526
482
for dirname in dir_list:
527
483
if is_inside(dirname, fname) or is_inside(fname, dirname):
532
def pumpfile(from_file, to_file, read_length=-1, buff_size=32768,
533
report_activity=None, direction='read'):
534
"""Copy contents of one file to another.
536
The read_length can either be -1 to read to end-of-file (EOF) or
537
it can specify the maximum number of bytes to read.
539
The buff_size represents the maximum size for each read operation
540
performed on from_file.
542
:param report_activity: Call this as bytes are read, see
543
Transport._report_activity
544
:param direction: Will be passed to report_activity
546
:return: The number of bytes copied.
550
# read specified number of bytes
552
while read_length > 0:
553
num_bytes_to_read = min(read_length, buff_size)
555
block = from_file.read(num_bytes_to_read)
559
if report_activity is not None:
560
report_activity(len(block), direction)
563
actual_bytes_read = len(block)
564
read_length -= actual_bytes_read
565
length += actual_bytes_read
569
block = from_file.read(buff_size)
573
if report_activity is not None:
574
report_activity(len(block), direction)
580
def pump_string_file(bytes, file_handle, segment_size=None):
581
"""Write bytes to file_handle in many smaller writes.
583
:param bytes: The string to write.
584
:param file_handle: The file to write to.
586
# Write data in chunks rather than all at once, because very large
587
# writes fail on some platforms (e.g. Windows with SMB mounted
590
segment_size = 5242880 # 5MB
591
segments = range(len(bytes) / segment_size + 1)
592
write = file_handle.write
593
for segment_index in segments:
594
segment = buffer(bytes, segment_index * segment_size, segment_size)
489
def pumpfile(fromfile, tofile):
490
"""Copy contents of one file to another."""
493
b = fromfile.read(BUFSIZE)
598
499
def file_iterator(input_file, readsize=32768):
682
555
def local_time_offset(t=None):
683
556
"""Return offset of local zone from GMT, either at present or at time t."""
557
# python2.3 localtime() can't take None
686
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
687
return offset.days * 86400 + offset.seconds
689
weekdays = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
691
def format_date(t, offset=0, timezone='original', date_fmt=None,
561
if time.localtime(t).tm_isdst and time.daylight:
564
return -time.timezone
567
def format_date(t, offset=0, timezone='original', date_fmt=None,
692
568
show_offset=True):
693
"""Return a formatted date string.
695
:param t: Seconds since the epoch.
696
:param offset: Timezone offset in seconds east of utc.
697
:param timezone: How to display the time: 'utc', 'original' for the
698
timezone specified by offset, or 'local' for the process's current
700
:param date_fmt: strftime format.
701
:param show_offset: Whether to append the timezone.
703
(date_fmt, tt, offset_str) = \
704
_format_date(t, offset, timezone, date_fmt, show_offset)
705
date_fmt = date_fmt.replace('%a', weekdays[tt[6]])
706
date_str = time.strftime(date_fmt, tt)
707
return date_str + offset_str
709
def format_local_date(t, offset=0, timezone='original', date_fmt=None,
711
"""Return an unicode date string formatted according to the current locale.
713
:param t: Seconds since the epoch.
714
:param offset: Timezone offset in seconds east of utc.
715
:param timezone: How to display the time: 'utc', 'original' for the
716
timezone specified by offset, or 'local' for the process's current
718
:param date_fmt: strftime format.
719
:param show_offset: Whether to append the timezone.
721
(date_fmt, tt, offset_str) = \
722
_format_date(t, offset, timezone, date_fmt, show_offset)
723
date_str = time.strftime(date_fmt, tt)
724
if not isinstance(date_str, unicode):
725
date_str = date_str.decode(get_user_encoding(), 'replace')
726
return date_str + offset_str
728
def _format_date(t, offset, timezone, date_fmt, show_offset):
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)
729
573
if timezone == 'utc':
730
574
tt = time.gmtime(t)
732
576
elif timezone == 'original':
735
579
tt = time.gmtime(t + offset)
736
580
elif timezone == 'local':
737
581
tt = time.localtime(t)
738
582
offset = local_time_offset(t)
740
raise errors.UnsupportedTimezoneFormat(timezone)
584
raise BzrError("unsupported timezone format %r" % timezone,
585
['options are "utc", "original", "local"'])
741
586
if date_fmt is None:
742
587
date_fmt = "%a %Y-%m-%d %H:%M:%S"
744
589
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
747
return (date_fmt, tt, offset_str)
592
return (time.strftime(date_fmt, tt) + offset_str)
750
595
def compact_date(when):
751
596
return time.strftime('%Y%m%d%H%M%S', time.gmtime(when))
754
def format_delta(delta):
755
"""Get a nice looking string for a time delta.
757
:param delta: The time difference in seconds, can be positive or negative.
758
positive indicates time in the past, negative indicates time in the
759
future. (usually time.time() - stored_time)
760
:return: String formatted to show approximate resolution
766
direction = 'in the future'
770
if seconds < 90: # print seconds up to 90 seconds
772
return '%d second %s' % (seconds, direction,)
774
return '%d seconds %s' % (seconds, direction)
776
minutes = int(seconds / 60)
777
seconds -= 60 * minutes
782
if minutes < 90: # print minutes, seconds up to 90 minutes
784
return '%d minute, %d second%s %s' % (
785
minutes, seconds, plural_seconds, direction)
787
return '%d minutes, %d second%s %s' % (
788
minutes, seconds, plural_seconds, direction)
790
hours = int(minutes / 60)
791
minutes -= 60 * hours
798
return '%d hour, %d minute%s %s' % (hours, minutes,
799
plural_minutes, direction)
800
return '%d hours, %d minute%s %s' % (hours, minutes,
801
plural_minutes, direction)
804
601
"""Return size of given open file."""
917
705
def link_or_copy(src, dest):
918
706
"""Hardlink a file, or copy it if it can't be hardlinked."""
919
707
if not hardlinks_good():
920
shutil.copyfile(src, dest)
923
711
os.link(src, dest)
924
712
except (OSError, IOError), e:
925
713
if e.errno != errno.EXDEV:
927
shutil.copyfile(src, dest)
930
def delete_any(path):
931
"""Delete a file, symlink or directory.
933
Will delete even if readonly.
717
def delete_any(full_path):
718
"""Delete a file or directory."""
936
_delete_file_or_dir(path)
937
except (OSError, IOError), e:
938
if e.errno in (errno.EPERM, errno.EACCES):
939
# make writable and try again
942
except (OSError, IOError):
944
_delete_file_or_dir(path)
722
# We may be renaming a dangling inventory id
723
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
949
def _delete_file_or_dir(path):
950
# Look Before You Leap (LBYL) is appropriate here instead of Easier to Ask for
951
# Forgiveness than Permission (EAFP) because:
952
# - root can damage a solaris file system by using unlink,
953
# - unlink raises different exceptions on different OSes (linux: EISDIR, win32:
954
# EACCES, OSX: EPERM) when invoked on a directory.
955
if isdir(path): # Takes care of symlinks
961
728
def has_symlinks():
962
if getattr(os, 'symlink', None) is not None:
969
if getattr(os, 'link', None) is not None:
975
def host_os_dereferences_symlinks():
976
return (has_symlinks()
977
and sys.platform not in ('cygwin', 'win32'))
980
def readlink(abspath):
981
"""Return a string representing the path to which the symbolic link points.
983
:param abspath: The link absolute unicode path.
985
This his guaranteed to return the symbolic link in unicode in all python
988
link = abspath.encode(_fs_enc)
989
target = os.readlink(link)
990
target = target.decode(_fs_enc)
729
if hasattr(os, 'symlink'):
994
735
def contains_whitespace(s):
995
736
"""True if there are any whitespace characters in s."""
996
# string.whitespace can include '\xa0' in certain locales, because it is
997
# considered "non-breaking-space" as part of ISO-8859-1. But it
998
# 1) Isn't a breaking whitespace
999
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
1001
# 3) '\xa0' isn't unicode safe since it is >128.
1003
# This should *not* be a unicode set of characters in case the source
1004
# string is not a Unicode string. We can auto-up-cast the characters since
1005
# they are ascii, but we don't want to auto-up-cast the string in case it
1007
for ch in ' \t\n\r\v\f':
737
for ch in string.whitespace:
1031
761
avoids that problem.
1034
if len(base) < MIN_ABS_PATHLENGTH:
1035
# must have space for e.g. a drive letter
1036
raise ValueError('%r is too short to calculate a relative path'
764
assert len(base) >= MIN_ABS_PATHLENGTH, ('Length of base must be equal or'
765
' exceed the platform minimum length (which is %d)' %
1039
768
rp = abspath(path)
1044
if len(head) <= len(base) and head != base:
1045
raise errors.PathNotChild(rp, base)
772
while len(head) >= len(base):
1046
773
if head == base:
1048
head, tail = split(head)
775
head, tail = os.path.split(head)
779
raise PathNotChild(rp, base)
1053
return pathjoin(*reversed(s))
1058
def _cicp_canonical_relpath(base, path):
1059
"""Return the canonical path relative to base.
1061
Like relpath, but on case-insensitive-case-preserving file-systems, this
1062
will return the relpath as stored on the file-system rather than in the
1063
case specified in the input string, for all existing portions of the path.
1065
This will cause O(N) behaviour if called for every path in a tree; if you
1066
have a number of paths to convert, you should use canonical_relpaths().
1068
# TODO: it should be possible to optimize this for Windows by using the
1069
# win32 API FindFiles function to look for the specified name - but using
1070
# os.listdir() still gives us the correct, platform agnostic semantics in
1073
rel = relpath(base, path)
1074
# '.' will have been turned into ''
1078
abs_base = abspath(base)
1080
_listdir = os.listdir
1082
# use an explicit iterator so we can easily consume the rest on early exit.
1083
bit_iter = iter(rel.split('/'))
1084
for bit in bit_iter:
1086
for look in _listdir(current):
1087
if lbit == look.lower():
1088
current = pathjoin(current, look)
1091
# got to the end, nothing matched, so we just return the
1092
# non-existing bits as they were specified (the filename may be
1093
# the target of a move, for example).
1094
current = pathjoin(current, bit, *list(bit_iter))
1096
return current[len(abs_base)+1:]
1098
# XXX - TODO - we need better detection/integration of case-insensitive
1099
# file-systems; Linux often sees FAT32 devices (or NFS-mounted OSX
1100
# filesystems), for example, so could probably benefit from the same basic
1101
# support there. For now though, only Windows and OSX get that support, and
1102
# they get it for *all* file-systems!
1103
if sys.platform in ('win32', 'darwin'):
1104
canonical_relpath = _cicp_canonical_relpath
1106
canonical_relpath = relpath
1108
def canonical_relpaths(base, paths):
1109
"""Create an iterable to canonicalize a sequence of relative paths.
1111
The intent is for this implementation to use a cache, vastly speeding
1112
up multiple transformations in the same directory.
1114
# but for now, we haven't optimized...
1115
return [canonical_relpath(base, p) for p in paths]
1117
787
def safe_unicode(unicode_or_utf8_string):
1118
788
"""Coerce unicode_or_utf8_string into unicode.
1120
790
If it is unicode, it is returned.
1121
Otherwise it is decoded from utf-8. If decoding fails, the exception is
1122
wrapped in a BzrBadParameterNotUnicode exception.
791
Otherwise it is decoded from utf-8. If a decoding error
792
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
793
as a BzrBadParameter exception.
1124
795
if isinstance(unicode_or_utf8_string, unicode):
1125
796
return unicode_or_utf8_string
1127
798
return unicode_or_utf8_string.decode('utf8')
1128
799
except UnicodeDecodeError:
1129
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
1132
def safe_utf8(unicode_or_utf8_string):
1133
"""Coerce unicode_or_utf8_string to a utf8 string.
1135
If it is a str, it is returned.
1136
If it is Unicode, it is encoded into a utf-8 string.
1138
if isinstance(unicode_or_utf8_string, str):
1139
# TODO: jam 20070209 This is overkill, and probably has an impact on
1140
# performance if we are dealing with lots of apis that want a
1143
# Make sure it is a valid utf-8 string
1144
unicode_or_utf8_string.decode('utf-8')
1145
except UnicodeDecodeError:
1146
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
1147
return unicode_or_utf8_string
1148
return unicode_or_utf8_string.encode('utf-8')
1151
_revision_id_warning = ('Unicode revision ids were deprecated in bzr 0.15.'
1152
' Revision id generators should be creating utf8'
1156
def safe_revision_id(unicode_or_utf8_string, warn=True):
1157
"""Revision ids should now be utf8, but at one point they were unicode.
1159
:param unicode_or_utf8_string: A possibly Unicode revision_id. (can also be
1161
:param warn: Functions that are sanitizing user data can set warn=False
1162
:return: None or a utf8 revision id.
1164
if (unicode_or_utf8_string is None
1165
or unicode_or_utf8_string.__class__ == str):
1166
return unicode_or_utf8_string
1168
symbol_versioning.warn(_revision_id_warning, DeprecationWarning,
1170
return cache_utf8.encode(unicode_or_utf8_string)
1173
_file_id_warning = ('Unicode file ids were deprecated in bzr 0.15. File id'
1174
' generators should be creating utf8 file ids.')
1177
def safe_file_id(unicode_or_utf8_string, warn=True):
1178
"""File ids should now be utf8, but at one point they were unicode.
1180
This is the same as safe_utf8, except it uses the cached encode functions
1181
to save a little bit of performance.
1183
:param unicode_or_utf8_string: A possibly Unicode file_id. (can also be
1185
:param warn: Functions that are sanitizing user data can set warn=False
1186
:return: None or a utf8 file id.
1188
if (unicode_or_utf8_string is None
1189
or unicode_or_utf8_string.__class__ == str):
1190
return unicode_or_utf8_string
1192
symbol_versioning.warn(_file_id_warning, DeprecationWarning,
1194
return cache_utf8.encode(unicode_or_utf8_string)
800
raise BzrBadParameterNotUnicode(unicode_or_utf8_string)
1197
803
_platform_normalizes_filenames = False
1264
870
def supports_executable():
1265
871
return sys.platform != "win32"
1268
def supports_posix_readonly():
1269
"""Return True if 'readonly' has POSIX semantics, False otherwise.
1271
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
1272
directory controls creation/deletion, etc.
1274
And under win32, readonly means that the directory itself cannot be
1275
deleted. The contents of a readonly directory can be changed, unlike POSIX
1276
where files in readonly directories cannot be added, deleted or renamed.
1278
return sys.platform != "win32"
1281
def set_or_unset_env(env_variable, value):
1282
"""Modify the environment, setting or removing the env_variable.
1284
:param env_variable: The environment variable in question
1285
:param value: The value to set the environment to. If None, then
1286
the variable will be removed.
1287
:return: The original value of the environment variable.
1289
orig_val = os.environ.get(env_variable)
1291
if orig_val is not None:
1292
del os.environ[env_variable]
1294
if isinstance(value, unicode):
1295
value = value.encode(get_user_encoding())
1296
os.environ[env_variable] = value
1300
874
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
1303
877
def check_legal_path(path):
1304
"""Check whether the supplied path is legal.
878
"""Check whether the supplied path is legal.
1305
879
This is only required on Windows, so we don't test on other platforms
1308
882
if sys.platform != "win32":
1310
884
if _validWin32PathRE.match(path) is None:
1311
raise errors.IllegalPath(path)
1314
_WIN32_ERROR_DIRECTORY = 267 # Similar to errno.ENOTDIR
1316
def _is_error_enotdir(e):
1317
"""Check if this exception represents ENOTDIR.
1319
Unfortunately, python is very inconsistent about the exception
1320
here. The cases are:
1321
1) Linux, Mac OSX all versions seem to set errno == ENOTDIR
1322
2) Windows, Python2.4, uses errno == ERROR_DIRECTORY (267)
1323
which is the windows error code.
1324
3) Windows, Python2.5 uses errno == EINVAL and
1325
winerror == ERROR_DIRECTORY
1327
:param e: An Exception object (expected to be OSError with an errno
1328
attribute, but we should be able to cope with anything)
1329
:return: True if this represents an ENOTDIR error. False otherwise.
1331
en = getattr(e, 'errno', None)
1332
if (en == errno.ENOTDIR
1333
or (sys.platform == 'win32'
1334
and (en == _WIN32_ERROR_DIRECTORY
1335
or (en == errno.EINVAL
1336
and getattr(e, 'winerror', None) == _WIN32_ERROR_DIRECTORY)
885
raise IllegalPath(path)
1342
888
def walkdirs(top, prefix=""):
1343
889
"""Yield data about all the directories in a tree.
1345
891
This yields all the data about the contents of a directory at a time.
1346
892
After each directory has been yielded, if the caller has mutated the list
1347
893
to exclude some directories, they are then not descended into.
1349
895
The data yielded is of the form:
1350
896
((directory-relpath, directory-path-from-top),
1351
[(relpath, basename, kind, lstat, path-from-top), ...]),
897
[(relpath, basename, kind, lstat), ...]),
1352
898
- directory-relpath is the relative path of the directory being returned
1353
899
with respect to top. prefix is prepended to this.
1354
- directory-path-from-root is the path including top for this directory.
900
- directory-path-from-root is the path including top for this directory.
1355
901
It is suitable for use with os functions.
1356
902
- relpath is the relative path within the subtree being walked.
1357
903
- basename is the basename of the path
1359
905
present within the tree - but it may be recorded as versioned. See
1361
907
- lstat is the stat data *if* the file was statted.
1362
- planned, not implemented:
908
- planned, not implemented:
1363
909
path_from_tree_root is the path from the root of the tree.
1365
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
911
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1366
912
allows one to walk a subtree but get paths that are relative to a tree
1367
913
rooted higher up.
1368
914
:return: an iterator over the dirs.
1370
916
#TODO there is a bit of a smell where the results of the directory-
1371
# summary in this, and the path from the root, may not agree
917
# summary in this, and the path from the root, may not agree
1372
918
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1373
919
# potentially confusing output. We should make this more robust - but
1374
920
# not at a speed cost. RBC 20060731
1376
923
_directory = _directory_kind
1377
_listdir = os.listdir
1378
_kind_from_mode = file_kind_from_stat_mode
1379
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
925
pending = [(prefix, "", _directory, None, top)]
928
currentdir = pending.pop()
1381
929
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1382
relroot, _, _, _, top = pending.pop()
1384
relprefix = relroot + u'/'
1387
top_slash = top + u'/'
1390
append = dirblock.append
1392
names = sorted(_listdir(top))
1394
if not _is_error_enotdir(e):
1398
abspath = top_slash + name
1399
statvalue = _lstat(abspath)
1400
kind = _kind_from_mode(statvalue.st_mode)
1401
append((relprefix + name, name, kind, statvalue, abspath))
1402
yield (relroot, top), dirblock
1404
# push the user specified dirs from dirblock
1405
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1408
class DirReader(object):
1409
"""An interface for reading directories."""
1411
def top_prefix_to_starting_dir(self, top, prefix=""):
1412
"""Converts top and prefix to a starting dir entry
1414
:param top: A utf8 path
1415
:param prefix: An optional utf8 path to prefix output relative paths
1417
:return: A tuple starting with prefix, and ending with the native
1420
raise NotImplementedError(self.top_prefix_to_starting_dir)
1422
def read_dir(self, prefix, top):
1423
"""Read a specific dir.
1425
:param prefix: A utf8 prefix to be preprended to the path basenames.
1426
:param top: A natively encoded path to read.
1427
:return: A list of the directories contents. Each item contains:
1428
(utf8_relpath, utf8_name, kind, lstatvalue, native_abspath)
1430
raise NotImplementedError(self.read_dir)
1433
_selected_dir_reader = None
1436
def _walkdirs_utf8(top, prefix=""):
1437
"""Yield data about all the directories in a tree.
1439
This yields the same information as walkdirs() only each entry is yielded
1440
in utf-8. On platforms which have a filesystem encoding of utf8 the paths
1441
are returned as exact byte-strings.
1443
:return: yields a tuple of (dir_info, [file_info])
1444
dir_info is (utf8_relpath, path-from-top)
1445
file_info is (utf8_relpath, utf8_name, kind, lstat, path-from-top)
1446
if top is an absolute path, path-from-top is also an absolute path.
1447
path-from-top might be unicode or utf8, but it is the correct path to
1448
pass to os functions to affect the file in question. (such as os.lstat)
1450
global _selected_dir_reader
1451
if _selected_dir_reader is None:
1452
fs_encoding = _fs_enc.upper()
1453
if sys.platform == "win32" and win32utils.winver == 'Windows NT':
1454
# Win98 doesn't have unicode apis like FindFirstFileW
1455
# TODO: We possibly could support Win98 by falling back to the
1456
# original FindFirstFile, and using TCHAR instead of WCHAR,
1457
# but that gets a bit tricky, and requires custom compiling
1460
from bzrlib._walkdirs_win32 import Win32ReadDir
1461
_selected_dir_reader = Win32ReadDir()
1464
elif fs_encoding in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1465
# ANSI_X3.4-1968 is a form of ASCII
1467
from bzrlib._readdir_pyx import UTF8DirReader
1468
_selected_dir_reader = UTF8DirReader()
1472
if _selected_dir_reader is None:
1473
# Fallback to the python version
1474
_selected_dir_reader = UnicodeDirReader()
1476
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1477
# But we don't actually uses 1-3 in pending, so set them to None
1478
pending = [[_selected_dir_reader.top_prefix_to_starting_dir(top, prefix)]]
1479
read_dir = _selected_dir_reader.read_dir
1480
_directory = _directory_kind
1482
relroot, _, _, _, top = pending[-1].pop()
1485
dirblock = sorted(read_dir(relroot, top))
1486
yield (relroot, top), dirblock
1487
# push the user specified dirs from dirblock
1488
next = [d for d in reversed(dirblock) if d[2] == _directory]
1490
pending.append(next)
1493
class UnicodeDirReader(DirReader):
1494
"""A dir reader for non-utf8 file systems, which transcodes."""
1496
__slots__ = ['_utf8_encode']
1499
self._utf8_encode = codecs.getencoder('utf8')
1501
def top_prefix_to_starting_dir(self, top, prefix=""):
1502
"""See DirReader.top_prefix_to_starting_dir."""
1503
return (safe_utf8(prefix), None, None, None, safe_unicode(top))
1505
def read_dir(self, prefix, top):
1506
"""Read a single directory from a non-utf8 file system.
1508
top, and the abspath element in the output are unicode, all other paths
1509
are utf8. Local disk IO is done via unicode calls to listdir etc.
1511
This is currently the fallback code path when the filesystem encoding is
1512
not UTF-8. It may be better to implement an alternative so that we can
1513
safely handle paths that are not properly decodable in the current
1516
See DirReader.read_dir for details.
1518
_utf8_encode = self._utf8_encode
1520
_listdir = os.listdir
1521
_kind_from_mode = file_kind_from_stat_mode
1524
relprefix = prefix + '/'
1527
top_slash = top + u'/'
1530
append = dirblock.append
932
relroot = currentdir[0] + '/'
1531
935
for name in sorted(_listdir(top)):
1533
name_utf8 = _utf8_encode(name)[0]
1534
except UnicodeDecodeError:
1535
raise errors.BadFilenameEncoding(
1536
_utf8_encode(relprefix)[0] + name, _fs_enc)
1537
abspath = top_slash + name
1538
statvalue = _lstat(abspath)
1539
kind = _kind_from_mode(statvalue.st_mode)
1540
append((relprefix + name_utf8, name_utf8, kind, statvalue, abspath))
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:
1544
948
def copy_tree(from_path, to_path, handlers={}):
1545
949
"""Copy all of the entries in from_path into to_path.
1547
:param from_path: The base directory to copy.
951
:param from_path: The base directory to copy.
1548
952
:param to_path: The target directory. If it does not exist, it will
1550
954
:param handlers: A dictionary of functions, which takes a source and
1596
1000
key_a = path_prefix_key(path_a)
1597
1001
key_b = path_prefix_key(path_b)
1598
1002
return cmp(key_a, key_b)
1601
_cached_user_encoding = None
1604
def get_user_encoding(use_cache=True):
1605
"""Find out what the preferred user encoding is.
1607
This is generally the encoding that is used for command line parameters
1608
and file contents. This may be different from the terminal encoding
1609
or the filesystem encoding.
1611
:param use_cache: Enable cache for detected encoding.
1612
(This parameter is turned on by default,
1613
and required only for selftesting)
1615
:return: A string defining the preferred user encoding
1617
global _cached_user_encoding
1618
if _cached_user_encoding is not None and use_cache:
1619
return _cached_user_encoding
1621
if sys.platform == 'darwin':
1622
# python locale.getpreferredencoding() always return
1623
# 'mac-roman' on darwin. That's a lie.
1624
sys.platform = 'posix'
1626
if os.environ.get('LANG', None) is None:
1627
# If LANG is not set, we end up with 'ascii', which is bad
1628
# ('mac-roman' is more than ascii), so we set a default which
1629
# will give us UTF-8 (which appears to work in all cases on
1630
# OSX). Users are still free to override LANG of course, as
1631
# long as it give us something meaningful. This work-around
1632
# *may* not be needed with python 3k and/or OSX 10.5, but will
1633
# work with them too -- vila 20080908
1634
os.environ['LANG'] = 'en_US.UTF-8'
1637
sys.platform = 'darwin'
1642
user_encoding = locale.getpreferredencoding()
1643
except locale.Error, e:
1644
sys.stderr.write('bzr: warning: %s\n'
1645
' Could not determine what text encoding to use.\n'
1646
' This error usually means your Python interpreter\n'
1647
' doesn\'t support the locale set by $LANG (%s)\n'
1648
" Continuing with ascii encoding.\n"
1649
% (e, os.environ.get('LANG')))
1650
user_encoding = 'ascii'
1652
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1653
# treat that as ASCII, and not support printing unicode characters to the
1656
# For python scripts run under vim, we get '', so also treat that as ASCII
1657
if user_encoding in (None, 'cp0', ''):
1658
user_encoding = 'ascii'
1662
codecs.lookup(user_encoding)
1664
sys.stderr.write('bzr: warning:'
1665
' unknown encoding %s.'
1666
' Continuing with ascii encoding.\n'
1669
user_encoding = 'ascii'
1672
_cached_user_encoding = user_encoding
1674
return user_encoding
1677
def get_host_name():
1678
"""Return the current unicode host name.
1680
This is meant to be used in place of socket.gethostname() because that
1681
behaves inconsistently on different platforms.
1683
if sys.platform == "win32":
1685
return win32utils.get_host_name()
1688
return socket.gethostname().decode(get_user_encoding())
1691
def recv_all(socket, bytes):
1692
"""Receive an exact number of bytes.
1694
Regular Socket.recv() may return less than the requested number of bytes,
1695
dependning on what's in the OS buffer. MSG_WAITALL is not available
1696
on all platforms, but this should work everywhere. This will return
1697
less than the requested amount if the remote end closes.
1699
This isn't optimized and is intended mostly for use in testing.
1702
while len(b) < bytes:
1703
new = until_no_eintr(socket.recv, bytes - len(b))
1710
def send_all(socket, bytes, report_activity=None):
1711
"""Send all bytes on a socket.
1713
Regular socket.sendall() can give socket error 10053 on Windows. This
1714
implementation sends no more than 64k at a time, which avoids this problem.
1716
:param report_activity: Call this as bytes are read, see
1717
Transport._report_activity
1720
for pos in xrange(0, len(bytes), chunk_size):
1721
block = bytes[pos:pos+chunk_size]
1722
if report_activity is not None:
1723
report_activity(len(block), 'write')
1724
until_no_eintr(socket.sendall, block)
1727
def dereference_path(path):
1728
"""Determine the real path to a file.
1730
All parent elements are dereferenced. But the file itself is not
1732
:param path: The original path. May be absolute or relative.
1733
:return: the real path *to* the file
1735
parent, base = os.path.split(path)
1736
# The pathjoin for '.' is a workaround for Python bug #1213894.
1737
# (initial path components aren't dereferenced)
1738
return pathjoin(realpath(pathjoin('.', parent)), base)
1741
def supports_mapi():
1742
"""Return True if we can use MAPI to launch a mail client."""
1743
return sys.platform == "win32"
1746
def resource_string(package, resource_name):
1747
"""Load a resource from a package and return it as a string.
1749
Note: Only packages that start with bzrlib are currently supported.
1751
This is designed to be a lightweight implementation of resource
1752
loading in a way which is API compatible with the same API from
1754
http://peak.telecommunity.com/DevCenter/PkgResources#basic-resource-access.
1755
If and when pkg_resources becomes a standard library, this routine
1758
# Check package name is within bzrlib
1759
if package == "bzrlib":
1760
resource_relpath = resource_name
1761
elif package.startswith("bzrlib."):
1762
package = package[len("bzrlib."):].replace('.', os.sep)
1763
resource_relpath = pathjoin(package, resource_name)
1765
raise errors.BzrError('resource package %s not in bzrlib' % package)
1767
# Map the resource to a file and read its contents
1768
base = dirname(bzrlib.__file__)
1769
if getattr(sys, 'frozen', None): # bzr.exe
1770
base = abspath(pathjoin(base, '..', '..'))
1771
filename = pathjoin(base, resource_relpath)
1772
return open(filename, 'rU').read()
1775
def file_kind_from_stat_mode_thunk(mode):
1776
global file_kind_from_stat_mode
1777
if file_kind_from_stat_mode is file_kind_from_stat_mode_thunk:
1779
from bzrlib._readdir_pyx import UTF8DirReader
1780
file_kind_from_stat_mode = UTF8DirReader().kind_from_mode
1782
from bzrlib._readdir_py import (
1783
_kind_from_mode as file_kind_from_stat_mode
1785
return file_kind_from_stat_mode(mode)
1786
file_kind_from_stat_mode = file_kind_from_stat_mode_thunk
1789
def file_kind(f, _lstat=os.lstat):
1791
return file_kind_from_stat_mode(_lstat(f).st_mode)
1793
if getattr(e, 'errno', None) in (errno.ENOENT, errno.ENOTDIR):
1794
raise errors.NoSuchFile(f)
1798
def until_no_eintr(f, *a, **kw):
1799
"""Run f(*a, **kw), retrying if an EINTR error occurs."""
1800
# Borrowed from Twisted's twisted.python.util.untilConcludes function.
1804
except (IOError, OSError), e:
1805
if e.errno == errno.EINTR:
1809
def re_compile_checked(re_string, flags=0, where=""):
1810
"""Return a compiled re, or raise a sensible error.
1812
This should only be used when compiling user-supplied REs.
1814
:param re_string: Text form of regular expression.
1815
:param flags: eg re.IGNORECASE
1816
:param where: Message explaining to the user the context where
1817
it occurred, eg 'log search filter'.
1819
# from https://bugs.launchpad.net/bzr/+bug/251352
1821
re_obj = re.compile(re_string, flags)
1826
where = ' in ' + where
1827
# despite the name 'error' is a type
1828
raise errors.BzrCommandError('Invalid regular expression%s: %r: %s'
1829
% (where, re_string, e))
1832
if sys.platform == "win32":
1835
return msvcrt.getch()
1840
fd = sys.stdin.fileno()
1841
settings = termios.tcgetattr(fd)
1844
ch = sys.stdin.read(1)
1846
termios.tcsetattr(fd, termios.TCSADRAIN, settings)
1850
if sys.platform == 'linux2':
1851
def _local_concurrency():
1853
prefix = 'processor'
1854
for line in file('/proc/cpuinfo', 'rb'):
1855
if line.startswith(prefix):
1856
concurrency = int(line[line.find(':')+1:]) + 1
1858
elif sys.platform == 'darwin':
1859
def _local_concurrency():
1860
return subprocess.Popen(['sysctl', '-n', 'hw.availcpu'],
1861
stdout=subprocess.PIPE).communicate()[0]
1862
elif sys.platform[0:7] == 'freebsd':
1863
def _local_concurrency():
1864
return subprocess.Popen(['sysctl', '-n', 'hw.ncpu'],
1865
stdout=subprocess.PIPE).communicate()[0]
1866
elif sys.platform == 'sunos5':
1867
def _local_concurrency():
1868
return subprocess.Popen(['psrinfo', '-p',],
1869
stdout=subprocess.PIPE).communicate()[0]
1870
elif sys.platform == "win32":
1871
def _local_concurrency():
1872
# This appears to return the number of cores.
1873
return os.environ.get('NUMBER_OF_PROCESSORS')
1875
def _local_concurrency():
1880
_cached_local_concurrency = None
1882
def local_concurrency(use_cache=True):
1883
"""Return how many processes can be run concurrently.
1885
Rely on platform specific implementations and default to 1 (one) if
1886
anything goes wrong.
1888
global _cached_local_concurrency
1889
if _cached_local_concurrency is not None and use_cache:
1890
return _cached_local_concurrency
1893
concurrency = _local_concurrency()
1894
except (OSError, IOError):
1897
concurrency = int(concurrency)
1898
except (TypeError, ValueError):
1901
_cached_concurrency = concurrency
added
removed
bzrlib/osutils.py
