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)
72
def make_readonly(filename):
73
"""Make a filename read-only."""
74
mod = os.lstat(filename).st_mode
75
if not stat.S_ISLNK(mod):
77
os.chmod(filename, mod)
80
def make_writable(filename):
81
mod = os.lstat(filename).st_mode
82
if not stat.S_ISLNK(mod):
84
os.chmod(filename, mod)
87
def minimum_path_selection(paths):
88
"""Return the smallset subset of paths which are outside paths.
90
:param paths: A container (and hence not None) of paths.
91
:return: A set of paths sufficient to include everything in paths via
92
is_inside_any, drawn from the paths parameter.
97
other_paths = paths.difference([path])
98
if not is_inside_any(other_paths, path):
99
# this is a top level path, we must check it.
100
search_paths.add(path)
108
"""Return a quoted filename filename
110
This previously used backslash quoting, but that works poorly on
112
# TODO: I'm not really sure this is the best format either.x
114
if _QUOTE_RE is None:
115
_QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')
117
if _QUOTE_RE.search(f):
123
_directory_kind = 'directory'
126
stat.S_IFDIR:_directory_kind,
127
stat.S_IFCHR:'chardev',
128
stat.S_IFBLK:'block',
131
stat.S_IFLNK:'symlink',
132
stat.S_IFSOCK:'socket',
136
def file_kind_from_stat_mode(stat_mode, _formats=_formats, _unknown='unknown'):
137
"""Generate a file kind from a stat mode. This is used in walkdirs.
139
Its performance is critical: Do not mutate without careful benchmarking.
142
return _formats[stat_mode & 0170000]
147
def file_kind(f, _lstat=os.lstat, _mapper=file_kind_from_stat_mode):
149
return _mapper(_lstat(f).st_mode)
151
if getattr(e, 'errno', None) in (errno.ENOENT, errno.ENOTDIR):
152
raise errors.NoSuchFile(f)
157
"""Return the current umask"""
158
# Assume that people aren't messing with the umask while running
159
# XXX: This is not thread safe, but there is no way to get the
160
# umask without setting it
168
_directory_kind: "/",
170
'tree-reference': '+',
174
def kind_marker(kind):
176
return _kind_marker_map[kind]
178
raise errors.BzrError('invalid file kind %r' % kind)
181
lexists = getattr(os.path, 'lexists', None)
185
stat = getattr(os, 'lstat', os.stat)
189
if e.errno == errno.ENOENT:
192
raise errors.BzrError("lstat/stat of (%r): %r" % (f, e))
195
def fancy_rename(old, new, rename_func, unlink_func):
196
"""A fancy rename, when you don't have atomic rename.
198
:param old: The old path, to rename from
199
:param new: The new path, to rename to
200
:param rename_func: The potentially non-atomic rename function
201
:param unlink_func: A way to delete the target file if the full rename succeeds
204
# sftp rename doesn't allow overwriting, so play tricks:
206
base = os.path.basename(new)
207
dirname = os.path.dirname(new)
208
tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))
209
tmp_name = pathjoin(dirname, tmp_name)
211
# Rename the file out of the way, but keep track if it didn't exist
212
# We don't want to grab just any exception
213
# something like EACCES should prevent us from continuing
214
# The downside is that the rename_func has to throw an exception
215
# with an errno = ENOENT, or NoSuchFile
218
rename_func(new, tmp_name)
219
except (errors.NoSuchFile,), e:
222
# RBC 20060103 abstraction leakage: the paramiko SFTP clients rename
223
# function raises an IOError with errno is None when a rename fails.
224
# This then gets caught here.
225
if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):
228
if (getattr(e, 'errno', None) is None
229
or e.errno not in (errno.ENOENT, errno.ENOTDIR)):
237
# This may throw an exception, in which case success will
239
rename_func(old, new)
241
except (IOError, OSError), e:
242
# source and target may be aliases of each other (e.g. on a
243
# case-insensitive filesystem), so we may have accidentally renamed
244
# source by when we tried to rename target
245
if not (file_existed and e.errno in (None, errno.ENOENT)):
249
# If the file used to exist, rename it back into place
250
# otherwise just delete it from the tmp location
252
unlink_func(tmp_name)
254
rename_func(tmp_name, new)
257
# In Python 2.4.2 and older, os.path.abspath and os.path.realpath
258
# choke on a Unicode string containing a relative path if
259
# os.getcwd() returns a non-sys.getdefaultencoding()-encoded
261
_fs_enc = sys.getfilesystemencoding() or 'utf-8'
262
def _posix_abspath(path):
263
# jam 20060426 rather than encoding to fsencoding
264
# copy posixpath.abspath, but use os.getcwdu instead
265
if not posixpath.isabs(path):
266
path = posixpath.join(getcwd(), path)
267
return posixpath.normpath(path)
270
def _posix_realpath(path):
271
return posixpath.realpath(path.encode(_fs_enc)).decode(_fs_enc)
274
def _win32_fixdrive(path):
275
"""Force drive letters to be consistent.
277
win32 is inconsistent whether it returns lower or upper case
278
and even if it was consistent the user might type the other
279
so we force it to uppercase
280
running python.exe under cmd.exe return capital C:\\
281
running win32 python inside a cygwin shell returns lowercase c:\\
283
drive, path = _nt_splitdrive(path)
284
return drive.upper() + path
287
def _win32_abspath(path):
288
# Real _nt_abspath doesn't have a problem with a unicode cwd
289
return _win32_fixdrive(_nt_abspath(unicode(path)).replace('\\', '/'))
292
def _win98_abspath(path):
293
"""Return the absolute version of a path.
294
Windows 98 safe implementation (python reimplementation
295
of Win32 API function GetFullPathNameW)
300
# \\HOST\path => //HOST/path
301
# //HOST/path => //HOST/path
302
# path => C:/cwd/path
305
# check for absolute path
306
drive = _nt_splitdrive(path)[0]
307
if drive == '' and path[:2] not in('//','\\\\'):
309
# we cannot simply os.path.join cwd and path
310
# because os.path.join('C:','/path') produce '/path'
311
# and this is incorrect
312
if path[:1] in ('/','\\'):
313
cwd = _nt_splitdrive(cwd)[0]
315
path = cwd + '\\' + path
316
return _win32_fixdrive(_nt_normpath(path).replace('\\', '/'))
318
if win32utils.winver == 'Windows 98':
319
_win32_abspath = _win98_abspath
322
def _win32_realpath(path):
323
# Real _nt_realpath doesn't have a problem with a unicode cwd
324
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
327
def _win32_pathjoin(*args):
328
return _nt_join(*args).replace('\\', '/')
331
def _win32_normpath(path):
332
return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))
336
return _win32_fixdrive(os.getcwdu().replace('\\', '/'))
339
def _win32_mkdtemp(*args, **kwargs):
340
return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))
343
def _win32_rename(old, new):
344
"""We expect to be able to atomically replace 'new' with old.
346
On win32, if new exists, it must be moved out of the way first,
350
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
352
if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):
353
# If we try to rename a non-existant file onto cwd, we get
354
# EPERM or EACCES instead of ENOENT, this will raise ENOENT
355
# if the old path doesn't exist, sometimes we get EACCES
356
# On Linux, we seem to get EBUSY, on Mac we get EINVAL
362
return unicodedata.normalize('NFC', os.getcwdu())
365
# Default is to just use the python builtins, but these can be rebound on
366
# particular platforms.
367
abspath = _posix_abspath
368
realpath = _posix_realpath
369
pathjoin = os.path.join
370
normpath = os.path.normpath
373
dirname = os.path.dirname
374
basename = os.path.basename
375
split = os.path.split
376
splitext = os.path.splitext
377
# These were already imported into local scope
378
# mkdtemp = tempfile.mkdtemp
379
# rmtree = shutil.rmtree
381
MIN_ABS_PATHLENGTH = 1
384
if sys.platform == 'win32':
385
abspath = _win32_abspath
386
realpath = _win32_realpath
387
pathjoin = _win32_pathjoin
388
normpath = _win32_normpath
389
getcwd = _win32_getcwd
390
mkdtemp = _win32_mkdtemp
391
rename = _win32_rename
393
MIN_ABS_PATHLENGTH = 3
395
def _win32_delete_readonly(function, path, excinfo):
396
"""Error handler for shutil.rmtree function [for win32]
397
Helps to remove files and dirs marked as read-only.
399
exception = excinfo[1]
400
if function in (os.remove, os.rmdir) \
401
and isinstance(exception, OSError) \
402
and exception.errno == errno.EACCES:
408
def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):
409
"""Replacer for shutil.rmtree: could remove readonly dirs/files"""
410
return shutil.rmtree(path, ignore_errors, onerror)
411
elif sys.platform == 'darwin':
415
def get_terminal_encoding():
416
"""Find the best encoding for printing to the screen.
418
This attempts to check both sys.stdout and sys.stdin to see
419
what encoding they are in, and if that fails it falls back to
420
bzrlib.user_encoding.
421
The problem is that on Windows, locale.getpreferredencoding()
422
is not the same encoding as that used by the console:
423
http://mail.python.org/pipermail/python-list/2003-May/162357.html
425
On my standard US Windows XP, the preferred encoding is
426
cp1252, but the console is cp437
428
output_encoding = getattr(sys.stdout, 'encoding', None)
429
if not output_encoding:
430
input_encoding = getattr(sys.stdin, 'encoding', None)
431
if not input_encoding:
432
output_encoding = bzrlib.user_encoding
433
mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)
435
output_encoding = input_encoding
436
mutter('encoding stdout as sys.stdin encoding %r', output_encoding)
438
mutter('encoding stdout as sys.stdout encoding %r', output_encoding)
439
if output_encoding == 'cp0':
440
# invalid encoding (cp0 means 'no codepage' on Windows)
441
output_encoding = bzrlib.user_encoding
442
mutter('cp0 is invalid encoding.'
443
' encoding stdout as bzrlib.user_encoding %r', output_encoding)
446
codecs.lookup(output_encoding)
448
sys.stderr.write('bzr: warning:'
449
' unknown terminal encoding %s.\n'
450
' Using encoding %s instead.\n'
451
% (output_encoding, bzrlib.user_encoding)
453
output_encoding = bzrlib.user_encoding
455
return output_encoding
458
def normalizepath(f):
459
if getattr(os.path, 'realpath', None) is not None:
463
[p,e] = os.path.split(f)
464
if e == "" or e == "." or e == "..":
467
return pathjoin(F(p), e)
471
"""True if f is an accessible directory."""
473
return S_ISDIR(os.lstat(f)[ST_MODE])
479
"""True if f is a regular file."""
481
return S_ISREG(os.lstat(f)[ST_MODE])
486
"""True if f is a symlink."""
488
return S_ISLNK(os.lstat(f)[ST_MODE])
492
def is_inside(dir, fname):
493
"""True if fname is inside dir.
495
The parameters should typically be passed to osutils.normpath first, so
496
that . and .. and repeated slashes are eliminated, and the separators
497
are canonical for the platform.
499
The empty string as a dir name is taken as top-of-tree and matches
502
# XXX: Most callers of this can actually do something smarter by
503
# looking at the inventory
513
return fname.startswith(dir)
516
def is_inside_any(dir_list, fname):
517
"""True if fname is inside any of given dirs."""
518
for dirname in dir_list:
519
if is_inside(dirname, fname):
524
def is_inside_or_parent_of_any(dir_list, fname):
525
"""True if fname is a child or a parent of any of the given files."""
526
for dirname in dir_list:
527
if is_inside(dirname, fname) or is_inside(fname, dirname):
532
def pumpfile(fromfile, tofile):
533
"""Copy contents of one file to another.
535
:return: The number of bytes copied.
540
b = fromfile.read(BUFSIZE)
548
def file_iterator(input_file, readsize=32768):
550
b = input_file.read(readsize)
557
if getattr(f, 'tell', None) is not None:
569
def sha_file_by_name(fname):
570
"""Calculate the SHA1 of a file by reading the full text"""
572
f = os.open(fname, os.O_RDONLY | O_BINARY)
575
b = os.read(f, 1<<16)
583
def sha_strings(strings, _factory=sha.new):
584
"""Return the sha-1 of concatenation of strings"""
586
map(s.update, strings)
590
def sha_string(f, _factory=sha.new):
591
return _factory(f).hexdigest()
594
def fingerprint_file(f):
596
return {'size': len(b),
597
'sha1': sha.new(b).hexdigest()}
600
def compare_files(a, b):
601
"""Returns true if equal in contents"""
612
def local_time_offset(t=None):
613
"""Return offset of local zone from GMT, either at present or at time t."""
616
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
617
return offset.days * 86400 + offset.seconds
620
def format_date(t, offset=0, timezone='original', date_fmt=None,
622
"""Return a formatted date string.
624
:param t: Seconds since the epoch.
625
:param offset: Timezone offset in seconds east of utc.
626
:param timezone: How to display the time: 'utc', 'original' for the
627
timezone specified by offset, or 'local' for the process's current
629
:param show_offset: Whether to append the timezone.
630
:param date_fmt: strftime format.
632
if timezone == 'utc':
635
elif timezone == 'original':
638
tt = time.gmtime(t + offset)
639
elif timezone == 'local':
640
tt = time.localtime(t)
641
offset = local_time_offset(t)
643
raise errors.UnsupportedTimezoneFormat(timezone)
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)
802
# Look Before You Leap (LBYL) is appropriate here instead of Easier to Ask for
803
# Forgiveness than Permission (EAFP) because:
804
# - root can damage a solaris file system by using unlink,
805
# - unlink raises different exceptions on different OSes (linux: EISDIR, win32:
806
# EACCES, OSX: EPERM) when invoked on a directory.
807
def delete_any(path):
808
"""Delete a file or directory."""
809
if isdir(path): # Takes care of symlinks
816
if getattr(os, 'symlink', None) is not None:
823
if getattr(os, 'link', None) is not None:
829
def contains_whitespace(s):
830
"""True if there are any whitespace characters in s."""
831
# string.whitespace can include '\xa0' in certain locales, because it is
832
# considered "non-breaking-space" as part of ISO-8859-1. But it
833
# 1) Isn't a breaking whitespace
834
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
836
# 3) '\xa0' isn't unicode safe since it is >128.
838
# This should *not* be a unicode set of characters in case the source
839
# string is not a Unicode string. We can auto-up-cast the characters since
840
# they are ascii, but we don't want to auto-up-cast the string in case it
842
for ch in ' \t\n\r\v\f':
849
def contains_linebreaks(s):
850
"""True if there is any vertical whitespace in s."""
858
def relpath(base, path):
859
"""Return path relative to base, or raise exception.
861
The path may be either an absolute path or a path relative to the
862
current working directory.
864
os.path.commonprefix (python2.4) has a bad bug that it works just
865
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
869
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
870
' exceed the platform minimum length (which is %d)' %
877
while len(head) >= len(base):
880
head, tail = os.path.split(head)
884
raise errors.PathNotChild(rp, base)
892
def safe_unicode(unicode_or_utf8_string):
893
"""Coerce unicode_or_utf8_string into unicode.
895
If it is unicode, it is returned.
896
Otherwise it is decoded from utf-8. If a decoding error
897
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
898
as a BzrBadParameter exception.
900
if isinstance(unicode_or_utf8_string, unicode):
901
return unicode_or_utf8_string
903
return unicode_or_utf8_string.decode('utf8')
904
except UnicodeDecodeError:
905
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
908
def safe_utf8(unicode_or_utf8_string):
909
"""Coerce unicode_or_utf8_string to a utf8 string.
911
If it is a str, it is returned.
912
If it is Unicode, it is encoded into a utf-8 string.
914
if isinstance(unicode_or_utf8_string, str):
915
# TODO: jam 20070209 This is overkill, and probably has an impact on
916
# performance if we are dealing with lots of apis that want a
919
# Make sure it is a valid utf-8 string
920
unicode_or_utf8_string.decode('utf-8')
921
except UnicodeDecodeError:
922
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
923
return unicode_or_utf8_string
924
return unicode_or_utf8_string.encode('utf-8')
927
_revision_id_warning = ('Unicode revision ids were deprecated in bzr 0.15.'
928
' Revision id generators should be creating utf8'
932
def safe_revision_id(unicode_or_utf8_string, warn=True):
933
"""Revision ids should now be utf8, but at one point they were unicode.
935
:param unicode_or_utf8_string: A possibly Unicode revision_id. (can also be
937
:param warn: Functions that are sanitizing user data can set warn=False
938
:return: None or a utf8 revision id.
940
if (unicode_or_utf8_string is None
941
or unicode_or_utf8_string.__class__ == str):
942
return unicode_or_utf8_string
944
symbol_versioning.warn(_revision_id_warning, DeprecationWarning,
946
return cache_utf8.encode(unicode_or_utf8_string)
949
_file_id_warning = ('Unicode file ids were deprecated in bzr 0.15. File id'
950
' generators should be creating utf8 file ids.')
953
def safe_file_id(unicode_or_utf8_string, warn=True):
954
"""File ids should now be utf8, but at one point they were unicode.
956
This is the same as safe_utf8, except it uses the cached encode functions
957
to save a little bit of performance.
959
:param unicode_or_utf8_string: A possibly Unicode file_id. (can also be
961
:param warn: Functions that are sanitizing user data can set warn=False
962
:return: None or a utf8 file id.
964
if (unicode_or_utf8_string is None
965
or unicode_or_utf8_string.__class__ == str):
966
return unicode_or_utf8_string
968
symbol_versioning.warn(_file_id_warning, DeprecationWarning,
970
return cache_utf8.encode(unicode_or_utf8_string)
973
_platform_normalizes_filenames = False
974
if sys.platform == 'darwin':
975
_platform_normalizes_filenames = True
978
def normalizes_filenames():
979
"""Return True if this platform normalizes unicode filenames.
981
Mac OSX does, Windows/Linux do not.
983
return _platform_normalizes_filenames
986
def _accessible_normalized_filename(path):
987
"""Get the unicode normalized path, and if you can access the file.
989
On platforms where the system normalizes filenames (Mac OSX),
990
you can access a file by any path which will normalize correctly.
991
On platforms where the system does not normalize filenames
992
(Windows, Linux), you have to access a file by its exact path.
994
Internally, bzr only supports NFC normalization, since that is
995
the standard for XML documents.
997
So return the normalized path, and a flag indicating if the file
998
can be accessed by that path.
1001
return unicodedata.normalize('NFC', unicode(path)), True
1004
def _inaccessible_normalized_filename(path):
1005
__doc__ = _accessible_normalized_filename.__doc__
1007
normalized = unicodedata.normalize('NFC', unicode(path))
1008
return normalized, normalized == path
1011
if _platform_normalizes_filenames:
1012
normalized_filename = _accessible_normalized_filename
1014
normalized_filename = _inaccessible_normalized_filename
1017
def terminal_width():
1018
"""Return estimated terminal width."""
1019
if sys.platform == 'win32':
1020
return win32utils.get_console_size()[0]
1023
import struct, fcntl, termios
1024
s = struct.pack('HHHH', 0, 0, 0, 0)
1025
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
1026
width = struct.unpack('HHHH', x)[1]
1031
width = int(os.environ['COLUMNS'])
1040
def supports_executable():
1041
return sys.platform != "win32"
1044
def supports_posix_readonly():
1045
"""Return True if 'readonly' has POSIX semantics, False otherwise.
1047
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
1048
directory controls creation/deletion, etc.
1050
And under win32, readonly means that the directory itself cannot be
1051
deleted. The contents of a readonly directory can be changed, unlike POSIX
1052
where files in readonly directories cannot be added, deleted or renamed.
1054
return sys.platform != "win32"
1057
def set_or_unset_env(env_variable, value):
1058
"""Modify the environment, setting or removing the env_variable.
1060
:param env_variable: The environment variable in question
1061
:param value: The value to set the environment to. If None, then
1062
the variable will be removed.
1063
:return: The original value of the environment variable.
1065
orig_val = os.environ.get(env_variable)
1067
if orig_val is not None:
1068
del os.environ[env_variable]
1070
if isinstance(value, unicode):
1071
value = value.encode(bzrlib.user_encoding)
1072
os.environ[env_variable] = value
1076
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
1079
def check_legal_path(path):
1080
"""Check whether the supplied path is legal.
1081
This is only required on Windows, so we don't test on other platforms
1084
if sys.platform != "win32":
1086
if _validWin32PathRE.match(path) is None:
1087
raise errors.IllegalPath(path)
1090
def walkdirs(top, prefix=""):
1091
"""Yield data about all the directories in a tree.
1093
This yields all the data about the contents of a directory at a time.
1094
After each directory has been yielded, if the caller has mutated the list
1095
to exclude some directories, they are then not descended into.
1097
The data yielded is of the form:
1098
((directory-relpath, directory-path-from-top),
1099
[(relpath, basename, kind, lstat, path-from-top), ...]),
1100
- directory-relpath is the relative path of the directory being returned
1101
with respect to top. prefix is prepended to this.
1102
- directory-path-from-root is the path including top for this directory.
1103
It is suitable for use with os functions.
1104
- relpath is the relative path within the subtree being walked.
1105
- basename is the basename of the path
1106
- kind is the kind of the file now. If unknown then the file is not
1107
present within the tree - but it may be recorded as versioned. See
1109
- lstat is the stat data *if* the file was statted.
1110
- planned, not implemented:
1111
path_from_tree_root is the path from the root of the tree.
1113
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1114
allows one to walk a subtree but get paths that are relative to a tree
1116
:return: an iterator over the dirs.
1118
#TODO there is a bit of a smell where the results of the directory-
1119
# summary in this, and the path from the root, may not agree
1120
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1121
# potentially confusing output. We should make this more robust - but
1122
# not at a speed cost. RBC 20060731
1124
_directory = _directory_kind
1125
_listdir = os.listdir
1126
_kind_from_mode = _formats.get
1127
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
1129
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1130
relroot, _, _, _, top = pending.pop()
1132
relprefix = relroot + u'/'
1135
top_slash = top + u'/'
1138
append = dirblock.append
1139
for name in sorted(_listdir(top)):
1140
abspath = top_slash + name
1141
statvalue = _lstat(abspath)
1142
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1143
append((relprefix + name, name, kind, statvalue, abspath))
1144
yield (relroot, top), dirblock
1146
# push the user specified dirs from dirblock
1147
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1150
def _walkdirs_utf8(top, prefix=""):
1151
"""Yield data about all the directories in a tree.
1153
This yields the same information as walkdirs() only each entry is yielded
1154
in utf-8. On platforms which have a filesystem encoding of utf8 the paths
1155
are returned as exact byte-strings.
1157
:return: yields a tuple of (dir_info, [file_info])
1158
dir_info is (utf8_relpath, path-from-top)
1159
file_info is (utf8_relpath, utf8_name, kind, lstat, path-from-top)
1160
if top is an absolute path, path-from-top is also an absolute path.
1161
path-from-top might be unicode or utf8, but it is the correct path to
1162
pass to os functions to affect the file in question. (such as os.lstat)
1164
fs_encoding = _fs_enc.upper()
1165
if (sys.platform == 'win32' or
1166
fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968')): # ascii
1167
return _walkdirs_unicode_to_utf8(top, prefix=prefix)
1169
return _walkdirs_fs_utf8(top, prefix=prefix)
1172
def _walkdirs_fs_utf8(top, prefix=""):
1173
"""See _walkdirs_utf8.
1175
This sub-function is called when we know the filesystem is already in utf8
1176
encoding. So we don't need to transcode filenames.
1179
_directory = _directory_kind
1180
_listdir = os.listdir
1181
_kind_from_mode = _formats.get
1183
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1184
# But we don't actually uses 1-3 in pending, so set them to None
1185
pending = [(safe_utf8(prefix), None, None, None, safe_utf8(top))]
1187
relroot, _, _, _, top = pending.pop()
1189
relprefix = relroot + '/'
1192
top_slash = top + '/'
1195
append = dirblock.append
1196
for name in sorted(_listdir(top)):
1197
abspath = top_slash + name
1198
statvalue = _lstat(abspath)
1199
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1200
append((relprefix + name, name, kind, statvalue, abspath))
1201
yield (relroot, top), dirblock
1203
# push the user specified dirs from dirblock
1204
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1207
def _walkdirs_unicode_to_utf8(top, prefix=""):
1208
"""See _walkdirs_utf8
1210
Because Win32 has a Unicode api, all of the 'path-from-top' entries will be
1212
This is currently the fallback code path when the filesystem encoding is
1213
not UTF-8. It may be better to implement an alternative so that we can
1214
safely handle paths that are not properly decodable in the current
1217
_utf8_encode = codecs.getencoder('utf8')
1219
_directory = _directory_kind
1220
_listdir = os.listdir
1221
_kind_from_mode = _formats.get
1223
pending = [(safe_utf8(prefix), None, None, None, safe_unicode(top))]
1225
relroot, _, _, _, top = pending.pop()
1227
relprefix = relroot + '/'
1230
top_slash = top + u'/'
1233
append = dirblock.append
1234
for name in sorted(_listdir(top)):
1235
name_utf8 = _utf8_encode(name)[0]
1236
abspath = top_slash + name
1237
statvalue = _lstat(abspath)
1238
kind = _kind_from_mode(statvalue.st_mode & 0170000, 'unknown')
1239
append((relprefix + name_utf8, name_utf8, kind, statvalue, abspath))
1240
yield (relroot, top), dirblock
1242
# push the user specified dirs from dirblock
1243
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1246
def copy_tree(from_path, to_path, handlers={}):
1247
"""Copy all of the entries in from_path into to_path.
1249
:param from_path: The base directory to copy.
1250
:param to_path: The target directory. If it does not exist, it will
1252
:param handlers: A dictionary of functions, which takes a source and
1253
destinations for files, directories, etc.
1254
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1255
'file', 'directory', and 'symlink' should always exist.
1256
If they are missing, they will be replaced with 'os.mkdir()',
1257
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1259
# Now, just copy the existing cached tree to the new location
1260
# We use a cheap trick here.
1261
# Absolute paths are prefixed with the first parameter
1262
# relative paths are prefixed with the second.
1263
# So we can get both the source and target returned
1264
# without any extra work.
1266
def copy_dir(source, dest):
1269
def copy_link(source, dest):
1270
"""Copy the contents of a symlink"""
1271
link_to = os.readlink(source)
1272
os.symlink(link_to, dest)
1274
real_handlers = {'file':shutil.copy2,
1275
'symlink':copy_link,
1276
'directory':copy_dir,
1278
real_handlers.update(handlers)
1280
if not os.path.exists(to_path):
1281
real_handlers['directory'](from_path, to_path)
1283
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1284
for relpath, name, kind, st, abspath in entries:
1285
real_handlers[kind](abspath, relpath)
1288
def path_prefix_key(path):
1289
"""Generate a prefix-order path key for path.
1291
This can be used to sort paths in the same way that walkdirs does.
1293
return (dirname(path) , path)
1296
def compare_paths_prefix_order(path_a, path_b):
1297
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1298
key_a = path_prefix_key(path_a)
1299
key_b = path_prefix_key(path_b)
1300
return cmp(key_a, key_b)
1303
_cached_user_encoding = None
1306
def get_user_encoding(use_cache=True):
1307
"""Find out what the preferred user encoding is.
1309
This is generally the encoding that is used for command line parameters
1310
and file contents. This may be different from the terminal encoding
1311
or the filesystem encoding.
1313
:param use_cache: Enable cache for detected encoding.
1314
(This parameter is turned on by default,
1315
and required only for selftesting)
1317
:return: A string defining the preferred user encoding
1319
global _cached_user_encoding
1320
if _cached_user_encoding is not None and use_cache:
1321
return _cached_user_encoding
1323
if sys.platform == 'darwin':
1324
# work around egregious python 2.4 bug
1325
sys.platform = 'posix'
1329
sys.platform = 'darwin'
1334
user_encoding = locale.getpreferredencoding()
1335
except locale.Error, e:
1336
sys.stderr.write('bzr: warning: %s\n'
1337
' Could not determine what text encoding to use.\n'
1338
' This error usually means your Python interpreter\n'
1339
' doesn\'t support the locale set by $LANG (%s)\n'
1340
" Continuing with ascii encoding.\n"
1341
% (e, os.environ.get('LANG')))
1342
user_encoding = 'ascii'
1344
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1345
# treat that as ASCII, and not support printing unicode characters to the
1347
if user_encoding in (None, 'cp0'):
1348
user_encoding = 'ascii'
1352
codecs.lookup(user_encoding)
1354
sys.stderr.write('bzr: warning:'
1355
' unknown encoding %s.'
1356
' Continuing with ascii encoding.\n'
1359
user_encoding = 'ascii'
1362
_cached_user_encoding = user_encoding
1364
return user_encoding
1367
def recv_all(socket, bytes):
1368
"""Receive an exact number of bytes.
1370
Regular Socket.recv() may return less than the requested number of bytes,
1371
dependning on what's in the OS buffer. MSG_WAITALL is not available
1372
on all platforms, but this should work everywhere. This will return
1373
less than the requested amount if the remote end closes.
1375
This isn't optimized and is intended mostly for use in testing.
1378
while len(b) < bytes:
1379
new = socket.recv(bytes - len(b))
1386
def send_all(socket, bytes):
1387
"""Send all bytes on a socket.
1389
Regular socket.sendall() can give socket error 10053 on Windows. This
1390
implementation sends no more than 64k at a time, which avoids this problem.
1393
for pos in xrange(0, len(bytes), chunk_size):
1394
socket.sendall(bytes[pos:pos+chunk_size])
1397
def dereference_path(path):
1398
"""Determine the real path to a file.
1400
All parent elements are dereferenced. But the file itself is not
1402
:param path: The original path. May be absolute or relative.
1403
:return: the real path *to* the file
1405
parent, base = os.path.split(path)
1406
# The pathjoin for '.' is a workaround for Python bug #1213894.
1407
# (initial path components aren't dereferenced)
1408
return pathjoin(realpath(pathjoin('.', parent)), base)
1411
def supports_mapi():
1412
"""Return True if we can use MAPI to launch a mail client."""
1413
return sys.platform == "win32"
1416
def resource_string(package, resource_name):
1417
"""Load a resource from a package and return it as a string.
1419
Note: Only packages that start with bzrlib are currently supported.
1421
This is designed to be a lightweight implementation of resource
1422
loading in a way which is API compatible with the same API from
1424
http://peak.telecommunity.com/DevCenter/PkgResources#basic-resource-access.
1425
If and when pkg_resources becomes a standard library, this routine
1428
# Check package name is within bzrlib
1429
if package == "bzrlib":
1430
resource_relpath = resource_name
1431
elif package.startswith("bzrlib."):
1432
package = package[len("bzrlib."):].replace('.', os.sep)
1433
resource_relpath = pathjoin(package, resource_name)
1435
raise errors.BzrError('resource package %s not in bzrlib' % package)
1437
# Map the resource to a file and read its contents
1438
base = dirname(bzrlib.__file__)
1439
if getattr(sys, 'frozen', None): # bzr.exe
1440
base = abspath(pathjoin(base, '..', '..'))
1441
filename = pathjoin(base, resource_relpath)
1442
return open(filename, 'rU').read()
added
removed
bzrlib/osutils.py
