1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from cStringIO import StringIO
21
from stat import (S_ISREG, S_ISDIR, S_ISLNK, ST_MODE, ST_SIZE,
22
S_ISCHR, S_ISBLK, S_ISFIFO, S_ISSOCK)
26
from bzrlib.lazy_import import lazy_import
27
lazy_import(globals(), """
28
from datetime import datetime
30
from ntpath import (abspath as _nt_abspath,
32
normpath as _nt_normpath,
33
realpath as _nt_realpath,
34
splitdrive as _nt_splitdrive,
44
from tempfile import (
55
from bzrlib.symbol_versioning import (
59
from bzrlib.trace import mutter
62
# On win32, O_BINARY is used to indicate the file should
63
# be opened in binary mode, rather than text mode.
64
# On other platforms, O_BINARY doesn't exist, because
65
# they always open in binary mode, so it is okay to
66
# OR with 0 on those platforms
67
O_BINARY = getattr(os, 'O_BINARY', 0)
70
def make_readonly(filename):
71
"""Make a filename read-only."""
72
mod = os.stat(filename).st_mode
74
os.chmod(filename, mod)
77
def make_writable(filename):
78
mod = os.stat(filename).st_mode
80
os.chmod(filename, mod)
87
"""Return a quoted filename filename
89
This previously used backslash quoting, but that works poorly on
91
# TODO: I'm not really sure this is the best format either.x
94
_QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')
96
if _QUOTE_RE.search(f):
102
_directory_kind = 'directory'
105
stat.S_IFDIR:_directory_kind,
106
stat.S_IFCHR:'chardev',
107
stat.S_IFBLK:'block',
110
stat.S_IFLNK:'symlink',
111
stat.S_IFSOCK:'socket',
115
def file_kind_from_stat_mode(stat_mode, _formats=_formats, _unknown='unknown'):
116
"""Generate a file kind from a stat mode. This is used in walkdirs.
118
Its performance is critical: Do not mutate without careful benchmarking.
121
return _formats[stat_mode & 0170000]
126
def file_kind(f, _lstat=os.lstat, _mapper=file_kind_from_stat_mode):
128
return _mapper(_lstat(f).st_mode)
130
if getattr(e, 'errno', None) == errno.ENOENT:
131
raise errors.NoSuchFile(f)
136
"""Return the current umask"""
137
# Assume that people aren't messing with the umask while running
138
# XXX: This is not thread safe, but there is no way to get the
139
# umask without setting it
145
def kind_marker(kind):
148
elif kind == _directory_kind:
150
elif kind == 'symlink':
153
raise errors.BzrError('invalid file kind %r' % kind)
155
lexists = getattr(os.path, 'lexists', None)
159
if getattr(os, 'lstat') is not None:
165
if e.errno == errno.ENOENT:
168
raise errors.BzrError("lstat/stat of (%r): %r" % (f, e))
171
def fancy_rename(old, new, rename_func, unlink_func):
172
"""A fancy rename, when you don't have atomic rename.
174
:param old: The old path, to rename from
175
:param new: The new path, to rename to
176
:param rename_func: The potentially non-atomic rename function
177
:param unlink_func: A way to delete the target file if the full rename succeeds
180
# sftp rename doesn't allow overwriting, so play tricks:
182
base = os.path.basename(new)
183
dirname = os.path.dirname(new)
184
tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))
185
tmp_name = pathjoin(dirname, tmp_name)
187
# Rename the file out of the way, but keep track if it didn't exist
188
# We don't want to grab just any exception
189
# something like EACCES should prevent us from continuing
190
# The downside is that the rename_func has to throw an exception
191
# with an errno = ENOENT, or NoSuchFile
194
rename_func(new, tmp_name)
195
except (errors.NoSuchFile,), e:
198
# RBC 20060103 abstraction leakage: the paramiko SFTP clients rename
199
# function raises an IOError with errno is None when a rename fails.
200
# This then gets caught here.
201
if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):
204
if (getattr(e, 'errno', None) is None
205
or e.errno not in (errno.ENOENT, errno.ENOTDIR)):
212
# This may throw an exception, in which case success will
214
rename_func(old, new)
218
# If the file used to exist, rename it back into place
219
# otherwise just delete it from the tmp location
221
unlink_func(tmp_name)
223
rename_func(tmp_name, new)
226
# In Python 2.4.2 and older, os.path.abspath and os.path.realpath
227
# choke on a Unicode string containing a relative path if
228
# os.getcwd() returns a non-sys.getdefaultencoding()-encoded
230
_fs_enc = sys.getfilesystemencoding() or 'utf-8'
231
def _posix_abspath(path):
232
# jam 20060426 rather than encoding to fsencoding
233
# copy posixpath.abspath, but use os.getcwdu instead
234
if not posixpath.isabs(path):
235
path = posixpath.join(getcwd(), path)
236
return posixpath.normpath(path)
239
def _posix_realpath(path):
240
return posixpath.realpath(path.encode(_fs_enc)).decode(_fs_enc)
243
def _win32_fixdrive(path):
244
"""Force drive letters to be consistent.
246
win32 is inconsistent whether it returns lower or upper case
247
and even if it was consistent the user might type the other
248
so we force it to uppercase
249
running python.exe under cmd.exe return capital C:\\
250
running win32 python inside a cygwin shell returns lowercase c:\\
252
drive, path = _nt_splitdrive(path)
253
return drive.upper() + path
256
def _win32_abspath(path):
257
# Real _nt_abspath doesn't have a problem with a unicode cwd
258
return _win32_fixdrive(_nt_abspath(unicode(path)).replace('\\', '/'))
261
def _win32_realpath(path):
262
# Real _nt_realpath doesn't have a problem with a unicode cwd
263
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
266
def _win32_pathjoin(*args):
267
return _nt_join(*args).replace('\\', '/')
270
def _win32_normpath(path):
271
return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))
275
return _win32_fixdrive(os.getcwdu().replace('\\', '/'))
278
def _win32_mkdtemp(*args, **kwargs):
279
return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))
282
def _win32_rename(old, new):
283
"""We expect to be able to atomically replace 'new' with old.
285
On win32, if new exists, it must be moved out of the way first,
289
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
291
if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):
292
# If we try to rename a non-existant file onto cwd, we get
293
# EPERM or EACCES instead of ENOENT, this will raise ENOENT
294
# if the old path doesn't exist, sometimes we get EACCES
295
# On Linux, we seem to get EBUSY, on Mac we get EINVAL
301
return unicodedata.normalize('NFKC', os.getcwdu())
304
# Default is to just use the python builtins, but these can be rebound on
305
# particular platforms.
306
abspath = _posix_abspath
307
realpath = _posix_realpath
308
pathjoin = os.path.join
309
normpath = os.path.normpath
312
dirname = os.path.dirname
313
basename = os.path.basename
314
split = os.path.split
315
splitext = os.path.splitext
316
# These were already imported into local scope
317
# mkdtemp = tempfile.mkdtemp
318
# rmtree = shutil.rmtree
320
MIN_ABS_PATHLENGTH = 1
323
if sys.platform == 'win32':
324
abspath = _win32_abspath
325
realpath = _win32_realpath
326
pathjoin = _win32_pathjoin
327
normpath = _win32_normpath
328
getcwd = _win32_getcwd
329
mkdtemp = _win32_mkdtemp
330
rename = _win32_rename
332
MIN_ABS_PATHLENGTH = 3
334
def _win32_delete_readonly(function, path, excinfo):
335
"""Error handler for shutil.rmtree function [for win32]
336
Helps to remove files and dirs marked as read-only.
338
exception = excinfo[1]
339
if function in (os.remove, os.rmdir) \
340
and isinstance(exception, OSError) \
341
and exception.errno == errno.EACCES:
347
def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):
348
"""Replacer for shutil.rmtree: could remove readonly dirs/files"""
349
return shutil.rmtree(path, ignore_errors, onerror)
350
elif sys.platform == 'darwin':
354
def get_terminal_encoding():
355
"""Find the best encoding for printing to the screen.
357
This attempts to check both sys.stdout and sys.stdin to see
358
what encoding they are in, and if that fails it falls back to
359
bzrlib.user_encoding.
360
The problem is that on Windows, locale.getpreferredencoding()
361
is not the same encoding as that used by the console:
362
http://mail.python.org/pipermail/python-list/2003-May/162357.html
364
On my standard US Windows XP, the preferred encoding is
365
cp1252, but the console is cp437
367
output_encoding = getattr(sys.stdout, 'encoding', None)
368
if not output_encoding:
369
input_encoding = getattr(sys.stdin, 'encoding', None)
370
if not input_encoding:
371
output_encoding = bzrlib.user_encoding
372
mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)
374
output_encoding = input_encoding
375
mutter('encoding stdout as sys.stdin encoding %r', output_encoding)
377
mutter('encoding stdout as sys.stdout encoding %r', output_encoding)
378
if output_encoding == 'cp0':
379
# invalid encoding (cp0 means 'no codepage' on Windows)
380
output_encoding = bzrlib.user_encoding
381
mutter('cp0 is invalid encoding.'
382
' encoding stdout as bzrlib.user_encoding %r', output_encoding)
383
return output_encoding
386
def normalizepath(f):
387
if getattr(os.path, 'realpath', None) is not None:
391
[p,e] = os.path.split(f)
392
if e == "" or e == "." or e == "..":
395
return pathjoin(F(p), e)
399
"""Copy a file to a backup.
401
Backups are named in GNU-style, with a ~ suffix.
403
If the file is already a backup, it's not copied.
409
if has_symlinks() and os.path.islink(fn):
410
target = os.readlink(fn)
411
os.symlink(target, bfn)
419
outf = file(bfn, 'wb')
427
"""True if f is an accessible directory."""
429
return S_ISDIR(os.lstat(f)[ST_MODE])
435
"""True if f is a regular file."""
437
return S_ISREG(os.lstat(f)[ST_MODE])
442
"""True if f is a symlink."""
444
return S_ISLNK(os.lstat(f)[ST_MODE])
448
def is_inside(dir, fname):
449
"""True if fname is inside dir.
451
The parameters should typically be passed to osutils.normpath first, so
452
that . and .. and repeated slashes are eliminated, and the separators
453
are canonical for the platform.
455
The empty string as a dir name is taken as top-of-tree and matches
458
# XXX: Most callers of this can actually do something smarter by
459
# looking at the inventory
469
return fname.startswith(dir)
472
def is_inside_any(dir_list, fname):
473
"""True if fname is inside any of given dirs."""
474
for dirname in dir_list:
475
if is_inside(dirname, fname):
481
def is_inside_or_parent_of_any(dir_list, fname):
482
"""True if fname is a child or a parent of any of the given files."""
483
for dirname in dir_list:
484
if is_inside(dirname, fname) or is_inside(fname, dirname):
490
def pumpfile(fromfile, tofile):
491
"""Copy contents of one file to another."""
494
b = fromfile.read(BUFSIZE)
500
def file_iterator(input_file, readsize=32768):
502
b = input_file.read(readsize)
509
if getattr(f, 'tell', None) is not None:
522
def sha_strings(strings):
523
"""Return the sha-1 of concatenation of strings"""
525
map(s.update, strings)
535
def fingerprint_file(f):
540
return {'size': size,
541
'sha1': s.hexdigest()}
544
def compare_files(a, b):
545
"""Returns true if equal in contents"""
556
def local_time_offset(t=None):
557
"""Return offset of local zone from GMT, either at present or at time t."""
560
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
561
return offset.days * 86400 + offset.seconds
564
def format_date(t, offset=0, timezone='original', date_fmt=None,
566
## TODO: Perhaps a global option to use either universal or local time?
567
## Or perhaps just let people set $TZ?
568
assert isinstance(t, float)
570
if timezone == 'utc':
573
elif timezone == 'original':
576
tt = time.gmtime(t + offset)
577
elif timezone == 'local':
578
tt = time.localtime(t)
579
offset = local_time_offset(t)
581
raise errors.BzrError("unsupported timezone format %r" % timezone,
582
['options are "utc", "original", "local"'])
584
date_fmt = "%a %Y-%m-%d %H:%M:%S"
586
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
589
return (time.strftime(date_fmt, tt) + offset_str)
592
def compact_date(when):
593
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
596
def format_delta(delta):
597
"""Get a nice looking string for a time delta.
599
:param delta: The time difference in seconds, can be positive or negative.
600
positive indicates time in the past, negative indicates time in the
601
future. (usually time.time() - stored_time)
602
:return: String formatted to show approximate resolution
608
direction = 'in the future'
612
if seconds < 90: # print seconds up to 90 seconds
614
return '%d second %s' % (seconds, direction,)
616
return '%d seconds %s' % (seconds, direction)
618
minutes = int(seconds / 60)
619
seconds -= 60 * minutes
624
if minutes < 90: # print minutes, seconds up to 90 minutes
626
return '%d minute, %d second%s %s' % (
627
minutes, seconds, plural_seconds, direction)
629
return '%d minutes, %d second%s %s' % (
630
minutes, seconds, plural_seconds, direction)
632
hours = int(minutes / 60)
633
minutes -= 60 * hours
640
return '%d hour, %d minute%s %s' % (hours, minutes,
641
plural_minutes, direction)
642
return '%d hours, %d minute%s %s' % (hours, minutes,
643
plural_minutes, direction)
646
"""Return size of given open file."""
647
return os.fstat(f.fileno())[ST_SIZE]
650
# Define rand_bytes based on platform.
652
# Python 2.4 and later have os.urandom,
653
# but it doesn't work on some arches
655
rand_bytes = os.urandom
656
except (NotImplementedError, AttributeError):
657
# If python doesn't have os.urandom, or it doesn't work,
658
# then try to first pull random data from /dev/urandom
660
rand_bytes = file('/dev/urandom', 'rb').read
661
# Otherwise, use this hack as a last resort
662
except (IOError, OSError):
663
# not well seeded, but better than nothing
668
s += chr(random.randint(0, 255))
673
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
675
"""Return a random string of num alphanumeric characters
677
The result only contains lowercase chars because it may be used on
678
case-insensitive filesystems.
681
for raw_byte in rand_bytes(num):
682
s += ALNUM[ord(raw_byte) % 36]
686
## TODO: We could later have path objects that remember their list
687
## decomposition (might be too tricksy though.)
690
"""Turn string into list of parts."""
691
assert isinstance(p, basestring)
693
# split on either delimiter because people might use either on
695
ps = re.split(r'[\\/]', p)
700
raise errors.BzrError("sorry, %r not allowed in path" % f)
701
elif (f == '.') or (f == ''):
708
assert isinstance(p, list)
710
if (f == '..') or (f is None) or (f == ''):
711
raise errors.BzrError("sorry, %r not allowed in path" % f)
715
@deprecated_function(zero_nine)
716
def appendpath(p1, p2):
720
return pathjoin(p1, p2)
724
"""Split s into lines, but without removing the newline characters."""
725
lines = s.split('\n')
726
result = [line + '\n' for line in lines[:-1]]
728
result.append(lines[-1])
732
def hardlinks_good():
733
return sys.platform not in ('win32', 'cygwin', 'darwin')
736
def link_or_copy(src, dest):
737
"""Hardlink a file, or copy it if it can't be hardlinked."""
738
if not hardlinks_good():
739
shutil.copyfile(src, dest)
743
except (OSError, IOError), e:
744
if e.errno != errno.EXDEV:
746
shutil.copyfile(src, dest)
748
def delete_any(full_path):
749
"""Delete a file or directory."""
753
# We may be renaming a dangling inventory id
754
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
760
if getattr(os, 'symlink', None) is not None:
766
def contains_whitespace(s):
767
"""True if there are any whitespace characters in s."""
768
for ch in string.whitespace:
775
def contains_linebreaks(s):
776
"""True if there is any vertical whitespace in s."""
784
def relpath(base, path):
785
"""Return path relative to base, or raise exception.
787
The path may be either an absolute path or a path relative to the
788
current working directory.
790
os.path.commonprefix (python2.4) has a bad bug that it works just
791
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
795
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
796
' exceed the platform minimum length (which is %d)' %
803
while len(head) >= len(base):
806
head, tail = os.path.split(head)
810
raise errors.PathNotChild(rp, base)
818
def safe_unicode(unicode_or_utf8_string):
819
"""Coerce unicode_or_utf8_string into unicode.
821
If it is unicode, it is returned.
822
Otherwise it is decoded from utf-8. If a decoding error
823
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
824
as a BzrBadParameter exception.
826
if isinstance(unicode_or_utf8_string, unicode):
827
return unicode_or_utf8_string
829
return unicode_or_utf8_string.decode('utf8')
830
except UnicodeDecodeError:
831
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
834
_platform_normalizes_filenames = False
835
if sys.platform == 'darwin':
836
_platform_normalizes_filenames = True
839
def normalizes_filenames():
840
"""Return True if this platform normalizes unicode filenames.
842
Mac OSX does, Windows/Linux do not.
844
return _platform_normalizes_filenames
847
def _accessible_normalized_filename(path):
848
"""Get the unicode normalized path, and if you can access the file.
850
On platforms where the system normalizes filenames (Mac OSX),
851
you can access a file by any path which will normalize correctly.
852
On platforms where the system does not normalize filenames
853
(Windows, Linux), you have to access a file by its exact path.
855
Internally, bzr only supports NFC/NFKC normalization, since that is
856
the standard for XML documents.
858
So return the normalized path, and a flag indicating if the file
859
can be accessed by that path.
862
return unicodedata.normalize('NFKC', unicode(path)), True
865
def _inaccessible_normalized_filename(path):
866
__doc__ = _accessible_normalized_filename.__doc__
868
normalized = unicodedata.normalize('NFKC', unicode(path))
869
return normalized, normalized == path
872
if _platform_normalizes_filenames:
873
normalized_filename = _accessible_normalized_filename
875
normalized_filename = _inaccessible_normalized_filename
878
def terminal_width():
879
"""Return estimated terminal width."""
880
if sys.platform == 'win32':
881
import bzrlib.win32console
882
return bzrlib.win32console.get_console_size()[0]
885
import struct, fcntl, termios
886
s = struct.pack('HHHH', 0, 0, 0, 0)
887
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
888
width = struct.unpack('HHHH', x)[1]
893
width = int(os.environ['COLUMNS'])
902
def supports_executable():
903
return sys.platform != "win32"
906
def set_or_unset_env(env_variable, value):
907
"""Modify the environment, setting or removing the env_variable.
909
:param env_variable: The environment variable in question
910
:param value: The value to set the environment to. If None, then
911
the variable will be removed.
912
:return: The original value of the environment variable.
914
orig_val = os.environ.get(env_variable)
916
if orig_val is not None:
917
del os.environ[env_variable]
919
if isinstance(value, unicode):
920
value = value.encode(bzrlib.user_encoding)
921
os.environ[env_variable] = value
925
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
928
def check_legal_path(path):
929
"""Check whether the supplied path is legal.
930
This is only required on Windows, so we don't test on other platforms
933
if sys.platform != "win32":
935
if _validWin32PathRE.match(path) is None:
936
raise errors.IllegalPath(path)
939
def walkdirs(top, prefix=""):
940
"""Yield data about all the directories in a tree.
942
This yields all the data about the contents of a directory at a time.
943
After each directory has been yielded, if the caller has mutated the list
944
to exclude some directories, they are then not descended into.
946
The data yielded is of the form:
947
((directory-relpath, directory-path-from-top),
948
[(relpath, basename, kind, lstat), ...]),
949
- directory-relpath is the relative path of the directory being returned
950
with respect to top. prefix is prepended to this.
951
- directory-path-from-root is the path including top for this directory.
952
It is suitable for use with os functions.
953
- relpath is the relative path within the subtree being walked.
954
- basename is the basename of the path
955
- kind is the kind of the file now. If unknown then the file is not
956
present within the tree - but it may be recorded as versioned. See
958
- lstat is the stat data *if* the file was statted.
959
- planned, not implemented:
960
path_from_tree_root is the path from the root of the tree.
962
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
963
allows one to walk a subtree but get paths that are relative to a tree
965
:return: an iterator over the dirs.
967
#TODO there is a bit of a smell where the results of the directory-
968
# summary in this, and the path from the root, may not agree
969
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
970
# potentially confusing output. We should make this more robust - but
971
# not at a speed cost. RBC 20060731
974
_directory = _directory_kind
975
_listdir = os.listdir
976
pending = [(prefix, "", _directory, None, top)]
979
currentdir = pending.pop()
980
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
983
relroot = currentdir[0] + '/'
986
for name in sorted(_listdir(top)):
987
abspath = top + '/' + name
988
statvalue = lstat(abspath)
989
dirblock.append((relroot + name, name,
990
file_kind_from_stat_mode(statvalue.st_mode),
992
yield (currentdir[0], top), dirblock
993
# push the user specified dirs from dirblock
994
for dir in reversed(dirblock):
995
if dir[2] == _directory:
999
def copy_tree(from_path, to_path, handlers={}):
1000
"""Copy all of the entries in from_path into to_path.
1002
:param from_path: The base directory to copy.
1003
:param to_path: The target directory. If it does not exist, it will
1005
:param handlers: A dictionary of functions, which takes a source and
1006
destinations for files, directories, etc.
1007
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1008
'file', 'directory', and 'symlink' should always exist.
1009
If they are missing, they will be replaced with 'os.mkdir()',
1010
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1012
# Now, just copy the existing cached tree to the new location
1013
# We use a cheap trick here.
1014
# Absolute paths are prefixed with the first parameter
1015
# relative paths are prefixed with the second.
1016
# So we can get both the source and target returned
1017
# without any extra work.
1019
def copy_dir(source, dest):
1022
def copy_link(source, dest):
1023
"""Copy the contents of a symlink"""
1024
link_to = os.readlink(source)
1025
os.symlink(link_to, dest)
1027
real_handlers = {'file':shutil.copy2,
1028
'symlink':copy_link,
1029
'directory':copy_dir,
1031
real_handlers.update(handlers)
1033
if not os.path.exists(to_path):
1034
real_handlers['directory'](from_path, to_path)
1036
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1037
for relpath, name, kind, st, abspath in entries:
1038
real_handlers[kind](abspath, relpath)
1041
def path_prefix_key(path):
1042
"""Generate a prefix-order path key for path.
1044
This can be used to sort paths in the same way that walkdirs does.
1046
return (dirname(path) , path)
1049
def compare_paths_prefix_order(path_a, path_b):
1050
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1051
key_a = path_prefix_key(path_a)
1052
key_b = path_prefix_key(path_b)
1053
return cmp(key_a, key_b)
1056
_cached_user_encoding = None
1059
def get_user_encoding():
1060
"""Find out what the preferred user encoding is.
1062
This is generally the encoding that is used for command line parameters
1063
and file contents. This may be different from the terminal encoding
1064
or the filesystem encoding.
1066
:return: A string defining the preferred user encoding
1068
global _cached_user_encoding
1069
if _cached_user_encoding is not None:
1070
return _cached_user_encoding
1072
if sys.platform == 'darwin':
1073
# work around egregious python 2.4 bug
1074
sys.platform = 'posix'
1078
sys.platform = 'darwin'
1083
_cached_user_encoding = locale.getpreferredencoding()
1084
except locale.Error, e:
1085
sys.stderr.write('bzr: warning: %s\n'
1086
' Could not determine what text encoding to use.\n'
1087
' This error usually means your Python interpreter\n'
1088
' doesn\'t support the locale set by $LANG (%s)\n'
1089
" Continuing with ascii encoding.\n"
1090
% (e, os.environ.get('LANG')))
1092
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1093
# treat that as ASCII, and not support printing unicode characters to the
1095
if _cached_user_encoding in (None, 'cp0'):
1096
_cached_user_encoding = 'ascii'
1097
return _cached_user_encoding
1100
def recv_all(socket, bytes):
1101
"""Receive an exact number of bytes.
1103
Regular Socket.recv() may return less than the requested number of bytes,
1104
dependning on what's in the OS buffer. MSG_WAITALL is not available
1105
on all platforms, but this should work everywhere. This will return
1106
less than the requested amount if the remote end closes.
1108
This isn't optimized and is intended mostly for use in testing.
1111
while len(b) < bytes:
1112
new = socket.recv(bytes - len(b))
1118
def dereference_path(path):
1119
"""Determine the real path to a file.
1121
All parent elements are dereferenced. But the file itself is not
1123
:param path: The original path. May be absolute or relative.
1124
:return: the real path *to* the file
1126
parent, base = os.path.split(path)
1127
# The pathjoin for '.' is a workaround for Python bug #1213894.
1128
# (initial path components aren't dereferenced)
1129
return pathjoin(realpath(pathjoin('.', parent)), base)