def __load_fake_tables(session):
df = pd.DataFrame([], columns=['TIME', 'OPEN_PRICE', 'MIN_PRICE', 'MAX_PRICE', 'CLOSE_PRICE', 'VOLUME'])
sessions_num = 12
today = date.today()
t = 0
for s in xrange(sessions_num):
dt = datetime.combine(today + timedelta(days=s), datetime.min.time())
while dt.time() < datetime(dt.year, dt.month, dt.day, hour=23, minute=59, second=0).time():
price = 1 + t % 2
data_row = [MarketSession.datetime_to_ms(dt), price, price, price, price, 1.0]
df.loc[t] = data_row
dt += timedelta(minutes=1)
t += 1
data_row = [MarketSession.datetime_to_ms(dt), price, price, price, price, 1.0]
df.loc[t] = data_row
return df, FxSingleCurrencyBroker.__split_sessions(df, session)
def __split_sessions(df, session_type):
"""
:param df: pandas data frame with currency exchange rates
:param session_type: type of session (Europe/Asia/America)
:return: list of extracted sessions
"""
stop = False
base_pointer = 0
sessions = []
while not stop:
time_point_ms = df.ix[base_pointer, 'TIME']
# get current day
day = MarketSession.ms_to_datetime(time_point_ms)
# get session's time limits for current day
begin_ms, end_ms = session_type.get_session_range(day)
# find time limits in database
begin_ids = df[df['TIME'] == begin_ms].index.tolist()
end_ids = df[df['TIME'] == end_ms + 1].index.tolist()
# there should be only single time point per each limit
if len(begin_ids) == 1 and len(end_ids) == 1:
interval_len = end_ids[0] - begin_ids[0]
volumes = df.ix[(interval_len/4 + begin_ids[0]):(3*interval_len/4 + begin_ids[0]), 'VOLUME']
prices = df.ix[(interval_len/4 + begin_ids[0]):(3*interval_len/4 + begin_ids[0]), 'CLOSE_PRICE']
# check real trading activity
if volumes.sum() and prices.max() != prices.min():
sessions.append((begin_ids[0], end_ids[0] - 1))
_, max_days = monthrange(day.year, day.month)
for i in xrange(1, max_days + 1):
next_point_ids = df[df['TIME'] == begin_ms + session_type.get_session_period()*i].index.tolist()
if len(next_point_ids) == 1:
base_pointer = next_point_ids[0]
stop = False
break
else:
stop = True
return sessions
