~bzr-pqm/bzr/bzr.dev

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
# Copyright (C) 2005-2010 Canonical Ltd
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

"""Transport is an abstraction layer to handle file access.

The abstraction is to allow access from the local filesystem, as well
as remote (such as http or sftp).

Transports are constructed from a string, being a URL or (as a degenerate
case) a local filesystem path.  This is typically the top directory of
a bzrdir, repository, or similar object we are interested in working with.
The Transport returned has methods to read, write and manipulate files within
it.
"""

from cStringIO import StringIO
import re
import sys

from bzrlib.lazy_import import lazy_import
lazy_import(globals(), """
import errno
from stat import S_ISDIR
import urllib
import urlparse

from bzrlib import (
    errors,
    osutils,
    symbol_versioning,
    ui,
    urlutils,
    )
""")

from bzrlib.symbol_versioning import (
        DEPRECATED_PARAMETER,
        )
from bzrlib.trace import (
    mutter,
    )
from bzrlib import registry


# a dictionary of open file streams. Keys are absolute paths, values are
# transport defined.
_file_streams = {}


def _get_protocol_handlers():
    """Return a dictionary of {urlprefix: [factory]}"""
    return transport_list_registry


def _set_protocol_handlers(new_handlers):
    """Replace the current protocol handlers dictionary.

    WARNING this will remove all build in protocols. Use with care.
    """
    global transport_list_registry
    transport_list_registry = new_handlers


def _clear_protocol_handlers():
    global transport_list_registry
    transport_list_registry = TransportListRegistry()


def _get_transport_modules():
    """Return a list of the modules providing transports."""
    modules = set()
    for prefix, factory_list in transport_list_registry.items():
        for factory in factory_list:
            if hasattr(factory, "_module_name"):
                modules.add(factory._module_name)
            else:
                modules.add(factory._obj.__module__)
    # Add chroot and pathfilter directly, because there is no handler
    # registered for it.
    modules.add('bzrlib.transport.chroot')
    modules.add('bzrlib.transport.pathfilter')
    result = list(modules)
    result.sort()
    return result


class TransportListRegistry(registry.Registry):
    """A registry which simplifies tracking available Transports.

    A registration of a new protocol requires two steps:
    1) register the prefix with the function register_transport( )
    2) register the protocol provider with the function
    register_transport_provider( ) ( and the "lazy" variant )

    This is needed because:
    a) a single provider can support multiple protocols ( like the ftp
    provider which supports both the ftp:// and the aftp:// protocols )
    b) a single protocol can have multiple providers ( like the http://
    protocol which is supported by both the urllib and pycurl provider )
    """

    def register_transport_provider(self, key, obj):
        self.get(key).insert(0, registry._ObjectGetter(obj))

    def register_lazy_transport_provider(self, key, module_name, member_name):
        self.get(key).insert(0,
                registry._LazyObjectGetter(module_name, member_name))

    def register_transport(self, key, help=None):
        self.register(key, [], help)

    def set_default_transport(self, key=None):
        """Return either 'key' or the default key if key is None"""
        self._default_key = key


transport_list_registry = TransportListRegistry()


def register_transport_proto(prefix, help=None, info=None,
                             register_netloc=False):
    transport_list_registry.register_transport(prefix, help)
    if register_netloc:
        if not prefix.endswith('://'):
            raise ValueError(prefix)
        register_urlparse_netloc_protocol(prefix[:-3])


def register_lazy_transport(prefix, module, classname):
    if not prefix in transport_list_registry:
        register_transport_proto(prefix)
    transport_list_registry.register_lazy_transport_provider(prefix, module, classname)


def register_transport(prefix, klass, override=DEPRECATED_PARAMETER):
    if not prefix in transport_list_registry:
        register_transport_proto(prefix)
    transport_list_registry.register_transport_provider(prefix, klass)


def register_urlparse_netloc_protocol(protocol):
    """Ensure that protocol is setup to be used with urlparse netloc parsing."""
    if protocol not in urlparse.uses_netloc:
        urlparse.uses_netloc.append(protocol)


def _unregister_urlparse_netloc_protocol(protocol):
    """Remove protocol from urlparse netloc parsing.

    Except for tests, you should never use that function. Using it with 'http',
    for example, will break all http transports.
    """
    if protocol in urlparse.uses_netloc:
        urlparse.uses_netloc.remove(protocol)


def unregister_transport(scheme, factory):
    """Unregister a transport."""
    l = transport_list_registry.get(scheme)
    for i in l:
        o = i.get_obj( )
        if o == factory:
            transport_list_registry.get(scheme).remove(i)
            break
    if len(l) == 0:
        transport_list_registry.remove(scheme)


class _CoalescedOffset(object):
    """A data container for keeping track of coalesced offsets."""

    __slots__ = ['start', 'length', 'ranges']

    def __init__(self, start, length, ranges):
        self.start = start
        self.length = length
        self.ranges = ranges

    def __cmp__(self, other):
        return cmp((self.start, self.length, self.ranges),
                   (other.start, other.length, other.ranges))

    def __repr__(self):
        return '%s(%r, %r, %r)' % (self.__class__.__name__,
            self.start, self.length, self.ranges)


class LateReadError(object):
    """A helper for transports which pretends to be a readable file.

    When read() is called, errors.ReadError is raised.
    """

    def __init__(self, path):
        self._path = path

    def close(self):
        """a no-op - do nothing."""

    def _fail(self):
        """Raise ReadError."""
        raise errors.ReadError(self._path)

    def __iter__(self):
        self._fail()

    def read(self, count=-1):
        self._fail()

    def readlines(self):
        self._fail()


class FileStream(object):
    """Base class for FileStreams."""

    def __init__(self, transport, relpath):
        """Create a FileStream for relpath on transport."""
        self.transport = transport
        self.relpath = relpath

    def _close(self):
        """A hook point for subclasses that need to take action on close."""

    def close(self):
        self._close()
        del _file_streams[self.transport.abspath(self.relpath)]


class FileFileStream(FileStream):
    """A file stream object returned by open_write_stream.

    This version uses a file like object to perform writes.
    """

    def __init__(self, transport, relpath, file_handle):
        FileStream.__init__(self, transport, relpath)
        self.file_handle = file_handle

    def _close(self):
        self.file_handle.close()

    def write(self, bytes):
        osutils.pump_string_file(bytes, self.file_handle)


class AppendBasedFileStream(FileStream):
    """A file stream object returned by open_write_stream.

    This version uses append on a transport to perform writes.
    """

    def write(self, bytes):
        self.transport.append_bytes(self.relpath, bytes)


class Transport(object):
    """This class encapsulates methods for retrieving or putting a file
    from/to a storage location.

    Most functions have a _multi variant, which allows you to queue up
    multiple requests. They generally have a dumb base implementation
    which just iterates over the arguments, but smart Transport
    implementations can do pipelining.
    In general implementations should support having a generator or a list
    as an argument (ie always iterate, never index)

    :ivar base: Base URL for the transport; should always end in a slash.
    """

    # implementations can override this if it is more efficient
    # for them to combine larger read chunks together
    _max_readv_combine = 50
    # It is better to read this much more data in order, rather
    # than doing another seek. Even for the local filesystem,
    # there is a benefit in just reading.
    # TODO: jam 20060714 Do some real benchmarking to figure out
    #       where the biggest benefit between combining reads and
    #       and seeking is. Consider a runtime auto-tune.
    _bytes_to_read_before_seek = 0

    def __init__(self, base):
        super(Transport, self).__init__()
        self.base = base

    def _translate_error(self, e, path, raise_generic=True):
        """Translate an IOError or OSError into an appropriate bzr error.

        This handles things like ENOENT, ENOTDIR, EEXIST, and EACCESS
        """
        if getattr(e, 'errno', None) is not None:
            if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
                raise errors.NoSuchFile(path, extra=e)
            # I would rather use errno.EFOO, but there doesn't seem to be
            # any matching for 267
            # This is the error when doing a listdir on a file:
            # WindowsError: [Errno 267] The directory name is invalid
            if sys.platform == 'win32' and e.errno in (errno.ESRCH, 267):
                raise errors.NoSuchFile(path, extra=e)
            if e.errno == errno.EEXIST:
                raise errors.FileExists(path, extra=e)
            if e.errno == errno.EACCES:
                raise errors.PermissionDenied(path, extra=e)
            if e.errno == errno.ENOTEMPTY:
                raise errors.DirectoryNotEmpty(path, extra=e)
            if e.errno == errno.EBUSY:
                raise errors.ResourceBusy(path, extra=e)
        if raise_generic:
            raise errors.TransportError(orig_error=e)

    def clone(self, offset=None):
        """Return a new Transport object, cloned from the current location,
        using a subdirectory or parent directory. This allows connections
        to be pooled, rather than a new one needed for each subdir.
        """
        raise NotImplementedError(self.clone)

    def create_prefix(self):
        """Create all the directories leading down to self.base."""
        cur_transport = self
        needed = [cur_transport]
        # Recurse upwards until we can create a directory successfully
        while True:
            new_transport = cur_transport.clone('..')
            if new_transport.base == cur_transport.base:
                raise errors.BzrCommandError(
                    "Failed to create path prefix for %s."
                    % cur_transport.base)
            try:
                new_transport.mkdir('.')
            except errors.NoSuchFile:
                needed.append(new_transport)
                cur_transport = new_transport
            except errors.FileExists:
                break
            else:
                break
        # Now we only need to create child directories
        while needed:
            cur_transport = needed.pop()
            cur_transport.ensure_base()

    def ensure_base(self):
        """Ensure that the directory this transport references exists.

        This will create a directory if it doesn't exist.
        :return: True if the directory was created, False otherwise.
        """
        # The default implementation just uses "Easier to ask for forgiveness
        # than permission". We attempt to create the directory, and just
        # suppress FileExists and PermissionDenied (for Windows) exceptions.
        try:
            self.mkdir('.')
        except (errors.FileExists, errors.PermissionDenied):
            return False
        else:
            return True

    def external_url(self):
        """Return a URL for self that can be given to an external process.

        There is no guarantee that the URL can be accessed from a different
        machine - e.g. file:/// urls are only usable on the local machine,
        sftp:/// urls when the server is only bound to localhost are only
        usable from localhost etc.

        NOTE: This method may remove security wrappers (e.g. on chroot
        transports) and thus should *only* be used when the result will not
        be used to obtain a new transport within bzrlib. Ideally chroot
        transports would know enough to cause the external url to be the exact
        one used that caused the chrooting in the first place, but that is not
        currently the case.

        :return: A URL that can be given to another process.
        :raises InProcessTransport: If the transport is one that cannot be
            accessed out of the current process (e.g. a MemoryTransport)
            then InProcessTransport is raised.
        """
        raise NotImplementedError(self.external_url)

    def _pump(self, from_file, to_file):
        """Most children will need to copy from one file-like
        object or string to another one.
        This just gives them something easy to call.
        """
        return osutils.pumpfile(from_file, to_file)

    def _get_total(self, multi):
        """Try to figure out how many entries are in multi,
        but if not possible, return None.
        """
        try:
            return len(multi)
        except TypeError: # We can't tell how many, because relpaths is a generator
            return None

    def _report_activity(self, bytes, direction):
        """Notify that this transport has activity.

        Implementations should call this from all methods that actually do IO.
        Be careful that it's not called twice, if one method is implemented on
        top of another.

        :param bytes: Number of bytes read or written.
        :param direction: 'read' or 'write' or None.
        """
        ui.ui_factory.report_transport_activity(self, bytes, direction)

    def _update_pb(self, pb, msg, count, total):
        """Update the progress bar based on the current count
        and total available, total may be None if it was
        not possible to determine.
        """
        if pb is None:
            return
        if total is None:
            pb.update(msg, count, count+1)
        else:
            pb.update(msg, count, total)

    def _iterate_over(self, multi, func, pb, msg, expand=True):
        """Iterate over all entries in multi, passing them to func,
        and update the progress bar as you go along.

        :param expand:  If True, the entries will be passed to the function
                        by expanding the tuple. If False, it will be passed
                        as a single parameter.
        """
        total = self._get_total(multi)
        result = []
        count = 0
        for entry in multi:
            self._update_pb(pb, msg, count, total)
            if expand:
                result.append(func(*entry))
            else:
                result.append(func(entry))
            count += 1
        return tuple(result)

    def abspath(self, relpath):
        """Return the full url to the given relative path.

        :param relpath: a string of a relative path
        """

        # XXX: Robert Collins 20051016 - is this really needed in the public
        # interface ?
        raise NotImplementedError(self.abspath)

    def _combine_paths(self, base_path, relpath):
        """Transform a Transport-relative path to a remote absolute path.

        This does not handle substitution of ~ but does handle '..' and '.'
        components.

        Examples::

            t._combine_paths('/home/sarah', 'project/foo')
                => '/home/sarah/project/foo'
            t._combine_paths('/home/sarah', '../../etc')
                => '/etc'
            t._combine_paths('/home/sarah', '/etc')
                => '/etc'

        :param base_path: urlencoded path for the transport root; typically a
             URL but need not contain scheme/host/etc.
        :param relpath: relative url string for relative part of remote path.
        :return: urlencoded string for final path.
        """
        if not isinstance(relpath, str):
            raise errors.InvalidURL(relpath)
        if relpath.startswith('/'):
            base_parts = []
        else:
            base_parts = base_path.split('/')
        if len(base_parts) > 0 and base_parts[-1] == '':
            base_parts = base_parts[:-1]
        for p in relpath.split('/'):
            if p == '..':
                if len(base_parts) == 0:
                    # In most filesystems, a request for the parent
                    # of root, just returns root.
                    continue
                base_parts.pop()
            elif p == '.':
                continue # No-op
            elif p != '':
                base_parts.append(p)
        path = '/'.join(base_parts)
        if not path.startswith('/'):
            path = '/' + path
        return path

    def recommended_page_size(self):
        """Return the recommended page size for this transport.

        This is potentially different for every path in a given namespace.
        For example, local transports might use an operating system call to
        get the block size for a given path, which can vary due to mount
        points.

        :return: The page size in bytes.
        """
        return 4 * 1024

    def relpath(self, abspath):
        """Return the local path portion from a given absolute path.

        This default implementation is not suitable for filesystems with
        aliasing, such as that given by symlinks, where a path may not
        start with our base, but still be a relpath once aliasing is
        resolved.
        """
        # TODO: This might want to use bzrlib.osutils.relpath
        #       but we have to watch out because of the prefix issues
        if not (abspath == self.base[:-1] or abspath.startswith(self.base)):
            raise errors.PathNotChild(abspath, self.base)
        pl = len(self.base)
        return abspath[pl:].strip('/')

    def local_abspath(self, relpath):
        """Return the absolute path on the local filesystem.

        This function will only be defined for Transports which have a
        physical local filesystem representation.

        :raises errors.NotLocalUrl: When no local path representation is
            available.
        """
        raise errors.NotLocalUrl(self.abspath(relpath))

    def has(self, relpath):
        """Does the file relpath exist?

        Note that some transports MAY allow querying on directories, but this
        is not part of the protocol.  In other words, the results of
        t.has("a_directory_name") are undefined.

        :rtype: bool
        """
        raise NotImplementedError(self.has)

    def has_multi(self, relpaths, pb=None):
        """Return True/False for each entry in relpaths"""
        total = self._get_total(relpaths)
        count = 0
        for relpath in relpaths:
            self._update_pb(pb, 'has', count, total)
            yield self.has(relpath)
            count += 1

    def has_any(self, relpaths):
        """Return True if any of the paths exist."""
        for relpath in relpaths:
            if self.has(relpath):
                return True
        return False

    def iter_files_recursive(self):
        """Iter the relative paths of files in the transports sub-tree.

        *NOTE*: This only lists *files*, not subdirectories!

        As with other listing functions, only some transports implement this,.
        you may check via listable() to determine if it will.
        """
        raise errors.TransportNotPossible("This transport has not "
                                          "implemented iter_files_recursive "
                                          "(but must claim to be listable "
                                          "to trigger this error).")

    def get(self, relpath):
        """Get the file at the given relative path.

        This may fail in a number of ways:
         - HTTP servers may return content for a directory. (unexpected
           content failure)
         - FTP servers may indicate NoSuchFile for a directory.
         - SFTP servers may give a file handle for a directory that will
           fail on read().

        For correct use of the interface, be sure to catch errors.PathError
        when calling it and catch errors.ReadError when reading from the
        returned object.

        :param relpath: The relative path to the file
        :rtype: File-like object.
        """
        raise NotImplementedError(self.get)

    def get_bytes(self, relpath):
        """Get a raw string of the bytes for a file at the given location.

        :param relpath: The relative path to the file
        """
        f = self.get(relpath)
        try:
            return f.read()
        finally:
            f.close()

    def get_smart_medium(self):
        """Return a smart client medium for this transport if possible.

        A smart medium doesn't imply the presence of a smart server: it implies
        that the smart protocol can be tunnelled via this transport.

        :raises NoSmartMedium: if no smart server medium is available.
        """
        raise errors.NoSmartMedium(self)

    def readv(self, relpath, offsets, adjust_for_latency=False,
        upper_limit=None):
        """Get parts of the file at the given relative path.

        :param relpath: The path to read data from.
        :param offsets: A list of (offset, size) tuples.
        :param adjust_for_latency: Adjust the requested offsets to accomodate
            transport latency. This may re-order the offsets, expand them to
            grab adjacent data when there is likely a high cost to requesting
            data relative to delivering it.
        :param upper_limit: When adjust_for_latency is True setting upper_limit
            allows the caller to tell the transport about the length of the
            file, so that requests are not issued for ranges beyond the end of
            the file. This matters because some servers and/or transports error
            in such a case rather than just satisfying the available ranges.
            upper_limit should always be provided when adjust_for_latency is
            True, and should be the size of the file in bytes.
        :return: A list or generator of (offset, data) tuples
        """
        if adjust_for_latency:
            # Design note: We may wish to have different algorithms for the
            # expansion of the offsets per-transport. E.g. for local disk to
            # use page-aligned expansion. If that is the case consider the
            # following structure:
            #  - a test that transport.readv uses self._offset_expander or some
            #    similar attribute, to do the expansion
            #  - a test for each transport that it has some known-good offset
            #    expander
            #  - unit tests for each offset expander
            #  - a set of tests for the offset expander interface, giving
            #    baseline behaviour (which the current transport
            #    adjust_for_latency tests could be repurposed to).
            offsets = self._sort_expand_and_combine(offsets, upper_limit)
        return self._readv(relpath, offsets)

    def _readv(self, relpath, offsets):
        """Get parts of the file at the given relative path.

        :param relpath: The path to read.
        :param offsets: A list of (offset, size) tuples.
        :return: A list or generator of (offset, data) tuples
        """
        if not offsets:
            return

        fp = self.get(relpath)
        return self._seek_and_read(fp, offsets, relpath)

    def _seek_and_read(self, fp, offsets, relpath='<unknown>'):
        """An implementation of readv that uses fp.seek and fp.read.

        This uses _coalesce_offsets to issue larger reads and fewer seeks.

        :param fp: A file-like object that supports seek() and read(size)
        :param offsets: A list of offsets to be read from the given file.
        :return: yield (pos, data) tuples for each request
        """
        # We are going to iterate multiple times, we need a list
        offsets = list(offsets)
        sorted_offsets = sorted(offsets)

        # turn the list of offsets into a stack
        offset_stack = iter(offsets)
        cur_offset_and_size = offset_stack.next()
        coalesced = self._coalesce_offsets(sorted_offsets,
                               limit=self._max_readv_combine,
                               fudge_factor=self._bytes_to_read_before_seek)

        # Cache the results, but only until they have been fulfilled
        data_map = {}
        for c_offset in coalesced:
            # TODO: jam 20060724 it might be faster to not issue seek if
            #       we are already at the right location. This should be
            #       benchmarked.
            fp.seek(c_offset.start)
            data = fp.read(c_offset.length)
            if len(data) < c_offset.length:
                raise errors.ShortReadvError(relpath, c_offset.start,
                            c_offset.length, actual=len(data))
            for suboffset, subsize in c_offset.ranges:
                key = (c_offset.start+suboffset, subsize)
                data_map[key] = data[suboffset:suboffset+subsize]

            # Now that we've read some data, see if we can yield anything back
            while cur_offset_and_size in data_map:
                this_data = data_map.pop(cur_offset_and_size)
                this_offset = cur_offset_and_size[0]
                try:
                    cur_offset_and_size = offset_stack.next()
                except StopIteration:
                    # Close the file handle as there will be no more data
                    # The handle would normally be cleaned up as this code goes
                    # out of scope, but as we are a generator, not all code
                    # will re-enter once we have consumed all the expected
                    # data. For example:
                    #   zip(range(len(requests)), readv(foo, requests))
                    # Will stop because the range is done, and not run the
                    # cleanup code for the readv().
                    fp.close()
                    cur_offset_and_size = None
                yield this_offset, this_data

    def _sort_expand_and_combine(self, offsets, upper_limit):
        """Helper for readv.

        :param offsets: A readv vector - (offset, length) tuples.
        :param upper_limit: The highest byte offset that may be requested.
        :return: A readv vector that will read all the regions requested by
            offsets, in start-to-end order, with no duplicated regions,
            expanded by the transports recommended page size.
        """
        offsets = sorted(offsets)
        # short circuit empty requests
        if len(offsets) == 0:
            def empty_yielder():
                # Quick thunk to stop this function becoming a generator
                # itself, rather we return a generator that has nothing to
                # yield.
                if False:
                    yield None
            return empty_yielder()
        # expand by page size at either end
        maximum_expansion = self.recommended_page_size()
        new_offsets = []
        for offset, length in offsets:
            expansion = maximum_expansion - length
            if expansion < 0:
                # we're asking for more than the minimum read anyway.
                expansion = 0
            reduction = expansion / 2
            new_offset = offset - reduction
            new_length = length + expansion
            if new_offset < 0:
                # don't ask for anything < 0
                new_offset = 0
            if (upper_limit is not None and
                new_offset + new_length > upper_limit):
                new_length = upper_limit - new_offset
            new_offsets.append((new_offset, new_length))
        # combine the expanded offsets
        offsets = []
        current_offset, current_length = new_offsets[0]
        current_finish = current_length + current_offset
        for offset, length in new_offsets[1:]:
            finish = offset + length
            if offset > current_finish:
                # there is a gap, output the current accumulator and start
                # a new one for the region we're examining.
                offsets.append((current_offset, current_length))
                current_offset = offset
                current_length = length
                current_finish = finish
                continue
            if finish > current_finish:
                # extend the current accumulator to the end of the region
                # we're examining.
                current_finish = finish
                current_length = finish - current_offset
        offsets.append((current_offset, current_length))
        return offsets

    @staticmethod
    def _coalesce_offsets(offsets, limit=0, fudge_factor=0, max_size=0):
        """Yield coalesced offsets.

        With a long list of neighboring requests, combine them
        into a single large request, while retaining the original
        offsets.
        Turns  [(15, 10), (25, 10)] => [(15, 20, [(0, 10), (10, 10)])]
        Note that overlapping requests are not permitted. (So [(15, 10), (20,
        10)] will raise a ValueError.) This is because the data we access never
        overlaps, and it allows callers to trust that we only need any byte of
        data for 1 request (so nothing needs to be buffered to fulfill a second
        request.)

        :param offsets: A list of (start, length) pairs
        :param limit: Only combine a maximum of this many pairs Some transports
                penalize multiple reads more than others, and sometimes it is
                better to return early.
                0 means no limit
        :param fudge_factor: All transports have some level of 'it is
                better to read some more data and throw it away rather
                than seek', so collapse if we are 'close enough'
        :param max_size: Create coalesced offsets no bigger than this size.
                When a single offset is bigger than 'max_size', it will keep
                its size and be alone in the coalesced offset.
                0 means no maximum size.
        :return: return a list of _CoalescedOffset objects, which have members
            for where to start, how much to read, and how to split those chunks
            back up
        """
        last_end = None
        cur = _CoalescedOffset(None, None, [])
        coalesced_offsets = []

        if max_size <= 0:
            # 'unlimited', but we actually take this to mean 100MB buffer limit
            max_size = 100*1024*1024

        for start, size in offsets:
            end = start + size
            if (last_end is not None
                and start <= last_end + fudge_factor
                and start >= cur.start
                and (limit <= 0 or len(cur.ranges) < limit)
                and (max_size <= 0 or end - cur.start <= max_size)):
                if start < last_end:
                    raise ValueError('Overlapping range not allowed:'
                        ' last range ended at %s, new one starts at %s'
                        % (last_end, start))
                cur.length = end - cur.start
                cur.ranges.append((start-cur.start, size))
            else:
                if cur.start is not None:
                    coalesced_offsets.append(cur)
                cur = _CoalescedOffset(start, size, [(0, size)])
            last_end = end

        if cur.start is not None:
            coalesced_offsets.append(cur)
        return coalesced_offsets

    def get_multi(self, relpaths, pb=None):
        """Get a list of file-like objects, one for each entry in relpaths.

        :param relpaths: A list of relative paths.
        :param pb:  An optional ProgressTask for indicating percent done.
        :return: A list or generator of file-like objects
        """
        # TODO: Consider having this actually buffer the requests,
        # in the default mode, it probably won't give worse performance,
        # and all children wouldn't have to implement buffering
        total = self._get_total(relpaths)
        count = 0
        for relpath in relpaths:
            self._update_pb(pb, 'get', count, total)
            yield self.get(relpath)
            count += 1

    def put_bytes(self, relpath, bytes, mode=None):
        """Atomically put the supplied bytes into the given location.

        :param relpath: The location to put the contents, relative to the
            transport base.
        :param bytes: A bytestring of data.
        :param mode: Create the file with the given mode.
        :return: None
        """
        if not isinstance(bytes, str):
            raise AssertionError(
                'bytes must be a plain string, not %s' % type(bytes))
        return self.put_file(relpath, StringIO(bytes), mode=mode)

    def put_bytes_non_atomic(self, relpath, bytes, mode=None,
                             create_parent_dir=False,
                             dir_mode=None):
        """Copy the string into the target location.

        This function is not strictly safe to use. See
        Transport.put_bytes_non_atomic for more information.

        :param relpath: The remote location to put the contents.
        :param bytes:   A string object containing the raw bytes to write into
                        the target file.
        :param mode:    Possible access permissions for new file.
                        None means do not set remote permissions.
        :param create_parent_dir: If we cannot create the target file because
                        the parent directory does not exist, go ahead and
                        create it, and then try again.
        :param dir_mode: Possible access permissions for new directories.
        """
        if not isinstance(bytes, str):
            raise AssertionError(
                'bytes must be a plain string, not %s' % type(bytes))
        self.put_file_non_atomic(relpath, StringIO(bytes), mode=mode,
                                 create_parent_dir=create_parent_dir,
                                 dir_mode=dir_mode)

    def put_file(self, relpath, f, mode=None):
        """Copy the file-like object into the location.

        :param relpath: Location to put the contents, relative to base.
        :param f:       File-like object.
        :param mode: The mode for the newly created file,
                     None means just use the default.
        :return: The length of the file that was written.
        """
        # We would like to mark this as NotImplemented, but most likely
        # transports have defined it in terms of the old api.
        symbol_versioning.warn('Transport %s should implement put_file,'
                               ' rather than implementing put() as of'
                               ' version 0.11.'
                               % (self.__class__.__name__,),
                               DeprecationWarning)
        return self.put(relpath, f, mode=mode)
        #raise NotImplementedError(self.put_file)

    def put_file_non_atomic(self, relpath, f, mode=None,
                            create_parent_dir=False,
                            dir_mode=None):
        """Copy the file-like object into the target location.

        This function is not strictly safe to use. It is only meant to
        be used when you already know that the target does not exist.
        It is not safe, because it will open and truncate the remote
        file. So there may be a time when the file has invalid contents.

        :param relpath: The remote location to put the contents.
        :param f:       File-like object.
        :param mode:    Possible access permissions for new file.
                        None means do not set remote permissions.
        :param create_parent_dir: If we cannot create the target file because
                        the parent directory does not exist, go ahead and
                        create it, and then try again.
        :param dir_mode: Possible access permissions for new directories.
        """
        # Default implementation just does an atomic put.
        try:
            return self.put_file(relpath, f, mode=mode)
        except errors.NoSuchFile:
            if not create_parent_dir:
                raise
            parent_dir = osutils.dirname(relpath)
            if parent_dir:
                self.mkdir(parent_dir, mode=dir_mode)
                return self.put_file(relpath, f, mode=mode)

    def mkdir(self, relpath, mode=None):
        """Create a directory at the given path."""
        raise NotImplementedError(self.mkdir)

    def mkdir_multi(self, relpaths, mode=None, pb=None):
        """Create a group of directories"""
        def mkdir(path):
            self.mkdir(path, mode=mode)
        return len(self._iterate_over(relpaths, mkdir, pb, 'mkdir', expand=False))

    def open_write_stream(self, relpath, mode=None):
        """Open a writable file stream at relpath.

        A file stream is a file like object with a write() method that accepts
        bytes to write.. Buffering may occur internally until the stream is
        closed with stream.close().  Calls to readv or the get_* methods will
        be synchronised with any internal buffering that may be present.

        :param relpath: The relative path to the file.
        :param mode: The mode for the newly created file,
                     None means just use the default
        :return: A FileStream. FileStream objects have two methods, write() and
            close(). There is no guarantee that data is committed to the file
            if close() has not been called (even if get() is called on the same
            path).
        """
        raise NotImplementedError(self.open_write_stream)

    def append_file(self, relpath, f, mode=None):
        """Append bytes from a file-like object to a file at relpath.

        The file is created if it does not already exist.

        :param f: a file-like object of the bytes to append.
        :param mode: Unix mode for newly created files.  This is not used for
            existing files.

        :returns: the length of relpath before the content was written to it.
        """
        symbol_versioning.warn('Transport %s should implement append_file,'
                               ' rather than implementing append() as of'
                               ' version 0.11.'
                               % (self.__class__.__name__,),
                               DeprecationWarning)
        return self.append(relpath, f, mode=mode)

    def append_bytes(self, relpath, bytes, mode=None):
        """Append bytes to a file at relpath.

        The file is created if it does not already exist.

        :type f: str
        :param f: a string of the bytes to append.
        :param mode: Unix mode for newly created files.  This is not used for
            existing files.

        :returns: the length of relpath before the content was written to it.
        """
        if not isinstance(bytes, str):
            raise TypeError(
                'bytes must be a plain string, not %s' % type(bytes))
        return self.append_file(relpath, StringIO(bytes), mode=mode)

    def append_multi(self, files, pb=None):
        """Append the text in each file-like or string object to
        the supplied location.

        :param files: A set of (path, f) entries
        :param pb:  An optional ProgressTask for indicating percent done.
        """
        return self._iterate_over(files, self.append_file, pb, 'append', expand=True)

    def copy(self, rel_from, rel_to):
        """Copy the item at rel_from to the location at rel_to.

        Override this for efficiency if a specific transport can do it
        faster than this default implementation.
        """
        self.put_file(rel_to, self.get(rel_from))

    def copy_multi(self, relpaths, pb=None):
        """Copy a bunch of entries.

        :param relpaths: A list of tuples of the form [(from, to), (from, to),...]
        """
        # This is the non-pipelined implementation, so that
        # implementors don't have to implement everything.
        return self._iterate_over(relpaths, self.copy, pb, 'copy', expand=True)

    def copy_to(self, relpaths, other, mode=None, pb=None):
        """Copy a set of entries from self into another Transport.

        :param relpaths: A list/generator of entries to be copied.
        :param mode: This is the target mode for the newly created files
        TODO: This interface needs to be updated so that the target location
              can be different from the source location.
        """
        # The dummy implementation just does a simple get + put
        def copy_entry(path):
            other.put_file(path, self.get(path), mode=mode)

        return len(self._iterate_over(relpaths, copy_entry, pb, 'copy_to', expand=False))

    def copy_tree(self, from_relpath, to_relpath):
        """Copy a subtree from one relpath to another.

        If a faster implementation is available, specific transports should
        implement it.
        """
        source = self.clone(from_relpath)
        target = self.clone(to_relpath)

        # create target directory with the same rwx bits as source.
        # use mask to ensure that bits other than rwx are ignored.
        stat = self.stat(from_relpath)
        target.mkdir('.', stat.st_mode & 0777)
        source.copy_tree_to_transport(target)

    def copy_tree_to_transport(self, to_transport):
        """Copy a subtree from one transport to another.

        self.base is used as the source tree root, and to_transport.base
        is used as the target.  to_transport.base must exist (and be a
        directory).
        """
        files = []
        directories = ['.']
        while directories:
            dir = directories.pop()
            if dir != '.':
                to_transport.mkdir(dir)
            for path in self.list_dir(dir):
                path = dir + '/' + path
                stat = self.stat(path)
                if S_ISDIR(stat.st_mode):
                    directories.append(path)
                else:
                    files.append(path)
        self.copy_to(files, to_transport)

    def rename(self, rel_from, rel_to):
        """Rename a file or directory.

        This *must* fail if the destination is a nonempty directory - it must
        not automatically remove it.  It should raise DirectoryNotEmpty, or
        some other PathError if the case can't be specifically detected.

        If the destination is an empty directory or a file this function may
        either fail or succeed, depending on the underlying transport.  It
        should not attempt to remove the destination if overwriting is not the
        native transport behaviour.  If at all possible the transport should
        ensure that the rename either completes or not, without leaving the
        destination deleted and the new file not moved in place.

        This is intended mainly for use in implementing LockDir.
        """
        # transports may need to override this
        raise NotImplementedError(self.rename)

    def move(self, rel_from, rel_to):
        """Move the item at rel_from to the location at rel_to.

        The destination is deleted if possible, even if it's a non-empty
        directory tree.

        If a transport can directly implement this it is suggested that
        it do so for efficiency.
        """
        if S_ISDIR(self.stat(rel_from).st_mode):
            self.copy_tree(rel_from, rel_to)
            self.delete_tree(rel_from)
        else:
            self.copy(rel_from, rel_to)
            self.delete(rel_from)

    def move_multi(self, relpaths, pb=None):
        """Move a bunch of entries.

        :param relpaths: A list of tuples of the form [(from1, to1), (from2, to2),...]
        """
        return self._iterate_over(relpaths, self.move, pb, 'move', expand=True)

    def move_multi_to(self, relpaths, rel_to):
        """Move a bunch of entries to a single location.
        This differs from move_multi in that you give a list of from, and
        a single destination, rather than multiple destinations.

        :param relpaths: A list of relative paths [from1, from2, from3, ...]
        :param rel_to: A directory where each entry should be placed.
        """
        # This is not implemented, because you need to do special tricks to
        # extract the basename, and add it to rel_to
        raise NotImplementedError(self.move_multi_to)

    def delete(self, relpath):
        """Delete the item at relpath"""
        raise NotImplementedError(self.delete)

    def delete_multi(self, relpaths, pb=None):
        """Queue up a bunch of deletes to be done.
        """
        return self._iterate_over(relpaths, self.delete, pb, 'delete', expand=False)

    def delete_tree(self, relpath):
        """Delete an entire tree. This may require a listable transport."""
        subtree = self.clone(relpath)
        files = []
        directories = ['.']
        pending_rmdirs = []
        while directories:
            dir = directories.pop()
            if dir != '.':
                pending_rmdirs.append(dir)
            for path in subtree.list_dir(dir):
                path = dir + '/' + path
                stat = subtree.stat(path)
                if S_ISDIR(stat.st_mode):
                    directories.append(path)
                else:
                    files.append(path)
        subtree.delete_multi(files)
        pending_rmdirs.reverse()
        for dir in pending_rmdirs:
            subtree.rmdir(dir)
        self.rmdir(relpath)

    def __repr__(self):
        return "<%s.%s url=%s>" % (self.__module__, self.__class__.__name__, self.base)

    def stat(self, relpath):
        """Return the stat information for a file.
        WARNING: This may not be implementable for all protocols, so use
        sparingly.
        NOTE: This returns an object with fields such as 'st_size'. It MAY
        or MAY NOT return the literal result of an os.stat() call, so all
        access should be via named fields.
        ALSO NOTE: Stats of directories may not be supported on some
        transports.
        """
        raise NotImplementedError(self.stat)

    def rmdir(self, relpath):
        """Remove a directory at the given path."""
        raise NotImplementedError

    def stat_multi(self, relpaths, pb=None):
        """Stat multiple files and return the information.
        """
        #TODO:  Is it worth making this a generator instead of a
        #       returning a list?
        stats = []
        def gather(path):
            stats.append(self.stat(path))

        count = self._iterate_over(relpaths, gather, pb, 'stat', expand=False)
        return stats

    def readlink(self, relpath):
        """Return a string representing the path to which the symbolic link points."""
        raise errors.TransportNotPossible("Dereferencing symlinks is not supported on %s" % self)

    def hardlink(self, source, link_name):
        """Create a hardlink pointing to source named link_name."""
        raise errors.TransportNotPossible("Hard links are not supported on %s" % self)

    def symlink(self, source, link_name):
        """Create a symlink pointing to source named link_name."""
        raise errors.TransportNotPossible("Symlinks are not supported on %s" % self)

    def listable(self):
        """Return True if this store supports listing."""
        raise NotImplementedError(self.listable)

    def list_dir(self, relpath):
        """Return a list of all files at the given location.
        WARNING: many transports do not support this, so trying avoid using
        it if at all possible.
        """
        raise errors.TransportNotPossible("Transport %r has not "
                                          "implemented list_dir "
                                          "(but must claim to be listable "
                                          "to trigger this error)."
                                          % (self))

    def lock_read(self, relpath):
        """Lock the given file for shared (read) access.

        WARNING: many transports do not support this, so trying avoid using it.
        These methods may be removed in the future.

        Transports may raise TransportNotPossible if OS-level locks cannot be
        taken over this transport.

        :return: A lock object, which should contain an unlock() function.
        """
        raise errors.TransportNotPossible("transport locks not supported on %s" % self)

    def lock_write(self, relpath):
        """Lock the given file for exclusive (write) access.

        WARNING: many transports do not support this, so trying avoid using it.
        These methods may be removed in the future.

        Transports may raise TransportNotPossible if OS-level locks cannot be
        taken over this transport.

        :return: A lock object, which should contain an unlock() function.
        """
        raise errors.TransportNotPossible("transport locks not supported on %s" % self)

    def is_readonly(self):
        """Return true if this connection cannot be written to."""
        return False

    def _can_roundtrip_unix_modebits(self):
        """Return true if this transport can store and retrieve unix modebits.

        (For example, 0700 to make a directory owner-private.)

        Note: most callers will not want to switch on this, but should rather
        just try and set permissions and let them be either stored or not.
        This is intended mainly for the use of the test suite.

        Warning: this is not guaranteed to be accurate as sometimes we can't
        be sure: for example with vfat mounted on unix, or a windows sftp
        server."""
        # TODO: Perhaps return a e.g. TransportCharacteristics that can answer
        # several questions about the transport.
        return False

    def _reuse_for(self, other_base):
        # This is really needed for ConnectedTransport only, but it's easier to
        # have Transport refuses to be reused than testing that the reuse
        # should be asked to ConnectedTransport only.
        return None

    def disconnect(self):
        # This is really needed for ConnectedTransport only, but it's easier to
        # have Transport do nothing than testing that the disconnect should be
        # asked to ConnectedTransport only.
        pass

    def _redirected_to(self, source, target):
        """Returns a transport suitable to re-issue a redirected request.

        :param source: The source url as returned by the server.
        :param target: The target url as returned by the server.

        The redirection can be handled only if the relpath involved is not
        renamed by the redirection.

        :returns: A transport or None.
        """
        # This returns None by default, meaning the transport can't handle the
        # redirection.
        return None



class _SharedConnection(object):
    """A connection shared between several transports."""

    def __init__(self, connection=None, credentials=None, base=None):
        """Constructor.

        :param connection: An opaque object specific to each transport.

        :param credentials: An opaque object containing the credentials used to
            create the connection.
        """
        self.connection = connection
        self.credentials = credentials
        self.base = base


class ConnectedTransport(Transport):
    """A transport connected to a remote server.

    This class provide the basis to implement transports that need to connect
    to a remote server.

    Host and credentials are available as private attributes, cloning preserves
    them and share the underlying, protocol specific, connection.
    """

    def __init__(self, base, _from_transport=None):
        """Constructor.

        The caller should ensure that _from_transport points at the same host
        as the new base.

        :param base: transport root URL

        :param _from_transport: optional transport to build from. The built
            transport will share the connection with this transport.
        """
        if not base.endswith('/'):
            base += '/'
        (self._scheme,
         self._user, self._password,
         self._host, self._port,
         self._path) = self._split_url(base)
        if _from_transport is not None:
            # Copy the password as it does not appear in base and will be lost
            # otherwise. It can appear in the _split_url above if the user
            # provided it on the command line. Otherwise, daughter classes will
            # prompt the user for one when appropriate.
            self._password = _from_transport._password

        base = self._unsplit_url(self._scheme,
                                 self._user, self._password,
                                 self._host, self._port,
                                 self._path)

        super(ConnectedTransport, self).__init__(base)
        if _from_transport is None:
            self._shared_connection = _SharedConnection()
        else:
            self._shared_connection = _from_transport._shared_connection

    def clone(self, offset=None):
        """Return a new transport with root at self.base + offset

        We leave the daughter classes take advantage of the hint
        that it's a cloning not a raw creation.
        """
        if offset is None:
            return self.__class__(self.base, _from_transport=self)
        else:
            return self.__class__(self.abspath(offset), _from_transport=self)

    @staticmethod
    def _split_url(url):
        return urlutils.parse_url(url)

    @staticmethod
    def _unsplit_url(scheme, user, password, host, port, path):
        """
        Build the full URL for the given already URL encoded path.

        user, password, host and path will be quoted if they contain reserved
        chars.

        :param scheme: protocol

        :param user: login

        :param password: associated password

        :param host: the server address

        :param port: the associated port

        :param path: the absolute path on the server

        :return: The corresponding URL.
        """
        netloc = urllib.quote(host)
        if user is not None:
            # Note that we don't put the password back even if we
            # have one so that it doesn't get accidentally
            # exposed.
            netloc = '%s@%s' % (urllib.quote(user), netloc)
        if port is not None:
            netloc = '%s:%d' % (netloc, port)
        path = urlutils.escape(path)
        return urlparse.urlunparse((scheme, netloc, path, None, None, None))

    def relpath(self, abspath):
        """Return the local path portion from a given absolute path"""
        scheme, user, password, host, port, path = self._split_url(abspath)
        error = []
        if (scheme != self._scheme):
            error.append('scheme mismatch')
        if (user != self._user):
            error.append('user name mismatch')
        if (host != self._host):
            error.append('host mismatch')
        if (port != self._port):
            error.append('port mismatch')
        if not (path == self._path[:-1] or path.startswith(self._path)):
            error.append('path mismatch')
        if error:
            extra = ', '.join(error)
            raise errors.PathNotChild(abspath, self.base, extra=extra)
        pl = len(self._path)
        return path[pl:].strip('/')

    def abspath(self, relpath):
        """Return the full url to the given relative path.

        :param relpath: the relative path urlencoded

        :returns: the Unicode version of the absolute path for relpath.
        """
        relative = urlutils.unescape(relpath).encode('utf-8')
        path = self._combine_paths(self._path, relative)
        return self._unsplit_url(self._scheme, self._user, self._password,
                                 self._host, self._port,
                                 path)

    def _remote_path(self, relpath):
        """Return the absolute path part of the url to the given relative path.

        This is the path that the remote server expect to receive in the
        requests, daughter classes should redefine this method if needed and
        use the result to build their requests.

        :param relpath: the path relative to the transport base urlencoded.

        :return: the absolute Unicode path on the server,
        """
        relative = urlutils.unescape(relpath).encode('utf-8')
        remote_path = self._combine_paths(self._path, relative)
        return remote_path

    def _get_shared_connection(self):
        """Get the object shared amongst cloned transports.

        This should be used only by classes that needs to extend the sharing
        with objects other than transports.

        Use _get_connection to get the connection itself.
        """
        return self._shared_connection

    def _set_connection(self, connection, credentials=None):
        """Record a newly created connection with its associated credentials.

        Note: To ensure that connection is still shared after a temporary
        failure and a new one needs to be created, daughter classes should
        always call this method to set the connection and do so each time a new
        connection is created.

        :param connection: An opaque object representing the connection used by
            the daughter class.

        :param credentials: An opaque object representing the credentials
            needed to create the connection.
        """
        self._shared_connection.connection = connection
        self._shared_connection.credentials = credentials

    def _get_connection(self):
        """Returns the transport specific connection object."""
        return self._shared_connection.connection

    def _get_credentials(self):
        """Returns the credentials used to establish the connection."""
        return self._shared_connection.credentials

    def _update_credentials(self, credentials):
        """Update the credentials of the current connection.

        Some protocols can renegociate the credentials within a connection,
        this method allows daughter classes to share updated credentials.

        :param credentials: the updated credentials.
        """
        # We don't want to call _set_connection here as we are only updating
        # the credentials not creating a new connection.
        self._shared_connection.credentials = credentials

    def _reuse_for(self, other_base):
        """Returns a transport sharing the same connection if possible.

        Note: we share the connection if the expected credentials are the
        same: (host, port, user). Some protocols may disagree and redefine the
        criteria in daughter classes.

        Note: we don't compare the passwords here because other_base may have
        been obtained from an existing transport.base which do not mention the
        password.

        :param other_base: the URL we want to share the connection with.

        :return: A new transport or None if the connection cannot be shared.
        """
        try:
            (scheme, user, password,
             host, port, path) = self._split_url(other_base)
        except errors.InvalidURL:
            # No hope in trying to reuse an existing transport for an invalid
            # URL
            return None

        transport = None
        # Don't compare passwords, they may be absent from other_base or from
        # self and they don't carry more information than user anyway.
        if (scheme == self._scheme
            and user == self._user
            and host == self._host
            and port == self._port):
            if not path.endswith('/'):
                # This normally occurs at __init__ time, but it's easier to do
                # it now to avoid creating two transports for the same base.
                path += '/'
            if self._path  == path:
                # shortcut, it's really the same transport
                return self
            # We don't call clone here because the intent is different: we
            # build a new transport on a different base (which may be totally
            # unrelated) but we share the connection.
            transport = self.__class__(other_base, _from_transport=self)
        return transport

    def disconnect(self):
        """Disconnect the transport.

        If and when required the transport willl reconnect automatically.
        """
        raise NotImplementedError(self.disconnect)


def get_transport(base, possible_transports=None):
    """Open a transport to access a URL or directory.

    :param base: either a URL or a directory name.

    :param transports: optional reusable transports list. If not None, created
        transports will be added to the list.

    :return: A new transport optionally sharing its connection with one of
        possible_transports.
    """
    if base is None:
        base = '.'
    last_err = None
    from bzrlib.directory_service import directories
    base = directories.dereference(base)

    def convert_path_to_url(base, error_str):
        if urlutils.is_url(base):
            # This looks like a URL, but we weren't able to
            # instantiate it as such raise an appropriate error
            # FIXME: we have a 'error_str' unused and we use last_err below
            raise errors.UnsupportedProtocol(base, last_err)
        # This doesn't look like a protocol, consider it a local path
        new_base = urlutils.local_path_to_url(base)
        # mutter('converting os path %r => url %s', base, new_base)
        return new_base

    # Catch any URLs which are passing Unicode rather than ASCII
    try:
        base = base.encode('ascii')
    except UnicodeError:
        # Only local paths can be Unicode
        base = convert_path_to_url(base,
            'URLs must be properly escaped (protocol: %s)')

    transport = None
    if possible_transports is not None:
        for t in possible_transports:
            t_same_connection = t._reuse_for(base)
            if t_same_connection is not None:
                # Add only new transports
                if t_same_connection not in possible_transports:
                    possible_transports.append(t_same_connection)
                return t_same_connection

    for proto, factory_list in transport_list_registry.items():
        if proto is not None and base.startswith(proto):
            transport, last_err = _try_transport_factories(base, factory_list)
            if transport:
                if possible_transports is not None:
                    if transport in possible_transports:
                        raise AssertionError()
                    possible_transports.append(transport)
                return transport

    # We tried all the different protocols, now try one last time
    # as a local protocol
    base = convert_path_to_url(base, 'Unsupported protocol: %s')

    # The default handler is the filesystem handler, stored as protocol None
    factory_list = transport_list_registry.get(None)
    transport, last_err = _try_transport_factories(base, factory_list)

    return transport


def _try_transport_factories(base, factory_list):
    last_err = None
    for factory in factory_list:
        try:
            return factory.get_obj()(base), None
        except errors.DependencyNotPresent, e:
            mutter("failed to instantiate transport %r for %r: %r" %
                    (factory, base, e))
            last_err = e
            continue
    return None, last_err


def do_catching_redirections(action, transport, redirected):
    """Execute an action with given transport catching redirections.

    This is a facility provided for callers needing to follow redirections
    silently. The silence is relative: it is the caller responsability to
    inform the user about each redirection or only inform the user of a user
    via the exception parameter.

    :param action: A callable, what the caller want to do while catching
                  redirections.
    :param transport: The initial transport used.
    :param redirected: A callable receiving the redirected transport and the
                  RedirectRequested exception.

    :return: Whatever 'action' returns
    """
    MAX_REDIRECTIONS = 8

    # If a loop occurs, there is little we can do. So we don't try to detect
    # them, just getting out if too much redirections occurs. The solution
    # is outside: where the loop is defined.
    for redirections in range(MAX_REDIRECTIONS):
        try:
            return action(transport)
        except errors.RedirectRequested, e:
            redirection_notice = '%s is%s redirected to %s' % (
                e.source, e.permanently, e.target)
            transport = redirected(transport, e, redirection_notice)
    else:
        # Loop exited without resolving redirect ? Either the
        # user has kept a very very very old reference or a loop
        # occurred in the redirections.  Nothing we can cure here:
        # tell the user. Note that as the user has been informed
        # about each redirection (it is the caller responsibility
        # to do that in redirected via the provided
        # redirection_notice). The caller may provide more
        # information if needed (like what file or directory we
        # were trying to act upon when the redirection loop
        # occurred).
        raise errors.TooManyRedirections


class Server(object):
    """A Transport Server.

    The Server interface provides a server for a given transport type.
    """

    def start_server(self):
        """Setup the server to service requests."""

    def stop_server(self):
        """Remove the server and cleanup any resources it owns."""


# None is the default transport, for things with no url scheme
register_transport_proto('file://',
            help="Access using the standard filesystem (default)")
register_lazy_transport('file://', 'bzrlib.transport.local', 'LocalTransport')
transport_list_registry.set_default_transport("file://")

register_transport_proto('sftp://',
            help="Access using SFTP (most SSH servers provide SFTP).",
            register_netloc=True)
register_lazy_transport('sftp://', 'bzrlib.transport.sftp', 'SFTPTransport')
# Decorated http transport
register_transport_proto('http+urllib://',
#                help="Read-only access of branches exported on the web."
                         register_netloc=True)
register_lazy_transport('http+urllib://', 'bzrlib.transport.http._urllib',
                        'HttpTransport_urllib')
register_transport_proto('https+urllib://',
#                help="Read-only access of branches exported on the web using SSL."
                         register_netloc=True)
register_lazy_transport('https+urllib://', 'bzrlib.transport.http._urllib',
                        'HttpTransport_urllib')
register_transport_proto('http+pycurl://',
#                help="Read-only access of branches exported on the web."
                         register_netloc=True)
register_lazy_transport('http+pycurl://', 'bzrlib.transport.http._pycurl',
                        'PyCurlTransport')
register_transport_proto('https+pycurl://',
#                help="Read-only access of branches exported on the web using SSL."
                         register_netloc=True)
register_lazy_transport('https+pycurl://', 'bzrlib.transport.http._pycurl',
                        'PyCurlTransport')
# Default http transports (last declared wins (if it can be imported))
register_transport_proto('http://',
                 help="Read-only access of branches exported on the web.")
register_transport_proto('https://',
            help="Read-only access of branches exported on the web using SSL.")
# The default http implementation is urllib, but https is pycurl if available
register_lazy_transport('http://', 'bzrlib.transport.http._pycurl',
                        'PyCurlTransport')
register_lazy_transport('http://', 'bzrlib.transport.http._urllib',
                        'HttpTransport_urllib')
register_lazy_transport('https://', 'bzrlib.transport.http._urllib',
                        'HttpTransport_urllib')
register_lazy_transport('https://', 'bzrlib.transport.http._pycurl',
                        'PyCurlTransport')

register_transport_proto('ftp://', help="Access using passive FTP.")
register_lazy_transport('ftp://', 'bzrlib.transport.ftp', 'FtpTransport')
register_transport_proto('aftp://', help="Access using active FTP.")
register_lazy_transport('aftp://', 'bzrlib.transport.ftp', 'FtpTransport')
register_transport_proto('gio+', help="Access using any GIO supported protocols.")
register_lazy_transport('gio+', 'bzrlib.transport.gio_transport', 'GioTransport')


# Default to trying GSSAPI authentication (if the kerberos module is
# available)
register_transport_proto('ftp+gssapi://', register_netloc=True)
register_transport_proto('aftp+gssapi://', register_netloc=True)
register_transport_proto('ftp+nogssapi://', register_netloc=True)
register_transport_proto('aftp+nogssapi://', register_netloc=True)
register_lazy_transport('ftp+gssapi://', 'bzrlib.transport.ftp._gssapi',
                        'GSSAPIFtpTransport')
register_lazy_transport('aftp+gssapi://', 'bzrlib.transport.ftp._gssapi',
                        'GSSAPIFtpTransport')
register_lazy_transport('ftp://', 'bzrlib.transport.ftp._gssapi',
                        'GSSAPIFtpTransport')
register_lazy_transport('aftp://', 'bzrlib.transport.ftp._gssapi',
                        'GSSAPIFtpTransport')
register_lazy_transport('ftp+nogssapi://', 'bzrlib.transport.ftp',
                        'FtpTransport')
register_lazy_transport('aftp+nogssapi://', 'bzrlib.transport.ftp',
                        'FtpTransport')

register_transport_proto('memory://')
register_lazy_transport('memory://', 'bzrlib.transport.memory',
                        'MemoryTransport')

# chroots cannot be implicitly accessed, they must be explicitly created:
register_transport_proto('chroot+')

register_transport_proto('readonly+',
#              help="This modifier converts any transport to be readonly."
            )
register_lazy_transport('readonly+', 'bzrlib.transport.readonly',
                        'ReadonlyTransportDecorator')

register_transport_proto('fakenfs+')
register_lazy_transport('fakenfs+', 'bzrlib.transport.fakenfs',
                        'FakeNFSTransportDecorator')

register_transport_proto('log+')
register_lazy_transport('log+', 'bzrlib.transport.log', 'TransportLogDecorator')

register_transport_proto('trace+')
register_lazy_transport('trace+', 'bzrlib.transport.trace',
                        'TransportTraceDecorator')

register_transport_proto('unlistable+')
register_lazy_transport('unlistable+', 'bzrlib.transport.unlistable',
                        'UnlistableTransportDecorator')

register_transport_proto('brokenrename+')
register_lazy_transport('brokenrename+', 'bzrlib.transport.brokenrename',
                        'BrokenRenameTransportDecorator')

register_transport_proto('vfat+')
register_lazy_transport('vfat+',
                        'bzrlib.transport.fakevfat',
                        'FakeVFATTransportDecorator')

register_transport_proto('nosmart+')
register_lazy_transport('nosmart+', 'bzrlib.transport.nosmart',
                        'NoSmartTransportDecorator')

# These two schemes were registered, but don't seem to have an actual transport
# protocol registered
for scheme in ['ssh', 'bzr+loopback']:
    register_urlparse_netloc_protocol(scheme)
del scheme

register_transport_proto('bzr://',
            help="Fast access using the Bazaar smart server.",
                         register_netloc=True)

register_lazy_transport('bzr://', 'bzrlib.transport.remote',
                        'RemoteTCPTransport')
register_transport_proto('bzr-v2://', register_netloc=True)

register_lazy_transport('bzr-v2://', 'bzrlib.transport.remote',
                        'RemoteTCPTransportV2Only')
register_transport_proto('bzr+http://',
#                help="Fast access using the Bazaar smart server over HTTP."
                         register_netloc=True)
register_lazy_transport('bzr+http://', 'bzrlib.transport.remote',
                        'RemoteHTTPTransport')
register_transport_proto('bzr+https://',
#                help="Fast access using the Bazaar smart server over HTTPS."
                         register_netloc=True)
register_lazy_transport('bzr+https://',
                        'bzrlib.transport.remote',
                        'RemoteHTTPTransport')
register_transport_proto('bzr+ssh://',
            help="Fast access using the Bazaar smart server over SSH.",
            register_netloc=True)
register_lazy_transport('bzr+ssh://', 'bzrlib.transport.remote',
                        'RemoteSSHTransport')

register_transport_proto('ssh:')
register_lazy_transport('ssh:', 'bzrlib.transport.remote',
                        'HintingSSHTransport')


transport_server_registry = registry.Registry()
transport_server_registry.register_lazy('bzr', 'bzrlib.smart.server',
    'serve_bzr', help="The Bazaar smart server protocol over TCP. (default port: 4155)")
transport_server_registry.default_key = 'bzr'