thrilltrack-explorer/docs/PHASE_5_TESTING.md

# Phase 5: Testing & Validation Guide

## Completed Implementation

✅ **Phase 1-2**: All 26 edge functions + 29 frontend calls have request tracking
✅ **Phase 3**: All 7 admin forms use `submissionReducer` for state management
✅ **Phase 4**: Ready for moderation state machine integration

## Manual Testing Checklist

### Test Suite 1: Form Submission Flow (30 min)

#### Test Case: RideForm submission with state machine
1. Navigate to `/admin` → Create Ride
2. Fill out form completely
3. **DevTools Check:**
   - React DevTools → Find `RideForm` component
   - Watch `submissionState` prop
   - Verify transitions: `draft` → `validating` → `submitting` → `complete`
4. Click Submit
5. **Expected Behavior:**
   - Button becomes disabled immediately
   - Text changes to "Saving..."
   - Success toast appears
   - Form redirects or resets

#### Test Case: Validation error handling
1. Fill out form with missing required field (e.g., no name)
2. Click Submit
3. **Expected Behavior:**
   - State transitions: `draft` → `validating` → `draft` (with errors)
   - Validation error toast appears
   - Button re-enables for retry
   - Form retains entered data

#### Test Case: Network error handling
1. Fill out form completely
2. Open DevTools → Network tab → Throttle to "Offline"
3. Click Submit
4. **Expected Behavior:**
   - State attempts transition
   - Error caught and handled
   - State resets to `draft`
   - Error toast with retry option
   - Button re-enables

### Test Suite 2: Request Tracking (30 min)

#### Test Case: Edge function correlation
1. Submit RideForm
2. **Browser Check:**
   - Network tab → Find POST request to edge function
   - Response Headers → Verify `X-Request-ID` present
   - Response Body → Verify `requestId` field present
3. Copy `requestId` value
4. **Database Check:**
   ```sql
   SELECT * FROM request_metadata 
   WHERE request_id = 'PASTE_REQUEST_ID_HERE';
   ```
5. **Expected:** Single row with matching endpoint, user_id, duration

#### Test Case: Toast notification with requestId
1. Trigger photo upload
2. **Expected:** Success toast displays:
   ```
   Upload Successful
   Request ID: abc12345
   ```

### Test Suite 3: Database Validation (1 hour)

#### Query 1: Request Metadata Coverage
```sql
SELECT 
  endpoint,
  COUNT(*) as request_count,
  COUNT(DISTINCT user_id) as unique_users,
  AVG(duration_ms) as avg_duration_ms,
  MAX(duration_ms) as max_duration_ms,
  MIN(duration_ms) as min_duration_ms,
  COUNT(CASE WHEN error_message IS NOT NULL THEN 1 END) as error_count,
  ROUND(100.0 * COUNT(CASE WHEN error_message IS NOT NULL THEN 1 END) / COUNT(*), 2) as error_rate_percent
FROM request_metadata
WHERE created_at > NOW() - INTERVAL '1 hour'
GROUP BY endpoint
ORDER BY request_count DESC;
```
**Expected:** All critical endpoints present (`process-selective-approval`, `upload-image`, etc.)

#### Query 2: Trace ID Correlation
```sql
SELECT 
  trace_id,
  COUNT(*) as operation_count,
  MIN(created_at) as first_operation,
  MAX(created_at) as last_operation,
  EXTRACT(EPOCH FROM (MAX(created_at) - MIN(created_at))) as total_duration_seconds,
  STRING_AGG(DISTINCT endpoint, ', ' ORDER BY endpoint) as endpoints_hit
FROM request_metadata
WHERE trace_id IS NOT NULL
  AND created_at > NOW() - INTERVAL '1 day'
GROUP BY trace_id
HAVING COUNT(*) > 1
ORDER BY operation_count DESC
LIMIT 20;
```
**Expected:** Batch approvals show 5-50 operations with same `trace_id`

#### Query 3: Status Type Safety
```sql
SELECT 
  'content_submissions' as table_name,
  status,
  COUNT(*) as count,
  CASE 
    WHEN status IN ('draft', 'pending', 'locked', 'reviewing', 'partially_approved', 'approved', 'rejected', 'escalated') 
    THEN 'VALID' 
    ELSE 'INVALID' 
  END as validity
FROM content_submissions
GROUP BY status

UNION ALL

SELECT 
  'submission_items' as table_name,
  status,
  COUNT(*) as count,
  CASE 
    WHEN status IN ('pending', 'approved', 'rejected', 'flagged', 'skipped') 
    THEN 'VALID' 
    ELSE 'INVALID' 
  END as validity
FROM submission_items
GROUP BY status
ORDER BY table_name, count DESC;
```
**Expected:** All rows show `VALID` in validity column

#### Query 4: Orphaned Data Check
```sql
SELECT 
  cs.id,
  cs.created_at,
  cs.status,
  cs.submission_type,
  cs.submitted_by,
  COUNT(si.id) as item_count
FROM content_submissions cs
LEFT JOIN submission_items si ON si.submission_id = cs.id
WHERE cs.created_at > NOW() - INTERVAL '2 hours'
GROUP BY cs.id, cs.created_at, cs.status, cs.submission_type, cs.submitted_by
HAVING COUNT(si.id) = 0
ORDER BY cs.created_at DESC;
```
**Expected:** 0 rows (or only very recent submissions < 1 hour old)

#### Query 5: Lock Duration Analysis
```sql
SELECT 
  DATE_TRUNC('hour', locked_at) as hour,
  COUNT(*) as locks_acquired,
  AVG(EXTRACT(EPOCH FROM (locked_until - locked_at))) / 60 as avg_lock_duration_minutes,
  COUNT(CASE WHEN locked_until < NOW() THEN 1 END) as expired_locks,
  COUNT(CASE WHEN status = 'locked' AND locked_until < NOW() THEN 1 END) as stuck_locks
FROM content_submissions
WHERE locked_at > NOW() - INTERVAL '24 hours'
GROUP BY DATE_TRUNC('hour', locked_at)
ORDER BY hour DESC;
```
**Expected:** 
- Average lock duration ~15 minutes
- Few expired locks
- Zero stuck locks

### Test Suite 4: Performance Testing (1 hour)

#### Test 1: State Machine Overhead
1. Open Chrome DevTools → Performance tab
2. Click "Record" (⚫)
3. Fill out and submit RideForm
4. Stop recording
5. **Analysis:**
   - Find "Reducer" or "submissionReducer" in flame graph
   - Measure total time in reducer calls
   - **Target:** < 5ms total overhead per submission

#### Test 2: Request Metadata Insert Performance
```sql
EXPLAIN ANALYZE 
INSERT INTO request_metadata (
  request_id, user_id, endpoint, method, status_code, duration_ms
) VALUES (
  gen_random_uuid(), 
  'test-user-id', 
  '/functions/test', 
  'POST', 
  200, 
  150
);
```
**Target:** Execution time < 50ms

#### Test 3: Memory Leak Detection
1. Open Chrome DevTools → Memory tab
2. Take heap snapshot (Baseline)
3. Perform 20 form submissions (RideForm)
4. Force garbage collection (🗑️ icon)
5. Take second heap snapshot
6. Compare snapshots
7. **Expected:** 
   - No significant memory retention from state machines
   - No dangling event listeners
   - No uncleaned timeouts/intervals

## Success Criteria

### Functional Requirements
- ✅ All 26 edge functions return `requestId` and `X-Request-ID` header
- ✅ All 29 `supabase.functions.invoke` calls use `invokeWithTracking`
- ✅ All 7 admin forms use `submissionReducer` for submission flow
- ⏳ `SubmissionReviewManager` uses `moderationReducer` for review flow
- ⏳ `useModerationQueue` uses `moderationReducer` for claim/release operations
- ⏳ Lock expiry monitoring active with warning toasts
- ✅ Error toasts display `requestId` for debugging support

### Quality Requirements
- ✅ Zero TypeScript errors in strict mode
- ✅ No illegal state transitions possible (enforced by reducers)
- ✅ 100% request correlation coverage for critical paths
- ⏳ Database queries validate no orphaned data or invalid statuses
- ⏳ Performance overhead within acceptable limits

## Next Steps

1. **Phase 4**: Integrate moderation state machine into `SubmissionReviewManager` and `useModerationQueueManager`
2. **Complete Testing**: Run all manual test scenarios
3. **Database Validation**: Execute all validation queries
4. **Performance Benchmarks**: Verify all metrics meet targets
5. **Memory Leak Testing**: Ensure no memory retention issues