def ochiaiList(dirPosition):
answerList = calculatorPublic.getAnswer(dirPosition)
gcovList = calculatorPublic.getGcov(dirPosition, 1).T
ef = calculatorPublic.getEF(gcovList, answerList)
ep = calculatorPublic.getEP(gcovList, answerList)
nf = calculatorPublic.getNF(gcovList, answerList)
return ef / numpy.sqrt((ef + nf) * (ef + ep))
def jaccardList(dirPosition):
answerList = calculatorPublic.getAnswer(dirPosition)
gcovList = calculatorPublic.getGcov(dirPosition, 1).T
ef = calculatorPublic.getEF(gcovList, answerList)
ep = calculatorPublic.getEP(gcovList, answerList)
nf = calculatorPublic.getNF(gcovList, answerList)
return 1 / (1 + (ep + nf) / ef)
def tarantulaList(dirPosition):
answerList = calculatorPublic.getAnswer(dirPosition)
gcovList = calculatorPublic.getGcov(dirPosition, 1).T
ef = calculatorPublic.getEF(gcovList, answerList)
ep = calculatorPublic.getEP(gcovList, answerList)
tf = calculatorPublic.getTF(answerList)
tp = calculatorPublic.getTP(answerList, tf)
return 1 / (1 + (numpy.dot(ep, tf)) / (numpy.dot(ef, tp)))
def networks(dirPosition, times):
answerList = calculatorPublic.getAnswer(dirPosition)
gcovList = calculatorPublic.getGcov(dirPosition, times)
training(dirPosition, answerList, gcovList)
model = tf.keras.models.load_model(dirPosition +
'/numpyDataDir/myModel.h5')
testList = numpy.identity(len(gcovList[0]))
return model.predict(testList)
def getTrueNumberList(dirPosition):
# 所有测试结果的簇
answerList = calculatorPublic.getAnswer(dirPosition)
# 存放成功轨迹的簇
trueNumberList = []
# 获取成功轨迹的簇
for i in range(answerList.__len__()):
if answerList[i] == 1:
trueNumberList.append(i)
return trueNumberList
def FLSFList(dirPosition):
answerList = calculatorPublic.getAnswer(dirPosition)
gcovList = calculatorPublic.getGcov(dirPosition, 2).T
fs = calculatorPublic.getFS(gcovList, answerList)
ps = calculatorPublic.getPS(gcovList, answerList)
tf = calculatorPublic.getTF(answerList)
tp = calculatorPublic.getTP(answerList, tf)
res = []
r1 = []
r2 = []
for i in range(0, len(fs)):
temp1 = 0
temp2 = 0
for j in range(0, len(fs[0])):
try:
flag1 = math.exp(fs[i][j] * 0.6)
except OverflowError:
flag1 = float('inf')
try:
flag2 = math.exp(-fs[i][j] * 0.6)
except OverflowError:
flag2 = 0
temp1 = temp1 + (flag1 - flag2) / (flag1 + flag2)
r1.append(temp1)
for k in range(0, len(ps[0])):
try:
flag1 = math.exp(ps[i][k] * 0.6)
except OverflowError:
flag1 = float('inf')
try:
flag2 = math.exp(-ps[i][k] * 0.6)
except OverflowError:
flag2 = 0
temp2 = temp2 + (flag1 - flag2) / (flag1 + flag2)
r2.append(temp2)
temp4 = r2[i] * tf
temp5 = r1[i] * tp
if temp5 == 0:
if temp4 == 0:
res = numpy.append(res, 1)
else:
res = numpy.append(res, 0)
else:
temp3 = 1 / (1 + temp4 / temp5)
res = numpy.append(res, temp3)
return res
def dbscan(dirPosition):
# tarantula算法得到结果
initList = tarantula.tarantulaList(dirPosition)
# 前２０％怀疑度矩阵
rankList = calculatorPublic.getResultList(initList)
# 将rankList矩阵装换为list类型方便之后计算等操作
rank = getRank(rankList)
# 程序是否通过的列表
answerList = calculatorPublic.getAnswer(dirPosition)
# 程序运行轨迹gcov矩阵
gcovList = calculatorPublic.getGcov(dirPosition, 1)
# 记录未通过测试用例编号(真实编号－１)
unPassList = []
# 未通过测试用例是否执行被怀疑语句的记录矩阵
initMap = []
# 获取记录矩阵（initMap）
for i in range(answerList.__len__()):
if answerList[i] == 0:
temp = []
unPassList.append(i)
for item in rank:
if gcovList[i][item] == 1:
temp.append(item)
if temp.__len__() != 0:
initMap.append(temp)
# 获取参数１：每个错误测试用例所包含的怀疑度行与怀疑度行总数的百分比
parameter_x = getParameter_x(initMap, rank)
# 获取参数２：每个错误测试用例的轨迹相对于第一个测试用例轨迹的变化值
parameter_y = getParameter_y(gcovList, unPassList)
# 对参数１和参数２进行整合形成一个对应的坐标矩阵
parameterMap = getParameterMap(parameter_x, parameter_y)
# 对得到的坐标矩阵进行DBSCAN聚类操作得到分类列表
clusteringList = getClusteringList(parameterMap)
# 对得到的分簇list进行分解
dbscanTestNumber = getDbscanTestNumber(clusteringList, unPassList)
print("dbscan:" + str(dbscanTestNumber.__len__()))
# 对得到的结果进行存放
numpy.save(dirPosition + "/numpyDataDir/dbscanTestNumber.npy",
dbscanTestNumber)
def tarantula(dirPosition, testsList):
# 载入源数据中的测试结果
answerList = calculatorPublic.getAnswer(dirPosition)
# 载入源数据中的gcov图谱集
gcovList = calculatorPublic.getGcov(dirPosition, 1)
# 存放本次计算需要的测试结果集
answer = []
# 存放本次计算需要的gcov图谱集
gcov = []
# 获取对应于testsList中所包含的测试用例的测试结果
for test in testsList:
answer.append(answerList[test])
gcov.append(gcovList[test])
# 将answer和gcov两个list都转换为numpy矩阵类型
answer = numpy.array(answer)
gcov = numpy.array(gcov)
# 对gcov进行翻转
gcov = gcov.T
ef = calculatorPublic.getEF(gcov, answer)
ep = calculatorPublic.getEP(gcov, answer)
tf = calculatorPublic.getTF(answer)
tp = calculatorPublic.getTP(answer, tf)
return 1 / (1 + (numpy.dot(ep, tf)) / (numpy.dot(ef, tp)))
def init(dirPosition):
# tarantula算法得到结果
initList = tarantula.tarantulaList(dirPosition)
# 前２０％怀疑度矩阵
rankList = calculatorPublic.getResultList(initList)
# 程序是否通过的列表
answerList = calculatorPublic.getAnswer(dirPosition)
# 程序运行轨迹gcov矩阵
gcovList = calculatorPublic.getGcov(dirPosition, 1)
# 记录未通过测试用例编号(真实编号－１)
unPassList = []
# 未通过测试用例是否执行被怀疑语句的记录矩阵
initMap = []
# 获取记录矩阵（initMap）
for i in range(answerList.__len__()):
if answerList[i] == 0:
temp = []
unPassList.append(i)
for j in range(rankList.__len__()):
for k in range(rankList[j].__len__()):
if gcovList[i][rankList[j][k]] == 1:
temp.append(rankList[j][k])
if temp.__len__() != 0:
initMap.append(temp)
# 相似度矩阵
similarMap = []
# 获取相似度矩阵
for i in range(initMap.__len__()):
temp = []
for j in range(initMap.__len__()):
if i == j:
temp.append(0)
else:
# 获取交集
len1 = list(set(initMap[i]).intersection(set(initMap[j]))).__len__()
# 获取并集
len2 = list(set(initMap[i]).union(set(initMap[j]))).__len__()
temp.append(len1 / len2)
similarMap.append(temp)
# 存放簇
cuMap = [[0]]
# 获取簇(cuMap)
for i in range(similarMap.__len__()):
position = check(cuMap, i)
if position == -1:
maxnum = 0
site = 0
for j in range(cuMap.__len__()):
for k in range(cuMap[j].__len__()):
if similarMap[cuMap[j][k]][i] > maxnum:
maxnum = similarMap[cuMap[j][k]][i]
site = j
if maxnum >= 0.7:
cuMap[site].append(i)
else:
cuMap.append([i])
# 存放簇中对应的测试用例编号（实际编号－１的值）
testNumberMap = []
# 将簇中对应编号进行转换
for i in range(cuMap.__len__()):
temp = []
for j in range(cuMap[i].__len__()):
temp.append(unPassList[cuMap[i][j]])
testNumberMap.append(temp)
print("clustering:" + str(testNumberMap.__len__()))
# 将分簇后的测试用例编号矩阵（testNumberMap）进行物理存储
numpy.save(dirPosition + "/numpyDataDir/testNumberMap.npy", testNumberMap)
