~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Ian Clatworthy
  • Date: 2007-12-11 02:07:30 UTC
  • mto: (3119.1.1 ianc-integration)
  • mto: This revision was merged to the branch mainline in revision 3120.
  • Revision ID: ian.clatworthy@internode.on.net-20071211020730-sdj4kj794dw0628e
make help topics more discoverable

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
154
        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
155
 
108
156
    @needs_tree_write_lock
109
157
    def _add(self, files, ids, kinds):
137
185
    @needs_tree_write_lock
138
186
    def add_reference(self, sub_tree):
139
187
        # use standard implementation, which calls back to self._add
140
 
        #
 
188
        # 
141
189
        # So we don't store the reference_revision in the working dirstate,
142
 
        # it's just recorded at the moment of commit.
 
190
        # it's just recorded at the moment of commit. 
143
191
        self._add_reference(sub_tree)
144
192
 
145
193
    def break_lock(self):
184
232
            WorkingTree3._comparison_data(self, entry, path)
185
233
        # it looks like a plain directory, but it's really a reference -- see
186
234
        # also kind()
187
 
        if (self._repo_supports_tree_reference and kind == 'directory'
188
 
            and entry is not None and entry.kind == 'tree-reference'):
 
235
        if (self._repo_supports_tree_reference and
 
236
            kind == 'directory' and
 
237
            self._directory_is_tree_reference(path)):
189
238
            kind = 'tree-reference'
190
239
        return kind, executable, stat_value
191
240
 
217
266
            return self._dirstate
218
267
        local_path = self.bzrdir.get_workingtree_transport(None
219
268
            ).local_abspath('dirstate')
220
 
        self._dirstate = dirstate.DirState.on_file(local_path,
221
 
            self._sha1_provider())
 
269
        self._dirstate = dirstate.DirState.on_file(local_path)
222
270
        return self._dirstate
223
271
 
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
272
    def filter_unversioned_files(self, paths):
237
273
        """Filter out paths that are versioned.
238
274
 
270
306
 
271
307
    def _generate_inventory(self):
272
308
        """Create and set self.inventory from the dirstate object.
273
 
 
 
309
        
274
310
        This is relatively expensive: we have to walk the entire dirstate.
275
311
        Ideally we would not, and can deprecate this function.
276
312
        """
280
316
        state._read_dirblocks_if_needed()
281
317
        root_key, current_entry = self._get_entry(path='')
282
318
        current_id = root_key[2]
283
 
        if not (current_entry[0][0] == 'd'): # directory
284
 
            raise AssertionError(current_entry)
 
319
        assert current_entry[0][0] == 'd' # directory
285
320
        inv = Inventory(root_id=current_id)
286
321
        # Turn some things into local variables
287
322
        minikind_to_kind = dirstate.DirState._minikind_to_kind
320
355
                    # add this entry to the parent map.
321
356
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
322
357
                elif kind == 'tree-reference':
323
 
                    if not self._repo_supports_tree_reference:
324
 
                        raise errors.UnsupportedOperation(
325
 
                            self._generate_inventory,
326
 
                            self.branch.repository)
 
358
                    assert self._repo_supports_tree_reference, \
 
359
                        "repository of %r " \
 
360
                        "doesn't support tree references " \
 
361
                        "required by entry %r" \
 
362
                        % (self, name)
327
363
                    inv_entry.reference_revision = link_or_sha1 or None
328
364
                elif kind != 'symlink':
329
365
                    raise AssertionError("unknown kind %r" % kind)
330
366
                # 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,))
 
367
                assert file_id not in inv_byid, ('file_id %s already in'
 
368
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
369
                assert name_unicode not in parent_ie.children
337
370
                inv_byid[file_id] = inv_entry
338
371
                parent_ie.children[name_unicode] = inv_entry
339
372
        self._inventory = inv
344
377
        If either file_id or path is supplied, it is used as the key to lookup.
345
378
        If both are supplied, the fastest lookup is used, and an error is
346
379
        raised if they do not both point at the same row.
347
 
 
 
380
        
348
381
        :param file_id: An optional unicode file_id to be looked up.
349
382
        :param path: An optional unicode path to be looked up.
350
383
        :return: The dirstate row tuple for path/file_id, or (None, None)
374
407
                    return None
375
408
                else:
376
409
                    raise
377
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
378
 
            stat_value=stat_value)
 
410
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
411
                                          stat_value=stat_value)
379
412
        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
 
413
            return link_or_sha1
390
414
        return None
391
415
 
392
416
    def _get_inventory(self):
406
430
    @needs_read_lock
407
431
    def get_parent_ids(self):
408
432
        """See Tree.get_parent_ids.
409
 
 
 
433
        
410
434
        This implementation requests the ids list from the dirstate file.
411
435
        """
412
436
        return self.current_dirstate().get_parent_ids()
434
458
        return osutils.lexists(pathjoin(
435
459
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
436
460
 
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
461
    @needs_read_lock
443
462
    def id2path(self, file_id):
444
463
        "Convert a file-id to a path."
474
493
 
475
494
            Note: The caller is expected to take a read-lock before calling this.
476
495
            """
477
 
            self._must_be_locked()
478
496
            if not path:
479
497
                path = self.id2path(file_id)
480
498
            mode = os.lstat(self.abspath(path)).st_mode
481
499
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
482
500
 
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
501
    @needs_read_lock
494
502
    def __iter__(self):
495
503
        """Iterate through file_ids for this tree.
508
516
        return iter(result)
509
517
 
510
518
    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
519
        for key, tree_details in self.current_dirstate()._iter_entries():
516
520
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
517
521
                # not relevant to the working tree
519
523
            if not key[1]:
520
524
                # the root is not a reference.
521
525
                continue
522
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
526
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
523
527
            try:
524
 
                if self._kind(relpath) == 'tree-reference':
525
 
                    yield relpath, key[2]
 
528
                if self._kind(path) == 'tree-reference':
 
529
                    yield path, key[2]
526
530
            except errors.NoSuchFile:
527
531
                # path is missing on disk.
528
532
                continue
529
533
 
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
534
    def kind(self, file_id):
537
535
        """Return the kind of a file.
538
536
 
542
540
        Note: The caller is expected to take a read-lock before calling this.
543
541
        """
544
542
        relpath = self.id2path(file_id)
545
 
        if relpath is None:
546
 
            raise AssertionError(
547
 
                "path for id {%s} is None!" % file_id)
 
543
        assert relpath != None, \
 
544
            "path for id {%s} is None!" % file_id
548
545
        return self._kind(relpath)
549
546
 
550
547
    def _kind(self, relpath):
551
548
        abspath = self.abspath(relpath)
552
549
        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'
 
550
        if (self._repo_supports_tree_reference and
 
551
            kind == 'directory' and
 
552
            self._directory_is_tree_reference(relpath)):
 
553
            kind = 'tree-reference'
558
554
        return kind
559
555
 
560
556
    @needs_read_lock
623
619
        result = []
624
620
        if not from_paths:
625
621
            return result
 
622
 
626
623
        state = self.current_dirstate()
627
 
        if isinstance(from_paths, basestring):
628
 
            raise ValueError()
 
624
 
 
625
        assert not isinstance(from_paths, basestring)
629
626
        to_dir_utf8 = to_dir.encode('utf8')
630
627
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
631
628
        id_index = state._get_id_index()
653
650
        if self._inventory is not None:
654
651
            update_inventory = True
655
652
            inv = self.inventory
 
653
            to_dir_ie = inv[to_dir_id]
656
654
            to_dir_id = to_entry[0][2]
657
 
            to_dir_ie = inv[to_dir_id]
658
655
        else:
659
656
            update_inventory = False
660
657
 
691
688
            from_entry = self._get_entry(path=from_rel)
692
689
            if from_entry == (None, None):
693
690
                raise errors.BzrMoveFailedError(from_rel,to_dir,
694
 
                    errors.NotVersionedError(path=from_rel))
 
691
                    errors.NotVersionedError(path=str(from_rel)))
695
692
 
696
693
            from_id = from_entry[0][2]
697
694
            to_rel = pathjoin(to_dir, from_tail)
786
783
                if minikind == 'd':
787
784
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
788
785
                        """Recursively update all entries in this dirblock."""
789
 
                        if from_dir == '':
790
 
                            raise AssertionError("renaming root not supported")
 
786
                        assert from_dir != '', "renaming root not supported"
791
787
                        from_key = (from_dir, '')
792
788
                        from_block_idx, present = \
793
789
                            state._find_block_index_from_key(from_key)
806
802
 
807
803
                        # Grab a copy since move_one may update the list.
808
804
                        for entry in from_block[1][:]:
809
 
                            if not (entry[0][0] == from_dir):
810
 
                                raise AssertionError()
 
805
                            assert entry[0][0] == from_dir
811
806
                            cur_details = entry[1][0]
812
807
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
813
808
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
872
867
        for tree in trees:
873
868
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
874
869
                parents):
875
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
876
 
                    trees, require_versioned)
 
870
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
877
871
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
878
872
        # -- make all paths utf8 --
879
873
        paths_utf8 = set()
940
934
                raise errors.PathsNotVersionedError(paths)
941
935
        # -- remove redundancy in supplied paths to prevent over-scanning --
942
936
        search_paths = osutils.minimum_path_selection(paths)
943
 
        # sketch:
 
937
        # sketch: 
944
938
        # for all search_indexs in each path at or under each element of
945
939
        # search_paths, if the detail is relocated: add the id, and add the
946
940
        # relocated path as one to search if its not searched already. If the
1002
996
 
1003
997
    def read_working_inventory(self):
1004
998
        """Read the working inventory.
1005
 
 
 
999
        
1006
1000
        This is a meaningless operation for dirstate, but we obey it anyhow.
1007
1001
        """
1008
1002
        return self.inventory
1026
1020
    def set_last_revision(self, new_revision):
1027
1021
        """Change the last revision in the working tree."""
1028
1022
        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.")
 
1023
        if new_revision in (NULL_REVISION, None):
 
1024
            assert len(parents) < 2, (
 
1025
                "setting the last parent to none with a pending merge is "
 
1026
                "unsupported.")
1034
1027
            self.set_parent_ids([])
1035
1028
        else:
1036
1029
            self.set_parent_ids([new_revision] + parents[1:],
1039
1032
    @needs_tree_write_lock
1040
1033
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1041
1034
        """Set the parent ids to revision_ids.
1042
 
 
 
1035
        
1043
1036
        See also set_parent_trees. This api will try to retrieve the tree data
1044
1037
        for each element of revision_ids from the trees repository. If you have
1045
1038
        tree data already available, it is more efficient to use
1077
1070
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1078
1071
        real_trees = []
1079
1072
        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
1073
        # convert absent trees to the null tree, which we convert back to
1087
1074
        # missing on access.
1088
1075
        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
1076
            _mod_revision.check_not_reserved_id(rev_id)
1096
1077
            if tree is not None:
1097
1078
                real_trees.append((rev_id, tree))
1098
1079
            else:
1099
1080
                real_trees.append((rev_id,
1100
 
                    self.branch.repository.revision_tree(
1101
 
                        _mod_revision.NULL_REVISION)))
 
1081
                    self.branch.repository.revision_tree(None)))
1102
1082
                ghosts.append(rev_id)
1103
 
            accepted_revisions.add(rev_id)
1104
1083
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
1105
1084
        self._make_dirty(reset_inventory=False)
1106
1085
 
1199
1178
                # just forget the whole block.
1200
1179
                entry_index = 0
1201
1180
                while entry_index < len(block[1]):
 
1181
                    # Mark this file id as having been removed
1202
1182
                    entry = block[1][entry_index]
1203
 
                    if entry[1][0][0] in 'ar':
1204
 
                        # don't remove absent or renamed entries
 
1183
                    ids_to_unversion.discard(entry[0][2])
 
1184
                    if (entry[1][0][0] in 'ar' # don't remove absent or renamed
 
1185
                                               # entries
 
1186
                        or not state._make_absent(entry)):
1205
1187
                        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
1188
                # go to the next block. (At the moment we dont delete empty
1213
1189
                # dirblocks)
1214
1190
                block_index += 1
1237
1213
            for file_id in file_ids:
1238
1214
                self._inventory.remove_recursive_id(file_id)
1239
1215
 
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
1216
    def update_basis_by_delta(self, new_revid, delta):
1254
1217
        """See MutableTree.update_basis_by_delta."""
1255
 
        if self.last_revision() == new_revid:
1256
 
            raise AssertionError()
 
1218
        assert self.last_revision() != new_revid
1257
1219
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
1258
1220
 
1259
1221
    @needs_read_lock
1263
1225
    @needs_tree_write_lock
1264
1226
    def _write_inventory(self, inv):
1265
1227
        """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:
 
1228
        assert not self._dirty, ("attempting to write an inventory when the "
 
1229
            "dirstate is dirty will cause data loss")
 
1230
        self.current_dirstate().set_state_from_inventory(inv)
 
1231
        self._make_dirty(reset_inventory=False)
 
1232
        if self._inventory is not None:
1280
1233
            self._inventory = inv
1281
1234
        self.flush()
1282
1235
 
1283
1236
 
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):
 
1237
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1238
    """The first consolidated dirstate working tree format.
 
1239
 
 
1240
    This format:
 
1241
        - exists within a metadir controlling .bzr
 
1242
        - includes an explicit version marker for the workingtree control
 
1243
          files, separate from the BzrDir format
 
1244
        - modifies the hash cache format
 
1245
        - is new in bzr 0.15
 
1246
        - uses a LockDir to guard access to it.
 
1247
    """
 
1248
 
 
1249
    upgrade_recommended = False
 
1250
 
 
1251
    def get_format_string(self):
 
1252
        """See WorkingTreeFormat.get_format_string()."""
 
1253
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1254
 
 
1255
    def get_format_description(self):
 
1256
        """See WorkingTreeFormat.get_format_description()."""
 
1257
        return "Working tree format 4"
 
1258
 
 
1259
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None):
1377
1260
        """See WorkingTreeFormat.initialize().
1378
1261
 
1379
1262
        :param revision_id: allows creating a working tree at a different
1380
1263
        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
1264
 
1388
1265
        These trees get an initial random root id, if their repository supports
1389
1266
        rich root data, TREE_ROOT otherwise.
1394
1271
        control_files = self._open_control_files(a_bzrdir)
1395
1272
        control_files.create_lock()
1396
1273
        control_files.lock_write()
1397
 
        transport.put_bytes('format', self.get_format_string(),
1398
 
            mode=a_bzrdir._get_file_mode())
 
1274
        control_files.put_utf8('format', self.get_format_string())
1399
1275
        if from_branch is not None:
1400
1276
            branch = from_branch
1401
1277
        else:
1407
1283
        state = dirstate.DirState.initialize(local_path)
1408
1284
        state.unlock()
1409
1285
        del state
1410
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1286
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1411
1287
                         branch,
1412
1288
                         _format=self,
1413
1289
                         _bzrdir=a_bzrdir,
1415
1291
        wt._new_tree()
1416
1292
        wt.lock_tree_write()
1417
1293
        try:
1418
 
            self._init_custom_control_files(wt)
1419
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
 
1294
            if revision_id in (None, NULL_REVISION):
1420
1295
                if branch.repository.supports_rich_root():
1421
1296
                    wt._set_root_id(generate_ids.gen_root_id())
1422
1297
                else:
1423
1298
                    wt._set_root_id(ROOT_ID)
1424
1299
                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)]
 
1300
            wt.set_last_revision(revision_id)
 
1301
            wt.flush()
 
1302
            basis = wt.basis_tree()
1440
1303
            basis.lock_read()
1441
 
            try:
1442
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1304
            # if the basis has a root id we have to use that; otherwise we use
 
1305
            # a new random one
 
1306
            basis_root_id = basis.get_root_id()
 
1307
            if basis_root_id is not None:
 
1308
                wt._set_root_id(basis_root_id)
1443
1309
                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()
 
1310
            transform.build_tree(basis, wt)
 
1311
            basis.unlock()
1465
1312
        finally:
1466
1313
            control_files.unlock()
1467
1314
            wt.unlock()
1468
1315
        return wt
1469
1316
 
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
1317
    def _open(self, a_bzrdir, control_files):
1480
1318
        """Open the tree itself.
1481
1319
 
1482
1320
        :param a_bzrdir: the dir for the tree.
1483
1321
        :param control_files: the control files for the tree.
1484
1322
        """
1485
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1323
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1486
1324
                           branch=a_bzrdir.open_branch(),
1487
1325
                           _format=self,
1488
1326
                           _bzrdir=a_bzrdir,
1489
1327
                           _control_files=control_files)
1490
1328
 
1491
1329
    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
1330
        # please test against something that will let us do tree references
1497
1331
        return bzrdir.format_registry.make_bzrdir(
1498
1332
            'dirstate-with-subtree')
1500
1334
    _matchingbzrdir = property(__get_matchingbzrdir)
1501
1335
 
1502
1336
 
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
1337
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
 
    """
 
1338
    """A revision tree pulling the inventory from a dirstate."""
1581
1339
 
1582
1340
    def __init__(self, dirstate, revision_id, repository):
1583
1341
        self._dirstate = dirstate
1586
1344
        self._inventory = None
1587
1345
        self._locked = 0
1588
1346
        self._dirstate_locked = False
1589
 
        self._repo_supports_tree_reference = getattr(
1590
 
            repository._format, "supports_tree_reference",
1591
 
            False)
1592
1347
 
1593
1348
    def __repr__(self):
1594
1349
        return "<%s of %s in %s>" % \
1597
1352
    def annotate_iter(self, file_id,
1598
1353
                      default_revision=_mod_revision.CURRENT_REVISION):
1599
1354
        """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]
 
1355
        w = self._get_weave(file_id)
 
1356
        return w.annotate_iter(self.inventory[file_id].revision)
1603
1357
 
1604
1358
    def _get_ancestors(self, default_revision):
1605
1359
        return set(self._repository.get_ancestry(self._revision_id,
1634
1388
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1635
1389
        return path_utf8.decode('utf8')
1636
1390
 
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
1391
    def _get_parent_index(self):
1646
1392
        """Return the index in the dirstate referenced by this tree."""
1647
1393
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1652
1398
        If either file_id or path is supplied, it is used as the key to lookup.
1653
1399
        If both are supplied, the fastest lookup is used, and an error is
1654
1400
        raised if they do not both point at the same row.
1655
 
 
 
1401
        
1656
1402
        :param file_id: An optional unicode file_id to be looked up.
1657
1403
        :param path: An optional unicode path to be looked up.
1658
1404
        :return: The dirstate row tuple for path/file_id, or (None, None)
1672
1418
 
1673
1419
        This is relatively expensive: we have to walk the entire dirstate.
1674
1420
        """
1675
 
        if not self._locked:
1676
 
            raise AssertionError(
1677
 
                'cannot generate inventory of an unlocked '
1678
 
                'dirstate revision tree')
 
1421
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1422
            'dirstate revision tree'
1679
1423
        # separate call for profiling - makes it clear where the costs are.
1680
1424
        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()))
 
1425
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1426
            'parent %s has disappeared from %s' % (
 
1427
            self._revision_id, self._dirstate.get_parent_ids())
1685
1428
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1686
1429
        # This is identical now to the WorkingTree _generate_inventory except
1687
1430
        # for the tree index use.
1688
1431
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1689
1432
        current_id = root_key[2]
1690
 
        if current_entry[parent_index][0] != 'd':
1691
 
            raise AssertionError()
 
1433
        assert current_entry[parent_index][0] == 'd'
1692
1434
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1693
1435
        inv.root.revision = current_entry[parent_index][4]
1694
1436
        # Turn some things into local variables
1734
1476
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1735
1477
                            % entry)
1736
1478
                # 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,))
 
1479
                assert file_id not in inv_byid
 
1480
                assert name_unicode not in parent_ie.children
1743
1481
                inv_byid[file_id] = inv_entry
1744
1482
                parent_ie.children[name_unicode] = inv_entry
1745
1483
        self._inventory = inv
1755
1493
            return None
1756
1494
        parent_index = self._get_parent_index()
1757
1495
        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
 
1496
        return self._repository.get_revision(last_changed_revision).timestamp
1763
1497
 
1764
1498
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1765
1499
        entry = self._get_entry(file_id=file_id, path=path)
1769
1503
            return parent_details[1]
1770
1504
        return None
1771
1505
 
 
1506
    @symbol_versioning.deprecated_method(symbol_versioning.zero_ninety)
 
1507
    def get_weave(self, file_id):
 
1508
        return self._get_weave(file_id)
 
1509
 
 
1510
    def _get_weave(self, file_id):
 
1511
        return self._repository.weave_store.get_weave(file_id,
 
1512
                self._repository.get_transaction())
 
1513
 
1772
1514
    def get_file(self, file_id, path=None):
1773
1515
        return StringIO(self.get_file_text(file_id))
1774
1516
 
 
1517
    def get_file_lines(self, file_id):
 
1518
        entry = self._get_entry(file_id=file_id)[1]
 
1519
        if entry == None:
 
1520
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1521
        return self._get_weave(file_id).get_lines(entry[1][4])
 
1522
 
1775
1523
    def get_file_size(self, file_id):
1776
 
        """See Tree.get_file_size"""
1777
1524
        return self.inventory[file_id].text_size
1778
1525
 
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)
 
1526
    def get_file_text(self, file_id):
 
1527
        return ''.join(self.get_file_lines(file_id))
1782
1528
 
1783
1529
    def get_reference_revision(self, file_id, path=None):
1784
1530
        return self.inventory[file_id].reference_revision
1803
1549
        if entry[1][parent_index][0] != 'l':
1804
1550
            return None
1805
1551
        else:
1806
 
            target = entry[1][parent_index][1]
1807
 
            target = target.decode('utf8')
1808
 
            return target
 
1552
            # At present, none of the tree implementations supports non-ascii
 
1553
            # symlink targets. So we will just assume that the dirstate path is
 
1554
            # correct.
 
1555
            return entry[1][parent_index][1]
1809
1556
 
1810
1557
    def get_revision_id(self):
1811
1558
        """Return the revision id for this tree."""
1830
1577
 
1831
1578
    def kind(self, file_id):
1832
1579
        entry = self._get_entry(file_id=file_id)[1]
1833
 
        if entry is None:
 
1580
        if entry == None:
1834
1581
            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)
 
1582
        return dirstate.DirState._minikind_to_kind[entry[1][0]]
1841
1583
 
1842
1584
    def path_content_summary(self, path):
1843
1585
        """See Tree.path_content_summary."""
1859
1601
            return None
1860
1602
        return ie.executable
1861
1603
 
1862
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1604
    def list_files(self, include_root=False):
1863
1605
        # We use a standard implementation, because DirStateRevisionTree is
1864
1606
        # dealing with one of the parents of the current state
1865
1607
        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:
 
1608
        entries = inv.iter_entries()
 
1609
        if self.inventory.root is not None and not include_root:
1875
1610
            entries.next()
1876
1611
        for path, entry in entries:
1877
1612
            yield path, 'V', entry.kind, entry.file_id, entry
1910
1645
                self._dirstate_locked = False
1911
1646
            self._repository.unlock()
1912
1647
 
1913
 
    @needs_read_lock
1914
 
    def supports_tree_reference(self):
1915
 
        return self._repo_supports_tree_reference
1916
 
 
1917
1648
    def walkdirs(self, prefix=""):
1918
1649
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1919
 
        # implementation based on an inventory.
 
1650
        # implementation based on an inventory.  
1920
1651
        # This should be cleaned up to use the much faster Dirstate code
1921
1652
        # So for now, we just build up the parent inventory, and extract
1922
1653
        # it the same way RevisionTree does.
1951
1682
 
1952
1683
class InterDirStateTree(InterTree):
1953
1684
    """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
 
1685
    
 
1686
    This is used only when both trees are in the dirstate working file, and 
 
1687
    the source is any parent within the dirstate, and the destination is 
1957
1688
    the current working tree of the same dirstate.
1958
1689
    """
1959
1690
    # this could be generalized to allow comparisons between any trees in the
1972
1703
        target.set_parent_ids([revid])
1973
1704
        return target.basis_tree(), target
1974
1705
 
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
1706
    _matching_from_tree_format = WorkingTreeFormat4()
1993
1707
    _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,
 
1708
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1709
 
 
1710
    def _iter_changes(self, include_unchanged=False,
2002
1711
                      specific_files=None, pb=None, extra_trees=[],
2003
1712
                      require_versioned=True, want_unversioned=False):
2004
1713
        """Return the changes from source to target.
2005
1714
 
2006
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1715
        :return: An iterator that yields tuples. See InterTree._iter_changes
2007
1716
            for details.
2008
1717
        :param specific_files: An optional list of file paths to restrict the
2009
1718
            comparison to. When mapping filenames to ids, all matches in all
2020
1729
            output. An unversioned file is defined as one with (False, False)
2021
1730
            for the versioned pair.
2022
1731
        """
 
1732
        utf8_decode = cache_utf8._utf8_decode
 
1733
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1734
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1735
        # NB: show_status depends on being able to pass in non-versioned files
 
1736
        # and report them as unknown
2023
1737
        # TODO: handle extra trees in the dirstate.
2024
 
        if (extra_trees or specific_files == []):
 
1738
        # TODO: handle comparisons as an empty tree as a different special
 
1739
        # case? mbp 20070226
 
1740
        if (extra_trees or (self.source._revision_id == NULL_REVISION)
 
1741
            or specific_files == []):
2025
1742
            # we can't fast-path these cases (yet)
2026
 
            return super(InterDirStateTree, self).iter_changes(
 
1743
            for f in super(InterDirStateTree, self)._iter_changes(
2027
1744
                include_unchanged, specific_files, pb, extra_trees,
2028
 
                require_versioned, want_unversioned=want_unversioned)
 
1745
                require_versioned, want_unversioned=want_unversioned):
 
1746
                yield f
 
1747
            return
2029
1748
        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))
 
1749
        assert (self.source._revision_id in parent_ids), \
 
1750
                "revision {%s} is not stored in {%s}, but %s " \
 
1751
                "can only be used for trees stored in the dirstate" \
 
1752
                % (self.source._revision_id, self.target, self._iter_changes)
2036
1753
        target_index = 0
2037
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1754
        if self.source._revision_id == NULL_REVISION:
2038
1755
            source_index = None
2039
1756
            indices = (target_index,)
2040
1757
        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))
 
1758
            assert (self.source._revision_id in parent_ids), \
 
1759
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1760
                self.source._revision_id, parent_ids)
2045
1761
            source_index = 1 + parent_ids.index(self.source._revision_id)
2046
 
            indices = (source_index, target_index)
 
1762
            indices = (source_index,target_index)
2047
1763
        # -- make all specific_files utf8 --
2048
1764
        if specific_files:
2049
1765
            specific_files_utf8 = set()
2050
1766
            for path in specific_files:
2051
 
                # Note, if there are many specific files, using cache_utf8
2052
 
                # would be good here.
2053
1767
                specific_files_utf8.add(path.encode('utf8'))
2054
1768
            specific_files = specific_files_utf8
2055
1769
        else:
2056
1770
            specific_files = set([''])
2057
1771
        # -- specific_files is now a utf8 path set --
2058
 
 
2059
1772
        # -- get the state object and prepare it.
2060
1773
        state = self.target.current_dirstate()
2061
1774
        state._read_dirblocks_if_needed()
 
1775
        def _entries_for_path(path):
 
1776
            """Return a list with all the entries that match path for all ids.
 
1777
            """
 
1778
            dirname, basename = os.path.split(path)
 
1779
            key = (dirname, basename, '')
 
1780
            block_index, present = state._find_block_index_from_key(key)
 
1781
            if not present:
 
1782
                # the block which should contain path is absent.
 
1783
                return []
 
1784
            result = []
 
1785
            block = state._dirblocks[block_index][1]
 
1786
            entry_index, _ = state._find_entry_index(key, block)
 
1787
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1788
            while (entry_index < len(block) and
 
1789
                block[entry_index][0][0:2] == key[0:2]):
 
1790
                result.append(block[entry_index])
 
1791
                entry_index += 1
 
1792
            return result
2062
1793
        if require_versioned:
2063
1794
            # -- check all supplied paths are versioned in a search tree. --
2064
 
            not_versioned = []
 
1795
            all_versioned = True
2065
1796
            for path in specific_files:
2066
 
                path_entries = state._entries_for_path(path)
 
1797
                path_entries = _entries_for_path(path)
2067
1798
                if not path_entries:
2068
1799
                    # this specified path is not present at all: error
2069
 
                    not_versioned.append(path)
2070
 
                    continue
 
1800
                    all_versioned = False
 
1801
                    break
2071
1802
                found_versioned = False
2072
1803
                # for each id at this path
2073
1804
                for entry in path_entries:
2080
1811
                if not found_versioned:
2081
1812
                    # none of the indexes was not 'absent' at all ids for this
2082
1813
                    # path.
2083
 
                    not_versioned.append(path)
2084
 
            if len(not_versioned) > 0:
2085
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1814
                    all_versioned = False
 
1815
                    break
 
1816
            if not all_versioned:
 
1817
                raise errors.PathsNotVersionedError(specific_files)
2086
1818
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2087
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1819
        search_specific_files = set()
 
1820
        for path in specific_files:
 
1821
            other_specific_files = specific_files.difference(set([path]))
 
1822
            if not osutils.is_inside_any(other_specific_files, path):
 
1823
                # this is a top level path, we must check it.
 
1824
                search_specific_files.add(path)
 
1825
        # sketch: 
 
1826
        # compare source_index and target_index at or under each element of search_specific_files.
 
1827
        # follow the following comparison table. Note that we only want to do diff operations when
 
1828
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1829
        # for the target.
 
1830
        # cases:
 
1831
        # 
 
1832
        # Source | Target | disk | action
 
1833
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1834
        #        |        |      | diff check on source-target
 
1835
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1836
        #        |        |      | ???
 
1837
        #   r    |  a     |      | add source to search
 
1838
        #   r    |  a     |  a   | 
 
1839
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1840
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1841
        #   a    | fdlt   |      | add new id
 
1842
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1843
        #   a    |  a     |      | not present in either tree, skip
 
1844
        #   a    |  a     |  a   | not present in any tree, skip
 
1845
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1846
        #        |        |      | may not be selected by the users list of paths.
 
1847
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1848
        #        |        |      | may not be selected by the users list of paths.
 
1849
        #  fdlt  | fdlt   |      | content in both: diff them
 
1850
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1851
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1852
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1853
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1854
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1855
        #        |        |      | this id at the other path.
 
1856
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1857
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1858
        #        |        |      | this id at the other path.
 
1859
 
 
1860
        # for all search_indexs in each path at or under each element of
 
1861
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1862
        # relocated path as one to search if its not searched already. If the
 
1863
        # detail is not relocated, add the id.
 
1864
        searched_specific_files = set()
 
1865
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1866
        # Using a list so that we can access the values and change them in
 
1867
        # nested scope. Each one is [path, file_id, entry]
 
1868
        last_source_parent = [None, None]
 
1869
        last_target_parent = [None, None]
2088
1870
 
2089
1871
        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()
 
1872
 
 
1873
        # Just a sentry, so that _process_entry can say that this
 
1874
        # record is handled, but isn't interesting to process (unchanged)
 
1875
        uninteresting = object()
 
1876
 
 
1877
 
 
1878
        old_dirname_to_file_id = {}
 
1879
        new_dirname_to_file_id = {}
 
1880
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1881
        #       keeping a cache of directories that we have seen.
 
1882
 
 
1883
        def _process_entry(entry, path_info):
 
1884
            """Compare an entry and real disk to generate delta information.
 
1885
 
 
1886
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1887
                the path of entry. If None, then the path is considered absent.
 
1888
                (Perhaps we should pass in a concrete entry for this ?)
 
1889
                Basename is returned as a utf8 string because we expect this
 
1890
                tuple will be ignored, and don't want to take the time to
 
1891
                decode.
 
1892
            :return: None if these don't match
 
1893
                     A tuple of information about the change, or
 
1894
                     the object 'uninteresting' if these match, but are
 
1895
                     basically identical.
 
1896
            """
 
1897
            if source_index is None:
 
1898
                source_details = NULL_PARENT_DETAILS
 
1899
            else:
 
1900
                source_details = entry[1][source_index]
 
1901
            target_details = entry[1][target_index]
 
1902
            target_minikind = target_details[0]
 
1903
            if path_info is not None and target_minikind in 'fdlt':
 
1904
                assert target_index == 0
 
1905
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1906
                                                  stat_value=path_info[3])
 
1907
                # The entry may have been modified by update_entry
 
1908
                target_details = entry[1][target_index]
 
1909
                target_minikind = target_details[0]
 
1910
            else:
 
1911
                link_or_sha1 = None
 
1912
            file_id = entry[0][2]
 
1913
            source_minikind = source_details[0]
 
1914
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1915
                # claimed content in both: diff
 
1916
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1917
                #        |        |      | diff check on source-target
 
1918
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1919
                #        |        |      | ???
 
1920
                if source_minikind in 'r':
 
1921
                    # add the source to the search path to find any children it
 
1922
                    # has.  TODO ? : only add if it is a container ?
 
1923
                    if not osutils.is_inside_any(searched_specific_files,
 
1924
                                                 source_details[1]):
 
1925
                        search_specific_files.add(source_details[1])
 
1926
                    # generate the old path; this is needed for stating later
 
1927
                    # as well.
 
1928
                    old_path = source_details[1]
 
1929
                    old_dirname, old_basename = os.path.split(old_path)
 
1930
                    path = pathjoin(entry[0][0], entry[0][1])
 
1931
                    old_entry = state._get_entry(source_index,
 
1932
                                                 path_utf8=old_path)
 
1933
                    # update the source details variable to be the real
 
1934
                    # location.
 
1935
                    source_details = old_entry[1][source_index]
 
1936
                    source_minikind = source_details[0]
 
1937
                else:
 
1938
                    old_dirname = entry[0][0]
 
1939
                    old_basename = entry[0][1]
 
1940
                    old_path = path = None
 
1941
                if path_info is None:
 
1942
                    # the file is missing on disk, show as removed.
 
1943
                    content_change = True
 
1944
                    target_kind = None
 
1945
                    target_exec = False
 
1946
                else:
 
1947
                    # source and target are both versioned and disk file is present.
 
1948
                    target_kind = path_info[2]
 
1949
                    if target_kind == 'directory':
 
1950
                        if path is None:
 
1951
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1952
                        new_dirname_to_file_id[path] = file_id
 
1953
                        if source_minikind != 'd':
 
1954
                            content_change = True
 
1955
                        else:
 
1956
                            # directories have no fingerprint
 
1957
                            content_change = False
 
1958
                        target_exec = False
 
1959
                    elif target_kind == 'file':
 
1960
                        if source_minikind != 'f':
 
1961
                            content_change = True
 
1962
                        else:
 
1963
                            # We could check the size, but we already have the
 
1964
                            # sha1 hash.
 
1965
                            content_change = (link_or_sha1 != source_details[1])
 
1966
                        # Target details is updated at update_entry time
 
1967
                        if use_filesystem_for_exec:
 
1968
                            # We don't need S_ISREG here, because we are sure
 
1969
                            # we are dealing with a file.
 
1970
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1971
                        else:
 
1972
                            target_exec = target_details[3]
 
1973
                    elif target_kind == 'symlink':
 
1974
                        if source_minikind != 'l':
 
1975
                            content_change = True
 
1976
                        else:
 
1977
                            content_change = (link_or_sha1 != source_details[1])
 
1978
                        target_exec = False
 
1979
                    elif target_kind == 'tree-reference':
 
1980
                        if source_minikind != 't':
 
1981
                            content_change = True
 
1982
                        else:
 
1983
                            content_change = False
 
1984
                        target_exec = False
 
1985
                    else:
 
1986
                        raise Exception, "unknown kind %s" % path_info[2]
 
1987
                if source_minikind == 'd':
 
1988
                    if path is None:
 
1989
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1990
                    old_dirname_to_file_id[old_path] = file_id
 
1991
                # parent id is the entry for the path in the target tree
 
1992
                if old_dirname == last_source_parent[0]:
 
1993
                    source_parent_id = last_source_parent[1]
 
1994
                else:
 
1995
                    try:
 
1996
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1997
                    except KeyError:
 
1998
                        source_parent_entry = state._get_entry(source_index,
 
1999
                                                               path_utf8=old_dirname)
 
2000
                        source_parent_id = source_parent_entry[0][2]
 
2001
                    if source_parent_id == entry[0][2]:
 
2002
                        # This is the root, so the parent is None
 
2003
                        source_parent_id = None
 
2004
                    else:
 
2005
                        last_source_parent[0] = old_dirname
 
2006
                        last_source_parent[1] = source_parent_id
 
2007
                new_dirname = entry[0][0]
 
2008
                if new_dirname == last_target_parent[0]:
 
2009
                    target_parent_id = last_target_parent[1]
 
2010
                else:
 
2011
                    try:
 
2012
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
2013
                    except KeyError:
 
2014
                        # TODO: We don't always need to do the lookup, because the
 
2015
                        #       parent entry will be the same as the source entry.
 
2016
                        target_parent_entry = state._get_entry(target_index,
 
2017
                                                               path_utf8=new_dirname)
 
2018
                        assert target_parent_entry != (None, None), (
 
2019
                            "Could not find target parent in wt: %s\nparent of: %s"
 
2020
                            % (new_dirname, entry))
 
2021
                        target_parent_id = target_parent_entry[0][2]
 
2022
                    if target_parent_id == entry[0][2]:
 
2023
                        # This is the root, so the parent is None
 
2024
                        target_parent_id = None
 
2025
                    else:
 
2026
                        last_target_parent[0] = new_dirname
 
2027
                        last_target_parent[1] = target_parent_id
 
2028
 
 
2029
                source_exec = source_details[3]
 
2030
                if (include_unchanged
 
2031
                    or content_change
 
2032
                    or source_parent_id != target_parent_id
 
2033
                    or old_basename != entry[0][1]
 
2034
                    or source_exec != target_exec
 
2035
                    ):
 
2036
                    if old_path is None:
 
2037
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2038
                        old_path_u = utf8_decode(old_path)[0]
 
2039
                        path_u = old_path_u
 
2040
                    else:
 
2041
                        old_path_u = utf8_decode(old_path)[0]
 
2042
                        if old_path == path:
 
2043
                            path_u = old_path_u
 
2044
                        else:
 
2045
                            path_u = utf8_decode(path)[0]
 
2046
                    source_kind = _minikind_to_kind[source_minikind]
 
2047
                    return (entry[0][2],
 
2048
                           (old_path_u, path_u),
 
2049
                           content_change,
 
2050
                           (True, True),
 
2051
                           (source_parent_id, target_parent_id),
 
2052
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2053
                           (source_kind, target_kind),
 
2054
                           (source_exec, target_exec))
 
2055
                else:
 
2056
                    return uninteresting
 
2057
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2058
                # looks like a new file
 
2059
                if path_info is not None:
 
2060
                    path = pathjoin(entry[0][0], entry[0][1])
 
2061
                    # parent id is the entry for the path in the target tree
 
2062
                    # TODO: these are the same for an entire directory: cache em.
 
2063
                    parent_id = state._get_entry(target_index,
 
2064
                                                 path_utf8=entry[0][0])[0][2]
 
2065
                    if parent_id == entry[0][2]:
 
2066
                        parent_id = None
 
2067
                    if use_filesystem_for_exec:
 
2068
                        # We need S_ISREG here, because we aren't sure if this
 
2069
                        # is a file or not.
 
2070
                        target_exec = bool(
 
2071
                            stat.S_ISREG(path_info[3].st_mode)
 
2072
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2073
                    else:
 
2074
                        target_exec = target_details[3]
 
2075
                    return (entry[0][2],
 
2076
                           (None, utf8_decode(path)[0]),
 
2077
                           True,
 
2078
                           (False, True),
 
2079
                           (None, parent_id),
 
2080
                           (None, utf8_decode(entry[0][1])[0]),
 
2081
                           (None, path_info[2]),
 
2082
                           (None, target_exec))
 
2083
                else:
 
2084
                    # but its not on disk: we deliberately treat this as just
 
2085
                    # never-present. (Why ?! - RBC 20070224)
 
2086
                    pass
 
2087
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2088
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2089
                # if its still on disk, *and* theres no other entry at this
 
2090
                # path [we dont know this in this routine at the moment -
 
2091
                # perhaps we should change this - then it would be an unknown.
 
2092
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2093
                # parent id is the entry for the path in the target tree
 
2094
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2095
                if parent_id == entry[0][2]:
 
2096
                    parent_id = None
 
2097
                return (entry[0][2],
 
2098
                       (utf8_decode(old_path)[0], None),
 
2099
                       True,
 
2100
                       (True, False),
 
2101
                       (parent_id, None),
 
2102
                       (utf8_decode(entry[0][1])[0], None),
 
2103
                       (_minikind_to_kind[source_minikind], None),
 
2104
                       (source_details[3], None))
 
2105
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2106
                # a rename; could be a true rename, or a rename inherited from
 
2107
                # a renamed parent. TODO: handle this efficiently. Its not
 
2108
                # common case to rename dirs though, so a correct but slow
 
2109
                # implementation will do.
 
2110
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2111
                    search_specific_files.add(target_details[1])
 
2112
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2113
                # neither of the selected trees contain this file,
 
2114
                # so skip over it. This is not currently directly tested, but
 
2115
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2116
                pass
 
2117
            else:
 
2118
                raise AssertionError("don't know how to compare "
 
2119
                    "source_minikind=%r, target_minikind=%r"
 
2120
                    % (source_minikind, target_minikind))
 
2121
                ## import pdb;pdb.set_trace()
 
2122
            return None
 
2123
 
 
2124
        while search_specific_files:
 
2125
            # TODO: the pending list should be lexically sorted?  the
 
2126
            # interface doesn't require it.
 
2127
            current_root = search_specific_files.pop()
 
2128
            current_root_unicode = current_root.decode('utf8')
 
2129
            searched_specific_files.add(current_root)
 
2130
            # process the entries for this containing directory: the rest will be
 
2131
            # found by their parents recursively.
 
2132
            root_entries = _entries_for_path(current_root)
 
2133
            root_abspath = self.target.abspath(current_root_unicode)
 
2134
            try:
 
2135
                root_stat = os.lstat(root_abspath)
 
2136
            except OSError, e:
 
2137
                if e.errno == errno.ENOENT:
 
2138
                    # the path does not exist: let _process_entry know that.
 
2139
                    root_dir_info = None
 
2140
                else:
 
2141
                    # some other random error: hand it up.
 
2142
                    raise
 
2143
            else:
 
2144
                root_dir_info = ('', current_root,
 
2145
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2146
                    root_abspath)
 
2147
                if root_dir_info[2] == 'directory':
 
2148
                    if self.target._directory_is_tree_reference(
 
2149
                        current_root.decode('utf8')):
 
2150
                        root_dir_info = root_dir_info[:2] + \
 
2151
                            ('tree-reference',) + root_dir_info[3:]
 
2152
 
 
2153
            if not root_entries and not root_dir_info:
 
2154
                # this specified path is not present at all, skip it.
 
2155
                continue
 
2156
            path_handled = False
 
2157
            for entry in root_entries:
 
2158
                result = _process_entry(entry, root_dir_info)
 
2159
                if result is not None:
 
2160
                    path_handled = True
 
2161
                    if result is not uninteresting:
 
2162
                        yield result
 
2163
            if want_unversioned and not path_handled and root_dir_info:
 
2164
                new_executable = bool(
 
2165
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2166
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2167
                yield (None,
 
2168
                       (None, current_root_unicode),
 
2169
                       True,
 
2170
                       (False, False),
 
2171
                       (None, None),
 
2172
                       (None, splitpath(current_root_unicode)[-1]),
 
2173
                       (None, root_dir_info[2]),
 
2174
                       (None, new_executable)
 
2175
                      )
 
2176
            initial_key = (current_root, '', '')
 
2177
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2178
            if block_index == 0:
 
2179
                # we have processed the total root already, but because the
 
2180
                # initial key matched it we should skip it here.
 
2181
                block_index +=1
 
2182
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2183
                current_dir_info = None
 
2184
            else:
 
2185
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2186
                try:
 
2187
                    current_dir_info = dir_iterator.next()
 
2188
                except OSError, e:
 
2189
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2190
                    # python 2.5 has e.errno == EINVAL,
 
2191
                    #            and e.winerror == ERROR_DIRECTORY
 
2192
                    e_winerror = getattr(e, 'winerror', None)
 
2193
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2194
                    # there may be directories in the inventory even though
 
2195
                    # this path is not a file on disk: so mark it as end of
 
2196
                    # iterator
 
2197
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2198
                        current_dir_info = None
 
2199
                    elif (sys.platform == 'win32'
 
2200
                          and (e.errno in win_errors
 
2201
                               or e_winerror in win_errors)):
 
2202
                        current_dir_info = None
 
2203
                    else:
 
2204
                        raise
 
2205
                else:
 
2206
                    if current_dir_info[0][0] == '':
 
2207
                        # remove .bzr from iteration
 
2208
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2209
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2210
                        del current_dir_info[1][bzr_index]
 
2211
            # walk until both the directory listing and the versioned metadata
 
2212
            # are exhausted. 
 
2213
            if (block_index < len(state._dirblocks) and
 
2214
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2215
                current_block = state._dirblocks[block_index]
 
2216
            else:
 
2217
                current_block = None
 
2218
            while (current_dir_info is not None or
 
2219
                   current_block is not None):
 
2220
                if (current_dir_info and current_block
 
2221
                    and current_dir_info[0][0] != current_block[0]):
 
2222
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2223
                        # filesystem data refers to paths not covered by the dirblock.
 
2224
                        # this has two possibilities:
 
2225
                        # A) it is versioned but empty, so there is no block for it
 
2226
                        # B) it is not versioned.
 
2227
 
 
2228
                        # if (A) then we need to recurse into it to check for
 
2229
                        # new unknown files or directories.
 
2230
                        # if (B) then we should ignore it, because we don't
 
2231
                        # recurse into unknown directories.
 
2232
                        path_index = 0
 
2233
                        while path_index < len(current_dir_info[1]):
 
2234
                                current_path_info = current_dir_info[1][path_index]
 
2235
                                if want_unversioned:
 
2236
                                    if current_path_info[2] == 'directory':
 
2237
                                        if self.target._directory_is_tree_reference(
 
2238
                                            current_path_info[0].decode('utf8')):
 
2239
                                            current_path_info = current_path_info[:2] + \
 
2240
                                                ('tree-reference',) + current_path_info[3:]
 
2241
                                    new_executable = bool(
 
2242
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2243
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2244
                                    yield (None,
 
2245
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2246
                                        True,
 
2247
                                        (False, False),
 
2248
                                        (None, None),
 
2249
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2250
                                        (None, current_path_info[2]),
 
2251
                                        (None, new_executable))
 
2252
                                # dont descend into this unversioned path if it is
 
2253
                                # a dir
 
2254
                                if current_path_info[2] in ('directory',
 
2255
                                                            'tree-reference'):
 
2256
                                    del current_dir_info[1][path_index]
 
2257
                                    path_index -= 1
 
2258
                                path_index += 1
 
2259
 
 
2260
                        # This dir info has been handled, go to the next
 
2261
                        try:
 
2262
                            current_dir_info = dir_iterator.next()
 
2263
                        except StopIteration:
 
2264
                            current_dir_info = None
 
2265
                    else:
 
2266
                        # We have a dirblock entry for this location, but there
 
2267
                        # is no filesystem path for this. This is most likely
 
2268
                        # because a directory was removed from the disk.
 
2269
                        # We don't have to report the missing directory,
 
2270
                        # because that should have already been handled, but we
 
2271
                        # need to handle all of the files that are contained
 
2272
                        # within.
 
2273
                        for current_entry in current_block[1]:
 
2274
                            # entry referring to file not present on disk.
 
2275
                            # advance the entry only, after processing.
 
2276
                            result = _process_entry(current_entry, None)
 
2277
                            if result is not None:
 
2278
                                if result is not uninteresting:
 
2279
                                    yield result
 
2280
                        block_index +=1
 
2281
                        if (block_index < len(state._dirblocks) and
 
2282
                            osutils.is_inside(current_root,
 
2283
                                              state._dirblocks[block_index][0])):
 
2284
                            current_block = state._dirblocks[block_index]
 
2285
                        else:
 
2286
                            current_block = None
 
2287
                    continue
 
2288
                entry_index = 0
 
2289
                if current_block and entry_index < len(current_block[1]):
 
2290
                    current_entry = current_block[1][entry_index]
 
2291
                else:
 
2292
                    current_entry = None
 
2293
                advance_entry = True
 
2294
                path_index = 0
 
2295
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2296
                    current_path_info = current_dir_info[1][path_index]
 
2297
                    if current_path_info[2] == 'directory':
 
2298
                        if self.target._directory_is_tree_reference(
 
2299
                            current_path_info[0].decode('utf8')):
 
2300
                            current_path_info = current_path_info[:2] + \
 
2301
                                ('tree-reference',) + current_path_info[3:]
 
2302
                else:
 
2303
                    current_path_info = None
 
2304
                advance_path = True
 
2305
                path_handled = False
 
2306
                while (current_entry is not None or
 
2307
                    current_path_info is not None):
 
2308
                    if current_entry is None:
 
2309
                        # the check for path_handled when the path is adnvaced
 
2310
                        # will yield this path if needed.
 
2311
                        pass
 
2312
                    elif current_path_info is None:
 
2313
                        # no path is fine: the per entry code will handle it.
 
2314
                        result = _process_entry(current_entry, current_path_info)
 
2315
                        if result is not None:
 
2316
                            if result is not uninteresting:
 
2317
                                yield result
 
2318
                    elif (current_entry[0][1] != current_path_info[1]
 
2319
                          or current_entry[1][target_index][0] in 'ar'):
 
2320
                        # The current path on disk doesn't match the dirblock
 
2321
                        # record. Either the dirblock is marked as absent, or
 
2322
                        # the file on disk is not present at all in the
 
2323
                        # dirblock. Either way, report about the dirblock
 
2324
                        # entry, and let other code handle the filesystem one.
 
2325
 
 
2326
                        # Compare the basename for these files to determine
 
2327
                        # which comes first
 
2328
                        if current_path_info[1] < current_entry[0][1]:
 
2329
                            # extra file on disk: pass for now, but only
 
2330
                            # increment the path, not the entry
 
2331
                            advance_entry = False
 
2332
                        else:
 
2333
                            # entry referring to file not present on disk.
 
2334
                            # advance the entry only, after processing.
 
2335
                            result = _process_entry(current_entry, None)
 
2336
                            if result is not None:
 
2337
                                if result is not uninteresting:
 
2338
                                    yield result
 
2339
                            advance_path = False
 
2340
                    else:
 
2341
                        result = _process_entry(current_entry, current_path_info)
 
2342
                        if result is not None:
 
2343
                            path_handled = True
 
2344
                            if result is not uninteresting:
 
2345
                                yield result
 
2346
                    if advance_entry and current_entry is not None:
 
2347
                        entry_index += 1
 
2348
                        if entry_index < len(current_block[1]):
 
2349
                            current_entry = current_block[1][entry_index]
 
2350
                        else:
 
2351
                            current_entry = None
 
2352
                    else:
 
2353
                        advance_entry = True # reset the advance flaga
 
2354
                    if advance_path and current_path_info is not None:
 
2355
                        if not path_handled:
 
2356
                            # unversioned in all regards
 
2357
                            if want_unversioned:
 
2358
                                new_executable = bool(
 
2359
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2360
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2361
                                yield (None,
 
2362
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2363
                                    True,
 
2364
                                    (False, False),
 
2365
                                    (None, None),
 
2366
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2367
                                    (None, current_path_info[2]),
 
2368
                                    (None, new_executable))
 
2369
                            # dont descend into this unversioned path if it is
 
2370
                            # a dir
 
2371
                            if current_path_info[2] in ('directory'):
 
2372
                                del current_dir_info[1][path_index]
 
2373
                                path_index -= 1
 
2374
                        # dont descend the disk iterator into any tree 
 
2375
                        # paths.
 
2376
                        if current_path_info[2] == 'tree-reference':
 
2377
                            del current_dir_info[1][path_index]
 
2378
                            path_index -= 1
 
2379
                        path_index += 1
 
2380
                        if path_index < len(current_dir_info[1]):
 
2381
                            current_path_info = current_dir_info[1][path_index]
 
2382
                            if current_path_info[2] == 'directory':
 
2383
                                if self.target._directory_is_tree_reference(
 
2384
                                    current_path_info[0].decode('utf8')):
 
2385
                                    current_path_info = current_path_info[:2] + \
 
2386
                                        ('tree-reference',) + current_path_info[3:]
 
2387
                        else:
 
2388
                            current_path_info = None
 
2389
                        path_handled = False
 
2390
                    else:
 
2391
                        advance_path = True # reset the advance flagg.
 
2392
                if current_block is not None:
 
2393
                    block_index += 1
 
2394
                    if (block_index < len(state._dirblocks) and
 
2395
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2396
                        current_block = state._dirblocks[block_index]
 
2397
                    else:
 
2398
                        current_block = None
 
2399
                if current_dir_info is not None:
 
2400
                    try:
 
2401
                        current_dir_info = dir_iterator.next()
 
2402
                    except StopIteration:
 
2403
                        current_dir_info = None
 
2404
 
2094
2405
 
2095
2406
    @staticmethod
2096
2407
    def is_compatible(source, target):
2097
2408
        # the target must be a dirstate working tree
2098
 
        if not isinstance(target, DirStateWorkingTree):
 
2409
        if not isinstance(target, WorkingTree4):
2099
2410
            return False
2100
 
        # the source must be a revtree or dirstate rev tree.
 
2411
        # the source must be a revtreee or dirstate rev tree.
2101
2412
        if not isinstance(source,
2102
2413
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2103
2414
            return False
2104
2415
        # the source revid must be in the target dirstate
2105
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2416
        if not (source._revision_id == NULL_REVISION or
2106
2417
            source._revision_id in target.get_parent_ids()):
2107
 
            # TODO: what about ghosts? it may well need to
 
2418
            # TODO: what about ghosts? it may well need to 
2108
2419
            # check for them explicitly.
2109
2420
            return False
2110
2421
        return True
2120
2431
 
2121
2432
    def convert(self, tree):
2122
2433
        # 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
 
2434
        # on-unlock behaviours, and so that noone else diddles with the 
2124
2435
        # tree during upgrade.
2125
2436
        tree._control_files.lock_write()
2126
2437
        try:
2152
2463
 
2153
2464
    def update_format(self, tree):
2154
2465
        """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())
 
2466
        tree._control_files.put_utf8('format',
 
2467
            self.target_format.get_format_string())