Support Resampling Weeks to/from FY5253Quarter

[cancan101]

52–53-week fiscal year are designed to end on the same day of the week each quarter / year. Therefore it should be possible to upsample / downsample to/from weeks. I made an initial attempt at this in #5148 but that PR was never merged. This logic still makes sense.

/CC @fdion

Currently attempts to do so yield:

import pandas as pd
from pandas.tseries.offsets import FY5253Quarter, Week, WeekDay
from pandas.tseries.period import Period, PeriodIndex, period_range
from datetime import datetime

offset_fyq = FY5253Quarter(weekday=WeekDay.SAT, startingMonth=12,
                       variation="last", qtr_with_extra_week=4)
freq_week = Week(weekday=WeekDay.SAT)

prange = period_range(datetime(2013, 1, 5),
                               periods=2 * 13,
                               freq=freq_week)

df = pd.DataFrame({"A": [1] * 13 + [2] * 13}, index=prange)
resampled = df.resample(offset_fyq, fill_method="mean")

ValueError                                Traceback (most recent call last)
<ipython-input-8-38fa44432b0b> in <module>()
     13 
     14 df = pd.DataFrame({"A": [1] * 13 + [2] * 13}, index=prange)
---> 15 resampled = df.resample(offset_fyq, fill_method="mean")

/Users/alex/.virtualenvs/work/lib/python2.7/site-packages/pandas/core/generic.pyc in resample(self, rule, how, axis, fill_method, closed, label, convention, kind, loffset, limit, base)
   3641                               fill_method=fill_method, convention=convention,
   3642                               limit=limit, base=base)
-> 3643         return sampler.resample(self).__finalize__(self)
   3644 
   3645     def first(self, offset):

/Users/alex/.virtualenvs/work/lib/python2.7/site-packages/pandas/tseries/resample.pyc in resample(self, obj)
     90 
     91             if self.kind is None or self.kind == 'period':
---> 92                 rs = self._resample_periods()
     93             else:
     94                 obj = self.obj.to_timestamp(how=self.convention)

/Users/alex/.virtualenvs/work/lib/python2.7/site-packages/pandas/tseries/resample.pyc in _resample_periods(self)
    340             return obj.reindex(new_index)
    341         else:
--> 342             start = axlabels[0].asfreq(self.freq, how=self.convention)
    343             end = axlabels[-1].asfreq(self.freq, how='end')
    344 

pandas/src/period.pyx in pandas._period.Period.asfreq (pandas/src/period.c:14801)()

/Users/alex/.virtualenvs/work/lib/python2.7/site-packages/pandas/tseries/frequencies.pyc in get_freq_code(freqstr)
    253             # e.g., freqstr = ('T', 5)
    254             try:
--> 255                 code = _period_str_to_code(freqstr[0])
    256                 stride = freqstr[1]
    257             except:

/Users/alex/.virtualenvs/work/lib/python2.7/site-packages/pandas/tseries/frequencies.pyc in _period_str_to_code(freqstr)
    809                           FutureWarning, stacklevel=3)
    810         except KeyError:
--> 811             raise ValueError("Unknown freqstr: %s" % freqstr)
    812 
    813         return _period_code_map[alias]

ValueError: Unknown freqstr: REQ-L-DEC-SAT-4

Also should be able to downsample to Day and upsample to FY5254

[jreback]

well you need to provide a parse these strings and convert them to proper offsets. That is the first step. Certainly would take a PR on that limited issue. Best to fix things like this in stages and not in big PR's.