Python Socket-Based File Transfer System with Separate Command and Data Channels

During recent Python studies the author explored socket programming and created a simple file‑transfer utility for Linux servers that natively support Python.

The project, hosted at GitHub , demonstrates sending files from a local host to a remote server.

Principle : The system relies on standard TCP sockets. While transferring a single file is straightforward, handling multiple files or entire directories requires a more sophisticated approach. The author introduced two separate socket channels—one for data and another for control commands—by creating distinct command thread and data thread components.

To transfer a whole folder, the program first traverses the directory structure, then sends each file sequentially over the data socket. A custom FileFinder class (blocking type) is used to enumerate files and directories, allowing the caller to pause the traversal until the previous file transfer completes.

Key source code (FileFinder):

import os,time
# 文件、文件夹寻找类 (阻塞型)
# 阻塞的设计：为了等待调用者的耗时操作【否则很快就完成了文件的遍历任务，调用者达不到顺序操作文件(夹)的意图】
class FileFinder:
   def __init__(self, finderCallback):
       self.finderCallback = finderCallback
       # 文件（夹）路径下所有文件的总大小
       self.sum_size = 0
       # 调用者控制的参数，若为False,则遍历工作继续进行，若为True，则阻塞任务，等待调用者完成它的其他耗时操作后在考虑是否改变此值
       self.recycle = True
       # 调用者控制的参数，若为False,则正常工作，若为True，则当recycle为False时遍历工作不阻塞快速完成，recycle为True时遍历工作阻塞
       self.off = False

   # 文件（夹）找到时的回调类
   class FinderCallback:
       # 找到文件夹
       def onFindDir(self, dir_path):
           pass
       # 找到文件
       def onFindFile(self, file_path, size):
           pass
       # 预留的刷新函数
       def onRefresh(self):
           pass

   # 查找文件（夹）方法
   def list_flie(self, root_dir):
       if os.path.isfile(root_dir):
           while self.recycle:
               time.sleep(0.05)
           if self.finderCallback:
               self.finderCallback.onFindFile(root_dir, os.path.getsize(root_dir))
               self.finderCallback.onRefresh()
               if not self.off:
                   self.recycle = True
       else:
           dirlist = os.listdir(root_dir)  # 列出文件夹下所有的目录与文件
           for dir in dirlist:
               path = os.path.join(root_dir, dir)
               if os.path.isfile(path):
                   while self.recycle:
                       time.sleep(0.05)
                   if self.finderCallback:
                       self.finderCallback.onFindFile(path, os.path.getsize(path))
                       self.finderCallback.onRefresh()
                       if not self.off:
                           self.recycle = True
               else:
                   while self.recycle:
                       time.sleep(0.05)
                   if self.finderCallback:
                       self.finderCallback.onFindDir(path)
                       self.finderCallback.onRefresh()
                       if not self.off:
                           self.recycle = True
                   # 递归调用（当遍历到文件夹时，继续遍历，直到当前文件夹下没有文件夹为止）
                   self.list_flie(path)

By inheriting FinderCallback and overriding onFindDir and onFindFile, the program sends appropriate control commands to the receiver during traversal.

Running the programs :

Start the receiver (server) on a host that has an accessible IP (preferably within a LAN):

python3 ftserver.py -i 10.135.xxx.xxx -d /home/ubuntu/downloads

Run the sender (client) on another machine, specifying the server’s public IP and the folder to transfer:

python3 ftclient.py -i 111.120.xxx.xxx -f /Users/capton/desktop/bilibili

Because many Chinese cloud providers use NAT, the public and private IPs must be correctly configured for the two endpoints to communicate.

Note: The system works best on a local network where IPs are directly reachable; using public networks requires knowledge of both the remote host’s public and internal IP addresses.

- END -