fixes

Signed-off-by: Dmitry Chigarev <[email protected]>

modin/core/dataframe/pandas/dataframe/dataframe.py

@@ -339,6 +339,8 @@ def set_dtypes_cache(self, dtypes):
         """
         dtypes = self._maybe_update_proxies(dtypes)
         if dtypes is None and self.has_materialized_columns:
+            # try to set a descriptor instead of 'None' to be more flexible in
+            # dtypes computing
             try:
                 self._dtypes = ModinDtypes(
                     DtypesDescriptor(

modin/core/dataframe/pandas/metadata/dtypes.py

@@ -53,7 +53,7 @@ class DtypesDescriptor:
         One can't pass `know_all_names=False` together with `remaining_dtype` as this creates ambiguity
         on how to interpret missing columns (whether they belong to `remaining_dtype` or not).
     _schema_is_known : bool, optional
-        Whether `known_dtypes` describe all the columns in the dataframe. This parameter intended mostly
+        Whether `known_dtypes` describe all columns in the dataframe. This parameter intended mostly
         for internal use.
     """
 
@@ -64,7 +64,7 @@ def __init__(
         remaining_dtype: Optional[np.dtype] = None,
         parent_df: Optional["PandasDataframe"] = None,
         columns_order: Optional[dict[int, IndexLabel]] = None,
-        know_all_names=True,
+        know_all_names: bool = True,
         _schema_is_known: Optional[bool] = None,
     ):
         if not know_all_names and remaining_dtype is not None:
@@ -89,7 +89,7 @@ def __init__(
         self._cols_with_unknown_dtypes: list[IndexLabel] = (
             [] if cols_with_unknown_dtypes is None else cols_with_unknown_dtypes
         )
-        # whether 'known_dtypes' describe all the columns in the dataframe
+        # whether 'known_dtypes' describe all columns in the dataframe
         if _schema_is_known is None:
             self._schema_is_known: bool = (
                 len(cols_with_unknown_dtypes) == 0
@@ -105,7 +105,7 @@ def __init__(
             )
         else:
             self._schema_is_known: bool = _schema_is_known
-        self._know_all_names = know_all_names
+        self._know_all_names: bool = know_all_names
         # a common dtype for columns that are not present in 'known_dtypes' nor in 'cols_with_unknown_dtypes'
         self._remaining_dtype: Optional[np.dtype] = remaining_dtype
         self._parent_df: Optional["PandasDataframe"] = parent_df
@@ -270,7 +270,9 @@ def copy(self) -> "DtypesDescriptor":
             self._cols_with_unknown_dtypes.copy(),
             self._remaining_dtype,
             self._parent_df,
-            columns_order=self.columns_order,
+            columns_order=None
+            if self.columns_order is None
+            else self.columns_order.copy(),
             know_all_names=self._know_all_names,
             _schema_is_known=self._schema_is_known,
         )

modin/core/storage_formats/base/query_compiler.py

@@ -4534,7 +4534,7 @@ def get_dtypes_set(self):
         -------
         set
         """
-        return set(self.dtypes.value)
+        return set(self.dtypes.values)
 
     # DateTime methods
     def between_time(self, **kwargs):  # noqa: PR01

modin/experimental/core/execution/native/implementations/hdk_on_native/dataframe/dataframe.py

@@ -511,6 +511,7 @@ def _maybe_update_proxies(self, dtypes, new_parent=None):
         table = self._partitions[0][0].get()
         if isinstance(table, pyarrow.Table):
             return super()._maybe_update_proxies(dtypes, new_parent=table)
+        return dtypes
 
     def groupby_agg(self, by, axis, agg, groupby_args, **kwargs):
         """

setup.cfg

@@ -12,7 +12,7 @@ tag_prefix =
 parentdir_prefix = modin-
 
 [tool:pytest]
-addopts =
+addopts = --cov-config=setup.cfg --cov=modin --cov-append --cov-report=
 xfail_strict=true
 markers =
     xfail_executions