python(11) : 同步目录

 来自GTP

支持忽略文件及文件夹

注意: clear_ingore_subfolder会删除过滤的文件夹, 若目标文件夹存才

# -*- coding: utf-8 -*-

# ====> 同步文件 <==== #

import os
import shutil
import time
import traceback

# 忽略同步的文件夹, 默认为源文件夹下面的文件夹名称, 源文件夹下面的子文件夹格式如: os.path.join('api', 'doc')
ignore_dirs = ['.git', '__pycache__', '.idea', 'files', os.path.join('api', 'doc')]

# 忽略同步的文件名称
ignore_files = ['app.log']

# 更新计数器
update_count = 0


# 同步文件
def sync_files(source_folder, destination_folder, source_ignore_dirs_relative_path):
    for root, dirs, files in os.walk(source_folder):
        # 处理文件
        for file in files:
            source_file = os.path.join(root, file)
            ignore_state = False
            for pp in source_ignore_dirs_relative_path:
                file_relative_path = os.path.relpath(source_file, pp)
                if not file_relative_path.startswith('..'):
                    ignore_state = True
                    break
            if not ignore_state:
                relative_path = os.path.relpath(source_file, source_folder)
                destination_file = os.path.join(destination_folder, relative_path)
                if not os.path.exists(destination_file) or (
                        os.path.exists(destination_file) and os.stat(source_file).st_mtime > os.stat(
                    destination_file).st_mtime):
                    if file not in ignore_files:
                        increase_counter()
                        shutil.copy(source_file, destination_file)
                        print(f"Copied 【file】 [{source_file}] => [{destination_file}]")

        # 处理子文件夹
        for subdir in dirs:
            source_subdir = os.path.join(root, subdir)
            relative_path = os.path.relpath(source_subdir, source_folder)
            destination_subdir = os.path.join(destination_folder, relative_path)
            if not os.path.exists(destination_subdir):
                ignore_state = False
                for pp in source_ignore_dirs_relative_path:
                    file_relative_path = os.path.relpath(source_subdir, pp)
                    if not file_relative_path.startswith('..'):
                        ignore_state = True
                        break
                if not ignore_state:
                    increase_counter()
                    shutil.copytree(source_subdir, destination_subdir)
                    print(f"Copied 【dir】 [{source_subdir}] => [{destination_subdir}]")
            else:
                sync_files(source_subdir, destination_subdir, source_ignore_dirs_relative_path)


# 清理忽略的子文件夹
def clear_ingore_subfolder(destination_folder):
    destination_ignore_dirs_relative_path = [f"{os.path.join(destination_folder, item)}" for item in ignore_dirs]
    for igp in destination_ignore_dirs_relative_path:
        if os.path.exists(igp) and os.path.isdir(igp):
            shutil.rmtree(igp)
            print(f"Cleard ignore 【dir】 {igp} ")


# 更新计数器(文件夹算1次)
def increase_counter():
    global update_count
    update_count += 1


# 执行
# source_folder: 源文件夹
# destination_folder: 目标文件夹
def run(source_folder, destination_folder):
    print(f'start sync file, source:【{source_folder}】, destination:【{destination_folder}】 ...')
    try:
        st = time.time()
        source_ignore_dirs_relative_path = [f"{os.path.join(source_folder, item)}" for item in ignore_dirs]
        sync_files(source_folder, destination_folder, source_ignore_dirs_relative_path)
        clear_ingore_subfolder(destination_folder)
        print(
            f'√√√ sync file completed, time consuming:[{round(time.time() - st, 2)}]second, updated file size:{update_count} √√√')
    except Exception:
        print(f'××× sync file error ×××')
        traceback.print_exc()


if __name__ == '__main__':
    source_folder = "D:\\synct\\s"
    destination_folder = "D:\\synct\\t"
    run(source_folder, destination_folder)