/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: Robert Collins
  • Date: 2007-03-29 06:03:18 UTC
  • mto: (2018.18.6 hpss-faster-copy)
  • mto: This revision was merged to the branch mainline in revision 2435.
  • Revision ID: robertc@robertcollins.net-20070329060318-uqqabpbnzsnaldx7
Make the InterRepository tests less abusive of the API for InterRepository. (Robert Collins, Andrew Bennetts)

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
# This is the Windows equivalent of ENOTDIR
 
100
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
101
# just an error code.
 
102
ERROR_DIRECTORY = 267
 
103
 
 
104
 
 
105
class WorkingTree4(WorkingTree3):
 
106
    """This is the Format 4 working tree.
 
107
 
 
108
    This differs from WorkingTree3 by:
 
109
     - Having a consolidated internal dirstate, stored in a
 
110
       randomly-accessible sorted file on disk.
 
111
     - Not having a regular inventory attribute.  One can be synthesized 
 
112
       on demand but this is expensive and should be avoided.
 
113
 
 
114
    This is new in bzr 0.15.
 
115
    """
 
116
 
72
117
    def __init__(self, basedir,
73
118
                 branch,
74
119
                 _control_files=None,
83
128
        """
84
129
        self._format = _format
85
130
        self.bzrdir = _bzrdir
 
131
        from bzrlib.trace import note, mutter
 
132
        assert isinstance(basedir, basestring), \
 
133
            "base directory %r is not a string" % basedir
86
134
        basedir = safe_unicode(basedir)
87
135
        mutter("opening working tree %r", basedir)
88
136
        self._branch = branch
 
137
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
138
            "branch %r is not a Branch" % self.branch
89
139
        self.basedir = realpath(basedir)
90
140
        # if branch is at our basedir and is a format 6 or less
91
141
        # assume all other formats have their own control files.
 
142
        assert isinstance(_control_files, LockableFiles), \
 
143
            "_control_files must be a LockableFiles, not %r" % _control_files
92
144
        self._control_files = _control_files
93
 
        self._transport = self._control_files._transport
94
145
        self._dirty = None
95
146
        #-------------
96
147
        # during a read or write lock these objects are set, and are
98
149
        self._dirstate = None
99
150
        self._inventory = None
100
151
        #-------------
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
152
 
108
153
    @needs_tree_write_lock
109
154
    def _add(self, files, ids, kinds):
111
156
        state = self.current_dirstate()
112
157
        for f, file_id, kind in zip(files, ids, kinds):
113
158
            f = f.strip('/')
 
159
            assert '//' not in f
 
160
            assert '..' not in f
114
161
            if self.path2id(f):
115
162
                # special case tree root handling.
116
163
                if f == '' and self.path2id(f) == ROOT_ID:
137
184
    @needs_tree_write_lock
138
185
    def add_reference(self, sub_tree):
139
186
        # use standard implementation, which calls back to self._add
140
 
        #
 
187
        # 
141
188
        # So we don't store the reference_revision in the working dirstate,
142
 
        # it's just recorded at the moment of commit.
 
189
        # it's just recorded at the moment of commit. 
143
190
        self._add_reference(sub_tree)
144
191
 
145
192
    def break_lock(self):
184
231
            WorkingTree3._comparison_data(self, entry, path)
185
232
        # it looks like a plain directory, but it's really a reference -- see
186
233
        # also kind()
187
 
        if (self._repo_supports_tree_reference and kind == 'directory'
188
 
            and entry is not None and entry.kind == 'tree-reference'):
 
234
        if (self._repo_supports_tree_reference and
 
235
            kind == 'directory' and
 
236
            self._directory_is_tree_reference(path)):
189
237
            kind = 'tree-reference'
190
238
        return kind, executable, stat_value
191
239
 
217
265
            return self._dirstate
218
266
        local_path = self.bzrdir.get_workingtree_transport(None
219
267
            ).local_abspath('dirstate')
220
 
        self._dirstate = dirstate.DirState.on_file(local_path,
221
 
            self._sha1_provider())
 
268
        self._dirstate = dirstate.DirState.on_file(local_path)
222
269
        return self._dirstate
223
270
 
224
 
    def _sha1_provider(self):
225
 
        """A function that returns a SHA1Provider suitable for this tree.
226
 
 
227
 
        :return: None if content filtering is not supported by this tree.
228
 
          Otherwise, a SHA1Provider is returned that sha's the canonical
229
 
          form of files, i.e. after read filters are applied.
230
 
        """
231
 
        if self.supports_content_filtering():
232
 
            return ContentFilterAwareSHA1Provider(self)
233
 
        else:
234
 
            return None
 
271
    def _directory_is_tree_reference(self, relpath):
 
272
        # as a special case, if a directory contains control files then 
 
273
        # it's a tree reference, except that the root of the tree is not
 
274
        return relpath and osutils.isdir(self.abspath(relpath) + u"/.bzr")
 
275
        # TODO: We could ask all the control formats whether they
 
276
        # recognize this directory, but at the moment there's no cheap api
 
277
        # to do that.  Since we probably can only nest bzr checkouts and
 
278
        # they always use this name it's ok for now.  -- mbp 20060306
 
279
        #
 
280
        # FIXME: There is an unhandled case here of a subdirectory
 
281
        # containing .bzr but not a branch; that will probably blow up
 
282
        # when you try to commit it.  It might happen if there is a
 
283
        # checkout in a subdirectory.  This can be avoided by not adding
 
284
        # it.  mbp 20070306
235
285
 
236
286
    def filter_unversioned_files(self, paths):
237
287
        """Filter out paths that are versioned.
270
320
 
271
321
    def _generate_inventory(self):
272
322
        """Create and set self.inventory from the dirstate object.
273
 
 
 
323
        
274
324
        This is relatively expensive: we have to walk the entire dirstate.
275
325
        Ideally we would not, and can deprecate this function.
276
326
        """
280
330
        state._read_dirblocks_if_needed()
281
331
        root_key, current_entry = self._get_entry(path='')
282
332
        current_id = root_key[2]
283
 
        if not (current_entry[0][0] == 'd'): # directory
284
 
            raise AssertionError(current_entry)
 
333
        assert current_entry[0][0] == 'd' # directory
285
334
        inv = Inventory(root_id=current_id)
286
335
        # Turn some things into local variables
287
336
        minikind_to_kind = dirstate.DirState._minikind_to_kind
320
369
                    # add this entry to the parent map.
321
370
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
322
371
                elif kind == 'tree-reference':
323
 
                    if not self._repo_supports_tree_reference:
324
 
                        raise errors.UnsupportedOperation(
325
 
                            self._generate_inventory,
326
 
                            self.branch.repository)
 
372
                    assert self._repo_supports_tree_reference, \
 
373
                        "repository of %r " \
 
374
                        "doesn't support tree references " \
 
375
                        "required by entry %r" \
 
376
                        % (self, name)
327
377
                    inv_entry.reference_revision = link_or_sha1 or None
328
378
                elif kind != 'symlink':
329
379
                    raise AssertionError("unknown kind %r" % kind)
330
380
                # 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,))
 
381
                assert file_id not in inv_byid, ('file_id %s already in'
 
382
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
383
                assert name_unicode not in parent_ie.children
337
384
                inv_byid[file_id] = inv_entry
338
385
                parent_ie.children[name_unicode] = inv_entry
339
386
        self._inventory = inv
344
391
        If either file_id or path is supplied, it is used as the key to lookup.
345
392
        If both are supplied, the fastest lookup is used, and an error is
346
393
        raised if they do not both point at the same row.
347
 
 
 
394
        
348
395
        :param file_id: An optional unicode file_id to be looked up.
349
396
        :param path: An optional unicode path to be looked up.
350
397
        :return: The dirstate row tuple for path/file_id, or (None, None)
359
406
    def get_file_sha1(self, file_id, path=None, stat_value=None):
360
407
        # check file id is valid unconditionally.
361
408
        entry = self._get_entry(file_id=file_id, path=path)
362
 
        if entry[0] is None:
363
 
            raise errors.NoSuchId(self, file_id)
 
409
        assert entry[0] is not None, 'what error should this raise'
364
410
        if path is None:
365
411
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
366
412
 
367
413
        file_abspath = self.abspath(path)
368
414
        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)
 
415
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
416
                                          stat_value=stat_value)
379
417
        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
 
418
            return link_or_sha1
390
419
        return None
391
420
 
392
421
    def _get_inventory(self):
393
422
        """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
423
        if self._inventory is not None:
398
424
            return self._inventory
399
425
        self._must_be_locked()
406
432
    @needs_read_lock
407
433
    def get_parent_ids(self):
408
434
        """See Tree.get_parent_ids.
409
 
 
 
435
        
410
436
        This implementation requests the ids list from the dirstate file.
411
437
        """
412
438
        return self.current_dirstate().get_parent_ids()
428
454
 
429
455
    def has_id(self, file_id):
430
456
        state = self.current_dirstate()
 
457
        file_id = osutils.safe_file_id(file_id)
431
458
        row, parents = self._get_entry(file_id=file_id)
432
459
        if row is None:
433
460
            return False
434
461
        return osutils.lexists(pathjoin(
435
462
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
436
463
 
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
464
    @needs_read_lock
443
465
    def id2path(self, file_id):
444
 
        "Convert a file-id to a path."
 
466
        file_id = osutils.safe_file_id(file_id)
445
467
        state = self.current_dirstate()
446
468
        entry = self._get_entry(file_id=file_id)
447
469
        if entry == (None, None):
449
471
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
450
472
        return path_utf8.decode('utf8')
451
473
 
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
474
    if not osutils.supports_executable():
 
475
        @needs_read_lock
459
476
        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
 
            """
 
477
            file_id = osutils.safe_file_id(file_id)
464
478
            entry = self._get_entry(file_id=file_id, path=path)
465
479
            if entry == (None, None):
466
480
                return False
467
481
            return entry[1][0][3]
468
 
 
469
 
        _is_executable_from_path_and_stat = \
470
 
            _is_executable_from_path_and_stat_from_basis
471
482
    else:
 
483
        @needs_read_lock
472
484
        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
485
            if not path:
 
486
                file_id = osutils.safe_file_id(file_id)
479
487
                path = self.id2path(file_id)
480
488
            mode = os.lstat(self.abspath(path)).st_mode
481
489
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
482
490
 
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
491
    @needs_read_lock
494
492
    def __iter__(self):
495
493
        """Iterate through file_ids for this tree.
508
506
        return iter(result)
509
507
 
510
508
    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
509
        for key, tree_details in self.current_dirstate()._iter_entries():
516
510
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
517
511
                # not relevant to the working tree
519
513
            if not key[1]:
520
514
                # the root is not a reference.
521
515
                continue
522
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
516
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
523
517
            try:
524
 
                if self._kind(relpath) == 'tree-reference':
525
 
                    yield relpath, key[2]
 
518
                if self._kind(path) == 'tree-reference':
 
519
                    yield path, key[2]
526
520
            except errors.NoSuchFile:
527
521
                # path is missing on disk.
528
522
                continue
529
523
 
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
 
 
 
524
    @needs_read_lock
536
525
    def kind(self, file_id):
537
526
        """Return the kind of a file.
538
527
 
539
528
        This is always the actual kind that's on disk, regardless of what it
540
529
        was added as.
541
 
 
542
 
        Note: The caller is expected to take a read-lock before calling this.
543
530
        """
544
531
        relpath = self.id2path(file_id)
545
 
        if relpath is None:
546
 
            raise AssertionError(
547
 
                "path for id {%s} is None!" % file_id)
 
532
        assert relpath != None, \
 
533
            "path for id {%s} is None!" % file_id
548
534
        return self._kind(relpath)
549
535
 
550
536
    def _kind(self, relpath):
551
537
        abspath = self.abspath(relpath)
552
538
        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'
 
539
        if (self._repo_supports_tree_reference and
 
540
            kind == 'directory' and
 
541
            self._directory_is_tree_reference(relpath)):
 
542
            kind = 'tree-reference'
558
543
        return kind
559
544
 
560
545
    @needs_read_lock
564
549
        if parent_ids:
565
550
            return parent_ids[0]
566
551
        else:
567
 
            return _mod_revision.NULL_REVISION
 
552
            return None
568
553
 
569
554
    def lock_read(self):
570
555
        """See Branch.lock_read, and WorkingTree.unlock."""
578
563
                # set our support for tree references from the repository in
579
564
                # use.
580
565
                self._repo_supports_tree_reference = getattr(
581
 
                    self.branch.repository._format, "supports_tree_reference",
 
566
                    self.branch.repository._format, "support_tree_reference",
582
567
                    False)
583
568
            except:
584
569
                self._control_files.unlock()
598
583
                # set our support for tree references from the repository in
599
584
                # use.
600
585
                self._repo_supports_tree_reference = getattr(
601
 
                    self.branch.repository._format, "supports_tree_reference",
 
586
                    self.branch.repository._format, "support_tree_reference",
602
587
                    False)
603
588
            except:
604
589
                self._control_files.unlock()
623
608
        result = []
624
609
        if not from_paths:
625
610
            return result
 
611
 
626
612
        state = self.current_dirstate()
627
 
        if isinstance(from_paths, basestring):
628
 
            raise ValueError()
 
613
 
 
614
        assert not isinstance(from_paths, basestring)
629
615
        to_dir_utf8 = to_dir.encode('utf8')
630
616
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
631
617
        id_index = state._get_id_index()
653
639
        if self._inventory is not None:
654
640
            update_inventory = True
655
641
            inv = self.inventory
 
642
            to_dir_ie = inv[to_dir_id]
656
643
            to_dir_id = to_entry[0][2]
657
 
            to_dir_ie = inv[to_dir_id]
658
644
        else:
659
645
            update_inventory = False
660
646
 
683
669
            new_entry = to_block[1][added_entry_index]
684
670
            rollbacks.append(lambda:state._make_absent(new_entry))
685
671
 
 
672
        # create rename entries and tuples
686
673
        for from_rel in from_paths:
687
674
            # from_rel is 'pathinroot/foo/bar'
688
675
            from_rel_utf8 = from_rel.encode('utf8')
691
678
            from_entry = self._get_entry(path=from_rel)
692
679
            if from_entry == (None, None):
693
680
                raise errors.BzrMoveFailedError(from_rel,to_dir,
694
 
                    errors.NotVersionedError(path=from_rel))
 
681
                    errors.NotVersionedError(path=str(from_rel)))
695
682
 
696
683
            from_id = from_entry[0][2]
697
684
            to_rel = pathjoin(to_dir, from_tail)
724
711
                if from_missing: # implicitly just update our path mapping
725
712
                    move_file = False
726
713
                elif not after:
727
 
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
714
                    raise errors.RenameFailedFilesExist(from_rel, to_rel,
 
715
                        extra="(Use --after to update the Bazaar id)")
728
716
 
729
717
            rollbacks = []
730
718
            def rollback_rename():
785
773
 
786
774
                if minikind == 'd':
787
775
                    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")
 
776
                        """all entries in this block need updating.
 
777
 
 
778
                        TODO: This is pretty ugly, and doesn't support
 
779
                        reverting, but it works.
 
780
                        """
 
781
                        assert from_dir != '', "renaming root not supported"
791
782
                        from_key = (from_dir, '')
792
783
                        from_block_idx, present = \
793
784
                            state._find_block_index_from_key(from_key)
803
794
                        to_block_index = state._ensure_block(
804
795
                            to_block_index, to_entry_index, to_dir_utf8)
805
796
                        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()
 
797
                        for entry in from_block[1]:
 
798
                            assert entry[0][0] == from_dir
811
799
                            cur_details = entry[1][0]
812
800
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
813
801
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
814
802
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
815
803
                            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
804
                            move_one(entry, from_path_utf8=from_path_utf8,
823
805
                                     minikind=minikind,
824
806
                                     executable=cur_details[3],
827
809
                                     size=cur_details[2],
828
810
                                     to_block=to_block,
829
811
                                     to_key=to_key,
830
 
                                     to_path_utf8=to_path_utf8)
 
812
                                     to_path_utf8=to_rel_utf8)
831
813
                            if minikind == 'd':
832
814
                                # We need to move all the children of this
833
815
                                # entry
872
854
        for tree in trees:
873
855
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
874
856
                parents):
875
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
876
 
                    trees, require_versioned)
 
857
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
877
858
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
878
859
        # -- make all paths utf8 --
879
860
        paths_utf8 = set()
939
920
            if not all_versioned:
940
921
                raise errors.PathsNotVersionedError(paths)
941
922
        # -- remove redundancy in supplied paths to prevent over-scanning --
942
 
        search_paths = osutils.minimum_path_selection(paths)
943
 
        # sketch:
 
923
        search_paths = set()
 
924
        for path in paths:
 
925
            other_paths = paths.difference(set([path]))
 
926
            if not osutils.is_inside_any(other_paths, path):
 
927
                # this is a top level path, we must check it.
 
928
                search_paths.add(path)
 
929
        # sketch: 
944
930
        # for all search_indexs in each path at or under each element of
945
931
        # search_paths, if the detail is relocated: add the id, and add the
946
932
        # relocated path as one to search if its not searched already. If the
1002
988
 
1003
989
    def read_working_inventory(self):
1004
990
        """Read the working inventory.
1005
 
 
 
991
        
1006
992
        This is a meaningless operation for dirstate, but we obey it anyhow.
1007
993
        """
1008
994
        return self.inventory
1013
999
 
1014
1000
        WorkingTree4 supplies revision_trees for any basis tree.
1015
1001
        """
 
1002
        revision_id = osutils.safe_revision_id(revision_id)
1016
1003
        dirstate = self.current_dirstate()
1017
1004
        parent_ids = dirstate.get_parent_ids()
1018
1005
        if revision_id not in parent_ids:
1025
1012
    @needs_tree_write_lock
1026
1013
    def set_last_revision(self, new_revision):
1027
1014
        """Change the last revision in the working tree."""
 
1015
        new_revision = osutils.safe_revision_id(new_revision)
1028
1016
        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.")
 
1017
        if new_revision in (NULL_REVISION, None):
 
1018
            assert len(parents) < 2, (
 
1019
                "setting the last parent to none with a pending merge is "
 
1020
                "unsupported.")
1034
1021
            self.set_parent_ids([])
1035
1022
        else:
1036
1023
            self.set_parent_ids([new_revision] + parents[1:],
1039
1026
    @needs_tree_write_lock
1040
1027
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1041
1028
        """Set the parent ids to revision_ids.
1042
 
 
 
1029
        
1043
1030
        See also set_parent_trees. This api will try to retrieve the tree data
1044
1031
        for each element of revision_ids from the trees repository. If you have
1045
1032
        tree data already available, it is more efficient to use
1049
1036
        :param revision_ids: The revision_ids to set as the parent ids of this
1050
1037
            working tree. Any of these may be ghosts.
1051
1038
        """
 
1039
        revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1052
1040
        trees = []
1053
1041
        for revision_id in revision_ids:
1054
1042
            try:
1060
1048
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
1061
1049
                revtree = None
1062
1050
            trees.append((revision_id, revtree))
 
1051
        self.current_dirstate()._validate()
1063
1052
        self.set_parent_trees(trees,
1064
1053
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1054
        self.current_dirstate()._validate()
1065
1055
 
1066
1056
    @needs_tree_write_lock
1067
1057
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
1072
1062
            parent tree - i.e. a ghost.
1073
1063
        """
1074
1064
        dirstate = self.current_dirstate()
 
1065
        dirstate._validate()
1075
1066
        if len(parents_list) > 0:
1076
1067
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
1077
1068
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1078
1069
        real_trees = []
1079
1070
        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
1071
        # convert absent trees to the null tree, which we convert back to
1087
1072
        # missing on access.
1088
1073
        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)
 
1074
            rev_id = osutils.safe_revision_id(rev_id)
1096
1075
            if tree is not None:
1097
1076
                real_trees.append((rev_id, tree))
1098
1077
            else:
1099
1078
                real_trees.append((rev_id,
1100
 
                    self.branch.repository.revision_tree(
1101
 
                        _mod_revision.NULL_REVISION)))
 
1079
                    self.branch.repository.revision_tree(None)))
1102
1080
                ghosts.append(rev_id)
1103
 
            accepted_revisions.add(rev_id)
 
1081
        dirstate._validate()
1104
1082
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1083
        dirstate._validate()
1105
1084
        self._make_dirty(reset_inventory=False)
 
1085
        dirstate._validate()
1106
1086
 
1107
1087
    def _set_root_id(self, file_id):
1108
1088
        """See WorkingTree.set_root_id."""
1111
1091
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
1112
1092
            self._make_dirty(reset_inventory=True)
1113
1093
 
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
1094
    @needs_read_lock
1125
1095
    def supports_tree_reference(self):
1126
1096
        return self._repo_supports_tree_reference
1127
1097
 
1128
1098
    def unlock(self):
1129
1099
        """Unlock in format 4 trees needs to write the entire dirstate."""
1130
 
        # do non-implementation specific cleanup
1131
 
        self._cleanup()
1132
 
 
1133
1100
        if self._control_files._lock_count == 1:
1134
1101
            # eventually we should do signature checking during read locks for
1135
1102
            # dirstate updates.
1166
1133
            return
1167
1134
        state = self.current_dirstate()
1168
1135
        state._read_dirblocks_if_needed()
1169
 
        ids_to_unversion = set(file_ids)
 
1136
        ids_to_unversion = set()
 
1137
        for file_id in file_ids:
 
1138
            ids_to_unversion.add(osutils.safe_file_id(file_id))
1170
1139
        paths_to_unversion = set()
1171
1140
        # sketch:
1172
1141
        # check if the root is to be unversioned, if so, assert for now.
1199
1168
                # just forget the whole block.
1200
1169
                entry_index = 0
1201
1170
                while entry_index < len(block[1]):
 
1171
                    # Mark this file id as having been removed
1202
1172
                    entry = block[1][entry_index]
1203
 
                    if entry[1][0][0] in 'ar':
1204
 
                        # don't remove absent or renamed entries
 
1173
                    ids_to_unversion.discard(entry[0][2])
 
1174
                    if (entry[1][0][0] == 'a'
 
1175
                        or not state._make_absent(entry)):
1205
1176
                        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
1177
                # go to the next block. (At the moment we dont delete empty
1213
1178
                # dirblocks)
1214
1179
                block_index += 1
1238
1203
                self._inventory.remove_recursive_id(file_id)
1239
1204
 
1240
1205
    @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
1206
    def _write_inventory(self, inv):
1265
1207
        """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:
 
1208
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1209
        self.current_dirstate().set_state_from_inventory(inv)
 
1210
        self._make_dirty(reset_inventory=False)
 
1211
        if self._inventory is not None:
1280
1212
            self._inventory = inv
1281
1213
        self.flush()
1282
1214
 
1283
1215
 
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):
 
1216
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1217
    """The first consolidated dirstate working tree format.
 
1218
 
 
1219
    This format:
 
1220
        - exists within a metadir controlling .bzr
 
1221
        - includes an explicit version marker for the workingtree control
 
1222
          files, separate from the BzrDir format
 
1223
        - modifies the hash cache format
 
1224
        - is new in bzr 0.15
 
1225
        - uses a LockDir to guard access to it.
 
1226
    """
 
1227
 
 
1228
    def get_format_string(self):
 
1229
        """See WorkingTreeFormat.get_format_string()."""
 
1230
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1231
 
 
1232
    def get_format_description(self):
 
1233
        """See WorkingTreeFormat.get_format_description()."""
 
1234
        return "Working tree format 4"
 
1235
 
 
1236
    def initialize(self, a_bzrdir, revision_id=None):
1377
1237
        """See WorkingTreeFormat.initialize().
1378
1238
 
1379
1239
        :param revision_id: allows creating a working tree at a different
1380
1240
        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
1241
 
1388
 
        These trees get an initial random root id, if their repository supports
1389
 
        rich root data, TREE_ROOT otherwise.
 
1242
        These trees get an initial random root id.
1390
1243
        """
 
1244
        revision_id = osutils.safe_revision_id(revision_id)
1391
1245
        if not isinstance(a_bzrdir.transport, LocalTransport):
1392
1246
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
1393
1247
        transport = a_bzrdir.get_workingtree_transport(self)
1394
1248
        control_files = self._open_control_files(a_bzrdir)
1395
1249
        control_files.create_lock()
1396
1250
        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()
 
1251
        control_files.put_utf8('format', self.get_format_string())
 
1252
        branch = a_bzrdir.open_branch()
1403
1253
        if revision_id is None:
1404
1254
            revision_id = branch.last_revision()
1405
1255
        local_path = transport.local_abspath('dirstate')
1406
1256
        # write out new dirstate (must exist when we create the tree)
1407
1257
        state = dirstate.DirState.initialize(local_path)
1408
1258
        state.unlock()
1409
 
        del state
1410
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1259
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1411
1260
                         branch,
1412
1261
                         _format=self,
1413
1262
                         _bzrdir=a_bzrdir,
1414
1263
                         _control_files=control_files)
1415
1264
        wt._new_tree()
1416
1265
        wt.lock_tree_write()
 
1266
        state._validate()
1417
1267
        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)
 
1268
            if revision_id in (None, NULL_REVISION):
 
1269
                wt._set_root_id(generate_ids.gen_root_id())
1424
1270
                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)]
 
1271
                wt.current_dirstate()._validate()
 
1272
            wt.set_last_revision(revision_id)
 
1273
            wt.flush()
 
1274
            basis = wt.basis_tree()
1440
1275
            basis.lock_read()
1441
 
            try:
1442
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1276
            # if the basis has a root id we have to use that; otherwise we use
 
1277
            # a new random one
 
1278
            basis_root_id = basis.get_root_id()
 
1279
            if basis_root_id is not None:
 
1280
                wt._set_root_id(basis_root_id)
1443
1281
                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()
 
1282
            transform.build_tree(basis, wt)
 
1283
            basis.unlock()
1465
1284
        finally:
1466
1285
            control_files.unlock()
1467
1286
            wt.unlock()
1468
1287
        return wt
1469
1288
 
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
1289
    def _open(self, a_bzrdir, control_files):
1480
1290
        """Open the tree itself.
1481
1291
 
1482
1292
        :param a_bzrdir: the dir for the tree.
1483
1293
        :param control_files: the control files for the tree.
1484
1294
        """
1485
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1295
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1486
1296
                           branch=a_bzrdir.open_branch(),
1487
1297
                           _format=self,
1488
1298
                           _bzrdir=a_bzrdir,
1489
1299
                           _control_files=control_files)
1490
1300
 
1491
1301
    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
1302
        # please test against something that will let us do tree references
1497
1303
        return bzrdir.format_registry.make_bzrdir(
1498
1304
            'dirstate-with-subtree')
1500
1306
    _matchingbzrdir = property(__get_matchingbzrdir)
1501
1307
 
1502
1308
 
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
1309
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
 
    """
 
1310
    """A revision tree pulling the inventory from a dirstate."""
1581
1311
 
1582
1312
    def __init__(self, dirstate, revision_id, repository):
1583
1313
        self._dirstate = dirstate
1584
 
        self._revision_id = revision_id
 
1314
        self._revision_id = osutils.safe_revision_id(revision_id)
1585
1315
        self._repository = repository
1586
1316
        self._inventory = None
1587
1317
        self._locked = 0
1588
1318
        self._dirstate_locked = False
1589
 
        self._repo_supports_tree_reference = getattr(
1590
 
            repository._format, "supports_tree_reference",
1591
 
            False)
1592
1319
 
1593
1320
    def __repr__(self):
1594
1321
        return "<%s of %s in %s>" % \
1595
1322
            (self.__class__.__name__, self._revision_id, self._dirstate)
1596
1323
 
1597
 
    def annotate_iter(self, file_id,
1598
 
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1324
    def annotate_iter(self, file_id):
1599
1325
        """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]
 
1326
        w = self._repository.weave_store.get_weave(file_id,
 
1327
                           self._repository.get_transaction())
 
1328
        return w.annotate_iter(self.inventory[file_id].revision)
1603
1329
 
1604
 
    def _get_ancestors(self, default_revision):
1605
 
        return set(self._repository.get_ancestry(self._revision_id,
1606
 
                                                 topo_sorted=False))
1607
1330
    def _comparison_data(self, entry, path):
1608
1331
        """See Tree._comparison_data."""
1609
1332
        if entry is None:
1626
1349
    def get_root_id(self):
1627
1350
        return self.path2id('')
1628
1351
 
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
1352
    def _get_parent_index(self):
1646
1353
        """Return the index in the dirstate referenced by this tree."""
1647
1354
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1652
1359
        If either file_id or path is supplied, it is used as the key to lookup.
1653
1360
        If both are supplied, the fastest lookup is used, and an error is
1654
1361
        raised if they do not both point at the same row.
1655
 
 
 
1362
        
1656
1363
        :param file_id: An optional unicode file_id to be looked up.
1657
1364
        :param path: An optional unicode path to be looked up.
1658
1365
        :return: The dirstate row tuple for path/file_id, or (None, None)
1659
1366
        """
1660
1367
        if file_id is None and path is None:
1661
1368
            raise errors.BzrError('must supply file_id or path')
 
1369
        file_id = osutils.safe_file_id(file_id)
1662
1370
        if path is not None:
1663
1371
            path = path.encode('utf8')
1664
1372
        parent_index = self._get_parent_index()
1672
1380
 
1673
1381
        This is relatively expensive: we have to walk the entire dirstate.
1674
1382
        """
1675
 
        if not self._locked:
1676
 
            raise AssertionError(
1677
 
                'cannot generate inventory of an unlocked '
1678
 
                'dirstate revision tree')
 
1383
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1384
            'dirstate revision tree'
1679
1385
        # separate call for profiling - makes it clear where the costs are.
1680
1386
        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()))
 
1387
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1388
            'parent %s has disappeared from %s' % (
 
1389
            self._revision_id, self._dirstate.get_parent_ids())
1685
1390
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1686
1391
        # This is identical now to the WorkingTree _generate_inventory except
1687
1392
        # for the tree index use.
1688
1393
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1689
1394
        current_id = root_key[2]
1690
 
        if current_entry[parent_index][0] != 'd':
1691
 
            raise AssertionError()
 
1395
        assert current_entry[parent_index][0] == 'd'
1692
1396
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1693
1397
        inv.root.revision = current_entry[parent_index][4]
1694
1398
        # Turn some things into local variables
1734
1438
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1735
1439
                            % entry)
1736
1440
                # 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,))
 
1441
                assert file_id not in inv_byid
 
1442
                assert name_unicode not in parent_ie.children
1743
1443
                inv_byid[file_id] = inv_entry
1744
1444
                parent_ie.children[name_unicode] = inv_entry
1745
1445
        self._inventory = inv
1755
1455
            return None
1756
1456
        parent_index = self._get_parent_index()
1757
1457
        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
 
1458
        return self._repository.get_revision(last_changed_revision).timestamp
1763
1459
 
1764
1460
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1765
1461
        entry = self._get_entry(file_id=file_id, path=path)
1769
1465
            return parent_details[1]
1770
1466
        return None
1771
1467
 
1772
 
    def get_file(self, file_id, path=None):
 
1468
    def get_file(self, file_id):
1773
1469
        return StringIO(self.get_file_text(file_id))
1774
1470
 
 
1471
    def get_file_lines(self, file_id):
 
1472
        ie = self.inventory[file_id]
 
1473
        return self._repository.weave_store.get_weave(file_id,
 
1474
                self._repository.get_transaction()).get_lines(ie.revision)
 
1475
 
1775
1476
    def get_file_size(self, file_id):
1776
 
        """See Tree.get_file_size"""
1777
1477
        return self.inventory[file_id].text_size
1778
1478
 
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)
 
1479
    def get_file_text(self, file_id):
 
1480
        return ''.join(self.get_file_lines(file_id))
1782
1481
 
1783
1482
    def get_reference_revision(self, file_id, path=None):
1784
1483
        return self.inventory[file_id].reference_revision
1785
1484
 
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
1485
    def get_symlink_target(self, file_id):
1801
1486
        entry = self._get_entry(file_id=file_id)
1802
1487
        parent_index = self._get_parent_index()
1803
1488
        if entry[1][parent_index][0] != 'l':
1804
1489
            return None
1805
1490
        else:
1806
 
            target = entry[1][parent_index][1]
1807
 
            target = target.decode('utf8')
1808
 
            return target
 
1491
            # At present, none of the tree implementations supports non-ascii
 
1492
            # symlink targets. So we will just assume that the dirstate path is
 
1493
            # correct.
 
1494
            return entry[1][parent_index][1]
1809
1495
 
1810
1496
    def get_revision_id(self):
1811
1497
        """Return the revision id for this tree."""
1829
1515
        return bool(self.path2id(filename))
1830
1516
 
1831
1517
    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)
 
1518
        return self.inventory[file_id].kind
1855
1519
 
1856
1520
    def is_executable(self, file_id, path=None):
1857
1521
        ie = self.inventory[file_id]
1859
1523
            return None
1860
1524
        return ie.executable
1861
1525
 
1862
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1526
    def list_files(self, include_root=False):
1863
1527
        # We use a standard implementation, because DirStateRevisionTree is
1864
1528
        # dealing with one of the parents of the current state
1865
1529
        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:
 
1530
        entries = inv.iter_entries()
 
1531
        if self.inventory.root is not None and not include_root:
1875
1532
            entries.next()
1876
1533
        for path, entry in entries:
1877
1534
            yield path, 'V', entry.kind, entry.file_id, entry
1910
1567
                self._dirstate_locked = False
1911
1568
            self._repository.unlock()
1912
1569
 
1913
 
    @needs_read_lock
1914
 
    def supports_tree_reference(self):
1915
 
        return self._repo_supports_tree_reference
1916
 
 
1917
1570
    def walkdirs(self, prefix=""):
1918
1571
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
1919
 
        # implementation based on an inventory.
 
1572
        # implementation based on an inventory.  
1920
1573
        # This should be cleaned up to use the much faster Dirstate code
1921
1574
        # So for now, we just build up the parent inventory, and extract
1922
1575
        # it the same way RevisionTree does.
1951
1604
 
1952
1605
class InterDirStateTree(InterTree):
1953
1606
    """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
 
1607
    
 
1608
    This is used only when both trees are in the dirstate working file, and 
 
1609
    the source is any parent within the dirstate, and the destination is 
1957
1610
    the current working tree of the same dirstate.
1958
1611
    """
1959
1612
    # this could be generalized to allow comparisons between any trees in the
1972
1625
        target.set_parent_ids([revid])
1973
1626
        return target.basis_tree(), target
1974
1627
 
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
1628
    _matching_from_tree_format = WorkingTreeFormat4()
1993
1629
    _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,
 
1630
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1631
 
 
1632
    def _iter_changes(self, include_unchanged=False,
2002
1633
                      specific_files=None, pb=None, extra_trees=[],
2003
1634
                      require_versioned=True, want_unversioned=False):
2004
1635
        """Return the changes from source to target.
2005
1636
 
2006
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1637
        :return: An iterator that yields tuples. See InterTree._iter_changes
2007
1638
            for details.
2008
1639
        :param specific_files: An optional list of file paths to restrict the
2009
1640
            comparison to. When mapping filenames to ids, all matches in all
2020
1651
            output. An unversioned file is defined as one with (False, False)
2021
1652
            for the versioned pair.
2022
1653
        """
 
1654
        utf8_decode_or_none = cache_utf8._utf8_decode_with_None
 
1655
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1656
        # NB: show_status depends on being able to pass in non-versioned files
 
1657
        # and report them as unknown
2023
1658
        # TODO: handle extra trees in the dirstate.
2024
 
        if (extra_trees or specific_files == []):
 
1659
        # TODO: handle comparisons as an empty tree as a different special
 
1660
        # case? mbp 20070226
 
1661
        if extra_trees or (self.source._revision_id == NULL_REVISION):
2025
1662
            # we can't fast-path these cases (yet)
2026
 
            return super(InterDirStateTree, self).iter_changes(
 
1663
            for f in super(InterDirStateTree, self)._iter_changes(
2027
1664
                include_unchanged, specific_files, pb, extra_trees,
2028
 
                require_versioned, want_unversioned=want_unversioned)
 
1665
                require_versioned, want_unversioned=want_unversioned):
 
1666
                yield f
 
1667
            return
2029
1668
        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))
 
1669
        assert (self.source._revision_id in parent_ids), \
 
1670
                "revision {%s} is not stored in {%s}, but %s " \
 
1671
                "can only be used for trees stored in the dirstate" \
 
1672
                % (self.source._revision_id, self.target, self._iter_changes)
2036
1673
        target_index = 0
2037
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1674
        if self.source._revision_id == NULL_REVISION:
2038
1675
            source_index = None
2039
1676
            indices = (target_index,)
2040
1677
        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))
 
1678
            assert (self.source._revision_id in parent_ids), \
 
1679
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1680
                self.source._revision_id, parent_ids)
2045
1681
            source_index = 1 + parent_ids.index(self.source._revision_id)
2046
 
            indices = (source_index, target_index)
 
1682
            indices = (source_index,target_index)
2047
1683
        # -- make all specific_files utf8 --
2048
1684
        if specific_files:
2049
1685
            specific_files_utf8 = set()
2050
1686
            for path in specific_files:
2051
 
                # Note, if there are many specific files, using cache_utf8
2052
 
                # would be good here.
2053
1687
                specific_files_utf8.add(path.encode('utf8'))
2054
1688
            specific_files = specific_files_utf8
2055
1689
        else:
2056
1690
            specific_files = set([''])
2057
1691
        # -- specific_files is now a utf8 path set --
2058
 
 
2059
1692
        # -- get the state object and prepare it.
2060
1693
        state = self.target.current_dirstate()
2061
1694
        state._read_dirblocks_if_needed()
 
1695
        def _entries_for_path(path):
 
1696
            """Return a list with all the entries that match path for all ids.
 
1697
            """
 
1698
            dirname, basename = os.path.split(path)
 
1699
            key = (dirname, basename, '')
 
1700
            block_index, present = state._find_block_index_from_key(key)
 
1701
            if not present:
 
1702
                # the block which should contain path is absent.
 
1703
                return []
 
1704
            result = []
 
1705
            block = state._dirblocks[block_index][1]
 
1706
            entry_index, _ = state._find_entry_index(key, block)
 
1707
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1708
            while (entry_index < len(block) and
 
1709
                block[entry_index][0][0:2] == key[0:2]):
 
1710
                result.append(block[entry_index])
 
1711
                entry_index += 1
 
1712
            return result
2062
1713
        if require_versioned:
2063
1714
            # -- check all supplied paths are versioned in a search tree. --
2064
 
            not_versioned = []
 
1715
            all_versioned = True
2065
1716
            for path in specific_files:
2066
 
                path_entries = state._entries_for_path(path)
 
1717
                path_entries = _entries_for_path(path)
2067
1718
                if not path_entries:
2068
1719
                    # this specified path is not present at all: error
2069
 
                    not_versioned.append(path)
2070
 
                    continue
 
1720
                    all_versioned = False
 
1721
                    break
2071
1722
                found_versioned = False
2072
1723
                # for each id at this path
2073
1724
                for entry in path_entries:
2080
1731
                if not found_versioned:
2081
1732
                    # none of the indexes was not 'absent' at all ids for this
2082
1733
                    # path.
2083
 
                    not_versioned.append(path)
2084
 
            if len(not_versioned) > 0:
2085
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1734
                    all_versioned = False
 
1735
                    break
 
1736
            if not all_versioned:
 
1737
                raise errors.PathsNotVersionedError(specific_files)
2086
1738
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2087
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1739
        search_specific_files = set()
 
1740
        for path in specific_files:
 
1741
            other_specific_files = specific_files.difference(set([path]))
 
1742
            if not osutils.is_inside_any(other_specific_files, path):
 
1743
                # this is a top level path, we must check it.
 
1744
                search_specific_files.add(path)
 
1745
        # sketch: 
 
1746
        # compare source_index and target_index at or under each element of search_specific_files.
 
1747
        # follow the following comparison table. Note that we only want to do diff operations when
 
1748
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1749
        # for the target.
 
1750
        # cases:
 
1751
        # 
 
1752
        # Source | Target | disk | action
 
1753
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1754
        #        |        |      | diff check on source-target
 
1755
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1756
        #        |        |      | ???
 
1757
        #   r    |  a     |      | add source to search
 
1758
        #   r    |  a     |  a   | 
 
1759
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1760
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1761
        #   a    | fdlt   |      | add new id
 
1762
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1763
        #   a    |  a     |      | not present in either tree, skip
 
1764
        #   a    |  a     |  a   | not present in any tree, skip
 
1765
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1766
        #        |        |      | may not be selected by the users list of paths.
 
1767
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1768
        #        |        |      | may not be selected by the users list of paths.
 
1769
        #  fdlt  | fdlt   |      | content in both: diff them
 
1770
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1771
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1772
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1773
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1774
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1775
        #        |        |      | this id at the other path.
 
1776
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1777
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1778
        #        |        |      | this id at the other path.
 
1779
 
 
1780
        # for all search_indexs in each path at or under each element of
 
1781
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1782
        # relocated path as one to search if its not searched already. If the
 
1783
        # detail is not relocated, add the id.
 
1784
        searched_specific_files = set()
 
1785
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1786
        # Using a list so that we can access the values and change them in
 
1787
        # nested scope. Each one is [path, file_id, entry]
 
1788
        last_source_parent = [None, None, None]
 
1789
        last_target_parent = [None, None, None]
2088
1790
 
2089
1791
        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()
 
1792
 
 
1793
        def _process_entry(entry, path_info):
 
1794
            """Compare an entry and real disk to generate delta information.
 
1795
 
 
1796
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1797
                the path of entry. If None, then the path is considered absent.
 
1798
                (Perhaps we should pass in a concrete entry for this ?)
 
1799
                Basename is returned as a utf8 string because we expect this
 
1800
                tuple will be ignored, and don't want to take the time to
 
1801
                decode.
 
1802
            """
 
1803
            if source_index is None:
 
1804
                source_details = NULL_PARENT_DETAILS
 
1805
            else:
 
1806
                source_details = entry[1][source_index]
 
1807
            target_details = entry[1][target_index]
 
1808
            target_minikind = target_details[0]
 
1809
            if path_info is not None and target_minikind in 'fdlt':
 
1810
                assert target_index == 0
 
1811
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1812
                                                  stat_value=path_info[3])
 
1813
                # The entry may have been modified by update_entry
 
1814
                target_details = entry[1][target_index]
 
1815
                target_minikind = target_details[0]
 
1816
            else:
 
1817
                link_or_sha1 = None
 
1818
            source_minikind = source_details[0]
 
1819
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1820
                # claimed content in both: diff
 
1821
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1822
                #        |        |      | diff check on source-target
 
1823
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1824
                #        |        |      | ???
 
1825
                if source_minikind in 'r':
 
1826
                    # add the source to the search path to find any children it
 
1827
                    # has.  TODO ? : only add if it is a container ?
 
1828
                    if not osutils.is_inside_any(searched_specific_files,
 
1829
                                                 source_details[1]):
 
1830
                        search_specific_files.add(source_details[1])
 
1831
                    # generate the old path; this is needed for stating later
 
1832
                    # as well.
 
1833
                    old_path = source_details[1]
 
1834
                    old_dirname, old_basename = os.path.split(old_path)
 
1835
                    path = pathjoin(entry[0][0], entry[0][1])
 
1836
                    old_entry = state._get_entry(source_index,
 
1837
                                                 path_utf8=old_path)
 
1838
                    # update the source details variable to be the real
 
1839
                    # location.
 
1840
                    source_details = old_entry[1][source_index]
 
1841
                    source_minikind = source_details[0]
 
1842
                else:
 
1843
                    old_dirname = entry[0][0]
 
1844
                    old_basename = entry[0][1]
 
1845
                    old_path = path = pathjoin(old_dirname, old_basename)
 
1846
                if path_info is None:
 
1847
                    # the file is missing on disk, show as removed.
 
1848
                    content_change = True
 
1849
                    target_kind = None
 
1850
                    target_exec = False
 
1851
                else:
 
1852
                    # source and target are both versioned and disk file is present.
 
1853
                    target_kind = path_info[2]
 
1854
                    if target_kind == 'directory':
 
1855
                        if source_minikind != 'd':
 
1856
                            content_change = True
 
1857
                        else:
 
1858
                            # directories have no fingerprint
 
1859
                            content_change = False
 
1860
                        target_exec = False
 
1861
                    elif target_kind == 'file':
 
1862
                        if source_minikind != 'f':
 
1863
                            content_change = True
 
1864
                        else:
 
1865
                            # We could check the size, but we already have the
 
1866
                            # sha1 hash.
 
1867
                            content_change = (link_or_sha1 != source_details[1])
 
1868
                        # Target details is updated at update_entry time
 
1869
                        if use_filesystem_for_exec:
 
1870
                            # We don't need S_ISREG here, because we are sure
 
1871
                            # we are dealing with a file.
 
1872
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1873
                        else:
 
1874
                            target_exec = target_details[3]
 
1875
                    elif target_kind == 'symlink':
 
1876
                        if source_minikind != 'l':
 
1877
                            content_change = True
 
1878
                        else:
 
1879
                            content_change = (link_or_sha1 != source_details[1])
 
1880
                        target_exec = False
 
1881
                    elif target_kind == 'tree-reference':
 
1882
                        if source_minikind != 't':
 
1883
                            content_change = True
 
1884
                        else:
 
1885
                            content_change = False
 
1886
                        target_exec = False
 
1887
                    else:
 
1888
                        raise Exception, "unknown kind %s" % path_info[2]
 
1889
                # parent id is the entry for the path in the target tree
 
1890
                if old_dirname == last_source_parent[0]:
 
1891
                    source_parent_id = last_source_parent[1]
 
1892
                else:
 
1893
                    source_parent_entry = state._get_entry(source_index,
 
1894
                                                           path_utf8=old_dirname)
 
1895
                    source_parent_id = source_parent_entry[0][2]
 
1896
                    if source_parent_id == entry[0][2]:
 
1897
                        # This is the root, so the parent is None
 
1898
                        source_parent_id = None
 
1899
                    else:
 
1900
                        last_source_parent[0] = old_dirname
 
1901
                        last_source_parent[1] = source_parent_id
 
1902
                        last_source_parent[2] = source_parent_entry
 
1903
                new_dirname = entry[0][0]
 
1904
                if new_dirname == last_target_parent[0]:
 
1905
                    target_parent_id = last_target_parent[1]
 
1906
                else:
 
1907
                    # TODO: We don't always need to do the lookup, because the
 
1908
                    #       parent entry will be the same as the source entry.
 
1909
                    target_parent_entry = state._get_entry(target_index,
 
1910
                                                           path_utf8=new_dirname)
 
1911
                    target_parent_id = target_parent_entry[0][2]
 
1912
                    if target_parent_id == entry[0][2]:
 
1913
                        # This is the root, so the parent is None
 
1914
                        target_parent_id = None
 
1915
                    else:
 
1916
                        last_target_parent[0] = new_dirname
 
1917
                        last_target_parent[1] = target_parent_id
 
1918
                        last_target_parent[2] = target_parent_entry
 
1919
 
 
1920
                source_exec = source_details[3]
 
1921
                return ((entry[0][2], (old_path, path), content_change,
 
1922
                        (True, True),
 
1923
                        (source_parent_id, target_parent_id),
 
1924
                        (old_basename, entry[0][1]),
 
1925
                        (_minikind_to_kind[source_minikind], target_kind),
 
1926
                        (source_exec, target_exec)),)
 
1927
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
1928
                # looks like a new file
 
1929
                if path_info is not None:
 
1930
                    path = pathjoin(entry[0][0], entry[0][1])
 
1931
                    # parent id is the entry for the path in the target tree
 
1932
                    # TODO: these are the same for an entire directory: cache em.
 
1933
                    parent_id = state._get_entry(target_index,
 
1934
                                                 path_utf8=entry[0][0])[0][2]
 
1935
                    if parent_id == entry[0][2]:
 
1936
                        parent_id = None
 
1937
                    if use_filesystem_for_exec:
 
1938
                        # We need S_ISREG here, because we aren't sure if this
 
1939
                        # is a file or not.
 
1940
                        target_exec = bool(
 
1941
                            stat.S_ISREG(path_info[3].st_mode)
 
1942
                            and stat.S_IEXEC & path_info[3].st_mode)
 
1943
                    else:
 
1944
                        target_exec = target_details[3]
 
1945
                    return ((entry[0][2], (None, path), True,
 
1946
                            (False, True),
 
1947
                            (None, parent_id),
 
1948
                            (None, entry[0][1]),
 
1949
                            (None, path_info[2]),
 
1950
                            (None, target_exec)),)
 
1951
                else:
 
1952
                    # but its not on disk: we deliberately treat this as just
 
1953
                    # never-present. (Why ?! - RBC 20070224)
 
1954
                    pass
 
1955
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
1956
                # unversioned, possibly, or possibly not deleted: we dont care.
 
1957
                # if its still on disk, *and* theres no other entry at this
 
1958
                # path [we dont know this in this routine at the moment -
 
1959
                # perhaps we should change this - then it would be an unknown.
 
1960
                old_path = pathjoin(entry[0][0], entry[0][1])
 
1961
                # parent id is the entry for the path in the target tree
 
1962
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
1963
                if parent_id == entry[0][2]:
 
1964
                    parent_id = None
 
1965
                return ((entry[0][2], (old_path, None), True,
 
1966
                        (True, False),
 
1967
                        (parent_id, None),
 
1968
                        (entry[0][1], None),
 
1969
                        (_minikind_to_kind[source_minikind], None),
 
1970
                        (source_details[3], None)),)
 
1971
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
1972
                # a rename; could be a true rename, or a rename inherited from
 
1973
                # a renamed parent. TODO: handle this efficiently. Its not
 
1974
                # common case to rename dirs though, so a correct but slow
 
1975
                # implementation will do.
 
1976
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
1977
                    search_specific_files.add(target_details[1])
 
1978
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
1979
                # neither of the selected trees contain this file,
 
1980
                # so skip over it. This is not currently directly tested, but
 
1981
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
1982
                pass
 
1983
            else:
 
1984
                raise AssertionError("don't know how to compare "
 
1985
                    "source_minikind=%r, target_minikind=%r"
 
1986
                    % (source_minikind, target_minikind))
 
1987
                ## import pdb;pdb.set_trace()
 
1988
            return ()
 
1989
 
 
1990
        while search_specific_files:
 
1991
            # TODO: the pending list should be lexically sorted?  the
 
1992
            # interface doesn't require it.
 
1993
            current_root = search_specific_files.pop()
 
1994
            current_root_unicode = current_root.decode('utf8')
 
1995
            searched_specific_files.add(current_root)
 
1996
            # process the entries for this containing directory: the rest will be
 
1997
            # found by their parents recursively.
 
1998
            root_entries = _entries_for_path(current_root)
 
1999
            root_abspath = self.target.abspath(current_root_unicode)
 
2000
            try:
 
2001
                root_stat = os.lstat(root_abspath)
 
2002
            except OSError, e:
 
2003
                if e.errno == errno.ENOENT:
 
2004
                    # the path does not exist: let _process_entry know that.
 
2005
                    root_dir_info = None
 
2006
                else:
 
2007
                    # some other random error: hand it up.
 
2008
                    raise
 
2009
            else:
 
2010
                root_dir_info = ('', current_root,
 
2011
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2012
                    root_abspath)
 
2013
                if root_dir_info[2] == 'directory':
 
2014
                    if self.target._directory_is_tree_reference(
 
2015
                        current_root.decode('utf8')):
 
2016
                        root_dir_info = root_dir_info[:2] + \
 
2017
                            ('tree-reference',) + root_dir_info[3:]
 
2018
 
 
2019
            if not root_entries and not root_dir_info:
 
2020
                # this specified path is not present at all, skip it.
 
2021
                continue
 
2022
            path_handled = False
 
2023
            for entry in root_entries:
 
2024
                for result in _process_entry(entry, root_dir_info):
 
2025
                    # this check should probably be outside the loop: one
 
2026
                    # 'iterate two trees' api, and then _iter_changes filters
 
2027
                    # unchanged pairs. - RBC 20070226
 
2028
                    path_handled = True
 
2029
                    if (include_unchanged
 
2030
                        or result[2]                    # content change
 
2031
                        or result[3][0] != result[3][1] # versioned status
 
2032
                        or result[4][0] != result[4][1] # parent id
 
2033
                        or result[5][0] != result[5][1] # name
 
2034
                        or result[6][0] != result[6][1] # kind
 
2035
                        or result[7][0] != result[7][1] # executable
 
2036
                        ):
 
2037
                        yield (result[0],
 
2038
                               (utf8_decode_or_none(result[1][0]),
 
2039
                                utf8_decode_or_none(result[1][1])),
 
2040
                               result[2],
 
2041
                               result[3],
 
2042
                               result[4],
 
2043
                               (utf8_decode_or_none(result[5][0]),
 
2044
                                utf8_decode_or_none(result[5][1])),
 
2045
                               result[6],
 
2046
                               result[7],
 
2047
                              )
 
2048
            if want_unversioned and not path_handled and root_dir_info:
 
2049
                new_executable = bool(
 
2050
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2051
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2052
                yield (None,
 
2053
                       (None, current_root_unicode),
 
2054
                       True,
 
2055
                       (False, False),
 
2056
                       (None, None),
 
2057
                       (None, splitpath(current_root_unicode)[-1]),
 
2058
                       (None, root_dir_info[2]),
 
2059
                       (None, new_executable)
 
2060
                      )
 
2061
            initial_key = (current_root, '', '')
 
2062
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2063
            if block_index == 0:
 
2064
                # we have processed the total root already, but because the
 
2065
                # initial key matched it we should skip it here.
 
2066
                block_index +=1
 
2067
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2068
                current_dir_info = None
 
2069
            else:
 
2070
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2071
                try:
 
2072
                    current_dir_info = dir_iterator.next()
 
2073
                except OSError, e:
 
2074
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2075
                    # python 2.5 has e.errno == EINVAL,
 
2076
                    #            and e.winerror == ERROR_DIRECTORY
 
2077
                    e_winerror = getattr(e, 'winerror', None)
 
2078
                    # there may be directories in the inventory even though
 
2079
                    # this path is not a file on disk: so mark it as end of
 
2080
                    # iterator
 
2081
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2082
                        current_dir_info = None
 
2083
                    elif (sys.platform == 'win32'
 
2084
                          and ERROR_DIRECTORY in (e.errno, e_winerror)):
 
2085
                        current_dir_info = None
 
2086
                    else:
 
2087
                        raise
 
2088
                else:
 
2089
                    if current_dir_info[0][0] == '':
 
2090
                        # remove .bzr from iteration
 
2091
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2092
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2093
                        del current_dir_info[1][bzr_index]
 
2094
            # walk until both the directory listing and the versioned metadata
 
2095
            # are exhausted. 
 
2096
            if (block_index < len(state._dirblocks) and
 
2097
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2098
                current_block = state._dirblocks[block_index]
 
2099
            else:
 
2100
                current_block = None
 
2101
            while (current_dir_info is not None or
 
2102
                   current_block is not None):
 
2103
                if (current_dir_info and current_block
 
2104
                    and current_dir_info[0][0] != current_block[0]):
 
2105
                    if current_dir_info[0][0] < current_block[0] :
 
2106
                        # filesystem data refers to paths not covered by the dirblock.
 
2107
                        # this has two possibilities:
 
2108
                        # A) it is versioned but empty, so there is no block for it
 
2109
                        # B) it is not versioned.
 
2110
                        # in either case it was processed by the containing directories walk:
 
2111
                        # if it is root/foo, when we walked root we emitted it,
 
2112
                        # or if we ere given root/foo to walk specifically, we
 
2113
                        # emitted it when checking the walk-root entries
 
2114
                        # advance the iterator and loop - we dont need to emit it.
 
2115
                        try:
 
2116
                            current_dir_info = dir_iterator.next()
 
2117
                        except StopIteration:
 
2118
                            current_dir_info = None
 
2119
                    else:
 
2120
                        # We have a dirblock entry for this location, but there
 
2121
                        # is no filesystem path for this. This is most likely
 
2122
                        # because a directory was removed from the disk.
 
2123
                        # We don't have to report the missing directory,
 
2124
                        # because that should have already been handled, but we
 
2125
                        # need to handle all of the files that are contained
 
2126
                        # within.
 
2127
                        for current_entry in current_block[1]:
 
2128
                            # entry referring to file not present on disk.
 
2129
                            # advance the entry only, after processing.
 
2130
                            for result in _process_entry(current_entry, None):
 
2131
                                # this check should probably be outside the loop: one
 
2132
                                # 'iterate two trees' api, and then _iter_changes filters
 
2133
                                # unchanged pairs. - RBC 20070226
 
2134
                                if (include_unchanged
 
2135
                                    or result[2]                    # content change
 
2136
                                    or result[3][0] != result[3][1] # versioned status
 
2137
                                    or result[4][0] != result[4][1] # parent id
 
2138
                                    or result[5][0] != result[5][1] # name
 
2139
                                    or result[6][0] != result[6][1] # kind
 
2140
                                    or result[7][0] != result[7][1] # executable
 
2141
                                    ):
 
2142
                                    yield (result[0],
 
2143
                                           (utf8_decode_or_none(result[1][0]),
 
2144
                                            utf8_decode_or_none(result[1][1])),
 
2145
                                           result[2],
 
2146
                                           result[3],
 
2147
                                           result[4],
 
2148
                                           (utf8_decode_or_none(result[5][0]),
 
2149
                                            utf8_decode_or_none(result[5][1])),
 
2150
                                           result[6],
 
2151
                                           result[7],
 
2152
                                          )
 
2153
                        block_index +=1
 
2154
                        if (block_index < len(state._dirblocks) and
 
2155
                            osutils.is_inside(current_root,
 
2156
                                              state._dirblocks[block_index][0])):
 
2157
                            current_block = state._dirblocks[block_index]
 
2158
                        else:
 
2159
                            current_block = None
 
2160
                    continue
 
2161
                entry_index = 0
 
2162
                if current_block and entry_index < len(current_block[1]):
 
2163
                    current_entry = current_block[1][entry_index]
 
2164
                else:
 
2165
                    current_entry = None
 
2166
                advance_entry = True
 
2167
                path_index = 0
 
2168
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2169
                    current_path_info = current_dir_info[1][path_index]
 
2170
                    if current_path_info[2] == 'directory':
 
2171
                        if self.target._directory_is_tree_reference(
 
2172
                            current_path_info[0].decode('utf8')):
 
2173
                            current_path_info = current_path_info[:2] + \
 
2174
                                ('tree-reference',) + current_path_info[3:]
 
2175
                else:
 
2176
                    current_path_info = None
 
2177
                advance_path = True
 
2178
                path_handled = False
 
2179
                while (current_entry is not None or
 
2180
                    current_path_info is not None):
 
2181
                    if current_entry is None:
 
2182
                        # the check for path_handled when the path is adnvaced
 
2183
                        # will yield this path if needed.
 
2184
                        pass
 
2185
                    elif current_path_info is None:
 
2186
                        # no path is fine: the per entry code will handle it.
 
2187
                        for result in _process_entry(current_entry, current_path_info):
 
2188
                            # this check should probably be outside the loop: one
 
2189
                            # 'iterate two trees' api, and then _iter_changes filters
 
2190
                            # unchanged pairs. - RBC 20070226
 
2191
                            if (include_unchanged
 
2192
                                or result[2]                    # content change
 
2193
                                or result[3][0] != result[3][1] # versioned status
 
2194
                                or result[4][0] != result[4][1] # parent id
 
2195
                                or result[5][0] != result[5][1] # name
 
2196
                                or result[6][0] != result[6][1] # kind
 
2197
                                or result[7][0] != result[7][1] # executable
 
2198
                                ):
 
2199
                                yield (result[0],
 
2200
                                       (utf8_decode_or_none(result[1][0]),
 
2201
                                        utf8_decode_or_none(result[1][1])),
 
2202
                                       result[2],
 
2203
                                       result[3],
 
2204
                                       result[4],
 
2205
                                       (utf8_decode_or_none(result[5][0]),
 
2206
                                        utf8_decode_or_none(result[5][1])),
 
2207
                                       result[6],
 
2208
                                       result[7],
 
2209
                                      )
 
2210
                    elif current_entry[0][1] != current_path_info[1]:
 
2211
                        if current_path_info[1] < current_entry[0][1]:
 
2212
                            # extra file on disk: pass for now, but only
 
2213
                            # increment the path, not the entry
 
2214
                            advance_entry = False
 
2215
                        else:
 
2216
                            # entry referring to file not present on disk.
 
2217
                            # advance the entry only, after processing.
 
2218
                            for result in _process_entry(current_entry, None):
 
2219
                                # this check should probably be outside the loop: one
 
2220
                                # 'iterate two trees' api, and then _iter_changes filters
 
2221
                                # unchanged pairs. - RBC 20070226
 
2222
                                path_handled = True
 
2223
                                if (include_unchanged
 
2224
                                    or result[2]                    # content change
 
2225
                                    or result[3][0] != result[3][1] # versioned status
 
2226
                                    or result[4][0] != result[4][1] # parent id
 
2227
                                    or result[5][0] != result[5][1] # name
 
2228
                                    or result[6][0] != result[6][1] # kind
 
2229
                                    or result[7][0] != result[7][1] # executable
 
2230
                                    ):
 
2231
                                    yield (result[0],
 
2232
                                           (utf8_decode_or_none(result[1][0]),
 
2233
                                            utf8_decode_or_none(result[1][1])),
 
2234
                                           result[2],
 
2235
                                           result[3],
 
2236
                                           result[4],
 
2237
                                           (utf8_decode_or_none(result[5][0]),
 
2238
                                            utf8_decode_or_none(result[5][1])),
 
2239
                                           result[6],
 
2240
                                           result[7],
 
2241
                                          )
 
2242
                            advance_path = False
 
2243
                    else:
 
2244
                        for result in _process_entry(current_entry, current_path_info):
 
2245
                            # this check should probably be outside the loop: one
 
2246
                            # 'iterate two trees' api, and then _iter_changes filters
 
2247
                            # unchanged pairs. - RBC 20070226
 
2248
                            path_handled = True
 
2249
                            if (include_unchanged
 
2250
                                or result[2]                    # content change
 
2251
                                or result[3][0] != result[3][1] # versioned status
 
2252
                                or result[4][0] != result[4][1] # parent id
 
2253
                                or result[5][0] != result[5][1] # name
 
2254
                                or result[6][0] != result[6][1] # kind
 
2255
                                or result[7][0] != result[7][1] # executable
 
2256
                                ):
 
2257
                                yield (result[0],
 
2258
                                       (utf8_decode_or_none(result[1][0]),
 
2259
                                        utf8_decode_or_none(result[1][1])),
 
2260
                                       result[2],
 
2261
                                       result[3],
 
2262
                                       result[4],
 
2263
                                       (utf8_decode_or_none(result[5][0]),
 
2264
                                        utf8_decode_or_none(result[5][1])),
 
2265
                                       result[6],
 
2266
                                       result[7],
 
2267
                                      )
 
2268
                    if advance_entry and current_entry is not None:
 
2269
                        entry_index += 1
 
2270
                        if entry_index < len(current_block[1]):
 
2271
                            current_entry = current_block[1][entry_index]
 
2272
                        else:
 
2273
                            current_entry = None
 
2274
                    else:
 
2275
                        advance_entry = True # reset the advance flaga
 
2276
                    if advance_path and current_path_info is not None:
 
2277
                        if not path_handled:
 
2278
                            # unversioned in all regards
 
2279
                            if want_unversioned:
 
2280
                                new_executable = bool(
 
2281
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2282
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2283
                                if want_unversioned:
 
2284
                                    yield (None,
 
2285
                                        (None, utf8_decode_or_none(current_path_info[0])),
 
2286
                                        True,
 
2287
                                        (False, False),
 
2288
                                        (None, None),
 
2289
                                        (None, utf8_decode_or_none(current_path_info[1])),
 
2290
                                        (None, current_path_info[2]),
 
2291
                                        (None, new_executable))
 
2292
                            # dont descend into this unversioned path if it is
 
2293
                            # a dir
 
2294
                            if current_path_info[2] in ('directory'):
 
2295
                                del current_dir_info[1][path_index]
 
2296
                                path_index -= 1
 
2297
                        # dont descend the disk iterator into any tree 
 
2298
                        # paths.
 
2299
                        if current_path_info[2] == 'tree-reference':
 
2300
                            del current_dir_info[1][path_index]
 
2301
                            path_index -= 1
 
2302
                        path_index += 1
 
2303
                        if path_index < len(current_dir_info[1]):
 
2304
                            current_path_info = current_dir_info[1][path_index]
 
2305
                            if current_path_info[2] == 'directory':
 
2306
                                if self.target._directory_is_tree_reference(
 
2307
                                    current_path_info[0].decode('utf8')):
 
2308
                                    current_path_info = current_path_info[:2] + \
 
2309
                                        ('tree-reference',) + current_path_info[3:]
 
2310
                        else:
 
2311
                            current_path_info = None
 
2312
                        path_handled = False
 
2313
                    else:
 
2314
                        advance_path = True # reset the advance flagg.
 
2315
                if current_block is not None:
 
2316
                    block_index += 1
 
2317
                    if (block_index < len(state._dirblocks) and
 
2318
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2319
                        current_block = state._dirblocks[block_index]
 
2320
                    else:
 
2321
                        current_block = None
 
2322
                if current_dir_info is not None:
 
2323
                    try:
 
2324
                        current_dir_info = dir_iterator.next()
 
2325
                    except StopIteration:
 
2326
                        current_dir_info = None
 
2327
 
2094
2328
 
2095
2329
    @staticmethod
2096
2330
    def is_compatible(source, target):
2097
2331
        # the target must be a dirstate working tree
2098
 
        if not isinstance(target, DirStateWorkingTree):
 
2332
        if not isinstance(target, WorkingTree4):
2099
2333
            return False
2100
 
        # the source must be a revtree or dirstate rev tree.
 
2334
        # the source must be a revtreee or dirstate rev tree.
2101
2335
        if not isinstance(source,
2102
2336
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2103
2337
            return False
2104
2338
        # the source revid must be in the target dirstate
2105
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2339
        if not (source._revision_id == NULL_REVISION or
2106
2340
            source._revision_id in target.get_parent_ids()):
2107
 
            # TODO: what about ghosts? it may well need to
 
2341
            # TODO: what about ghosts? it may well need to 
2108
2342
            # check for them explicitly.
2109
2343
            return False
2110
2344
        return True
2120
2354
 
2121
2355
    def convert(self, tree):
2122
2356
        # 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
 
2357
        # on-unlock behaviours, and so that noone else diddles with the 
2124
2358
        # tree during upgrade.
2125
2359
        tree._control_files.lock_write()
2126
2360
        try:
2152
2386
 
2153
2387
    def update_format(self, tree):
2154
2388
        """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())
 
2389
        tree._control_files.put_utf8('format',
 
2390
            self.target_format.get_format_string())