1
# Copyright (C) 2005, 2006, 2007 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 (
57
from bzrlib import symbol_versioning
58
from bzrlib.symbol_versioning import (
61
from bzrlib.trace import mutter
64
# On win32, O_BINARY is used to indicate the file should
65
# be opened in binary mode, rather than text mode.
66
# On other platforms, O_BINARY doesn't exist, because
67
# they always open in binary mode, so it is okay to
68
# OR with 0 on those platforms
69
O_BINARY = getattr(os, 'O_BINARY', 0)
71
# On posix, use lstat instead of stat so that we can
72
# operate on broken symlinks. On Windows revert to stat.
73
lstat = getattr(os, 'lstat', os.stat)
75
def make_readonly(filename):
76
"""Make a filename read-only."""
77
mod = lstat(filename).st_mode
78
if not stat.S_ISLNK(mod):
80
os.chmod(filename, mod)
83
def make_writable(filename):
84
mod = lstat(filename).st_mode
85
if not stat.S_ISLNK(mod):
87
os.chmod(filename, mod)
94
"""Return a quoted filename filename
96
This previously used backslash quoting, but that works poorly on
98
# TODO: I'm not really sure this is the best format either.x
100
if _QUOTE_RE is None:
101
_QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')
103
if _QUOTE_RE.search(f):
109
_directory_kind = 'directory'
112
stat.S_IFDIR:_directory_kind,
113
stat.S_IFCHR:'chardev',
114
stat.S_IFBLK:'block',
117
stat.S_IFLNK:'symlink',
118
stat.S_IFSOCK:'socket',
122
def file_kind_from_stat_mode(stat_mode, _formats=_formats, _unknown='unknown'):
123
"""Generate a file kind from a stat mode. This is used in walkdirs.
125
Its performance is critical: Do not mutate without careful benchmarking.
128
return _formats[stat_mode & 0170000]
133
def file_kind(f, _lstat=os.lstat, _mapper=file_kind_from_stat_mode):
135
return _mapper(_lstat(f).st_mode)
137
if getattr(e, 'errno', None) == errno.ENOENT:
138
raise errors.NoSuchFile(f)
143
"""Return the current umask"""
144
# Assume that people aren't messing with the umask while running
145
# XXX: This is not thread safe, but there is no way to get the
146
# umask without setting it
154
_directory_kind: "/",
156
'tree-reference': '+',
160
def kind_marker(kind):
162
return _kind_marker_map[kind]
164
raise errors.BzrError('invalid file kind %r' % kind)
167
lexists = getattr(os.path, 'lexists', None)
171
stat = getattr(os, 'lstat', os.stat)
175
if e.errno == errno.ENOENT:
178
raise errors.BzrError("lstat/stat of (%r): %r" % (f, e))
181
def fancy_rename(old, new, rename_func, unlink_func):
182
"""A fancy rename, when you don't have atomic rename.
184
:param old: The old path, to rename from
185
:param new: The new path, to rename to
186
:param rename_func: The potentially non-atomic rename function
187
:param unlink_func: A way to delete the target file if the full rename succeeds
190
# sftp rename doesn't allow overwriting, so play tricks:
192
base = os.path.basename(new)
193
dirname = os.path.dirname(new)
194
tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))
195
tmp_name = pathjoin(dirname, tmp_name)
197
# Rename the file out of the way, but keep track if it didn't exist
198
# We don't want to grab just any exception
199
# something like EACCES should prevent us from continuing
200
# The downside is that the rename_func has to throw an exception
201
# with an errno = ENOENT, or NoSuchFile
204
rename_func(new, tmp_name)
205
except (errors.NoSuchFile,), e:
208
# RBC 20060103 abstraction leakage: the paramiko SFTP clients rename
209
# function raises an IOError with errno is None when a rename fails.
210
# This then gets caught here.
211
if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):
214
if (getattr(e, 'errno', None) is None
215
or e.errno not in (errno.ENOENT, errno.ENOTDIR)):
222
# This may throw an exception, in which case success will
224
rename_func(old, new)
228
# If the file used to exist, rename it back into place
229
# otherwise just delete it from the tmp location
231
unlink_func(tmp_name)
233
rename_func(tmp_name, new)
236
# In Python 2.4.2 and older, os.path.abspath and os.path.realpath
237
# choke on a Unicode string containing a relative path if
238
# os.getcwd() returns a non-sys.getdefaultencoding()-encoded
240
_fs_enc = sys.getfilesystemencoding() or 'utf-8'
241
def _posix_abspath(path):
242
# jam 20060426 rather than encoding to fsencoding
243
# copy posixpath.abspath, but use os.getcwdu instead
244
if not posixpath.isabs(path):
245
path = posixpath.join(getcwd(), path)
246
return posixpath.normpath(path)
249
def _posix_realpath(path):
250
return posixpath.realpath(path.encode(_fs_enc)).decode(_fs_enc)
253
def _win32_fixdrive(path):
254
"""Force drive letters to be consistent.
256
win32 is inconsistent whether it returns lower or upper case
257
and even if it was consistent the user might type the other
258
so we force it to uppercase
259
running python.exe under cmd.exe return capital C:\\
260
running win32 python inside a cygwin shell returns lowercase c:\\
262
drive, path = _nt_splitdrive(path)
263
return drive.upper() + path
266
def _win32_abspath(path):
267
# Real _nt_abspath doesn't have a problem with a unicode cwd
268
return _win32_fixdrive(_nt_abspath(unicode(path)).replace('\\', '/'))
271
def _win98_abspath(path):
272
"""Return the absolute version of a path.
273
Windows 98 safe implementation (python reimplementation
274
of Win32 API function GetFullPathNameW)
279
# \\HOST\path => //HOST/path
280
# //HOST/path => //HOST/path
281
# path => C:/cwd/path
284
# check for absolute path
285
drive = _nt_splitdrive(path)[0]
286
if drive == '' and path[:2] not in('//','\\\\'):
288
# we cannot simply os.path.join cwd and path
289
# because os.path.join('C:','/path') produce '/path'
290
# and this is incorrect
291
if path[:1] in ('/','\\'):
292
cwd = _nt_splitdrive(cwd)[0]
294
path = cwd + '\\' + path
295
return _win32_fixdrive(_nt_normpath(path).replace('\\', '/'))
297
if win32utils.winver == 'Windows 98':
298
_win32_abspath = _win98_abspath
301
def _win32_realpath(path):
302
# Real _nt_realpath doesn't have a problem with a unicode cwd
303
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
306
def _win32_pathjoin(*args):
307
return _nt_join(*args).replace('\\', '/')
310
def _win32_normpath(path):
311
return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))
315
return _win32_fixdrive(os.getcwdu().replace('\\', '/'))
318
def _win32_mkdtemp(*args, **kwargs):
319
return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))
322
def _win32_rename(old, new):
323
"""We expect to be able to atomically replace 'new' with old.
325
On win32, if new exists, it must be moved out of the way first,
329
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
331
if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):
332
# If we try to rename a non-existant file onto cwd, we get
333
# EPERM or EACCES instead of ENOENT, this will raise ENOENT
334
# if the old path doesn't exist, sometimes we get EACCES
335
# On Linux, we seem to get EBUSY, on Mac we get EINVAL
341
return unicodedata.normalize('NFKC', os.getcwdu())
344
# Default is to just use the python builtins, but these can be rebound on
345
# particular platforms.
346
abspath = _posix_abspath
347
realpath = _posix_realpath
348
pathjoin = os.path.join
349
normpath = os.path.normpath
352
dirname = os.path.dirname
353
basename = os.path.basename
354
split = os.path.split
355
splitext = os.path.splitext
356
# These were already imported into local scope
357
# mkdtemp = tempfile.mkdtemp
358
# rmtree = shutil.rmtree
360
MIN_ABS_PATHLENGTH = 1
363
if sys.platform == 'win32':
364
abspath = _win32_abspath
365
realpath = _win32_realpath
366
pathjoin = _win32_pathjoin
367
normpath = _win32_normpath
368
getcwd = _win32_getcwd
369
mkdtemp = _win32_mkdtemp
370
rename = _win32_rename
372
MIN_ABS_PATHLENGTH = 3
374
def _win32_delete_readonly(function, path, excinfo):
375
"""Error handler for shutil.rmtree function [for win32]
376
Helps to remove files and dirs marked as read-only.
378
exception = excinfo[1]
379
if function in (os.remove, os.rmdir) \
380
and isinstance(exception, OSError) \
381
and exception.errno == errno.EACCES:
387
def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):
388
"""Replacer for shutil.rmtree: could remove readonly dirs/files"""
389
return shutil.rmtree(path, ignore_errors, onerror)
390
elif sys.platform == 'darwin':
394
def get_terminal_encoding():
395
"""Find the best encoding for printing to the screen.
397
This attempts to check both sys.stdout and sys.stdin to see
398
what encoding they are in, and if that fails it falls back to
399
bzrlib.user_encoding.
400
The problem is that on Windows, locale.getpreferredencoding()
401
is not the same encoding as that used by the console:
402
http://mail.python.org/pipermail/python-list/2003-May/162357.html
404
On my standard US Windows XP, the preferred encoding is
405
cp1252, but the console is cp437
407
output_encoding = getattr(sys.stdout, 'encoding', None)
408
if not output_encoding:
409
input_encoding = getattr(sys.stdin, 'encoding', None)
410
if not input_encoding:
411
output_encoding = bzrlib.user_encoding
412
mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)
414
output_encoding = input_encoding
415
mutter('encoding stdout as sys.stdin encoding %r', output_encoding)
417
mutter('encoding stdout as sys.stdout encoding %r', output_encoding)
418
if output_encoding == 'cp0':
419
# invalid encoding (cp0 means 'no codepage' on Windows)
420
output_encoding = bzrlib.user_encoding
421
mutter('cp0 is invalid encoding.'
422
' encoding stdout as bzrlib.user_encoding %r', output_encoding)
425
codecs.lookup(output_encoding)
427
sys.stderr.write('bzr: warning:'
428
' unknown terminal encoding %s.\n'
429
' Using encoding %s instead.\n'
430
% (output_encoding, bzrlib.user_encoding)
432
output_encoding = bzrlib.user_encoding
434
return output_encoding
437
def normalizepath(f):
438
if getattr(os.path, 'realpath', None) is not None:
442
[p,e] = os.path.split(f)
443
if e == "" or e == "." or e == "..":
446
return pathjoin(F(p), e)
450
"""Copy a file to a backup.
452
Backups are named in GNU-style, with a ~ suffix.
454
If the file is already a backup, it's not copied.
460
if has_symlinks() and os.path.islink(fn):
461
target = os.readlink(fn)
462
os.symlink(target, bfn)
470
outf = file(bfn, 'wb')
478
"""True if f is an accessible directory."""
480
return S_ISDIR(os.lstat(f)[ST_MODE])
486
"""True if f is a regular file."""
488
return S_ISREG(os.lstat(f)[ST_MODE])
493
"""True if f is a symlink."""
495
return S_ISLNK(os.lstat(f)[ST_MODE])
499
def is_inside(dir, fname):
500
"""True if fname is inside dir.
502
The parameters should typically be passed to osutils.normpath first, so
503
that . and .. and repeated slashes are eliminated, and the separators
504
are canonical for the platform.
506
The empty string as a dir name is taken as top-of-tree and matches
509
# XXX: Most callers of this can actually do something smarter by
510
# looking at the inventory
520
return fname.startswith(dir)
523
def is_inside_any(dir_list, fname):
524
"""True if fname is inside any of given dirs."""
525
for dirname in dir_list:
526
if is_inside(dirname, fname):
531
def is_inside_or_parent_of_any(dir_list, fname):
532
"""True if fname is a child or a parent of any of the given files."""
533
for dirname in dir_list:
534
if is_inside(dirname, fname) or is_inside(fname, dirname):
539
def pumpfile(fromfile, tofile):
540
"""Copy contents of one file to another.
542
:return: The number of bytes copied.
547
b = fromfile.read(BUFSIZE)
555
def file_iterator(input_file, readsize=32768):
557
b = input_file.read(readsize)
564
if getattr(f, 'tell', None) is not None:
577
def sha_strings(strings):
578
"""Return the sha-1 of concatenation of strings"""
580
map(s.update, strings)
590
def fingerprint_file(f):
595
return {'size': size,
596
'sha1': s.hexdigest()}
599
def compare_files(a, b):
600
"""Returns true if equal in contents"""
611
def local_time_offset(t=None):
612
"""Return offset of local zone from GMT, either at present or at time t."""
615
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
616
return offset.days * 86400 + offset.seconds
619
def format_date(t, offset=0, timezone='original', date_fmt=None,
621
"""Return a formatted date string.
623
:param t: Seconds since the epoch.
624
:param offset: Timezone offset in seconds east of utc.
625
:param timezone: How to display the time: 'utc', 'original' for the
626
timezone specified by offset, or 'local' for the process's current
628
:param show_offset: Whether to append the timezone.
629
:param date_fmt: strftime format.
631
if timezone == 'utc':
634
elif timezone == 'original':
637
tt = time.gmtime(t + offset)
638
elif timezone == 'local':
639
tt = time.localtime(t)
640
offset = local_time_offset(t)
642
raise errors.BzrError("unsupported timezone format %r" % timezone,
643
['options are "utc", "original", "local"'])
645
date_fmt = "%a %Y-%m-%d %H:%M:%S"
647
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
650
return (time.strftime(date_fmt, tt) + offset_str)
653
def compact_date(when):
654
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
657
def format_delta(delta):
658
"""Get a nice looking string for a time delta.
660
:param delta: The time difference in seconds, can be positive or negative.
661
positive indicates time in the past, negative indicates time in the
662
future. (usually time.time() - stored_time)
663
:return: String formatted to show approximate resolution
669
direction = 'in the future'
673
if seconds < 90: # print seconds up to 90 seconds
675
return '%d second %s' % (seconds, direction,)
677
return '%d seconds %s' % (seconds, direction)
679
minutes = int(seconds / 60)
680
seconds -= 60 * minutes
685
if minutes < 90: # print minutes, seconds up to 90 minutes
687
return '%d minute, %d second%s %s' % (
688
minutes, seconds, plural_seconds, direction)
690
return '%d minutes, %d second%s %s' % (
691
minutes, seconds, plural_seconds, direction)
693
hours = int(minutes / 60)
694
minutes -= 60 * hours
701
return '%d hour, %d minute%s %s' % (hours, minutes,
702
plural_minutes, direction)
703
return '%d hours, %d minute%s %s' % (hours, minutes,
704
plural_minutes, direction)
707
"""Return size of given open file."""
708
return os.fstat(f.fileno())[ST_SIZE]
711
# Define rand_bytes based on platform.
713
# Python 2.4 and later have os.urandom,
714
# but it doesn't work on some arches
716
rand_bytes = os.urandom
717
except (NotImplementedError, AttributeError):
718
# If python doesn't have os.urandom, or it doesn't work,
719
# then try to first pull random data from /dev/urandom
721
rand_bytes = file('/dev/urandom', 'rb').read
722
# Otherwise, use this hack as a last resort
723
except (IOError, OSError):
724
# not well seeded, but better than nothing
729
s += chr(random.randint(0, 255))
734
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
736
"""Return a random string of num alphanumeric characters
738
The result only contains lowercase chars because it may be used on
739
case-insensitive filesystems.
742
for raw_byte in rand_bytes(num):
743
s += ALNUM[ord(raw_byte) % 36]
747
## TODO: We could later have path objects that remember their list
748
## decomposition (might be too tricksy though.)
751
"""Turn string into list of parts."""
752
assert isinstance(p, basestring)
754
# split on either delimiter because people might use either on
756
ps = re.split(r'[\\/]', p)
761
raise errors.BzrError("sorry, %r not allowed in path" % f)
762
elif (f == '.') or (f == ''):
769
assert isinstance(p, (list, tuple))
771
if (f == '..') or (f is None) or (f == ''):
772
raise errors.BzrError("sorry, %r not allowed in path" % f)
777
"""Split s into lines, but without removing the newline characters."""
778
lines = s.split('\n')
779
result = [line + '\n' for line in lines[:-1]]
781
result.append(lines[-1])
785
def hardlinks_good():
786
return sys.platform not in ('win32', 'cygwin', 'darwin')
789
def link_or_copy(src, dest):
790
"""Hardlink a file, or copy it if it can't be hardlinked."""
791
if not hardlinks_good():
792
shutil.copyfile(src, dest)
796
except (OSError, IOError), e:
797
if e.errno != errno.EXDEV:
799
shutil.copyfile(src, dest)
801
def delete_any(full_path):
802
"""Delete a file or directory."""
806
# We may be renaming a dangling inventory id
807
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
813
if getattr(os, 'symlink', None) is not None:
819
def contains_whitespace(s):
820
"""True if there are any whitespace characters in s."""
821
# string.whitespace can include '\xa0' in certain locales, because it is
822
# considered "non-breaking-space" as part of ISO-8859-1. But it
823
# 1) Isn't a breaking whitespace
824
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
826
# 3) '\xa0' isn't unicode safe since it is >128.
828
# This should *not* be a unicode set of characters in case the source
829
# string is not a Unicode string. We can auto-up-cast the characters since
830
# they are ascii, but we don't want to auto-up-cast the string in case it
832
for ch in ' \t\n\r\v\f':
839
def contains_linebreaks(s):
840
"""True if there is any vertical whitespace in s."""
848
def relpath(base, path):
849
"""Return path relative to base, or raise exception.
851
The path may be either an absolute path or a path relative to the
852
current working directory.
854
os.path.commonprefix (python2.4) has a bad bug that it works just
855
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
859
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
860
' exceed the platform minimum length (which is %d)' %
867
while len(head) >= len(base):
870
head, tail = os.path.split(head)
874
raise errors.PathNotChild(rp, base)
882
def safe_unicode(unicode_or_utf8_string):
883
"""Coerce unicode_or_utf8_string into unicode.
885
If it is unicode, it is returned.
886
Otherwise it is decoded from utf-8. If a decoding error
887
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
888
as a BzrBadParameter exception.
890
if isinstance(unicode_or_utf8_string, unicode):
891
return unicode_or_utf8_string
893
return unicode_or_utf8_string.decode('utf8')
894
except UnicodeDecodeError:
895
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
898
def safe_utf8(unicode_or_utf8_string):
899
"""Coerce unicode_or_utf8_string to a utf8 string.
901
If it is a str, it is returned.
902
If it is Unicode, it is encoded into a utf-8 string.
904
if isinstance(unicode_or_utf8_string, str):
905
# TODO: jam 20070209 This is overkill, and probably has an impact on
906
# performance if we are dealing with lots of apis that want a
909
# Make sure it is a valid utf-8 string
910
unicode_or_utf8_string.decode('utf-8')
911
except UnicodeDecodeError:
912
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
913
return unicode_or_utf8_string
914
return unicode_or_utf8_string.encode('utf-8')
917
_revision_id_warning = ('Unicode revision ids were deprecated in bzr 0.15.'
918
' Revision id generators should be creating utf8'
922
def safe_revision_id(unicode_or_utf8_string, warn=True):
923
"""Revision ids should now be utf8, but at one point they were unicode.
925
:param unicode_or_utf8_string: A possibly Unicode revision_id. (can also be
927
:param warn: Functions that are sanitizing user data can set warn=False
928
:return: None or a utf8 revision id.
930
if (unicode_or_utf8_string is None
931
or unicode_or_utf8_string.__class__ == str):
932
return unicode_or_utf8_string
934
symbol_versioning.warn(_revision_id_warning, DeprecationWarning,
936
return cache_utf8.encode(unicode_or_utf8_string)
939
_file_id_warning = ('Unicode file ids were deprecated in bzr 0.15. File id'
940
' generators should be creating utf8 file ids.')
943
def safe_file_id(unicode_or_utf8_string, warn=True):
944
"""File ids should now be utf8, but at one point they were unicode.
946
This is the same as safe_utf8, except it uses the cached encode functions
947
to save a little bit of performance.
949
:param unicode_or_utf8_string: A possibly Unicode file_id. (can also be
951
:param warn: Functions that are sanitizing user data can set warn=False
952
:return: None or a utf8 file id.
954
if (unicode_or_utf8_string is None
955
or unicode_or_utf8_string.__class__ == str):
956
return unicode_or_utf8_string
958
symbol_versioning.warn(_file_id_warning, DeprecationWarning,
960
return cache_utf8.encode(unicode_or_utf8_string)
963
_platform_normalizes_filenames = False
964
if sys.platform == 'darwin':
965
_platform_normalizes_filenames = True
968
def normalizes_filenames():
969
"""Return True if this platform normalizes unicode filenames.
971
Mac OSX does, Windows/Linux do not.
973
return _platform_normalizes_filenames
976
def _accessible_normalized_filename(path):
977
"""Get the unicode normalized path, and if you can access the file.
979
On platforms where the system normalizes filenames (Mac OSX),
980
you can access a file by any path which will normalize correctly.
981
On platforms where the system does not normalize filenames
982
(Windows, Linux), you have to access a file by its exact path.
984
Internally, bzr only supports NFC/NFKC normalization, since that is
985
the standard for XML documents.
987
So return the normalized path, and a flag indicating if the file
988
can be accessed by that path.
991
return unicodedata.normalize('NFKC', unicode(path)), True
994
def _inaccessible_normalized_filename(path):
995
__doc__ = _accessible_normalized_filename.__doc__
997
normalized = unicodedata.normalize('NFKC', unicode(path))
998
return normalized, normalized == path
1001
if _platform_normalizes_filenames:
1002
normalized_filename = _accessible_normalized_filename
1004
normalized_filename = _inaccessible_normalized_filename
1007
def terminal_width():
1008
"""Return estimated terminal width."""
1009
if sys.platform == 'win32':
1010
return win32utils.get_console_size()[0]
1013
import struct, fcntl, termios
1014
s = struct.pack('HHHH', 0, 0, 0, 0)
1015
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
1016
width = struct.unpack('HHHH', x)[1]
1021
width = int(os.environ['COLUMNS'])
1030
def supports_executable():
1031
return sys.platform != "win32"
1034
def supports_posix_readonly():
1035
"""Return True if 'readonly' has POSIX semantics, False otherwise.
1037
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
1038
directory controls creation/deletion, etc.
1040
And under win32, readonly means that the directory itself cannot be
1041
deleted. The contents of a readonly directory can be changed, unlike POSIX
1042
where files in readonly directories cannot be added, deleted or renamed.
1044
return sys.platform != "win32"
1047
def set_or_unset_env(env_variable, value):
1048
"""Modify the environment, setting or removing the env_variable.
1050
:param env_variable: The environment variable in question
1051
:param value: The value to set the environment to. If None, then
1052
the variable will be removed.
1053
:return: The original value of the environment variable.
1055
orig_val = os.environ.get(env_variable)
1057
if orig_val is not None:
1058
del os.environ[env_variable]
1060
if isinstance(value, unicode):
1061
value = value.encode(bzrlib.user_encoding)
1062
os.environ[env_variable] = value
1066
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
1069
def check_legal_path(path):
1070
"""Check whether the supplied path is legal.
1071
This is only required on Windows, so we don't test on other platforms
1074
if sys.platform != "win32":
1076
if _validWin32PathRE.match(path) is None:
1077
raise errors.IllegalPath(path)
1080
def walkdirs(top, prefix=""):
1081
"""Yield data about all the directories in a tree.
1083
This yields all the data about the contents of a directory at a time.
1084
After each directory has been yielded, if the caller has mutated the list
1085
to exclude some directories, they are then not descended into.
1087
The data yielded is of the form:
1088
((directory-relpath, directory-path-from-top),
1089
[(relpath, basename, kind, lstat, path-from-top), ...]),
1090
- directory-relpath is the relative path of the directory being returned
1091
with respect to top. prefix is prepended to this.
1092
- directory-path-from-root is the path including top for this directory.
1093
It is suitable for use with os functions.
1094
- relpath is the relative path within the subtree being walked.
1095
- basename is the basename of the path
1096
- kind is the kind of the file now. If unknown then the file is not
1097
present within the tree - but it may be recorded as versioned. See
1099
- lstat is the stat data *if* the file was statted.
1100
- planned, not implemented:
1101
path_from_tree_root is the path from the root of the tree.
1103
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1104
allows one to walk a subtree but get paths that are relative to a tree
1106
:return: an iterator over the dirs.
1108
#TODO there is a bit of a smell where the results of the directory-
1109
# summary in this, and the path from the root, may not agree
1110
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1111
# potentially confusing output. We should make this more robust - but
1112
# not at a speed cost. RBC 20060731
1114
_directory = _directory_kind
1115
_listdir = os.listdir
1116
_kind_from_mode = _formats.get
1117
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
1119
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1120
relroot, _, _, _, top = pending.pop()
1122
relprefix = relroot + u'/'
1125
top_slash = top + u'/'
1128
append = dirblock.append
1129
for name in sorted(_listdir(top)):
1130
abspath = top_slash + name
1131
statvalue = _lstat(abspath)
1132
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1133
append((relprefix + name, name, kind, statvalue, abspath))
1134
yield (relroot, top), dirblock
1136
# push the user specified dirs from dirblock
1137
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1140
def _walkdirs_utf8(top, prefix=""):
1141
"""Yield data about all the directories in a tree.
1143
This yields the same information as walkdirs() only each entry is yielded
1144
in utf-8. On platforms which have a filesystem encoding of utf8 the paths
1145
are returned as exact byte-strings.
1147
:return: yields a tuple of (dir_info, [file_info])
1148
dir_info is (utf8_relpath, path-from-top)
1149
file_info is (utf8_relpath, utf8_name, kind, lstat, path-from-top)
1150
if top is an absolute path, path-from-top is also an absolute path.
1151
path-from-top might be unicode or utf8, but it is the correct path to
1152
pass to os functions to affect the file in question. (such as os.lstat)
1154
fs_encoding = _fs_enc.upper()
1155
if (sys.platform == 'win32' or
1156
fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968')): # ascii
1157
return _walkdirs_unicode_to_utf8(top, prefix=prefix)
1159
return _walkdirs_fs_utf8(top, prefix=prefix)
1162
def _walkdirs_fs_utf8(top, prefix=""):
1163
"""See _walkdirs_utf8.
1165
This sub-function is called when we know the filesystem is already in utf8
1166
encoding. So we don't need to transcode filenames.
1169
_directory = _directory_kind
1170
_listdir = os.listdir
1171
_kind_from_mode = _formats.get
1173
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1174
# But we don't actually uses 1-3 in pending, so set them to None
1175
pending = [(safe_utf8(prefix), None, None, None, safe_utf8(top))]
1177
relroot, _, _, _, top = pending.pop()
1179
relprefix = relroot + '/'
1182
top_slash = top + '/'
1185
append = dirblock.append
1186
for name in sorted(_listdir(top)):
1187
abspath = top_slash + name
1188
statvalue = _lstat(abspath)
1189
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1190
append((relprefix + name, name, kind, statvalue, abspath))
1191
yield (relroot, top), dirblock
1193
# push the user specified dirs from dirblock
1194
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1197
def _walkdirs_unicode_to_utf8(top, prefix=""):
1198
"""See _walkdirs_utf8
1200
Because Win32 has a Unicode api, all of the 'path-from-top' entries will be
1202
This is currently the fallback code path when the filesystem encoding is
1203
not UTF-8. It may be better to implement an alternative so that we can
1204
safely handle paths that are not properly decodable in the current
1207
_utf8_encode = codecs.getencoder('utf8')
1209
_directory = _directory_kind
1210
_listdir = os.listdir
1211
_kind_from_mode = _formats.get
1213
pending = [(safe_utf8(prefix), None, None, None, safe_unicode(top))]
1215
relroot, _, _, _, top = pending.pop()
1217
relprefix = relroot + '/'
1220
top_slash = top + u'/'
1223
append = dirblock.append
1224
for name in sorted(_listdir(top)):
1225
name_utf8 = _utf8_encode(name)[0]
1226
abspath = top_slash + name
1227
statvalue = _lstat(abspath)
1228
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1229
append((relprefix + name_utf8, name_utf8, kind, statvalue, abspath))
1230
yield (relroot, top), dirblock
1232
# push the user specified dirs from dirblock
1233
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1236
def copy_tree(from_path, to_path, handlers={}):
1237
"""Copy all of the entries in from_path into to_path.
1239
:param from_path: The base directory to copy.
1240
:param to_path: The target directory. If it does not exist, it will
1242
:param handlers: A dictionary of functions, which takes a source and
1243
destinations for files, directories, etc.
1244
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1245
'file', 'directory', and 'symlink' should always exist.
1246
If they are missing, they will be replaced with 'os.mkdir()',
1247
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1249
# Now, just copy the existing cached tree to the new location
1250
# We use a cheap trick here.
1251
# Absolute paths are prefixed with the first parameter
1252
# relative paths are prefixed with the second.
1253
# So we can get both the source and target returned
1254
# without any extra work.
1256
def copy_dir(source, dest):
1259
def copy_link(source, dest):
1260
"""Copy the contents of a symlink"""
1261
link_to = os.readlink(source)
1262
os.symlink(link_to, dest)
1264
real_handlers = {'file':shutil.copy2,
1265
'symlink':copy_link,
1266
'directory':copy_dir,
1268
real_handlers.update(handlers)
1270
if not os.path.exists(to_path):
1271
real_handlers['directory'](from_path, to_path)
1273
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1274
for relpath, name, kind, st, abspath in entries:
1275
real_handlers[kind](abspath, relpath)
1278
def path_prefix_key(path):
1279
"""Generate a prefix-order path key for path.
1281
This can be used to sort paths in the same way that walkdirs does.
1283
return (dirname(path) , path)
1286
def compare_paths_prefix_order(path_a, path_b):
1287
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1288
key_a = path_prefix_key(path_a)
1289
key_b = path_prefix_key(path_b)
1290
return cmp(key_a, key_b)
1293
_cached_user_encoding = None
1296
def get_user_encoding(use_cache=True):
1297
"""Find out what the preferred user encoding is.
1299
This is generally the encoding that is used for command line parameters
1300
and file contents. This may be different from the terminal encoding
1301
or the filesystem encoding.
1303
:param use_cache: Enable cache for detected encoding.
1304
(This parameter is turned on by default,
1305
and required only for selftesting)
1307
:return: A string defining the preferred user encoding
1309
global _cached_user_encoding
1310
if _cached_user_encoding is not None and use_cache:
1311
return _cached_user_encoding
1313
if sys.platform == 'darwin':
1314
# work around egregious python 2.4 bug
1315
sys.platform = 'posix'
1319
sys.platform = 'darwin'
1324
user_encoding = locale.getpreferredencoding()
1325
except locale.Error, e:
1326
sys.stderr.write('bzr: warning: %s\n'
1327
' Could not determine what text encoding to use.\n'
1328
' This error usually means your Python interpreter\n'
1329
' doesn\'t support the locale set by $LANG (%s)\n'
1330
" Continuing with ascii encoding.\n"
1331
% (e, os.environ.get('LANG')))
1332
user_encoding = 'ascii'
1334
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1335
# treat that as ASCII, and not support printing unicode characters to the
1337
if user_encoding in (None, 'cp0'):
1338
user_encoding = 'ascii'
1342
codecs.lookup(user_encoding)
1344
sys.stderr.write('bzr: warning:'
1345
' unknown encoding %s.'
1346
' Continuing with ascii encoding.\n'
1349
user_encoding = 'ascii'
1352
_cached_user_encoding = user_encoding
1354
return user_encoding
1357
def recv_all(socket, bytes):
1358
"""Receive an exact number of bytes.
1360
Regular Socket.recv() may return less than the requested number of bytes,
1361
dependning on what's in the OS buffer. MSG_WAITALL is not available
1362
on all platforms, but this should work everywhere. This will return
1363
less than the requested amount if the remote end closes.
1365
This isn't optimized and is intended mostly for use in testing.
1368
while len(b) < bytes:
1369
new = socket.recv(bytes - len(b))
1375
def dereference_path(path):
1376
"""Determine the real path to a file.
1378
All parent elements are dereferenced. But the file itself is not
1380
:param path: The original path. May be absolute or relative.
1381
:return: the real path *to* the file
1383
parent, base = os.path.split(path)
1384
# The pathjoin for '.' is a workaround for Python bug #1213894.
1385
# (initial path components aren't dereferenced)
1386
return pathjoin(realpath(pathjoin('.', parent)), base)
1389
def supports_mapi():
1390
"""Return True if we can use MAPI to launch a mail client."""
1391
return sys.platform == "win32"
added
removed
bzrlib/osutils.py
