ChatGPT解决这个技术问题 Extra ChatGPT

How to scp in Python?

What's the most pythonic way to scp a file in Python? The only route I'm aware of is

os.system('scp "%s" "%s:%s"' % (localfile, remotehost, remotefile) )

which is a hack, and which doesn't work outside Linux-like systems, and which needs help from the Pexpect module to avoid password prompts unless you already have passwordless SSH set up to the remote host.

I'm aware of Twisted's conch, but I'd prefer to avoid implementing scp myself via low-level ssh modules.

I'm aware of paramiko, a Python module that supports SSH and SFTP; but it doesn't support SCP.

Background: I'm connecting to a router which doesn't support SFTP but does support SSH/SCP, so SFTP isn't an option.

EDIT: This is a duplicate of How to copy a file to a remote server in Python using SCP or SSH?. However, that question doesn't give an scp-specific answer that deals with keys from within Python. I'm hoping for a way to run code kind of like

import scp

client = scp.Client(host=host, user=user, keyfile=keyfile)
# or
client = scp.Client(host=host, user=user)
client.use_system_keys()
# or
client = scp.Client(host=host, user=user, password=password)

# and then
client.transfer('/etc/local/filename', '/etc/remote/filename')

M
Martin Prikryl

Try the Python scp module for Paramiko. It's very easy to use. See the following example:

import paramiko
from scp import SCPClient

def createSSHClient(server, port, user, password):
    client = paramiko.SSHClient()
    client.load_system_host_keys()
    client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
    client.connect(server, port, user, password)
    return client

ssh = createSSHClient(server, port, user, password)
scp = SCPClient(ssh.get_transport())

Then call scp.get() or scp.put() to do SCP operations.

(SCPClient code)


Except that module doesn't actually use paramiko - it's a lot of code that in the end for scp ends up actually calling the scp command line which only works on *nix.
Agreed, what you're seeing in the source code is the remote call to scp -t or scp -f ('to' and 'from' modes, which exist for this server-side purpose). This is essentially how scp works - it relies on another copy of scp existing on the server. This article explains it very well: blogs.oracle.com/janp/entry/how_the_scp_protocol_works
This solution worked for me with two caveats: 1. these are the import statements I used: import paramiko from scp import SCPClient 2. Here is an example of the scp.get() command: scp.get(r'/nfs_home/appers/xxxx/test2.txt', r'C:\Users\xxxx\Desktop\MR_Test')
This not only works great, but also takes advantage of SSH keys if they exist.
I'm getting an error scp.SCPException: scp Protocol not available . I'm trying to copy files from Windows to Mac. The same command works fine on the terminal.
O
Oddstr13

You might be interested in trying Pexpect (source code). This would allow you to deal with interactive prompts for your password.

Here's a snip of example usage (for ftp) from the main website:

# This connects to the openbsd ftp site and
# downloads the recursive directory listing.
import pexpect
child = pexpect.spawn ('ftp ftp.openbsd.org')
child.expect ('Name .*: ')
child.sendline ('anonymous')
child.expect ('Password:')
child.sendline ('noah@example.com')
child.expect ('ftp> ')
child.sendline ('cd pub')
child.expect('ftp> ')
child.sendline ('get ls-lR.gz')
child.expect('ftp> ')
child.sendline ('bye')

M
Maviles

Couldn't find a straight answer, and this "scp.Client" module doesn't exist. Instead, this suits me:

from paramiko import SSHClient
from scp import SCPClient

ssh = SSHClient()
ssh.load_system_host_keys()
ssh.connect('example.com')

with SCPClient(ssh.get_transport()) as scp:
   scp.put('test.txt', 'test2.txt')
   scp.get('test2.txt')

J
JimB

You could also check out paramiko. There's no scp module (yet), but it fully supports sftp.

[EDIT] Sorry, missed the line where you mentioned paramiko. The following module is simply an implementation of the scp protocol for paramiko. If you don't want to use paramiko or conch (the only ssh implementations I know of for python), you could rework this to run over a regular ssh session using pipes.

scp.py for paramiko


Are you saying that the attached solution will securely transfer files to any machine running sshd, even if it's not running sftpd? That's exactly what I'm looking for, but I can't tell from your comment whether this just wraps sftp in an scp-like facade.
This will transfer files with any machine running sshd, which has scp in the PATH (scp isn't part of the ssh spec, but it's fairly ubiquitous). This invokes "scp -t" on the server, and uses the scp protocol to transfer files, which has nothing to do with sftp.
Please note that I used openssh's implementation of scp as the model, so this isn't guaranteed to work with other versions. Some versions of sshd may also use the scp2 protocol, which is basically the same as sftp.
Could you please see this question: stackoverflow.com/questions/20111242/… I am wondering if paramiko uses subprocess or comething else like sockets. I am having memory issues using subprocess.check_output('ssh blah@blah.com "cat /data/file*") due to clone/fork issues and am wondering if paramiko will have the same issues or not?
@Paul: paramiko will not have the same issues, since it does not use a subprocess.
B
Blauohr

if you install putty on win32 you get an pscp (putty scp).

so you can use the os.system hack on win32 too.

(and you can use the putty-agent for key-managment)

sorry it is only a hack (but you can wrap it in a python class)


The only reason the 'nix one works is, you have SCP on the path; as Blauohr points out, that's not too hard to fix. +1
L
Loïc

As of today, the best solution is probably AsyncSSH

https://asyncssh.readthedocs.io/en/latest/#scp-client

async with asyncssh.connect('host.tld') as conn:
    await asyncssh.scp((conn, 'example.txt'), '.', recurse=True)

Hi. You make a claim about AsyncSSH being the best, but could you make a statement or 2 to support this claim? Maybe differentiate it from scp(), or how it mattered for your use case. (This said, it's vastly less code - in your example at least)
@Crossfit_and_Beer hello. So I said "probably the best", I'll let the judgement to anyone :) I've not much time right now to compare AsyncSSH to other libraries. But very quickly : it's simple to use, it's asynchronous, it's maintained and the maintener is very nice and helpful, it is quite complete and very well documented.
I went with this solution and it performs well, with over 100 simultaneous transfers.
u
user7529863

You can use the package subprocess and the command call to use the scp command from the shell.

from subprocess import call

cmd = "scp user1@host1:files user2@host2:files"
call(cmd.split(" "))

How is this substantially different from the base case mentioned by the questioner? Yes, subprocess is better than os.system, but in both cases, it doesn't work outside linux-like systems, has issues with password prompts, etc.
u
user443854

Have a look at fabric.transfer.

from fabric import Connection

with Connection(host="hostname", 
                user="admin", 
                connect_kwargs={"key_filename": "/home/myuser/.ssh/private.key"}
               ) as c:
    c.get('/foo/bar/file.txt', '/tmp/')

Could you be more specific?
Fabric does not support scp as of now. (Reference: github.com/fabric/fabric/issues/1948)
s
shrikkanth roxor
import paramiko

client = paramiko.SSHClient()
client.set_missing_host_key_policy(paramiko.AutoAddPolicy())

client.connect('<IP Address>', username='<User Name>',password='' ,key_filename='<.PEM File path')

#Setup sftp connection and transmit this script 
print ("copying")

sftp = client.open_sftp() 
sftp.put(<Source>, <Destination>)


sftp.close()

It would help if you added some commentary to your answer explaining why your solution is a good one and help readers compare it with the other solutions available. Just posting code does not always help learner know how to recognise better solutions.
This is SFTP, not SCP.
s
some bits flipped

It has been quite a while since this question was asked, and in the meantime, another library that can handle this has cropped up: You can use the copy function included in the Plumbum library:

import plumbum
r = plumbum.machines.SshMachine("example.net")
   # this will use your ssh config as `ssh` from shell
   # depending on your config, you might also need additional
   # params, eg: `user="username", keyfile=".ssh/some_key"`
fro = plumbum.local.path("some_file")
to = r.path("/path/to/destination/")
plumbum.path.utils.copy(fro, to)

How can I enter the passphrase for the private key with Plumbum ? p
@ekta: You will be prompted for it on the command line. If you want to provide it from your own code, I don't think plumbum's API supports this. But plumbum's SshMachine does have access to ssh-agent, so if you just want to avoid typing it in every time, that's one way to do it. You could also file a feature request on plumbum's github page.
u
user178047

If you are on *nix you can use sshpass

sshpass -p password scp -o User=username -o StrictHostKeyChecking=no src dst:/path

P
Pat Notz

Hmmm, perhaps another option would be to use something like sshfs (there an sshfs for Mac too). Once your router is mounted you can just copy the files outright. I'm not sure if that works for your particular application but it's a nice solution to keep handy.


c
ccpizza

I while ago I put together a python SCP copy script that depends on paramiko. It includes code to handle connections with a private key or SSH key agent with a fallback to password authentication.

http://code.activestate.com/recipes/576810-copy-files-over-ssh-using-paramiko/


Thanks for the link and for taking time to publish good example of using paramiko.