Window and CTE

Common Table Expressions (CTE)
PostgreSQL offers writing of Common Table Expressions (CTE) . Common Table Expressions are useful to define temporary SQL query used in a bigger query.
For this example, let's assume we want to count the new user creation per day during the month of September. One way would be to Group by EXTRACT('day' FROM created_at) , but days without new user will return not rows, where we want it to return zero.
In this case, using joins onto a generated series of day is the way to go. CTE makes it very simple to write and manage:
1
dates_in_september = Clear::SQL.select({
2
    day_start: "generate_series(date '2018-09-01', date '2018-09-30', '1 day'::interval)",
3
    day_end: "generate_series(date '2018-09-01', date '2018-09-30', '1 day'::interval) + '1 day'::interval";
4
})
5
​
6
Clear::SQL.select({
7
    count: "COUNT(users.*)",
8
    day: "dates.day_start"
9
})
10
  .with_cte(dates: dates_in_septembers)
11
  .from("dates")
12
  .left_joins(User.table){ (users.created_at >= day_start) & (users.created_at < day_end) }
13
  .group_by("dates.day_start")
14
  .order_by("dates.day_start")
15
  .fetch do |hash|
16
      puts "users created the #{hash["day"]}: #{hash["count"]}"
17
  end
Copied!
Since all model collections are SQL query, you can pass collection as parameter of with_cte block.
Window
You can pass window using window method:
1
Clear::SQL
2
    .select("sum(salary) OVER w", "avg(salary) OVER w")
3
    .from("empsalary")
4
    .window({w: "(PARTITION BY depname ORDER BY salary DESC)"})
Copied!

Eager Loading

Scopes

Last modified 2yr ago

Copy link

Contents

Common Table Expressions (CTE)

Window