def test_leakage_limit_nrecords():
data = np.array(
[(pd.Timestamp("2019-01-01 01:02:03").value / 10**9, 1, 2, 0)],
dtype=[("Epoch", "i8"), ("Bid", "f4"), ("Ask", "f4"),
("Nanoseconds", "i4")],
)
data2 = np.array(
[(pd.Timestamp("2019-01-01 01:02:03").value / 10**9, 3, 4, 100_000_000)
],
dtype=[("Epoch", "i8"), ("Bid", "f4"), ("Ask", "f4"),
("Nanoseconds", "i4")],
)
client.write(data, "DEBUG/1Sec/TICK", isvariablelength=True)
client.write(data2, "DEBUG/1Sec/TICK", isvariablelength=True)
param = pymkts.Param(
"DEBUG",
"1Sec",
"TICK",
start=pd.Timestamp("2019-01-01 01:02:03.100000000"),
limit=1,
limit_from_start=True,
)
reply = client.query(param)
df = reply.first().df()
print(df)
# if wrong, it means the first tick leaks outside of the interval starting within
# `start` parameter
cond1 = ((df.shape[0] == 1) and (df.Bid[0] == 3)
and (df.Nanoseconds[0] == 100_000_000))
assert cond1
param = pymkts.Param(
"DEBUG",
"1Sec",
"TICK",
end=pd.Timestamp("2019-01-01 01:02:03.000000000"),
limit=1,
limit_from_start=False,
)
reply = client.query(param)
df = reply.first().df()
print(df)
# if wrong, it means the second tick leaks outside of the interval ending within
# `end` parameter
cond2 = (df.shape[0] == 1) and (df.Bid[0] == 1) and (df.Nanoseconds[0]
== 0)
assert cond2
def random_query(self, symbol: str, timeframe: str, attribute_group: str,
size: int, num: int, limit_from_start: bool,
time_range: str, limit: str) -> int:
# TODO: add start, end, limit, limit_from_start options
if num <= 0:
return 0
params = []
for k in range(num):
start, end = get_range(time_range)
params.append(
pymkts.Param(symbol,
timeframe,
attribute_group,
start=start,
end=end,
limit=get_limit(limit, size),
limit_from_start=limit_from_start))
now = time.time_ns()
for k in range(num):
self.cli.query(params[k])
# reply = self.cli.query(params[k])
# print(reply.first().df())
elapsed = time.time_ns() - now
return elapsed
def random_query(self, symbol: str, timeframe: str, attribute_group: str,
size: int, num: int, limit_from_start: bool,
time_range: str, limit: str, grpc: bool) -> int:
if num <= 0:
return 0
params = []
for k in range(num):
start, end = get_range(time_range)
params.append(
pymkts.Param(symbols=symbol,
timeframe=timeframe,
attrgroup=attribute_group,
start=start,
end=end,
limit=get_limit(limit, size),
limit_from_start=limit_from_start))
c = self.get_client(grpc)
# first query takes a lot of time compared to the following queries.
# query once to remove the effect.
if num > 0:
c.query(params=params[0])
now = time.time_ns()
for k in range(num):
c.query(params=params[k])
# reply = self.cli.query(params[k])
# print(reply.first().df())
elapsed = time.time_ns() - now
return elapsed
def get_bulk_history(self, symbols, timeframe, attrgroup='OHLCV',
start=None, end=None,
limit=None, limit_from_start=False):
if isinstance(symbols, str):
symbols = [symbols]
p = pymkts.Param(','.join(symbols), get_timeframe(timeframe), attrgroup,
start=start, end=end,
limit=limit, limit_from_start=limit_from_start)
r = self.client.query(p)
return {symbol: ds.df() for symbol, ds in r.by_symbols().items()}
def get_history(self, symbol, timeframe, attrgroup='OHLCV',
start=None, end=None,
limit=None, limit_from_start=False):
p = pymkts.Param(symbol, get_timeframe(timeframe), attrgroup,
start=start, end=end,
limit=limit, limit_from_start=limit_from_start)
try:
return self.client.query(p).first().df()
except Exception as e:
if 'No files returned from query parse' in e.args[0]:
return None
raise e
def get_df(symbol, timeframe='1D', attrgroup='OHLCV'):
client = pymkts.Client()
p = pymkts.Param(symbol, timeframe, attrgroup, start='2015-01-01')
return client.query(p).first().df()