Shubham Jakhete - Create way to see deactivated people's last week on the Weekly Summaries Reports page (backend)

[shubhamjakhete]

Description

Bug list: "Create way to see deactivated people's last week on the Weekly Summaries Reports page" (priority high, Jae).

When people are deactivated, they should still show up on the "Last Week" tab of the Weekly Summaries Report with a clear final week notice, so they get credit for their final work in the weekly blog reporting and admins know they will no longer be reporting on them.

PRs #1850 and OneCommunityGlobal/HighestGoodNetworkApp#4270 added this, but it doesn't work. While fixing it I found three problems in weeklySummaries:

• finalWeekIndex was measured from the requested week with Math.abs, so it changed with the tab being loaded. A user deactivated last week matched no tab.
• The index was set after an early return for users with no time entries, so anyone who logged 0 hours in their final week never got one.
• Active users with a scheduled end date got an index, which the frontend uses to hide them from every other tab.

Related PRS (if any):

This backend PR is related to frontend PR OneCommunityGlobal/HighestGoodNetworkApp#5577.
To test this backend PR you need to check out frontend PR OneCommunityGlobal/HighestGoodNetworkApp#5577.
Link: OneCommunityGlobal/HighestGoodNetworkApp#5577

Main changes explained:

• Create src/utilities/finalWeekIndex.js: returns which tab (0 to 3) an end date falls in, measured from the current week in Pacific time. Returns null for future, out-of-range, or invalid dates.
• Update src/helpers/reporthelper.js:
  • Use the new helper, set the index before the early return, and set it to null for active users.
  • Narrow the query from "all users" to active users plus inactive, non-paused users whose end date falls in the requested weeks. It excludes Paused rather than requiring Separated, because 46 of 54 inactive users on dev have no inactiveReason.
• Add unit tests for the helper (Pacific week boundaries, both DST changes, bad input) and for the report query. All 20 pass; the 4 report tests fail on the old code.

How to test:

1. Check out this branch and frontend PR OneCommunityGlobal/HighestGoodNetworkApp#FRONTEND.
2. Run npm install, then npm run dev.
3. Clear site data/cache and log in as an Admin or Owner.
4. Deactivate a test user with a final day in last week, and schedule a separation for another test user in a future week.
5. Go to Dashboard → Reports → Weekly Summaries Reports and click Refresh on each tab.
6. Verify:
   • The deactivated user appears only on Last Week, with the final week notice, even if they logged 0 hours.
   • The scheduled user appears on every tab, with no notice.
   • Paused users appear on no tab.
7. Run npx jest src/utilities/__tests__/finalWeekIndex.test.js src/helpers/__tests__/reporthelper.test.js.

Note:

• Tested as Owner (local seed data) and Admin (dev data). On dev, before the fix the Last Week tab showed 2,660 of 2,663 active users (3 active users with an end date were hidden). After the fix it shows 2,666 of 2,666 (3 accounts were added in between). No inactive users on dev currently have an end date in the last four weeks, so I tested the notice with a local MongoDB seeded with one user per case: left last week, left with 0 hours, left two weeks ago, old data with no inactiveReason, scheduled separation, paused, and normal active. I can share the seed script if it helps review.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[adit24dhaya]

Reviewed f19b661 with frontend #5577 at 7276aa42. The final-week index is now measured from the current Pacific week rather than the requested tab, is assigned before the no-time-entries return, and stays null for active users. The narrowed query includes active users and non-paused inactive users whose final date falls in the requested range, while retaining legacy inactive records without inactiveReason. I found no blocking issue.

Validation: 20/20 tests passed in the two changed suites on Node 20.19.4. Coverage includes Pacific Saturday/Sunday boundaries, both DST transitions, invalid/future/out-of-window dates, zero-hour users and scheduled active separations. Babel compiled 774 files; changed-file ESLint had zero errors/nine warnings, and git diff --check passed.

Twenty additional paired assertions exercised the real weeklySummaries helper and actual frontend visibility rules across four tabs using six fixtures. Active scheduled users remain visible throughout; zero-hour and legacy inactive users appear only in their final week; paused and old inactive users are excluded. Persistence is mocked, and the fixture interprets the first-match query predicates; this is not a real MongoDB integration test. I checked the requested-week controller calls and the weekly-email caller's reuse of this helper; the documented reduction of historical inactive users in that email is intentional scope, not a new mutation. No permission route or account-deactivation write is added here.

Paired frontend validation: 41/41 tests in seven files, build 9,156 modules, eight real-component browser cases in light/dark with 24 visibility and four notice assertions, plus five theme/contrast assertions. The actual dark notice measures 5.12:1 with the route stylesheet loaded. Frontend JS lint/diff checks passed; separate SCSS lint could not run because its parser is unavailable.

No backend server or real database was used, no accounts were changed and no external API calls were made. Full authenticated navigation, actual MongoDB aggregation and email rendering were not exercised.

Commit audit: both commits are focused, imperative, under 72 characters, without trailing periods, and include blank-separated what/why bodies. The history references the relevant #1850/#4270 work and explains the email effect. Approved.

[AaditTrivedi]

Reviewed by Aadit Trivedi.

Checked out the PR, installed with npm ci, and ran the two test files it adds: finalWeekIndex.test.js and reporthelper.test.js pass 20/20. I reviewed the backend on its own; the frontend PR #5577 needs this backend running, so I have not tested the report tabs end to end.

Verified

• getFinalWeekIndex computes the week in company time (America/Los_Angeles), rounds a calendar-day difference so a 167- or 169-hour DST week cannot shift the result, parses strings strictly as ISO 8601, and returns null for missing, invalid, future, or out-of-window dates.
• The report query now includes active users, plus inactive users whose endDate falls within the requested weeks, excluding those with inactiveReason 'Paused'. Active users with a scheduled separation get no final week index.
• The index is now set before the early return, so a user who logged no time in their final week is still placed correctly.

Pointers, not blocking

1. Inactive users with no endDate are now excluded from the report entirely. Normal deactivation always sets endDate, but a profile edit can clear it, so those users would disappear from the report. Probably fine, worth knowing.
2. The 'Paused' string is written out in the query; using the shared InactiveReason constant, as userProfileController.js does, would keep it consistent if the value ever changes.

Approving.

[vidiyala99]

Tested f19b661 together with frontend #5577 (7276aa4); details and screenshots are in my review there. Ran this branch on a test server against the shared dev data (read-only; nothing was written) and compared GET /api/reports/weeklysummaries?week=0..3 with development:

• development: every tab returns all 56 inactive users, and the user who left last week gets finalWeekIndex: 0 on the Last Week request.
• This PR: each tab returns active users plus only the people who left that week, with finalWeekIndex 1 (Last Week), 2 (Week Before Last) and 0 (This Week) as expected. Active users with a scheduled end date get null.

finalWeekIndex.test.js and reporthelper.test.js pass (20/20).

Note for the description: reporthelper.weeklySummaries also feeds the weekly summaries email job (userHelper.emailWeeklySummariesForAllUsers), which will now list only people who left that week instead of every inactive user. This should merge together with #5577, since the frontend relies on the new index.

Approving.

[one-community]

Thank you all, merging!