def test_remote_data_service_industry():
from quantos.data.align import align
import pandas as pd
ds = RemoteDataService()
arr = ds.get_index_comp(index='000300.SH', start_date=20130101, end_date=20170505)
df = ds.get_industry_raw(symbol=','.join(arr), type_='ZZ')
df = df.astype(dtype={'in_date': int})
# df_ann = df.loc[:, ['in_date', 'symbol']]
# df_ann = df_ann.set_index(['symbol', 'in_date'])
# df_ann = df_ann.unstack(level='symbol')
from quantos.data.dataview import DataView
dic_sec = DataView._group_df_to_dict(df, by='symbol')
dic_sec = {sec: df.reset_index() for sec, df in dic_sec.viewitems()}
df_ann = pd.concat([df.loc[:, 'in_date'].rename(sec) for sec, df in dic_sec.viewitems()], axis=1)
df_value = pd.concat([df.loc[:, 'industry1_code'].rename(sec) for sec, df in dic_sec.viewitems()], axis=1)
dates_arr = ds.get_trade_date(20140101, 20170505)
res = align(df_value, df_ann, dates_arr)
print
# df_ann = df.pivot(index='in_date', columns='symbol', values='in_date')
# df_value = df.pivot(index=None, columns='symbol', values='industry1_code')
def align_single_df(df_one_sec):
df_value = df_one_sec.loc[:, ['industry1_code']]
df_ann = df_one_sec.loc[:, ['in_date']]
res = align(df_value, df_ann, dates_arr)
return res
# res_list = [align_single_df(df) for sec, df in dic_sec.viewitems()]
res_list = [align_single_df(df) for sec, df in dic_sec.items()[:10]]
res = pd.concat(res_list, axis=1)
print res
def test_align():
# -------------------------------------------------------------------------------------
# input and pre-process demo data
ds = RemoteDataService()
raw, msg = ds.query_lb_fin_stat('income', '600000.SH', 20151225, 20170501,
'oper_rev')
assert msg == '0,'
idx_list = ['report_date', 'symbol']
raw_idx = raw.set_index(idx_list)
raw_idx.sort_index(axis=0, level=idx_list, inplace=True)
df_ann = raw_idx.loc[pd.IndexSlice[:, :], 'ann_date']
df_ann = df_ann.unstack(level=1)
df_value = raw_idx.loc[pd.IndexSlice[:, :], 'oper_rev']
df_value = df_value.unstack(level=1)
date_arr = ds.get_trade_date(20160101, 20170501)
df_close = pd.DataFrame(index=date_arr, columns=df_value.columns, data=1e3)
# -------------------------------------------------------------------------------------
# demo usage of parser
parser = Parser()
parser.register_function(
'Myfunc',
lambda x: x * 0 + 1) # simultaneously test register function and align
expr_formula = 'revenue / Myfunc(close)'
expression = parser.parse(expr_formula)
for i in range(100):
df_res = parser.evaluate({
'revenue': df_value,
'close': df_close
}, df_ann, date_arr)
# -------------------------------------------------------------------------------------
sec = '600000.SH'
"""
# print to validate results
print "\n======Expression Formula:\n{:s}".format(expr_formula)
print "\n======Report date, ann_date and evaluation value:"
tmp = pd.concat([df_ann.loc[:, sec], df_value.loc[:, sec]], axis=1)
tmp.columns = ['df_ann', 'df_value']
print tmp
print "\n======Selection of result of expansion:"
print "20161028 {:.4f}".format(df_res.loc[20161028, sec])
print "20161031 {:.4f}".format(df_res.loc[20161031, sec])
print "20170427 {:.4f}".format(df_res.loc[20170427, sec])
"""
assert abs(df_res.loc[20161028, sec] - 82172000000) < 1
assert abs(df_res.loc[20161031, sec] - 120928000000) < 1
assert abs(df_res.loc[20170427, sec] - 42360000000) < 1
