feat: backfill licensed users #8791
Conversation
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
1 Skipped Deployment
|
Dependency Review✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.OpenSSF Scorecard
Scanned Files |
| @@ -0,0 +1,85 @@ | |||
| exports.up = (db, cb) => { | |||
| db.runSql(` | |||
| WITH user_events AS ( | |||
There was a problem hiding this comment.
Just gets the correct events
| WHERE | ||
| type IN ('user-created', 'user-deleted') | ||
| ), | ||
| dates AS ( |
There was a problem hiding this comment.
Generates needed dates from first event to today.
| FROM | ||
| generated_dates | ||
| ), | ||
| active_emails AS ( |
There was a problem hiding this comment.
For each date, for each email, find the most recent created and deleted event before the date
| d.date, | ||
| ue.email | ||
| ), | ||
| result AS ( |
There was a problem hiding this comment.
Count the emails on date
- add 1 if the deleted date is empty, delete happened within last 30 days, delete happened before the creation
- add 0 else, so the email was not active during that day.
| d.date | ||
| ORDER BY | ||
| d.date | ||
| ) INSERT INTO licensed_users (date, count) |
There was a problem hiding this comment.
Insert all the date results into db, not overwriting existing results.
| expect(rows.find((row) => row.date === '2024-11-04').count).toBe(1); | ||
| expect(rows.find((row) => row.date === '2024-11-05').count).toBe(0); // 30 days has passed |
There was a problem hiding this comment.
This might be the same thing as I pointed out in a separate PR, so feel free to ignore if you don't think it's relevant, but October is a 31-day month. So if you create the user on the 1st, and delete them on the 2nd, the 31st will be the 30th day after deletion. Extrapolating, that means that november 3rd is the 30 day mark. Should we switch this to check November 4th? Or should it be 3rd? Again, because this will in reality differentiate on milliseconds, it probably doesn't matter much. I'm sure youv'e got it covered.
If someone is deleted on the 5th, does the 5th count as the first day of 30 or is the 6th the first day?
This migration introduces a query that calculates the licensed user counts and inserts them into the licensed_users table.
The logic ensures that:
Logic Breakdown:
Identify User Events (user_events): Extracts email addresses from user-related events (user-created and user-deleted) and tracks the type and timestamp of the event. This step ensures the ability to differentiate between user creation and deletion activities.
Generate a Date Range (dates): Creates a continuous range of dates spanning from the earliest recorded event up to the current date. This ensures we analyze every date, even those without events.
Determine Active Users (active_emails): Links dates with user events to calculate the status of each email address (active or deleted) on a given day. This step handles:
Calculate Daily Active User Counts (result):
For each date, counts the distinct email addresses that are active based on the conditions: