| 33 | |
| 34 | |
| 35 | class divideDataSet: |
| 36 | # EXTRACT_RATIO = 0.2 每个家族抽取样本的比例 ,0.2表示按照8:2划分训练集和验证集 |
| 37 | |
| 38 | def __init__(self, origin_path, goal_path, EXTRACT_RATIO=0.2): |
| 39 | self.origin_path = origin_path |
| 40 | self.goal_path = goal_path |
| 41 | self.EXTRACT_RATIO = EXTRACT_RATIO |
| 42 | |
| 43 | def checkFileisExist(self): |
| 44 | try: |
| 45 | createBlankFile(self.origin_path, self.goal_path).createNewFiles() |
| 46 | Flag = True |
| 47 | except: |
| 48 | print('something wrong in checking File is exist') |
| 49 | Flag = False |
| 50 | return Flag |
| 51 | |
| 52 | def copyFile(self): |
| 53 | list = readFilenameList(self.origin_path) |
| 54 | # print(list) |
| 55 | flag = self.checkFileisExist() |
| 56 | if flag == False: return 'moveFile Failed' |
| 57 | for malFamily in list: # 对train文件夹下的每个子文件夹，随机取出EXTRACT_RATIO比例的文件放入validation文件夹中 |
| 58 | file_abs_path = self.origin_path + "\\" + malFamily |
| 59 | pathDir = readFilenameList(file_abs_path) |
| 60 | # filenumber = len(pathDir) |
| 61 | # picknumber = int(filenumber * self.EXTRACT_RATIO) |
| 62 | picknumber = dict_num[malFamily] |
| 63 | sample = random.sample(pathDir, picknumber) |
| 64 | # print(sample) |
| 65 | for name in sample: |
| 66 | # shutil.move(self.origin_path + '\\' + malFamily + '\\' + name, |
| 67 | # self.goal_path + '\\' + malFamily + '\\' + name) |
| 68 | shutil.copy(self.origin_path + '\\' + malFamily + '\\' + name, |
| 69 | self.goal_path + '\\' + malFamily + '\\' + name) |
| 70 | return 'moveFile Succeed!' |
| 71 | |
| 72 | if __name__ == '__main__': |
| 73 | divideDataSet(origin_path=r'K:\数据库\ISCX-IDS-2012\2_extractFlowFeature',goal_path=r'K:\数据库\ISCX-IDS-2012\3_sampleDataset').copyFile() |