Multiple Full Joins On a User Table in a Single Query - sql

Question

I have a schema of a database where we have a a user table and several fact tables.
The schema is goes something like this:
╔════════════╗
║ USERS ║
╠════════════╣
║ sales_id ║ PK
║ traveler_id║ PK
╚════════════╝
╔════════════╗
║ Sales ║
╠════════════╣
║ sales_id ║ --> FK to User sales_id PK
║ date ║
║ metric_1 ║
║ metric_2 ║
╚════════════╝
╔════════════╗
║ Travels ║
╠════════════╣
║ traveler_id║ --> FK to User traveler_id PK
║ date ║
║ metric_3 ║
║ metric_4 ║
╚════════════╝
I'm trying to join them as a single fact table to get all the metrics and their dates in a inqle query. It would return something like below:
╔════════════╦══════════════╦══════════╦══════════╦══════════╦══════════╦══════════╗
║ sales_id ║ traveler_id ║ date ║ metric_1 ║ metric_2 ║ metric_3 ║ metric_4 ║
╠════════════╬══════════════╬══════════╬══════════╬══════════╬══════════╬══════════╣
║ 1 ║ A ║ Jan ║ x1 ║ a1 ║ null ║ null ║
║ 1 ║ A ║ Feb ║ x2 ║ a2 ║ null ║ null ║
║ 1 ║ A ║ Jan ║ x3 ║ a3 ║ b1 ║ null ║
║ 2 ║ B ║ Mar ║ x4 ║ a4 ║ null ║ c1 ║
╚════════════╩══════════════╩══════════╩══════════╩══════════╩══════════╩══════════╝
Here is the sql I wrote:
select
coalesce(s.date, t.date) as date,
s.sales_id,
t.traveler_id,
max(coalesce(s.metric_1, 0)) as metric_1,
max(coalesce(s.metric_2, 0)) as metric_2,
max(coalesce(t.metric_3, 0)) as metric_3,
max(coalesce(t.metric_4, 0)) as metric_4,
from users as u
full join sales as s on (u.sales_id = s.sales_id)
full join travels as t on (u.traveler_id = t.traveler_id)
group by 1, 2, 3
However, my joining logic ends up returning the same values for the rows that are as null.
╔════════════╦══════════════╦══════════╦══════════╦══════════╦══════════╦══════════╗
║ sales_id ║ traveler_id ║ date ║ metric_1 ║ metric_2 ║ metric_3 ║ metric_4 ║
╠════════════╬══════════════╬══════════╬══════════╬══════════╬══════════╬══════════╣
║ 1 ║ A ║ Jan ║ x1 ║ a1 ║ b1 ║ null ║
║ 1 ║ A ║ Feb ║ x2 ║ a2 ║ b1 ║ null ║
║ 1 ║ A ║ Jan ║ x3 ║ a3 ║ b1 ║ null ║
║ 2 ║ B ║ Mar ║ x4 ║ a4 ║ null ║ null ║
╚════════════╩══════════════╩══════════╩══════════╩══════════╩══════════╩══════════╝
My suspicion is that it's due to the second full join. I would assume I could break these down into two other joins, and match their id, but I'm having a hard time trying to understand how to incorporate the dates, properly.

Answer 1

select
coalesce(s.date, t.date) as date,
s.sales_id,
t.traveler_id,
max(coalesce(s.metric_1, 0)) as metric_1,
max(coalesce(s.metric_2, 0)) as metric_2,
max(coalesce(t.metric_3, 0)) as metric_3,
max(coalesce(t.metric_4, 0)) as metric_4
from users as u
left join sales as s on (u.sales_id = s.sales_id and s.date = coalesce(s.date, t.date))
left join travels as t on (u.traveler_id = t.traveler_id and t.date = coalesce(s.date, t.date))
group by 1, 2, 3
This is what the table looks like:
╔════════════╦══════════════╦══════════╦══════════╦══════════╦══════════╦══════════╗
║ sales_id ║ traveler_id ║ date ║ metric_1 ║ metric_2 ║ metric_3 ║ metric_4 ║
╠════════════╬══════════════╬══════════╬══════════╬══════════╬══════════╬══════════╣
║ 1 ║ A ║ Jan ║ x1 ║ a1 ║ b1 ║ null ║
║ 1 ║ A ║ Feb ║ x2 ║ a2 ║ null ║ null ║
║ 1 ║ A ║ Jan ║ x3 ║ a3 ║ b1 ║ null ║
║ 2 ║ B ║ Mar ║ x4 ║ a4 ║ null ║ c1 ║
╚════════════╩══════════════╩══════════╩══════════╩══════════╩══════════╩══════════╝