def str_to_dates_test(period, exp):
""" test module for str_to_dates() """
res = fetch.str_to_dates(period)
if res == exp:
print('pass')
else:
print('fail')
def new_price_in_n_days_multi(stock, period, params):
"""
filter out price if it's new high/low in specified multiple windows
"""
# parse params
windows = params[0]
windows.sort() # e.g. [20, 60, 120, 240]
# -1 for it's included
res = fetch.fetch_price('index', period, windows[-1] - 1) # just get data once, so get it big
if res[0] == False: # empty files for half of a year
return False
else:
idx_df = res[1]
# -1 for it's included
res = fetch.fetch_price(stock, period, windows[-1] - 1) # just get data once, so get it big
if res[0] == False: # empty files for half of a year
return False
else:
df = res[1]
# be compatible between twse/otc
if fetch.stock_table.loc[stock]['TWSE/OTC'] == '上櫃':
price_label = '收盤'
else:
price_label = '收盤價'
df = pandas.DataFrame(index=idx_df.index, data=df[price_label])
period = fetch.str_to_dates(period)
if period[0] != [False]:
start_dt, end_dt = period
else:
return False
res_list = []
for win in windows:
for high_low in ['high', 'low']:
if high_low == 'high':
res = pandas.stats.moments.rolling_max(df[price_label], win, min_periods=1)
else:
res = pandas.stats.moments.rolling_min(df[price_label], win, min_periods=1)
df[str(win) + high_low] = res == df[price_label]
df = df.drop(price_label, 1)
df = df[str(start_dt) <= res.index]
df = df[df.index <= str(end_dt)]
return df.astype(int)