FIX-modin-project#4652: Support categorical data in from_dataframe

Signed-off-by: Karthik Velayutham <vkarthik@ponder.io>

modin/core/dataframe/pandas/exchange/dataframe_protocol/column.py

@@ -43,6 +43,13 @@
 from .exception import NoValidityBuffer, NoOffsetsBuffer
 
 
+_NO_VALIDITY_BUFFER = {
+    ColumnNullType.NON_NULLABLE: "This column is non-nullable so does not have a mask",
+    ColumnNullType.USE_NAN: "This column uses NaN as null so does not have a separate mask",
+    ColumnNullType.USE_SENTINEL: "This column uses a sentinel value so does not have a mask",
+}
+
+
 @_inherit_docstrings(ProtocolColumn)
 class PandasProtocolColumn(ProtocolColumn):
     """
@@ -414,11 +421,9 @@ def _get_validity_buffer(self) -> Tuple[PandasProtocolBuffer, Any]:
             self._validity_buffer_cache = (buffer, dtype)
             return self._validity_buffer_cache
 
-        if null == ColumnNullType.NON_NULLABLE:
-            msg = "This column is non-nullable so does not have a mask"
-        elif null == ColumnNullType.USE_NAN:
-            msg = "This column uses NaN as null so does not have a separate mask"
-        else:
+        try:
+            msg = _NO_VALIDITY_BUFFER[null]
+        except KeyError:
             raise NotImplementedError("See self.describe_null")
 
         raise NoValidityBuffer(msg)

modin/test/exchange/dataframe_protocol/pandas/test_protocol.py

@@ -19,8 +19,7 @@
 from modin.test.test_utils import warns_that_defaulting_to_pandas
 
 
-def test_simple_import():
-    modin_df_producer = pd.DataFrame(test_data["int_data"])
+def eval_df_protocol(modin_df_producer):
     internal_modin_df_producer = modin_df_producer.__dataframe__()
     # Our configuration in pytest.ini requires that we explicitly catch all
     # instances of defaulting to pandas, this one raises a warning on `.from_dataframe`
@@ -43,3 +42,13 @@ def test_simple_import():
 
     df_equals(modin_df_producer, modin_df_consumer)
     df_equals(modin_df_producer, internal_modin_df_consumer)
+
+
+def test_simple_import():
+    modin_df = pd.DataFrame(test_data["int_data"])
+    eval_df_protocol(modin_df)
+
+
+def test_categorical_from_dataframe():
+    modin_df = pd.DataFrame({"foo": pd.Series([0, 1, 2, 3], dtype="category")})
+    eval_df_protocol(modin_df)