def getTestApiSequence(file="Data/SourceData/test.csv"):
print(file)
data_iter = loaddata(file)
lines = next(data_iter)
data = {}
for lines in data_iter:
# print(lines[0])
for line in lines:
user_id = line[0]
api_name = line[1]
# return_value = line[-2]
order_index = int(line[-1])
tid = line[3]
d = list((tid, order_index, api_name))
if user_id in data:
data[user_id].append(d)
else:
data[user_id] = [d]
# break
new_data = {}
for user_id in data:
user_data = data[user_id]
if not user_data:
continue
user_data = sorted(user_data, key=lambda s: s[1])
user_data = sorted(user_data, key=lambda s: s[0])
api_features = [apiMap[line[2]]5 for line in user_data]
def devidefile(input=input_file, decide_size=60, output_dir="Data/SourceData/"+file+"/"):
data_iter = loaddata(input)
files_fps = [open(output_dir+("%05d"%i)+".txt", "w") for i in range(decide_size)]
for lines in data_iter:
for line in lines:
value = int(int(line[0])%decide_size)
write_line = ",".join(line)+"\n"
files_fps[value].write(write_line)
def getLabel(input_dir="Data/SourceData", output_dir="Data/static", train=True):
input_file = input_dir + "/train.csv"
output_file = output_dir + "/label.txt"
if not train:
return None
data_iter = loaddata(input_file)
data_lab = {}
with open(output_file, "w") as fp:
for lines in data_iter:
for line in lines:
if line[0] not in data_lab:
data_lab[line[0]] = line[1]
fp.write(line[0] + "\t" + line[1] + "\n")
return data_lab
def getStaticfeature(input_dir="Data/SourceData", output_dir=output_path_dir, train=True):
file = input_dir + "/train.csv"
if not train:
file = input_dir + "/test.csv"
data_iter = loaddata(file)
api_static = None
return_static = {}
api_index = 0
return_index = 0
api_map = {}
return_map = {}
filename = output_dir + "/static_feature.txt"
if not train:
filename = output_dir + "/test_static_feature.txt"
fp = open(filename, "w")
for lines in data_iter:
for line in lines:
api_feature = str(line[0] + "_api_" + line[2])
if api_static and api_feature in api_static:
api_static[api_feature] += 1
elif api_static and api_feature not in api_static:
for key in api_static:
fp.write(key + "\t" + str(api_static[key]) + "\n")
api_static = {api_feature: 1}
else:
api_static = {api_feature: 1}
apiName = str(line[2])
if apiName not in api_map:
api_index += 1
api_map[apiName] = api_index
return_feature = str(line[0] + "_return_" + line[-2])
if return_static and return_feature in return_static:
return_static[return_feature] += 1
elif return_static and return_feature not in return_static:
for key in return_static:
fp.write(key + "\t" + str(return_static[key]) + "\n")
return_static = {return_feature: 1}
else:
return_static = {return_feature: 1}
returnValue = str(line[-1])
if returnValue not in return_map:
return_index += 1
return_map[returnValue] = return_index
fp.close()
return api_map, return_map
def getApipv(file="Data/SourceData/train.csv"):
data_iter = loaddata(file)
# lines = next(data_iter)
api_static_result = {}
return_result_static = {}
api_id = {}
return_id = {}
api_index = 1
return_index = 1
for lines in data_iter:
print(lines[0])
for line in lines:
feature_name = str(line[0] + "_" + line[2])
if feature_name in api_static_result:
api_static_result[feature_name] += 1
else:
api_static_result[feature_name] = 1
api_name = str(line[2])
if api_name not in api_id:
api_index = api_index + 1
api_id[api_name] = api_index
return_result = str(line[-2])
if return_result not in return_id:
return_id[return_result] = return_index
return_index += 1
feature_name_2 = str(line[0] + "_" + line[-2])
if feature_name_2 in return_result_static:
return_result_static[feature_name_2] += 1
else:
return_result_static[feature_name_2] = 1
# lines = next(data_iter)
# break
return api_static_result, return_result_static, api_id, return_id