def beamStep(self, re):
newRegexes = list()
# generate merged steps
for stateID1 in list((s.ID_) for s in re.states_.values()):
for stateID2 in list((s.ID_) for s in re.states_.values()):
if stateID1 == stateID2:
continue
newRe = re.copy()
newRe.mergeRandom(stateID1, stateID2)
newRegexes.append((newRe, self.likelihood(newRe)))
# generate wildcard steps
for stateID1 in list((s.ID_) for s in re.states_.values()):
for wildcard in ['S', 'N', 'A']:
for k, s in re.states_[stateID1].next_:
if keysOverlap(k,
wildcard) and keyMinus(wildcard, k) != '':
newRe = re.copy()
newRe.wildcardize(stateID1, wildcard)
newRegexes.append((newRe, self.likelihood(newRe)))
# only replace one of the transitions for a wildcard
break
return newRegexes
def beamStep(self, re): newRegexes = list() # generate merged steps for stateID1 in list((s.ID_) for s in re.states_.values()): for stateID2 in list((s.ID_) for s in re.states_.values()): if stateID1 == stateID2: continue newRe = re.copy() newRe.mergeRandom(stateID1, stateID2) newRegexes.append((newRe, self.likelihood(newRe))) # generate wildcard steps for stateID1 in list((s.ID_) for s in re.states_.values()): for wildcard in ['S', 'N', 'A']: for k, s in re.states_[stateID1].next_: if keysOverlap(k, wildcard) and keyMinus(wildcard, k) != '': newRe = re.copy() newRe.wildcardize(stateID1, wildcard) newRegexes.append((newRe, self.likelihood(newRe))) # only replace one of the transitions for a wildcard break return newRegexes
def subsetConstruction(self):
# copy old
old = self.copy()
# clear values
self.mergeQueue_ = list()
self.states_ = dict()
self.lastStateID_ = -1
# initialize DFA
self.start_ = State(regex=self, start=True)
self.start_.accept_ = old.start_.accept_
stateMapping = {0: [0]}
stateQueue = [0]
while len(stateQueue) > 0:
stateID = stateQueue.pop(0)
state = self.states_[stateID]
if DEBUG: print "\nstate", stateID, "=", stateMapping[stateID]
# create distinct transitons
newNext = list()
for mappedID in stateMapping[stateID]:
for key1, nextStateID in list(
(key, s.ID_) for key, s in old.states_[mappedID].next_):
add = True
newNext2 = list()
for key2, nextStateIDs in newNext:
# if there is overlap, break transitions into pieces
if keysOverlap(key1, key2):
newNext2.append((keyIntersect(key1, key2),
nextStateIDs + [nextStateID]))
if DEBUG:
print "intersect leads to", (keyIntersect(
key1, key2), nextStateIDs + [nextStateID])
temp = keyMinus(key2, key1)
if len(temp) > 0:
if DEBUG:
print "modified to lead to", (temp,
nextStateIDs)
newNext2.append((temp, nextStateIDs))
key1 = keyMinus(key1, key2)
# otherwise, keep the transition
else:
newNext2.append((key2, nextStateIDs))
if len(key1) > 0:
if DEBUG:
print "added to lead to", (key1, [nextStateID])
newNext2.append((key1, [nextStateID]))
newNext = newNext2
if DEBUG: print "adding states", newNext
# add state mappings
for key, nextStateIDs1 in newNext:
addSet = True
# map to preexising state if there is one
for nextStateID2, nextStateIDs2 in stateMapping.items():
if set(nextStateIDs1) == set(nextStateIDs2):
addSet = False
state.nextIs(key, self.states_[nextStateID2])
break
# otherwise, create one
if addSet:
newState = State(self)
state.nextIs(key, newState)
stateMapping[newState.ID_] = nextStateIDs1
for mappedID in nextStateIDs1:
newState.accept_ = old.states_[mappedID].accept_
if newState.accept_:
break
stateQueue.append(newState.ID_)
def subsetConstruction(self):
# copy old
old = self.copy()
# clear values
self.mergeQueue_ = list()
self.states_ = dict()
self.lastStateID_ = -1
# initialize DFA
self.start_ = State(regex=self, start=True)
self.start_.accept_ = old.start_.accept_
stateMapping = {0 : [0]}
stateQueue = [0]
while len(stateQueue) > 0:
stateID = stateQueue.pop(0)
state = self.states_[stateID]
if DEBUG: print "\nstate", stateID, "=", stateMapping[stateID]
# create distinct transitons
newNext = list()
for mappedID in stateMapping[stateID]:
for key1, nextStateID in list((key, s.ID_) for key,s in old.states_[mappedID].next_):
add = True
newNext2 = list()
for key2, nextStateIDs in newNext:
# if there is overlap, break transitions into pieces
if keysOverlap(key1, key2):
newNext2.append((keyIntersect(key1, key2), nextStateIDs + [nextStateID]))
if DEBUG: print "intersect leads to", (keyIntersect(key1, key2), nextStateIDs + [nextStateID])
temp = keyMinus(key2, key1)
if len(temp) > 0:
if DEBUG: print "modified to lead to", (temp, nextStateIDs)
newNext2.append((temp, nextStateIDs))
key1 = keyMinus(key1, key2)
# otherwise, keep the transition
else:
newNext2.append((key2, nextStateIDs))
if len(key1) > 0:
if DEBUG: print "added to lead to", (key1, [nextStateID])
newNext2.append((key1, [nextStateID]))
newNext = newNext2
if DEBUG: print "adding states", newNext
# add state mappings
for key, nextStateIDs1 in newNext:
addSet = True
# map to preexising state if there is one
for nextStateID2, nextStateIDs2 in stateMapping.items():
if set(nextStateIDs1) == set(nextStateIDs2):
addSet = False
state.nextIs(key, self.states_[nextStateID2])
break
# otherwise, create one
if addSet:
newState = State(self)
state.nextIs(key, newState)
stateMapping[newState.ID_] = nextStateIDs1
for mappedID in nextStateIDs1:
newState.accept_ = old.states_[mappedID].accept_
if newState.accept_:
break
stateQueue.append(newState.ID_)