/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: 2007-03-14 20:15:52 UTC
  • mto: (2353.4.2 locking)
  • mto: This revision was merged to the branch mainline in revision 2360.
  • Revision ID: john@arbash-meinel.com-20070314201552-bjtfua57456dviep
Update the lock code and test code so that if more than one
lock implementation is available, they will both be tested.

It is quite a bit of overhead, for a case where we are likely to only have 1
real lock implementation per platform, but hey, for now we have 2.

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
 
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
56
73
from bzrlib.mutabletree import needs_tree_write_lock
57
74
from bzrlib.osutils import (
58
75
    file_kind,
59
76
    isdir,
 
77
    normpath,
60
78
    pathjoin,
 
79
    rand_chars,
61
80
    realpath,
62
81
    safe_unicode,
 
82
    splitpath,
63
83
    )
64
 
from bzrlib.trace import mutter
 
84
from bzrlib.trace import mutter, note
65
85
from bzrlib.transport.local import LocalTransport
66
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
        )
67
95
from bzrlib.tree import Tree
68
96
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
69
97
 
70
98
 
71
 
class DirStateWorkingTree(WorkingTree3):
 
99
class WorkingTree4(WorkingTree3):
 
100
    """This is the Format 4 working tree.
 
101
 
 
102
    This differs from WorkingTree3 by:
 
103
     - Having a consolidated internal dirstate, stored in a
 
104
       randomly-accessible sorted file on disk.
 
105
     - Not having a regular inventory attribute.  One can be synthesized 
 
106
       on demand but this is expensive and should be avoided.
 
107
 
 
108
    This is new in bzr 0.15.
 
109
    """
 
110
 
72
111
    def __init__(self, basedir,
73
112
                 branch,
74
113
                 _control_files=None,
83
122
        """
84
123
        self._format = _format
85
124
        self.bzrdir = _bzrdir
 
125
        from bzrlib.trace import note, mutter
 
126
        assert isinstance(basedir, basestring), \
 
127
            "base directory %r is not a string" % basedir
86
128
        basedir = safe_unicode(basedir)
87
129
        mutter("opening working tree %r", basedir)
88
130
        self._branch = branch
 
131
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
132
            "branch %r is not a Branch" % self.branch
89
133
        self.basedir = realpath(basedir)
90
134
        # if branch is at our basedir and is a format 6 or less
91
135
        # assume all other formats have their own control files.
 
136
        assert isinstance(_control_files, LockableFiles), \
 
137
            "_control_files must be a LockableFiles, not %r" % _control_files
92
138
        self._control_files = _control_files
93
 
        self._transport = self._control_files._transport
94
139
        self._dirty = None
95
140
        #-------------
96
141
        # during a read or write lock these objects are set, and are
98
143
        self._dirstate = None
99
144
        self._inventory = None
100
145
        #-------------
101
 
        self._setup_directory_is_tree_reference()
102
 
        self._detect_case_handling()
103
 
        self._rules_searcher = None
104
 
        self.views = self._make_views()
105
 
        #--- allow tests to select the dirstate iter_changes implementation
106
 
        self._iter_changes = dirstate._process_entry
107
146
 
108
147
    @needs_tree_write_lock
109
148
    def _add(self, files, ids, kinds):
111
150
        state = self.current_dirstate()
112
151
        for f, file_id, kind in zip(files, ids, kinds):
113
152
            f = f.strip('/')
 
153
            assert '//' not in f
 
154
            assert '..' not in f
114
155
            if self.path2id(f):
115
156
                # special case tree root handling.
116
157
                if f == '' and self.path2id(f) == ROOT_ID:
137
178
    @needs_tree_write_lock
138
179
    def add_reference(self, sub_tree):
139
180
        # use standard implementation, which calls back to self._add
140
 
        #
 
181
        # 
141
182
        # So we don't store the reference_revision in the working dirstate,
142
 
        # it's just recorded at the moment of commit.
 
183
        # it's just recorded at the moment of commit. 
143
184
        self._add_reference(sub_tree)
144
185
 
145
186
    def break_lock(self):
184
225
            WorkingTree3._comparison_data(self, entry, path)
185
226
        # it looks like a plain directory, but it's really a reference -- see
186
227
        # also kind()
187
 
        if (self._repo_supports_tree_reference and kind == 'directory'
188
 
            and entry is not None and entry.kind == 'tree-reference'):
 
228
        if (self._repo_supports_tree_reference and
 
229
            kind == 'directory' and
 
230
            self._directory_is_tree_reference(path)):
189
231
            kind = 'tree-reference'
190
232
        return kind, executable, stat_value
191
233
 
217
259
            return self._dirstate
218
260
        local_path = self.bzrdir.get_workingtree_transport(None
219
261
            ).local_abspath('dirstate')
220
 
        self._dirstate = dirstate.DirState.on_file(local_path,
221
 
            self._sha1_provider())
 
262
        self._dirstate = dirstate.DirState.on_file(local_path)
222
263
        return self._dirstate
223
264
 
224
 
    def _sha1_provider(self):
225
 
        """A function that returns a SHA1Provider suitable for this tree.
226
 
 
227
 
        :return: None if content filtering is not supported by this tree.
228
 
          Otherwise, a SHA1Provider is returned that sha's the canonical
229
 
          form of files, i.e. after read filters are applied.
230
 
        """
231
 
        if self.supports_content_filtering():
232
 
            return ContentFilterAwareSHA1Provider(self)
233
 
        else:
234
 
            return None
 
265
    def _directory_is_tree_reference(self, relpath):
 
266
        # as a special case, if a directory contains control files then 
 
267
        # it's a tree reference, except that the root of the tree is not
 
268
        return relpath and osutils.isdir(self.abspath(relpath) + u"/.bzr")
 
269
        # TODO: We could ask all the control formats whether they
 
270
        # recognize this directory, but at the moment there's no cheap api
 
271
        # to do that.  Since we probably can only nest bzr checkouts and
 
272
        # they always use this name it's ok for now.  -- mbp 20060306
 
273
        #
 
274
        # FIXME: There is an unhandled case here of a subdirectory
 
275
        # containing .bzr but not a branch; that will probably blow up
 
276
        # when you try to commit it.  It might happen if there is a
 
277
        # checkout in a subdirectory.  This can be avoided by not adding
 
278
        # it.  mbp 20070306
235
279
 
236
280
    def filter_unversioned_files(self, paths):
237
281
        """Filter out paths that are versioned.
270
314
 
271
315
    def _generate_inventory(self):
272
316
        """Create and set self.inventory from the dirstate object.
273
 
 
 
317
        
274
318
        This is relatively expensive: we have to walk the entire dirstate.
275
319
        Ideally we would not, and can deprecate this function.
276
320
        """
280
324
        state._read_dirblocks_if_needed()
281
325
        root_key, current_entry = self._get_entry(path='')
282
326
        current_id = root_key[2]
283
 
        if not (current_entry[0][0] == 'd'): # directory
284
 
            raise AssertionError(current_entry)
 
327
        assert current_entry[0][0] == 'd' # directory
285
328
        inv = Inventory(root_id=current_id)
286
329
        # Turn some things into local variables
287
330
        minikind_to_kind = dirstate.DirState._minikind_to_kind
320
363
                    # add this entry to the parent map.
321
364
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
322
365
                elif kind == 'tree-reference':
323
 
                    if not self._repo_supports_tree_reference:
324
 
                        raise errors.UnsupportedOperation(
325
 
                            self._generate_inventory,
326
 
                            self.branch.repository)
 
366
                    assert self._repo_supports_tree_reference, \
 
367
                        "repository of %r " \
 
368
                        "doesn't support tree references " \
 
369
                        "required by entry %r" \
 
370
                        % (self, name)
327
371
                    inv_entry.reference_revision = link_or_sha1 or None
328
372
                elif kind != 'symlink':
329
373
                    raise AssertionError("unknown kind %r" % kind)
330
374
                # These checks cost us around 40ms on a 55k entry tree
331
 
                if file_id in inv_byid:
332
 
                    raise AssertionError('file_id %s already in'
333
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
334
 
                if name_unicode in parent_ie.children:
335
 
                    raise AssertionError('name %r already in parent'
336
 
                        % (name_unicode,))
 
375
                assert file_id not in inv_byid, ('file_id %s already in'
 
376
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
377
                assert name_unicode not in parent_ie.children
337
378
                inv_byid[file_id] = inv_entry
338
379
                parent_ie.children[name_unicode] = inv_entry
339
380
        self._inventory = inv
344
385
        If either file_id or path is supplied, it is used as the key to lookup.
345
386
        If both are supplied, the fastest lookup is used, and an error is
346
387
        raised if they do not both point at the same row.
347
 
 
 
388
        
348
389
        :param file_id: An optional unicode file_id to be looked up.
349
390
        :param path: An optional unicode path to be looked up.
350
391
        :return: The dirstate row tuple for path/file_id, or (None, None)
359
400
    def get_file_sha1(self, file_id, path=None, stat_value=None):
360
401
        # check file id is valid unconditionally.
361
402
        entry = self._get_entry(file_id=file_id, path=path)
362
 
        if entry[0] is None:
363
 
            raise errors.NoSuchId(self, file_id)
 
403
        assert entry[0] is not None, 'what error should this raise'
364
404
        if path is None:
365
405
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
366
406
 
367
407
        file_abspath = self.abspath(path)
368
408
        state = self.current_dirstate()
369
 
        if stat_value is None:
370
 
            try:
371
 
                stat_value = os.lstat(file_abspath)
372
 
            except OSError, e:
373
 
                if e.errno == errno.ENOENT:
374
 
                    return None
375
 
                else:
376
 
                    raise
377
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
378
 
            stat_value=stat_value)
 
409
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
410
                                          stat_value=stat_value)
379
411
        if entry[1][0][0] == 'f':
380
 
            if link_or_sha1 is None:
381
 
                file_obj, statvalue = self.get_file_with_stat(file_id, path)
382
 
                try:
383
 
                    sha1 = osutils.sha_file(file_obj)
384
 
                finally:
385
 
                    file_obj.close()
386
 
                self._observed_sha1(file_id, path, (sha1, statvalue))
387
 
                return sha1
388
 
            else:
389
 
                return link_or_sha1
 
412
            return link_or_sha1
390
413
        return None
391
414
 
392
415
    def _get_inventory(self):
393
416
        """Get the inventory for the tree. This is only valid within a lock."""
394
 
        if 'evil' in debug.debug_flags:
395
 
            trace.mutter_callsite(2,
396
 
                "accessing .inventory forces a size of tree translation.")
397
417
        if self._inventory is not None:
398
418
            return self._inventory
399
419
        self._must_be_locked()
406
426
    @needs_read_lock
407
427
    def get_parent_ids(self):
408
428
        """See Tree.get_parent_ids.
409
 
 
 
429
        
410
430
        This implementation requests the ids list from the dirstate file.
411
431
        """
412
432
        return self.current_dirstate().get_parent_ids()
428
448
 
429
449
    def has_id(self, file_id):
430
450
        state = self.current_dirstate()
 
451
        file_id = osutils.safe_file_id(file_id)
431
452
        row, parents = self._get_entry(file_id=file_id)
432
453
        if row is None:
433
454
            return False
434
455
        return osutils.lexists(pathjoin(
435
456
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
436
457
 
437
 
    def has_or_had_id(self, file_id):
438
 
        state = self.current_dirstate()
439
 
        row, parents = self._get_entry(file_id=file_id)
440
 
        return row is not None
441
 
 
442
458
    @needs_read_lock
443
459
    def id2path(self, file_id):
444
 
        "Convert a file-id to a path."
 
460
        file_id = osutils.safe_file_id(file_id)
445
461
        state = self.current_dirstate()
446
462
        entry = self._get_entry(file_id=file_id)
447
463
        if entry == (None, None):
449
465
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
450
466
        return path_utf8.decode('utf8')
451
467
 
452
 
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
453
 
        entry = self._get_entry(path=path)
454
 
        if entry == (None, None):
455
 
            return False # Missing entries are not executable
456
 
        return entry[1][0][3] # Executable?
457
 
 
458
468
    if not osutils.supports_executable():
 
469
        @needs_read_lock
459
470
        def is_executable(self, file_id, path=None):
460
 
            """Test if a file is executable or not.
461
 
 
462
 
            Note: The caller is expected to take a read-lock before calling this.
463
 
            """
 
471
            file_id = osutils.safe_file_id(file_id)
464
472
            entry = self._get_entry(file_id=file_id, path=path)
465
473
            if entry == (None, None):
466
474
                return False
467
475
            return entry[1][0][3]
468
 
 
469
 
        _is_executable_from_path_and_stat = \
470
 
            _is_executable_from_path_and_stat_from_basis
471
476
    else:
 
477
        @needs_read_lock
472
478
        def is_executable(self, file_id, path=None):
473
 
            """Test if a file is executable or not.
474
 
 
475
 
            Note: The caller is expected to take a read-lock before calling this.
476
 
            """
477
 
            self._must_be_locked()
478
479
            if not path:
 
480
                file_id = osutils.safe_file_id(file_id)
479
481
                path = self.id2path(file_id)
480
482
            mode = os.lstat(self.abspath(path)).st_mode
481
483
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
482
484
 
483
 
    def all_file_ids(self):
484
 
        """See Tree.iter_all_file_ids"""
485
 
        self._must_be_locked()
486
 
        result = set()
487
 
        for key, tree_details in self.current_dirstate()._iter_entries():
488
 
            if tree_details[0][0] in ('a', 'r'): # relocated
489
 
                continue
490
 
            result.add(key[2])
491
 
        return result
492
 
 
493
485
    @needs_read_lock
494
486
    def __iter__(self):
495
487
        """Iterate through file_ids for this tree.
508
500
        return iter(result)
509
501
 
510
502
    def iter_references(self):
511
 
        if not self._repo_supports_tree_reference:
512
 
            # When the repo doesn't support references, we will have nothing to
513
 
            # return
514
 
            return
515
503
        for key, tree_details in self.current_dirstate()._iter_entries():
516
504
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
517
505
                # not relevant to the working tree
519
507
            if not key[1]:
520
508
                # the root is not a reference.
521
509
                continue
522
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
510
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
523
511
            try:
524
 
                if self._kind(relpath) == 'tree-reference':
525
 
                    yield relpath, key[2]
 
512
                if self._kind(path) == 'tree-reference':
 
513
                    yield path, key[2]
526
514
            except errors.NoSuchFile:
527
515
                # path is missing on disk.
528
516
                continue
529
517
 
530
 
    def _observed_sha1(self, file_id, path, (sha1, statvalue)):
531
 
        """See MutableTree._observed_sha1."""
532
 
        state = self.current_dirstate()
533
 
        entry = self._get_entry(file_id=file_id, path=path)
534
 
        state._observed_sha1(entry, sha1, statvalue)
535
 
 
 
518
    @needs_read_lock
536
519
    def kind(self, file_id):
537
520
        """Return the kind of a file.
538
521
 
539
522
        This is always the actual kind that's on disk, regardless of what it
540
523
        was added as.
541
 
 
542
 
        Note: The caller is expected to take a read-lock before calling this.
543
524
        """
544
525
        relpath = self.id2path(file_id)
545
 
        if relpath is None:
546
 
            raise AssertionError(
547
 
                "path for id {%s} is None!" % file_id)
 
526
        assert relpath != None, \
 
527
            "path for id {%s} is None!" % file_id
548
528
        return self._kind(relpath)
549
529
 
550
530
    def _kind(self, relpath):
551
531
        abspath = self.abspath(relpath)
552
532
        kind = file_kind(abspath)
553
 
        if (self._repo_supports_tree_reference and kind == 'directory'):
554
 
            entry = self._get_entry(path=relpath)
555
 
            if entry[1] is not None:
556
 
                if entry[1][0][0] == 't':
557
 
                    kind = 'tree-reference'
 
533
        if (self._repo_supports_tree_reference and
 
534
            kind == 'directory' and
 
535
            self._directory_is_tree_reference(relpath)):
 
536
            kind = 'tree-reference'
558
537
        return kind
559
538
 
560
539
    @needs_read_lock
564
543
        if parent_ids:
565
544
            return parent_ids[0]
566
545
        else:
567
 
            return _mod_revision.NULL_REVISION
 
546
            return None
568
547
 
569
548
    def lock_read(self):
570
549
        """See Branch.lock_read, and WorkingTree.unlock."""
578
557
                # set our support for tree references from the repository in
579
558
                # use.
580
559
                self._repo_supports_tree_reference = getattr(
581
 
                    self.branch.repository._format, "supports_tree_reference",
 
560
                    self.branch.repository._format, "support_tree_reference",
582
561
                    False)
583
562
            except:
584
563
                self._control_files.unlock()
598
577
                # set our support for tree references from the repository in
599
578
                # use.
600
579
                self._repo_supports_tree_reference = getattr(
601
 
                    self.branch.repository._format, "supports_tree_reference",
 
580
                    self.branch.repository._format, "support_tree_reference",
602
581
                    False)
603
582
            except:
604
583
                self._control_files.unlock()
623
602
        result = []
624
603
        if not from_paths:
625
604
            return result
 
605
 
626
606
        state = self.current_dirstate()
627
 
        if isinstance(from_paths, basestring):
628
 
            raise ValueError()
 
607
 
 
608
        assert not isinstance(from_paths, basestring)
629
609
        to_dir_utf8 = to_dir.encode('utf8')
630
610
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
631
611
        id_index = state._get_id_index()
653
633
        if self._inventory is not None:
654
634
            update_inventory = True
655
635
            inv = self.inventory
 
636
            to_dir_ie = inv[to_dir_id]
656
637
            to_dir_id = to_entry[0][2]
657
 
            to_dir_ie = inv[to_dir_id]
658
638
        else:
659
639
            update_inventory = False
660
640
 
683
663
            new_entry = to_block[1][added_entry_index]
684
664
            rollbacks.append(lambda:state._make_absent(new_entry))
685
665
 
 
666
        # create rename entries and tuples
686
667
        for from_rel in from_paths:
687
668
            # from_rel is 'pathinroot/foo/bar'
688
669
            from_rel_utf8 = from_rel.encode('utf8')
691
672
            from_entry = self._get_entry(path=from_rel)
692
673
            if from_entry == (None, None):
693
674
                raise errors.BzrMoveFailedError(from_rel,to_dir,
694
 
                    errors.NotVersionedError(path=from_rel))
 
675
                    errors.NotVersionedError(path=str(from_rel)))
695
676
 
696
677
            from_id = from_entry[0][2]
697
678
            to_rel = pathjoin(to_dir, from_tail)
724
705
                if from_missing: # implicitly just update our path mapping
725
706
                    move_file = False
726
707
                elif not after:
727
 
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
708
                    raise errors.RenameFailedFilesExist(from_rel, to_rel,
 
709
                        extra="(Use --after to update the Bazaar id)")
728
710
 
729
711
            rollbacks = []
730
712
            def rollback_rename():
785
767
 
786
768
                if minikind == 'd':
787
769
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
788
 
                        """Recursively update all entries in this dirblock."""
789
 
                        if from_dir == '':
790
 
                            raise AssertionError("renaming root not supported")
 
770
                        """all entries in this block need updating.
 
771
 
 
772
                        TODO: This is pretty ugly, and doesn't support
 
773
                        reverting, but it works.
 
774
                        """
 
775
                        assert from_dir != '', "renaming root not supported"
791
776
                        from_key = (from_dir, '')
792
777
                        from_block_idx, present = \
793
778
                            state._find_block_index_from_key(from_key)
803
788
                        to_block_index = state._ensure_block(
804
789
                            to_block_index, to_entry_index, to_dir_utf8)
805
790
                        to_block = state._dirblocks[to_block_index]
806
 
 
807
 
                        # Grab a copy since move_one may update the list.
808
 
                        for entry in from_block[1][:]:
809
 
                            if not (entry[0][0] == from_dir):
810
 
                                raise AssertionError()
 
791
                        for entry in from_block[1]:
 
792
                            assert entry[0][0] == from_dir
811
793
                            cur_details = entry[1][0]
812
794
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
813
795
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
814
796
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
815
797
                            minikind = cur_details[0]
816
 
                            if minikind in 'ar':
817
 
                                # Deleted children of a renamed directory
818
 
                                # Do not need to be updated.
819
 
                                # Children that have been renamed out of this
820
 
                                # directory should also not be updated
821
 
                                continue
822
798
                            move_one(entry, from_path_utf8=from_path_utf8,
823
799
                                     minikind=minikind,
824
800
                                     executable=cur_details[3],
827
803
                                     size=cur_details[2],
828
804
                                     to_block=to_block,
829
805
                                     to_key=to_key,
830
 
                                     to_path_utf8=to_path_utf8)
 
806
                                     to_path_utf8=to_rel_utf8)
831
807
                            if minikind == 'd':
832
808
                                # We need to move all the children of this
833
809
                                # entry
872
848
        for tree in trees:
873
849
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
874
850
                parents):
875
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
876
 
                    trees, require_versioned)
 
851
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
877
852
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
878
853
        # -- make all paths utf8 --
879
854
        paths_utf8 = set()
939
914
            if not all_versioned:
940
915
                raise errors.PathsNotVersionedError(paths)
941
916
        # -- remove redundancy in supplied paths to prevent over-scanning --
942
 
        search_paths = osutils.minimum_path_selection(paths)
943
 
        # sketch:
 
917
        search_paths = set()
 
918
        for path in paths:
 
919
            other_paths = paths.difference(set([path]))
 
920
            if not osutils.is_inside_any(other_paths, path):
 
921
                # this is a top level path, we must check it.
 
922
                search_paths.add(path)
 
923
        # sketch: 
944
924
        # for all search_indexs in each path at or under each element of
945
925
        # search_paths, if the detail is relocated: add the id, and add the
946
926
        # relocated path as one to search if its not searched already. If the
1002
982
 
1003
983
    def read_working_inventory(self):
1004
984
        """Read the working inventory.
1005
 
 
 
985
        
1006
986
        This is a meaningless operation for dirstate, but we obey it anyhow.
1007
987
        """
1008
988
        return self.inventory
1013
993
 
1014
994
        WorkingTree4 supplies revision_trees for any basis tree.
1015
995
        """
 
996
        revision_id = osutils.safe_revision_id(revision_id)
1016
997
        dirstate = self.current_dirstate()
1017
998
        parent_ids = dirstate.get_parent_ids()
1018
999
        if revision_id not in parent_ids:
1025
1006
    @needs_tree_write_lock
1026
1007
    def set_last_revision(self, new_revision):
1027
1008
        """Change the last revision in the working tree."""
 
1009
        new_revision = osutils.safe_revision_id(new_revision)
1028
1010
        parents = self.get_parent_ids()
1029
 
        if new_revision in (_mod_revision.NULL_REVISION, None):
1030
 
            if len(parents) >= 2:
1031
 
                raise AssertionError(
1032
 
                    "setting the last parent to none with a pending merge is "
1033
 
                    "unsupported.")
 
1011
        if new_revision in (NULL_REVISION, None):
 
1012
            assert len(parents) < 2, (
 
1013
                "setting the last parent to none with a pending merge is "
 
1014
                "unsupported.")
1034
1015
            self.set_parent_ids([])
1035
1016
        else:
1036
1017
            self.set_parent_ids([new_revision] + parents[1:],
1039
1020
    @needs_tree_write_lock
1040
1021
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1041
1022
        """Set the parent ids to revision_ids.
1042
 
 
 
1023
        
1043
1024
        See also set_parent_trees. This api will try to retrieve the tree data
1044
1025
        for each element of revision_ids from the trees repository. If you have
1045
1026
        tree data already available, it is more efficient to use
1049
1030
        :param revision_ids: The revision_ids to set as the parent ids of this
1050
1031
            working tree. Any of these may be ghosts.
1051
1032
        """
 
1033
        revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1052
1034
        trees = []
1053
1035
        for revision_id in revision_ids:
1054
1036
            try:
1060
1042
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
1061
1043
                revtree = None
1062
1044
            trees.append((revision_id, revtree))
 
1045
        self.current_dirstate()._validate()
1063
1046
        self.set_parent_trees(trees,
1064
1047
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1048
        self.current_dirstate()._validate()
1065
1049
 
1066
1050
    @needs_tree_write_lock
1067
1051
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
1072
1056
            parent tree - i.e. a ghost.
1073
1057
        """
1074
1058
        dirstate = self.current_dirstate()
 
1059
        dirstate._validate()
1075
1060
        if len(parents_list) > 0:
1076
1061
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
1077
1062
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1078
1063
        real_trees = []
1079
1064
        ghosts = []
1080
 
 
1081
 
        parent_ids = [rev_id for rev_id, tree in parents_list]
1082
 
        graph = self.branch.repository.get_graph()
1083
 
        heads = graph.heads(parent_ids)
1084
 
        accepted_revisions = set()
1085
 
 
1086
1065
        # convert absent trees to the null tree, which we convert back to
1087
1066
        # missing on access.
1088
1067
        for rev_id, tree in parents_list:
1089
 
            if len(accepted_revisions) > 0:
1090
 
                # we always accept the first tree
1091
 
                if rev_id in accepted_revisions or rev_id not in heads:
1092
 
                    # We have already included either this tree, or its
1093
 
                    # descendent, so we skip it.
1094
 
                    continue
1095
 
            _mod_revision.check_not_reserved_id(rev_id)
 
1068
            rev_id = osutils.safe_revision_id(rev_id)
1096
1069
            if tree is not None:
1097
1070
                real_trees.append((rev_id, tree))
1098
1071
            else:
1099
1072
                real_trees.append((rev_id,
1100
 
                    self.branch.repository.revision_tree(
1101
 
                        _mod_revision.NULL_REVISION)))
 
1073
                    self.branch.repository.revision_tree(None)))
1102
1074
                ghosts.append(rev_id)
1103
 
            accepted_revisions.add(rev_id)
 
1075
        dirstate._validate()
1104
1076
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1077
        dirstate._validate()
1105
1078
        self._make_dirty(reset_inventory=False)
 
1079
        dirstate._validate()
1106
1080
 
1107
1081
    def _set_root_id(self, file_id):
1108
1082
        """See WorkingTree.set_root_id."""
1111
1085
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
1112
1086
            self._make_dirty(reset_inventory=True)
1113
1087
 
1114
 
    def _sha_from_stat(self, path, stat_result):
1115
 
        """Get a sha digest from the tree's stat cache.
1116
 
 
1117
 
        The default implementation assumes no stat cache is present.
1118
 
 
1119
 
        :param path: The path.
1120
 
        :param stat_result: The stat result being looked up.
1121
 
        """
1122
 
        return self.current_dirstate().sha1_from_stat(path, stat_result)
1123
 
 
1124
1088
    @needs_read_lock
1125
1089
    def supports_tree_reference(self):
1126
1090
        return self._repo_supports_tree_reference
1127
1091
 
1128
1092
    def unlock(self):
1129
1093
        """Unlock in format 4 trees needs to write the entire dirstate."""
1130
 
        # do non-implementation specific cleanup
1131
 
        self._cleanup()
1132
 
 
1133
1094
        if self._control_files._lock_count == 1:
1134
1095
            # eventually we should do signature checking during read locks for
1135
1096
            # dirstate updates.
1166
1127
            return
1167
1128
        state = self.current_dirstate()
1168
1129
        state._read_dirblocks_if_needed()
1169
 
        ids_to_unversion = set(file_ids)
 
1130
        ids_to_unversion = set()
 
1131
        for file_id in file_ids:
 
1132
            ids_to_unversion.add(osutils.safe_file_id(file_id))
1170
1133
        paths_to_unversion = set()
1171
1134
        # sketch:
1172
1135
        # check if the root is to be unversioned, if so, assert for now.
1199
1162
                # just forget the whole block.
1200
1163
                entry_index = 0
1201
1164
                while entry_index < len(block[1]):
1202
 
                    entry = block[1][entry_index]
1203
 
                    if entry[1][0][0] in 'ar':
1204
 
                        # don't remove absent or renamed entries
 
1165
                    # Mark this file id as having been removed
 
1166
                    ids_to_unversion.discard(block[1][entry_index][0][2])
 
1167
                    if not state._make_absent(block[1][entry_index]):
1205
1168
                        entry_index += 1
1206
 
                    else:
1207
 
                        # Mark this file id as having been removed
1208
 
                        ids_to_unversion.discard(entry[0][2])
1209
 
                        if not state._make_absent(entry):
1210
 
                            # The block has not shrunk.
1211
 
                            entry_index += 1
1212
1169
                # go to the next block. (At the moment we dont delete empty
1213
1170
                # dirblocks)
1214
1171
                block_index += 1
1238
1195
                self._inventory.remove_recursive_id(file_id)
1239
1196
 
1240
1197
    @needs_tree_write_lock
1241
 
    def rename_one(self, from_rel, to_rel, after=False):
1242
 
        """See WorkingTree.rename_one"""
1243
 
        self.flush()
1244
 
        WorkingTree.rename_one(self, from_rel, to_rel, after)
1245
 
 
1246
 
    @needs_tree_write_lock
1247
 
    def apply_inventory_delta(self, changes):
1248
 
        """See MutableTree.apply_inventory_delta"""
1249
 
        state = self.current_dirstate()
1250
 
        state.update_by_delta(changes)
1251
 
        self._make_dirty(reset_inventory=True)
1252
 
 
1253
 
    def update_basis_by_delta(self, new_revid, delta):
1254
 
        """See MutableTree.update_basis_by_delta."""
1255
 
        if self.last_revision() == new_revid:
1256
 
            raise AssertionError()
1257
 
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
1258
 
 
1259
 
    @needs_read_lock
1260
 
    def _validate(self):
1261
 
        self._dirstate._validate()
1262
 
 
1263
 
    @needs_tree_write_lock
1264
1198
    def _write_inventory(self, inv):
1265
1199
        """Write inventory as the current inventory."""
1266
 
        if self._dirty:
1267
 
            raise AssertionError("attempting to write an inventory when the "
1268
 
                "dirstate is dirty will lose pending changes")
1269
 
        had_inventory = self._inventory is not None
1270
 
        # Setting self._inventory = None forces the dirstate to regenerate the
1271
 
        # working inventory. We do this because self.inventory may be inv, or
1272
 
        # may have been modified, and either case would prevent a clean delta
1273
 
        # being created.
1274
 
        self._inventory = None
1275
 
        # generate a delta,
1276
 
        delta = inv._make_delta(self.inventory)
1277
 
        # and apply it.
1278
 
        self.apply_inventory_delta(delta)
1279
 
        if had_inventory:
 
1200
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1201
        self.current_dirstate().set_state_from_inventory(inv)
 
1202
        self._make_dirty(reset_inventory=False)
 
1203
        if self._inventory is not None:
1280
1204
            self._inventory = inv
1281
1205
        self.flush()
1282
1206
 
1283
1207
 
1284
 
class ContentFilterAwareSHA1Provider(dirstate.SHA1Provider):
1285
 
 
1286
 
    def __init__(self, tree):
1287
 
        self.tree = tree
1288
 
 
1289
 
    def sha1(self, abspath):
1290
 
        """See dirstate.SHA1Provider.sha1()."""
1291
 
        filters = self.tree._content_filter_stack(
1292
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1293
 
        return internal_size_sha_file_byname(abspath, filters)[1]
1294
 
 
1295
 
    def stat_and_sha1(self, abspath):
1296
 
        """See dirstate.SHA1Provider.stat_and_sha1()."""
1297
 
        filters = self.tree._content_filter_stack(
1298
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1299
 
        file_obj = file(abspath, 'rb', 65000)
1300
 
        try:
1301
 
            statvalue = os.fstat(file_obj.fileno())
1302
 
            if filters:
1303
 
                file_obj = filtered_input_file(file_obj, filters)
1304
 
            sha1 = osutils.size_sha_file(file_obj)[1]
1305
 
        finally:
1306
 
            file_obj.close()
1307
 
        return statvalue, sha1
1308
 
 
1309
 
 
1310
 
class ContentFilteringDirStateWorkingTree(DirStateWorkingTree):
1311
 
    """Dirstate working tree that supports content filtering.
1312
 
 
1313
 
    The dirstate holds the hash and size of the canonical form of the file, 
1314
 
    and most methods must return that.
1315
 
    """
1316
 
 
1317
 
    def _file_content_summary(self, path, stat_result):
1318
 
        # This is to support the somewhat obsolete path_content_summary method
1319
 
        # with content filtering: see
1320
 
        # <https://bugs.edge.launchpad.net/bzr/+bug/415508>.
1321
 
        #
1322
 
        # If the dirstate cache is up to date and knows the hash and size,
1323
 
        # return that.
1324
 
        # Otherwise if there are no content filters, return the on-disk size
1325
 
        # and leave the hash blank.
1326
 
        # Otherwise, read and filter the on-disk file and use its size and
1327
 
        # hash.
1328
 
        #
1329
 
        # The dirstate doesn't store the size of the canonical form so we
1330
 
        # can't trust it for content-filtered trees.  We just return None.
1331
 
        dirstate_sha1 = self._dirstate.sha1_from_stat(path, stat_result)
1332
 
        executable = self._is_executable_from_path_and_stat(path, stat_result)
1333
 
        return ('file', None, executable, dirstate_sha1)
1334
 
 
1335
 
 
1336
 
class WorkingTree4(DirStateWorkingTree):
1337
 
    """This is the Format 4 working tree.
1338
 
 
1339
 
    This differs from WorkingTree3 by:
1340
 
     - Having a consolidated internal dirstate, stored in a
1341
 
       randomly-accessible sorted file on disk.
1342
 
     - Not having a regular inventory attribute.  One can be synthesized
1343
 
       on demand but this is expensive and should be avoided.
1344
 
 
1345
 
    This is new in bzr 0.15.
1346
 
    """
1347
 
 
1348
 
 
1349
 
class WorkingTree5(ContentFilteringDirStateWorkingTree):
1350
 
    """This is the Format 5 working tree.
1351
 
 
1352
 
    This differs from WorkingTree4 by:
1353
 
     - Supporting content filtering.
1354
 
 
1355
 
    This is new in bzr 1.11.
1356
 
    """
1357
 
 
1358
 
 
1359
 
class WorkingTree6(ContentFilteringDirStateWorkingTree):
1360
 
    """This is the Format 6 working tree.
1361
 
 
1362
 
    This differs from WorkingTree5 by:
1363
 
     - Supporting a current view that may mask the set of files in a tree
1364
 
       impacted by most user operations.
1365
 
 
1366
 
    This is new in bzr 1.14.
1367
 
    """
1368
 
 
1369
 
    def _make_views(self):
1370
 
        return views.PathBasedViews(self)
1371
 
 
1372
 
 
1373
 
class DirStateWorkingTreeFormat(WorkingTreeFormat3):
1374
 
 
1375
 
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
1376
 
                   accelerator_tree=None, hardlink=False):
 
1208
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1209
    """The first consolidated dirstate working tree format.
 
1210
 
 
1211
    This format:
 
1212
        - exists within a metadir controlling .bzr
 
1213
        - includes an explicit version marker for the workingtree control
 
1214
          files, separate from the BzrDir format
 
1215
        - modifies the hash cache format
 
1216
        - is new in bzr 0.15
 
1217
        - uses a LockDir to guard access to it.
 
1218
    """
 
1219
 
 
1220
    def get_format_string(self):
 
1221
        """See WorkingTreeFormat.get_format_string()."""
 
1222
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1223
 
 
1224
    def get_format_description(self):
 
1225
        """See WorkingTreeFormat.get_format_description()."""
 
1226
        return "Working tree format 4"
 
1227
 
 
1228
    def initialize(self, a_bzrdir, revision_id=None):
1377
1229
        """See WorkingTreeFormat.initialize().
1378
1230
 
1379
1231
        :param revision_id: allows creating a working tree at a different
1380
1232
        revision than the branch is at.
1381
 
        :param accelerator_tree: A tree which can be used for retrieving file
1382
 
            contents more quickly than the revision tree, i.e. a workingtree.
1383
 
            The revision tree will be used for cases where accelerator_tree's
1384
 
            content is different.
1385
 
        :param hardlink: If true, hard-link files from accelerator_tree,
1386
 
            where possible.
1387
1233
 
1388
 
        These trees get an initial random root id, if their repository supports
1389
 
        rich root data, TREE_ROOT otherwise.
 
1234
        These trees get an initial random root id.
1390
1235
        """
 
1236
        revision_id = osutils.safe_revision_id(revision_id)
1391
1237
        if not isinstance(a_bzrdir.transport, LocalTransport):
1392
1238
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
1393
1239
        transport = a_bzrdir.get_workingtree_transport(self)
1394
1240
        control_files = self._open_control_files(a_bzrdir)
1395
1241
        control_files.create_lock()
1396
1242
        control_files.lock_write()
1397
 
        transport.put_bytes('format', self.get_format_string(),
1398
 
            mode=a_bzrdir._get_file_mode())
1399
 
        if from_branch is not None:
1400
 
            branch = from_branch
1401
 
        else:
1402
 
            branch = a_bzrdir.open_branch()
 
1243
        control_files.put_utf8('format', self.get_format_string())
 
1244
        branch = a_bzrdir.open_branch()
1403
1245
        if revision_id is None:
1404
1246
            revision_id = branch.last_revision()
1405
1247
        local_path = transport.local_abspath('dirstate')
1406
1248
        # write out new dirstate (must exist when we create the tree)
1407
1249
        state = dirstate.DirState.initialize(local_path)
1408
1250
        state.unlock()
1409
 
        del state
1410
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1251
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1411
1252
                         branch,
1412
1253
                         _format=self,
1413
1254
                         _bzrdir=a_bzrdir,
1414
1255
                         _control_files=control_files)
1415
1256
        wt._new_tree()
1416
1257
        wt.lock_tree_write()
 
1258
        state._validate()
1417
1259
        try:
1418
 
            self._init_custom_control_files(wt)
1419
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
1420
 
                if branch.repository.supports_rich_root():
1421
 
                    wt._set_root_id(generate_ids.gen_root_id())
1422
 
                else:
1423
 
                    wt._set_root_id(ROOT_ID)
 
1260
            if revision_id in (None, NULL_REVISION):
 
1261
                wt._set_root_id(generate_ids.gen_root_id())
1424
1262
                wt.flush()
1425
 
            basis = None
1426
 
            # frequently, we will get here due to branching.  The accelerator
1427
 
            # tree will be the tree from the branch, so the desired basis
1428
 
            # tree will often be a parent of the accelerator tree.
1429
 
            if accelerator_tree is not None:
1430
 
                try:
1431
 
                    basis = accelerator_tree.revision_tree(revision_id)
1432
 
                except errors.NoSuchRevision:
1433
 
                    pass
1434
 
            if basis is None:
1435
 
                basis = branch.repository.revision_tree(revision_id)
1436
 
            if revision_id == _mod_revision.NULL_REVISION:
1437
 
                parents_list = []
1438
 
            else:
1439
 
                parents_list = [(revision_id, basis)]
 
1263
                wt.current_dirstate()._validate()
 
1264
            wt.set_last_revision(revision_id)
 
1265
            wt.flush()
 
1266
            basis = wt.basis_tree()
1440
1267
            basis.lock_read()
1441
 
            try:
1442
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1268
            # if the basis has a root id we have to use that; otherwise we use
 
1269
            # a new random one
 
1270
            basis_root_id = basis.get_root_id()
 
1271
            if basis_root_id is not None:
 
1272
                wt._set_root_id(basis_root_id)
1443
1273
                wt.flush()
1444
 
                # if the basis has a root id we have to use that; otherwise we
1445
 
                # use a new random one
1446
 
                basis_root_id = basis.get_root_id()
1447
 
                if basis_root_id is not None:
1448
 
                    wt._set_root_id(basis_root_id)
1449
 
                    wt.flush()
1450
 
                if wt.supports_content_filtering():
1451
 
                    # The original tree may not have the same content filters
1452
 
                    # applied so we can't safely build the inventory delta from
1453
 
                    # the source tree.
1454
 
                    delta_from_tree = False
1455
 
                else:
1456
 
                    delta_from_tree = True
1457
 
                # delta_from_tree is safe even for DirStateRevisionTrees,
1458
 
                # because wt4.apply_inventory_delta does not mutate the input
1459
 
                # inventory entries.
1460
 
                transform.build_tree(basis, wt, accelerator_tree,
1461
 
                                     hardlink=hardlink,
1462
 
                                     delta_from_tree=delta_from_tree)
1463
 
            finally:
1464
 
                basis.unlock()
 
1274
            transform.build_tree(basis, wt)
 
1275
            basis.unlock()
1465
1276
        finally:
1466
1277
            control_files.unlock()
1467
1278
            wt.unlock()
1468
1279
        return wt
1469
1280
 
1470
 
    def _init_custom_control_files(self, wt):
1471
 
        """Subclasses with custom control files should override this method.
1472
 
 
1473
 
        The working tree and control files are locked for writing when this
1474
 
        method is called.
1475
 
 
1476
 
        :param wt: the WorkingTree object
1477
 
        """
1478
 
 
1479
1281
    def _open(self, a_bzrdir, control_files):
1480
1282
        """Open the tree itself.
1481
1283
 
1482
1284
        :param a_bzrdir: the dir for the tree.
1483
1285
        :param control_files: the control files for the tree.
1484
1286
        """
1485
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1287
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1486
1288
                           branch=a_bzrdir.open_branch(),
1487
1289
                           _format=self,
1488
1290
                           _bzrdir=a_bzrdir,
1489
1291
                           _control_files=control_files)
1490
1292
 
1491
1293
    def __get_matchingbzrdir(self):
1492
 
        return self._get_matchingbzrdir()
1493
 
 
1494
 
    def _get_matchingbzrdir(self):
1495
 
        """Overrideable method to get a bzrdir for testing."""
1496
1294
        # please test against something that will let us do tree references
1497
1295
        return bzrdir.format_registry.make_bzrdir(
1498
1296
            'dirstate-with-subtree')
1500
1298
    _matchingbzrdir = property(__get_matchingbzrdir)
1501
1299
 
1502
1300
 
1503
 
class WorkingTreeFormat4(DirStateWorkingTreeFormat):
1504
 
    """The first consolidated dirstate working tree format.
1505
 
 
1506
 
    This format:
1507
 
        - exists within a metadir controlling .bzr
1508
 
        - includes an explicit version marker for the workingtree control
1509
 
          files, separate from the BzrDir format
1510
 
        - modifies the hash cache format
1511
 
        - is new in bzr 0.15
1512
 
        - uses a LockDir to guard access to it.
1513
 
    """
1514
 
 
1515
 
    upgrade_recommended = False
1516
 
 
1517
 
    _tree_class = WorkingTree4
1518
 
 
1519
 
    def get_format_string(self):
1520
 
        """See WorkingTreeFormat.get_format_string()."""
1521
 
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
1522
 
 
1523
 
    def get_format_description(self):
1524
 
        """See WorkingTreeFormat.get_format_description()."""
1525
 
        return "Working tree format 4"
1526
 
 
1527
 
 
1528
 
class WorkingTreeFormat5(DirStateWorkingTreeFormat):
1529
 
    """WorkingTree format supporting content filtering.
1530
 
    """
1531
 
 
1532
 
    upgrade_recommended = False
1533
 
 
1534
 
    _tree_class = WorkingTree5
1535
 
 
1536
 
    def get_format_string(self):
1537
 
        """See WorkingTreeFormat.get_format_string()."""
1538
 
        return "Bazaar Working Tree Format 5 (bzr 1.11)\n"
1539
 
 
1540
 
    def get_format_description(self):
1541
 
        """See WorkingTreeFormat.get_format_description()."""
1542
 
        return "Working tree format 5"
1543
 
 
1544
 
    def supports_content_filtering(self):
1545
 
        return True
1546
 
 
1547
 
 
1548
 
class WorkingTreeFormat6(DirStateWorkingTreeFormat):
1549
 
    """WorkingTree format supporting views.
1550
 
    """
1551
 
 
1552
 
    upgrade_recommended = False
1553
 
 
1554
 
    _tree_class = WorkingTree6
1555
 
 
1556
 
    def get_format_string(self):
1557
 
        """See WorkingTreeFormat.get_format_string()."""
1558
 
        return "Bazaar Working Tree Format 6 (bzr 1.14)\n"
1559
 
 
1560
 
    def get_format_description(self):
1561
 
        """See WorkingTreeFormat.get_format_description()."""
1562
 
        return "Working tree format 6"
1563
 
 
1564
 
    def _init_custom_control_files(self, wt):
1565
 
        """Subclasses with custom control files should override this method."""
1566
 
        wt._transport.put_bytes('views', '', mode=wt.bzrdir._get_file_mode())
1567
 
 
1568
 
    def supports_content_filtering(self):
1569
 
        return True
1570
 
 
1571
 
    def supports_views(self):
1572
 
        return True
1573
 
 
1574
 
 
1575
1301
class DirStateRevisionTree(Tree):
1576
 
    """A revision tree pulling the inventory from a dirstate.
1577
 
    
1578
 
    Note that this is one of the historical (ie revision) trees cached in the
1579
 
    dirstate for easy access, not the workingtree.
1580
 
    """
 
1302
    """A revision tree pulling the inventory from a dirstate."""
1581
1303
 
1582
1304
    def __init__(self, dirstate, revision_id, repository):
1583
1305
        self._dirstate = dirstate
1584
 
        self._revision_id = revision_id
 
1306
        self._revision_id = osutils.safe_revision_id(revision_id)
1585
1307
        self._repository = repository
1586
1308
        self._inventory = None
1587
1309
        self._locked = 0
1588
1310
        self._dirstate_locked = False
1589
 
        self._repo_supports_tree_reference = getattr(
1590
 
            repository._format, "supports_tree_reference",
1591
 
            False)
1592
1311
 
1593
1312
    def __repr__(self):
1594
1313
        return "<%s of %s in %s>" % \
1595
1314
            (self.__class__.__name__, self._revision_id, self._dirstate)
1596
1315
 
1597
 
    def annotate_iter(self, file_id,
1598
 
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1316
    def annotate_iter(self, file_id):
1599
1317
        """See Tree.annotate_iter"""
1600
 
        text_key = (file_id, self.inventory[file_id].revision)
1601
 
        annotations = self._repository.texts.annotate(text_key)
1602
 
        return [(key[-1], line) for (key, line) in annotations]
 
1318
        w = self._repository.weave_store.get_weave(file_id,
 
1319
                           self._repository.get_transaction())
 
1320
        return w.annotate_iter(self.inventory[file_id].revision)
1603
1321
 
1604
 
    def _get_ancestors(self, default_revision):
1605
 
        return set(self._repository.get_ancestry(self._revision_id,
1606
 
                                                 topo_sorted=False))
1607
1322
    def _comparison_data(self, entry, path):
1608
1323
        """See Tree._comparison_data."""
1609
1324
        if entry is None:
1626
1341
    def get_root_id(self):
1627
1342
        return self.path2id('')
1628
1343
 
1629
 
    def id2path(self, file_id):
1630
 
        "Convert a file-id to a path."
1631
 
        entry = self._get_entry(file_id=file_id)
1632
 
        if entry == (None, None):
1633
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1634
 
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1635
 
        return path_utf8.decode('utf8')
1636
 
 
1637
 
    def iter_references(self):
1638
 
        if not self._repo_supports_tree_reference:
1639
 
            # When the repo doesn't support references, we will have nothing to
1640
 
            # return
1641
 
            return iter([])
1642
 
        # Otherwise, fall back to the default implementation
1643
 
        return super(DirStateRevisionTree, self).iter_references()
1644
 
 
1645
1344
    def _get_parent_index(self):
1646
1345
        """Return the index in the dirstate referenced by this tree."""
1647
1346
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1652
1351
        If either file_id or path is supplied, it is used as the key to lookup.
1653
1352
        If both are supplied, the fastest lookup is used, and an error is
1654
1353
        raised if they do not both point at the same row.
1655
 
 
 
1354
        
1656
1355
        :param file_id: An optional unicode file_id to be looked up.
1657
1356
        :param path: An optional unicode path to be looked up.
1658
1357
        :return: The dirstate row tuple for path/file_id, or (None, None)
1659
1358
        """
1660
1359
        if file_id is None and path is None:
1661
1360
            raise errors.BzrError('must supply file_id or path')
 
1361
        file_id = osutils.safe_file_id(file_id)
1662
1362
        if path is not None:
1663
1363
            path = path.encode('utf8')
1664
1364
        parent_index = self._get_parent_index()
1672
1372
 
1673
1373
        This is relatively expensive: we have to walk the entire dirstate.
1674
1374
        """
1675
 
        if not self._locked:
1676
 
            raise AssertionError(
1677
 
                'cannot generate inventory of an unlocked '
1678
 
                'dirstate revision tree')
 
1375
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1376
            'dirstate revision tree'
1679
1377
        # separate call for profiling - makes it clear where the costs are.
1680
1378
        self._dirstate._read_dirblocks_if_needed()
1681
 
        if self._revision_id not in self._dirstate.get_parent_ids():
1682
 
            raise AssertionError(
1683
 
                'parent %s has disappeared from %s' % (
1684
 
                self._revision_id, self._dirstate.get_parent_ids()))
 
1379
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1380
            'parent %s has disappeared from %s' % (
 
1381
            self._revision_id, self._dirstate.get_parent_ids())
1685
1382
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1686
1383
        # This is identical now to the WorkingTree _generate_inventory except
1687
1384
        # for the tree index use.
1688
1385
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1689
1386
        current_id = root_key[2]
1690
 
        if current_entry[parent_index][0] != 'd':
1691
 
            raise AssertionError()
 
1387
        assert current_entry[parent_index][0] == 'd'
1692
1388
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1693
1389
        inv.root.revision = current_entry[parent_index][4]
1694
1390
        # Turn some things into local variables
1734
1430
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1735
1431
                            % entry)
1736
1432
                # These checks cost us around 40ms on a 55k entry tree
1737
 
                if file_id in inv_byid:
1738
 
                    raise AssertionError('file_id %s already in'
1739
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
1740
 
                if name_unicode in parent_ie.children:
1741
 
                    raise AssertionError('name %r already in parent'
1742
 
                        % (name_unicode,))
 
1433
                assert file_id not in inv_byid
 
1434
                assert name_unicode not in parent_ie.children
1743
1435
                inv_byid[file_id] = inv_entry
1744
1436
                parent_ie.children[name_unicode] = inv_entry
1745
1437
        self._inventory = inv
1755
1447
            return None
1756
1448
        parent_index = self._get_parent_index()
1757
1449
        last_changed_revision = entry[1][parent_index][4]
1758
 
        try:
1759
 
            rev = self._repository.get_revision(last_changed_revision)
1760
 
        except errors.NoSuchRevision:
1761
 
            raise errors.FileTimestampUnavailable(self.id2path(file_id))
1762
 
        return rev.timestamp
 
1450
        return self._repository.get_revision(last_changed_revision).timestamp
1763
1451
 
1764
1452
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1765
1453
        entry = self._get_entry(file_id=file_id, path=path)
1769
1457
            return parent_details[1]
1770
1458
        return None
1771
1459
 
1772
 
    def get_file(self, file_id, path=None):
 
1460
    def get_file(self, file_id):
1773
1461
        return StringIO(self.get_file_text(file_id))
1774
1462
 
 
1463
    def get_file_lines(self, file_id):
 
1464
        ie = self.inventory[file_id]
 
1465
        return self._repository.weave_store.get_weave(file_id,
 
1466
                self._repository.get_transaction()).get_lines(ie.revision)
 
1467
 
1775
1468
    def get_file_size(self, file_id):
1776
 
        """See Tree.get_file_size"""
1777
1469
        return self.inventory[file_id].text_size
1778
1470
 
1779
 
    def get_file_text(self, file_id, path=None):
1780
 
        _, content = list(self.iter_files_bytes([(file_id, None)]))[0]
1781
 
        return ''.join(content)
 
1471
    def get_file_text(self, file_id):
 
1472
        return ''.join(self.get_file_lines(file_id))
1782
1473
 
1783
1474
    def get_reference_revision(self, file_id, path=None):
1784
1475
        return self.inventory[file_id].reference_revision
1785
1476
 
1786
 
    def iter_files_bytes(self, desired_files):
1787
 
        """See Tree.iter_files_bytes.
1788
 
 
1789
 
        This version is implemented on top of Repository.iter_files_bytes"""
1790
 
        parent_index = self._get_parent_index()
1791
 
        repo_desired_files = []
1792
 
        for file_id, identifier in desired_files:
1793
 
            entry = self._get_entry(file_id)
1794
 
            if entry == (None, None):
1795
 
                raise errors.NoSuchId(self, file_id)
1796
 
            repo_desired_files.append((file_id, entry[1][parent_index][4],
1797
 
                                       identifier))
1798
 
        return self._repository.iter_files_bytes(repo_desired_files)
1799
 
 
1800
1477
    def get_symlink_target(self, file_id):
1801
1478
        entry = self._get_entry(file_id=file_id)
1802
1479
        parent_index = self._get_parent_index()
1803
1480
        if entry[1][parent_index][0] != 'l':
1804
1481
            return None
1805
1482
        else:
1806
 
            target = entry[1][parent_index][1]
1807
 
            target = target.decode('utf8')
1808
 
            return target
 
1483
            # At present, none of the tree implementations supports non-ascii
 
1484
            # symlink targets. So we will just assume that the dirstate path is
 
1485
            # correct.
 
1486
            return entry[1][parent_index][1]
1809
1487
 
1810
1488
    def get_revision_id(self):
1811
1489
        """Return the revision id for this tree."""
1829
1507
        return bool(self.path2id(filename))
1830
1508
 
1831
1509
    def kind(self, file_id):
1832
 
        entry = self._get_entry(file_id=file_id)[1]
1833
 
        if entry is None:
1834
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1835
 
        parent_index = self._get_parent_index()
1836
 
        return dirstate.DirState._minikind_to_kind[entry[parent_index][0]]
1837
 
 
1838
 
    def stored_kind(self, file_id):
1839
 
        """See Tree.stored_kind"""
1840
 
        return self.kind(file_id)
1841
 
 
1842
 
    def path_content_summary(self, path):
1843
 
        """See Tree.path_content_summary."""
1844
 
        id = self.inventory.path2id(path)
1845
 
        if id is None:
1846
 
            return ('missing', None, None, None)
1847
 
        entry = self._inventory[id]
1848
 
        kind = entry.kind
1849
 
        if kind == 'file':
1850
 
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
1851
 
        elif kind == 'symlink':
1852
 
            return (kind, None, None, entry.symlink_target)
1853
 
        else:
1854
 
            return (kind, None, None, None)
 
1510
        return self.inventory[file_id].kind
1855
1511
 
1856
1512
    def is_executable(self, file_id, path=None):
1857
1513
        ie = self.inventory[file_id]
1859
1515
            return None
1860
1516
        return ie.executable
1861
1517
 
1862
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1518
    def list_files(self, include_root=False):
1863
1519
        # We use a standard implementation, because DirStateRevisionTree is
1864
1520
        # dealing with one of the parents of the current state
1865
1521
        inv = self._get_inventory()
1866
 
        if from_dir is None:
1867
 
            from_dir_id = None
1868
 
        else:
1869
 
            from_dir_id = inv.path2id(from_dir)
1870
 
            if from_dir_id is None:
1871
 
                # Directory not versioned
1872
 
                return
1873
 
        entries = inv.iter_entries(from_dir=from_dir_id, recursive=recursive)
1874
 
        if inv.root is not None and not include_root and from_dir is None:
 
1522
        entries = inv.iter_entries()
 
1523
        if self.inventory.root is not None and not include_root:
1875
1524
            entries.next()
1876
1525
        for path, entry in entries:
1877
1526
            yield path, 'V', entry.kind, entry.file_id, entry
1910
1559
                self._dirstate_locked = False
1911
1560
            self._repository.unlock()
1912
1561
 
1913
 
    @needs_read_lock
1914
 
    def supports_tree_reference(self):
1915
 
        return self._repo_supports_tree_reference
1916
 
 
1917
1562
    def walkdirs(self, prefix=""):
1918
1563
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1919
 
        # implementation based on an inventory.
 
1564
        # implementation based on an inventory.  
1920
1565
        # This should be cleaned up to use the much faster Dirstate code
1921
1566
        # So for now, we just build up the parent inventory, and extract
1922
1567
        # it the same way RevisionTree does.
1951
1596
 
1952
1597
class InterDirStateTree(InterTree):
1953
1598
    """Fast path optimiser for changes_from with dirstate trees.
1954
 
 
1955
 
    This is used only when both trees are in the dirstate working file, and
1956
 
    the source is any parent within the dirstate, and the destination is
 
1599
    
 
1600
    This is used only when both trees are in the dirstate working file, and 
 
1601
    the source is any parent within the dirstate, and the destination is 
1957
1602
    the current working tree of the same dirstate.
1958
1603
    """
1959
1604
    # this could be generalized to allow comparisons between any trees in the
1972
1617
        target.set_parent_ids([revid])
1973
1618
        return target.basis_tree(), target
1974
1619
 
1975
 
    @classmethod
1976
 
    def make_source_parent_tree_python_dirstate(klass, test_case, source, target):
1977
 
        result = klass.make_source_parent_tree(source, target)
1978
 
        result[1]._iter_changes = dirstate.ProcessEntryPython
1979
 
        return result
1980
 
 
1981
 
    @classmethod
1982
 
    def make_source_parent_tree_compiled_dirstate(klass, test_case, source,
1983
 
                                                  target):
1984
 
        from bzrlib.tests.test__dirstate_helpers import \
1985
 
            compiled_dirstate_helpers_feature
1986
 
        test_case.requireFeature(compiled_dirstate_helpers_feature)
1987
 
        from bzrlib._dirstate_helpers_pyx import ProcessEntryC
1988
 
        result = klass.make_source_parent_tree(source, target)
1989
 
        result[1]._iter_changes = ProcessEntryC
1990
 
        return result
1991
 
 
1992
1620
    _matching_from_tree_format = WorkingTreeFormat4()
1993
1621
    _matching_to_tree_format = WorkingTreeFormat4()
1994
 
 
1995
 
    @classmethod
1996
 
    def _test_mutable_trees_to_test_trees(klass, test_case, source, target):
1997
 
        # This method shouldn't be called, because we have python and C
1998
 
        # specific flavours.
1999
 
        raise NotImplementedError
2000
 
 
2001
 
    def iter_changes(self, include_unchanged=False,
 
1622
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1623
 
 
1624
    def _iter_changes(self, include_unchanged=False,
2002
1625
                      specific_files=None, pb=None, extra_trees=[],
2003
1626
                      require_versioned=True, want_unversioned=False):
2004
1627
        """Return the changes from source to target.
2005
1628
 
2006
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1629
        :return: An iterator that yields tuples. See InterTree._iter_changes
2007
1630
            for details.
2008
1631
        :param specific_files: An optional list of file paths to restrict the
2009
1632
            comparison to. When mapping filenames to ids, all matches in all
2020
1643
            output. An unversioned file is defined as one with (False, False)
2021
1644
            for the versioned pair.
2022
1645
        """
 
1646
        utf8_decode = cache_utf8._utf8_decode_with_None
 
1647
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1648
        # NB: show_status depends on being able to pass in non-versioned files
 
1649
        # and report them as unknown
2023
1650
        # TODO: handle extra trees in the dirstate.
2024
 
        if (extra_trees or specific_files == []):
 
1651
        # TODO: handle comparisons as an empty tree as a different special
 
1652
        # case? mbp 20070226
 
1653
        if extra_trees or (self.source._revision_id == NULL_REVISION):
2025
1654
            # we can't fast-path these cases (yet)
2026
 
            return super(InterDirStateTree, self).iter_changes(
 
1655
            for f in super(InterDirStateTree, self)._iter_changes(
2027
1656
                include_unchanged, specific_files, pb, extra_trees,
2028
 
                require_versioned, want_unversioned=want_unversioned)
 
1657
                require_versioned, want_unversioned=want_unversioned):
 
1658
                yield f
 
1659
            return
2029
1660
        parent_ids = self.target.get_parent_ids()
2030
 
        if not (self.source._revision_id in parent_ids
2031
 
                or self.source._revision_id == _mod_revision.NULL_REVISION):
2032
 
            raise AssertionError(
2033
 
                "revision {%s} is not stored in {%s}, but %s "
2034
 
                "can only be used for trees stored in the dirstate"
2035
 
                % (self.source._revision_id, self.target, self.iter_changes))
 
1661
        assert (self.source._revision_id in parent_ids), \
 
1662
                "revision {%s} is not stored in {%s}, but %s " \
 
1663
                "can only be used for trees stored in the dirstate" \
 
1664
                % (self.source._revision_id, self.target, self._iter_changes)
2036
1665
        target_index = 0
2037
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1666
        if self.source._revision_id == NULL_REVISION:
2038
1667
            source_index = None
2039
1668
            indices = (target_index,)
2040
1669
        else:
2041
 
            if not (self.source._revision_id in parent_ids):
2042
 
                raise AssertionError(
2043
 
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
2044
 
                    self.source._revision_id, parent_ids))
 
1670
            assert (self.source._revision_id in parent_ids), \
 
1671
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1672
                self.source._revision_id, parent_ids)
2045
1673
            source_index = 1 + parent_ids.index(self.source._revision_id)
2046
 
            indices = (source_index, target_index)
 
1674
            indices = (source_index,target_index)
2047
1675
        # -- make all specific_files utf8 --
2048
1676
        if specific_files:
2049
1677
            specific_files_utf8 = set()
2050
1678
            for path in specific_files:
2051
 
                # Note, if there are many specific files, using cache_utf8
2052
 
                # would be good here.
2053
1679
                specific_files_utf8.add(path.encode('utf8'))
2054
1680
            specific_files = specific_files_utf8
2055
1681
        else:
2056
1682
            specific_files = set([''])
2057
1683
        # -- specific_files is now a utf8 path set --
2058
 
 
2059
1684
        # -- get the state object and prepare it.
2060
1685
        state = self.target.current_dirstate()
2061
1686
        state._read_dirblocks_if_needed()
 
1687
        def _entries_for_path(path):
 
1688
            """Return a list with all the entries that match path for all ids.
 
1689
            """
 
1690
            dirname, basename = os.path.split(path)
 
1691
            key = (dirname, basename, '')
 
1692
            block_index, present = state._find_block_index_from_key(key)
 
1693
            if not present:
 
1694
                # the block which should contain path is absent.
 
1695
                return []
 
1696
            result = []
 
1697
            block = state._dirblocks[block_index][1]
 
1698
            entry_index, _ = state._find_entry_index(key, block)
 
1699
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1700
            while (entry_index < len(block) and
 
1701
                block[entry_index][0][0:2] == key[0:2]):
 
1702
                result.append(block[entry_index])
 
1703
                entry_index += 1
 
1704
            return result
2062
1705
        if require_versioned:
2063
1706
            # -- check all supplied paths are versioned in a search tree. --
2064
 
            not_versioned = []
 
1707
            all_versioned = True
2065
1708
            for path in specific_files:
2066
 
                path_entries = state._entries_for_path(path)
 
1709
                path_entries = _entries_for_path(path)
2067
1710
                if not path_entries:
2068
1711
                    # this specified path is not present at all: error
2069
 
                    not_versioned.append(path)
2070
 
                    continue
 
1712
                    all_versioned = False
 
1713
                    break
2071
1714
                found_versioned = False
2072
1715
                # for each id at this path
2073
1716
                for entry in path_entries:
2080
1723
                if not found_versioned:
2081
1724
                    # none of the indexes was not 'absent' at all ids for this
2082
1725
                    # path.
2083
 
                    not_versioned.append(path)
2084
 
            if len(not_versioned) > 0:
2085
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1726
                    all_versioned = False
 
1727
                    break
 
1728
            if not all_versioned:
 
1729
                raise errors.PathsNotVersionedError(specific_files)
2086
1730
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2087
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1731
        search_specific_files = set()
 
1732
        for path in specific_files:
 
1733
            other_specific_files = specific_files.difference(set([path]))
 
1734
            if not osutils.is_inside_any(other_specific_files, path):
 
1735
                # this is a top level path, we must check it.
 
1736
                search_specific_files.add(path)
 
1737
        # sketch: 
 
1738
        # compare source_index and target_index at or under each element of search_specific_files.
 
1739
        # follow the following comparison table. Note that we only want to do diff operations when
 
1740
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1741
        # for the target.
 
1742
        # cases:
 
1743
        # 
 
1744
        # Source | Target | disk | action
 
1745
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1746
        #        |        |      | diff check on source-target
 
1747
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1748
        #        |        |      | ???
 
1749
        #   r    |  a     |      | add source to search
 
1750
        #   r    |  a     |  a   | 
 
1751
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1752
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1753
        #   a    | fdlt   |      | add new id
 
1754
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1755
        #   a    |  a     |      | not present in either tree, skip
 
1756
        #   a    |  a     |  a   | not present in any tree, skip
 
1757
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1758
        #        |        |      | may not be selected by the users list of paths.
 
1759
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1760
        #        |        |      | may not be selected by the users list of paths.
 
1761
        #  fdlt  | fdlt   |      | content in both: diff them
 
1762
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1763
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1764
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1765
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1766
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1767
        #        |        |      | this id at the other path.
 
1768
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1769
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1770
        #        |        |      | this id at the other path.
 
1771
 
 
1772
        # for all search_indexs in each path at or under each element of
 
1773
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1774
        # relocated path as one to search if its not searched already. If the
 
1775
        # detail is not relocated, add the id.
 
1776
        searched_specific_files = set()
 
1777
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1778
        # Using a list so that we can access the values and change them in
 
1779
        # nested scope. Each one is [path, file_id, entry]
 
1780
        last_source_parent = [None, None, None]
 
1781
        last_target_parent = [None, None, None]
2088
1782
 
2089
1783
        use_filesystem_for_exec = (sys.platform != 'win32')
2090
 
        iter_changes = self.target._iter_changes(include_unchanged,
2091
 
            use_filesystem_for_exec, search_specific_files, state,
2092
 
            source_index, target_index, want_unversioned, self.target)
2093
 
        return iter_changes.iter_changes()
 
1784
 
 
1785
        def _process_entry(entry, path_info):
 
1786
            """Compare an entry and real disk to generate delta information.
 
1787
 
 
1788
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1789
                the path of entry. If None, then the path is considered absent.
 
1790
                (Perhaps we should pass in a concrete entry for this ?)
 
1791
                Basename is returned as a utf8 string because we expect this
 
1792
                tuple will be ignored, and don't want to take the time to
 
1793
                decode.
 
1794
            """
 
1795
            if source_index is None:
 
1796
                source_details = NULL_PARENT_DETAILS
 
1797
            else:
 
1798
                source_details = entry[1][source_index]
 
1799
            target_details = entry[1][target_index]
 
1800
            target_minikind = target_details[0]
 
1801
            if path_info is not None and target_minikind in 'fdlt':
 
1802
                assert target_index == 0
 
1803
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1804
                                                  stat_value=path_info[3])
 
1805
                # The entry may have been modified by update_entry
 
1806
                target_details = entry[1][target_index]
 
1807
                target_minikind = target_details[0]
 
1808
            else:
 
1809
                link_or_sha1 = None
 
1810
            source_minikind = source_details[0]
 
1811
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1812
                # claimed content in both: diff
 
1813
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1814
                #        |        |      | diff check on source-target
 
1815
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1816
                #        |        |      | ???
 
1817
                if source_minikind in 'r':
 
1818
                    # add the source to the search path to find any children it
 
1819
                    # has.  TODO ? : only add if it is a container ?
 
1820
                    if not osutils.is_inside_any(searched_specific_files,
 
1821
                                                 source_details[1]):
 
1822
                        search_specific_files.add(source_details[1])
 
1823
                    # generate the old path; this is needed for stating later
 
1824
                    # as well.
 
1825
                    old_path = source_details[1]
 
1826
                    old_dirname, old_basename = os.path.split(old_path)
 
1827
                    path = pathjoin(entry[0][0], entry[0][1])
 
1828
                    old_entry = state._get_entry(source_index,
 
1829
                                                 path_utf8=old_path)
 
1830
                    # update the source details variable to be the real
 
1831
                    # location.
 
1832
                    source_details = old_entry[1][source_index]
 
1833
                    source_minikind = source_details[0]
 
1834
                else:
 
1835
                    old_dirname = entry[0][0]
 
1836
                    old_basename = entry[0][1]
 
1837
                    old_path = path = pathjoin(old_dirname, old_basename)
 
1838
                if path_info is None:
 
1839
                    # the file is missing on disk, show as removed.
 
1840
                    content_change = True
 
1841
                    target_kind = None
 
1842
                    target_exec = False
 
1843
                else:
 
1844
                    # source and target are both versioned and disk file is present.
 
1845
                    target_kind = path_info[2]
 
1846
                    if target_kind == 'directory':
 
1847
                        if source_minikind != 'd':
 
1848
                            content_change = True
 
1849
                        else:
 
1850
                            # directories have no fingerprint
 
1851
                            content_change = False
 
1852
                        target_exec = False
 
1853
                    elif target_kind == 'file':
 
1854
                        if source_minikind != 'f':
 
1855
                            content_change = True
 
1856
                        else:
 
1857
                            # We could check the size, but we already have the
 
1858
                            # sha1 hash.
 
1859
                            content_change = (link_or_sha1 != source_details[1])
 
1860
                        # Target details is updated at update_entry time
 
1861
                        if use_filesystem_for_exec:
 
1862
                            # We don't need S_ISREG here, because we are sure
 
1863
                            # we are dealing with a file.
 
1864
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1865
                        else:
 
1866
                            target_exec = target_details[3]
 
1867
                    elif target_kind == 'symlink':
 
1868
                        if source_minikind != 'l':
 
1869
                            content_change = True
 
1870
                        else:
 
1871
                            content_change = (link_or_sha1 != source_details[1])
 
1872
                        target_exec = False
 
1873
                    elif target_kind == 'tree-reference':
 
1874
                        if source_minikind != 't':
 
1875
                            content_change = True
 
1876
                        else:
 
1877
                            content_change = False
 
1878
                        target_exec = False
 
1879
                    else:
 
1880
                        raise Exception, "unknown kind %s" % path_info[2]
 
1881
                # parent id is the entry for the path in the target tree
 
1882
                if old_dirname == last_source_parent[0]:
 
1883
                    source_parent_id = last_source_parent[1]
 
1884
                else:
 
1885
                    source_parent_entry = state._get_entry(source_index,
 
1886
                                                           path_utf8=old_dirname)
 
1887
                    source_parent_id = source_parent_entry[0][2]
 
1888
                    if source_parent_id == entry[0][2]:
 
1889
                        # This is the root, so the parent is None
 
1890
                        source_parent_id = None
 
1891
                    else:
 
1892
                        last_source_parent[0] = old_dirname
 
1893
                        last_source_parent[1] = source_parent_id
 
1894
                        last_source_parent[2] = source_parent_entry
 
1895
                new_dirname = entry[0][0]
 
1896
                if new_dirname == last_target_parent[0]:
 
1897
                    target_parent_id = last_target_parent[1]
 
1898
                else:
 
1899
                    # TODO: We don't always need to do the lookup, because the
 
1900
                    #       parent entry will be the same as the source entry.
 
1901
                    target_parent_entry = state._get_entry(target_index,
 
1902
                                                           path_utf8=new_dirname)
 
1903
                    target_parent_id = target_parent_entry[0][2]
 
1904
                    if target_parent_id == entry[0][2]:
 
1905
                        # This is the root, so the parent is None
 
1906
                        target_parent_id = None
 
1907
                    else:
 
1908
                        last_target_parent[0] = new_dirname
 
1909
                        last_target_parent[1] = target_parent_id
 
1910
                        last_target_parent[2] = target_parent_entry
 
1911
 
 
1912
                source_exec = source_details[3]
 
1913
                return ((entry[0][2], (old_path, path), content_change,
 
1914
                        (True, True),
 
1915
                        (source_parent_id, target_parent_id),
 
1916
                        (old_basename, entry[0][1]),
 
1917
                        (_minikind_to_kind[source_minikind], target_kind),
 
1918
                        (source_exec, target_exec)),)
 
1919
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
1920
                # looks like a new file
 
1921
                if path_info is not None:
 
1922
                    path = pathjoin(entry[0][0], entry[0][1])
 
1923
                    # parent id is the entry for the path in the target tree
 
1924
                    # TODO: these are the same for an entire directory: cache em.
 
1925
                    parent_id = state._get_entry(target_index,
 
1926
                                                 path_utf8=entry[0][0])[0][2]
 
1927
                    if parent_id == entry[0][2]:
 
1928
                        parent_id = None
 
1929
                    if use_filesystem_for_exec:
 
1930
                        # We need S_ISREG here, because we aren't sure if this
 
1931
                        # is a file or not.
 
1932
                        target_exec = bool(
 
1933
                            stat.S_ISREG(path_info[3].st_mode)
 
1934
                            and stat.S_IEXEC & path_info[3].st_mode)
 
1935
                    else:
 
1936
                        target_exec = target_details[3]
 
1937
                    return ((entry[0][2], (None, path), True,
 
1938
                            (False, True),
 
1939
                            (None, parent_id),
 
1940
                            (None, entry[0][1]),
 
1941
                            (None, path_info[2]),
 
1942
                            (None, target_exec)),)
 
1943
                else:
 
1944
                    # but its not on disk: we deliberately treat this as just
 
1945
                    # never-present. (Why ?! - RBC 20070224)
 
1946
                    pass
 
1947
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
1948
                # unversioned, possibly, or possibly not deleted: we dont care.
 
1949
                # if its still on disk, *and* theres no other entry at this
 
1950
                # path [we dont know this in this routine at the moment -
 
1951
                # perhaps we should change this - then it would be an unknown.
 
1952
                old_path = pathjoin(entry[0][0], entry[0][1])
 
1953
                # parent id is the entry for the path in the target tree
 
1954
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
1955
                if parent_id == entry[0][2]:
 
1956
                    parent_id = None
 
1957
                return ((entry[0][2], (old_path, None), True,
 
1958
                        (True, False),
 
1959
                        (parent_id, None),
 
1960
                        (entry[0][1], None),
 
1961
                        (_minikind_to_kind[source_minikind], None),
 
1962
                        (source_details[3], None)),)
 
1963
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
1964
                # a rename; could be a true rename, or a rename inherited from
 
1965
                # a renamed parent. TODO: handle this efficiently. Its not
 
1966
                # common case to rename dirs though, so a correct but slow
 
1967
                # implementation will do.
 
1968
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
1969
                    search_specific_files.add(target_details[1])
 
1970
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
1971
                # neither of the selected trees contain this file,
 
1972
                # so skip over it. This is not currently directly tested, but
 
1973
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
1974
                pass
 
1975
            else:
 
1976
                raise AssertionError("don't know how to compare "
 
1977
                    "source_minikind=%r, target_minikind=%r"
 
1978
                    % (source_minikind, target_minikind))
 
1979
                ## import pdb;pdb.set_trace()
 
1980
            return ()
 
1981
 
 
1982
        while search_specific_files:
 
1983
            # TODO: the pending list should be lexically sorted?  the
 
1984
            # interface doesn't require it.
 
1985
            current_root = search_specific_files.pop()
 
1986
            searched_specific_files.add(current_root)
 
1987
            # process the entries for this containing directory: the rest will be
 
1988
            # found by their parents recursively.
 
1989
            root_entries = _entries_for_path(current_root)
 
1990
            root_abspath = self.target.abspath(current_root)
 
1991
            try:
 
1992
                root_stat = os.lstat(root_abspath)
 
1993
            except OSError, e:
 
1994
                if e.errno == errno.ENOENT:
 
1995
                    # the path does not exist: let _process_entry know that.
 
1996
                    root_dir_info = None
 
1997
                else:
 
1998
                    # some other random error: hand it up.
 
1999
                    raise
 
2000
            else:
 
2001
                root_dir_info = ('', current_root,
 
2002
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2003
                    root_abspath)
 
2004
                if root_dir_info[2] == 'directory':
 
2005
                    if self.target._directory_is_tree_reference(
 
2006
                        current_root.decode('utf8')):
 
2007
                        root_dir_info = root_dir_info[:2] + \
 
2008
                            ('tree-reference',) + root_dir_info[3:]
 
2009
 
 
2010
            if not root_entries and not root_dir_info:
 
2011
                # this specified path is not present at all, skip it.
 
2012
                continue
 
2013
            path_handled = False
 
2014
            for entry in root_entries:
 
2015
                for result in _process_entry(entry, root_dir_info):
 
2016
                    # this check should probably be outside the loop: one
 
2017
                    # 'iterate two trees' api, and then _iter_changes filters
 
2018
                    # unchanged pairs. - RBC 20070226
 
2019
                    path_handled = True
 
2020
                    if (include_unchanged
 
2021
                        or result[2]                    # content change
 
2022
                        or result[3][0] != result[3][1] # versioned status
 
2023
                        or result[4][0] != result[4][1] # parent id
 
2024
                        or result[5][0] != result[5][1] # name
 
2025
                        or result[6][0] != result[6][1] # kind
 
2026
                        or result[7][0] != result[7][1] # executable
 
2027
                        ):
 
2028
                        result = (result[0],
 
2029
                            ((utf8_decode(result[1][0])[0]),
 
2030
                             utf8_decode(result[1][1])[0]),) + result[2:]
 
2031
                        yield result
 
2032
            if want_unversioned and not path_handled:
 
2033
                new_executable = bool(
 
2034
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2035
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2036
                yield (None, (None, current_root), True, (False, False),
 
2037
                    (None, None),
 
2038
                    (None, splitpath(current_root)[-1]),
 
2039
                    (None, root_dir_info[2]), (None, new_executable))
 
2040
            initial_key = (current_root, '', '')
 
2041
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2042
            if block_index == 0:
 
2043
                # we have processed the total root already, but because the
 
2044
                # initial key matched it we should skip it here.
 
2045
                block_index +=1
 
2046
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2047
                current_dir_info = None
 
2048
            else:
 
2049
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2050
                try:
 
2051
                    current_dir_info = dir_iterator.next()
 
2052
                except OSError, e:
 
2053
                    if e.errno in (errno.ENOENT, errno.ENOTDIR):
 
2054
                        # there may be directories in the inventory even though
 
2055
                        # this path is not a file on disk: so mark it as end of
 
2056
                        # iterator
 
2057
                        current_dir_info = None
 
2058
                    else:
 
2059
                        raise
 
2060
                else:
 
2061
                    if current_dir_info[0][0] == '':
 
2062
                        # remove .bzr from iteration
 
2063
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2064
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2065
                        del current_dir_info[1][bzr_index]
 
2066
            # walk until both the directory listing and the versioned metadata
 
2067
            # are exhausted. 
 
2068
            if (block_index < len(state._dirblocks) and
 
2069
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2070
                current_block = state._dirblocks[block_index]
 
2071
            else:
 
2072
                current_block = None
 
2073
            while (current_dir_info is not None or
 
2074
                   current_block is not None):
 
2075
                if (current_dir_info and current_block
 
2076
                    and current_dir_info[0][0] != current_block[0]):
 
2077
                    if current_dir_info[0][0] < current_block[0] :
 
2078
                        # filesystem data refers to paths not covered by the dirblock.
 
2079
                        # this has two possibilities:
 
2080
                        # A) it is versioned but empty, so there is no block for it
 
2081
                        # B) it is not versioned.
 
2082
                        # in either case it was processed by the containing directories walk:
 
2083
                        # if it is root/foo, when we walked root we emitted it,
 
2084
                        # or if we ere given root/foo to walk specifically, we
 
2085
                        # emitted it when checking the walk-root entries
 
2086
                        # advance the iterator and loop - we dont need to emit it.
 
2087
                        try:
 
2088
                            current_dir_info = dir_iterator.next()
 
2089
                        except StopIteration:
 
2090
                            current_dir_info = None
 
2091
                    else:
 
2092
                        # We have a dirblock entry for this location, but there
 
2093
                        # is no filesystem path for this. This is most likely
 
2094
                        # because a directory was removed from the disk.
 
2095
                        # We don't have to report the missing directory,
 
2096
                        # because that should have already been handled, but we
 
2097
                        # need to handle all of the files that are contained
 
2098
                        # within.
 
2099
                        for current_entry in current_block[1]:
 
2100
                            # entry referring to file not present on disk.
 
2101
                            # advance the entry only, after processing.
 
2102
                            for result in _process_entry(current_entry, None):
 
2103
                                # this check should probably be outside the loop: one
 
2104
                                # 'iterate two trees' api, and then _iter_changes filters
 
2105
                                # unchanged pairs. - RBC 20070226
 
2106
                                if (include_unchanged
 
2107
                                    or result[2]                    # content change
 
2108
                                    or result[3][0] != result[3][1] # versioned status
 
2109
                                    or result[4][0] != result[4][1] # parent id
 
2110
                                    or result[5][0] != result[5][1] # name
 
2111
                                    or result[6][0] != result[6][1] # kind
 
2112
                                    or result[7][0] != result[7][1] # executable
 
2113
                                    ):
 
2114
                                    result = (result[0],
 
2115
                                        ((utf8_decode(result[1][0])[0]),
 
2116
                                         utf8_decode(result[1][1])[0]),) + result[2:]
 
2117
                                    yield result
 
2118
                        block_index +=1
 
2119
                        if (block_index < len(state._dirblocks) and
 
2120
                            osutils.is_inside(current_root,
 
2121
                                              state._dirblocks[block_index][0])):
 
2122
                            current_block = state._dirblocks[block_index]
 
2123
                        else:
 
2124
                            current_block = None
 
2125
                    continue
 
2126
                entry_index = 0
 
2127
                if current_block and entry_index < len(current_block[1]):
 
2128
                    current_entry = current_block[1][entry_index]
 
2129
                else:
 
2130
                    current_entry = None
 
2131
                advance_entry = True
 
2132
                path_index = 0
 
2133
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2134
                    current_path_info = current_dir_info[1][path_index]
 
2135
                    if current_path_info[2] == 'directory':
 
2136
                        if self.target._directory_is_tree_reference(
 
2137
                            current_path_info[0].decode('utf8')):
 
2138
                            current_path_info = current_path_info[:2] + \
 
2139
                                ('tree-reference',) + current_path_info[3:]
 
2140
                else:
 
2141
                    current_path_info = None
 
2142
                advance_path = True
 
2143
                path_handled = False
 
2144
                while (current_entry is not None or
 
2145
                    current_path_info is not None):
 
2146
                    if current_entry is None:
 
2147
                        # the check for path_handled when the path is adnvaced
 
2148
                        # will yield this path if needed.
 
2149
                        pass
 
2150
                    elif current_path_info is None:
 
2151
                        # no path is fine: the per entry code will handle it.
 
2152
                        for result in _process_entry(current_entry, current_path_info):
 
2153
                            # this check should probably be outside the loop: one
 
2154
                            # 'iterate two trees' api, and then _iter_changes filters
 
2155
                            # unchanged pairs. - RBC 20070226
 
2156
                            if (include_unchanged
 
2157
                                or result[2]                    # content change
 
2158
                                or result[3][0] != result[3][1] # versioned status
 
2159
                                or result[4][0] != result[4][1] # parent id
 
2160
                                or result[5][0] != result[5][1] # name
 
2161
                                or result[6][0] != result[6][1] # kind
 
2162
                                or result[7][0] != result[7][1] # executable
 
2163
                                ):
 
2164
                                result = (result[0],
 
2165
                                    ((utf8_decode(result[1][0])[0]),
 
2166
                                     utf8_decode(result[1][1])[0]),) + result[2:]
 
2167
                                yield result
 
2168
                    elif current_entry[0][1] != current_path_info[1]:
 
2169
                        if current_path_info[1] < current_entry[0][1]:
 
2170
                            # extra file on disk: pass for now, but only
 
2171
                            # increment the path, not the entry
 
2172
                            advance_entry = False
 
2173
                        else:
 
2174
                            # entry referring to file not present on disk.
 
2175
                            # advance the entry only, after processing.
 
2176
                            for result in _process_entry(current_entry, None):
 
2177
                                # this check should probably be outside the loop: one
 
2178
                                # 'iterate two trees' api, and then _iter_changes filters
 
2179
                                # unchanged pairs. - RBC 20070226
 
2180
                                path_handled = True
 
2181
                                if (include_unchanged
 
2182
                                    or result[2]                    # content change
 
2183
                                    or result[3][0] != result[3][1] # versioned status
 
2184
                                    or result[4][0] != result[4][1] # parent id
 
2185
                                    or result[5][0] != result[5][1] # name
 
2186
                                    or result[6][0] != result[6][1] # kind
 
2187
                                    or result[7][0] != result[7][1] # executable
 
2188
                                    ):
 
2189
                                    result = (result[0],
 
2190
                                        ((utf8_decode(result[1][0])[0]),
 
2191
                                         utf8_decode(result[1][1])[0]),) + result[2:]
 
2192
                                    yield result
 
2193
                            advance_path = False
 
2194
                    else:
 
2195
                        for result in _process_entry(current_entry, current_path_info):
 
2196
                            # this check should probably be outside the loop: one
 
2197
                            # 'iterate two trees' api, and then _iter_changes filters
 
2198
                            # unchanged pairs. - RBC 20070226
 
2199
                            path_handled = True
 
2200
                            if (include_unchanged
 
2201
                                or result[2]                    # content change
 
2202
                                or result[3][0] != result[3][1] # versioned status
 
2203
                                or result[4][0] != result[4][1] # parent id
 
2204
                                or result[5][0] != result[5][1] # name
 
2205
                                or result[6][0] != result[6][1] # kind
 
2206
                                or result[7][0] != result[7][1] # executable
 
2207
                                ):
 
2208
                                result = (result[0],
 
2209
                                    ((utf8_decode(result[1][0])[0]),
 
2210
                                     utf8_decode(result[1][1])[0]),) + result[2:]
 
2211
                                yield result
 
2212
                    if advance_entry and current_entry is not None:
 
2213
                        entry_index += 1
 
2214
                        if entry_index < len(current_block[1]):
 
2215
                            current_entry = current_block[1][entry_index]
 
2216
                        else:
 
2217
                            current_entry = None
 
2218
                    else:
 
2219
                        advance_entry = True # reset the advance flaga
 
2220
                    if advance_path and current_path_info is not None:
 
2221
                        if not path_handled:
 
2222
                            # unversioned in all regards
 
2223
                            if want_unversioned:
 
2224
                                new_executable = bool(
 
2225
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2226
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2227
                                if want_unversioned:
 
2228
                                    yield (None, (None, current_path_info[0]),
 
2229
                                        True,
 
2230
                                        (False, False),
 
2231
                                        (None, None),
 
2232
                                        (None, current_path_info[1]),
 
2233
                                        (None, current_path_info[2]),
 
2234
                                        (None, new_executable))
 
2235
                            # dont descend into this unversioned path if it is
 
2236
                            # a dir
 
2237
                            if current_path_info[2] in ('directory'):
 
2238
                                del current_dir_info[1][path_index]
 
2239
                                path_index -= 1
 
2240
                        # dont descend the disk iterator into any tree 
 
2241
                        # paths.
 
2242
                        if current_path_info[2] == 'tree-reference':
 
2243
                            del current_dir_info[1][path_index]
 
2244
                            path_index -= 1
 
2245
                        path_index += 1
 
2246
                        if path_index < len(current_dir_info[1]):
 
2247
                            current_path_info = current_dir_info[1][path_index]
 
2248
                            if current_path_info[2] == 'directory':
 
2249
                                if self.target._directory_is_tree_reference(
 
2250
                                    current_path_info[0].decode('utf8')):
 
2251
                                    current_path_info = current_path_info[:2] + \
 
2252
                                        ('tree-reference',) + current_path_info[3:]
 
2253
                        else:
 
2254
                            current_path_info = None
 
2255
                        path_handled = False
 
2256
                    else:
 
2257
                        advance_path = True # reset the advance flagg.
 
2258
                if current_block is not None:
 
2259
                    block_index += 1
 
2260
                    if (block_index < len(state._dirblocks) and
 
2261
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2262
                        current_block = state._dirblocks[block_index]
 
2263
                    else:
 
2264
                        current_block = None
 
2265
                if current_dir_info is not None:
 
2266
                    try:
 
2267
                        current_dir_info = dir_iterator.next()
 
2268
                    except StopIteration:
 
2269
                        current_dir_info = None
 
2270
 
2094
2271
 
2095
2272
    @staticmethod
2096
2273
    def is_compatible(source, target):
2097
2274
        # the target must be a dirstate working tree
2098
 
        if not isinstance(target, DirStateWorkingTree):
 
2275
        if not isinstance(target, WorkingTree4):
2099
2276
            return False
2100
 
        # the source must be a revtree or dirstate rev tree.
 
2277
        # the source must be a revtreee or dirstate rev tree.
2101
2278
        if not isinstance(source,
2102
2279
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2103
2280
            return False
2104
2281
        # the source revid must be in the target dirstate
2105
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2282
        if not (source._revision_id == NULL_REVISION or
2106
2283
            source._revision_id in target.get_parent_ids()):
2107
 
            # TODO: what about ghosts? it may well need to
 
2284
            # TODO: what about ghosts? it may well need to 
2108
2285
            # check for them explicitly.
2109
2286
            return False
2110
2287
        return True
2120
2297
 
2121
2298
    def convert(self, tree):
2122
2299
        # lock the control files not the tree, so that we dont get tree
2123
 
        # on-unlock behaviours, and so that noone else diddles with the
 
2300
        # on-unlock behaviours, and so that noone else diddles with the 
2124
2301
        # tree during upgrade.
2125
2302
        tree._control_files.lock_write()
2126
2303
        try:
2127
 
            tree.read_working_inventory()
2128
2304
            self.create_dirstate_data(tree)
2129
2305
            self.update_format(tree)
2130
2306
            self.remove_xml_files(tree)
2152
2328
 
2153
2329
    def update_format(self, tree):
2154
2330
        """Change the format marker."""
2155
 
        tree._transport.put_bytes('format',
2156
 
            self.target_format.get_format_string(),
2157
 
            mode=tree.bzrdir._get_file_mode())
2158
 
 
2159
 
 
2160
 
class Converter4to5(object):
2161
 
    """Perform an in-place upgrade of format 4 to format 5 trees."""
2162
 
 
2163
 
    def __init__(self):
2164
 
        self.target_format = WorkingTreeFormat5()
2165
 
 
2166
 
    def convert(self, tree):
2167
 
        # lock the control files not the tree, so that we don't get tree
2168
 
        # on-unlock behaviours, and so that no-one else diddles with the
2169
 
        # tree during upgrade.
2170
 
        tree._control_files.lock_write()
2171
 
        try:
2172
 
            self.update_format(tree)
2173
 
        finally:
2174
 
            tree._control_files.unlock()
2175
 
 
2176
 
    def update_format(self, tree):
2177
 
        """Change the format marker."""
2178
 
        tree._transport.put_bytes('format',
2179
 
            self.target_format.get_format_string(),
2180
 
            mode=tree.bzrdir._get_file_mode())
2181
 
 
2182
 
 
2183
 
class Converter4or5to6(object):
2184
 
    """Perform an in-place upgrade of format 4 or 5 to format 6 trees."""
2185
 
 
2186
 
    def __init__(self):
2187
 
        self.target_format = WorkingTreeFormat6()
2188
 
 
2189
 
    def convert(self, tree):
2190
 
        # lock the control files not the tree, so that we don't get tree
2191
 
        # on-unlock behaviours, and so that no-one else diddles with the
2192
 
        # tree during upgrade.
2193
 
        tree._control_files.lock_write()
2194
 
        try:
2195
 
            self.init_custom_control_files(tree)
2196
 
            self.update_format(tree)
2197
 
        finally:
2198
 
            tree._control_files.unlock()
2199
 
 
2200
 
    def init_custom_control_files(self, tree):
2201
 
        """Initialize custom control files."""
2202
 
        tree._transport.put_bytes('views', '',
2203
 
            mode=tree.bzrdir._get_file_mode())
2204
 
 
2205
 
    def update_format(self, tree):
2206
 
        """Change the format marker."""
2207
 
        tree._transport.put_bytes('format',
2208
 
            self.target_format.get_format_string(),
2209
 
            mode=tree.bzrdir._get_file_mode())
 
2331
        tree._control_files.put_utf8('format',
 
2332
            self.target_format.get_format_string())