def test_test_thread(path: str,
thread_num: int,
is_save_intermediate_file: bool = False) -> None:
# 有bug，废弃
Log(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()))
Log("开始测试", path, "的数据")
exe = ExecuteCode(is_save_intermediate_file=is_save_intermediate_file)
exe_res, compile_succ = [], 0
for root, dirs, files in os.walk(path):
exe_res, compile_succ = exe.execute_code_thread(
PathFunc.to_linux(path), files, thread_num)
break
Log("编译成功数量：", compile_succ, "/", len(files))
with open("build/execute_info", "wb") as f:
pickle.dump(exe_res, f)
g = GenerateAns.GenerateAns("ans.csvc", "sample_submission.csv")
while True:
t = g.get()
if t is None:
break
g.add(1 if compare_test(exe_res[t[0]], exe_res[t[1]]) else 0)
Log(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()))
def test_test(path: str, is_save_intermediate_file: bool = False) -> None:
Log("开始测试", path, "的数据")
exe_res, compile_succ = dict(), 0
exe = ExecuteCode(is_save_intermediate_file=is_save_intermediate_file)
for root, dirs, files in os.walk(path):
total_num, cur_num = len(files), 1
for name in files:
Log("开始编译", name, "：[", str(cur_num), "/", str(total_num), "]")
cur_num += 1
p = PathFunc.join(PathFunc.to_linux(path), name)
exe_res[name.split(".")[0]], cs = exe.execute_code(p)
compile_succ += cs
break
Log("编译成功数量：", compile_succ, "/", len(files))
with open("../build/execute_info", "wb") as f:
pickle.dump(exe_res, f)
with open("../build/execute_info", "rb") as f:
exe_res = pickle.load(f)
g = GenerateAns.GenerateAns("../ans.csv", "../sample_submission.csv")
while True:
t = g.get()
if t is None:
break
g.add(1 if compare_test(exe_res[t[0]], exe_res[t[1]]) else 0)
def test_train_in_one_dir(path: str,
test_num: int = -1,
is_save_intermediate_file=False) -> None:
exe_res = []
exe = ExecuteCode(is_save_intermediate_file=is_save_intermediate_file)
Log("开始测试", path, "的数据")
count = 0
for root, dirs, files in os.walk(path):
for name in files:
if count == test_num:
break
count += 1
Log("开始编译", name)
p = PathFunc.join(PathFunc.to_linux(path), name)
exe_res.append({"res": exe.execute_code(p), "name": name})
break
Log("开始比较输出")
length = len(exe_res)
for i in range(length):
for j in range(i + 1, length):
if not compare_test(exe_res[i]["res"], exe_res[j]["res"]):
l1, l2 = exe_res[i], exe_res[j]
print("文件", exe_res[i]["name"], "和", exe_res[j]["name"],
"结果不一致")
continue
def test_train_random_from_same_dir(dir_path: str,
test_num: int,
is_save_intermediate_file: bool = False):
dir_list = os.listdir(dir_path)
dir_list_len = len(dir_list)
dir_file = []
for name in dir_list:
dir_file_list = os.listdir(os.path.join(dir_path, name))
dir_file.append([name, len(dir_file_list), dir_file_list])
exe = ExecuteCode(is_save_intermediate_file=is_save_intermediate_file)
is_same_num, succ_num = 0, 0
for _ in range(test_num):
i = random.randint(0, dir_list_len - 1)
j1 = random.randint(0, dir_file[i][1] - 1)
j2 = random.randint(0, dir_file[i][1] - 1)
Log("开始比较 ", dir_file[i][0], " 的 ", dir_file[i][2][j1], " 和 ",
dir_file[i][2][j2])
p1 = PathFunc.join(dir_path, dir_file[i][0], dir_file[i][2][j1])
p2 = PathFunc.join(dir_path, dir_file[i][0], dir_file[i][2][j2])
ret_list1, _ = exe.execute_code(p1)
ret_list2, _ = exe.execute_code(p2)
if not compare_test(ret_list1, ret_list2):
GLog("文件不相同")
for r1, r2 in zip(ret_list1, ret_list2):
GLog(r1, r2)
if sum([1 if x.result == 0 else 0 for x in ret_list1]) != len(ret_list1) \
or sum([1 if x.result == 0 else 0 for x in ret_list2]) != len(ret_list2):
print("diff")
else:
succ_num += 1
else:
is_same_num += 1
succ_num += 1
return is_same_num, succ_num
def act(mask: int, file_list: list) -> None:
exe_res, total_num, cur_num = dict(), len(file_list), 1
for name in file_list:
Log(str(mask), "开始编译", name, "：[", str(cur_num), "/", str(total_num), "]")
cur_num += 1
p = PathFunc.join(dir_path, name)
exe_res[name.split(".")[0]], cs = self.execute_code(p, str(mask))
exe_res_list[mask] = exe_res
compile_succ_list[mask] = compile_succ_list.get(mask, 0) + cs
def get_trained_xgb(train_path: str, save_path: str):
Log("开始训练模型")
for i in range(83):
dtrain = xgb.DMatrix(
os.path.join(train_path,
str(i) + ".csv") + "?format=csv&label_column=0")
param = {
'max_depth': 16,
'nthread': 4,
'gamma': 0.00001,
'objective': 'binary:logistic'
}
bst = xgb.train(param, dtrain, 16)
bst.save_model(os.path.join(save_path, str(i) + ".model"))
train_preds = bst.predict(dtrain)
train_predictions = [round(value) for value in train_preds]
y_train = dtrain.get_label() # 值为输入数据的第一行
train_accuracy = f1_score(y_train, train_predictions)
Log("Train f1: %.2f%%" % (train_accuracy * 100.0))
def get_trained_xgb(train_path: str, save_path: str, prec: float):
get_what = []
for i in range(83):
for j in range(int(500 * prec)):
get_what.append(i * 500 + j)
Log("开始训练模型")
for i in range(83):
csv = pd.read_csv(os.path.join(train_path, str(i) + ".csv"))
train = csv.iloc[get_what, 1:].values
labels = csv.iloc[get_what, :1].values
dtrain = xgb.DMatrix(train, label=labels)
param = {'max_depth': 16, 'nthread': 4, 'gamma': 0.00001, 'objective': 'binary:logistic'}
bst = xgb.train(param, dtrain, 16)
bst.save_model(os.path.join(save_path, str(i) + "_report.model"))
train_preds = bst.predict(dtrain)
train_predictions = [round(value) for value in train_preds]
y_train = dtrain.get_label() # 值为输入数据的第一行
train_accuracy = f1_score(y_train, train_predictions)
Log("Train f1: %.2f%%" % (train_accuracy * 100.0))