help="Using test data.") parser.add_argument("-l", "--headless", action="store_true", dest="headless", help="Cast headless mode.") args = parser.parse_args() path = args.path THREADS_NUM = args.thread_num RETRIES = args.retries TIMEOUT = args.timeout RENDER_TIME = args.render_time is_debug = args.debug is_test = args.test is_headless = args.headless RUN_PATH = sys.path[0] LOG, RQ = Logger('CrawlerLog', is_debug).getlog() sch = Scheduler(path) try: sch.scheduling() except KeyboardInterrupt: LOG.info("Quit.") finally: LOG.info('Done.') for w, _, _ in sch.d.workers_info: w.exit() # TODO 非阻塞退出 ex_cmd('TASKKILL /IM chromedriver.exe /F', None, True)
from main import ex_cmd if __name__ == '__main__': ex_cmd("rm -rf */datas/*") ex_cmd("rm -rf */Logs/*") ex_cmd("rm -rf */logs/*") ex_cmd("rm -rf */tmp/*") ex_cmd("rm -rf */pics/*") ex_cmd("rm -rf 4/noiseFigure/*") ex_cmd("rm -rf 4/rawFigure/*") ex_cmd("rm -rf 4/result_figure/*")
sbin_path = getcwd() chdir(abslt_root_path) tgz_list = glob("*.tgz") for tgz in tgz_list: decrypt(tgz) db_tgz_list = glob("_*.tgz") for _tgz in db_tgz_list: # _分割 num = _tgz[1:].split("_")[0] _tgz_name = splitext(_tgz)[0] if not exists(_tgz_name): mkdir(_tgz_name) cmd = "tar -zxvf {} -C {}".format(_tgz, _tgz_name) print(cmd) # 需要等待子线程结束,要不然下面重命名会找不到文件直接主进程关掉了 ex_cmd(cmd) # 重命名 dot_files_path = join(_tgz_name, "_queryResult_db_") chdir(dot_files_path) print("change dir to {}".format(dot_files_path)) ex_cmd("ls result*. | sh -c \"xargs -n1 -i mv {} {}tar\"") # Popen("ls |sh -c \"xargs -n1 -i sed -i '1d' {}\"", shell=True).communicate() ex_cmd("ls result*.tar | sh -c \"xargs -n1 -i tar -zxvf {}\"") ex_cmd("ls _queryResult_db_/result*.txt | sh -c \"xargs -n1 -i cat {} >>../../" + num + ".txt\"") print("change dir to {}".format(abslt_root_path)) chdir(abslt_root_path)
from main import ex_cmd from os import mkdir, makedirs from os.path import exists, join import time def safe_mkdir(path: str) -> None: if not exists(path): mkdir(path) def safe_makedirs(path: str) -> None: if not exists(path): makedirs(path) if __name__ == '__main__': safe_mkdir("backup") date = time.strftime('%Y%m%d%H%M', time.localtime(time.time())) root = join("backup", date) safe_makedirs(root) for path in (join(root, "1", "datas"), join(root, "2", "datas"), join(root, "3", "datas")): safe_makedirs(path) ex_cmd("cp -r 2/datas/*.txt backup/{}/2/datas".format(date)) ex_cmd("cp -r 2/datas/[!_]*_ret.csv backup/{}/2/datas".format(date)) ex_cmd("cp -r 3/datas/* backup/{}/3/datas".format(date))