1
1
# Copyright (C) 2005 Robey Pointer <robey@lag.net>
2
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
2
# Copyright (C) 2005, 2006 Canonical Ltd
4
4
# This program is free software; you can redistribute it and/or modify
5
5
# it under the terms of the GNU General Public License as published by
6
6
# the Free Software Foundation; either version 2 of the License, or
7
7
# (at your option) any later version.
9
9
# This program is distributed in the hope that it will be useful,
10
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
12
# GNU General Public License for more details.
14
14
# You should have received a copy of the GNU General Public License
15
15
# along with this program; if not, write to the Free Software
16
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
18
18
"""Implementation of Transport over SFTP, using paramiko."""
20
# TODO: Remove the transport-based lock_read and lock_write methods. They'll
21
# then raise TransportNotPossible, which will break remote access to any
22
# formats which rely on OS-level locks. That should be fine as those formats
23
# are pretty old, but these combinations may have to be removed from the test
24
# suite. Those formats all date back to 0.7; so we should be able to remove
25
# these methods when we officially drop support for those formats.
46
from bzrlib.errors import (FileExists,
47
NoSuchFile, PathNotChild,
33
from bzrlib.config import config_dir, ensure_config_dir_exists
34
from bzrlib.errors import (ConnectionError,
36
TransportNotPossible, NoSuchFile, PathNotChild,
51
40
ParamikoNotPresent,
53
from bzrlib.osutils import pathjoin, fancy_rename, getcwd
54
from bzrlib.symbol_versioning import (
57
from bzrlib.trace import mutter, warning
42
from bzrlib.osutils import pathjoin, fancy_rename
43
from bzrlib.trace import mutter, warning, error
58
44
from bzrlib.transport import (
45
register_urlparse_netloc_protocol,
67
# Disable one particular warning that comes from paramiko in Python2.5; if
68
# this is emitted at the wrong time it tends to cause spurious test failures
69
# or at least noise in the test case::
71
# [1770/7639 in 86s, 1 known failures, 50 skipped, 2 missing features]
72
# test_permissions.TestSftpPermissions.test_new_files
73
# /var/lib/python-support/python2.5/paramiko/message.py:226: DeprecationWarning: integer argument expected, got float
74
# self.packet.write(struct.pack('>I', n))
75
warnings.filterwarnings('ignore',
76
'integer argument expected, got float',
77
category=DeprecationWarning,
78
module='paramiko.message')
51
import bzrlib.urlutils as urlutils
87
60
CMD_HANDLE, CMD_OPEN)
88
61
from paramiko.sftp_attr import SFTPAttributes
89
62
from paramiko.sftp_file import SFTPFile
92
_paramiko_version = getattr(paramiko, '__version_info__', (0, 0, 0))
93
# don't use prefetch unless paramiko version >= 1.5.5 (there were bugs earlier)
94
_default_do_prefetch = (_paramiko_version >= (1, 5, 5))
63
from paramiko.sftp_client import SFTPClient
66
register_urlparse_netloc_protocol('sftp')
70
# TODO: This should possibly ignore SIGHUP as well, but bzr currently
71
# doesn't handle it itself.
72
# <https://launchpad.net/products/bzr/+bug/41433/+index>
74
signal.signal(signal.SIGINT, signal.SIG_IGN)
77
def os_specific_subprocess_params():
78
"""Get O/S specific subprocess parameters."""
79
if sys.platform == 'win32':
80
# setting the process group and closing fds is not supported on
84
# We close fds other than the pipes as the child process does not need
87
# We also set the child process to ignore SIGINT. Normally the signal
88
# would be sent to every process in the foreground process group, but
89
# this causes it to be seen only by bzr and not by ssh. Python will
90
# generate a KeyboardInterrupt in bzr, and we will then have a chance
91
# to release locks or do other cleanup over ssh before the connection
93
# <https://launchpad.net/products/bzr/+bug/5987>
95
# Running it in a separate process group is not good because then it
96
# can't get non-echoed input of a password or passphrase.
97
# <https://launchpad.net/products/bzr/+bug/40508>
98
return {'preexec_fn': _ignore_sigint,
103
# don't use prefetch unless paramiko version >= 1.5.2 (there were bugs earlier)
104
_default_do_prefetch = False
105
if getattr(paramiko, '__version_info__', (0, 0, 0)) >= (1, 5, 5):
106
_default_do_prefetch = True
110
def _get_ssh_vendor():
111
"""Find out what version of SSH is on the system."""
113
if _ssh_vendor is not None:
118
if 'BZR_SSH' in os.environ:
119
_ssh_vendor = os.environ['BZR_SSH']
120
if _ssh_vendor == 'paramiko':
125
p = subprocess.Popen(['ssh', '-V'],
126
stdin=subprocess.PIPE,
127
stdout=subprocess.PIPE,
128
stderr=subprocess.PIPE,
129
**os_specific_subprocess_params())
130
returncode = p.returncode
131
stdout, stderr = p.communicate()
135
if 'OpenSSH' in stderr:
136
mutter('ssh implementation is OpenSSH')
137
_ssh_vendor = 'openssh'
138
elif 'SSH Secure Shell' in stderr:
139
mutter('ssh implementation is SSH Corp.')
142
if _ssh_vendor != 'none':
145
# XXX: 20051123 jamesh
146
# A check for putty's plink or lsh would go here.
148
mutter('falling back to paramiko implementation')
152
class SFTPSubprocess:
153
"""A socket-like object that talks to an ssh subprocess via pipes."""
154
def __init__(self, hostname, vendor, port=None, user=None):
155
assert vendor in ['openssh', 'ssh']
156
if vendor == 'openssh':
158
'-oForwardX11=no', '-oForwardAgent=no',
159
'-oClearAllForwardings=yes', '-oProtocol=2',
160
'-oNoHostAuthenticationForLocalhost=yes']
162
args.extend(['-p', str(port)])
164
args.extend(['-l', user])
165
args.extend(['-s', hostname, 'sftp'])
166
elif vendor == 'ssh':
169
args.extend(['-p', str(port)])
171
args.extend(['-l', user])
172
args.extend(['-s', 'sftp', hostname])
174
self.proc = subprocess.Popen(args,
175
stdin=subprocess.PIPE,
176
stdout=subprocess.PIPE,
177
**os_specific_subprocess_params())
179
def send(self, data):
180
return os.write(self.proc.stdin.fileno(), data)
182
def recv_ready(self):
183
# TODO: jam 20051215 this function is necessary to support the
184
# pipelined() function. In reality, it probably should use
185
# poll() or select() to actually return if there is data
186
# available, otherwise we probably don't get any benefit
189
def recv(self, count):
190
return os.read(self.proc.stdout.fileno(), count)
193
self.proc.stdin.close()
194
self.proc.stdout.close()
198
class LoopbackSFTP(object):
199
"""Simple wrapper for a socket that pretends to be a paramiko Channel."""
201
def __init__(self, sock):
204
def send(self, data):
205
return self.__socket.send(data)
208
return self.__socket.recv(n)
210
def recv_ready(self):
214
self.__socket.close()
220
# This is a weakref dictionary, so that we can reuse connections
221
# that are still active. Long term, it might be nice to have some
222
# sort of expiration policy, such as disconnect if inactive for
223
# X seconds. But that requires a lot more fanciness.
224
_connected_hosts = weakref.WeakValueDictionary()
226
def clear_connection_cache():
227
"""Remove all hosts from the SFTP connection cache.
229
Primarily useful for test cases wanting to force garbage collection.
231
_connected_hosts.clear()
234
def load_host_keys():
236
Load system host keys (probably doesn't work on windows) and any
237
"discovered" keys from previous sessions.
239
global SYSTEM_HOSTKEYS, BZR_HOSTKEYS
241
SYSTEM_HOSTKEYS = paramiko.util.load_host_keys(os.path.expanduser('~/.ssh/known_hosts'))
243
mutter('failed to load system host keys: ' + str(e))
244
bzr_hostkey_path = pathjoin(config_dir(), 'ssh_host_keys')
246
BZR_HOSTKEYS = paramiko.util.load_host_keys(bzr_hostkey_path)
248
mutter('failed to load bzr host keys: ' + str(e))
252
def save_host_keys():
254
Save "discovered" host keys in $(config)/ssh_host_keys/.
256
global SYSTEM_HOSTKEYS, BZR_HOSTKEYS
257
bzr_hostkey_path = pathjoin(config_dir(), 'ssh_host_keys')
258
ensure_config_dir_exists()
261
f = open(bzr_hostkey_path, 'w')
262
f.write('# SSH host keys collected by bzr\n')
263
for hostname, keys in BZR_HOSTKEYS.iteritems():
264
for keytype, key in keys.iteritems():
265
f.write('%s %s %s\n' % (hostname, keytype, key.get_base64()))
268
mutter('failed to save bzr host keys: ' + str(e))
97
271
class SFTPLock(object):
98
"""This fakes a lock in a remote location.
100
A present lock is indicated just by the existence of a file. This
101
doesn't work well on all transports and they are only used in
102
deprecated storage formats.
272
"""This fakes a lock in a remote location."""
105
273
__slots__ = ['path', 'lock_path', 'lock_file', 'transport']
107
274
def __init__(self, path, transport):
275
assert isinstance(transport, SFTPTransport)
108
277
self.lock_file = None
110
279
self.lock_path = path + '.write-lock'
133
302
# What specific errors should we catch here?
137
class _SFTPReadvHelper(object):
138
"""A class to help with managing the state of a readv request."""
140
# See _get_requests for an explanation.
141
_max_request_size = 32768
143
def __init__(self, original_offsets, relpath):
144
"""Create a new readv helper.
146
:param original_offsets: The original requests given by the caller of
148
:param relpath: The name of the file (if known)
150
self.original_offsets = list(original_offsets)
151
self.relpath = relpath
153
def _get_requests(self):
154
"""Break up the offsets into individual requests over sftp.
156
The SFTP spec only requires implementers to support 32kB requests. We
157
could try something larger (openssh supports 64kB), but then we have to
158
handle requests that fail.
159
So instead, we just break up our maximum chunks into 32kB chunks, and
160
asyncronously requests them.
161
Newer versions of paramiko would do the chunking for us, but we want to
162
start processing results right away, so we do it ourselves.
164
# TODO: Because we issue async requests, we don't 'fudge' any extra
165
# data. I'm not 100% sure that is the best choice.
167
# The first thing we do, is to collapse the individual requests as much
168
# as possible, so we don't issues requests <32kB
169
sorted_offsets = sorted(self.original_offsets)
170
coalesced = list(ConnectedTransport._coalesce_offsets(sorted_offsets,
171
limit=0, fudge_factor=0))
173
for c_offset in coalesced:
174
start = c_offset.start
175
size = c_offset.length
177
# Break this up into 32kB requests
179
next_size = min(size, self._max_request_size)
180
requests.append((start, next_size))
183
mutter('SFTP.readv(%s) %s offsets => %s coalesced => %s requests',
184
self.relpath, len(sorted_offsets), len(coalesced),
188
def request_and_yield_offsets(self, fp):
189
"""Request the data from the remote machine, yielding the results.
191
:param fp: A Paramiko SFTPFile object that supports readv.
192
:return: Yield the data requested by the original readv caller, one by
195
requests = self._get_requests()
196
offset_iter = iter(self.original_offsets)
197
cur_offset, cur_size = offset_iter.next()
198
# paramiko .readv() yields strings that are in the order of the requests
199
# So we track the current request to know where the next data is
200
# being returned from.
206
# This is used to buffer chunks which we couldn't process yet
207
# It is (start, end, data) tuples.
209
# Create an 'unlimited' data stream, so we stop based on requests,
210
# rather than just because the data stream ended. This lets us detect
212
data_stream = itertools.chain(fp.readv(requests),
213
itertools.repeat(None))
214
for (start, length), data in itertools.izip(requests, data_stream):
216
if cur_coalesced is not None:
217
raise errors.ShortReadvError(self.relpath,
218
start, length, len(data))
219
if len(data) != length:
220
raise errors.ShortReadvError(self.relpath,
221
start, length, len(data))
223
# This is the first request, just buffer it
224
buffered_data = [data]
225
buffered_len = length
227
elif start == last_end:
228
# The data we are reading fits neatly on the previous
229
# buffer, so this is all part of a larger coalesced range.
230
buffered_data.append(data)
231
buffered_len += length
233
# We have an 'interrupt' in the data stream. So we know we are
234
# at a request boundary.
236
# We haven't consumed the buffer so far, so put it into
237
# data_chunks, and continue.
238
buffered = ''.join(buffered_data)
239
data_chunks.append((input_start, buffered))
241
buffered_data = [data]
242
buffered_len = length
243
last_end = start + length
244
if input_start == cur_offset and cur_size <= buffered_len:
245
# Simplify the next steps a bit by transforming buffered_data
246
# into a single string. We also have the nice property that
247
# when there is only one string ''.join([x]) == x, so there is
249
buffered = ''.join(buffered_data)
250
# Clean out buffered data so that we keep memory
254
# TODO: We *could* also consider the case where cur_offset is in
255
# in the buffered range, even though it doesn't *start*
256
# the buffered range. But for packs we pretty much always
257
# read in order, so you won't get any extra data in the
259
while (input_start == cur_offset
260
and (buffered_offset + cur_size) <= buffered_len):
261
# We've buffered enough data to process this request, spit it
263
cur_data = buffered[buffered_offset:buffered_offset + cur_size]
264
# move the direct pointer into our buffered data
265
buffered_offset += cur_size
266
# Move the start-of-buffer pointer
267
input_start += cur_size
268
# Yield the requested data
269
yield cur_offset, cur_data
270
cur_offset, cur_size = offset_iter.next()
271
# at this point, we've consumed as much of buffered as we can,
272
# so break off the portion that we consumed
273
if buffered_offset == len(buffered_data):
274
# No tail to leave behind
278
buffered = buffered[buffered_offset:]
279
buffered_data = [buffered]
280
buffered_len = len(buffered)
282
buffered = ''.join(buffered_data)
284
data_chunks.append((input_start, buffered))
286
mutter('SFTP readv left with %d out-of-order bytes',
287
sum(map(lambda x: len(x[1]), data_chunks)))
288
# We've processed all the readv data, at this point, anything we
289
# couldn't process is in data_chunks. This doesn't happen often, so
290
# this code path isn't optimized
291
# We use an interesting process for data_chunks
292
# Specifically if we have "bisect_left([(start, len, entries)],
294
# If start == qstart, then we get the specific node. Otherwise we
295
# get the previous node
297
idx = bisect.bisect_left(data_chunks, (cur_offset,))
298
if data_chunks[idx][0] == cur_offset: # The data starts here
299
data = data_chunks[idx][1][:cur_size]
301
# The data is in a portion of a previous page
303
sub_offset = cur_offset - data_chunks[idx][0]
304
data = data_chunks[idx][1]
305
data = data[sub_offset:sub_offset + cur_size]
307
# We are missing the page where the data should be found,
310
if len(data) != cur_size:
311
raise AssertionError('We must have miscalulated.'
312
' We expected %d bytes, but only found %d'
313
% (cur_size, len(data)))
314
yield cur_offset, data
315
cur_offset, cur_size = offset_iter.next()
318
class SFTPTransport(ConnectedTransport):
319
"""Transport implementation for SFTP access."""
305
class SFTPTransport (Transport):
307
Transport implementation for SFTP access.
321
309
_do_prefetch = _default_do_prefetch
322
# TODO: jam 20060717 Conceivably these could be configurable, either
323
# by auto-tuning at run-time, or by a configuration (per host??)
324
# but the performance curve is pretty flat, so just going with
325
# reasonable defaults.
326
_max_readv_combine = 200
327
# Having to round trip to the server means waiting for a response,
328
# so it is better to download extra bytes.
329
# 8KiB had good performance for both local and remote network operations
330
_bytes_to_read_before_seek = 8192
332
# The sftp spec says that implementations SHOULD allow reads
333
# to be at least 32K. paramiko.readv() does an async request
334
# for the chunks. So we need to keep it within a single request
335
# size for paramiko <= 1.6.1. paramiko 1.6.2 will probably chop
336
# up the request itself, rather than us having to worry about it
337
_max_request_size = 32768
339
def __init__(self, base, _from_transport=None):
340
super(SFTPTransport, self).__init__(base,
341
_from_transport=_from_transport)
311
def __init__(self, base, clone_from=None):
312
assert base.startswith('sftp://')
313
self._parse_url(base)
314
base = self._unparse_url()
317
super(SFTPTransport, self).__init__(base)
318
if clone_from is None:
321
# use the same ssh connection, etc
322
self._sftp = clone_from._sftp
323
# super saves 'self.base'
325
def should_cache(self):
327
Return True if the data pulled across should be cached locally.
331
def clone(self, offset=None):
333
Return a new SFTPTransport with root at self.base + offset.
334
We share the same SFTP session between such transports, because it's
335
fairly expensive to set them up.
338
return SFTPTransport(self.base, self)
340
return SFTPTransport(self.abspath(offset), self)
342
def abspath(self, relpath):
344
Return the full url to the given relative path.
346
@param relpath: the relative path or path components
347
@type relpath: str or list
349
return self._unparse_url(self._remote_path(relpath))
343
351
def _remote_path(self, relpath):
344
352
"""Return the path to be passed along the sftp protocol for relpath.
346
:param relpath: is a urlencoded string.
348
relative = urlutils.unescape(relpath).encode('utf-8')
349
remote_path = self._combine_paths(self._path, relative)
350
# the initial slash should be removed from the path, and treated as a
351
# homedir relative path (the path begins with a double slash if it is
352
# absolute). see draft-ietf-secsh-scp-sftp-ssh-uri-03.txt
353
# RBC 20060118 we are not using this as its too user hostile. instead
354
# we are following lftp and using /~/foo to mean '~/foo'
355
# vila--20070602 and leave absolute paths begin with a single slash.
356
if remote_path.startswith('/~/'):
357
remote_path = remote_path[3:]
358
elif remote_path == '/~':
362
def _create_connection(self, credentials=None):
363
"""Create a new connection with the provided credentials.
365
:param credentials: The credentials needed to establish the connection.
367
:return: The created connection and its associated credentials.
369
The credentials are only the password as it may have been entered
370
interactively by the user and may be different from the one provided
371
in base url at transport creation time.
373
if credentials is None:
374
password = self._password
376
password = credentials
378
vendor = ssh._get_ssh_vendor()
381
auth = config.AuthenticationConfig()
382
user = auth.get_user('ssh', self._host, self._port)
383
connection = vendor.connect_sftp(self._user, password,
384
self._host, self._port)
385
return connection, (user, password)
388
"""Ensures that a connection is established"""
389
connection = self._get_connection()
390
if connection is None:
391
# First connection ever
392
connection, credentials = self._create_connection()
393
self._set_connection(connection, credentials)
354
relpath is a urlencoded string.
356
# FIXME: share the common code across transports
357
assert isinstance(relpath, basestring)
358
relpath = urlutils.unescape(relpath).split('/')
359
basepath = self._path.split('/')
360
if len(basepath) > 0 and basepath[-1] == '':
361
basepath = basepath[:-1]
365
if len(basepath) == 0:
366
# In most filesystems, a request for the parent
367
# of root, just returns root.
375
path = '/'.join(basepath)
378
def relpath(self, abspath):
379
username, password, host, port, path = self._split_url(abspath)
381
if (username != self._username):
382
error.append('username mismatch')
383
if (host != self._host):
384
error.append('host mismatch')
385
if (port != self._port):
386
error.append('port mismatch')
387
if (not path.startswith(self._path)):
388
error.append('path mismatch')
390
extra = ': ' + ', '.join(error)
391
raise PathNotChild(abspath, self.base, extra=extra)
393
return path[pl:].strip('/')
396
395
def has(self, relpath):
398
397
Does the target location exist?
401
self._get_sftp().stat(self._remote_path(relpath))
400
self._sftp.stat(self._remote_path(relpath))
413
412
path = self._remote_path(relpath)
414
f = self._get_sftp().file(path, mode='rb')
413
f = self._sftp.file(path, mode='rb')
415
414
if self._do_prefetch and (getattr(f, 'prefetch', None) is not None):
418
417
except (IOError, paramiko.SSHException), e:
419
self._translate_io_exception(e, path, ': error retrieving',
420
failure_exc=errors.ReadError)
422
def _readv(self, relpath, offsets):
423
"""See Transport.readv()"""
424
# We overload the default readv() because we want to use a file
425
# that does not have prefetch enabled.
426
# Also, if we have a new paramiko, it implements an async readv()
431
path = self._remote_path(relpath)
432
fp = self._get_sftp().file(path, mode='rb')
433
readv = getattr(fp, 'readv', None)
435
return self._sftp_readv(fp, offsets, relpath)
436
mutter('seek and read %s offsets', len(offsets))
437
return self._seek_and_read(fp, offsets, relpath)
438
except (IOError, paramiko.SSHException), e:
439
418
self._translate_io_exception(e, path, ': error retrieving')
441
def recommended_page_size(self):
442
"""See Transport.recommended_page_size().
444
For SFTP we suggest a large page size to reduce the overhead
445
introduced by latency.
449
def _sftp_readv(self, fp, offsets, relpath='<unknown>'):
450
"""Use the readv() member of fp to do async readv.
452
And then read them using paramiko.readv(). paramiko.readv()
453
does not support ranges > 64K, so it caps the request size, and
454
just reads until it gets all the stuff it wants
456
helper = _SFTPReadvHelper(offsets, relpath)
457
return helper.request_and_yield_offsets(fp)
459
def put_file(self, relpath, f, mode=None):
461
Copy the file-like object into the location.
420
def get_partial(self, relpath, start, length=None):
422
Get just part of a file.
424
:param relpath: Path to the file, relative to base
425
:param start: The starting position to read from
426
:param length: The length to read. A length of None indicates
427
read to the end of the file.
428
:return: A file-like object containing at least the specified bytes.
429
Some implementations may return objects which can be read
430
past this length, but this is not guaranteed.
432
# TODO: implement get_partial_multi to help with knit support
433
f = self.get(relpath)
435
if self._do_prefetch and hasattr(f, 'prefetch'):
439
def put(self, relpath, f, mode=None):
441
Copy the file-like or string object into the location.
463
443
:param relpath: Location to put the contents, relative to base.
464
:param f: File-like object.
444
:param f: File-like or string object.
465
445
:param mode: The final mode for the file
467
447
final_path = self._remote_path(relpath)
468
return self._put(final_path, f, mode=mode)
448
self._put(final_path, f, mode=mode)
470
450
def _put(self, abspath, f, mode=None):
471
451
"""Helper function so both put() and copy_abspaths can reuse the code"""
512
self._get_sftp().remove(tmp_abspath)
478
self._sftp.remove(tmp_abspath)
514
480
# raise the saved except
516
482
# raise the original with its traceback if we can.
519
def _put_non_atomic_helper(self, relpath, writer, mode=None,
520
create_parent_dir=False,
522
abspath = self._remote_path(relpath)
524
# TODO: jam 20060816 paramiko doesn't publicly expose a way to
525
# set the file mode at create time. If it does, use it.
526
# But for now, we just chmod later anyway.
528
def _open_and_write_file():
529
"""Try to open the target file, raise error on failure"""
533
fout = self._get_sftp().file(abspath, mode='wb')
534
fout.set_pipelined(True)
536
except (paramiko.SSHException, IOError), e:
537
self._translate_io_exception(e, abspath,
540
# This is designed to chmod() right before we close.
541
# Because we set_pipelined() earlier, theoretically we might
542
# avoid the round trip for fout.close()
544
self._get_sftp().chmod(abspath, mode)
549
if not create_parent_dir:
550
_open_and_write_file()
553
# Try error handling to create the parent directory if we need to
555
_open_and_write_file()
557
# Try to create the parent directory, and then go back to
559
parent_dir = os.path.dirname(abspath)
560
self._mkdir(parent_dir, dir_mode)
561
_open_and_write_file()
563
def put_file_non_atomic(self, relpath, f, mode=None,
564
create_parent_dir=False,
566
"""Copy the file-like object into the target location.
568
This function is not strictly safe to use. It is only meant to
569
be used when you already know that the target does not exist.
570
It is not safe, because it will open and truncate the remote
571
file. So there may be a time when the file has invalid contents.
573
:param relpath: The remote location to put the contents.
574
:param f: File-like object.
575
:param mode: Possible access permissions for new file.
576
None means do not set remote permissions.
577
:param create_parent_dir: If we cannot create the target file because
578
the parent directory does not exist, go ahead and
579
create it, and then try again.
583
self._put_non_atomic_helper(relpath, writer, mode=mode,
584
create_parent_dir=create_parent_dir,
587
def put_bytes_non_atomic(self, relpath, bytes, mode=None,
588
create_parent_dir=False,
592
self._put_non_atomic_helper(relpath, writer, mode=mode,
593
create_parent_dir=create_parent_dir,
596
485
def iter_files_recursive(self):
597
486
"""Walk the relative paths of all files in this transport."""
598
487
queue = list(self.list_dir('.'))
600
relpath = queue.pop(0)
489
relpath = urllib.quote(queue.pop(0))
601
490
st = self.stat(relpath)
602
491
if stat.S_ISDIR(st.st_mode):
603
492
for i, basename in enumerate(self.list_dir(relpath)):
608
def _mkdir(self, abspath, mode=None):
614
self._get_sftp().mkdir(abspath, local_mode)
616
# chmod a dir through sftp will erase any sgid bit set
617
# on the server side. So, if the bit mode are already
618
# set, avoid the chmod. If the mode is not fine but
619
# the sgid bit is set, report a warning to the user
620
# with the umask fix.
621
stat = self._get_sftp().lstat(abspath)
622
mode = mode & 0777 # can't set special bits anyway
623
if mode != stat.st_mode & 0777:
624
if stat.st_mode & 06000:
625
warning('About to chmod %s over sftp, which will result'
626
' in its suid or sgid bits being cleared. If'
627
' you want to preserve those bits, change your '
628
' environment on the server to use umask 0%03o.'
629
% (abspath, 0777 - mode))
630
self._get_sftp().chmod(abspath, mode=mode)
631
except (paramiko.SSHException, IOError), e:
632
self._translate_io_exception(e, abspath, ': unable to mkdir',
633
failure_exc=FileExists)
635
497
def mkdir(self, relpath, mode=None):
636
498
"""Create a directory at the given path."""
637
self._mkdir(self._remote_path(relpath), mode=mode)
639
def open_write_stream(self, relpath, mode=None):
640
"""See Transport.open_write_stream."""
641
# initialise the file to zero-length
642
# this is three round trips, but we don't use this
643
# api more than once per write_group at the moment so
644
# it is a tolerable overhead. Better would be to truncate
645
# the file after opening. RBC 20070805
646
self.put_bytes_non_atomic(relpath, "", mode)
647
abspath = self._remote_path(relpath)
648
# TODO: jam 20060816 paramiko doesn't publicly expose a way to
649
# set the file mode at create time. If it does, use it.
650
# But for now, we just chmod later anyway.
653
handle = self._get_sftp().file(abspath, mode='wb')
654
handle.set_pipelined(True)
500
path = self._remote_path(relpath)
501
# In the paramiko documentation, it says that passing a mode flag
502
# will filtered against the server umask.
503
# StubSFTPServer does not do this, which would be nice, because it is
504
# what we really want :)
505
# However, real servers do use umask, so we really should do it that way
506
self._sftp.mkdir(path)
508
self._sftp.chmod(path, mode=mode)
655
509
except (paramiko.SSHException, IOError), e:
656
self._translate_io_exception(e, abspath,
658
_file_streams[self.abspath(relpath)] = handle
659
return FileFileStream(self, relpath, handle)
510
self._translate_io_exception(e, path, ': unable to mkdir',
511
failure_exc=FileExists)
661
def _translate_io_exception(self, e, path, more_info='',
513
def _translate_io_exception(self, e, path, more_info='',
662
514
failure_exc=PathError):
663
515
"""Translate a paramiko or IOError into a friendlier exception.
808
648
# that we have taken the lock.
809
649
return SFTPLock(relpath, self)
651
def _unparse_url(self, path=None):
654
path = urllib.quote(path)
655
# handle homedir paths
656
if not path.startswith('/'):
658
netloc = urllib.quote(self._host)
659
if self._username is not None:
660
netloc = '%s@%s' % (urllib.quote(self._username), netloc)
661
if self._port is not None:
662
netloc = '%s:%d' % (netloc, self._port)
663
return urlparse.urlunparse(('sftp', netloc, path, '', '', ''))
665
def _split_url(self, url):
666
(scheme, username, password, host, port, path) = split_url(url)
667
assert scheme == 'sftp'
669
# the initial slash should be removed from the path, and treated
670
# as a homedir relative path (the path begins with a double slash
671
# if it is absolute).
672
# see draft-ietf-secsh-scp-sftp-ssh-uri-03.txt
673
# RBC 20060118 we are not using this as its too user hostile. instead
674
# we are following lftp and using /~/foo to mean '~/foo'.
675
# handle homedir paths
676
if path.startswith('/~/'):
680
return (username, password, host, port, path)
682
def _parse_url(self, url):
683
(self._username, self._password,
684
self._host, self._port, self._path) = self._split_url(url)
686
def _sftp_connect(self):
687
"""Connect to the remote sftp server.
688
After this, self._sftp should have a valid connection (or
689
we raise an TransportError 'could not connect').
691
TODO: Raise a more reasonable ConnectionFailed exception
693
global _connected_hosts
695
idx = (self._host, self._port, self._username)
697
self._sftp = _connected_hosts[idx]
702
vendor = _get_ssh_vendor()
703
if vendor == 'loopback':
704
sock = socket.socket()
705
sock.connect((self._host, self._port))
706
self._sftp = SFTPClient(LoopbackSFTP(sock))
707
elif vendor != 'none':
708
sock = SFTPSubprocess(self._host, vendor, self._port,
710
self._sftp = SFTPClient(sock)
712
self._paramiko_connect()
714
_connected_hosts[idx] = self._sftp
716
def _paramiko_connect(self):
717
global SYSTEM_HOSTKEYS, BZR_HOSTKEYS
722
t = paramiko.Transport((self._host, self._port or 22))
723
t.set_log_channel('bzr.paramiko')
725
except paramiko.SSHException, e:
726
raise ConnectionError('Unable to reach SSH host %s:%d' %
727
(self._host, self._port), e)
729
server_key = t.get_remote_server_key()
730
server_key_hex = paramiko.util.hexify(server_key.get_fingerprint())
731
keytype = server_key.get_name()
732
if SYSTEM_HOSTKEYS.has_key(self._host) and SYSTEM_HOSTKEYS[self._host].has_key(keytype):
733
our_server_key = SYSTEM_HOSTKEYS[self._host][keytype]
734
our_server_key_hex = paramiko.util.hexify(our_server_key.get_fingerprint())
735
elif BZR_HOSTKEYS.has_key(self._host) and BZR_HOSTKEYS[self._host].has_key(keytype):
736
our_server_key = BZR_HOSTKEYS[self._host][keytype]
737
our_server_key_hex = paramiko.util.hexify(our_server_key.get_fingerprint())
739
warning('Adding %s host key for %s: %s' % (keytype, self._host, server_key_hex))
740
if not BZR_HOSTKEYS.has_key(self._host):
741
BZR_HOSTKEYS[self._host] = {}
742
BZR_HOSTKEYS[self._host][keytype] = server_key
743
our_server_key = server_key
744
our_server_key_hex = paramiko.util.hexify(our_server_key.get_fingerprint())
746
if server_key != our_server_key:
747
filename1 = os.path.expanduser('~/.ssh/known_hosts')
748
filename2 = pathjoin(config_dir(), 'ssh_host_keys')
749
raise TransportError('Host keys for %s do not match! %s != %s' % \
750
(self._host, our_server_key_hex, server_key_hex),
751
['Try editing %s or %s' % (filename1, filename2)])
756
self._sftp = t.open_sftp_client()
757
except paramiko.SSHException, e:
758
raise ConnectionError('Unable to start sftp client %s:%d' %
759
(self._host, self._port), e)
761
def _sftp_auth(self, transport):
762
# paramiko requires a username, but it might be none if nothing was supplied
763
# use the local username, just in case.
764
# We don't override self._username, because if we aren't using paramiko,
765
# the username might be specified in ~/.ssh/config and we don't want to
766
# force it to something else
767
# Also, it would mess up the self.relpath() functionality
768
username = self._username or getpass.getuser()
770
# Paramiko tries to open a socket.AF_UNIX in order to connect
771
# to ssh-agent. That attribute doesn't exist on win32 (it does in cygwin)
772
# so we get an AttributeError exception. For now, just don't try to
773
# connect to an agent if we are on win32
774
if sys.platform != 'win32':
775
agent = paramiko.Agent()
776
for key in agent.get_keys():
777
mutter('Trying SSH agent key %s' % paramiko.util.hexify(key.get_fingerprint()))
779
transport.auth_publickey(username, key)
781
except paramiko.SSHException, e:
784
# okay, try finding id_rsa or id_dss? (posix only)
785
if self._try_pkey_auth(transport, paramiko.RSAKey, username, 'id_rsa'):
787
if self._try_pkey_auth(transport, paramiko.DSSKey, username, 'id_dsa'):
792
transport.auth_password(username, self._password)
794
except paramiko.SSHException, e:
797
# FIXME: Don't keep a password held in memory if you can help it
798
#self._password = None
800
# give up and ask for a password
801
password = bzrlib.ui.ui_factory.get_password(
802
prompt='SSH %(user)s@%(host)s password',
803
user=username, host=self._host)
805
transport.auth_password(username, password)
806
except paramiko.SSHException, e:
807
raise ConnectionError('Unable to authenticate to SSH host as %s@%s' %
808
(username, self._host), e)
810
def _try_pkey_auth(self, transport, pkey_class, username, filename):
811
filename = os.path.expanduser('~/.ssh/' + filename)
813
key = pkey_class.from_private_key_file(filename)
814
transport.auth_publickey(username, key)
816
except paramiko.PasswordRequiredException:
817
password = bzrlib.ui.ui_factory.get_password(
818
prompt='SSH %(filename)s password',
821
key = pkey_class.from_private_key_file(filename, password)
822
transport.auth_publickey(username, key)
824
except paramiko.SSHException:
825
mutter('SSH authentication via %s key failed.' % (os.path.basename(filename),))
826
except paramiko.SSHException:
827
mutter('SSH authentication via %s key failed.' % (os.path.basename(filename),))
811
832
def _sftp_open_exclusive(self, abspath, mode=None):
812
833
"""Open a remote path exclusively.
884
895
self._socket.bind(('localhost', 0))
885
896
self._socket.listen(1)
886
897
self.port = self._socket.getsockname()[1]
887
self._stop_event = threading.Event()
898
self.stop_event = threading.Event()
901
s, _ = self._socket.accept()
902
# now close the listen socket
905
self._callback(s, self.stop_event)
907
pass #Ignore socket errors
909
# probably a failed test
910
warning('Exception from within unit test server thread: %r' % x)
890
# called from outside this thread
891
self._stop_event.set()
913
self.stop_event.set()
892
914
# use a timeout here, because if the test fails, the server thread may
893
915
# never notice the stop_event.
899
readable, writable_unused, exception_unused = \
900
select.select([self._socket], [], [], 0.1)
901
if self._stop_event.isSet():
903
if len(readable) == 0:
906
s, addr_unused = self._socket.accept()
907
# because the loopback socket is inline, and transports are
908
# never explicitly closed, best to launch a new thread.
909
threading.Thread(target=self._callback, args=(s,)).start()
910
except socket.error, x:
911
sys.excepthook(*sys.exc_info())
912
warning('Socket error during accept() within unit test server'
915
# probably a failed test; unit test thread will log the
917
sys.excepthook(*sys.exc_info())
918
warning('Exception from within unit test server thread: %r' %
922
class SocketDelay(object):
923
"""A socket decorator to make TCP appear slower.
925
This changes recv, send, and sendall to add a fixed latency to each python
926
call if a new roundtrip is detected. That is, when a recv is called and the
927
flag new_roundtrip is set, latency is charged. Every send and send_all
930
In addition every send, sendall and recv sleeps a bit per character send to
933
Not all methods are implemented, this is deliberate as this class is not a
934
replacement for the builtin sockets layer. fileno is not implemented to
935
prevent the proxy being bypassed.
939
_proxied_arguments = dict.fromkeys([
940
"close", "getpeername", "getsockname", "getsockopt", "gettimeout",
941
"setblocking", "setsockopt", "settimeout", "shutdown"])
943
def __init__(self, sock, latency, bandwidth=1.0,
946
:param bandwith: simulated bandwith (MegaBit)
947
:param really_sleep: If set to false, the SocketDelay will just
948
increase a counter, instead of calling time.sleep. This is useful for
949
unittesting the SocketDelay.
952
self.latency = latency
953
self.really_sleep = really_sleep
954
self.time_per_byte = 1 / (bandwidth / 8.0 * 1024 * 1024)
955
self.new_roundtrip = False
958
if self.really_sleep:
961
SocketDelay.simulated_time += s
963
def __getattr__(self, attr):
964
if attr in SocketDelay._proxied_arguments:
965
return getattr(self.sock, attr)
966
raise AttributeError("'SocketDelay' object has no attribute %r" %
970
return SocketDelay(self.sock.dup(), self.latency, self.time_per_byte,
973
def recv(self, *args):
974
data = self.sock.recv(*args)
975
if data and self.new_roundtrip:
976
self.new_roundtrip = False
977
self.sleep(self.latency)
978
self.sleep(len(data) * self.time_per_byte)
981
def sendall(self, data, flags=0):
982
if not self.new_roundtrip:
983
self.new_roundtrip = True
984
self.sleep(self.latency)
985
self.sleep(len(data) * self.time_per_byte)
986
return self.sock.sendall(data, flags)
988
def send(self, data, flags=0):
989
if not self.new_roundtrip:
990
self.new_roundtrip = True
991
self.sleep(self.latency)
992
bytes_sent = self.sock.send(data, flags)
993
self.sleep(bytes_sent * self.time_per_byte)
997
919
class SFTPServer(Server):
998
920
"""Common code for SFTP server facilities."""
1000
def __init__(self, server_interface=StubServer):
1001
923
self._original_vendor = None
1002
924
self._homedir = None
1003
925
self._server_homedir = None
1004
926
self._listener = None
1005
927
self._root = None
1006
self._vendor = ssh.ParamikoVendor()
1007
self._server_interface = server_interface
928
self._vendor = 'none'
1008
929
# sftp server logs
1010
self.add_latency = 0
1012
932
def _get_sftp_url(self, path):
1013
933
"""Calculate an sftp url to this server for path."""
1017
937
"""StubServer uses this to log when a new server is created."""
1018
938
self.logs.append(message)
1020
def _run_server_entry(self, sock):
1021
"""Entry point for all implementations of _run_server.
1023
If self.add_latency is > 0.000001 then sock is given a latency adding
1026
if self.add_latency > 0.000001:
1027
sock = SocketDelay(sock, self.add_latency)
1028
return self._run_server(sock)
1030
def _run_server(self, s):
940
def _run_server(self, s, stop_event):
1031
941
ssh_server = paramiko.Transport(s)
1032
key_file = pathjoin(self._homedir, 'test_rsa.key')
1033
f = open(key_file, 'w')
1034
f.write(STUB_SERVER_KEY)
942
key_file = os.path.join(self._homedir, 'test_rsa.key')
943
file(key_file, 'w').write(STUB_SERVER_KEY)
1036
944
host_key = paramiko.RSAKey.from_private_key_file(key_file)
1037
945
ssh_server.add_server_key(host_key)
1038
server = self._server_interface(self)
946
server = StubServer(self)
1039
947
ssh_server.set_subsystem_handler('sftp', paramiko.SFTPServer,
1040
948
StubSFTPServer, root=self._root,
1041
949
home=self._server_homedir)
1042
950
event = threading.Event()
1043
951
ssh_server.start_server(event, server)
953
stop_event.wait(30.0)
1046
def setUp(self, backing_server=None):
1047
# XXX: TODO: make sftpserver back onto backing_server rather than local
1049
if not (backing_server is None or
1050
isinstance(backing_server, local.LocalURLServer)):
1051
raise AssertionError(
1052
"backing_server should not be %r, because this can only serve the "
1053
"local current working directory." % (backing_server,))
1054
self._original_vendor = ssh._ssh_vendor_manager._cached_ssh_vendor
1055
ssh._ssh_vendor_manager._cached_ssh_vendor = self._vendor
1056
if sys.platform == 'win32':
1057
# Win32 needs to use the UNICODE api
1058
self._homedir = getcwd()
1060
# But Linux SFTP servers should just deal in bytestreams
1061
self._homedir = os.getcwd()
957
self._original_vendor = _ssh_vendor
958
_ssh_vendor = self._vendor
959
self._homedir = os.getcwdu()
1062
960
if self._server_homedir is None:
1063
961
self._server_homedir = self._homedir
1064
962
self._root = '/'
1065
if sys.platform == 'win32':
1067
self._listener = SocketListener(self._run_server_entry)
963
# FIXME WINDOWS: _root should be _server_homedir[0]:/
964
self._listener = SingleListener(self._run_server)
1068
965
self._listener.setDaemon(True)
1069
966
self._listener.start()
1071
968
def tearDown(self):
1072
969
"""See bzrlib.transport.Server.tearDown."""
1073
971
self._listener.stop()
1074
ssh._ssh_vendor_manager._cached_ssh_vendor = self._original_vendor
1076
def get_bogus_url(self):
1077
"""See bzrlib.transport.Server.get_bogus_url."""
1078
# this is chosen to try to prevent trouble with proxies, wierd dns, etc
1079
# we bind a random socket, so that we get a guaranteed unused port
1080
# we just never listen on that port
1082
s.bind(('localhost', 0))
1083
return 'sftp://%s:%s/' % s.getsockname()
972
_ssh_vendor = self._original_vendor
1086
975
class SFTPFullAbsoluteServer(SFTPServer):
added
removed
bzrlib/transport/sftp.py
