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]
285
path = cwd + '\\' + path
286
return _win32_fixdrive(_nt_normpath(path).replace('\\', '/'))
288
if win32utils.winver == 'Windows 98':
289
_win32_abspath = _win98_abspath
292
def _win32_realpath(path):
293
# Real _nt_realpath doesn't have a problem with a unicode cwd
294
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
297
def _win32_pathjoin(*args):
298
return _nt_join(*args).replace('\\', '/')
301
def _win32_normpath(path):
302
return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))
306
return _win32_fixdrive(os.getcwdu().replace('\\', '/'))
309
def _win32_mkdtemp(*args, **kwargs):
310
return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))
313
def _win32_rename(old, new):
314
"""We expect to be able to atomically replace 'new' with old.
316
On win32, if new exists, it must be moved out of the way first,
320
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
322
if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):
323
# If we try to rename a non-existant file onto cwd, we get
324
# EPERM or EACCES instead of ENOENT, this will raise ENOENT
325
# if the old path doesn't exist, sometimes we get EACCES
326
# On Linux, we seem to get EBUSY, on Mac we get EINVAL
332
return unicodedata.normalize('NFKC', os.getcwdu())
335
# Default is to just use the python builtins, but these can be rebound on
336
# particular platforms.
337
abspath = _posix_abspath
338
realpath = _posix_realpath
339
pathjoin = os.path.join
340
normpath = os.path.normpath
343
dirname = os.path.dirname
344
basename = os.path.basename
345
split = os.path.split
346
splitext = os.path.splitext
347
# These were already imported into local scope
348
# mkdtemp = tempfile.mkdtemp
349
# rmtree = shutil.rmtree
351
MIN_ABS_PATHLENGTH = 1
354
if sys.platform == 'win32':
355
abspath = _win32_abspath
356
realpath = _win32_realpath
357
pathjoin = _win32_pathjoin
358
normpath = _win32_normpath
359
getcwd = _win32_getcwd
360
mkdtemp = _win32_mkdtemp
361
rename = _win32_rename
363
MIN_ABS_PATHLENGTH = 3
365
def _win32_delete_readonly(function, path, excinfo):
366
"""Error handler for shutil.rmtree function [for win32]
367
Helps to remove files and dirs marked as read-only.
369
exception = excinfo[1]
370
if function in (os.remove, os.rmdir) \
371
and isinstance(exception, OSError) \
372
and exception.errno == errno.EACCES:
378
def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):
379
"""Replacer for shutil.rmtree: could remove readonly dirs/files"""
380
return shutil.rmtree(path, ignore_errors, onerror)
381
elif sys.platform == 'darwin':
385
def get_terminal_encoding():
386
"""Find the best encoding for printing to the screen.
388
This attempts to check both sys.stdout and sys.stdin to see
389
what encoding they are in, and if that fails it falls back to
390
bzrlib.user_encoding.
391
The problem is that on Windows, locale.getpreferredencoding()
392
is not the same encoding as that used by the console:
393
http://mail.python.org/pipermail/python-list/2003-May/162357.html
395
On my standard US Windows XP, the preferred encoding is
396
cp1252, but the console is cp437
398
output_encoding = getattr(sys.stdout, 'encoding', None)
399
if not output_encoding:
400
input_encoding = getattr(sys.stdin, 'encoding', None)
401
if not input_encoding:
402
output_encoding = bzrlib.user_encoding
403
mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)
405
output_encoding = input_encoding
406
mutter('encoding stdout as sys.stdin encoding %r', output_encoding)
408
mutter('encoding stdout as sys.stdout encoding %r', output_encoding)
409
if output_encoding == 'cp0':
410
# invalid encoding (cp0 means 'no codepage' on Windows)
411
output_encoding = bzrlib.user_encoding
412
mutter('cp0 is invalid encoding.'
413
' encoding stdout as bzrlib.user_encoding %r', output_encoding)
416
codecs.lookup(output_encoding)
418
sys.stderr.write('bzr: warning:'
419
' unknown terminal encoding %s.\n'
420
' Using encoding %s instead.\n'
421
% (output_encoding, bzrlib.user_encoding)
423
output_encoding = bzrlib.user_encoding
425
return output_encoding
428
def normalizepath(f):
429
if getattr(os.path, 'realpath', None) is not None:
433
[p,e] = os.path.split(f)
434
if e == "" or e == "." or e == "..":
437
return pathjoin(F(p), e)
441
"""Copy a file to a backup.
443
Backups are named in GNU-style, with a ~ suffix.
445
If the file is already a backup, it's not copied.
451
if has_symlinks() and os.path.islink(fn):
452
target = os.readlink(fn)
453
os.symlink(target, bfn)
461
outf = file(bfn, 'wb')
469
"""True if f is an accessible directory."""
471
return S_ISDIR(os.lstat(f)[ST_MODE])
477
"""True if f is a regular file."""
479
return S_ISREG(os.lstat(f)[ST_MODE])
484
"""True if f is a symlink."""
486
return S_ISLNK(os.lstat(f)[ST_MODE])
490
def is_inside(dir, fname):
491
"""True if fname is inside dir.
493
The parameters should typically be passed to osutils.normpath first, so
494
that . and .. and repeated slashes are eliminated, and the separators
495
are canonical for the platform.
497
The empty string as a dir name is taken as top-of-tree and matches
500
# XXX: Most callers of this can actually do something smarter by
501
# looking at the inventory
511
return fname.startswith(dir)
514
def is_inside_any(dir_list, fname):
515
"""True if fname is inside any of given dirs."""
516
for dirname in dir_list:
517
if is_inside(dirname, fname):
523
def is_inside_or_parent_of_any(dir_list, fname):
524
"""True if fname is a child or a parent of any of the given files."""
525
for dirname in dir_list:
526
if is_inside(dirname, fname) or is_inside(fname, dirname):
532
def pumpfile(fromfile, tofile):
533
"""Copy contents of one file to another."""
536
b = fromfile.read(BUFSIZE)
542
def file_iterator(input_file, readsize=32768):
544
b = input_file.read(readsize)
551
if getattr(f, 'tell', None) is not None:
564
def sha_strings(strings):
565
"""Return the sha-1 of concatenation of strings"""
567
map(s.update, strings)
577
def fingerprint_file(f):
582
return {'size': size,
583
'sha1': s.hexdigest()}
586
def compare_files(a, b):
587
"""Returns true if equal in contents"""
598
def local_time_offset(t=None):
599
"""Return offset of local zone from GMT, either at present or at time t."""
602
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
603
return offset.days * 86400 + offset.seconds
606
def format_date(t, offset=0, timezone='original', date_fmt=None,
608
## TODO: Perhaps a global option to use either universal or local time?
609
## Or perhaps just let people set $TZ?
610
assert isinstance(t, float)
612
if timezone == 'utc':
615
elif timezone == 'original':
618
tt = time.gmtime(t + offset)
619
elif timezone == 'local':
620
tt = time.localtime(t)
621
offset = local_time_offset(t)
623
raise errors.BzrError("unsupported timezone format %r" % timezone,
624
['options are "utc", "original", "local"'])
626
date_fmt = "%a %Y-%m-%d %H:%M:%S"
628
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
631
return (time.strftime(date_fmt, tt) + offset_str)
634
def compact_date(when):
635
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
638
def format_delta(delta):
639
"""Get a nice looking string for a time delta.
641
:param delta: The time difference in seconds, can be positive or negative.
642
positive indicates time in the past, negative indicates time in the
643
future. (usually time.time() - stored_time)
644
:return: String formatted to show approximate resolution
650
direction = 'in the future'
654
if seconds < 90: # print seconds up to 90 seconds
656
return '%d second %s' % (seconds, direction,)
658
return '%d seconds %s' % (seconds, direction)
660
minutes = int(seconds / 60)
661
seconds -= 60 * minutes
666
if minutes < 90: # print minutes, seconds up to 90 minutes
668
return '%d minute, %d second%s %s' % (
669
minutes, seconds, plural_seconds, direction)
671
return '%d minutes, %d second%s %s' % (
672
minutes, seconds, plural_seconds, direction)
674
hours = int(minutes / 60)
675
minutes -= 60 * hours
682
return '%d hour, %d minute%s %s' % (hours, minutes,
683
plural_minutes, direction)
684
return '%d hours, %d minute%s %s' % (hours, minutes,
685
plural_minutes, direction)
688
"""Return size of given open file."""
689
return os.fstat(f.fileno())[ST_SIZE]
692
# Define rand_bytes based on platform.
694
# Python 2.4 and later have os.urandom,
695
# but it doesn't work on some arches
697
rand_bytes = os.urandom
698
except (NotImplementedError, AttributeError):
699
# If python doesn't have os.urandom, or it doesn't work,
700
# then try to first pull random data from /dev/urandom
702
rand_bytes = file('/dev/urandom', 'rb').read
703
# Otherwise, use this hack as a last resort
704
except (IOError, OSError):
705
# not well seeded, but better than nothing
710
s += chr(random.randint(0, 255))
715
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
717
"""Return a random string of num alphanumeric characters
719
The result only contains lowercase chars because it may be used on
720
case-insensitive filesystems.
723
for raw_byte in rand_bytes(num):
724
s += ALNUM[ord(raw_byte) % 36]
728
## TODO: We could later have path objects that remember their list
729
## decomposition (might be too tricksy though.)
732
"""Turn string into list of parts."""
733
assert isinstance(p, basestring)
735
# split on either delimiter because people might use either on
737
ps = re.split(r'[\\/]', p)
742
raise errors.BzrError("sorry, %r not allowed in path" % f)
743
elif (f == '.') or (f == ''):
750
assert isinstance(p, list)
752
if (f == '..') or (f is None) or (f == ''):
753
raise errors.BzrError("sorry, %r not allowed in path" % f)
757
@deprecated_function(zero_nine)
758
def appendpath(p1, p2):
762
return pathjoin(p1, p2)
766
"""Split s into lines, but without removing the newline characters."""
767
lines = s.split('\n')
768
result = [line + '\n' for line in lines[:-1]]
770
result.append(lines[-1])
774
def hardlinks_good():
775
return sys.platform not in ('win32', 'cygwin', 'darwin')
778
def link_or_copy(src, dest):
779
"""Hardlink a file, or copy it if it can't be hardlinked."""
780
if not hardlinks_good():
781
shutil.copyfile(src, dest)
785
except (OSError, IOError), e:
786
if e.errno != errno.EXDEV:
788
shutil.copyfile(src, dest)
790
def delete_any(full_path):
791
"""Delete a file or directory."""
795
# We may be renaming a dangling inventory id
796
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
802
if getattr(os, 'symlink', None) is not None:
808
def contains_whitespace(s):
809
"""True if there are any whitespace characters in s."""
810
# string.whitespace can include '\xa0' in certain locales, because it is
811
# considered "non-breaking-space" as part of ISO-8859-1. But it
812
# 1) Isn't a breaking whitespace
813
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
815
# 3) '\xa0' isn't unicode safe since it is >128.
816
# So we are following textwrap's example and hard-coding our own.
817
# We probably could ignore \v and \f, too.
818
for ch in u' \t\n\r\v\f':
825
def contains_linebreaks(s):
826
"""True if there is any vertical whitespace in s."""
834
def relpath(base, path):
835
"""Return path relative to base, or raise exception.
837
The path may be either an absolute path or a path relative to the
838
current working directory.
840
os.path.commonprefix (python2.4) has a bad bug that it works just
841
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
845
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
846
' exceed the platform minimum length (which is %d)' %
853
while len(head) >= len(base):
856
head, tail = os.path.split(head)
860
raise errors.PathNotChild(rp, base)
868
def safe_unicode(unicode_or_utf8_string):
869
"""Coerce unicode_or_utf8_string into unicode.
871
If it is unicode, it is returned.
872
Otherwise it is decoded from utf-8. If a decoding error
873
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
874
as a BzrBadParameter exception.
876
if isinstance(unicode_or_utf8_string, unicode):
877
return unicode_or_utf8_string
879
return unicode_or_utf8_string.decode('utf8')
880
except UnicodeDecodeError:
881
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
884
_platform_normalizes_filenames = False
885
if sys.platform == 'darwin':
886
_platform_normalizes_filenames = True
889
def normalizes_filenames():
890
"""Return True if this platform normalizes unicode filenames.
892
Mac OSX does, Windows/Linux do not.
894
return _platform_normalizes_filenames
897
def _accessible_normalized_filename(path):
898
"""Get the unicode normalized path, and if you can access the file.
900
On platforms where the system normalizes filenames (Mac OSX),
901
you can access a file by any path which will normalize correctly.
902
On platforms where the system does not normalize filenames
903
(Windows, Linux), you have to access a file by its exact path.
905
Internally, bzr only supports NFC/NFKC normalization, since that is
906
the standard for XML documents.
908
So return the normalized path, and a flag indicating if the file
909
can be accessed by that path.
912
return unicodedata.normalize('NFKC', unicode(path)), True
915
def _inaccessible_normalized_filename(path):
916
__doc__ = _accessible_normalized_filename.__doc__
918
normalized = unicodedata.normalize('NFKC', unicode(path))
919
return normalized, normalized == path
922
if _platform_normalizes_filenames:
923
normalized_filename = _accessible_normalized_filename
925
normalized_filename = _inaccessible_normalized_filename
928
def terminal_width():
929
"""Return estimated terminal width."""
930
if sys.platform == 'win32':
931
return win32utils.get_console_size()[0]
934
import struct, fcntl, termios
935
s = struct.pack('HHHH', 0, 0, 0, 0)
936
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
937
width = struct.unpack('HHHH', x)[1]
942
width = int(os.environ['COLUMNS'])
951
def supports_executable():
952
return sys.platform != "win32"
955
def supports_posix_readonly():
956
"""Return True if 'readonly' has POSIX semantics, False otherwise.
958
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
959
directory controls creation/deletion, etc.
961
And under win32, readonly means that the directory itself cannot be
962
deleted. The contents of a readonly directory can be changed, unlike POSIX
963
where files in readonly directories cannot be added, deleted or renamed.
965
return sys.platform != "win32"
968
def set_or_unset_env(env_variable, value):
969
"""Modify the environment, setting or removing the env_variable.
971
:param env_variable: The environment variable in question
972
:param value: The value to set the environment to. If None, then
973
the variable will be removed.
974
:return: The original value of the environment variable.
976
orig_val = os.environ.get(env_variable)
978
if orig_val is not None:
979
del os.environ[env_variable]
981
if isinstance(value, unicode):
982
value = value.encode(bzrlib.user_encoding)
983
os.environ[env_variable] = value
987
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
990
def check_legal_path(path):
991
"""Check whether the supplied path is legal.
992
This is only required on Windows, so we don't test on other platforms
995
if sys.platform != "win32":
997
if _validWin32PathRE.match(path) is None:
998
raise errors.IllegalPath(path)
1001
def walkdirs(top, prefix=""):
1002
"""Yield data about all the directories in a tree.
1004
This yields all the data about the contents of a directory at a time.
1005
After each directory has been yielded, if the caller has mutated the list
1006
to exclude some directories, they are then not descended into.
1008
The data yielded is of the form:
1009
((directory-relpath, directory-path-from-top),
1010
[(relpath, basename, kind, lstat), ...]),
1011
- directory-relpath is the relative path of the directory being returned
1012
with respect to top. prefix is prepended to this.
1013
- directory-path-from-root is the path including top for this directory.
1014
It is suitable for use with os functions.
1015
- relpath is the relative path within the subtree being walked.
1016
- basename is the basename of the path
1017
- kind is the kind of the file now. If unknown then the file is not
1018
present within the tree - but it may be recorded as versioned. See
1020
- lstat is the stat data *if* the file was statted.
1021
- planned, not implemented:
1022
path_from_tree_root is the path from the root of the tree.
1024
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1025
allows one to walk a subtree but get paths that are relative to a tree
1027
:return: an iterator over the dirs.
1029
#TODO there is a bit of a smell where the results of the directory-
1030
# summary in this, and the path from the root, may not agree
1031
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1032
# potentially confusing output. We should make this more robust - but
1033
# not at a speed cost. RBC 20060731
1036
_directory = _directory_kind
1037
_listdir = os.listdir
1038
pending = [(prefix, "", _directory, None, top)]
1041
currentdir = pending.pop()
1042
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1045
relroot = currentdir[0] + '/'
1048
for name in sorted(_listdir(top)):
1049
abspath = top + '/' + name
1050
statvalue = lstat(abspath)
1051
dirblock.append((relroot + name, name,
1052
file_kind_from_stat_mode(statvalue.st_mode),
1053
statvalue, abspath))
1054
yield (currentdir[0], top), dirblock
1055
# push the user specified dirs from dirblock
1056
for dir in reversed(dirblock):
1057
if dir[2] == _directory:
1061
def copy_tree(from_path, to_path, handlers={}):
1062
"""Copy all of the entries in from_path into to_path.
1064
:param from_path: The base directory to copy.
1065
:param to_path: The target directory. If it does not exist, it will
1067
:param handlers: A dictionary of functions, which takes a source and
1068
destinations for files, directories, etc.
1069
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1070
'file', 'directory', and 'symlink' should always exist.
1071
If they are missing, they will be replaced with 'os.mkdir()',
1072
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1074
# Now, just copy the existing cached tree to the new location
1075
# We use a cheap trick here.
1076
# Absolute paths are prefixed with the first parameter
1077
# relative paths are prefixed with the second.
1078
# So we can get both the source and target returned
1079
# without any extra work.
1081
def copy_dir(source, dest):
1084
def copy_link(source, dest):
1085
"""Copy the contents of a symlink"""
1086
link_to = os.readlink(source)
1087
os.symlink(link_to, dest)
1089
real_handlers = {'file':shutil.copy2,
1090
'symlink':copy_link,
1091
'directory':copy_dir,
1093
real_handlers.update(handlers)
1095
if not os.path.exists(to_path):
1096
real_handlers['directory'](from_path, to_path)
1098
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1099
for relpath, name, kind, st, abspath in entries:
1100
real_handlers[kind](abspath, relpath)
1103
def path_prefix_key(path):
1104
"""Generate a prefix-order path key for path.
1106
This can be used to sort paths in the same way that walkdirs does.
1108
return (dirname(path) , path)
1111
def compare_paths_prefix_order(path_a, path_b):
1112
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1113
key_a = path_prefix_key(path_a)
1114
key_b = path_prefix_key(path_b)
1115
return cmp(key_a, key_b)
1118
_cached_user_encoding = None
1121
def get_user_encoding(use_cache=True):
1122
"""Find out what the preferred user encoding is.
1124
This is generally the encoding that is used for command line parameters
1125
and file contents. This may be different from the terminal encoding
1126
or the filesystem encoding.
1128
:param use_cache: Enable cache for detected encoding.
1129
(This parameter is turned on by default,
1130
and required only for selftesting)
1132
:return: A string defining the preferred user encoding
1134
global _cached_user_encoding
1135
if _cached_user_encoding is not None and use_cache:
1136
return _cached_user_encoding
1138
if sys.platform == 'darwin':
1139
# work around egregious python 2.4 bug
1140
sys.platform = 'posix'
1144
sys.platform = 'darwin'
1149
user_encoding = locale.getpreferredencoding()
1150
except locale.Error, e:
1151
sys.stderr.write('bzr: warning: %s\n'
1152
' Could not determine what text encoding to use.\n'
1153
' This error usually means your Python interpreter\n'
1154
' doesn\'t support the locale set by $LANG (%s)\n'
1155
" Continuing with ascii encoding.\n"
1156
% (e, os.environ.get('LANG')))
1157
user_encoding = 'ascii'
1159
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1160
# treat that as ASCII, and not support printing unicode characters to the
1162
if user_encoding in (None, 'cp0'):
1163
user_encoding = 'ascii'
1167
codecs.lookup(user_encoding)
1169
sys.stderr.write('bzr: warning:'
1170
' unknown encoding %s.'
1171
' Continuing with ascii encoding.\n'
1174
user_encoding = 'ascii'
1177
_cached_user_encoding = user_encoding
1179
return user_encoding
1182
def recv_all(socket, bytes):
1183
"""Receive an exact number of bytes.
1185
Regular Socket.recv() may return less than the requested number of bytes,
1186
dependning on what's in the OS buffer. MSG_WAITALL is not available
1187
on all platforms, but this should work everywhere. This will return
1188
less than the requested amount if the remote end closes.
1190
This isn't optimized and is intended mostly for use in testing.
1193
while len(b) < bytes:
1194
new = socket.recv(bytes - len(b))
1200
def dereference_path(path):
1201
"""Determine the real path to a file.
1203
All parent elements are dereferenced. But the file itself is not
1205
:param path: The original path. May be absolute or relative.
1206
:return: the real path *to* the file
1208
parent, base = os.path.split(path)
1209
# The pathjoin for '.' is a workaround for Python bug #1213894.
1210
# (initial path components aren't dereferenced)
1211
return pathjoin(realpath(pathjoin('.', parent)), base)
added
removed
bzrlib/osutils.py
