fix(ia): restore bulk-import Class-column links for legacy imports (#1624)

[JasonWildMe]

Issue

Fixes #1624. Since the 10.2 update, the bulk-import task page (/react/bulk-import-task) shows an empty Class column — no Match Results links — for imports that existed before 10.2. Reported on Whiskerbook and GiraffeSpotter; re-sending the import to ID works around it.

Root cause

The Class column is fed by ImportTask.statsAnnotations(), which selects the task to link via Task.getPreferredMatchResultsTaskForAnnotation(ann, importTaskId, …), passing the import's id. The C15 selector restricts candidates to tasks whose parameters.importTaskId equals that id and returned null when none matched.

importTaskId is stamped onto tasks only by the modern bulk-import path (BulkImport.initiateIA()). Pre-10.2 imports' tasks never carry it, so the filter emptied the candidate list, the selector returned null, encounterTaskInfo got no entry, and the frontend rendered -. (Re-sending to ID creates freshly-stamped tasks, which is why that works around it.)

Fix

A single fallback in Task.getPreferredMatchResultsTaskForAnnotation: when no task carries this import's id, fall back to the annotation's un-stamped tasks (no importTaskId) instead of returning null. Candidates are bucketed three ways:

Bucket	Meaning	Used?
scoped	importTaskId == this import	first choice (preserves the 10.2 freeze)
legacy	no importTaskId at all	fallback when scoped empty
other import	a different import's id	silently dropped

Dropping the third bucket prevents a legacy import's page from linking to an unrelated, newer import's task.

Why it's safe — strict superset, no regression

• scoped non-empty → tasks = scoped, byte-identical to the old tasks = filtered and every downstream decision. Any imperfect selection on this path is pre-existing C15 behavior, unchanged here.
• scoped empty, legacy non-empty → returns a legacy task where the old code returned null. The only behavioral delta — the fix.
• scoped empty, legacy empty → null, same as before.

The method never returns a different non-null task than before; it only converts specific null results into the legacy task. The encounter page is unaffected (it calls the 2-arg overload with importTaskId == null and never enters this branch).

Also normalizes a blank importTaskId to unscoped (defensive).

Review (Codex, 2 rounds)

Reviewed adversarially with a focus on unexpected fallback behavior. Findings addressed: blank-id edge fixed; doc-ordering wording corrected; the freeze-on-import limitation for the WBIA path (match tasks are created un-stamped in IBEISIA.processCallback) is pre-existing, not introduced here and documented.

Recommended follow-up (separate PR): propagate importTaskId into the taskParameters built in IBEISIA.processCallback so future WBIA match tasks are stamped and the freeze holds for them too. It does not repair already-persisted tasks, so this fallback is still required for existing data.

Full design notes and the no-regression analysis: docs/superpowers/specs/2026-06-17-issue-1624-bulk-import-class-column-fix.md.

Testing

No schema change, no data migration; repairs existing data on all installs at once.

• Pre-10.2 import: Class-column links render again.
• Modern (stamped) import: link still frozen on that import's task after an encounter-page re-ID.
• Encounter page Match Results link: unchanged.

🤖 Generated with Claude Code

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 51.61%. Comparing base (f6c25b8) to head (06eb730).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1625   +/-   ##
=======================================
  Coverage   51.61%   51.61%           
=======================================
  Files         308      308           
  Lines       12140    12140           
  Branches     3816     3816           
=======================================
  Hits         6266     6266           
  Misses       5591     5591           
  Partials      283      283           

Flag	Coverage Δ
backend	51.61% <ø> (ø)
frontend	51.61% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[naknomum]

lgtm