def update(self, conf):
ld = log_db.LogData(conf)
db_top_dt, db_end_dt = ld.dt_term()
if self._end_dt is not None and \
self._end_dt + self._binsize < db_end_dt:
_logger.warning("New data is too small or not found")
return
self._evmap = log2event.generate_evmap(conf, ld, None, None)
if self._end_dt is None:
top_dt = dtutil.adj_sep(db_top_dt, self._binsize)
else:
top_dt = self._end_dt
# The last bin will not be added, because it may be uncompleted
end_dt = dtutil.adj_sep(db_end_dt, self._binsize)
l_label = dtutil.label(top_dt, end_dt, self._binsize)
_logger.info("updating changepoint data ({0} - {1})".format(
top_dt, end_dt))
for eid in self._evmap.iter_eid():
evdef = self._evmap.info(eid)
_logger.info("processing {0}".format(self._evmap.info_str(eid)))
cf = self._load_cf(evdef)
l_data, l_score = self._load_data(evdef)
if cf is None:
cf = self._new_cf()
l_data = []
l_score = []
l_dt = [
line.dt for line in ld.iter_lines(
**self._evmap.iterline_args(eid, top_dt, end_dt))
]
if len(l_dt) > 0:
_logger.info("{0} messages in given term".format(len(l_dt)))
l_val = dtutil.discretize(l_dt, l_label, binarize=False)
for val in l_val:
l_data.append(val)
score = cf.update(val)
l_score.append(score)
self._dump_cf(evdef, cf)
self._dump_data(evdef, l_data, l_score)
else:
_logger.info("no message found in processing term, passed")
self._end_dt = end_dt
self._dt_label += l_label
if self._top_dt is None:
self._top_dt = top_dt
_logger.info("task completed")
def update(self, conf):
ld = log_db.LogData(conf)
db_top_dt, db_end_dt = ld.dt_term()
if self._end_dt is not None and \
self._end_dt + self._binsize < db_end_dt:
_logger.warning("New data is too small or not found")
return
self._evmap = log2event.generate_evmap(conf, ld, None, None)
if self._end_dt is None:
top_dt = dtutil.adj_sep(db_top_dt, self._binsize)
else:
top_dt = self._end_dt
# The last bin will not be added, because it may be uncompleted
end_dt = dtutil.adj_sep(db_end_dt, self._binsize)
l_label = dtutil.label(top_dt, end_dt, self._binsize)
_logger.info("updating changepoint data ({0} - {1})".format(
top_dt, end_dt))
for eid in self._evmap.iter_eid():
evdef = self._evmap.info(eid)
_logger.info("processing {0}".format(self._evmap.info_str(eid)))
cf = self._load_cf(evdef)
l_data, l_score = self._load_data(evdef)
if cf is None:
cf = self._new_cf()
l_data = []
l_score = []
l_dt = [line.dt for line in ld.iter_lines(
**self._evmap.iterline_args(eid, top_dt, end_dt))]
if len(l_dt) > 0:
_logger.info("{0} messages in given term".format(len(l_dt)))
l_val = dtutil.discretize(l_dt, l_label, binarize = False)
for val in l_val:
l_data.append(val)
score = cf.update(val)
l_score.append(score)
self._dump_cf(evdef, cf)
self._dump_data(evdef, l_data, l_score)
else:
_logger.info("no message found in processing term, passed")
self._end_dt = end_dt
self._dt_label += l_label
if self._top_dt is None:
self._top_dt = top_dt
_logger.info("task completed")
def get(self, evdef, top_dt = None, end_dt = None):
if top_dt is None:
top_dt = self._top_dt
if end_dt is None:
end_dt = self._end_dt
l_label = self._dt_label
l_data, l_score = self._load_data(evdef)
if len(l_data) == 0:
return None
if dtutil.is_sep(top_dt, self._binsize):
top_index = l_label.index(top_dt)
else:
top_index = l_label.index(dtutil.adj_sep(top_dt, self._binsize))
if dtutil.is_sep(end_dt, self._binsize):
end_index = l_label.index(end_dt)
else:
end_index = l_label.index(dtutil.radj_sep(end_dt, self._binsize))
return zip(self._dt_label, l_data, l_score)[top_index:end_index]
def get(self, evdef, top_dt=None, end_dt=None):
if top_dt is None:
top_dt = self._top_dt
if end_dt is None:
end_dt = self._end_dt
l_label = self._dt_label
l_data, l_score = self._load_data(evdef)
if len(l_data) == 0:
return None
if dtutil.is_sep(top_dt, self._binsize):
top_index = l_label.index(top_dt)
else:
top_index = l_label.index(dtutil.adj_sep(top_dt, self._binsize))
if dtutil.is_sep(end_dt, self._binsize):
end_index = l_label.index(end_dt)
else:
end_index = l_label.index(dtutil.radj_sep(end_dt, self._binsize))
return zip(self._dt_label, l_data, l_score)[top_index:end_index]
def dt_ticks(value_term, dt_term, dt_bin, duration = None, recent = False):
"""
Args:
value_term (int, int)
dt_term (datetime.datetime, datetime.datetime):
The range of datetime for ticks, corresponding to value_term.
dt_bin (datetime.timedelta): Bin length in datetime.
duration (datetime.timedelta): Duration of ticks. If None,
automatically decided in this function.
recent (bool): In default, ticks are decided from old datetime.
If recent is true, ticks are decided from recent datetime.
"""
import dtutil
top_dt, end_dt = dt_term
if duration is None:
whole_term = end_dt - top_dt
if whole_term <= datetime.timedelta(minutes = 3):
duration = datetime.timedelta(seconds = 10)
elif whole_term <= datetime.timedelta(minutes = 10):
duration = datetime.timedelta(minutes = 1)
elif whole_term <= datetime.timedelta(minutes = 30):
duration = datetime.timedelta(minutes = 3)
elif whole_term <= datetime.timedelta(hours = 1):
duration = datetime.timedelta(minutes = 5)
elif whole_term <= datetime.timedelta(hours = 6):
duration = datetime.timedelta(minutes = 10)
elif whole_term <= datetime.timedelta(days = 1):
duration = datetime.timedelta(hours = 6)
elif whole_term <= datetime.timedelta(days = 7):
duration = datetime.timedelta(days = 1)
elif whole_term <= datetime.timedelta(days = 14):
duration = datetime.timedelta(days = 2)
elif whole_term <= datetime.timedelta(days = 60):
duration = datetime.timedelta(days = 7)
elif whole_term <= datetime.timedelta(days = 300):
duration = datetime.timedelta(days = 20)
else:
duration = datetime.timedelta(days = 50)
if duration >= datetime.timedelta(days = 1):
adjsearch_dur = datetime.timedelta(days = 1)
dtstr = lambda dt: dt.strftime("%Y-%m-%d")
else:
adjsearch_dur = duration
dtstr = lambda dt: dt.strftime("%H:%M:%S")
ticks_label = []
if recent is False:
temp_dt = dtutil.radj_sep(top_dt, adjsearch_dur)
assert(temp_dt <= end_dt)
ticks_label.append(temp_dt)
while 1:
temp_dt = temp_dt + duration
if temp_dt > end_dt:
break
else:
ticks_label.append(temp_dt)
else:
temp_dt = dtutil.adj_sep(end_dt, adjsearch_dur)
assert(temp_dt >= top_dt)
ticks_label.append(temp_dt)
while 1:
temp_dt = temp_dt - duration
if temp_dt < top_dt:
break
else:
ticks_label.append(temp_dt)
ticks_label.sort()
val_label = range(value_term[0], value_term[1] + 1, 1)
dt_label = dtutil.dtrange(top_dt, end_dt, dt_bin, include_end = True)
ticks_values = []
ticks_dts = ticks_label[:]
for val, dt in zip(val_label, dt_label):
if dt == ticks_dts[0]:
ticks_dts.pop(0)
ticks_values.append(val)
if len(ticks_dts) == 0:
break
return ticks_values, [dtstr(l) for l in ticks_label]
