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(), """
29
from datetime import datetime
31
from ntpath import (abspath as _nt_abspath,
33
normpath as _nt_normpath,
34
realpath as _nt_realpath,
35
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)
385
codecs.lookup(output_encoding)
387
sys.stderr.write('bzr: warning:'
388
' unknown terminal encoding %s.\n'
389
' Using encoding %s instead.\n'
390
% (output_encoding, bzrlib.user_encoding)
392
output_encoding = bzrlib.user_encoding
394
return output_encoding
397
def normalizepath(f):
398
if getattr(os.path, 'realpath', None) is not None:
402
[p,e] = os.path.split(f)
403
if e == "" or e == "." or e == "..":
406
return pathjoin(F(p), e)
410
"""Copy a file to a backup.
412
Backups are named in GNU-style, with a ~ suffix.
414
If the file is already a backup, it's not copied.
420
if has_symlinks() and os.path.islink(fn):
421
target = os.readlink(fn)
422
os.symlink(target, bfn)
430
outf = file(bfn, 'wb')
438
"""True if f is an accessible directory."""
440
return S_ISDIR(os.lstat(f)[ST_MODE])
446
"""True if f is a regular file."""
448
return S_ISREG(os.lstat(f)[ST_MODE])
453
"""True if f is a symlink."""
455
return S_ISLNK(os.lstat(f)[ST_MODE])
459
def is_inside(dir, fname):
460
"""True if fname is inside dir.
462
The parameters should typically be passed to osutils.normpath first, so
463
that . and .. and repeated slashes are eliminated, and the separators
464
are canonical for the platform.
466
The empty string as a dir name is taken as top-of-tree and matches
469
# XXX: Most callers of this can actually do something smarter by
470
# looking at the inventory
480
return fname.startswith(dir)
483
def is_inside_any(dir_list, fname):
484
"""True if fname is inside any of given dirs."""
485
for dirname in dir_list:
486
if is_inside(dirname, fname):
492
def is_inside_or_parent_of_any(dir_list, fname):
493
"""True if fname is a child or a parent of any of the given files."""
494
for dirname in dir_list:
495
if is_inside(dirname, fname) or is_inside(fname, dirname):
501
def pumpfile(fromfile, tofile):
502
"""Copy contents of one file to another."""
505
b = fromfile.read(BUFSIZE)
511
def file_iterator(input_file, readsize=32768):
513
b = input_file.read(readsize)
520
if getattr(f, 'tell', None) is not None:
533
def sha_strings(strings):
534
"""Return the sha-1 of concatenation of strings"""
536
map(s.update, strings)
546
def fingerprint_file(f):
551
return {'size': size,
552
'sha1': s.hexdigest()}
555
def compare_files(a, b):
556
"""Returns true if equal in contents"""
567
def local_time_offset(t=None):
568
"""Return offset of local zone from GMT, either at present or at time t."""
571
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
572
return offset.days * 86400 + offset.seconds
575
def format_date(t, offset=0, timezone='original', date_fmt=None,
577
## TODO: Perhaps a global option to use either universal or local time?
578
## Or perhaps just let people set $TZ?
579
assert isinstance(t, float)
581
if timezone == 'utc':
584
elif timezone == 'original':
587
tt = time.gmtime(t + offset)
588
elif timezone == 'local':
589
tt = time.localtime(t)
590
offset = local_time_offset(t)
592
raise errors.BzrError("unsupported timezone format %r" % timezone,
593
['options are "utc", "original", "local"'])
595
date_fmt = "%a %Y-%m-%d %H:%M:%S"
597
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
600
return (time.strftime(date_fmt, tt) + offset_str)
603
def compact_date(when):
604
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
607
def format_delta(delta):
608
"""Get a nice looking string for a time delta.
610
:param delta: The time difference in seconds, can be positive or negative.
611
positive indicates time in the past, negative indicates time in the
612
future. (usually time.time() - stored_time)
613
:return: String formatted to show approximate resolution
619
direction = 'in the future'
623
if seconds < 90: # print seconds up to 90 seconds
625
return '%d second %s' % (seconds, direction,)
627
return '%d seconds %s' % (seconds, direction)
629
minutes = int(seconds / 60)
630
seconds -= 60 * minutes
635
if minutes < 90: # print minutes, seconds up to 90 minutes
637
return '%d minute, %d second%s %s' % (
638
minutes, seconds, plural_seconds, direction)
640
return '%d minutes, %d second%s %s' % (
641
minutes, seconds, plural_seconds, direction)
643
hours = int(minutes / 60)
644
minutes -= 60 * hours
651
return '%d hour, %d minute%s %s' % (hours, minutes,
652
plural_minutes, direction)
653
return '%d hours, %d minute%s %s' % (hours, minutes,
654
plural_minutes, direction)
657
"""Return size of given open file."""
658
return os.fstat(f.fileno())[ST_SIZE]
661
# Define rand_bytes based on platform.
663
# Python 2.4 and later have os.urandom,
664
# but it doesn't work on some arches
666
rand_bytes = os.urandom
667
except (NotImplementedError, AttributeError):
668
# If python doesn't have os.urandom, or it doesn't work,
669
# then try to first pull random data from /dev/urandom
671
rand_bytes = file('/dev/urandom', 'rb').read
672
# Otherwise, use this hack as a last resort
673
except (IOError, OSError):
674
# not well seeded, but better than nothing
679
s += chr(random.randint(0, 255))
684
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
686
"""Return a random string of num alphanumeric characters
688
The result only contains lowercase chars because it may be used on
689
case-insensitive filesystems.
692
for raw_byte in rand_bytes(num):
693
s += ALNUM[ord(raw_byte) % 36]
697
## TODO: We could later have path objects that remember their list
698
## decomposition (might be too tricksy though.)
701
"""Turn string into list of parts."""
702
assert isinstance(p, basestring)
704
# split on either delimiter because people might use either on
706
ps = re.split(r'[\\/]', p)
711
raise errors.BzrError("sorry, %r not allowed in path" % f)
712
elif (f == '.') or (f == ''):
719
assert isinstance(p, list)
721
if (f == '..') or (f is None) or (f == ''):
722
raise errors.BzrError("sorry, %r not allowed in path" % f)
726
@deprecated_function(zero_nine)
727
def appendpath(p1, p2):
731
return pathjoin(p1, p2)
735
"""Split s into lines, but without removing the newline characters."""
736
lines = s.split('\n')
737
result = [line + '\n' for line in lines[:-1]]
739
result.append(lines[-1])
743
def hardlinks_good():
744
return sys.platform not in ('win32', 'cygwin', 'darwin')
747
def link_or_copy(src, dest):
748
"""Hardlink a file, or copy it if it can't be hardlinked."""
749
if not hardlinks_good():
750
shutil.copyfile(src, dest)
754
except (OSError, IOError), e:
755
if e.errno != errno.EXDEV:
757
shutil.copyfile(src, dest)
759
def delete_any(full_path):
760
"""Delete a file or directory."""
764
# We may be renaming a dangling inventory id
765
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
771
if getattr(os, 'symlink', None) is not None:
777
def contains_whitespace(s):
778
"""True if there are any whitespace characters in s."""
779
# string.whitespace can include '\xa0' in certain locales, because it is
780
# considered "non-breaking-space" as part of ISO-8859-1. But it
781
# 1) Isn't a breaking whitespace
782
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
784
# 3) '\xa0' isn't unicode safe since it is >128.
785
# So we are following textwrap's example and hard-coding our own.
786
# We probably could ignore \v and \f, too.
787
for ch in u' \t\n\r\v\f':
794
def contains_linebreaks(s):
795
"""True if there is any vertical whitespace in s."""
803
def relpath(base, path):
804
"""Return path relative to base, or raise exception.
806
The path may be either an absolute path or a path relative to the
807
current working directory.
809
os.path.commonprefix (python2.4) has a bad bug that it works just
810
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
814
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
815
' exceed the platform minimum length (which is %d)' %
822
while len(head) >= len(base):
825
head, tail = os.path.split(head)
829
raise errors.PathNotChild(rp, base)
837
def safe_unicode(unicode_or_utf8_string):
838
"""Coerce unicode_or_utf8_string into unicode.
840
If it is unicode, it is returned.
841
Otherwise it is decoded from utf-8. If a decoding error
842
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
843
as a BzrBadParameter exception.
845
if isinstance(unicode_or_utf8_string, unicode):
846
return unicode_or_utf8_string
848
return unicode_or_utf8_string.decode('utf8')
849
except UnicodeDecodeError:
850
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
853
_platform_normalizes_filenames = False
854
if sys.platform == 'darwin':
855
_platform_normalizes_filenames = True
858
def normalizes_filenames():
859
"""Return True if this platform normalizes unicode filenames.
861
Mac OSX does, Windows/Linux do not.
863
return _platform_normalizes_filenames
866
def _accessible_normalized_filename(path):
867
"""Get the unicode normalized path, and if you can access the file.
869
On platforms where the system normalizes filenames (Mac OSX),
870
you can access a file by any path which will normalize correctly.
871
On platforms where the system does not normalize filenames
872
(Windows, Linux), you have to access a file by its exact path.
874
Internally, bzr only supports NFC/NFKC normalization, since that is
875
the standard for XML documents.
877
So return the normalized path, and a flag indicating if the file
878
can be accessed by that path.
881
return unicodedata.normalize('NFKC', unicode(path)), True
884
def _inaccessible_normalized_filename(path):
885
__doc__ = _accessible_normalized_filename.__doc__
887
normalized = unicodedata.normalize('NFKC', unicode(path))
888
return normalized, normalized == path
891
if _platform_normalizes_filenames:
892
normalized_filename = _accessible_normalized_filename
894
normalized_filename = _inaccessible_normalized_filename
897
def terminal_width():
898
"""Return estimated terminal width."""
899
if sys.platform == 'win32':
900
import bzrlib.win32console
901
return bzrlib.win32console.get_console_size()[0]
904
import struct, fcntl, termios
905
s = struct.pack('HHHH', 0, 0, 0, 0)
906
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
907
width = struct.unpack('HHHH', x)[1]
912
width = int(os.environ['COLUMNS'])
921
def supports_executable():
922
return sys.platform != "win32"
925
def supports_posix_readonly():
926
"""Return True if 'readonly' has POSIX semantics, False otherwise.
928
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
929
directory controls creation/deletion, etc.
931
And under win32, readonly means that the directory itself cannot be
932
deleted. The contents of a readonly directory can be changed, unlike POSIX
933
where files in readonly directories cannot be added, deleted or renamed.
935
return sys.platform != "win32"
938
def set_or_unset_env(env_variable, value):
939
"""Modify the environment, setting or removing the env_variable.
941
:param env_variable: The environment variable in question
942
:param value: The value to set the environment to. If None, then
943
the variable will be removed.
944
:return: The original value of the environment variable.
946
orig_val = os.environ.get(env_variable)
948
if orig_val is not None:
949
del os.environ[env_variable]
951
if isinstance(value, unicode):
952
value = value.encode(bzrlib.user_encoding)
953
os.environ[env_variable] = value
957
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
960
def check_legal_path(path):
961
"""Check whether the supplied path is legal.
962
This is only required on Windows, so we don't test on other platforms
965
if sys.platform != "win32":
967
if _validWin32PathRE.match(path) is None:
968
raise errors.IllegalPath(path)
971
def walkdirs(top, prefix=""):
972
"""Yield data about all the directories in a tree.
974
This yields all the data about the contents of a directory at a time.
975
After each directory has been yielded, if the caller has mutated the list
976
to exclude some directories, they are then not descended into.
978
The data yielded is of the form:
979
((directory-relpath, directory-path-from-top),
980
[(relpath, basename, kind, lstat), ...]),
981
- directory-relpath is the relative path of the directory being returned
982
with respect to top. prefix is prepended to this.
983
- directory-path-from-root is the path including top for this directory.
984
It is suitable for use with os functions.
985
- relpath is the relative path within the subtree being walked.
986
- basename is the basename of the path
987
- kind is the kind of the file now. If unknown then the file is not
988
present within the tree - but it may be recorded as versioned. See
990
- lstat is the stat data *if* the file was statted.
991
- planned, not implemented:
992
path_from_tree_root is the path from the root of the tree.
994
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
995
allows one to walk a subtree but get paths that are relative to a tree
997
:return: an iterator over the dirs.
999
#TODO there is a bit of a smell where the results of the directory-
1000
# summary in this, and the path from the root, may not agree
1001
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1002
# potentially confusing output. We should make this more robust - but
1003
# not at a speed cost. RBC 20060731
1006
_directory = _directory_kind
1007
_listdir = os.listdir
1008
pending = [(prefix, "", _directory, None, top)]
1011
currentdir = pending.pop()
1012
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1015
relroot = currentdir[0] + '/'
1018
for name in sorted(_listdir(top)):
1019
abspath = top + '/' + name
1020
statvalue = lstat(abspath)
1021
dirblock.append((relroot + name, name,
1022
file_kind_from_stat_mode(statvalue.st_mode),
1023
statvalue, abspath))
1024
yield (currentdir[0], top), dirblock
1025
# push the user specified dirs from dirblock
1026
for dir in reversed(dirblock):
1027
if dir[2] == _directory:
1031
def copy_tree(from_path, to_path, handlers={}):
1032
"""Copy all of the entries in from_path into to_path.
1034
:param from_path: The base directory to copy.
1035
:param to_path: The target directory. If it does not exist, it will
1037
:param handlers: A dictionary of functions, which takes a source and
1038
destinations for files, directories, etc.
1039
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1040
'file', 'directory', and 'symlink' should always exist.
1041
If they are missing, they will be replaced with 'os.mkdir()',
1042
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1044
# Now, just copy the existing cached tree to the new location
1045
# We use a cheap trick here.
1046
# Absolute paths are prefixed with the first parameter
1047
# relative paths are prefixed with the second.
1048
# So we can get both the source and target returned
1049
# without any extra work.
1051
def copy_dir(source, dest):
1054
def copy_link(source, dest):
1055
"""Copy the contents of a symlink"""
1056
link_to = os.readlink(source)
1057
os.symlink(link_to, dest)
1059
real_handlers = {'file':shutil.copy2,
1060
'symlink':copy_link,
1061
'directory':copy_dir,
1063
real_handlers.update(handlers)
1065
if not os.path.exists(to_path):
1066
real_handlers['directory'](from_path, to_path)
1068
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1069
for relpath, name, kind, st, abspath in entries:
1070
real_handlers[kind](abspath, relpath)
1073
def path_prefix_key(path):
1074
"""Generate a prefix-order path key for path.
1076
This can be used to sort paths in the same way that walkdirs does.
1078
return (dirname(path) , path)
1081
def compare_paths_prefix_order(path_a, path_b):
1082
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1083
key_a = path_prefix_key(path_a)
1084
key_b = path_prefix_key(path_b)
1085
return cmp(key_a, key_b)
1088
_cached_user_encoding = None
1091
def get_user_encoding(use_cache=True):
1092
"""Find out what the preferred user encoding is.
1094
This is generally the encoding that is used for command line parameters
1095
and file contents. This may be different from the terminal encoding
1096
or the filesystem encoding.
1098
:param use_cache: Enable cache for detected encoding.
1099
(This parameter is turned on by default,
1100
and required only for selftesting)
1102
:return: A string defining the preferred user encoding
1104
global _cached_user_encoding
1105
if _cached_user_encoding is not None and use_cache:
1106
return _cached_user_encoding
1108
if sys.platform == 'darwin':
1109
# work around egregious python 2.4 bug
1110
sys.platform = 'posix'
1114
sys.platform = 'darwin'
1119
user_encoding = locale.getpreferredencoding()
1120
except locale.Error, e:
1121
sys.stderr.write('bzr: warning: %s\n'
1122
' Could not determine what text encoding to use.\n'
1123
' This error usually means your Python interpreter\n'
1124
' doesn\'t support the locale set by $LANG (%s)\n'
1125
" Continuing with ascii encoding.\n"
1126
% (e, os.environ.get('LANG')))
1127
user_encoding = 'ascii'
1129
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1130
# treat that as ASCII, and not support printing unicode characters to the
1132
if user_encoding in (None, 'cp0'):
1133
user_encoding = 'ascii'
1137
codecs.lookup(user_encoding)
1139
sys.stderr.write('bzr: warning:'
1140
' unknown encoding %s.'
1141
' Continuing with ascii encoding.\n'
1144
user_encoding = 'ascii'
1147
_cached_user_encoding = user_encoding
1149
return user_encoding
1152
def recv_all(socket, bytes):
1153
"""Receive an exact number of bytes.
1155
Regular Socket.recv() may return less than the requested number of bytes,
1156
dependning on what's in the OS buffer. MSG_WAITALL is not available
1157
on all platforms, but this should work everywhere. This will return
1158
less than the requested amount if the remote end closes.
1160
This isn't optimized and is intended mostly for use in testing.
1163
while len(b) < bytes:
1164
new = socket.recv(bytes - len(b))
1170
def dereference_path(path):
1171
"""Determine the real path to a file.
1173
All parent elements are dereferenced. But the file itself is not
1175
:param path: The original path. May be absolute or relative.
1176
:return: the real path *to* the file
1178
parent, base = os.path.split(path)
1179
# The pathjoin for '.' is a workaround for Python bug #1213894.
1180
# (initial path components aren't dereferenced)
1181
return pathjoin(realpath(pathjoin('.', parent)), base)
added
removed
bzrlib/osutils.py
