def event2stat(edict, top_dt, end_dt, dur, binarize = True):
d_stat = {}
l_label = dtutil.label(top_dt, end_dt, dur)
for eid, l_ev in edict.iteritems():
d_stat[eid] = dtutil.discretize(l_ev, l_label, binarize)
return d_stat
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")