Higher-level ftplib

ftplib in itself is a bit raw, as it leaves details about the protocol for the user to handle. ftptool abstracts that away, and even provides a neat interface for file management.

Code says more than words, so let's look at an example: connecting.

>>> a_host = FTPHost.connect("ftp.python.org", user="foo", password="bar")

connect is a classmethod that lets you create an FTPHost instance with an underlying ftplib.FTP instance.

Changing and getting the current directory is implemented as a property called current_directory. It is lazy; it won't ask the server which the current directory is until you ask for it.

Note that since it's a property, this will actually go one level up:

>>> a_host.current_directory = ".."

Similarly, this will descend into the "foo" directory:

>>> a_host.current_directory = "foo"

In most cases, it's easier to just specify absolute paths:

>>> a_host.current_directory = "/foo"

current_directory will always be the server-side representation; when you change directory, it ends up sending a CWD and then a PWD to get the result of the operation (since the FTP protocol doesn't define what the reply text to a CWD is.)

A os.walk interface is implemented for walking the directory tree:

>>> for (dirname, subdirs, files) in a_host.walk("/a_dir"):
...     print dirname, "has file(s)", ", ".join(files)
...
/a_dir has file(s) foo, bar
/a_dir/other_dir has file(s) hello
/a_dir/some_dir has file(s)

Just like os.walk, you can remove entries in the subdirs list to avoid descending into them:

>>> for (dirname, subdirs, files) in a_host.walk("/a_dir"):
...     for subdir in subdirs:
...         if subdir.startswith("other_"):
...             subdirs.remove(subdir)
...     print dirname, "has file(s)", ", ".join(files)
...
/a_dir has file(s) foo, bar
/a_dir/some_dir has file(s)

You can non-recursively list a directory using listdir:

>>> a_host.listdir("/a_dir")
(['other_dir', 'some_dir'], ['foo', 'bar'])

The most simple form of creating a directory is mkdir. You simply give it a directory to create, and so it does:

>>> a_host.mkdir("/new_dir")

If you just want to ascertain that a directory is ready, i.e., exists for an upload, you could use makedirs which tries to create every part of the directory, piece by piece.

>>> a_host.makedirs("/a_dir/some_dir/a_new_dir/other_new_dir")

Would, hypothetically, create a_new_dir and other_new_dir.

ftptool implements it by first trying to change directory into the given path, to see if it exists, and then changes back. If it does, it simply returns, otherwise it creates the directories piece by piece.

Files in ftptool are implemented using proxy objects called FTPFileProxy. They represent a file on a remote host. Using them is easy as pie!

>>> a_host.file_proxy("/a_dir/foo").download_to_str()
'This is the file "foo".'
>>> a_host.file_proxy("/a_dir/new_file").upload_from_str("Hello world!")

ftptool provides three ways of uploading or downloading files: * to/from_str: using a str object, * to/from_file: using a filename, * and the default: using a file-like object.

Given:

>>> f = a_host.file_proxy("/foo.txt")

You could upload and download from str using these two:

>>> f.upload_from_str("Hi!")
>>> f.download_to_str()
'Hi!'

And using a filename like this:

>>> f.upload_from_file("/etc/motd")
>>> f.download_to_file("/tmp/motd")

And lastly, using file-like objects:

>>> f.upload(StringIO("Test!"))
>>> fp = StringIO()
>>> f.download(fp)
>>> fp.getvalue()
'Test!'

Renaming is a method of the file proxies, called rename. It returns a new file proxy for the renamed-to file, so the common pattern will be:

>>> a_file = a_host.file_proxy("hello_world")
>>> a_file = a_file.rename("foobar")

This will issue a rename command, too, so a_file will essentially be the same as before, with a new name and a new instance ID.

Deleting a file is much like renaming it: it's a method of the file proxies, called delete. It, however, doesn't have a meaningful return value.

>>> a_file.delete()

ftptool supports two types of mirroring: local to remote, and remote to local. As in, it can download a whole directory and all descendants into a local directory, for you to play with. It can also upload a whole directory to a remote host.

The first one, downloading, is called mirror_to_local. It's used like so:

>>> a_host.mirror_to_local('/a_dir', 'my_copy_of_a_dir')

The cousin, mirror_to_remote, has the same signature; source first, then destination.

>>> a_host.mirror_to_remote('my_copy_of_a_dir', '/a_dir')

If the local working directory is the one you want to upload, you can just give mirror_to_remote an empty string or a dot.