/brz/remove-bazaar

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

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: John Arbash Meinel
  • Date: 2007-07-12 05:15:03 UTC
  • mto: This revision was merged to the branch mainline in revision 2643.
  • Revision ID: john@arbash-meinel.com-20070712051503-ntboo0z3prcrcg3t
Fix dirstate benchmarks for new layout.

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
 
    revision as _mod_revision,
44
55
    revisiontree,
45
 
    trace,
 
56
    textui,
46
57
    transform,
47
 
    views,
 
58
    urlutils,
 
59
    xml5,
 
60
    xml6,
48
61
    )
49
62
import bzrlib.branch
 
63
from bzrlib.transport import get_transport
50
64
import bzrlib.ui
51
65
""")
52
66
 
 
67
from bzrlib import symbol_versioning
53
68
from bzrlib.decorators import needs_read_lock, needs_write_lock
54
 
from bzrlib.filters import filtered_input_file, internal_size_sha_file_byname
55
 
from bzrlib.inventory import Inventory, ROOT_ID, entry_factory
56
 
from bzrlib.lock import LogicalLockResult
 
69
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
70
from bzrlib.lockable_files import LockableFiles, TransportLock
 
71
from bzrlib.lockdir import LockDir
 
72
import bzrlib.mutabletree
57
73
from bzrlib.mutabletree import needs_tree_write_lock
58
74
from bzrlib.osutils import (
59
75
    file_kind,
60
76
    isdir,
 
77
    normpath,
61
78
    pathjoin,
 
79
    rand_chars,
62
80
    realpath,
63
81
    safe_unicode,
 
82
    splitpath,
64
83
    )
65
 
from bzrlib.trace import mutter
 
84
from bzrlib.trace import mutter, note
66
85
from bzrlib.transport.local import LocalTransport
67
86
from bzrlib.tree import InterTree
 
87
from bzrlib.progress import DummyProgress, ProgressPhase
 
88
from bzrlib.revision import NULL_REVISION, CURRENT_REVISION
 
89
from bzrlib.rio import RioReader, rio_file, Stanza
 
90
from bzrlib.symbol_versioning import (deprecated_passed,
 
91
        deprecated_method,
 
92
        deprecated_function,
 
93
        DEPRECATED_PARAMETER,
 
94
        )
68
95
from bzrlib.tree import Tree
69
96
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
70
97
 
71
98
 
72
 
class DirStateWorkingTree(WorkingTree3):
 
99
# This is the Windows equivalent of ENOTDIR
 
100
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
101
# just an error code.
 
102
ERROR_PATH_NOT_FOUND = 3
 
103
ERROR_DIRECTORY = 267
 
104
 
 
105
 
 
106
class WorkingTree4(WorkingTree3):
 
107
    """This is the Format 4 working tree.
 
108
 
 
109
    This differs from WorkingTree3 by:
 
110
     - Having a consolidated internal dirstate, stored in a
 
111
       randomly-accessible sorted file on disk.
 
112
     - Not having a regular inventory attribute.  One can be synthesized 
 
113
       on demand but this is expensive and should be avoided.
 
114
 
 
115
    This is new in bzr 0.15.
 
116
    """
 
117
 
73
118
    def __init__(self, basedir,
74
119
                 branch,
75
120
                 _control_files=None,
84
129
        """
85
130
        self._format = _format
86
131
        self.bzrdir = _bzrdir
 
132
        from bzrlib.trace import note, mutter
 
133
        assert isinstance(basedir, basestring), \
 
134
            "base directory %r is not a string" % basedir
87
135
        basedir = safe_unicode(basedir)
88
136
        mutter("opening working tree %r", basedir)
89
137
        self._branch = branch
 
138
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
139
            "branch %r is not a Branch" % self.branch
90
140
        self.basedir = realpath(basedir)
91
141
        # if branch is at our basedir and is a format 6 or less
92
142
        # assume all other formats have their own control files.
 
143
        assert isinstance(_control_files, LockableFiles), \
 
144
            "_control_files must be a LockableFiles, not %r" % _control_files
93
145
        self._control_files = _control_files
94
 
        self._transport = self._control_files._transport
95
146
        self._dirty = None
96
147
        #-------------
97
148
        # during a read or write lock these objects are set, and are
99
150
        self._dirstate = None
100
151
        self._inventory = None
101
152
        #-------------
102
 
        self._setup_directory_is_tree_reference()
103
 
        self._detect_case_handling()
104
 
        self._rules_searcher = None
105
 
        self.views = self._make_views()
106
 
        #--- allow tests to select the dirstate iter_changes implementation
107
 
        self._iter_changes = dirstate._process_entry
108
153
 
109
154
    @needs_tree_write_lock
110
155
    def _add(self, files, ids, kinds):
112
157
        state = self.current_dirstate()
113
158
        for f, file_id, kind in zip(files, ids, kinds):
114
159
            f = f.strip('/')
 
160
            assert '//' not in f
 
161
            assert '..' not in f
115
162
            if self.path2id(f):
116
163
                # special case tree root handling.
117
164
                if f == '' and self.path2id(f) == ROOT_ID:
138
185
    @needs_tree_write_lock
139
186
    def add_reference(self, sub_tree):
140
187
        # use standard implementation, which calls back to self._add
141
 
        #
 
188
        # 
142
189
        # So we don't store the reference_revision in the working dirstate,
143
 
        # it's just recorded at the moment of commit.
 
190
        # it's just recorded at the moment of commit. 
144
191
        self._add_reference(sub_tree)
145
192
 
146
193
    def break_lock(self):
185
232
            WorkingTree3._comparison_data(self, entry, path)
186
233
        # it looks like a plain directory, but it's really a reference -- see
187
234
        # also kind()
188
 
        if (self._repo_supports_tree_reference and kind == 'directory'
189
 
            and entry is not None and entry.kind == 'tree-reference'):
 
235
        if (self._repo_supports_tree_reference and
 
236
            kind == 'directory' and
 
237
            self._directory_is_tree_reference(path)):
190
238
            kind = 'tree-reference'
191
239
        return kind, executable, stat_value
192
240
 
218
266
            return self._dirstate
219
267
        local_path = self.bzrdir.get_workingtree_transport(None
220
268
            ).local_abspath('dirstate')
221
 
        self._dirstate = dirstate.DirState.on_file(local_path,
222
 
            self._sha1_provider())
 
269
        self._dirstate = dirstate.DirState.on_file(local_path)
223
270
        return self._dirstate
224
271
 
225
 
    def _sha1_provider(self):
226
 
        """A function that returns a SHA1Provider suitable for this tree.
227
 
 
228
 
        :return: None if content filtering is not supported by this tree.
229
 
          Otherwise, a SHA1Provider is returned that sha's the canonical
230
 
          form of files, i.e. after read filters are applied.
231
 
        """
232
 
        if self.supports_content_filtering():
233
 
            return ContentFilterAwareSHA1Provider(self)
234
 
        else:
235
 
            return None
 
272
    def _directory_is_tree_reference(self, relpath):
 
273
        # as a special case, if a directory contains control files then 
 
274
        # it's a tree reference, except that the root of the tree is not
 
275
        return relpath and osutils.isdir(self.abspath(relpath) + u"/.bzr")
 
276
        # TODO: We could ask all the control formats whether they
 
277
        # recognize this directory, but at the moment there's no cheap api
 
278
        # to do that.  Since we probably can only nest bzr checkouts and
 
279
        # they always use this name it's ok for now.  -- mbp 20060306
 
280
        #
 
281
        # FIXME: There is an unhandled case here of a subdirectory
 
282
        # containing .bzr but not a branch; that will probably blow up
 
283
        # when you try to commit it.  It might happen if there is a
 
284
        # checkout in a subdirectory.  This can be avoided by not adding
 
285
        # it.  mbp 20070306
236
286
 
237
287
    def filter_unversioned_files(self, paths):
238
288
        """Filter out paths that are versioned.
271
321
 
272
322
    def _generate_inventory(self):
273
323
        """Create and set self.inventory from the dirstate object.
274
 
 
 
324
        
275
325
        This is relatively expensive: we have to walk the entire dirstate.
276
326
        Ideally we would not, and can deprecate this function.
277
327
        """
281
331
        state._read_dirblocks_if_needed()
282
332
        root_key, current_entry = self._get_entry(path='')
283
333
        current_id = root_key[2]
284
 
        if not (current_entry[0][0] == 'd'): # directory
285
 
            raise AssertionError(current_entry)
 
334
        assert current_entry[0][0] == 'd' # directory
286
335
        inv = Inventory(root_id=current_id)
287
336
        # Turn some things into local variables
288
337
        minikind_to_kind = dirstate.DirState._minikind_to_kind
321
370
                    # add this entry to the parent map.
322
371
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
323
372
                elif kind == 'tree-reference':
324
 
                    if not self._repo_supports_tree_reference:
325
 
                        raise errors.UnsupportedOperation(
326
 
                            self._generate_inventory,
327
 
                            self.branch.repository)
 
373
                    assert self._repo_supports_tree_reference, \
 
374
                        "repository of %r " \
 
375
                        "doesn't support tree references " \
 
376
                        "required by entry %r" \
 
377
                        % (self, name)
328
378
                    inv_entry.reference_revision = link_or_sha1 or None
329
379
                elif kind != 'symlink':
330
380
                    raise AssertionError("unknown kind %r" % kind)
331
381
                # These checks cost us around 40ms on a 55k entry tree
332
 
                if file_id in inv_byid:
333
 
                    raise AssertionError('file_id %s already in'
334
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
335
 
                if name_unicode in parent_ie.children:
336
 
                    raise AssertionError('name %r already in parent'
337
 
                        % (name_unicode,))
 
382
                assert file_id not in inv_byid, ('file_id %s already in'
 
383
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
384
                assert name_unicode not in parent_ie.children
338
385
                inv_byid[file_id] = inv_entry
339
386
                parent_ie.children[name_unicode] = inv_entry
340
387
        self._inventory = inv
345
392
        If either file_id or path is supplied, it is used as the key to lookup.
346
393
        If both are supplied, the fastest lookup is used, and an error is
347
394
        raised if they do not both point at the same row.
348
 
 
 
395
        
349
396
        :param file_id: An optional unicode file_id to be looked up.
350
397
        :param path: An optional unicode path to be looked up.
351
398
        :return: The dirstate row tuple for path/file_id, or (None, None)
360
407
    def get_file_sha1(self, file_id, path=None, stat_value=None):
361
408
        # check file id is valid unconditionally.
362
409
        entry = self._get_entry(file_id=file_id, path=path)
363
 
        if entry[0] is None:
364
 
            raise errors.NoSuchId(self, file_id)
 
410
        assert entry[0] is not None, 'what error should this raise'
365
411
        if path is None:
366
412
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
367
413
 
375
421
                    return None
376
422
                else:
377
423
                    raise
378
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
379
 
            stat_value=stat_value)
 
424
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
425
                                          stat_value=stat_value)
380
426
        if entry[1][0][0] == 'f':
381
 
            if link_or_sha1 is None:
382
 
                file_obj, statvalue = self.get_file_with_stat(file_id, path)
383
 
                try:
384
 
                    sha1 = osutils.sha_file(file_obj)
385
 
                finally:
386
 
                    file_obj.close()
387
 
                self._observed_sha1(file_id, path, (sha1, statvalue))
388
 
                return sha1
389
 
            else:
390
 
                return link_or_sha1
 
427
            return link_or_sha1
391
428
        return None
392
429
 
393
430
    def _get_inventory(self):
394
431
        """Get the inventory for the tree. This is only valid within a lock."""
395
 
        if 'evil' in debug.debug_flags:
396
 
            trace.mutter_callsite(2,
397
 
                "accessing .inventory forces a size of tree translation.")
398
432
        if self._inventory is not None:
399
433
            return self._inventory
400
434
        self._must_be_locked()
407
441
    @needs_read_lock
408
442
    def get_parent_ids(self):
409
443
        """See Tree.get_parent_ids.
410
 
 
 
444
        
411
445
        This implementation requests the ids list from the dirstate file.
412
446
        """
413
447
        return self.current_dirstate().get_parent_ids()
429
463
 
430
464
    def has_id(self, file_id):
431
465
        state = self.current_dirstate()
 
466
        file_id = osutils.safe_file_id(file_id)
432
467
        row, parents = self._get_entry(file_id=file_id)
433
468
        if row is None:
434
469
            return False
435
470
        return osutils.lexists(pathjoin(
436
471
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
437
472
 
438
 
    def has_or_had_id(self, file_id):
439
 
        state = self.current_dirstate()
440
 
        row, parents = self._get_entry(file_id=file_id)
441
 
        return row is not None
442
 
 
443
473
    @needs_read_lock
444
474
    def id2path(self, file_id):
445
475
        "Convert a file-id to a path."
 
476
        file_id = osutils.safe_file_id(file_id)
446
477
        state = self.current_dirstate()
447
478
        entry = self._get_entry(file_id=file_id)
448
479
        if entry == (None, None):
450
481
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
451
482
        return path_utf8.decode('utf8')
452
483
 
453
 
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
454
 
        entry = self._get_entry(path=path)
455
 
        if entry == (None, None):
456
 
            return False # Missing entries are not executable
457
 
        return entry[1][0][3] # Executable?
458
 
 
459
484
    if not osutils.supports_executable():
460
485
        def is_executable(self, file_id, path=None):
461
486
            """Test if a file is executable or not.
462
487
 
463
488
            Note: The caller is expected to take a read-lock before calling this.
464
489
            """
 
490
            file_id = osutils.safe_file_id(file_id)
465
491
            entry = self._get_entry(file_id=file_id, path=path)
466
492
            if entry == (None, None):
467
493
                return False
468
494
            return entry[1][0][3]
469
 
 
470
 
        _is_executable_from_path_and_stat = \
471
 
            _is_executable_from_path_and_stat_from_basis
472
495
    else:
473
496
        def is_executable(self, file_id, path=None):
474
497
            """Test if a file is executable or not.
475
498
 
476
499
            Note: The caller is expected to take a read-lock before calling this.
477
500
            """
478
 
            self._must_be_locked()
479
501
            if not path:
 
502
                file_id = osutils.safe_file_id(file_id)
480
503
                path = self.id2path(file_id)
481
504
            mode = os.lstat(self.abspath(path)).st_mode
482
505
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
483
506
 
484
 
    def all_file_ids(self):
485
 
        """See Tree.iter_all_file_ids"""
486
 
        self._must_be_locked()
487
 
        result = set()
488
 
        for key, tree_details in self.current_dirstate()._iter_entries():
489
 
            if tree_details[0][0] in ('a', 'r'): # relocated
490
 
                continue
491
 
            result.add(key[2])
492
 
        return result
493
 
 
494
507
    @needs_read_lock
495
508
    def __iter__(self):
496
509
        """Iterate through file_ids for this tree.
509
522
        return iter(result)
510
523
 
511
524
    def iter_references(self):
512
 
        if not self._repo_supports_tree_reference:
513
 
            # When the repo doesn't support references, we will have nothing to
514
 
            # return
515
 
            return
516
525
        for key, tree_details in self.current_dirstate()._iter_entries():
517
526
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
518
527
                # not relevant to the working tree
520
529
            if not key[1]:
521
530
                # the root is not a reference.
522
531
                continue
523
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
532
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
524
533
            try:
525
 
                if self._kind(relpath) == 'tree-reference':
526
 
                    yield relpath, key[2]
 
534
                if self._kind(path) == 'tree-reference':
 
535
                    yield path, key[2]
527
536
            except errors.NoSuchFile:
528
537
                # path is missing on disk.
529
538
                continue
530
539
 
531
 
    def _observed_sha1(self, file_id, path, (sha1, statvalue)):
532
 
        """See MutableTree._observed_sha1."""
533
 
        state = self.current_dirstate()
534
 
        entry = self._get_entry(file_id=file_id, path=path)
535
 
        state._observed_sha1(entry, sha1, statvalue)
536
 
 
537
540
    def kind(self, file_id):
538
541
        """Return the kind of a file.
539
542
 
543
546
        Note: The caller is expected to take a read-lock before calling this.
544
547
        """
545
548
        relpath = self.id2path(file_id)
546
 
        if relpath is None:
547
 
            raise AssertionError(
548
 
                "path for id {%s} is None!" % file_id)
 
549
        assert relpath != None, \
 
550
            "path for id {%s} is None!" % file_id
549
551
        return self._kind(relpath)
550
552
 
551
553
    def _kind(self, relpath):
552
554
        abspath = self.abspath(relpath)
553
555
        kind = file_kind(abspath)
554
 
        if (self._repo_supports_tree_reference and kind == 'directory'):
555
 
            entry = self._get_entry(path=relpath)
556
 
            if entry[1] is not None:
557
 
                if entry[1][0][0] == 't':
558
 
                    kind = 'tree-reference'
 
556
        if (self._repo_supports_tree_reference and
 
557
            kind == 'directory' and
 
558
            self._directory_is_tree_reference(relpath)):
 
559
            kind = 'tree-reference'
559
560
        return kind
560
561
 
561
562
    @needs_read_lock
565
566
        if parent_ids:
566
567
            return parent_ids[0]
567
568
        else:
568
 
            return _mod_revision.NULL_REVISION
 
569
            return None
569
570
 
570
571
    def lock_read(self):
571
 
        """See Branch.lock_read, and WorkingTree.unlock.
572
 
 
573
 
        :return: A bzrlib.lock.LogicalLockResult.
574
 
        """
 
572
        """See Branch.lock_read, and WorkingTree.unlock."""
575
573
        self.branch.lock_read()
576
574
        try:
577
575
            self._control_files.lock_read()
590
588
        except:
591
589
            self.branch.unlock()
592
590
            raise
593
 
        return LogicalLockResult(self.unlock)
594
591
 
595
592
    def _lock_self_write(self):
596
593
        """This should be called after the branch is locked."""
611
608
        except:
612
609
            self.branch.unlock()
613
610
            raise
614
 
        return LogicalLockResult(self.unlock)
615
611
 
616
612
    def lock_tree_write(self):
617
 
        """See MutableTree.lock_tree_write, and WorkingTree.unlock.
618
 
 
619
 
        :return: A bzrlib.lock.LogicalLockResult.
620
 
        """
 
613
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
621
614
        self.branch.lock_read()
622
 
        return self._lock_self_write()
 
615
        self._lock_self_write()
623
616
 
624
617
    def lock_write(self):
625
 
        """See MutableTree.lock_write, and WorkingTree.unlock.
626
 
 
627
 
        :return: A bzrlib.lock.LogicalLockResult.
628
 
        """
 
618
        """See MutableTree.lock_write, and WorkingTree.unlock."""
629
619
        self.branch.lock_write()
630
 
        return self._lock_self_write()
 
620
        self._lock_self_write()
631
621
 
632
622
    @needs_tree_write_lock
633
623
    def move(self, from_paths, to_dir, after=False):
635
625
        result = []
636
626
        if not from_paths:
637
627
            return result
 
628
 
638
629
        state = self.current_dirstate()
639
 
        if isinstance(from_paths, basestring):
640
 
            raise ValueError()
 
630
 
 
631
        assert not isinstance(from_paths, basestring)
641
632
        to_dir_utf8 = to_dir.encode('utf8')
642
633
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
643
634
        id_index = state._get_id_index()
665
656
        if self._inventory is not None:
666
657
            update_inventory = True
667
658
            inv = self.inventory
 
659
            to_dir_ie = inv[to_dir_id]
668
660
            to_dir_id = to_entry[0][2]
669
 
            to_dir_ie = inv[to_dir_id]
670
661
        else:
671
662
            update_inventory = False
672
663
 
703
694
            from_entry = self._get_entry(path=from_rel)
704
695
            if from_entry == (None, None):
705
696
                raise errors.BzrMoveFailedError(from_rel,to_dir,
706
 
                    errors.NotVersionedError(path=from_rel))
 
697
                    errors.NotVersionedError(path=str(from_rel)))
707
698
 
708
699
            from_id = from_entry[0][2]
709
700
            to_rel = pathjoin(to_dir, from_tail)
736
727
                if from_missing: # implicitly just update our path mapping
737
728
                    move_file = False
738
729
                elif not after:
739
 
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
730
                    raise errors.RenameFailedFilesExist(from_rel, to_rel,
 
731
                        extra="(Use --after to update the Bazaar id)")
740
732
 
741
733
            rollbacks = []
742
734
            def rollback_rename():
798
790
                if minikind == 'd':
799
791
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
800
792
                        """Recursively update all entries in this dirblock."""
801
 
                        if from_dir == '':
802
 
                            raise AssertionError("renaming root not supported")
 
793
                        assert from_dir != '', "renaming root not supported"
803
794
                        from_key = (from_dir, '')
804
795
                        from_block_idx, present = \
805
796
                            state._find_block_index_from_key(from_key)
818
809
 
819
810
                        # Grab a copy since move_one may update the list.
820
811
                        for entry in from_block[1][:]:
821
 
                            if not (entry[0][0] == from_dir):
822
 
                                raise AssertionError()
 
812
                            assert entry[0][0] == from_dir
823
813
                            cur_details = entry[1][0]
824
814
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
825
815
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
884
874
        for tree in trees:
885
875
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
886
876
                parents):
887
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
888
 
                    trees, require_versioned)
 
877
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
889
878
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
890
879
        # -- make all paths utf8 --
891
880
        paths_utf8 = set()
951
940
            if not all_versioned:
952
941
                raise errors.PathsNotVersionedError(paths)
953
942
        # -- remove redundancy in supplied paths to prevent over-scanning --
954
 
        search_paths = osutils.minimum_path_selection(paths)
955
 
        # sketch:
 
943
        search_paths = set()
 
944
        for path in paths:
 
945
            other_paths = paths.difference(set([path]))
 
946
            if not osutils.is_inside_any(other_paths, path):
 
947
                # this is a top level path, we must check it.
 
948
                search_paths.add(path)
 
949
        # sketch: 
956
950
        # for all search_indexs in each path at or under each element of
957
951
        # search_paths, if the detail is relocated: add the id, and add the
958
952
        # relocated path as one to search if its not searched already. If the
1014
1008
 
1015
1009
    def read_working_inventory(self):
1016
1010
        """Read the working inventory.
1017
 
 
 
1011
        
1018
1012
        This is a meaningless operation for dirstate, but we obey it anyhow.
1019
1013
        """
1020
1014
        return self.inventory
1025
1019
 
1026
1020
        WorkingTree4 supplies revision_trees for any basis tree.
1027
1021
        """
 
1022
        revision_id = osutils.safe_revision_id(revision_id)
1028
1023
        dirstate = self.current_dirstate()
1029
1024
        parent_ids = dirstate.get_parent_ids()
1030
1025
        if revision_id not in parent_ids:
1037
1032
    @needs_tree_write_lock
1038
1033
    def set_last_revision(self, new_revision):
1039
1034
        """Change the last revision in the working tree."""
 
1035
        new_revision = osutils.safe_revision_id(new_revision)
1040
1036
        parents = self.get_parent_ids()
1041
 
        if new_revision in (_mod_revision.NULL_REVISION, None):
1042
 
            if len(parents) >= 2:
1043
 
                raise AssertionError(
1044
 
                    "setting the last parent to none with a pending merge is "
1045
 
                    "unsupported.")
 
1037
        if new_revision in (NULL_REVISION, None):
 
1038
            assert len(parents) < 2, (
 
1039
                "setting the last parent to none with a pending merge is "
 
1040
                "unsupported.")
1046
1041
            self.set_parent_ids([])
1047
1042
        else:
1048
1043
            self.set_parent_ids([new_revision] + parents[1:],
1051
1046
    @needs_tree_write_lock
1052
1047
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1053
1048
        """Set the parent ids to revision_ids.
1054
 
 
 
1049
        
1055
1050
        See also set_parent_trees. This api will try to retrieve the tree data
1056
1051
        for each element of revision_ids from the trees repository. If you have
1057
1052
        tree data already available, it is more efficient to use
1061
1056
        :param revision_ids: The revision_ids to set as the parent ids of this
1062
1057
            working tree. Any of these may be ghosts.
1063
1058
        """
 
1059
        revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1064
1060
        trees = []
1065
1061
        for revision_id in revision_ids:
1066
1062
            try:
1089
1085
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1090
1086
        real_trees = []
1091
1087
        ghosts = []
1092
 
 
1093
 
        parent_ids = [rev_id for rev_id, tree in parents_list]
1094
 
        graph = self.branch.repository.get_graph()
1095
 
        heads = graph.heads(parent_ids)
1096
 
        accepted_revisions = set()
1097
 
 
1098
1088
        # convert absent trees to the null tree, which we convert back to
1099
1089
        # missing on access.
1100
1090
        for rev_id, tree in parents_list:
1101
 
            if len(accepted_revisions) > 0:
1102
 
                # we always accept the first tree
1103
 
                if rev_id in accepted_revisions or rev_id not in heads:
1104
 
                    # We have already included either this tree, or its
1105
 
                    # descendent, so we skip it.
1106
 
                    continue
1107
 
            _mod_revision.check_not_reserved_id(rev_id)
 
1091
            rev_id = osutils.safe_revision_id(rev_id)
1108
1092
            if tree is not None:
1109
1093
                real_trees.append((rev_id, tree))
1110
1094
            else:
1111
1095
                real_trees.append((rev_id,
1112
 
                    self.branch.repository.revision_tree(
1113
 
                        _mod_revision.NULL_REVISION)))
 
1096
                    self.branch.repository.revision_tree(None)))
1114
1097
                ghosts.append(rev_id)
1115
 
            accepted_revisions.add(rev_id)
1116
1098
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
1117
1099
        self._make_dirty(reset_inventory=False)
1118
1100
 
1123
1105
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
1124
1106
            self._make_dirty(reset_inventory=True)
1125
1107
 
1126
 
    def _sha_from_stat(self, path, stat_result):
1127
 
        """Get a sha digest from the tree's stat cache.
1128
 
 
1129
 
        The default implementation assumes no stat cache is present.
1130
 
 
1131
 
        :param path: The path.
1132
 
        :param stat_result: The stat result being looked up.
1133
 
        """
1134
 
        return self.current_dirstate().sha1_from_stat(path, stat_result)
1135
 
 
1136
1108
    @needs_read_lock
1137
1109
    def supports_tree_reference(self):
1138
1110
        return self._repo_supports_tree_reference
1139
1111
 
1140
1112
    def unlock(self):
1141
1113
        """Unlock in format 4 trees needs to write the entire dirstate."""
1142
 
        # do non-implementation specific cleanup
1143
 
        self._cleanup()
1144
 
 
1145
1114
        if self._control_files._lock_count == 1:
1146
1115
            # eventually we should do signature checking during read locks for
1147
1116
            # dirstate updates.
1178
1147
            return
1179
1148
        state = self.current_dirstate()
1180
1149
        state._read_dirblocks_if_needed()
1181
 
        ids_to_unversion = set(file_ids)
 
1150
        ids_to_unversion = set()
 
1151
        for file_id in file_ids:
 
1152
            ids_to_unversion.add(osutils.safe_file_id(file_id))
1182
1153
        paths_to_unversion = set()
1183
1154
        # sketch:
1184
1155
        # check if the root is to be unversioned, if so, assert for now.
1211
1182
                # just forget the whole block.
1212
1183
                entry_index = 0
1213
1184
                while entry_index < len(block[1]):
 
1185
                    # Mark this file id as having been removed
1214
1186
                    entry = block[1][entry_index]
1215
 
                    if entry[1][0][0] in 'ar':
1216
 
                        # don't remove absent or renamed entries
 
1187
                    ids_to_unversion.discard(entry[0][2])
 
1188
                    if (entry[1][0][0] == 'a'
 
1189
                        or not state._make_absent(entry)):
1217
1190
                        entry_index += 1
1218
 
                    else:
1219
 
                        # Mark this file id as having been removed
1220
 
                        ids_to_unversion.discard(entry[0][2])
1221
 
                        if not state._make_absent(entry):
1222
 
                            # The block has not shrunk.
1223
 
                            entry_index += 1
1224
1191
                # go to the next block. (At the moment we dont delete empty
1225
1192
                # dirblocks)
1226
1193
                block_index += 1
1249
1216
            for file_id in file_ids:
1250
1217
                self._inventory.remove_recursive_id(file_id)
1251
1218
 
1252
 
    @needs_tree_write_lock
1253
 
    def rename_one(self, from_rel, to_rel, after=False):
1254
 
        """See WorkingTree.rename_one"""
1255
 
        self.flush()
1256
 
        WorkingTree.rename_one(self, from_rel, to_rel, after)
1257
 
 
1258
 
    @needs_tree_write_lock
1259
 
    def apply_inventory_delta(self, changes):
1260
 
        """See MutableTree.apply_inventory_delta"""
1261
 
        state = self.current_dirstate()
1262
 
        state.update_by_delta(changes)
1263
 
        self._make_dirty(reset_inventory=True)
1264
 
 
1265
 
    def update_basis_by_delta(self, new_revid, delta):
1266
 
        """See MutableTree.update_basis_by_delta."""
1267
 
        if self.last_revision() == new_revid:
1268
 
            raise AssertionError()
1269
 
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
1270
 
 
1271
1219
    @needs_read_lock
1272
1220
    def _validate(self):
1273
1221
        self._dirstate._validate()
1275
1223
    @needs_tree_write_lock
1276
1224
    def _write_inventory(self, inv):
1277
1225
        """Write inventory as the current inventory."""
1278
 
        if self._dirty:
1279
 
            raise AssertionError("attempting to write an inventory when the "
1280
 
                "dirstate is dirty will lose pending changes")
1281
 
        had_inventory = self._inventory is not None
1282
 
        # Setting self._inventory = None forces the dirstate to regenerate the
1283
 
        # working inventory. We do this because self.inventory may be inv, or
1284
 
        # may have been modified, and either case would prevent a clean delta
1285
 
        # being created.
1286
 
        self._inventory = None
1287
 
        # generate a delta,
1288
 
        delta = inv._make_delta(self.inventory)
1289
 
        # and apply it.
1290
 
        self.apply_inventory_delta(delta)
1291
 
        if had_inventory:
 
1226
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1227
        self.current_dirstate().set_state_from_inventory(inv)
 
1228
        self._make_dirty(reset_inventory=False)
 
1229
        if self._inventory is not None:
1292
1230
            self._inventory = inv
1293
1231
        self.flush()
1294
1232
 
1295
1233
 
1296
 
class ContentFilterAwareSHA1Provider(dirstate.SHA1Provider):
1297
 
 
1298
 
    def __init__(self, tree):
1299
 
        self.tree = tree
1300
 
 
1301
 
    def sha1(self, abspath):
1302
 
        """See dirstate.SHA1Provider.sha1()."""
1303
 
        filters = self.tree._content_filter_stack(
1304
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1305
 
        return internal_size_sha_file_byname(abspath, filters)[1]
1306
 
 
1307
 
    def stat_and_sha1(self, abspath):
1308
 
        """See dirstate.SHA1Provider.stat_and_sha1()."""
1309
 
        filters = self.tree._content_filter_stack(
1310
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1311
 
        file_obj = file(abspath, 'rb', 65000)
1312
 
        try:
1313
 
            statvalue = os.fstat(file_obj.fileno())
1314
 
            if filters:
1315
 
                file_obj = filtered_input_file(file_obj, filters)
1316
 
            sha1 = osutils.size_sha_file(file_obj)[1]
1317
 
        finally:
1318
 
            file_obj.close()
1319
 
        return statvalue, sha1
1320
 
 
1321
 
 
1322
 
class ContentFilteringDirStateWorkingTree(DirStateWorkingTree):
1323
 
    """Dirstate working tree that supports content filtering.
1324
 
 
1325
 
    The dirstate holds the hash and size of the canonical form of the file, 
1326
 
    and most methods must return that.
1327
 
    """
1328
 
 
1329
 
    def _file_content_summary(self, path, stat_result):
1330
 
        # This is to support the somewhat obsolete path_content_summary method
1331
 
        # with content filtering: see
1332
 
        # <https://bugs.edge.launchpad.net/bzr/+bug/415508>.
1333
 
        #
1334
 
        # If the dirstate cache is up to date and knows the hash and size,
1335
 
        # return that.
1336
 
        # Otherwise if there are no content filters, return the on-disk size
1337
 
        # and leave the hash blank.
1338
 
        # Otherwise, read and filter the on-disk file and use its size and
1339
 
        # hash.
1340
 
        #
1341
 
        # The dirstate doesn't store the size of the canonical form so we
1342
 
        # can't trust it for content-filtered trees.  We just return None.
1343
 
        dirstate_sha1 = self._dirstate.sha1_from_stat(path, stat_result)
1344
 
        executable = self._is_executable_from_path_and_stat(path, stat_result)
1345
 
        return ('file', None, executable, dirstate_sha1)
1346
 
 
1347
 
 
1348
 
class WorkingTree4(DirStateWorkingTree):
1349
 
    """This is the Format 4 working tree.
1350
 
 
1351
 
    This differs from WorkingTree3 by:
1352
 
     - Having a consolidated internal dirstate, stored in a
1353
 
       randomly-accessible sorted file on disk.
1354
 
     - Not having a regular inventory attribute.  One can be synthesized
1355
 
       on demand but this is expensive and should be avoided.
1356
 
 
1357
 
    This is new in bzr 0.15.
1358
 
    """
1359
 
 
1360
 
 
1361
 
class WorkingTree5(ContentFilteringDirStateWorkingTree):
1362
 
    """This is the Format 5 working tree.
1363
 
 
1364
 
    This differs from WorkingTree4 by:
1365
 
     - Supporting content filtering.
1366
 
 
1367
 
    This is new in bzr 1.11.
1368
 
    """
1369
 
 
1370
 
 
1371
 
class WorkingTree6(ContentFilteringDirStateWorkingTree):
1372
 
    """This is the Format 6 working tree.
1373
 
 
1374
 
    This differs from WorkingTree5 by:
1375
 
     - Supporting a current view that may mask the set of files in a tree
1376
 
       impacted by most user operations.
1377
 
 
1378
 
    This is new in bzr 1.14.
1379
 
    """
1380
 
 
1381
 
    def _make_views(self):
1382
 
        return views.PathBasedViews(self)
1383
 
 
1384
 
 
1385
 
class DirStateWorkingTreeFormat(WorkingTreeFormat3):
1386
 
 
1387
 
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
1388
 
                   accelerator_tree=None, hardlink=False):
 
1234
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1235
    """The first consolidated dirstate working tree format.
 
1236
 
 
1237
    This format:
 
1238
        - exists within a metadir controlling .bzr
 
1239
        - includes an explicit version marker for the workingtree control
 
1240
          files, separate from the BzrDir format
 
1241
        - modifies the hash cache format
 
1242
        - is new in bzr 0.15
 
1243
        - uses a LockDir to guard access to it.
 
1244
    """
 
1245
 
 
1246
    upgrade_recommended = False
 
1247
 
 
1248
    def get_format_string(self):
 
1249
        """See WorkingTreeFormat.get_format_string()."""
 
1250
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1251
 
 
1252
    def get_format_description(self):
 
1253
        """See WorkingTreeFormat.get_format_description()."""
 
1254
        return "Working tree format 4"
 
1255
 
 
1256
    def initialize(self, a_bzrdir, revision_id=None):
1389
1257
        """See WorkingTreeFormat.initialize().
1390
1258
 
1391
1259
        :param revision_id: allows creating a working tree at a different
1392
1260
        revision than the branch is at.
1393
 
        :param accelerator_tree: A tree which can be used for retrieving file
1394
 
            contents more quickly than the revision tree, i.e. a workingtree.
1395
 
            The revision tree will be used for cases where accelerator_tree's
1396
 
            content is different.
1397
 
        :param hardlink: If true, hard-link files from accelerator_tree,
1398
 
            where possible.
1399
1261
 
1400
1262
        These trees get an initial random root id, if their repository supports
1401
1263
        rich root data, TREE_ROOT otherwise.
1402
1264
        """
 
1265
        revision_id = osutils.safe_revision_id(revision_id)
1403
1266
        if not isinstance(a_bzrdir.transport, LocalTransport):
1404
1267
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
1405
1268
        transport = a_bzrdir.get_workingtree_transport(self)
1406
1269
        control_files = self._open_control_files(a_bzrdir)
1407
1270
        control_files.create_lock()
1408
1271
        control_files.lock_write()
1409
 
        transport.put_bytes('format', self.get_format_string(),
1410
 
            mode=a_bzrdir._get_file_mode())
1411
 
        if from_branch is not None:
1412
 
            branch = from_branch
1413
 
        else:
1414
 
            branch = a_bzrdir.open_branch()
 
1272
        control_files.put_utf8('format', self.get_format_string())
 
1273
        branch = a_bzrdir.open_branch()
1415
1274
        if revision_id is None:
1416
1275
            revision_id = branch.last_revision()
1417
1276
        local_path = transport.local_abspath('dirstate')
1419
1278
        state = dirstate.DirState.initialize(local_path)
1420
1279
        state.unlock()
1421
1280
        del state
1422
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1281
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1423
1282
                         branch,
1424
1283
                         _format=self,
1425
1284
                         _bzrdir=a_bzrdir,
1427
1286
        wt._new_tree()
1428
1287
        wt.lock_tree_write()
1429
1288
        try:
1430
 
            self._init_custom_control_files(wt)
1431
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
 
1289
            if revision_id in (None, NULL_REVISION):
1432
1290
                if branch.repository.supports_rich_root():
1433
1291
                    wt._set_root_id(generate_ids.gen_root_id())
1434
1292
                else:
1435
1293
                    wt._set_root_id(ROOT_ID)
1436
1294
                wt.flush()
1437
 
            basis = None
1438
 
            # frequently, we will get here due to branching.  The accelerator
1439
 
            # tree will be the tree from the branch, so the desired basis
1440
 
            # tree will often be a parent of the accelerator tree.
1441
 
            if accelerator_tree is not None:
1442
 
                try:
1443
 
                    basis = accelerator_tree.revision_tree(revision_id)
1444
 
                except errors.NoSuchRevision:
1445
 
                    pass
1446
 
            if basis is None:
1447
 
                basis = branch.repository.revision_tree(revision_id)
1448
 
            if revision_id == _mod_revision.NULL_REVISION:
1449
 
                parents_list = []
1450
 
            else:
1451
 
                parents_list = [(revision_id, basis)]
 
1295
            wt.set_last_revision(revision_id)
 
1296
            wt.flush()
 
1297
            basis = wt.basis_tree()
1452
1298
            basis.lock_read()
1453
 
            try:
1454
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1299
            # if the basis has a root id we have to use that; otherwise we use
 
1300
            # a new random one
 
1301
            basis_root_id = basis.get_root_id()
 
1302
            if basis_root_id is not None:
 
1303
                wt._set_root_id(basis_root_id)
1455
1304
                wt.flush()
1456
 
                # if the basis has a root id we have to use that; otherwise we
1457
 
                # use a new random one
1458
 
                basis_root_id = basis.get_root_id()
1459
 
                if basis_root_id is not None:
1460
 
                    wt._set_root_id(basis_root_id)
1461
 
                    wt.flush()
1462
 
                if wt.supports_content_filtering():
1463
 
                    # The original tree may not have the same content filters
1464
 
                    # applied so we can't safely build the inventory delta from
1465
 
                    # the source tree.
1466
 
                    delta_from_tree = False
1467
 
                else:
1468
 
                    delta_from_tree = True
1469
 
                # delta_from_tree is safe even for DirStateRevisionTrees,
1470
 
                # because wt4.apply_inventory_delta does not mutate the input
1471
 
                # inventory entries.
1472
 
                transform.build_tree(basis, wt, accelerator_tree,
1473
 
                                     hardlink=hardlink,
1474
 
                                     delta_from_tree=delta_from_tree)
1475
 
            finally:
1476
 
                basis.unlock()
 
1305
            transform.build_tree(basis, wt)
 
1306
            basis.unlock()
1477
1307
        finally:
1478
1308
            control_files.unlock()
1479
1309
            wt.unlock()
1480
1310
        return wt
1481
1311
 
1482
 
    def _init_custom_control_files(self, wt):
1483
 
        """Subclasses with custom control files should override this method.
1484
 
 
1485
 
        The working tree and control files are locked for writing when this
1486
 
        method is called.
1487
 
 
1488
 
        :param wt: the WorkingTree object
1489
 
        """
1490
 
 
1491
1312
    def _open(self, a_bzrdir, control_files):
1492
1313
        """Open the tree itself.
1493
1314
 
1494
1315
        :param a_bzrdir: the dir for the tree.
1495
1316
        :param control_files: the control files for the tree.
1496
1317
        """
1497
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1318
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1498
1319
                           branch=a_bzrdir.open_branch(),
1499
1320
                           _format=self,
1500
1321
                           _bzrdir=a_bzrdir,
1501
1322
                           _control_files=control_files)
1502
1323
 
1503
1324
    def __get_matchingbzrdir(self):
1504
 
        return self._get_matchingbzrdir()
1505
 
 
1506
 
    def _get_matchingbzrdir(self):
1507
 
        """Overrideable method to get a bzrdir for testing."""
1508
1325
        # please test against something that will let us do tree references
1509
1326
        return bzrdir.format_registry.make_bzrdir(
1510
1327
            'dirstate-with-subtree')
1512
1329
    _matchingbzrdir = property(__get_matchingbzrdir)
1513
1330
 
1514
1331
 
1515
 
class WorkingTreeFormat4(DirStateWorkingTreeFormat):
1516
 
    """The first consolidated dirstate working tree format.
1517
 
 
1518
 
    This format:
1519
 
        - exists within a metadir controlling .bzr
1520
 
        - includes an explicit version marker for the workingtree control
1521
 
          files, separate from the BzrDir format
1522
 
        - modifies the hash cache format
1523
 
        - is new in bzr 0.15
1524
 
        - uses a LockDir to guard access to it.
1525
 
    """
1526
 
 
1527
 
    upgrade_recommended = False
1528
 
 
1529
 
    _tree_class = WorkingTree4
1530
 
 
1531
 
    def get_format_string(self):
1532
 
        """See WorkingTreeFormat.get_format_string()."""
1533
 
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
1534
 
 
1535
 
    def get_format_description(self):
1536
 
        """See WorkingTreeFormat.get_format_description()."""
1537
 
        return "Working tree format 4"
1538
 
 
1539
 
 
1540
 
class WorkingTreeFormat5(DirStateWorkingTreeFormat):
1541
 
    """WorkingTree format supporting content filtering.
1542
 
    """
1543
 
 
1544
 
    upgrade_recommended = False
1545
 
 
1546
 
    _tree_class = WorkingTree5
1547
 
 
1548
 
    def get_format_string(self):
1549
 
        """See WorkingTreeFormat.get_format_string()."""
1550
 
        return "Bazaar Working Tree Format 5 (bzr 1.11)\n"
1551
 
 
1552
 
    def get_format_description(self):
1553
 
        """See WorkingTreeFormat.get_format_description()."""
1554
 
        return "Working tree format 5"
1555
 
 
1556
 
    def supports_content_filtering(self):
1557
 
        return True
1558
 
 
1559
 
 
1560
 
class WorkingTreeFormat6(DirStateWorkingTreeFormat):
1561
 
    """WorkingTree format supporting views.
1562
 
    """
1563
 
 
1564
 
    upgrade_recommended = False
1565
 
 
1566
 
    _tree_class = WorkingTree6
1567
 
 
1568
 
    def get_format_string(self):
1569
 
        """See WorkingTreeFormat.get_format_string()."""
1570
 
        return "Bazaar Working Tree Format 6 (bzr 1.14)\n"
1571
 
 
1572
 
    def get_format_description(self):
1573
 
        """See WorkingTreeFormat.get_format_description()."""
1574
 
        return "Working tree format 6"
1575
 
 
1576
 
    def _init_custom_control_files(self, wt):
1577
 
        """Subclasses with custom control files should override this method."""
1578
 
        wt._transport.put_bytes('views', '', mode=wt.bzrdir._get_file_mode())
1579
 
 
1580
 
    def supports_content_filtering(self):
1581
 
        return True
1582
 
 
1583
 
    def supports_views(self):
1584
 
        return True
1585
 
 
1586
 
 
1587
1332
class DirStateRevisionTree(Tree):
1588
 
    """A revision tree pulling the inventory from a dirstate.
1589
 
    
1590
 
    Note that this is one of the historical (ie revision) trees cached in the
1591
 
    dirstate for easy access, not the workingtree.
1592
 
    """
 
1333
    """A revision tree pulling the inventory from a dirstate."""
1593
1334
 
1594
1335
    def __init__(self, dirstate, revision_id, repository):
1595
1336
        self._dirstate = dirstate
1596
 
        self._revision_id = revision_id
 
1337
        self._revision_id = osutils.safe_revision_id(revision_id)
1597
1338
        self._repository = repository
1598
1339
        self._inventory = None
1599
1340
        self._locked = 0
1600
1341
        self._dirstate_locked = False
1601
 
        self._repo_supports_tree_reference = getattr(
1602
 
            repository._format, "supports_tree_reference",
1603
 
            False)
1604
1342
 
1605
1343
    def __repr__(self):
1606
1344
        return "<%s of %s in %s>" % \
1607
1345
            (self.__class__.__name__, self._revision_id, self._dirstate)
1608
1346
 
1609
 
    def annotate_iter(self, file_id,
1610
 
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1347
    def annotate_iter(self, file_id):
1611
1348
        """See Tree.annotate_iter"""
1612
 
        text_key = (file_id, self.inventory[file_id].revision)
1613
 
        annotations = self._repository.texts.annotate(text_key)
1614
 
        return [(key[-1], line) for (key, line) in annotations]
 
1349
        w = self._repository.weave_store.get_weave(file_id,
 
1350
                           self._repository.get_transaction())
 
1351
        return w.annotate_iter(self.inventory[file_id].revision)
1615
1352
 
1616
 
    def _get_ancestors(self, default_revision):
1617
 
        return set(self._repository.get_ancestry(self._revision_id,
1618
 
                                                 topo_sorted=False))
1619
1353
    def _comparison_data(self, entry, path):
1620
1354
        """See Tree._comparison_data."""
1621
1355
        if entry is None:
1638
1372
    def get_root_id(self):
1639
1373
        return self.path2id('')
1640
1374
 
1641
 
    def id2path(self, file_id):
1642
 
        "Convert a file-id to a path."
1643
 
        entry = self._get_entry(file_id=file_id)
1644
 
        if entry == (None, None):
1645
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1646
 
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1647
 
        return path_utf8.decode('utf8')
1648
 
 
1649
 
    def iter_references(self):
1650
 
        if not self._repo_supports_tree_reference:
1651
 
            # When the repo doesn't support references, we will have nothing to
1652
 
            # return
1653
 
            return iter([])
1654
 
        # Otherwise, fall back to the default implementation
1655
 
        return super(DirStateRevisionTree, self).iter_references()
1656
 
 
1657
1375
    def _get_parent_index(self):
1658
1376
        """Return the index in the dirstate referenced by this tree."""
1659
1377
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1664
1382
        If either file_id or path is supplied, it is used as the key to lookup.
1665
1383
        If both are supplied, the fastest lookup is used, and an error is
1666
1384
        raised if they do not both point at the same row.
1667
 
 
 
1385
        
1668
1386
        :param file_id: An optional unicode file_id to be looked up.
1669
1387
        :param path: An optional unicode path to be looked up.
1670
1388
        :return: The dirstate row tuple for path/file_id, or (None, None)
1671
1389
        """
1672
1390
        if file_id is None and path is None:
1673
1391
            raise errors.BzrError('must supply file_id or path')
 
1392
        file_id = osutils.safe_file_id(file_id)
1674
1393
        if path is not None:
1675
1394
            path = path.encode('utf8')
1676
1395
        parent_index = self._get_parent_index()
1684
1403
 
1685
1404
        This is relatively expensive: we have to walk the entire dirstate.
1686
1405
        """
1687
 
        if not self._locked:
1688
 
            raise AssertionError(
1689
 
                'cannot generate inventory of an unlocked '
1690
 
                'dirstate revision tree')
 
1406
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1407
            'dirstate revision tree'
1691
1408
        # separate call for profiling - makes it clear where the costs are.
1692
1409
        self._dirstate._read_dirblocks_if_needed()
1693
 
        if self._revision_id not in self._dirstate.get_parent_ids():
1694
 
            raise AssertionError(
1695
 
                'parent %s has disappeared from %s' % (
1696
 
                self._revision_id, self._dirstate.get_parent_ids()))
 
1410
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1411
            'parent %s has disappeared from %s' % (
 
1412
            self._revision_id, self._dirstate.get_parent_ids())
1697
1413
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1698
1414
        # This is identical now to the WorkingTree _generate_inventory except
1699
1415
        # for the tree index use.
1700
1416
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1701
1417
        current_id = root_key[2]
1702
 
        if current_entry[parent_index][0] != 'd':
1703
 
            raise AssertionError()
 
1418
        assert current_entry[parent_index][0] == 'd'
1704
1419
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1705
1420
        inv.root.revision = current_entry[parent_index][4]
1706
1421
        # Turn some things into local variables
1746
1461
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1747
1462
                            % entry)
1748
1463
                # These checks cost us around 40ms on a 55k entry tree
1749
 
                if file_id in inv_byid:
1750
 
                    raise AssertionError('file_id %s already in'
1751
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
1752
 
                if name_unicode in parent_ie.children:
1753
 
                    raise AssertionError('name %r already in parent'
1754
 
                        % (name_unicode,))
 
1464
                assert file_id not in inv_byid
 
1465
                assert name_unicode not in parent_ie.children
1755
1466
                inv_byid[file_id] = inv_entry
1756
1467
                parent_ie.children[name_unicode] = inv_entry
1757
1468
        self._inventory = inv
1767
1478
            return None
1768
1479
        parent_index = self._get_parent_index()
1769
1480
        last_changed_revision = entry[1][parent_index][4]
1770
 
        try:
1771
 
            rev = self._repository.get_revision(last_changed_revision)
1772
 
        except errors.NoSuchRevision:
1773
 
            raise errors.FileTimestampUnavailable(self.id2path(file_id))
1774
 
        return rev.timestamp
 
1481
        return self._repository.get_revision(last_changed_revision).timestamp
1775
1482
 
1776
1483
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1777
1484
        entry = self._get_entry(file_id=file_id, path=path)
1781
1488
            return parent_details[1]
1782
1489
        return None
1783
1490
 
1784
 
    def get_file(self, file_id, path=None):
 
1491
    def get_file(self, file_id):
1785
1492
        return StringIO(self.get_file_text(file_id))
1786
1493
 
 
1494
    def get_file_lines(self, file_id):
 
1495
        ie = self.inventory[file_id]
 
1496
        return self._repository.weave_store.get_weave(file_id,
 
1497
                self._repository.get_transaction()).get_lines(ie.revision)
 
1498
 
1787
1499
    def get_file_size(self, file_id):
1788
 
        """See Tree.get_file_size"""
1789
1500
        return self.inventory[file_id].text_size
1790
1501
 
1791
 
    def get_file_text(self, file_id, path=None):
1792
 
        _, content = list(self.iter_files_bytes([(file_id, None)]))[0]
1793
 
        return ''.join(content)
 
1502
    def get_file_text(self, file_id):
 
1503
        return ''.join(self.get_file_lines(file_id))
1794
1504
 
1795
1505
    def get_reference_revision(self, file_id, path=None):
1796
1506
        return self.inventory[file_id].reference_revision
1797
1507
 
1798
 
    def iter_files_bytes(self, desired_files):
1799
 
        """See Tree.iter_files_bytes.
1800
 
 
1801
 
        This version is implemented on top of Repository.iter_files_bytes"""
1802
 
        parent_index = self._get_parent_index()
1803
 
        repo_desired_files = []
1804
 
        for file_id, identifier in desired_files:
1805
 
            entry = self._get_entry(file_id)
1806
 
            if entry == (None, None):
1807
 
                raise errors.NoSuchId(self, file_id)
1808
 
            repo_desired_files.append((file_id, entry[1][parent_index][4],
1809
 
                                       identifier))
1810
 
        return self._repository.iter_files_bytes(repo_desired_files)
1811
 
 
1812
1508
    def get_symlink_target(self, file_id):
1813
1509
        entry = self._get_entry(file_id=file_id)
1814
1510
        parent_index = self._get_parent_index()
1815
1511
        if entry[1][parent_index][0] != 'l':
1816
1512
            return None
1817
1513
        else:
1818
 
            target = entry[1][parent_index][1]
1819
 
            target = target.decode('utf8')
1820
 
            return target
 
1514
            # At present, none of the tree implementations supports non-ascii
 
1515
            # symlink targets. So we will just assume that the dirstate path is
 
1516
            # correct.
 
1517
            return entry[1][parent_index][1]
1821
1518
 
1822
1519
    def get_revision_id(self):
1823
1520
        """Return the revision id for this tree."""
1841
1538
        return bool(self.path2id(filename))
1842
1539
 
1843
1540
    def kind(self, file_id):
1844
 
        entry = self._get_entry(file_id=file_id)[1]
1845
 
        if entry is None:
1846
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1847
 
        parent_index = self._get_parent_index()
1848
 
        return dirstate.DirState._minikind_to_kind[entry[parent_index][0]]
1849
 
 
1850
 
    def stored_kind(self, file_id):
1851
 
        """See Tree.stored_kind"""
1852
 
        return self.kind(file_id)
1853
 
 
1854
 
    def path_content_summary(self, path):
1855
 
        """See Tree.path_content_summary."""
1856
 
        id = self.inventory.path2id(path)
1857
 
        if id is None:
1858
 
            return ('missing', None, None, None)
1859
 
        entry = self._inventory[id]
1860
 
        kind = entry.kind
1861
 
        if kind == 'file':
1862
 
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
1863
 
        elif kind == 'symlink':
1864
 
            return (kind, None, None, entry.symlink_target)
1865
 
        else:
1866
 
            return (kind, None, None, None)
 
1541
        return self.inventory[file_id].kind
1867
1542
 
1868
1543
    def is_executable(self, file_id, path=None):
1869
1544
        ie = self.inventory[file_id]
1871
1546
            return None
1872
1547
        return ie.executable
1873
1548
 
1874
 
    def is_locked(self):
1875
 
        return self._locked
1876
 
 
1877
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1549
    def list_files(self, include_root=False):
1878
1550
        # We use a standard implementation, because DirStateRevisionTree is
1879
1551
        # dealing with one of the parents of the current state
1880
1552
        inv = self._get_inventory()
1881
 
        if from_dir is None:
1882
 
            from_dir_id = None
1883
 
        else:
1884
 
            from_dir_id = inv.path2id(from_dir)
1885
 
            if from_dir_id is None:
1886
 
                # Directory not versioned
1887
 
                return
1888
 
        entries = inv.iter_entries(from_dir=from_dir_id, recursive=recursive)
1889
 
        if inv.root is not None and not include_root and from_dir is None:
 
1553
        entries = inv.iter_entries()
 
1554
        if self.inventory.root is not None and not include_root:
1890
1555
            entries.next()
1891
1556
        for path, entry in entries:
1892
1557
            yield path, 'V', entry.kind, entry.file_id, entry
1893
1558
 
1894
1559
    def lock_read(self):
1895
 
        """Lock the tree for a set of operations.
1896
 
 
1897
 
        :return: A bzrlib.lock.LogicalLockResult.
1898
 
        """
 
1560
        """Lock the tree for a set of operations."""
1899
1561
        if not self._locked:
1900
1562
            self._repository.lock_read()
1901
1563
            if self._dirstate._lock_token is None:
1902
1564
                self._dirstate.lock_read()
1903
1565
                self._dirstate_locked = True
1904
1566
        self._locked += 1
1905
 
        return LogicalLockResult(self.unlock)
1906
1567
 
1907
1568
    def _must_be_locked(self):
1908
1569
        if not self._locked:
1929
1590
                self._dirstate_locked = False
1930
1591
            self._repository.unlock()
1931
1592
 
1932
 
    @needs_read_lock
1933
 
    def supports_tree_reference(self):
1934
 
        return self._repo_supports_tree_reference
1935
 
 
1936
1593
    def walkdirs(self, prefix=""):
1937
1594
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1938
 
        # implementation based on an inventory.
 
1595
        # implementation based on an inventory.  
1939
1596
        # This should be cleaned up to use the much faster Dirstate code
1940
1597
        # So for now, we just build up the parent inventory, and extract
1941
1598
        # it the same way RevisionTree does.
1970
1627
 
1971
1628
class InterDirStateTree(InterTree):
1972
1629
    """Fast path optimiser for changes_from with dirstate trees.
1973
 
 
1974
 
    This is used only when both trees are in the dirstate working file, and
1975
 
    the source is any parent within the dirstate, and the destination is
 
1630
    
 
1631
    This is used only when both trees are in the dirstate working file, and 
 
1632
    the source is any parent within the dirstate, and the destination is 
1976
1633
    the current working tree of the same dirstate.
1977
1634
    """
1978
1635
    # this could be generalized to allow comparisons between any trees in the
1991
1648
        target.set_parent_ids([revid])
1992
1649
        return target.basis_tree(), target
1993
1650
 
1994
 
    @classmethod
1995
 
    def make_source_parent_tree_python_dirstate(klass, test_case, source, target):
1996
 
        result = klass.make_source_parent_tree(source, target)
1997
 
        result[1]._iter_changes = dirstate.ProcessEntryPython
1998
 
        return result
1999
 
 
2000
 
    @classmethod
2001
 
    def make_source_parent_tree_compiled_dirstate(klass, test_case, source,
2002
 
                                                  target):
2003
 
        from bzrlib.tests.test__dirstate_helpers import \
2004
 
            compiled_dirstate_helpers_feature
2005
 
        test_case.requireFeature(compiled_dirstate_helpers_feature)
2006
 
        from bzrlib._dirstate_helpers_pyx import ProcessEntryC
2007
 
        result = klass.make_source_parent_tree(source, target)
2008
 
        result[1]._iter_changes = ProcessEntryC
2009
 
        return result
2010
 
 
2011
1651
    _matching_from_tree_format = WorkingTreeFormat4()
2012
1652
    _matching_to_tree_format = WorkingTreeFormat4()
2013
 
 
2014
 
    @classmethod
2015
 
    def _test_mutable_trees_to_test_trees(klass, test_case, source, target):
2016
 
        # This method shouldn't be called, because we have python and C
2017
 
        # specific flavours.
2018
 
        raise NotImplementedError
2019
 
 
2020
 
    def iter_changes(self, include_unchanged=False,
 
1653
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1654
 
 
1655
    def _iter_changes(self, include_unchanged=False,
2021
1656
                      specific_files=None, pb=None, extra_trees=[],
2022
1657
                      require_versioned=True, want_unversioned=False):
2023
1658
        """Return the changes from source to target.
2024
1659
 
2025
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1660
        :return: An iterator that yields tuples. See InterTree._iter_changes
2026
1661
            for details.
2027
1662
        :param specific_files: An optional list of file paths to restrict the
2028
1663
            comparison to. When mapping filenames to ids, all matches in all
2039
1674
            output. An unversioned file is defined as one with (False, False)
2040
1675
            for the versioned pair.
2041
1676
        """
 
1677
        utf8_decode = cache_utf8._utf8_decode
 
1678
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1679
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1680
        # NB: show_status depends on being able to pass in non-versioned files
 
1681
        # and report them as unknown
2042
1682
        # TODO: handle extra trees in the dirstate.
2043
 
        if (extra_trees or specific_files == []):
 
1683
        # TODO: handle comparisons as an empty tree as a different special
 
1684
        # case? mbp 20070226
 
1685
        if extra_trees or (self.source._revision_id == NULL_REVISION):
2044
1686
            # we can't fast-path these cases (yet)
2045
 
            return super(InterDirStateTree, self).iter_changes(
 
1687
            for f in super(InterDirStateTree, self)._iter_changes(
2046
1688
                include_unchanged, specific_files, pb, extra_trees,
2047
 
                require_versioned, want_unversioned=want_unversioned)
 
1689
                require_versioned, want_unversioned=want_unversioned):
 
1690
                yield f
 
1691
            return
2048
1692
        parent_ids = self.target.get_parent_ids()
2049
 
        if not (self.source._revision_id in parent_ids
2050
 
                or self.source._revision_id == _mod_revision.NULL_REVISION):
2051
 
            raise AssertionError(
2052
 
                "revision {%s} is not stored in {%s}, but %s "
2053
 
                "can only be used for trees stored in the dirstate"
2054
 
                % (self.source._revision_id, self.target, self.iter_changes))
 
1693
        assert (self.source._revision_id in parent_ids), \
 
1694
                "revision {%s} is not stored in {%s}, but %s " \
 
1695
                "can only be used for trees stored in the dirstate" \
 
1696
                % (self.source._revision_id, self.target, self._iter_changes)
2055
1697
        target_index = 0
2056
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1698
        if self.source._revision_id == NULL_REVISION:
2057
1699
            source_index = None
2058
1700
            indices = (target_index,)
2059
1701
        else:
2060
 
            if not (self.source._revision_id in parent_ids):
2061
 
                raise AssertionError(
2062
 
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
2063
 
                    self.source._revision_id, parent_ids))
 
1702
            assert (self.source._revision_id in parent_ids), \
 
1703
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1704
                self.source._revision_id, parent_ids)
2064
1705
            source_index = 1 + parent_ids.index(self.source._revision_id)
2065
 
            indices = (source_index, target_index)
 
1706
            indices = (source_index,target_index)
2066
1707
        # -- make all specific_files utf8 --
2067
1708
        if specific_files:
2068
1709
            specific_files_utf8 = set()
2069
1710
            for path in specific_files:
2070
 
                # Note, if there are many specific files, using cache_utf8
2071
 
                # would be good here.
2072
1711
                specific_files_utf8.add(path.encode('utf8'))
2073
1712
            specific_files = specific_files_utf8
2074
1713
        else:
2075
1714
            specific_files = set([''])
2076
1715
        # -- specific_files is now a utf8 path set --
2077
 
 
2078
1716
        # -- get the state object and prepare it.
2079
1717
        state = self.target.current_dirstate()
2080
1718
        state._read_dirblocks_if_needed()
 
1719
        def _entries_for_path(path):
 
1720
            """Return a list with all the entries that match path for all ids.
 
1721
            """
 
1722
            dirname, basename = os.path.split(path)
 
1723
            key = (dirname, basename, '')
 
1724
            block_index, present = state._find_block_index_from_key(key)
 
1725
            if not present:
 
1726
                # the block which should contain path is absent.
 
1727
                return []
 
1728
            result = []
 
1729
            block = state._dirblocks[block_index][1]
 
1730
            entry_index, _ = state._find_entry_index(key, block)
 
1731
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1732
            while (entry_index < len(block) and
 
1733
                block[entry_index][0][0:2] == key[0:2]):
 
1734
                result.append(block[entry_index])
 
1735
                entry_index += 1
 
1736
            return result
2081
1737
        if require_versioned:
2082
1738
            # -- check all supplied paths are versioned in a search tree. --
2083
 
            not_versioned = []
 
1739
            all_versioned = True
2084
1740
            for path in specific_files:
2085
 
                path_entries = state._entries_for_path(path)
 
1741
                path_entries = _entries_for_path(path)
2086
1742
                if not path_entries:
2087
1743
                    # this specified path is not present at all: error
2088
 
                    not_versioned.append(path)
2089
 
                    continue
 
1744
                    all_versioned = False
 
1745
                    break
2090
1746
                found_versioned = False
2091
1747
                # for each id at this path
2092
1748
                for entry in path_entries:
2099
1755
                if not found_versioned:
2100
1756
                    # none of the indexes was not 'absent' at all ids for this
2101
1757
                    # path.
2102
 
                    not_versioned.append(path)
2103
 
            if len(not_versioned) > 0:
2104
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1758
                    all_versioned = False
 
1759
                    break
 
1760
            if not all_versioned:
 
1761
                raise errors.PathsNotVersionedError(specific_files)
2105
1762
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2106
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1763
        search_specific_files = set()
 
1764
        for path in specific_files:
 
1765
            other_specific_files = specific_files.difference(set([path]))
 
1766
            if not osutils.is_inside_any(other_specific_files, path):
 
1767
                # this is a top level path, we must check it.
 
1768
                search_specific_files.add(path)
 
1769
        # sketch: 
 
1770
        # compare source_index and target_index at or under each element of search_specific_files.
 
1771
        # follow the following comparison table. Note that we only want to do diff operations when
 
1772
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1773
        # for the target.
 
1774
        # cases:
 
1775
        # 
 
1776
        # Source | Target | disk | action
 
1777
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1778
        #        |        |      | diff check on source-target
 
1779
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1780
        #        |        |      | ???
 
1781
        #   r    |  a     |      | add source to search
 
1782
        #   r    |  a     |  a   | 
 
1783
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1784
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1785
        #   a    | fdlt   |      | add new id
 
1786
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1787
        #   a    |  a     |      | not present in either tree, skip
 
1788
        #   a    |  a     |  a   | not present in any tree, skip
 
1789
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1790
        #        |        |      | may not be selected by the users list of paths.
 
1791
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1792
        #        |        |      | may not be selected by the users list of paths.
 
1793
        #  fdlt  | fdlt   |      | content in both: diff them
 
1794
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1795
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1796
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1797
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1798
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1799
        #        |        |      | this id at the other path.
 
1800
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1801
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1802
        #        |        |      | this id at the other path.
 
1803
 
 
1804
        # for all search_indexs in each path at or under each element of
 
1805
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1806
        # relocated path as one to search if its not searched already. If the
 
1807
        # detail is not relocated, add the id.
 
1808
        searched_specific_files = set()
 
1809
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1810
        # Using a list so that we can access the values and change them in
 
1811
        # nested scope. Each one is [path, file_id, entry]
 
1812
        last_source_parent = [None, None]
 
1813
        last_target_parent = [None, None]
2107
1814
 
2108
1815
        use_filesystem_for_exec = (sys.platform != 'win32')
2109
 
        iter_changes = self.target._iter_changes(include_unchanged,
2110
 
            use_filesystem_for_exec, search_specific_files, state,
2111
 
            source_index, target_index, want_unversioned, self.target)
2112
 
        return iter_changes.iter_changes()
 
1816
 
 
1817
        # Just a sentry, so that _process_entry can say that this
 
1818
        # record is handled, but isn't interesting to process (unchanged)
 
1819
        uninteresting = object()
 
1820
 
 
1821
 
 
1822
        old_dirname_to_file_id = {}
 
1823
        new_dirname_to_file_id = {}
 
1824
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1825
        #       keeping a cache of directories that we have seen.
 
1826
 
 
1827
        def _process_entry(entry, path_info):
 
1828
            """Compare an entry and real disk to generate delta information.
 
1829
 
 
1830
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1831
                the path of entry. If None, then the path is considered absent.
 
1832
                (Perhaps we should pass in a concrete entry for this ?)
 
1833
                Basename is returned as a utf8 string because we expect this
 
1834
                tuple will be ignored, and don't want to take the time to
 
1835
                decode.
 
1836
            :return: None if these don't match
 
1837
                     A tuple of information about the change, or
 
1838
                     the object 'uninteresting' if these match, but are
 
1839
                     basically identical.
 
1840
            """
 
1841
            if source_index is None:
 
1842
                source_details = NULL_PARENT_DETAILS
 
1843
            else:
 
1844
                source_details = entry[1][source_index]
 
1845
            target_details = entry[1][target_index]
 
1846
            target_minikind = target_details[0]
 
1847
            if path_info is not None and target_minikind in 'fdlt':
 
1848
                assert target_index == 0
 
1849
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1850
                                                  stat_value=path_info[3])
 
1851
                # The entry may have been modified by update_entry
 
1852
                target_details = entry[1][target_index]
 
1853
                target_minikind = target_details[0]
 
1854
            else:
 
1855
                link_or_sha1 = None
 
1856
            file_id = entry[0][2]
 
1857
            source_minikind = source_details[0]
 
1858
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1859
                # claimed content in both: diff
 
1860
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1861
                #        |        |      | diff check on source-target
 
1862
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1863
                #        |        |      | ???
 
1864
                if source_minikind in 'r':
 
1865
                    # add the source to the search path to find any children it
 
1866
                    # has.  TODO ? : only add if it is a container ?
 
1867
                    if not osutils.is_inside_any(searched_specific_files,
 
1868
                                                 source_details[1]):
 
1869
                        search_specific_files.add(source_details[1])
 
1870
                    # generate the old path; this is needed for stating later
 
1871
                    # as well.
 
1872
                    old_path = source_details[1]
 
1873
                    old_dirname, old_basename = os.path.split(old_path)
 
1874
                    path = pathjoin(entry[0][0], entry[0][1])
 
1875
                    old_entry = state._get_entry(source_index,
 
1876
                                                 path_utf8=old_path)
 
1877
                    # update the source details variable to be the real
 
1878
                    # location.
 
1879
                    source_details = old_entry[1][source_index]
 
1880
                    source_minikind = source_details[0]
 
1881
                else:
 
1882
                    old_dirname = entry[0][0]
 
1883
                    old_basename = entry[0][1]
 
1884
                    old_path = path = None
 
1885
                if path_info is None:
 
1886
                    # the file is missing on disk, show as removed.
 
1887
                    content_change = True
 
1888
                    target_kind = None
 
1889
                    target_exec = False
 
1890
                else:
 
1891
                    # source and target are both versioned and disk file is present.
 
1892
                    target_kind = path_info[2]
 
1893
                    if target_kind == 'directory':
 
1894
                        if path is None:
 
1895
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1896
                        new_dirname_to_file_id[path] = file_id
 
1897
                        if source_minikind != 'd':
 
1898
                            content_change = True
 
1899
                        else:
 
1900
                            # directories have no fingerprint
 
1901
                            content_change = False
 
1902
                        target_exec = False
 
1903
                    elif target_kind == 'file':
 
1904
                        if source_minikind != 'f':
 
1905
                            content_change = True
 
1906
                        else:
 
1907
                            # We could check the size, but we already have the
 
1908
                            # sha1 hash.
 
1909
                            content_change = (link_or_sha1 != source_details[1])
 
1910
                        # Target details is updated at update_entry time
 
1911
                        if use_filesystem_for_exec:
 
1912
                            # We don't need S_ISREG here, because we are sure
 
1913
                            # we are dealing with a file.
 
1914
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1915
                        else:
 
1916
                            target_exec = target_details[3]
 
1917
                    elif target_kind == 'symlink':
 
1918
                        if source_minikind != 'l':
 
1919
                            content_change = True
 
1920
                        else:
 
1921
                            content_change = (link_or_sha1 != source_details[1])
 
1922
                        target_exec = False
 
1923
                    elif target_kind == 'tree-reference':
 
1924
                        if source_minikind != 't':
 
1925
                            content_change = True
 
1926
                        else:
 
1927
                            content_change = False
 
1928
                        target_exec = False
 
1929
                    else:
 
1930
                        raise Exception, "unknown kind %s" % path_info[2]
 
1931
                if source_minikind == 'd':
 
1932
                    if path is None:
 
1933
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1934
                    old_dirname_to_file_id[old_path] = file_id
 
1935
                # parent id is the entry for the path in the target tree
 
1936
                if old_dirname == last_source_parent[0]:
 
1937
                    source_parent_id = last_source_parent[1]
 
1938
                else:
 
1939
                    try:
 
1940
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1941
                    except KeyError:
 
1942
                        source_parent_entry = state._get_entry(source_index,
 
1943
                                                               path_utf8=old_dirname)
 
1944
                        source_parent_id = source_parent_entry[0][2]
 
1945
                    if source_parent_id == entry[0][2]:
 
1946
                        # This is the root, so the parent is None
 
1947
                        source_parent_id = None
 
1948
                    else:
 
1949
                        last_source_parent[0] = old_dirname
 
1950
                        last_source_parent[1] = source_parent_id
 
1951
                new_dirname = entry[0][0]
 
1952
                if new_dirname == last_target_parent[0]:
 
1953
                    target_parent_id = last_target_parent[1]
 
1954
                else:
 
1955
                    try:
 
1956
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
1957
                    except KeyError:
 
1958
                        # TODO: We don't always need to do the lookup, because the
 
1959
                        #       parent entry will be the same as the source entry.
 
1960
                        target_parent_entry = state._get_entry(target_index,
 
1961
                                                               path_utf8=new_dirname)
 
1962
                        assert target_parent_entry != (None, None), (
 
1963
                            "Could not find target parent in wt: %s\nparent of: %s"
 
1964
                            % (new_dirname, entry))
 
1965
                        target_parent_id = target_parent_entry[0][2]
 
1966
                    if target_parent_id == entry[0][2]:
 
1967
                        # This is the root, so the parent is None
 
1968
                        target_parent_id = None
 
1969
                    else:
 
1970
                        last_target_parent[0] = new_dirname
 
1971
                        last_target_parent[1] = target_parent_id
 
1972
 
 
1973
                source_exec = source_details[3]
 
1974
                if (include_unchanged
 
1975
                    or content_change
 
1976
                    or source_parent_id != target_parent_id
 
1977
                    or old_basename != entry[0][1]
 
1978
                    or source_exec != target_exec
 
1979
                    ):
 
1980
                    if old_path is None:
 
1981
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1982
                        old_path_u = utf8_decode(old_path)[0]
 
1983
                        path_u = old_path_u
 
1984
                    else:
 
1985
                        old_path_u = utf8_decode(old_path)[0]
 
1986
                        if old_path == path:
 
1987
                            path_u = old_path_u
 
1988
                        else:
 
1989
                            path_u = utf8_decode(path)[0]
 
1990
                    source_kind = _minikind_to_kind[source_minikind]
 
1991
                    return (entry[0][2],
 
1992
                           (old_path_u, path_u),
 
1993
                           content_change,
 
1994
                           (True, True),
 
1995
                           (source_parent_id, target_parent_id),
 
1996
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
1997
                           (source_kind, target_kind),
 
1998
                           (source_exec, target_exec))
 
1999
                else:
 
2000
                    return uninteresting
 
2001
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2002
                # looks like a new file
 
2003
                if path_info is not None:
 
2004
                    path = pathjoin(entry[0][0], entry[0][1])
 
2005
                    # parent id is the entry for the path in the target tree
 
2006
                    # TODO: these are the same for an entire directory: cache em.
 
2007
                    parent_id = state._get_entry(target_index,
 
2008
                                                 path_utf8=entry[0][0])[0][2]
 
2009
                    if parent_id == entry[0][2]:
 
2010
                        parent_id = None
 
2011
                    if use_filesystem_for_exec:
 
2012
                        # We need S_ISREG here, because we aren't sure if this
 
2013
                        # is a file or not.
 
2014
                        target_exec = bool(
 
2015
                            stat.S_ISREG(path_info[3].st_mode)
 
2016
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2017
                    else:
 
2018
                        target_exec = target_details[3]
 
2019
                    return (entry[0][2],
 
2020
                           (None, utf8_decode(path)[0]),
 
2021
                           True,
 
2022
                           (False, True),
 
2023
                           (None, parent_id),
 
2024
                           (None, utf8_decode(entry[0][1])[0]),
 
2025
                           (None, path_info[2]),
 
2026
                           (None, target_exec))
 
2027
                else:
 
2028
                    # but its not on disk: we deliberately treat this as just
 
2029
                    # never-present. (Why ?! - RBC 20070224)
 
2030
                    pass
 
2031
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2032
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2033
                # if its still on disk, *and* theres no other entry at this
 
2034
                # path [we dont know this in this routine at the moment -
 
2035
                # perhaps we should change this - then it would be an unknown.
 
2036
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2037
                # parent id is the entry for the path in the target tree
 
2038
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2039
                if parent_id == entry[0][2]:
 
2040
                    parent_id = None
 
2041
                return (entry[0][2],
 
2042
                       (utf8_decode(old_path)[0], None),
 
2043
                       True,
 
2044
                       (True, False),
 
2045
                       (parent_id, None),
 
2046
                       (utf8_decode(entry[0][1])[0], None),
 
2047
                       (_minikind_to_kind[source_minikind], None),
 
2048
                       (source_details[3], None))
 
2049
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2050
                # a rename; could be a true rename, or a rename inherited from
 
2051
                # a renamed parent. TODO: handle this efficiently. Its not
 
2052
                # common case to rename dirs though, so a correct but slow
 
2053
                # implementation will do.
 
2054
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2055
                    search_specific_files.add(target_details[1])
 
2056
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2057
                # neither of the selected trees contain this file,
 
2058
                # so skip over it. This is not currently directly tested, but
 
2059
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2060
                pass
 
2061
            else:
 
2062
                raise AssertionError("don't know how to compare "
 
2063
                    "source_minikind=%r, target_minikind=%r"
 
2064
                    % (source_minikind, target_minikind))
 
2065
                ## import pdb;pdb.set_trace()
 
2066
            return None
 
2067
 
 
2068
        while search_specific_files:
 
2069
            # TODO: the pending list should be lexically sorted?  the
 
2070
            # interface doesn't require it.
 
2071
            current_root = search_specific_files.pop()
 
2072
            current_root_unicode = current_root.decode('utf8')
 
2073
            searched_specific_files.add(current_root)
 
2074
            # process the entries for this containing directory: the rest will be
 
2075
            # found by their parents recursively.
 
2076
            root_entries = _entries_for_path(current_root)
 
2077
            root_abspath = self.target.abspath(current_root_unicode)
 
2078
            try:
 
2079
                root_stat = os.lstat(root_abspath)
 
2080
            except OSError, e:
 
2081
                if e.errno == errno.ENOENT:
 
2082
                    # the path does not exist: let _process_entry know that.
 
2083
                    root_dir_info = None
 
2084
                else:
 
2085
                    # some other random error: hand it up.
 
2086
                    raise
 
2087
            else:
 
2088
                root_dir_info = ('', current_root,
 
2089
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2090
                    root_abspath)
 
2091
                if root_dir_info[2] == 'directory':
 
2092
                    if self.target._directory_is_tree_reference(
 
2093
                        current_root.decode('utf8')):
 
2094
                        root_dir_info = root_dir_info[:2] + \
 
2095
                            ('tree-reference',) + root_dir_info[3:]
 
2096
 
 
2097
            if not root_entries and not root_dir_info:
 
2098
                # this specified path is not present at all, skip it.
 
2099
                continue
 
2100
            path_handled = False
 
2101
            for entry in root_entries:
 
2102
                result = _process_entry(entry, root_dir_info)
 
2103
                if result is not None:
 
2104
                    path_handled = True
 
2105
                    if result is not uninteresting:
 
2106
                        yield result
 
2107
            if want_unversioned and not path_handled and root_dir_info:
 
2108
                new_executable = bool(
 
2109
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2110
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2111
                yield (None,
 
2112
                       (None, current_root_unicode),
 
2113
                       True,
 
2114
                       (False, False),
 
2115
                       (None, None),
 
2116
                       (None, splitpath(current_root_unicode)[-1]),
 
2117
                       (None, root_dir_info[2]),
 
2118
                       (None, new_executable)
 
2119
                      )
 
2120
            initial_key = (current_root, '', '')
 
2121
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2122
            if block_index == 0:
 
2123
                # we have processed the total root already, but because the
 
2124
                # initial key matched it we should skip it here.
 
2125
                block_index +=1
 
2126
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2127
                current_dir_info = None
 
2128
            else:
 
2129
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2130
                try:
 
2131
                    current_dir_info = dir_iterator.next()
 
2132
                except OSError, e:
 
2133
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2134
                    # python 2.5 has e.errno == EINVAL,
 
2135
                    #            and e.winerror == ERROR_DIRECTORY
 
2136
                    e_winerror = getattr(e, 'winerror', None)
 
2137
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2138
                    # there may be directories in the inventory even though
 
2139
                    # this path is not a file on disk: so mark it as end of
 
2140
                    # iterator
 
2141
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2142
                        current_dir_info = None
 
2143
                    elif (sys.platform == 'win32'
 
2144
                          and (e.errno in win_errors
 
2145
                               or e_winerror in win_errors)):
 
2146
                        current_dir_info = None
 
2147
                    else:
 
2148
                        raise
 
2149
                else:
 
2150
                    if current_dir_info[0][0] == '':
 
2151
                        # remove .bzr from iteration
 
2152
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2153
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2154
                        del current_dir_info[1][bzr_index]
 
2155
            # walk until both the directory listing and the versioned metadata
 
2156
            # are exhausted. 
 
2157
            if (block_index < len(state._dirblocks) and
 
2158
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2159
                current_block = state._dirblocks[block_index]
 
2160
            else:
 
2161
                current_block = None
 
2162
            while (current_dir_info is not None or
 
2163
                   current_block is not None):
 
2164
                if (current_dir_info and current_block
 
2165
                    and current_dir_info[0][0] != current_block[0]):
 
2166
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2167
                        # filesystem data refers to paths not covered by the dirblock.
 
2168
                        # this has two possibilities:
 
2169
                        # A) it is versioned but empty, so there is no block for it
 
2170
                        # B) it is not versioned.
 
2171
 
 
2172
                        # if (A) then we need to recurse into it to check for
 
2173
                        # new unknown files or directories.
 
2174
                        # if (B) then we should ignore it, because we don't
 
2175
                        # recurse into unknown directories.
 
2176
                        path_index = 0
 
2177
                        while path_index < len(current_dir_info[1]):
 
2178
                                current_path_info = current_dir_info[1][path_index]
 
2179
                                if want_unversioned:
 
2180
                                    if current_path_info[2] == 'directory':
 
2181
                                        if self.target._directory_is_tree_reference(
 
2182
                                            current_path_info[0].decode('utf8')):
 
2183
                                            current_path_info = current_path_info[:2] + \
 
2184
                                                ('tree-reference',) + current_path_info[3:]
 
2185
                                    new_executable = bool(
 
2186
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2187
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2188
                                    yield (None,
 
2189
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2190
                                        True,
 
2191
                                        (False, False),
 
2192
                                        (None, None),
 
2193
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2194
                                        (None, current_path_info[2]),
 
2195
                                        (None, new_executable))
 
2196
                                # dont descend into this unversioned path if it is
 
2197
                                # a dir
 
2198
                                if current_path_info[2] in ('directory',
 
2199
                                                            'tree-reference'):
 
2200
                                    del current_dir_info[1][path_index]
 
2201
                                    path_index -= 1
 
2202
                                path_index += 1
 
2203
 
 
2204
                        # This dir info has been handled, go to the next
 
2205
                        try:
 
2206
                            current_dir_info = dir_iterator.next()
 
2207
                        except StopIteration:
 
2208
                            current_dir_info = None
 
2209
                    else:
 
2210
                        # We have a dirblock entry for this location, but there
 
2211
                        # is no filesystem path for this. This is most likely
 
2212
                        # because a directory was removed from the disk.
 
2213
                        # We don't have to report the missing directory,
 
2214
                        # because that should have already been handled, but we
 
2215
                        # need to handle all of the files that are contained
 
2216
                        # within.
 
2217
                        for current_entry in current_block[1]:
 
2218
                            # entry referring to file not present on disk.
 
2219
                            # advance the entry only, after processing.
 
2220
                            result = _process_entry(current_entry, None)
 
2221
                            if result is not None:
 
2222
                                if result is not uninteresting:
 
2223
                                    yield result
 
2224
                        block_index +=1
 
2225
                        if (block_index < len(state._dirblocks) and
 
2226
                            osutils.is_inside(current_root,
 
2227
                                              state._dirblocks[block_index][0])):
 
2228
                            current_block = state._dirblocks[block_index]
 
2229
                        else:
 
2230
                            current_block = None
 
2231
                    continue
 
2232
                entry_index = 0
 
2233
                if current_block and entry_index < len(current_block[1]):
 
2234
                    current_entry = current_block[1][entry_index]
 
2235
                else:
 
2236
                    current_entry = None
 
2237
                advance_entry = True
 
2238
                path_index = 0
 
2239
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2240
                    current_path_info = current_dir_info[1][path_index]
 
2241
                    if current_path_info[2] == 'directory':
 
2242
                        if self.target._directory_is_tree_reference(
 
2243
                            current_path_info[0].decode('utf8')):
 
2244
                            current_path_info = current_path_info[:2] + \
 
2245
                                ('tree-reference',) + current_path_info[3:]
 
2246
                else:
 
2247
                    current_path_info = None
 
2248
                advance_path = True
 
2249
                path_handled = False
 
2250
                while (current_entry is not None or
 
2251
                    current_path_info is not None):
 
2252
                    if current_entry is None:
 
2253
                        # the check for path_handled when the path is adnvaced
 
2254
                        # will yield this path if needed.
 
2255
                        pass
 
2256
                    elif current_path_info is None:
 
2257
                        # no path is fine: the per entry code will handle it.
 
2258
                        result = _process_entry(current_entry, current_path_info)
 
2259
                        if result is not None:
 
2260
                            if result is not uninteresting:
 
2261
                                yield result
 
2262
                    elif (current_entry[0][1] != current_path_info[1]
 
2263
                          or current_entry[1][target_index][0] in 'ar'):
 
2264
                        # The current path on disk doesn't match the dirblock
 
2265
                        # record. Either the dirblock is marked as absent, or
 
2266
                        # the file on disk is not present at all in the
 
2267
                        # dirblock. Either way, report about the dirblock
 
2268
                        # entry, and let other code handle the filesystem one.
 
2269
 
 
2270
                        # Compare the basename for these files to determine
 
2271
                        # which comes first
 
2272
                        if current_path_info[1] < current_entry[0][1]:
 
2273
                            # extra file on disk: pass for now, but only
 
2274
                            # increment the path, not the entry
 
2275
                            advance_entry = False
 
2276
                        else:
 
2277
                            # entry referring to file not present on disk.
 
2278
                            # advance the entry only, after processing.
 
2279
                            result = _process_entry(current_entry, None)
 
2280
                            if result is not None:
 
2281
                                if result is not uninteresting:
 
2282
                                    yield result
 
2283
                            advance_path = False
 
2284
                    else:
 
2285
                        result = _process_entry(current_entry, current_path_info)
 
2286
                        if result is not None:
 
2287
                            path_handled = True
 
2288
                            if result is not uninteresting:
 
2289
                                yield result
 
2290
                    if advance_entry and current_entry is not None:
 
2291
                        entry_index += 1
 
2292
                        if entry_index < len(current_block[1]):
 
2293
                            current_entry = current_block[1][entry_index]
 
2294
                        else:
 
2295
                            current_entry = None
 
2296
                    else:
 
2297
                        advance_entry = True # reset the advance flaga
 
2298
                    if advance_path and current_path_info is not None:
 
2299
                        if not path_handled:
 
2300
                            # unversioned in all regards
 
2301
                            if want_unversioned:
 
2302
                                new_executable = bool(
 
2303
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2304
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2305
                                yield (None,
 
2306
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2307
                                    True,
 
2308
                                    (False, False),
 
2309
                                    (None, None),
 
2310
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2311
                                    (None, current_path_info[2]),
 
2312
                                    (None, new_executable))
 
2313
                            # dont descend into this unversioned path if it is
 
2314
                            # a dir
 
2315
                            if current_path_info[2] in ('directory'):
 
2316
                                del current_dir_info[1][path_index]
 
2317
                                path_index -= 1
 
2318
                        # dont descend the disk iterator into any tree 
 
2319
                        # paths.
 
2320
                        if current_path_info[2] == 'tree-reference':
 
2321
                            del current_dir_info[1][path_index]
 
2322
                            path_index -= 1
 
2323
                        path_index += 1
 
2324
                        if path_index < len(current_dir_info[1]):
 
2325
                            current_path_info = current_dir_info[1][path_index]
 
2326
                            if current_path_info[2] == 'directory':
 
2327
                                if self.target._directory_is_tree_reference(
 
2328
                                    current_path_info[0].decode('utf8')):
 
2329
                                    current_path_info = current_path_info[:2] + \
 
2330
                                        ('tree-reference',) + current_path_info[3:]
 
2331
                        else:
 
2332
                            current_path_info = None
 
2333
                        path_handled = False
 
2334
                    else:
 
2335
                        advance_path = True # reset the advance flagg.
 
2336
                if current_block is not None:
 
2337
                    block_index += 1
 
2338
                    if (block_index < len(state._dirblocks) and
 
2339
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2340
                        current_block = state._dirblocks[block_index]
 
2341
                    else:
 
2342
                        current_block = None
 
2343
                if current_dir_info is not None:
 
2344
                    try:
 
2345
                        current_dir_info = dir_iterator.next()
 
2346
                    except StopIteration:
 
2347
                        current_dir_info = None
 
2348
 
2113
2349
 
2114
2350
    @staticmethod
2115
2351
    def is_compatible(source, target):
2116
2352
        # the target must be a dirstate working tree
2117
 
        if not isinstance(target, DirStateWorkingTree):
 
2353
        if not isinstance(target, WorkingTree4):
2118
2354
            return False
2119
 
        # the source must be a revtree or dirstate rev tree.
 
2355
        # the source must be a revtreee or dirstate rev tree.
2120
2356
        if not isinstance(source,
2121
2357
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2122
2358
            return False
2123
2359
        # the source revid must be in the target dirstate
2124
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2360
        if not (source._revision_id == NULL_REVISION or
2125
2361
            source._revision_id in target.get_parent_ids()):
2126
 
            # TODO: what about ghosts? it may well need to
 
2362
            # TODO: what about ghosts? it may well need to 
2127
2363
            # check for them explicitly.
2128
2364
            return False
2129
2365
        return True
2139
2375
 
2140
2376
    def convert(self, tree):
2141
2377
        # lock the control files not the tree, so that we dont get tree
2142
 
        # on-unlock behaviours, and so that noone else diddles with the
 
2378
        # on-unlock behaviours, and so that noone else diddles with the 
2143
2379
        # tree during upgrade.
2144
2380
        tree._control_files.lock_write()
2145
2381
        try:
2171
2407
 
2172
2408
    def update_format(self, tree):
2173
2409
        """Change the format marker."""
2174
 
        tree._transport.put_bytes('format',
2175
 
            self.target_format.get_format_string(),
2176
 
            mode=tree.bzrdir._get_file_mode())
2177
 
 
2178
 
 
2179
 
class Converter4to5(object):
2180
 
    """Perform an in-place upgrade of format 4 to format 5 trees."""
2181
 
 
2182
 
    def __init__(self):
2183
 
        self.target_format = WorkingTreeFormat5()
2184
 
 
2185
 
    def convert(self, tree):
2186
 
        # lock the control files not the tree, so that we don't get tree
2187
 
        # on-unlock behaviours, and so that no-one else diddles with the
2188
 
        # tree during upgrade.
2189
 
        tree._control_files.lock_write()
2190
 
        try:
2191
 
            self.update_format(tree)
2192
 
        finally:
2193
 
            tree._control_files.unlock()
2194
 
 
2195
 
    def update_format(self, tree):
2196
 
        """Change the format marker."""
2197
 
        tree._transport.put_bytes('format',
2198
 
            self.target_format.get_format_string(),
2199
 
            mode=tree.bzrdir._get_file_mode())
2200
 
 
2201
 
 
2202
 
class Converter4or5to6(object):
2203
 
    """Perform an in-place upgrade of format 4 or 5 to format 6 trees."""
2204
 
 
2205
 
    def __init__(self):
2206
 
        self.target_format = WorkingTreeFormat6()
2207
 
 
2208
 
    def convert(self, tree):
2209
 
        # lock the control files not the tree, so that we don't get tree
2210
 
        # on-unlock behaviours, and so that no-one else diddles with the
2211
 
        # tree during upgrade.
2212
 
        tree._control_files.lock_write()
2213
 
        try:
2214
 
            self.init_custom_control_files(tree)
2215
 
            self.update_format(tree)
2216
 
        finally:
2217
 
            tree._control_files.unlock()
2218
 
 
2219
 
    def init_custom_control_files(self, tree):
2220
 
        """Initialize custom control files."""
2221
 
        tree._transport.put_bytes('views', '',
2222
 
            mode=tree.bzrdir._get_file_mode())
2223
 
 
2224
 
    def update_format(self, tree):
2225
 
        """Change the format marker."""
2226
 
        tree._transport.put_bytes('format',
2227
 
            self.target_format.get_format_string(),
2228
 
            mode=tree.bzrdir._get_file_mode())
 
2410
        tree._control_files.put_utf8('format',
 
2411
            self.target_format.get_format_string())