def similar_block_dag(conf, top_dt, end_dt, area, method, ignore_same=True):
src_dir = conf.get("dag", "output_dir")
ig_direction = conf.getboolean("search", "dag_ig_direction")
wflag = conf.getboolean("search", "dag_weight")
dur = conf.getdur("dag", "stat_bin")
name = pc_log.thread_name(conf, top_dt, end_dt, dur, area)
if name in common.rep_dir(src_dir):
r_temp = pcresult.PCOutput(conf).load(name)
else:
r_temp = pc_log.pc_log(conf, top_dt, end_dt, dur, area, dump=False)
src_dir = conf.get("dag", "output_dir")
l_r = pcresult.results_in_area(conf, src_dir, area)
weight = None
if wflag:
weight = pcresult.EdgeTFIDF(l_r)
result = []
for r in l_r:
if ignore_same and (r.end_dt > top_dt and r.top_dt < end_dt):
# ignore if common term included
pass
else:
if method == "dag_ed":
dist = pcresult.graph_edit_distance(r_temp, r, ig_direction, weight)
elif method == "dag_mcs":
dist = pcresult.mcs_size_ratio(r_temp, r, ig_direction, weight)
else:
raise NotImplementedError
result.append((r, dist))
return result
def test_dag_search(conf, method, area = None):
import cg_dag
if method is None:
method = conf.get("search", "method")
src_dir = conf.get("dag", "output_dir")
if area is None:
l_area = pcresult.result_areas(conf)
else:
l_area = [area]
for area in l_area:
l_r = pcresult.results_in_area(conf, src_dir, area)
result = []
for r in l_r:
if method == "log":
result = similar_block_log(conf, r.top_dt, r.end_dt, r.area,
ignore_same = True)
elif method in ("dag_ed", "dag_mcs"):
result = cg_dag.similar_block_dag(conf, r.top_dt, r.end_dt,
r.area, method, ignore_same = True)
else:
raise NotImplementedError
print r.cond_str()
if len(result) > 10:
result = ex_sorted(result,
key = lambda x: x[1], reverse = False)[:10]
for r_found, val in result:
print val, r_found.cond_str()
print
def similar_block_log(conf, top_dt, end_dt, area, ignore_same = True):
#assert conf.get("search", "method") == "log"
ld = log_db.LogData(conf)
dagc = DAGComparison(conf, area)
edict = dagc.data_for_cond(top_dt, end_dt, area)
if edict is None:
edict = {}
for line in ld.iter_lines(top_dt = top_dt, end_dt = end_dt,
area = area):
weid = dagc.w_evmap.process_line(line)
edict[weid] = edict.get(weid, 0) + 1
l_weid = edict.keys()
src_evv = _event_vector(l_weid, edict, dagc)
print("{0} - {1} ({2}) : {3}".format(top_dt, end_dt, area, src_evv))
src_dir = conf.get("dag", "output_dir")
l_r = pcresult.results_in_area(conf, src_dir, area)
result = []
for r in l_r:
if ignore_same and (r.end_dt > top_dt and r.top_dt < end_dt):
# ignore if common term included
pass
else:
edict_r = dagc.data_for_r(r)
r_evv = _event_vector(l_weid, edict_r, dagc)
print("{0} - {1} ({2}) : {3}".format(r.top_dt, r.end_dt,
r.area, r_evv))
dist = _evv_distance(src_evv, r_evv)
result.append((r, dist))
return result
def similar_block_dag(conf, top_dt, end_dt, area, method, ignore_same=True):
src_dir = conf.get("dag", "output_dir")
ig_direction = conf.getboolean("search", "dag_ig_direction")
wflag = conf.getboolean("search", "dag_weight")
dur = conf.getdur("dag", "stat_bin")
name = pc_log.thread_name(conf, top_dt, end_dt, dur, area)
if name in common.rep_dir(src_dir):
r_temp = pcresult.PCOutput(conf).load(name)
else:
r_temp = pc_log.pc_log(conf, top_dt, end_dt, dur, area, dump=False)
src_dir = conf.get("dag", "output_dir")
l_r = pcresult.results_in_area(conf, src_dir, area)
weight = None
if wflag:
weight = pcresult.EdgeTFIDF(l_r)
result = []
for r in l_r:
if ignore_same and (r.end_dt > top_dt and r.top_dt < end_dt):
# ignore if common term included
pass
else:
if method == "dag_ed":
dist = pcresult.graph_edit_distance(r_temp, r, ig_direction,
weight)
elif method == "dag_mcs":
dist = pcresult.mcs_size_ratio(r_temp, r, ig_direction, weight)
else:
raise NotImplementedError
result.append((r, dist))
return result
def _init_event_stat(self, conf):
ld = log_db.LogData(conf)
w_top_dt, w_end_dt = pc_log.whole_term(conf, ld)
gid_name = conf.get("dag", "event_gid")
self.w_evmap = log2event.EventDefinitionMap(w_top_dt, w_end_dt,
gid_name)
src_dir = conf.get("dag", "output_dir")
l_r = pcresult.results_in_area(conf, src_dir, self.area)
for r in l_r:
edict = {}
for line in ld.iter_lines(top_dt = r.top_dt, end_dt = r.end_dt,
area = r.area):
weid = self.w_evmap.process_line(line)
edict[weid] = edict.get(weid, 0) + 1
self.d_ev[(r.top_dt, r.end_dt, r.area)] = edict
def heatmap(conf, method, area, fn):
def result2data(result, l_label):
d_temp = {}
for r, dist in result:
d_temp[r.get_fn()] = dist
return [d_temp[label] for label in l_label]
import cg_dag
if method is None:
method = conf.get("search", "method")
if area is None:
area = "all"
src_dir = conf.get("dag", "output_dir")
l_r = pcresult.results_in_area(conf, src_dir, area)
l_label = [r.get_fn() for r in l_r]
l_result = []
for r in l_r:
if method == "log":
result = similar_block_log(conf, r.top_dt, r.end_dt,
r.area, ignore_same = False)
elif method in ("dag_ed", "dag_mcs"):
result = cg_dag.similar_block_dag(conf,
r.top_dt, r.end_dt, r.area, method, ignore_same = False)
else:
raise NotImplementedError
l_result.append(result2data(result, l_label))
# replace None to max value of whole result
mval = max([max(result) for result in l_result])
data = np.array([[mval if i is None else i for i in result] for result
in l_result])
# data = np.array(l_result)
n = len(l_r)
assert data.shape == (n, n)
x, y = np.meshgrid(np.arange(n + 1), np.arange(n + 1))
explot.dump(fn + ".temp", (x, y, data))
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
cm = explot.generate_cmap(["orangered", "white"])
plt.pcolor(x, y, data, cmap = cm)
plt.colorbar()
plt.savefig(fn)
