cpython/Tools/scripts/ftpmirror.py

401 lines
13 KiB
Python
Raw Normal View History

#! /usr/bin/env python
1995-06-20 16:31:37 -03:00
"""Mirror a remote ftp subtree into a local directory tree.
1995-06-20 16:31:37 -03:00
usage: ftpmirror [-v] [-q] [-i] [-m] [-n] [-r] [-s pat]
[-l username [-p passwd [-a account]]]
hostname[:port] [remotedir [localdir]]
1995-06-20 16:31:37 -03:00
-v: verbose
-q: quiet
-i: interactive mode
-m: macintosh server (NCSA telnet 2.4) (implies -n -s '*.o')
-n: don't log in
-r: remove local files/directories no longer pertinent
-l username [-p passwd [-a account]]: login info (default .netrc or anonymous)
1995-06-20 16:31:37 -03:00
-s pat: skip files matching pattern
hostname: remote host w/ optional port separated by ':'
1995-06-20 16:31:37 -03:00
remotedir: remote directory (default initial)
localdir: local directory (default current)
"""
import os
import sys
import time
import getopt
import ftplib
import netrc
from fnmatch import fnmatch
# Print usage message and exit
1995-06-20 16:31:37 -03:00
def usage(*args):
2001-01-17 04:48:39 -04:00
sys.stdout = sys.stderr
for msg in args: print msg
print __doc__
sys.exit(2)
1995-06-20 16:31:37 -03:00
verbose = 1 # 0 for -q, 2 for -v
interactive = 0
mac = 0
rmok = 0
nologin = 0
1996-02-05 14:49:00 -04:00
skippats = ['.', '..', '.mirrorinfo']
1995-06-20 16:31:37 -03:00
# Main program: parse command line and start processing
1995-06-20 16:31:37 -03:00
def main():
2001-01-17 04:48:39 -04:00
global verbose, interactive, mac, rmok, nologin
try:
opts, args = getopt.getopt(sys.argv[1:], 'a:bil:mnp:qrs:v')
except getopt.error, msg:
usage(msg)
login = ''
passwd = ''
account = ''
if not args: usage('hostname missing')
host = args[0]
port = 0
if ':' in host:
host, port = host.split(':', 1)
port = int(port)
try:
auth = netrc.netrc().authenticators(host)
if auth is not None:
login, account, passwd = auth
except (netrc.NetrcParseError, IOError):
pass
2001-01-17 04:48:39 -04:00
for o, a in opts:
if o == '-l': login = a
if o == '-p': passwd = a
if o == '-a': account = a
if o == '-v': verbose = verbose + 1
if o == '-q': verbose = 0
if o == '-i': interactive = 1
if o == '-m': mac = 1; nologin = 1; skippats.append('*.o')
if o == '-n': nologin = 1
if o == '-r': rmok = 1
if o == '-s': skippats.append(a)
remotedir = ''
localdir = ''
if args[1:]:
remotedir = args[1]
if args[2:]:
localdir = args[2]
if args[3:]: usage('too many arguments')
#
f = ftplib.FTP()
if verbose: print "Connecting to '%s%s'..." % (host,
(port and ":%d"%port or ""))
f.connect(host,port)
2001-01-17 04:48:39 -04:00
if not nologin:
if verbose:
print 'Logging in as %r...' % (login or 'anonymous')
2001-01-17 04:48:39 -04:00
f.login(login, passwd, account)
if verbose: print 'OK.'
pwd = f.pwd()
if verbose > 1: print 'PWD =', repr(pwd)
2001-01-17 04:48:39 -04:00
if remotedir:
if verbose > 1: print 'cwd(%s)' % repr(remotedir)
2001-01-17 04:48:39 -04:00
f.cwd(remotedir)
if verbose > 1: print 'OK.'
pwd = f.pwd()
if verbose > 1: print 'PWD =', repr(pwd)
2001-01-17 04:48:39 -04:00
#
mirrorsubdir(f, localdir)
1995-06-20 16:31:37 -03:00
# Core logic: mirror one subdirectory (recursively)
1995-06-20 16:31:37 -03:00
def mirrorsubdir(f, localdir):
2001-01-17 04:48:39 -04:00
pwd = f.pwd()
if localdir and not os.path.isdir(localdir):
if verbose: print 'Creating local directory', repr(localdir)
2001-01-17 04:48:39 -04:00
try:
makedir(localdir)
except os.error, msg:
print "Failed to establish local directory", repr(localdir)
2001-01-17 04:48:39 -04:00
return
infofilename = os.path.join(localdir, '.mirrorinfo')
try:
text = open(infofilename, 'r').read()
except IOError, msg:
text = '{}'
try:
info = eval(text)
except (SyntaxError, NameError):
print 'Bad mirror info in', repr(infofilename)
2001-01-17 04:48:39 -04:00
info = {}
subdirs = []
listing = []
if verbose: print 'Listing remote directory %r...' % (pwd,)
2001-01-17 04:48:39 -04:00
f.retrlines('LIST', listing.append)
filesfound = []
for line in listing:
if verbose > 1: print '-->', repr(line)
2001-01-17 04:48:39 -04:00
if mac:
# Mac listing has just filenames;
# trailing / means subdirectory
filename = line.strip()
2001-01-17 04:48:39 -04:00
mode = '-'
if filename[-1:] == '/':
filename = filename[:-1]
mode = 'd'
infostuff = ''
else:
# Parse, assuming a UNIX listing
words = line.split(None, 8)
2001-01-17 04:48:39 -04:00
if len(words) < 6:
if verbose > 1: print 'Skipping short line'
continue
filename = words[-1].lstrip()
i = filename.find(" -> ")
2001-01-17 04:48:39 -04:00
if i >= 0:
# words[0] had better start with 'l'...
if verbose > 1:
print 'Found symbolic link %r' % (filename,)
2001-01-17 04:48:39 -04:00
linkto = filename[i+4:]
filename = filename[:i]
infostuff = words[-5:-1]
mode = words[0]
skip = 0
for pat in skippats:
if fnmatch(filename, pat):
if verbose > 1:
print 'Skip pattern', repr(pat),
print 'matches', repr(filename)
2001-01-17 04:48:39 -04:00
skip = 1
break
if skip:
continue
if mode[0] == 'd':
if verbose > 1:
print 'Remembering subdirectory', repr(filename)
2001-01-17 04:48:39 -04:00
subdirs.append(filename)
continue
filesfound.append(filename)
if info.has_key(filename) and info[filename] == infostuff:
if verbose > 1:
print 'Already have this version of',repr(filename)
2001-01-17 04:48:39 -04:00
continue
fullname = os.path.join(localdir, filename)
tempname = os.path.join(localdir, '@'+filename)
if interactive:
doit = askabout('file', filename, pwd)
if not doit:
if not info.has_key(filename):
info[filename] = 'Not retrieved'
continue
try:
os.unlink(tempname)
except os.error:
pass
if mode[0] == 'l':
if verbose:
print "Creating symlink %r -> %r" % (filename, linkto)
2001-01-17 04:48:39 -04:00
try:
os.symlink(linkto, tempname)
except IOError, msg:
print "Can't create %r: %s" % (tempname, msg)
2001-01-17 04:48:39 -04:00
continue
else:
try:
fp = open(tempname, 'wb')
except IOError, msg:
print "Can't create %r: %s" % (tempname, msg)
2001-01-17 04:48:39 -04:00
continue
if verbose:
print 'Retrieving %r from %r as %r...' % (filename, pwd, fullname)
2001-01-17 04:48:39 -04:00
if verbose:
fp1 = LoggingFile(fp, 1024, sys.stdout)
else:
fp1 = fp
t0 = time.time()
try:
f.retrbinary('RETR ' + filename,
fp1.write, 8*1024)
except ftplib.error_perm, msg:
print msg
t1 = time.time()
bytes = fp.tell()
fp.close()
if fp1 != fp:
fp1.close()
try:
os.unlink(fullname)
except os.error:
pass # Ignore the error
try:
os.rename(tempname, fullname)
except os.error, msg:
print "Can't rename %r to %r: %s" % (tempname, fullname, msg)
2001-01-17 04:48:39 -04:00
continue
info[filename] = infostuff
writedict(info, infofilename)
if verbose and mode[0] != 'l':
dt = t1 - t0
kbytes = bytes / 1024.0
print int(round(kbytes)),
print 'Kbytes in',
print int(round(dt)),
print 'seconds',
if t1 > t0:
print '(~%d Kbytes/sec)' % \
int(round(kbytes/dt),)
print
#
# Remove files from info that are no longer remote
deletions = 0
for filename in info.keys():
if filename not in filesfound:
if verbose:
print "Removing obsolete info entry for",
print repr(filename), "in", repr(localdir or ".")
2001-01-17 04:48:39 -04:00
del info[filename]
deletions = deletions + 1
if deletions:
writedict(info, infofilename)
#
# Remove local files that are no longer in the remote directory
try:
if not localdir: names = os.listdir(os.curdir)
else: names = os.listdir(localdir)
except os.error:
names = []
for name in names:
if name[0] == '.' or info.has_key(name) or name in subdirs:
continue
skip = 0
for pat in skippats:
if fnmatch(name, pat):
if verbose > 1:
print 'Skip pattern', repr(pat),
print 'matches', repr(name)
2001-01-17 04:48:39 -04:00
skip = 1
break
if skip:
continue
fullname = os.path.join(localdir, name)
if not rmok:
if verbose:
print 'Local file', repr(fullname),
2001-01-17 04:48:39 -04:00
print 'is no longer pertinent'
continue
if verbose: print 'Removing local file/dir', repr(fullname)
2001-01-17 04:48:39 -04:00
remove(fullname)
#
# Recursively mirror subdirectories
for subdir in subdirs:
if interactive:
doit = askabout('subdirectory', subdir, pwd)
if not doit: continue
if verbose: print 'Processing subdirectory', repr(subdir)
2001-01-17 04:48:39 -04:00
localsubdir = os.path.join(localdir, subdir)
pwd = f.pwd()
if verbose > 1:
print 'Remote directory now:', repr(pwd)
print 'Remote cwd', repr(subdir)
2001-01-17 04:48:39 -04:00
try:
f.cwd(subdir)
except ftplib.error_perm, msg:
print "Can't chdir to", repr(subdir), ":", repr(msg)
2001-01-17 04:48:39 -04:00
else:
if verbose: print 'Mirroring as', repr(localsubdir)
2001-01-17 04:48:39 -04:00
mirrorsubdir(f, localsubdir)
if verbose > 1: print 'Remote cwd ..'
f.cwd('..')
newpwd = f.pwd()
if newpwd != pwd:
print 'Ended up in wrong directory after cd + cd ..'
print 'Giving up now.'
break
else:
if verbose > 1: print 'OK.'
1995-06-20 16:31:37 -03:00
# Helper to remove a file or directory tree
def remove(fullname):
2001-01-17 04:48:39 -04:00
if os.path.isdir(fullname) and not os.path.islink(fullname):
try:
names = os.listdir(fullname)
except os.error:
names = []
ok = 1
for name in names:
if not remove(os.path.join(fullname, name)):
ok = 0
if not ok:
return 0
try:
os.rmdir(fullname)
except os.error, msg:
print "Can't remove local directory %r: %s" % (fullname, msg)
2001-01-17 04:48:39 -04:00
return 0
else:
try:
os.unlink(fullname)
except os.error, msg:
print "Can't remove local file %r: %s" % (fullname, msg)
2001-01-17 04:48:39 -04:00
return 0
return 1
1995-06-20 16:31:37 -03:00
# Wrapper around a file for writing to write a hash sign every block.
class LoggingFile:
2001-01-17 04:48:39 -04:00
def __init__(self, fp, blocksize, outfp):
self.fp = fp
self.bytes = 0
self.hashes = 0
self.blocksize = blocksize
self.outfp = outfp
def write(self, data):
self.bytes = self.bytes + len(data)
hashes = int(self.bytes) / self.blocksize
while hashes > self.hashes:
self.outfp.write('#')
self.outfp.flush()
self.hashes = self.hashes + 1
self.fp.write(data)
def close(self):
self.outfp.write('\n')
1995-06-20 16:31:37 -03:00
# Ask permission to download a file.
def askabout(filetype, filename, pwd):
2001-01-17 04:48:39 -04:00
prompt = 'Retrieve %s %s from %s ? [ny] ' % (filetype, filename, pwd)
while 1:
reply = raw_input(prompt).strip().lower()
2001-01-17 04:48:39 -04:00
if reply in ['y', 'ye', 'yes']:
return 1
if reply in ['', 'n', 'no', 'nop', 'nope']:
return 0
print 'Please answer yes or no.'
1995-06-20 16:31:37 -03:00
# Create a directory if it doesn't exist. Recursively create the
# parent directory as well if needed.
def makedir(pathname):
2001-01-17 04:48:39 -04:00
if os.path.isdir(pathname):
return
dirname = os.path.dirname(pathname)
if dirname: makedir(dirname)
os.mkdir(pathname, 0777)
1995-06-20 16:31:37 -03:00
# Write a dictionary to a file in a way that can be read back using
# rval() but is still somewhat readable (i.e. not a single long line).
# Also creates a backup file.
1995-06-20 16:31:37 -03:00
def writedict(dict, filename):
dir, fname = os.path.split(filename)
tempname = os.path.join(dir, '@' + fname)
backup = os.path.join(dir, fname + '~')
2001-01-17 04:48:39 -04:00
try:
os.unlink(backup)
except os.error:
pass
fp = open(tempname, 'w')
fp.write('{\n')
for key, value in dict.items():
fp.write('%r: %r,\n' % (key, value))
2001-01-17 04:48:39 -04:00
fp.write('}\n')
fp.close()
try:
os.rename(filename, backup)
except os.error:
pass
os.rename(tempname, filename)
1995-06-20 16:31:37 -03:00
if __name__ == '__main__':
2001-01-17 04:48:39 -04:00
main()