~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Vincent Ladeuil
  • Date: 2007-07-18 09:43:41 UTC
  • mto: (2778.5.1 vila)
  • mto: This revision was merged to the branch mainline in revision 2789.
  • Revision ID: v.ladeuil+lp@free.fr-20070718094341-edmgsog3el06yqow
Add performance analysis of missing.

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,
38
 
    debug,
 
45
    conflicts as _mod_conflicts,
 
46
    delta,
39
47
    dirstate,
40
48
    errors,
41
49
    generate_ids,
 
50
    globbing,
 
51
    hashcache,
 
52
    ignores,
 
53
    merge,
42
54
    osutils,
43
55
    revision as _mod_revision,
44
56
    revisiontree,
45
 
    trace,
 
57
    textui,
46
58
    transform,
47
 
    views,
 
59
    urlutils,
 
60
    xml5,
 
61
    xml6,
48
62
    )
49
63
import bzrlib.branch
 
64
from bzrlib.transport import get_transport
50
65
import bzrlib.ui
51
66
""")
52
67
 
 
68
from bzrlib import symbol_versioning
53
69
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
 
70
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
71
from bzrlib.lockable_files import LockableFiles, TransportLock
 
72
from bzrlib.lockdir import LockDir
 
73
import bzrlib.mutabletree
56
74
from bzrlib.mutabletree import needs_tree_write_lock
57
75
from bzrlib.osutils import (
58
76
    file_kind,
59
77
    isdir,
 
78
    normpath,
60
79
    pathjoin,
 
80
    rand_chars,
61
81
    realpath,
62
82
    safe_unicode,
 
83
    splitpath,
63
84
    )
64
 
from bzrlib.trace import mutter
 
85
from bzrlib.trace import mutter, note
65
86
from bzrlib.transport.local import LocalTransport
66
87
from bzrlib.tree import InterTree
 
88
from bzrlib.progress import DummyProgress, ProgressPhase
 
89
from bzrlib.revision import NULL_REVISION, CURRENT_REVISION
 
90
from bzrlib.rio import RioReader, rio_file, Stanza
 
91
from bzrlib.symbol_versioning import (deprecated_passed,
 
92
        deprecated_method,
 
93
        deprecated_function,
 
94
        DEPRECATED_PARAMETER,
 
95
        )
67
96
from bzrlib.tree import Tree
68
97
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
69
98
 
70
99
 
71
 
class DirStateWorkingTree(WorkingTree3):
 
100
# This is the Windows equivalent of ENOTDIR
 
101
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
102
# just an error code.
 
103
ERROR_PATH_NOT_FOUND = 3
 
104
ERROR_DIRECTORY = 267
 
105
 
 
106
 
 
107
class WorkingTree4(WorkingTree3):
 
108
    """This is the Format 4 working tree.
 
109
 
 
110
    This differs from WorkingTree3 by:
 
111
     - Having a consolidated internal dirstate, stored in a
 
112
       randomly-accessible sorted file on disk.
 
113
     - Not having a regular inventory attribute.  One can be synthesized 
 
114
       on demand but this is expensive and should be avoided.
 
115
 
 
116
    This is new in bzr 0.15.
 
117
    """
 
118
 
72
119
    def __init__(self, basedir,
73
120
                 branch,
74
121
                 _control_files=None,
83
130
        """
84
131
        self._format = _format
85
132
        self.bzrdir = _bzrdir
 
133
        from bzrlib.trace import note, mutter
 
134
        assert isinstance(basedir, basestring), \
 
135
            "base directory %r is not a string" % basedir
86
136
        basedir = safe_unicode(basedir)
87
137
        mutter("opening working tree %r", basedir)
88
138
        self._branch = branch
 
139
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
140
            "branch %r is not a Branch" % self.branch
89
141
        self.basedir = realpath(basedir)
90
142
        # if branch is at our basedir and is a format 6 or less
91
143
        # assume all other formats have their own control files.
 
144
        assert isinstance(_control_files, LockableFiles), \
 
145
            "_control_files must be a LockableFiles, not %r" % _control_files
92
146
        self._control_files = _control_files
93
 
        self._transport = self._control_files._transport
94
147
        self._dirty = None
95
148
        #-------------
96
149
        # during a read or write lock these objects are set, and are
98
151
        self._dirstate = None
99
152
        self._inventory = None
100
153
        #-------------
101
 
        self._setup_directory_is_tree_reference()
102
 
        self._detect_case_handling()
103
 
        self._rules_searcher = None
104
 
        self.views = self._make_views()
105
 
        #--- allow tests to select the dirstate iter_changes implementation
106
 
        self._iter_changes = dirstate._process_entry
107
154
 
108
155
    @needs_tree_write_lock
109
156
    def _add(self, files, ids, kinds):
111
158
        state = self.current_dirstate()
112
159
        for f, file_id, kind in zip(files, ids, kinds):
113
160
            f = f.strip('/')
 
161
            assert '//' not in f
 
162
            assert '..' not in f
114
163
            if self.path2id(f):
115
164
                # special case tree root handling.
116
165
                if f == '' and self.path2id(f) == ROOT_ID:
137
186
    @needs_tree_write_lock
138
187
    def add_reference(self, sub_tree):
139
188
        # use standard implementation, which calls back to self._add
140
 
        #
 
189
        # 
141
190
        # So we don't store the reference_revision in the working dirstate,
142
 
        # it's just recorded at the moment of commit.
 
191
        # it's just recorded at the moment of commit. 
143
192
        self._add_reference(sub_tree)
144
193
 
145
194
    def break_lock(self):
184
233
            WorkingTree3._comparison_data(self, entry, path)
185
234
        # it looks like a plain directory, but it's really a reference -- see
186
235
        # also kind()
187
 
        if (self._repo_supports_tree_reference and kind == 'directory'
188
 
            and entry is not None and entry.kind == 'tree-reference'):
 
236
        if (self._repo_supports_tree_reference and
 
237
            kind == 'directory' and
 
238
            self._directory_is_tree_reference(path)):
189
239
            kind = 'tree-reference'
190
240
        return kind, executable, stat_value
191
241
 
217
267
            return self._dirstate
218
268
        local_path = self.bzrdir.get_workingtree_transport(None
219
269
            ).local_abspath('dirstate')
220
 
        self._dirstate = dirstate.DirState.on_file(local_path,
221
 
            self._sha1_provider())
 
270
        self._dirstate = dirstate.DirState.on_file(local_path)
222
271
        return self._dirstate
223
272
 
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
 
273
    def _directory_is_tree_reference(self, relpath):
 
274
        # as a special case, if a directory contains control files then 
 
275
        # it's a tree reference, except that the root of the tree is not
 
276
        return relpath and osutils.isdir(self.abspath(relpath) + u"/.bzr")
 
277
        # TODO: We could ask all the control formats whether they
 
278
        # recognize this directory, but at the moment there's no cheap api
 
279
        # to do that.  Since we probably can only nest bzr checkouts and
 
280
        # they always use this name it's ok for now.  -- mbp 20060306
 
281
        #
 
282
        # FIXME: There is an unhandled case here of a subdirectory
 
283
        # containing .bzr but not a branch; that will probably blow up
 
284
        # when you try to commit it.  It might happen if there is a
 
285
        # checkout in a subdirectory.  This can be avoided by not adding
 
286
        # it.  mbp 20070306
235
287
 
236
288
    def filter_unversioned_files(self, paths):
237
289
        """Filter out paths that are versioned.
270
322
 
271
323
    def _generate_inventory(self):
272
324
        """Create and set self.inventory from the dirstate object.
273
 
 
 
325
        
274
326
        This is relatively expensive: we have to walk the entire dirstate.
275
327
        Ideally we would not, and can deprecate this function.
276
328
        """
280
332
        state._read_dirblocks_if_needed()
281
333
        root_key, current_entry = self._get_entry(path='')
282
334
        current_id = root_key[2]
283
 
        if not (current_entry[0][0] == 'd'): # directory
284
 
            raise AssertionError(current_entry)
 
335
        assert current_entry[0][0] == 'd' # directory
285
336
        inv = Inventory(root_id=current_id)
286
337
        # Turn some things into local variables
287
338
        minikind_to_kind = dirstate.DirState._minikind_to_kind
320
371
                    # add this entry to the parent map.
321
372
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
322
373
                elif kind == 'tree-reference':
323
 
                    if not self._repo_supports_tree_reference:
324
 
                        raise errors.UnsupportedOperation(
325
 
                            self._generate_inventory,
326
 
                            self.branch.repository)
 
374
                    assert self._repo_supports_tree_reference, \
 
375
                        "repository of %r " \
 
376
                        "doesn't support tree references " \
 
377
                        "required by entry %r" \
 
378
                        % (self, name)
327
379
                    inv_entry.reference_revision = link_or_sha1 or None
328
380
                elif kind != 'symlink':
329
381
                    raise AssertionError("unknown kind %r" % kind)
330
382
                # 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,))
 
383
                assert file_id not in inv_byid, ('file_id %s already in'
 
384
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
385
                assert name_unicode not in parent_ie.children
337
386
                inv_byid[file_id] = inv_entry
338
387
                parent_ie.children[name_unicode] = inv_entry
339
388
        self._inventory = inv
344
393
        If either file_id or path is supplied, it is used as the key to lookup.
345
394
        If both are supplied, the fastest lookup is used, and an error is
346
395
        raised if they do not both point at the same row.
347
 
 
 
396
        
348
397
        :param file_id: An optional unicode file_id to be looked up.
349
398
        :param path: An optional unicode path to be looked up.
350
399
        :return: The dirstate row tuple for path/file_id, or (None, None)
359
408
    def get_file_sha1(self, file_id, path=None, stat_value=None):
360
409
        # check file id is valid unconditionally.
361
410
        entry = self._get_entry(file_id=file_id, path=path)
362
 
        if entry[0] is None:
363
 
            raise errors.NoSuchId(self, file_id)
 
411
        assert entry[0] is not None, 'what error should this raise'
364
412
        if path is None:
365
413
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
366
414
 
374
422
                    return None
375
423
                else:
376
424
                    raise
377
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
378
 
            stat_value=stat_value)
 
425
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
426
                                          stat_value=stat_value)
379
427
        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
 
428
            return link_or_sha1
390
429
        return None
391
430
 
392
431
    def _get_inventory(self):
393
432
        """Get the inventory for the tree. This is only valid within a lock."""
394
 
        if 'evil' in debug.debug_flags:
395
 
            trace.mutter_callsite(2,
396
 
                "accessing .inventory forces a size of tree translation.")
397
433
        if self._inventory is not None:
398
434
            return self._inventory
399
435
        self._must_be_locked()
406
442
    @needs_read_lock
407
443
    def get_parent_ids(self):
408
444
        """See Tree.get_parent_ids.
409
 
 
 
445
        
410
446
        This implementation requests the ids list from the dirstate file.
411
447
        """
412
448
        return self.current_dirstate().get_parent_ids()
428
464
 
429
465
    def has_id(self, file_id):
430
466
        state = self.current_dirstate()
 
467
        file_id = osutils.safe_file_id(file_id)
431
468
        row, parents = self._get_entry(file_id=file_id)
432
469
        if row is None:
433
470
            return False
434
471
        return osutils.lexists(pathjoin(
435
472
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
436
473
 
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
474
    @needs_read_lock
443
475
    def id2path(self, file_id):
444
476
        "Convert a file-id to a path."
 
477
        file_id = osutils.safe_file_id(file_id)
445
478
        state = self.current_dirstate()
446
479
        entry = self._get_entry(file_id=file_id)
447
480
        if entry == (None, None):
449
482
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
450
483
        return path_utf8.decode('utf8')
451
484
 
452
 
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
453
 
        entry = self._get_entry(path=path)
454
 
        if entry == (None, None):
455
 
            return False # Missing entries are not executable
456
 
        return entry[1][0][3] # Executable?
457
 
 
458
485
    if not osutils.supports_executable():
459
486
        def is_executable(self, file_id, path=None):
460
487
            """Test if a file is executable or not.
461
488
 
462
489
            Note: The caller is expected to take a read-lock before calling this.
463
490
            """
 
491
            file_id = osutils.safe_file_id(file_id)
464
492
            entry = self._get_entry(file_id=file_id, path=path)
465
493
            if entry == (None, None):
466
494
                return False
467
495
            return entry[1][0][3]
468
 
 
469
 
        _is_executable_from_path_and_stat = \
470
 
            _is_executable_from_path_and_stat_from_basis
471
496
    else:
472
497
        def is_executable(self, file_id, path=None):
473
498
            """Test if a file is executable or not.
474
499
 
475
500
            Note: The caller is expected to take a read-lock before calling this.
476
501
            """
477
 
            self._must_be_locked()
478
502
            if not path:
 
503
                file_id = osutils.safe_file_id(file_id)
479
504
                path = self.id2path(file_id)
480
505
            mode = os.lstat(self.abspath(path)).st_mode
481
506
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
482
507
 
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
508
    @needs_read_lock
494
509
    def __iter__(self):
495
510
        """Iterate through file_ids for this tree.
508
523
        return iter(result)
509
524
 
510
525
    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
526
        for key, tree_details in self.current_dirstate()._iter_entries():
516
527
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
517
528
                # not relevant to the working tree
519
530
            if not key[1]:
520
531
                # the root is not a reference.
521
532
                continue
522
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
533
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
523
534
            try:
524
 
                if self._kind(relpath) == 'tree-reference':
525
 
                    yield relpath, key[2]
 
535
                if self._kind(path) == 'tree-reference':
 
536
                    yield path, key[2]
526
537
            except errors.NoSuchFile:
527
538
                # path is missing on disk.
528
539
                continue
529
540
 
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
541
    def kind(self, file_id):
537
542
        """Return the kind of a file.
538
543
 
542
547
        Note: The caller is expected to take a read-lock before calling this.
543
548
        """
544
549
        relpath = self.id2path(file_id)
545
 
        if relpath is None:
546
 
            raise AssertionError(
547
 
                "path for id {%s} is None!" % file_id)
 
550
        assert relpath != None, \
 
551
            "path for id {%s} is None!" % file_id
548
552
        return self._kind(relpath)
549
553
 
550
554
    def _kind(self, relpath):
551
555
        abspath = self.abspath(relpath)
552
556
        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'
 
557
        if (self._repo_supports_tree_reference and
 
558
            kind == 'directory' and
 
559
            self._directory_is_tree_reference(relpath)):
 
560
            kind = 'tree-reference'
558
561
        return kind
559
562
 
560
563
    @needs_read_lock
564
567
        if parent_ids:
565
568
            return parent_ids[0]
566
569
        else:
567
 
            return _mod_revision.NULL_REVISION
 
570
            return None
568
571
 
569
572
    def lock_read(self):
570
 
        """See Branch.lock_read, and WorkingTree.unlock.
571
 
 
572
 
        :return: An object with an unlock method which will release the lock
573
 
            obtained.
574
 
        """
 
573
        """See Branch.lock_read, and WorkingTree.unlock."""
575
574
        self.branch.lock_read()
576
575
        try:
577
576
            self._control_files.lock_read()
590
589
        except:
591
590
            self.branch.unlock()
592
591
            raise
593
 
        return self
594
592
 
595
593
    def _lock_self_write(self):
596
594
        """This should be called after the branch is locked."""
611
609
        except:
612
610
            self.branch.unlock()
613
611
            raise
614
 
        return self
615
612
 
616
613
    def lock_tree_write(self):
617
 
        """See MutableTree.lock_tree_write, and WorkingTree.unlock.
618
 
 
619
 
        :return: An object with an unlock method which will release the lock
620
 
            obtained.
621
 
        """
 
614
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
622
615
        self.branch.lock_read()
623
 
        return self._lock_self_write()
 
616
        self._lock_self_write()
624
617
 
625
618
    def lock_write(self):
626
 
        """See MutableTree.lock_write, and WorkingTree.unlock.
627
 
 
628
 
        :return: An object with an unlock method which will release the lock
629
 
            obtained.
630
 
        """
 
619
        """See MutableTree.lock_write, and WorkingTree.unlock."""
631
620
        self.branch.lock_write()
632
 
        return self._lock_self_write()
 
621
        self._lock_self_write()
633
622
 
634
623
    @needs_tree_write_lock
635
624
    def move(self, from_paths, to_dir, after=False):
637
626
        result = []
638
627
        if not from_paths:
639
628
            return result
 
629
 
640
630
        state = self.current_dirstate()
641
 
        if isinstance(from_paths, basestring):
642
 
            raise ValueError()
 
631
 
 
632
        assert not isinstance(from_paths, basestring)
643
633
        to_dir_utf8 = to_dir.encode('utf8')
644
634
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
645
635
        id_index = state._get_id_index()
667
657
        if self._inventory is not None:
668
658
            update_inventory = True
669
659
            inv = self.inventory
 
660
            to_dir_ie = inv[to_dir_id]
670
661
            to_dir_id = to_entry[0][2]
671
 
            to_dir_ie = inv[to_dir_id]
672
662
        else:
673
663
            update_inventory = False
674
664
 
705
695
            from_entry = self._get_entry(path=from_rel)
706
696
            if from_entry == (None, None):
707
697
                raise errors.BzrMoveFailedError(from_rel,to_dir,
708
 
                    errors.NotVersionedError(path=from_rel))
 
698
                    errors.NotVersionedError(path=str(from_rel)))
709
699
 
710
700
            from_id = from_entry[0][2]
711
701
            to_rel = pathjoin(to_dir, from_tail)
738
728
                if from_missing: # implicitly just update our path mapping
739
729
                    move_file = False
740
730
                elif not after:
741
 
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
731
                    raise errors.RenameFailedFilesExist(from_rel, to_rel,
 
732
                        extra="(Use --after to update the Bazaar id)")
742
733
 
743
734
            rollbacks = []
744
735
            def rollback_rename():
800
791
                if minikind == 'd':
801
792
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
802
793
                        """Recursively update all entries in this dirblock."""
803
 
                        if from_dir == '':
804
 
                            raise AssertionError("renaming root not supported")
 
794
                        assert from_dir != '', "renaming root not supported"
805
795
                        from_key = (from_dir, '')
806
796
                        from_block_idx, present = \
807
797
                            state._find_block_index_from_key(from_key)
820
810
 
821
811
                        # Grab a copy since move_one may update the list.
822
812
                        for entry in from_block[1][:]:
823
 
                            if not (entry[0][0] == from_dir):
824
 
                                raise AssertionError()
 
813
                            assert entry[0][0] == from_dir
825
814
                            cur_details = entry[1][0]
826
815
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
827
816
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
886
875
        for tree in trees:
887
876
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
888
877
                parents):
889
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
890
 
                    trees, require_versioned)
 
878
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
891
879
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
892
880
        # -- make all paths utf8 --
893
881
        paths_utf8 = set()
953
941
            if not all_versioned:
954
942
                raise errors.PathsNotVersionedError(paths)
955
943
        # -- remove redundancy in supplied paths to prevent over-scanning --
956
 
        search_paths = osutils.minimum_path_selection(paths)
957
 
        # sketch:
 
944
        search_paths = set()
 
945
        for path in paths:
 
946
            other_paths = paths.difference(set([path]))
 
947
            if not osutils.is_inside_any(other_paths, path):
 
948
                # this is a top level path, we must check it.
 
949
                search_paths.add(path)
 
950
        # sketch: 
958
951
        # for all search_indexs in each path at or under each element of
959
952
        # search_paths, if the detail is relocated: add the id, and add the
960
953
        # relocated path as one to search if its not searched already. If the
1016
1009
 
1017
1010
    def read_working_inventory(self):
1018
1011
        """Read the working inventory.
1019
 
 
 
1012
        
1020
1013
        This is a meaningless operation for dirstate, but we obey it anyhow.
1021
1014
        """
1022
1015
        return self.inventory
1027
1020
 
1028
1021
        WorkingTree4 supplies revision_trees for any basis tree.
1029
1022
        """
 
1023
        revision_id = osutils.safe_revision_id(revision_id)
1030
1024
        dirstate = self.current_dirstate()
1031
1025
        parent_ids = dirstate.get_parent_ids()
1032
1026
        if revision_id not in parent_ids:
1039
1033
    @needs_tree_write_lock
1040
1034
    def set_last_revision(self, new_revision):
1041
1035
        """Change the last revision in the working tree."""
 
1036
        new_revision = osutils.safe_revision_id(new_revision)
1042
1037
        parents = self.get_parent_ids()
1043
 
        if new_revision in (_mod_revision.NULL_REVISION, None):
1044
 
            if len(parents) >= 2:
1045
 
                raise AssertionError(
1046
 
                    "setting the last parent to none with a pending merge is "
1047
 
                    "unsupported.")
 
1038
        if new_revision in (NULL_REVISION, None):
 
1039
            assert len(parents) < 2, (
 
1040
                "setting the last parent to none with a pending merge is "
 
1041
                "unsupported.")
1048
1042
            self.set_parent_ids([])
1049
1043
        else:
1050
1044
            self.set_parent_ids([new_revision] + parents[1:],
1053
1047
    @needs_tree_write_lock
1054
1048
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1055
1049
        """Set the parent ids to revision_ids.
1056
 
 
 
1050
        
1057
1051
        See also set_parent_trees. This api will try to retrieve the tree data
1058
1052
        for each element of revision_ids from the trees repository. If you have
1059
1053
        tree data already available, it is more efficient to use
1063
1057
        :param revision_ids: The revision_ids to set as the parent ids of this
1064
1058
            working tree. Any of these may be ghosts.
1065
1059
        """
 
1060
        revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1066
1061
        trees = []
1067
1062
        for revision_id in revision_ids:
1068
1063
            try:
1091
1086
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1092
1087
        real_trees = []
1093
1088
        ghosts = []
1094
 
 
1095
 
        parent_ids = [rev_id for rev_id, tree in parents_list]
1096
 
        graph = self.branch.repository.get_graph()
1097
 
        heads = graph.heads(parent_ids)
1098
 
        accepted_revisions = set()
1099
 
 
1100
1089
        # convert absent trees to the null tree, which we convert back to
1101
1090
        # missing on access.
1102
1091
        for rev_id, tree in parents_list:
1103
 
            if len(accepted_revisions) > 0:
1104
 
                # we always accept the first tree
1105
 
                if rev_id in accepted_revisions or rev_id not in heads:
1106
 
                    # We have already included either this tree, or its
1107
 
                    # descendent, so we skip it.
1108
 
                    continue
 
1092
            rev_id = osutils.safe_revision_id(rev_id)
1109
1093
            _mod_revision.check_not_reserved_id(rev_id)
1110
1094
            if tree is not None:
1111
1095
                real_trees.append((rev_id, tree))
1112
1096
            else:
1113
1097
                real_trees.append((rev_id,
1114
 
                    self.branch.repository.revision_tree(
1115
 
                        _mod_revision.NULL_REVISION)))
 
1098
                    self.branch.repository.revision_tree(None)))
1116
1099
                ghosts.append(rev_id)
1117
 
            accepted_revisions.add(rev_id)
1118
1100
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
1119
1101
        self._make_dirty(reset_inventory=False)
1120
1102
 
1125
1107
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
1126
1108
            self._make_dirty(reset_inventory=True)
1127
1109
 
1128
 
    def _sha_from_stat(self, path, stat_result):
1129
 
        """Get a sha digest from the tree's stat cache.
1130
 
 
1131
 
        The default implementation assumes no stat cache is present.
1132
 
 
1133
 
        :param path: The path.
1134
 
        :param stat_result: The stat result being looked up.
1135
 
        """
1136
 
        return self.current_dirstate().sha1_from_stat(path, stat_result)
1137
 
 
1138
1110
    @needs_read_lock
1139
1111
    def supports_tree_reference(self):
1140
1112
        return self._repo_supports_tree_reference
1141
1113
 
1142
1114
    def unlock(self):
1143
1115
        """Unlock in format 4 trees needs to write the entire dirstate."""
1144
 
        # do non-implementation specific cleanup
1145
 
        self._cleanup()
1146
 
 
1147
1116
        if self._control_files._lock_count == 1:
1148
1117
            # eventually we should do signature checking during read locks for
1149
1118
            # dirstate updates.
1180
1149
            return
1181
1150
        state = self.current_dirstate()
1182
1151
        state._read_dirblocks_if_needed()
1183
 
        ids_to_unversion = set(file_ids)
 
1152
        ids_to_unversion = set()
 
1153
        for file_id in file_ids:
 
1154
            ids_to_unversion.add(osutils.safe_file_id(file_id))
1184
1155
        paths_to_unversion = set()
1185
1156
        # sketch:
1186
1157
        # check if the root is to be unversioned, if so, assert for now.
1213
1184
                # just forget the whole block.
1214
1185
                entry_index = 0
1215
1186
                while entry_index < len(block[1]):
 
1187
                    # Mark this file id as having been removed
1216
1188
                    entry = block[1][entry_index]
1217
 
                    if entry[1][0][0] in 'ar':
1218
 
                        # don't remove absent or renamed entries
 
1189
                    ids_to_unversion.discard(entry[0][2])
 
1190
                    if (entry[1][0][0] == 'a'
 
1191
                        or not state._make_absent(entry)):
1219
1192
                        entry_index += 1
1220
 
                    else:
1221
 
                        # Mark this file id as having been removed
1222
 
                        ids_to_unversion.discard(entry[0][2])
1223
 
                        if not state._make_absent(entry):
1224
 
                            # The block has not shrunk.
1225
 
                            entry_index += 1
1226
1193
                # go to the next block. (At the moment we dont delete empty
1227
1194
                # dirblocks)
1228
1195
                block_index += 1
1251
1218
            for file_id in file_ids:
1252
1219
                self._inventory.remove_recursive_id(file_id)
1253
1220
 
1254
 
    @needs_tree_write_lock
1255
 
    def rename_one(self, from_rel, to_rel, after=False):
1256
 
        """See WorkingTree.rename_one"""
1257
 
        self.flush()
1258
 
        WorkingTree.rename_one(self, from_rel, to_rel, after)
1259
 
 
1260
 
    @needs_tree_write_lock
1261
 
    def apply_inventory_delta(self, changes):
1262
 
        """See MutableTree.apply_inventory_delta"""
1263
 
        state = self.current_dirstate()
1264
 
        state.update_by_delta(changes)
1265
 
        self._make_dirty(reset_inventory=True)
1266
 
 
1267
 
    def update_basis_by_delta(self, new_revid, delta):
1268
 
        """See MutableTree.update_basis_by_delta."""
1269
 
        if self.last_revision() == new_revid:
1270
 
            raise AssertionError()
1271
 
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
1272
 
 
1273
1221
    @needs_read_lock
1274
1222
    def _validate(self):
1275
1223
        self._dirstate._validate()
1277
1225
    @needs_tree_write_lock
1278
1226
    def _write_inventory(self, inv):
1279
1227
        """Write inventory as the current inventory."""
1280
 
        if self._dirty:
1281
 
            raise AssertionError("attempting to write an inventory when the "
1282
 
                "dirstate is dirty will lose pending changes")
1283
 
        had_inventory = self._inventory is not None
1284
 
        # Setting self._inventory = None forces the dirstate to regenerate the
1285
 
        # working inventory. We do this because self.inventory may be inv, or
1286
 
        # may have been modified, and either case would prevent a clean delta
1287
 
        # being created.
1288
 
        self._inventory = None
1289
 
        # generate a delta,
1290
 
        delta = inv._make_delta(self.inventory)
1291
 
        # and apply it.
1292
 
        self.apply_inventory_delta(delta)
1293
 
        if had_inventory:
 
1228
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1229
        self.current_dirstate().set_state_from_inventory(inv)
 
1230
        self._make_dirty(reset_inventory=False)
 
1231
        if self._inventory is not None:
1294
1232
            self._inventory = inv
1295
1233
        self.flush()
1296
1234
 
1297
1235
 
1298
 
class ContentFilterAwareSHA1Provider(dirstate.SHA1Provider):
1299
 
 
1300
 
    def __init__(self, tree):
1301
 
        self.tree = tree
1302
 
 
1303
 
    def sha1(self, abspath):
1304
 
        """See dirstate.SHA1Provider.sha1()."""
1305
 
        filters = self.tree._content_filter_stack(
1306
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1307
 
        return internal_size_sha_file_byname(abspath, filters)[1]
1308
 
 
1309
 
    def stat_and_sha1(self, abspath):
1310
 
        """See dirstate.SHA1Provider.stat_and_sha1()."""
1311
 
        filters = self.tree._content_filter_stack(
1312
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1313
 
        file_obj = file(abspath, 'rb', 65000)
1314
 
        try:
1315
 
            statvalue = os.fstat(file_obj.fileno())
1316
 
            if filters:
1317
 
                file_obj = filtered_input_file(file_obj, filters)
1318
 
            sha1 = osutils.size_sha_file(file_obj)[1]
1319
 
        finally:
1320
 
            file_obj.close()
1321
 
        return statvalue, sha1
1322
 
 
1323
 
 
1324
 
class ContentFilteringDirStateWorkingTree(DirStateWorkingTree):
1325
 
    """Dirstate working tree that supports content filtering.
1326
 
 
1327
 
    The dirstate holds the hash and size of the canonical form of the file, 
1328
 
    and most methods must return that.
1329
 
    """
1330
 
 
1331
 
    def _file_content_summary(self, path, stat_result):
1332
 
        # This is to support the somewhat obsolete path_content_summary method
1333
 
        # with content filtering: see
1334
 
        # <https://bugs.edge.launchpad.net/bzr/+bug/415508>.
1335
 
        #
1336
 
        # If the dirstate cache is up to date and knows the hash and size,
1337
 
        # return that.
1338
 
        # Otherwise if there are no content filters, return the on-disk size
1339
 
        # and leave the hash blank.
1340
 
        # Otherwise, read and filter the on-disk file and use its size and
1341
 
        # hash.
1342
 
        #
1343
 
        # The dirstate doesn't store the size of the canonical form so we
1344
 
        # can't trust it for content-filtered trees.  We just return None.
1345
 
        dirstate_sha1 = self._dirstate.sha1_from_stat(path, stat_result)
1346
 
        executable = self._is_executable_from_path_and_stat(path, stat_result)
1347
 
        return ('file', None, executable, dirstate_sha1)
1348
 
 
1349
 
 
1350
 
class WorkingTree4(DirStateWorkingTree):
1351
 
    """This is the Format 4 working tree.
1352
 
 
1353
 
    This differs from WorkingTree3 by:
1354
 
     - Having a consolidated internal dirstate, stored in a
1355
 
       randomly-accessible sorted file on disk.
1356
 
     - Not having a regular inventory attribute.  One can be synthesized
1357
 
       on demand but this is expensive and should be avoided.
1358
 
 
1359
 
    This is new in bzr 0.15.
1360
 
    """
1361
 
 
1362
 
 
1363
 
class WorkingTree5(ContentFilteringDirStateWorkingTree):
1364
 
    """This is the Format 5 working tree.
1365
 
 
1366
 
    This differs from WorkingTree4 by:
1367
 
     - Supporting content filtering.
1368
 
 
1369
 
    This is new in bzr 1.11.
1370
 
    """
1371
 
 
1372
 
 
1373
 
class WorkingTree6(ContentFilteringDirStateWorkingTree):
1374
 
    """This is the Format 6 working tree.
1375
 
 
1376
 
    This differs from WorkingTree5 by:
1377
 
     - Supporting a current view that may mask the set of files in a tree
1378
 
       impacted by most user operations.
1379
 
 
1380
 
    This is new in bzr 1.14.
1381
 
    """
1382
 
 
1383
 
    def _make_views(self):
1384
 
        return views.PathBasedViews(self)
1385
 
 
1386
 
 
1387
 
class DirStateWorkingTreeFormat(WorkingTreeFormat3):
1388
 
 
1389
 
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
1390
 
                   accelerator_tree=None, hardlink=False):
 
1236
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1237
    """The first consolidated dirstate working tree format.
 
1238
 
 
1239
    This format:
 
1240
        - exists within a metadir controlling .bzr
 
1241
        - includes an explicit version marker for the workingtree control
 
1242
          files, separate from the BzrDir format
 
1243
        - modifies the hash cache format
 
1244
        - is new in bzr 0.15
 
1245
        - uses a LockDir to guard access to it.
 
1246
    """
 
1247
 
 
1248
    upgrade_recommended = False
 
1249
 
 
1250
    def get_format_string(self):
 
1251
        """See WorkingTreeFormat.get_format_string()."""
 
1252
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1253
 
 
1254
    def get_format_description(self):
 
1255
        """See WorkingTreeFormat.get_format_description()."""
 
1256
        return "Working tree format 4"
 
1257
 
 
1258
    def initialize(self, a_bzrdir, revision_id=None):
1391
1259
        """See WorkingTreeFormat.initialize().
1392
1260
 
1393
1261
        :param revision_id: allows creating a working tree at a different
1394
1262
        revision than the branch is at.
1395
 
        :param accelerator_tree: A tree which can be used for retrieving file
1396
 
            contents more quickly than the revision tree, i.e. a workingtree.
1397
 
            The revision tree will be used for cases where accelerator_tree's
1398
 
            content is different.
1399
 
        :param hardlink: If true, hard-link files from accelerator_tree,
1400
 
            where possible.
1401
1263
 
1402
1264
        These trees get an initial random root id, if their repository supports
1403
1265
        rich root data, TREE_ROOT otherwise.
1404
1266
        """
 
1267
        revision_id = osutils.safe_revision_id(revision_id)
1405
1268
        if not isinstance(a_bzrdir.transport, LocalTransport):
1406
1269
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
1407
1270
        transport = a_bzrdir.get_workingtree_transport(self)
1408
1271
        control_files = self._open_control_files(a_bzrdir)
1409
1272
        control_files.create_lock()
1410
1273
        control_files.lock_write()
1411
 
        transport.put_bytes('format', self.get_format_string(),
1412
 
            mode=a_bzrdir._get_file_mode())
1413
 
        if from_branch is not None:
1414
 
            branch = from_branch
1415
 
        else:
1416
 
            branch = a_bzrdir.open_branch()
 
1274
        control_files.put_utf8('format', self.get_format_string())
 
1275
        branch = a_bzrdir.open_branch()
1417
1276
        if revision_id is None:
1418
1277
            revision_id = branch.last_revision()
1419
1278
        local_path = transport.local_abspath('dirstate')
1421
1280
        state = dirstate.DirState.initialize(local_path)
1422
1281
        state.unlock()
1423
1282
        del state
1424
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1283
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1425
1284
                         branch,
1426
1285
                         _format=self,
1427
1286
                         _bzrdir=a_bzrdir,
1429
1288
        wt._new_tree()
1430
1289
        wt.lock_tree_write()
1431
1290
        try:
1432
 
            self._init_custom_control_files(wt)
1433
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
 
1291
            if revision_id in (None, NULL_REVISION):
1434
1292
                if branch.repository.supports_rich_root():
1435
1293
                    wt._set_root_id(generate_ids.gen_root_id())
1436
1294
                else:
1437
1295
                    wt._set_root_id(ROOT_ID)
1438
1296
                wt.flush()
1439
 
            basis = None
1440
 
            # frequently, we will get here due to branching.  The accelerator
1441
 
            # tree will be the tree from the branch, so the desired basis
1442
 
            # tree will often be a parent of the accelerator tree.
1443
 
            if accelerator_tree is not None:
1444
 
                try:
1445
 
                    basis = accelerator_tree.revision_tree(revision_id)
1446
 
                except errors.NoSuchRevision:
1447
 
                    pass
1448
 
            if basis is None:
1449
 
                basis = branch.repository.revision_tree(revision_id)
1450
 
            if revision_id == _mod_revision.NULL_REVISION:
1451
 
                parents_list = []
1452
 
            else:
1453
 
                parents_list = [(revision_id, basis)]
 
1297
            wt.set_last_revision(revision_id)
 
1298
            wt.flush()
 
1299
            basis = wt.basis_tree()
1454
1300
            basis.lock_read()
1455
 
            try:
1456
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1301
            # if the basis has a root id we have to use that; otherwise we use
 
1302
            # a new random one
 
1303
            basis_root_id = basis.get_root_id()
 
1304
            if basis_root_id is not None:
 
1305
                wt._set_root_id(basis_root_id)
1457
1306
                wt.flush()
1458
 
                # if the basis has a root id we have to use that; otherwise we
1459
 
                # use a new random one
1460
 
                basis_root_id = basis.get_root_id()
1461
 
                if basis_root_id is not None:
1462
 
                    wt._set_root_id(basis_root_id)
1463
 
                    wt.flush()
1464
 
                if wt.supports_content_filtering():
1465
 
                    # The original tree may not have the same content filters
1466
 
                    # applied so we can't safely build the inventory delta from
1467
 
                    # the source tree.
1468
 
                    delta_from_tree = False
1469
 
                else:
1470
 
                    delta_from_tree = True
1471
 
                # delta_from_tree is safe even for DirStateRevisionTrees,
1472
 
                # because wt4.apply_inventory_delta does not mutate the input
1473
 
                # inventory entries.
1474
 
                transform.build_tree(basis, wt, accelerator_tree,
1475
 
                                     hardlink=hardlink,
1476
 
                                     delta_from_tree=delta_from_tree)
1477
 
            finally:
1478
 
                basis.unlock()
 
1307
            transform.build_tree(basis, wt)
 
1308
            basis.unlock()
1479
1309
        finally:
1480
1310
            control_files.unlock()
1481
1311
            wt.unlock()
1482
1312
        return wt
1483
1313
 
1484
 
    def _init_custom_control_files(self, wt):
1485
 
        """Subclasses with custom control files should override this method.
1486
 
 
1487
 
        The working tree and control files are locked for writing when this
1488
 
        method is called.
1489
 
 
1490
 
        :param wt: the WorkingTree object
1491
 
        """
1492
 
 
1493
1314
    def _open(self, a_bzrdir, control_files):
1494
1315
        """Open the tree itself.
1495
1316
 
1496
1317
        :param a_bzrdir: the dir for the tree.
1497
1318
        :param control_files: the control files for the tree.
1498
1319
        """
1499
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1320
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1500
1321
                           branch=a_bzrdir.open_branch(),
1501
1322
                           _format=self,
1502
1323
                           _bzrdir=a_bzrdir,
1503
1324
                           _control_files=control_files)
1504
1325
 
1505
1326
    def __get_matchingbzrdir(self):
1506
 
        return self._get_matchingbzrdir()
1507
 
 
1508
 
    def _get_matchingbzrdir(self):
1509
 
        """Overrideable method to get a bzrdir for testing."""
1510
1327
        # please test against something that will let us do tree references
1511
1328
        return bzrdir.format_registry.make_bzrdir(
1512
1329
            'dirstate-with-subtree')
1514
1331
    _matchingbzrdir = property(__get_matchingbzrdir)
1515
1332
 
1516
1333
 
1517
 
class WorkingTreeFormat4(DirStateWorkingTreeFormat):
1518
 
    """The first consolidated dirstate working tree format.
1519
 
 
1520
 
    This format:
1521
 
        - exists within a metadir controlling .bzr
1522
 
        - includes an explicit version marker for the workingtree control
1523
 
          files, separate from the BzrDir format
1524
 
        - modifies the hash cache format
1525
 
        - is new in bzr 0.15
1526
 
        - uses a LockDir to guard access to it.
1527
 
    """
1528
 
 
1529
 
    upgrade_recommended = False
1530
 
 
1531
 
    _tree_class = WorkingTree4
1532
 
 
1533
 
    def get_format_string(self):
1534
 
        """See WorkingTreeFormat.get_format_string()."""
1535
 
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
1536
 
 
1537
 
    def get_format_description(self):
1538
 
        """See WorkingTreeFormat.get_format_description()."""
1539
 
        return "Working tree format 4"
1540
 
 
1541
 
 
1542
 
class WorkingTreeFormat5(DirStateWorkingTreeFormat):
1543
 
    """WorkingTree format supporting content filtering.
1544
 
    """
1545
 
 
1546
 
    upgrade_recommended = False
1547
 
 
1548
 
    _tree_class = WorkingTree5
1549
 
 
1550
 
    def get_format_string(self):
1551
 
        """See WorkingTreeFormat.get_format_string()."""
1552
 
        return "Bazaar Working Tree Format 5 (bzr 1.11)\n"
1553
 
 
1554
 
    def get_format_description(self):
1555
 
        """See WorkingTreeFormat.get_format_description()."""
1556
 
        return "Working tree format 5"
1557
 
 
1558
 
    def supports_content_filtering(self):
1559
 
        return True
1560
 
 
1561
 
 
1562
 
class WorkingTreeFormat6(DirStateWorkingTreeFormat):
1563
 
    """WorkingTree format supporting views.
1564
 
    """
1565
 
 
1566
 
    upgrade_recommended = False
1567
 
 
1568
 
    _tree_class = WorkingTree6
1569
 
 
1570
 
    def get_format_string(self):
1571
 
        """See WorkingTreeFormat.get_format_string()."""
1572
 
        return "Bazaar Working Tree Format 6 (bzr 1.14)\n"
1573
 
 
1574
 
    def get_format_description(self):
1575
 
        """See WorkingTreeFormat.get_format_description()."""
1576
 
        return "Working tree format 6"
1577
 
 
1578
 
    def _init_custom_control_files(self, wt):
1579
 
        """Subclasses with custom control files should override this method."""
1580
 
        wt._transport.put_bytes('views', '', mode=wt.bzrdir._get_file_mode())
1581
 
 
1582
 
    def supports_content_filtering(self):
1583
 
        return True
1584
 
 
1585
 
    def supports_views(self):
1586
 
        return True
1587
 
 
1588
 
 
1589
1334
class DirStateRevisionTree(Tree):
1590
 
    """A revision tree pulling the inventory from a dirstate.
1591
 
    
1592
 
    Note that this is one of the historical (ie revision) trees cached in the
1593
 
    dirstate for easy access, not the workingtree.
1594
 
    """
 
1335
    """A revision tree pulling the inventory from a dirstate."""
1595
1336
 
1596
1337
    def __init__(self, dirstate, revision_id, repository):
1597
1338
        self._dirstate = dirstate
1598
 
        self._revision_id = revision_id
 
1339
        self._revision_id = osutils.safe_revision_id(revision_id)
1599
1340
        self._repository = repository
1600
1341
        self._inventory = None
1601
1342
        self._locked = 0
1602
1343
        self._dirstate_locked = False
1603
 
        self._repo_supports_tree_reference = getattr(
1604
 
            repository._format, "supports_tree_reference",
1605
 
            False)
1606
1344
 
1607
1345
    def __repr__(self):
1608
1346
        return "<%s of %s in %s>" % \
1609
1347
            (self.__class__.__name__, self._revision_id, self._dirstate)
1610
1348
 
1611
 
    def annotate_iter(self, file_id,
1612
 
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1349
    def annotate_iter(self, file_id):
1613
1350
        """See Tree.annotate_iter"""
1614
 
        text_key = (file_id, self.inventory[file_id].revision)
1615
 
        annotations = self._repository.texts.annotate(text_key)
1616
 
        return [(key[-1], line) for (key, line) in annotations]
 
1351
        w = self._repository.weave_store.get_weave(file_id,
 
1352
                           self._repository.get_transaction())
 
1353
        return w.annotate_iter(self.inventory[file_id].revision)
1617
1354
 
1618
 
    def _get_ancestors(self, default_revision):
1619
 
        return set(self._repository.get_ancestry(self._revision_id,
1620
 
                                                 topo_sorted=False))
1621
1355
    def _comparison_data(self, entry, path):
1622
1356
        """See Tree._comparison_data."""
1623
1357
        if entry is None:
1640
1374
    def get_root_id(self):
1641
1375
        return self.path2id('')
1642
1376
 
1643
 
    def id2path(self, file_id):
1644
 
        "Convert a file-id to a path."
1645
 
        entry = self._get_entry(file_id=file_id)
1646
 
        if entry == (None, None):
1647
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1648
 
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1649
 
        return path_utf8.decode('utf8')
1650
 
 
1651
 
    def iter_references(self):
1652
 
        if not self._repo_supports_tree_reference:
1653
 
            # When the repo doesn't support references, we will have nothing to
1654
 
            # return
1655
 
            return iter([])
1656
 
        # Otherwise, fall back to the default implementation
1657
 
        return super(DirStateRevisionTree, self).iter_references()
1658
 
 
1659
1377
    def _get_parent_index(self):
1660
1378
        """Return the index in the dirstate referenced by this tree."""
1661
1379
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1666
1384
        If either file_id or path is supplied, it is used as the key to lookup.
1667
1385
        If both are supplied, the fastest lookup is used, and an error is
1668
1386
        raised if they do not both point at the same row.
1669
 
 
 
1387
        
1670
1388
        :param file_id: An optional unicode file_id to be looked up.
1671
1389
        :param path: An optional unicode path to be looked up.
1672
1390
        :return: The dirstate row tuple for path/file_id, or (None, None)
1673
1391
        """
1674
1392
        if file_id is None and path is None:
1675
1393
            raise errors.BzrError('must supply file_id or path')
 
1394
        file_id = osutils.safe_file_id(file_id)
1676
1395
        if path is not None:
1677
1396
            path = path.encode('utf8')
1678
1397
        parent_index = self._get_parent_index()
1686
1405
 
1687
1406
        This is relatively expensive: we have to walk the entire dirstate.
1688
1407
        """
1689
 
        if not self._locked:
1690
 
            raise AssertionError(
1691
 
                'cannot generate inventory of an unlocked '
1692
 
                'dirstate revision tree')
 
1408
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1409
            'dirstate revision tree'
1693
1410
        # separate call for profiling - makes it clear where the costs are.
1694
1411
        self._dirstate._read_dirblocks_if_needed()
1695
 
        if self._revision_id not in self._dirstate.get_parent_ids():
1696
 
            raise AssertionError(
1697
 
                'parent %s has disappeared from %s' % (
1698
 
                self._revision_id, self._dirstate.get_parent_ids()))
 
1412
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1413
            'parent %s has disappeared from %s' % (
 
1414
            self._revision_id, self._dirstate.get_parent_ids())
1699
1415
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1700
1416
        # This is identical now to the WorkingTree _generate_inventory except
1701
1417
        # for the tree index use.
1702
1418
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1703
1419
        current_id = root_key[2]
1704
 
        if current_entry[parent_index][0] != 'd':
1705
 
            raise AssertionError()
 
1420
        assert current_entry[parent_index][0] == 'd'
1706
1421
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1707
1422
        inv.root.revision = current_entry[parent_index][4]
1708
1423
        # Turn some things into local variables
1748
1463
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1749
1464
                            % entry)
1750
1465
                # These checks cost us around 40ms on a 55k entry tree
1751
 
                if file_id in inv_byid:
1752
 
                    raise AssertionError('file_id %s already in'
1753
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
1754
 
                if name_unicode in parent_ie.children:
1755
 
                    raise AssertionError('name %r already in parent'
1756
 
                        % (name_unicode,))
 
1466
                assert file_id not in inv_byid
 
1467
                assert name_unicode not in parent_ie.children
1757
1468
                inv_byid[file_id] = inv_entry
1758
1469
                parent_ie.children[name_unicode] = inv_entry
1759
1470
        self._inventory = inv
1769
1480
            return None
1770
1481
        parent_index = self._get_parent_index()
1771
1482
        last_changed_revision = entry[1][parent_index][4]
1772
 
        try:
1773
 
            rev = self._repository.get_revision(last_changed_revision)
1774
 
        except errors.NoSuchRevision:
1775
 
            raise errors.FileTimestampUnavailable(self.id2path(file_id))
1776
 
        return rev.timestamp
 
1483
        return self._repository.get_revision(last_changed_revision).timestamp
1777
1484
 
1778
1485
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1779
1486
        entry = self._get_entry(file_id=file_id, path=path)
1783
1490
            return parent_details[1]
1784
1491
        return None
1785
1492
 
1786
 
    def get_file(self, file_id, path=None):
 
1493
    def get_weave(self, file_id):
 
1494
        return self._repository.weave_store.get_weave(file_id,
 
1495
                self._repository.get_transaction())
 
1496
 
 
1497
    def get_file(self, file_id):
1787
1498
        return StringIO(self.get_file_text(file_id))
1788
1499
 
 
1500
    def get_file_lines(self, file_id):
 
1501
        ie = self.inventory[file_id]
 
1502
        return self._repository.weave_store.get_weave(file_id,
 
1503
                self._repository.get_transaction()).get_lines(ie.revision)
 
1504
 
1789
1505
    def get_file_size(self, file_id):
1790
 
        """See Tree.get_file_size"""
1791
1506
        return self.inventory[file_id].text_size
1792
1507
 
1793
 
    def get_file_text(self, file_id, path=None):
1794
 
        _, content = list(self.iter_files_bytes([(file_id, None)]))[0]
1795
 
        return ''.join(content)
 
1508
    def get_file_text(self, file_id):
 
1509
        return ''.join(self.get_file_lines(file_id))
1796
1510
 
1797
1511
    def get_reference_revision(self, file_id, path=None):
1798
1512
        return self.inventory[file_id].reference_revision
1799
1513
 
1800
 
    def iter_files_bytes(self, desired_files):
1801
 
        """See Tree.iter_files_bytes.
1802
 
 
1803
 
        This version is implemented on top of Repository.iter_files_bytes"""
1804
 
        parent_index = self._get_parent_index()
1805
 
        repo_desired_files = []
1806
 
        for file_id, identifier in desired_files:
1807
 
            entry = self._get_entry(file_id)
1808
 
            if entry == (None, None):
1809
 
                raise errors.NoSuchId(self, file_id)
1810
 
            repo_desired_files.append((file_id, entry[1][parent_index][4],
1811
 
                                       identifier))
1812
 
        return self._repository.iter_files_bytes(repo_desired_files)
1813
 
 
1814
1514
    def get_symlink_target(self, file_id):
1815
1515
        entry = self._get_entry(file_id=file_id)
1816
1516
        parent_index = self._get_parent_index()
1817
1517
        if entry[1][parent_index][0] != 'l':
1818
1518
            return None
1819
1519
        else:
1820
 
            target = entry[1][parent_index][1]
1821
 
            target = target.decode('utf8')
1822
 
            return target
 
1520
            # At present, none of the tree implementations supports non-ascii
 
1521
            # symlink targets. So we will just assume that the dirstate path is
 
1522
            # correct.
 
1523
            return entry[1][parent_index][1]
1823
1524
 
1824
1525
    def get_revision_id(self):
1825
1526
        """Return the revision id for this tree."""
1843
1544
        return bool(self.path2id(filename))
1844
1545
 
1845
1546
    def kind(self, file_id):
1846
 
        entry = self._get_entry(file_id=file_id)[1]
1847
 
        if entry is None:
1848
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1849
 
        parent_index = self._get_parent_index()
1850
 
        return dirstate.DirState._minikind_to_kind[entry[parent_index][0]]
1851
 
 
1852
 
    def stored_kind(self, file_id):
1853
 
        """See Tree.stored_kind"""
1854
 
        return self.kind(file_id)
1855
 
 
1856
 
    def path_content_summary(self, path):
1857
 
        """See Tree.path_content_summary."""
1858
 
        id = self.inventory.path2id(path)
1859
 
        if id is None:
1860
 
            return ('missing', None, None, None)
1861
 
        entry = self._inventory[id]
1862
 
        kind = entry.kind
1863
 
        if kind == 'file':
1864
 
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
1865
 
        elif kind == 'symlink':
1866
 
            return (kind, None, None, entry.symlink_target)
1867
 
        else:
1868
 
            return (kind, None, None, None)
 
1547
        return self.inventory[file_id].kind
1869
1548
 
1870
1549
    def is_executable(self, file_id, path=None):
1871
1550
        ie = self.inventory[file_id]
1873
1552
            return None
1874
1553
        return ie.executable
1875
1554
 
1876
 
    def is_locked(self):
1877
 
        return self._locked
1878
 
 
1879
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1555
    def list_files(self, include_root=False):
1880
1556
        # We use a standard implementation, because DirStateRevisionTree is
1881
1557
        # dealing with one of the parents of the current state
1882
1558
        inv = self._get_inventory()
1883
 
        if from_dir is None:
1884
 
            from_dir_id = None
1885
 
        else:
1886
 
            from_dir_id = inv.path2id(from_dir)
1887
 
            if from_dir_id is None:
1888
 
                # Directory not versioned
1889
 
                return
1890
 
        entries = inv.iter_entries(from_dir=from_dir_id, recursive=recursive)
1891
 
        if inv.root is not None and not include_root and from_dir is None:
 
1559
        entries = inv.iter_entries()
 
1560
        if self.inventory.root is not None and not include_root:
1892
1561
            entries.next()
1893
1562
        for path, entry in entries:
1894
1563
            yield path, 'V', entry.kind, entry.file_id, entry
1895
1564
 
1896
1565
    def lock_read(self):
1897
 
        """Lock the tree for a set of operations.
1898
 
 
1899
 
        :return: An object with an unlock method which will release the lock
1900
 
            obtained.
1901
 
        """
 
1566
        """Lock the tree for a set of operations."""
1902
1567
        if not self._locked:
1903
1568
            self._repository.lock_read()
1904
1569
            if self._dirstate._lock_token is None:
1905
1570
                self._dirstate.lock_read()
1906
1571
                self._dirstate_locked = True
1907
1572
        self._locked += 1
1908
 
        return self
1909
1573
 
1910
1574
    def _must_be_locked(self):
1911
1575
        if not self._locked:
1932
1596
                self._dirstate_locked = False
1933
1597
            self._repository.unlock()
1934
1598
 
1935
 
    @needs_read_lock
1936
 
    def supports_tree_reference(self):
1937
 
        return self._repo_supports_tree_reference
1938
 
 
1939
1599
    def walkdirs(self, prefix=""):
1940
1600
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1941
 
        # implementation based on an inventory.
 
1601
        # implementation based on an inventory.  
1942
1602
        # This should be cleaned up to use the much faster Dirstate code
1943
1603
        # So for now, we just build up the parent inventory, and extract
1944
1604
        # it the same way RevisionTree does.
1973
1633
 
1974
1634
class InterDirStateTree(InterTree):
1975
1635
    """Fast path optimiser for changes_from with dirstate trees.
1976
 
 
1977
 
    This is used only when both trees are in the dirstate working file, and
1978
 
    the source is any parent within the dirstate, and the destination is
 
1636
    
 
1637
    This is used only when both trees are in the dirstate working file, and 
 
1638
    the source is any parent within the dirstate, and the destination is 
1979
1639
    the current working tree of the same dirstate.
1980
1640
    """
1981
1641
    # this could be generalized to allow comparisons between any trees in the
1994
1654
        target.set_parent_ids([revid])
1995
1655
        return target.basis_tree(), target
1996
1656
 
1997
 
    @classmethod
1998
 
    def make_source_parent_tree_python_dirstate(klass, test_case, source, target):
1999
 
        result = klass.make_source_parent_tree(source, target)
2000
 
        result[1]._iter_changes = dirstate.ProcessEntryPython
2001
 
        return result
2002
 
 
2003
 
    @classmethod
2004
 
    def make_source_parent_tree_compiled_dirstate(klass, test_case, source,
2005
 
                                                  target):
2006
 
        from bzrlib.tests.test__dirstate_helpers import \
2007
 
            compiled_dirstate_helpers_feature
2008
 
        test_case.requireFeature(compiled_dirstate_helpers_feature)
2009
 
        from bzrlib._dirstate_helpers_pyx import ProcessEntryC
2010
 
        result = klass.make_source_parent_tree(source, target)
2011
 
        result[1]._iter_changes = ProcessEntryC
2012
 
        return result
2013
 
 
2014
1657
    _matching_from_tree_format = WorkingTreeFormat4()
2015
1658
    _matching_to_tree_format = WorkingTreeFormat4()
2016
 
 
2017
 
    @classmethod
2018
 
    def _test_mutable_trees_to_test_trees(klass, test_case, source, target):
2019
 
        # This method shouldn't be called, because we have python and C
2020
 
        # specific flavours.
2021
 
        raise NotImplementedError
2022
 
 
2023
 
    def iter_changes(self, include_unchanged=False,
 
1659
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1660
 
 
1661
    def _iter_changes(self, include_unchanged=False,
2024
1662
                      specific_files=None, pb=None, extra_trees=[],
2025
1663
                      require_versioned=True, want_unversioned=False):
2026
1664
        """Return the changes from source to target.
2027
1665
 
2028
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1666
        :return: An iterator that yields tuples. See InterTree._iter_changes
2029
1667
            for details.
2030
1668
        :param specific_files: An optional list of file paths to restrict the
2031
1669
            comparison to. When mapping filenames to ids, all matches in all
2042
1680
            output. An unversioned file is defined as one with (False, False)
2043
1681
            for the versioned pair.
2044
1682
        """
 
1683
        utf8_decode = cache_utf8._utf8_decode
 
1684
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1685
        # NB: show_status depends on being able to pass in non-versioned files
 
1686
        # and report them as unknown
2045
1687
        # TODO: handle extra trees in the dirstate.
2046
 
        if (extra_trees or specific_files == []):
 
1688
        # TODO: handle comparisons as an empty tree as a different special
 
1689
        # case? mbp 20070226
 
1690
        if extra_trees or (self.source._revision_id == NULL_REVISION):
2047
1691
            # we can't fast-path these cases (yet)
2048
 
            return super(InterDirStateTree, self).iter_changes(
 
1692
            for f in super(InterDirStateTree, self)._iter_changes(
2049
1693
                include_unchanged, specific_files, pb, extra_trees,
2050
 
                require_versioned, want_unversioned=want_unversioned)
 
1694
                require_versioned, want_unversioned=want_unversioned):
 
1695
                yield f
 
1696
            return
2051
1697
        parent_ids = self.target.get_parent_ids()
2052
 
        if not (self.source._revision_id in parent_ids
2053
 
                or self.source._revision_id == _mod_revision.NULL_REVISION):
2054
 
            raise AssertionError(
2055
 
                "revision {%s} is not stored in {%s}, but %s "
2056
 
                "can only be used for trees stored in the dirstate"
2057
 
                % (self.source._revision_id, self.target, self.iter_changes))
 
1698
        assert (self.source._revision_id in parent_ids), \
 
1699
                "revision {%s} is not stored in {%s}, but %s " \
 
1700
                "can only be used for trees stored in the dirstate" \
 
1701
                % (self.source._revision_id, self.target, self._iter_changes)
2058
1702
        target_index = 0
2059
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1703
        if self.source._revision_id == NULL_REVISION:
2060
1704
            source_index = None
2061
1705
            indices = (target_index,)
2062
1706
        else:
2063
 
            if not (self.source._revision_id in parent_ids):
2064
 
                raise AssertionError(
2065
 
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
2066
 
                    self.source._revision_id, parent_ids))
 
1707
            assert (self.source._revision_id in parent_ids), \
 
1708
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1709
                self.source._revision_id, parent_ids)
2067
1710
            source_index = 1 + parent_ids.index(self.source._revision_id)
2068
 
            indices = (source_index, target_index)
 
1711
            indices = (source_index,target_index)
2069
1712
        # -- make all specific_files utf8 --
2070
1713
        if specific_files:
2071
1714
            specific_files_utf8 = set()
2072
1715
            for path in specific_files:
2073
 
                # Note, if there are many specific files, using cache_utf8
2074
 
                # would be good here.
2075
1716
                specific_files_utf8.add(path.encode('utf8'))
2076
1717
            specific_files = specific_files_utf8
2077
1718
        else:
2078
1719
            specific_files = set([''])
2079
1720
        # -- specific_files is now a utf8 path set --
2080
 
 
2081
1721
        # -- get the state object and prepare it.
2082
1722
        state = self.target.current_dirstate()
2083
1723
        state._read_dirblocks_if_needed()
 
1724
        def _entries_for_path(path):
 
1725
            """Return a list with all the entries that match path for all ids.
 
1726
            """
 
1727
            dirname, basename = os.path.split(path)
 
1728
            key = (dirname, basename, '')
 
1729
            block_index, present = state._find_block_index_from_key(key)
 
1730
            if not present:
 
1731
                # the block which should contain path is absent.
 
1732
                return []
 
1733
            result = []
 
1734
            block = state._dirblocks[block_index][1]
 
1735
            entry_index, _ = state._find_entry_index(key, block)
 
1736
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1737
            while (entry_index < len(block) and
 
1738
                block[entry_index][0][0:2] == key[0:2]):
 
1739
                result.append(block[entry_index])
 
1740
                entry_index += 1
 
1741
            return result
2084
1742
        if require_versioned:
2085
1743
            # -- check all supplied paths are versioned in a search tree. --
2086
 
            not_versioned = []
 
1744
            all_versioned = True
2087
1745
            for path in specific_files:
2088
 
                path_entries = state._entries_for_path(path)
 
1746
                path_entries = _entries_for_path(path)
2089
1747
                if not path_entries:
2090
1748
                    # this specified path is not present at all: error
2091
 
                    not_versioned.append(path)
2092
 
                    continue
 
1749
                    all_versioned = False
 
1750
                    break
2093
1751
                found_versioned = False
2094
1752
                # for each id at this path
2095
1753
                for entry in path_entries:
2102
1760
                if not found_versioned:
2103
1761
                    # none of the indexes was not 'absent' at all ids for this
2104
1762
                    # path.
2105
 
                    not_versioned.append(path)
2106
 
            if len(not_versioned) > 0:
2107
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1763
                    all_versioned = False
 
1764
                    break
 
1765
            if not all_versioned:
 
1766
                raise errors.PathsNotVersionedError(specific_files)
2108
1767
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2109
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1768
        search_specific_files = set()
 
1769
        for path in specific_files:
 
1770
            other_specific_files = specific_files.difference(set([path]))
 
1771
            if not osutils.is_inside_any(other_specific_files, path):
 
1772
                # this is a top level path, we must check it.
 
1773
                search_specific_files.add(path)
 
1774
        # sketch: 
 
1775
        # compare source_index and target_index at or under each element of search_specific_files.
 
1776
        # follow the following comparison table. Note that we only want to do diff operations when
 
1777
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1778
        # for the target.
 
1779
        # cases:
 
1780
        # 
 
1781
        # Source | Target | disk | action
 
1782
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1783
        #        |        |      | diff check on source-target
 
1784
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1785
        #        |        |      | ???
 
1786
        #   r    |  a     |      | add source to search
 
1787
        #   r    |  a     |  a   | 
 
1788
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1789
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1790
        #   a    | fdlt   |      | add new id
 
1791
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1792
        #   a    |  a     |      | not present in either tree, skip
 
1793
        #   a    |  a     |  a   | not present in any tree, skip
 
1794
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1795
        #        |        |      | may not be selected by the users list of paths.
 
1796
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1797
        #        |        |      | may not be selected by the users list of paths.
 
1798
        #  fdlt  | fdlt   |      | content in both: diff them
 
1799
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1800
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1801
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1802
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1803
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1804
        #        |        |      | this id at the other path.
 
1805
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1806
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1807
        #        |        |      | this id at the other path.
 
1808
 
 
1809
        # for all search_indexs in each path at or under each element of
 
1810
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1811
        # relocated path as one to search if its not searched already. If the
 
1812
        # detail is not relocated, add the id.
 
1813
        searched_specific_files = set()
 
1814
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1815
        # Using a list so that we can access the values and change them in
 
1816
        # nested scope. Each one is [path, file_id, entry]
 
1817
        last_source_parent = [None, None]
 
1818
        last_target_parent = [None, None]
2110
1819
 
2111
1820
        use_filesystem_for_exec = (sys.platform != 'win32')
2112
 
        iter_changes = self.target._iter_changes(include_unchanged,
2113
 
            use_filesystem_for_exec, search_specific_files, state,
2114
 
            source_index, target_index, want_unversioned, self.target)
2115
 
        return iter_changes.iter_changes()
 
1821
 
 
1822
        # Just a sentry, so that _process_entry can say that this
 
1823
        # record is handled, but isn't interesting to process (unchanged)
 
1824
        uninteresting = object()
 
1825
 
 
1826
 
 
1827
        old_dirname_to_file_id = {}
 
1828
        new_dirname_to_file_id = {}
 
1829
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1830
        #       keeping a cache of directories that we have seen.
 
1831
 
 
1832
        def _process_entry(entry, path_info):
 
1833
            """Compare an entry and real disk to generate delta information.
 
1834
 
 
1835
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1836
                the path of entry. If None, then the path is considered absent.
 
1837
                (Perhaps we should pass in a concrete entry for this ?)
 
1838
                Basename is returned as a utf8 string because we expect this
 
1839
                tuple will be ignored, and don't want to take the time to
 
1840
                decode.
 
1841
            :return: None if these don't match
 
1842
                     A tuple of information about the change, or
 
1843
                     the object 'uninteresting' if these match, but are
 
1844
                     basically identical.
 
1845
            """
 
1846
            if source_index is None:
 
1847
                source_details = NULL_PARENT_DETAILS
 
1848
            else:
 
1849
                source_details = entry[1][source_index]
 
1850
            target_details = entry[1][target_index]
 
1851
            target_minikind = target_details[0]
 
1852
            if path_info is not None and target_minikind in 'fdlt':
 
1853
                assert target_index == 0
 
1854
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1855
                                                  stat_value=path_info[3])
 
1856
                # The entry may have been modified by update_entry
 
1857
                target_details = entry[1][target_index]
 
1858
                target_minikind = target_details[0]
 
1859
            else:
 
1860
                link_or_sha1 = None
 
1861
            file_id = entry[0][2]
 
1862
            source_minikind = source_details[0]
 
1863
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1864
                # claimed content in both: diff
 
1865
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1866
                #        |        |      | diff check on source-target
 
1867
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1868
                #        |        |      | ???
 
1869
                if source_minikind in 'r':
 
1870
                    # add the source to the search path to find any children it
 
1871
                    # has.  TODO ? : only add if it is a container ?
 
1872
                    if not osutils.is_inside_any(searched_specific_files,
 
1873
                                                 source_details[1]):
 
1874
                        search_specific_files.add(source_details[1])
 
1875
                    # generate the old path; this is needed for stating later
 
1876
                    # as well.
 
1877
                    old_path = source_details[1]
 
1878
                    old_dirname, old_basename = os.path.split(old_path)
 
1879
                    path = pathjoin(entry[0][0], entry[0][1])
 
1880
                    old_entry = state._get_entry(source_index,
 
1881
                                                 path_utf8=old_path)
 
1882
                    # update the source details variable to be the real
 
1883
                    # location.
 
1884
                    source_details = old_entry[1][source_index]
 
1885
                    source_minikind = source_details[0]
 
1886
                else:
 
1887
                    old_dirname = entry[0][0]
 
1888
                    old_basename = entry[0][1]
 
1889
                    old_path = path = None
 
1890
                if path_info is None:
 
1891
                    # the file is missing on disk, show as removed.
 
1892
                    content_change = True
 
1893
                    target_kind = None
 
1894
                    target_exec = False
 
1895
                else:
 
1896
                    # source and target are both versioned and disk file is present.
 
1897
                    target_kind = path_info[2]
 
1898
                    if target_kind == 'directory':
 
1899
                        if path is None:
 
1900
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1901
                        new_dirname_to_file_id[path] = file_id
 
1902
                        if source_minikind != 'd':
 
1903
                            content_change = True
 
1904
                        else:
 
1905
                            # directories have no fingerprint
 
1906
                            content_change = False
 
1907
                        target_exec = False
 
1908
                    elif target_kind == 'file':
 
1909
                        if source_minikind != 'f':
 
1910
                            content_change = True
 
1911
                        else:
 
1912
                            # We could check the size, but we already have the
 
1913
                            # sha1 hash.
 
1914
                            content_change = (link_or_sha1 != source_details[1])
 
1915
                        # Target details is updated at update_entry time
 
1916
                        if use_filesystem_for_exec:
 
1917
                            # We don't need S_ISREG here, because we are sure
 
1918
                            # we are dealing with a file.
 
1919
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1920
                        else:
 
1921
                            target_exec = target_details[3]
 
1922
                    elif target_kind == 'symlink':
 
1923
                        if source_minikind != 'l':
 
1924
                            content_change = True
 
1925
                        else:
 
1926
                            content_change = (link_or_sha1 != source_details[1])
 
1927
                        target_exec = False
 
1928
                    elif target_kind == 'tree-reference':
 
1929
                        if source_minikind != 't':
 
1930
                            content_change = True
 
1931
                        else:
 
1932
                            content_change = False
 
1933
                        target_exec = False
 
1934
                    else:
 
1935
                        raise Exception, "unknown kind %s" % path_info[2]
 
1936
                if source_minikind == 'd':
 
1937
                    if path is None:
 
1938
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1939
                    old_dirname_to_file_id[old_path] = file_id
 
1940
                # parent id is the entry for the path in the target tree
 
1941
                if old_dirname == last_source_parent[0]:
 
1942
                    source_parent_id = last_source_parent[1]
 
1943
                else:
 
1944
                    try:
 
1945
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1946
                    except KeyError:
 
1947
                        source_parent_entry = state._get_entry(source_index,
 
1948
                                                               path_utf8=old_dirname)
 
1949
                        source_parent_id = source_parent_entry[0][2]
 
1950
                    if source_parent_id == entry[0][2]:
 
1951
                        # This is the root, so the parent is None
 
1952
                        source_parent_id = None
 
1953
                    else:
 
1954
                        last_source_parent[0] = old_dirname
 
1955
                        last_source_parent[1] = source_parent_id
 
1956
                new_dirname = entry[0][0]
 
1957
                if new_dirname == last_target_parent[0]:
 
1958
                    target_parent_id = last_target_parent[1]
 
1959
                else:
 
1960
                    try:
 
1961
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
1962
                    except KeyError:
 
1963
                        # TODO: We don't always need to do the lookup, because the
 
1964
                        #       parent entry will be the same as the source entry.
 
1965
                        target_parent_entry = state._get_entry(target_index,
 
1966
                                                               path_utf8=new_dirname)
 
1967
                        assert target_parent_entry != (None, None), (
 
1968
                            "Could not find target parent in wt: %s\nparent of: %s"
 
1969
                            % (new_dirname, entry))
 
1970
                        target_parent_id = target_parent_entry[0][2]
 
1971
                    if target_parent_id == entry[0][2]:
 
1972
                        # This is the root, so the parent is None
 
1973
                        target_parent_id = None
 
1974
                    else:
 
1975
                        last_target_parent[0] = new_dirname
 
1976
                        last_target_parent[1] = target_parent_id
 
1977
 
 
1978
                source_exec = source_details[3]
 
1979
                if (include_unchanged
 
1980
                    or content_change
 
1981
                    or source_parent_id != target_parent_id
 
1982
                    or old_basename != entry[0][1]
 
1983
                    or source_exec != target_exec
 
1984
                    ):
 
1985
                    if old_path is None:
 
1986
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1987
                        old_path_u = utf8_decode(old_path)[0]
 
1988
                        path_u = old_path_u
 
1989
                    else:
 
1990
                        old_path_u = utf8_decode(old_path)[0]
 
1991
                        if old_path == path:
 
1992
                            path_u = old_path_u
 
1993
                        else:
 
1994
                            path_u = utf8_decode(path)[0]
 
1995
                    source_kind = _minikind_to_kind[source_minikind]
 
1996
                    return (entry[0][2],
 
1997
                           (old_path_u, path_u),
 
1998
                           content_change,
 
1999
                           (True, True),
 
2000
                           (source_parent_id, target_parent_id),
 
2001
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2002
                           (source_kind, target_kind),
 
2003
                           (source_exec, target_exec))
 
2004
                else:
 
2005
                    return uninteresting
 
2006
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2007
                # looks like a new file
 
2008
                if path_info is not None:
 
2009
                    path = pathjoin(entry[0][0], entry[0][1])
 
2010
                    # parent id is the entry for the path in the target tree
 
2011
                    # TODO: these are the same for an entire directory: cache em.
 
2012
                    parent_id = state._get_entry(target_index,
 
2013
                                                 path_utf8=entry[0][0])[0][2]
 
2014
                    if parent_id == entry[0][2]:
 
2015
                        parent_id = None
 
2016
                    if use_filesystem_for_exec:
 
2017
                        # We need S_ISREG here, because we aren't sure if this
 
2018
                        # is a file or not.
 
2019
                        target_exec = bool(
 
2020
                            stat.S_ISREG(path_info[3].st_mode)
 
2021
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2022
                    else:
 
2023
                        target_exec = target_details[3]
 
2024
                    return (entry[0][2],
 
2025
                           (None, utf8_decode(path)[0]),
 
2026
                           True,
 
2027
                           (False, True),
 
2028
                           (None, parent_id),
 
2029
                           (None, utf8_decode(entry[0][1])[0]),
 
2030
                           (None, path_info[2]),
 
2031
                           (None, target_exec))
 
2032
                else:
 
2033
                    # but its not on disk: we deliberately treat this as just
 
2034
                    # never-present. (Why ?! - RBC 20070224)
 
2035
                    pass
 
2036
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2037
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2038
                # if its still on disk, *and* theres no other entry at this
 
2039
                # path [we dont know this in this routine at the moment -
 
2040
                # perhaps we should change this - then it would be an unknown.
 
2041
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2042
                # parent id is the entry for the path in the target tree
 
2043
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2044
                if parent_id == entry[0][2]:
 
2045
                    parent_id = None
 
2046
                return (entry[0][2],
 
2047
                       (utf8_decode(old_path)[0], None),
 
2048
                       True,
 
2049
                       (True, False),
 
2050
                       (parent_id, None),
 
2051
                       (utf8_decode(entry[0][1])[0], None),
 
2052
                       (_minikind_to_kind[source_minikind], None),
 
2053
                       (source_details[3], None))
 
2054
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2055
                # a rename; could be a true rename, or a rename inherited from
 
2056
                # a renamed parent. TODO: handle this efficiently. Its not
 
2057
                # common case to rename dirs though, so a correct but slow
 
2058
                # implementation will do.
 
2059
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2060
                    search_specific_files.add(target_details[1])
 
2061
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2062
                # neither of the selected trees contain this file,
 
2063
                # so skip over it. This is not currently directly tested, but
 
2064
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2065
                pass
 
2066
            else:
 
2067
                raise AssertionError("don't know how to compare "
 
2068
                    "source_minikind=%r, target_minikind=%r"
 
2069
                    % (source_minikind, target_minikind))
 
2070
                ## import pdb;pdb.set_trace()
 
2071
            return None
 
2072
 
 
2073
        while search_specific_files:
 
2074
            # TODO: the pending list should be lexically sorted?  the
 
2075
            # interface doesn't require it.
 
2076
            current_root = search_specific_files.pop()
 
2077
            current_root_unicode = current_root.decode('utf8')
 
2078
            searched_specific_files.add(current_root)
 
2079
            # process the entries for this containing directory: the rest will be
 
2080
            # found by their parents recursively.
 
2081
            root_entries = _entries_for_path(current_root)
 
2082
            root_abspath = self.target.abspath(current_root_unicode)
 
2083
            try:
 
2084
                root_stat = os.lstat(root_abspath)
 
2085
            except OSError, e:
 
2086
                if e.errno == errno.ENOENT:
 
2087
                    # the path does not exist: let _process_entry know that.
 
2088
                    root_dir_info = None
 
2089
                else:
 
2090
                    # some other random error: hand it up.
 
2091
                    raise
 
2092
            else:
 
2093
                root_dir_info = ('', current_root,
 
2094
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2095
                    root_abspath)
 
2096
                if root_dir_info[2] == 'directory':
 
2097
                    if self.target._directory_is_tree_reference(
 
2098
                        current_root.decode('utf8')):
 
2099
                        root_dir_info = root_dir_info[:2] + \
 
2100
                            ('tree-reference',) + root_dir_info[3:]
 
2101
 
 
2102
            if not root_entries and not root_dir_info:
 
2103
                # this specified path is not present at all, skip it.
 
2104
                continue
 
2105
            path_handled = False
 
2106
            for entry in root_entries:
 
2107
                result = _process_entry(entry, root_dir_info)
 
2108
                if result is not None:
 
2109
                    path_handled = True
 
2110
                    if result is not uninteresting:
 
2111
                        yield result
 
2112
            if want_unversioned and not path_handled and root_dir_info:
 
2113
                new_executable = bool(
 
2114
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2115
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2116
                yield (None,
 
2117
                       (None, current_root_unicode),
 
2118
                       True,
 
2119
                       (False, False),
 
2120
                       (None, None),
 
2121
                       (None, splitpath(current_root_unicode)[-1]),
 
2122
                       (None, root_dir_info[2]),
 
2123
                       (None, new_executable)
 
2124
                      )
 
2125
            initial_key = (current_root, '', '')
 
2126
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2127
            if block_index == 0:
 
2128
                # we have processed the total root already, but because the
 
2129
                # initial key matched it we should skip it here.
 
2130
                block_index +=1
 
2131
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2132
                current_dir_info = None
 
2133
            else:
 
2134
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2135
                try:
 
2136
                    current_dir_info = dir_iterator.next()
 
2137
                except OSError, e:
 
2138
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2139
                    # python 2.5 has e.errno == EINVAL,
 
2140
                    #            and e.winerror == ERROR_DIRECTORY
 
2141
                    e_winerror = getattr(e, 'winerror', None)
 
2142
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2143
                    # there may be directories in the inventory even though
 
2144
                    # this path is not a file on disk: so mark it as end of
 
2145
                    # iterator
 
2146
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2147
                        current_dir_info = None
 
2148
                    elif (sys.platform == 'win32'
 
2149
                          and (e.errno in win_errors
 
2150
                               or e_winerror in win_errors)):
 
2151
                        current_dir_info = None
 
2152
                    else:
 
2153
                        raise
 
2154
                else:
 
2155
                    if current_dir_info[0][0] == '':
 
2156
                        # remove .bzr from iteration
 
2157
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2158
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2159
                        del current_dir_info[1][bzr_index]
 
2160
            # walk until both the directory listing and the versioned metadata
 
2161
            # are exhausted. 
 
2162
            if (block_index < len(state._dirblocks) and
 
2163
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2164
                current_block = state._dirblocks[block_index]
 
2165
            else:
 
2166
                current_block = None
 
2167
            while (current_dir_info is not None or
 
2168
                   current_block is not None):
 
2169
                if (current_dir_info and current_block
 
2170
                    and current_dir_info[0][0] != current_block[0]):
 
2171
                    if current_dir_info[0][0].split('/') < current_block[0].split('/'):
 
2172
                        # filesystem data refers to paths not covered by the dirblock.
 
2173
                        # this has two possibilities:
 
2174
                        # A) it is versioned but empty, so there is no block for it
 
2175
                        # B) it is not versioned.
 
2176
 
 
2177
                        # if (A) then we need to recurse into it to check for
 
2178
                        # new unknown files or directories.
 
2179
                        # if (B) then we should ignore it, because we don't
 
2180
                        # recurse into unknown directories.
 
2181
                        path_index = 0
 
2182
                        while path_index < len(current_dir_info[1]):
 
2183
                                current_path_info = current_dir_info[1][path_index]
 
2184
                                if want_unversioned:
 
2185
                                    if current_path_info[2] == 'directory':
 
2186
                                        if self.target._directory_is_tree_reference(
 
2187
                                            current_path_info[0].decode('utf8')):
 
2188
                                            current_path_info = current_path_info[:2] + \
 
2189
                                                ('tree-reference',) + current_path_info[3:]
 
2190
                                    new_executable = bool(
 
2191
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2192
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2193
                                    yield (None,
 
2194
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2195
                                        True,
 
2196
                                        (False, False),
 
2197
                                        (None, None),
 
2198
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2199
                                        (None, current_path_info[2]),
 
2200
                                        (None, new_executable))
 
2201
                                # dont descend into this unversioned path if it is
 
2202
                                # a dir
 
2203
                                if current_path_info[2] in ('directory',
 
2204
                                                            'tree-reference'):
 
2205
                                    del current_dir_info[1][path_index]
 
2206
                                    path_index -= 1
 
2207
                                path_index += 1
 
2208
 
 
2209
                        # This dir info has been handled, go to the next
 
2210
                        try:
 
2211
                            current_dir_info = dir_iterator.next()
 
2212
                        except StopIteration:
 
2213
                            current_dir_info = None
 
2214
                    else:
 
2215
                        # We have a dirblock entry for this location, but there
 
2216
                        # is no filesystem path for this. This is most likely
 
2217
                        # because a directory was removed from the disk.
 
2218
                        # We don't have to report the missing directory,
 
2219
                        # because that should have already been handled, but we
 
2220
                        # need to handle all of the files that are contained
 
2221
                        # within.
 
2222
                        for current_entry in current_block[1]:
 
2223
                            # entry referring to file not present on disk.
 
2224
                            # advance the entry only, after processing.
 
2225
                            result = _process_entry(current_entry, None)
 
2226
                            if result is not None:
 
2227
                                if result is not uninteresting:
 
2228
                                    yield result
 
2229
                        block_index +=1
 
2230
                        if (block_index < len(state._dirblocks) and
 
2231
                            osutils.is_inside(current_root,
 
2232
                                              state._dirblocks[block_index][0])):
 
2233
                            current_block = state._dirblocks[block_index]
 
2234
                        else:
 
2235
                            current_block = None
 
2236
                    continue
 
2237
                entry_index = 0
 
2238
                if current_block and entry_index < len(current_block[1]):
 
2239
                    current_entry = current_block[1][entry_index]
 
2240
                else:
 
2241
                    current_entry = None
 
2242
                advance_entry = True
 
2243
                path_index = 0
 
2244
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2245
                    current_path_info = current_dir_info[1][path_index]
 
2246
                    if current_path_info[2] == 'directory':
 
2247
                        if self.target._directory_is_tree_reference(
 
2248
                            current_path_info[0].decode('utf8')):
 
2249
                            current_path_info = current_path_info[:2] + \
 
2250
                                ('tree-reference',) + current_path_info[3:]
 
2251
                else:
 
2252
                    current_path_info = None
 
2253
                advance_path = True
 
2254
                path_handled = False
 
2255
                while (current_entry is not None or
 
2256
                    current_path_info is not None):
 
2257
                    if current_entry is None:
 
2258
                        # the check for path_handled when the path is adnvaced
 
2259
                        # will yield this path if needed.
 
2260
                        pass
 
2261
                    elif current_path_info is None:
 
2262
                        # no path is fine: the per entry code will handle it.
 
2263
                        result = _process_entry(current_entry, current_path_info)
 
2264
                        if result is not None:
 
2265
                            if result is not uninteresting:
 
2266
                                yield result
 
2267
                    elif (current_entry[0][1] != current_path_info[1]
 
2268
                          or current_entry[1][target_index][0] in 'ar'):
 
2269
                        # The current path on disk doesn't match the dirblock
 
2270
                        # record. Either the dirblock is marked as absent, or
 
2271
                        # the file on disk is not present at all in the
 
2272
                        # dirblock. Either way, report about the dirblock
 
2273
                        # entry, and let other code handle the filesystem one.
 
2274
 
 
2275
                        # Compare the basename for these files to determine
 
2276
                        # which comes first
 
2277
                        if current_path_info[1] < current_entry[0][1]:
 
2278
                            # extra file on disk: pass for now, but only
 
2279
                            # increment the path, not the entry
 
2280
                            advance_entry = False
 
2281
                        else:
 
2282
                            # entry referring to file not present on disk.
 
2283
                            # advance the entry only, after processing.
 
2284
                            result = _process_entry(current_entry, None)
 
2285
                            if result is not None:
 
2286
                                if result is not uninteresting:
 
2287
                                    yield result
 
2288
                            advance_path = False
 
2289
                    else:
 
2290
                        result = _process_entry(current_entry, current_path_info)
 
2291
                        if result is not None:
 
2292
                            path_handled = True
 
2293
                            if result is not uninteresting:
 
2294
                                yield result
 
2295
                    if advance_entry and current_entry is not None:
 
2296
                        entry_index += 1
 
2297
                        if entry_index < len(current_block[1]):
 
2298
                            current_entry = current_block[1][entry_index]
 
2299
                        else:
 
2300
                            current_entry = None
 
2301
                    else:
 
2302
                        advance_entry = True # reset the advance flaga
 
2303
                    if advance_path and current_path_info is not None:
 
2304
                        if not path_handled:
 
2305
                            # unversioned in all regards
 
2306
                            if want_unversioned:
 
2307
                                new_executable = bool(
 
2308
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2309
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2310
                                yield (None,
 
2311
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2312
                                    True,
 
2313
                                    (False, False),
 
2314
                                    (None, None),
 
2315
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2316
                                    (None, current_path_info[2]),
 
2317
                                    (None, new_executable))
 
2318
                            # dont descend into this unversioned path if it is
 
2319
                            # a dir
 
2320
                            if current_path_info[2] in ('directory'):
 
2321
                                del current_dir_info[1][path_index]
 
2322
                                path_index -= 1
 
2323
                        # dont descend the disk iterator into any tree 
 
2324
                        # paths.
 
2325
                        if current_path_info[2] == 'tree-reference':
 
2326
                            del current_dir_info[1][path_index]
 
2327
                            path_index -= 1
 
2328
                        path_index += 1
 
2329
                        if path_index < len(current_dir_info[1]):
 
2330
                            current_path_info = current_dir_info[1][path_index]
 
2331
                            if current_path_info[2] == 'directory':
 
2332
                                if self.target._directory_is_tree_reference(
 
2333
                                    current_path_info[0].decode('utf8')):
 
2334
                                    current_path_info = current_path_info[:2] + \
 
2335
                                        ('tree-reference',) + current_path_info[3:]
 
2336
                        else:
 
2337
                            current_path_info = None
 
2338
                        path_handled = False
 
2339
                    else:
 
2340
                        advance_path = True # reset the advance flagg.
 
2341
                if current_block is not None:
 
2342
                    block_index += 1
 
2343
                    if (block_index < len(state._dirblocks) and
 
2344
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2345
                        current_block = state._dirblocks[block_index]
 
2346
                    else:
 
2347
                        current_block = None
 
2348
                if current_dir_info is not None:
 
2349
                    try:
 
2350
                        current_dir_info = dir_iterator.next()
 
2351
                    except StopIteration:
 
2352
                        current_dir_info = None
 
2353
 
2116
2354
 
2117
2355
    @staticmethod
2118
2356
    def is_compatible(source, target):
2119
2357
        # the target must be a dirstate working tree
2120
 
        if not isinstance(target, DirStateWorkingTree):
 
2358
        if not isinstance(target, WorkingTree4):
2121
2359
            return False
2122
 
        # the source must be a revtree or dirstate rev tree.
 
2360
        # the source must be a revtreee or dirstate rev tree.
2123
2361
        if not isinstance(source,
2124
2362
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2125
2363
            return False
2126
2364
        # the source revid must be in the target dirstate
2127
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2365
        if not (source._revision_id == NULL_REVISION or
2128
2366
            source._revision_id in target.get_parent_ids()):
2129
 
            # TODO: what about ghosts? it may well need to
 
2367
            # TODO: what about ghosts? it may well need to 
2130
2368
            # check for them explicitly.
2131
2369
            return False
2132
2370
        return True
2142
2380
 
2143
2381
    def convert(self, tree):
2144
2382
        # lock the control files not the tree, so that we dont get tree
2145
 
        # on-unlock behaviours, and so that noone else diddles with the
 
2383
        # on-unlock behaviours, and so that noone else diddles with the 
2146
2384
        # tree during upgrade.
2147
2385
        tree._control_files.lock_write()
2148
2386
        try:
2174
2412
 
2175
2413
    def update_format(self, tree):
2176
2414
        """Change the format marker."""
2177
 
        tree._transport.put_bytes('format',
2178
 
            self.target_format.get_format_string(),
2179
 
            mode=tree.bzrdir._get_file_mode())
2180
 
 
2181
 
 
2182
 
class Converter4to5(object):
2183
 
    """Perform an in-place upgrade of format 4 to format 5 trees."""
2184
 
 
2185
 
    def __init__(self):
2186
 
        self.target_format = WorkingTreeFormat5()
2187
 
 
2188
 
    def convert(self, tree):
2189
 
        # lock the control files not the tree, so that we don't get tree
2190
 
        # on-unlock behaviours, and so that no-one else diddles with the
2191
 
        # tree during upgrade.
2192
 
        tree._control_files.lock_write()
2193
 
        try:
2194
 
            self.update_format(tree)
2195
 
        finally:
2196
 
            tree._control_files.unlock()
2197
 
 
2198
 
    def update_format(self, tree):
2199
 
        """Change the format marker."""
2200
 
        tree._transport.put_bytes('format',
2201
 
            self.target_format.get_format_string(),
2202
 
            mode=tree.bzrdir._get_file_mode())
2203
 
 
2204
 
 
2205
 
class Converter4or5to6(object):
2206
 
    """Perform an in-place upgrade of format 4 or 5 to format 6 trees."""
2207
 
 
2208
 
    def __init__(self):
2209
 
        self.target_format = WorkingTreeFormat6()
2210
 
 
2211
 
    def convert(self, tree):
2212
 
        # lock the control files not the tree, so that we don't get tree
2213
 
        # on-unlock behaviours, and so that no-one else diddles with the
2214
 
        # tree during upgrade.
2215
 
        tree._control_files.lock_write()
2216
 
        try:
2217
 
            self.init_custom_control_files(tree)
2218
 
            self.update_format(tree)
2219
 
        finally:
2220
 
            tree._control_files.unlock()
2221
 
 
2222
 
    def init_custom_control_files(self, tree):
2223
 
        """Initialize custom control files."""
2224
 
        tree._transport.put_bytes('views', '',
2225
 
            mode=tree.bzrdir._get_file_mode())
2226
 
 
2227
 
    def update_format(self, tree):
2228
 
        """Change the format marker."""
2229
 
        tree._transport.put_bytes('format',
2230
 
            self.target_format.get_format_string(),
2231
 
            mode=tree.bzrdir._get_file_mode())
 
2415
        tree._control_files.put_utf8('format',
 
2416
            self.target_format.get_format_string())