From eb5736156bdb736f8197d92fc83b799586ae492e Mon Sep 17 00:00:00 2001
From: Peter Wu <lekensteyn@gmail.com>
Date: Wed, 26 Jun 2013 20:00:08 +0200
Subject: ftp*.py: add FTP scripts

- ftp-get.py: sequentially download a list of files with one connection.
- ftp-list.py: recursively show directory contents.
- ftp.py: supports "MLSD" which is missing in inetutils.
---
 ftp-get.py | 110 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 110 insertions(+)
 create mode 100755 ftp-get.py

(limited to 'ftp-get.py')

diff --git a/ftp-get.py b/ftp-get.py
new file mode 100755
index 0000000..ed47402
--- /dev/null
+++ b/ftp-get.py
@@ -0,0 +1,110 @@
+#!/usr/bin/python
+# Retrieve a list of files from one server. Saves modification time and supports
+# download resumption.
+#
+# Copyright (C) 2013 Peter Wu <lekensteyn@gmail.com>
+
+import sys
+from ftplib import FTP
+import re
+import os, os.path
+from datetime import datetime, timezone
+
+user = "anonymous"
+passwd = "anon"
+host = None
+port = 21
+path = "/"
+
+outdir = os.getcwd()
+
+patt_url = re.compile(r"""
+^
+(?:ftp://)?
+(?:
+	(?P<user>.+?)
+	(?:
+		:(?P<passwd>.+?)
+	)?
+	@
+)?
+(?P<host>[a-zA-Z0-9.-]+)
+(:(?P<port>\d+))?
+(?P<path>/.*)?
+$
+""", re.VERBOSE)
+
+if len(sys.argv) >= 2:
+	m = patt_url.match(sys.argv[1])
+	if m:
+		if m.group("user") is not None:
+			user = m.group("user")
+		if m.group("passwd") is not None:
+			passwd = m.group("passwd")
+		host = m.group("host")
+		if m.group("port") is not None:
+			port = int(m.group("port"))
+		if m.group("path") is not None:
+			path = m.group("path")
+			print("Warning: path component is ignored", file=sys.stderr)
+
+if host is None:
+	print("Usage: python", sys.argv[0],
+		"[ftp://][user[:pass]@]ftp.example.com[:21][/path]",
+		file=sys.stderr)
+	print("A file list is read from stdin", file=sys.stderr)
+	sys.exit(1)
+
+files = []
+for line in sys.stdin:
+	file_path = line.rstrip("\r\n")
+	# note: os.path.join(outdir, file_path) throws away if file_path
+	# is absolute. That is *not* intended here.
+	local_file_path = outdir + "/" + file_path
+	files.append((file_path, local_file_path))
+
+if not files:
+	print("Nothing to do", file=sys.stderr)
+	sys.exit(1)
+
+with FTP() as ftp:
+	ftp.connect(host, port)
+	ftp.login(user, passwd)
+	print(ftp.getwelcome(), file=sys.stderr)
+
+	current = 0
+	total = len(files)
+
+	for file_path, local_file_path in files:
+		current = current + 1
+		print("Processing %d/%d: %s" % (current, total, file_path))
+
+		# TODO: do something with path parameter as passed in a URL?
+		remote_file_path = file_path
+
+		# do not send a REST(art) command when starting at the beginning
+		offset = None
+		file_size = ftp.size(remote_file_path)
+		if file_size is not None:
+			try:
+				local_size = os.path.getsize(local_file_path)
+				if local_size > file_size:
+					print("Local size %d is larger than remote %d, restarting %s" % (local_size, file_size, file_path))
+				elif local_size == file_size:
+					# assume fully downloaded. Maybe check mtime?
+					print("Already completed:", file_path)
+					continue
+				elif local_size > 0:
+					offset = local_size
+					print("Downloading %d remaining bytes of %s" % file_size - offset, file_path)
+			except OSError:
+				pass
+
+		os.makedirs(os.path.dirname(local_file_path), exist_ok=True)
+		with open(local_file_path, "ab") as local_file:
+			ftp.retrbinary("RETR " + remote_file_path, local_file.write, rest=offset)
+
+		# adjust modification times (server returns UTC)
+		timeval = ftp.sendcmd("MDTM " + remote_file_path)[4:]
+		mtime = datetime.strptime(timeval, "%Y%m%d%H%M%S").replace(tzinfo=timezone.utc).timestamp()
+		os.utime(local_file_path, times=(mtime, mtime))
-- 
cgit v1.2.1