~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
# Copyright (C) 2007-2011 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

"""Knit-based pack repository formats."""

from bzrlib.lazy_import import lazy_import
lazy_import(globals(), """
from itertools import izip
import time

from bzrlib import (
    bzrdir,
    debug,
    errors,
    knit,
    osutils,
    pack,
    revision as _mod_revision,
    trace,
    tsort,
    ui,
    xml5,
    xml6,
    xml7,
    )
from bzrlib.knit import (
    _KnitGraphIndex,
    KnitPlainFactory,
    KnitVersionedFiles,
    )
""")

from bzrlib import (
    btree_index,
    )
from bzrlib.index import (
    CombinedGraphIndex,
    GraphIndex,
    GraphIndexPrefixAdapter,
    InMemoryGraphIndex,
    )
from bzrlib.repofmt.knitrepo import (
    KnitRepository,
    )
from bzrlib.repofmt.pack_repo import (
    _DirectPackAccess,
    NewPack,
    RepositoryFormatPack,
    ResumedPack,
    Packer,
    PackCommitBuilder,
    PackRepository,
    PackRootCommitBuilder,
    RepositoryPackCollection,
    )
from bzrlib.repository import (
    StreamSource,
    )


class KnitPackRepository(PackRepository, KnitRepository):

    def __init__(self, _format, a_bzrdir, control_files, _commit_builder_class,
        _serializer):
        PackRepository.__init__(self, _format, a_bzrdir, control_files,
            _commit_builder_class, _serializer)
        if self._format.supports_chks:
            raise AssertionError("chk not supported")
        index_transport = self._transport.clone('indices')
        self._pack_collection = KnitRepositoryPackCollection(self,
            self._transport,
            index_transport,
            self._transport.clone('upload'),
            self._transport.clone('packs'),
            _format.index_builder_class,
            _format.index_class,
            use_chk_index=False,
            )
        self.inventories = KnitVersionedFiles(
            _KnitGraphIndex(self._pack_collection.inventory_index.combined_index,
                add_callback=self._pack_collection.inventory_index.add_callback,
                deltas=True, parents=True, is_locked=self.is_locked),
            data_access=self._pack_collection.inventory_index.data_access,
            max_delta_chain=200)
        self.revisions = KnitVersionedFiles(
            _KnitGraphIndex(self._pack_collection.revision_index.combined_index,
                add_callback=self._pack_collection.revision_index.add_callback,
                deltas=False, parents=True, is_locked=self.is_locked,
                track_external_parent_refs=True),
            data_access=self._pack_collection.revision_index.data_access,
            max_delta_chain=0)
        self.signatures = KnitVersionedFiles(
            _KnitGraphIndex(self._pack_collection.signature_index.combined_index,
                add_callback=self._pack_collection.signature_index.add_callback,
                deltas=False, parents=False, is_locked=self.is_locked),
            data_access=self._pack_collection.signature_index.data_access,
            max_delta_chain=0)
        self.texts = KnitVersionedFiles(
            _KnitGraphIndex(self._pack_collection.text_index.combined_index,
                add_callback=self._pack_collection.text_index.add_callback,
                deltas=True, parents=True, is_locked=self.is_locked),
            data_access=self._pack_collection.text_index.data_access,
            max_delta_chain=200)
        self.chk_bytes = None
        # True when the repository object is 'write locked' (as opposed to the
        # physical lock only taken out around changes to the pack-names list.)
        # Another way to represent this would be a decorator around the control
        # files object that presents logical locks as physical ones - if this
        # gets ugly consider that alternative design. RBC 20071011
        self._write_lock_count = 0
        self._transaction = None
        # for tests
        self._reconcile_does_inventory_gc = True
        self._reconcile_fixes_text_parents = True
        self._reconcile_backsup_inventory = False

    def _get_source(self, to_format):
        if to_format.network_name() == self._format.network_name():
            return KnitPackStreamSource(self, to_format)
        return PackRepository._get_source(self, to_format)

    def _reconcile_pack(self, collection, packs, extension, revs, pb):
        packer = KnitReconcilePacker(collection, packs, extension, revs)
        return packer.pack(pb)


class RepositoryFormatKnitPack1(RepositoryFormatPack):
    """A no-subtrees parameterized Pack repository.

    This format was introduced in 0.92.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackCommitBuilder
    @property
    def _serializer(self):
        return xml5.serializer_v5
    # What index classes to use
    index_builder_class = InMemoryGraphIndex
    index_class = GraphIndex

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir('pack-0.92')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar pack repository format 1 (needs bzr 0.92)\n"

    def get_format_description(self):
        """See RepositoryFormat.get_format_description()."""
        return "Packs containing knits without subtree support"


class RepositoryFormatKnitPack3(RepositoryFormatPack):
    """A subtrees parameterized Pack repository.

    This repository format uses the xml7 serializer to get:
     - support for recording full info about the tree root
     - support for recording tree-references

    This format was introduced in 0.92.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackRootCommitBuilder
    rich_root_data = True
    experimental = True
    supports_tree_reference = True
    @property
    def _serializer(self):
        return xml7.serializer_v7
    # What index classes to use
    index_builder_class = InMemoryGraphIndex
    index_class = GraphIndex

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir(
            'pack-0.92-subtree')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar pack repository format 1 with subtree support (needs bzr 0.92)\n"

    def get_format_description(self):
        """See RepositoryFormat.get_format_description()."""
        return "Packs containing knits with subtree support\n"


class RepositoryFormatKnitPack4(RepositoryFormatPack):
    """A rich-root, no subtrees parameterized Pack repository.

    This repository format uses the xml6 serializer to get:
     - support for recording full info about the tree root

    This format was introduced in 1.0.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackRootCommitBuilder
    rich_root_data = True
    supports_tree_reference = False
    @property
    def _serializer(self):
        return xml6.serializer_v6
    # What index classes to use
    index_builder_class = InMemoryGraphIndex
    index_class = GraphIndex

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir(
            'rich-root-pack')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return ("Bazaar pack repository format 1 with rich root"
                " (needs bzr 1.0)\n")

    def get_format_description(self):
        """See RepositoryFormat.get_format_description()."""
        return "Packs containing knits with rich root support\n"


class RepositoryFormatKnitPack5(RepositoryFormatPack):
    """Repository that supports external references to allow stacking.

    New in release 1.6.

    Supports external lookups, which results in non-truncated ghosts after
    reconcile compared to pack-0.92 formats.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackCommitBuilder
    supports_external_lookups = True
    # What index classes to use
    index_builder_class = InMemoryGraphIndex
    index_class = GraphIndex

    @property
    def _serializer(self):
        return xml5.serializer_v5

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir('1.6')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar RepositoryFormatKnitPack5 (bzr 1.6)\n"

    def get_format_description(self):
        """See RepositoryFormat.get_format_description()."""
        return "Packs 5 (adds stacking support, requires bzr 1.6)"


class RepositoryFormatKnitPack5RichRoot(RepositoryFormatPack):
    """A repository with rich roots and stacking.

    New in release 1.6.1.

    Supports stacking on other repositories, allowing data to be accessed
    without being stored locally.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackRootCommitBuilder
    rich_root_data = True
    supports_tree_reference = False # no subtrees
    supports_external_lookups = True
    # What index classes to use
    index_builder_class = InMemoryGraphIndex
    index_class = GraphIndex

    @property
    def _serializer(self):
        return xml6.serializer_v6

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir(
            '1.6.1-rich-root')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6.1)\n"

    def get_format_description(self):
        return "Packs 5 rich-root (adds stacking support, requires bzr 1.6.1)"


class RepositoryFormatKnitPack5RichRootBroken(RepositoryFormatPack):
    """A repository with rich roots and external references.

    New in release 1.6.

    Supports external lookups, which results in non-truncated ghosts after
    reconcile compared to pack-0.92 formats.

    This format was deprecated because the serializer it uses accidentally
    supported subtrees, when the format was not intended to. This meant that
    someone could accidentally fetch from an incorrect repository.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackRootCommitBuilder
    rich_root_data = True
    supports_tree_reference = False # no subtrees

    supports_external_lookups = True
    # What index classes to use
    index_builder_class = InMemoryGraphIndex
    index_class = GraphIndex

    @property
    def _serializer(self):
        return xml7.serializer_v7

    def _get_matching_bzrdir(self):
        matching = bzrdir.format_registry.make_bzrdir(
            '1.6.1-rich-root')
        matching.repository_format = self
        return matching

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar RepositoryFormatKnitPack5RichRoot (bzr 1.6)\n"

    def get_format_description(self):
        return ("Packs 5 rich-root (adds stacking support, requires bzr 1.6)"
                " (deprecated)")

    def is_deprecated(self):
        return True


class RepositoryFormatKnitPack6(RepositoryFormatPack):
    """A repository with stacking and btree indexes,
    without rich roots or subtrees.

    This is equivalent to pack-1.6 with B+Tree indices.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackCommitBuilder
    supports_external_lookups = True
    # What index classes to use
    index_builder_class = btree_index.BTreeBuilder
    index_class = btree_index.BTreeGraphIndex

    @property
    def _serializer(self):
        return xml5.serializer_v5

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir('1.9')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar RepositoryFormatKnitPack6 (bzr 1.9)\n"

    def get_format_description(self):
        """See RepositoryFormat.get_format_description()."""
        return "Packs 6 (uses btree indexes, requires bzr 1.9)"


class RepositoryFormatKnitPack6RichRoot(RepositoryFormatPack):
    """A repository with rich roots, no subtrees, stacking and btree indexes.

    1.6-rich-root with B+Tree indices.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackRootCommitBuilder
    rich_root_data = True
    supports_tree_reference = False # no subtrees
    supports_external_lookups = True
    # What index classes to use
    index_builder_class = btree_index.BTreeBuilder
    index_class = btree_index.BTreeGraphIndex

    @property
    def _serializer(self):
        return xml6.serializer_v6

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir(
            '1.9-rich-root')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return "Bazaar RepositoryFormatKnitPack6RichRoot (bzr 1.9)\n"

    def get_format_description(self):
        return "Packs 6 rich-root (uses btree indexes, requires bzr 1.9)"


class RepositoryFormatPackDevelopment2Subtree(RepositoryFormatPack):
    """A subtrees development repository.

    This format should be retained in 2.3, to provide an upgrade path from this
    to RepositoryFormat2aSubtree.  It can be removed in later releases.

    1.6.1-subtree[as it might have been] with B+Tree indices.
    """

    repository_class = KnitPackRepository
    _commit_builder_class = PackRootCommitBuilder
    rich_root_data = True
    experimental = True
    supports_tree_reference = True
    supports_external_lookups = True
    # What index classes to use
    index_builder_class = btree_index.BTreeBuilder
    index_class = btree_index.BTreeGraphIndex

    @property
    def _serializer(self):
        return xml7.serializer_v7

    def _get_matching_bzrdir(self):
        return bzrdir.format_registry.make_bzrdir(
            'development5-subtree')

    def _ignore_setting_bzrdir(self, format):
        pass

    _matchingbzrdir = property(_get_matching_bzrdir, _ignore_setting_bzrdir)

    def get_format_string(self):
        """See RepositoryFormat.get_format_string()."""
        return ("Bazaar development format 2 with subtree support "
            "(needs bzr.dev from before 1.8)\n")

    def get_format_description(self):
        """See RepositoryFormat.get_format_description()."""
        return ("Development repository format, currently the same as "
            "1.6.1-subtree with B+Tree indices.\n")


class KnitPackStreamSource(StreamSource):
    """A StreamSource used to transfer data between same-format KnitPack repos.

    This source assumes:
        1) Same serialization format for all objects
        2) Same root information
        3) XML format inventories
        4) Atomic inserts (so we can stream inventory texts before text
           content)
        5) No chk_bytes
    """

    def __init__(self, from_repository, to_format):
        super(KnitPackStreamSource, self).__init__(from_repository, to_format)
        self._text_keys = None
        self._text_fetch_order = 'unordered'

    def _get_filtered_inv_stream(self, revision_ids):
        from_repo = self.from_repository
        parent_ids = from_repo._find_parent_ids_of_revisions(revision_ids)
        parent_keys = [(p,) for p in parent_ids]
        find_text_keys = from_repo._serializer._find_text_key_references
        parent_text_keys = set(find_text_keys(
            from_repo._inventory_xml_lines_for_keys(parent_keys)))
        content_text_keys = set()
        knit = KnitVersionedFiles(None, None)
        factory = KnitPlainFactory()
        def find_text_keys_from_content(record):
            if record.storage_kind not in ('knit-delta-gz', 'knit-ft-gz'):
                raise ValueError("Unknown content storage kind for"
                    " inventory text: %s" % (record.storage_kind,))
            # It's a knit record, it has a _raw_record field (even if it was
            # reconstituted from a network stream).
            raw_data = record._raw_record
            # read the entire thing
            revision_id = record.key[-1]
            content, _ = knit._parse_record(revision_id, raw_data)
            if record.storage_kind == 'knit-delta-gz':
                line_iterator = factory.get_linedelta_content(content)
            elif record.storage_kind == 'knit-ft-gz':
                line_iterator = factory.get_fulltext_content(content)
            content_text_keys.update(find_text_keys(
                [(line, revision_id) for line in line_iterator]))
        revision_keys = [(r,) for r in revision_ids]
        def _filtered_inv_stream():
            source_vf = from_repo.inventories
            stream = source_vf.get_record_stream(revision_keys,
                                                 'unordered', False)
            for record in stream:
                if record.storage_kind == 'absent':
                    raise errors.NoSuchRevision(from_repo, record.key)
                find_text_keys_from_content(record)
                yield record
            self._text_keys = content_text_keys - parent_text_keys
        return ('inventories', _filtered_inv_stream())

    def _get_text_stream(self):
        # Note: We know we don't have to handle adding root keys, because both
        # the source and target are the identical network name.
        text_stream = self.from_repository.texts.get_record_stream(
                        self._text_keys, self._text_fetch_order, False)
        return ('texts', text_stream)

    def get_stream(self, search):
        revision_ids = search.get_keys()
        for stream_info in self._fetch_revision_texts(revision_ids):
            yield stream_info
        self._revision_keys = [(rev_id,) for rev_id in revision_ids]
        yield self._get_filtered_inv_stream(revision_ids)
        yield self._get_text_stream()


class KnitPacker(Packer):
    """Packer that works with knit packs."""

    def __init__(self, pack_collection, packs, suffix, revision_ids=None,
                 reload_func=None):
        super(KnitPacker, self).__init__(pack_collection, packs, suffix,
                                          revision_ids=revision_ids,
                                          reload_func=reload_func)

    def _pack_map_and_index_list(self, index_attribute):
        """Convert a list of packs to an index pack map and index list.

        :param index_attribute: The attribute that the desired index is found
            on.
        :return: A tuple (map, list) where map contains the dict from
            index:pack_tuple, and list contains the indices in the preferred
            access order.
        """
        indices = []
        pack_map = {}
        for pack_obj in self.packs:
            index = getattr(pack_obj, index_attribute)
            indices.append(index)
            pack_map[index] = pack_obj
        return pack_map, indices

    def _index_contents(self, indices, key_filter=None):
        """Get an iterable of the index contents from a pack_map.

        :param indices: The list of indices to query
        :param key_filter: An optional filter to limit the keys returned.
        """
        all_index = CombinedGraphIndex(indices)
        if key_filter is None:
            return all_index.iter_all_entries()
        else:
            return all_index.iter_entries(key_filter)

    def _copy_nodes(self, nodes, index_map, writer, write_index,
        output_lines=None):
        """Copy knit nodes between packs with no graph references.

        :param output_lines: Output full texts of copied items.
        """
        pb = ui.ui_factory.nested_progress_bar()
        try:
            return self._do_copy_nodes(nodes, index_map, writer,
                write_index, pb, output_lines=output_lines)
        finally:
            pb.finished()

    def _do_copy_nodes(self, nodes, index_map, writer, write_index, pb,
        output_lines=None):
        # for record verification
        knit = KnitVersionedFiles(None, None)
        # plan a readv on each source pack:
        # group by pack
        nodes = sorted(nodes)
        # how to map this into knit.py - or knit.py into this?
        # we don't want the typical knit logic, we want grouping by pack
        # at this point - perhaps a helper library for the following code
        # duplication points?
        request_groups = {}
        for index, key, value in nodes:
            if index not in request_groups:
                request_groups[index] = []
            request_groups[index].append((key, value))
        record_index = 0
        pb.update("Copied record", record_index, len(nodes))
        for index, items in request_groups.iteritems():
            pack_readv_requests = []
            for key, value in items:
                # ---- KnitGraphIndex.get_position
                bits = value[1:].split(' ')
                offset, length = int(bits[0]), int(bits[1])
                pack_readv_requests.append((offset, length, (key, value[0])))
            # linear scan up the pack
            pack_readv_requests.sort()
            # copy the data
            pack_obj = index_map[index]
            transport, path = pack_obj.access_tuple()
            try:
                reader = pack.make_readv_reader(transport, path,
                    [offset[0:2] for offset in pack_readv_requests])
            except errors.NoSuchFile:
                if self._reload_func is not None:
                    self._reload_func()
                raise
            for (names, read_func), (_1, _2, (key, eol_flag)) in \
                izip(reader.iter_records(), pack_readv_requests):
                raw_data = read_func(None)
                # check the header only
                if output_lines is not None:
                    output_lines(knit._parse_record(key[-1], raw_data)[0])
                else:
                    df, _ = knit._parse_record_header(key, raw_data)
                    df.close()
                pos, size = writer.add_bytes_record(raw_data, names)
                write_index.add_node(key, eol_flag + "%d %d" % (pos, size))
                pb.update("Copied record", record_index)
                record_index += 1

    def _copy_nodes_graph(self, index_map, writer, write_index,
        readv_group_iter, total_items, output_lines=False):
        """Copy knit nodes between packs.

        :param output_lines: Return lines present in the copied data as
            an iterator of line,version_id.
        """
        pb = ui.ui_factory.nested_progress_bar()
        try:
            for result in self._do_copy_nodes_graph(index_map, writer,
                write_index, output_lines, pb, readv_group_iter, total_items):
                yield result
        except Exception:
            # Python 2.4 does not permit try:finally: in a generator.
            pb.finished()
            raise
        else:
            pb.finished()

    def _do_copy_nodes_graph(self, index_map, writer, write_index,
        output_lines, pb, readv_group_iter, total_items):
        # for record verification
        knit = KnitVersionedFiles(None, None)
        # for line extraction when requested (inventories only)
        if output_lines:
            factory = KnitPlainFactory()
        record_index = 0
        pb.update("Copied record", record_index, total_items)
        for index, readv_vector, node_vector in readv_group_iter:
            # copy the data
            pack_obj = index_map[index]
            transport, path = pack_obj.access_tuple()
            try:
                reader = pack.make_readv_reader(transport, path, readv_vector)
            except errors.NoSuchFile:
                if self._reload_func is not None:
                    self._reload_func()
                raise
            for (names, read_func), (key, eol_flag, references) in \
                izip(reader.iter_records(), node_vector):
                raw_data = read_func(None)
                if output_lines:
                    # read the entire thing
                    content, _ = knit._parse_record(key[-1], raw_data)
                    if len(references[-1]) == 0:
                        line_iterator = factory.get_fulltext_content(content)
                    else:
                        line_iterator = factory.get_linedelta_content(content)
                    for line in line_iterator:
                        yield line, key
                else:
                    # check the header only
                    df, _ = knit._parse_record_header(key, raw_data)
                    df.close()
                pos, size = writer.add_bytes_record(raw_data, names)
                write_index.add_node(key, eol_flag + "%d %d" % (pos, size), references)
                pb.update("Copied record", record_index)
                record_index += 1

    def _process_inventory_lines(self, inv_lines):
        """Use up the inv_lines generator and setup a text key filter."""
        repo = self._pack_collection.repo
        fileid_revisions = repo._find_file_ids_from_xml_inventory_lines(
            inv_lines, self.revision_keys)
        text_filter = []
        for fileid, file_revids in fileid_revisions.iteritems():
            text_filter.extend([(fileid, file_revid) for file_revid in file_revids])
        self._text_filter = text_filter

    def _copy_inventory_texts(self):
        # select inventory keys
        inv_keys = self._revision_keys # currently the same keyspace, and note that
        # querying for keys here could introduce a bug where an inventory item
        # is missed, so do not change it to query separately without cross
        # checking like the text key check below.
        inventory_index_map, inventory_indices = self._pack_map_and_index_list(
            'inventory_index')
        inv_nodes = self._index_contents(inventory_indices, inv_keys)
        # copy inventory keys and adjust values
        # XXX: Should be a helper function to allow different inv representation
        # at this point.
        self.pb.update("Copying inventory texts", 2)
        total_items, readv_group_iter = self._least_readv_node_readv(inv_nodes)
        # Only grab the output lines if we will be processing them
        output_lines = bool(self.revision_ids)
        inv_lines = self._copy_nodes_graph(inventory_index_map,
            self.new_pack._writer, self.new_pack.inventory_index,
            readv_group_iter, total_items, output_lines=output_lines)
        if self.revision_ids:
            self._process_inventory_lines(inv_lines)
        else:
            # eat the iterator to cause it to execute.
            list(inv_lines)
            self._text_filter = None
        if 'pack' in debug.debug_flags:
            trace.mutter('%s: create_pack: inventories copied: %s%s %d items t+%6.3fs',
                time.ctime(), self._pack_collection._upload_transport.base,
                self.new_pack.random_name,
                self.new_pack.inventory_index.key_count(),
                time.time() - self.new_pack.start_time)

    def _update_pack_order(self, entries, index_to_pack_map):
        """Determine how we want our packs to be ordered.

        This changes the sort order of the self.packs list so that packs unused
        by 'entries' will be at the end of the list, so that future requests
        can avoid probing them.  Used packs will be at the front of the
        self.packs list, in the order of their first use in 'entries'.

        :param entries: A list of (index, ...) tuples
        :param index_to_pack_map: A mapping from index objects to pack objects.
        """
        packs = []
        seen_indexes = set()
        for entry in entries:
            index = entry[0]
            if index not in seen_indexes:
                packs.append(index_to_pack_map[index])
                seen_indexes.add(index)
        if len(packs) == len(self.packs):
            if 'pack' in debug.debug_flags:
                trace.mutter('Not changing pack list, all packs used.')
            return
        seen_packs = set(packs)
        for pack in self.packs:
            if pack not in seen_packs:
                packs.append(pack)
                seen_packs.add(pack)
        if 'pack' in debug.debug_flags:
            old_names = [p.access_tuple()[1] for p in self.packs]
            new_names = [p.access_tuple()[1] for p in packs]
            trace.mutter('Reordering packs\nfrom: %s\n  to: %s',
                   old_names, new_names)
        self.packs = packs

    def _copy_revision_texts(self):
        # select revisions
        if self.revision_ids:
            revision_keys = [(revision_id,) for revision_id in self.revision_ids]
        else:
            revision_keys = None
        # select revision keys
        revision_index_map, revision_indices = self._pack_map_and_index_list(
            'revision_index')
        revision_nodes = self._index_contents(revision_indices, revision_keys)
        revision_nodes = list(revision_nodes)
        self._update_pack_order(revision_nodes, revision_index_map)
        # copy revision keys and adjust values
        self.pb.update("Copying revision texts", 1)
        total_items, readv_group_iter = self._revision_node_readv(revision_nodes)
        list(self._copy_nodes_graph(revision_index_map, self.new_pack._writer,
            self.new_pack.revision_index, readv_group_iter, total_items))
        if 'pack' in debug.debug_flags:
            trace.mutter('%s: create_pack: revisions copied: %s%s %d items t+%6.3fs',
                time.ctime(), self._pack_collection._upload_transport.base,
                self.new_pack.random_name,
                self.new_pack.revision_index.key_count(),
                time.time() - self.new_pack.start_time)
        self._revision_keys = revision_keys

    def _get_text_nodes(self):
        text_index_map, text_indices = self._pack_map_and_index_list(
            'text_index')
        return text_index_map, self._index_contents(text_indices,
            self._text_filter)

    def _copy_text_texts(self):
        # select text keys
        text_index_map, text_nodes = self._get_text_nodes()
        if self._text_filter is not None:
            # We could return the keys copied as part of the return value from
            # _copy_nodes_graph but this doesn't work all that well with the
            # need to get line output too, so we check separately, and as we're
            # going to buffer everything anyway, we check beforehand, which
            # saves reading knit data over the wire when we know there are
            # mising records.
            text_nodes = set(text_nodes)
            present_text_keys = set(_node[1] for _node in text_nodes)
            missing_text_keys = set(self._text_filter) - present_text_keys
            if missing_text_keys:
                # TODO: raise a specific error that can handle many missing
                # keys.
                trace.mutter("missing keys during fetch: %r", missing_text_keys)
                a_missing_key = missing_text_keys.pop()
                raise errors.RevisionNotPresent(a_missing_key[1],
                    a_missing_key[0])
        # copy text keys and adjust values
        self.pb.update("Copying content texts", 3)
        total_items, readv_group_iter = self._least_readv_node_readv(text_nodes)
        list(self._copy_nodes_graph(text_index_map, self.new_pack._writer,
            self.new_pack.text_index, readv_group_iter, total_items))
        self._log_copied_texts()

    def _create_pack_from_packs(self):
        self.pb.update("Opening pack", 0, 5)
        self.new_pack = self.open_pack()
        new_pack = self.new_pack
        # buffer data - we won't be reading-back during the pack creation and
        # this makes a significant difference on sftp pushes.
        new_pack.set_write_cache_size(1024*1024)
        if 'pack' in debug.debug_flags:
            plain_pack_list = ['%s%s' % (a_pack.pack_transport.base, a_pack.name)
                for a_pack in self.packs]
            if self.revision_ids is not None:
                rev_count = len(self.revision_ids)
            else:
                rev_count = 'all'
            trace.mutter('%s: create_pack: creating pack from source packs: '
                '%s%s %s revisions wanted %s t=0',
                time.ctime(), self._pack_collection._upload_transport.base, new_pack.random_name,
                plain_pack_list, rev_count)
        self._copy_revision_texts()
        self._copy_inventory_texts()
        self._copy_text_texts()
        # select signature keys
        signature_filter = self._revision_keys # same keyspace
        signature_index_map, signature_indices = self._pack_map_and_index_list(
            'signature_index')
        signature_nodes = self._index_contents(signature_indices,
            signature_filter)
        # copy signature keys and adjust values
        self.pb.update("Copying signature texts", 4)
        self._copy_nodes(signature_nodes, signature_index_map, new_pack._writer,
            new_pack.signature_index)
        if 'pack' in debug.debug_flags:
            trace.mutter('%s: create_pack: revision signatures copied: %s%s %d items t+%6.3fs',
                time.ctime(), self._pack_collection._upload_transport.base, new_pack.random_name,
                new_pack.signature_index.key_count(),
                time.time() - new_pack.start_time)
        new_pack._check_references()
        if not self._use_pack(new_pack):
            new_pack.abort()
            return None
        self.pb.update("Finishing pack", 5)
        new_pack.finish()
        self._pack_collection.allocate(new_pack)
        return new_pack

    def _least_readv_node_readv(self, nodes):
        """Generate request groups for nodes using the least readv's.

        :param nodes: An iterable of graph index nodes.
        :return: Total node count and an iterator of the data needed to perform
            readvs to obtain the data for nodes. Each item yielded by the
            iterator is a tuple with:
            index, readv_vector, node_vector. readv_vector is a list ready to
            hand to the transport readv method, and node_vector is a list of
            (key, eol_flag, references) for the node retrieved by the
            matching readv_vector.
        """
        # group by pack so we do one readv per pack
        nodes = sorted(nodes)
        total = len(nodes)
        request_groups = {}
        for index, key, value, references in nodes:
            if index not in request_groups:
                request_groups[index] = []
            request_groups[index].append((key, value, references))
        result = []
        for index, items in request_groups.iteritems():
            pack_readv_requests = []
            for key, value, references in items:
                # ---- KnitGraphIndex.get_position
                bits = value[1:].split(' ')
                offset, length = int(bits[0]), int(bits[1])
                pack_readv_requests.append(
                    ((offset, length), (key, value[0], references)))
            # linear scan up the pack to maximum range combining.
            pack_readv_requests.sort()
            # split out the readv and the node data.
            pack_readv = [readv for readv, node in pack_readv_requests]
            node_vector = [node for readv, node in pack_readv_requests]
            result.append((index, pack_readv, node_vector))
        return total, result

    def _revision_node_readv(self, revision_nodes):
        """Return the total revisions and the readv's to issue.

        :param revision_nodes: The revision index contents for the packs being
            incorporated into the new pack.
        :return: As per _least_readv_node_readv.
        """
        return self._least_readv_node_readv(revision_nodes)


class KnitReconcilePacker(KnitPacker):
    """A packer which regenerates indices etc as it copies.

    This is used by ``bzr reconcile`` to cause parent text pointers to be
    regenerated.
    """

    def __init__(self, *args, **kwargs):
        super(KnitReconcilePacker, self).__init__(*args, **kwargs)
        self._data_changed = False

    def _process_inventory_lines(self, inv_lines):
        """Generate a text key reference map rather for reconciling with."""
        repo = self._pack_collection.repo
        refs = repo._serializer._find_text_key_references(inv_lines)
        self._text_refs = refs
        # during reconcile we:
        #  - convert unreferenced texts to full texts
        #  - correct texts which reference a text not copied to be full texts
        #  - copy all others as-is but with corrected parents.
        #  - so at this point we don't know enough to decide what becomes a full
        #    text.
        self._text_filter = None

    def _copy_text_texts(self):
        """generate what texts we should have and then copy."""
        self.pb.update("Copying content texts", 3)
        # we have three major tasks here:
        # 1) generate the ideal index
        repo = self._pack_collection.repo
        ancestors = dict([(key[0], tuple(ref[0] for ref in refs[0])) for
            _1, key, _2, refs in
            self.new_pack.revision_index.iter_all_entries()])
        ideal_index = repo._generate_text_key_index(self._text_refs, ancestors)
        # 2) generate a text_nodes list that contains all the deltas that can
        #    be used as-is, with corrected parents.
        ok_nodes = []
        bad_texts = []
        discarded_nodes = []
        NULL_REVISION = _mod_revision.NULL_REVISION
        text_index_map, text_nodes = self._get_text_nodes()
        for node in text_nodes:
            # 0 - index
            # 1 - key
            # 2 - value
            # 3 - refs
            try:
                ideal_parents = tuple(ideal_index[node[1]])
            except KeyError:
                discarded_nodes.append(node)
                self._data_changed = True
            else:
                if ideal_parents == (NULL_REVISION,):
                    ideal_parents = ()
                if ideal_parents == node[3][0]:
                    # no change needed.
                    ok_nodes.append(node)
                elif ideal_parents[0:1] == node[3][0][0:1]:
                    # the left most parent is the same, or there are no parents
                    # today. Either way, we can preserve the representation as
                    # long as we change the refs to be inserted.
                    self._data_changed = True
                    ok_nodes.append((node[0], node[1], node[2],
                        (ideal_parents, node[3][1])))
                    self._data_changed = True
                else:
                    # Reinsert this text completely
                    bad_texts.append((node[1], ideal_parents))
                    self._data_changed = True
        # we're finished with some data.
        del ideal_index
        del text_nodes
        # 3) bulk copy the ok data
        total_items, readv_group_iter = self._least_readv_node_readv(ok_nodes)
        list(self._copy_nodes_graph(text_index_map, self.new_pack._writer,
            self.new_pack.text_index, readv_group_iter, total_items))
        # 4) adhoc copy all the other texts.
        # We have to topologically insert all texts otherwise we can fail to
        # reconcile when parts of a single delta chain are preserved intact,
        # and other parts are not. E.g. Discarded->d1->d2->d3. d1 will be
        # reinserted, and if d3 has incorrect parents it will also be
        # reinserted. If we insert d3 first, d2 is present (as it was bulk
        # copied), so we will try to delta, but d2 is not currently able to be
        # extracted because its basis d1 is not present. Topologically sorting
        # addresses this. The following generates a sort for all the texts that
        # are being inserted without having to reference the entire text key
        # space (we only topo sort the revisions, which is smaller).
        topo_order = tsort.topo_sort(ancestors)
        rev_order = dict(zip(topo_order, range(len(topo_order))))
        bad_texts.sort(key=lambda key:rev_order.get(key[0][1], 0))
        transaction = repo.get_transaction()
        file_id_index = GraphIndexPrefixAdapter(
            self.new_pack.text_index,
            ('blank', ), 1,
            add_nodes_callback=self.new_pack.text_index.add_nodes)
        data_access = _DirectPackAccess(
                {self.new_pack.text_index:self.new_pack.access_tuple()})
        data_access.set_writer(self.new_pack._writer, self.new_pack.text_index,
            self.new_pack.access_tuple())
        output_texts = KnitVersionedFiles(
            _KnitGraphIndex(self.new_pack.text_index,
                add_callback=self.new_pack.text_index.add_nodes,
                deltas=True, parents=True, is_locked=repo.is_locked),
            data_access=data_access, max_delta_chain=200)
        for key, parent_keys in bad_texts:
            # We refer to the new pack to delta data being output.
            # A possible improvement would be to catch errors on short reads
            # and only flush then.
            self.new_pack.flush()
            parents = []
            for parent_key in parent_keys:
                if parent_key[0] != key[0]:
                    # Graph parents must match the fileid
                    raise errors.BzrError('Mismatched key parent %r:%r' %
                        (key, parent_keys))
                parents.append(parent_key[1])
            text_lines = osutils.split_lines(repo.texts.get_record_stream(
                [key], 'unordered', True).next().get_bytes_as('fulltext'))
            output_texts.add_lines(key, parent_keys, text_lines,
                random_id=True, check_content=False)
        # 5) check that nothing inserted has a reference outside the keyspace.
        missing_text_keys = self.new_pack.text_index._external_references()
        if missing_text_keys:
            raise errors.BzrCheckError('Reference to missing compression parents %r'
                % (missing_text_keys,))
        self._log_copied_texts()

    def _use_pack(self, new_pack):
        """Override _use_pack to check for reconcile having changed content."""
        # XXX: we might be better checking this at the copy time.
        original_inventory_keys = set()
        inv_index = self._pack_collection.inventory_index.combined_index
        for entry in inv_index.iter_all_entries():
            original_inventory_keys.add(entry[1])
        new_inventory_keys = set()
        for entry in new_pack.inventory_index.iter_all_entries():
            new_inventory_keys.add(entry[1])
        if new_inventory_keys != original_inventory_keys:
            self._data_changed = True
        return new_pack.data_inserted() and self._data_changed


class OptimisingKnitPacker(KnitPacker):
    """A packer which spends more time to create better disk layouts."""

    def _revision_node_readv(self, revision_nodes):
        """Return the total revisions and the readv's to issue.

        This sort places revisions in topological order with the ancestors
        after the children.

        :param revision_nodes: The revision index contents for the packs being
            incorporated into the new pack.
        :return: As per _least_readv_node_readv.
        """
        # build an ancestors dict
        ancestors = {}
        by_key = {}
        for index, key, value, references in revision_nodes:
            ancestors[key] = references[0]
            by_key[key] = (index, value, references)
        order = tsort.topo_sort(ancestors)
        total = len(order)
        # Single IO is pathological, but it will work as a starting point.
        requests = []
        for key in reversed(order):
            index, value, references = by_key[key]
            # ---- KnitGraphIndex.get_position
            bits = value[1:].split(' ')
            offset, length = int(bits[0]), int(bits[1])
            requests.append(
                (index, [(offset, length)], [(key, value[0], references)]))
        # TODO: combine requests in the same index that are in ascending order.
        return total, requests

    def open_pack(self):
        """Open a pack for the pack we are creating."""
        new_pack = super(OptimisingKnitPacker, self).open_pack()
        # Turn on the optimization flags for all the index builders.
        new_pack.revision_index.set_optimize(for_size=True)
        new_pack.inventory_index.set_optimize(for_size=True)
        new_pack.text_index.set_optimize(for_size=True)
        new_pack.signature_index.set_optimize(for_size=True)
        return new_pack


class KnitRepositoryPackCollection(RepositoryPackCollection):
    """A knit pack collection."""

    pack_factory = NewPack
    resumed_pack_factory = ResumedPack
    normal_packer_class = KnitPacker
    optimising_packer_class = OptimisingKnitPacker