Incorrect result of nlargest function

[dchigarev]

System information

• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Windows 10
• Modin version (modin.__version__): 0.7.3+200.g8e7a5682
• Python version: 3.7.5
• Code we can use to reproduce:

if __name__ == "__main__":
    import modin.pandas as pd
    import pandas
    
     # test case from test_nlargest
    data = {
            "population": [
                59000000,
                65000000,
                434000,
                434000,
                434000,
                337000,
                11300,
                11300,
                11300,
            ],
            "GDP": [1937894, 2583560, 12011, 4520, 12128, 17036, 182, 38, 311],
            "alpha-2": ["IT", "FR", "MT", "MV", "BN", "IS", "NR", "TV", "AI"],
        }
    index = [
        "Italy",
        "France",
        "Malta",
        "Maldives",
        "Brunei",
        "Iceland",
        "Nauru",
        "Tuvalu",
        "Anguilla",
    ]
    modin_df = pd.DataFrame(data=data, index=index)
    pandas_df = pandas.DataFrame(data=data, index=index)

    md_res = modin_df.nlargest(3, "population")
    pd_res = pandas_df.nlargest(3, "population")

    print("pd_res:\n", pd_res, sep="")
    print("\nmd_res_partitions:\n", md_res._query_compiler._modin_frame._partitions[0][0].to_pandas(), sep="")
    print("\nmd_res:\n", md_res, sep="") # Exception: IndexError('positional indexers are out-of-bounds')

Output:

pd_res:
        population      GDP alpha-2
France    65000000  2583560      FR
Italy     59000000  1937894      IT
Malta       434000    12011      MT

md_res_partitions:
        population      GDP alpha-2
France    65000000  2583560      FR
Italy     59000000  1937894      IT
Malta       434000    12011      MT

md_res:
distributed.worker - WARNING -  Compute Failed
Function:  apply_list_of_funcs
args:      ([[b"\x80\x04\x95\xb0\x03\x00\x00\x00\x00\x00\x00\x8c\x17cloudpickle.cloudpickle\x94\x8c\x0e_fill_function\x94\x93\x94(h\x00\x8c\x0f_make_skel_func\x94\x93\x94h\x00\x8c\r_builtin_type\x94\x93\x94\x8c\x08CodeType\x94\x85\x94R\x94(K\x01K\x00K\x01K\x05K\x13C\x14t\x00\xa0\x01|\x00j\x02\x88\x01\x88\x00f\x02\x19\x00\xa1\x01S\x00\x94N\x85\x94\x8c\x06pandas\x94\x8c\tDataFrame\x94\x8c\x04iloc\x94\x87\x94\x8c\x02df\x94\x85\x94\x8cZc:\\users\\dchigare\\desktop\\repos\\modin\\modin\\engines\\dask\\pandas_on_dask\\frame\\partition.py\x94\x8c\x08<lambda>\x94KcC\x00\x94\x8c\x0bcol_indices\x94\x8c\x0brow_indices\x94\x86\x94)t\x94R\x94K\x02}\x94(\x8c\x0b__package__\x94\x8c'modin.engines.dask.pandas_on_dask.frame\x94\x8c\x08__name__\x94\x8c1modin.engines.dask.pandas_on_dask.frame.partition\x94\x8c\x08__file__\x94h\x12u\x87\x94R\x94}\x94(\x8c\x07globals\x94}\x94h\x0ch\x00\x8c\tsubimport\x94\x93\x94h\x0c\x85\x94R\x94s\x8c\x08defaults\x94N\x8c\x04dict\x94}\x94\x8c\x0eclosure_values\x94]\x94(\x8c\x15numpy.c
kwargs:    {}
Exception: IndexError('positional indexers are out-of-bounds')

Traceback (most recent call last):
  File "C:\Users\dchigare\Desktop\REPOS\TESTS\reprod.py", line 39, in <module>
    print("\nmd_res:\n", md_res, sep="") # Exception: IndexError('positional indexers are out-of-bounds')
  File "c:\users\dchigare\desktop\repos\modin\modin\pandas\base.py", line 3465, in __str__
    return repr(self)
  File "c:\users\dchigare\desktop\repos\modin\modin\pandas\dataframe.py", line 162, in __repr__
    result = repr(self._build_repr_df(num_rows, num_cols))
  File "c:\users\dchigare\desktop\repos\modin\modin\pandas\base.py", line 100, in _build_repr_df
    return self.iloc[indexer]._query_compiler.to_pandas()
  File "c:\users\dchigare\desktop\repos\modin\modin\backends\pandas\query_compiler.py", line 188, in to_pandas
    return self._modin_frame.to_pandas()
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\base\frame\data.py", line 1350, in to_pandas
    df = self._frame_mgr_cls.to_pandas(self._partitions)
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\base\frame\partition_manager.py", line 258, in to_pandas
    retrieved_objects = [[obj.to_pandas() for obj in part] for part in partitions]
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\base\frame\partition_manager.py", line 258, in <listcomp>
    retrieved_objects = [[obj.to_pandas() for obj in part] for part in partitions]
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\base\frame\partition_manager.py", line 258, in <listcomp>
    retrieved_objects = [[obj.to_pandas() for obj in part] for part in partitions]
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\dask\pandas_on_dask\frame\partition.py", line 127, in to_pandas
    dataframe = self.get()
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\dask\pandas_on_dask\frame\partition.py", line 63, in get
    return self.future.result()
  File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\distributed\client.py", line 218, in result
    raise exc.with_traceback(tb)
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\dask\pandas_on_dask\frame\partition.py", line 27, in apply_list_of_funcs
    df = func(df, **kwargs)
  File "c:\users\dchigare\desktop\repos\modin\modin\engines\dask\pandas_on_dask\frame\partition.py", line 99, in <lambda>
    lambda df: pandas.DataFrame(df.iloc[row_indices, col_indices])
  File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\pandas\core\indexing.py", line 1762, in __getitem__
    return self._getitem_tuple(key)
  File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\pandas\core\indexing.py", line 2067, in _getitem_tuple
    self._has_valid_tuple(tup)
  File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\pandas\core\indexing.py", line 703, in _has_valid_tuple
    self._validate_key(k, i)
  File "C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\lib\site-packages\pandas\core\indexing.py", line 2009, in _validate_key
    raise IndexError("positional indexers are out-of-bounds")
IndexError: positional indexers are out-of-bounds

Describe the problem

If we print columns and index of md_res in debug, we can see that it contains incorrect values. But as we see from the output of above code, partitions itself contains the correct result, that's actually the reason why that tests don't fail in CI (df_equals that used in tests converts modin result via to_pandas which consider information only from partitions)

(Pdb) md_res.columns
Index(['__reduced__'], dtype='object')
(Pdb) md_res.index
Index(['Italy', 'France', 'Malta', 'Maldives', 'Brunei', 'Iceland', 'Nauru',
       'Tuvalu', 'Anguilla'],
      dtype='object')