT NULL DEFAULT now()
|
||||
)
|
||||
DISTRIBUTED BY (airplane_sk);
|
||||
@@ -0,0 +1,83 @@
|
||||
-- DQ для DDS dim_airplanes.
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_dup_sk BIGINT;
|
||||
v_dup_bk BIGINT;
|
||||
v_missing_bk BIGINT;
|
||||
v_null_count BIGINT;
|
||||
BEGIN
|
||||
-- Таблица не пуста.
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.dim_airplanes;
|
||||
|
||||
IF v_row_count = 0 THEN
|
||||
RAISE EXCEPTION 'DQ FAILED: dds.dim_airplanes пуста.';
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по SK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT airplane_sk)
|
||||
INTO v_dup_sk
|
||||
FROM dds.dim_airplanes;
|
||||
|
||||
IF v_dup_sk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airplanes найдены дубликаты airplane_sk: %',
|
||||
v_dup_sk;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по BK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT airplane_bk)
|
||||
INTO v_dup_bk
|
||||
FROM dds.dim_airplanes;
|
||||
|
||||
IF v_dup_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airplanes найдены дубликаты airplane_bk: %',
|
||||
v_dup_bk;
|
||||
END IF;
|
||||
|
||||
-- Покрытие ODS: все airplane_code из ODS есть в DDS.
|
||||
SELECT COUNT(*)
|
||||
INTO v_missing_bk
|
||||
FROM (SELECT DISTINCT airplane_code FROM ods.airplanes) AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_airplanes AS d
|
||||
WHERE d.airplane_bk = s.airplane_code
|
||||
);
|
||||
|
||||
IF v_missing_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airplanes отсутствуют ключи из ods.airplanes: %',
|
||||
v_missing_bk;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_airplanes
|
||||
WHERE airplane_sk IS NULL
|
||||
OR airplane_bk IS NULL
|
||||
OR airplane_bk = ''
|
||||
OR model IS NULL
|
||||
OR model = ''
|
||||
OR total_seats IS NULL
|
||||
OR created_at IS NULL
|
||||
OR updated_at IS NULL
|
||||
OR _load_id IS NULL
|
||||
OR _load_id = ''
|
||||
OR _load_ts IS NULL;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airplanes найдены NULL/пустые обязательные поля: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_airplanes ок, строк=%',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,92 @@
|
||||
-- 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
|
||||
),
|
||||
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;
|
||||
@@ -0,0 +1,18 @@
|
||||
-- DDL для DDS-слоя по таблице dim_airports (SCD1-измерение).
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.dim_airports (
|
||||
airport_sk INTEGER NOT NULL,
|
||||
airport_bk TEXT NOT NULL,
|
||||
airport_name TEXT NOT NULL,
|
||||
city TEXT NOT NULL,
|
||||
country TEXT NOT NULL,
|
||||
timezone TEXT NOT NULL,
|
||||
coordinates TEXT,
|
||||
created_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
_load_id TEXT NOT NULL,
|
||||
_load_ts TIMESTAMP NOT NULL DEFAULT now()
|
||||
)
|
||||
DISTRIBUTED BY (airport_sk);
|
||||
@@ -0,0 +1,88 @@
|
||||
-- DQ для DDS dim_airports.
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_dup_sk BIGINT;
|
||||
v_dup_bk BIGINT;
|
||||
v_missing_bk BIGINT;
|
||||
v_null_count BIGINT;
|
||||
BEGIN
|
||||
-- Таблица не пуста.
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.dim_airports;
|
||||
|
||||
IF v_row_count = 0 THEN
|
||||
RAISE EXCEPTION 'DQ FAILED: dds.dim_airports пуста.';
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по SK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT airport_sk)
|
||||
INTO v_dup_sk
|
||||
FROM dds.dim_airports;
|
||||
|
||||
IF v_dup_sk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airports найдены дубликаты airport_sk: %',
|
||||
v_dup_sk;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по BK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT airport_bk)
|
||||
INTO v_dup_bk
|
||||
FROM dds.dim_airports;
|
||||
|
||||
IF v_dup_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airports найдены дубликаты airport_bk: %',
|
||||
v_dup_bk;
|
||||
END IF;
|
||||
|
||||
-- Покрытие ODS: все airport_code из ODS есть в DDS.
|
||||
SELECT COUNT(*)
|
||||
INTO v_missing_bk
|
||||
FROM (SELECT DISTINCT airport_code FROM ods.airports) AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_airports AS d
|
||||
WHERE d.airport_bk = s.airport_code
|
||||
);
|
||||
|
||||
IF v_missing_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airports отсутствуют ключи из ods.airports: %',
|
||||
v_missing_bk;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_airports
|
||||
WHERE airport_sk IS NULL
|
||||
OR airport_bk IS NULL
|
||||
OR airport_bk = ''
|
||||
OR airport_name IS NULL
|
||||
OR airport_name = ''
|
||||
OR city IS NULL
|
||||
OR city = ''
|
||||
OR country IS NULL
|
||||
OR country = ''
|
||||
OR timezone IS NULL
|
||||
OR timezone = ''
|
||||
OR created_at IS NULL
|
||||
OR updated_at IS NULL
|
||||
OR _load_id IS NULL
|
||||
OR _load_id = ''
|
||||
OR _load_ts IS NULL;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_airports найдены NULL/пустые обязательные поля: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_airports ок, строк=%',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,60 @@
|
||||
-- Загрузка DDS dim_airports: SCD1 UPSERT (UPDATE изменившихся + INSERT новых).
|
||||
|
||||
-- Statement 1: UPDATE существующих записей (если атрибуты изменились).
|
||||
UPDATE dds.dim_airports AS d
|
||||
SET airport_name = s.airport_name,
|
||||
city = s.city,
|
||||
country = s.country,
|
||||
timezone = s.timezone,
|
||||
coordinates = s.coordinates,
|
||||
updated_at = now(),
|
||||
_load_id = '{{ run_id }}',
|
||||
_load_ts = now()
|
||||
FROM ods.airports AS s
|
||||
WHERE d.airport_bk = s.airport_code
|
||||
AND (
|
||||
d.airport_name IS DISTINCT FROM s.airport_name
|
||||
OR d.city IS DISTINCT FROM s.city
|
||||
OR d.country IS DISTINCT FROM s.country
|
||||
OR d.timezone IS DISTINCT FROM s.timezone
|
||||
OR d.coordinates IS DISTINCT FROM s.coordinates
|
||||
);
|
||||
|
||||
-- Statement 2: INSERT новых записей (MAX(sk) + ROW_NUMBER()).
|
||||
WITH max_sk AS (
|
||||
SELECT COALESCE(MAX(airport_sk), 0) AS v
|
||||
FROM dds.dim_airports
|
||||
)
|
||||
INSERT INTO dds.dim_airports (
|
||||
airport_sk,
|
||||
airport_bk,
|
||||
airport_name,
|
||||
city,
|
||||
country,
|
||||
timezone,
|
||||
coordinates,
|
||||
created_at,
|
||||
updated_at,
|
||||
_load_id,
|
||||
_load_ts
|
||||
)
|
||||
SELECT
|
||||
(SELECT v FROM max_sk) + ROW_NUMBER() OVER (ORDER BY s.airport_code)::INTEGER,
|
||||
s.airport_code,
|
||||
s.airport_name,
|
||||
s.city,
|
||||
s.country,
|
||||
s.timezone,
|
||||
s.coordinates,
|
||||
now(),
|
||||
now(),
|
||||
'{{ run_id }}',
|
||||
now()
|
||||
FROM ods.airports AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_airports AS d
|
||||
WHERE d.airport_bk = s.airport_code
|
||||
);
|
||||
|
||||
ANALYZE dds.dim_airports;
|
||||
@@ -0,0 +1,15 @@
|
||||
-- DDL для DDS-слоя по таблице dim_calendar (статическое измерение дат).
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.dim_calendar (
|
||||
calendar_sk INTEGER NOT NULL,
|
||||
date_actual DATE NOT NULL,
|
||||
year_actual INTEGER NOT NULL,
|
||||
month_actual INTEGER NOT NULL,
|
||||
day_actual INTEGER NOT NULL,
|
||||
day_of_week INTEGER NOT NULL,
|
||||
day_name TEXT NOT NULL,
|
||||
is_weekend BOOLEAN NOT NULL
|
||||
)
|
||||
DISTRIBUTED BY (calendar_sk);
|
||||
@@ -0,0 +1,87 @@
|
||||
-- DQ для DDS dim_calendar.
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_dup_sk BIGINT;
|
||||
v_dup_date BIGINT;
|
||||
v_null_count BIGINT;
|
||||
v_missing_flight_dates BIGINT;
|
||||
BEGIN
|
||||
-- Таблица должна быть достаточно заполнена.
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.dim_calendar;
|
||||
|
||||
IF v_row_count < 1000 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: dds.dim_calendar содержит слишком мало строк: %',
|
||||
v_row_count;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по surrogate key.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT calendar_sk)
|
||||
INTO v_dup_sk
|
||||
FROM dds.dim_calendar;
|
||||
|
||||
IF v_dup_sk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_calendar найдены дубликаты calendar_sk: %',
|
||||
v_dup_sk;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по business key (date_actual).
|
||||
SELECT COUNT(*) - COUNT(DISTINCT date_actual)
|
||||
INTO v_dup_date
|
||||
FROM dds.dim_calendar;
|
||||
|
||||
IF v_dup_date <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_calendar найдены дубликаты date_actual: %',
|
||||
v_dup_date;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля не NULL.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_calendar
|
||||
WHERE calendar_sk IS NULL
|
||||
OR date_actual IS NULL
|
||||
OR year_actual IS NULL
|
||||
OR month_actual IS NULL
|
||||
OR day_actual IS NULL
|
||||
OR day_of_week IS NULL
|
||||
OR day_name IS NULL
|
||||
OR day_name = ''
|
||||
OR is_weekend IS NULL;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_calendar найдены NULL/пустые обязательные поля: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
-- Календарь покрывает даты вылета из ODS (где scheduled_departure не NULL).
|
||||
SELECT COUNT(*)
|
||||
INTO v_missing_flight_dates
|
||||
FROM (
|
||||
SELECT DISTINCT f.scheduled_departure::DATE AS departure_date
|
||||
FROM ods.flights AS f
|
||||
WHERE f.scheduled_departure IS NOT NULL
|
||||
) AS src
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_calendar AS c
|
||||
WHERE c.date_actual = src.departure_date
|
||||
);
|
||||
|
||||
IF v_missing_flight_dates <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: dds.dim_calendar не покрывает даты вылета из ods.flights: %',
|
||||
v_missing_flight_dates;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_calendar ок, строк=%',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,29 @@
|
||||
-- Загрузка DDS dim_calendar: статическое измерение (генерация дат).
|
||||
-- Заполняем только если таблица пуста (идемпотентно).
|
||||
|
||||
INSERT INTO dds.dim_calendar (
|
||||
calendar_sk,
|
||||
date_actual,
|
||||
year_actual,
|
||||
month_actual,
|
||||
day_actual,
|
||||
day_of_week,
|
||||
day_name,
|
||||
is_weekend
|
||||
)
|
||||
SELECT
|
||||
ROW_NUMBER() OVER (ORDER BY d.date_actual)::INTEGER AS calendar_sk,
|
||||
d.date_actual,
|
||||
EXTRACT(YEAR FROM d.date_actual)::INTEGER AS year_actual,
|
||||
EXTRACT(MONTH FROM d.date_actual)::INTEGER AS month_actual,
|
||||
EXTRACT(DAY FROM d.date_actual)::INTEGER AS day_actual,
|
||||
EXTRACT(ISODOW FROM d.date_actual)::INTEGER AS day_of_week,
|
||||
TO_CHAR(d.date_actual, 'FMDay') AS day_name,
|
||||
EXTRACT(ISODOW FROM d.date_actual) IN (6, 7) AS is_weekend
|
||||
FROM (
|
||||
SELECT generate_series('2016-01-01'::DATE, '2030-12-31'::DATE, '1 day'::INTERVAL)::DATE
|
||||
AS date_actual
|
||||
) AS d
|
||||
WHERE NOT EXISTS (SELECT 1 FROM dds.dim_calendar LIMIT 1);
|
||||
|
||||
ANALYZE dds.dim_calendar;
|
||||
@@ -0,0 +1,14 @@
|
||||
-- DDL для DDS-слоя по таблице dim_passengers (SCD1-измерение).
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.dim_passengers (
|
||||
passenger_sk INTEGER NOT NULL,
|
||||
passenger_bk TEXT NOT NULL,
|
||||
passenger_name TEXT NOT NULL,
|
||||
created_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
_load_id TEXT NOT NULL,
|
||||
_load_ts TIMESTAMP NOT NULL DEFAULT now()
|
||||
)
|
||||
DISTRIBUTED BY (passenger_sk);
|
||||
@@ -0,0 +1,100 @@
|
||||
-- DQ для DDS dim_passengers.
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_src_count BIGINT;
|
||||
v_dup_sk BIGINT;
|
||||
v_dup_bk BIGINT;
|
||||
v_missing_bk BIGINT;
|
||||
v_null_count BIGINT;
|
||||
BEGIN
|
||||
-- Для инкрементальных периодов без новых билетов допускаем пустую dim_passengers.
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.dim_passengers;
|
||||
|
||||
SELECT COUNT(DISTINCT passenger_id)
|
||||
INTO v_src_count
|
||||
FROM ods.tickets
|
||||
WHERE passenger_id IS NOT NULL
|
||||
AND passenger_id <> '';
|
||||
|
||||
IF v_row_count = 0 AND v_src_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: dds.dim_passengers пуста при непустом источнике ods.tickets (passenger_id=%).',
|
||||
v_src_count;
|
||||
ELSIF v_row_count = 0 AND v_src_count = 0 THEN
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_passengers пуста, т.к. в ods.tickets нет passenger_id для загрузки.';
|
||||
RETURN;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по SK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT passenger_sk)
|
||||
INTO v_dup_sk
|
||||
FROM dds.dim_passengers;
|
||||
|
||||
IF v_dup_sk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_passengers найдены дубликаты passenger_sk: %',
|
||||
v_dup_sk;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по BK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT passenger_bk)
|
||||
INTO v_dup_bk
|
||||
FROM dds.dim_passengers;
|
||||
|
||||
IF v_dup_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_passengers найдены дубликаты passenger_bk: %',
|
||||
v_dup_bk;
|
||||
END IF;
|
||||
|
||||
-- Покрытие ODS: все passenger_id из ods.tickets есть в DDS.
|
||||
SELECT COUNT(*)
|
||||
INTO v_missing_bk
|
||||
FROM (
|
||||
SELECT DISTINCT passenger_id
|
||||
FROM ods.tickets
|
||||
WHERE passenger_id IS NOT NULL
|
||||
AND passenger_id <> ''
|
||||
) AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_passengers AS d
|
||||
WHERE d.passenger_bk = s.passenger_id
|
||||
);
|
||||
|
||||
IF v_missing_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_passengers отсутствуют passenger_id из ods.tickets: %',
|
||||
v_missing_bk;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_passengers
|
||||
WHERE passenger_sk IS NULL
|
||||
OR passenger_bk IS NULL
|
||||
OR passenger_bk = ''
|
||||
OR passenger_name IS NULL
|
||||
OR passenger_name = ''
|
||||
OR created_at IS NULL
|
||||
OR updated_at IS NULL
|
||||
OR _load_id IS NULL
|
||||
OR _load_id = ''
|
||||
OR _load_ts IS NULL;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_passengers найдены NULL/пустые обязательные поля: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_passengers ок, строк=%',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,82 @@
|
||||
-- Загрузка DDS dim_passengers: SCD1 UPSERT (UPDATE изменившихся + INSERT новых).
|
||||
|
||||
-- Statement 1: UPDATE существующих записей (если атрибуты изменились).
|
||||
WITH src AS (
|
||||
SELECT
|
||||
d.passenger_id,
|
||||
d.passenger_name
|
||||
FROM (
|
||||
SELECT
|
||||
t.passenger_id,
|
||||
t.passenger_name,
|
||||
ROW_NUMBER() OVER (
|
||||
PARTITION BY t.passenger_id
|
||||
ORDER BY t.event_ts DESC NULLS LAST, t._load_ts DESC, t.ticket_no DESC
|
||||
) AS rn
|
||||
FROM ods.tickets AS t
|
||||
WHERE t.passenger_id IS NOT NULL
|
||||
AND t.passenger_id <> ''
|
||||
AND t.passenger_name IS NOT NULL
|
||||
AND t.passenger_name <> ''
|
||||
) AS d
|
||||
WHERE d.rn = 1
|
||||
)
|
||||
UPDATE dds.dim_passengers AS d
|
||||
SET passenger_name = s.passenger_name,
|
||||
updated_at = now(),
|
||||
_load_id = '{{ run_id }}',
|
||||
_load_ts = now()
|
||||
FROM src AS s
|
||||
WHERE d.passenger_bk = s.passenger_id
|
||||
AND d.passenger_name IS DISTINCT FROM s.passenger_name;
|
||||
|
||||
-- Statement 2: INSERT новых записей (MAX(sk) + ROW_NUMBER()).
|
||||
WITH src AS (
|
||||
SELECT
|
||||
d.passenger_id,
|
||||
d.passenger_name
|
||||
FROM (
|
||||
SELECT
|
||||
t.passenger_id,
|
||||
t.passenger_name,
|
||||
ROW_NUMBER() OVER (
|
||||
PARTITION BY t.passenger_id
|
||||
ORDER BY t.event_ts DESC NULLS LAST, t._load_ts DESC, t.ticket_no DESC
|
||||
) AS rn
|
||||
FROM ods.tickets AS t
|
||||
WHERE t.passenger_id IS NOT NULL
|
||||
AND t.passenger_id <> ''
|
||||
AND t.passenger_name IS NOT NULL
|
||||
AND t.passenger_name <> ''
|
||||
) AS d
|
||||
WHERE d.rn = 1
|
||||
),
|
||||
max_sk AS (
|
||||
SELECT COALESCE(MAX(passenger_sk), 0) AS v
|
||||
FROM dds.dim_passengers
|
||||
)
|
||||
INSERT INTO dds.dim_passengers (
|
||||
passenger_sk,
|
||||
passenger_bk,
|
||||
passenger_name,
|
||||
created_at,
|
||||
updated_at,
|
||||
_load_id,
|
||||
_load_ts
|
||||
)
|
||||
SELECT
|
||||
(SELECT v FROM max_sk) + ROW_NUMBER() OVER (ORDER BY s.passenger_id)::INTEGER,
|
||||
s.passenger_id,
|
||||
s.passenger_name,
|
||||
now(),
|
||||
now(),
|
||||
'{{ run_id }}',
|
||||
now()
|
||||
FROM src AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_passengers AS d
|
||||
WHERE d.passenger_bk = s.passenger_id
|
||||
);
|
||||
|
||||
ANALYZE dds.dim_passengers;
|
||||
@@ -0,0 +1,22 @@
|
||||
-- DDL для DDS-слоя по таблице dim_routes (SCD2-измерение).
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.dim_routes (
|
||||
route_sk INTEGER NOT NULL,
|
||||
route_bk TEXT NOT NULL,
|
||||
departure_airport TEXT NOT NULL,
|
||||
arrival_airport TEXT NOT NULL,
|
||||
airplane_code TEXT NOT NULL,
|
||||
days_of_week TEXT,
|
||||
departure_time TIME,
|
||||
duration INTERVAL,
|
||||
hashdiff TEXT NOT NULL,
|
||||
valid_from DATE NOT NULL,
|
||||
valid_to DATE,
|
||||
created_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
_load_id TEXT NOT NULL,
|
||||
_load_ts TIMESTAMP NOT NULL DEFAULT now()
|
||||
)
|
||||
DISTRIBUTED BY (route_sk);
|
||||
@@ -0,0 +1,149 @@
|
||||
-- DQ для DDS dim_routes (SCD2).
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_dup_sk BIGINT;
|
||||
v_dup_current BIGINT;
|
||||
v_overlap_count BIGINT;
|
||||
v_missing_count BIGINT;
|
||||
v_orphan_current BIGINT;
|
||||
v_null_count BIGINT;
|
||||
BEGIN
|
||||
-- Таблица не пуста.
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.dim_routes;
|
||||
|
||||
IF v_row_count = 0 THEN
|
||||
RAISE EXCEPTION 'DQ FAILED: dds.dim_routes пуста.';
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по SK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT route_sk)
|
||||
INTO v_dup_sk
|
||||
FROM dds.dim_routes;
|
||||
|
||||
IF v_dup_sk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_routes найдены дубликаты route_sk: %',
|
||||
v_dup_sk;
|
||||
END IF;
|
||||
|
||||
-- Корректность интервалов (valid_from <= valid_to для закрытых версий).
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_routes
|
||||
WHERE valid_to IS NOT NULL
|
||||
AND valid_from > valid_to;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_routes найдены версии с valid_from > valid_to: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
-- Нет перекрытий интервалов для одного route_bk.
|
||||
SELECT COUNT(*)
|
||||
INTO v_overlap_count
|
||||
FROM (
|
||||
SELECT 1
|
||||
FROM dds.dim_routes AS d1
|
||||
JOIN dds.dim_routes AS d2
|
||||
ON d1.route_bk = d2.route_bk
|
||||
AND d1.route_sk < d2.route_sk
|
||||
AND d1.valid_from < COALESCE(d2.valid_to, DATE '9999-12-31')
|
||||
AND d2.valid_from < COALESCE(d1.valid_to, DATE '9999-12-31')
|
||||
) AS overlap_rows;
|
||||
|
||||
IF v_overlap_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_routes найдены перекрытия SCD2-интервалов: %',
|
||||
v_overlap_count;
|
||||
END IF;
|
||||
|
||||
-- Не более одной текущей версии на route_bk.
|
||||
SELECT COUNT(*)
|
||||
INTO v_dup_current
|
||||
FROM (
|
||||
SELECT route_bk
|
||||
FROM dds.dim_routes
|
||||
WHERE valid_to IS NULL
|
||||
GROUP BY route_bk
|
||||
HAVING COUNT(*) > 1
|
||||
) AS d;
|
||||
|
||||
IF v_dup_current <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_routes найдены route_bk с > 1 текущей версией: %',
|
||||
v_dup_current;
|
||||
END IF;
|
||||
|
||||
-- Покрытие ODS: все route_no имеют хотя бы одну версию в DDS.
|
||||
SELECT COUNT(*)
|
||||
INTO v_missing_count
|
||||
FROM (SELECT DISTINCT route_no FROM ods.routes) AS o
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_routes AS d
|
||||
WHERE d.route_bk = o.route_no
|
||||
);
|
||||
|
||||
IF v_missing_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_routes отсутствуют маршруты из ODS: %',
|
||||
v_missing_count;
|
||||
END IF;
|
||||
|
||||
-- Current-срез DDS не содержит route_bk, которых нет в ODS.
|
||||
SELECT COUNT(*)
|
||||
INTO v_orphan_current
|
||||
FROM (
|
||||
SELECT DISTINCT route_bk
|
||||
FROM dds.dim_routes
|
||||
WHERE valid_to IS NULL
|
||||
) AS d
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM ods.routes AS o
|
||||
WHERE o.route_no = d.route_bk
|
||||
);
|
||||
|
||||
IF v_orphan_current <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в current-срезе dds.dim_routes есть route_bk вне ODS: %',
|
||||
v_orphan_current;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_routes
|
||||
WHERE route_sk IS NULL
|
||||
OR route_bk IS NULL
|
||||
OR route_bk = ''
|
||||
OR departure_airport IS NULL
|
||||
OR departure_airport = ''
|
||||
OR arrival_airport IS NULL
|
||||
OR arrival_airport = ''
|
||||
OR airplane_code IS NULL
|
||||
OR airplane_code = ''
|
||||
OR hashdiff IS NULL
|
||||
OR hashdiff = ''
|
||||
OR valid_from IS NULL
|
||||
OR created_at IS NULL
|
||||
OR updated_at IS NULL
|
||||
OR _load_id IS NULL
|
||||
OR _load_id = ''
|
||||
OR _load_ts IS NULL;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_routes найдены NULL обязательные поля: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_routes ок, строк=% (версий)',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,130 @@
|
||||
-- Загрузка DDS dim_routes: SCD2 с hashdiff.
|
||||
|
||||
-- Statement 1: Закрыть устаревшие версии (valid_to = текущая дата).
|
||||
WITH src AS (
|
||||
SELECT
|
||||
route_no,
|
||||
departure_airport,
|
||||
arrival_airport,
|
||||
airplane_code,
|
||||
days_of_week,
|
||||
departure_time,
|
||||
duration,
|
||||
md5(
|
||||
COALESCE(departure_airport, '') || '|' ||
|
||||
COALESCE(arrival_airport, '') || '|' ||
|
||||
COALESCE(airplane_code, '') || '|' ||
|
||||
COALESCE(days_of_week, '') || '|' ||
|
||||
COALESCE(departure_time::TEXT, '') || '|' ||
|
||||
COALESCE(duration::TEXT, '')
|
||||
) AS hashdiff,
|
||||
ROW_NUMBER() OVER (PARTITION BY route_no ORDER BY validity DESC) AS rn
|
||||
FROM ods.routes
|
||||
)
|
||||
UPDATE dds.dim_routes AS d
|
||||
SET valid_to = CURRENT_DATE,
|
||||
updated_at = now(),
|
||||
_load_id = '{{ run_id }}',
|
||||
_load_ts = now()
|
||||
FROM src AS s
|
||||
WHERE s.rn = 1
|
||||
AND d.route_bk = s.route_no
|
||||
AND d.valid_to IS NULL
|
||||
AND d.hashdiff <> s.hashdiff;
|
||||
|
||||
-- Statement 1.1: Закрыть "исчезнувшие" маршруты.
|
||||
WITH src AS (
|
||||
SELECT
|
||||
route_no,
|
||||
ROW_NUMBER() OVER (PARTITION BY route_no ORDER BY validity DESC) AS rn
|
||||
FROM ods.routes
|
||||
)
|
||||
UPDATE dds.dim_routes AS d
|
||||
SET valid_to = CURRENT_DATE,
|
||||
updated_at = now(),
|
||||
_load_id = '{{ run_id }}',
|
||||
_load_ts = now()
|
||||
WHERE d.valid_to IS NULL
|
||||
AND NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM src AS s
|
||||
WHERE s.rn = 1
|
||||
AND s.route_no = d.route_bk
|
||||
);
|
||||
|
||||
-- Statement 2: Вставить новые версии (для изменённых и новых route_no).
|
||||
WITH src AS (
|
||||
SELECT
|
||||
route_no,
|
||||
departure_airport,
|
||||
arrival_airport,
|
||||
airplane_code,
|
||||
days_of_week,
|
||||
departure_time,
|
||||
duration,
|
||||
md5(
|
||||
COALESCE(departure_airport, '') || '|' ||
|
||||
COALESCE(arrival_airport, '') || '|' ||
|
||||
COALESCE(airplane_code, '') || '|' ||
|
||||
COALESCE(days_of_week, '') || '|' ||
|
||||
COALESCE(departure_time::TEXT, '') || '|' ||
|
||||
COALESCE(duration::TEXT, '')
|
||||
) AS hashdiff,
|
||||
ROW_NUMBER() OVER (PARTITION BY route_no ORDER BY validity DESC) AS rn
|
||||
FROM ods.routes
|
||||
),
|
||||
max_sk AS (
|
||||
SELECT COALESCE(MAX(route_sk), 0) AS v
|
||||
FROM dds.dim_routes
|
||||
)
|
||||
INSERT INTO dds.dim_routes (
|
||||
route_sk,
|
||||
route_bk,
|
||||
departure_airport,
|
||||
arrival_airport,
|
||||
airplane_code,
|
||||
days_of_week,
|
||||
departure_time,
|
||||
duration,
|
||||
hashdiff,
|
||||
valid_from,
|
||||
valid_to,
|
||||
created_at,
|
||||
updated_at,
|
||||
_load_id,
|
||||
_load_ts
|
||||
)
|
||||
SELECT
|
||||
(SELECT v FROM max_sk) + ROW_NUMBER() OVER (ORDER BY s.route_no)::INTEGER,
|
||||
s.route_no,
|
||||
s.departure_airport,
|
||||
s.arrival_airport,
|
||||
s.airplane_code,
|
||||
s.days_of_week,
|
||||
s.departure_time,
|
||||
s.duration,
|
||||
s.hashdiff,
|
||||
CASE
|
||||
WHEN EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_routes AS d2
|
||||
WHERE d2.route_bk = s.route_no
|
||||
) THEN CURRENT_DATE
|
||||
ELSE '1900-01-01'::DATE
|
||||
END AS valid_from,
|
||||
NULL,
|
||||
now(),
|
||||
now(),
|
||||
'{{ run_id }}',
|
||||
now()
|
||||
FROM src AS s
|
||||
WHERE s.rn = 1
|
||||
AND NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_routes AS d
|
||||
WHERE d.route_bk = s.route_no
|
||||
AND d.valid_to IS NULL
|
||||
AND d.hashdiff = s.hashdiff
|
||||
);
|
||||
|
||||
ANALYZE dds.dim_routes;
|
||||
@@ -0,0 +1,13 @@
|
||||
-- DDL для DDS-слоя по таблице dim_tariffs (SCD1-измерение).
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.dim_tariffs (
|
||||
tariff_sk INTEGER NOT NULL,
|
||||
fare_conditions TEXT NOT NULL,
|
||||
created_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMP NOT NULL DEFAULT now(),
|
||||
_load_id TEXT NOT NULL,
|
||||
_load_ts TIMESTAMP NOT NULL DEFAULT now()
|
||||
)
|
||||
DISTRIBUTED BY (tariff_sk);
|
||||
@@ -0,0 +1,98 @@
|
||||
-- DQ для DDS dim_tariffs.
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_src_count BIGINT;
|
||||
v_dup_sk BIGINT;
|
||||
v_dup_bk BIGINT;
|
||||
v_missing_bk BIGINT;
|
||||
v_null_count BIGINT;
|
||||
BEGIN
|
||||
-- Для инкрементальных периодов без новых сегментов допускаем пустую dim_tariffs.
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.dim_tariffs;
|
||||
|
||||
SELECT COUNT(DISTINCT fare_conditions)
|
||||
INTO v_src_count
|
||||
FROM ods.segments
|
||||
WHERE fare_conditions IS NOT NULL
|
||||
AND fare_conditions <> '';
|
||||
|
||||
IF v_row_count = 0 AND v_src_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: dds.dim_tariffs пуста при непустом источнике ods.segments (fare_conditions=%).',
|
||||
v_src_count;
|
||||
ELSIF v_row_count = 0 AND v_src_count = 0 THEN
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_tariffs пуста, т.к. в ods.segments нет тарифов для загрузки.';
|
||||
RETURN;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по SK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT tariff_sk)
|
||||
INTO v_dup_sk
|
||||
FROM dds.dim_tariffs;
|
||||
|
||||
IF v_dup_sk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_tariffs найдены дубликаты tariff_sk: %',
|
||||
v_dup_sk;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по BK.
|
||||
SELECT COUNT(*) - COUNT(DISTINCT fare_conditions)
|
||||
INTO v_dup_bk
|
||||
FROM dds.dim_tariffs;
|
||||
|
||||
IF v_dup_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_tariffs найдены дубликаты fare_conditions: %',
|
||||
v_dup_bk;
|
||||
END IF;
|
||||
|
||||
-- Покрытие ODS: все fare_conditions из ods.segments есть в DDS.
|
||||
SELECT COUNT(*)
|
||||
INTO v_missing_bk
|
||||
FROM (
|
||||
SELECT DISTINCT fare_conditions
|
||||
FROM ods.segments
|
||||
WHERE fare_conditions IS NOT NULL
|
||||
AND fare_conditions <> ''
|
||||
) AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_tariffs AS d
|
||||
WHERE d.fare_conditions = s.fare_conditions
|
||||
);
|
||||
|
||||
IF v_missing_bk <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_tariffs отсутствуют значения fare_conditions из ods.segments: %',
|
||||
v_missing_bk;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_count
|
||||
FROM dds.dim_tariffs
|
||||
WHERE tariff_sk IS NULL
|
||||
OR fare_conditions IS NULL
|
||||
OR fare_conditions = ''
|
||||
OR created_at IS NULL
|
||||
OR updated_at IS NULL
|
||||
OR _load_id IS NULL
|
||||
OR _load_id = ''
|
||||
OR _load_ts IS NULL;
|
||||
|
||||
IF v_null_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.dim_tariffs найдены NULL/пустые обязательные поля: %',
|
||||
v_null_count;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.dim_tariffs ок, строк=%',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,36 @@
|
||||
-- Загрузка DDS dim_tariffs: SCD1 UPSERT (INSERT новых тарифов).
|
||||
|
||||
WITH src AS (
|
||||
SELECT DISTINCT
|
||||
s.fare_conditions
|
||||
FROM ods.segments AS s
|
||||
WHERE s.fare_conditions IS NOT NULL
|
||||
AND s.fare_conditions <> ''
|
||||
),
|
||||
max_sk AS (
|
||||
SELECT COALESCE(MAX(tariff_sk), 0) AS v
|
||||
FROM dds.dim_tariffs
|
||||
)
|
||||
INSERT INTO dds.dim_tariffs (
|
||||
tariff_sk,
|
||||
fare_conditions,
|
||||
created_at,
|
||||
updated_at,
|
||||
_load_id,
|
||||
_load_ts
|
||||
)
|
||||
SELECT
|
||||
(SELECT v FROM max_sk) + ROW_NUMBER() OVER (ORDER BY s.fare_conditions)::INTEGER,
|
||||
s.fare_conditions,
|
||||
now(),
|
||||
now(),
|
||||
'{{ run_id }}',
|
||||
now()
|
||||
FROM src AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.dim_tariffs AS d
|
||||
WHERE d.fare_conditions = s.fare_conditions
|
||||
);
|
||||
|
||||
ANALYZE dds.dim_tariffs;
|
||||
@@ -0,0 +1,23 @@
|
||||
-- DDL для DDS-слоя по таблице fact_flight_sales.
|
||||
|
||||
CREATE SCHEMA IF NOT EXISTS dds;
|
||||
|
||||
CREATE TABLE IF NOT EXISTS dds.fact_flight_sales (
|
||||
calendar_sk INTEGER,
|
||||
departure_airport_sk INTEGER,
|
||||
arrival_airport_sk INTEGER,
|
||||
airplane_sk INTEGER,
|
||||
tariff_sk INTEGER,
|
||||
passenger_sk INTEGER,
|
||||
route_sk INTEGER,
|
||||
book_ref TEXT NOT NULL,
|
||||
ticket_no TEXT NOT NULL,
|
||||
flight_id INTEGER NOT NULL,
|
||||
book_date DATE,
|
||||
seat_no TEXT,
|
||||
price NUMERIC(10,2),
|
||||
is_boarded BOOLEAN NOT NULL,
|
||||
_load_id TEXT NOT NULL,
|
||||
_load_ts TIMESTAMP NOT NULL DEFAULT now()
|
||||
)
|
||||
DISTRIBUTED BY (ticket_no);
|
||||
@@ -0,0 +1,151 @@
|
||||
-- DQ для DDS fact_flight_sales.
|
||||
|
||||
DO $$
|
||||
DECLARE
|
||||
v_row_count BIGINT;
|
||||
v_ods_count BIGINT;
|
||||
v_dup_count BIGINT;
|
||||
v_null_passenger BIGINT;
|
||||
v_null_tariff BIGINT;
|
||||
v_null_route_related BIGINT;
|
||||
v_null_calendar BIGINT;
|
||||
v_null_required BIGINT;
|
||||
BEGIN
|
||||
-- Для пустого инкрементального окна (ods.segments) допускаем пустой факт.
|
||||
SELECT COUNT(*)
|
||||
INTO v_ods_count
|
||||
FROM ods.segments;
|
||||
|
||||
SELECT COUNT(*)
|
||||
INTO v_row_count
|
||||
FROM dds.fact_flight_sales;
|
||||
|
||||
IF v_ods_count = 0 THEN
|
||||
IF v_row_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: ods.segments пустая, но в dds.fact_flight_sales есть строки: %',
|
||||
v_row_count;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: ods.segments и dds.fact_flight_sales пустые (инкрементальное окно без сегментов).';
|
||||
RETURN;
|
||||
END IF;
|
||||
|
||||
IF v_row_count = 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: dds.fact_flight_sales пуста при непустом источнике ods.segments (%).',
|
||||
v_ods_count;
|
||||
END IF;
|
||||
|
||||
-- Покрытие: количество строк = ods.segments.
|
||||
IF v_row_count <> v_ods_count THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: dds.fact_flight_sales (%) <> ods.segments (%). Потеряны строки.',
|
||||
v_row_count,
|
||||
v_ods_count;
|
||||
END IF;
|
||||
|
||||
-- Нет дублей по зерну.
|
||||
SELECT COUNT(*)
|
||||
INTO v_dup_count
|
||||
FROM (
|
||||
SELECT ticket_no, flight_id
|
||||
FROM dds.fact_flight_sales
|
||||
GROUP BY ticket_no, flight_id
|
||||
HAVING COUNT(*) > 1
|
||||
) AS d;
|
||||
|
||||
IF v_dup_count <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в dds.fact_flight_sales дубликаты (ticket_no, flight_id): %',
|
||||
v_dup_count;
|
||||
END IF;
|
||||
|
||||
-- Ссылочная целостность: passenger_sk.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_passenger
|
||||
FROM dds.fact_flight_sales
|
||||
WHERE passenger_sk IS NULL;
|
||||
|
||||
IF v_null_passenger <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в fact_flight_sales строки без passenger_sk: %',
|
||||
v_null_passenger;
|
||||
END IF;
|
||||
|
||||
-- Ссылочная целостность: tariff_sk.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_tariff
|
||||
FROM dds.fact_flight_sales
|
||||
WHERE tariff_sk IS NULL;
|
||||
|
||||
IF v_null_tariff <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в fact_flight_sales строки без tariff_sk: %',
|
||||
v_null_tariff;
|
||||
END IF;
|
||||
|
||||
-- Route-related FK: допустимо при аномалиях, фейлим если > 1%.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_route_related
|
||||
FROM dds.fact_flight_sales
|
||||
WHERE route_sk IS NULL
|
||||
OR departure_airport_sk IS NULL
|
||||
OR arrival_airport_sk IS NULL
|
||||
OR airplane_sk IS NULL;
|
||||
|
||||
IF v_null_route_related > 0 THEN
|
||||
IF v_null_route_related * 100.0 / NULLIF(v_row_count, 0) > 1.0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в fact_flight_sales слишком много строк с NULL в route-related FK: % (>1%%)',
|
||||
v_null_route_related;
|
||||
ELSE
|
||||
RAISE NOTICE
|
||||
'DQ WARNING: в fact_flight_sales строк с NULL в route-related FK: % (<=1%%, допустимо)',
|
||||
v_null_route_related;
|
||||
END IF;
|
||||
END IF;
|
||||
|
||||
-- Calendar: допустимо если scheduled_departure IS NULL, фейлим если > 1%.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_calendar
|
||||
FROM dds.fact_flight_sales
|
||||
WHERE calendar_sk IS NULL;
|
||||
|
||||
IF v_null_calendar > 0 THEN
|
||||
IF v_null_calendar * 100.0 / NULLIF(v_row_count, 0) > 1.0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в fact_flight_sales слишком много строк без calendar_sk: % (>1%%)',
|
||||
v_null_calendar;
|
||||
ELSE
|
||||
RAISE NOTICE
|
||||
'DQ WARNING: в fact_flight_sales строк без calendar_sk: % (<=1%%, допустимо)',
|
||||
v_null_calendar;
|
||||
END IF;
|
||||
END IF;
|
||||
|
||||
-- Обязательные поля.
|
||||
SELECT COUNT(*)
|
||||
INTO v_null_required
|
||||
FROM dds.fact_flight_sales
|
||||
WHERE book_ref IS NULL
|
||||
OR book_ref = ''
|
||||
OR ticket_no IS NULL
|
||||
OR ticket_no = ''
|
||||
OR flight_id IS NULL
|
||||
OR is_boarded IS NULL
|
||||
OR _load_id IS NULL
|
||||
OR _load_id = ''
|
||||
OR _load_ts IS NULL;
|
||||
|
||||
IF v_null_required <> 0 THEN
|
||||
RAISE EXCEPTION
|
||||
'DQ FAILED: в fact_flight_sales NULL обязательные поля: %',
|
||||
v_null_required;
|
||||
END IF;
|
||||
|
||||
RAISE NOTICE
|
||||
'DQ PASSED: dds.fact_flight_sales ок, строк=%',
|
||||
v_row_count;
|
||||
END $$;
|
||||
@@ -0,0 +1,111 @@
|
||||
-- Загрузка DDS fact_flight_sales: инкрементальный UPSERT по (ticket_no, flight_id).
|
||||
|
||||
-- Statement 1: UPDATE существующих строк факта.
|
||||
-- Обновляем только мутабельные поля; SK измерений не перезаписываем.
|
||||
UPDATE dds.fact_flight_sales AS f
|
||||
SET seat_no = bp.seat_no,
|
||||
price = seg.segment_amount,
|
||||
is_boarded = (bp.ticket_no IS NOT NULL),
|
||||
_load_id = '{{ run_id }}',
|
||||
_load_ts = now()
|
||||
FROM ods.segments AS seg
|
||||
LEFT JOIN ods.boarding_passes AS bp
|
||||
ON bp.ticket_no = seg.ticket_no
|
||||
AND bp.flight_id = seg.flight_id
|
||||
WHERE f.ticket_no = seg.ticket_no
|
||||
AND f.flight_id = seg.flight_id
|
||||
AND (
|
||||
f.is_boarded IS DISTINCT FROM (bp.ticket_no IS NOT NULL)
|
||||
OR f.price IS DISTINCT FROM seg.segment_amount
|
||||
OR f.seat_no IS DISTINCT FROM bp.seat_no
|
||||
);
|
||||
|
||||
-- Statement 2: INSERT новых строк факта.
|
||||
-- Dimension SK фиксируются на момент вставки (point-in-time для SCD2 routes).
|
||||
WITH fact_src AS (
|
||||
SELECT
|
||||
seg.ticket_no,
|
||||
seg.flight_id,
|
||||
cal.calendar_sk,
|
||||
dep.airport_sk AS departure_airport_sk,
|
||||
arr.airport_sk AS arrival_airport_sk,
|
||||
ap.airplane_sk,
|
||||
tar.tariff_sk,
|
||||
pax.passenger_sk,
|
||||
rte.route_sk,
|
||||
tkt.book_ref,
|
||||
bkg.book_date::DATE AS book_date,
|
||||
bp.seat_no,
|
||||
seg.segment_amount AS price,
|
||||
(bp.ticket_no IS NOT NULL) AS is_boarded
|
||||
FROM ods.segments AS seg
|
||||
JOIN ods.tickets AS tkt
|
||||
ON tkt.ticket_no = seg.ticket_no
|
||||
JOIN ods.bookings AS bkg
|
||||
ON bkg.book_ref = tkt.book_ref
|
||||
JOIN ods.flights AS flt
|
||||
ON flt.flight_id = seg.flight_id
|
||||
LEFT JOIN dds.dim_routes AS rte
|
||||
ON rte.route_bk = flt.route_no
|
||||
AND flt.scheduled_departure::DATE >= rte.valid_from
|
||||
AND (rte.valid_to IS NULL OR flt.scheduled_departure::DATE < rte.valid_to)
|
||||
LEFT JOIN dds.dim_calendar AS cal
|
||||
ON cal.date_actual = flt.scheduled_departure::DATE
|
||||
LEFT JOIN dds.dim_airports AS dep
|
||||
ON dep.airport_bk = rte.departure_airport
|
||||
LEFT JOIN dds.dim_airports AS arr
|
||||
ON arr.airport_bk = rte.arrival_airport
|
||||
LEFT JOIN dds.dim_airplanes AS ap
|
||||
ON ap.airplane_bk = rte.airplane_code
|
||||
LEFT JOIN dds.dim_tariffs AS tar
|
||||
ON tar.fare_conditions = seg.fare_conditions
|
||||
LEFT JOIN dds.dim_passengers AS pax
|
||||
ON pax.passenger_bk = tkt.passenger_id
|
||||
LEFT JOIN ods.boarding_passes AS bp
|
||||
ON bp.ticket_no = seg.ticket_no
|
||||
AND bp.flight_id = seg.flight_id
|
||||
)
|
||||
INSERT INTO dds.fact_flight_sales (
|
||||
calendar_sk,
|
||||
departure_airport_sk,
|
||||
arrival_airport_sk,
|
||||
airplane_sk,
|
||||
tariff_sk,
|
||||
passenger_sk,
|
||||
route_sk,
|
||||
book_ref,
|
||||
ticket_no,
|
||||
flight_id,
|
||||
book_date,
|
||||
seat_no,
|
||||
price,
|
||||
is_boarded,
|
||||
_load_id,
|
||||
_load_ts
|
||||
)
|
||||
SELECT
|
||||
s.calendar_sk,
|
||||
s.departure_airport_sk,
|
||||
s.arrival_airport_sk,
|
||||
s.airplane_sk,
|
||||
s.tariff_sk,
|
||||
s.passenger_sk,
|
||||
s.route_sk,
|
||||
s.book_ref,
|
||||
s.ticket_no,
|
||||
s.flight_id,
|
||||
s.book_date,
|
||||
s.seat_no,
|
||||
s.price,
|
||||
s.is_boarded,
|
||||
'{{ run_id }}',
|
||||
now()
|
||||
FROM fact_src AS s
|
||||
WHERE NOT EXISTS (
|
||||
SELECT 1
|
||||
FROM dds.fact_flight_sales AS f
|
||||
WHERE f.ticket_no = s.ticket_no
|
||||
AND f.flight_id = s.flight_id
|
||||
);
|
||||
|
||||
ANALYZE dds.fact_flight_sales;
|
||||
Reference in New Issue
Block a user