使用Python Paramiko通过多个连接/线程将大文件上传到SFTP服务器

3ks5zfa0  于 2023-03-04  发布在  Python
关注(0)|答案(1)|浏览(327)

我正在尝试使用线程和python paramiko库将文件分块SFTP到远程服务器。
它在不同的线程中打开一个本地文件和sftp块到远程服务器。
我基本上遵循这个解决方案,它使用相同的方法通过SFTP下载大文件。我想发送大文件代替。Downloading solution
但是,我在chunk in infile.readv(chunks):的行上输入write_chunks()时遇到以下错误:
属性错误:"_io. BufferedReader"对象没有属性"readv"
有谁能帮助解决这个错误吗?我以为infile是一个文件描述符。我不明白为什么它是一个_io.BufferedReader object

import threading, os, time, paramiko

import time, paramiko

MAX_RETRIES = 10

ftp_server = "server.com"
port = 22
remote_file = "/home/filecopy.bin"
local_file = "/home/file.bin"
ssh_conn = sftp_client = None
username = "none"
password = "none"

#you could make the number of threads relative to file size
NUM_THREADS = 2
MAX_RETRIES = 10

def make_filepart_path(file_path, part_number):
    """creates filepart path from filepath"""
    return "%s.filepart.%s" % (file_path, part_number+1)

def write_chunks(chunks, tnum, remote_file_part, username, password, ftp_server, max_retries):
    ssh_conn = sftp_client = None
    for retry in range(max_retries):
        try:
            ssh_conn = paramiko.Transport((ftp_server, port))
            ssh_conn.connect(username=username, password=password)
            sftp_client = paramiko.SFTPClient.from_transport(ssh_conn)
            with sftp_client.open(remote_file_part, "wb") as outfile:
                with open(local_file, "rb") as infile:
                    for chunk in infile.readv(chunks):
                        outfile.write(chunk)
            break
        except (EOFError, paramiko.ssh_exception.SSHException, OSError) as x:
            retry += 1
            print("%s %s Thread %s - > retrying %s..." % (type(x), x, tnum, retry))
            time.sleep(abs(retry) * 10)
        finally:
            if hasattr(sftp_client, "close") and callable(sftp_client.close):
                sftp_client.close()
            if hasattr(ssh_conn, "close") and callable(ssh_conn.close):
                ssh_conn.close()


start_time = time.time()

for retry in range(MAX_RETRIES):
    try:
        ssh_conn = paramiko.Transport((ftp_server, port))
        ssh_conn.connect(username=username, password=password)
        sftp_client = paramiko.SFTPClient.from_transport(ssh_conn)
        # connect to get the file's size in order to calculate chunks
        #filesize = sftp_client.stat(remote_file).st_size
        filesize = os.stat(local_file).st_size
        sftp_client.close()
        ssh_conn.close()
        chunksize = pow(2, 12)
        chunks = [(offset, chunksize) for offset in range(0, filesize, chunksize)]
        thread_chunk_size = (len(chunks) // NUM_THREADS) + 1
        # break the chunks into sub lists to hand off to threads
        thread_chunks = [chunks[i:i+thread_chunk_size] for i in range(0, len(chunks) - 1, thread_chunk_size)]
        threads = []
        fileparts = []
        for thread_num in range(len(thread_chunks)):
            remote_file_part = make_filepart_path(remote_file, thread_num) 
            args = (thread_chunks[thread_num], thread_num, remote_file_part, username, password, ftp_server, MAX_RETRIES)
            threads.append(threading.Thread(target=write_chunks, args=args))
            fileparts.append(remote_file_part)
        for thread in threads:
            thread.start()
        for thread in threads:
            thread.join()
        # join file parts into one file, remove fileparts
        with sftp_client.open(remote_file_part, "wb") as outfile:
            for filepart in fileparts:
                with open(filepart, "rb") as infile:
                    outfile.write(infile.read())
                os.remove(filepart)
        break
    except (EOFError, paramiko.ssh_exception.SSHException, OSError) as x:
        retry += 1
        print("%s %s - > retrying %s..." % (type(x), x, retry))
        time.sleep(abs(retry) * 10)
    finally:
       if hasattr(sftp_client, "close") and callable(sftp_client.close):
           sftp_client.close()
       if hasattr(ssh_conn, "close") and callable(ssh_conn.close):
           ssh_conn.close()

print("Loading File %s Took %d seconds " % (sftp_file, time.time() - start_time))

堆栈跟踪:

Exception in thread Thread-4:
Traceback (most recent call last):
  File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.6/threading.py", line 864, in run
    self._target(*self._args, **self._kwargs)
  File "simpleNNInference.py", line 210, in write_chunks
    for chunk in infile.readv(chunks):
AttributeError: '_io.BufferedReader' object has no attribute 'readv'

Exception in thread Thread-3:
Traceback (most recent call last):
  File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.6/threading.py", line 864, in run
    self._target(*self._args, **self._kwargs)
  File "simpleNNInference.py", line 210, in write_chunks
    for chunk in infile.readv(chunks):
AttributeError: '_io.BufferedReader' object has no attribute 'readv'
w8biq8rn

w8biq8rn1#

有关如何对一个大文件执行并行多部分上载的示例,请参见以下示例。
请注意,大多数SFTP服务器(包括OpenSSH,直到最新的9.0)不允许远程合并文件,因此您必须使用revert to shell command for that

import os
import threading
import paramiko

sftp_server = "example.com"
username = "username"
password = "password"

local_path = "/local/path/file.dat"
remote_path = "/remote/path/file.dat"

threads_count = 4

size = os.path.getsize(local_path)
part_size = int(size / threads_count)

def open_ssh():
    ssh = paramiko.SSHClient()
    ssh.connect(sftp_server, username=username, password=password)
    return ssh

def upload_part(num, offset, part_size, remote_path_part):
    print(f"Running thread {num}")
    try:
        ssh = open_ssh()
        sftp = ssh.open_sftp()
        with open(local_path, "rb") as fl:
            fl.seek(offset)
            with sftp.open(remote_path_part, "wb") as fr:
                fr.set_pipelined(True)
                size = 0
                while size < part_size:
                    s = 32768
                    if size + s > part_size:
                        s = part_size - size
                    data = fl.read(s)
                    fr.write(data)
                    size += len(data)
                    if len(data) == 0:
                        break
    except (paramiko.ssh_exception.SSHException) as x:
        print(f"Thread {num} failed: {x}")
    print(f"Thread {num} done")

print("Starting")
offset = 0
threads = []
part_filenames = []
for num in range(threads_count):
    if num == threads_count - 1:
        part_size = size - offset
    remote_path_part = f"{remote_path}.{num}"
    args = (num, offset, part_size, remote_path_part)
    print(f"Starting thread {num} offset {offset} size {part_size} " +
          f"part name {remote_path_part}")
    thread = threading.Thread(target=upload_part, args=args)
    threads.append(thread)
    part_filenames.append(remote_path_part)
    thread.start()
    print(f"Started thread {num}")
    offset += part_size

for num in range(len(threads)):
    print(f"Waiting for thread {num}")
    threads[num].join()

print("All thread done")

parts_list = " ".join(part_filenames)
merge_command =
    f"rm \"{remote_path}\" 2> /dev/null ; " + \
    f"for i in {parts_list} ; do cat \"$i\" >> {remote_path} && " + \
     "rm \"$i\" || break ; done"
print(f"Merge command: {merge_command}");

ssh = open_ssh()
stdin, stdout, stderr = ssh.exec_command(merge_command)
print(stdout.read().decode("utf-8"))
print(stderr.read().decode("utf-8"))

我不知道SFTP规范对此有多大支持,但许多SFTP服务器(包括OpenSSH)允许从多个连接并行写入同一个文件,因此您甚至可以不合并文件,直接上传到目标文件的各个部分:

import os
import threading
import paramiko

sftp_server = "example.com"
username = "username"
password = "password"

local_path = "/local/path/file.dat"
remote_path = "/remote/path/file.dat"

threads_count = 4

size = os.path.getsize(local_path)
part_size = int(size / threads_count)
lock = threading.Lock()
created = False

def upload_part(num, offset, part_size):
    print(f"Running thread {num}")
    try:
        ssh = paramiko.SSHClient()
        ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
        ssh.connect(sftp_server, port=port, username=username, password=password)
        sftp = ssh.open_sftp()
        with open(local_path, "rb") as fl:
            fl.seek(offset)
            with lock:
                global created
                m = "r+" if created else "w"
                created = True
                fr = sftp.open(remote_path, m)
            with fr:
                fr.seek(offset)
                fr.set_pipelined(True)
                size = 0
                while size < part_size:
                    s = 32768
                    if size + s > part_size:
                        s = part_size - size
                    data = fl.read(s)
                    fr.write(data)
                    size += len(data)
                    if len(data) == 0:
                        break
    except (paramiko.ssh_exception.SSHException) as x:
        print(f"Thread {num} failed: {x}")
    print(f"Thread {num} done")

print("Starting")
offset = 0
threads = []
for num in range(threads_count):
    if num == threads_count - 1:
        part_size = size - offset
    args = (num, offset, part_size)
    print(f"Starting thread {num} offset {offset} size {part_size}")
    thread = threading.Thread(target=upload_part, args=args)
    threads.append(thread)
    thread.start()
    print(f"Started thread {num}")
    offset += part_size

for num in range(len(threads)):
    print(f"Waiting for thread {num}")
    threads[num].join()

print("All thread done")

相关问题