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)
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)
111
"""Return a quoted filename filename
113
This previously used backslash quoting, but that works poorly on
115
# TODO: I'm not really sure this is the best format either.x
117
if _QUOTE_RE is None:
118
_QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')
120
if _QUOTE_RE.search(f):
126
_directory_kind = 'directory'
129
stat.S_IFDIR:_directory_kind,
130
stat.S_IFCHR:'chardev',
131
stat.S_IFBLK:'block',
134
stat.S_IFLNK:'symlink',
135
stat.S_IFSOCK:'socket',
139
def file_kind_from_stat_mode(stat_mode, _formats=_formats, _unknown='unknown'):
140
"""Generate a file kind from a stat mode. This is used in walkdirs.
142
Its performance is critical: Do not mutate without careful benchmarking.
145
return _formats[stat_mode & 0170000]
150
def file_kind(f, _lstat=os.lstat, _mapper=file_kind_from_stat_mode):
152
return _mapper(_lstat(f).st_mode)
154
if getattr(e, 'errno', None) == errno.ENOENT:
155
raise errors.NoSuchFile(f)
160
"""Return the current umask"""
161
# Assume that people aren't messing with the umask while running
162
# XXX: This is not thread safe, but there is no way to get the
163
# umask without setting it
171
_directory_kind: "/",
173
'tree-reference': '+',
177
def kind_marker(kind):
179
return _kind_marker_map[kind]
181
raise errors.BzrError('invalid file kind %r' % kind)
184
lexists = getattr(os.path, 'lexists', None)
188
stat = getattr(os, 'lstat', os.stat)
192
if e.errno == errno.ENOENT:
195
raise errors.BzrError("lstat/stat of (%r): %r" % (f, e))
198
def fancy_rename(old, new, rename_func, unlink_func):
199
"""A fancy rename, when you don't have atomic rename.
201
:param old: The old path, to rename from
202
:param new: The new path, to rename to
203
:param rename_func: The potentially non-atomic rename function
204
:param unlink_func: A way to delete the target file if the full rename succeeds
207
# sftp rename doesn't allow overwriting, so play tricks:
209
base = os.path.basename(new)
210
dirname = os.path.dirname(new)
211
tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))
212
tmp_name = pathjoin(dirname, tmp_name)
214
# Rename the file out of the way, but keep track if it didn't exist
215
# We don't want to grab just any exception
216
# something like EACCES should prevent us from continuing
217
# The downside is that the rename_func has to throw an exception
218
# with an errno = ENOENT, or NoSuchFile
221
rename_func(new, tmp_name)
222
except (errors.NoSuchFile,), e:
225
# RBC 20060103 abstraction leakage: the paramiko SFTP clients rename
226
# function raises an IOError with errno is None when a rename fails.
227
# This then gets caught here.
228
if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):
231
if (getattr(e, 'errno', None) is None
232
or e.errno not in (errno.ENOENT, errno.ENOTDIR)):
239
# This may throw an exception, in which case success will
241
rename_func(old, new)
245
# If the file used to exist, rename it back into place
246
# otherwise just delete it from the tmp location
248
unlink_func(tmp_name)
250
rename_func(tmp_name, new)
253
# In Python 2.4.2 and older, os.path.abspath and os.path.realpath
254
# choke on a Unicode string containing a relative path if
255
# os.getcwd() returns a non-sys.getdefaultencoding()-encoded
257
_fs_enc = sys.getfilesystemencoding() or 'utf-8'
258
def _posix_abspath(path):
259
# jam 20060426 rather than encoding to fsencoding
260
# copy posixpath.abspath, but use os.getcwdu instead
261
if not posixpath.isabs(path):
262
path = posixpath.join(getcwd(), path)
263
return posixpath.normpath(path)
266
def _posix_realpath(path):
267
return posixpath.realpath(path.encode(_fs_enc)).decode(_fs_enc)
270
def _win32_fixdrive(path):
271
"""Force drive letters to be consistent.
273
win32 is inconsistent whether it returns lower or upper case
274
and even if it was consistent the user might type the other
275
so we force it to uppercase
276
running python.exe under cmd.exe return capital C:\\
277
running win32 python inside a cygwin shell returns lowercase c:\\
279
drive, path = _nt_splitdrive(path)
280
return drive.upper() + path
283
def _win32_abspath(path):
284
# Real _nt_abspath doesn't have a problem with a unicode cwd
285
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
def _win32_realpath(path):
319
# Real _nt_realpath doesn't have a problem with a unicode cwd
320
return _win32_fixdrive(_nt_realpath(unicode(path)).replace('\\', '/'))
323
def _win32_pathjoin(*args):
324
return _nt_join(*args).replace('\\', '/')
327
def _win32_normpath(path):
328
return _win32_fixdrive(_nt_normpath(unicode(path)).replace('\\', '/'))
332
return _win32_fixdrive(os.getcwdu().replace('\\', '/'))
335
def _win32_mkdtemp(*args, **kwargs):
336
return _win32_fixdrive(tempfile.mkdtemp(*args, **kwargs).replace('\\', '/'))
339
def _win32_rename(old, new):
340
"""We expect to be able to atomically replace 'new' with old.
342
On win32, if new exists, it must be moved out of the way first,
346
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
348
if e.errno in (errno.EPERM, errno.EACCES, errno.EBUSY, errno.EINVAL):
349
# If we try to rename a non-existant file onto cwd, we get
350
# EPERM or EACCES instead of ENOENT, this will raise ENOENT
351
# if the old path doesn't exist, sometimes we get EACCES
352
# On Linux, we seem to get EBUSY, on Mac we get EINVAL
358
return unicodedata.normalize('NFKC', os.getcwdu())
361
# Default is to just use the python builtins, but these can be rebound on
362
# particular platforms.
363
abspath = _posix_abspath
364
realpath = _posix_realpath
365
pathjoin = os.path.join
366
normpath = os.path.normpath
369
dirname = os.path.dirname
370
basename = os.path.basename
371
split = os.path.split
372
splitext = os.path.splitext
373
# These were already imported into local scope
374
# mkdtemp = tempfile.mkdtemp
375
# rmtree = shutil.rmtree
377
MIN_ABS_PATHLENGTH = 1
380
if sys.platform == 'win32':
381
abspath = _win32_abspath
382
realpath = _win32_realpath
383
pathjoin = _win32_pathjoin
384
normpath = _win32_normpath
385
getcwd = _win32_getcwd
386
mkdtemp = _win32_mkdtemp
387
rename = _win32_rename
389
MIN_ABS_PATHLENGTH = 3
391
def _win32_delete_readonly(function, path, excinfo):
392
"""Error handler for shutil.rmtree function [for win32]
393
Helps to remove files and dirs marked as read-only.
395
exception = excinfo[1]
396
if function in (os.remove, os.rmdir) \
397
and isinstance(exception, OSError) \
398
and exception.errno == errno.EACCES:
404
def rmtree(path, ignore_errors=False, onerror=_win32_delete_readonly):
405
"""Replacer for shutil.rmtree: could remove readonly dirs/files"""
406
return shutil.rmtree(path, ignore_errors, onerror)
407
elif sys.platform == 'darwin':
411
def get_terminal_encoding():
412
"""Find the best encoding for printing to the screen.
414
This attempts to check both sys.stdout and sys.stdin to see
415
what encoding they are in, and if that fails it falls back to
416
bzrlib.user_encoding.
417
The problem is that on Windows, locale.getpreferredencoding()
418
is not the same encoding as that used by the console:
419
http://mail.python.org/pipermail/python-list/2003-May/162357.html
421
On my standard US Windows XP, the preferred encoding is
422
cp1252, but the console is cp437
424
output_encoding = getattr(sys.stdout, 'encoding', None)
425
if not output_encoding:
426
input_encoding = getattr(sys.stdin, 'encoding', None)
427
if not input_encoding:
428
output_encoding = bzrlib.user_encoding
429
mutter('encoding stdout as bzrlib.user_encoding %r', output_encoding)
431
output_encoding = input_encoding
432
mutter('encoding stdout as sys.stdin encoding %r', output_encoding)
434
mutter('encoding stdout as sys.stdout encoding %r', output_encoding)
435
if output_encoding == 'cp0':
436
# invalid encoding (cp0 means 'no codepage' on Windows)
437
output_encoding = bzrlib.user_encoding
438
mutter('cp0 is invalid encoding.'
439
' encoding stdout as bzrlib.user_encoding %r', output_encoding)
442
codecs.lookup(output_encoding)
444
sys.stderr.write('bzr: warning:'
445
' unknown terminal encoding %s.\n'
446
' Using encoding %s instead.\n'
447
% (output_encoding, bzrlib.user_encoding)
449
output_encoding = bzrlib.user_encoding
451
return output_encoding
454
def normalizepath(f):
455
if getattr(os.path, 'realpath', None) is not None:
459
[p,e] = os.path.split(f)
460
if e == "" or e == "." or e == "..":
463
return pathjoin(F(p), e)
467
"""Copy a file to a backup.
469
Backups are named in GNU-style, with a ~ suffix.
471
If the file is already a backup, it's not copied.
477
if has_symlinks() and os.path.islink(fn):
478
target = os.readlink(fn)
479
os.symlink(target, bfn)
487
outf = file(bfn, 'wb')
495
"""True if f is an accessible directory."""
497
return S_ISDIR(os.lstat(f)[ST_MODE])
503
"""True if f is a regular file."""
505
return S_ISREG(os.lstat(f)[ST_MODE])
510
"""True if f is a symlink."""
512
return S_ISLNK(os.lstat(f)[ST_MODE])
516
def is_inside(dir, fname):
517
"""True if fname is inside dir.
519
The parameters should typically be passed to osutils.normpath first, so
520
that . and .. and repeated slashes are eliminated, and the separators
521
are canonical for the platform.
523
The empty string as a dir name is taken as top-of-tree and matches
526
# XXX: Most callers of this can actually do something smarter by
527
# looking at the inventory
537
return fname.startswith(dir)
540
def is_inside_any(dir_list, fname):
541
"""True if fname is inside any of given dirs."""
542
for dirname in dir_list:
543
if is_inside(dirname, fname):
548
def is_inside_or_parent_of_any(dir_list, fname):
549
"""True if fname is a child or a parent of any of the given files."""
550
for dirname in dir_list:
551
if is_inside(dirname, fname) or is_inside(fname, dirname):
556
def pumpfile(fromfile, tofile):
557
"""Copy contents of one file to another.
559
:return: The number of bytes copied.
564
b = fromfile.read(BUFSIZE)
572
def file_iterator(input_file, readsize=32768):
574
b = input_file.read(readsize)
581
if getattr(f, 'tell', None) is not None:
593
def sha_file_by_name(fname):
594
"""Calculate the SHA1 of a file by reading the full text"""
596
f = os.open(fname, os.O_RDONLY)
599
b = os.read(f, 1<<16)
607
def sha_strings(strings, _factory=sha.new):
608
"""Return the sha-1 of concatenation of strings"""
610
map(s.update, strings)
614
def sha_string(f, _factory=sha.new):
615
return _factory(f).hexdigest()
618
def fingerprint_file(f):
620
return {'size': len(b),
621
'sha1': sha.new(b).hexdigest()}
624
def compare_files(a, b):
625
"""Returns true if equal in contents"""
636
def local_time_offset(t=None):
637
"""Return offset of local zone from GMT, either at present or at time t."""
640
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
641
return offset.days * 86400 + offset.seconds
644
def format_date(t, offset=0, timezone='original', date_fmt=None,
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.
656
if timezone == 'utc':
659
elif timezone == 'original':
662
tt = time.gmtime(t + offset)
663
elif timezone == 'local':
664
tt = time.localtime(t)
665
offset = local_time_offset(t)
667
raise errors.BzrError("unsupported timezone format %r" % timezone,
668
['options are "utc", "original", "local"'])
670
date_fmt = "%a %Y-%m-%d %H:%M:%S"
672
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
675
return (time.strftime(date_fmt, tt) + offset_str)
678
def compact_date(when):
679
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
"""Return size of given open file."""
733
return os.fstat(f.fileno())[ST_SIZE]
736
# Define rand_bytes based on platform.
738
# Python 2.4 and later have os.urandom,
739
# but it doesn't work on some arches
741
rand_bytes = os.urandom
742
except (NotImplementedError, AttributeError):
743
# If python doesn't have os.urandom, or it doesn't work,
744
# then try to first pull random data from /dev/urandom
746
rand_bytes = file('/dev/urandom', 'rb').read
747
# Otherwise, use this hack as a last resort
748
except (IOError, OSError):
749
# not well seeded, but better than nothing
754
s += chr(random.randint(0, 255))
759
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
761
"""Return a random string of num alphanumeric characters
763
The result only contains lowercase chars because it may be used on
764
case-insensitive filesystems.
767
for raw_byte in rand_bytes(num):
768
s += ALNUM[ord(raw_byte) % 36]
772
## TODO: We could later have path objects that remember their list
773
## decomposition (might be too tricksy though.)
776
"""Turn string into list of parts."""
777
assert isinstance(p, basestring)
779
# split on either delimiter because people might use either on
781
ps = re.split(r'[\\/]', p)
786
raise errors.BzrError("sorry, %r not allowed in path" % f)
787
elif (f == '.') or (f == ''):
794
assert isinstance(p, (list, tuple))
796
if (f == '..') or (f is None) or (f == ''):
797
raise errors.BzrError("sorry, %r not allowed in path" % f)
802
"""Split s into lines, but without removing the newline characters."""
803
lines = s.split('\n')
804
result = [line + '\n' for line in lines[:-1]]
806
result.append(lines[-1])
810
def hardlinks_good():
811
return sys.platform not in ('win32', 'cygwin', 'darwin')
814
def link_or_copy(src, dest):
815
"""Hardlink a file, or copy it if it can't be hardlinked."""
816
if not hardlinks_good():
817
shutil.copyfile(src, dest)
821
except (OSError, IOError), e:
822
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):
833
"""Delete a file or directory."""
834
if isdir(path): # Takes care of symlinks
841
if getattr(os, 'symlink', None) is not None:
847
def contains_whitespace(s):
848
"""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':
867
def contains_linebreaks(s):
868
"""True if there is any vertical whitespace in s."""
876
def relpath(base, path):
877
"""Return path relative to base, or raise exception.
879
The path may be either an absolute path or a path relative to the
880
current working directory.
882
os.path.commonprefix (python2.4) has a bad bug that it works just
883
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
887
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
888
' exceed the platform minimum length (which is %d)' %
895
while len(head) >= len(base):
898
head, tail = os.path.split(head)
902
raise errors.PathNotChild(rp, base)
910
def safe_unicode(unicode_or_utf8_string):
911
"""Coerce unicode_or_utf8_string into unicode.
913
If it is unicode, it is returned.
914
Otherwise it is decoded from utf-8. If a decoding error
915
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
916
as a BzrBadParameter exception.
918
if isinstance(unicode_or_utf8_string, unicode):
919
return unicode_or_utf8_string
921
return unicode_or_utf8_string.decode('utf8')
922
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)
991
_platform_normalizes_filenames = False
992
if sys.platform == 'darwin':
993
_platform_normalizes_filenames = True
996
def normalizes_filenames():
997
"""Return True if this platform normalizes unicode filenames.
999
Mac OSX does, Windows/Linux do not.
1001
return _platform_normalizes_filenames
1004
def _accessible_normalized_filename(path):
1005
"""Get the unicode normalized path, and if you can access the file.
1007
On platforms where the system normalizes filenames (Mac OSX),
1008
you can access a file by any path which will normalize correctly.
1009
On platforms where the system does not normalize filenames
1010
(Windows, Linux), you have to access a file by its exact path.
1012
Internally, bzr only supports NFC/NFKC normalization, since that is
1013
the standard for XML documents.
1015
So return the normalized path, and a flag indicating if the file
1016
can be accessed by that path.
1019
return unicodedata.normalize('NFKC', unicode(path)), True
1022
def _inaccessible_normalized_filename(path):
1023
__doc__ = _accessible_normalized_filename.__doc__
1025
normalized = unicodedata.normalize('NFKC', unicode(path))
1026
return normalized, normalized == path
1029
if _platform_normalizes_filenames:
1030
normalized_filename = _accessible_normalized_filename
1032
normalized_filename = _inaccessible_normalized_filename
1035
def terminal_width():
1036
"""Return estimated terminal width."""
1037
if sys.platform == 'win32':
1038
return win32utils.get_console_size()[0]
1041
import struct, fcntl, termios
1042
s = struct.pack('HHHH', 0, 0, 0, 0)
1043
x = fcntl.ioctl(1, termios.TIOCGWINSZ, s)
1044
width = struct.unpack('HHHH', x)[1]
1049
width = int(os.environ['COLUMNS'])
1058
def supports_executable():
1059
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
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
1097
def check_legal_path(path):
1098
"""Check whether the supplied path is legal.
1099
This is only required on Windows, so we don't test on other platforms
1102
if sys.platform != "win32":
1104
if _validWin32PathRE.match(path) is None:
1105
raise errors.IllegalPath(path)
1108
def walkdirs(top, prefix=""):
1109
"""Yield data about all the directories in a tree.
1111
This yields all the data about the contents of a directory at a time.
1112
After each directory has been yielded, if the caller has mutated the list
1113
to exclude some directories, they are then not descended into.
1115
The data yielded is of the form:
1116
((directory-relpath, directory-path-from-top),
1117
[(relpath, basename, kind, lstat, path-from-top), ...]),
1118
- directory-relpath is the relative path of the directory being returned
1119
with respect to top. prefix is prepended to this.
1120
- directory-path-from-root is the path including top for this directory.
1121
It is suitable for use with os functions.
1122
- relpath is the relative path within the subtree being walked.
1123
- basename is the basename of the path
1124
- kind is the kind of the file now. If unknown then the file is not
1125
present within the tree - but it may be recorded as versioned. See
1127
- lstat is the stat data *if* the file was statted.
1128
- planned, not implemented:
1129
path_from_tree_root is the path from the root of the tree.
1131
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1132
allows one to walk a subtree but get paths that are relative to a tree
1134
:return: an iterator over the dirs.
1136
#TODO there is a bit of a smell where the results of the directory-
1137
# summary in this, and the path from the root, may not agree
1138
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1139
# potentially confusing output. We should make this more robust - but
1140
# not at a speed cost. RBC 20060731
1142
_directory = _directory_kind
1143
_listdir = os.listdir
1144
_kind_from_mode = _formats.get
1145
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
1147
# 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)
1264
def copy_tree(from_path, to_path, handlers={}):
1265
"""Copy all of the entries in from_path into to_path.
1267
:param from_path: The base directory to copy.
1268
:param to_path: The target directory. If it does not exist, it will
1270
:param handlers: A dictionary of functions, which takes a source and
1271
destinations for files, directories, etc.
1272
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1273
'file', 'directory', and 'symlink' should always exist.
1274
If they are missing, they will be replaced with 'os.mkdir()',
1275
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1277
# Now, just copy the existing cached tree to the new location
1278
# We use a cheap trick here.
1279
# Absolute paths are prefixed with the first parameter
1280
# relative paths are prefixed with the second.
1281
# So we can get both the source and target returned
1282
# without any extra work.
1284
def copy_dir(source, dest):
1287
def copy_link(source, dest):
1288
"""Copy the contents of a symlink"""
1289
link_to = os.readlink(source)
1290
os.symlink(link_to, dest)
1292
real_handlers = {'file':shutil.copy2,
1293
'symlink':copy_link,
1294
'directory':copy_dir,
1296
real_handlers.update(handlers)
1298
if not os.path.exists(to_path):
1299
real_handlers['directory'](from_path, to_path)
1301
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1302
for relpath, name, kind, st, abspath in entries:
1303
real_handlers[kind](abspath, relpath)
1306
def path_prefix_key(path):
1307
"""Generate a prefix-order path key for path.
1309
This can be used to sort paths in the same way that walkdirs does.
1311
return (dirname(path) , path)
1314
def compare_paths_prefix_order(path_a, path_b):
1315
"""Compare path_a and path_b to generate the same order walkdirs uses."""
1316
key_a = path_prefix_key(path_a)
1317
key_b = path_prefix_key(path_b)
1318
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
