1
# Bazaar-NG -- distributed version control
3
# Copyright (C) 2005 by Canonical Ltd
5
# This program is free software; you can redistribute it and/or modify
6
# it under the terms of the GNU General Public License as published by
7
# the Free Software Foundation; either version 2 of the License, or
8
# (at your option) any later version.
10
# This program is distributed in the hope that it will be useful,
11
# but WITHOUT ANY WARRANTY; without even the implied warranty of
12
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13
# GNU General Public License for more details.
15
# You should have received a copy of the GNU General Public License
16
# along with this program; if not, write to the Free Software
17
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19
from shutil import copyfile
20
from stat import (S_ISREG, S_ISDIR, S_ISLNK, ST_MODE, ST_SIZE,
21
S_ISCHR, S_ISBLK, S_ISFIFO, S_ISSOCK)
22
from cStringIO import StringIO
36
from bzrlib.errors import (BzrError,
37
BzrBadParameterNotUnicode,
43
from bzrlib.trace import mutter
46
def make_readonly(filename):
47
"""Make a filename read-only."""
48
mod = os.stat(filename).st_mode
50
os.chmod(filename, mod)
53
def make_writable(filename):
54
mod = os.stat(filename).st_mode
56
os.chmod(filename, mod)
63
"""Return a quoted filename filename
65
This previously used backslash quoting, but that works poorly on
67
# TODO: I'm not really sure this is the best format either.x
70
_QUOTE_RE = re.compile(r'([^a-zA-Z0-9.,:/\\_~-])')
72
if _QUOTE_RE.search(f):
79
mode = os.lstat(f)[ST_MODE]
98
def kind_marker(kind):
101
elif kind == 'directory':
103
elif kind == 'symlink':
106
raise BzrError('invalid file kind %r' % kind)
109
if hasattr(os.path, 'lexists'):
110
return os.path.lexists(f)
112
if hasattr(os, 'lstat'):
118
if e.errno == errno.ENOENT:
121
raise BzrError("lstat/stat of (%r): %r" % (f, e))
123
def fancy_rename(old, new, rename_func, unlink_func):
124
"""A fancy rename, when you don't have atomic rename.
126
:param old: The old path, to rename from
127
:param new: The new path, to rename to
128
:param rename_func: The potentially non-atomic rename function
129
:param unlink_func: A way to delete the target file if the full rename succeeds
132
# sftp rename doesn't allow overwriting, so play tricks:
134
base = os.path.basename(new)
135
dirname = os.path.dirname(new)
136
tmp_name = u'tmp.%s.%.9f.%d.%s' % (base, time.time(), os.getpid(), rand_chars(10))
137
tmp_name = pathjoin(dirname, tmp_name)
139
# Rename the file out of the way, but keep track if it didn't exist
140
# We don't want to grab just any exception
141
# something like EACCES should prevent us from continuing
142
# The downside is that the rename_func has to throw an exception
143
# with an errno = ENOENT, or NoSuchFile
146
rename_func(new, tmp_name)
147
except (NoSuchFile,), e:
150
# RBC 20060103 abstraction leakage: the paramiko SFTP clients rename
151
# function raises an IOError with errno == None when a rename fails.
152
# This then gets caught here.
153
if e.errno not in (None, errno.ENOENT, errno.ENOTDIR):
156
if (not hasattr(e, 'errno')
157
or e.errno not in (errno.ENOENT, errno.ENOTDIR)):
164
# This may throw an exception, in which case success will
166
rename_func(old, new)
170
# If the file used to exist, rename it back into place
171
# otherwise just delete it from the tmp location
173
unlink_func(tmp_name)
175
rename_func(tmp_name, new)
178
def urlescape(relpath):
179
"""Escape relpath to be a valid url."""
180
if isinstance(relpath, unicode):
181
relpath = relpath.encode('utf-8')
182
# After quoting and encoding, the path should be perfectly
183
# safe as a plain ASCII string, str() just enforces this
184
return str(urllib.quote(relpath))
187
def urlunescape(url):
188
"""Unescape relpath from url format.
190
This returns a Unicode path from a URL
192
# jam 20060427 URLs are supposed to be ASCII only strings
193
# If they are passed in as unicode, urllib.unquote
194
# will return a UNICODE string, which actually contains
195
# utf-8 bytes. So we have to ensure that they are
196
# plain ASCII strings, or the final .decode will
197
# try to encode the UNICODE => ASCII, and then decode
199
unquoted = urllib.unquote(str(url))
201
unicode_path = unquoted.decode('utf-8')
202
except UnicodeError, e:
203
raise InvalidURL(url, e)
207
def _posix_local_path_to_url(path):
208
"""Convert a local path like ./foo into a URL like file:///path/to/foo
210
This also handles transforming escaping unicode characters, etc.
212
# importing directly from posixpath allows us to test this
213
# on non-posix platforms
214
from posixpath import normpath
215
return 'file://' + urlescape(normpath(_posix_abspath(path)))
218
def _posix_local_path_from_url(url):
219
"""Convert a url like file:///path/to/foo into /path/to/foo"""
220
if not url.startswith('file:///'):
221
raise InvalidURL(url, 'local urls must start with file:///')
222
# We only strip off 2 slashes
223
return urlunescape(url[len('file://'):])
226
def _win32_local_path_to_url(path):
227
"""Convert a local path like ./foo into a URL like file:///C|/path/to/foo
229
This also handles transforming escaping unicode characters, etc.
231
# importing directly from ntpath allows us to test this
232
# on non-win32 platforms
233
# TODO: jam 20060426 consider moving this import outside of the function
234
from ntpath import normpath
235
win32_path = normpath(abspath(path)).replace('\\', '/')
236
return 'file:///' + win32_path[0] + '|' + urlescape(win32_path[2:])
239
def _win32_local_path_from_url(url):
240
"""Convert a url like file:///C|/path/to/foo into C:/path/to/foo"""
241
if not url.startswith('file:///'):
242
raise InvalidURL(url, 'local urls must start with file:///')
243
# We strip off all 3 slashes
244
win32_url = url[len('file:///'):]
245
if (win32_url[0] not in 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
246
or win32_url[1] != '|'
247
or win32_url[2] != '/'):
248
raise InvalidURL(url, 'Win32 file urls start with file:///X|/, where X is a valid drive letter')
249
# TODO: jam 20060426, we could .upper() or .lower() the drive letter
250
# for better consistency.
251
return win32_url[0] + u':' + urlunescape(win32_url[2:])
254
# In Python 2.4.2 and older, os.path.abspath and os.path.realpath
255
# choke on a Unicode string containing a relative path if
256
# os.getcwd() returns a non-sys.getdefaultencoding()-encoded
258
_fs_enc = sys.getfilesystemencoding()
259
def _posix_abspath(path):
260
return os.path.abspath(path.encode(_fs_enc)).decode(_fs_enc)
263
def _posix_realpath(path):
264
return os.path.realpath(path.encode(_fs_enc)).decode(_fs_enc)
267
def _win32_abspath(path):
268
return _posix_abspath(path).replace('\\', '/')
271
def _win32_realpath(path):
272
return _posix_realpath(path).replace('\\', '/')
275
def _win32_pathjoin(*args):
276
return os.path.join(*args).replace('\\', '/')
279
def _win32_normpath(path):
280
return os.path.normpath(path).replace('\\', '/')
284
return os.getcwdu().replace('\\', '/')
287
def _win32_mkdtemp(*args, **kwargs):
288
return tempfile.mkdtemp(*args, **kwargs).replace('\\', '/')
291
def _win32_rename(old, new):
292
fancy_rename(old, new, rename_func=os.rename, unlink_func=os.unlink)
295
# Default is to just use the python builtins
296
abspath = _posix_abspath
297
realpath = _posix_realpath
298
pathjoin = os.path.join
299
normpath = os.path.normpath
301
mkdtemp = tempfile.mkdtemp
303
dirname = os.path.dirname
304
basename = os.path.basename
305
local_path_to_url = _posix_local_path_to_url
306
local_path_from_url = _posix_local_path_from_url
308
MIN_ABS_PATHLENGTH = 1
309
MIN_ABS_URLPATHLENGTH = len('file:///')
313
if sys.platform == 'win32':
314
abspath = _win32_abspath
315
realpath = _win32_realpath
316
pathjoin = _win32_pathjoin
317
normpath = _win32_normpath
318
getcwd = _win32_getcwd
319
mkdtemp = _win32_mkdtemp
320
rename = _win32_rename
322
local_path_to_url = _win32_local_path_to_url
323
local_path_from_url = _win32_local_path_from_url
325
MIN_ABS_PATHLENGTH = 3
326
MIN_ABS_URLPATHLENGTH = len('file:///C|/')
328
def normalizepath(f):
329
if hasattr(os.path, 'realpath'):
333
[p,e] = os.path.split(f)
334
if e == "" or e == "." or e == "..":
337
return pathjoin(F(p), e)
341
"""Copy a file to a backup.
343
Backups are named in GNU-style, with a ~ suffix.
345
If the file is already a backup, it's not copied.
351
if has_symlinks() and os.path.islink(fn):
352
target = os.readlink(fn)
353
os.symlink(target, bfn)
361
outf = file(bfn, 'wb')
369
"""True if f is an accessible directory."""
371
return S_ISDIR(os.lstat(f)[ST_MODE])
377
"""True if f is a regular file."""
379
return S_ISREG(os.lstat(f)[ST_MODE])
384
"""True if f is a symlink."""
386
return S_ISLNK(os.lstat(f)[ST_MODE])
390
def is_inside(dir, fname):
391
"""True if fname is inside dir.
393
The parameters should typically be passed to osutils.normpath first, so
394
that . and .. and repeated slashes are eliminated, and the separators
395
are canonical for the platform.
397
The empty string as a dir name is taken as top-of-tree and matches
400
>>> is_inside('src', pathjoin('src', 'foo.c'))
402
>>> is_inside('src', 'srccontrol')
404
>>> is_inside('src', pathjoin('src', 'a', 'a', 'a', 'foo.c'))
406
>>> is_inside('foo.c', 'foo.c')
408
>>> is_inside('foo.c', '')
410
>>> is_inside('', 'foo.c')
413
# XXX: Most callers of this can actually do something smarter by
414
# looking at the inventory
424
return fname.startswith(dir)
427
def is_inside_any(dir_list, fname):
428
"""True if fname is inside any of given dirs."""
429
for dirname in dir_list:
430
if is_inside(dirname, fname):
436
def pumpfile(fromfile, tofile):
437
"""Copy contents of one file to another."""
440
b = fromfile.read(BUFSIZE)
446
def file_iterator(input_file, readsize=32768):
448
b = input_file.read(readsize)
455
if hasattr(f, 'tell'):
468
def sha_strings(strings):
469
"""Return the sha-1 of concatenation of strings"""
471
map(s.update, strings)
481
def fingerprint_file(f):
486
return {'size': size,
487
'sha1': s.hexdigest()}
490
def compare_files(a, b):
491
"""Returns true if equal in contents"""
502
def local_time_offset(t=None):
503
"""Return offset of local zone from GMT, either at present or at time t."""
504
# python2.3 localtime() can't take None
508
if time.localtime(t).tm_isdst and time.daylight:
511
return -time.timezone
514
def format_date(t, offset=0, timezone='original', date_fmt=None,
516
## TODO: Perhaps a global option to use either universal or local time?
517
## Or perhaps just let people set $TZ?
518
assert isinstance(t, float)
520
if timezone == 'utc':
523
elif timezone == 'original':
526
tt = time.gmtime(t + offset)
527
elif timezone == 'local':
528
tt = time.localtime(t)
529
offset = local_time_offset(t)
531
raise BzrError("unsupported timezone format %r" % timezone,
532
['options are "utc", "original", "local"'])
534
date_fmt = "%a %Y-%m-%d %H:%M:%S"
536
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
539
return (time.strftime(date_fmt, tt) + offset_str)
542
def compact_date(when):
543
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
548
"""Return size of given open file."""
549
return os.fstat(f.fileno())[ST_SIZE]
552
# Define rand_bytes based on platform.
554
# Python 2.4 and later have os.urandom,
555
# but it doesn't work on some arches
557
rand_bytes = os.urandom
558
except (NotImplementedError, AttributeError):
559
# If python doesn't have os.urandom, or it doesn't work,
560
# then try to first pull random data from /dev/urandom
561
if os.path.exists("/dev/urandom"):
562
rand_bytes = file('/dev/urandom', 'rb').read
563
# Otherwise, use this hack as a last resort
565
# not well seeded, but better than nothing
570
s += chr(random.randint(0, 255))
575
ALNUM = '0123456789abcdefghijklmnopqrstuvwxyz'
577
"""Return a random string of num alphanumeric characters
579
The result only contains lowercase chars because it may be used on
580
case-insensitive filesystems.
583
for raw_byte in rand_bytes(num):
584
s += ALNUM[ord(raw_byte) % 36]
588
## TODO: We could later have path objects that remember their list
589
## decomposition (might be too tricksy though.)
592
"""Turn string into list of parts.
598
>>> splitpath('a/./b')
600
>>> splitpath('a/.b')
602
>>> splitpath('a/../b')
603
Traceback (most recent call last):
605
BzrError: sorry, '..' not allowed in path
607
assert isinstance(p, types.StringTypes)
609
# split on either delimiter because people might use either on
611
ps = re.split(r'[\\/]', p)
616
raise BzrError("sorry, %r not allowed in path" % f)
617
elif (f == '.') or (f == ''):
624
assert isinstance(p, list)
626
if (f == '..') or (f == None) or (f == ''):
627
raise BzrError("sorry, %r not allowed in path" % f)
631
def appendpath(p1, p2):
635
return pathjoin(p1, p2)
639
"""Split s into lines, but without removing the newline characters."""
640
lines = s.split('\n')
641
result = [line + '\n' for line in lines[:-1]]
643
result.append(lines[-1])
647
def hardlinks_good():
648
return sys.platform not in ('win32', 'cygwin', 'darwin')
651
def link_or_copy(src, dest):
652
"""Hardlink a file, or copy it if it can't be hardlinked."""
653
if not hardlinks_good():
658
except (OSError, IOError), e:
659
if e.errno != errno.EXDEV:
663
def delete_any(full_path):
664
"""Delete a file or directory."""
668
# We may be renaming a dangling inventory id
669
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
675
if hasattr(os, 'symlink'):
681
def contains_whitespace(s):
682
"""True if there are any whitespace characters in s."""
683
for ch in string.whitespace:
690
def contains_linebreaks(s):
691
"""True if there is any vertical whitespace in s."""
699
def relpath(base, path):
700
"""Return path relative to base, or raise exception.
702
The path may be either an absolute path or a path relative to the
703
current working directory.
705
os.path.commonprefix (python2.4) has a bad bug that it works just
706
on string prefixes, assuming that '/u' is a prefix of '/u2'. This
710
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
711
' exceed the platform minimum length (which is %d)' %
718
while len(head) >= len(base):
721
head, tail = os.path.split(head)
725
raise PathNotChild(rp, base)
733
def urlrelpath(base, path):
734
"""Compute just the relative sub-portion of a url
736
This assumes that both paths are already fully specified file:// URLs.
738
assert len(base) >= MIN_ABS_URLPATHLENGTH, ('Length of base must be equal or'
739
' exceed the platform minimum url length (which is %d)' %
740
MIN_ABS_URLPATHLENGTH)
742
base = local_path_from_url(base)
743
path = local_path_from_url(path)
744
return urlescape(relpath(base, path))
747
def safe_unicode(unicode_or_utf8_string):
748
"""Coerce unicode_or_utf8_string into unicode.
750
If it is unicode, it is returned.
751
Otherwise it is decoded from utf-8. If a decoding error
752
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
753
as a BzrBadParameter exception.
755
if isinstance(unicode_or_utf8_string, unicode):
756
return unicode_or_utf8_string
758
return unicode_or_utf8_string.decode('utf8')
759
except UnicodeDecodeError:
760
raise BzrBadParameterNotUnicode(unicode_or_utf8_string)
763
_platform_normalizes_filenames = False
764
if sys.platform == 'darwin':
765
_platform_normalizes_filenames = True
768
def normalizes_filenames():
769
"""Return True if this platform normalizes unicode filenames.
771
Mac OSX does, Windows/Linux do not.
773
return _platform_normalizes_filenames
776
if _platform_normalizes_filenames:
777
def unicode_filename(path):
778
"""Make sure 'path' is a properly normalized filename.
780
On platforms where the system normalizes filenames (Mac OSX),
781
you can access a file by any path which will normalize
783
Internally, bzr only supports NFC/NFKC normalization, since
784
that is the standard for XML documents.
785
So we return an normalized path, and indicate this has been
788
:return: (path, is_normalized) Return a path which can
789
access the file, and whether or not this path is
792
return unicodedata.normalize('NFKC', path), True
794
def unicode_filename(path):
795
"""Make sure 'path' is a properly normalized filename.
797
On platforms where the system does not normalize filenames
798
(Windows, Linux), you have to access a file by its exact path.
799
Internally, bzr only supports NFC/NFKC normalization, since
800
that is the standard for XML documents.
801
So we return the original path, and indicate if this is
804
:return: (path, is_normalized) Return a path which can
805
access the file, and whether or not this path is
808
return path, unicodedata.normalize('NFKC', path) == path
811
def terminal_width():
812
"""Return estimated terminal width."""
814
# TODO: Do something smart on Windows?
816
# TODO: Is there anything that gets a better update when the window
817
# is resized while the program is running? We could use the Python termcap
820
return int(os.environ['COLUMNS'])
821
except (IndexError, KeyError, ValueError):
824
def supports_executable():
825
return sys.platform != "win32"
828
def strip_url_trailing_slash(path):
829
"""Strip trailing slash, except for root paths.
830
The definition of 'root path' is platform-dependent.
832
assert path.startswith('file:///'), \
833
'strip_url_trailing_slash expects file:// urls (%s)' % path
834
if len(path) != MIN_ABS_URLPATHLENGTH and path[-1] == '/':
840
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
843
def check_legal_path(path):
844
"""Check whether the supplied path is legal.
845
This is only required on Windows, so we don't test on other platforms
848
if sys.platform != "win32":
850
if _validWin32PathRE.match(path) is None:
851
raise IllegalPath(path)
added
removed
bzrlib/osutils.py
