def comparator(self, a, b):
if len(a) < len(b):
return -1
elif len(a) > len(b):
return 1
else:
return Utils.seqLength(a) - Utils.seqLength(b)
def MSCandidateGenSPM(self, F):
logging.debug('MSCandidateGenSPM: %s', F)
cs = []
for s1 in F:
for s2 in F:
if self.MS[s1[0][0]] == self.getStrictlyMinimumMIS(s1):
if (Utils.removeItem(s1, 1) == Utils.removeItem(s2, Utils.seqLength(s2)-1)) and (self.MS[s2[-1][-1]] >= self.MS[s1[0][0]]): #TODO: need to check why >= here?
nc = self.extendSequence(s1, s2, MSCandidateJoinCriteria.FORWARD)
for c in nc:
cs.append(c)
logging.debug('join: %s %s -> %s %d', s1, s2, c, MSCandidateJoinCriteria.FORWARD)
elif self.MS[s2[-1][-1]] == self.getStrictlyMinimumMIS(s2):
if (Utils.removeItem(s2, Utils.seqLength(s2)-2) == Utils.removeItem(s1, 0)) and (self.MS[s1[0][0]] > self.MS[s2[-1][-1]]):
nc = self.extendSequence(s1, s2, MSCandidateJoinCriteria.REVERSE)
for c in nc:
cs.append(c)
logging.debug('join: %s %s -> %s %d', s1, s2, c, MSCandidateJoinCriteria.REVERSE)
else:
if Utils.removeItem(s1, 0) == Utils.removeItem(s2, Utils.seqLength(s2)-1):
nc = self.extendSequence(s1, s2, MSCandidateJoinCriteria.APRIORI)
for c in nc:
cs.append(c)
logging.debug('join: %s %s -> %s %d', s1, s2, c, MSCandidateJoinCriteria.APRIORI)
return [c for c in cs if self.canPrune(c) is False]
def test_getitem(self):
s = []
self.assertEqual(Utils.getItem(s, 4), None)
s = [[30, 70, 80]]
self.assertEqual(Utils.getItem(s, 0), 30)
s = [[70, 80], [90]]
self.assertEqual(Utils.getItem(s, Utils.seqLength(s)-1), 90)
def extendSequence(self, s1, s2, criteria):
newCandidates = []
if criteria == MSCandidateJoinCriteria.FORWARD:
if len(s2[-1]) == 1:
s1copy = copy.deepcopy(s1)
s1copy.append(s2[-1])
newCandidates.append(s1copy)
if Utils.seqLength(s1) == 2 and len(s1) == 2 and s2[-1][-1] > s1[-1][-1]:
s1copy1 = copy.deepcopy(s1)
s1copy1[-1].append(s2[-1][-1])
newCandidates.append(s1copy1)
elif (Utils.seqLength(s1) == 2 and len(s1) == 1 and s2[-1][-1] > s1[-1][-1]) or Utils.seqLength(s1) > 2:
s1copy = copy.deepcopy(s1)
s1copy[-1].append(s2[-1][-1])
newCandidates.append(s1copy)
elif criteria == MSCandidateJoinCriteria.REVERSE:
if len(s1[0]) == 1:
s2copy = copy.deepcopy(s2)
s2copy.insert(0, s1[0])
newCandidates.append(s2copy)
if Utils.seqLength(s2) == 2 and len(s2) == 2 and s1[0][0] > s2[0][0]:
s2copy1 = copy.deepcopy(s2)
s2copy1[0].insert(0, s1[0][0])
newCandidates.append(s2copy1)
elif (Utils.seqLength(s2) == 2 and len(s2) == 1 and s1[0][0] > s2[0][0]) or Utils.seqLength(s2) > 2:
s2copy1 = copy.deepcopy(s2)
s2copy1[0].insert(0, s1[0][0])
newCandidates.append(s2copy1)
elif criteria == MSCandidateJoinCriteria.APRIORI:
if len(s2[-1]) == 1:
s1copy = copy.deepcopy(s1)
s1copy.append(s2[-1])
newCandidates.append(s1copy)
else:
s1copy = copy.deepcopy(s1)
s1copy[-1].append(s2[-1][-1])
newCandidates.append(s1copy)
return newCandidates
def test_removeitem(self):
s = []
ts = s
self.assertEqual(Utils.removeItem(s, 0), [])
self.assertEqual(s, ts)
s = [[30, 70, 80]]
ts = s
self.assertEqual(Utils.removeItem(s, 0), [[70, 80]])
self.assertEqual(s, ts)
s = [[70, 80], [90]]
ts = s
self.assertEqual(Utils.removeItem(s, Utils.seqLength(s)-1), [[70, 80]])
self.assertEqual(s, ts)
def canPrune(self, seq):
sLowestMIS = self.getStrictlyMinimumMIS(seq)
k = Utils.seqLength(seq)
for i in range(k):
item = Utils.getItem(seq, i)
if self.MS[item] == sLowestMIS:
continue
k_1_subseq = Utils.removeItem(seq, i)
count = 0
for d in self.T:
if Utils.isSubsequence(k_1_subseq, d):
count += 1
if float(count) / len(self.T) < self.getMinMIS(k_1_subseq):
return True
return False
def test_seqlength(self):
s = []
self.assertEqual(Utils.seqLength(s), 0)
s = [[1]]
self.assertEqual(Utils.seqLength(s), 1)
s = [[2, 3]]
self.assertEqual(Utils.seqLength(s), 2)
s = [[1], [2, 3]]
self.assertEqual(Utils.seqLength(s), 3)
s = [[1], [2, 3], [4]]
self.assertEqual(Utils.seqLength(s), 4)
s = [[1], [1, 1], [1]]
self.assertEqual(Utils.seqLength(s), 4)
data_file = sys.argv[2]
para_file = sys.argv[3]
result_file = sys.argv[4]
DP = DataProcessor(data_file, para_file, result_file, False)
inputData = DP.loadInput()
startTime = datetime.now()
print 'Execution started at:', startTime
algo = pyMSGSP(inputData["T"], inputData["MS"], inputData["SDC"], logging.INFO)
# algo = BruteForceSPM(inputData["T"], inputData["MS"], inputData["SDC"])
outputData = algo.run()
print 'Execution time:', datetime.now() - startTime
outputDict = defaultdict(list)
for seq in outputData:
count = 0
for d in inputData["T"]:
if Utils.isSubsequence(seq, d):
count += 1
outputDict[Utils.seqLength(seq)].append((seq, count))
DP.printOutput(outputDict)
