Improve test data generation for 4-year span with realistic policies

- Extend data range to 4 years (2022-2025) with 450 expenses
- Add annual plan years starting Jan 1st each year:
  - 2022: HSA only
  - 2023: HSA+DCFSA
  - 2024-2025: FSA+DCFSA
  - 2026: FSA (future year)
- HSA expenses always remain Open (never reimbursed)
- Cap FSA/DCFSA reimbursements at election amounts
- Add HSA balance history with proper timestamps showing $4k/year contributions
- Reduce average expense to ~$89 for ~$10k/year total

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <[email protected]>

Author: JavaDogWebDesign

scripts/generate_test_data.py

@@ -236,9 +236,9 @@ def create_placeholder_pdf_bytes(expense_data=None):
 PLACEHOLDER_IMAGE = None  # Will be generated on first use
 
 # Configuration
-NUM_EXPENSES = 350
+NUM_EXPENSES = 450  # Increased for 4 years of data
 START_DATE = datetime(2022, 1, 1)
-END_DATE = datetime(2024, 12, 31)
+END_DATE = datetime(2025, 12, 31)  # 4 years of expenses
 
 # Sample data
 PROVIDERS = [
@@ -382,40 +382,78 @@ def generate_test_data():
             'user_id': 1
         })
 
-    # Create plan years with a mix of plan types:
-    # 2022: HSA only (no FSA/DCFSA)
-    # 2023: HSA+DCFSA combined (HSA with dependent care)
-    # 2024: FSA+DCFSA combined (no HSA, using FSA instead)
+    # Create plan years - ONE policy type per year cycle, starting January 1st:
+    # Policy types: HSA, HSA+DCFSA, FSA, or FSA+DCFSA
+    # User saves $4,000/year to HSA when HSA-eligible (HSA or HSA+DCFSA policy)
+    #
+    # 2022: HSA (simple HSA only)
+    # 2023: HSA+DCFSA (HSA with dependent care)
+    # 2024: FSA+DCFSA (switched to FSA)
+    # 2025: FSA+DCFSA (continuing FSA)
+    # 2026: FSA (future year policy - no DCFSA)
     plan_years = []
     plan_year_id = 1
 
-    # 2022: No FSA/DCFSA plans - HSA only year
-    # (HSA expenses don't need a plan_year_id)
+    # 2022: HSA only
+    plan_years.append({
+        'id': plan_year_id,
+        'user_id': 1,
+        'plan_type': 'HSA',
+        'plan_year_start': '2022-01-01',
+        'plan_year_end': '2022-12-31',
+        'election_amount': None,  # HSA doesn't use election_amount
+        'dcfsa_election_amount': None  # No DCFSA this year
+    })
+    plan_year_id += 1
 
-    # 2023: HSA+DCFSA combined plan
+    # 2023: HSA+DCFSA
     plan_years.append({
         'id': plan_year_id,
         'user_id': 1,
         'plan_type': 'HSA+DCFSA',
         'plan_year_start': '2023-01-01',
         'plan_year_end': '2023-12-31',
-        'election_amount': 3850.00,  # HSA family contribution limit 2023
+        'election_amount': None,  # HSA doesn't use election_amount
         'dcfsa_election_amount': 5000.00  # DCFSA election
     })
     plan_year_id += 1
 
-    # 2024: FSA+DCFSA combined plan (switched from HSA to FSA)
+    # 2024: FSA+DCFSA (switched from HSA to FSA)
     plan_years.append({
         'id': plan_year_id,
         'user_id': 1,
         'plan_type': 'FSA+DCFSA',
         'plan_year_start': '2024-01-01',
         'plan_year_end': '2024-12-31',
-        'election_amount': 3050.00,  # FSA limit 2024
+        'election_amount': 3200.00,  # FSA election
         'dcfsa_election_amount': 5000.00  # DCFSA election
     })
     plan_year_id += 1
 
+    # 2025: FSA+DCFSA (current year)
+    plan_years.append({
+        'id': plan_year_id,
+        'user_id': 1,
+        'plan_type': 'FSA+DCFSA',
+        'plan_year_start': '2025-01-01',
+        'plan_year_end': '2025-12-31',
+        'election_amount': 3300.00,  # FSA election
+        'dcfsa_election_amount': 5000.00  # DCFSA election
+    })
+    plan_year_id += 1
+
+    # 2026: FSA only (future year)
+    plan_years.append({
+        'id': plan_year_id,
+        'user_id': 1,
+        'plan_type': 'FSA',
+        'plan_year_start': '2026-01-01',
+        'plan_year_end': '2026-12-31',
+        'election_amount': 3400.00,  # FSA election
+        'dcfsa_election_amount': None  # No DCFSA this year
+    })
+    plan_year_id += 1
+
     # Build plan year lookup
     # HSA expenses don't need plan_year_id (always NULL)
     # FSA expenses map to FSA or FSA+DCFSA plans
@@ -431,11 +469,31 @@ def generate_test_data():
         elif plan_type == 'HSA+DCFSA':
             # HSA expenses don't use plan_year_id, only DCFSA does
             plan_year_lookup[('DCFSA', year)] = py['id']
+        elif plan_type == 'HSA':
+            # HSA-only plan - HSA expenses don't use plan_year_id
+            pass
         elif plan_type == 'FSA':
             plan_year_lookup[('FSA', year)] = py['id']
         elif plan_type == 'DCFSA':
             plan_year_lookup[('DCFSA', year)] = py['id']
 
+    # Build election amount lookup by (category, year)
+    # FSA/DCFSA reimbursements cannot exceed election amounts
+    election_lookup = {}
+    for py in plan_years:
+        year = int(py['plan_year_start'][:4])
+        plan_type = py['plan_type']
+
+        if plan_type in ['FSA', 'FSA+DCFSA']:
+            if py['election_amount']:
+                election_lookup[('FSA', year)] = py['election_amount']
+        if plan_type in ['DCFSA', 'FSA+DCFSA', 'HSA+DCFSA']:
+            if py['dcfsa_election_amount']:
+                election_lookup[('DCFSA', year)] = py['dcfsa_election_amount']
+
+    # Track cumulative reimbursements per (category, year)
+    reimbursement_totals = {}
+
     # Generate expenses
     expenses = []
     attachments = []
@@ -451,36 +509,64 @@ def generate_test_data():
         date_str = date_of_service.strftime('%Y-%m-%d')
         year = date_of_service.year
 
-        # Category depends on year and available plans:
-        # 2022: HSA only (no FSA/DCFSA plans available)
-        # 2023: HSA + DCFSA (HSA+DCFSA plan, no FSA)
-        # 2024: FSA + DCFSA (FSA+DCFSA plan, no HSA)
+        # Category depends on year's policy type:
+        # 2022: HSA only
+        # 2023: HSA+DCFSA (can use HSA or DCFSA)
+        # 2024-2025: FSA+DCFSA (can use FSA or DCFSA)
         if year == 2022:
-            # HSA only year
+            # HSA only policy
             category = 'HSA'
         elif year == 2023:
-            # HSA+DCFSA plan - can use HSA or DCFSA
+            # HSA+DCFSA policy - can use HSA or DCFSA
             category = random.choices(['HSA', 'DCFSA'], weights=[0.7, 0.3])[0]
-        else:  # 2024
-            # FSA+DCFSA plan - can use FSA or DCFSA
+        else:  # 2024, 2025
+            # FSA+DCFSA policy - can use FSA or DCFSA
             category = random.choices(['FSA', 'DCFSA'], weights=[0.7, 0.3])[0]
 
-        # Random status with weights
-        status = random.choices(STATUSES, weights=STATUS_WEIGHTS)[0]
-
-        # Random amount
-        if random.random() < 0.3:
-            amount = round(random.uniform(5, 50), 2)
-        elif random.random() < 0.7:
-            amount = round(random.uniform(50, 300), 2)
+        # Status depends on category - HSA expenses are never reimbursed
+        if category == 'HSA':
+            status = 'Open'
+        else:
+            status = random.choices(STATUSES, weights=STATUS_WEIGHTS)[0]
+
+        # Random amount - targeting ~$10k/year average across all expenses
+        # With ~112 expenses/year, avg expense should be ~$89
+        rand_val = random.random()
+        if rand_val < 0.50:
+            amount = round(random.uniform(10, 50), 2)      # 50%: small expenses
+        elif rand_val < 0.85:
+            amount = round(random.uniform(50, 150), 2)     # 35%: medium expenses
         else:
-            amount = round(random.uniform(300, 2000), 2)
+            amount = round(random.uniform(150, 350), 2)    # 15%: larger expenses
+
+        # Calculate reimbursement - HSA expenses are never reimbursed
+        # FSA/DCFSA reimbursements cannot exceed election amounts
+        if category == 'HSA':
+            amount_inflow = 0
+        elif category in ['FSA', 'DCFSA']:
+            # Check election limit
+            election_key = (category, year)
+            election_limit = election_lookup.get(election_key, 0)
+            current_total = reimbursement_totals.get(election_key, 0)
+            remaining = election_limit - current_total
+
+            if remaining <= 0:
+                # Election exhausted - no more reimbursements this year
+                status = 'Open'
+                amount_inflow = 0
+            elif status in ['Reimbursed', 'Closed']:
+                amount_inflow = min(amount, remaining)
+                if amount_inflow < amount:
+                    status = 'Partially Reimbursed'
+            elif status == 'Partially Reimbursed':
+                desired = round(amount * random.uniform(0.3, 0.8), 2)
+                amount_inflow = min(desired, remaining)
+            else:
+                amount_inflow = 0
 
-        # Calculate reimbursement
-        if status in ['Reimbursed', 'Closed']:
-            amount_inflow = amount
-        elif status == 'Partially Reimbursed':
-            amount_inflow = round(amount * random.uniform(0.3, 0.8), 2)
+            # Track the reimbursement
+            if amount_inflow > 0:
+                reimbursement_totals[election_key] = current_total + amount_inflow
         else:
             amount_inflow = 0
 
@@ -582,22 +668,26 @@ def generate_test_data():
         if expense_id % 50 == 0:
             print(f"  Generated {expense_id}/{NUM_EXPENSES} expenses...")
 
-    # HSA balance history - realistic progression over years
+    # HSA balance history - realistic progression over 4 years
+    # User saves $4,000/year to HSA when HSA-eligible (2022 HSA, 2023 HSA+DCFSA)
+    # 2024-2025 switched to FSA, so no new HSA contributions, but balance carries over
+    # HSA expenses remain open (not reimbursed) - balance is available for future use
     hsa_balance_history = []
     history_id = 1
     running_balance = 0
+    # (amount, change_type, notes, date)
+    # Add entries each year so the chart shows the balance over time
     transactions = [
-        (5000.00, 'initial_balance', 'Initial HSA rollover from previous employer'),
-        (3850.00, 'contribution', '2022 annual HSA contribution'),
-        (1500.00, 'contribution', '2022 employer contribution'),
-        (3850.00, 'contribution', '2023 annual HSA contribution'),
-        (1500.00, 'contribution', '2023 employer contribution'),
-        (1000.00, 'contribution', '2023 catch-up contribution'),
-        (-500.00, 'reimbursement_added', 'HSA reimbursement for dental work'),
-        (1200.00, 'contribution', '2024 partial year contribution'),
+        (2500.00, 'initial_balance', 'Initial HSA rollover from previous employer', '2022-01-01'),
+        (4000.00, 'contribution', '2022 annual HSA contribution', '2022-01-15'),
+        (4000.00, 'contribution', '2023 annual HSA contribution', '2023-01-15'),
+        # No HSA contributions 2024-2025 (switched to FSA plan)
+        # Add $0 entries to show balance carrying over unchanged
+        (0.00, 'contribution', '2024 - No HSA contribution (FSA plan year)', '2024-01-15'),
+        (0.00, 'contribution', '2025 - No HSA contribution (FSA plan year)', '2025-01-15'),
     ]
 
-    for amount, change_type, notes in transactions:
+    for amount, change_type, notes, date_str in transactions:
         previous = running_balance
         running_balance += amount
         hsa_balance_history.append({
@@ -609,11 +699,11 @@ def generate_test_data():
             'change_type': change_type,
             'related_expense_id': None,
             'notes': notes,
-            'recorded_at': datetime.now().isoformat()
+            'recorded_at': f'{date_str}T12:00:00'
         })
         history_id += 1
 
-    # Final balance should be ~$17,400
+    # Final balance should be $10,500
     final_hsa_balance = running_balance
 
     return {