def test_historical_regular_holidays_fall_into_precomputed_holidays(self):
from pandas import DatetimeIndex
precomputed_holidays = DatetimeIndex(self.calendar.adhoc_holidays)
# precomputed holidays won't include weekends (saturday, sunday)
self.assertTrue(all(d.weekday() < 5 for d in precomputed_holidays))
generated_holidays = self.calendar.regular_holidays.holidays(
precomputed_holidays.min(),
precomputed_holidays.max(),
return_name=True)
# generated holidays include weekends
self.assertFalse(all(d.weekday() < 5
for d in generated_holidays.index))
# filter non weekend generated holidays
non_weekend_mask = DatetimeIndex(
[d for d in generated_holidays.index if d.weekday() < 5])
non_weekend_generated_holidays = generated_holidays[non_weekend_mask]
# generated holidays should generally fall into one of the precomputed holidays
# except the future holidays that are not precomputed yet
isin = non_weekend_generated_holidays.index.isin(precomputed_holidays)
missing = non_weekend_generated_holidays[~isin]
self.assertTrue(all(isin), "missing holidays = \n%s" % missing)
def read_eng_file(private_file: str,
date_index: bool = True,
compute_date: bool = True,
allowed_flags: Sequence[int] = (0, ),
dates: pd.DatetimeIndex = None) -> pd.DataFrame:
"""Read a .eof.csv (engineering output file, comma-separated value format) file
Parameters
----------
private_file:
the path to the private netCDF file or the .eof.csv file
date_index:
if `True` then the returned dataframe is indexed by date.
compute_date:
if `True` and `date_index` is `False`, then a 'date' column is added to the dataframe containing the observation
datetimes. Has no effect if reading a netCDF file.
allowed_flags:
which quality flags are kept in the dataframe. If `None` or the string `'all'` any flag is valid.
dates:
a date array indicating the date range of data to retain. Data between the min and max of this array will be
kept. If this is `None`, no date limiting is done. If this is given, `compute_date` is considered `True` regardless
of its actual value.
Returns
-------
pd.DataFrame:
dataframe with all the information from the .eof.csv file
"""
if private_file.endswith('.nc') or private_file.endswith('.nc4'):
df = _read_private_nc(private_file, date_index=date_index)
else:
df = _read_eof_csv(private_file,
date_index=date_index,
compute_date=compute_date)
if allowed_flags is None or allowed_flags == 'all':
xx = df['flag'] > -99
else:
xx = df['flag'].isin(allowed_flags)
if dates is not None:
if date_index:
df_dates = this_df.index
else:
df_dates = this_df['date']
xx &= (this_df['date'] >= dates.min()) & (this_df['date'] <=
dates.max())
return df[xx]
def to_tree(cls, node: pd.DatetimeIndex, ctx):
"""Serialize DatetimeIndex to tree."""
tree = {}
if node.inferred_freq is not None:
tree["freq"] = node.inferred_freq
else:
tree["values"] = node.values.astype(np.int64)
tree["start"] = node[0]
tree["end"] = node[-1]
tree["min"] = node.min()
tree["max"] = node.max()
return tree
def to_yaml_tree(self, obj: pd.DatetimeIndex, tag: str, ctx) -> dict:
"""Convert to python dict."""
tree = {}
if obj.inferred_freq is not None:
tree["freq"] = obj.inferred_freq
else:
tree["values"] = obj.values.astype(np.int64)
tree["start"] = obj[0]
tree["end"] = obj[-1]
tree["min"] = obj.min()
tree["max"] = obj.max()
return tree
def create_cal(trade_dates: pd.DatetimeIndex) -> pd.DataFrame:
"""构造交易日历
Args:
trade_dates (pd.DatatimeIndex, optional): 交易日. Defaults to None.
Returns:
pd.DataFrame: 交易日历表
"""
min_date = trade_dates.min()
max_date = trade_dates.max()
dates = pd.date_range(min_date, max_date)
df = pd.DataFrame(index=dates)
df['is_tradeday'] = False
df.loc[trade_dates, 'is_tradeday'] = True
return df
