BUG: rolling.median/quantile does not work with (some) BaseIndexer subclasses

[mroeschke]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Using the example from computation.rst.

In [1]: In [70]: use_expanding = [True, False, True, False, True]
   ...:

In [2]: In [72]: df = pd.DataFrame({'values': range(5)})
   ...:

In [3]: In [2]: from pandas.api.indexers import BaseIndexer
   ...:

In [4]: ...: class CustomIndexer(BaseIndexer):
   ...: ...:
   ...: ...:    def get_window_bounds(self, num_values, min_periods, center, closed):
   ...: ...:        start = np.empty(num_values, dtype=np.int64)
   ...: ...:        end = np.empty(num_values, dtype=np.int64)
   ...: ...:        for i in range(num_values):
   ...: ...:            if self.use_expanding[i]:
   ...: ...:                start[i] = 0
   ...: ...:                end[i] = i + 1
   ...: ...:            else:
   ...: ...:                start[i] = i
   ...: ...:                end[i] = i + self.window_size
   ...: ...:        return start, end
   ...:

In [5]: indexer = CustomIndexer(window_size=1, use_expanding=use_expanding)

# Each window
In [7]: for window in df.rolling(indexer):
   ...:     print(window)
   ...:
   values
0       0
   values
1       1
   values
0       0
1       1
2       2
   values
3       3
   values
0       0
1       1
2       2
3       3
4       4

# Should be [0, 1, 1, 3, 2]
In [8]: df.rolling(indexer).median()
Out[8]:
   values
0     0.0
1     1.0
2     1.5
3     NaN
4     3.0

# Should be [0, 1, 1, 3, 2]
In [10]: df.rolling(indexer).quantile(0.5)
Out[10]:
   values
0     0.0
1     1.0
2     1.5
3     NaN
4     3.0

The cython aggregations probably assume some sort of property of the windows that may not hold with the custom indexers

[phofl]

Problem are the non monotonic starting points of the windows. Same problem with non monotonic ending points

use_expanding = [True, False, True, False, True]

df = pd.DataFrame({'values': range(5)})
from pandas.api.indexers import BaseIndexer

class CustomIndexer(BaseIndexer):
   
    def get_window_bounds(self, num_values, min_periods, center, closed):
       start = np.empty(num_values, dtype=np.int64)
       end = np.empty(num_values, dtype=np.int64)
       for i in range(num_values):
           if self.use_expanding[i]:
               start[i] = i
               end[i] = max(i -1, 1)
           else:
               start[i] = i
               end[i] = i + self.window_size
       return start, end

indexer = CustomIndexer(window_size=1, use_expanding=use_expanding)

# Each window
for window in df.rolling(indexer):
    print(window)


# print(df)
rolled = df.rolling(indexer)
print(rolled.median())