Fix ParquetDataCatalog bar queries by instrument_id

RELEASES.md

@@ -9,7 +9,7 @@ None
 None
 
 ### Fixes
-None
+- Fixed `ParquetDataCatalog` bar queries by `instrument_id` which were no longer returning data (the intent is to use `bar_type`, however using `instrument_id` now returns all matching bars)
 
 ---
 

nautilus_trader/persistence/catalog/parquet.py

@@ -421,10 +421,23 @@ def backend_session(
             # Parse the parent directory which *should* be the instrument ID,
             # this prevents us matching all instrument ID substrings.
             dir = path.split("/")[-2]
-            if instrument_ids and not any(dir == urisafe_instrument_id(x) for x in instrument_ids):
-                continue
+
+            # Filter by instrument ID
+            if data_cls == Bar:
+                if instrument_ids and not any(
+                    dir.startswith(urisafe_instrument_id(x) + "-") for x in instrument_ids
+                ):
+                    continue
+            else:
+                if instrument_ids and not any(
+                    dir == urisafe_instrument_id(x) for x in instrument_ids
+                ):
+                    continue
+
+            # Filter by bar type
             if bar_types and not any(dir == urisafe_instrument_id(x) for x in bar_types):
                 continue
+
             table = f"{file_prefix}_{idx}"
             query = self._build_query(
                 table,

tests/unit_tests/persistence/test_catalog.py

@@ -241,7 +241,7 @@ def test_catalog_custom_data(catalog: ParquetDataCatalog) -> None:
     assert isinstance(data[0], CustomData)
 
 
-def test_catalog_bars(catalog: ParquetDataCatalog) -> None:
+def test_catalog_bars_querying_by_bar_type(catalog: ParquetDataCatalog) -> None:
     # Arrange
     bar_type = TestDataStubs.bartype_adabtc_binance_1min_last()
     instrument = TestInstrumentProvider.adabtc_binance()
@@ -261,6 +261,24 @@ def test_catalog_bars(catalog: ParquetDataCatalog) -> None:
     assert len(bars) == len(stub_bars) == 10
 
 
+def test_catalog_bars_querying_by_instrument_id(catalog: ParquetDataCatalog) -> None:
+    # Arrange
+    bar_type = TestDataStubs.bartype_adabtc_binance_1min_last()
+    instrument = TestInstrumentProvider.adabtc_binance()
+    stub_bars = TestDataStubs.binance_bars_from_csv(
+        "ADABTC-1m-2021-11-27.csv",
+        bar_type,
+        instrument,
+    )
+
+    # Act
+    catalog.write_data(stub_bars)
+
+    # Assert
+    bars = catalog.bars(instrument_ids=[instrument.id.value])
+    assert len(bars) == len(stub_bars) == 10
+
+
 def test_catalog_write_pyo3_order_book_depth10(catalog: ParquetDataCatalog) -> None:
     # Arrange
     instrument = TestInstrumentProvider.ethusdt_binance()
@@ -339,9 +357,11 @@ def test_catalog_multiple_bar_types(catalog: ParquetDataCatalog) -> None:
     # Assert
     bars1 = catalog.bars(bar_types=[str(bar_type1)])
     bars2 = catalog.bars(bar_types=[str(bar_type2)])
+    bars3 = catalog.bars(instrument_ids=[instrument1.id.value])
     all_bars = catalog.bars()
     assert len(bars1) == 10
     assert len(bars2) == 10
+    assert len(bars3) == 10
     assert len(all_bars) == 20