Этап 3. Слой Core — обогащённые данные. Пет проект: DWH. РАБОТА НАД ОШИБКАМИ - изменил логику загрузи и обновления измерений.

This commit is contained in:
mnoskov
2026-04-24 18:10:49 +03:00
parent 5c2037a49a
commit 95456999b3
@@ -23,6 +23,7 @@ CREATE TABLE core.dim_lesson (
duration_min smallint, duration_min smallint,
course_id bigint NOT NULL REFERENCES core.dim_course(course_id) course_id bigint NOT NULL REFERENCES core.dim_course(course_id)
) DISTRIBUTED BY (lesson_id); ) DISTRIBUTED BY (lesson_id);
CREATE INDEX idx_dim_lesson_course_id ON core.dim_lesson(course_id);
--DROP TABLE IF EXISTS core.dim_date; --DROP TABLE IF EXISTS core.dim_date;
CREATE TABLE core.dim_date ( CREATE TABLE core.dim_date (
@@ -52,34 +53,92 @@ CREATE TABLE core.enrollments (
--6 --6
-- dim_user
INSERT INTO core.dim_user (user_id, name, age, email, registration_date) INSERT INTO core.dim_user (user_id, name, age, email, registration_date)
SELECT DISTINCT SELECT
id::int AS user_id, raw_u.id,
name::text, raw_u.name,
age::int, raw_u.age,
email::text, raw_u.email,
registration_date::date raw_u.registration_date
FROM raw.users
WHERE id IS NOT NULL AND email IS NOT NULL;
INSERT INTO core.dim_course (course_id, title, category)
SELECT DISTINCT
id::int AS course_id,
title::text,
category::text
FROM FROM
raw.courses raw.users raw_u
WHERE id IS NOT NULL AND title IS NOT NULL; WHERE NOT EXISTS
(SELECT 1 FROM core.dim_user core_u WHERE core_u.user_id = raw_u.id)
AND raw_u.email IS NOT NULL AND raw_u.registration_date IS NOT NULL;
UPDATE core.dim_user core_u
SET
name = raw_u.name,
age = raw_u.age,
email = raw_u.email,
registration_date = raw_u.registration_date
FROM
raw.users raw_u
WHERE
core_u.user_id = raw_u.id
AND (
core_u.name != raw_u.name OR
core_u.age != raw_u.age OR
core_u.email != raw_u.email OR
core_u.registration_date != raw_u.registration_date
);
-- dim_course
INSERT INTO core.dim_course (course_id, title, category, created_at)
SELECT
raw_c.id,
raw_c.title,
raw_c.category,
raw_c.created_at
FROM
raw.courses raw_c
WHERE NOT EXISTS
(SELECT 1 FROM core.dim_course core_c WHERE core_c.course_id = raw_c.id)
AND raw_c.title IS NOT NULL AND raw_c;
UPDATE core.dim_course core_c
SET
title = raw_c.title,
category = raw_c.category,
created_at = raw_c.created_at
FROM
raw.courses raw_c
WHERE
core_c.course_id = raw_c.id
AND (
core_c.title != raw_c.title OR
core_c.category != raw_c.category OR
core_c.created_at != raw_c.created_at
);
-- dim_lesson
INSERT INTO core.dim_lesson (lesson_id, title, duration_min, course_id) INSERT INTO core.dim_lesson (lesson_id, title, duration_min, course_id)
SELECT DISTINCT SELECT
id::int AS lesson_id, raw_l.id,
title::text, raw_l.title,
duration_min::int, raw_l.duration_min,
course_id::int raw_l.course_id
FROM FROM
raw.lessons raw.lessons raw_l
WHERE id IS NOT NULL; WHERE NOT EXISTS
(SELECT 1 FROM core.dim_lesson core_l WHERE core_l.lesson_id = raw_l.id)
AND raw_l.title IS NOT NULL AND raw_l.course_id IS NOT NULL;
UPDATE core.dim_lesson core_l
SET
title = raw_l.title,
duration_min = raw_l.duration_min,
course_id = raw_l.course_id -- решил оставить, тк чисто теоретические же урок может начать принадлежать другому курсу
FROM
raw.lessons raw_l
WHERE
core_l.lesson_id = raw_l.id
AND (
core_l.title != raw_l.title OR
core_l.duration_min != raw_l.duration_min OR
core_l.course_id != raw_l.course_id
);
-- загуглил как заполнять: -- загуглил как заполнять:
INSERT INTO core.dim_date (date, year, month, day, weekday) INSERT INTO core.dim_date (date, year, month, day, weekday)
@@ -90,7 +149,8 @@ SELECT
EXTRACT(DAY FROM d)::int, EXTRACT(DAY FROM d)::int,
to_char(d, 'day') to_char(d, 'day')
FROM FROM
generate_series('2023-01-01'::date, '2025-12-31'::date, INTERVAL '1 day') AS d; generate_series('2023-01-01'::date, '2030-12-31'::date, INTERVAL '1 day') AS d;
-- получается dim_date загружается один раз, и должно храниться где-то отдельно?
INSERT INTO core.fact_lesson_views (user_id, lesson_id, course_id, viewed_at) INSERT INTO core.fact_lesson_views (user_id, lesson_id, course_id, viewed_at)
SELECT SELECT