def remake_lt(ld, ltid):
# use iter_words, because iter_lines needs ltline registered in table
# but ltline on new_ltid is now on construction in this function...
new_lt = None
cnt = 0
for l_w in ld.db.iter_words(ltid=ltid):
print l_w
if new_lt is None:
new_lt = l_w
else:
new_lt = lt_common.merge_lt(new_lt, l_w, sym)
cnt += 1
print "result : " + str(new_lt)
print
return new_lt, cnt
def remake_lt(ld, ltid):
# use iter_words, because iter_lines needs ltline registered in table
# but ltline on new_ltid is now on construction in this function...
new_lt = None
cnt = 0
for l_w in ld.db.iter_words(ltid = ltid):
print l_w
if new_lt is None:
new_lt = l_w
else:
new_lt = lt_common.merge_lt(new_lt, l_w, sym)
cnt += 1
print "result : " + str(new_lt)
print
return new_lt, cnt
def process_line(self, l_w, l_s):
n_parent = self.n_root
while True:
for n_child in n_parent:
_logger.debug("comparing with ltid {0}".format(
n_child.lt.ltid))
nc_lt = n_child.lt.ltw
sr = self.seq_ratio(nc_lt, l_w)
_logger.debug("seq_ratio : {0}".format(sr))
if sr >= self.threshold:
_logger.debug("merged with ltid {0}".format(
n_child.lt.ltid))
new_lt = lt_common.merge_lt(nc_lt, l_w, self.sym)
if new_lt == nc_lt:
self.ltm.count_lt(n_child.lt.ltid)
else:
self.ltm.replace_and_count_lt(n_child.lt.ltid, new_lt)
_logger.debug("ltid {0} replaced".format(
n_child.lt.ltid))
_logger.debug("-> {0}".format(str(n_child.lt)))
return n_child.lt, False
else:
if self.equal(nc_lt, l_w):
_logger.warning(
"comparing same line, but seqratio is small...")
else:
if len(n_parent) < self.max_child:
_logger.debug("no node to be merged, add new node")
n = self._new_node(l_w, l_s)
n_parent.join(n)
return n.lt, True
else:
_logger.debug("children : {0}".format(
[e.lt.ltid for e in n_parent.l_child]))
l_sim = [(edit_distance(n_child.lt.ltw, l_w,
self.sym), n_child)
for n_child in n_parent]
n_parent = max(l_sim, key=lambda x: x[0])[1]
_logger.debug("go down to node(ltid {0})".format(
n_parent.lt.ltid))
def process_line(self, l_w, l_s):
n_parent = self.n_root
while True:
for n_child in n_parent:
_logger.debug(
"comparing with ltid {0}".format(n_child.lt.ltid))
nc_lt = n_child.lt.ltw
sr = self.seq_ratio(nc_lt, l_w)
_logger.debug("seq_ratio : {0}".format(sr))
if sr >= self.threshold:
_logger.debug(
"merged with ltid {0}".format(n_child.lt.ltid))
new_lt = lt_common.merge_lt(nc_lt, l_w, self.sym)
if new_lt == nc_lt:
self.ltm.count_lt(n_child.lt.ltid)
else:
self.ltm.replace_and_count_lt(n_child.lt.ltid, new_lt)
_logger.debug(
"ltid {0} replaced".format(n_child.lt.ltid))
_logger.debug("-> {0}".format(str(n_child.lt)))
return n_child.lt, False
else:
if self.equal(nc_lt, l_w):
_logger.warning(
"comparing same line, but seqratio is small...")
else:
if len(n_parent) < self.max_child:
_logger.debug("no node to be merged, add new node")
n = self._new_node(l_w, l_s)
n_parent.join(n)
return n.lt, True
else:
_logger.debug("children : {0}".format(
[e.lt.ltid for e in n_parent.l_child]))
l_sim = [(edit_distance(n_child.lt.ltw, l_w, self.sym),
n_child) for n_child in n_parent]
n_parent = max(l_sim, key=lambda x: x[0])[1]
_logger.debug("go down to node(ltid {0})".format(
n_parent.lt.ltid))
def merge_ltid(ld, ltid1, ltid2, sym):
ld.set_ltm()
print("merge following log templates...")
print _str_lt(ltid1)
print _str_lt(ltid2)
print
ltw1 = ld.lt(ltid1).ltw
cnt1 = ld.lt(ltid1).cnt
l_s = ld.lt(ltid1).lts
ltw2 = ld.lt(ltid2).ltw
cnt2 = ld.lt(ltid2).cnt
if not len(ltw1) == len(ltw2):
sys.exit("log template length is different, failed")
new_ltw = lt_common.merge_lt(ltw1, ltw2, sym)
ld.ltm.replace_lt(ltid1, new_ltw, l_s, cnt1 + cnt2)
ld.ltm.remove_lt(ltid2)
ld.db.update_log({"ltid": ltid2}, {"ltid": ltid1})
ld.commit_db()
print("> new log template : ltid {0}".format(ltid1))
print _str_lt(ltid1)
def merge_ltid(ld, ltid1, ltid2, sym):
ld.set_ltm()
print("merge following log templates...")
print _str_lt(ltid1)
print _str_lt(ltid2)
print
ltw1 = ld.lt(ltid1).ltw
cnt1 = ld.lt(ltid1).cnt
l_s = ld.lt(ltid1).lts
ltw2 = ld.lt(ltid2).ltw
cnt2 = ld.lt(ltid2).cnt
if not len(ltw1) == len(ltw2):
sys.exit("log template length is different, failed")
new_ltw = lt_common.merge_lt(ltw1, ltw2, sym)
ld.ltm.replace_lt(ltid1, new_ltw, l_s, cnt1 + cnt2)
ld.ltm.remove_lt(ltid2)
ld.db.update_log({"ltid" : ltid2}, {"ltid" : ltid1})
ld.commit_db()
print("> new log template : ltid {0}".format(ltid1))
print _str_lt(ltid1)