1
# Copyright (C) 2005-2010 Canonical Ltd
1
# Copyright (C) 2005 Robey Pointer <robey@lag.net>, Canonical Ltd
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
11
# GNU General Public License for more details.
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
17
"""Implementation of Transport over SFTP, using paramiko."""
19
# TODO: Remove the transport-based lock_read and lock_write methods. They'll
20
# then raise TransportNotPossible, which will break remote access to any
21
# formats which rely on OS-level locks. That should be fine as those formats
22
# are pretty old, but these combinations may have to be removed from the test
23
# suite. Those formats all date back to 0.7; so we should be able to remove
24
# these methods when we officially drop support for those formats.
44
from bzrlib.errors import (FileExists,
45
NoSuchFile, PathNotChild,
32
from bzrlib.errors import (FileExists,
33
TransportNotPossible, NoSuchFile, NonRelativePath,
51
from bzrlib.osutils import pathjoin, fancy_rename, getcwd
52
from bzrlib.symbol_versioning import (
55
from bzrlib.trace import mutter, warning
56
from bzrlib.transport import (
65
# Disable one particular warning that comes from paramiko in Python2.5; if
66
# this is emitted at the wrong time it tends to cause spurious test failures
67
# or at least noise in the test case::
69
# [1770/7639 in 86s, 1 known failures, 50 skipped, 2 missing features]
70
# test_permissions.TestSftpPermissions.test_new_files
71
# /var/lib/python-support/python2.5/paramiko/message.py:226: DeprecationWarning: integer argument expected, got float
72
# self.packet.write(struct.pack('>I', n))
73
warnings.filterwarnings('ignore',
74
'integer argument expected, got float',
75
category=DeprecationWarning,
76
module='paramiko.message')
36
from bzrlib.config import config_dir
37
from bzrlib.trace import mutter, warning, error
38
from bzrlib.transport import Transport, register_transport
80
except ImportError, e:
81
raise ParamikoNotPresent(e)
43
error('The SFTP transport requires paramiko.')
83
46
from paramiko.sftp import (SFTP_FLAG_WRITE, SFTP_FLAG_CREATE,
84
47
SFTP_FLAG_EXCL, SFTP_FLAG_TRUNC,
85
SFTP_OK, CMD_HANDLE, CMD_OPEN)
86
49
from paramiko.sftp_attr import SFTPAttributes
87
50
from paramiko.sftp_file import SFTPFile
90
_paramiko_version = getattr(paramiko, '__version_info__', (0, 0, 0))
91
# don't use prefetch unless paramiko version >= 1.5.5 (there were bugs earlier)
92
_default_do_prefetch = (_paramiko_version >= (1, 5, 5))
51
from paramiko.sftp_client import SFTPClient
53
if 'sftp' not in urlparse.uses_netloc: urlparse.uses_netloc.append('sftp')
57
def _get_ssh_vendor():
58
"""Find out what version of SSH is on the system."""
60
if _ssh_vendor is not None:
66
p = subprocess.Popen(['ssh', '-V'],
68
stdin=subprocess.PIPE,
69
stdout=subprocess.PIPE,
70
stderr=subprocess.PIPE)
71
returncode = p.returncode
72
stdout, stderr = p.communicate()
76
if 'OpenSSH' in stderr:
77
mutter('ssh implementation is OpenSSH')
78
_ssh_vendor = 'openssh'
79
elif 'SSH Secure Shell' in stderr:
80
mutter('ssh implementation is SSH Corp.')
83
if _ssh_vendor != 'none':
86
# XXX: 20051123 jamesh
87
# A check for putty's plink or lsh would go here.
89
mutter('falling back to paramiko implementation')
94
"""A socket-like object that talks to an ssh subprocess via pipes."""
95
def __init__(self, hostname, port=None, user=None):
96
vendor = _get_ssh_vendor()
97
assert vendor in ['openssh', 'ssh']
98
if vendor == 'openssh':
100
'-oForwardX11=no', '-oForwardAgent=no',
101
'-oClearAllForwardings=yes', '-oProtocol=2',
102
'-oNoHostAuthenticationForLocalhost=yes']
104
args.extend(['-p', str(port)])
106
args.extend(['-l', user])
107
args.extend(['-s', hostname, 'sftp'])
108
elif vendor == 'ssh':
111
args.extend(['-p', str(port)])
113
args.extend(['-l', user])
114
args.extend(['-s', 'sftp', hostname])
116
self.proc = subprocess.Popen(args, close_fds=True,
117
stdin=subprocess.PIPE,
118
stdout=subprocess.PIPE)
120
def send(self, data):
121
return os.write(self.proc.stdin.fileno(), data)
123
def recv(self, count):
124
return os.read(self.proc.stdout.fileno(), count)
127
self.proc.stdin.close()
128
self.proc.stdout.close()
135
# This is a weakref dictionary, so that we can reuse connections
136
# that are still active. Long term, it might be nice to have some
137
# sort of expiration policy, such as disconnect if inactive for
138
# X seconds. But that requires a lot more fanciness.
139
_connected_hosts = weakref.WeakValueDictionary()
141
def load_host_keys():
143
Load system host keys (probably doesn't work on windows) and any
144
"discovered" keys from previous sessions.
146
global SYSTEM_HOSTKEYS, BZR_HOSTKEYS
148
SYSTEM_HOSTKEYS = paramiko.util.load_host_keys(os.path.expanduser('~/.ssh/known_hosts'))
150
mutter('failed to load system host keys: ' + str(e))
151
bzr_hostkey_path = os.path.join(config_dir(), 'ssh_host_keys')
153
BZR_HOSTKEYS = paramiko.util.load_host_keys(bzr_hostkey_path)
155
mutter('failed to load bzr host keys: ' + str(e))
158
def save_host_keys():
160
Save "discovered" host keys in $(config)/ssh_host_keys/.
162
global SYSTEM_HOSTKEYS, BZR_HOSTKEYS
163
bzr_hostkey_path = os.path.join(config_dir(), 'ssh_host_keys')
164
if not os.path.isdir(config_dir()):
165
os.mkdir(config_dir())
167
f = open(bzr_hostkey_path, 'w')
168
f.write('# SSH host keys collected by bzr\n')
169
for hostname, keys in BZR_HOSTKEYS.iteritems():
170
for keytype, key in keys.iteritems():
171
f.write('%s %s %s\n' % (hostname, keytype, key.get_base64()))
174
mutter('failed to save bzr host keys: ' + str(e))
178
class SFTPTransportError (TransportError):
95
181
class SFTPLock(object):
96
"""This fakes a lock in a remote location.
98
A present lock is indicated just by the existence of a file. This
99
doesn't work well on all transports and they are only used in
100
deprecated storage formats.
182
"""This fakes a lock in a remote location."""
103
183
__slots__ = ['path', 'lock_path', 'lock_file', 'transport']
105
184
def __init__(self, path, transport):
185
assert isinstance(transport, SFTPTransport)
106
187
self.lock_file = None
108
189
self.lock_path = path + '.write-lock'
109
190
self.transport = transport
111
# RBC 20060103 FIXME should we be using private methods here ?
112
abspath = transport._remote_path(self.lock_path)
113
self.lock_file = transport._sftp_open_exclusive(abspath)
192
self.lock_file = transport._sftp_open_exclusive(self.lock_path)
114
193
except FileExists:
115
194
raise LockError('File %r already locked' % (self.path,))
117
196
def __del__(self):
118
197
"""Should this warn, or actually try to cleanup?"""
119
198
if self.lock_file:
120
warning("SFTPLock %r not explicitly unlocked" % (self.path,))
199
warn("SFTPLock %r not explicitly unlocked" % (self.path,))
123
202
def unlock(self):
131
210
# What specific errors should we catch here?
135
class _SFTPReadvHelper(object):
136
"""A class to help with managing the state of a readv request."""
138
# See _get_requests for an explanation.
139
_max_request_size = 32768
141
def __init__(self, original_offsets, relpath, _report_activity):
142
"""Create a new readv helper.
144
:param original_offsets: The original requests given by the caller of
146
:param relpath: The name of the file (if known)
147
:param _report_activity: A Transport._report_activity bound method,
148
to be called as data arrives.
150
self.original_offsets = list(original_offsets)
151
self.relpath = relpath
152
self._report_activity = _report_activity
154
def _get_requests(self):
155
"""Break up the offsets into individual requests over sftp.
157
The SFTP spec only requires implementers to support 32kB requests. We
158
could try something larger (openssh supports 64kB), but then we have to
159
handle requests that fail.
160
So instead, we just break up our maximum chunks into 32kB chunks, and
161
asyncronously requests them.
162
Newer versions of paramiko would do the chunking for us, but we want to
163
start processing results right away, so we do it ourselves.
165
# TODO: Because we issue async requests, we don't 'fudge' any extra
166
# data. I'm not 100% sure that is the best choice.
168
# The first thing we do, is to collapse the individual requests as much
169
# as possible, so we don't issues requests <32kB
170
sorted_offsets = sorted(self.original_offsets)
171
coalesced = list(ConnectedTransport._coalesce_offsets(sorted_offsets,
172
limit=0, fudge_factor=0))
174
for c_offset in coalesced:
175
start = c_offset.start
176
size = c_offset.length
178
# Break this up into 32kB requests
180
next_size = min(size, self._max_request_size)
181
requests.append((start, next_size))
184
if 'sftp' in debug.debug_flags:
185
mutter('SFTP.readv(%s) %s offsets => %s coalesced => %s requests',
186
self.relpath, len(sorted_offsets), len(coalesced),
190
def request_and_yield_offsets(self, fp):
191
"""Request the data from the remote machine, yielding the results.
193
:param fp: A Paramiko SFTPFile object that supports readv.
194
:return: Yield the data requested by the original readv caller, one by
197
requests = self._get_requests()
198
offset_iter = iter(self.original_offsets)
199
cur_offset, cur_size = offset_iter.next()
200
# paramiko .readv() yields strings that are in the order of the requests
201
# So we track the current request to know where the next data is
202
# being returned from.
208
# This is used to buffer chunks which we couldn't process yet
209
# It is (start, end, data) tuples.
211
# Create an 'unlimited' data stream, so we stop based on requests,
212
# rather than just because the data stream ended. This lets us detect
214
data_stream = itertools.chain(fp.readv(requests),
215
itertools.repeat(None))
216
for (start, length), data in itertools.izip(requests, data_stream):
218
if cur_coalesced is not None:
219
raise errors.ShortReadvError(self.relpath,
220
start, length, len(data))
221
if len(data) != length:
222
raise errors.ShortReadvError(self.relpath,
223
start, length, len(data))
224
self._report_activity(length, 'read')
226
# This is the first request, just buffer it
227
buffered_data = [data]
228
buffered_len = length
230
elif start == last_end:
231
# The data we are reading fits neatly on the previous
232
# buffer, so this is all part of a larger coalesced range.
233
buffered_data.append(data)
234
buffered_len += length
213
class SFTPTransport (Transport):
215
Transport implementation for SFTP access.
218
def __init__(self, base, clone_from=None):
219
assert base.startswith('sftp://')
220
self._parse_url(base)
221
base = self._unparse_url()
222
super(SFTPTransport, self).__init__(base)
223
if clone_from is None:
226
# use the same ssh connection, etc
227
self._sftp = clone_from._sftp
228
# super saves 'self.base'
230
def should_cache(self):
232
Return True if the data pulled across should be cached locally.
236
def clone(self, offset=None):
238
Return a new SFTPTransport with root at self.base + offset.
239
We share the same SFTP session between such transports, because it's
240
fairly expensive to set them up.
243
return SFTPTransport(self.base, self)
245
return SFTPTransport(self.abspath(offset), self)
247
def abspath(self, relpath):
249
Return the full url to the given relative path.
251
@param relpath: the relative path or path components
252
@type relpath: str or list
254
return self._unparse_url(self._abspath(relpath))
256
def _abspath(self, relpath):
257
"""Return the absolute path segment without the SFTP URL."""
258
# FIXME: share the common code across transports
259
assert isinstance(relpath, basestring)
260
relpath = [urllib.unquote(relpath)]
261
basepath = self._path.split('/')
262
if len(basepath) > 0 and basepath[-1] == '':
263
basepath = basepath[:-1]
267
if len(basepath) == 0:
268
# In most filesystems, a request for the parent
269
# of root, just returns root.
236
# We have an 'interrupt' in the data stream. So we know we are
237
# at a request boundary.
239
# We haven't consumed the buffer so far, so put it into
240
# data_chunks, and continue.
241
buffered = ''.join(buffered_data)
242
data_chunks.append((input_start, buffered))
244
buffered_data = [data]
245
buffered_len = length
246
last_end = start + length
247
if input_start == cur_offset and cur_size <= buffered_len:
248
# Simplify the next steps a bit by transforming buffered_data
249
# into a single string. We also have the nice property that
250
# when there is only one string ''.join([x]) == x, so there is
252
buffered = ''.join(buffered_data)
253
# Clean out buffered data so that we keep memory
257
# TODO: We *could* also consider the case where cur_offset is in
258
# in the buffered range, even though it doesn't *start*
259
# the buffered range. But for packs we pretty much always
260
# read in order, so you won't get any extra data in the
262
while (input_start == cur_offset
263
and (buffered_offset + cur_size) <= buffered_len):
264
# We've buffered enough data to process this request, spit it
266
cur_data = buffered[buffered_offset:buffered_offset + cur_size]
267
# move the direct pointer into our buffered data
268
buffered_offset += cur_size
269
# Move the start-of-buffer pointer
270
input_start += cur_size
271
# Yield the requested data
272
yield cur_offset, cur_data
273
cur_offset, cur_size = offset_iter.next()
274
# at this point, we've consumed as much of buffered as we can,
275
# so break off the portion that we consumed
276
if buffered_offset == len(buffered_data):
277
# No tail to leave behind
281
buffered = buffered[buffered_offset:]
282
buffered_data = [buffered]
283
buffered_len = len(buffered)
285
buffered = ''.join(buffered_data)
287
data_chunks.append((input_start, buffered))
289
if 'sftp' in debug.debug_flags:
290
mutter('SFTP readv left with %d out-of-order bytes',
291
sum(map(lambda x: len(x[1]), data_chunks)))
292
# We've processed all the readv data, at this point, anything we
293
# couldn't process is in data_chunks. This doesn't happen often, so
294
# this code path isn't optimized
295
# We use an interesting process for data_chunks
296
# Specifically if we have "bisect_left([(start, len, entries)],
298
# If start == qstart, then we get the specific node. Otherwise we
299
# get the previous node
301
idx = bisect.bisect_left(data_chunks, (cur_offset,))
302
if idx < len(data_chunks) and data_chunks[idx][0] == cur_offset:
303
# The data starts here
304
data = data_chunks[idx][1][:cur_size]
306
# The data is in a portion of a previous page
308
sub_offset = cur_offset - data_chunks[idx][0]
309
data = data_chunks[idx][1]
310
data = data[sub_offset:sub_offset + cur_size]
312
# We are missing the page where the data should be found,
315
if len(data) != cur_size:
316
raise AssertionError('We must have miscalulated.'
317
' We expected %d bytes, but only found %d'
318
% (cur_size, len(data)))
319
yield cur_offset, data
320
cur_offset, cur_size = offset_iter.next()
323
class SFTPTransport(ConnectedTransport):
324
"""Transport implementation for SFTP access."""
326
_do_prefetch = _default_do_prefetch
327
# TODO: jam 20060717 Conceivably these could be configurable, either
328
# by auto-tuning at run-time, or by a configuration (per host??)
329
# but the performance curve is pretty flat, so just going with
330
# reasonable defaults.
331
_max_readv_combine = 200
332
# Having to round trip to the server means waiting for a response,
333
# so it is better to download extra bytes.
334
# 8KiB had good performance for both local and remote network operations
335
_bytes_to_read_before_seek = 8192
337
# The sftp spec says that implementations SHOULD allow reads
338
# to be at least 32K. paramiko.readv() does an async request
339
# for the chunks. So we need to keep it within a single request
340
# size for paramiko <= 1.6.1. paramiko 1.6.2 will probably chop
341
# up the request itself, rather than us having to worry about it
342
_max_request_size = 32768
344
def __init__(self, base, _from_transport=None):
345
super(SFTPTransport, self).__init__(base,
346
_from_transport=_from_transport)
348
def _remote_path(self, relpath):
349
"""Return the path to be passed along the sftp protocol for relpath.
351
:param relpath: is a urlencoded string.
353
relative = urlutils.unescape(relpath).encode('utf-8')
354
remote_path = self._combine_paths(self._path, relative)
355
# the initial slash should be removed from the path, and treated as a
356
# homedir relative path (the path begins with a double slash if it is
357
# absolute). see draft-ietf-secsh-scp-sftp-ssh-uri-03.txt
358
# RBC 20060118 we are not using this as its too user hostile. instead
359
# we are following lftp and using /~/foo to mean '~/foo'
360
# vila--20070602 and leave absolute paths begin with a single slash.
361
if remote_path.startswith('/~/'):
362
remote_path = remote_path[3:]
363
elif remote_path == '/~':
367
def _create_connection(self, credentials=None):
368
"""Create a new connection with the provided credentials.
370
:param credentials: The credentials needed to establish the connection.
372
:return: The created connection and its associated credentials.
374
The credentials are only the password as it may have been entered
375
interactively by the user and may be different from the one provided
376
in base url at transport creation time.
378
if credentials is None:
379
password = self._password
381
password = credentials
383
vendor = ssh._get_ssh_vendor()
386
auth = config.AuthenticationConfig()
387
user = auth.get_user('ssh', self._host, self._port)
388
connection = vendor.connect_sftp(self._user, password,
389
self._host, self._port)
390
return connection, (user, password)
393
"""Ensures that a connection is established"""
394
connection = self._get_connection()
395
if connection is None:
396
# First connection ever
397
connection, credentials = self._create_connection()
398
self._set_connection(connection, credentials)
277
path = '/'.join(basepath)
278
# could still be a "relative" path here, but relative on the sftp server
281
def relpath(self, abspath):
282
username, password, host, port, path = self._split_url(abspath)
283
if (username != self._username or host != self._host or
284
port != self._port or not path.startswith(self._path)):
285
raise NonRelativePath('path %r is not under base URL %r'
286
% (abspath, self.base))
288
return path[pl:].lstrip('/')
401
290
def has(self, relpath):
403
292
Does the target location exist?
406
self._get_sftp().stat(self._remote_path(relpath))
407
# stat result is about 20 bytes, let's say
408
self._report_activity(20, 'read')
295
self._sftp.stat(self._abspath(relpath))
413
def get(self, relpath):
414
"""Get the file at the given relative path.
300
def get(self, relpath, decode=False):
302
Get the file at the given relative path.
416
304
:param relpath: The relative path to the file
419
# FIXME: by returning the file directly, we don't pass this
420
# through to report_activity. We could try wrapping the object
421
# before it's returned. For readv and get_bytes it's handled in
422
# the higher-level function.
424
path = self._remote_path(relpath)
425
f = self._get_sftp().file(path, mode='rb')
426
if self._do_prefetch and (getattr(f, 'prefetch', None) is not None):
307
path = self._abspath(relpath)
308
f = self._sftp.file(path)
309
# TODO: Don't prefetch until paramiko fixes itself
310
if hasattr(f, 'prefetch'):
429
except (IOError, paramiko.SSHException), e:
430
self._translate_io_exception(e, path, ': error retrieving',
431
failure_exc=errors.ReadError)
433
def get_bytes(self, relpath):
434
# reimplement this here so that we can report how many bytes came back
313
except (IOError, paramiko.SSHException), x:
314
raise NoSuchFile('Error retrieving %s: %s' % (path, str(x)), x)
316
def get_partial(self, relpath, start, length=None):
318
Get just part of a file.
320
:param relpath: Path to the file, relative to base
321
:param start: The starting position to read from
322
:param length: The length to read. A length of None indicates
323
read to the end of the file.
324
:return: A file-like object containing at least the specified bytes.
325
Some implementations may return objects which can be read
326
past this length, but this is not guaranteed.
328
# TODO: implement get_partial_multi to help with knit support
435
329
f = self.get(relpath)
438
self._report_activity(len(bytes), 'read')
443
def _readv(self, relpath, offsets):
444
"""See Transport.readv()"""
445
# We overload the default readv() because we want to use a file
446
# that does not have prefetch enabled.
447
# Also, if we have a new paramiko, it implements an async readv()
452
path = self._remote_path(relpath)
453
fp = self._get_sftp().file(path, mode='rb')
454
readv = getattr(fp, 'readv', None)
456
return self._sftp_readv(fp, offsets, relpath)
457
if 'sftp' in debug.debug_flags:
458
mutter('seek and read %s offsets', len(offsets))
459
return self._seek_and_read(fp, offsets, relpath)
460
except (IOError, paramiko.SSHException), e:
461
self._translate_io_exception(e, path, ': error retrieving')
463
def recommended_page_size(self):
464
"""See Transport.recommended_page_size().
466
For SFTP we suggest a large page size to reduce the overhead
467
introduced by latency.
471
def _sftp_readv(self, fp, offsets, relpath):
472
"""Use the readv() member of fp to do async readv.
474
Then read them using paramiko.readv(). paramiko.readv()
475
does not support ranges > 64K, so it caps the request size, and
476
just reads until it gets all the stuff it wants.
478
helper = _SFTPReadvHelper(offsets, relpath, self._report_activity)
479
return helper.request_and_yield_offsets(fp)
481
def put_file(self, relpath, f, mode=None):
483
Copy the file-like object into the location.
331
if hasattr(f, 'prefetch'):
335
def put(self, relpath, f):
337
Copy the file-like or string object into the location.
485
339
:param relpath: Location to put the contents, relative to base.
486
:param f: File-like object.
487
:param mode: The final mode for the file
340
:param f: File-like or string object.
489
final_path = self._remote_path(relpath)
490
return self._put(final_path, f, mode=mode)
492
def _put(self, abspath, f, mode=None):
493
"""Helper function so both put() and copy_abspaths can reuse the code"""
494
tmp_abspath = '%s.tmp.%.9f.%d.%d' % (abspath, time.time(),
342
final_path = self._abspath(relpath)
343
tmp_relpath = '%s.tmp.%.9f.%d.%d' % (relpath, time.time(),
495
344
os.getpid(), random.randint(0,0x7FFFFFFF))
496
fout = self._sftp_open_exclusive(tmp_abspath, mode=mode)
345
tmp_abspath = self._abspath(tmp_relpath)
346
fout = self._sftp_open_exclusive(tmp_relpath)
500
fout.set_pipelined(True)
501
length = self._pump(f, fout)
502
except (IOError, paramiko.SSHException), e:
503
self._translate_io_exception(e, tmp_abspath)
504
# XXX: This doesn't truly help like we would like it to.
505
# The problem is that openssh strips sticky bits. So while we
506
# can properly set group write permission, we lose the group
507
# sticky bit. So it is probably best to stop chmodding, and
508
# just tell users that they need to set the umask correctly.
509
# The attr.st_mode = mode, in _sftp_open_exclusive
510
# will handle when the user wants the final mode to be more
511
# restrictive. And then we avoid a round trip. Unless
512
# paramiko decides to expose an async chmod()
514
# This is designed to chmod() right before we close.
515
# Because we set_pipelined() earlier, theoretically we might
516
# avoid the round trip for fout.close()
518
self._get_sftp().chmod(tmp_abspath, mode)
521
self._rename_and_overwrite(tmp_abspath, abspath)
352
self._translate_io_exception(e, relpath)
353
except paramiko.SSHException, x:
354
raise SFTPTransportError('Unable to write file %r' % (relpath,), x)
523
355
except Exception, e:
524
356
# If we fail, try to clean up the temporary file
525
357
# before we throw the exception
526
358
# but don't let another exception mess things up
527
# Write out the traceback, because otherwise
528
# the catch and throw destroys it
530
mutter(traceback.format_exc())
534
self._get_sftp().remove(tmp_abspath)
536
# raise the saved except
538
# raise the original with its traceback if we can.
541
def _put_non_atomic_helper(self, relpath, writer, mode=None,
542
create_parent_dir=False,
544
abspath = self._remote_path(relpath)
546
# TODO: jam 20060816 paramiko doesn't publicly expose a way to
547
# set the file mode at create time. If it does, use it.
548
# But for now, we just chmod later anyway.
550
def _open_and_write_file():
551
"""Try to open the target file, raise error on failure"""
361
self._sftp.remove(tmp_abspath)
366
# sftp rename doesn't allow overwriting, so play tricks:
367
tmp_safety = 'bzr.tmp.%.9f.%d.%d' % (time.time(), os.getpid(), random.randint(0, 0x7FFFFFFF))
368
tmp_safety = self._abspath(tmp_safety)
370
self._sftp.rename(final_path, tmp_safety)
555
fout = self._get_sftp().file(abspath, mode='wb')
556
fout.set_pipelined(True)
558
except (paramiko.SSHException, IOError), e:
559
self._translate_io_exception(e, abspath,
562
# This is designed to chmod() right before we close.
563
# Because we set_pipelined() earlier, theoretically we might
564
# avoid the round trip for fout.close()
566
self._get_sftp().chmod(abspath, mode)
377
self._sftp.rename(tmp_abspath, final_path)
379
self._translate_io_exception(e, relpath)
380
except paramiko.SSHException, x:
381
raise SFTPTransportError('Unable to rename into file %r' % (path,), x)
571
if not create_parent_dir:
572
_open_and_write_file()
575
# Try error handling to create the parent directory if we need to
577
_open_and_write_file()
579
# Try to create the parent directory, and then go back to
581
parent_dir = os.path.dirname(abspath)
582
self._mkdir(parent_dir, dir_mode)
583
_open_and_write_file()
585
def put_file_non_atomic(self, relpath, f, mode=None,
586
create_parent_dir=False,
588
"""Copy the file-like object into the target location.
590
This function is not strictly safe to use. It is only meant to
591
be used when you already know that the target does not exist.
592
It is not safe, because it will open and truncate the remote
593
file. So there may be a time when the file has invalid contents.
595
:param relpath: The remote location to put the contents.
596
:param f: File-like object.
597
:param mode: Possible access permissions for new file.
598
None means do not set remote permissions.
599
:param create_parent_dir: If we cannot create the target file because
600
the parent directory does not exist, go ahead and
601
create it, and then try again.
605
self._put_non_atomic_helper(relpath, writer, mode=mode,
606
create_parent_dir=create_parent_dir,
609
def put_bytes_non_atomic(self, relpath, bytes, mode=None,
610
create_parent_dir=False,
614
self._put_non_atomic_helper(relpath, writer, mode=mode,
615
create_parent_dir=create_parent_dir,
387
self._sftp.unlink(tmp_safety)
389
self._sftp.rename(tmp_safety, final_path)
618
391
def iter_files_recursive(self):
619
392
"""Walk the relative paths of all files in this transport."""
620
# progress is handled by list_dir
621
393
queue = list(self.list_dir('.'))
623
relpath = queue.pop(0)
395
relpath = urllib.quote(queue.pop(0))
624
396
st = self.stat(relpath)
625
397
if stat.S_ISDIR(st.st_mode):
626
398
for i, basename in enumerate(self.list_dir(relpath)):
631
def _mkdir(self, abspath, mode=None):
637
self._report_activity(len(abspath), 'write')
638
self._get_sftp().mkdir(abspath, local_mode)
639
self._report_activity(1, 'read')
641
# chmod a dir through sftp will erase any sgid bit set
642
# on the server side. So, if the bit mode are already
643
# set, avoid the chmod. If the mode is not fine but
644
# the sgid bit is set, report a warning to the user
645
# with the umask fix.
646
stat = self._get_sftp().lstat(abspath)
647
mode = mode & 0777 # can't set special bits anyway
648
if mode != stat.st_mode & 0777:
649
if stat.st_mode & 06000:
650
warning('About to chmod %s over sftp, which will result'
651
' in its suid or sgid bits being cleared. If'
652
' you want to preserve those bits, change your '
653
' environment on the server to use umask 0%03o.'
654
% (abspath, 0777 - mode))
655
self._get_sftp().chmod(abspath, mode=mode)
656
except (paramiko.SSHException, IOError), e:
657
self._translate_io_exception(e, abspath, ': unable to mkdir',
658
failure_exc=FileExists)
660
def mkdir(self, relpath, mode=None):
403
def mkdir(self, relpath):
661
404
"""Create a directory at the given path."""
662
self._mkdir(self._remote_path(relpath), mode=mode)
664
def open_write_stream(self, relpath, mode=None):
665
"""See Transport.open_write_stream."""
666
# initialise the file to zero-length
667
# this is three round trips, but we don't use this
668
# api more than once per write_group at the moment so
669
# it is a tolerable overhead. Better would be to truncate
670
# the file after opening. RBC 20070805
671
self.put_bytes_non_atomic(relpath, "", mode)
672
abspath = self._remote_path(relpath)
673
# TODO: jam 20060816 paramiko doesn't publicly expose a way to
674
# set the file mode at create time. If it does, use it.
675
# But for now, we just chmod later anyway.
678
handle = self._get_sftp().file(abspath, mode='wb')
679
handle.set_pipelined(True)
680
except (paramiko.SSHException, IOError), e:
681
self._translate_io_exception(e, abspath,
683
_file_streams[self.abspath(relpath)] = handle
684
return FileFileStream(self, relpath, handle)
686
def _translate_io_exception(self, e, path, more_info='',
687
failure_exc=PathError):
688
"""Translate a paramiko or IOError into a friendlier exception.
690
:param e: The original exception
691
:param path: The path in question when the error is raised
692
:param more_info: Extra information that can be included,
693
such as what was going on
694
:param failure_exc: Paramiko has the super fun ability to raise completely
695
opaque errors that just set "e.args = ('Failure',)" with
697
If this parameter is set, it defines the exception
698
to raise in these cases.
406
path = self._abspath(relpath)
407
self._sftp.mkdir(path)
409
self._translate_io_exception(e, relpath)
410
except (IOError, paramiko.SSHException), x:
411
raise SFTPTransportError('Unable to mkdir %r' % (path,), x)
413
def _translate_io_exception(self, e, relpath):
700
414
# paramiko seems to generate detailless errors.
701
self._translate_error(e, path, raise_generic=False)
702
if getattr(e, 'args', None) is not None:
703
if (e.args == ('No such file or directory',) or
704
e.args == ('No such file',)):
705
raise NoSuchFile(path, str(e) + more_info)
706
if (e.args == ('mkdir failed',) or
707
e.args[0].startswith('syserr: File exists')):
708
raise FileExists(path, str(e) + more_info)
709
# strange but true, for the paramiko server.
710
if (e.args == ('Failure',)):
711
raise failure_exc(path, str(e) + more_info)
712
# Can be something like args = ('Directory not empty:
713
# '/srv/bazaar.launchpad.net/blah...: '
714
# [Errno 39] Directory not empty',)
715
if (e.args[0].startswith('Directory not empty: ')
716
or getattr(e, 'errno', None) == errno.ENOTEMPTY):
717
raise errors.DirectoryNotEmpty(path, str(e))
718
mutter('Raising exception with args %s', e.args)
719
if getattr(e, 'errno', None) is not None:
720
mutter('Raising exception with errno %s', e.errno)
415
if (e.errno == errno.ENOENT or
416
e.args == ('No such file or directory',) or
417
e.args == ('No such file',)):
418
raise NoSuchFile(relpath)
419
if (e.args == ('mkdir failed',)):
420
raise FileExists(relpath)
421
# strange but true, for the paramiko server.
422
if (e.args == ('Failure',)):
423
raise FileExists(relpath)
723
def append_file(self, relpath, f, mode=None):
426
def append(self, relpath, f):
725
428
Append the text in the file-like object into the final
729
path = self._remote_path(relpath)
730
fout = self._get_sftp().file(path, 'ab')
732
self._get_sftp().chmod(path, mode)
432
path = self._abspath(relpath)
433
fout = self._sftp.file(path, 'ab')
734
434
self._pump(f, fout)
736
except (IOError, paramiko.SSHException), e:
737
self._translate_io_exception(e, relpath, ': unable to append')
739
def rename(self, rel_from, rel_to):
740
"""Rename without special overwriting"""
742
self._get_sftp().rename(self._remote_path(rel_from),
743
self._remote_path(rel_to))
744
except (IOError, paramiko.SSHException), e:
745
self._translate_io_exception(e, rel_from,
746
': unable to rename to %r' % (rel_to))
748
def _rename_and_overwrite(self, abs_from, abs_to):
749
"""Do a fancy rename on the remote server.
751
Using the implementation provided by osutils.
754
sftp = self._get_sftp()
755
fancy_rename(abs_from, abs_to,
756
rename_func=sftp.rename,
757
unlink_func=sftp.remove)
758
except (IOError, paramiko.SSHException), e:
759
self._translate_io_exception(e, abs_from,
760
': unable to rename to %r' % (abs_to))
435
except (IOError, paramiko.SSHException), x:
436
raise SFTPTransportError('Unable to append file %r' % (path,), x)
438
def copy(self, rel_from, rel_to):
439
"""Copy the item at rel_from to the location at rel_to"""
440
path_from = self._abspath(rel_from)
441
path_to = self._abspath(rel_to)
442
self._copy_abspaths(path_from, path_to)
444
def _copy_abspaths(self, path_from, path_to):
445
"""Copy files given an absolute path
447
:param path_from: Path on remote server to read
448
:param path_to: Path on remote server to write
451
TODO: Should the destination location be atomically created?
452
This has not been specified
453
TODO: This should use some sort of remote copy, rather than
454
pulling the data locally, and then writing it remotely
457
fin = self._sftp.file(path_from, 'rb')
459
fout = self._sftp.file(path_to, 'wb')
461
fout.set_pipelined(True)
462
self._pump(fin, fout)
467
except (IOError, paramiko.SSHException), x:
468
raise SFTPTransportError('Unable to copy %r to %r' % (path_from, path_to), x)
470
def copy_to(self, relpaths, other, pb=None):
471
"""Copy a set of entries from self into another Transport.
473
:param relpaths: A list/generator of entries to be copied.
475
if isinstance(other, SFTPTransport) and other._sftp is self._sftp:
476
# Both from & to are on the same remote filesystem
477
# We can use a remote copy, instead of pulling locally, and pushing
479
total = self._get_total(relpaths)
481
for path in relpaths:
482
path_from = self._abspath(relpath)
483
path_to = other._abspath(relpath)
484
self._update_pb(pb, 'copy-to', count, total)
485
self._copy_abspaths(path_from, path_to)
489
return super(SFTPTransport, self).copy_to(relpaths, other, pb=pb)
491
# The dummy implementation just does a simple get + put
492
def copy_entry(path):
493
other.put(path, self.get(path))
495
return self._iterate_over(relpaths, copy_entry, pb, 'copy_to', expand=False)
762
497
def move(self, rel_from, rel_to):
763
498
"""Move the item at rel_from to the location at rel_to"""
764
path_from = self._remote_path(rel_from)
765
path_to = self._remote_path(rel_to)
766
self._rename_and_overwrite(path_from, path_to)
499
path_from = self._abspath(rel_from)
500
path_to = self._abspath(rel_to)
502
self._sftp.rename(path_from, path_to)
503
except (IOError, paramiko.SSHException), x:
504
raise SFTPTransportError('Unable to move %r to %r' % (path_from, path_to), x)
768
506
def delete(self, relpath):
769
507
"""Delete the item at relpath"""
770
path = self._remote_path(relpath)
508
path = self._abspath(relpath)
772
self._get_sftp().remove(path)
773
except (IOError, paramiko.SSHException), e:
774
self._translate_io_exception(e, path, ': unable to delete')
776
def external_url(self):
777
"""See bzrlib.transport.Transport.external_url."""
778
# the external path for SFTP is the base
510
self._sftp.remove(path)
511
except (IOError, paramiko.SSHException), x:
512
raise SFTPTransportError('Unable to delete %r' % (path,), x)
781
514
def listable(self):
782
515
"""Return True if this store supports listing."""
862
560
# that we have taken the lock.
863
561
return SFTPLock(relpath, self)
865
def _sftp_open_exclusive(self, abspath, mode=None):
564
def _unparse_url(self, path=None):
567
path = urllib.quote(path)
568
if path.startswith('/'):
569
path = '/%2F' + path[1:]
572
netloc = urllib.quote(self._host)
573
if self._username is not None:
574
netloc = '%s@%s' % (urllib.quote(self._username), netloc)
575
if self._port is not None:
576
netloc = '%s:%d' % (netloc, self._port)
578
return urlparse.urlunparse(('sftp', netloc, path, '', '', ''))
580
def _split_url(self, url):
581
if isinstance(url, unicode):
582
url = url.encode('utf-8')
583
(scheme, netloc, path, params,
584
query, fragment) = urlparse.urlparse(url, allow_fragments=False)
585
assert scheme == 'sftp'
586
username = password = host = port = None
588
username, host = netloc.split('@', 1)
590
username, password = username.split(':', 1)
591
password = urllib.unquote(password)
592
username = urllib.unquote(username)
597
host, port = host.rsplit(':', 1)
601
raise SFTPTransportError('%s: invalid port number' % port)
602
host = urllib.unquote(host)
604
path = urllib.unquote(path)
606
# the initial slash should be removed from the path, and treated
607
# as a homedir relative path (the path begins with a double slash
608
# if it is absolute).
609
# see draft-ietf-secsh-scp-sftp-ssh-uri-03.txt
610
if path.startswith('/'):
613
return (username, password, host, port, path)
615
def _parse_url(self, url):
616
(self._username, self._password,
617
self._host, self._port, self._path) = self._split_url(url)
618
if self._port is None:
621
def _sftp_connect(self):
622
"""Connect to the remote sftp server.
623
After this, self._sftp should have a valid connection (or
624
we raise an SFTPTransportError 'could not connect').
626
TODO: Raise a more reasonable ConnectionFailed exception
628
global _connected_hosts
630
idx = (self._host, self._port, self._username)
632
self._sftp = _connected_hosts[idx]
637
vendor = _get_ssh_vendor()
639
sock = SFTPSubprocess(self._host, self._port, self._username)
640
self._sftp = SFTPClient(sock)
642
self._paramiko_connect()
644
_connected_hosts[idx] = self._sftp
646
def _paramiko_connect(self):
647
global SYSTEM_HOSTKEYS, BZR_HOSTKEYS
652
t = paramiko.Transport((self._host, self._port))
654
except paramiko.SSHException:
655
raise SFTPTransportError('Unable to reach SSH host %s:%d' % (self._host, self._port))
657
server_key = t.get_remote_server_key()
658
server_key_hex = paramiko.util.hexify(server_key.get_fingerprint())
659
keytype = server_key.get_name()
660
if SYSTEM_HOSTKEYS.has_key(self._host) and SYSTEM_HOSTKEYS[self._host].has_key(keytype):
661
our_server_key = SYSTEM_HOSTKEYS[self._host][keytype]
662
our_server_key_hex = paramiko.util.hexify(our_server_key.get_fingerprint())
663
elif BZR_HOSTKEYS.has_key(self._host) and BZR_HOSTKEYS[self._host].has_key(keytype):
664
our_server_key = BZR_HOSTKEYS[self._host][keytype]
665
our_server_key_hex = paramiko.util.hexify(our_server_key.get_fingerprint())
667
warning('Adding %s host key for %s: %s' % (keytype, self._host, server_key_hex))
668
if not BZR_HOSTKEYS.has_key(self._host):
669
BZR_HOSTKEYS[self._host] = {}
670
BZR_HOSTKEYS[self._host][keytype] = server_key
671
our_server_key = server_key
672
our_server_key_hex = paramiko.util.hexify(our_server_key.get_fingerprint())
674
if server_key != our_server_key:
675
filename1 = os.path.expanduser('~/.ssh/known_hosts')
676
filename2 = os.path.join(config_dir(), 'ssh_host_keys')
677
raise SFTPTransportError('Host keys for %s do not match! %s != %s' % \
678
(self._host, our_server_key_hex, server_key_hex),
679
['Try editing %s or %s' % (filename1, filename2)])
681
self._sftp_auth(t, self._username or getpass.getuser(), self._host)
684
self._sftp = t.open_sftp_client()
685
except paramiko.SSHException:
686
raise BzrError('Unable to find path %s on SFTP server %s' % \
687
(self._path, self._host))
689
def _sftp_auth(self, transport, username, host):
690
agent = paramiko.Agent()
691
for key in agent.get_keys():
692
mutter('Trying SSH agent key %s' % paramiko.util.hexify(key.get_fingerprint()))
694
transport.auth_publickey(self._username, key)
696
except paramiko.SSHException, e:
699
# okay, try finding id_rsa or id_dss? (posix only)
700
if self._try_pkey_auth(transport, paramiko.RSAKey, 'id_rsa'):
702
if self._try_pkey_auth(transport, paramiko.DSSKey, 'id_dsa'):
707
transport.auth_password(self._username, self._password)
709
except paramiko.SSHException, e:
712
# give up and ask for a password
713
# FIXME: shouldn't be implementing UI this deep into bzrlib
714
enc = sys.stdout.encoding
715
password = getpass.getpass('SSH %s@%s password: ' %
716
(self._username.encode(enc, 'replace'), self._host.encode(enc, 'replace')))
718
transport.auth_password(self._username, password)
719
except paramiko.SSHException:
720
raise SFTPTransportError('Unable to authenticate to SSH host as %s@%s' % \
721
(self._username, self._host))
723
def _try_pkey_auth(self, transport, pkey_class, filename):
724
filename = os.path.expanduser('~/.ssh/' + filename)
726
key = pkey_class.from_private_key_file(filename)
727
transport.auth_publickey(self._username, key)
729
except paramiko.PasswordRequiredException:
730
# FIXME: shouldn't be implementing UI this deep into bzrlib
731
enc = sys.stdout.encoding
732
password = getpass.getpass('SSH %s password: ' %
733
(os.path.basename(filename).encode(enc, 'replace'),))
735
key = pkey_class.from_private_key_file(filename, password)
736
transport.auth_publickey(self._username, key)
738
except paramiko.SSHException:
739
mutter('SSH authentication via %s key failed.' % (os.path.basename(filename),))
740
except paramiko.SSHException:
741
mutter('SSH authentication via %s key failed.' % (os.path.basename(filename),))
746
def _sftp_open_exclusive(self, relpath):
866
747
"""Open a remote path exclusively.
868
749
SFTP supports O_EXCL (SFTP_FLAG_EXCL), which fails if
added
removed
bzrlib/transport/sftp.py
