def modelAEM(etrs):
print("Modeling AEM ...")
# Modeling traffic with AEM
aem = AEM(etrs, cut_gap)
withUA = aem.sessionsWithUA()
withoutUA = aem.sessionsWithoutUA()
forest = {}
for ua in withUA:
ss = withUA[ua]
for el in ss[1:]:
trees = aem.model(el)
if ua not in forest:
forest[ua] = []
forest[ua].extend(trees)
for host in withoutUA:
nss = withoutUA[host]
for el in nss[1:]:
trees = aem.model(el)
if host not in forest:
forest[host] = []
forest[host].extend(trees)
return forest
input_folder = sys.argv[1]
hl = os.path.join(input_folder, "http_logs")
if not os.path.exists(hl):
raise Exception("Sry, I do not find *http_logs*.out in given folder.")
# read http logs
etrs = []
for entry in HTTPLogReader(hl):
if entry is not None:
etrs.append(entry)
etrs.sort(key=lambda x: x.rqtstart())
if len(etrs) == 0:
print("No http logs")
sys.exit(-1)
aem = AEM(etrs)
withUA = aem.sessionsWithUA()
withoutUA = aem.sessionsWithoutUA()
# output file
output = os.path.join(input_folder, "aem.out")
try:
os.remove(output)
except OSError:
pass
for ua in withUA:
ss = withUA[ua]
open(output, "ab").write("\n****************** %s\n" % ua)
for el in ss[1:]:
trees = aem.model(el)
