FEAT-modin-project#2491: fixes

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

modin/backends/pandas/query_compiler.py

@@ -23,7 +23,7 @@
 )
 from pandas.core.base import DataError
 from typing import Type, Callable
-from collections.abc import Container
+from collections.abc import Iterable
 import warnings
 
 
@@ -2462,7 +2462,7 @@ def _groupby_dict_reduce(
         map_dict = {}
         reduce_dict = {}
         rename_columns = any(
-            not isinstance(fn, str) and isinstance(fn, Container)
+            not isinstance(fn, str) and isinstance(fn, Iterable)
             for fn in agg_func.values()
         )
         for col, col_funcs in agg_func.items():
@@ -2475,11 +2475,8 @@ def _groupby_dict_reduce(
 
             map_fns = []
             for i, fn in enumerate(col_funcs):
-                if not isinstance(fn, str) and isinstance(fn, Container):
-                    assert (
-                        len(fn) == 2
-                    ), f"Incorrect number of values to unpack. (got {len(fn)} expected 2)"
-                    future_col_name, func = fn[0], fn[1]
+                if not isinstance(fn, str) and isinstance(fn, Iterable):
+                    future_col_name, func = fn
                 elif isinstance(fn, str):
                     future_col_name, func = fn, fn
                 else:

modin/pandas/groupby.py

@@ -29,6 +29,7 @@
 from pandas.core.aggregation import reconstruct_func
 import pandas.core.common as com
 from types import BuiltinFunctionType
+from collections.abc import Iterable
 
 from modin.error_message import ErrorMessage
 from modin.utils import _inherit_docstrings, try_cast_to_pandas, wrap_udf_function
@@ -376,22 +377,15 @@ def aggregate(self, func=None, *args, **kwargs):
         relabeling_required = False
         if isinstance(func, dict) or func is None:
 
-            def _reconstruct_func(func, **kwargs):
-                relabeling_required, func, new_columns, order = reconstruct_func(
-                    func, **kwargs
-                )
-                # We convert to the string version of the function for simplicity.
-                func = {
-                    k: v
-                    if not callable(v) or v.__name__ not in dir(self)
-                    else v.__name__
-                    for k, v in func.items()
-                }
-                return relabeling_required, func, new_columns, order
-
-            relabeling_required, func_dict, new_columns, order = _reconstruct_func(
+            def try_get_str_func(o):
+                if not isinstance(o, str) and isinstance(o, Iterable):
+                    return [try_get_str_func(v) for v in o]
+                return o.__name__ if callable(o) and o.__name__ in dir(self) else o
+
+            relabeling_required, func_dict, new_columns, order = reconstruct_func(
                 func, **kwargs
             )
+            func_dict = {k: try_get_str_func(v) for k, v in func_dict.items()}
 
             if any(i not in self._df.columns for i in func_dict.keys()):
                 from pandas.core.base import SpecificationError

modin/pandas/test/test_groupby.py

@@ -641,6 +641,7 @@ def test_large_row_groupby(is_by_category):
         min,
         sum,
         {"A": "sum"},
+        {"A": lambda df: df.sum()},
         {"A": "max", "B": "sum", "C": "min"},
     ]
     for func in agg_functions:
@@ -1437,7 +1438,7 @@ def get_columns(df):
         ),
         lambda grp: grp.agg(
             {
-                df_from_grp(grp).columns[0]: (max, min, sum),
+                df_from_grp(grp).columns[0]: (("new_max", max), min, sum),
                 df_from_grp(grp).columns[-1]: (sum, min, max),
             }
         ),