Fix and Refine TradeTickDataWrangler.process_bar_data (#1586)

nautilus_trader/persistence/wranglers.pyx

@@ -200,8 +200,8 @@ def prepare_tick_data_from_bars(
 
     # Merge tick data
     df_ticks_final = pd.concat([df_ticks_o, df_ticks_h, df_ticks_l, df_ticks_c])
-    df_ticks_final.dropna(inplace=True)
-    df_ticks_final.sort_index(axis=0, kind="mergesort", inplace=True)
+    df_ticks_final = df_ticks_final.dropna()
+    df_ticks_final = df_ticks_final.sort_index(axis=0, kind="mergesort")
 
     cdef int i
     # Randomly shift high low prices
@@ -572,32 +572,24 @@ cdef class TradeTickDataWrangler:
         # Ensure index is tz-aware UTC
         data = as_utc_index(data)
 
-        # Determine the Aggressor Side based on Close vs Open
-        if "side" not in data and "buyer_maker" not in data:
-            data['side'] = ['BUY' if close > open_ else 'SELL' for open_, close in zip(data['open'], data['close'])]
-
         cdef dict data_open = {
             "price": data["open"],
             "size": data["volume"] / 4,
-            "side": data["side"],
         }
 
         cdef dict data_high = {
             "price": data["high"],
             "size": data["volume"] / 4,
-            "side": data["side"],
         }
 
         cdef dict data_low = {
             "price": data["low"],
             "size": data["volume"] / 4,
-            "side": data["side"],
         }
 
         cdef dict data_close = {
             "price": data["close"],
             "size": data["volume"] / 4,
-            "side": data["side"],
         }
 
         df_ticks_final, ts_events, ts_inits = prepare_tick_data_from_bars(
@@ -612,12 +604,19 @@ cdef class TradeTickDataWrangler:
         )
         df_ticks_final["trade_id"] = df_ticks_final.index.view(np.uint64).astype(str)
 
+        # Adjust size precision
+        size_precision = self.instrument.size_precision
+        if is_raw:
+            df_ticks_final["size"] = df_ticks_final["size"].apply(lambda x: round(x, size_precision - 9))
+        else:
+            df_ticks_final["size"] = df_ticks_final["size"].round(size_precision)
+
         if is_raw:
             return list(map(
                 self._build_tick_from_raw,
                 df_ticks_final["price"],
                 df_ticks_final["size"],
-                self._create_side_if_not_exist(data),
+                self._create_side_if_not_exist(df_ticks_final),
                 df_ticks_final["trade_id"],
                 ts_events,
                 ts_inits,
@@ -627,7 +626,7 @@ cdef class TradeTickDataWrangler:
                 self._build_tick,
                 df_ticks_final["price"],
                 df_ticks_final["size"],
-                self._create_side_if_not_exist(data),
+                self._create_side_if_not_exist(df_ticks_final),
                 df_ticks_final["trade_id"],
                 ts_events,
                 ts_inits,
@@ -636,8 +635,10 @@ cdef class TradeTickDataWrangler:
     def _create_side_if_not_exist(self, data):
         if "side" in data.columns:
             return data["side"].apply(lambda x: AggressorSide.BUYER if str(x).upper() == "BUY" else AggressorSide.SELLER)
-        else:
+        elif "buyer_maker" in data.columns:
             return data["buyer_maker"].apply(lambda x: AggressorSide.SELLER if x is True else AggressorSide.BUYER)
+        else:
+            return [AggressorSide.NO_AGGRESSOR] * len(data)
 
     # cpdef method for Python wrap() (called with map)
     cpdef TradeTick _build_tick_from_raw(

tests/unit_tests/persistence/test_wranglers.py

@@ -12,7 +12,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 # -------------------------------------------------------------------------------------------------
-
+import pandas as pd
 import pytest
 
 from nautilus_trader.model.enums import BookAction
@@ -111,6 +111,7 @@ def test_trade_bar_data_wrangler(
     provider = TestDataProvider()
     data = provider.read_csv_bars("fxcm/usdjpy-m1-bid-2013.csv")
     data.loc[:, "volume"] = 100_0000
+    expected_ticks_count = len(data) * 4
 
     # Act
     ticks = wrangler.process_bar_data(
@@ -124,3 +125,42 @@ def test_trade_bar_data_wrangler(
     assert ticks[1].ts_event == ts_event2
     assert ticks[2].ts_event == ts_event3
     assert ticks[3].ts_event == ts_event4
+    assert len(ticks) == expected_ticks_count
+
+
+@pytest.mark.parametrize("is_raw", [False, True])
+def test_trade_bar_data_wrangler_size_precision(is_raw: bool) -> None:
+    # Arrange
+    spy = TestInstrumentProvider.equity("SPY", "ARCA")
+    wrangler = TradeTickDataWrangler(instrument=spy)
+    factor = 1e9 if is_raw else 1
+    ts = pd.Timestamp("2024-01-05 21:00:00+0000", tz="UTC")
+    data = pd.DataFrame(
+        {
+            "open": {ts: 468.01 * factor},
+            "high": {ts: 468.08 * factor},
+            "low": {ts: 467.81 * factor},
+            "close": {ts: 467.96 * factor},
+            "volume": {ts: 18735.0 * factor},
+        },
+    )
+
+    # Calculate expected_size
+    if is_raw:
+        # For raw data, adjust precision by -9
+        expected_size = round(data["volume"].iloc[0] / 4, spy.size_precision - 9)
+    else:
+        # For non-raw data, apply standard precision and scale back up to compare with raw
+        expected_size = round(data["volume"].iloc[0] / 4, spy.size_precision) * 1e9
+
+    # Act
+    ticks = wrangler.process_bar_data(
+        data=data,
+        offset_interval_ms=0,
+        timestamp_is_close=True,
+        is_raw=is_raw,
+    )
+
+    # Assert
+    for tick in ticks:
+        assert tick.size.raw == expected_size