Index Monotonic Sort Bug in class DateChunker

[qcyfred]

Index Monotonic Sort Bug in class DateChunker (in file date_chunker.py)

If the df's index is not monotonic increasing, arctic will sort the df by index. BUT the variable dates is still not in order.

I suggest arctic to put the code dates = df.index.get_level_values('date') after the if sentence.

def to_chunks(self, df, chunk_size='D', func=None, **kwargs):
    """
    chunks the dataframe/series by dates

    Parameters
    ----------
    df: pandas dataframe or series
    chunk_size: str
        any valid Pandas frequency string
    func: function
        func will be applied to each `chunk` generated by the chunker.
        This function CANNOT modify the date column of the dataframe!

    Returns
    -------
    generator that produces tuples: (start date, end date,
              chunk_size, dataframe/series)
    """
    if 'date' in df.index.names:
        dates = df.index.get_level_values('date')
        if not df.index.is_monotonic_increasing:
            df = df.sort_index()
        # TODO dates won't be sorted, which will cause data store error.
        
      # dates = df.index.get_level_values('date')

Anyway, arctic is an excellent project !

这是我第一次在github上留言。蹩脚的英文。