~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2007-11-28 01:35:49 UTC
  • mfrom: (3036.1.4 check)
  • Revision ID: pqm@pqm.ubuntu.com-20071128013549-w54seazrf4oeywb9
(robertc) Fix check to correctly report unreferenced texts. (Robert
        Collins, John A Meinel, 162931, 165071)

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
 
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
56
75
from bzrlib.mutabletree import needs_tree_write_lock
57
76
from bzrlib.osutils import (
58
77
    file_kind,
59
78
    isdir,
 
79
    normpath,
60
80
    pathjoin,
 
81
    rand_chars,
61
82
    realpath,
62
83
    safe_unicode,
 
84
    splitpath,
63
85
    )
64
 
from bzrlib.trace import mutter
 
86
from bzrlib.trace import mutter, note
65
87
from bzrlib.transport.local import LocalTransport
66
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
        )
67
97
from bzrlib.tree import Tree
68
98
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
69
99
 
70
100
 
71
 
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
 
72
120
    def __init__(self, basedir,
73
121
                 branch,
74
122
                 _control_files=None,
83
131
        """
84
132
        self._format = _format
85
133
        self.bzrdir = _bzrdir
 
134
        assert isinstance(basedir, basestring), \
 
135
            "base directory %r is not a string" % basedir
86
136
        basedir = safe_unicode(basedir)
87
137
        mutter("opening working tree %r", basedir)
88
138
        self._branch = branch
 
139
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
140
            "branch %r is not a Branch" % self.branch
89
141
        self.basedir = realpath(basedir)
90
142
        # if branch is at our basedir and is a format 6 or less
91
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
92
146
        self._control_files = _control_files
93
 
        self._transport = self._control_files._transport
94
147
        self._dirty = None
95
148
        #-------------
96
149
        # during a read or write lock these objects are set, and are
98
151
        self._dirstate = None
99
152
        self._inventory = None
100
153
        #-------------
101
 
        self._setup_directory_is_tree_reference()
102
 
        self._detect_case_handling()
103
 
        self._rules_searcher = None
104
 
        self.views = self._make_views()
105
 
        #--- allow tests to select the dirstate iter_changes implementation
106
 
        self._iter_changes = dirstate._process_entry
107
154
 
108
155
    @needs_tree_write_lock
109
156
    def _add(self, files, ids, kinds):
137
184
    @needs_tree_write_lock
138
185
    def add_reference(self, sub_tree):
139
186
        # use standard implementation, which calls back to self._add
140
 
        #
 
187
        # 
141
188
        # So we don't store the reference_revision in the working dirstate,
142
 
        # it's just recorded at the moment of commit.
 
189
        # it's just recorded at the moment of commit. 
143
190
        self._add_reference(sub_tree)
144
191
 
145
192
    def break_lock(self):
184
231
            WorkingTree3._comparison_data(self, entry, path)
185
232
        # it looks like a plain directory, but it's really a reference -- see
186
233
        # also kind()
187
 
        if (self._repo_supports_tree_reference and kind == 'directory'
188
 
            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)):
189
237
            kind = 'tree-reference'
190
238
        return kind, executable, stat_value
191
239
 
217
265
            return self._dirstate
218
266
        local_path = self.bzrdir.get_workingtree_transport(None
219
267
            ).local_abspath('dirstate')
220
 
        self._dirstate = dirstate.DirState.on_file(local_path,
221
 
            self._sha1_provider())
 
268
        self._dirstate = dirstate.DirState.on_file(local_path)
222
269
        return self._dirstate
223
270
 
224
 
    def _sha1_provider(self):
225
 
        """A function that returns a SHA1Provider suitable for this tree.
226
 
 
227
 
        :return: None if content filtering is not supported by this tree.
228
 
          Otherwise, a SHA1Provider is returned that sha's the canonical
229
 
          form of files, i.e. after read filters are applied.
230
 
        """
231
 
        if self.supports_content_filtering():
232
 
            return ContentFilterAwareSHA1Provider(self)
233
 
        else:
234
 
            return None
235
 
 
236
271
    def filter_unversioned_files(self, paths):
237
272
        """Filter out paths that are versioned.
238
273
 
270
305
 
271
306
    def _generate_inventory(self):
272
307
        """Create and set self.inventory from the dirstate object.
273
 
 
 
308
        
274
309
        This is relatively expensive: we have to walk the entire dirstate.
275
310
        Ideally we would not, and can deprecate this function.
276
311
        """
280
315
        state._read_dirblocks_if_needed()
281
316
        root_key, current_entry = self._get_entry(path='')
282
317
        current_id = root_key[2]
283
 
        if not (current_entry[0][0] == 'd'): # directory
284
 
            raise AssertionError(current_entry)
 
318
        assert current_entry[0][0] == 'd' # directory
285
319
        inv = Inventory(root_id=current_id)
286
320
        # Turn some things into local variables
287
321
        minikind_to_kind = dirstate.DirState._minikind_to_kind
320
354
                    # add this entry to the parent map.
321
355
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
322
356
                elif kind == 'tree-reference':
323
 
                    if not self._repo_supports_tree_reference:
324
 
                        raise errors.UnsupportedOperation(
325
 
                            self._generate_inventory,
326
 
                            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)
327
362
                    inv_entry.reference_revision = link_or_sha1 or None
328
363
                elif kind != 'symlink':
329
364
                    raise AssertionError("unknown kind %r" % kind)
330
365
                # These checks cost us around 40ms on a 55k entry tree
331
 
                if file_id in inv_byid:
332
 
                    raise AssertionError('file_id %s already in'
333
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
334
 
                if name_unicode in parent_ie.children:
335
 
                    raise AssertionError('name %r already in parent'
336
 
                        % (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
337
369
                inv_byid[file_id] = inv_entry
338
370
                parent_ie.children[name_unicode] = inv_entry
339
371
        self._inventory = inv
344
376
        If either file_id or path is supplied, it is used as the key to lookup.
345
377
        If both are supplied, the fastest lookup is used, and an error is
346
378
        raised if they do not both point at the same row.
347
 
 
 
379
        
348
380
        :param file_id: An optional unicode file_id to be looked up.
349
381
        :param path: An optional unicode path to be looked up.
350
382
        :return: The dirstate row tuple for path/file_id, or (None, None)
374
406
                    return None
375
407
                else:
376
408
                    raise
377
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
378
 
            stat_value=stat_value)
 
409
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
410
                                          stat_value=stat_value)
379
411
        if entry[1][0][0] == 'f':
380
 
            if link_or_sha1 is None:
381
 
                file_obj, statvalue = self.get_file_with_stat(file_id, path)
382
 
                try:
383
 
                    sha1 = osutils.sha_file(file_obj)
384
 
                finally:
385
 
                    file_obj.close()
386
 
                self._observed_sha1(file_id, path, (sha1, statvalue))
387
 
                return sha1
388
 
            else:
389
 
                return link_or_sha1
 
412
            return link_or_sha1
390
413
        return None
391
414
 
392
415
    def _get_inventory(self):
406
429
    @needs_read_lock
407
430
    def get_parent_ids(self):
408
431
        """See Tree.get_parent_ids.
409
 
 
 
432
        
410
433
        This implementation requests the ids list from the dirstate file.
411
434
        """
412
435
        return self.current_dirstate().get_parent_ids()
434
457
        return osutils.lexists(pathjoin(
435
458
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
436
459
 
437
 
    def has_or_had_id(self, file_id):
438
 
        state = self.current_dirstate()
439
 
        row, parents = self._get_entry(file_id=file_id)
440
 
        return row is not None
441
 
 
442
460
    @needs_read_lock
443
461
    def id2path(self, file_id):
444
462
        "Convert a file-id to a path."
474
492
 
475
493
            Note: The caller is expected to take a read-lock before calling this.
476
494
            """
477
 
            self._must_be_locked()
478
495
            if not path:
479
496
                path = self.id2path(file_id)
480
497
            mode = os.lstat(self.abspath(path)).st_mode
481
498
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
482
499
 
483
 
    def all_file_ids(self):
484
 
        """See Tree.iter_all_file_ids"""
485
 
        self._must_be_locked()
486
 
        result = set()
487
 
        for key, tree_details in self.current_dirstate()._iter_entries():
488
 
            if tree_details[0][0] in ('a', 'r'): # relocated
489
 
                continue
490
 
            result.add(key[2])
491
 
        return result
492
 
 
493
500
    @needs_read_lock
494
501
    def __iter__(self):
495
502
        """Iterate through file_ids for this tree.
508
515
        return iter(result)
509
516
 
510
517
    def iter_references(self):
511
 
        if not self._repo_supports_tree_reference:
512
 
            # When the repo doesn't support references, we will have nothing to
513
 
            # return
514
 
            return
515
518
        for key, tree_details in self.current_dirstate()._iter_entries():
516
519
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
517
520
                # not relevant to the working tree
519
522
            if not key[1]:
520
523
                # the root is not a reference.
521
524
                continue
522
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
525
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
523
526
            try:
524
 
                if self._kind(relpath) == 'tree-reference':
525
 
                    yield relpath, key[2]
 
527
                if self._kind(path) == 'tree-reference':
 
528
                    yield path, key[2]
526
529
            except errors.NoSuchFile:
527
530
                # path is missing on disk.
528
531
                continue
529
532
 
530
 
    def _observed_sha1(self, file_id, path, (sha1, statvalue)):
531
 
        """See MutableTree._observed_sha1."""
532
 
        state = self.current_dirstate()
533
 
        entry = self._get_entry(file_id=file_id, path=path)
534
 
        state._observed_sha1(entry, sha1, statvalue)
535
 
 
536
533
    def kind(self, file_id):
537
534
        """Return the kind of a file.
538
535
 
542
539
        Note: The caller is expected to take a read-lock before calling this.
543
540
        """
544
541
        relpath = self.id2path(file_id)
545
 
        if relpath is None:
546
 
            raise AssertionError(
547
 
                "path for id {%s} is None!" % file_id)
 
542
        assert relpath != None, \
 
543
            "path for id {%s} is None!" % file_id
548
544
        return self._kind(relpath)
549
545
 
550
546
    def _kind(self, relpath):
551
547
        abspath = self.abspath(relpath)
552
548
        kind = file_kind(abspath)
553
 
        if (self._repo_supports_tree_reference and kind == 'directory'):
554
 
            entry = self._get_entry(path=relpath)
555
 
            if entry[1] is not None:
556
 
                if entry[1][0][0] == 't':
557
 
                    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'
558
553
        return kind
559
554
 
560
555
    @needs_read_lock
623
618
        result = []
624
619
        if not from_paths:
625
620
            return result
 
621
 
626
622
        state = self.current_dirstate()
627
 
        if isinstance(from_paths, basestring):
628
 
            raise ValueError()
 
623
 
 
624
        assert not isinstance(from_paths, basestring)
629
625
        to_dir_utf8 = to_dir.encode('utf8')
630
626
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
631
627
        id_index = state._get_id_index()
653
649
        if self._inventory is not None:
654
650
            update_inventory = True
655
651
            inv = self.inventory
 
652
            to_dir_ie = inv[to_dir_id]
656
653
            to_dir_id = to_entry[0][2]
657
 
            to_dir_ie = inv[to_dir_id]
658
654
        else:
659
655
            update_inventory = False
660
656
 
691
687
            from_entry = self._get_entry(path=from_rel)
692
688
            if from_entry == (None, None):
693
689
                raise errors.BzrMoveFailedError(from_rel,to_dir,
694
 
                    errors.NotVersionedError(path=from_rel))
 
690
                    errors.NotVersionedError(path=str(from_rel)))
695
691
 
696
692
            from_id = from_entry[0][2]
697
693
            to_rel = pathjoin(to_dir, from_tail)
786
782
                if minikind == 'd':
787
783
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
788
784
                        """Recursively update all entries in this dirblock."""
789
 
                        if from_dir == '':
790
 
                            raise AssertionError("renaming root not supported")
 
785
                        assert from_dir != '', "renaming root not supported"
791
786
                        from_key = (from_dir, '')
792
787
                        from_block_idx, present = \
793
788
                            state._find_block_index_from_key(from_key)
806
801
 
807
802
                        # Grab a copy since move_one may update the list.
808
803
                        for entry in from_block[1][:]:
809
 
                            if not (entry[0][0] == from_dir):
810
 
                                raise AssertionError()
 
804
                            assert entry[0][0] == from_dir
811
805
                            cur_details = entry[1][0]
812
806
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
813
807
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
872
866
        for tree in trees:
873
867
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
874
868
                parents):
875
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
876
 
                    trees, require_versioned)
 
869
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
877
870
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
878
871
        # -- make all paths utf8 --
879
872
        paths_utf8 = set()
940
933
                raise errors.PathsNotVersionedError(paths)
941
934
        # -- remove redundancy in supplied paths to prevent over-scanning --
942
935
        search_paths = osutils.minimum_path_selection(paths)
943
 
        # sketch:
 
936
        # sketch: 
944
937
        # for all search_indexs in each path at or under each element of
945
938
        # search_paths, if the detail is relocated: add the id, and add the
946
939
        # relocated path as one to search if its not searched already. If the
1002
995
 
1003
996
    def read_working_inventory(self):
1004
997
        """Read the working inventory.
1005
 
 
 
998
        
1006
999
        This is a meaningless operation for dirstate, but we obey it anyhow.
1007
1000
        """
1008
1001
        return self.inventory
1026
1019
    def set_last_revision(self, new_revision):
1027
1020
        """Change the last revision in the working tree."""
1028
1021
        parents = self.get_parent_ids()
1029
 
        if new_revision in (_mod_revision.NULL_REVISION, None):
1030
 
            if len(parents) >= 2:
1031
 
                raise AssertionError(
1032
 
                    "setting the last parent to none with a pending merge is "
1033
 
                    "unsupported.")
 
1022
        if new_revision in (NULL_REVISION, None):
 
1023
            assert len(parents) < 2, (
 
1024
                "setting the last parent to none with a pending merge is "
 
1025
                "unsupported.")
1034
1026
            self.set_parent_ids([])
1035
1027
        else:
1036
1028
            self.set_parent_ids([new_revision] + parents[1:],
1039
1031
    @needs_tree_write_lock
1040
1032
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1041
1033
        """Set the parent ids to revision_ids.
1042
 
 
 
1034
        
1043
1035
        See also set_parent_trees. This api will try to retrieve the tree data
1044
1036
        for each element of revision_ids from the trees repository. If you have
1045
1037
        tree data already available, it is more efficient to use
1077
1069
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1078
1070
        real_trees = []
1079
1071
        ghosts = []
1080
 
 
1081
 
        parent_ids = [rev_id for rev_id, tree in parents_list]
1082
 
        graph = self.branch.repository.get_graph()
1083
 
        heads = graph.heads(parent_ids)
1084
 
        accepted_revisions = set()
1085
 
 
1086
1072
        # convert absent trees to the null tree, which we convert back to
1087
1073
        # missing on access.
1088
1074
        for rev_id, tree in parents_list:
1089
 
            if len(accepted_revisions) > 0:
1090
 
                # we always accept the first tree
1091
 
                if rev_id in accepted_revisions or rev_id not in heads:
1092
 
                    # We have already included either this tree, or its
1093
 
                    # descendent, so we skip it.
1094
 
                    continue
1095
1075
            _mod_revision.check_not_reserved_id(rev_id)
1096
1076
            if tree is not None:
1097
1077
                real_trees.append((rev_id, tree))
1098
1078
            else:
1099
1079
                real_trees.append((rev_id,
1100
 
                    self.branch.repository.revision_tree(
1101
 
                        _mod_revision.NULL_REVISION)))
 
1080
                    self.branch.repository.revision_tree(None)))
1102
1081
                ghosts.append(rev_id)
1103
 
            accepted_revisions.add(rev_id)
1104
1082
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
1105
1083
        self._make_dirty(reset_inventory=False)
1106
1084
 
1199
1177
                # just forget the whole block.
1200
1178
                entry_index = 0
1201
1179
                while entry_index < len(block[1]):
 
1180
                    # Mark this file id as having been removed
1202
1181
                    entry = block[1][entry_index]
1203
 
                    if entry[1][0][0] in 'ar':
1204
 
                        # don't remove absent or renamed entries
 
1182
                    ids_to_unversion.discard(entry[0][2])
 
1183
                    if (entry[1][0][0] in 'ar' # don't remove absent or renamed
 
1184
                                               # entries
 
1185
                        or not state._make_absent(entry)):
1205
1186
                        entry_index += 1
1206
 
                    else:
1207
 
                        # Mark this file id as having been removed
1208
 
                        ids_to_unversion.discard(entry[0][2])
1209
 
                        if not state._make_absent(entry):
1210
 
                            # The block has not shrunk.
1211
 
                            entry_index += 1
1212
1187
                # go to the next block. (At the moment we dont delete empty
1213
1188
                # dirblocks)
1214
1189
                block_index += 1
1237
1212
            for file_id in file_ids:
1238
1213
                self._inventory.remove_recursive_id(file_id)
1239
1214
 
1240
 
    @needs_tree_write_lock
1241
 
    def rename_one(self, from_rel, to_rel, after=False):
1242
 
        """See WorkingTree.rename_one"""
1243
 
        self.flush()
1244
 
        WorkingTree.rename_one(self, from_rel, to_rel, after)
1245
 
 
1246
 
    @needs_tree_write_lock
1247
 
    def apply_inventory_delta(self, changes):
1248
 
        """See MutableTree.apply_inventory_delta"""
1249
 
        state = self.current_dirstate()
1250
 
        state.update_by_delta(changes)
1251
 
        self._make_dirty(reset_inventory=True)
1252
 
 
1253
1215
    def update_basis_by_delta(self, new_revid, delta):
1254
1216
        """See MutableTree.update_basis_by_delta."""
1255
 
        if self.last_revision() == new_revid:
1256
 
            raise AssertionError()
 
1217
        assert self.last_revision() != new_revid
1257
1218
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
1258
1219
 
1259
1220
    @needs_read_lock
1263
1224
    @needs_tree_write_lock
1264
1225
    def _write_inventory(self, inv):
1265
1226
        """Write inventory as the current inventory."""
1266
 
        if self._dirty:
1267
 
            raise AssertionError("attempting to write an inventory when the "
1268
 
                "dirstate is dirty will lose pending changes")
1269
 
        had_inventory = self._inventory is not None
1270
 
        # Setting self._inventory = None forces the dirstate to regenerate the
1271
 
        # working inventory. We do this because self.inventory may be inv, or
1272
 
        # may have been modified, and either case would prevent a clean delta
1273
 
        # being created.
1274
 
        self._inventory = None
1275
 
        # generate a delta,
1276
 
        delta = inv._make_delta(self.inventory)
1277
 
        # and apply it.
1278
 
        self.apply_inventory_delta(delta)
1279
 
        if had_inventory:
 
1227
        assert not self._dirty, ("attempting to write an inventory when the "
 
1228
            "dirstate is dirty will cause data loss")
 
1229
        self.current_dirstate().set_state_from_inventory(inv)
 
1230
        self._make_dirty(reset_inventory=False)
 
1231
        if self._inventory is not None:
1280
1232
            self._inventory = inv
1281
1233
        self.flush()
1282
1234
 
1283
1235
 
1284
 
class ContentFilterAwareSHA1Provider(dirstate.SHA1Provider):
1285
 
 
1286
 
    def __init__(self, tree):
1287
 
        self.tree = tree
1288
 
 
1289
 
    def sha1(self, abspath):
1290
 
        """See dirstate.SHA1Provider.sha1()."""
1291
 
        filters = self.tree._content_filter_stack(
1292
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1293
 
        return internal_size_sha_file_byname(abspath, filters)[1]
1294
 
 
1295
 
    def stat_and_sha1(self, abspath):
1296
 
        """See dirstate.SHA1Provider.stat_and_sha1()."""
1297
 
        filters = self.tree._content_filter_stack(
1298
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1299
 
        file_obj = file(abspath, 'rb', 65000)
1300
 
        try:
1301
 
            statvalue = os.fstat(file_obj.fileno())
1302
 
            if filters:
1303
 
                file_obj = filtered_input_file(file_obj, filters)
1304
 
            sha1 = osutils.size_sha_file(file_obj)[1]
1305
 
        finally:
1306
 
            file_obj.close()
1307
 
        return statvalue, sha1
1308
 
 
1309
 
 
1310
 
class ContentFilteringDirStateWorkingTree(DirStateWorkingTree):
1311
 
    """Dirstate working tree that supports content filtering.
1312
 
 
1313
 
    The dirstate holds the hash and size of the canonical form of the file, 
1314
 
    and most methods must return that.
1315
 
    """
1316
 
 
1317
 
    def _file_content_summary(self, path, stat_result):
1318
 
        # This is to support the somewhat obsolete path_content_summary method
1319
 
        # with content filtering: see
1320
 
        # <https://bugs.edge.launchpad.net/bzr/+bug/415508>.
1321
 
        #
1322
 
        # If the dirstate cache is up to date and knows the hash and size,
1323
 
        # return that.
1324
 
        # Otherwise if there are no content filters, return the on-disk size
1325
 
        # and leave the hash blank.
1326
 
        # Otherwise, read and filter the on-disk file and use its size and
1327
 
        # hash.
1328
 
        #
1329
 
        # The dirstate doesn't store the size of the canonical form so we
1330
 
        # can't trust it for content-filtered trees.  We just return None.
1331
 
        dirstate_sha1 = self._dirstate.sha1_from_stat(path, stat_result)
1332
 
        executable = self._is_executable_from_path_and_stat(path, stat_result)
1333
 
        return ('file', None, executable, dirstate_sha1)
1334
 
 
1335
 
 
1336
 
class WorkingTree4(DirStateWorkingTree):
1337
 
    """This is the Format 4 working tree.
1338
 
 
1339
 
    This differs from WorkingTree3 by:
1340
 
     - Having a consolidated internal dirstate, stored in a
1341
 
       randomly-accessible sorted file on disk.
1342
 
     - Not having a regular inventory attribute.  One can be synthesized
1343
 
       on demand but this is expensive and should be avoided.
1344
 
 
1345
 
    This is new in bzr 0.15.
1346
 
    """
1347
 
 
1348
 
 
1349
 
class WorkingTree5(ContentFilteringDirStateWorkingTree):
1350
 
    """This is the Format 5 working tree.
1351
 
 
1352
 
    This differs from WorkingTree4 by:
1353
 
     - Supporting content filtering.
1354
 
 
1355
 
    This is new in bzr 1.11.
1356
 
    """
1357
 
 
1358
 
 
1359
 
class WorkingTree6(ContentFilteringDirStateWorkingTree):
1360
 
    """This is the Format 6 working tree.
1361
 
 
1362
 
    This differs from WorkingTree5 by:
1363
 
     - Supporting a current view that may mask the set of files in a tree
1364
 
       impacted by most user operations.
1365
 
 
1366
 
    This is new in bzr 1.14.
1367
 
    """
1368
 
 
1369
 
    def _make_views(self):
1370
 
        return views.PathBasedViews(self)
1371
 
 
1372
 
 
1373
 
class DirStateWorkingTreeFormat(WorkingTreeFormat3):
1374
 
 
1375
 
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
1376
 
                   accelerator_tree=None, hardlink=False):
 
1236
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1237
    """The first consolidated dirstate working tree format.
 
1238
 
 
1239
    This format:
 
1240
        - exists within a metadir controlling .bzr
 
1241
        - includes an explicit version marker for the workingtree control
 
1242
          files, separate from the BzrDir format
 
1243
        - modifies the hash cache format
 
1244
        - is new in bzr 0.15
 
1245
        - uses a LockDir to guard access to it.
 
1246
    """
 
1247
 
 
1248
    upgrade_recommended = False
 
1249
 
 
1250
    def get_format_string(self):
 
1251
        """See WorkingTreeFormat.get_format_string()."""
 
1252
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1253
 
 
1254
    def get_format_description(self):
 
1255
        """See WorkingTreeFormat.get_format_description()."""
 
1256
        return "Working tree format 4"
 
1257
 
 
1258
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None):
1377
1259
        """See WorkingTreeFormat.initialize().
1378
1260
 
1379
1261
        :param revision_id: allows creating a working tree at a different
1380
1262
        revision than the branch is at.
1381
 
        :param accelerator_tree: A tree which can be used for retrieving file
1382
 
            contents more quickly than the revision tree, i.e. a workingtree.
1383
 
            The revision tree will be used for cases where accelerator_tree's
1384
 
            content is different.
1385
 
        :param hardlink: If true, hard-link files from accelerator_tree,
1386
 
            where possible.
1387
1263
 
1388
1264
        These trees get an initial random root id, if their repository supports
1389
1265
        rich root data, TREE_ROOT otherwise.
1394
1270
        control_files = self._open_control_files(a_bzrdir)
1395
1271
        control_files.create_lock()
1396
1272
        control_files.lock_write()
1397
 
        transport.put_bytes('format', self.get_format_string(),
1398
 
            mode=a_bzrdir._get_file_mode())
 
1273
        control_files.put_utf8('format', self.get_format_string())
1399
1274
        if from_branch is not None:
1400
1275
            branch = from_branch
1401
1276
        else:
1407
1282
        state = dirstate.DirState.initialize(local_path)
1408
1283
        state.unlock()
1409
1284
        del state
1410
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1285
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1411
1286
                         branch,
1412
1287
                         _format=self,
1413
1288
                         _bzrdir=a_bzrdir,
1415
1290
        wt._new_tree()
1416
1291
        wt.lock_tree_write()
1417
1292
        try:
1418
 
            self._init_custom_control_files(wt)
1419
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
 
1293
            if revision_id in (None, NULL_REVISION):
1420
1294
                if branch.repository.supports_rich_root():
1421
1295
                    wt._set_root_id(generate_ids.gen_root_id())
1422
1296
                else:
1423
1297
                    wt._set_root_id(ROOT_ID)
1424
1298
                wt.flush()
1425
 
            basis = None
1426
 
            # frequently, we will get here due to branching.  The accelerator
1427
 
            # tree will be the tree from the branch, so the desired basis
1428
 
            # tree will often be a parent of the accelerator tree.
1429
 
            if accelerator_tree is not None:
1430
 
                try:
1431
 
                    basis = accelerator_tree.revision_tree(revision_id)
1432
 
                except errors.NoSuchRevision:
1433
 
                    pass
1434
 
            if basis is None:
1435
 
                basis = branch.repository.revision_tree(revision_id)
1436
 
            if revision_id == _mod_revision.NULL_REVISION:
1437
 
                parents_list = []
1438
 
            else:
1439
 
                parents_list = [(revision_id, basis)]
 
1299
            wt.set_last_revision(revision_id)
 
1300
            wt.flush()
 
1301
            basis = wt.basis_tree()
1440
1302
            basis.lock_read()
1441
 
            try:
1442
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1303
            # if the basis has a root id we have to use that; otherwise we use
 
1304
            # a new random one
 
1305
            basis_root_id = basis.get_root_id()
 
1306
            if basis_root_id is not None:
 
1307
                wt._set_root_id(basis_root_id)
1443
1308
                wt.flush()
1444
 
                # if the basis has a root id we have to use that; otherwise we
1445
 
                # use a new random one
1446
 
                basis_root_id = basis.get_root_id()
1447
 
                if basis_root_id is not None:
1448
 
                    wt._set_root_id(basis_root_id)
1449
 
                    wt.flush()
1450
 
                if wt.supports_content_filtering():
1451
 
                    # The original tree may not have the same content filters
1452
 
                    # applied so we can't safely build the inventory delta from
1453
 
                    # the source tree.
1454
 
                    delta_from_tree = False
1455
 
                else:
1456
 
                    delta_from_tree = True
1457
 
                # delta_from_tree is safe even for DirStateRevisionTrees,
1458
 
                # because wt4.apply_inventory_delta does not mutate the input
1459
 
                # inventory entries.
1460
 
                transform.build_tree(basis, wt, accelerator_tree,
1461
 
                                     hardlink=hardlink,
1462
 
                                     delta_from_tree=delta_from_tree)
1463
 
            finally:
1464
 
                basis.unlock()
 
1309
            transform.build_tree(basis, wt)
 
1310
            basis.unlock()
1465
1311
        finally:
1466
1312
            control_files.unlock()
1467
1313
            wt.unlock()
1468
1314
        return wt
1469
1315
 
1470
 
    def _init_custom_control_files(self, wt):
1471
 
        """Subclasses with custom control files should override this method.
1472
 
 
1473
 
        The working tree and control files are locked for writing when this
1474
 
        method is called.
1475
 
 
1476
 
        :param wt: the WorkingTree object
1477
 
        """
1478
 
 
1479
1316
    def _open(self, a_bzrdir, control_files):
1480
1317
        """Open the tree itself.
1481
1318
 
1482
1319
        :param a_bzrdir: the dir for the tree.
1483
1320
        :param control_files: the control files for the tree.
1484
1321
        """
1485
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1322
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1486
1323
                           branch=a_bzrdir.open_branch(),
1487
1324
                           _format=self,
1488
1325
                           _bzrdir=a_bzrdir,
1489
1326
                           _control_files=control_files)
1490
1327
 
1491
1328
    def __get_matchingbzrdir(self):
1492
 
        return self._get_matchingbzrdir()
1493
 
 
1494
 
    def _get_matchingbzrdir(self):
1495
 
        """Overrideable method to get a bzrdir for testing."""
1496
1329
        # please test against something that will let us do tree references
1497
1330
        return bzrdir.format_registry.make_bzrdir(
1498
1331
            'dirstate-with-subtree')
1500
1333
    _matchingbzrdir = property(__get_matchingbzrdir)
1501
1334
 
1502
1335
 
1503
 
class WorkingTreeFormat4(DirStateWorkingTreeFormat):
1504
 
    """The first consolidated dirstate working tree format.
1505
 
 
1506
 
    This format:
1507
 
        - exists within a metadir controlling .bzr
1508
 
        - includes an explicit version marker for the workingtree control
1509
 
          files, separate from the BzrDir format
1510
 
        - modifies the hash cache format
1511
 
        - is new in bzr 0.15
1512
 
        - uses a LockDir to guard access to it.
1513
 
    """
1514
 
 
1515
 
    upgrade_recommended = False
1516
 
 
1517
 
    _tree_class = WorkingTree4
1518
 
 
1519
 
    def get_format_string(self):
1520
 
        """See WorkingTreeFormat.get_format_string()."""
1521
 
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
1522
 
 
1523
 
    def get_format_description(self):
1524
 
        """See WorkingTreeFormat.get_format_description()."""
1525
 
        return "Working tree format 4"
1526
 
 
1527
 
 
1528
 
class WorkingTreeFormat5(DirStateWorkingTreeFormat):
1529
 
    """WorkingTree format supporting content filtering.
1530
 
    """
1531
 
 
1532
 
    upgrade_recommended = False
1533
 
 
1534
 
    _tree_class = WorkingTree5
1535
 
 
1536
 
    def get_format_string(self):
1537
 
        """See WorkingTreeFormat.get_format_string()."""
1538
 
        return "Bazaar Working Tree Format 5 (bzr 1.11)\n"
1539
 
 
1540
 
    def get_format_description(self):
1541
 
        """See WorkingTreeFormat.get_format_description()."""
1542
 
        return "Working tree format 5"
1543
 
 
1544
 
    def supports_content_filtering(self):
1545
 
        return True
1546
 
 
1547
 
 
1548
 
class WorkingTreeFormat6(DirStateWorkingTreeFormat):
1549
 
    """WorkingTree format supporting views.
1550
 
    """
1551
 
 
1552
 
    upgrade_recommended = False
1553
 
 
1554
 
    _tree_class = WorkingTree6
1555
 
 
1556
 
    def get_format_string(self):
1557
 
        """See WorkingTreeFormat.get_format_string()."""
1558
 
        return "Bazaar Working Tree Format 6 (bzr 1.14)\n"
1559
 
 
1560
 
    def get_format_description(self):
1561
 
        """See WorkingTreeFormat.get_format_description()."""
1562
 
        return "Working tree format 6"
1563
 
 
1564
 
    def _init_custom_control_files(self, wt):
1565
 
        """Subclasses with custom control files should override this method."""
1566
 
        wt._transport.put_bytes('views', '', mode=wt.bzrdir._get_file_mode())
1567
 
 
1568
 
    def supports_content_filtering(self):
1569
 
        return True
1570
 
 
1571
 
    def supports_views(self):
1572
 
        return True
1573
 
 
1574
 
 
1575
1336
class DirStateRevisionTree(Tree):
1576
 
    """A revision tree pulling the inventory from a dirstate.
1577
 
    
1578
 
    Note that this is one of the historical (ie revision) trees cached in the
1579
 
    dirstate for easy access, not the workingtree.
1580
 
    """
 
1337
    """A revision tree pulling the inventory from a dirstate."""
1581
1338
 
1582
1339
    def __init__(self, dirstate, revision_id, repository):
1583
1340
        self._dirstate = dirstate
1586
1343
        self._inventory = None
1587
1344
        self._locked = 0
1588
1345
        self._dirstate_locked = False
1589
 
        self._repo_supports_tree_reference = getattr(
1590
 
            repository._format, "supports_tree_reference",
1591
 
            False)
1592
1346
 
1593
1347
    def __repr__(self):
1594
1348
        return "<%s of %s in %s>" % \
1597
1351
    def annotate_iter(self, file_id,
1598
1352
                      default_revision=_mod_revision.CURRENT_REVISION):
1599
1353
        """See Tree.annotate_iter"""
1600
 
        text_key = (file_id, self.inventory[file_id].revision)
1601
 
        annotations = self._repository.texts.annotate(text_key)
1602
 
        return [(key[-1], line) for (key, line) in annotations]
 
1354
        w = self._get_weave(file_id)
 
1355
        return w.annotate_iter(self.inventory[file_id].revision)
1603
1356
 
1604
1357
    def _get_ancestors(self, default_revision):
1605
1358
        return set(self._repository.get_ancestry(self._revision_id,
1634
1387
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1635
1388
        return path_utf8.decode('utf8')
1636
1389
 
1637
 
    def iter_references(self):
1638
 
        if not self._repo_supports_tree_reference:
1639
 
            # When the repo doesn't support references, we will have nothing to
1640
 
            # return
1641
 
            return iter([])
1642
 
        # Otherwise, fall back to the default implementation
1643
 
        return super(DirStateRevisionTree, self).iter_references()
1644
 
 
1645
1390
    def _get_parent_index(self):
1646
1391
        """Return the index in the dirstate referenced by this tree."""
1647
1392
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1652
1397
        If either file_id or path is supplied, it is used as the key to lookup.
1653
1398
        If both are supplied, the fastest lookup is used, and an error is
1654
1399
        raised if they do not both point at the same row.
1655
 
 
 
1400
        
1656
1401
        :param file_id: An optional unicode file_id to be looked up.
1657
1402
        :param path: An optional unicode path to be looked up.
1658
1403
        :return: The dirstate row tuple for path/file_id, or (None, None)
1672
1417
 
1673
1418
        This is relatively expensive: we have to walk the entire dirstate.
1674
1419
        """
1675
 
        if not self._locked:
1676
 
            raise AssertionError(
1677
 
                'cannot generate inventory of an unlocked '
1678
 
                'dirstate revision tree')
 
1420
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1421
            'dirstate revision tree'
1679
1422
        # separate call for profiling - makes it clear where the costs are.
1680
1423
        self._dirstate._read_dirblocks_if_needed()
1681
 
        if self._revision_id not in self._dirstate.get_parent_ids():
1682
 
            raise AssertionError(
1683
 
                'parent %s has disappeared from %s' % (
1684
 
                self._revision_id, self._dirstate.get_parent_ids()))
 
1424
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1425
            'parent %s has disappeared from %s' % (
 
1426
            self._revision_id, self._dirstate.get_parent_ids())
1685
1427
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1686
1428
        # This is identical now to the WorkingTree _generate_inventory except
1687
1429
        # for the tree index use.
1688
1430
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1689
1431
        current_id = root_key[2]
1690
 
        if current_entry[parent_index][0] != 'd':
1691
 
            raise AssertionError()
 
1432
        assert current_entry[parent_index][0] == 'd'
1692
1433
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1693
1434
        inv.root.revision = current_entry[parent_index][4]
1694
1435
        # Turn some things into local variables
1734
1475
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1735
1476
                            % entry)
1736
1477
                # These checks cost us around 40ms on a 55k entry tree
1737
 
                if file_id in inv_byid:
1738
 
                    raise AssertionError('file_id %s already in'
1739
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
1740
 
                if name_unicode in parent_ie.children:
1741
 
                    raise AssertionError('name %r already in parent'
1742
 
                        % (name_unicode,))
 
1478
                assert file_id not in inv_byid
 
1479
                assert name_unicode not in parent_ie.children
1743
1480
                inv_byid[file_id] = inv_entry
1744
1481
                parent_ie.children[name_unicode] = inv_entry
1745
1482
        self._inventory = inv
1755
1492
            return None
1756
1493
        parent_index = self._get_parent_index()
1757
1494
        last_changed_revision = entry[1][parent_index][4]
1758
 
        try:
1759
 
            rev = self._repository.get_revision(last_changed_revision)
1760
 
        except errors.NoSuchRevision:
1761
 
            raise errors.FileTimestampUnavailable(self.id2path(file_id))
1762
 
        return rev.timestamp
 
1495
        return self._repository.get_revision(last_changed_revision).timestamp
1763
1496
 
1764
1497
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1765
1498
        entry = self._get_entry(file_id=file_id, path=path)
1769
1502
            return parent_details[1]
1770
1503
        return None
1771
1504
 
 
1505
    @symbol_versioning.deprecated_method(symbol_versioning.zero_ninety)
 
1506
    def get_weave(self, file_id):
 
1507
        return self._get_weave(file_id)
 
1508
 
 
1509
    def _get_weave(self, file_id):
 
1510
        return self._repository.weave_store.get_weave(file_id,
 
1511
                self._repository.get_transaction())
 
1512
 
1772
1513
    def get_file(self, file_id, path=None):
1773
1514
        return StringIO(self.get_file_text(file_id))
1774
1515
 
 
1516
    def get_file_lines(self, file_id):
 
1517
        entry = self._get_entry(file_id=file_id)[1]
 
1518
        if entry == None:
 
1519
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1520
        return self._get_weave(file_id).get_lines(entry[1][4])
 
1521
 
1775
1522
    def get_file_size(self, file_id):
1776
 
        """See Tree.get_file_size"""
1777
1523
        return self.inventory[file_id].text_size
1778
1524
 
1779
 
    def get_file_text(self, file_id, path=None):
1780
 
        _, content = list(self.iter_files_bytes([(file_id, None)]))[0]
1781
 
        return ''.join(content)
 
1525
    def get_file_text(self, file_id):
 
1526
        return ''.join(self.get_file_lines(file_id))
1782
1527
 
1783
1528
    def get_reference_revision(self, file_id, path=None):
1784
1529
        return self.inventory[file_id].reference_revision
1803
1548
        if entry[1][parent_index][0] != 'l':
1804
1549
            return None
1805
1550
        else:
1806
 
            target = entry[1][parent_index][1]
1807
 
            target = target.decode('utf8')
1808
 
            return target
 
1551
            # At present, none of the tree implementations supports non-ascii
 
1552
            # symlink targets. So we will just assume that the dirstate path is
 
1553
            # correct.
 
1554
            return entry[1][parent_index][1]
1809
1555
 
1810
1556
    def get_revision_id(self):
1811
1557
        """Return the revision id for this tree."""
1830
1576
 
1831
1577
    def kind(self, file_id):
1832
1578
        entry = self._get_entry(file_id=file_id)[1]
1833
 
        if entry is None:
 
1579
        if entry == None:
1834
1580
            raise errors.NoSuchId(tree=self, file_id=file_id)
1835
 
        parent_index = self._get_parent_index()
1836
 
        return dirstate.DirState._minikind_to_kind[entry[parent_index][0]]
1837
 
 
1838
 
    def stored_kind(self, file_id):
1839
 
        """See Tree.stored_kind"""
1840
 
        return self.kind(file_id)
 
1581
        return dirstate.DirState._minikind_to_kind[entry[1][0]]
1841
1582
 
1842
1583
    def path_content_summary(self, path):
1843
1584
        """See Tree.path_content_summary."""
1859
1600
            return None
1860
1601
        return ie.executable
1861
1602
 
1862
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1603
    def list_files(self, include_root=False):
1863
1604
        # We use a standard implementation, because DirStateRevisionTree is
1864
1605
        # dealing with one of the parents of the current state
1865
1606
        inv = self._get_inventory()
1866
 
        if from_dir is None:
1867
 
            from_dir_id = None
1868
 
        else:
1869
 
            from_dir_id = inv.path2id(from_dir)
1870
 
            if from_dir_id is None:
1871
 
                # Directory not versioned
1872
 
                return
1873
 
        entries = inv.iter_entries(from_dir=from_dir_id, recursive=recursive)
1874
 
        if inv.root is not None and not include_root and from_dir is None:
 
1607
        entries = inv.iter_entries()
 
1608
        if self.inventory.root is not None and not include_root:
1875
1609
            entries.next()
1876
1610
        for path, entry in entries:
1877
1611
            yield path, 'V', entry.kind, entry.file_id, entry
1910
1644
                self._dirstate_locked = False
1911
1645
            self._repository.unlock()
1912
1646
 
1913
 
    @needs_read_lock
1914
 
    def supports_tree_reference(self):
1915
 
        return self._repo_supports_tree_reference
1916
 
 
1917
1647
    def walkdirs(self, prefix=""):
1918
1648
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1919
 
        # implementation based on an inventory.
 
1649
        # implementation based on an inventory.  
1920
1650
        # This should be cleaned up to use the much faster Dirstate code
1921
1651
        # So for now, we just build up the parent inventory, and extract
1922
1652
        # it the same way RevisionTree does.
1951
1681
 
1952
1682
class InterDirStateTree(InterTree):
1953
1683
    """Fast path optimiser for changes_from with dirstate trees.
1954
 
 
1955
 
    This is used only when both trees are in the dirstate working file, and
1956
 
    the source is any parent within the dirstate, and the destination is
 
1684
    
 
1685
    This is used only when both trees are in the dirstate working file, and 
 
1686
    the source is any parent within the dirstate, and the destination is 
1957
1687
    the current working tree of the same dirstate.
1958
1688
    """
1959
1689
    # this could be generalized to allow comparisons between any trees in the
1972
1702
        target.set_parent_ids([revid])
1973
1703
        return target.basis_tree(), target
1974
1704
 
1975
 
    @classmethod
1976
 
    def make_source_parent_tree_python_dirstate(klass, test_case, source, target):
1977
 
        result = klass.make_source_parent_tree(source, target)
1978
 
        result[1]._iter_changes = dirstate.ProcessEntryPython
1979
 
        return result
1980
 
 
1981
 
    @classmethod
1982
 
    def make_source_parent_tree_compiled_dirstate(klass, test_case, source,
1983
 
                                                  target):
1984
 
        from bzrlib.tests.test__dirstate_helpers import \
1985
 
            compiled_dirstate_helpers_feature
1986
 
        test_case.requireFeature(compiled_dirstate_helpers_feature)
1987
 
        from bzrlib._dirstate_helpers_pyx import ProcessEntryC
1988
 
        result = klass.make_source_parent_tree(source, target)
1989
 
        result[1]._iter_changes = ProcessEntryC
1990
 
        return result
1991
 
 
1992
1705
    _matching_from_tree_format = WorkingTreeFormat4()
1993
1706
    _matching_to_tree_format = WorkingTreeFormat4()
1994
 
 
1995
 
    @classmethod
1996
 
    def _test_mutable_trees_to_test_trees(klass, test_case, source, target):
1997
 
        # This method shouldn't be called, because we have python and C
1998
 
        # specific flavours.
1999
 
        raise NotImplementedError
2000
 
 
2001
 
    def iter_changes(self, include_unchanged=False,
 
1707
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1708
 
 
1709
    def _iter_changes(self, include_unchanged=False,
2002
1710
                      specific_files=None, pb=None, extra_trees=[],
2003
1711
                      require_versioned=True, want_unversioned=False):
2004
1712
        """Return the changes from source to target.
2005
1713
 
2006
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1714
        :return: An iterator that yields tuples. See InterTree._iter_changes
2007
1715
            for details.
2008
1716
        :param specific_files: An optional list of file paths to restrict the
2009
1717
            comparison to. When mapping filenames to ids, all matches in all
2020
1728
            output. An unversioned file is defined as one with (False, False)
2021
1729
            for the versioned pair.
2022
1730
        """
 
1731
        utf8_decode = cache_utf8._utf8_decode
 
1732
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1733
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1734
        # NB: show_status depends on being able to pass in non-versioned files
 
1735
        # and report them as unknown
2023
1736
        # TODO: handle extra trees in the dirstate.
2024
 
        if (extra_trees or specific_files == []):
 
1737
        # TODO: handle comparisons as an empty tree as a different special
 
1738
        # case? mbp 20070226
 
1739
        if (extra_trees or (self.source._revision_id == NULL_REVISION)
 
1740
            or specific_files == []):
2025
1741
            # we can't fast-path these cases (yet)
2026
 
            return super(InterDirStateTree, self).iter_changes(
 
1742
            for f in super(InterDirStateTree, self)._iter_changes(
2027
1743
                include_unchanged, specific_files, pb, extra_trees,
2028
 
                require_versioned, want_unversioned=want_unversioned)
 
1744
                require_versioned, want_unversioned=want_unversioned):
 
1745
                yield f
 
1746
            return
2029
1747
        parent_ids = self.target.get_parent_ids()
2030
 
        if not (self.source._revision_id in parent_ids
2031
 
                or self.source._revision_id == _mod_revision.NULL_REVISION):
2032
 
            raise AssertionError(
2033
 
                "revision {%s} is not stored in {%s}, but %s "
2034
 
                "can only be used for trees stored in the dirstate"
2035
 
                % (self.source._revision_id, self.target, self.iter_changes))
 
1748
        assert (self.source._revision_id in parent_ids), \
 
1749
                "revision {%s} is not stored in {%s}, but %s " \
 
1750
                "can only be used for trees stored in the dirstate" \
 
1751
                % (self.source._revision_id, self.target, self._iter_changes)
2036
1752
        target_index = 0
2037
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1753
        if self.source._revision_id == NULL_REVISION:
2038
1754
            source_index = None
2039
1755
            indices = (target_index,)
2040
1756
        else:
2041
 
            if not (self.source._revision_id in parent_ids):
2042
 
                raise AssertionError(
2043
 
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
2044
 
                    self.source._revision_id, parent_ids))
 
1757
            assert (self.source._revision_id in parent_ids), \
 
1758
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1759
                self.source._revision_id, parent_ids)
2045
1760
            source_index = 1 + parent_ids.index(self.source._revision_id)
2046
 
            indices = (source_index, target_index)
 
1761
            indices = (source_index,target_index)
2047
1762
        # -- make all specific_files utf8 --
2048
1763
        if specific_files:
2049
1764
            specific_files_utf8 = set()
2050
1765
            for path in specific_files:
2051
 
                # Note, if there are many specific files, using cache_utf8
2052
 
                # would be good here.
2053
1766
                specific_files_utf8.add(path.encode('utf8'))
2054
1767
            specific_files = specific_files_utf8
2055
1768
        else:
2056
1769
            specific_files = set([''])
2057
1770
        # -- specific_files is now a utf8 path set --
2058
 
 
2059
1771
        # -- get the state object and prepare it.
2060
1772
        state = self.target.current_dirstate()
2061
1773
        state._read_dirblocks_if_needed()
 
1774
        def _entries_for_path(path):
 
1775
            """Return a list with all the entries that match path for all ids.
 
1776
            """
 
1777
            dirname, basename = os.path.split(path)
 
1778
            key = (dirname, basename, '')
 
1779
            block_index, present = state._find_block_index_from_key(key)
 
1780
            if not present:
 
1781
                # the block which should contain path is absent.
 
1782
                return []
 
1783
            result = []
 
1784
            block = state._dirblocks[block_index][1]
 
1785
            entry_index, _ = state._find_entry_index(key, block)
 
1786
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1787
            while (entry_index < len(block) and
 
1788
                block[entry_index][0][0:2] == key[0:2]):
 
1789
                result.append(block[entry_index])
 
1790
                entry_index += 1
 
1791
            return result
2062
1792
        if require_versioned:
2063
1793
            # -- check all supplied paths are versioned in a search tree. --
2064
 
            not_versioned = []
 
1794
            all_versioned = True
2065
1795
            for path in specific_files:
2066
 
                path_entries = state._entries_for_path(path)
 
1796
                path_entries = _entries_for_path(path)
2067
1797
                if not path_entries:
2068
1798
                    # this specified path is not present at all: error
2069
 
                    not_versioned.append(path)
2070
 
                    continue
 
1799
                    all_versioned = False
 
1800
                    break
2071
1801
                found_versioned = False
2072
1802
                # for each id at this path
2073
1803
                for entry in path_entries:
2080
1810
                if not found_versioned:
2081
1811
                    # none of the indexes was not 'absent' at all ids for this
2082
1812
                    # path.
2083
 
                    not_versioned.append(path)
2084
 
            if len(not_versioned) > 0:
2085
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1813
                    all_versioned = False
 
1814
                    break
 
1815
            if not all_versioned:
 
1816
                raise errors.PathsNotVersionedError(specific_files)
2086
1817
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2087
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1818
        search_specific_files = set()
 
1819
        for path in specific_files:
 
1820
            other_specific_files = specific_files.difference(set([path]))
 
1821
            if not osutils.is_inside_any(other_specific_files, path):
 
1822
                # this is a top level path, we must check it.
 
1823
                search_specific_files.add(path)
 
1824
        # sketch: 
 
1825
        # compare source_index and target_index at or under each element of search_specific_files.
 
1826
        # follow the following comparison table. Note that we only want to do diff operations when
 
1827
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1828
        # for the target.
 
1829
        # cases:
 
1830
        # 
 
1831
        # Source | Target | disk | action
 
1832
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1833
        #        |        |      | diff check on source-target
 
1834
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1835
        #        |        |      | ???
 
1836
        #   r    |  a     |      | add source to search
 
1837
        #   r    |  a     |  a   | 
 
1838
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1839
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1840
        #   a    | fdlt   |      | add new id
 
1841
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1842
        #   a    |  a     |      | not present in either tree, skip
 
1843
        #   a    |  a     |  a   | not present in any tree, skip
 
1844
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1845
        #        |        |      | may not be selected by the users list of paths.
 
1846
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1847
        #        |        |      | may not be selected by the users list of paths.
 
1848
        #  fdlt  | fdlt   |      | content in both: diff them
 
1849
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1850
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1851
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1852
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1853
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1854
        #        |        |      | this id at the other path.
 
1855
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1856
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1857
        #        |        |      | this id at the other path.
 
1858
 
 
1859
        # for all search_indexs in each path at or under each element of
 
1860
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1861
        # relocated path as one to search if its not searched already. If the
 
1862
        # detail is not relocated, add the id.
 
1863
        searched_specific_files = set()
 
1864
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1865
        # Using a list so that we can access the values and change them in
 
1866
        # nested scope. Each one is [path, file_id, entry]
 
1867
        last_source_parent = [None, None]
 
1868
        last_target_parent = [None, None]
2088
1869
 
2089
1870
        use_filesystem_for_exec = (sys.platform != 'win32')
2090
 
        iter_changes = self.target._iter_changes(include_unchanged,
2091
 
            use_filesystem_for_exec, search_specific_files, state,
2092
 
            source_index, target_index, want_unversioned, self.target)
2093
 
        return iter_changes.iter_changes()
 
1871
 
 
1872
        # Just a sentry, so that _process_entry can say that this
 
1873
        # record is handled, but isn't interesting to process (unchanged)
 
1874
        uninteresting = object()
 
1875
 
 
1876
 
 
1877
        old_dirname_to_file_id = {}
 
1878
        new_dirname_to_file_id = {}
 
1879
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1880
        #       keeping a cache of directories that we have seen.
 
1881
 
 
1882
        def _process_entry(entry, path_info):
 
1883
            """Compare an entry and real disk to generate delta information.
 
1884
 
 
1885
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1886
                the path of entry. If None, then the path is considered absent.
 
1887
                (Perhaps we should pass in a concrete entry for this ?)
 
1888
                Basename is returned as a utf8 string because we expect this
 
1889
                tuple will be ignored, and don't want to take the time to
 
1890
                decode.
 
1891
            :return: None if these don't match
 
1892
                     A tuple of information about the change, or
 
1893
                     the object 'uninteresting' if these match, but are
 
1894
                     basically identical.
 
1895
            """
 
1896
            if source_index is None:
 
1897
                source_details = NULL_PARENT_DETAILS
 
1898
            else:
 
1899
                source_details = entry[1][source_index]
 
1900
            target_details = entry[1][target_index]
 
1901
            target_minikind = target_details[0]
 
1902
            if path_info is not None and target_minikind in 'fdlt':
 
1903
                assert target_index == 0
 
1904
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1905
                                                  stat_value=path_info[3])
 
1906
                # The entry may have been modified by update_entry
 
1907
                target_details = entry[1][target_index]
 
1908
                target_minikind = target_details[0]
 
1909
            else:
 
1910
                link_or_sha1 = None
 
1911
            file_id = entry[0][2]
 
1912
            source_minikind = source_details[0]
 
1913
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1914
                # claimed content in both: diff
 
1915
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1916
                #        |        |      | diff check on source-target
 
1917
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1918
                #        |        |      | ???
 
1919
                if source_minikind in 'r':
 
1920
                    # add the source to the search path to find any children it
 
1921
                    # has.  TODO ? : only add if it is a container ?
 
1922
                    if not osutils.is_inside_any(searched_specific_files,
 
1923
                                                 source_details[1]):
 
1924
                        search_specific_files.add(source_details[1])
 
1925
                    # generate the old path; this is needed for stating later
 
1926
                    # as well.
 
1927
                    old_path = source_details[1]
 
1928
                    old_dirname, old_basename = os.path.split(old_path)
 
1929
                    path = pathjoin(entry[0][0], entry[0][1])
 
1930
                    old_entry = state._get_entry(source_index,
 
1931
                                                 path_utf8=old_path)
 
1932
                    # update the source details variable to be the real
 
1933
                    # location.
 
1934
                    source_details = old_entry[1][source_index]
 
1935
                    source_minikind = source_details[0]
 
1936
                else:
 
1937
                    old_dirname = entry[0][0]
 
1938
                    old_basename = entry[0][1]
 
1939
                    old_path = path = None
 
1940
                if path_info is None:
 
1941
                    # the file is missing on disk, show as removed.
 
1942
                    content_change = True
 
1943
                    target_kind = None
 
1944
                    target_exec = False
 
1945
                else:
 
1946
                    # source and target are both versioned and disk file is present.
 
1947
                    target_kind = path_info[2]
 
1948
                    if target_kind == 'directory':
 
1949
                        if path is None:
 
1950
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1951
                        new_dirname_to_file_id[path] = file_id
 
1952
                        if source_minikind != 'd':
 
1953
                            content_change = True
 
1954
                        else:
 
1955
                            # directories have no fingerprint
 
1956
                            content_change = False
 
1957
                        target_exec = False
 
1958
                    elif target_kind == 'file':
 
1959
                        if source_minikind != 'f':
 
1960
                            content_change = True
 
1961
                        else:
 
1962
                            # We could check the size, but we already have the
 
1963
                            # sha1 hash.
 
1964
                            content_change = (link_or_sha1 != source_details[1])
 
1965
                        # Target details is updated at update_entry time
 
1966
                        if use_filesystem_for_exec:
 
1967
                            # We don't need S_ISREG here, because we are sure
 
1968
                            # we are dealing with a file.
 
1969
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1970
                        else:
 
1971
                            target_exec = target_details[3]
 
1972
                    elif target_kind == 'symlink':
 
1973
                        if source_minikind != 'l':
 
1974
                            content_change = True
 
1975
                        else:
 
1976
                            content_change = (link_or_sha1 != source_details[1])
 
1977
                        target_exec = False
 
1978
                    elif target_kind == 'tree-reference':
 
1979
                        if source_minikind != 't':
 
1980
                            content_change = True
 
1981
                        else:
 
1982
                            content_change = False
 
1983
                        target_exec = False
 
1984
                    else:
 
1985
                        raise Exception, "unknown kind %s" % path_info[2]
 
1986
                if source_minikind == 'd':
 
1987
                    if path is None:
 
1988
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1989
                    old_dirname_to_file_id[old_path] = file_id
 
1990
                # parent id is the entry for the path in the target tree
 
1991
                if old_dirname == last_source_parent[0]:
 
1992
                    source_parent_id = last_source_parent[1]
 
1993
                else:
 
1994
                    try:
 
1995
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1996
                    except KeyError:
 
1997
                        source_parent_entry = state._get_entry(source_index,
 
1998
                                                               path_utf8=old_dirname)
 
1999
                        source_parent_id = source_parent_entry[0][2]
 
2000
                    if source_parent_id == entry[0][2]:
 
2001
                        # This is the root, so the parent is None
 
2002
                        source_parent_id = None
 
2003
                    else:
 
2004
                        last_source_parent[0] = old_dirname
 
2005
                        last_source_parent[1] = source_parent_id
 
2006
                new_dirname = entry[0][0]
 
2007
                if new_dirname == last_target_parent[0]:
 
2008
                    target_parent_id = last_target_parent[1]
 
2009
                else:
 
2010
                    try:
 
2011
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
2012
                    except KeyError:
 
2013
                        # TODO: We don't always need to do the lookup, because the
 
2014
                        #       parent entry will be the same as the source entry.
 
2015
                        target_parent_entry = state._get_entry(target_index,
 
2016
                                                               path_utf8=new_dirname)
 
2017
                        assert target_parent_entry != (None, None), (
 
2018
                            "Could not find target parent in wt: %s\nparent of: %s"
 
2019
                            % (new_dirname, entry))
 
2020
                        target_parent_id = target_parent_entry[0][2]
 
2021
                    if target_parent_id == entry[0][2]:
 
2022
                        # This is the root, so the parent is None
 
2023
                        target_parent_id = None
 
2024
                    else:
 
2025
                        last_target_parent[0] = new_dirname
 
2026
                        last_target_parent[1] = target_parent_id
 
2027
 
 
2028
                source_exec = source_details[3]
 
2029
                if (include_unchanged
 
2030
                    or content_change
 
2031
                    or source_parent_id != target_parent_id
 
2032
                    or old_basename != entry[0][1]
 
2033
                    or source_exec != target_exec
 
2034
                    ):
 
2035
                    if old_path is None:
 
2036
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2037
                        old_path_u = utf8_decode(old_path)[0]
 
2038
                        path_u = old_path_u
 
2039
                    else:
 
2040
                        old_path_u = utf8_decode(old_path)[0]
 
2041
                        if old_path == path:
 
2042
                            path_u = old_path_u
 
2043
                        else:
 
2044
                            path_u = utf8_decode(path)[0]
 
2045
                    source_kind = _minikind_to_kind[source_minikind]
 
2046
                    return (entry[0][2],
 
2047
                           (old_path_u, path_u),
 
2048
                           content_change,
 
2049
                           (True, True),
 
2050
                           (source_parent_id, target_parent_id),
 
2051
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2052
                           (source_kind, target_kind),
 
2053
                           (source_exec, target_exec))
 
2054
                else:
 
2055
                    return uninteresting
 
2056
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2057
                # looks like a new file
 
2058
                if path_info is not None:
 
2059
                    path = pathjoin(entry[0][0], entry[0][1])
 
2060
                    # parent id is the entry for the path in the target tree
 
2061
                    # TODO: these are the same for an entire directory: cache em.
 
2062
                    parent_id = state._get_entry(target_index,
 
2063
                                                 path_utf8=entry[0][0])[0][2]
 
2064
                    if parent_id == entry[0][2]:
 
2065
                        parent_id = None
 
2066
                    if use_filesystem_for_exec:
 
2067
                        # We need S_ISREG here, because we aren't sure if this
 
2068
                        # is a file or not.
 
2069
                        target_exec = bool(
 
2070
                            stat.S_ISREG(path_info[3].st_mode)
 
2071
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2072
                    else:
 
2073
                        target_exec = target_details[3]
 
2074
                    return (entry[0][2],
 
2075
                           (None, utf8_decode(path)[0]),
 
2076
                           True,
 
2077
                           (False, True),
 
2078
                           (None, parent_id),
 
2079
                           (None, utf8_decode(entry[0][1])[0]),
 
2080
                           (None, path_info[2]),
 
2081
                           (None, target_exec))
 
2082
                else:
 
2083
                    # but its not on disk: we deliberately treat this as just
 
2084
                    # never-present. (Why ?! - RBC 20070224)
 
2085
                    pass
 
2086
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2087
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2088
                # if its still on disk, *and* theres no other entry at this
 
2089
                # path [we dont know this in this routine at the moment -
 
2090
                # perhaps we should change this - then it would be an unknown.
 
2091
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2092
                # parent id is the entry for the path in the target tree
 
2093
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2094
                if parent_id == entry[0][2]:
 
2095
                    parent_id = None
 
2096
                return (entry[0][2],
 
2097
                       (utf8_decode(old_path)[0], None),
 
2098
                       True,
 
2099
                       (True, False),
 
2100
                       (parent_id, None),
 
2101
                       (utf8_decode(entry[0][1])[0], None),
 
2102
                       (_minikind_to_kind[source_minikind], None),
 
2103
                       (source_details[3], None))
 
2104
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2105
                # a rename; could be a true rename, or a rename inherited from
 
2106
                # a renamed parent. TODO: handle this efficiently. Its not
 
2107
                # common case to rename dirs though, so a correct but slow
 
2108
                # implementation will do.
 
2109
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2110
                    search_specific_files.add(target_details[1])
 
2111
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2112
                # neither of the selected trees contain this file,
 
2113
                # so skip over it. This is not currently directly tested, but
 
2114
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2115
                pass
 
2116
            else:
 
2117
                raise AssertionError("don't know how to compare "
 
2118
                    "source_minikind=%r, target_minikind=%r"
 
2119
                    % (source_minikind, target_minikind))
 
2120
                ## import pdb;pdb.set_trace()
 
2121
            return None
 
2122
 
 
2123
        while search_specific_files:
 
2124
            # TODO: the pending list should be lexically sorted?  the
 
2125
            # interface doesn't require it.
 
2126
            current_root = search_specific_files.pop()
 
2127
            current_root_unicode = current_root.decode('utf8')
 
2128
            searched_specific_files.add(current_root)
 
2129
            # process the entries for this containing directory: the rest will be
 
2130
            # found by their parents recursively.
 
2131
            root_entries = _entries_for_path(current_root)
 
2132
            root_abspath = self.target.abspath(current_root_unicode)
 
2133
            try:
 
2134
                root_stat = os.lstat(root_abspath)
 
2135
            except OSError, e:
 
2136
                if e.errno == errno.ENOENT:
 
2137
                    # the path does not exist: let _process_entry know that.
 
2138
                    root_dir_info = None
 
2139
                else:
 
2140
                    # some other random error: hand it up.
 
2141
                    raise
 
2142
            else:
 
2143
                root_dir_info = ('', current_root,
 
2144
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2145
                    root_abspath)
 
2146
                if root_dir_info[2] == 'directory':
 
2147
                    if self.target._directory_is_tree_reference(
 
2148
                        current_root.decode('utf8')):
 
2149
                        root_dir_info = root_dir_info[:2] + \
 
2150
                            ('tree-reference',) + root_dir_info[3:]
 
2151
 
 
2152
            if not root_entries and not root_dir_info:
 
2153
                # this specified path is not present at all, skip it.
 
2154
                continue
 
2155
            path_handled = False
 
2156
            for entry in root_entries:
 
2157
                result = _process_entry(entry, root_dir_info)
 
2158
                if result is not None:
 
2159
                    path_handled = True
 
2160
                    if result is not uninteresting:
 
2161
                        yield result
 
2162
            if want_unversioned and not path_handled and root_dir_info:
 
2163
                new_executable = bool(
 
2164
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2165
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2166
                yield (None,
 
2167
                       (None, current_root_unicode),
 
2168
                       True,
 
2169
                       (False, False),
 
2170
                       (None, None),
 
2171
                       (None, splitpath(current_root_unicode)[-1]),
 
2172
                       (None, root_dir_info[2]),
 
2173
                       (None, new_executable)
 
2174
                      )
 
2175
            initial_key = (current_root, '', '')
 
2176
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2177
            if block_index == 0:
 
2178
                # we have processed the total root already, but because the
 
2179
                # initial key matched it we should skip it here.
 
2180
                block_index +=1
 
2181
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2182
                current_dir_info = None
 
2183
            else:
 
2184
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2185
                try:
 
2186
                    current_dir_info = dir_iterator.next()
 
2187
                except OSError, e:
 
2188
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2189
                    # python 2.5 has e.errno == EINVAL,
 
2190
                    #            and e.winerror == ERROR_DIRECTORY
 
2191
                    e_winerror = getattr(e, 'winerror', None)
 
2192
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2193
                    # there may be directories in the inventory even though
 
2194
                    # this path is not a file on disk: so mark it as end of
 
2195
                    # iterator
 
2196
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2197
                        current_dir_info = None
 
2198
                    elif (sys.platform == 'win32'
 
2199
                          and (e.errno in win_errors
 
2200
                               or e_winerror in win_errors)):
 
2201
                        current_dir_info = None
 
2202
                    else:
 
2203
                        raise
 
2204
                else:
 
2205
                    if current_dir_info[0][0] == '':
 
2206
                        # remove .bzr from iteration
 
2207
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2208
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2209
                        del current_dir_info[1][bzr_index]
 
2210
            # walk until both the directory listing and the versioned metadata
 
2211
            # are exhausted. 
 
2212
            if (block_index < len(state._dirblocks) and
 
2213
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2214
                current_block = state._dirblocks[block_index]
 
2215
            else:
 
2216
                current_block = None
 
2217
            while (current_dir_info is not None or
 
2218
                   current_block is not None):
 
2219
                if (current_dir_info and current_block
 
2220
                    and current_dir_info[0][0] != current_block[0]):
 
2221
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2222
                        # filesystem data refers to paths not covered by the dirblock.
 
2223
                        # this has two possibilities:
 
2224
                        # A) it is versioned but empty, so there is no block for it
 
2225
                        # B) it is not versioned.
 
2226
 
 
2227
                        # if (A) then we need to recurse into it to check for
 
2228
                        # new unknown files or directories.
 
2229
                        # if (B) then we should ignore it, because we don't
 
2230
                        # recurse into unknown directories.
 
2231
                        path_index = 0
 
2232
                        while path_index < len(current_dir_info[1]):
 
2233
                                current_path_info = current_dir_info[1][path_index]
 
2234
                                if want_unversioned:
 
2235
                                    if current_path_info[2] == 'directory':
 
2236
                                        if self.target._directory_is_tree_reference(
 
2237
                                            current_path_info[0].decode('utf8')):
 
2238
                                            current_path_info = current_path_info[:2] + \
 
2239
                                                ('tree-reference',) + current_path_info[3:]
 
2240
                                    new_executable = bool(
 
2241
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2242
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2243
                                    yield (None,
 
2244
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2245
                                        True,
 
2246
                                        (False, False),
 
2247
                                        (None, None),
 
2248
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2249
                                        (None, current_path_info[2]),
 
2250
                                        (None, new_executable))
 
2251
                                # dont descend into this unversioned path if it is
 
2252
                                # a dir
 
2253
                                if current_path_info[2] in ('directory',
 
2254
                                                            'tree-reference'):
 
2255
                                    del current_dir_info[1][path_index]
 
2256
                                    path_index -= 1
 
2257
                                path_index += 1
 
2258
 
 
2259
                        # This dir info has been handled, go to the next
 
2260
                        try:
 
2261
                            current_dir_info = dir_iterator.next()
 
2262
                        except StopIteration:
 
2263
                            current_dir_info = None
 
2264
                    else:
 
2265
                        # We have a dirblock entry for this location, but there
 
2266
                        # is no filesystem path for this. This is most likely
 
2267
                        # because a directory was removed from the disk.
 
2268
                        # We don't have to report the missing directory,
 
2269
                        # because that should have already been handled, but we
 
2270
                        # need to handle all of the files that are contained
 
2271
                        # within.
 
2272
                        for current_entry in current_block[1]:
 
2273
                            # entry referring to file not present on disk.
 
2274
                            # advance the entry only, after processing.
 
2275
                            result = _process_entry(current_entry, None)
 
2276
                            if result is not None:
 
2277
                                if result is not uninteresting:
 
2278
                                    yield result
 
2279
                        block_index +=1
 
2280
                        if (block_index < len(state._dirblocks) and
 
2281
                            osutils.is_inside(current_root,
 
2282
                                              state._dirblocks[block_index][0])):
 
2283
                            current_block = state._dirblocks[block_index]
 
2284
                        else:
 
2285
                            current_block = None
 
2286
                    continue
 
2287
                entry_index = 0
 
2288
                if current_block and entry_index < len(current_block[1]):
 
2289
                    current_entry = current_block[1][entry_index]
 
2290
                else:
 
2291
                    current_entry = None
 
2292
                advance_entry = True
 
2293
                path_index = 0
 
2294
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2295
                    current_path_info = current_dir_info[1][path_index]
 
2296
                    if current_path_info[2] == 'directory':
 
2297
                        if self.target._directory_is_tree_reference(
 
2298
                            current_path_info[0].decode('utf8')):
 
2299
                            current_path_info = current_path_info[:2] + \
 
2300
                                ('tree-reference',) + current_path_info[3:]
 
2301
                else:
 
2302
                    current_path_info = None
 
2303
                advance_path = True
 
2304
                path_handled = False
 
2305
                while (current_entry is not None or
 
2306
                    current_path_info is not None):
 
2307
                    if current_entry is None:
 
2308
                        # the check for path_handled when the path is adnvaced
 
2309
                        # will yield this path if needed.
 
2310
                        pass
 
2311
                    elif current_path_info is None:
 
2312
                        # no path is fine: the per entry code will handle it.
 
2313
                        result = _process_entry(current_entry, current_path_info)
 
2314
                        if result is not None:
 
2315
                            if result is not uninteresting:
 
2316
                                yield result
 
2317
                    elif (current_entry[0][1] != current_path_info[1]
 
2318
                          or current_entry[1][target_index][0] in 'ar'):
 
2319
                        # The current path on disk doesn't match the dirblock
 
2320
                        # record. Either the dirblock is marked as absent, or
 
2321
                        # the file on disk is not present at all in the
 
2322
                        # dirblock. Either way, report about the dirblock
 
2323
                        # entry, and let other code handle the filesystem one.
 
2324
 
 
2325
                        # Compare the basename for these files to determine
 
2326
                        # which comes first
 
2327
                        if current_path_info[1] < current_entry[0][1]:
 
2328
                            # extra file on disk: pass for now, but only
 
2329
                            # increment the path, not the entry
 
2330
                            advance_entry = False
 
2331
                        else:
 
2332
                            # entry referring to file not present on disk.
 
2333
                            # advance the entry only, after processing.
 
2334
                            result = _process_entry(current_entry, None)
 
2335
                            if result is not None:
 
2336
                                if result is not uninteresting:
 
2337
                                    yield result
 
2338
                            advance_path = False
 
2339
                    else:
 
2340
                        result = _process_entry(current_entry, current_path_info)
 
2341
                        if result is not None:
 
2342
                            path_handled = True
 
2343
                            if result is not uninteresting:
 
2344
                                yield result
 
2345
                    if advance_entry and current_entry is not None:
 
2346
                        entry_index += 1
 
2347
                        if entry_index < len(current_block[1]):
 
2348
                            current_entry = current_block[1][entry_index]
 
2349
                        else:
 
2350
                            current_entry = None
 
2351
                    else:
 
2352
                        advance_entry = True # reset the advance flaga
 
2353
                    if advance_path and current_path_info is not None:
 
2354
                        if not path_handled:
 
2355
                            # unversioned in all regards
 
2356
                            if want_unversioned:
 
2357
                                new_executable = bool(
 
2358
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2359
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2360
                                yield (None,
 
2361
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2362
                                    True,
 
2363
                                    (False, False),
 
2364
                                    (None, None),
 
2365
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2366
                                    (None, current_path_info[2]),
 
2367
                                    (None, new_executable))
 
2368
                            # dont descend into this unversioned path if it is
 
2369
                            # a dir
 
2370
                            if current_path_info[2] in ('directory'):
 
2371
                                del current_dir_info[1][path_index]
 
2372
                                path_index -= 1
 
2373
                        # dont descend the disk iterator into any tree 
 
2374
                        # paths.
 
2375
                        if current_path_info[2] == 'tree-reference':
 
2376
                            del current_dir_info[1][path_index]
 
2377
                            path_index -= 1
 
2378
                        path_index += 1
 
2379
                        if path_index < len(current_dir_info[1]):
 
2380
                            current_path_info = current_dir_info[1][path_index]
 
2381
                            if current_path_info[2] == 'directory':
 
2382
                                if self.target._directory_is_tree_reference(
 
2383
                                    current_path_info[0].decode('utf8')):
 
2384
                                    current_path_info = current_path_info[:2] + \
 
2385
                                        ('tree-reference',) + current_path_info[3:]
 
2386
                        else:
 
2387
                            current_path_info = None
 
2388
                        path_handled = False
 
2389
                    else:
 
2390
                        advance_path = True # reset the advance flagg.
 
2391
                if current_block is not None:
 
2392
                    block_index += 1
 
2393
                    if (block_index < len(state._dirblocks) and
 
2394
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2395
                        current_block = state._dirblocks[block_index]
 
2396
                    else:
 
2397
                        current_block = None
 
2398
                if current_dir_info is not None:
 
2399
                    try:
 
2400
                        current_dir_info = dir_iterator.next()
 
2401
                    except StopIteration:
 
2402
                        current_dir_info = None
 
2403
 
2094
2404
 
2095
2405
    @staticmethod
2096
2406
    def is_compatible(source, target):
2097
2407
        # the target must be a dirstate working tree
2098
 
        if not isinstance(target, DirStateWorkingTree):
 
2408
        if not isinstance(target, WorkingTree4):
2099
2409
            return False
2100
 
        # the source must be a revtree or dirstate rev tree.
 
2410
        # the source must be a revtreee or dirstate rev tree.
2101
2411
        if not isinstance(source,
2102
2412
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2103
2413
            return False
2104
2414
        # the source revid must be in the target dirstate
2105
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2415
        if not (source._revision_id == NULL_REVISION or
2106
2416
            source._revision_id in target.get_parent_ids()):
2107
 
            # TODO: what about ghosts? it may well need to
 
2417
            # TODO: what about ghosts? it may well need to 
2108
2418
            # check for them explicitly.
2109
2419
            return False
2110
2420
        return True
2120
2430
 
2121
2431
    def convert(self, tree):
2122
2432
        # lock the control files not the tree, so that we dont get tree
2123
 
        # on-unlock behaviours, and so that noone else diddles with the
 
2433
        # on-unlock behaviours, and so that noone else diddles with the 
2124
2434
        # tree during upgrade.
2125
2435
        tree._control_files.lock_write()
2126
2436
        try:
2152
2462
 
2153
2463
    def update_format(self, tree):
2154
2464
        """Change the format marker."""
2155
 
        tree._transport.put_bytes('format',
2156
 
            self.target_format.get_format_string(),
2157
 
            mode=tree.bzrdir._get_file_mode())
2158
 
 
2159
 
 
2160
 
class Converter4to5(object):
2161
 
    """Perform an in-place upgrade of format 4 to format 5 trees."""
2162
 
 
2163
 
    def __init__(self):
2164
 
        self.target_format = WorkingTreeFormat5()
2165
 
 
2166
 
    def convert(self, tree):
2167
 
        # lock the control files not the tree, so that we don't get tree
2168
 
        # on-unlock behaviours, and so that no-one else diddles with the
2169
 
        # tree during upgrade.
2170
 
        tree._control_files.lock_write()
2171
 
        try:
2172
 
            self.update_format(tree)
2173
 
        finally:
2174
 
            tree._control_files.unlock()
2175
 
 
2176
 
    def update_format(self, tree):
2177
 
        """Change the format marker."""
2178
 
        tree._transport.put_bytes('format',
2179
 
            self.target_format.get_format_string(),
2180
 
            mode=tree.bzrdir._get_file_mode())
2181
 
 
2182
 
 
2183
 
class Converter4or5to6(object):
2184
 
    """Perform an in-place upgrade of format 4 or 5 to format 6 trees."""
2185
 
 
2186
 
    def __init__(self):
2187
 
        self.target_format = WorkingTreeFormat6()
2188
 
 
2189
 
    def convert(self, tree):
2190
 
        # lock the control files not the tree, so that we don't get tree
2191
 
        # on-unlock behaviours, and so that no-one else diddles with the
2192
 
        # tree during upgrade.
2193
 
        tree._control_files.lock_write()
2194
 
        try:
2195
 
            self.init_custom_control_files(tree)
2196
 
            self.update_format(tree)
2197
 
        finally:
2198
 
            tree._control_files.unlock()
2199
 
 
2200
 
    def init_custom_control_files(self, tree):
2201
 
        """Initialize custom control files."""
2202
 
        tree._transport.put_bytes('views', '',
2203
 
            mode=tree.bzrdir._get_file_mode())
2204
 
 
2205
 
    def update_format(self, tree):
2206
 
        """Change the format marker."""
2207
 
        tree._transport.put_bytes('format',
2208
 
            self.target_format.get_format_string(),
2209
 
            mode=tree.bzrdir._get_file_mode())
 
2465
        tree._control_files.put_utf8('format',
 
2466
            self.target_format.get_format_string())