68
60
# OR with 0 on those platforms
69
61
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
64
def make_readonly(filename):
76
65
"""Make a filename read-only."""
77
mod = lstat(filename).st_mode
78
if not stat.S_ISLNK(mod):
80
os.chmod(filename, mod)
66
mod = os.stat(filename).st_mode
68
os.chmod(filename, mod)
83
71
def make_writable(filename):
84
mod = lstat(filename).st_mode
85
if not stat.S_ISLNK(mod):
87
os.chmod(filename, mod)
90
def minimum_path_selection(paths):
91
"""Return the smallset subset of paths which are outside paths.
93
:param paths: A container (and hence not None) of paths.
94
:return: A set of paths sufficient to include everything in paths via
95
is_inside_any, drawn from the paths parameter.
100
other_paths = paths.difference([path])
101
if not is_inside_any(other_paths, path):
102
# this is a top level path, we must check it.
103
search_paths.add(path)
72
mod = os.stat(filename).st_mode
74
os.chmod(filename, mod)
285
252
return _win32_fixdrive(_nt_abspath(unicode(path)).replace('\\', '/'))
288
def _win98_abspath(path):
289
"""Return the absolute version of a path.
290
Windows 98 safe implementation (python reimplementation
291
of Win32 API function GetFullPathNameW)
296
# \\HOST\path => //HOST/path
297
# //HOST/path => //HOST/path
298
# path => C:/cwd/path
301
# check for absolute path
302
drive = _nt_splitdrive(path)[0]
303
if drive == '' and path[:2] not in('//','\\\\'):
305
# we cannot simply os.path.join cwd and path
306
# because os.path.join('C:','/path') produce '/path'
307
# and this is incorrect
308
if path[:1] in ('/','\\'):
309
cwd = _nt_splitdrive(cwd)[0]
311
path = cwd + '\\' + path
312
return _win32_fixdrive(_nt_normpath(path).replace('\\', '/'))
314
if win32utils.winver == 'Windows 98':
315
_win32_abspath = _win98_abspath
318
255
def _win32_realpath(path):
319
256
# Real _nt_realpath doesn't have a problem with a unicode cwd
320
257
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
636
555
def local_time_offset(t=None):
637
556
"""Return offset of local zone from GMT, either at present or at time t."""
557
# python2.3 localtime() can't take None
640
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
641
return offset.days * 86400 + offset.seconds
561
if time.localtime(t).tm_isdst and time.daylight:
564
return -time.timezone
644
def format_date(t, offset=0, timezone='original', date_fmt=None,
567
def format_date(t, offset=0, timezone='original', date_fmt=None,
645
568
show_offset=True):
646
"""Return a formatted date string.
648
:param t: Seconds since the epoch.
649
:param offset: Timezone offset in seconds east of utc.
650
:param timezone: How to display the time: 'utc', 'original' for the
651
timezone specified by offset, or 'local' for the process's current
653
:param show_offset: Whether to append the timezone.
654
:param date_fmt: strftime format.
569
## TODO: Perhaps a global option to use either universal or local time?
570
## Or perhaps just let people set $TZ?
571
assert isinstance(t, float)
656
573
if timezone == 'utc':
657
574
tt = time.gmtime(t)
659
576
elif timezone == 'original':
662
579
tt = time.gmtime(t + offset)
663
580
elif timezone == 'local':
664
581
tt = time.localtime(t)
665
582
offset = local_time_offset(t)
667
raise errors.BzrError("unsupported timezone format %r" % timezone,
668
['options are "utc", "original", "local"'])
584
raise BzrError("unsupported timezone format %r" % timezone,
585
['options are "utc", "original", "local"'])
669
586
if date_fmt is None:
670
587
date_fmt = "%a %Y-%m-%d %H:%M:%S"
679
596
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
682
def format_delta(delta):
683
"""Get a nice looking string for a time delta.
685
:param delta: The time difference in seconds, can be positive or negative.
686
positive indicates time in the past, negative indicates time in the
687
future. (usually time.time() - stored_time)
688
:return: String formatted to show approximate resolution
694
direction = 'in the future'
698
if seconds < 90: # print seconds up to 90 seconds
700
return '%d second %s' % (seconds, direction,)
702
return '%d seconds %s' % (seconds, direction)
704
minutes = int(seconds / 60)
705
seconds -= 60 * minutes
710
if minutes < 90: # print minutes, seconds up to 90 minutes
712
return '%d minute, %d second%s %s' % (
713
minutes, seconds, plural_seconds, direction)
715
return '%d minutes, %d second%s %s' % (
716
minutes, seconds, plural_seconds, direction)
718
hours = int(minutes / 60)
719
minutes -= 60 * hours
726
return '%d hour, %d minute%s %s' % (hours, minutes,
727
plural_minutes, direction)
728
return '%d hours, %d minute%s %s' % (hours, minutes,
729
plural_minutes, direction)
732
601
"""Return size of given open file."""
814
705
def link_or_copy(src, dest):
815
706
"""Hardlink a file, or copy it if it can't be hardlinked."""
816
707
if not hardlinks_good():
817
shutil.copyfile(src, dest)
820
711
os.link(src, dest)
821
712
except (OSError, IOError), e:
822
713
if e.errno != errno.EXDEV:
824
shutil.copyfile(src, dest)
827
# Look Before You Leap (LBYL) is appropriate here instead of Easier to Ask for
828
# Forgiveness than Permission (EAFP) because:
829
# - root can damage a solaris file system by using unlink,
830
# - unlink raises different exceptions on different OSes (linux: EISDIR, win32:
831
# EACCES, OSX: EPERM) when invoked on a directory.
832
def delete_any(path):
717
def delete_any(full_path):
833
718
"""Delete a file or directory."""
834
if isdir(path): # Takes care of symlinks
722
# We may be renaming a dangling inventory id
723
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
840
728
def has_symlinks():
841
if getattr(os, 'symlink', None) is not None:
729
if hasattr(os, 'symlink'):
847
735
def contains_whitespace(s):
848
736
"""True if there are any whitespace characters in s."""
849
# string.whitespace can include '\xa0' in certain locales, because it is
850
# considered "non-breaking-space" as part of ISO-8859-1. But it
851
# 1) Isn't a breaking whitespace
852
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
854
# 3) '\xa0' isn't unicode safe since it is >128.
856
# This should *not* be a unicode set of characters in case the source
857
# string is not a Unicode string. We can auto-up-cast the characters since
858
# they are ascii, but we don't want to auto-up-cast the string in case it
860
for ch in ' \t\n\r\v\f':
737
for ch in string.whitespace:
921
798
return unicode_or_utf8_string.decode('utf8')
922
799
except UnicodeDecodeError:
923
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
926
def safe_utf8(unicode_or_utf8_string):
927
"""Coerce unicode_or_utf8_string to a utf8 string.
929
If it is a str, it is returned.
930
If it is Unicode, it is encoded into a utf-8 string.
932
if isinstance(unicode_or_utf8_string, str):
933
# TODO: jam 20070209 This is overkill, and probably has an impact on
934
# performance if we are dealing with lots of apis that want a
937
# Make sure it is a valid utf-8 string
938
unicode_or_utf8_string.decode('utf-8')
939
except UnicodeDecodeError:
940
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
941
return unicode_or_utf8_string
942
return unicode_or_utf8_string.encode('utf-8')
945
_revision_id_warning = ('Unicode revision ids were deprecated in bzr 0.15.'
946
' Revision id generators should be creating utf8'
950
def safe_revision_id(unicode_or_utf8_string, warn=True):
951
"""Revision ids should now be utf8, but at one point they were unicode.
953
:param unicode_or_utf8_string: A possibly Unicode revision_id. (can also be
955
:param warn: Functions that are sanitizing user data can set warn=False
956
:return: None or a utf8 revision id.
958
if (unicode_or_utf8_string is None
959
or unicode_or_utf8_string.__class__ == str):
960
return unicode_or_utf8_string
962
symbol_versioning.warn(_revision_id_warning, DeprecationWarning,
964
return cache_utf8.encode(unicode_or_utf8_string)
967
_file_id_warning = ('Unicode file ids were deprecated in bzr 0.15. File id'
968
' generators should be creating utf8 file ids.')
971
def safe_file_id(unicode_or_utf8_string, warn=True):
972
"""File ids should now be utf8, but at one point they were unicode.
974
This is the same as safe_utf8, except it uses the cached encode functions
975
to save a little bit of performance.
977
:param unicode_or_utf8_string: A possibly Unicode file_id. (can also be
979
:param warn: Functions that are sanitizing user data can set warn=False
980
:return: None or a utf8 file id.
982
if (unicode_or_utf8_string is None
983
or unicode_or_utf8_string.__class__ == str):
984
return unicode_or_utf8_string
986
symbol_versioning.warn(_file_id_warning, DeprecationWarning,
988
return cache_utf8.encode(unicode_or_utf8_string)
800
raise BzrBadParameterNotUnicode(unicode_or_utf8_string)
991
803
_platform_normalizes_filenames = False
1058
870
def supports_executable():
1059
871
return sys.platform != "win32"
1062
def supports_posix_readonly():
1063
"""Return True if 'readonly' has POSIX semantics, False otherwise.
1065
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
1066
directory controls creation/deletion, etc.
1068
And under win32, readonly means that the directory itself cannot be
1069
deleted. The contents of a readonly directory can be changed, unlike POSIX
1070
where files in readonly directories cannot be added, deleted or renamed.
1072
return sys.platform != "win32"
1075
def set_or_unset_env(env_variable, value):
1076
"""Modify the environment, setting or removing the env_variable.
1078
:param env_variable: The environment variable in question
1079
:param value: The value to set the environment to. If None, then
1080
the variable will be removed.
1081
:return: The original value of the environment variable.
1083
orig_val = os.environ.get(env_variable)
1085
if orig_val is not None:
1086
del os.environ[env_variable]
1088
if isinstance(value, unicode):
1089
value = value.encode(bzrlib.user_encoding)
1090
os.environ[env_variable] = value
1094
874
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
1138
918
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1139
919
# potentially confusing output. We should make this more robust - but
1140
920
# not at a speed cost. RBC 20060731
1142
923
_directory = _directory_kind
1143
_listdir = os.listdir
1144
_kind_from_mode = _formats.get
1145
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
925
pending = [(prefix, "", _directory, None, top)]
928
currentdir = pending.pop()
1147
929
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1148
relroot, _, _, _, top = pending.pop()
1150
relprefix = relroot + u'/'
1153
top_slash = top + u'/'
1156
append = dirblock.append
1157
for name in sorted(_listdir(top)):
1158
abspath = top_slash + name
1159
statvalue = _lstat(abspath)
1160
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1161
append((relprefix + name, name, kind, statvalue, abspath))
1162
yield (relroot, top), dirblock
1164
# push the user specified dirs from dirblock
1165
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1168
def _walkdirs_utf8(top, prefix=""):
1169
"""Yield data about all the directories in a tree.
1171
This yields the same information as walkdirs() only each entry is yielded
1172
in utf-8. On platforms which have a filesystem encoding of utf8 the paths
1173
are returned as exact byte-strings.
1175
:return: yields a tuple of (dir_info, [file_info])
1176
dir_info is (utf8_relpath, path-from-top)
1177
file_info is (utf8_relpath, utf8_name, kind, lstat, path-from-top)
1178
if top is an absolute path, path-from-top is also an absolute path.
1179
path-from-top might be unicode or utf8, but it is the correct path to
1180
pass to os functions to affect the file in question. (such as os.lstat)
1182
fs_encoding = _fs_enc.upper()
1183
if (sys.platform == 'win32' or
1184
fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968')): # ascii
1185
return _walkdirs_unicode_to_utf8(top, prefix=prefix)
1187
return _walkdirs_fs_utf8(top, prefix=prefix)
1190
def _walkdirs_fs_utf8(top, prefix=""):
1191
"""See _walkdirs_utf8.
1193
This sub-function is called when we know the filesystem is already in utf8
1194
encoding. So we don't need to transcode filenames.
1197
_directory = _directory_kind
1198
_listdir = os.listdir
1199
_kind_from_mode = _formats.get
1201
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1202
# But we don't actually uses 1-3 in pending, so set them to None
1203
pending = [(safe_utf8(prefix), None, None, None, safe_utf8(top))]
1205
relroot, _, _, _, top = pending.pop()
1207
relprefix = relroot + '/'
1210
top_slash = top + '/'
1213
append = dirblock.append
1214
for name in sorted(_listdir(top)):
1215
abspath = top_slash + name
1216
statvalue = _lstat(abspath)
1217
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1218
append((relprefix + name, name, kind, statvalue, abspath))
1219
yield (relroot, top), dirblock
1221
# push the user specified dirs from dirblock
1222
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1225
def _walkdirs_unicode_to_utf8(top, prefix=""):
1226
"""See _walkdirs_utf8
1228
Because Win32 has a Unicode api, all of the 'path-from-top' entries will be
1230
This is currently the fallback code path when the filesystem encoding is
1231
not UTF-8. It may be better to implement an alternative so that we can
1232
safely handle paths that are not properly decodable in the current
1235
_utf8_encode = codecs.getencoder('utf8')
1237
_directory = _directory_kind
1238
_listdir = os.listdir
1239
_kind_from_mode = _formats.get
1241
pending = [(safe_utf8(prefix), None, None, None, safe_unicode(top))]
1243
relroot, _, _, _, top = pending.pop()
1245
relprefix = relroot + '/'
1248
top_slash = top + u'/'
1251
append = dirblock.append
1252
for name in sorted(_listdir(top)):
1253
name_utf8 = _utf8_encode(name)[0]
1254
abspath = top_slash + name
1255
statvalue = _lstat(abspath)
1256
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1257
append((relprefix + name_utf8, name_utf8, kind, statvalue, abspath))
1258
yield (relroot, top), dirblock
1260
# push the user specified dirs from dirblock
1261
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
932
relroot = currentdir[0] + '/'
935
for name in sorted(_listdir(top)):
936
abspath = top + '/' + name
937
statvalue = lstat(abspath)
938
dirblock.append((relroot + name, name,
939
file_kind_from_stat_mode(statvalue.st_mode),
941
yield (currentdir[0], top), dirblock
942
# push the user specified dirs from dirblock
943
for dir in reversed(dirblock):
944
if dir[2] == _directory:
1264
948
def copy_tree(from_path, to_path, handlers={}):
1316
1000
key_a = path_prefix_key(path_a)
1317
1001
key_b = path_prefix_key(path_b)
1318
1002
return cmp(key_a, key_b)
1321
_cached_user_encoding = None
1324
def get_user_encoding(use_cache=True):
1325
"""Find out what the preferred user encoding is.
1327
This is generally the encoding that is used for command line parameters
1328
and file contents. This may be different from the terminal encoding
1329
or the filesystem encoding.
1331
:param use_cache: Enable cache for detected encoding.
1332
(This parameter is turned on by default,
1333
and required only for selftesting)
1335
:return: A string defining the preferred user encoding
1337
global _cached_user_encoding
1338
if _cached_user_encoding is not None and use_cache:
1339
return _cached_user_encoding
1341
if sys.platform == 'darwin':
1342
# work around egregious python 2.4 bug
1343
sys.platform = 'posix'
1347
sys.platform = 'darwin'
1352
user_encoding = locale.getpreferredencoding()
1353
except locale.Error, e:
1354
sys.stderr.write('bzr: warning: %s\n'
1355
' Could not determine what text encoding to use.\n'
1356
' This error usually means your Python interpreter\n'
1357
' doesn\'t support the locale set by $LANG (%s)\n'
1358
" Continuing with ascii encoding.\n"
1359
% (e, os.environ.get('LANG')))
1360
user_encoding = 'ascii'
1362
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1363
# treat that as ASCII, and not support printing unicode characters to the
1365
if user_encoding in (None, 'cp0'):
1366
user_encoding = 'ascii'
1370
codecs.lookup(user_encoding)
1372
sys.stderr.write('bzr: warning:'
1373
' unknown encoding %s.'
1374
' Continuing with ascii encoding.\n'
1377
user_encoding = 'ascii'
1380
_cached_user_encoding = user_encoding
1382
return user_encoding
1385
def recv_all(socket, bytes):
1386
"""Receive an exact number of bytes.
1388
Regular Socket.recv() may return less than the requested number of bytes,
1389
dependning on what's in the OS buffer. MSG_WAITALL is not available
1390
on all platforms, but this should work everywhere. This will return
1391
less than the requested amount if the remote end closes.
1393
This isn't optimized and is intended mostly for use in testing.
1396
while len(b) < bytes:
1397
new = socket.recv(bytes - len(b))
1403
def dereference_path(path):
1404
"""Determine the real path to a file.
1406
All parent elements are dereferenced. But the file itself is not
1408
:param path: The original path. May be absolute or relative.
1409
:return: the real path *to* the file
1411
parent, base = os.path.split(path)
1412
# The pathjoin for '.' is a workaround for Python bug #1213894.
1413
# (initial path components aren't dereferenced)
1414
return pathjoin(realpath(pathjoin('.', parent)), base)
1417
def supports_mapi():
1418
"""Return True if we can use MAPI to launch a mail client."""
1419
return sys.platform == "win32"
added
removed
bzrlib/osutils.py
