pandas (original) (raw)

Consider this example, where there is a DataFrame with a non-unique DatetimeIndex:

In [8]: df = pd.DataFrame(np.random.randn(5, 3), columns=['a', 'b', 'c'], index=pd.date_range("2012", freq='H', periods=5))

In [9]: df = df.iloc[[0, 2, 2, 3]] 

In [10]: df
Out[10]: 
                            a         b         c
2012-01-01 00:00:00 -1.534726 -0.559295  0.207194
2012-01-01 02:00:00 -1.072027  0.376595  0.407512
2012-01-01 02:00:00 -1.072027  0.376595  0.407512
2012-01-01 03:00:00  0.581614  1.782635 -0.678197

If you then use drop to drop some columns, but forget to specify columns= or axis=1 (so you are actually dropping rows), you get a wrong error and very confusing error message:

In [10]: df.drop(['a', 'b'])
...

~/scipy/pandas/pandas/core/indexes/base.py in get_indexer_non_unique(self, target)
   4559             tgt_values = target._ndarray_values
   4560 
-> 4561         indexer, missing = self._engine.get_indexer_non_unique(tgt_values)
   4562         return ensure_platform_int(indexer), missing
   4563 

~/scipy/pandas/pandas/_libs/index.pyx in pandas._libs.index.IndexEngine.get_indexer_non_unique()

TypeError: 'NoneType' object is not iterable

Tested with pandas 0.25 and pandas master.