feat: add backfill job for historical price data

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

backend/tests/e2e/test_collection_job.py

@@ -91,8 +91,8 @@ def test_run_backfill_generates_yearly_chunks():
     assert price_ranges[0][0] == "20230101"  # First chunk starts at start_year
 
 
-def test_run_backfill_skips_already_collected_range():
-    """Backfill should start from earliest existing data backwards."""
+def test_run_backfill_with_existing_data_only_fills_gaps():
+    """Backfill should only collect before earliest and after latest existing data."""
     collected_ranges = []
 
     def make_price_collector(name):
@@ -112,9 +112,30 @@ def test_run_backfill_skips_already_collected_range():
         mock_db = MagicMock()
         mock_session_local.return_value = mock_db
 
-        # Simulate no existing data so both targets get backfilled
-        mock_db.query.return_value.scalar.return_value = None
-        # We'll verify the function runs without error
+        # Simulate: data exists from 2024-06-01 to 2024-12-31
+        call_count = [0]
+        def scalar_side_effect():
+            call_count[0] += 1
+            # func.min returns earliest date, func.max returns latest date
+            # Calls alternate: min for Price, (then max for Price forward fill),
+            # min for ETFPrice, (then max for ETFPrice forward fill)
+            if call_count[0] == 1:  # min(Price.date)
+                return date(2024, 6, 1)
+            elif call_count[0] == 2:  # max(Price.date) for forward fill
+                return date(2024, 12, 31)
+            elif call_count[0] == 3:  # min(ETFPrice.date)
+                return date(2024, 6, 1)
+            elif call_count[0] == 4:  # max(ETFPrice.date) for forward fill
+                return date(2024, 12, 31)
+            return None
+
+        mock_db.query.return_value.scalar.side_effect = scalar_side_effect
         run_backfill(start_year=2023)
 
-    assert len(collected_ranges) > 0
+    # Price backfill: should collect 2023-01-01 to 2024-05-31 (before earliest)
+    price_ranges = [(s, e) for name, s, e in collected_ranges if name == "price"]
+    assert len(price_ranges) >= 2  # At least backward chunks + forward fill
+    assert price_ranges[0][0] == "20230101"
+    # Last backward chunk should end at or before 2024-05-31
+    backward_chunks = [r for r in price_ranges if r[1] <= "20240531"]
+    assert len(backward_chunks) >= 1