def filter(self, startdate, enddate=None, parent=None, **kwargs):
if 'datetime_index' not in kwargs:
kwargs.update({'datetime_index': self.datetime_index})
kwargs.update({'reindex': self.reindex})
if 'date_check' not in kwargs:
kwargs.update({'date_check': self.date_check})
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None)
startdate, enddate = univ_window[0], univ_window[-1]
# at first, the universe is full, i.e. all sids are included
parent = pd.DataFrame(True,
index=pd.to_datetime(univ_window)
if self.datetime_index else univ_window,
columns=SIDS)
for elem in self._filters:
# if the item is a tuple of filter, we first call their filter method and then intersect the results
if isinstance(elem, tuple) or isinstance(elem, list):
res = [
f.filter(startdate, enddate, parent=parent, **kwargs)
for f in elem
]
for df in res:
parent = parent & df
# simply call its filter method
else:
parent = elem.filter(startdate,
enddate,
parent=parent,
**kwargs)
return parent
def filter(self, startdate, enddate=None, parent=None, **kwargs):
if 'datetime_index' not in kwargs:
kwargs.update({'datetime_index': self.datetime_index})
kwargs.update({'reindex': self.reindex})
if 'date_check' not in kwargs:
kwargs.update({'date_check': self.date_check})
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None)
startdate, enddate = univ_window[0], univ_window[-1]
# at first, the universe is full, i.e. all sids are included
parent = pd.DataFrame(True,
index=pd.to_datetime(univ_window) if self.datetime_index else univ_window,
columns=SIDS)
for elem in self._filters:
# if the item is a tuple of filter, we first call their filter method and then intersect the results
if isinstance(elem, tuple) or isinstance(elem, list):
res = [f.filter(startdate, enddate, parent=parent, **kwargs) for f in elem]
for df in res:
parent = parent & df
# simply call its filter method
else:
parent = elem.filter(startdate, enddate, parent=parent, **kwargs)
return parent
def filter(self, startdate, enddate=None, parent=None, return_parent=False, **kwargs):
datetime_index = kwargs.get('datetime_index', self.datetime_index)
reindex = kwargs.get('reindex', self.reindex)
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None)
si, ei = map(DATES.index, [univ_window[0], univ_window[-1]])
data_window = DATES[si-self.delay-(self.window-1): ei-self.delay+1]
dfs = []
for fetcher, dname in self.datas:
df = fetcher.fetch_window(dname, data_window)
df.index = DATES[si-(self.window-1): ei+1]
dfs.append(df)
df = self.synth(*dfs)
parent = self.comply(df, parent)
df = self.rule(self.window)(df)
self.comply(df, parent, False)
df = df.iloc[self.window-1:]
df.index = univ_window
if return_parent is True:
return (self.format(parent, datetime_index=datetime_index, reindex=reindex),
self.format(df, datetime_index=datetime_index, reindex=reindex))
return self.format(df, datetime_index=datetime_index, reindex=reindex)
def generate_dates(startdate, enddate, num):
if enddate is None:
enddate = DATES[-1]
dates = dateutil.cut_window(
DATES, dateutil.compliment_datestring(str(startdate), -1, True),
dateutil.compliment_datestring(str(enddate), 1, True))
chksize = len(dates) / num + (len(dates) % num > 0)
return [dates[i:i + chksize] for i in range(0, len(dates), chksize)]
def interval_fetch(pl, times, startdate, enddate, backdays=0, date_check=False):
window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check),
backdays
)
return AlphaBase.interval_fetch_window(pl, times, window)
def record_fetch(df, startdate, enddate, backdays, date_check=False):
window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check),
backdays
)
return AlphaBase.record_fetch_window(df, window)
def fetch(self, indicator, startdate, enddate=None, backdays=0, **kwargs):
date_check = kwargs.get('date_check', self.date_check)
window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None,
backdays=backdays)
return self.fetch_window(indicator, window, **kwargs)
def fetch(self, indicator, startdate, enddate=None, backdays=0, **kwargs):
date_check = kwargs.get('date_check', self.date_check)
window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None,
backdays=backdays)
return self.fetch_window(indicator, window, **kwargs)
def fetch(self, dname, times, startdate, enddate=None, backdays=0, **kwargs):
"""Use :py:meth:`fetch_window` behind the scene."""
date_check = kwargs.get('date_check', self.date_check)
window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None,
backdays=backdays)
return self.fetch_window(dname, times, window, **kwargs)
def generate_dates(startdate, enddate, num):
if enddate is None:
enddate = DATES[-1]
dates = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, True),
dateutil.compliment_datestring(str(enddate), 1, True)
)
chksize = len(dates) / num + (len(dates) % num > 0)
return [dates[i: i+chksize] for i in range(0, len(dates), chksize)]
def fetch(self, dname, startdate, enddate=None, backdays=0, **kwargs):
"""Use :py:meth:`fetch_window` behind the scene."""
date_check = kwargs.get('date_check', self.date_check)
window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None,
backdays=backdays)
return self.fetch_window(dname, window, **kwargs)
def filter(self, startdate, enddate=None, parent=None, **kwargs):
datetime_index = kwargs.get('datetime_index', self.datetime_index)
reindex = kwargs.get('reindex', self.reindex)
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None)
index, columns = univ_window, SIDS
df = pd.DataFrame(0, index, columns)
df = df.add(self.series.astype(int), axis=1)
self.comply(df, parent, False)
return self.format(df, datetime_index, reindex)
def filter(self, startdate, enddate=None, parent=None, **kwargs):
datetime_index = kwargs.get('datetime_index', self.datetime_index)
reindex = kwargs.get('reindex', self.reindex)
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None)
index, columns = univ_window, SIDS
df = pd.DataFrame(0, index, columns)
df = df.add(self.series.astype(int), axis=1)
self.comply(df, parent, False)
return self.format(df, datetime_index, reindex)
def generate_dates(startdate, enddate, parts=None):
startdate, enddate = str(startdate), str(enddate)
if enddate[:5].lower() == 'today':
enddate = DATES[-1-int(enddate[6:])]
dates = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, True),
dateutil.compliment_datestring(str(enddate), 1, True)
)
if parts is None:
return dates
chksize = len(dates)/parts
if len(dates) > chksize * parts:
chksize += 1
return [dates[i: i+chksize] for i in range(0, len(dates), chksize)]
def filter(self, startdate, enddate=None, parent=None, **kwargs):
if 'datetime_index' not in kwargs:
kwargs.update({'datetime_index': self.datetime_index})
kwargs.update({'reindex': self.reindex})
if 'date_check' not in kwargs:
kwargs.update({'date_check': self.date_check})
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None)
startdate, enddate = univ_window[0], univ_window[-1]
res = [elem.filter(startdate, enddate=enddate, parent=parent, **kwargs) for elem in self._filters]
seed = res.pop()
for df in res:
seed = seed | df
return seed
def fetch(self, *args, **kwargs):
date_check = kwargs.get('date_check', self.date_check)
enddate = kwargs.get('enddate', None)
backdays = kwargs.get('backdays', 0)
try:
startdate = dateutil.compliment_datestring(str(args[0]), -1, date_check)
dname = self._dname
args = args[1:]
except ValueError:
startdate = dateutil.compliment_datestring(str(args[1]), -1, date_check)
dname = args[0]
args = args[2:]
if args:
enddate = args[0]
window = dateutil.cut_window(
DATES,
startdate,
dateutil.compliment_datestring(str(enddate), 1, date_check) if enddate is not None else None,
backdays=backdays)
return self.fetch_window(dname, window, **kwargs)
def filter(self,
startdate,
enddate=None,
parent=None,
return_parent=False,
**kwargs):
datetime_index = kwargs.get('datetime_index', self.datetime_index)
reindex = kwargs.get('reindex', self.reindex)
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None)
si, ei = map(DATES.index, [univ_window[0], univ_window[-1]])
data_window = DATES[si - self.delay - (self.window - 1):ei -
self.delay + 1]
dfs = []
for fetcher, dname in self.datas:
df = fetcher.fetch_window(dname, data_window)
df.index = DATES[si - (self.window - 1):ei + 1]
dfs.append(df)
df = self.synth(*dfs)
parent = self.comply(df, parent)
df = self.rule(self.window)(df)
self.comply(df, parent, False)
df = df.iloc[self.window - 1:]
df.index = univ_window
if return_parent is True:
return (self.format(parent,
datetime_index=datetime_index,
reindex=reindex),
self.format(df,
datetime_index=datetime_index,
reindex=reindex))
return self.format(df, datetime_index=datetime_index, reindex=reindex)
def filter(self, startdate, enddate=None, parent=None, **kwargs):
if 'datetime_index' not in kwargs:
kwargs.update({'datetime_index': self.datetime_index})
kwargs.update({'reindex': self.reindex})
if 'date_check' not in kwargs:
kwargs.update({'date_check': self.date_check})
date_check = kwargs.get('date_check', self.date_check)
univ_window = dateutil.cut_window(
DATES,
dateutil.compliment_datestring(str(startdate), -1, date_check),
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None)
startdate, enddate = univ_window[0], univ_window[-1]
res = [
elem.filter(startdate, enddate=enddate, parent=parent, **kwargs)
for elem in self._filters
]
seed = res.pop()
for df in res:
seed = seed | df
return seed
def fetch(self, *args, **kwargs):
date_check = kwargs.get('date_check', self.date_check)
enddate = kwargs.get('enddate', None)
backdays = kwargs.get('backdays', 0)
try:
startdate = dateutil.compliment_datestring(str(args[0]), -1,
date_check)
dname = self._dname
args = args[1:]
except ValueError:
startdate = dateutil.compliment_datestring(str(args[1]), -1,
date_check)
dname = args[0]
args = args[2:]
if args:
enddate = args[0]
window = dateutil.cut_window(
DATES,
startdate,
dateutil.compliment_datestring(str(enddate), 1, date_check)
if enddate is not None else None,
backdays=backdays)
return self.fetch_window(dname, window, **kwargs)
def test_cut_window_2(self):
l = [2, 3, 5, 7, 11, 13, 17, 19]
self.assertEqual(dateutil.cut_window(l, 9, None, 2),
[5, 7, 11, 13, 17, 19])
