Select with limited join - sql

Question

I have two tables: products and products_prices.
products table:
id
name
user_id
1
Headphones
1
2
Phone
1
products_prices table:
id
product_id
price
time
1
1
10
1
2
1
15
2
3
1
20
3
4
2
10
4
5
2
15
5
6
2
20
6
I have a simple query:
SELECT * FROM products WHERE (user_id = 1) LIMIT 1 OFFSET 1
So I need to get limited rows from products table with only two prices values from table product_prices ordered by time for each row in products.
(I need to get product with two latest prices).
This is example of what I want to get:
id
user_id
name
curr_price
prev_price
2
1
Phone
20
15
And example of my query:
select products.*,
(SELECT price FROM products_prices WHERE product_id = products.id ORDER BY time asc LIMIT 1 OFFSET 0) as curr_price,
(SELECT price FROM products_prices WHERE product_id = products.id ORDER BY time asc LIMIT 1 OFFSET 1) as prev_price
from "products"
where (products."user_id" = 1)
limit 1 offset 1
Is it possible to do it without subqueries?

Answer 1

Not sure I find any of these easier to read...
0th approach using window functions and a CTE Demo
With products as (SELECT 1 ID, 'Headphones' name, 1 user_id UNION ALL
SELECT 2 ID, 'Phone' name, 1 user_id ),
products_Prices as (SELECT 1 ID, 1 Product_ID, 10 price, 1 time UNION ALL
SELECT 2 ID, 1 Product_ID, 15 price, 2 time UNION ALL
SELECT 3 ID, 1 Product_ID, 20 price, 3 time UNION ALL
SELECT 4 ID, 2 Product_ID, 33 price, 4 time UNION ALL
SELECT 5 ID, 2 Product_ID, 22 price, 5 time UNION ALL
SELECT 6 ID, 2 Product_ID, 11 price, 6 time),
STEP1 as (
SELECT P.ID, P.Name, P.user_ID,
price as CurrentPrice, lead(price) over (partition by P.ID order by time desc) Prev_Price, time,
row_number() over (Partition by P.ID order by time Desc) RN
FROM Products P
LEFT JOIN Products_Prices Z
on Z.Product_ID = P.ID)
SELECT Id, Name, User_ID, CurrentPRice, PRev_Price
From Step1 where RN = 1
Giving us:
+----+------------+---------+--------------+------------+
| id | name | user_id | currentprice | prev_price |
+----+------------+---------+--------------+------------+
| 1 | Headphones | 1 | 20 | 15 |
| 2 | Phone | 1 | 11 | 22 |
+----+------------+---------+--------------+------------+
1st approach using analytics and a CTE: note I changed price numbers to show variance.
DEMO
With products as (SELECT 1 ID, 'Headphones' name, 1 user_id UNION ALL
SELECT 2 ID, 'Phone' name, 1 user_id ),
products_Prices as (SELECT 1 ID, 1 Product_ID, 10 price, 1 time UNION ALL
SELECT 2 ID, 1 Product_ID, 15 price, 2 time UNION ALL
SELECT 3 ID, 1 Product_ID, 20 price, 3 time UNION ALL
SELECT 4 ID, 2 Product_ID, 33 price, 4 time UNION ALL
SELECT 5 ID, 2 Product_ID, 22 price, 5 time UNION ALL
SELECT 6 ID, 2 Product_ID, 11 price, 6 time),
STEP1 as (SELECT P.ID, P.Name, P.user_ID, PP.price, row_number() over (partition by PP.product_ID order by time desc) RN
FROM Products P
LEFT JOIN products_prices PP
on P.ID = PP.Product_ID)
SELECT ID, Name, User_ID, max(case when RN = 1 then Price end) as Current_price, max(case when RN=2 then price end) as Last_price
FROM STEP1
WHERE RN <=2
GROUP BY ID, name, User_ID
Giving us:
+----+------------+---------+---------------+------------+
| id | name | user_id | current_price | last_price |
+----+------------+---------+---------------+------------+
| 2 | Phone | 1 | 11 | 22 |
| 1 | Headphones | 1 | 20 | 15 |
+----+------------+---------+---------------+------------+
Option 2 using lateral.
demo
With products as (SELECT 1 ID, 'Headphones' name, 1 user_id UNION ALL
SELECT 2 ID, 'Phone' name, 1 user_id ),
products_Prices as (SELECT 1 ID, 1 Product_ID, 10 price, 1 time UNION ALL
SELECT 2 ID, 1 Product_ID, 15 price, 2 time UNION ALL
SELECT 3 ID, 1 Product_ID, 20 price, 3 time UNION ALL
SELECT 4 ID, 2 Product_ID, 33 price, 4 time UNION ALL
SELECT 5 ID, 2 Product_ID, 22 price, 5 time UNION ALL
SELECT 6 ID, 2 Product_ID, 11 price, 6 time)
SELECT P.ID, P.Name, P.user_ID, PP.price, time
FROM Products P
LEFT JOIN lateral (SELECT Product_ID, Price, time
FROM Products_Prices Z
WHERE Z.Product_ID = P.ID
ORDER BY Time Desc LIMIT 2) PP
on TRUE
ORDER BY TIME DESC;
Givng us : (unpivoted) and using the row number logic above we could pivot.
+----+------------+---------+-------+------+
| id | name | user_id | price | time |
+----+------------+---------+-------+------+
| 2 | Phone | 1 | 11 | 6 |
| 2 | Phone | 1 | 22 | 5 |
| 1 | Headphones | 1 | 20 | 3 |
| 1 | Headphones | 1 | 15 | 2 |
+----+------------+---------+-------+------+