482
526
for dirname in dir_list:
483
527
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
489
def pumpfile(fromfile, tofile):
490
"""Copy contents of one file to another."""
493
b = fromfile.read(BUFSIZE)
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)
499
598
def file_iterator(input_file, readsize=32768):
555
682
def local_time_offset(t=None):
556
683
"""Return offset of local zone from GMT, either at present or at time t."""
557
# python2.3 localtime() can't take 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,
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,
568
692
show_offset=True):
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)
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):
573
729
if timezone == 'utc':
574
730
tt = time.gmtime(t)
576
732
elif timezone == 'original':
579
735
tt = time.gmtime(t + offset)
580
736
elif timezone == 'local':
581
737
tt = time.localtime(t)
582
738
offset = local_time_offset(t)
584
raise BzrError("unsupported timezone format %r" % timezone,
585
['options are "utc", "original", "local"'])
740
raise errors.UnsupportedTimezoneFormat(timezone)
586
741
if date_fmt is None:
587
742
date_fmt = "%a %Y-%m-%d %H:%M:%S"
589
744
offset_str = ' %+03d%02d' % (offset / 3600, (offset / 60) % 60)
592
return (time.strftime(date_fmt, tt) + offset_str)
747
return (date_fmt, tt, offset_str)
595
750
def compact_date(when):
596
751
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)
601
804
"""Return size of given open file."""
705
917
def link_or_copy(src, dest):
706
918
"""Hardlink a file, or copy it if it can't be hardlinked."""
707
919
if not hardlinks_good():
920
shutil.copyfile(src, dest)
711
923
os.link(src, dest)
712
924
except (OSError, IOError), e:
713
925
if e.errno != errno.EXDEV:
717
def delete_any(full_path):
927
shutil.copyfile(src, dest)
930
# Look Before You Leap (LBYL) is appropriate here instead of Easier to Ask for
931
# Forgiveness than Permission (EAFP) because:
932
# - root can damage a solaris file system by using unlink,
933
# - unlink raises different exceptions on different OSes (linux: EISDIR, win32:
934
# EACCES, OSX: EPERM) when invoked on a directory.
935
def delete_any(path):
718
936
"""Delete a file or directory."""
722
# We may be renaming a dangling inventory id
723
if e.errno not in (errno.EISDIR, errno.EACCES, errno.EPERM):
937
if isdir(path): # Takes care of symlinks
728
943
def has_symlinks():
729
if hasattr(os, 'symlink'):
944
if getattr(os, 'symlink', None) is not None:
951
if getattr(os, 'link', None) is not None:
957
def host_os_dereferences_symlinks():
958
return (has_symlinks()
959
and sys.platform not in ('cygwin', 'win32'))
962
def readlink(abspath):
963
"""Return a string representing the path to which the symbolic link points.
965
:param abspath: The link absolute unicode path.
967
This his guaranteed to return the symbolic link in unicode in all python
970
link = abspath.encode(_fs_enc)
971
target = os.readlink(link)
972
target = target.decode(_fs_enc)
735
976
def contains_whitespace(s):
736
977
"""True if there are any whitespace characters in s."""
737
for ch in string.whitespace:
978
# string.whitespace can include '\xa0' in certain locales, because it is
979
# considered "non-breaking-space" as part of ISO-8859-1. But it
980
# 1) Isn't a breaking whitespace
981
# 2) Isn't one of ' \t\r\n' which are characters we sometimes use as
983
# 3) '\xa0' isn't unicode safe since it is >128.
985
# This should *not* be a unicode set of characters in case the source
986
# string is not a Unicode string. We can auto-up-cast the characters since
987
# they are ascii, but we don't want to auto-up-cast the string in case it
989
for ch in ' \t\n\r\v\f':
1040
def _cicp_canonical_relpath(base, path):
1041
"""Return the canonical path relative to base.
1043
Like relpath, but on case-insensitive-case-preserving file-systems, this
1044
will return the relpath as stored on the file-system rather than in the
1045
case specified in the input string, for all existing portions of the path.
1047
This will cause O(N) behaviour if called for every path in a tree; if you
1048
have a number of paths to convert, you should use canonical_relpaths().
1050
# TODO: it should be possible to optimize this for Windows by using the
1051
# win32 API FindFiles function to look for the specified name - but using
1052
# os.listdir() still gives us the correct, platform agnostic semantics in
1055
rel = relpath(base, path)
1056
# '.' will have been turned into ''
1060
abs_base = abspath(base)
1062
_listdir = os.listdir
1064
# use an explicit iterator so we can easily consume the rest on early exit.
1065
bit_iter = iter(rel.split('/'))
1066
for bit in bit_iter:
1068
for look in _listdir(current):
1069
if lbit == look.lower():
1070
current = pathjoin(current, look)
1073
# got to the end, nothing matched, so we just return the
1074
# non-existing bits as they were specified (the filename may be
1075
# the target of a move, for example).
1076
current = pathjoin(current, bit, *list(bit_iter))
1078
return current[len(abs_base)+1:]
1080
# XXX - TODO - we need better detection/integration of case-insensitive
1081
# file-systems; Linux often sees FAT32 devices (or NFS-mounted OSX
1082
# filesystems), for example, so could probably benefit from the same basic
1083
# support there. For now though, only Windows and OSX get that support, and
1084
# they get it for *all* file-systems!
1085
if sys.platform in ('win32', 'darwin'):
1086
canonical_relpath = _cicp_canonical_relpath
1088
canonical_relpath = relpath
1090
def canonical_relpaths(base, paths):
1091
"""Create an iterable to canonicalize a sequence of relative paths.
1093
The intent is for this implementation to use a cache, vastly speeding
1094
up multiple transformations in the same directory.
1096
# but for now, we haven't optimized...
1097
return [canonical_relpath(base, p) for p in paths]
787
1099
def safe_unicode(unicode_or_utf8_string):
788
1100
"""Coerce unicode_or_utf8_string into unicode.
790
1102
If it is unicode, it is returned.
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.
1103
Otherwise it is decoded from utf-8. If decoding fails, the exception is
1104
wrapped in a BzrBadParameterNotUnicode exception.
795
1106
if isinstance(unicode_or_utf8_string, unicode):
796
1107
return unicode_or_utf8_string
798
1109
return unicode_or_utf8_string.decode('utf8')
799
1110
except UnicodeDecodeError:
800
raise BzrBadParameterNotUnicode(unicode_or_utf8_string)
1111
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
1114
def safe_utf8(unicode_or_utf8_string):
1115
"""Coerce unicode_or_utf8_string to a utf8 string.
1117
If it is a str, it is returned.
1118
If it is Unicode, it is encoded into a utf-8 string.
1120
if isinstance(unicode_or_utf8_string, str):
1121
# TODO: jam 20070209 This is overkill, and probably has an impact on
1122
# performance if we are dealing with lots of apis that want a
1125
# Make sure it is a valid utf-8 string
1126
unicode_or_utf8_string.decode('utf-8')
1127
except UnicodeDecodeError:
1128
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
1129
return unicode_or_utf8_string
1130
return unicode_or_utf8_string.encode('utf-8')
1133
_revision_id_warning = ('Unicode revision ids were deprecated in bzr 0.15.'
1134
' Revision id generators should be creating utf8'
1138
def safe_revision_id(unicode_or_utf8_string, warn=True):
1139
"""Revision ids should now be utf8, but at one point they were unicode.
1141
:param unicode_or_utf8_string: A possibly Unicode revision_id. (can also be
1143
:param warn: Functions that are sanitizing user data can set warn=False
1144
:return: None or a utf8 revision id.
1146
if (unicode_or_utf8_string is None
1147
or unicode_or_utf8_string.__class__ == str):
1148
return unicode_or_utf8_string
1150
symbol_versioning.warn(_revision_id_warning, DeprecationWarning,
1152
return cache_utf8.encode(unicode_or_utf8_string)
1155
_file_id_warning = ('Unicode file ids were deprecated in bzr 0.15. File id'
1156
' generators should be creating utf8 file ids.')
1159
def safe_file_id(unicode_or_utf8_string, warn=True):
1160
"""File ids should now be utf8, but at one point they were unicode.
1162
This is the same as safe_utf8, except it uses the cached encode functions
1163
to save a little bit of performance.
1165
:param unicode_or_utf8_string: A possibly Unicode file_id. (can also be
1167
:param warn: Functions that are sanitizing user data can set warn=False
1168
:return: None or a utf8 file id.
1170
if (unicode_or_utf8_string is None
1171
or unicode_or_utf8_string.__class__ == str):
1172
return unicode_or_utf8_string
1174
symbol_versioning.warn(_file_id_warning, DeprecationWarning,
1176
return cache_utf8.encode(unicode_or_utf8_string)
803
1179
_platform_normalizes_filenames = False
870
1246
def supports_executable():
871
1247
return sys.platform != "win32"
1250
def supports_posix_readonly():
1251
"""Return True if 'readonly' has POSIX semantics, False otherwise.
1253
Notably, a win32 readonly file cannot be deleted, unlike POSIX where the
1254
directory controls creation/deletion, etc.
1256
And under win32, readonly means that the directory itself cannot be
1257
deleted. The contents of a readonly directory can be changed, unlike POSIX
1258
where files in readonly directories cannot be added, deleted or renamed.
1260
return sys.platform != "win32"
1263
def set_or_unset_env(env_variable, value):
1264
"""Modify the environment, setting or removing the env_variable.
1266
:param env_variable: The environment variable in question
1267
:param value: The value to set the environment to. If None, then
1268
the variable will be removed.
1269
:return: The original value of the environment variable.
1271
orig_val = os.environ.get(env_variable)
1273
if orig_val is not None:
1274
del os.environ[env_variable]
1276
if isinstance(value, unicode):
1277
value = value.encode(get_user_encoding())
1278
os.environ[env_variable] = value
874
1282
_validWin32PathRE = re.compile(r'^([A-Za-z]:[/\\])?[^:<>*"?\|]*$')
877
1285
def check_legal_path(path):
878
"""Check whether the supplied path is legal.
1286
"""Check whether the supplied path is legal.
879
1287
This is only required on Windows, so we don't test on other platforms
882
1290
if sys.platform != "win32":
884
1292
if _validWin32PathRE.match(path) is None:
885
raise IllegalPath(path)
1293
raise errors.IllegalPath(path)
1296
_WIN32_ERROR_DIRECTORY = 267 # Similar to errno.ENOTDIR
1298
def _is_error_enotdir(e):
1299
"""Check if this exception represents ENOTDIR.
1301
Unfortunately, python is very inconsistent about the exception
1302
here. The cases are:
1303
1) Linux, Mac OSX all versions seem to set errno == ENOTDIR
1304
2) Windows, Python2.4, uses errno == ERROR_DIRECTORY (267)
1305
which is the windows error code.
1306
3) Windows, Python2.5 uses errno == EINVAL and
1307
winerror == ERROR_DIRECTORY
1309
:param e: An Exception object (expected to be OSError with an errno
1310
attribute, but we should be able to cope with anything)
1311
:return: True if this represents an ENOTDIR error. False otherwise.
1313
en = getattr(e, 'errno', None)
1314
if (en == errno.ENOTDIR
1315
or (sys.platform == 'win32'
1316
and (en == _WIN32_ERROR_DIRECTORY
1317
or (en == errno.EINVAL
1318
and getattr(e, 'winerror', None) == _WIN32_ERROR_DIRECTORY)
888
1324
def walkdirs(top, prefix=""):
889
1325
"""Yield data about all the directories in a tree.
891
1327
This yields all the data about the contents of a directory at a time.
892
1328
After each directory has been yielded, if the caller has mutated the list
893
1329
to exclude some directories, they are then not descended into.
895
1331
The data yielded is of the form:
896
1332
((directory-relpath, directory-path-from-top),
897
[(relpath, basename, kind, lstat), ...]),
1333
[(relpath, basename, kind, lstat, path-from-top), ...]),
898
1334
- directory-relpath is the relative path of the directory being returned
899
1335
with respect to top. prefix is prepended to this.
900
- directory-path-from-root is the path including top for this directory.
1336
- directory-path-from-root is the path including top for this directory.
901
1337
It is suitable for use with os functions.
902
1338
- relpath is the relative path within the subtree being walked.
903
1339
- basename is the basename of the path
905
1341
present within the tree - but it may be recorded as versioned. See
907
1343
- lstat is the stat data *if* the file was statted.
908
- planned, not implemented:
1344
- planned, not implemented:
909
1345
path_from_tree_root is the path from the root of the tree.
911
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1347
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
912
1348
allows one to walk a subtree but get paths that are relative to a tree
913
1349
rooted higher up.
914
1350
:return: an iterator over the dirs.
916
1352
#TODO there is a bit of a smell where the results of the directory-
917
# summary in this, and the path from the root, may not agree
1353
# summary in this, and the path from the root, may not agree
918
1354
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
919
1355
# potentially confusing output. We should make this more robust - but
920
1356
# not at a speed cost. RBC 20060731
923
1358
_directory = _directory_kind
925
pending = [(prefix, "", _directory, None, top)]
1359
_listdir = os.listdir
1360
_kind_from_mode = file_kind_from_stat_mode
1361
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
928
currentdir = pending.pop()
929
1363
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
932
relroot = currentdir[0] + '/'
1364
relroot, _, _, _, top = pending.pop()
1366
relprefix = relroot + u'/'
1369
top_slash = top + u'/'
1372
append = dirblock.append
1374
names = sorted(_listdir(top))
1376
if not _is_error_enotdir(e):
1380
abspath = top_slash + name
1381
statvalue = _lstat(abspath)
1382
kind = _kind_from_mode(statvalue.st_mode)
1383
append((relprefix + name, name, kind, statvalue, abspath))
1384
yield (relroot, top), dirblock
1386
# push the user specified dirs from dirblock
1387
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1390
class DirReader(object):
1391
"""An interface for reading directories."""
1393
def top_prefix_to_starting_dir(self, top, prefix=""):
1394
"""Converts top and prefix to a starting dir entry
1396
:param top: A utf8 path
1397
:param prefix: An optional utf8 path to prefix output relative paths
1399
:return: A tuple starting with prefix, and ending with the native
1402
raise NotImplementedError(self.top_prefix_to_starting_dir)
1404
def read_dir(self, prefix, top):
1405
"""Read a specific dir.
1407
:param prefix: A utf8 prefix to be preprended to the path basenames.
1408
:param top: A natively encoded path to read.
1409
:return: A list of the directories contents. Each item contains:
1410
(utf8_relpath, utf8_name, kind, lstatvalue, native_abspath)
1412
raise NotImplementedError(self.read_dir)
1415
_selected_dir_reader = None
1418
def _walkdirs_utf8(top, prefix=""):
1419
"""Yield data about all the directories in a tree.
1421
This yields the same information as walkdirs() only each entry is yielded
1422
in utf-8. On platforms which have a filesystem encoding of utf8 the paths
1423
are returned as exact byte-strings.
1425
:return: yields a tuple of (dir_info, [file_info])
1426
dir_info is (utf8_relpath, path-from-top)
1427
file_info is (utf8_relpath, utf8_name, kind, lstat, path-from-top)
1428
if top is an absolute path, path-from-top is also an absolute path.
1429
path-from-top might be unicode or utf8, but it is the correct path to
1430
pass to os functions to affect the file in question. (such as os.lstat)
1432
global _selected_dir_reader
1433
if _selected_dir_reader is None:
1434
fs_encoding = _fs_enc.upper()
1435
if sys.platform == "win32" and win32utils.winver == 'Windows NT':
1436
# Win98 doesn't have unicode apis like FindFirstFileW
1437
# TODO: We possibly could support Win98 by falling back to the
1438
# original FindFirstFile, and using TCHAR instead of WCHAR,
1439
# but that gets a bit tricky, and requires custom compiling
1442
from bzrlib._walkdirs_win32 import Win32ReadDir
1443
_selected_dir_reader = Win32ReadDir()
1446
elif fs_encoding in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1447
# ANSI_X3.4-1968 is a form of ASCII
1449
from bzrlib._readdir_pyx import UTF8DirReader
1450
_selected_dir_reader = UTF8DirReader()
1454
if _selected_dir_reader is None:
1455
# Fallback to the python version
1456
_selected_dir_reader = UnicodeDirReader()
1458
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1459
# But we don't actually uses 1-3 in pending, so set them to None
1460
pending = [[_selected_dir_reader.top_prefix_to_starting_dir(top, prefix)]]
1461
read_dir = _selected_dir_reader.read_dir
1462
_directory = _directory_kind
1464
relroot, _, _, _, top = pending[-1].pop()
1467
dirblock = sorted(read_dir(relroot, top))
1468
yield (relroot, top), dirblock
1469
# push the user specified dirs from dirblock
1470
next = [d for d in reversed(dirblock) if d[2] == _directory]
1472
pending.append(next)
1475
class UnicodeDirReader(DirReader):
1476
"""A dir reader for non-utf8 file systems, which transcodes."""
1478
__slots__ = ['_utf8_encode']
1481
self._utf8_encode = codecs.getencoder('utf8')
1483
def top_prefix_to_starting_dir(self, top, prefix=""):
1484
"""See DirReader.top_prefix_to_starting_dir."""
1485
return (safe_utf8(prefix), None, None, None, safe_unicode(top))
1487
def read_dir(self, prefix, top):
1488
"""Read a single directory from a non-utf8 file system.
1490
top, and the abspath element in the output are unicode, all other paths
1491
are utf8. Local disk IO is done via unicode calls to listdir etc.
1493
This is currently the fallback code path when the filesystem encoding is
1494
not UTF-8. It may be better to implement an alternative so that we can
1495
safely handle paths that are not properly decodable in the current
1498
See DirReader.read_dir for details.
1500
_utf8_encode = self._utf8_encode
1502
_listdir = os.listdir
1503
_kind_from_mode = file_kind_from_stat_mode
1506
relprefix = prefix + '/'
1509
top_slash = top + u'/'
1512
append = dirblock.append
935
1513
for name in sorted(_listdir(top)):
936
abspath = top + '/' + name
937
statvalue = lstat(abspath)
938
dirblock.append((relroot + name, name,
939
file_kind_from_stat_mode(statvalue.st_mode),
941
yield (currentdir[0], top), dirblock
942
# push the user specified dirs from dirblock
943
for dir in reversed(dirblock):
944
if dir[2] == _directory:
1515
name_utf8 = _utf8_encode(name)[0]
1516
except UnicodeDecodeError:
1517
raise errors.BadFilenameEncoding(
1518
_utf8_encode(relprefix)[0] + name, _fs_enc)
1519
abspath = top_slash + name
1520
statvalue = _lstat(abspath)
1521
kind = _kind_from_mode(statvalue.st_mode)
1522
append((relprefix + name_utf8, name_utf8, kind, statvalue, abspath))
1526
def copy_tree(from_path, to_path, handlers={}):
1527
"""Copy all of the entries in from_path into to_path.
1529
:param from_path: The base directory to copy.
1530
:param to_path: The target directory. If it does not exist, it will
1532
:param handlers: A dictionary of functions, which takes a source and
1533
destinations for files, directories, etc.
1534
It is keyed on the file kind, such as 'directory', 'symlink', or 'file'
1535
'file', 'directory', and 'symlink' should always exist.
1536
If they are missing, they will be replaced with 'os.mkdir()',
1537
'os.readlink() + os.symlink()', and 'shutil.copy2()', respectively.
1539
# Now, just copy the existing cached tree to the new location
1540
# We use a cheap trick here.
1541
# Absolute paths are prefixed with the first parameter
1542
# relative paths are prefixed with the second.
1543
# So we can get both the source and target returned
1544
# without any extra work.
1546
def copy_dir(source, dest):
1549
def copy_link(source, dest):
1550
"""Copy the contents of a symlink"""
1551
link_to = os.readlink(source)
1552
os.symlink(link_to, dest)
1554
real_handlers = {'file':shutil.copy2,
1555
'symlink':copy_link,
1556
'directory':copy_dir,
1558
real_handlers.update(handlers)
1560
if not os.path.exists(to_path):
1561
real_handlers['directory'](from_path, to_path)
1563
for dir_info, entries in walkdirs(from_path, prefix=to_path):
1564
for relpath, name, kind, st, abspath in entries:
1565
real_handlers[kind](abspath, relpath)
948
1568
def path_prefix_key(path):
958
1578
key_a = path_prefix_key(path_a)
959
1579
key_b = path_prefix_key(path_b)
960
1580
return cmp(key_a, key_b)
1583
_cached_user_encoding = None
1586
def get_user_encoding(use_cache=True):
1587
"""Find out what the preferred user encoding is.
1589
This is generally the encoding that is used for command line parameters
1590
and file contents. This may be different from the terminal encoding
1591
or the filesystem encoding.
1593
:param use_cache: Enable cache for detected encoding.
1594
(This parameter is turned on by default,
1595
and required only for selftesting)
1597
:return: A string defining the preferred user encoding
1599
global _cached_user_encoding
1600
if _cached_user_encoding is not None and use_cache:
1601
return _cached_user_encoding
1603
if sys.platform == 'darwin':
1604
# python locale.getpreferredencoding() always return
1605
# 'mac-roman' on darwin. That's a lie.
1606
sys.platform = 'posix'
1608
if os.environ.get('LANG', None) is None:
1609
# If LANG is not set, we end up with 'ascii', which is bad
1610
# ('mac-roman' is more than ascii), so we set a default which
1611
# will give us UTF-8 (which appears to work in all cases on
1612
# OSX). Users are still free to override LANG of course, as
1613
# long as it give us something meaningful. This work-around
1614
# *may* not be needed with python 3k and/or OSX 10.5, but will
1615
# work with them too -- vila 20080908
1616
os.environ['LANG'] = 'en_US.UTF-8'
1619
sys.platform = 'darwin'
1624
user_encoding = locale.getpreferredencoding()
1625
except locale.Error, e:
1626
sys.stderr.write('bzr: warning: %s\n'
1627
' Could not determine what text encoding to use.\n'
1628
' This error usually means your Python interpreter\n'
1629
' doesn\'t support the locale set by $LANG (%s)\n'
1630
" Continuing with ascii encoding.\n"
1631
% (e, os.environ.get('LANG')))
1632
user_encoding = 'ascii'
1634
# Windows returns 'cp0' to indicate there is no code page. So we'll just
1635
# treat that as ASCII, and not support printing unicode characters to the
1638
# For python scripts run under vim, we get '', so also treat that as ASCII
1639
if user_encoding in (None, 'cp0', ''):
1640
user_encoding = 'ascii'
1644
codecs.lookup(user_encoding)
1646
sys.stderr.write('bzr: warning:'
1647
' unknown encoding %s.'
1648
' Continuing with ascii encoding.\n'
1651
user_encoding = 'ascii'
1654
_cached_user_encoding = user_encoding
1656
return user_encoding
1659
def get_host_name():
1660
"""Return the current unicode host name.
1662
This is meant to be used in place of socket.gethostname() because that
1663
behaves inconsistently on different platforms.
1665
if sys.platform == "win32":
1667
return win32utils.get_host_name()
1670
return socket.gethostname().decode(get_user_encoding())
1673
def recv_all(socket, bytes):
1674
"""Receive an exact number of bytes.
1676
Regular Socket.recv() may return less than the requested number of bytes,
1677
dependning on what's in the OS buffer. MSG_WAITALL is not available
1678
on all platforms, but this should work everywhere. This will return
1679
less than the requested amount if the remote end closes.
1681
This isn't optimized and is intended mostly for use in testing.
1684
while len(b) < bytes:
1685
new = until_no_eintr(socket.recv, bytes - len(b))
1692
def send_all(socket, bytes, report_activity=None):
1693
"""Send all bytes on a socket.
1695
Regular socket.sendall() can give socket error 10053 on Windows. This
1696
implementation sends no more than 64k at a time, which avoids this problem.
1698
:param report_activity: Call this as bytes are read, see
1699
Transport._report_activity
1702
for pos in xrange(0, len(bytes), chunk_size):
1703
block = bytes[pos:pos+chunk_size]
1704
if report_activity is not None:
1705
report_activity(len(block), 'write')
1706
until_no_eintr(socket.sendall, block)
1709
def dereference_path(path):
1710
"""Determine the real path to a file.
1712
All parent elements are dereferenced. But the file itself is not
1714
:param path: The original path. May be absolute or relative.
1715
:return: the real path *to* the file
1717
parent, base = os.path.split(path)
1718
# The pathjoin for '.' is a workaround for Python bug #1213894.
1719
# (initial path components aren't dereferenced)
1720
return pathjoin(realpath(pathjoin('.', parent)), base)
1723
def supports_mapi():
1724
"""Return True if we can use MAPI to launch a mail client."""
1725
return sys.platform == "win32"
1728
def resource_string(package, resource_name):
1729
"""Load a resource from a package and return it as a string.
1731
Note: Only packages that start with bzrlib are currently supported.
1733
This is designed to be a lightweight implementation of resource
1734
loading in a way which is API compatible with the same API from
1736
http://peak.telecommunity.com/DevCenter/PkgResources#basic-resource-access.
1737
If and when pkg_resources becomes a standard library, this routine
1740
# Check package name is within bzrlib
1741
if package == "bzrlib":
1742
resource_relpath = resource_name
1743
elif package.startswith("bzrlib."):
1744
package = package[len("bzrlib."):].replace('.', os.sep)
1745
resource_relpath = pathjoin(package, resource_name)
1747
raise errors.BzrError('resource package %s not in bzrlib' % package)
1749
# Map the resource to a file and read its contents
1750
base = dirname(bzrlib.__file__)
1751
if getattr(sys, 'frozen', None): # bzr.exe
1752
base = abspath(pathjoin(base, '..', '..'))
1753
filename = pathjoin(base, resource_relpath)
1754
return open(filename, 'rU').read()
1757
def file_kind_from_stat_mode_thunk(mode):
1758
global file_kind_from_stat_mode
1759
if file_kind_from_stat_mode is file_kind_from_stat_mode_thunk:
1761
from bzrlib._readdir_pyx import UTF8DirReader
1762
file_kind_from_stat_mode = UTF8DirReader().kind_from_mode
1764
from bzrlib._readdir_py import (
1765
_kind_from_mode as file_kind_from_stat_mode
1767
return file_kind_from_stat_mode(mode)
1768
file_kind_from_stat_mode = file_kind_from_stat_mode_thunk
1771
def file_kind(f, _lstat=os.lstat):
1773
return file_kind_from_stat_mode(_lstat(f).st_mode)
1775
if getattr(e, 'errno', None) in (errno.ENOENT, errno.ENOTDIR):
1776
raise errors.NoSuchFile(f)
1780
def until_no_eintr(f, *a, **kw):
1781
"""Run f(*a, **kw), retrying if an EINTR error occurs."""
1782
# Borrowed from Twisted's twisted.python.util.untilConcludes function.
1786
except (IOError, OSError), e:
1787
if e.errno == errno.EINTR:
1791
def re_compile_checked(re_string, flags=0, where=""):
1792
"""Return a compiled re, or raise a sensible error.
1794
This should only be used when compiling user-supplied REs.
1796
:param re_string: Text form of regular expression.
1797
:param flags: eg re.IGNORECASE
1798
:param where: Message explaining to the user the context where
1799
it occurred, eg 'log search filter'.
1801
# from https://bugs.launchpad.net/bzr/+bug/251352
1803
re_obj = re.compile(re_string, flags)
1808
where = ' in ' + where
1809
# despite the name 'error' is a type
1810
raise errors.BzrCommandError('Invalid regular expression%s: %r: %s'
1811
% (where, re_string, e))
1814
if sys.platform == "win32":
1817
return msvcrt.getch()
1822
fd = sys.stdin.fileno()
1823
settings = termios.tcgetattr(fd)
1826
ch = sys.stdin.read(1)
1828
termios.tcsetattr(fd, termios.TCSADRAIN, settings)
1832
if sys.platform == 'linux2':
1833
def _local_concurrency():
1835
prefix = 'processor'
1836
for line in file('/proc/cpuinfo', 'rb'):
1837
if line.startswith(prefix):
1838
concurrency = int(line[line.find(':')+1:]) + 1
1840
elif sys.platform == 'darwin':
1841
def _local_concurrency():
1842
return subprocess.Popen(['sysctl', '-n', 'hw.availcpu'],
1843
stdout=subprocess.PIPE).communicate()[0]
1844
elif sys.platform[0:7] == 'freebsd':
1845
def _local_concurrency():
1846
return subprocess.Popen(['sysctl', '-n', 'hw.ncpu'],
1847
stdout=subprocess.PIPE).communicate()[0]
1848
elif sys.platform == 'sunos5':
1849
def _local_concurrency():
1850
return subprocess.Popen(['psrinfo', '-p',],
1851
stdout=subprocess.PIPE).communicate()[0]
1852
elif sys.platform == "win32":
1853
def _local_concurrency():
1854
# This appears to return the number of cores.
1855
return os.environ.get('NUMBER_OF_PROCESSORS')
1857
def _local_concurrency():
1862
_cached_local_concurrency = None
1864
def local_concurrency(use_cache=True):
1865
"""Return how many processes can be run concurrently.
1867
Rely on platform specific implementations and default to 1 (one) if
1868
anything goes wrong.
1870
global _cached_local_concurrency
1871
if _cached_local_concurrency is not None and use_cache:
1872
return _cached_local_concurrency
1875
concurrency = _local_concurrency()
1876
except (OSError, IOError):
1879
concurrency = int(concurrency)
1880
except (TypeError, ValueError):
1883
_cached_concurrency = concurrency