import pandas as pd

# 将meta文件转换为模型可以直接使用的label文件
meta_csv_path = "cicdos2017/dataset/meta.csv"
label_csv_path = "cicdos2017/label/label.csv"

df = pd.read_csv(meta_csv_path)
mixed_row_num =0
high_row_num = 0
low_row_num = 0
normal_row_num =0

label_ls =[]
for index, row in df.iterrows():
    if index % 20000 == 0:
        print("processing index:",index)
    high_num = int(row[0])
    low_num = int(row[1])
    normal_num = int(row[2])

    features = row[3:43]
    
    label = "normal"
    if low_num==0 and normal_num==0:
        label = "high"
        high_row_num +=1
    elif high_num==0 and normal_num==0:
        label = "low"
        low_row_num+=1
    elif high_num==0 and low_num==0:
        label = "normal"
        normal_row_num+=1
    else:
        mixed_row_num+=1
        #print("high,low,normal:",high_num,low_num,normal_num)
        continue #混合的row直接丢弃
    label_row =[label]
    label_row.extend(features)
    label_ls.append(label_row)

print("mix row num:",mixed_row_num)
print("high row num:",high_row_num)
print("low row num:",low_row_num)
print("normal row num:",normal_row_num)


# write to csv file
data = pd.DataFrame(data=label_ls)
# print(total_data)
data.to_csv(label_csv_path, index=False,
                    encoding="utf-8", sep=',', mode='w', header=True)