import os
from Sample_cleaning import de_PE
from Sample_cleaning import de_oversize
import shutil
from Sample_cleaning import de_repeatSample
from Sample_cleaning import de_shell
from Sample_cleaning import de_cfg
if __name__ == "__main__":
    # 获取清洗样本的存放目录
    # shutil.copyfile(r"C:\Users\Administrator\Desktop\大规模复杂软件无效样本清洗\demo1\T\ccs2023a-paper48.pdf",
    #                 r"C:\Users\Administrator\Desktop\大规模复杂软件无效样本清洗\demo1\save_dir")
    sampleFile_dir = os.path.join(os.getcwd(), "Sample")
    # out_dir=os.path.join(os.getcwd(), "save_dir")
    # print(sampleFile_dir)
    # print(out_dir)
    # 1.清洗非可执行程序
    # notexe_save_dir=os.path.join(os.getcwd(),"notexe_dir")
    # test = de_PE.deleteNotPE(sampleFile_dir=sampleFile_dir,save_dir=notexe_save_dir)
    # test.notPEfile_Filter()
    # 2.清洗加壳程序
    # shellsave_dir=os.path.join(os.getcwd(),"shell_dir")
    # test2=de_shell.de_shell(sample_path=sampleFile_dir,save_dir=shellsave_dir)
    # test2.fileFilter()
    # 3.清洗不可表征cfg样本
    notcfg_dir=os.path.join(os.getcwd(),"notcfg_dir")
    test4=de_cfg.de_notcfg(sampleFile_dir=sampleFile_dir,save_dir=notcfg_dir)
    test4.notcfgfile_Filter()
    # 4.清洗过大样本文件
    # oversizefile_save_dir=os.path.join(os.getcwd(),"oversizefile_dir")
    # test4=de_oversize.de_oversizefile(sampleFile_dir=sampleFile_dir,max_file_size=10,min_file_size=0.01,save_dir=oversizefile_save_dir)
    # test4.filesizeFilter()
    # 5.删除重复的文件
    # repeatfilesave_dir=os.path.join(os.getcwd(),"repeatfile_dir")
    # test5=de_repeatSample.de_repeatfile(sampleFile_dir=sampleFile_dir,save_dir=repeatfilesave_dir)
    # test5.remove_duplicate_files()