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 (
56
from bzrlib.symbol_versioning import (
60
from bzrlib.trace import mutter
63
# On win32, O_BINARY is used to indicate the file should
64
# be opened in binary mode, rather than text mode.
65
# On other platforms, O_BINARY doesn't exist, because
66
# they always open in binary mode, so it is okay to
67
# OR with 0 on those platforms
68
O_BINARY = getattr(os, 'O_BINARY', 0)
71
def make_readonly(filename):
72
"""Make a filename read-only."""
73
mod = os.stat(filename).st_mode
75
os.chmod(filename, mod)
78
def make_writable(filename):
79
mod = os.stat(filename).st_mode
81
os.chmod(filename, mod)
88
"""Return a quoted filename filename
90
This previously used backslash quoting, but that works poorly on
92
# TODO: I'm not really sure this is the best format either.x
95
_QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')
97
if _QUOTE_RE.search(f):
103
_directory_kind = 'directory'
106
stat.S_IFDIR:_directory_kind,
107
stat.S_IFCHR:'chardev',
108
stat.S_IFBLK:'block',
111
stat.S_IFLNK:'symlink',
112
stat.S_IFSOCK:'socket',
116
def file_kind_from_stat_mode(stat_mode, _formats=_formats, _unknown='unknown'):
117
"""Generate a file kind from a stat mode. This is used in walkdirs.
119
Its performance is critical: Do not mutate without careful benchmarking.
122
return _formats[stat_mode & 0170000]
127
def file_kind(f, _lstat=os.lstat, _mapper=file_kind_from_stat_mode):
129
return _mapper(_lstat(f).st_mode)
131
if getattr(e, 'errno', None) == errno.ENOENT:
132
raise errors.NoSuchFile(f)
137
"""Return the current umask"""
138
# Assume that people aren't messing with the umask while running
139
# XXX: This is not thread safe, but there is no way to get the
140
# umask without setting it
146
def kind_marker(kind):
149
elif kind == _directory_kind:
151
elif kind == 'symlink':
154
raise errors.BzrError('invalid file kind %r' % kind)
156
lexists = getattr(os.path, 'lexists', None)
160
if getattr(os, 'lstat') is not None:
166
if e.errno == errno.ENOENT:
169
raise errors.BzrError("lstat/stat of (%r): %r" % (f, e))
172
def fancy_rename(old, new, rename_func, unlink_func):
173
"""A fancy rename, when you don't have atomic rename.
175
:param old: The old path, to rename from
176
:param new: The new path, to rename to
177
:param rename_func: The potentially non-atomic rename function
178
:param unlink_func: A way to delete the target file if the full rename succeeds
181
# sftp rename doesn't allow overwriting, so play tricks:
183
base = os.path.basename(new)
184
dirname = os.path.dirname(new)
185
tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))
186
tmp_name = pathjoin(dirname, tmp_name)
188
# Rename the file out of the way, but keep track if it didn't exist
189
# We don't want to grab just any exception
190
# something like EACCES should prevent us from continuing
191
# The downside is that the rename_func has to throw an exception
192
# with an errno = ENOENT, or NoSuchFile
195
rename_func(new, tmp_name)
196
except (errors.NoSuchFile,), e:
199
# RBC 20060103 abstraction leakage: the paramiko SFTP clients rename
200
# function raises an IOError with errno is None when a rename fails.
201
# This then gets caught here.
202
if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):
205
if (getattr(e, 'errno', None) is None
206
or e.errno not in (errno.ENOENT, errno.ENOTDIR)):
213
# This may throw an exception, in which case success will
215
rename_func(old, new)
219
# If the file used to exist, rename it back into place
220
# otherwise just delete it from the tmp location
222
unlink_func(tmp_name)
224
rename_func(tmp_name, new)
227
# In Python 2.4.2 and older, os.path.abspath and os.path.realpath
228
# choke on a Unicode string containing a relative path if
229
# os.getcwd() returns a non-sys.getdefaultencoding()-encoded
231
_fs_enc = sys.getfilesystemencoding() or 'utf-8'
232
def _posix_abspath(path):
233
# jam 20060426 rather than encoding to fsencoding
234
# copy posixpath.abspath, but use os.getcwdu instead
235
if not posixpath.isabs(path):
236
path = posixpath.join(getcwd(), path)
237
return posixpath.normpath(path)
240
def _posix_realpath(path):
241
return posixpath.realpath(path.encode(_fs_enc)).decode(_fs_enc)
244
def _win32_fixdrive(path):
245
"""Force drive letters to be consistent.
247
win32 is inconsistent whether it returns lower or upper case
248
and even if it was consistent the user might type the other
249
so we force it to uppercase
250
running python.exe under cmd.exe return capital C:\\
251
running win32 python inside a cygwin shell returns lowercase c:\\
253
drive, path = _nt_splitdrive(path)
254
return drive.upper() + path
257
def _win32_abspath(path):
258
# Real _nt_abspath doesn't have a problem with a unicode cwd
259
return _win32_fixdrive(_nt_abspath(unicode(path)).replace('\\', '/'))
262
def _win98_abspath(path):
263
"""Return the absolute version of a path.
264
Windows 98 safe implementation (python reimplementation
265
of Win32 API function GetFullPathNameW)
270
# \\HOST\path => //HOST/path
271
# //HOST/path => //HOST/path
272
# path => C:/cwd/path
275
# check for absolute path
276
drive = _nt_splitdrive(path)[0]
277
if drive == '' and path[:2] not in('//','\\\\'):
279
# we cannot simply os.path.join cwd and path
280
# because os.path.join('C:','/path') produce '/path'
281
# and this is incorrect
282
if path[:1] in ('/','\\'):
283
cwd = _nt_splitdrive(cwd)[0]
284
path = cwd + '\\' + path
285
return _win32_fixdrive(_nt_normpath(path).replace('\\', '/'))
287
if win32utils.winver == 'Windows 98':
288
_win32_abspath = _win98_abspath
291
def _win32_realpath(path):
292
# Real _nt_realpath doesn't have a problem with a unicode cwd
293
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
296
def _win32_pathjoin(*args):
297
return _nt_join(*args).replace('\\', '/')
300
def _win32_normpath(path):
301
return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))
305
return _win32_fixdrive(os.getcwdu().replace('\\', '/'))
308
def _win32_mkdtemp(*args, **kwargs):
309
return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))
312
def _win32_rename(old, new):
313
"""We expect to be able to atomically replace 'new' with old.
315
On win32, if new exists, it must be moved out of the way first,
319
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
321
if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):
322
# If we try to rename a non-existant file onto cwd, we get
323
# EPERM or EACCES instead of ENOENT, this will raise ENOENT
324
# if the old path doesn't exist, sometimes we get EACCES
325
# On Linux, we seem to get EBUSY, on Mac we get EINVAL
331
return unicodedata.normalize('NFKC', os.getcwdu())
334
# Default is to just use the python builtins, but these can be rebound on
335
# particular platforms.
336
abspath = _posix_abspath
337
realpath = _posix_realpath
338
pathjoin = os.path.join
339
normpath = os.path.normpath
342
dirname = os.path.dirname
343
basename = os.path.basename
344
split = os.path.split
345
splitext = os.path.splitext
346
# These were already imported into local scope
347
# mkdtemp = tempfile.mkdtemp
348
# rmtree = shutil.rmtree
350
MIN_ABS_PATHLENGTH = 1
353
if sys.platform == 'win32':
354
abspath = _win32_abspath
355
realpath = _win32_realpath
356
pathjoin = _win32_pathjoin
357
normpath = _win32_normpath
358
getcwd = _win32_getcwd
359
mkdtemp = _win32_mkdtemp
360
rename = _win32_rename
362
MIN_ABS_PATHLENGTH = 3
364
def _win32_delete_readonly(function, path, excinfo):
365
"""Error handler for shutil.rmtree function [for win32]
366
Helps to remove files and dirs marked as read-only.
368
exception = excinfo[1]
369
if function in (os.remove, os.rmdir) \
370
and isinstance(exception, OSError) \
371
and exception.errno == errno.EACCES:
377
def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):
378
"""Replacer for shutil.rmtree: could remove readonly dirs/files"""
379
return shutil.rmtree(path, ignore_errors, onerror)
380
elif sys.platform == 'darwin':
384
def get_terminal_encoding():
385
"""Find the best encoding for printing to the screen.
387
This attempts to check both sys.stdout and sys.stdin to see
388
what encoding they are in, and if that fails it falls back to
389
bzrlib.user_encoding.
390
The problem is that on Windows, locale.getpreferredencoding()
391
is not the same encoding as that used by the console:
392
http://mail.python.org/pipermail/python-list/2003-May/162357.html
394
On my standard US Windows XP, the preferred encoding is
395
cp1252, but the console is cp437
397
output_encoding = getattr(sys.stdout, 'encoding', None)
398
if not output_encoding:
399
input_encoding = getattr(sys.stdin, 'encoding', None)
400
if not input_encoding:
401
output_encoding = bzrlib.user_encoding
402
mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)
404
output_encoding = input_encoding
405
mutter('encoding stdout as sys.stdin encoding %r', output_encoding)
407
mutter('encoding stdout as sys.stdout encoding %r', output_encoding)
408
if output_encoding == 'cp0':
409
# invalid encoding (cp0 means 'no codepage' on Windows)
410
output_encoding = bzrlib.user_encoding
411
mutter('cp0 is invalid encoding.'
412
' encoding stdout as bzrlib.user_encoding %r', output_encoding)
415
codecs.lookup(output_encoding)
417
sys.stderr.write('bzr: warning:'
418
' unknown terminal encoding %s.\n'
419
' Using encoding %s instead.\n'
420
% (output_encoding, bzrlib.user_encoding)
422
output_encoding = bzrlib.user_encoding
424
return output_encoding
427
def normalizepath(f):
428
if getattr(os.path, 'realpath', None) is not None:
432
[p,e] = os.path.split(f)
433
if e == "" or e == "." or e == "..":
436
return pathjoin(F(p), e)
440
"""Copy a file to a backup.
442
Backups are named in GNU-style, with a ~ suffix.
444
If the file is already a backup, it's not copied.
450
if has_symlinks() and os.path.islink(fn):
451
target = os.readlink(fn)
452
os.symlink(target, bfn)
460
outf = file(bfn, 'wb')
468
"""True if f is an accessible directory."""
470
return S_ISDIR(os.lstat(f)[ST_MODE])
476
"""True if f is a regular file."""
478
return S_ISREG(os.lstat(f)[ST_MODE])
483
"""True if f is a symlink."""
485
return S_ISLNK(os.lstat(f)[ST_MODE])
489
def is_inside(dir, fname):
490
"""True if fname is inside dir.
492
The parameters should typically be passed to osutils.normpath first, so
493
that . and .. and repeated slashes are eliminated, and the separators
494
are canonical for the platform.
496
The empty string as a dir name is taken as top-of-tree and matches
499
# XXX: Most callers of this can actually do something smarter by
500
# looking at the inventory
510
return fname.startswith(dir)
513
def is_inside_any(dir_list, fname):
514
"""True if fname is inside any of given dirs."""
515
for dirname in dir_list:
516
if is_inside(dirname, fname):
522
def is_inside_or_parent_of_any(dir_list, fname):
523
"""True if fname is a child or a parent of any of the given files."""
524
for dirname in dir_list:
525
if is_inside(dirname, fname) or is_inside(fname, dirname):
531
def pumpfile(fromfile, tofile):
532
"""Copy contents of one file to another."""
535
b = fromfile.read(BUFSIZE)
541
def file_iterator(input_file, readsize=32768):
543
b = input_file.read(readsize)
550
if getattr(f, 'tell', None) is not None:
563
def sha_strings(strings):
564
"""Return the sha-1 of concatenation of strings"""
566
map(s.update, strings)
576
def fingerprint_file(f):
581
return {'size': size,
582
'sha1': s.hexdigest()}
585
def compare_files(a, b):
586
"""Returns true if equal in contents"""
597
def local_time_offset(t=None):
598
"""Return offset of local zone from GMT, either at present or at time t."""
601
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
602
return offset.days * 86400 + offset.seconds
605
def format_date(t, offset=0, timezone='original', date_fmt=None,
607
## TODO: Perhaps a global option to use either universal or local time?
608
## Or perhaps just let people set $TZ?
609
assert isinstance(t, float)
611
if timezone == 'utc':
614
elif timezone == 'original':
617
tt = time.gmtime(t + offset)
618
elif timezone == 'local':
619
tt = time.localtime(t)
620
offset = local_time_offset(t)
622
raise errors.BzrError("unsupported timezone format %r" % timezone,
623
['options are "utc", "original", "local"'])
625
date_fmt = "%a %Y-%m-%d %H:%M:%S"
627
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
630
return (time.strftime(date_fmt, tt) + offset_str)
633
def compact_date(when):
634
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
637
def format_delta(delta):
638
"""Get a nice looking string for a time delta.
640
:param delta: The time difference in seconds, can be positive or negative.
641
positive indicates time in the past, negative indicates time in the
642
future. (usually time.time() - stored_time)
643
:return: String formatted to show approximate resolution
649
direction = 'in the future'
653
if seconds < 90: # print seconds up to 90 seconds
655
return '%d second %s' % (seconds, direction,)
657
return '%d seconds %s' % (seconds, direction)
659
minutes = int(seconds / 60)
660
seconds -= 60 * minutes
665
if minutes < 90: # print minutes, seconds up to 90 minutes
667
return '%d minute, %d second%s %s' % (
668
minutes, seconds, plural_seconds, direction)
670
return '%d minutes, %d second%s %s' % (
671
minutes, seconds, plural_seconds, direction)
673
hours = int(minutes / 60)
674
minutes -= 60 * hours
681
return '%d hour, %d minute%s %s' % (hours, minutes,
682
plural_minutes, direction)
683
return '%d hours, %d minute%s %s' % (hours, minutes,
684
plural_minutes, direction)
687
"""Return size of given open file."""
688
return os.fstat(f.fileno())[ST_SIZE]
691
# Define rand_bytes based on platform.
693
# Python 2.4 and later have os.urandom,
694
# but it doesn't work on some arches
696
rand_bytes = os.urandom
697
except (NotImplementedError, AttributeError):
698
# If python doesn't have os.urandom, or it doesn't work,
699
# then try to first pull random data from /dev/urandom
701
rand_bytes = file('/dev/urandom', 'rb').read
702
# Otherwise, use this hack as a last resort
703
except (IOError, OSError):
704
# not well seeded, but better than nothing
709
s += chr(random.randint(0, 255))
714
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
716
"""Return a random string of num alphanumeric characters
718
The result only contains lowercase chars because it may be used on
719
case-insensitive filesystems.
722
for raw_byte in rand_bytes(num):
723
s += ALNUM[ord(raw_byte) % 36]
727
## TODO: We could later have path objects that remember their list
728
## decomposition (might be too tricksy though.)
731
"""Turn string into list of parts."""
732
assert isinstance(p, basestring)
734
# split on either delimiter because people might use either on
736
ps = re.split(r'[\\/]', p)
741
raise errors.BzrError("sorry, %r not allowed in path" % f)
742
elif (f == '.') or (f == ''):
749
assert isinstance(p, list)
751
if (f == '..') or (f is None) or (f == ''):
752
raise errors.BzrError("sorry, %r not allowed in path" % f)
756
@deprecated_function(zero_nine)
757
def appendpath(p1, p2):
761
return pathjoin(p1, p2)
765
"""Split s into lines, but without removing the newline characters."""
766
lines = s.split('\n')
767
result = [line + '\n' for line in lines[:-1]]
769
result.append(lines[-1])
773
def hardlinks_good():
774
return sys.platform not in ('win32', 'cygwin', 'darwin')
777
def link_or_copy(src, dest):
778
"""Hardlink a file, or copy it if it can't be hardlinked."""
779
if not hardlinks_good():
780
shutil.copyfile(src, dest)
784
except (OSError, IOError), e:
785
if e.errno != errno.EXDEV:
787
shutil.copyfile(src, dest)
789
def delete_any(full_path):
790
"""Delete a file or directory."""
794
# We may be renaming a dangling inventory id
795
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
801
if getattr(os, 'symlink', None) is not None:
807
def contains_whitespace(s):
808
"""True if there are any whitespace characters in s."""
809
# string.whitespace can include '\xa0' in certain locales, because it is
810
# considered "non-breaking-space" as part of ISO-8859-1. But it
811
# 1) Isn't a breaking whitespace
812
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
814
# 3) '\xa0' isn't unicode safe since it is >128.
815
# So we are following textwrap's example and hard-coding our own.
816
# We probably could ignore \v and \f, too.
817
for ch in u' \t\n\r\v\f':
824
def contains_linebreaks(s):
825
"""True if there is any vertical whitespace in s."""
833
def relpath(base, path):
834
"""Return path relative to base, or raise exception.
836
The path may be either an absolute path or a path relative to the
837
current working directory.
839
os.path.commonprefix (python2.4) has a bad bug that it works just
840
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
844
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
845
' exceed the platform minimum length (which is %d)' %
852
while len(head) >= len(base):
855
head, tail = os.path.split(head)
859
raise errors.PathNotChild(rp, base)
867
def safe_unicode(unicode_or_utf8_string):
868
"""Coerce unicode_or_utf8_string into unicode.
870
If it is unicode, it is returned.
871
Otherwise it is decoded from utf-8. If a decoding error
872
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
873
as a BzrBadParameter exception.
875
if isinstance(unicode_or_utf8_string, unicode):
876
return unicode_or_utf8_string
878
return unicode_or_utf8_string.decode('utf8')
879
except UnicodeDecodeError:
880
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
883
_platform_normalizes_filenames = False
884
if sys.platform == 'darwin':
885
_platform_normalizes_filenames = True
888
def normalizes_filenames():
889
"""Return True if this platform normalizes unicode filenames.
891
Mac OSX does, Windows/Linux do not.
893
return _platform_normalizes_filenames
896
def _accessible_normalized_filename(path):
897
"""Get the unicode normalized path, and if you can access the file.
899
On platforms where the system normalizes filenames (Mac OSX),
900
you can access a file by any path which will normalize correctly.
901
On platforms where the system does not normalize filenames
902
(Windows, Linux), you have to access a file by its exact path.
904
Internally, bzr only supports NFC/NFKC normalization, since that is
905
the standard for XML documents.
907
So return the normalized path, and a flag indicating if the file
908
can be accessed by that path.
911
return unicodedata.normalize('NFKC', unicode(path)), True
914
def _inaccessible_normalized_filename(path):
915
__doc__ = _accessible_normalized_filename.__doc__
917
normalized = unicodedata.normalize('NFKC', unicode(path))
918
return normalized, normalized == path
921
if _platform_normalizes_filenames:
922
normalized_filename = _accessible_normalized_filename
924
normalized_filename = _inaccessible_normalized_filename
927
def terminal_width():
928
"""Return estimated terminal width."""
929
if sys.platform == 'win32':
930
return win32utils.get_console_size()[0]
933
import struct, fcntl, termios
934
s = struct.pack('HHHH', 0, 0, 0, 0)
935
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
936
width = struct.unpack('HHHH', x)[1]
941
width = int(os.environ['COLUMNS'])
950
def supports_executable():
951
return sys.platform != "win32"
954
def supports_posix_readonly():
955
"""Return True if 'readonly' has POSIX semantics, False otherwise.
957
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
958
directory controls creation/deletion, etc.
960
And under win32, readonly means that the directory itself cannot be
961
deleted. The contents of a readonly directory can be changed, unlike POSIX
962
where files in readonly directories cannot be added, deleted or renamed.
964
return sys.platform != "win32"
967
def set_or_unset_env(env_variable, value):
968
"""Modify the environment, setting or removing the env_variable.
970
:param env_variable: The environment variable in question
971
:param value: The value to set the environment to. If None, then
972
the variable will be removed.
973
:return: The original value of the environment variable.
975
orig_val = os.environ.get(env_variable)
977
if orig_val is not None:
978
del os.environ[env_variable]
980
if isinstance(value, unicode):
981
value = value.encode(bzrlib.user_encoding)
982
os.environ[env_variable] = value
986
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
989
def check_legal_path(path):
990
"""Check whether the supplied path is legal.
991
This is only required on Windows, so we don't test on other platforms
994
if sys.platform != "win32":
996
if _validWin32PathRE.match(path) is None:
997
raise errors.IllegalPath(path)
1000
def walkdirs(top, prefix=""):
1001
"""Yield data about all the directories in a tree.
1003
This yields all the data about the contents of a directory at a time.
1004
After each directory has been yielded, if the caller has mutated the list
1005
to exclude some directories, they are then not descended into.
1007
The data yielded is of the form:
1008
((directory-relpath, directory-path-from-top),
1009
[(relpath, basename, kind, lstat), ...]),
1010
- directory-relpath is the relative path of the directory being returned
1011
with respect to top. prefix is prepended to this.
1012
- directory-path-from-root is the path including top for this directory.
1013
It is suitable for use with os functions.
1014
- relpath is the relative path within the subtree being walked.
1015
- basename is the basename of the path
1016
- kind is the kind of the file now. If unknown then the file is not
1017
present within the tree - but it may be recorded as versioned. See
1019
- lstat is the stat data *if* the file was statted.
1020
- planned, not implemented:
1021
path_from_tree_root is the path from the root of the tree.
1023
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1024
allows one to walk a subtree but get paths that are relative to a tree
1026
:return: an iterator over the dirs.
1028
#TODO there is a bit of a smell where the results of the directory-
1029
# summary in this, and the path from the root, may not agree
1030
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1031
# potentially confusing output. We should make this more robust - but
1032
# not at a speed cost. RBC 20060731
1035
_directory = _directory_kind
1036
_listdir = os.listdir
1037
pending = [(prefix, "", _directory, None, top)]
1040
currentdir = pending.pop()
1041
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1044
relroot = currentdir[0] + '/'
1047
for name in sorted(_listdir(top)):
1048
abspath = top + '/' + name
1049
statvalue = lstat(abspath)
1050
dirblock.append((relroot + name, name,
1051
file_kind_from_stat_mode(statvalue.st_mode),
1052
statvalue, abspath))
1053
yield (currentdir[0], top), dirblock
1054
# push the user specified dirs from dirblock
1055
for dir in reversed(dirblock):
1056
if dir[2] == _directory:
1060
def copy_tree(from_path, to_path, handlers={}):
1061
"""Copy all of the entries in from_path into to_path.
1063
:param from_path: The base directory to copy.
1064
:param to_path: The target directory. If it does not exist, it will
1066
:param handlers: A dictionary of functions, which takes a source and
1067
destinations for files, directories, etc.
1068
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1069
'file', 'directory', and 'symlink' should always exist.
1070
If they are missing, they will be replaced with 'os.mkdir()',
1071
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1073
# Now, just copy the existing cached tree to the new location
1074
# We use a cheap trick here.
1075
# Absolute paths are prefixed with the first parameter
1076
# relative paths are prefixed with the second.
1077
# So we can get both the source and target returned
1078
# without any extra work.
1080
def copy_dir(source, dest):
1083
def copy_link(source, dest):
1084
"""Copy the contents of a symlink"""
1085
link_to = os.readlink(source)
1086
os.symlink(link_to, dest)
1088
real_handlers = {'file':shutil.copy2,
1089
'symlink':copy_link,
1090
'directory':copy_dir,
1092
real_handlers.update(handlers)
1094
if not os.path.exists(to_path):
1095
real_handlers['directory'](from_path, to_path)
1097
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1098
for relpath, name, kind, st, abspath in entries:
1099
real_handlers[kind](abspath, relpath)
1102
def path_prefix_key(path):
1103
"""Generate a prefix-order path key for path.
1105
This can be used to sort paths in the same way that walkdirs does.
1107
return (dirname(path) , path)
1110
def compare_paths_prefix_order(path_a, path_b):
1111
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1112
key_a = path_prefix_key(path_a)
1113
key_b = path_prefix_key(path_b)
1114
return cmp(key_a, key_b)
1117
_cached_user_encoding = None
1120
def get_user_encoding(use_cache=True):
1121
"""Find out what the preferred user encoding is.
1123
This is generally the encoding that is used for command line parameters
1124
and file contents. This may be different from the terminal encoding
1125
or the filesystem encoding.
1127
:param use_cache: Enable cache for detected encoding.
1128
(This parameter is turned on by default,
1129
and required only for selftesting)
1131
:return: A string defining the preferred user encoding
1133
global _cached_user_encoding
1134
if _cached_user_encoding is not None and use_cache:
1135
return _cached_user_encoding
1137
if sys.platform == 'darwin':
1138
# work around egregious python 2.4 bug
1139
sys.platform = 'posix'
1143
sys.platform = 'darwin'
1148
user_encoding = locale.getpreferredencoding()
1149
except locale.Error, e:
1150
sys.stderr.write('bzr: warning: %s\n'
1151
' Could not determine what text encoding to use.\n'
1152
' This error usually means your Python interpreter\n'
1153
' doesn\'t support the locale set by $LANG (%s)\n'
1154
" Continuing with ascii encoding.\n"
1155
% (e, os.environ.get('LANG')))
1156
user_encoding = 'ascii'
1158
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1159
# treat that as ASCII, and not support printing unicode characters to the
1161
if user_encoding in (None, 'cp0'):
1162
user_encoding = 'ascii'
1166
codecs.lookup(user_encoding)
1168
sys.stderr.write('bzr: warning:'
1169
' unknown encoding %s.'
1170
' Continuing with ascii encoding.\n'
1173
user_encoding = 'ascii'
1176
_cached_user_encoding = user_encoding
1178
return user_encoding
1181
def recv_all(socket, bytes):
1182
"""Receive an exact number of bytes.
1184
Regular Socket.recv() may return less than the requested number of bytes,
1185
dependning on what's in the OS buffer. MSG_WAITALL is not available
1186
on all platforms, but this should work everywhere. This will return
1187
less than the requested amount if the remote end closes.
1189
This isn't optimized and is intended mostly for use in testing.
1192
while len(b) < bytes:
1193
new = socket.recv(bytes - len(b))
1199
def dereference_path(path):
1200
"""Determine the real path to a file.
1202
All parent elements are dereferenced. But the file itself is not
1204
:param path: The original path. May be absolute or relative.
1205
:return: the real path *to* the file
1207
parent, base = os.path.split(path)
1208
# The pathjoin for '.' is a workaround for Python bug #1213894.
1209
# (initial path components aren't dereferenced)
1210
return pathjoin(realpath(pathjoin('.', parent)), base)
added
removed
bzrlib/osutils.py
