REFACTOR-modin-project#2009: avoid index access in is_scalar calls (m…

…odin-project#2010)

Signed-off-by: ienkovich <ilya.enkovich@intel.com>

modin/pandas/base.py

@@ -15,7 +15,6 @@
 import numpy as np
 from numpy import nan
 import pandas
-from pandas.api.types import is_scalar
 from pandas.compat import numpy as numpy_compat
 from pandas.core.common import count_not_none, pipe
 from pandas.core.dtypes.common import (
@@ -33,7 +32,7 @@
 import pickle as pkl
 
 from modin.error_message import ErrorMessage
-from modin.pandas.utils import try_cast_to_pandas
+from modin.pandas.utils import try_cast_to_pandas, is_scalar
 
 # Similar to pandas, sentinel value to use as kwarg in place of None when None has
 # special meaning and needs to be distinguished from a user explicitly passing None.

modin/pandas/indexing.py

@@ -13,12 +13,13 @@
 
 import numpy as np
 import pandas
-from pandas.api.types import is_scalar, is_list_like, is_bool
+from pandas.api.types import is_list_like, is_bool
 from pandas.core.dtypes.common import is_integer
 from pandas.core.indexing import IndexingError
 
 from .dataframe import DataFrame
 from .series import Series
+from .utils import is_scalar
 
 """Indexing Helper Class works as follows:
 

modin/pandas/series.py

@@ -20,15 +20,14 @@
 from pandas.core.dtypes.common import (
     is_dict_like,
     is_list_like,
-    is_scalar,
 )
 import sys
 import warnings
 
 from .base import BasePandasDataset
 from .iterator import PartitionIterator
 from .utils import _inherit_docstrings
-from .utils import from_pandas, to_pandas
+from .utils import from_pandas, to_pandas, is_scalar
 
 if sys.version_info[0] == 3 and sys.version_info[1] >= 7:
     # Python >= 3.7

modin/pandas/utils.py

@@ -155,3 +155,30 @@ def hashable(obj):
     except TypeError:
         return False
     return True
+
+
+def is_scalar(obj):
+    """
+    Return True if given object is scalar.
+
+    This method wrks the same as is_scalar method from Pandas but
+    it is optimized for Modin frames. For BasePandasDataset objects
+    Pandas version of is_scalar tries to access missing attribute
+    causing index scan. This tiggers execution for lazy frames and
+    we avoid it by handling BasePandasDataset objects separately.
+
+    Parameters
+    ----------
+    val : object
+        Object to check.
+
+    Returns
+    -------
+    bool
+        True if given object is scalar and False otherwise.
+    """
+
+    from pandas.api.types import is_scalar as pandas_is_scalar
+    from .base import BasePandasDataset
+
+    return not isinstance(obj, BasePandasDataset) and pandas_is_scalar(obj)