505
480
for dirname in dir_list:
506
481
if is_inside(dirname, fname) or is_inside(fname, dirname):
511
def pumpfile(from_file, to_file, read_length=-1, buff_size=32768,
512
report_activity=None, direction='read'):
513
"""Copy contents of one file to another.
515
The read_length can either be -1 to read to end-of-file (EOF) or
516
it can specify the maximum number of bytes to read.
518
The buff_size represents the maximum size for each read operation
519
performed on from_file.
521
:param report_activity: Call this as bytes are read, see
522
Transport._report_activity
523
:param direction: Will be passed to report_activity
525
:return: The number of bytes copied.
529
# read specified number of bytes
531
while read_length > 0:
532
num_bytes_to_read = min(read_length, buff_size)
534
block = from_file.read(num_bytes_to_read)
538
if report_activity is not None:
539
report_activity(len(block), direction)
542
actual_bytes_read = len(block)
543
read_length -= actual_bytes_read
544
length += actual_bytes_read
548
block = from_file.read(buff_size)
552
if report_activity is not None:
553
report_activity(len(block), direction)
559
def pump_string_file(bytes, file_handle, segment_size=None):
560
"""Write bytes to file_handle in many smaller writes.
562
:param bytes: The string to write.
563
:param file_handle: The file to write to.
565
# Write data in chunks rather than all at once, because very large
566
# writes fail on some platforms (e.g. Windows with SMB mounted
569
segment_size = 5242880 # 5MB
570
segments = range(len(bytes) / segment_size + 1)
571
write = file_handle.write
572
for segment_index in segments:
573
segment = buffer(bytes, segment_index * segment_size, segment_size)
487
def pumpfile(fromfile, tofile):
488
"""Copy contents of one file to another."""
491
b = fromfile.read(BUFSIZE)
577
497
def file_iterator(input_file, readsize=32768):
661
553
def local_time_offset(t=None):
662
554
"""Return offset of local zone from GMT, either at present or at time t."""
555
# python2.3 localtime() can't take None
665
offset = datetime.fromtimestamp(t) - datetime.utcfromtimestamp(t)
666
return offset.days * 86400 + offset.seconds
668
weekdays = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
670
def format_date(t, offset=0, timezone='original', date_fmt=None,
559
if time.localtime(t).tm_isdst and time.daylight:
562
return -time.timezone
565
def format_date(t, offset=0, timezone='original', date_fmt=None,
671
566
show_offset=True):
672
"""Return a formatted date string.
674
:param t: Seconds since the epoch.
675
:param offset: Timezone offset in seconds east of utc.
676
:param timezone: How to display the time: 'utc', 'original' for the
677
timezone specified by offset, or 'local' for the process's current
679
:param date_fmt: strftime format.
680
:param show_offset: Whether to append the timezone.
682
(date_fmt, tt, offset_str) = \
683
_format_date(t, offset, timezone, date_fmt, show_offset)
684
date_fmt = date_fmt.replace('%a', weekdays[tt[6]])
685
date_str = time.strftime(date_fmt, tt)
686
return date_str + offset_str
688
def format_local_date(t, offset=0, timezone='original', date_fmt=None,
690
"""Return an unicode date string formatted according to the current locale.
692
:param t: Seconds since the epoch.
693
:param offset: Timezone offset in seconds east of utc.
694
:param timezone: How to display the time: 'utc', 'original' for the
695
timezone specified by offset, or 'local' for the process's current
697
:param date_fmt: strftime format.
698
:param show_offset: Whether to append the timezone.
700
(date_fmt, tt, offset_str) = \
701
_format_date(t, offset, timezone, date_fmt, show_offset)
702
date_str = time.strftime(date_fmt, tt)
703
if not isinstance(date_str, unicode):
704
date_str = date_str.decode(bzrlib.user_encoding, 'replace')
705
return date_str + offset_str
707
def _format_date(t, offset, timezone, date_fmt, show_offset):
567
## TODO: Perhaps a global option to use either universal or local time?
568
## Or perhaps just let people set $TZ?
569
assert isinstance(t, float)
708
571
if timezone == 'utc':
709
572
tt = time.gmtime(t)
992
def _cicp_canonical_relpath(base, path):
993
"""Return the canonical path relative to base.
995
Like relpath, but on case-insensitive-case-preserving file-systems, this
996
will return the relpath as stored on the file-system rather than in the
997
case specified in the input string, for all existing portions of the path.
999
This will cause O(N) behaviour if called for every path in a tree; if you
1000
have a number of paths to convert, you should use canonical_relpaths().
1002
# TODO: it should be possible to optimize this for Windows by using the
1003
# win32 API FindFiles function to look for the specified name - but using
1004
# os.listdir() still gives us the correct, platform agnostic semantics in
1007
rel = relpath(base, path)
1008
# '.' will have been turned into ''
1012
abs_base = abspath(base)
1014
_listdir = os.listdir
1016
# use an explicit iterator so we can easily consume the rest on early exit.
1017
bit_iter = iter(rel.split('/'))
1018
for bit in bit_iter:
1020
for look in _listdir(current):
1021
if lbit == look.lower():
1022
current = pathjoin(current, look)
1025
# got to the end, nothing matched, so we just return the
1026
# non-existing bits as they were specified (the filename may be
1027
# the target of a move, for example).
1028
current = pathjoin(current, bit, *list(bit_iter))
1030
return current[len(abs_base)+1:]
1032
# XXX - TODO - we need better detection/integration of case-insensitive
1033
# file-systems; Linux often sees FAT32 devices, for example, so could
1034
# probably benefit from the same basic support there. For now though, only
1035
# Windows gets that support, and it gets it for *all* file-systems!
1036
if sys.platform == "win32":
1037
canonical_relpath = _cicp_canonical_relpath
1039
canonical_relpath = relpath
1041
def canonical_relpaths(base, paths):
1042
"""Create an iterable to canonicalize a sequence of relative paths.
1044
The intent is for this implementation to use a cache, vastly speeding
1045
up multiple transformations in the same directory.
1047
# but for now, we haven't optimized...
1048
return [canonical_relpath(base, p) for p in paths]
1050
819
def safe_unicode(unicode_or_utf8_string):
1051
820
"""Coerce unicode_or_utf8_string into unicode.
1053
822
If it is unicode, it is returned.
1054
Otherwise it is decoded from utf-8. If decoding fails, the exception is
1055
wrapped in a BzrBadParameterNotUnicode exception.
823
Otherwise it is decoded from utf-8. If a decoding error
824
occurs, it is wrapped as a If the decoding fails, the exception is wrapped
825
as a BzrBadParameter exception.
1057
827
if isinstance(unicode_or_utf8_string, unicode):
1058
828
return unicode_or_utf8_string
1062
832
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
1065
def safe_utf8(unicode_or_utf8_string):
1066
"""Coerce unicode_or_utf8_string to a utf8 string.
1068
If it is a str, it is returned.
1069
If it is Unicode, it is encoded into a utf-8 string.
1071
if isinstance(unicode_or_utf8_string, str):
1072
# TODO: jam 20070209 This is overkill, and probably has an impact on
1073
# performance if we are dealing with lots of apis that want a
1076
# Make sure it is a valid utf-8 string
1077
unicode_or_utf8_string.decode('utf-8')
1078
except UnicodeDecodeError:
1079
raise errors.BzrBadParameterNotUnicode(unicode_or_utf8_string)
1080
return unicode_or_utf8_string
1081
return unicode_or_utf8_string.encode('utf-8')
1084
_revision_id_warning = ('Unicode revision ids were deprecated in bzr 0.15.'
1085
' Revision id generators should be creating utf8'
1089
def safe_revision_id(unicode_or_utf8_string, warn=True):
1090
"""Revision ids should now be utf8, but at one point they were unicode.
1092
:param unicode_or_utf8_string: A possibly Unicode revision_id. (can also be
1094
:param warn: Functions that are sanitizing user data can set warn=False
1095
:return: None or a utf8 revision id.
1097
if (unicode_or_utf8_string is None
1098
or unicode_or_utf8_string.__class__ == str):
1099
return unicode_or_utf8_string
1101
symbol_versioning.warn(_revision_id_warning, DeprecationWarning,
1103
return cache_utf8.encode(unicode_or_utf8_string)
1106
_file_id_warning = ('Unicode file ids were deprecated in bzr 0.15. File id'
1107
' generators should be creating utf8 file ids.')
1110
def safe_file_id(unicode_or_utf8_string, warn=True):
1111
"""File ids should now be utf8, but at one point they were unicode.
1113
This is the same as safe_utf8, except it uses the cached encode functions
1114
to save a little bit of performance.
1116
:param unicode_or_utf8_string: A possibly Unicode file_id. (can also be
1118
:param warn: Functions that are sanitizing user data can set warn=False
1119
:return: None or a utf8 file id.
1121
if (unicode_or_utf8_string is None
1122
or unicode_or_utf8_string.__class__ == str):
1123
return unicode_or_utf8_string
1125
symbol_versioning.warn(_file_id_warning, DeprecationWarning,
1127
return cache_utf8.encode(unicode_or_utf8_string)
1130
835
_platform_normalizes_filenames = False
1131
836
if sys.platform == 'darwin':
1132
837
_platform_normalizes_filenames = True
1244
937
raise errors.IllegalPath(path)
1247
_WIN32_ERROR_DIRECTORY = 267 # Similar to errno.ENOTDIR
1249
def _is_error_enotdir(e):
1250
"""Check if this exception represents ENOTDIR.
1252
Unfortunately, python is very inconsistent about the exception
1253
here. The cases are:
1254
1) Linux, Mac OSX all versions seem to set errno == ENOTDIR
1255
2) Windows, Python2.4, uses errno == ERROR_DIRECTORY (267)
1256
which is the windows error code.
1257
3) Windows, Python2.5 uses errno == EINVAL and
1258
winerror == ERROR_DIRECTORY
1260
:param e: An Exception object (expected to be OSError with an errno
1261
attribute, but we should be able to cope with anything)
1262
:return: True if this represents an ENOTDIR error. False otherwise.
1264
en = getattr(e, 'errno', None)
1265
if (en == errno.ENOTDIR
1266
or (sys.platform == 'win32'
1267
and (en == _WIN32_ERROR_DIRECTORY
1268
or (en == errno.EINVAL
1269
and getattr(e, 'winerror', None) == _WIN32_ERROR_DIRECTORY)
1275
940
def walkdirs(top, prefix=""):
1276
941
"""Yield data about all the directories in a tree.
1278
943
This yields all the data about the contents of a directory at a time.
1279
944
After each directory has been yielded, if the caller has mutated the list
1280
945
to exclude some directories, they are then not descended into.
1282
947
The data yielded is of the form:
1283
948
((directory-relpath, directory-path-from-top),
1284
[(relpath, basename, kind, lstat, path-from-top), ...]),
949
[(relpath, basename, kind, lstat), ...]),
1285
950
- directory-relpath is the relative path of the directory being returned
1286
951
with respect to top. prefix is prepended to this.
1287
- directory-path-from-root is the path including top for this directory.
952
- directory-path-from-root is the path including top for this directory.
1288
953
It is suitable for use with os functions.
1289
954
- relpath is the relative path within the subtree being walked.
1290
955
- basename is the basename of the path
1292
957
present within the tree - but it may be recorded as versioned. See
1294
959
- lstat is the stat data *if* the file was statted.
1295
- planned, not implemented:
960
- planned, not implemented:
1296
961
path_from_tree_root is the path from the root of the tree.
1298
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
963
:param prefix: Prefix the relpaths that are yielded with 'prefix'. This
1299
964
allows one to walk a subtree but get paths that are relative to a tree
1300
965
rooted higher up.
1301
966
:return: an iterator over the dirs.
1303
968
#TODO there is a bit of a smell where the results of the directory-
1304
# summary in this, and the path from the root, may not agree
969
# summary in this, and the path from the root, may not agree
1305
970
# depending on top and prefix - i.e. ./foo and foo as a pair leads to
1306
971
# potentially confusing output. We should make this more robust - but
1307
972
# not at a speed cost. RBC 20060731
1309
975
_directory = _directory_kind
1310
976
_listdir = os.listdir
1311
_kind_from_mode = file_kind_from_stat_mode
1312
pending = [(safe_unicode(prefix), "", _directory, None, safe_unicode(top))]
977
pending = [(prefix, "", _directory, None, top)]
980
currentdir = pending.pop()
1314
981
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1315
relroot, _, _, _, top = pending.pop()
1317
relprefix = relroot + u'/'
1320
top_slash = top + u'/'
1323
append = dirblock.append
1325
names = sorted(_listdir(top))
1327
if not _is_error_enotdir(e):
1331
abspath = top_slash + name
1332
statvalue = _lstat(abspath)
1333
kind = _kind_from_mode(statvalue.st_mode)
1334
append((relprefix + name, name, kind, statvalue, abspath))
1335
yield (relroot, top), dirblock
1337
# push the user specified dirs from dirblock
1338
pending.extend(d for d in reversed(dirblock) if d[2] == _directory)
1341
class DirReader(object):
1342
"""An interface for reading directories."""
1344
def top_prefix_to_starting_dir(self, top, prefix=""):
1345
"""Converts top and prefix to a starting dir entry
1347
:param top: A utf8 path
1348
:param prefix: An optional utf8 path to prefix output relative paths
1350
:return: A tuple starting with prefix, and ending with the native
1353
raise NotImplementedError(self.top_prefix_to_starting_dir)
1355
def read_dir(self, prefix, top):
1356
"""Read a specific dir.
1358
:param prefix: A utf8 prefix to be preprended to the path basenames.
1359
:param top: A natively encoded path to read.
1360
:return: A list of the directories contents. Each item contains:
1361
(utf8_relpath, utf8_name, kind, lstatvalue, native_abspath)
1363
raise NotImplementedError(self.read_dir)
1366
_selected_dir_reader = None
1369
def _walkdirs_utf8(top, prefix=""):
1370
"""Yield data about all the directories in a tree.
1372
This yields the same information as walkdirs() only each entry is yielded
1373
in utf-8. On platforms which have a filesystem encoding of utf8 the paths
1374
are returned as exact byte-strings.
1376
:return: yields a tuple of (dir_info, [file_info])
1377
dir_info is (utf8_relpath, path-from-top)
1378
file_info is (utf8_relpath, utf8_name, kind, lstat, path-from-top)
1379
if top is an absolute path, path-from-top is also an absolute path.
1380
path-from-top might be unicode or utf8, but it is the correct path to
1381
pass to os functions to affect the file in question. (such as os.lstat)
1383
global _selected_dir_reader
1384
if _selected_dir_reader is None:
1385
fs_encoding = _fs_enc.upper()
1386
if sys.platform == "win32" and win32utils.winver == 'Windows NT':
1387
# Win98 doesn't have unicode apis like FindFirstFileW
1388
# TODO: We possibly could support Win98 by falling back to the
1389
# original FindFirstFile, and using TCHAR instead of WCHAR,
1390
# but that gets a bit tricky, and requires custom compiling
1393
from bzrlib._walkdirs_win32 import Win32ReadDir
1395
_selected_dir_reader = UnicodeDirReader()
1397
_selected_dir_reader = Win32ReadDir()
1398
elif fs_encoding not in ('UTF-8', 'US-ASCII', 'ANSI_X3.4-1968'):
1399
# ANSI_X3.4-1968 is a form of ASCII
1400
_selected_dir_reader = UnicodeDirReader()
1403
from bzrlib._readdir_pyx import UTF8DirReader
1405
# No optimised code path
1406
_selected_dir_reader = UnicodeDirReader()
1408
_selected_dir_reader = UTF8DirReader()
1409
# 0 - relpath, 1- basename, 2- kind, 3- stat, 4-toppath
1410
# But we don't actually uses 1-3 in pending, so set them to None
1411
pending = [[_selected_dir_reader.top_prefix_to_starting_dir(top, prefix)]]
1412
read_dir = _selected_dir_reader.read_dir
1413
_directory = _directory_kind
1415
relroot, _, _, _, top = pending[-1].pop()
1418
dirblock = sorted(read_dir(relroot, top))
1419
yield (relroot, top), dirblock
1420
# push the user specified dirs from dirblock
1421
next = [d for d in reversed(dirblock) if d[2] == _directory]
1423
pending.append(next)
1426
class UnicodeDirReader(DirReader):
1427
"""A dir reader for non-utf8 file systems, which transcodes."""
1429
__slots__ = ['_utf8_encode']
1432
self._utf8_encode = codecs.getencoder('utf8')
1434
def top_prefix_to_starting_dir(self, top, prefix=""):
1435
"""See DirReader.top_prefix_to_starting_dir."""
1436
return (safe_utf8(prefix), None, None, None, safe_unicode(top))
1438
def read_dir(self, prefix, top):
1439
"""Read a single directory from a non-utf8 file system.
1441
top, and the abspath element in the output are unicode, all other paths
1442
are utf8. Local disk IO is done via unicode calls to listdir etc.
1444
This is currently the fallback code path when the filesystem encoding is
1445
not UTF-8. It may be better to implement an alternative so that we can
1446
safely handle paths that are not properly decodable in the current
1449
See DirReader.read_dir for details.
1451
_utf8_encode = self._utf8_encode
1453
_listdir = os.listdir
1454
_kind_from_mode = file_kind_from_stat_mode
1457
relprefix = prefix + '/'
1460
top_slash = top + u'/'
1463
append = dirblock.append
984
relroot = currentdir[0] + '/'
1464
987
for name in sorted(_listdir(top)):
1466
name_utf8 = _utf8_encode(name)[0]
1467
except UnicodeDecodeError:
1468
raise errors.BadFilenameEncoding(
1469
_utf8_encode(relprefix)[0] + name, _fs_enc)
1470
abspath = top_slash + name
1471
statvalue = _lstat(abspath)
1472
kind = _kind_from_mode(statvalue.st_mode)
1473
append((relprefix + name_utf8, name_utf8, kind, statvalue, abspath))
988
abspath = top + '/' + name
989
statvalue = lstat(abspath)
990
dirblock.append((relroot + name, name,
991
file_kind_from_stat_mode(statvalue.st_mode),
993
yield (currentdir[0], top), dirblock
994
# push the user specified dirs from dirblock
995
for dir in reversed(dirblock):
996
if dir[2] == _directory:
1477
1000
def copy_tree(from_path, to_path, handlers={}):
1478
1001
"""Copy all of the entries in from_path into to_path.
1480
:param from_path: The base directory to copy.
1003
:param from_path: The base directory to copy.
1481
1004
:param to_path: The target directory. If it does not exist, it will
1483
1006
:param handlers: A dictionary of functions, which takes a source and
1534
1057
_cached_user_encoding = None
1537
def get_user_encoding(use_cache=True):
1060
def get_user_encoding():
1538
1061
"""Find out what the preferred user encoding is.
1540
1063
This is generally the encoding that is used for command line parameters
1541
1064
and file contents. This may be different from the terminal encoding
1542
1065
or the filesystem encoding.
1544
:param use_cache: Enable cache for detected encoding.
1545
(This parameter is turned on by default,
1546
and required only for selftesting)
1548
1067
:return: A string defining the preferred user encoding
1550
1069
global _cached_user_encoding
1551
if _cached_user_encoding is not None and use_cache:
1070
if _cached_user_encoding is not None:
1552
1071
return _cached_user_encoding
1554
1073
if sys.platform == 'darwin':
1555
# python locale.getpreferredencoding() always return
1556
# 'mac-roman' on darwin. That's a lie.
1074
# work around egregious python 2.4 bug
1557
1075
sys.platform = 'posix'
1559
if os.environ.get('LANG', None) is None:
1560
# If LANG is not set, we end up with 'ascii', which is bad
1561
# ('mac-roman' is more than ascii), so we set a default which
1562
# will give us UTF-8 (which appears to work in all cases on
1563
# OSX). Users are still free to override LANG of course, as
1564
# long as it give us something meaningful. This work-around
1565
# *may* not be needed with python 3k and/or OSX 10.5, but will
1566
# work with them too -- vila 20080908
1567
os.environ['LANG'] = 'en_US.UTF-8'
1570
1079
sys.platform = 'darwin'
1669
1128
# The pathjoin for '.' is a workaround for Python bug #1213894.
1670
1129
# (initial path components aren't dereferenced)
1671
1130
return pathjoin(realpath(pathjoin('.', parent)), base)
1674
def supports_mapi():
1675
"""Return True if we can use MAPI to launch a mail client."""
1676
return sys.platform == "win32"
1679
def resource_string(package, resource_name):
1680
"""Load a resource from a package and return it as a string.
1682
Note: Only packages that start with bzrlib are currently supported.
1684
This is designed to be a lightweight implementation of resource
1685
loading in a way which is API compatible with the same API from
1687
http://peak.telecommunity.com/DevCenter/PkgResources#basic-resource-access.
1688
If and when pkg_resources becomes a standard library, this routine
1691
# Check package name is within bzrlib
1692
if package == "bzrlib":
1693
resource_relpath = resource_name
1694
elif package.startswith("bzrlib."):
1695
package = package[len("bzrlib."):].replace('.', os.sep)
1696
resource_relpath = pathjoin(package, resource_name)
1698
raise errors.BzrError('resource package %s not in bzrlib' % package)
1700
# Map the resource to a file and read its contents
1701
base = dirname(bzrlib.__file__)
1702
if getattr(sys, 'frozen', None): # bzr.exe
1703
base = abspath(pathjoin(base, '..', '..'))
1704
filename = pathjoin(base, resource_relpath)
1705
return open(filename, 'rU').read()
1708
def file_kind_from_stat_mode_thunk(mode):
1709
global file_kind_from_stat_mode
1710
if file_kind_from_stat_mode is file_kind_from_stat_mode_thunk:
1712
from bzrlib._readdir_pyx import UTF8DirReader
1713
file_kind_from_stat_mode = UTF8DirReader().kind_from_mode
1715
from bzrlib._readdir_py import (
1716
_kind_from_mode as file_kind_from_stat_mode
1718
return file_kind_from_stat_mode(mode)
1719
file_kind_from_stat_mode = file_kind_from_stat_mode_thunk
1722
def file_kind(f, _lstat=os.lstat):
1724
return file_kind_from_stat_mode(_lstat(f).st_mode)
1726
if getattr(e, 'errno', None) in (errno.ENOENT, errno.ENOTDIR):
1727
raise errors.NoSuchFile(f)
1731
def until_no_eintr(f, *a, **kw):
1732
"""Run f(*a, **kw), retrying if an EINTR error occurs."""
1733
# Borrowed from Twisted's twisted.python.util.untilConcludes function.
1737
except (IOError, OSError), e:
1738
if e.errno == errno.EINTR:
1742
def re_compile_checked(re_string, flags=0, where=""):
1743
"""Return a compiled re, or raise a sensible error.
1745
This should only be used when compiling user-supplied REs.
1747
:param re_string: Text form of regular expression.
1748
:param flags: eg re.IGNORECASE
1749
:param where: Message explaining to the user the context where
1750
it occurred, eg 'log search filter'.
1752
# from https://bugs.launchpad.net/bzr/+bug/251352
1754
re_obj = re.compile(re_string, flags)
1759
where = ' in ' + where
1760
# despite the name 'error' is a type
1761
raise errors.BzrCommandError('Invalid regular expression%s: %r: %s'
1762
% (where, re_string, e))
1765
if sys.platform == "win32":
1768
return msvcrt.getch()
1773
fd = sys.stdin.fileno()
1774
settings = termios.tcgetattr(fd)
1777
ch = sys.stdin.read(1)
1779
termios.tcsetattr(fd, termios.TCSADRAIN, settings)