py-dir-watcher/clone.py

#!/usr/bin/env python3

import os
import logging
import string
import hashlib
import time


class Cloner:
    def __init__(self, targetDir: str, backupDir: str):
        self.targetDir = targetDir
        self.backupDir = backupDir
        self.interval = 1

    def handleDir(self, dirPath: str):
        backupDirPath = dirPath.replace(self.targetDir, self.backupDir)
        if not os.path.isdir(backupDirPath):
            os.system('mkdir {path}'.format(path=backupDirPath))
        with os.scandir(dirPath) as it:
            for entry in it:
                if entry.is_dir():
                    self.handleDir(entry.path)
                elif entry.is_file():
                    self.handleFile(entry.path)

    def handleFile(self, fileName: str):
        cloneFileName = fileName.replace(self.targetDir, self.backupDir)

        if os.path.exists(cloneFileName):
            with open(cloneFileName + '.md5', 'r') as cloneHashFile, open(fileName, 'rb') as targetFile:
                cloneHashFileContent = cloneHashFile.read()
                targetFileContent = targetFile.read()
                realHash = hashlib.md5(targetFileContent).hexdigest()
                if realHash != cloneHashFileContent:
                    os.system('cp {backup} {real}'.format(
                        backup=cloneFileName, real=fileName))
                    logging.warning(
                        '{file} failed check, restoring...'.format(file=fileName))
                else:
                    logging.info('{file} is fine'.format(file=fileName))
        else:
            logging.warning(
                '{file} is a new file, added to backup dir'.format(file=fileName))
            os.system('cp {real} {clone}'.format(
                real=fileName, clone=cloneFileName))
            with open(cloneFileName + '.md5', 'w') as cloneHashFile, open(fileName, 'rb') as targetFile:
                targetFileContent = targetFile.read()
                cloneHashFile.write(hashlib.md5(targetFileContent).hexdigest())

    def setLog(self, logFileName: str):
        logFormatter = logging.Formatter(
            fmt="[%(asctime)s][%(levelname)s] %(message)s",
            datefmt='%d-%b-%y %H:%M:%S')

        rootLogger = logging.getLogger()

        fileHandler = logging.FileHandler(logFileName, mode='a')
        fileHandler.setFormatter(logFormatter)
        rootLogger.addHandler(fileHandler)

        # # uncomment this to enable console logging
        # consoleHandler = logging.StreamHandler()
        # consoleHandler.setFormatter(logFormatter)
        # rootLogger.addHandler(consoleHandler)

        rootLogger.setLevel(logging.WARNING)

    def setInterval(self, interval: int):
        self.interval = interval

    def start(self):
        while True:
            self.handleDir(self.targetDir)
            time.sleep(5)


if __name__ == "__main__":
    import argparse

    parser = argparse.ArgumentParser()
    parser.add_argument(
        '--target',
        metavar='PATH',
        type=str,
        required=True,
        help='directory to clone and monitor')
    parser.add_argument(
        '--backup',
        metavar='PATH',
        type=str,
        required=True,
        help='backup directory to store clone'
    )
    parser.add_argument(
        '--interval',
        metavar='SECOND',
        type=int,
        default=1,
        help='interval in second(s) which the scripts run, default is 1'
    )
    parser.add_argument(
        '--log',
        metavar='LOGFILE',
        type=str,
        default='/var/log/py-dir-watcher/log.txt',
        help='specify custom logfile target, default is \'/var/log/py-dir-watcher/log.txt\' '
    )

    args = parser.parse_args()

    app = Cloner(args.target, args.backup)
    app.setLog(args.log)
    app.setInterval(args.interval)
    app.start()