blob: c3469d06b863cee1178d1d5c7ae133a8baab6c46 [file] [log] [blame]
Guido van Rossumf06ee5f1996-11-27 19:52:01 +00001#! /usr/bin/env python
Guido van Rossume830e551995-06-20 19:31:37 +00002
Guido van Rossumfdc5adc1998-01-04 02:03:12 +00003"""Mirror a remote ftp subtree into a local directory tree.
4
5usage: ftpmirror [-v] [-q] [-i] [-m] [-n] [-r] [-s pat]
6 [-l username [-p passwd [-a account]]]
Tim Peters70c43782001-01-17 08:48:39 +00007 hostname [remotedir [localdir]]
Guido van Rossumfdc5adc1998-01-04 02:03:12 +00008-v: verbose
9-q: quiet
10-i: interactive mode
11-m: macintosh server (NCSA telnet 2.4) (implies -n -s '*.o')
12-n: don't log in
13-r: remove local files/directories no longer pertinent
Martin v. Löwisa4debff2001-06-07 17:17:00 +000014-l username [-p passwd [-a account]]: login info (default .netrc or anonymous)
Guido van Rossumfdc5adc1998-01-04 02:03:12 +000015-s pat: skip files matching pattern
16hostname: remote host
17remotedir: remote directory (default initial)
18localdir: local directory (default current)
19"""
20
Guido van Rossume830e551995-06-20 19:31:37 +000021import os
22import sys
23import time
24import getopt
Guido van Rossume830e551995-06-20 19:31:37 +000025import ftplib
Martin v. Löwisa4debff2001-06-07 17:17:00 +000026import netrc
Guido van Rossume830e551995-06-20 19:31:37 +000027from fnmatch import fnmatch
28
Guido van Rossumfdc5adc1998-01-04 02:03:12 +000029# Print usage message and exit
Guido van Rossume830e551995-06-20 19:31:37 +000030def usage(*args):
Tim Peters70c43782001-01-17 08:48:39 +000031 sys.stdout = sys.stderr
32 for msg in args: print msg
33 print __doc__
34 sys.exit(2)
Guido van Rossume830e551995-06-20 19:31:37 +000035
36verbose = 1 # 0 for -q, 2 for -v
37interactive = 0
38mac = 0
39rmok = 0
40nologin = 0
Guido van Rossumd2966cb1996-02-05 18:49:00 +000041skippats = ['.', '..', '.mirrorinfo']
Guido van Rossume830e551995-06-20 19:31:37 +000042
Guido van Rossumfdc5adc1998-01-04 02:03:12 +000043# Main program: parse command line and start processing
Guido van Rossume830e551995-06-20 19:31:37 +000044def main():
Tim Peters70c43782001-01-17 08:48:39 +000045 global verbose, interactive, mac, rmok, nologin
46 try:
47 opts, args = getopt.getopt(sys.argv[1:], 'a:bil:mnp:qrs:v')
48 except getopt.error, msg:
49 usage(msg)
50 login = ''
51 passwd = ''
52 account = ''
Martin v. Löwisa4debff2001-06-07 17:17:00 +000053 if not args: usage('hostname missing')
54 host = args[0]
55 try:
56 auth = netrc.netrc().authenticators(host)
57 if auth is not None:
58 login, account, passwd = auth
59 except (netrc.NetrcParseError, IOError):
60 pass
Tim Peters70c43782001-01-17 08:48:39 +000061 for o, a in opts:
62 if o == '-l': login = a
63 if o == '-p': passwd = a
64 if o == '-a': account = a
65 if o == '-v': verbose = verbose + 1
66 if o == '-q': verbose = 0
67 if o == '-i': interactive = 1
68 if o == '-m': mac = 1; nologin = 1; skippats.append('*.o')
69 if o == '-n': nologin = 1
70 if o == '-r': rmok = 1
71 if o == '-s': skippats.append(a)
Tim Peters70c43782001-01-17 08:48:39 +000072 remotedir = ''
73 localdir = ''
74 if args[1:]:
75 remotedir = args[1]
76 if args[2:]:
77 localdir = args[2]
78 if args[3:]: usage('too many arguments')
79 #
80 f = ftplib.FTP()
81 if verbose: print 'Connecting to %s...' % `host`
82 f.connect(host)
83 if not nologin:
84 if verbose:
85 print 'Logging in as %s...' % `login or 'anonymous'`
86 f.login(login, passwd, account)
87 if verbose: print 'OK.'
88 pwd = f.pwd()
89 if verbose > 1: print 'PWD =', `pwd`
90 if remotedir:
91 if verbose > 1: print 'cwd(%s)' % `remotedir`
92 f.cwd(remotedir)
93 if verbose > 1: print 'OK.'
94 pwd = f.pwd()
95 if verbose > 1: print 'PWD =', `pwd`
96 #
97 mirrorsubdir(f, localdir)
Guido van Rossume830e551995-06-20 19:31:37 +000098
Guido van Rossumfdc5adc1998-01-04 02:03:12 +000099# Core logic: mirror one subdirectory (recursively)
Guido van Rossume830e551995-06-20 19:31:37 +0000100def mirrorsubdir(f, localdir):
Tim Peters70c43782001-01-17 08:48:39 +0000101 pwd = f.pwd()
102 if localdir and not os.path.isdir(localdir):
103 if verbose: print 'Creating local directory', `localdir`
104 try:
105 makedir(localdir)
106 except os.error, msg:
107 print "Failed to establish local directory", `localdir`
108 return
109 infofilename = os.path.join(localdir, '.mirrorinfo')
110 try:
111 text = open(infofilename, 'r').read()
112 except IOError, msg:
113 text = '{}'
114 try:
115 info = eval(text)
116 except (SyntaxError, NameError):
117 print 'Bad mirror info in %s' % `infofilename`
118 info = {}
119 subdirs = []
120 listing = []
121 if verbose: print 'Listing remote directory %s...' % `pwd`
122 f.retrlines('LIST', listing.append)
123 filesfound = []
124 for line in listing:
125 if verbose > 1: print '-->', `line`
126 if mac:
127 # Mac listing has just filenames;
128 # trailing / means subdirectory
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000129 filename = line.strip()
Tim Peters70c43782001-01-17 08:48:39 +0000130 mode = '-'
131 if filename[-1:] == '/':
132 filename = filename[:-1]
133 mode = 'd'
134 infostuff = ''
135 else:
136 # Parse, assuming a UNIX listing
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000137 words = line.split(None, 8)
Tim Peters70c43782001-01-17 08:48:39 +0000138 if len(words) < 6:
139 if verbose > 1: print 'Skipping short line'
140 continue
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000141 filename = words[-1].lstrip()
142 i = filename.find(" -> ")
Tim Peters70c43782001-01-17 08:48:39 +0000143 if i >= 0:
144 # words[0] had better start with 'l'...
145 if verbose > 1:
146 print 'Found symbolic link %s' % `filename`
147 linkto = filename[i+4:]
148 filename = filename[:i]
149 infostuff = words[-5:-1]
150 mode = words[0]
151 skip = 0
152 for pat in skippats:
153 if fnmatch(filename, pat):
154 if verbose > 1:
155 print 'Skip pattern', `pat`,
156 print 'matches', `filename`
157 skip = 1
158 break
159 if skip:
160 continue
161 if mode[0] == 'd':
162 if verbose > 1:
163 print 'Remembering subdirectory', `filename`
164 subdirs.append(filename)
165 continue
166 filesfound.append(filename)
167 if info.has_key(filename) and info[filename] == infostuff:
168 if verbose > 1:
169 print 'Already have this version of',`filename`
170 continue
171 fullname = os.path.join(localdir, filename)
172 tempname = os.path.join(localdir, '@'+filename)
173 if interactive:
174 doit = askabout('file', filename, pwd)
175 if not doit:
176 if not info.has_key(filename):
177 info[filename] = 'Not retrieved'
178 continue
179 try:
180 os.unlink(tempname)
181 except os.error:
182 pass
183 if mode[0] == 'l':
184 if verbose:
185 print "Creating symlink %s -> %s" % (
186 `filename`, `linkto`)
187 try:
188 os.symlink(linkto, tempname)
189 except IOError, msg:
190 print "Can't create %s: %s" % (
191 `tempname`, str(msg))
192 continue
193 else:
194 try:
195 fp = open(tempname, 'wb')
196 except IOError, msg:
197 print "Can't create %s: %s" % (
198 `tempname`, str(msg))
199 continue
200 if verbose:
201 print 'Retrieving %s from %s as %s...' % \
202 (`filename`, `pwd`, `fullname`)
203 if verbose:
204 fp1 = LoggingFile(fp, 1024, sys.stdout)
205 else:
206 fp1 = fp
207 t0 = time.time()
208 try:
209 f.retrbinary('RETR ' + filename,
210 fp1.write, 8*1024)
211 except ftplib.error_perm, msg:
212 print msg
213 t1 = time.time()
214 bytes = fp.tell()
215 fp.close()
216 if fp1 != fp:
217 fp1.close()
218 try:
219 os.unlink(fullname)
220 except os.error:
221 pass # Ignore the error
222 try:
223 os.rename(tempname, fullname)
224 except os.error, msg:
225 print "Can't rename %s to %s: %s" % (`tempname`,
226 `fullname`,
227 str(msg))
228 continue
229 info[filename] = infostuff
230 writedict(info, infofilename)
231 if verbose and mode[0] != 'l':
232 dt = t1 - t0
233 kbytes = bytes / 1024.0
234 print int(round(kbytes)),
235 print 'Kbytes in',
236 print int(round(dt)),
237 print 'seconds',
238 if t1 > t0:
239 print '(~%d Kbytes/sec)' % \
240 int(round(kbytes/dt),)
241 print
242 #
243 # Remove files from info that are no longer remote
244 deletions = 0
245 for filename in info.keys():
246 if filename not in filesfound:
247 if verbose:
248 print "Removing obsolete info entry for",
249 print `filename`, "in", `localdir or "."`
250 del info[filename]
251 deletions = deletions + 1
252 if deletions:
253 writedict(info, infofilename)
254 #
255 # Remove local files that are no longer in the remote directory
256 try:
257 if not localdir: names = os.listdir(os.curdir)
258 else: names = os.listdir(localdir)
259 except os.error:
260 names = []
261 for name in names:
262 if name[0] == '.' or info.has_key(name) or name in subdirs:
263 continue
264 skip = 0
265 for pat in skippats:
266 if fnmatch(name, pat):
267 if verbose > 1:
268 print 'Skip pattern', `pat`,
269 print 'matches', `name`
270 skip = 1
271 break
272 if skip:
273 continue
274 fullname = os.path.join(localdir, name)
275 if not rmok:
276 if verbose:
277 print 'Local file', `fullname`,
278 print 'is no longer pertinent'
279 continue
280 if verbose: print 'Removing local file/dir', `fullname`
281 remove(fullname)
282 #
283 # Recursively mirror subdirectories
284 for subdir in subdirs:
285 if interactive:
286 doit = askabout('subdirectory', subdir, pwd)
287 if not doit: continue
288 if verbose: print 'Processing subdirectory', `subdir`
289 localsubdir = os.path.join(localdir, subdir)
290 pwd = f.pwd()
291 if verbose > 1:
292 print 'Remote directory now:', `pwd`
293 print 'Remote cwd', `subdir`
294 try:
295 f.cwd(subdir)
296 except ftplib.error_perm, msg:
297 print "Can't chdir to", `subdir`, ":", `msg`
298 else:
299 if verbose: print 'Mirroring as', `localsubdir`
300 mirrorsubdir(f, localsubdir)
301 if verbose > 1: print 'Remote cwd ..'
302 f.cwd('..')
303 newpwd = f.pwd()
304 if newpwd != pwd:
305 print 'Ended up in wrong directory after cd + cd ..'
306 print 'Giving up now.'
307 break
308 else:
309 if verbose > 1: print 'OK.'
Guido van Rossume830e551995-06-20 19:31:37 +0000310
Guido van Rossumfdc5adc1998-01-04 02:03:12 +0000311# Helper to remove a file or directory tree
312def remove(fullname):
Tim Peters70c43782001-01-17 08:48:39 +0000313 if os.path.isdir(fullname) and not os.path.islink(fullname):
314 try:
315 names = os.listdir(fullname)
316 except os.error:
317 names = []
318 ok = 1
319 for name in names:
320 if not remove(os.path.join(fullname, name)):
321 ok = 0
322 if not ok:
323 return 0
324 try:
325 os.rmdir(fullname)
326 except os.error, msg:
327 print "Can't remove local directory %s: %s" % \
328 (`fullname`, str(msg))
329 return 0
330 else:
331 try:
332 os.unlink(fullname)
333 except os.error, msg:
334 print "Can't remove local file %s: %s" % \
335 (`fullname`, str(msg))
336 return 0
337 return 1
Guido van Rossumfdc5adc1998-01-04 02:03:12 +0000338
Guido van Rossume830e551995-06-20 19:31:37 +0000339# Wrapper around a file for writing to write a hash sign every block.
340class LoggingFile:
Tim Peters70c43782001-01-17 08:48:39 +0000341 def __init__(self, fp, blocksize, outfp):
342 self.fp = fp
343 self.bytes = 0
344 self.hashes = 0
345 self.blocksize = blocksize
346 self.outfp = outfp
347 def write(self, data):
348 self.bytes = self.bytes + len(data)
349 hashes = int(self.bytes) / self.blocksize
350 while hashes > self.hashes:
351 self.outfp.write('#')
352 self.outfp.flush()
353 self.hashes = self.hashes + 1
354 self.fp.write(data)
355 def close(self):
356 self.outfp.write('\n')
Guido van Rossume830e551995-06-20 19:31:37 +0000357
358# Ask permission to download a file.
359def askabout(filetype, filename, pwd):
Tim Peters70c43782001-01-17 08:48:39 +0000360 prompt = 'Retrieve %s %s from %s ? [ny] ' % (filetype, filename, pwd)
361 while 1:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000362 reply = raw_input(prompt).strip().lower()
Tim Peters70c43782001-01-17 08:48:39 +0000363 if reply in ['y', 'ye', 'yes']:
364 return 1
365 if reply in ['', 'n', 'no', 'nop', 'nope']:
366 return 0
367 print 'Please answer yes or no.'
Guido van Rossume830e551995-06-20 19:31:37 +0000368
369# Create a directory if it doesn't exist. Recursively create the
370# parent directory as well if needed.
371def makedir(pathname):
Tim Peters70c43782001-01-17 08:48:39 +0000372 if os.path.isdir(pathname):
373 return
374 dirname = os.path.dirname(pathname)
375 if dirname: makedir(dirname)
376 os.mkdir(pathname, 0777)
Guido van Rossume830e551995-06-20 19:31:37 +0000377
378# Write a dictionary to a file in a way that can be read back using
379# rval() but is still somewhat readable (i.e. not a single long line).
Guido van Rossuma2596961998-06-29 23:17:16 +0000380# Also creates a backup file.
Guido van Rossume830e551995-06-20 19:31:37 +0000381def writedict(dict, filename):
Tim Peters70c43782001-01-17 08:48:39 +0000382 dir, file = os.path.split(filename)
383 tempname = os.path.join(dir, '@' + file)
384 backup = os.path.join(dir, file + '~')
385 try:
386 os.unlink(backup)
387 except os.error:
388 pass
389 fp = open(tempname, 'w')
390 fp.write('{\n')
391 for key, value in dict.items():
392 fp.write('%s: %s,\n' % (`key`, `value`))
393 fp.write('}\n')
394 fp.close()
395 try:
396 os.rename(filename, backup)
397 except os.error:
398 pass
399 os.rename(tempname, filename)
Guido van Rossume830e551995-06-20 19:31:37 +0000400
Barry Warsaw64850ef2000-09-16 22:09:51 +0000401
402if __name__ == '__main__':
Tim Peters70c43782001-01-17 08:48:39 +0000403 main()