/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2007-10-26 08:56:09 UTC
  • mfrom: (2592.3.247 mbp-writegroups)
  • Revision ID: pqm@pqm.ubuntu.com-20071026085609-c3r8skfrmjj21j0m
Unlock while in a write group now aborts the write group, unlocks, and errors.  Also includes the knit extraction one-liner tweak.

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2007-2010 Canonical Ltd
 
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
17
17
"""WorkingTree4 format and implementation.
18
18
 
28
28
 
29
29
from bzrlib.lazy_import import lazy_import
30
30
lazy_import(globals(), """
 
31
from bisect import bisect_left
 
32
import collections
 
33
from copy import deepcopy
31
34
import errno
 
35
import itertools
 
36
import operator
32
37
import stat
 
38
from time import time
 
39
import warnings
33
40
 
34
41
import bzrlib
35
42
from bzrlib import (
36
43
    bzrdir,
37
44
    cache_utf8,
 
45
    conflicts as _mod_conflicts,
38
46
    debug,
 
47
    delta,
39
48
    dirstate,
40
49
    errors,
41
50
    generate_ids,
 
51
    globbing,
 
52
    ignores,
 
53
    merge,
42
54
    osutils,
43
55
    revision as _mod_revision,
44
56
    revisiontree,
 
57
    textui,
45
58
    trace,
46
59
    transform,
47
 
    views,
 
60
    urlutils,
 
61
    xml5,
 
62
    xml6,
48
63
    )
49
64
import bzrlib.branch
 
65
from bzrlib.transport import get_transport
50
66
import bzrlib.ui
51
67
""")
52
68
 
 
69
from bzrlib import symbol_versioning
53
70
from bzrlib.decorators import needs_read_lock, needs_write_lock
54
 
from bzrlib.filters import filtered_input_file, internal_size_sha_file_byname
55
 
from bzrlib.inventory import Inventory, ROOT_ID, entry_factory
56
 
from bzrlib.lock import LogicalLockResult
 
71
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
72
from bzrlib.lockable_files import LockableFiles, TransportLock
 
73
from bzrlib.lockdir import LockDir
 
74
import bzrlib.mutabletree
57
75
from bzrlib.mutabletree import needs_tree_write_lock
58
76
from bzrlib.osutils import (
59
77
    file_kind,
60
78
    isdir,
 
79
    normpath,
61
80
    pathjoin,
 
81
    rand_chars,
62
82
    realpath,
63
83
    safe_unicode,
 
84
    splitpath,
64
85
    )
65
 
from bzrlib.trace import mutter
 
86
from bzrlib.trace import mutter, note
66
87
from bzrlib.transport.local import LocalTransport
67
88
from bzrlib.tree import InterTree
 
89
from bzrlib.progress import DummyProgress, ProgressPhase
 
90
from bzrlib.revision import NULL_REVISION, CURRENT_REVISION
 
91
from bzrlib.rio import RioReader, rio_file, Stanza
 
92
from bzrlib.symbol_versioning import (deprecated_passed,
 
93
        deprecated_method,
 
94
        deprecated_function,
 
95
        DEPRECATED_PARAMETER,
 
96
        )
68
97
from bzrlib.tree import Tree
69
98
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
70
99
 
71
100
 
72
 
class DirStateWorkingTree(WorkingTree3):
 
101
# This is the Windows equivalent of ENOTDIR
 
102
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
103
# just an error code.
 
104
ERROR_PATH_NOT_FOUND = 3
 
105
ERROR_DIRECTORY = 267
 
106
 
 
107
 
 
108
class WorkingTree4(WorkingTree3):
 
109
    """This is the Format 4 working tree.
 
110
 
 
111
    This differs from WorkingTree3 by:
 
112
     - Having a consolidated internal dirstate, stored in a
 
113
       randomly-accessible sorted file on disk.
 
114
     - Not having a regular inventory attribute.  One can be synthesized 
 
115
       on demand but this is expensive and should be avoided.
 
116
 
 
117
    This is new in bzr 0.15.
 
118
    """
 
119
 
73
120
    def __init__(self, basedir,
74
121
                 branch,
75
122
                 _control_files=None,
84
131
        """
85
132
        self._format = _format
86
133
        self.bzrdir = _bzrdir
 
134
        assert isinstance(basedir, basestring), \
 
135
            "base directory %r is not a string" % basedir
87
136
        basedir = safe_unicode(basedir)
88
137
        mutter("opening working tree %r", basedir)
89
138
        self._branch = branch
 
139
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
140
            "branch %r is not a Branch" % self.branch
90
141
        self.basedir = realpath(basedir)
91
142
        # if branch is at our basedir and is a format 6 or less
92
143
        # assume all other formats have their own control files.
 
144
        assert isinstance(_control_files, LockableFiles), \
 
145
            "_control_files must be a LockableFiles, not %r" % _control_files
93
146
        self._control_files = _control_files
94
 
        self._transport = self._control_files._transport
95
147
        self._dirty = None
96
148
        #-------------
97
149
        # during a read or write lock these objects are set, and are
99
151
        self._dirstate = None
100
152
        self._inventory = None
101
153
        #-------------
102
 
        self._setup_directory_is_tree_reference()
103
 
        self._detect_case_handling()
104
 
        self._rules_searcher = None
105
 
        self.views = self._make_views()
106
 
        #--- allow tests to select the dirstate iter_changes implementation
107
 
        self._iter_changes = dirstate._process_entry
108
154
 
109
155
    @needs_tree_write_lock
110
156
    def _add(self, files, ids, kinds):
138
184
    @needs_tree_write_lock
139
185
    def add_reference(self, sub_tree):
140
186
        # use standard implementation, which calls back to self._add
141
 
        #
 
187
        # 
142
188
        # So we don't store the reference_revision in the working dirstate,
143
 
        # it's just recorded at the moment of commit.
 
189
        # it's just recorded at the moment of commit. 
144
190
        self._add_reference(sub_tree)
145
191
 
146
192
    def break_lock(self):
185
231
            WorkingTree3._comparison_data(self, entry, path)
186
232
        # it looks like a plain directory, but it's really a reference -- see
187
233
        # also kind()
188
 
        if (self._repo_supports_tree_reference and kind == 'directory'
189
 
            and entry is not None and entry.kind == 'tree-reference'):
 
234
        if (self._repo_supports_tree_reference and
 
235
            kind == 'directory' and
 
236
            self._directory_is_tree_reference(path)):
190
237
            kind = 'tree-reference'
191
238
        return kind, executable, stat_value
192
239
 
218
265
            return self._dirstate
219
266
        local_path = self.bzrdir.get_workingtree_transport(None
220
267
            ).local_abspath('dirstate')
221
 
        self._dirstate = dirstate.DirState.on_file(local_path,
222
 
            self._sha1_provider())
 
268
        self._dirstate = dirstate.DirState.on_file(local_path)
223
269
        return self._dirstate
224
270
 
225
 
    def _sha1_provider(self):
226
 
        """A function that returns a SHA1Provider suitable for this tree.
227
 
 
228
 
        :return: None if content filtering is not supported by this tree.
229
 
          Otherwise, a SHA1Provider is returned that sha's the canonical
230
 
          form of files, i.e. after read filters are applied.
231
 
        """
232
 
        if self.supports_content_filtering():
233
 
            return ContentFilterAwareSHA1Provider(self)
234
 
        else:
235
 
            return None
236
 
 
237
271
    def filter_unversioned_files(self, paths):
238
272
        """Filter out paths that are versioned.
239
273
 
271
305
 
272
306
    def _generate_inventory(self):
273
307
        """Create and set self.inventory from the dirstate object.
274
 
 
 
308
        
275
309
        This is relatively expensive: we have to walk the entire dirstate.
276
310
        Ideally we would not, and can deprecate this function.
277
311
        """
281
315
        state._read_dirblocks_if_needed()
282
316
        root_key, current_entry = self._get_entry(path='')
283
317
        current_id = root_key[2]
284
 
        if not (current_entry[0][0] == 'd'): # directory
285
 
            raise AssertionError(current_entry)
 
318
        assert current_entry[0][0] == 'd' # directory
286
319
        inv = Inventory(root_id=current_id)
287
320
        # Turn some things into local variables
288
321
        minikind_to_kind = dirstate.DirState._minikind_to_kind
321
354
                    # add this entry to the parent map.
322
355
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
323
356
                elif kind == 'tree-reference':
324
 
                    if not self._repo_supports_tree_reference:
325
 
                        raise errors.UnsupportedOperation(
326
 
                            self._generate_inventory,
327
 
                            self.branch.repository)
 
357
                    assert self._repo_supports_tree_reference, \
 
358
                        "repository of %r " \
 
359
                        "doesn't support tree references " \
 
360
                        "required by entry %r" \
 
361
                        % (self, name)
328
362
                    inv_entry.reference_revision = link_or_sha1 or None
329
363
                elif kind != 'symlink':
330
364
                    raise AssertionError("unknown kind %r" % kind)
331
365
                # These checks cost us around 40ms on a 55k entry tree
332
 
                if file_id in inv_byid:
333
 
                    raise AssertionError('file_id %s already in'
334
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
335
 
                if name_unicode in parent_ie.children:
336
 
                    raise AssertionError('name %r already in parent'
337
 
                        % (name_unicode,))
 
366
                assert file_id not in inv_byid, ('file_id %s already in'
 
367
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
368
                assert name_unicode not in parent_ie.children
338
369
                inv_byid[file_id] = inv_entry
339
370
                parent_ie.children[name_unicode] = inv_entry
340
371
        self._inventory = inv
345
376
        If either file_id or path is supplied, it is used as the key to lookup.
346
377
        If both are supplied, the fastest lookup is used, and an error is
347
378
        raised if they do not both point at the same row.
348
 
 
 
379
        
349
380
        :param file_id: An optional unicode file_id to be looked up.
350
381
        :param path: An optional unicode path to be looked up.
351
382
        :return: The dirstate row tuple for path/file_id, or (None, None)
375
406
                    return None
376
407
                else:
377
408
                    raise
378
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
379
 
            stat_value=stat_value)
 
409
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
410
                                          stat_value=stat_value)
380
411
        if entry[1][0][0] == 'f':
381
 
            if link_or_sha1 is None:
382
 
                file_obj, statvalue = self.get_file_with_stat(file_id, path)
383
 
                try:
384
 
                    sha1 = osutils.sha_file(file_obj)
385
 
                finally:
386
 
                    file_obj.close()
387
 
                self._observed_sha1(file_id, path, (sha1, statvalue))
388
 
                return sha1
389
 
            else:
390
 
                return link_or_sha1
 
412
            return link_or_sha1
391
413
        return None
392
414
 
393
415
    def _get_inventory(self):
407
429
    @needs_read_lock
408
430
    def get_parent_ids(self):
409
431
        """See Tree.get_parent_ids.
410
 
 
 
432
        
411
433
        This implementation requests the ids list from the dirstate file.
412
434
        """
413
435
        return self.current_dirstate().get_parent_ids()
435
457
        return osutils.lexists(pathjoin(
436
458
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
437
459
 
438
 
    def has_or_had_id(self, file_id):
439
 
        state = self.current_dirstate()
440
 
        row, parents = self._get_entry(file_id=file_id)
441
 
        return row is not None
442
 
 
443
460
    @needs_read_lock
444
461
    def id2path(self, file_id):
445
462
        "Convert a file-id to a path."
475
492
 
476
493
            Note: The caller is expected to take a read-lock before calling this.
477
494
            """
478
 
            self._must_be_locked()
479
495
            if not path:
480
496
                path = self.id2path(file_id)
481
497
            mode = os.lstat(self.abspath(path)).st_mode
482
498
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
483
499
 
484
 
    def all_file_ids(self):
485
 
        """See Tree.iter_all_file_ids"""
486
 
        self._must_be_locked()
487
 
        result = set()
488
 
        for key, tree_details in self.current_dirstate()._iter_entries():
489
 
            if tree_details[0][0] in ('a', 'r'): # relocated
490
 
                continue
491
 
            result.add(key[2])
492
 
        return result
493
 
 
494
500
    @needs_read_lock
495
501
    def __iter__(self):
496
502
        """Iterate through file_ids for this tree.
509
515
        return iter(result)
510
516
 
511
517
    def iter_references(self):
512
 
        if not self._repo_supports_tree_reference:
513
 
            # When the repo doesn't support references, we will have nothing to
514
 
            # return
515
 
            return
516
518
        for key, tree_details in self.current_dirstate()._iter_entries():
517
519
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
518
520
                # not relevant to the working tree
520
522
            if not key[1]:
521
523
                # the root is not a reference.
522
524
                continue
523
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
525
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
524
526
            try:
525
 
                if self._kind(relpath) == 'tree-reference':
526
 
                    yield relpath, key[2]
 
527
                if self._kind(path) == 'tree-reference':
 
528
                    yield path, key[2]
527
529
            except errors.NoSuchFile:
528
530
                # path is missing on disk.
529
531
                continue
530
532
 
531
 
    def _observed_sha1(self, file_id, path, (sha1, statvalue)):
532
 
        """See MutableTree._observed_sha1."""
533
 
        state = self.current_dirstate()
534
 
        entry = self._get_entry(file_id=file_id, path=path)
535
 
        state._observed_sha1(entry, sha1, statvalue)
536
 
 
537
533
    def kind(self, file_id):
538
534
        """Return the kind of a file.
539
535
 
543
539
        Note: The caller is expected to take a read-lock before calling this.
544
540
        """
545
541
        relpath = self.id2path(file_id)
546
 
        if relpath is None:
547
 
            raise AssertionError(
548
 
                "path for id {%s} is None!" % file_id)
 
542
        assert relpath != None, \
 
543
            "path for id {%s} is None!" % file_id
549
544
        return self._kind(relpath)
550
545
 
551
546
    def _kind(self, relpath):
552
547
        abspath = self.abspath(relpath)
553
548
        kind = file_kind(abspath)
554
 
        if (self._repo_supports_tree_reference and kind == 'directory'):
555
 
            entry = self._get_entry(path=relpath)
556
 
            if entry[1] is not None:
557
 
                if entry[1][0][0] == 't':
558
 
                    kind = 'tree-reference'
 
549
        if (self._repo_supports_tree_reference and
 
550
            kind == 'directory' and
 
551
            self._directory_is_tree_reference(relpath)):
 
552
            kind = 'tree-reference'
559
553
        return kind
560
554
 
561
555
    @needs_read_lock
568
562
            return _mod_revision.NULL_REVISION
569
563
 
570
564
    def lock_read(self):
571
 
        """See Branch.lock_read, and WorkingTree.unlock.
572
 
 
573
 
        :return: A bzrlib.lock.LogicalLockResult.
574
 
        """
 
565
        """See Branch.lock_read, and WorkingTree.unlock."""
575
566
        self.branch.lock_read()
576
567
        try:
577
568
            self._control_files.lock_read()
590
581
        except:
591
582
            self.branch.unlock()
592
583
            raise
593
 
        return LogicalLockResult(self.unlock)
594
584
 
595
585
    def _lock_self_write(self):
596
586
        """This should be called after the branch is locked."""
611
601
        except:
612
602
            self.branch.unlock()
613
603
            raise
614
 
        return LogicalLockResult(self.unlock)
615
604
 
616
605
    def lock_tree_write(self):
617
 
        """See MutableTree.lock_tree_write, and WorkingTree.unlock.
618
 
 
619
 
        :return: A bzrlib.lock.LogicalLockResult.
620
 
        """
 
606
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
621
607
        self.branch.lock_read()
622
 
        return self._lock_self_write()
 
608
        self._lock_self_write()
623
609
 
624
610
    def lock_write(self):
625
 
        """See MutableTree.lock_write, and WorkingTree.unlock.
626
 
 
627
 
        :return: A bzrlib.lock.LogicalLockResult.
628
 
        """
 
611
        """See MutableTree.lock_write, and WorkingTree.unlock."""
629
612
        self.branch.lock_write()
630
 
        return self._lock_self_write()
 
613
        self._lock_self_write()
631
614
 
632
615
    @needs_tree_write_lock
633
616
    def move(self, from_paths, to_dir, after=False):
635
618
        result = []
636
619
        if not from_paths:
637
620
            return result
 
621
 
638
622
        state = self.current_dirstate()
639
 
        if isinstance(from_paths, basestring):
640
 
            raise ValueError()
 
623
 
 
624
        assert not isinstance(from_paths, basestring)
641
625
        to_dir_utf8 = to_dir.encode('utf8')
642
626
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
643
627
        id_index = state._get_id_index()
665
649
        if self._inventory is not None:
666
650
            update_inventory = True
667
651
            inv = self.inventory
 
652
            to_dir_ie = inv[to_dir_id]
668
653
            to_dir_id = to_entry[0][2]
669
 
            to_dir_ie = inv[to_dir_id]
670
654
        else:
671
655
            update_inventory = False
672
656
 
703
687
            from_entry = self._get_entry(path=from_rel)
704
688
            if from_entry == (None, None):
705
689
                raise errors.BzrMoveFailedError(from_rel,to_dir,
706
 
                    errors.NotVersionedError(path=from_rel))
 
690
                    errors.NotVersionedError(path=str(from_rel)))
707
691
 
708
692
            from_id = from_entry[0][2]
709
693
            to_rel = pathjoin(to_dir, from_tail)
736
720
                if from_missing: # implicitly just update our path mapping
737
721
                    move_file = False
738
722
                elif not after:
739
 
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
723
                    raise errors.RenameFailedFilesExist(from_rel, to_rel,
 
724
                        extra="(Use --after to update the Bazaar id)")
740
725
 
741
726
            rollbacks = []
742
727
            def rollback_rename():
798
783
                if minikind == 'd':
799
784
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
800
785
                        """Recursively update all entries in this dirblock."""
801
 
                        if from_dir == '':
802
 
                            raise AssertionError("renaming root not supported")
 
786
                        assert from_dir != '', "renaming root not supported"
803
787
                        from_key = (from_dir, '')
804
788
                        from_block_idx, present = \
805
789
                            state._find_block_index_from_key(from_key)
818
802
 
819
803
                        # Grab a copy since move_one may update the list.
820
804
                        for entry in from_block[1][:]:
821
 
                            if not (entry[0][0] == from_dir):
822
 
                                raise AssertionError()
 
805
                            assert entry[0][0] == from_dir
823
806
                            cur_details = entry[1][0]
824
807
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
825
808
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
884
867
        for tree in trees:
885
868
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
886
869
                parents):
887
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
888
 
                    trees, require_versioned)
 
870
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
889
871
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
890
872
        # -- make all paths utf8 --
891
873
        paths_utf8 = set()
952
934
                raise errors.PathsNotVersionedError(paths)
953
935
        # -- remove redundancy in supplied paths to prevent over-scanning --
954
936
        search_paths = osutils.minimum_path_selection(paths)
955
 
        # sketch:
 
937
        # sketch: 
956
938
        # for all search_indexs in each path at or under each element of
957
939
        # search_paths, if the detail is relocated: add the id, and add the
958
940
        # relocated path as one to search if its not searched already. If the
1014
996
 
1015
997
    def read_working_inventory(self):
1016
998
        """Read the working inventory.
1017
 
 
 
999
        
1018
1000
        This is a meaningless operation for dirstate, but we obey it anyhow.
1019
1001
        """
1020
1002
        return self.inventory
1038
1020
    def set_last_revision(self, new_revision):
1039
1021
        """Change the last revision in the working tree."""
1040
1022
        parents = self.get_parent_ids()
1041
 
        if new_revision in (_mod_revision.NULL_REVISION, None):
1042
 
            if len(parents) >= 2:
1043
 
                raise AssertionError(
1044
 
                    "setting the last parent to none with a pending merge is "
1045
 
                    "unsupported.")
 
1023
        if new_revision in (NULL_REVISION, None):
 
1024
            assert len(parents) < 2, (
 
1025
                "setting the last parent to none with a pending merge is "
 
1026
                "unsupported.")
1046
1027
            self.set_parent_ids([])
1047
1028
        else:
1048
1029
            self.set_parent_ids([new_revision] + parents[1:],
1051
1032
    @needs_tree_write_lock
1052
1033
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1053
1034
        """Set the parent ids to revision_ids.
1054
 
 
 
1035
        
1055
1036
        See also set_parent_trees. This api will try to retrieve the tree data
1056
1037
        for each element of revision_ids from the trees repository. If you have
1057
1038
        tree data already available, it is more efficient to use
1089
1070
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1090
1071
        real_trees = []
1091
1072
        ghosts = []
1092
 
 
1093
 
        parent_ids = [rev_id for rev_id, tree in parents_list]
1094
 
        graph = self.branch.repository.get_graph()
1095
 
        heads = graph.heads(parent_ids)
1096
 
        accepted_revisions = set()
1097
 
 
1098
1073
        # convert absent trees to the null tree, which we convert back to
1099
1074
        # missing on access.
1100
1075
        for rev_id, tree in parents_list:
1101
 
            if len(accepted_revisions) > 0:
1102
 
                # we always accept the first tree
1103
 
                if rev_id in accepted_revisions or rev_id not in heads:
1104
 
                    # We have already included either this tree, or its
1105
 
                    # descendent, so we skip it.
1106
 
                    continue
1107
1076
            _mod_revision.check_not_reserved_id(rev_id)
1108
1077
            if tree is not None:
1109
1078
                real_trees.append((rev_id, tree))
1110
1079
            else:
1111
1080
                real_trees.append((rev_id,
1112
 
                    self.branch.repository.revision_tree(
1113
 
                        _mod_revision.NULL_REVISION)))
 
1081
                    self.branch.repository.revision_tree(None)))
1114
1082
                ghosts.append(rev_id)
1115
 
            accepted_revisions.add(rev_id)
1116
1083
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
1117
1084
        self._make_dirty(reset_inventory=False)
1118
1085
 
1211
1178
                # just forget the whole block.
1212
1179
                entry_index = 0
1213
1180
                while entry_index < len(block[1]):
 
1181
                    # Mark this file id as having been removed
1214
1182
                    entry = block[1][entry_index]
1215
 
                    if entry[1][0][0] in 'ar':
1216
 
                        # don't remove absent or renamed entries
 
1183
                    ids_to_unversion.discard(entry[0][2])
 
1184
                    if (entry[1][0][0] in 'ar' # don't remove absent or renamed
 
1185
                                               # entries
 
1186
                        or not state._make_absent(entry)):
1217
1187
                        entry_index += 1
1218
 
                    else:
1219
 
                        # Mark this file id as having been removed
1220
 
                        ids_to_unversion.discard(entry[0][2])
1221
 
                        if not state._make_absent(entry):
1222
 
                            # The block has not shrunk.
1223
 
                            entry_index += 1
1224
1188
                # go to the next block. (At the moment we dont delete empty
1225
1189
                # dirblocks)
1226
1190
                block_index += 1
1249
1213
            for file_id in file_ids:
1250
1214
                self._inventory.remove_recursive_id(file_id)
1251
1215
 
1252
 
    @needs_tree_write_lock
1253
 
    def rename_one(self, from_rel, to_rel, after=False):
1254
 
        """See WorkingTree.rename_one"""
1255
 
        self.flush()
1256
 
        WorkingTree.rename_one(self, from_rel, to_rel, after)
1257
 
 
1258
 
    @needs_tree_write_lock
1259
 
    def apply_inventory_delta(self, changes):
1260
 
        """See MutableTree.apply_inventory_delta"""
1261
 
        state = self.current_dirstate()
1262
 
        state.update_by_delta(changes)
1263
 
        self._make_dirty(reset_inventory=True)
1264
 
 
1265
 
    def update_basis_by_delta(self, new_revid, delta):
1266
 
        """See MutableTree.update_basis_by_delta."""
1267
 
        if self.last_revision() == new_revid:
1268
 
            raise AssertionError()
1269
 
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
1270
 
 
1271
1216
    @needs_read_lock
1272
1217
    def _validate(self):
1273
1218
        self._dirstate._validate()
1275
1220
    @needs_tree_write_lock
1276
1221
    def _write_inventory(self, inv):
1277
1222
        """Write inventory as the current inventory."""
1278
 
        if self._dirty:
1279
 
            raise AssertionError("attempting to write an inventory when the "
1280
 
                "dirstate is dirty will lose pending changes")
1281
 
        had_inventory = self._inventory is not None
1282
 
        # Setting self._inventory = None forces the dirstate to regenerate the
1283
 
        # working inventory. We do this because self.inventory may be inv, or
1284
 
        # may have been modified, and either case would prevent a clean delta
1285
 
        # being created.
1286
 
        self._inventory = None
1287
 
        # generate a delta,
1288
 
        delta = inv._make_delta(self.inventory)
1289
 
        # and apply it.
1290
 
        self.apply_inventory_delta(delta)
1291
 
        if had_inventory:
 
1223
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1224
        self.current_dirstate().set_state_from_inventory(inv)
 
1225
        self._make_dirty(reset_inventory=False)
 
1226
        if self._inventory is not None:
1292
1227
            self._inventory = inv
1293
1228
        self.flush()
1294
1229
 
1295
1230
 
1296
 
class ContentFilterAwareSHA1Provider(dirstate.SHA1Provider):
1297
 
 
1298
 
    def __init__(self, tree):
1299
 
        self.tree = tree
1300
 
 
1301
 
    def sha1(self, abspath):
1302
 
        """See dirstate.SHA1Provider.sha1()."""
1303
 
        filters = self.tree._content_filter_stack(
1304
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1305
 
        return internal_size_sha_file_byname(abspath, filters)[1]
1306
 
 
1307
 
    def stat_and_sha1(self, abspath):
1308
 
        """See dirstate.SHA1Provider.stat_and_sha1()."""
1309
 
        filters = self.tree._content_filter_stack(
1310
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1311
 
        file_obj = file(abspath, 'rb', 65000)
1312
 
        try:
1313
 
            statvalue = os.fstat(file_obj.fileno())
1314
 
            if filters:
1315
 
                file_obj = filtered_input_file(file_obj, filters)
1316
 
            sha1 = osutils.size_sha_file(file_obj)[1]
1317
 
        finally:
1318
 
            file_obj.close()
1319
 
        return statvalue, sha1
1320
 
 
1321
 
 
1322
 
class ContentFilteringDirStateWorkingTree(DirStateWorkingTree):
1323
 
    """Dirstate working tree that supports content filtering.
1324
 
 
1325
 
    The dirstate holds the hash and size of the canonical form of the file, 
1326
 
    and most methods must return that.
1327
 
    """
1328
 
 
1329
 
    def _file_content_summary(self, path, stat_result):
1330
 
        # This is to support the somewhat obsolete path_content_summary method
1331
 
        # with content filtering: see
1332
 
        # <https://bugs.edge.launchpad.net/bzr/+bug/415508>.
1333
 
        #
1334
 
        # If the dirstate cache is up to date and knows the hash and size,
1335
 
        # return that.
1336
 
        # Otherwise if there are no content filters, return the on-disk size
1337
 
        # and leave the hash blank.
1338
 
        # Otherwise, read and filter the on-disk file and use its size and
1339
 
        # hash.
1340
 
        #
1341
 
        # The dirstate doesn't store the size of the canonical form so we
1342
 
        # can't trust it for content-filtered trees.  We just return None.
1343
 
        dirstate_sha1 = self._dirstate.sha1_from_stat(path, stat_result)
1344
 
        executable = self._is_executable_from_path_and_stat(path, stat_result)
1345
 
        return ('file', None, executable, dirstate_sha1)
1346
 
 
1347
 
 
1348
 
class WorkingTree4(DirStateWorkingTree):
1349
 
    """This is the Format 4 working tree.
1350
 
 
1351
 
    This differs from WorkingTree3 by:
1352
 
     - Having a consolidated internal dirstate, stored in a
1353
 
       randomly-accessible sorted file on disk.
1354
 
     - Not having a regular inventory attribute.  One can be synthesized
1355
 
       on demand but this is expensive and should be avoided.
1356
 
 
1357
 
    This is new in bzr 0.15.
1358
 
    """
1359
 
 
1360
 
 
1361
 
class WorkingTree5(ContentFilteringDirStateWorkingTree):
1362
 
    """This is the Format 5 working tree.
1363
 
 
1364
 
    This differs from WorkingTree4 by:
1365
 
     - Supporting content filtering.
1366
 
 
1367
 
    This is new in bzr 1.11.
1368
 
    """
1369
 
 
1370
 
 
1371
 
class WorkingTree6(ContentFilteringDirStateWorkingTree):
1372
 
    """This is the Format 6 working tree.
1373
 
 
1374
 
    This differs from WorkingTree5 by:
1375
 
     - Supporting a current view that may mask the set of files in a tree
1376
 
       impacted by most user operations.
1377
 
 
1378
 
    This is new in bzr 1.14.
1379
 
    """
1380
 
 
1381
 
    def _make_views(self):
1382
 
        return views.PathBasedViews(self)
1383
 
 
1384
 
 
1385
 
class DirStateWorkingTreeFormat(WorkingTreeFormat3):
1386
 
 
1387
 
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
1388
 
                   accelerator_tree=None, hardlink=False):
 
1231
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1232
    """The first consolidated dirstate working tree format.
 
1233
 
 
1234
    This format:
 
1235
        - exists within a metadir controlling .bzr
 
1236
        - includes an explicit version marker for the workingtree control
 
1237
          files, separate from the BzrDir format
 
1238
        - modifies the hash cache format
 
1239
        - is new in bzr 0.15
 
1240
        - uses a LockDir to guard access to it.
 
1241
    """
 
1242
 
 
1243
    upgrade_recommended = False
 
1244
 
 
1245
    def get_format_string(self):
 
1246
        """See WorkingTreeFormat.get_format_string()."""
 
1247
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1248
 
 
1249
    def get_format_description(self):
 
1250
        """See WorkingTreeFormat.get_format_description()."""
 
1251
        return "Working tree format 4"
 
1252
 
 
1253
    def initialize(self, a_bzrdir, revision_id=None):
1389
1254
        """See WorkingTreeFormat.initialize().
1390
1255
 
1391
1256
        :param revision_id: allows creating a working tree at a different
1392
1257
        revision than the branch is at.
1393
 
        :param accelerator_tree: A tree which can be used for retrieving file
1394
 
            contents more quickly than the revision tree, i.e. a workingtree.
1395
 
            The revision tree will be used for cases where accelerator_tree's
1396
 
            content is different.
1397
 
        :param hardlink: If true, hard-link files from accelerator_tree,
1398
 
            where possible.
1399
1258
 
1400
1259
        These trees get an initial random root id, if their repository supports
1401
1260
        rich root data, TREE_ROOT otherwise.
1406
1265
        control_files = self._open_control_files(a_bzrdir)
1407
1266
        control_files.create_lock()
1408
1267
        control_files.lock_write()
1409
 
        transport.put_bytes('format', self.get_format_string(),
1410
 
            mode=a_bzrdir._get_file_mode())
1411
 
        if from_branch is not None:
1412
 
            branch = from_branch
1413
 
        else:
1414
 
            branch = a_bzrdir.open_branch()
 
1268
        control_files.put_utf8('format', self.get_format_string())
 
1269
        branch = a_bzrdir.open_branch()
1415
1270
        if revision_id is None:
1416
1271
            revision_id = branch.last_revision()
1417
1272
        local_path = transport.local_abspath('dirstate')
1419
1274
        state = dirstate.DirState.initialize(local_path)
1420
1275
        state.unlock()
1421
1276
        del state
1422
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1277
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1423
1278
                         branch,
1424
1279
                         _format=self,
1425
1280
                         _bzrdir=a_bzrdir,
1427
1282
        wt._new_tree()
1428
1283
        wt.lock_tree_write()
1429
1284
        try:
1430
 
            self._init_custom_control_files(wt)
1431
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
 
1285
            if revision_id in (None, NULL_REVISION):
1432
1286
                if branch.repository.supports_rich_root():
1433
1287
                    wt._set_root_id(generate_ids.gen_root_id())
1434
1288
                else:
1435
1289
                    wt._set_root_id(ROOT_ID)
1436
1290
                wt.flush()
1437
 
            basis = None
1438
 
            # frequently, we will get here due to branching.  The accelerator
1439
 
            # tree will be the tree from the branch, so the desired basis
1440
 
            # tree will often be a parent of the accelerator tree.
1441
 
            if accelerator_tree is not None:
1442
 
                try:
1443
 
                    basis = accelerator_tree.revision_tree(revision_id)
1444
 
                except errors.NoSuchRevision:
1445
 
                    pass
1446
 
            if basis is None:
1447
 
                basis = branch.repository.revision_tree(revision_id)
1448
 
            if revision_id == _mod_revision.NULL_REVISION:
1449
 
                parents_list = []
1450
 
            else:
1451
 
                parents_list = [(revision_id, basis)]
 
1291
            wt.set_last_revision(revision_id)
 
1292
            wt.flush()
 
1293
            basis = wt.basis_tree()
1452
1294
            basis.lock_read()
1453
 
            try:
1454
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1295
            # if the basis has a root id we have to use that; otherwise we use
 
1296
            # a new random one
 
1297
            basis_root_id = basis.get_root_id()
 
1298
            if basis_root_id is not None:
 
1299
                wt._set_root_id(basis_root_id)
1455
1300
                wt.flush()
1456
 
                # if the basis has a root id we have to use that; otherwise we
1457
 
                # use a new random one
1458
 
                basis_root_id = basis.get_root_id()
1459
 
                if basis_root_id is not None:
1460
 
                    wt._set_root_id(basis_root_id)
1461
 
                    wt.flush()
1462
 
                if wt.supports_content_filtering():
1463
 
                    # The original tree may not have the same content filters
1464
 
                    # applied so we can't safely build the inventory delta from
1465
 
                    # the source tree.
1466
 
                    delta_from_tree = False
1467
 
                else:
1468
 
                    delta_from_tree = True
1469
 
                # delta_from_tree is safe even for DirStateRevisionTrees,
1470
 
                # because wt4.apply_inventory_delta does not mutate the input
1471
 
                # inventory entries.
1472
 
                transform.build_tree(basis, wt, accelerator_tree,
1473
 
                                     hardlink=hardlink,
1474
 
                                     delta_from_tree=delta_from_tree)
1475
 
            finally:
1476
 
                basis.unlock()
 
1301
            transform.build_tree(basis, wt)
 
1302
            basis.unlock()
1477
1303
        finally:
1478
1304
            control_files.unlock()
1479
1305
            wt.unlock()
1480
1306
        return wt
1481
1307
 
1482
 
    def _init_custom_control_files(self, wt):
1483
 
        """Subclasses with custom control files should override this method.
1484
 
 
1485
 
        The working tree and control files are locked for writing when this
1486
 
        method is called.
1487
 
 
1488
 
        :param wt: the WorkingTree object
1489
 
        """
1490
 
 
1491
1308
    def _open(self, a_bzrdir, control_files):
1492
1309
        """Open the tree itself.
1493
1310
 
1494
1311
        :param a_bzrdir: the dir for the tree.
1495
1312
        :param control_files: the control files for the tree.
1496
1313
        """
1497
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1314
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1498
1315
                           branch=a_bzrdir.open_branch(),
1499
1316
                           _format=self,
1500
1317
                           _bzrdir=a_bzrdir,
1501
1318
                           _control_files=control_files)
1502
1319
 
1503
1320
    def __get_matchingbzrdir(self):
1504
 
        return self._get_matchingbzrdir()
1505
 
 
1506
 
    def _get_matchingbzrdir(self):
1507
 
        """Overrideable method to get a bzrdir for testing."""
1508
1321
        # please test against something that will let us do tree references
1509
1322
        return bzrdir.format_registry.make_bzrdir(
1510
1323
            'dirstate-with-subtree')
1512
1325
    _matchingbzrdir = property(__get_matchingbzrdir)
1513
1326
 
1514
1327
 
1515
 
class WorkingTreeFormat4(DirStateWorkingTreeFormat):
1516
 
    """The first consolidated dirstate working tree format.
1517
 
 
1518
 
    This format:
1519
 
        - exists within a metadir controlling .bzr
1520
 
        - includes an explicit version marker for the workingtree control
1521
 
          files, separate from the BzrDir format
1522
 
        - modifies the hash cache format
1523
 
        - is new in bzr 0.15
1524
 
        - uses a LockDir to guard access to it.
1525
 
    """
1526
 
 
1527
 
    upgrade_recommended = False
1528
 
 
1529
 
    _tree_class = WorkingTree4
1530
 
 
1531
 
    def get_format_string(self):
1532
 
        """See WorkingTreeFormat.get_format_string()."""
1533
 
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
1534
 
 
1535
 
    def get_format_description(self):
1536
 
        """See WorkingTreeFormat.get_format_description()."""
1537
 
        return "Working tree format 4"
1538
 
 
1539
 
 
1540
 
class WorkingTreeFormat5(DirStateWorkingTreeFormat):
1541
 
    """WorkingTree format supporting content filtering.
1542
 
    """
1543
 
 
1544
 
    upgrade_recommended = False
1545
 
 
1546
 
    _tree_class = WorkingTree5
1547
 
 
1548
 
    def get_format_string(self):
1549
 
        """See WorkingTreeFormat.get_format_string()."""
1550
 
        return "Bazaar Working Tree Format 5 (bzr 1.11)\n"
1551
 
 
1552
 
    def get_format_description(self):
1553
 
        """See WorkingTreeFormat.get_format_description()."""
1554
 
        return "Working tree format 5"
1555
 
 
1556
 
    def supports_content_filtering(self):
1557
 
        return True
1558
 
 
1559
 
 
1560
 
class WorkingTreeFormat6(DirStateWorkingTreeFormat):
1561
 
    """WorkingTree format supporting views.
1562
 
    """
1563
 
 
1564
 
    upgrade_recommended = False
1565
 
 
1566
 
    _tree_class = WorkingTree6
1567
 
 
1568
 
    def get_format_string(self):
1569
 
        """See WorkingTreeFormat.get_format_string()."""
1570
 
        return "Bazaar Working Tree Format 6 (bzr 1.14)\n"
1571
 
 
1572
 
    def get_format_description(self):
1573
 
        """See WorkingTreeFormat.get_format_description()."""
1574
 
        return "Working tree format 6"
1575
 
 
1576
 
    def _init_custom_control_files(self, wt):
1577
 
        """Subclasses with custom control files should override this method."""
1578
 
        wt._transport.put_bytes('views', '', mode=wt.bzrdir._get_file_mode())
1579
 
 
1580
 
    def supports_content_filtering(self):
1581
 
        return True
1582
 
 
1583
 
    def supports_views(self):
1584
 
        return True
1585
 
 
1586
 
 
1587
1328
class DirStateRevisionTree(Tree):
1588
 
    """A revision tree pulling the inventory from a dirstate.
1589
 
    
1590
 
    Note that this is one of the historical (ie revision) trees cached in the
1591
 
    dirstate for easy access, not the workingtree.
1592
 
    """
 
1329
    """A revision tree pulling the inventory from a dirstate."""
1593
1330
 
1594
1331
    def __init__(self, dirstate, revision_id, repository):
1595
1332
        self._dirstate = dirstate
1598
1335
        self._inventory = None
1599
1336
        self._locked = 0
1600
1337
        self._dirstate_locked = False
1601
 
        self._repo_supports_tree_reference = getattr(
1602
 
            repository._format, "supports_tree_reference",
1603
 
            False)
1604
1338
 
1605
1339
    def __repr__(self):
1606
1340
        return "<%s of %s in %s>" % \
1609
1343
    def annotate_iter(self, file_id,
1610
1344
                      default_revision=_mod_revision.CURRENT_REVISION):
1611
1345
        """See Tree.annotate_iter"""
1612
 
        text_key = (file_id, self.inventory[file_id].revision)
1613
 
        annotations = self._repository.texts.annotate(text_key)
1614
 
        return [(key[-1], line) for (key, line) in annotations]
 
1346
        w = self._get_weave(file_id)
 
1347
        return w.annotate_iter(self.inventory[file_id].revision)
1615
1348
 
1616
1349
    def _get_ancestors(self, default_revision):
1617
1350
        return set(self._repository.get_ancestry(self._revision_id,
1638
1371
    def get_root_id(self):
1639
1372
        return self.path2id('')
1640
1373
 
1641
 
    def id2path(self, file_id):
1642
 
        "Convert a file-id to a path."
1643
 
        entry = self._get_entry(file_id=file_id)
1644
 
        if entry == (None, None):
1645
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1646
 
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1647
 
        return path_utf8.decode('utf8')
1648
 
 
1649
 
    def iter_references(self):
1650
 
        if not self._repo_supports_tree_reference:
1651
 
            # When the repo doesn't support references, we will have nothing to
1652
 
            # return
1653
 
            return iter([])
1654
 
        # Otherwise, fall back to the default implementation
1655
 
        return super(DirStateRevisionTree, self).iter_references()
1656
 
 
1657
1374
    def _get_parent_index(self):
1658
1375
        """Return the index in the dirstate referenced by this tree."""
1659
1376
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1664
1381
        If either file_id or path is supplied, it is used as the key to lookup.
1665
1382
        If both are supplied, the fastest lookup is used, and an error is
1666
1383
        raised if they do not both point at the same row.
1667
 
 
 
1384
        
1668
1385
        :param file_id: An optional unicode file_id to be looked up.
1669
1386
        :param path: An optional unicode path to be looked up.
1670
1387
        :return: The dirstate row tuple for path/file_id, or (None, None)
1684
1401
 
1685
1402
        This is relatively expensive: we have to walk the entire dirstate.
1686
1403
        """
1687
 
        if not self._locked:
1688
 
            raise AssertionError(
1689
 
                'cannot generate inventory of an unlocked '
1690
 
                'dirstate revision tree')
 
1404
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1405
            'dirstate revision tree'
1691
1406
        # separate call for profiling - makes it clear where the costs are.
1692
1407
        self._dirstate._read_dirblocks_if_needed()
1693
 
        if self._revision_id not in self._dirstate.get_parent_ids():
1694
 
            raise AssertionError(
1695
 
                'parent %s has disappeared from %s' % (
1696
 
                self._revision_id, self._dirstate.get_parent_ids()))
 
1408
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1409
            'parent %s has disappeared from %s' % (
 
1410
            self._revision_id, self._dirstate.get_parent_ids())
1697
1411
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1698
1412
        # This is identical now to the WorkingTree _generate_inventory except
1699
1413
        # for the tree index use.
1700
1414
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1701
1415
        current_id = root_key[2]
1702
 
        if current_entry[parent_index][0] != 'd':
1703
 
            raise AssertionError()
 
1416
        assert current_entry[parent_index][0] == 'd'
1704
1417
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1705
1418
        inv.root.revision = current_entry[parent_index][4]
1706
1419
        # Turn some things into local variables
1746
1459
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1747
1460
                            % entry)
1748
1461
                # These checks cost us around 40ms on a 55k entry tree
1749
 
                if file_id in inv_byid:
1750
 
                    raise AssertionError('file_id %s already in'
1751
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
1752
 
                if name_unicode in parent_ie.children:
1753
 
                    raise AssertionError('name %r already in parent'
1754
 
                        % (name_unicode,))
 
1462
                assert file_id not in inv_byid
 
1463
                assert name_unicode not in parent_ie.children
1755
1464
                inv_byid[file_id] = inv_entry
1756
1465
                parent_ie.children[name_unicode] = inv_entry
1757
1466
        self._inventory = inv
1767
1476
            return None
1768
1477
        parent_index = self._get_parent_index()
1769
1478
        last_changed_revision = entry[1][parent_index][4]
1770
 
        try:
1771
 
            rev = self._repository.get_revision(last_changed_revision)
1772
 
        except errors.NoSuchRevision:
1773
 
            raise errors.FileTimestampUnavailable(self.id2path(file_id))
1774
 
        return rev.timestamp
 
1479
        return self._repository.get_revision(last_changed_revision).timestamp
1775
1480
 
1776
1481
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1777
1482
        entry = self._get_entry(file_id=file_id, path=path)
1781
1486
            return parent_details[1]
1782
1487
        return None
1783
1488
 
 
1489
    @symbol_versioning.deprecated_method(symbol_versioning.zero_ninety)
 
1490
    def get_weave(self, file_id):
 
1491
        return self._get_weave(file_id)
 
1492
 
 
1493
    def _get_weave(self, file_id):
 
1494
        return self._repository.weave_store.get_weave(file_id,
 
1495
                self._repository.get_transaction())
 
1496
 
1784
1497
    def get_file(self, file_id, path=None):
1785
1498
        return StringIO(self.get_file_text(file_id))
1786
1499
 
 
1500
    def get_file_lines(self, file_id):
 
1501
        ie = self.inventory[file_id]
 
1502
        return self._get_weave(file_id).get_lines(ie.revision)
 
1503
 
1787
1504
    def get_file_size(self, file_id):
1788
 
        """See Tree.get_file_size"""
1789
1505
        return self.inventory[file_id].text_size
1790
1506
 
1791
 
    def get_file_text(self, file_id, path=None):
1792
 
        _, content = list(self.iter_files_bytes([(file_id, None)]))[0]
1793
 
        return ''.join(content)
 
1507
    def get_file_text(self, file_id):
 
1508
        return ''.join(self.get_file_lines(file_id))
1794
1509
 
1795
1510
    def get_reference_revision(self, file_id, path=None):
1796
1511
        return self.inventory[file_id].reference_revision
1815
1530
        if entry[1][parent_index][0] != 'l':
1816
1531
            return None
1817
1532
        else:
1818
 
            target = entry[1][parent_index][1]
1819
 
            target = target.decode('utf8')
1820
 
            return target
 
1533
            # At present, none of the tree implementations supports non-ascii
 
1534
            # symlink targets. So we will just assume that the dirstate path is
 
1535
            # correct.
 
1536
            return entry[1][parent_index][1]
1821
1537
 
1822
1538
    def get_revision_id(self):
1823
1539
        """Return the revision id for this tree."""
1841
1557
        return bool(self.path2id(filename))
1842
1558
 
1843
1559
    def kind(self, file_id):
1844
 
        entry = self._get_entry(file_id=file_id)[1]
1845
 
        if entry is None:
1846
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1847
 
        parent_index = self._get_parent_index()
1848
 
        return dirstate.DirState._minikind_to_kind[entry[parent_index][0]]
1849
 
 
1850
 
    def stored_kind(self, file_id):
1851
 
        """See Tree.stored_kind"""
1852
 
        return self.kind(file_id)
 
1560
        return self.inventory[file_id].kind
1853
1561
 
1854
1562
    def path_content_summary(self, path):
1855
1563
        """See Tree.path_content_summary."""
1871
1579
            return None
1872
1580
        return ie.executable
1873
1581
 
1874
 
    def is_locked(self):
1875
 
        return self._locked
1876
 
 
1877
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1582
    def list_files(self, include_root=False):
1878
1583
        # We use a standard implementation, because DirStateRevisionTree is
1879
1584
        # dealing with one of the parents of the current state
1880
1585
        inv = self._get_inventory()
1881
 
        if from_dir is None:
1882
 
            from_dir_id = None
1883
 
        else:
1884
 
            from_dir_id = inv.path2id(from_dir)
1885
 
            if from_dir_id is None:
1886
 
                # Directory not versioned
1887
 
                return
1888
 
        entries = inv.iter_entries(from_dir=from_dir_id, recursive=recursive)
1889
 
        if inv.root is not None and not include_root and from_dir is None:
 
1586
        entries = inv.iter_entries()
 
1587
        if self.inventory.root is not None and not include_root:
1890
1588
            entries.next()
1891
1589
        for path, entry in entries:
1892
1590
            yield path, 'V', entry.kind, entry.file_id, entry
1893
1591
 
1894
1592
    def lock_read(self):
1895
 
        """Lock the tree for a set of operations.
1896
 
 
1897
 
        :return: A bzrlib.lock.LogicalLockResult.
1898
 
        """
 
1593
        """Lock the tree for a set of operations."""
1899
1594
        if not self._locked:
1900
1595
            self._repository.lock_read()
1901
1596
            if self._dirstate._lock_token is None:
1902
1597
                self._dirstate.lock_read()
1903
1598
                self._dirstate_locked = True
1904
1599
        self._locked += 1
1905
 
        return LogicalLockResult(self.unlock)
1906
1600
 
1907
1601
    def _must_be_locked(self):
1908
1602
        if not self._locked:
1929
1623
                self._dirstate_locked = False
1930
1624
            self._repository.unlock()
1931
1625
 
1932
 
    @needs_read_lock
1933
 
    def supports_tree_reference(self):
1934
 
        return self._repo_supports_tree_reference
1935
 
 
1936
1626
    def walkdirs(self, prefix=""):
1937
1627
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1938
 
        # implementation based on an inventory.
 
1628
        # implementation based on an inventory.  
1939
1629
        # This should be cleaned up to use the much faster Dirstate code
1940
1630
        # So for now, we just build up the parent inventory, and extract
1941
1631
        # it the same way RevisionTree does.
1970
1660
 
1971
1661
class InterDirStateTree(InterTree):
1972
1662
    """Fast path optimiser for changes_from with dirstate trees.
1973
 
 
1974
 
    This is used only when both trees are in the dirstate working file, and
1975
 
    the source is any parent within the dirstate, and the destination is
 
1663
    
 
1664
    This is used only when both trees are in the dirstate working file, and 
 
1665
    the source is any parent within the dirstate, and the destination is 
1976
1666
    the current working tree of the same dirstate.
1977
1667
    """
1978
1668
    # this could be generalized to allow comparisons between any trees in the
1991
1681
        target.set_parent_ids([revid])
1992
1682
        return target.basis_tree(), target
1993
1683
 
1994
 
    @classmethod
1995
 
    def make_source_parent_tree_python_dirstate(klass, test_case, source, target):
1996
 
        result = klass.make_source_parent_tree(source, target)
1997
 
        result[1]._iter_changes = dirstate.ProcessEntryPython
1998
 
        return result
1999
 
 
2000
 
    @classmethod
2001
 
    def make_source_parent_tree_compiled_dirstate(klass, test_case, source,
2002
 
                                                  target):
2003
 
        from bzrlib.tests.test__dirstate_helpers import \
2004
 
            compiled_dirstate_helpers_feature
2005
 
        test_case.requireFeature(compiled_dirstate_helpers_feature)
2006
 
        from bzrlib._dirstate_helpers_pyx import ProcessEntryC
2007
 
        result = klass.make_source_parent_tree(source, target)
2008
 
        result[1]._iter_changes = ProcessEntryC
2009
 
        return result
2010
 
 
2011
1684
    _matching_from_tree_format = WorkingTreeFormat4()
2012
1685
    _matching_to_tree_format = WorkingTreeFormat4()
2013
 
 
2014
 
    @classmethod
2015
 
    def _test_mutable_trees_to_test_trees(klass, test_case, source, target):
2016
 
        # This method shouldn't be called, because we have python and C
2017
 
        # specific flavours.
2018
 
        raise NotImplementedError
2019
 
 
2020
 
    def iter_changes(self, include_unchanged=False,
 
1686
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1687
 
 
1688
    def _iter_changes(self, include_unchanged=False,
2021
1689
                      specific_files=None, pb=None, extra_trees=[],
2022
1690
                      require_versioned=True, want_unversioned=False):
2023
1691
        """Return the changes from source to target.
2024
1692
 
2025
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1693
        :return: An iterator that yields tuples. See InterTree._iter_changes
2026
1694
            for details.
2027
1695
        :param specific_files: An optional list of file paths to restrict the
2028
1696
            comparison to. When mapping filenames to ids, all matches in all
2039
1707
            output. An unversioned file is defined as one with (False, False)
2040
1708
            for the versioned pair.
2041
1709
        """
 
1710
        utf8_decode = cache_utf8._utf8_decode
 
1711
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1712
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1713
        # NB: show_status depends on being able to pass in non-versioned files
 
1714
        # and report them as unknown
2042
1715
        # TODO: handle extra trees in the dirstate.
2043
 
        if (extra_trees or specific_files == []):
 
1716
        # TODO: handle comparisons as an empty tree as a different special
 
1717
        # case? mbp 20070226
 
1718
        if (extra_trees or (self.source._revision_id == NULL_REVISION)
 
1719
            or specific_files == []):
2044
1720
            # we can't fast-path these cases (yet)
2045
 
            return super(InterDirStateTree, self).iter_changes(
 
1721
            for f in super(InterDirStateTree, self)._iter_changes(
2046
1722
                include_unchanged, specific_files, pb, extra_trees,
2047
 
                require_versioned, want_unversioned=want_unversioned)
 
1723
                require_versioned, want_unversioned=want_unversioned):
 
1724
                yield f
 
1725
            return
2048
1726
        parent_ids = self.target.get_parent_ids()
2049
 
        if not (self.source._revision_id in parent_ids
2050
 
                or self.source._revision_id == _mod_revision.NULL_REVISION):
2051
 
            raise AssertionError(
2052
 
                "revision {%s} is not stored in {%s}, but %s "
2053
 
                "can only be used for trees stored in the dirstate"
2054
 
                % (self.source._revision_id, self.target, self.iter_changes))
 
1727
        assert (self.source._revision_id in parent_ids), \
 
1728
                "revision {%s} is not stored in {%s}, but %s " \
 
1729
                "can only be used for trees stored in the dirstate" \
 
1730
                % (self.source._revision_id, self.target, self._iter_changes)
2055
1731
        target_index = 0
2056
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1732
        if self.source._revision_id == NULL_REVISION:
2057
1733
            source_index = None
2058
1734
            indices = (target_index,)
2059
1735
        else:
2060
 
            if not (self.source._revision_id in parent_ids):
2061
 
                raise AssertionError(
2062
 
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
2063
 
                    self.source._revision_id, parent_ids))
 
1736
            assert (self.source._revision_id in parent_ids), \
 
1737
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1738
                self.source._revision_id, parent_ids)
2064
1739
            source_index = 1 + parent_ids.index(self.source._revision_id)
2065
 
            indices = (source_index, target_index)
 
1740
            indices = (source_index,target_index)
2066
1741
        # -- make all specific_files utf8 --
2067
1742
        if specific_files:
2068
1743
            specific_files_utf8 = set()
2069
1744
            for path in specific_files:
2070
 
                # Note, if there are many specific files, using cache_utf8
2071
 
                # would be good here.
2072
1745
                specific_files_utf8.add(path.encode('utf8'))
2073
1746
            specific_files = specific_files_utf8
2074
1747
        else:
2075
1748
            specific_files = set([''])
2076
1749
        # -- specific_files is now a utf8 path set --
2077
 
 
2078
1750
        # -- get the state object and prepare it.
2079
1751
        state = self.target.current_dirstate()
2080
1752
        state._read_dirblocks_if_needed()
 
1753
        def _entries_for_path(path):
 
1754
            """Return a list with all the entries that match path for all ids.
 
1755
            """
 
1756
            dirname, basename = os.path.split(path)
 
1757
            key = (dirname, basename, '')
 
1758
            block_index, present = state._find_block_index_from_key(key)
 
1759
            if not present:
 
1760
                # the block which should contain path is absent.
 
1761
                return []
 
1762
            result = []
 
1763
            block = state._dirblocks[block_index][1]
 
1764
            entry_index, _ = state._find_entry_index(key, block)
 
1765
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1766
            while (entry_index < len(block) and
 
1767
                block[entry_index][0][0:2] == key[0:2]):
 
1768
                result.append(block[entry_index])
 
1769
                entry_index += 1
 
1770
            return result
2081
1771
        if require_versioned:
2082
1772
            # -- check all supplied paths are versioned in a search tree. --
2083
 
            not_versioned = []
 
1773
            all_versioned = True
2084
1774
            for path in specific_files:
2085
 
                path_entries = state._entries_for_path(path)
 
1775
                path_entries = _entries_for_path(path)
2086
1776
                if not path_entries:
2087
1777
                    # this specified path is not present at all: error
2088
 
                    not_versioned.append(path)
2089
 
                    continue
 
1778
                    all_versioned = False
 
1779
                    break
2090
1780
                found_versioned = False
2091
1781
                # for each id at this path
2092
1782
                for entry in path_entries:
2099
1789
                if not found_versioned:
2100
1790
                    # none of the indexes was not 'absent' at all ids for this
2101
1791
                    # path.
2102
 
                    not_versioned.append(path)
2103
 
            if len(not_versioned) > 0:
2104
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1792
                    all_versioned = False
 
1793
                    break
 
1794
            if not all_versioned:
 
1795
                raise errors.PathsNotVersionedError(specific_files)
2105
1796
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2106
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1797
        search_specific_files = set()
 
1798
        for path in specific_files:
 
1799
            other_specific_files = specific_files.difference(set([path]))
 
1800
            if not osutils.is_inside_any(other_specific_files, path):
 
1801
                # this is a top level path, we must check it.
 
1802
                search_specific_files.add(path)
 
1803
        # sketch: 
 
1804
        # compare source_index and target_index at or under each element of search_specific_files.
 
1805
        # follow the following comparison table. Note that we only want to do diff operations when
 
1806
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1807
        # for the target.
 
1808
        # cases:
 
1809
        # 
 
1810
        # Source | Target | disk | action
 
1811
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1812
        #        |        |      | diff check on source-target
 
1813
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1814
        #        |        |      | ???
 
1815
        #   r    |  a     |      | add source to search
 
1816
        #   r    |  a     |  a   | 
 
1817
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1818
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1819
        #   a    | fdlt   |      | add new id
 
1820
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1821
        #   a    |  a     |      | not present in either tree, skip
 
1822
        #   a    |  a     |  a   | not present in any tree, skip
 
1823
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1824
        #        |        |      | may not be selected by the users list of paths.
 
1825
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1826
        #        |        |      | may not be selected by the users list of paths.
 
1827
        #  fdlt  | fdlt   |      | content in both: diff them
 
1828
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1829
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1830
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1831
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1832
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1833
        #        |        |      | this id at the other path.
 
1834
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1835
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1836
        #        |        |      | this id at the other path.
 
1837
 
 
1838
        # for all search_indexs in each path at or under each element of
 
1839
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1840
        # relocated path as one to search if its not searched already. If the
 
1841
        # detail is not relocated, add the id.
 
1842
        searched_specific_files = set()
 
1843
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1844
        # Using a list so that we can access the values and change them in
 
1845
        # nested scope. Each one is [path, file_id, entry]
 
1846
        last_source_parent = [None, None]
 
1847
        last_target_parent = [None, None]
2107
1848
 
2108
1849
        use_filesystem_for_exec = (sys.platform != 'win32')
2109
 
        iter_changes = self.target._iter_changes(include_unchanged,
2110
 
            use_filesystem_for_exec, search_specific_files, state,
2111
 
            source_index, target_index, want_unversioned, self.target)
2112
 
        return iter_changes.iter_changes()
 
1850
 
 
1851
        # Just a sentry, so that _process_entry can say that this
 
1852
        # record is handled, but isn't interesting to process (unchanged)
 
1853
        uninteresting = object()
 
1854
 
 
1855
 
 
1856
        old_dirname_to_file_id = {}
 
1857
        new_dirname_to_file_id = {}
 
1858
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1859
        #       keeping a cache of directories that we have seen.
 
1860
 
 
1861
        def _process_entry(entry, path_info):
 
1862
            """Compare an entry and real disk to generate delta information.
 
1863
 
 
1864
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1865
                the path of entry. If None, then the path is considered absent.
 
1866
                (Perhaps we should pass in a concrete entry for this ?)
 
1867
                Basename is returned as a utf8 string because we expect this
 
1868
                tuple will be ignored, and don't want to take the time to
 
1869
                decode.
 
1870
            :return: None if these don't match
 
1871
                     A tuple of information about the change, or
 
1872
                     the object 'uninteresting' if these match, but are
 
1873
                     basically identical.
 
1874
            """
 
1875
            if source_index is None:
 
1876
                source_details = NULL_PARENT_DETAILS
 
1877
            else:
 
1878
                source_details = entry[1][source_index]
 
1879
            target_details = entry[1][target_index]
 
1880
            target_minikind = target_details[0]
 
1881
            if path_info is not None and target_minikind in 'fdlt':
 
1882
                assert target_index == 0
 
1883
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1884
                                                  stat_value=path_info[3])
 
1885
                # The entry may have been modified by update_entry
 
1886
                target_details = entry[1][target_index]
 
1887
                target_minikind = target_details[0]
 
1888
            else:
 
1889
                link_or_sha1 = None
 
1890
            file_id = entry[0][2]
 
1891
            source_minikind = source_details[0]
 
1892
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1893
                # claimed content in both: diff
 
1894
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1895
                #        |        |      | diff check on source-target
 
1896
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1897
                #        |        |      | ???
 
1898
                if source_minikind in 'r':
 
1899
                    # add the source to the search path to find any children it
 
1900
                    # has.  TODO ? : only add if it is a container ?
 
1901
                    if not osutils.is_inside_any(searched_specific_files,
 
1902
                                                 source_details[1]):
 
1903
                        search_specific_files.add(source_details[1])
 
1904
                    # generate the old path; this is needed for stating later
 
1905
                    # as well.
 
1906
                    old_path = source_details[1]
 
1907
                    old_dirname, old_basename = os.path.split(old_path)
 
1908
                    path = pathjoin(entry[0][0], entry[0][1])
 
1909
                    old_entry = state._get_entry(source_index,
 
1910
                                                 path_utf8=old_path)
 
1911
                    # update the source details variable to be the real
 
1912
                    # location.
 
1913
                    source_details = old_entry[1][source_index]
 
1914
                    source_minikind = source_details[0]
 
1915
                else:
 
1916
                    old_dirname = entry[0][0]
 
1917
                    old_basename = entry[0][1]
 
1918
                    old_path = path = None
 
1919
                if path_info is None:
 
1920
                    # the file is missing on disk, show as removed.
 
1921
                    content_change = True
 
1922
                    target_kind = None
 
1923
                    target_exec = False
 
1924
                else:
 
1925
                    # source and target are both versioned and disk file is present.
 
1926
                    target_kind = path_info[2]
 
1927
                    if target_kind == 'directory':
 
1928
                        if path is None:
 
1929
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1930
                        new_dirname_to_file_id[path] = file_id
 
1931
                        if source_minikind != 'd':
 
1932
                            content_change = True
 
1933
                        else:
 
1934
                            # directories have no fingerprint
 
1935
                            content_change = False
 
1936
                        target_exec = False
 
1937
                    elif target_kind == 'file':
 
1938
                        if source_minikind != 'f':
 
1939
                            content_change = True
 
1940
                        else:
 
1941
                            # We could check the size, but we already have the
 
1942
                            # sha1 hash.
 
1943
                            content_change = (link_or_sha1 != source_details[1])
 
1944
                        # Target details is updated at update_entry time
 
1945
                        if use_filesystem_for_exec:
 
1946
                            # We don't need S_ISREG here, because we are sure
 
1947
                            # we are dealing with a file.
 
1948
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1949
                        else:
 
1950
                            target_exec = target_details[3]
 
1951
                    elif target_kind == 'symlink':
 
1952
                        if source_minikind != 'l':
 
1953
                            content_change = True
 
1954
                        else:
 
1955
                            content_change = (link_or_sha1 != source_details[1])
 
1956
                        target_exec = False
 
1957
                    elif target_kind == 'tree-reference':
 
1958
                        if source_minikind != 't':
 
1959
                            content_change = True
 
1960
                        else:
 
1961
                            content_change = False
 
1962
                        target_exec = False
 
1963
                    else:
 
1964
                        raise Exception, "unknown kind %s" % path_info[2]
 
1965
                if source_minikind == 'd':
 
1966
                    if path is None:
 
1967
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1968
                    old_dirname_to_file_id[old_path] = file_id
 
1969
                # parent id is the entry for the path in the target tree
 
1970
                if old_dirname == last_source_parent[0]:
 
1971
                    source_parent_id = last_source_parent[1]
 
1972
                else:
 
1973
                    try:
 
1974
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1975
                    except KeyError:
 
1976
                        source_parent_entry = state._get_entry(source_index,
 
1977
                                                               path_utf8=old_dirname)
 
1978
                        source_parent_id = source_parent_entry[0][2]
 
1979
                    if source_parent_id == entry[0][2]:
 
1980
                        # This is the root, so the parent is None
 
1981
                        source_parent_id = None
 
1982
                    else:
 
1983
                        last_source_parent[0] = old_dirname
 
1984
                        last_source_parent[1] = source_parent_id
 
1985
                new_dirname = entry[0][0]
 
1986
                if new_dirname == last_target_parent[0]:
 
1987
                    target_parent_id = last_target_parent[1]
 
1988
                else:
 
1989
                    try:
 
1990
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
1991
                    except KeyError:
 
1992
                        # TODO: We don't always need to do the lookup, because the
 
1993
                        #       parent entry will be the same as the source entry.
 
1994
                        target_parent_entry = state._get_entry(target_index,
 
1995
                                                               path_utf8=new_dirname)
 
1996
                        assert target_parent_entry != (None, None), (
 
1997
                            "Could not find target parent in wt: %s\nparent of: %s"
 
1998
                            % (new_dirname, entry))
 
1999
                        target_parent_id = target_parent_entry[0][2]
 
2000
                    if target_parent_id == entry[0][2]:
 
2001
                        # This is the root, so the parent is None
 
2002
                        target_parent_id = None
 
2003
                    else:
 
2004
                        last_target_parent[0] = new_dirname
 
2005
                        last_target_parent[1] = target_parent_id
 
2006
 
 
2007
                source_exec = source_details[3]
 
2008
                if (include_unchanged
 
2009
                    or content_change
 
2010
                    or source_parent_id != target_parent_id
 
2011
                    or old_basename != entry[0][1]
 
2012
                    or source_exec != target_exec
 
2013
                    ):
 
2014
                    if old_path is None:
 
2015
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2016
                        old_path_u = utf8_decode(old_path)[0]
 
2017
                        path_u = old_path_u
 
2018
                    else:
 
2019
                        old_path_u = utf8_decode(old_path)[0]
 
2020
                        if old_path == path:
 
2021
                            path_u = old_path_u
 
2022
                        else:
 
2023
                            path_u = utf8_decode(path)[0]
 
2024
                    source_kind = _minikind_to_kind[source_minikind]
 
2025
                    return (entry[0][2],
 
2026
                           (old_path_u, path_u),
 
2027
                           content_change,
 
2028
                           (True, True),
 
2029
                           (source_parent_id, target_parent_id),
 
2030
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2031
                           (source_kind, target_kind),
 
2032
                           (source_exec, target_exec))
 
2033
                else:
 
2034
                    return uninteresting
 
2035
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2036
                # looks like a new file
 
2037
                if path_info is not None:
 
2038
                    path = pathjoin(entry[0][0], entry[0][1])
 
2039
                    # parent id is the entry for the path in the target tree
 
2040
                    # TODO: these are the same for an entire directory: cache em.
 
2041
                    parent_id = state._get_entry(target_index,
 
2042
                                                 path_utf8=entry[0][0])[0][2]
 
2043
                    if parent_id == entry[0][2]:
 
2044
                        parent_id = None
 
2045
                    if use_filesystem_for_exec:
 
2046
                        # We need S_ISREG here, because we aren't sure if this
 
2047
                        # is a file or not.
 
2048
                        target_exec = bool(
 
2049
                            stat.S_ISREG(path_info[3].st_mode)
 
2050
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2051
                    else:
 
2052
                        target_exec = target_details[3]
 
2053
                    return (entry[0][2],
 
2054
                           (None, utf8_decode(path)[0]),
 
2055
                           True,
 
2056
                           (False, True),
 
2057
                           (None, parent_id),
 
2058
                           (None, utf8_decode(entry[0][1])[0]),
 
2059
                           (None, path_info[2]),
 
2060
                           (None, target_exec))
 
2061
                else:
 
2062
                    # but its not on disk: we deliberately treat this as just
 
2063
                    # never-present. (Why ?! - RBC 20070224)
 
2064
                    pass
 
2065
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2066
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2067
                # if its still on disk, *and* theres no other entry at this
 
2068
                # path [we dont know this in this routine at the moment -
 
2069
                # perhaps we should change this - then it would be an unknown.
 
2070
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2071
                # parent id is the entry for the path in the target tree
 
2072
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2073
                if parent_id == entry[0][2]:
 
2074
                    parent_id = None
 
2075
                return (entry[0][2],
 
2076
                       (utf8_decode(old_path)[0], None),
 
2077
                       True,
 
2078
                       (True, False),
 
2079
                       (parent_id, None),
 
2080
                       (utf8_decode(entry[0][1])[0], None),
 
2081
                       (_minikind_to_kind[source_minikind], None),
 
2082
                       (source_details[3], None))
 
2083
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2084
                # a rename; could be a true rename, or a rename inherited from
 
2085
                # a renamed parent. TODO: handle this efficiently. Its not
 
2086
                # common case to rename dirs though, so a correct but slow
 
2087
                # implementation will do.
 
2088
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2089
                    search_specific_files.add(target_details[1])
 
2090
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2091
                # neither of the selected trees contain this file,
 
2092
                # so skip over it. This is not currently directly tested, but
 
2093
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2094
                pass
 
2095
            else:
 
2096
                raise AssertionError("don't know how to compare "
 
2097
                    "source_minikind=%r, target_minikind=%r"
 
2098
                    % (source_minikind, target_minikind))
 
2099
                ## import pdb;pdb.set_trace()
 
2100
            return None
 
2101
 
 
2102
        while search_specific_files:
 
2103
            # TODO: the pending list should be lexically sorted?  the
 
2104
            # interface doesn't require it.
 
2105
            current_root = search_specific_files.pop()
 
2106
            current_root_unicode = current_root.decode('utf8')
 
2107
            searched_specific_files.add(current_root)
 
2108
            # process the entries for this containing directory: the rest will be
 
2109
            # found by their parents recursively.
 
2110
            root_entries = _entries_for_path(current_root)
 
2111
            root_abspath = self.target.abspath(current_root_unicode)
 
2112
            try:
 
2113
                root_stat = os.lstat(root_abspath)
 
2114
            except OSError, e:
 
2115
                if e.errno == errno.ENOENT:
 
2116
                    # the path does not exist: let _process_entry know that.
 
2117
                    root_dir_info = None
 
2118
                else:
 
2119
                    # some other random error: hand it up.
 
2120
                    raise
 
2121
            else:
 
2122
                root_dir_info = ('', current_root,
 
2123
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2124
                    root_abspath)
 
2125
                if root_dir_info[2] == 'directory':
 
2126
                    if self.target._directory_is_tree_reference(
 
2127
                        current_root.decode('utf8')):
 
2128
                        root_dir_info = root_dir_info[:2] + \
 
2129
                            ('tree-reference',) + root_dir_info[3:]
 
2130
 
 
2131
            if not root_entries and not root_dir_info:
 
2132
                # this specified path is not present at all, skip it.
 
2133
                continue
 
2134
            path_handled = False
 
2135
            for entry in root_entries:
 
2136
                result = _process_entry(entry, root_dir_info)
 
2137
                if result is not None:
 
2138
                    path_handled = True
 
2139
                    if result is not uninteresting:
 
2140
                        yield result
 
2141
            if want_unversioned and not path_handled and root_dir_info:
 
2142
                new_executable = bool(
 
2143
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2144
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2145
                yield (None,
 
2146
                       (None, current_root_unicode),
 
2147
                       True,
 
2148
                       (False, False),
 
2149
                       (None, None),
 
2150
                       (None, splitpath(current_root_unicode)[-1]),
 
2151
                       (None, root_dir_info[2]),
 
2152
                       (None, new_executable)
 
2153
                      )
 
2154
            initial_key = (current_root, '', '')
 
2155
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2156
            if block_index == 0:
 
2157
                # we have processed the total root already, but because the
 
2158
                # initial key matched it we should skip it here.
 
2159
                block_index +=1
 
2160
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2161
                current_dir_info = None
 
2162
            else:
 
2163
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2164
                try:
 
2165
                    current_dir_info = dir_iterator.next()
 
2166
                except OSError, e:
 
2167
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2168
                    # python 2.5 has e.errno == EINVAL,
 
2169
                    #            and e.winerror == ERROR_DIRECTORY
 
2170
                    e_winerror = getattr(e, 'winerror', None)
 
2171
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2172
                    # there may be directories in the inventory even though
 
2173
                    # this path is not a file on disk: so mark it as end of
 
2174
                    # iterator
 
2175
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2176
                        current_dir_info = None
 
2177
                    elif (sys.platform == 'win32'
 
2178
                          and (e.errno in win_errors
 
2179
                               or e_winerror in win_errors)):
 
2180
                        current_dir_info = None
 
2181
                    else:
 
2182
                        raise
 
2183
                else:
 
2184
                    if current_dir_info[0][0] == '':
 
2185
                        # remove .bzr from iteration
 
2186
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2187
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2188
                        del current_dir_info[1][bzr_index]
 
2189
            # walk until both the directory listing and the versioned metadata
 
2190
            # are exhausted. 
 
2191
            if (block_index < len(state._dirblocks) and
 
2192
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2193
                current_block = state._dirblocks[block_index]
 
2194
            else:
 
2195
                current_block = None
 
2196
            while (current_dir_info is not None or
 
2197
                   current_block is not None):
 
2198
                if (current_dir_info and current_block
 
2199
                    and current_dir_info[0][0] != current_block[0]):
 
2200
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2201
                        # filesystem data refers to paths not covered by the dirblock.
 
2202
                        # this has two possibilities:
 
2203
                        # A) it is versioned but empty, so there is no block for it
 
2204
                        # B) it is not versioned.
 
2205
 
 
2206
                        # if (A) then we need to recurse into it to check for
 
2207
                        # new unknown files or directories.
 
2208
                        # if (B) then we should ignore it, because we don't
 
2209
                        # recurse into unknown directories.
 
2210
                        path_index = 0
 
2211
                        while path_index < len(current_dir_info[1]):
 
2212
                                current_path_info = current_dir_info[1][path_index]
 
2213
                                if want_unversioned:
 
2214
                                    if current_path_info[2] == 'directory':
 
2215
                                        if self.target._directory_is_tree_reference(
 
2216
                                            current_path_info[0].decode('utf8')):
 
2217
                                            current_path_info = current_path_info[:2] + \
 
2218
                                                ('tree-reference',) + current_path_info[3:]
 
2219
                                    new_executable = bool(
 
2220
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2221
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2222
                                    yield (None,
 
2223
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2224
                                        True,
 
2225
                                        (False, False),
 
2226
                                        (None, None),
 
2227
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2228
                                        (None, current_path_info[2]),
 
2229
                                        (None, new_executable))
 
2230
                                # dont descend into this unversioned path if it is
 
2231
                                # a dir
 
2232
                                if current_path_info[2] in ('directory',
 
2233
                                                            'tree-reference'):
 
2234
                                    del current_dir_info[1][path_index]
 
2235
                                    path_index -= 1
 
2236
                                path_index += 1
 
2237
 
 
2238
                        # This dir info has been handled, go to the next
 
2239
                        try:
 
2240
                            current_dir_info = dir_iterator.next()
 
2241
                        except StopIteration:
 
2242
                            current_dir_info = None
 
2243
                    else:
 
2244
                        # We have a dirblock entry for this location, but there
 
2245
                        # is no filesystem path for this. This is most likely
 
2246
                        # because a directory was removed from the disk.
 
2247
                        # We don't have to report the missing directory,
 
2248
                        # because that should have already been handled, but we
 
2249
                        # need to handle all of the files that are contained
 
2250
                        # within.
 
2251
                        for current_entry in current_block[1]:
 
2252
                            # entry referring to file not present on disk.
 
2253
                            # advance the entry only, after processing.
 
2254
                            result = _process_entry(current_entry, None)
 
2255
                            if result is not None:
 
2256
                                if result is not uninteresting:
 
2257
                                    yield result
 
2258
                        block_index +=1
 
2259
                        if (block_index < len(state._dirblocks) and
 
2260
                            osutils.is_inside(current_root,
 
2261
                                              state._dirblocks[block_index][0])):
 
2262
                            current_block = state._dirblocks[block_index]
 
2263
                        else:
 
2264
                            current_block = None
 
2265
                    continue
 
2266
                entry_index = 0
 
2267
                if current_block and entry_index < len(current_block[1]):
 
2268
                    current_entry = current_block[1][entry_index]
 
2269
                else:
 
2270
                    current_entry = None
 
2271
                advance_entry = True
 
2272
                path_index = 0
 
2273
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2274
                    current_path_info = current_dir_info[1][path_index]
 
2275
                    if current_path_info[2] == 'directory':
 
2276
                        if self.target._directory_is_tree_reference(
 
2277
                            current_path_info[0].decode('utf8')):
 
2278
                            current_path_info = current_path_info[:2] + \
 
2279
                                ('tree-reference',) + current_path_info[3:]
 
2280
                else:
 
2281
                    current_path_info = None
 
2282
                advance_path = True
 
2283
                path_handled = False
 
2284
                while (current_entry is not None or
 
2285
                    current_path_info is not None):
 
2286
                    if current_entry is None:
 
2287
                        # the check for path_handled when the path is adnvaced
 
2288
                        # will yield this path if needed.
 
2289
                        pass
 
2290
                    elif current_path_info is None:
 
2291
                        # no path is fine: the per entry code will handle it.
 
2292
                        result = _process_entry(current_entry, current_path_info)
 
2293
                        if result is not None:
 
2294
                            if result is not uninteresting:
 
2295
                                yield result
 
2296
                    elif (current_entry[0][1] != current_path_info[1]
 
2297
                          or current_entry[1][target_index][0] in 'ar'):
 
2298
                        # The current path on disk doesn't match the dirblock
 
2299
                        # record. Either the dirblock is marked as absent, or
 
2300
                        # the file on disk is not present at all in the
 
2301
                        # dirblock. Either way, report about the dirblock
 
2302
                        # entry, and let other code handle the filesystem one.
 
2303
 
 
2304
                        # Compare the basename for these files to determine
 
2305
                        # which comes first
 
2306
                        if current_path_info[1] < current_entry[0][1]:
 
2307
                            # extra file on disk: pass for now, but only
 
2308
                            # increment the path, not the entry
 
2309
                            advance_entry = False
 
2310
                        else:
 
2311
                            # entry referring to file not present on disk.
 
2312
                            # advance the entry only, after processing.
 
2313
                            result = _process_entry(current_entry, None)
 
2314
                            if result is not None:
 
2315
                                if result is not uninteresting:
 
2316
                                    yield result
 
2317
                            advance_path = False
 
2318
                    else:
 
2319
                        result = _process_entry(current_entry, current_path_info)
 
2320
                        if result is not None:
 
2321
                            path_handled = True
 
2322
                            if result is not uninteresting:
 
2323
                                yield result
 
2324
                    if advance_entry and current_entry is not None:
 
2325
                        entry_index += 1
 
2326
                        if entry_index < len(current_block[1]):
 
2327
                            current_entry = current_block[1][entry_index]
 
2328
                        else:
 
2329
                            current_entry = None
 
2330
                    else:
 
2331
                        advance_entry = True # reset the advance flaga
 
2332
                    if advance_path and current_path_info is not None:
 
2333
                        if not path_handled:
 
2334
                            # unversioned in all regards
 
2335
                            if want_unversioned:
 
2336
                                new_executable = bool(
 
2337
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2338
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2339
                                yield (None,
 
2340
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2341
                                    True,
 
2342
                                    (False, False),
 
2343
                                    (None, None),
 
2344
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2345
                                    (None, current_path_info[2]),
 
2346
                                    (None, new_executable))
 
2347
                            # dont descend into this unversioned path if it is
 
2348
                            # a dir
 
2349
                            if current_path_info[2] in ('directory'):
 
2350
                                del current_dir_info[1][path_index]
 
2351
                                path_index -= 1
 
2352
                        # dont descend the disk iterator into any tree 
 
2353
                        # paths.
 
2354
                        if current_path_info[2] == 'tree-reference':
 
2355
                            del current_dir_info[1][path_index]
 
2356
                            path_index -= 1
 
2357
                        path_index += 1
 
2358
                        if path_index < len(current_dir_info[1]):
 
2359
                            current_path_info = current_dir_info[1][path_index]
 
2360
                            if current_path_info[2] == 'directory':
 
2361
                                if self.target._directory_is_tree_reference(
 
2362
                                    current_path_info[0].decode('utf8')):
 
2363
                                    current_path_info = current_path_info[:2] + \
 
2364
                                        ('tree-reference',) + current_path_info[3:]
 
2365
                        else:
 
2366
                            current_path_info = None
 
2367
                        path_handled = False
 
2368
                    else:
 
2369
                        advance_path = True # reset the advance flagg.
 
2370
                if current_block is not None:
 
2371
                    block_index += 1
 
2372
                    if (block_index < len(state._dirblocks) and
 
2373
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2374
                        current_block = state._dirblocks[block_index]
 
2375
                    else:
 
2376
                        current_block = None
 
2377
                if current_dir_info is not None:
 
2378
                    try:
 
2379
                        current_dir_info = dir_iterator.next()
 
2380
                    except StopIteration:
 
2381
                        current_dir_info = None
 
2382
 
2113
2383
 
2114
2384
    @staticmethod
2115
2385
    def is_compatible(source, target):
2116
2386
        # the target must be a dirstate working tree
2117
 
        if not isinstance(target, DirStateWorkingTree):
 
2387
        if not isinstance(target, WorkingTree4):
2118
2388
            return False
2119
 
        # the source must be a revtree or dirstate rev tree.
 
2389
        # the source must be a revtreee or dirstate rev tree.
2120
2390
        if not isinstance(source,
2121
2391
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2122
2392
            return False
2123
2393
        # the source revid must be in the target dirstate
2124
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2394
        if not (source._revision_id == NULL_REVISION or
2125
2395
            source._revision_id in target.get_parent_ids()):
2126
 
            # TODO: what about ghosts? it may well need to
 
2396
            # TODO: what about ghosts? it may well need to 
2127
2397
            # check for them explicitly.
2128
2398
            return False
2129
2399
        return True
2139
2409
 
2140
2410
    def convert(self, tree):
2141
2411
        # lock the control files not the tree, so that we dont get tree
2142
 
        # on-unlock behaviours, and so that noone else diddles with the
 
2412
        # on-unlock behaviours, and so that noone else diddles with the 
2143
2413
        # tree during upgrade.
2144
2414
        tree._control_files.lock_write()
2145
2415
        try:
2171
2441
 
2172
2442
    def update_format(self, tree):
2173
2443
        """Change the format marker."""
2174
 
        tree._transport.put_bytes('format',
2175
 
            self.target_format.get_format_string(),
2176
 
            mode=tree.bzrdir._get_file_mode())
2177
 
 
2178
 
 
2179
 
class Converter4to5(object):
2180
 
    """Perform an in-place upgrade of format 4 to format 5 trees."""
2181
 
 
2182
 
    def __init__(self):
2183
 
        self.target_format = WorkingTreeFormat5()
2184
 
 
2185
 
    def convert(self, tree):
2186
 
        # lock the control files not the tree, so that we don't get tree
2187
 
        # on-unlock behaviours, and so that no-one else diddles with the
2188
 
        # tree during upgrade.
2189
 
        tree._control_files.lock_write()
2190
 
        try:
2191
 
            self.update_format(tree)
2192
 
        finally:
2193
 
            tree._control_files.unlock()
2194
 
 
2195
 
    def update_format(self, tree):
2196
 
        """Change the format marker."""
2197
 
        tree._transport.put_bytes('format',
2198
 
            self.target_format.get_format_string(),
2199
 
            mode=tree.bzrdir._get_file_mode())
2200
 
 
2201
 
 
2202
 
class Converter4or5to6(object):
2203
 
    """Perform an in-place upgrade of format 4 or 5 to format 6 trees."""
2204
 
 
2205
 
    def __init__(self):
2206
 
        self.target_format = WorkingTreeFormat6()
2207
 
 
2208
 
    def convert(self, tree):
2209
 
        # lock the control files not the tree, so that we don't get tree
2210
 
        # on-unlock behaviours, and so that no-one else diddles with the
2211
 
        # tree during upgrade.
2212
 
        tree._control_files.lock_write()
2213
 
        try:
2214
 
            self.init_custom_control_files(tree)
2215
 
            self.update_format(tree)
2216
 
        finally:
2217
 
            tree._control_files.unlock()
2218
 
 
2219
 
    def init_custom_control_files(self, tree):
2220
 
        """Initialize custom control files."""
2221
 
        tree._transport.put_bytes('views', '',
2222
 
            mode=tree.bzrdir._get_file_mode())
2223
 
 
2224
 
    def update_format(self, tree):
2225
 
        """Change the format marker."""
2226
 
        tree._transport.put_bytes('format',
2227
 
            self.target_format.get_format_string(),
2228
 
            mode=tree.bzrdir._get_file_mode())
 
2444
        tree._control_files.put_utf8('format',
 
2445
            self.target_format.get_format_string())