FIX: Fix client map_symbols for stype_out

Author: nmacholl

CHANGELOG.md

@@ -7,7 +7,8 @@
 - Added `map_symbols` option for `batch.submit_job(...)`, default `False`. When `True` this appends a requested symbol field to each record (available for CSV and JSON text encodings)
 
 #### Bug fixes
-- Fixed an issue where no disconnection exception were raised when iterating the `Live` client.
+- Fixed an issue where no disconnection exception were raised when iterating the `Live` client
+- Fixed an issue where calling `DBNStore.to_df`, `DBNStore.to_json`, or `DBNStore.to_csv` with `map_symbols=True` would cause a `TypeError`
 
 #### Breaking changes
 - Removed `default_value` parameter from `Historical.symbology.resolve`

databento/common/dbnstore.py

@@ -828,7 +828,7 @@ def to_csv(
         path: Path | str,
         pretty_px: bool = True,
         pretty_ts: bool = True,
-        map_symbols: bool = True,
+        map_symbols: bool | None = None,
         schema: Schema | str | None = None,
     ) -> None:
         """
@@ -883,7 +883,7 @@ def to_df(
         self,
         pretty_px: bool = ...,
         pretty_ts: bool = ...,
-        map_symbols: bool = ...,
+        map_symbols: bool | None = ...,
         schema: Schema | str | None = ...,
         count: None = ...,
     ) -> pd.DataFrame:
@@ -894,7 +894,7 @@ def to_df(
         self,
         pretty_px: bool = ...,
         pretty_ts: bool = ...,
-        map_symbols: bool = ...,
+        map_symbols: bool | None = ...,
         schema: Schema | str | None = ...,
         count: int = ...,
     ) -> DataFrameIterator:
@@ -904,7 +904,7 @@ def to_df(
         self,
         pretty_px: bool = True,
         pretty_ts: bool = True,
-        map_symbols: bool = True,
+        map_symbols: bool | None = None,
         schema: Schema | str | None = None,
         count: int | None = None,
     ) -> pd.DataFrame | DataFrameIterator:
@@ -951,8 +951,16 @@ def to_df(
                 raise ValueError("a schema must be specified for mixed DBN data")
             schema = self.schema
 
-        if not self._instrument_id_index:
-            self._instrument_id_index = self._build_instrument_id_index()
+        if map_symbols is None:
+            map_symbols = self.stype_out == SType.INSTRUMENT_ID
+
+        if map_symbols:
+            if self.stype_out != SType.INSTRUMENT_ID:
+                raise ValueError(
+                    "`map_symbols` is not supported when `stype_out` is not 'instrument_id'",
+                )
+            if not self._instrument_id_index:
+                self._instrument_id_index = self._build_instrument_id_index()
 
         if count is None:
             records = iter([self.to_ndarray(schema)])
@@ -1002,7 +1010,7 @@ def to_json(
         path: Path | str,
         pretty_px: bool = True,
         pretty_ts: bool = True,
-        map_symbols: bool = True,
+        map_symbols: bool | None = None,
         schema: Schema | str | None = None,
     ) -> None:
         """

tests/test_historical_bento.py

@@ -5,6 +5,7 @@
 from io import BytesIO
 from pathlib import Path
 from typing import Any, Callable
+from unittest.mock import MagicMock
 
 import databento
 import numpy as np
@@ -1140,3 +1141,50 @@ def test_dbnstore_to_df_with_count_empty(
 
     # Assert
     assert next(df_iter).empty
+
+
+def test_dbnstore_to_df_cannot_map_symbols(
+    test_data: Callable[[Schema], bytes],
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """
+    Test that calling to_df with map_symbols=True on a DBNStore with a
+    stype_out other than 'instrument_id' raises a ValueError.
+    """
+    # Arrange
+    dbn_stub_data = (
+        zstandard.ZstdDecompressor().stream_reader(test_data(Schema.TRADES)).read()
+    )
+
+    # Act
+    dbnstore = DBNStore.from_bytes(data=dbn_stub_data)
+    monkeypatch.setattr(DBNStore, "stype_out", MagicMock(return_type=SType.RAW_SYMBOL))
+
+    # Assert
+    with pytest.raises(ValueError):
+        dbnstore.to_df(
+            map_symbols=True,
+        )
+
+def test_dbnstore_to_df_cannot_map_symbols_default_to_false(
+    test_data: Callable[[Schema], bytes],
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """
+    Test that calling to_df with on a DBNStore with a stype_out other than
+    'instrument_id' won't raise a ValueError if `map_symbols` is not explicitly
+    set.
+    """
+    # Arrange
+    dbn_stub_data = (
+        zstandard.ZstdDecompressor().stream_reader(test_data(Schema.TRADES)).read()
+    )
+
+    # Act
+    dbnstore = DBNStore.from_bytes(data=dbn_stub_data)
+    monkeypatch.setattr(DBNStore, "stype_out", MagicMock(return_type=SType.RAW_SYMBOL))
+
+    df_iter = dbnstore.to_df()
+
+    # Assert
+    assert len(df_iter) == 4