Evaluations — Testing

Get your hash

node -e "const crypto = require('crypto'); console.log(crypto.createHash('sha256').update('yourlogin'.toLowerCase().trim()).digest('hex'))"

Register for evaluations

The extension auto-registers on install. To verify:

npx wrangler d1 execute better-intra-d1 --remote --command="SELECT * FROM eval_users WHERE hash = '<YOUR_HASH>'"

To manually register / unregister (simulates extension):

npx wrangler d1 execute better-intra-d1 --remote --command="INSERT OR IGNORE INTO eval_users (hash) VALUES ('<YOUR_HASH>')"
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM eval_users WHERE hash = '<YOUR_HASH>'"

Manual notification test

Create test-notif.sql:

-- booked: roles are hidden, just project + time
INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 1, 'evaluator',
'{"type":"booked","role":"evaluator","id":1,"projectName":"Libft","beginAt":"2026-06-14T15:00:00.000Z","endAt":"2026-06-14T15:15:00.000Z","teamName":null}');

-- revealed: names are visible (15min before)
INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 2, 'evaluator',
'{"type":"revealed","role":"evaluator","id":2,"projectName":"Libft","beginAt":"2026-06-14T15:00:00.000Z","endAt":"2026-06-14T15:15:00.000Z","logins":"elmo","teamName":null}');

Run:

npx wrangler d1 execute better-intra-d1 --remote --file=test-notif.sql

Wait up to 5 min for the extension to poll. You’ll get two Chrome notifications:

“Evaluation booked — Evaluation for Libft at 15:00”
“Evaluation — You’re correcting elmo for Libft at 15:00”

“evaluated” role (being corrected)

Same structure, different role and response fields:

-- booked: someone booked you, name hidden
INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 3, 'evaluated',
'{"type":"booked","role":"evaluated","id":3,"projectName":"Minishell","beginAt":"2026-06-14T16:00:00.000Z","endAt":"2026-06-14T16:15:00.000Z","teamName":null}');

-- revealed: evaluator's name appears
INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 4, 'evaluated',
'{"type":"revealed","role":"evaluated","id":4,"projectName":"Minishell","beginAt":"2026-06-14T16:00:00.000Z","endAt":"2026-06-14T16:15:00.000Z","login":"elmo","teamName":null}');

Expected Chrome notifications:

“Evaluation booked — Minishell will be evaluated at 16:00”
“Evaluation — elmo is about to evaluate Minishell at 16:00”

Notification field reference

Field	“booked”	“revealed”	Notes
`type`	`"booked"`	`"revealed"`	Required
`role`	`"evaluator"`	`"evaluator"`	`"evaluated"` for being corrected
`id`	scale_team id	scale_team id	Int, unique per eval per role per user
`projectName`	string or null	string or null	Displayed in notification
`beginAt`	ISO string	ISO string	Parsed for HH:MM in notification
`endAt`	ISO string	ISO string	Not displayed, used internally
`teamName`	string or null	string or null	Not displayed currently
`logins`	—	string	evaluator role: comma-separated corrected logins
`login`	—	string	evaluated role: single evaluator login

Notification preference toggles

These chrome.storage.local keys control which notifs show (all default true):

Key	Suppressed when `false`
`EVALUATIONS_NOTIFY_AS_EVALUATOR`	Skips all notifs with `role: "evaluator"`
`EVALUATIONS_NOTIFY_AS_EVALUATED`	Skips all notifs with `role: "evaluated"`
`EVALUATIONS_NOTIFY_REVEAL`	Skips all notifs with `type: "revealed"`

To test suppression: open the Better Intra popup, go to the Evaluations tab, toggle one off, then re-insert a matching notif. It should be silently consumed (marked consumed = 1) without a Chrome notification.

Duplicate prevention

pending_notifs has PRIMARY KEY (hash, eval_id, role). INSERT OR IGNORE blocks duplicates:

-- Insert the same row twice
INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 99, 'evaluator',
'{"type":"booked","role":"evaluator","id":99,"projectName":"DedupTest","beginAt":"2026-06-14T17:00:00.000Z","endAt":"2026-06-14T17:15:00.000Z","teamName":null}');

-- This is silently ignored (no error, no insert)
INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 99, 'evaluator',
'{"type":"booked","role":"evaluator","id":99,"projectName":"DedupTest","beginAt":"2026-06-14T17:00:00.000Z","endAt":"2026-06-14T17:15:00.000Z","teamName":null}');

-- Verify only 1 row stored
SELECT hash, eval_id, role FROM pending_notifs WHERE eval_id = 99;

The same constraint protects eval_states — no duplicate state transitions.

Full flow simulation (booked → revealed)

Simulate what the cron does for a real eval lifecycle:

-- Step 1: cron detects new booked eval, writes state + pending
INSERT INTO eval_states (hash, eval_id, role, state)
VALUES ('<YOUR_HASH>', 100, 'evaluator', 'booked');

INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 100, 'evaluator',
'{"type":"booked","role":"evaluator","id":100,"projectName":"ft_irc","beginAt":"2026-06-14T18:00:00.000Z","endAt":"2026-06-14T18:30:00.000Z","teamName":null}');

-- Step 2: cron detects names revealed (15min before), updates state, inserts new notif
UPDATE eval_states SET state = 'revealed', updated_at = unixepoch()
WHERE hash = '<YOUR_HASH>' AND eval_id = 100 AND role = 'evaluator';

INSERT INTO pending_notifs (hash, eval_id, role, data)
VALUES ('<YOUR_HASH>', 100, 'evaluator',
'{"type":"revealed","role":"evaluator","id":100,"projectName":"ft_irc","beginAt":"2026-06-14T18:00:00.000Z","endAt":"2026-06-14T18:30:00.000Z","logins":"elmo,kermit","teamName":"IRC"}');

-- Verify both states exist
SELECT hash, eval_id, role, state, updated_at FROM eval_states WHERE eval_id = 100;

Debug queries

# All registered eval users
npx wrangler d1 execute better-intra-d1 --remote --command="SELECT hash, datetime(created_at, 'unixepoch') FROM eval_users"

# All eval states ever recorded
npx wrangler d1 execute better-intra-d1 --remote --command="SELECT hash, eval_id, role, state, datetime(updated_at, 'unixepoch') FROM eval_states ORDER BY updated_at DESC"

# All pending notifs (not yet consumed by extension)
npx wrangler d1 execute better-intra-d1 --remote --command="SELECT hash, eval_id, role, consumed, datetime(created_at, 'unixepoch') FROM pending_notifs WHERE consumed = 0"

# All notifs including consumed
npx wrangler d1 execute better-intra-d1 --remote --command="SELECT hash, eval_id, role, consumed, datetime(created_at, 'unixepoch') FROM pending_notifs ORDER BY created_at DESC"

# Counts per table
npx wrangler d1 execute better-intra-d1 --remote --command="SELECT 'eval_users' AS tbl, COUNT(*) FROM eval_users UNION ALL SELECT 'eval_states', COUNT(*) FROM eval_states UNION ALL SELECT 'pending_notifs', COUNT(*) FROM pending_notifs"

Cleanup

# Delete all test data for a specific hash
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM pending_notifs WHERE hash = '<YOUR_HASH>'"
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM eval_states WHERE hash = '<YOUR_HASH>'"
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM eval_users WHERE hash = '<YOUR_HASH>'"

# Reset everything (destructive — wipes all eval data for all users)
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM pending_notifs"
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM eval_states"
npx wrangler d1 execute better-intra-d1 --remote --command="DELETE FROM eval_users"