~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

merge bzr.dev

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2007-2011 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
 
 
41
import bzrlib
34
42
from bzrlib import (
35
43
    bzrdir,
36
44
    cache_utf8,
37
45
    conflicts as _mod_conflicts,
38
 
    debug,
 
46
    delta,
39
47
    dirstate,
40
48
    errors,
41
 
    filters as _mod_filters,
42
49
    generate_ids,
 
50
    globbing,
 
51
    hashcache,
 
52
    ignores,
 
53
    merge,
43
54
    osutils,
44
 
    revision as _mod_revision,
45
55
    revisiontree,
46
 
    trace,
 
56
    textui,
47
57
    transform,
48
 
    views,
 
58
    urlutils,
 
59
    xml5,
 
60
    xml6,
49
61
    )
 
62
import bzrlib.branch
 
63
from bzrlib.transport import get_transport
 
64
import bzrlib.ui
50
65
""")
51
66
 
 
67
from bzrlib import symbol_versioning
52
68
from bzrlib.decorators import needs_read_lock, needs_write_lock
53
 
from bzrlib.inventory import Inventory, ROOT_ID, entry_factory
54
 
from bzrlib.lock import LogicalLockResult
55
 
from bzrlib.lockable_files import LockableFiles
 
69
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
70
from bzrlib.lockable_files import LockableFiles, TransportLock
56
71
from bzrlib.lockdir import LockDir
 
72
import bzrlib.mutabletree
57
73
from bzrlib.mutabletree import needs_tree_write_lock
58
74
from bzrlib.osutils import (
59
75
    file_kind,
60
76
    isdir,
 
77
    normpath,
61
78
    pathjoin,
 
79
    rand_chars,
62
80
    realpath,
63
81
    safe_unicode,
 
82
    splitpath,
64
83
    )
 
84
from bzrlib.trace import mutter, note
65
85
from bzrlib.transport.local import LocalTransport
66
 
from bzrlib.tree import (
67
 
    InterTree,
68
 
    InventoryTree,
69
 
    )
70
 
from bzrlib.workingtree import (
71
 
    InventoryWorkingTree,
72
 
    WorkingTree,
73
 
    WorkingTreeFormat,
74
 
    )
75
 
 
76
 
 
77
 
class DirStateWorkingTree(InventoryWorkingTree):
78
 
 
79
 
    _DEFAULT_WORTH_SAVING_LIMIT = 10
 
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
        )
 
95
from bzrlib.tree import Tree
 
96
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
 
97
 
 
98
 
 
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
    """
80
116
 
81
117
    def __init__(self, basedir,
82
118
                 branch,
92
128
        """
93
129
        self._format = _format
94
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
95
134
        basedir = safe_unicode(basedir)
96
 
        trace.mutter("opening working tree %r", basedir)
 
135
        mutter("opening working tree %r", basedir)
97
136
        self._branch = branch
 
137
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
138
            "branch %r is not a Branch" % self.branch
98
139
        self.basedir = realpath(basedir)
99
140
        # if branch is at our basedir and is a format 6 or less
100
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
101
144
        self._control_files = _control_files
102
 
        self._transport = self._control_files._transport
103
145
        self._dirty = None
104
146
        #-------------
105
147
        # during a read or write lock these objects are set, and are
107
149
        self._dirstate = None
108
150
        self._inventory = None
109
151
        #-------------
110
 
        self._setup_directory_is_tree_reference()
111
 
        self._detect_case_handling()
112
 
        self._rules_searcher = None
113
 
        self.views = self._make_views()
114
 
        #--- allow tests to select the dirstate iter_changes implementation
115
 
        self._iter_changes = dirstate._process_entry
116
152
 
117
153
    @needs_tree_write_lock
118
154
    def _add(self, files, ids, kinds):
120
156
        state = self.current_dirstate()
121
157
        for f, file_id, kind in zip(files, ids, kinds):
122
158
            f = f.strip('/')
 
159
            assert '//' not in f
 
160
            assert '..' not in f
123
161
            if self.path2id(f):
124
162
                # special case tree root handling.
125
163
                if f == '' and self.path2id(f) == ROOT_ID:
133
171
            state.add(f, file_id, kind, None, '')
134
172
        self._make_dirty(reset_inventory=True)
135
173
 
136
 
    def _get_check_refs(self):
137
 
        """Return the references needed to perform a check of this tree."""
138
 
        return [('trees', self.last_revision())]
139
 
 
140
174
    def _make_dirty(self, reset_inventory):
141
175
        """Make the tree state dirty.
142
176
 
150
184
    @needs_tree_write_lock
151
185
    def add_reference(self, sub_tree):
152
186
        # use standard implementation, which calls back to self._add
153
 
        #
 
187
        # 
154
188
        # So we don't store the reference_revision in the working dirstate,
155
 
        # it's just recorded at the moment of commit.
 
189
        # it's just recorded at the moment of commit. 
156
190
        self._add_reference(sub_tree)
157
191
 
158
192
    def break_lock(self):
194
228
 
195
229
    def _comparison_data(self, entry, path):
196
230
        kind, executable, stat_value = \
197
 
            WorkingTree._comparison_data(self, entry, path)
 
231
            WorkingTree3._comparison_data(self, entry, path)
198
232
        # it looks like a plain directory, but it's really a reference -- see
199
233
        # also kind()
200
 
        if (self._repo_supports_tree_reference and kind == 'directory'
201
 
            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)):
202
237
            kind = 'tree-reference'
203
238
        return kind, executable, stat_value
204
239
 
206
241
    def commit(self, message=None, revprops=None, *args, **kwargs):
207
242
        # mark the tree as dirty post commit - commit
208
243
        # can change the current versioned list by doing deletes.
209
 
        result = WorkingTree.commit(self, message, revprops, *args, **kwargs)
 
244
        result = WorkingTree3.commit(self, message, revprops, *args, **kwargs)
210
245
        self._make_dirty(reset_inventory=True)
211
246
        return result
212
247
 
230
265
            return self._dirstate
231
266
        local_path = self.bzrdir.get_workingtree_transport(None
232
267
            ).local_abspath('dirstate')
233
 
        self._dirstate = dirstate.DirState.on_file(local_path,
234
 
            self._sha1_provider(), self._worth_saving_limit())
 
268
        self._dirstate = dirstate.DirState.on_file(local_path)
235
269
        return self._dirstate
236
270
 
237
 
    def _sha1_provider(self):
238
 
        """A function that returns a SHA1Provider suitable for this tree.
239
 
 
240
 
        :return: None if content filtering is not supported by this tree.
241
 
          Otherwise, a SHA1Provider is returned that sha's the canonical
242
 
          form of files, i.e. after read filters are applied.
243
 
        """
244
 
        if self.supports_content_filtering():
245
 
            return ContentFilterAwareSHA1Provider(self)
246
 
        else:
247
 
            return None
248
 
 
249
 
    def _worth_saving_limit(self):
250
 
        """How many hash changes are ok before we must save the dirstate.
251
 
 
252
 
        :return: an integer. -1 means never save.
253
 
        """
254
 
        config = self.branch.get_config()
255
 
        val = config.get_user_option('bzr.workingtree.worth_saving_limit')
256
 
        if val is None:
257
 
            val = self._DEFAULT_WORTH_SAVING_LIMIT
258
 
        else:
259
 
            try:
260
 
                val = int(val)
261
 
            except ValueError, e:
262
 
                trace.warning('Invalid config value for'
263
 
                              ' "bzr.workingtree.worth_saving_limit"'
264
 
                              ' value %r is not an integer.'
265
 
                              % (val,))
266
 
                val = self._DEFAULT_WORTH_SAVING_LIMIT
267
 
        return val
 
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
268
285
 
269
286
    def filter_unversioned_files(self, paths):
270
287
        """Filter out paths that are versioned.
303
320
 
304
321
    def _generate_inventory(self):
305
322
        """Create and set self.inventory from the dirstate object.
306
 
 
 
323
        
307
324
        This is relatively expensive: we have to walk the entire dirstate.
308
325
        Ideally we would not, and can deprecate this function.
309
326
        """
313
330
        state._read_dirblocks_if_needed()
314
331
        root_key, current_entry = self._get_entry(path='')
315
332
        current_id = root_key[2]
316
 
        if not (current_entry[0][0] == 'd'): # directory
317
 
            raise AssertionError(current_entry)
 
333
        assert current_entry[0][0] == 'd' # directory
318
334
        inv = Inventory(root_id=current_id)
319
335
        # Turn some things into local variables
320
336
        minikind_to_kind = dirstate.DirState._minikind_to_kind
353
369
                    # add this entry to the parent map.
354
370
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
355
371
                elif kind == 'tree-reference':
356
 
                    if not self._repo_supports_tree_reference:
357
 
                        raise errors.UnsupportedOperation(
358
 
                            self._generate_inventory,
359
 
                            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)
360
377
                    inv_entry.reference_revision = link_or_sha1 or None
361
378
                elif kind != 'symlink':
362
379
                    raise AssertionError("unknown kind %r" % kind)
363
380
                # These checks cost us around 40ms on a 55k entry tree
364
 
                if file_id in inv_byid:
365
 
                    raise AssertionError('file_id %s already in'
366
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
367
 
                if name_unicode in parent_ie.children:
368
 
                    raise AssertionError('name %r already in parent'
369
 
                        % (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
370
384
                inv_byid[file_id] = inv_entry
371
385
                parent_ie.children[name_unicode] = inv_entry
372
386
        self._inventory = inv
377
391
        If either file_id or path is supplied, it is used as the key to lookup.
378
392
        If both are supplied, the fastest lookup is used, and an error is
379
393
        raised if they do not both point at the same row.
380
 
 
 
394
        
381
395
        :param file_id: An optional unicode file_id to be looked up.
382
396
        :param path: An optional unicode path to be looked up.
383
397
        :return: The dirstate row tuple for path/file_id, or (None, None)
392
406
    def get_file_sha1(self, file_id, path=None, stat_value=None):
393
407
        # check file id is valid unconditionally.
394
408
        entry = self._get_entry(file_id=file_id, path=path)
395
 
        if entry[0] is None:
396
 
            raise errors.NoSuchId(self, file_id)
 
409
        assert entry[0] is not None, 'what error should this raise'
397
410
        if path is None:
398
411
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
399
412
 
400
413
        file_abspath = self.abspath(path)
401
414
        state = self.current_dirstate()
402
 
        if stat_value is None:
403
 
            try:
404
 
                stat_value = osutils.lstat(file_abspath)
405
 
            except OSError, e:
406
 
                if e.errno == errno.ENOENT:
407
 
                    return None
408
 
                else:
409
 
                    raise
410
 
        link_or_sha1 = dirstate.update_entry(state, entry, file_abspath,
411
 
            stat_value=stat_value)
 
415
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
416
                                          stat_value=stat_value)
412
417
        if entry[1][0][0] == 'f':
413
 
            if link_or_sha1 is None:
414
 
                file_obj, statvalue = self.get_file_with_stat(file_id, path)
415
 
                try:
416
 
                    sha1 = osutils.sha_file(file_obj)
417
 
                finally:
418
 
                    file_obj.close()
419
 
                self._observed_sha1(file_id, path, (sha1, statvalue))
420
 
                return sha1
421
 
            else:
422
 
                return link_or_sha1
 
418
            return link_or_sha1
423
419
        return None
424
420
 
425
421
    def _get_inventory(self):
426
422
        """Get the inventory for the tree. This is only valid within a lock."""
427
 
        if 'evil' in debug.debug_flags:
428
 
            trace.mutter_callsite(2,
429
 
                "accessing .inventory forces a size of tree translation.")
430
423
        if self._inventory is not None:
431
424
            return self._inventory
432
425
        self._must_be_locked()
439
432
    @needs_read_lock
440
433
    def get_parent_ids(self):
441
434
        """See Tree.get_parent_ids.
442
 
 
 
435
        
443
436
        This implementation requests the ids list from the dirstate file.
444
437
        """
445
438
        return self.current_dirstate().get_parent_ids()
461
454
 
462
455
    def has_id(self, file_id):
463
456
        state = self.current_dirstate()
 
457
        file_id = osutils.safe_file_id(file_id)
464
458
        row, parents = self._get_entry(file_id=file_id)
465
459
        if row is None:
466
460
            return False
467
461
        return osutils.lexists(pathjoin(
468
462
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
469
463
 
470
 
    def has_or_had_id(self, file_id):
471
 
        state = self.current_dirstate()
472
 
        row, parents = self._get_entry(file_id=file_id)
473
 
        return row is not None
474
 
 
475
464
    @needs_read_lock
476
465
    def id2path(self, file_id):
477
 
        "Convert a file-id to a path."
 
466
        file_id = osutils.safe_file_id(file_id)
478
467
        state = self.current_dirstate()
479
468
        entry = self._get_entry(file_id=file_id)
480
469
        if entry == (None, None):
482
471
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
483
472
        return path_utf8.decode('utf8')
484
473
 
485
 
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
486
 
        entry = self._get_entry(path=path)
487
 
        if entry == (None, None):
488
 
            return False # Missing entries are not executable
489
 
        return entry[1][0][3] # Executable?
490
 
 
491
474
    if not osutils.supports_executable():
 
475
        @needs_read_lock
492
476
        def is_executable(self, file_id, path=None):
493
 
            """Test if a file is executable or not.
494
 
 
495
 
            Note: The caller is expected to take a read-lock before calling this.
496
 
            """
 
477
            file_id = osutils.safe_file_id(file_id)
497
478
            entry = self._get_entry(file_id=file_id, path=path)
498
479
            if entry == (None, None):
499
480
                return False
500
481
            return entry[1][0][3]
501
 
 
502
 
        _is_executable_from_path_and_stat = \
503
 
            _is_executable_from_path_and_stat_from_basis
504
482
    else:
 
483
        @needs_read_lock
505
484
        def is_executable(self, file_id, path=None):
506
 
            """Test if a file is executable or not.
507
 
 
508
 
            Note: The caller is expected to take a read-lock before calling this.
509
 
            """
510
 
            self._must_be_locked()
511
485
            if not path:
 
486
                file_id = osutils.safe_file_id(file_id)
512
487
                path = self.id2path(file_id)
513
 
            mode = osutils.lstat(self.abspath(path)).st_mode
 
488
            mode = os.lstat(self.abspath(path)).st_mode
514
489
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
515
490
 
516
 
    def all_file_ids(self):
517
 
        """See Tree.iter_all_file_ids"""
518
 
        self._must_be_locked()
519
 
        result = set()
520
 
        for key, tree_details in self.current_dirstate()._iter_entries():
521
 
            if tree_details[0][0] in ('a', 'r'): # relocated
522
 
                continue
523
 
            result.add(key[2])
524
 
        return result
525
 
 
526
491
    @needs_read_lock
527
492
    def __iter__(self):
528
493
        """Iterate through file_ids for this tree.
541
506
        return iter(result)
542
507
 
543
508
    def iter_references(self):
544
 
        if not self._repo_supports_tree_reference:
545
 
            # When the repo doesn't support references, we will have nothing to
546
 
            # return
547
 
            return
548
509
        for key, tree_details in self.current_dirstate()._iter_entries():
549
510
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
550
511
                # not relevant to the working tree
552
513
            if not key[1]:
553
514
                # the root is not a reference.
554
515
                continue
555
 
            relpath = pathjoin(key[0].decode('utf8'), key[1].decode('utf8'))
 
516
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
556
517
            try:
557
 
                if self._kind(relpath) == 'tree-reference':
558
 
                    yield relpath, key[2]
 
518
                if self._kind(path) == 'tree-reference':
 
519
                    yield path, key[2]
559
520
            except errors.NoSuchFile:
560
521
                # path is missing on disk.
561
522
                continue
562
523
 
563
 
    def _observed_sha1(self, file_id, path, (sha1, statvalue)):
564
 
        """See MutableTree._observed_sha1."""
565
 
        state = self.current_dirstate()
566
 
        entry = self._get_entry(file_id=file_id, path=path)
567
 
        state._observed_sha1(entry, sha1, statvalue)
568
 
 
 
524
    @needs_read_lock
569
525
    def kind(self, file_id):
570
526
        """Return the kind of a file.
571
527
 
572
528
        This is always the actual kind that's on disk, regardless of what it
573
529
        was added as.
574
 
 
575
 
        Note: The caller is expected to take a read-lock before calling this.
576
530
        """
577
531
        relpath = self.id2path(file_id)
578
 
        if relpath is None:
579
 
            raise AssertionError(
580
 
                "path for id {%s} is None!" % file_id)
 
532
        assert relpath != None, \
 
533
            "path for id {%s} is None!" % file_id
581
534
        return self._kind(relpath)
582
535
 
583
536
    def _kind(self, relpath):
584
537
        abspath = self.abspath(relpath)
585
538
        kind = file_kind(abspath)
586
 
        if (self._repo_supports_tree_reference and kind == 'directory'):
587
 
            entry = self._get_entry(path=relpath)
588
 
            if entry[1] is not None:
589
 
                if entry[1][0][0] == 't':
590
 
                    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'
591
543
        return kind
592
544
 
593
545
    @needs_read_lock
597
549
        if parent_ids:
598
550
            return parent_ids[0]
599
551
        else:
600
 
            return _mod_revision.NULL_REVISION
 
552
            return None
601
553
 
602
554
    def lock_read(self):
603
 
        """See Branch.lock_read, and WorkingTree.unlock.
604
 
 
605
 
        :return: A bzrlib.lock.LogicalLockResult.
606
 
        """
 
555
        """See Branch.lock_read, and WorkingTree.unlock."""
607
556
        self.branch.lock_read()
608
557
        try:
609
558
            self._control_files.lock_read()
622
571
        except:
623
572
            self.branch.unlock()
624
573
            raise
625
 
        return LogicalLockResult(self.unlock)
626
574
 
627
575
    def _lock_self_write(self):
628
576
        """This should be called after the branch is locked."""
643
591
        except:
644
592
            self.branch.unlock()
645
593
            raise
646
 
        return LogicalLockResult(self.unlock)
647
594
 
648
595
    def lock_tree_write(self):
649
 
        """See MutableTree.lock_tree_write, and WorkingTree.unlock.
650
 
 
651
 
        :return: A bzrlib.lock.LogicalLockResult.
652
 
        """
 
596
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
653
597
        self.branch.lock_read()
654
 
        return self._lock_self_write()
 
598
        self._lock_self_write()
655
599
 
656
600
    def lock_write(self):
657
 
        """See MutableTree.lock_write, and WorkingTree.unlock.
658
 
 
659
 
        :return: A bzrlib.lock.LogicalLockResult.
660
 
        """
 
601
        """See MutableTree.lock_write, and WorkingTree.unlock."""
661
602
        self.branch.lock_write()
662
 
        return self._lock_self_write()
 
603
        self._lock_self_write()
663
604
 
664
605
    @needs_tree_write_lock
665
606
    def move(self, from_paths, to_dir, after=False):
667
608
        result = []
668
609
        if not from_paths:
669
610
            return result
 
611
 
670
612
        state = self.current_dirstate()
671
 
        if isinstance(from_paths, basestring):
672
 
            raise ValueError()
 
613
 
 
614
        assert not isinstance(from_paths, basestring)
673
615
        to_dir_utf8 = to_dir.encode('utf8')
674
616
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
675
617
        id_index = state._get_id_index()
697
639
        if self._inventory is not None:
698
640
            update_inventory = True
699
641
            inv = self.inventory
 
642
            to_dir_ie = inv[to_dir_id]
700
643
            to_dir_id = to_entry[0][2]
701
 
            to_dir_ie = inv[to_dir_id]
702
644
        else:
703
645
            update_inventory = False
704
646
 
727
669
            new_entry = to_block[1][added_entry_index]
728
670
            rollbacks.append(lambda:state._make_absent(new_entry))
729
671
 
 
672
        # create rename entries and tuples
730
673
        for from_rel in from_paths:
731
674
            # from_rel is 'pathinroot/foo/bar'
732
675
            from_rel_utf8 = from_rel.encode('utf8')
735
678
            from_entry = self._get_entry(path=from_rel)
736
679
            if from_entry == (None, None):
737
680
                raise errors.BzrMoveFailedError(from_rel,to_dir,
738
 
                    errors.NotVersionedError(path=from_rel))
 
681
                    errors.NotVersionedError(path=str(from_rel)))
739
682
 
740
683
            from_id = from_entry[0][2]
741
684
            to_rel = pathjoin(to_dir, from_tail)
768
711
                if from_missing: # implicitly just update our path mapping
769
712
                    move_file = False
770
713
                elif not after:
771
 
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
714
                    raise errors.RenameFailedFilesExist(from_rel, to_rel,
 
715
                        extra="(Use --after to update the Bazaar id)")
772
716
 
773
717
            rollbacks = []
774
718
            def rollback_rename():
829
773
 
830
774
                if minikind == 'd':
831
775
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
832
 
                        """Recursively update all entries in this dirblock."""
833
 
                        if from_dir == '':
834
 
                            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"
835
782
                        from_key = (from_dir, '')
836
783
                        from_block_idx, present = \
837
784
                            state._find_block_index_from_key(from_key)
847
794
                        to_block_index = state._ensure_block(
848
795
                            to_block_index, to_entry_index, to_dir_utf8)
849
796
                        to_block = state._dirblocks[to_block_index]
850
 
 
851
 
                        # Grab a copy since move_one may update the list.
852
 
                        for entry in from_block[1][:]:
853
 
                            if not (entry[0][0] == from_dir):
854
 
                                raise AssertionError()
 
797
                        for entry in from_block[1]:
 
798
                            assert entry[0][0] == from_dir
855
799
                            cur_details = entry[1][0]
856
800
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
857
801
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
858
802
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
859
803
                            minikind = cur_details[0]
860
 
                            if minikind in 'ar':
861
 
                                # Deleted children of a renamed directory
862
 
                                # Do not need to be updated.
863
 
                                # Children that have been renamed out of this
864
 
                                # directory should also not be updated
865
 
                                continue
866
804
                            move_one(entry, from_path_utf8=from_path_utf8,
867
805
                                     minikind=minikind,
868
806
                                     executable=cur_details[3],
882
820
                rollback_rename()
883
821
                raise
884
822
            result.append((from_rel, to_rel))
885
 
            state._mark_modified()
 
823
            state._dirblock_state = dirstate.DirState.IN_MEMORY_MODIFIED
886
824
            self._make_dirty(reset_inventory=False)
887
825
 
888
826
        return result
916
854
        for tree in trees:
917
855
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
918
856
                parents):
919
 
                return super(DirStateWorkingTree, self).paths2ids(paths,
920
 
                    trees, require_versioned)
 
857
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
921
858
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
922
859
        # -- make all paths utf8 --
923
860
        paths_utf8 = set()
983
920
            if not all_versioned:
984
921
                raise errors.PathsNotVersionedError(paths)
985
922
        # -- remove redundancy in supplied paths to prevent over-scanning --
986
 
        search_paths = osutils.minimum_path_selection(paths)
987
 
        # 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: 
988
930
        # for all search_indexs in each path at or under each element of
989
931
        # search_paths, if the detail is relocated: add the id, and add the
990
932
        # relocated path as one to search if its not searched already. If the
1046
988
 
1047
989
    def read_working_inventory(self):
1048
990
        """Read the working inventory.
1049
 
 
 
991
        
1050
992
        This is a meaningless operation for dirstate, but we obey it anyhow.
1051
993
        """
1052
994
        return self.inventory
1057
999
 
1058
1000
        WorkingTree4 supplies revision_trees for any basis tree.
1059
1001
        """
 
1002
        revision_id = osutils.safe_revision_id(revision_id)
1060
1003
        dirstate = self.current_dirstate()
1061
1004
        parent_ids = dirstate.get_parent_ids()
1062
1005
        if revision_id not in parent_ids:
1069
1012
    @needs_tree_write_lock
1070
1013
    def set_last_revision(self, new_revision):
1071
1014
        """Change the last revision in the working tree."""
 
1015
        new_revision = osutils.safe_revision_id(new_revision)
1072
1016
        parents = self.get_parent_ids()
1073
 
        if new_revision in (_mod_revision.NULL_REVISION, None):
1074
 
            if len(parents) >= 2:
1075
 
                raise AssertionError(
1076
 
                    "setting the last parent to none with a pending merge is "
1077
 
                    "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.")
1078
1021
            self.set_parent_ids([])
1079
1022
        else:
1080
1023
            self.set_parent_ids([new_revision] + parents[1:],
1083
1026
    @needs_tree_write_lock
1084
1027
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
1085
1028
        """Set the parent ids to revision_ids.
1086
 
 
 
1029
        
1087
1030
        See also set_parent_trees. This api will try to retrieve the tree data
1088
1031
        for each element of revision_ids from the trees repository. If you have
1089
1032
        tree data already available, it is more efficient to use
1093
1036
        :param revision_ids: The revision_ids to set as the parent ids of this
1094
1037
            working tree. Any of these may be ghosts.
1095
1038
        """
 
1039
        revision_ids = [osutils.safe_revision_id(r) for r in revision_ids]
1096
1040
        trees = []
1097
1041
        for revision_id in revision_ids:
1098
1042
            try:
1104
1048
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
1105
1049
                revtree = None
1106
1050
            trees.append((revision_id, revtree))
 
1051
        self.current_dirstate()._validate()
1107
1052
        self.set_parent_trees(trees,
1108
1053
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1054
        self.current_dirstate()._validate()
1109
1055
 
1110
1056
    @needs_tree_write_lock
1111
1057
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
1116
1062
            parent tree - i.e. a ghost.
1117
1063
        """
1118
1064
        dirstate = self.current_dirstate()
 
1065
        dirstate._validate()
1119
1066
        if len(parents_list) > 0:
1120
1067
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
1121
1068
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
1122
1069
        real_trees = []
1123
1070
        ghosts = []
1124
 
 
1125
 
        parent_ids = [rev_id for rev_id, tree in parents_list]
1126
 
        graph = self.branch.repository.get_graph()
1127
 
        heads = graph.heads(parent_ids)
1128
 
        accepted_revisions = set()
1129
 
 
1130
1071
        # convert absent trees to the null tree, which we convert back to
1131
1072
        # missing on access.
1132
1073
        for rev_id, tree in parents_list:
1133
 
            if len(accepted_revisions) > 0:
1134
 
                # we always accept the first tree
1135
 
                if rev_id in accepted_revisions or rev_id not in heads:
1136
 
                    # We have already included either this tree, or its
1137
 
                    # descendent, so we skip it.
1138
 
                    continue
1139
 
            _mod_revision.check_not_reserved_id(rev_id)
 
1074
            rev_id = osutils.safe_revision_id(rev_id)
1140
1075
            if tree is not None:
1141
1076
                real_trees.append((rev_id, tree))
1142
1077
            else:
1143
1078
                real_trees.append((rev_id,
1144
 
                    self.branch.repository.revision_tree(
1145
 
                        _mod_revision.NULL_REVISION)))
 
1079
                    self.branch.repository.revision_tree(None)))
1146
1080
                ghosts.append(rev_id)
1147
 
            accepted_revisions.add(rev_id)
1148
 
        updated = False
1149
 
        if (len(real_trees) == 1
1150
 
            and not ghosts
1151
 
            and self.branch.repository._format.fast_deltas
1152
 
            and isinstance(real_trees[0][1],
1153
 
                revisiontree.InventoryRevisionTree)
1154
 
            and self.get_parent_ids()):
1155
 
            rev_id, rev_tree = real_trees[0]
1156
 
            basis_id = self.get_parent_ids()[0]
1157
 
            # There are times when basis_tree won't be in
1158
 
            # self.branch.repository, (switch, for example)
1159
 
            try:
1160
 
                basis_tree = self.branch.repository.revision_tree(basis_id)
1161
 
            except errors.NoSuchRevision:
1162
 
                # Fall back to the set_parent_trees(), since we can't use
1163
 
                # _make_delta if we can't get the RevisionTree
1164
 
                pass
1165
 
            else:
1166
 
                delta = rev_tree.inventory._make_delta(basis_tree.inventory)
1167
 
                dirstate.update_basis_by_delta(delta, rev_id)
1168
 
                updated = True
1169
 
        if not updated:
1170
 
            dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1081
        dirstate._validate()
 
1082
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1083
        dirstate._validate()
1171
1084
        self._make_dirty(reset_inventory=False)
 
1085
        dirstate._validate()
1172
1086
 
1173
1087
    def _set_root_id(self, file_id):
1174
1088
        """See WorkingTree.set_root_id."""
1177
1091
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
1178
1092
            self._make_dirty(reset_inventory=True)
1179
1093
 
1180
 
    def _sha_from_stat(self, path, stat_result):
1181
 
        """Get a sha digest from the tree's stat cache.
1182
 
 
1183
 
        The default implementation assumes no stat cache is present.
1184
 
 
1185
 
        :param path: The path.
1186
 
        :param stat_result: The stat result being looked up.
1187
 
        """
1188
 
        return self.current_dirstate().sha1_from_stat(path, stat_result)
1189
 
 
1190
1094
    @needs_read_lock
1191
1095
    def supports_tree_reference(self):
1192
1096
        return self._repo_supports_tree_reference
1194
1098
    def unlock(self):
1195
1099
        """Unlock in format 4 trees needs to write the entire dirstate."""
1196
1100
        if self._control_files._lock_count == 1:
1197
 
            # do non-implementation specific cleanup
1198
 
            self._cleanup()
1199
 
 
1200
1101
            # eventually we should do signature checking during read locks for
1201
1102
            # dirstate updates.
1202
1103
            if self._control_files._lock_mode == 'w':
1232
1133
            return
1233
1134
        state = self.current_dirstate()
1234
1135
        state._read_dirblocks_if_needed()
1235
 
        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))
1236
1139
        paths_to_unversion = set()
1237
1140
        # sketch:
1238
1141
        # check if the root is to be unversioned, if so, assert for now.
1265
1168
                # just forget the whole block.
1266
1169
                entry_index = 0
1267
1170
                while entry_index < len(block[1]):
 
1171
                    # Mark this file id as having been removed
1268
1172
                    entry = block[1][entry_index]
1269
 
                    if entry[1][0][0] in 'ar':
1270
 
                        # 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)):
1271
1176
                        entry_index += 1
1272
 
                    else:
1273
 
                        # Mark this file id as having been removed
1274
 
                        ids_to_unversion.discard(entry[0][2])
1275
 
                        if not state._make_absent(entry):
1276
 
                            # The block has not shrunk.
1277
 
                            entry_index += 1
1278
1177
                # go to the next block. (At the moment we dont delete empty
1279
1178
                # dirblocks)
1280
1179
                block_index += 1
1301
1200
        # have to change the legacy inventory too.
1302
1201
        if self._inventory is not None:
1303
1202
            for file_id in file_ids:
1304
 
                if self._inventory.has_id(file_id):
1305
 
                    self._inventory.remove_recursive_id(file_id)
1306
 
 
1307
 
    @needs_tree_write_lock
1308
 
    def rename_one(self, from_rel, to_rel, after=False):
1309
 
        """See WorkingTree.rename_one"""
1310
 
        self.flush()
1311
 
        super(DirStateWorkingTree, self).rename_one(from_rel, to_rel, after)
1312
 
 
1313
 
    @needs_tree_write_lock
1314
 
    def apply_inventory_delta(self, changes):
1315
 
        """See MutableTree.apply_inventory_delta"""
1316
 
        state = self.current_dirstate()
1317
 
        state.update_by_delta(changes)
1318
 
        self._make_dirty(reset_inventory=True)
1319
 
 
1320
 
    def update_basis_by_delta(self, new_revid, delta):
1321
 
        """See MutableTree.update_basis_by_delta."""
1322
 
        if self.last_revision() == new_revid:
1323
 
            raise AssertionError()
1324
 
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
 
1203
                self._inventory.remove_recursive_id(file_id)
1325
1204
 
1326
1205
    @needs_read_lock
1327
1206
    def _validate(self):
1330
1209
    @needs_tree_write_lock
1331
1210
    def _write_inventory(self, inv):
1332
1211
        """Write inventory as the current inventory."""
1333
 
        if self._dirty:
1334
 
            raise AssertionError("attempting to write an inventory when the "
1335
 
                "dirstate is dirty will lose pending changes")
1336
 
        had_inventory = self._inventory is not None
1337
 
        # Setting self._inventory = None forces the dirstate to regenerate the
1338
 
        # working inventory. We do this because self.inventory may be inv, or
1339
 
        # may have been modified, and either case would prevent a clean delta
1340
 
        # being created.
1341
 
        self._inventory = None
1342
 
        # generate a delta,
1343
 
        delta = inv._make_delta(self.inventory)
1344
 
        # and apply it.
1345
 
        self.apply_inventory_delta(delta)
1346
 
        if had_inventory:
 
1212
        assert not self._dirty, "attempting to write an inventory when the dirstate is dirty will cause data loss"
 
1213
        self.current_dirstate().set_state_from_inventory(inv)
 
1214
        self._make_dirty(reset_inventory=False)
 
1215
        if self._inventory is not None:
1347
1216
            self._inventory = inv
1348
1217
        self.flush()
1349
1218
 
1350
 
    @needs_tree_write_lock
1351
 
    def reset_state(self, revision_ids=None):
1352
 
        """Reset the state of the working tree.
1353
 
 
1354
 
        This does a hard-reset to a last-known-good state. This is a way to
1355
 
        fix if something got corrupted (like the .bzr/checkout/dirstate file)
1356
 
        """
1357
 
        if revision_ids is None:
1358
 
            revision_ids = self.get_parent_ids()
1359
 
        if not revision_ids:
1360
 
            base_tree = self.branch.repository.revision_tree(
1361
 
                _mod_revision.NULL_REVISION)
1362
 
            trees = []
1363
 
        else:
1364
 
            trees = zip(revision_ids,
1365
 
                        self.branch.repository.revision_trees(revision_ids))
1366
 
            base_tree = trees[0][1]
1367
 
        state = self.current_dirstate()
1368
 
        # We don't support ghosts yet
1369
 
        state.set_state_from_scratch(base_tree.inventory, trees, [])
1370
 
 
1371
 
 
1372
 
class ContentFilterAwareSHA1Provider(dirstate.SHA1Provider):
1373
 
 
1374
 
    def __init__(self, tree):
1375
 
        self.tree = tree
1376
 
 
1377
 
    def sha1(self, abspath):
1378
 
        """See dirstate.SHA1Provider.sha1()."""
1379
 
        filters = self.tree._content_filter_stack(
1380
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1381
 
        return _mod_filters.internal_size_sha_file_byname(abspath, filters)[1]
1382
 
 
1383
 
    def stat_and_sha1(self, abspath):
1384
 
        """See dirstate.SHA1Provider.stat_and_sha1()."""
1385
 
        filters = self.tree._content_filter_stack(
1386
 
            self.tree.relpath(osutils.safe_unicode(abspath)))
1387
 
        file_obj = file(abspath, 'rb', 65000)
1388
 
        try:
1389
 
            statvalue = os.fstat(file_obj.fileno())
1390
 
            if filters:
1391
 
                file_obj = _mod_filters.filtered_input_file(file_obj, filters)
1392
 
            sha1 = osutils.size_sha_file(file_obj)[1]
1393
 
        finally:
1394
 
            file_obj.close()
1395
 
        return statvalue, sha1
1396
 
 
1397
 
 
1398
 
class ContentFilteringDirStateWorkingTree(DirStateWorkingTree):
1399
 
    """Dirstate working tree that supports content filtering.
1400
 
 
1401
 
    The dirstate holds the hash and size of the canonical form of the file, 
1402
 
    and most methods must return that.
1403
 
    """
1404
 
 
1405
 
    def _file_content_summary(self, path, stat_result):
1406
 
        # This is to support the somewhat obsolete path_content_summary method
1407
 
        # with content filtering: see
1408
 
        # <https://bugs.launchpad.net/bzr/+bug/415508>.
1409
 
        #
1410
 
        # If the dirstate cache is up to date and knows the hash and size,
1411
 
        # return that.
1412
 
        # Otherwise if there are no content filters, return the on-disk size
1413
 
        # and leave the hash blank.
1414
 
        # Otherwise, read and filter the on-disk file and use its size and
1415
 
        # hash.
1416
 
        #
1417
 
        # The dirstate doesn't store the size of the canonical form so we
1418
 
        # can't trust it for content-filtered trees.  We just return None.
1419
 
        dirstate_sha1 = self._dirstate.sha1_from_stat(path, stat_result)
1420
 
        executable = self._is_executable_from_path_and_stat(path, stat_result)
1421
 
        return ('file', None, executable, dirstate_sha1)
1422
 
 
1423
 
 
1424
 
class WorkingTree4(DirStateWorkingTree):
1425
 
    """This is the Format 4 working tree.
1426
 
 
1427
 
    This differs from WorkingTree by:
1428
 
     - Having a consolidated internal dirstate, stored in a
1429
 
       randomly-accessible sorted file on disk.
1430
 
     - Not having a regular inventory attribute.  One can be synthesized
1431
 
       on demand but this is expensive and should be avoided.
1432
 
 
1433
 
    This is new in bzr 0.15.
1434
 
    """
1435
 
 
1436
 
 
1437
 
class WorkingTree5(ContentFilteringDirStateWorkingTree):
1438
 
    """This is the Format 5 working tree.
1439
 
 
1440
 
    This differs from WorkingTree4 by:
1441
 
     - Supporting content filtering.
1442
 
 
1443
 
    This is new in bzr 1.11.
1444
 
    """
1445
 
 
1446
 
 
1447
 
class WorkingTree6(ContentFilteringDirStateWorkingTree):
1448
 
    """This is the Format 6 working tree.
1449
 
 
1450
 
    This differs from WorkingTree5 by:
1451
 
     - Supporting a current view that may mask the set of files in a tree
1452
 
       impacted by most user operations.
1453
 
 
1454
 
    This is new in bzr 1.14.
1455
 
    """
1456
 
 
1457
 
    def _make_views(self):
1458
 
        return views.PathBasedViews(self)
1459
 
 
1460
 
 
1461
 
class DirStateWorkingTreeFormat(WorkingTreeFormat):
1462
 
 
1463
 
    missing_parent_conflicts = True
1464
 
 
1465
 
    supports_versioned_directories = True
1466
 
 
1467
 
    _lock_class = LockDir
1468
 
    _lock_file_name = 'lock'
1469
 
 
1470
 
    def _open_control_files(self, a_bzrdir):
1471
 
        transport = a_bzrdir.get_workingtree_transport(None)
1472
 
        return LockableFiles(transport, self._lock_file_name,
1473
 
                             self._lock_class)
1474
 
 
1475
 
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
1476
 
                   accelerator_tree=None, hardlink=False):
 
1219
 
 
1220
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1221
    """The first consolidated dirstate working tree format.
 
1222
 
 
1223
    This format:
 
1224
        - exists within a metadir controlling .bzr
 
1225
        - includes an explicit version marker for the workingtree control
 
1226
          files, separate from the BzrDir format
 
1227
        - modifies the hash cache format
 
1228
        - is new in bzr 0.15
 
1229
        - uses a LockDir to guard access to it.
 
1230
    """
 
1231
 
 
1232
    upgrade_recommended = False
 
1233
 
 
1234
    def get_format_string(self):
 
1235
        """See WorkingTreeFormat.get_format_string()."""
 
1236
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1237
 
 
1238
    def get_format_description(self):
 
1239
        """See WorkingTreeFormat.get_format_description()."""
 
1240
        return "Working tree format 4"
 
1241
 
 
1242
    def initialize(self, a_bzrdir, revision_id=None):
1477
1243
        """See WorkingTreeFormat.initialize().
1478
1244
 
1479
1245
        :param revision_id: allows creating a working tree at a different
1480
 
            revision than the branch is at.
1481
 
        :param accelerator_tree: A tree which can be used for retrieving file
1482
 
            contents more quickly than the revision tree, i.e. a workingtree.
1483
 
            The revision tree will be used for cases where accelerator_tree's
1484
 
            content is different.
1485
 
        :param hardlink: If true, hard-link files from accelerator_tree,
1486
 
            where possible.
 
1246
        revision than the branch is at.
1487
1247
 
1488
 
        These trees get an initial random root id, if their repository supports
1489
 
        rich root data, TREE_ROOT otherwise.
 
1248
        These trees get an initial random root id.
1490
1249
        """
 
1250
        revision_id = osutils.safe_revision_id(revision_id)
1491
1251
        if not isinstance(a_bzrdir.transport, LocalTransport):
1492
1252
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
1493
1253
        transport = a_bzrdir.get_workingtree_transport(self)
1494
1254
        control_files = self._open_control_files(a_bzrdir)
1495
1255
        control_files.create_lock()
1496
1256
        control_files.lock_write()
1497
 
        transport.put_bytes('format', self.get_format_string(),
1498
 
            mode=a_bzrdir._get_file_mode())
1499
 
        if from_branch is not None:
1500
 
            branch = from_branch
1501
 
        else:
1502
 
            branch = a_bzrdir.open_branch()
 
1257
        control_files.put_utf8('format', self.get_format_string())
 
1258
        branch = a_bzrdir.open_branch()
1503
1259
        if revision_id is None:
1504
1260
            revision_id = branch.last_revision()
1505
1261
        local_path = transport.local_abspath('dirstate')
1507
1263
        state = dirstate.DirState.initialize(local_path)
1508
1264
        state.unlock()
1509
1265
        del state
1510
 
        wt = self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1266
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1511
1267
                         branch,
1512
1268
                         _format=self,
1513
1269
                         _bzrdir=a_bzrdir,
1514
1270
                         _control_files=control_files)
1515
1271
        wt._new_tree()
1516
1272
        wt.lock_tree_write()
 
1273
        wt.current_dirstate()._validate()
1517
1274
        try:
1518
 
            self._init_custom_control_files(wt)
1519
 
            if revision_id in (None, _mod_revision.NULL_REVISION):
1520
 
                if branch.repository.supports_rich_root():
1521
 
                    wt._set_root_id(generate_ids.gen_root_id())
1522
 
                else:
1523
 
                    wt._set_root_id(ROOT_ID)
 
1275
            if revision_id in (None, NULL_REVISION):
 
1276
                wt._set_root_id(generate_ids.gen_root_id())
1524
1277
                wt.flush()
1525
 
            basis = None
1526
 
            # frequently, we will get here due to branching.  The accelerator
1527
 
            # tree will be the tree from the branch, so the desired basis
1528
 
            # tree will often be a parent of the accelerator tree.
1529
 
            if accelerator_tree is not None:
1530
 
                try:
1531
 
                    basis = accelerator_tree.revision_tree(revision_id)
1532
 
                except errors.NoSuchRevision:
1533
 
                    pass
1534
 
            if basis is None:
1535
 
                basis = branch.repository.revision_tree(revision_id)
1536
 
            if revision_id == _mod_revision.NULL_REVISION:
1537
 
                parents_list = []
1538
 
            else:
1539
 
                parents_list = [(revision_id, basis)]
 
1278
                wt.current_dirstate()._validate()
 
1279
            wt.set_last_revision(revision_id)
 
1280
            wt.flush()
 
1281
            basis = wt.basis_tree()
1540
1282
            basis.lock_read()
1541
 
            try:
1542
 
                wt.set_parent_trees(parents_list, allow_leftmost_as_ghost=True)
 
1283
            # if the basis has a root id we have to use that; otherwise we use
 
1284
            # a new random one
 
1285
            basis_root_id = basis.get_root_id()
 
1286
            if basis_root_id is not None:
 
1287
                wt._set_root_id(basis_root_id)
1543
1288
                wt.flush()
1544
 
                # if the basis has a root id we have to use that; otherwise we
1545
 
                # use a new random one
1546
 
                basis_root_id = basis.get_root_id()
1547
 
                if basis_root_id is not None:
1548
 
                    wt._set_root_id(basis_root_id)
1549
 
                    wt.flush()
1550
 
                if wt.supports_content_filtering():
1551
 
                    # The original tree may not have the same content filters
1552
 
                    # applied so we can't safely build the inventory delta from
1553
 
                    # the source tree.
1554
 
                    delta_from_tree = False
1555
 
                else:
1556
 
                    delta_from_tree = True
1557
 
                # delta_from_tree is safe even for DirStateRevisionTrees,
1558
 
                # because wt4.apply_inventory_delta does not mutate the input
1559
 
                # inventory entries.
1560
 
                transform.build_tree(basis, wt, accelerator_tree,
1561
 
                                     hardlink=hardlink,
1562
 
                                     delta_from_tree=delta_from_tree)
1563
 
            finally:
1564
 
                basis.unlock()
 
1289
            transform.build_tree(basis, wt)
 
1290
            basis.unlock()
1565
1291
        finally:
1566
1292
            control_files.unlock()
1567
1293
            wt.unlock()
1568
1294
        return wt
1569
1295
 
1570
 
    def _init_custom_control_files(self, wt):
1571
 
        """Subclasses with custom control files should override this method.
1572
 
 
1573
 
        The working tree and control files are locked for writing when this
1574
 
        method is called.
1575
 
 
1576
 
        :param wt: the WorkingTree object
1577
 
        """
1578
 
 
1579
 
    def open(self, a_bzrdir, _found=False):
1580
 
        """Return the WorkingTree object for a_bzrdir
1581
 
 
1582
 
        _found is a private parameter, do not use it. It is used to indicate
1583
 
               if format probing has already been done.
1584
 
        """
1585
 
        if not _found:
1586
 
            # we are being called directly and must probe.
1587
 
            raise NotImplementedError
1588
 
        if not isinstance(a_bzrdir.transport, LocalTransport):
1589
 
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
1590
 
        wt = self._open(a_bzrdir, self._open_control_files(a_bzrdir))
1591
 
        return wt
1592
 
 
1593
1296
    def _open(self, a_bzrdir, control_files):
1594
1297
        """Open the tree itself.
1595
1298
 
1596
1299
        :param a_bzrdir: the dir for the tree.
1597
1300
        :param control_files: the control files for the tree.
1598
1301
        """
1599
 
        return self._tree_class(a_bzrdir.root_transport.local_abspath('.'),
 
1302
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
1600
1303
                           branch=a_bzrdir.open_branch(),
1601
1304
                           _format=self,
1602
1305
                           _bzrdir=a_bzrdir,
1603
1306
                           _control_files=control_files)
1604
1307
 
1605
1308
    def __get_matchingbzrdir(self):
1606
 
        return self._get_matchingbzrdir()
1607
 
 
1608
 
    def _get_matchingbzrdir(self):
1609
 
        """Overrideable method to get a bzrdir for testing."""
1610
1309
        # please test against something that will let us do tree references
1611
1310
        return bzrdir.format_registry.make_bzrdir(
1612
1311
            'dirstate-with-subtree')
1614
1313
    _matchingbzrdir = property(__get_matchingbzrdir)
1615
1314
 
1616
1315
 
1617
 
class WorkingTreeFormat4(DirStateWorkingTreeFormat):
1618
 
    """The first consolidated dirstate working tree format.
1619
 
 
1620
 
    This format:
1621
 
        - exists within a metadir controlling .bzr
1622
 
        - includes an explicit version marker for the workingtree control
1623
 
          files, separate from the BzrDir format
1624
 
        - modifies the hash cache format
1625
 
        - is new in bzr 0.15
1626
 
        - uses a LockDir to guard access to it.
1627
 
    """
1628
 
 
1629
 
    upgrade_recommended = False
1630
 
 
1631
 
    _tree_class = WorkingTree4
1632
 
 
1633
 
    def get_format_string(self):
1634
 
        """See WorkingTreeFormat.get_format_string()."""
1635
 
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
1636
 
 
1637
 
    def get_format_description(self):
1638
 
        """See WorkingTreeFormat.get_format_description()."""
1639
 
        return "Working tree format 4"
1640
 
 
1641
 
 
1642
 
class WorkingTreeFormat5(DirStateWorkingTreeFormat):
1643
 
    """WorkingTree format supporting content filtering.
1644
 
    """
1645
 
 
1646
 
    upgrade_recommended = False
1647
 
 
1648
 
    _tree_class = WorkingTree5
1649
 
 
1650
 
    def get_format_string(self):
1651
 
        """See WorkingTreeFormat.get_format_string()."""
1652
 
        return "Bazaar Working Tree Format 5 (bzr 1.11)\n"
1653
 
 
1654
 
    def get_format_description(self):
1655
 
        """See WorkingTreeFormat.get_format_description()."""
1656
 
        return "Working tree format 5"
1657
 
 
1658
 
    def supports_content_filtering(self):
1659
 
        return True
1660
 
 
1661
 
 
1662
 
class WorkingTreeFormat6(DirStateWorkingTreeFormat):
1663
 
    """WorkingTree format supporting views.
1664
 
    """
1665
 
 
1666
 
    upgrade_recommended = False
1667
 
 
1668
 
    _tree_class = WorkingTree6
1669
 
 
1670
 
    def get_format_string(self):
1671
 
        """See WorkingTreeFormat.get_format_string()."""
1672
 
        return "Bazaar Working Tree Format 6 (bzr 1.14)\n"
1673
 
 
1674
 
    def get_format_description(self):
1675
 
        """See WorkingTreeFormat.get_format_description()."""
1676
 
        return "Working tree format 6"
1677
 
 
1678
 
    def _init_custom_control_files(self, wt):
1679
 
        """Subclasses with custom control files should override this method."""
1680
 
        wt._transport.put_bytes('views', '', mode=wt.bzrdir._get_file_mode())
1681
 
 
1682
 
    def supports_content_filtering(self):
1683
 
        return True
1684
 
 
1685
 
    def supports_views(self):
1686
 
        return True
1687
 
 
1688
 
 
1689
 
class DirStateRevisionTree(InventoryTree):
1690
 
    """A revision tree pulling the inventory from a dirstate.
1691
 
    
1692
 
    Note that this is one of the historical (ie revision) trees cached in the
1693
 
    dirstate for easy access, not the workingtree.
1694
 
    """
 
1316
class DirStateRevisionTree(Tree):
 
1317
    """A revision tree pulling the inventory from a dirstate."""
1695
1318
 
1696
1319
    def __init__(self, dirstate, revision_id, repository):
1697
1320
        self._dirstate = dirstate
1698
 
        self._revision_id = revision_id
 
1321
        self._revision_id = osutils.safe_revision_id(revision_id)
1699
1322
        self._repository = repository
1700
1323
        self._inventory = None
1701
1324
        self._locked = 0
1702
1325
        self._dirstate_locked = False
1703
 
        self._repo_supports_tree_reference = getattr(
1704
 
            repository._format, "supports_tree_reference",
1705
 
            False)
1706
1326
 
1707
1327
    def __repr__(self):
1708
1328
        return "<%s of %s in %s>" % \
1709
1329
            (self.__class__.__name__, self._revision_id, self._dirstate)
1710
1330
 
1711
 
    def annotate_iter(self, file_id,
1712
 
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1331
    def annotate_iter(self, file_id):
1713
1332
        """See Tree.annotate_iter"""
1714
 
        text_key = (file_id, self.get_file_revision(file_id))
1715
 
        annotations = self._repository.texts.annotate(text_key)
1716
 
        return [(key[-1], line) for (key, line) in annotations]
 
1333
        w = self._repository.weave_store.get_weave(file_id,
 
1334
                           self._repository.get_transaction())
 
1335
        return w.annotate_iter(self.inventory[file_id].revision)
1717
1336
 
1718
1337
    def _comparison_data(self, entry, path):
1719
1338
        """See Tree._comparison_data."""
1737
1356
    def get_root_id(self):
1738
1357
        return self.path2id('')
1739
1358
 
1740
 
    def id2path(self, file_id):
1741
 
        "Convert a file-id to a path."
1742
 
        entry = self._get_entry(file_id=file_id)
1743
 
        if entry == (None, None):
1744
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1745
 
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
1746
 
        return path_utf8.decode('utf8')
1747
 
 
1748
 
    def iter_references(self):
1749
 
        if not self._repo_supports_tree_reference:
1750
 
            # When the repo doesn't support references, we will have nothing to
1751
 
            # return
1752
 
            return iter([])
1753
 
        # Otherwise, fall back to the default implementation
1754
 
        return super(DirStateRevisionTree, self).iter_references()
1755
 
 
1756
1359
    def _get_parent_index(self):
1757
1360
        """Return the index in the dirstate referenced by this tree."""
1758
1361
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
1763
1366
        If either file_id or path is supplied, it is used as the key to lookup.
1764
1367
        If both are supplied, the fastest lookup is used, and an error is
1765
1368
        raised if they do not both point at the same row.
1766
 
 
 
1369
        
1767
1370
        :param file_id: An optional unicode file_id to be looked up.
1768
1371
        :param path: An optional unicode path to be looked up.
1769
1372
        :return: The dirstate row tuple for path/file_id, or (None, None)
1770
1373
        """
1771
1374
        if file_id is None and path is None:
1772
1375
            raise errors.BzrError('must supply file_id or path')
 
1376
        file_id = osutils.safe_file_id(file_id)
1773
1377
        if path is not None:
1774
1378
            path = path.encode('utf8')
1775
1379
        parent_index = self._get_parent_index()
1783
1387
 
1784
1388
        This is relatively expensive: we have to walk the entire dirstate.
1785
1389
        """
1786
 
        if not self._locked:
1787
 
            raise AssertionError(
1788
 
                'cannot generate inventory of an unlocked '
1789
 
                'dirstate revision tree')
 
1390
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1391
            'dirstate revision tree'
1790
1392
        # separate call for profiling - makes it clear where the costs are.
1791
1393
        self._dirstate._read_dirblocks_if_needed()
1792
 
        if self._revision_id not in self._dirstate.get_parent_ids():
1793
 
            raise AssertionError(
1794
 
                'parent %s has disappeared from %s' % (
1795
 
                self._revision_id, self._dirstate.get_parent_ids()))
 
1394
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1395
            'parent %s has disappeared from %s' % (
 
1396
            self._revision_id, self._dirstate.get_parent_ids())
1796
1397
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
1797
1398
        # This is identical now to the WorkingTree _generate_inventory except
1798
1399
        # for the tree index use.
1799
1400
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
1800
1401
        current_id = root_key[2]
1801
 
        if current_entry[parent_index][0] != 'd':
1802
 
            raise AssertionError()
 
1402
        assert current_entry[parent_index][0] == 'd'
1803
1403
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
1804
1404
        inv.root.revision = current_entry[parent_index][4]
1805
1405
        # Turn some things into local variables
1836
1436
                elif kind == 'directory':
1837
1437
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
1838
1438
                elif kind == 'symlink':
 
1439
                    inv_entry.executable = False
 
1440
                    inv_entry.text_size = None
1839
1441
                    inv_entry.symlink_target = utf8_decode(fingerprint)[0]
1840
1442
                elif kind == 'tree-reference':
1841
1443
                    inv_entry.reference_revision = fingerprint or None
1843
1445
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
1844
1446
                            % entry)
1845
1447
                # These checks cost us around 40ms on a 55k entry tree
1846
 
                if file_id in inv_byid:
1847
 
                    raise AssertionError('file_id %s already in'
1848
 
                        ' inventory as %s' % (file_id, inv_byid[file_id]))
1849
 
                if name_unicode in parent_ie.children:
1850
 
                    raise AssertionError('name %r already in parent'
1851
 
                        % (name_unicode,))
 
1448
                assert file_id not in inv_byid
 
1449
                assert name_unicode not in parent_ie.children
1852
1450
                inv_byid[file_id] = inv_entry
1853
1451
                parent_ie.children[name_unicode] = inv_entry
1854
1452
        self._inventory = inv
1864
1462
            return None
1865
1463
        parent_index = self._get_parent_index()
1866
1464
        last_changed_revision = entry[1][parent_index][4]
1867
 
        try:
1868
 
            rev = self._repository.get_revision(last_changed_revision)
1869
 
        except errors.NoSuchRevision:
1870
 
            raise errors.FileTimestampUnavailable(self.id2path(file_id))
1871
 
        return rev.timestamp
 
1465
        return self._repository.get_revision(last_changed_revision).timestamp
1872
1466
 
1873
1467
    def get_file_sha1(self, file_id, path=None, stat_value=None):
1874
1468
        entry = self._get_entry(file_id=file_id, path=path)
1878
1472
            return parent_details[1]
1879
1473
        return None
1880
1474
 
1881
 
    @needs_read_lock
1882
 
    def get_file_revision(self, file_id):
1883
 
        return self.inventory[file_id].revision
1884
 
 
1885
 
    def get_file(self, file_id, path=None):
 
1475
    def get_file(self, file_id):
1886
1476
        return StringIO(self.get_file_text(file_id))
1887
1477
 
 
1478
    def get_file_lines(self, file_id):
 
1479
        ie = self.inventory[file_id]
 
1480
        return self._repository.weave_store.get_weave(file_id,
 
1481
                self._repository.get_transaction()).get_lines(ie.revision)
 
1482
 
1888
1483
    def get_file_size(self, file_id):
1889
 
        """See Tree.get_file_size"""
1890
1484
        return self.inventory[file_id].text_size
1891
1485
 
1892
 
    def get_file_text(self, file_id, path=None):
1893
 
        _, content = list(self.iter_files_bytes([(file_id, None)]))[0]
1894
 
        return ''.join(content)
 
1486
    def get_file_text(self, file_id):
 
1487
        return ''.join(self.get_file_lines(file_id))
1895
1488
 
1896
1489
    def get_reference_revision(self, file_id, path=None):
1897
1490
        return self.inventory[file_id].reference_revision
1898
1491
 
1899
 
    def iter_files_bytes(self, desired_files):
1900
 
        """See Tree.iter_files_bytes.
1901
 
 
1902
 
        This version is implemented on top of Repository.iter_files_bytes"""
1903
 
        parent_index = self._get_parent_index()
1904
 
        repo_desired_files = []
1905
 
        for file_id, identifier in desired_files:
1906
 
            entry = self._get_entry(file_id)
1907
 
            if entry == (None, None):
1908
 
                raise errors.NoSuchId(self, file_id)
1909
 
            repo_desired_files.append((file_id, entry[1][parent_index][4],
1910
 
                                       identifier))
1911
 
        return self._repository.iter_files_bytes(repo_desired_files)
1912
 
 
1913
 
    def get_symlink_target(self, file_id, path=None):
 
1492
    def get_symlink_target(self, file_id):
1914
1493
        entry = self._get_entry(file_id=file_id)
1915
1494
        parent_index = self._get_parent_index()
1916
1495
        if entry[1][parent_index][0] != 'l':
1917
1496
            return None
1918
1497
        else:
1919
 
            target = entry[1][parent_index][1]
1920
 
            target = target.decode('utf8')
1921
 
            return target
 
1498
            # At present, none of the tree implementations supports non-ascii
 
1499
            # symlink targets. So we will just assume that the dirstate path is
 
1500
            # correct.
 
1501
            return entry[1][parent_index][1]
1922
1502
 
1923
1503
    def get_revision_id(self):
1924
1504
        """Return the revision id for this tree."""
1942
1522
        return bool(self.path2id(filename))
1943
1523
 
1944
1524
    def kind(self, file_id):
1945
 
        entry = self._get_entry(file_id=file_id)[1]
1946
 
        if entry is None:
1947
 
            raise errors.NoSuchId(tree=self, file_id=file_id)
1948
 
        parent_index = self._get_parent_index()
1949
 
        return dirstate.DirState._minikind_to_kind[entry[parent_index][0]]
1950
 
 
1951
 
    def stored_kind(self, file_id):
1952
 
        """See Tree.stored_kind"""
1953
 
        return self.kind(file_id)
1954
 
 
1955
 
    def path_content_summary(self, path):
1956
 
        """See Tree.path_content_summary."""
1957
 
        id = self.inventory.path2id(path)
1958
 
        if id is None:
1959
 
            return ('missing', None, None, None)
1960
 
        entry = self._inventory[id]
1961
 
        kind = entry.kind
1962
 
        if kind == 'file':
1963
 
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
1964
 
        elif kind == 'symlink':
1965
 
            return (kind, None, None, entry.symlink_target)
1966
 
        else:
1967
 
            return (kind, None, None, None)
 
1525
        return self.inventory[file_id].kind
1968
1526
 
1969
1527
    def is_executable(self, file_id, path=None):
1970
1528
        ie = self.inventory[file_id]
1971
1529
        if ie.kind != "file":
1972
 
            return False
 
1530
            return None
1973
1531
        return ie.executable
1974
1532
 
1975
 
    def is_locked(self):
1976
 
        return self._locked
1977
 
 
1978
 
    def list_files(self, include_root=False, from_dir=None, recursive=True):
 
1533
    def list_files(self, include_root=False):
1979
1534
        # We use a standard implementation, because DirStateRevisionTree is
1980
1535
        # dealing with one of the parents of the current state
1981
1536
        inv = self._get_inventory()
1982
 
        if from_dir is None:
1983
 
            from_dir_id = None
1984
 
        else:
1985
 
            from_dir_id = inv.path2id(from_dir)
1986
 
            if from_dir_id is None:
1987
 
                # Directory not versioned
1988
 
                return
1989
 
        entries = inv.iter_entries(from_dir=from_dir_id, recursive=recursive)
1990
 
        if inv.root is not None and not include_root and from_dir is None:
 
1537
        entries = inv.iter_entries()
 
1538
        if self.inventory.root is not None and not include_root:
1991
1539
            entries.next()
1992
1540
        for path, entry in entries:
1993
1541
            yield path, 'V', entry.kind, entry.file_id, entry
1994
1542
 
1995
1543
    def lock_read(self):
1996
 
        """Lock the tree for a set of operations.
1997
 
 
1998
 
        :return: A bzrlib.lock.LogicalLockResult.
1999
 
        """
 
1544
        """Lock the tree for a set of operations."""
2000
1545
        if not self._locked:
2001
1546
            self._repository.lock_read()
2002
1547
            if self._dirstate._lock_token is None:
2003
1548
                self._dirstate.lock_read()
2004
1549
                self._dirstate_locked = True
2005
1550
        self._locked += 1
2006
 
        return LogicalLockResult(self.unlock)
2007
1551
 
2008
1552
    def _must_be_locked(self):
2009
1553
        if not self._locked:
2030
1574
                self._dirstate_locked = False
2031
1575
            self._repository.unlock()
2032
1576
 
2033
 
    @needs_read_lock
2034
 
    def supports_tree_reference(self):
2035
 
        return self._repo_supports_tree_reference
2036
 
 
2037
1577
    def walkdirs(self, prefix=""):
2038
1578
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
2039
 
        # implementation based on an inventory.
 
1579
        # implementation based on an inventory.  
2040
1580
        # This should be cleaned up to use the much faster Dirstate code
2041
1581
        # So for now, we just build up the parent inventory, and extract
2042
1582
        # it the same way RevisionTree does.
2071
1611
 
2072
1612
class InterDirStateTree(InterTree):
2073
1613
    """Fast path optimiser for changes_from with dirstate trees.
2074
 
 
2075
 
    This is used only when both trees are in the dirstate working file, and
2076
 
    the source is any parent within the dirstate, and the destination is
 
1614
    
 
1615
    This is used only when both trees are in the dirstate working file, and 
 
1616
    the source is any parent within the dirstate, and the destination is 
2077
1617
    the current working tree of the same dirstate.
2078
1618
    """
2079
1619
    # this could be generalized to allow comparisons between any trees in the
2088
1628
    def make_source_parent_tree(source, target):
2089
1629
        """Change the source tree into a parent of the target."""
2090
1630
        revid = source.commit('record tree')
2091
 
        target.branch.fetch(source.branch, revid)
 
1631
        target.branch.repository.fetch(source.branch.repository, revid)
2092
1632
        target.set_parent_ids([revid])
2093
1633
        return target.basis_tree(), target
2094
1634
 
2095
 
    @classmethod
2096
 
    def make_source_parent_tree_python_dirstate(klass, test_case, source, target):
2097
 
        result = klass.make_source_parent_tree(source, target)
2098
 
        result[1]._iter_changes = dirstate.ProcessEntryPython
2099
 
        return result
2100
 
 
2101
 
    @classmethod
2102
 
    def make_source_parent_tree_compiled_dirstate(klass, test_case, source,
2103
 
                                                  target):
2104
 
        from bzrlib.tests.test__dirstate_helpers import \
2105
 
            compiled_dirstate_helpers_feature
2106
 
        test_case.requireFeature(compiled_dirstate_helpers_feature)
2107
 
        from bzrlib._dirstate_helpers_pyx import ProcessEntryC
2108
 
        result = klass.make_source_parent_tree(source, target)
2109
 
        result[1]._iter_changes = ProcessEntryC
2110
 
        return result
2111
 
 
2112
1635
    _matching_from_tree_format = WorkingTreeFormat4()
2113
1636
    _matching_to_tree_format = WorkingTreeFormat4()
2114
 
 
2115
 
    @classmethod
2116
 
    def _test_mutable_trees_to_test_trees(klass, test_case, source, target):
2117
 
        # This method shouldn't be called, because we have python and C
2118
 
        # specific flavours.
2119
 
        raise NotImplementedError
2120
 
 
2121
 
    def iter_changes(self, include_unchanged=False,
 
1637
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1638
 
 
1639
    def _iter_changes(self, include_unchanged=False,
2122
1640
                      specific_files=None, pb=None, extra_trees=[],
2123
1641
                      require_versioned=True, want_unversioned=False):
2124
1642
        """Return the changes from source to target.
2125
1643
 
2126
 
        :return: An iterator that yields tuples. See InterTree.iter_changes
 
1644
        :return: An iterator that yields tuples. See InterTree._iter_changes
2127
1645
            for details.
2128
1646
        :param specific_files: An optional list of file paths to restrict the
2129
1647
            comparison to. When mapping filenames to ids, all matches in all
2140
1658
            output. An unversioned file is defined as one with (False, False)
2141
1659
            for the versioned pair.
2142
1660
        """
 
1661
        utf8_decode_or_none = cache_utf8._utf8_decode_with_None
 
1662
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1663
        # NB: show_status depends on being able to pass in non-versioned files
 
1664
        # and report them as unknown
2143
1665
        # TODO: handle extra trees in the dirstate.
2144
 
        if (extra_trees or specific_files == []):
 
1666
        # TODO: handle comparisons as an empty tree as a different special
 
1667
        # case? mbp 20070226
 
1668
        if extra_trees or (self.source._revision_id == NULL_REVISION):
2145
1669
            # we can't fast-path these cases (yet)
2146
 
            return super(InterDirStateTree, self).iter_changes(
 
1670
            for f in super(InterDirStateTree, self)._iter_changes(
2147
1671
                include_unchanged, specific_files, pb, extra_trees,
2148
 
                require_versioned, want_unversioned=want_unversioned)
 
1672
                require_versioned, want_unversioned=want_unversioned):
 
1673
                yield f
 
1674
            return
2149
1675
        parent_ids = self.target.get_parent_ids()
2150
 
        if not (self.source._revision_id in parent_ids
2151
 
                or self.source._revision_id == _mod_revision.NULL_REVISION):
2152
 
            raise AssertionError(
2153
 
                "revision {%s} is not stored in {%s}, but %s "
2154
 
                "can only be used for trees stored in the dirstate"
2155
 
                % (self.source._revision_id, self.target, self.iter_changes))
 
1676
        assert (self.source._revision_id in parent_ids), \
 
1677
                "revision {%s} is not stored in {%s}, but %s " \
 
1678
                "can only be used for trees stored in the dirstate" \
 
1679
                % (self.source._revision_id, self.target, self._iter_changes)
2156
1680
        target_index = 0
2157
 
        if self.source._revision_id == _mod_revision.NULL_REVISION:
 
1681
        if self.source._revision_id == NULL_REVISION:
2158
1682
            source_index = None
2159
1683
            indices = (target_index,)
2160
1684
        else:
2161
 
            if not (self.source._revision_id in parent_ids):
2162
 
                raise AssertionError(
2163
 
                    "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
2164
 
                    self.source._revision_id, parent_ids))
 
1685
            assert (self.source._revision_id in parent_ids), \
 
1686
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1687
                self.source._revision_id, parent_ids)
2165
1688
            source_index = 1 + parent_ids.index(self.source._revision_id)
2166
 
            indices = (source_index, target_index)
 
1689
            indices = (source_index,target_index)
2167
1690
        # -- make all specific_files utf8 --
2168
1691
        if specific_files:
2169
1692
            specific_files_utf8 = set()
2170
1693
            for path in specific_files:
2171
 
                # Note, if there are many specific files, using cache_utf8
2172
 
                # would be good here.
2173
1694
                specific_files_utf8.add(path.encode('utf8'))
2174
1695
            specific_files = specific_files_utf8
2175
1696
        else:
2176
1697
            specific_files = set([''])
2177
1698
        # -- specific_files is now a utf8 path set --
2178
 
 
2179
1699
        # -- get the state object and prepare it.
2180
1700
        state = self.target.current_dirstate()
2181
1701
        state._read_dirblocks_if_needed()
 
1702
        def _entries_for_path(path):
 
1703
            """Return a list with all the entries that match path for all ids.
 
1704
            """
 
1705
            dirname, basename = os.path.split(path)
 
1706
            key = (dirname, basename, '')
 
1707
            block_index, present = state._find_block_index_from_key(key)
 
1708
            if not present:
 
1709
                # the block which should contain path is absent.
 
1710
                return []
 
1711
            result = []
 
1712
            block = state._dirblocks[block_index][1]
 
1713
            entry_index, _ = state._find_entry_index(key, block)
 
1714
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1715
            while (entry_index < len(block) and
 
1716
                block[entry_index][0][0:2] == key[0:2]):
 
1717
                result.append(block[entry_index])
 
1718
                entry_index += 1
 
1719
            return result
2182
1720
        if require_versioned:
2183
1721
            # -- check all supplied paths are versioned in a search tree. --
2184
 
            not_versioned = []
 
1722
            all_versioned = True
2185
1723
            for path in specific_files:
2186
 
                path_entries = state._entries_for_path(path)
 
1724
                path_entries = _entries_for_path(path)
2187
1725
                if not path_entries:
2188
1726
                    # this specified path is not present at all: error
2189
 
                    not_versioned.append(path)
2190
 
                    continue
 
1727
                    all_versioned = False
 
1728
                    break
2191
1729
                found_versioned = False
2192
1730
                # for each id at this path
2193
1731
                for entry in path_entries:
2200
1738
                if not found_versioned:
2201
1739
                    # none of the indexes was not 'absent' at all ids for this
2202
1740
                    # path.
2203
 
                    not_versioned.append(path)
2204
 
            if len(not_versioned) > 0:
2205
 
                raise errors.PathsNotVersionedError(not_versioned)
 
1741
                    all_versioned = False
 
1742
                    break
 
1743
            if not all_versioned:
 
1744
                raise errors.PathsNotVersionedError(specific_files)
2206
1745
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
2207
 
        search_specific_files = osutils.minimum_path_selection(specific_files)
 
1746
        search_specific_files = set()
 
1747
        for path in specific_files:
 
1748
            other_specific_files = specific_files.difference(set([path]))
 
1749
            if not osutils.is_inside_any(other_specific_files, path):
 
1750
                # this is a top level path, we must check it.
 
1751
                search_specific_files.add(path)
 
1752
        # sketch: 
 
1753
        # compare source_index and target_index at or under each element of search_specific_files.
 
1754
        # follow the following comparison table. Note that we only want to do diff operations when
 
1755
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1756
        # for the target.
 
1757
        # cases:
 
1758
        # 
 
1759
        # Source | Target | disk | action
 
1760
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1761
        #        |        |      | diff check on source-target
 
1762
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1763
        #        |        |      | ???
 
1764
        #   r    |  a     |      | add source to search
 
1765
        #   r    |  a     |  a   | 
 
1766
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1767
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1768
        #   a    | fdlt   |      | add new id
 
1769
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1770
        #   a    |  a     |      | not present in either tree, skip
 
1771
        #   a    |  a     |  a   | not present in any tree, skip
 
1772
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1773
        #        |        |      | may not be selected by the users list of paths.
 
1774
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1775
        #        |        |      | may not be selected by the users list of paths.
 
1776
        #  fdlt  | fdlt   |      | content in both: diff them
 
1777
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1778
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1779
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1780
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1781
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1782
        #        |        |      | this id at the other path.
 
1783
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1784
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1785
        #        |        |      | this id at the other path.
 
1786
 
 
1787
        # for all search_indexs in each path at or under each element of
 
1788
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1789
        # relocated path as one to search if its not searched already. If the
 
1790
        # detail is not relocated, add the id.
 
1791
        searched_specific_files = set()
 
1792
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1793
        # Using a list so that we can access the values and change them in
 
1794
        # nested scope. Each one is [path, file_id, entry]
 
1795
        last_source_parent = [None, None, None]
 
1796
        last_target_parent = [None, None, None]
2208
1797
 
2209
1798
        use_filesystem_for_exec = (sys.platform != 'win32')
2210
 
        iter_changes = self.target._iter_changes(include_unchanged,
2211
 
            use_filesystem_for_exec, search_specific_files, state,
2212
 
            source_index, target_index, want_unversioned, self.target)
2213
 
        return iter_changes.iter_changes()
 
1799
 
 
1800
        def _process_entry(entry, path_info):
 
1801
            """Compare an entry and real disk to generate delta information.
 
1802
 
 
1803
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1804
                the path of entry. If None, then the path is considered absent.
 
1805
                (Perhaps we should pass in a concrete entry for this ?)
 
1806
                Basename is returned as a utf8 string because we expect this
 
1807
                tuple will be ignored, and don't want to take the time to
 
1808
                decode.
 
1809
            """
 
1810
            if source_index is None:
 
1811
                source_details = NULL_PARENT_DETAILS
 
1812
            else:
 
1813
                source_details = entry[1][source_index]
 
1814
            target_details = entry[1][target_index]
 
1815
            target_minikind = target_details[0]
 
1816
            if path_info is not None and target_minikind in 'fdlt':
 
1817
                assert target_index == 0
 
1818
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1819
                                                  stat_value=path_info[3])
 
1820
                # The entry may have been modified by update_entry
 
1821
                target_details = entry[1][target_index]
 
1822
                target_minikind = target_details[0]
 
1823
            else:
 
1824
                link_or_sha1 = None
 
1825
            source_minikind = source_details[0]
 
1826
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1827
                # claimed content in both: diff
 
1828
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1829
                #        |        |      | diff check on source-target
 
1830
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1831
                #        |        |      | ???
 
1832
                if source_minikind in 'r':
 
1833
                    # add the source to the search path to find any children it
 
1834
                    # has.  TODO ? : only add if it is a container ?
 
1835
                    if not osutils.is_inside_any(searched_specific_files,
 
1836
                                                 source_details[1]):
 
1837
                        search_specific_files.add(source_details[1])
 
1838
                    # generate the old path; this is needed for stating later
 
1839
                    # as well.
 
1840
                    old_path = source_details[1]
 
1841
                    old_dirname, old_basename = os.path.split(old_path)
 
1842
                    path = pathjoin(entry[0][0], entry[0][1])
 
1843
                    old_entry = state._get_entry(source_index,
 
1844
                                                 path_utf8=old_path)
 
1845
                    # update the source details variable to be the real
 
1846
                    # location.
 
1847
                    source_details = old_entry[1][source_index]
 
1848
                    source_minikind = source_details[0]
 
1849
                else:
 
1850
                    old_dirname = entry[0][0]
 
1851
                    old_basename = entry[0][1]
 
1852
                    old_path = path = pathjoin(old_dirname, old_basename)
 
1853
                if path_info is None:
 
1854
                    # the file is missing on disk, show as removed.
 
1855
                    content_change = True
 
1856
                    target_kind = None
 
1857
                    target_exec = False
 
1858
                else:
 
1859
                    # source and target are both versioned and disk file is present.
 
1860
                    target_kind = path_info[2]
 
1861
                    if target_kind == 'directory':
 
1862
                        if source_minikind != 'd':
 
1863
                            content_change = True
 
1864
                        else:
 
1865
                            # directories have no fingerprint
 
1866
                            content_change = False
 
1867
                        target_exec = False
 
1868
                    elif target_kind == 'file':
 
1869
                        if source_minikind != 'f':
 
1870
                            content_change = True
 
1871
                        else:
 
1872
                            # We could check the size, but we already have the
 
1873
                            # sha1 hash.
 
1874
                            content_change = (link_or_sha1 != source_details[1])
 
1875
                        # Target details is updated at update_entry time
 
1876
                        if use_filesystem_for_exec:
 
1877
                            # We don't need S_ISREG here, because we are sure
 
1878
                            # we are dealing with a file.
 
1879
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1880
                        else:
 
1881
                            target_exec = target_details[3]
 
1882
                    elif target_kind == 'symlink':
 
1883
                        if source_minikind != 'l':
 
1884
                            content_change = True
 
1885
                        else:
 
1886
                            content_change = (link_or_sha1 != source_details[1])
 
1887
                        target_exec = False
 
1888
                    elif target_kind == 'tree-reference':
 
1889
                        if source_minikind != 't':
 
1890
                            content_change = True
 
1891
                        else:
 
1892
                            content_change = False
 
1893
                        target_exec = False
 
1894
                    else:
 
1895
                        raise Exception, "unknown kind %s" % path_info[2]
 
1896
                # parent id is the entry for the path in the target tree
 
1897
                if old_dirname == last_source_parent[0]:
 
1898
                    source_parent_id = last_source_parent[1]
 
1899
                else:
 
1900
                    source_parent_entry = state._get_entry(source_index,
 
1901
                                                           path_utf8=old_dirname)
 
1902
                    source_parent_id = source_parent_entry[0][2]
 
1903
                    if source_parent_id == entry[0][2]:
 
1904
                        # This is the root, so the parent is None
 
1905
                        source_parent_id = None
 
1906
                    else:
 
1907
                        last_source_parent[0] = old_dirname
 
1908
                        last_source_parent[1] = source_parent_id
 
1909
                        last_source_parent[2] = source_parent_entry
 
1910
                new_dirname = entry[0][0]
 
1911
                if new_dirname == last_target_parent[0]:
 
1912
                    target_parent_id = last_target_parent[1]
 
1913
                else:
 
1914
                    # TODO: We don't always need to do the lookup, because the
 
1915
                    #       parent entry will be the same as the source entry.
 
1916
                    target_parent_entry = state._get_entry(target_index,
 
1917
                                                           path_utf8=new_dirname)
 
1918
                    target_parent_id = target_parent_entry[0][2]
 
1919
                    if target_parent_id == entry[0][2]:
 
1920
                        # This is the root, so the parent is None
 
1921
                        target_parent_id = None
 
1922
                    else:
 
1923
                        last_target_parent[0] = new_dirname
 
1924
                        last_target_parent[1] = target_parent_id
 
1925
                        last_target_parent[2] = target_parent_entry
 
1926
 
 
1927
                source_exec = source_details[3]
 
1928
                return ((entry[0][2], (old_path, path), content_change,
 
1929
                        (True, True),
 
1930
                        (source_parent_id, target_parent_id),
 
1931
                        (old_basename, entry[0][1]),
 
1932
                        (_minikind_to_kind[source_minikind], target_kind),
 
1933
                        (source_exec, target_exec)),)
 
1934
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
1935
                # looks like a new file
 
1936
                if path_info is not None:
 
1937
                    path = pathjoin(entry[0][0], entry[0][1])
 
1938
                    # parent id is the entry for the path in the target tree
 
1939
                    # TODO: these are the same for an entire directory: cache em.
 
1940
                    parent_id = state._get_entry(target_index,
 
1941
                                                 path_utf8=entry[0][0])[0][2]
 
1942
                    if parent_id == entry[0][2]:
 
1943
                        parent_id = None
 
1944
                    if use_filesystem_for_exec:
 
1945
                        # We need S_ISREG here, because we aren't sure if this
 
1946
                        # is a file or not.
 
1947
                        target_exec = bool(
 
1948
                            stat.S_ISREG(path_info[3].st_mode)
 
1949
                            and stat.S_IEXEC & path_info[3].st_mode)
 
1950
                    else:
 
1951
                        target_exec = target_details[3]
 
1952
                    return ((entry[0][2], (None, path), True,
 
1953
                            (False, True),
 
1954
                            (None, parent_id),
 
1955
                            (None, entry[0][1]),
 
1956
                            (None, path_info[2]),
 
1957
                            (None, target_exec)),)
 
1958
                else:
 
1959
                    # but its not on disk: we deliberately treat this as just
 
1960
                    # never-present. (Why ?! - RBC 20070224)
 
1961
                    pass
 
1962
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
1963
                # unversioned, possibly, or possibly not deleted: we dont care.
 
1964
                # if its still on disk, *and* theres no other entry at this
 
1965
                # path [we dont know this in this routine at the moment -
 
1966
                # perhaps we should change this - then it would be an unknown.
 
1967
                old_path = pathjoin(entry[0][0], entry[0][1])
 
1968
                # parent id is the entry for the path in the target tree
 
1969
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
1970
                if parent_id == entry[0][2]:
 
1971
                    parent_id = None
 
1972
                return ((entry[0][2], (old_path, None), True,
 
1973
                        (True, False),
 
1974
                        (parent_id, None),
 
1975
                        (entry[0][1], None),
 
1976
                        (_minikind_to_kind[source_minikind], None),
 
1977
                        (source_details[3], None)),)
 
1978
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
1979
                # a rename; could be a true rename, or a rename inherited from
 
1980
                # a renamed parent. TODO: handle this efficiently. Its not
 
1981
                # common case to rename dirs though, so a correct but slow
 
1982
                # implementation will do.
 
1983
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
1984
                    search_specific_files.add(target_details[1])
 
1985
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
1986
                # neither of the selected trees contain this file,
 
1987
                # so skip over it. This is not currently directly tested, but
 
1988
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
1989
                pass
 
1990
            else:
 
1991
                raise AssertionError("don't know how to compare "
 
1992
                    "source_minikind=%r, target_minikind=%r"
 
1993
                    % (source_minikind, target_minikind))
 
1994
                ## import pdb;pdb.set_trace()
 
1995
            return ()
 
1996
 
 
1997
        while search_specific_files:
 
1998
            # TODO: the pending list should be lexically sorted?  the
 
1999
            # interface doesn't require it.
 
2000
            current_root = search_specific_files.pop()
 
2001
            current_root_unicode = current_root.decode('utf8')
 
2002
            searched_specific_files.add(current_root)
 
2003
            # process the entries for this containing directory: the rest will be
 
2004
            # found by their parents recursively.
 
2005
            root_entries = _entries_for_path(current_root)
 
2006
            root_abspath = self.target.abspath(current_root_unicode)
 
2007
            try:
 
2008
                root_stat = os.lstat(root_abspath)
 
2009
            except OSError, e:
 
2010
                if e.errno == errno.ENOENT:
 
2011
                    # the path does not exist: let _process_entry know that.
 
2012
                    root_dir_info = None
 
2013
                else:
 
2014
                    # some other random error: hand it up.
 
2015
                    raise
 
2016
            else:
 
2017
                root_dir_info = ('', current_root,
 
2018
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2019
                    root_abspath)
 
2020
                if root_dir_info[2] == 'directory':
 
2021
                    if self.target._directory_is_tree_reference(
 
2022
                        current_root.decode('utf8')):
 
2023
                        root_dir_info = root_dir_info[:2] + \
 
2024
                            ('tree-reference',) + root_dir_info[3:]
 
2025
 
 
2026
            if not root_entries and not root_dir_info:
 
2027
                # this specified path is not present at all, skip it.
 
2028
                continue
 
2029
            path_handled = False
 
2030
            for entry in root_entries:
 
2031
                for result in _process_entry(entry, root_dir_info):
 
2032
                    # this check should probably be outside the loop: one
 
2033
                    # 'iterate two trees' api, and then _iter_changes filters
 
2034
                    # unchanged pairs. - RBC 20070226
 
2035
                    path_handled = True
 
2036
                    if (include_unchanged
 
2037
                        or result[2]                    # content change
 
2038
                        or result[3][0] != result[3][1] # versioned status
 
2039
                        or result[4][0] != result[4][1] # parent id
 
2040
                        or result[5][0] != result[5][1] # name
 
2041
                        or result[6][0] != result[6][1] # kind
 
2042
                        or result[7][0] != result[7][1] # executable
 
2043
                        ):
 
2044
                        yield (result[0],
 
2045
                               (utf8_decode_or_none(result[1][0]),
 
2046
                                utf8_decode_or_none(result[1][1])),
 
2047
                               result[2],
 
2048
                               result[3],
 
2049
                               result[4],
 
2050
                               (utf8_decode_or_none(result[5][0]),
 
2051
                                utf8_decode_or_none(result[5][1])),
 
2052
                               result[6],
 
2053
                               result[7],
 
2054
                              )
 
2055
            if want_unversioned and not path_handled and root_dir_info:
 
2056
                new_executable = bool(
 
2057
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2058
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2059
                yield (None,
 
2060
                       (None, current_root_unicode),
 
2061
                       True,
 
2062
                       (False, False),
 
2063
                       (None, None),
 
2064
                       (None, splitpath(current_root_unicode)[-1]),
 
2065
                       (None, root_dir_info[2]),
 
2066
                       (None, new_executable)
 
2067
                      )
 
2068
            initial_key = (current_root, '', '')
 
2069
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2070
            if block_index == 0:
 
2071
                # we have processed the total root already, but because the
 
2072
                # initial key matched it we should skip it here.
 
2073
                block_index +=1
 
2074
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2075
                current_dir_info = None
 
2076
            else:
 
2077
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2078
                try:
 
2079
                    current_dir_info = dir_iterator.next()
 
2080
                except OSError, e:
 
2081
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2082
                    # python 2.5 has e.errno == EINVAL,
 
2083
                    #            and e.winerror == ERROR_DIRECTORY
 
2084
                    e_winerror = getattr(e, 'winerror', None)
 
2085
                    # there may be directories in the inventory even though
 
2086
                    # this path is not a file on disk: so mark it as end of
 
2087
                    # iterator
 
2088
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2089
                        current_dir_info = None
 
2090
                    elif (sys.platform == 'win32'
 
2091
                          and ERROR_DIRECTORY in (e.errno, e_winerror)):
 
2092
                        current_dir_info = None
 
2093
                    else:
 
2094
                        raise
 
2095
                else:
 
2096
                    if current_dir_info[0][0] == '':
 
2097
                        # remove .bzr from iteration
 
2098
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2099
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2100
                        del current_dir_info[1][bzr_index]
 
2101
            # walk until both the directory listing and the versioned metadata
 
2102
            # are exhausted. 
 
2103
            if (block_index < len(state._dirblocks) and
 
2104
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2105
                current_block = state._dirblocks[block_index]
 
2106
            else:
 
2107
                current_block = None
 
2108
            while (current_dir_info is not None or
 
2109
                   current_block is not None):
 
2110
                if (current_dir_info and current_block
 
2111
                    and current_dir_info[0][0] != current_block[0]):
 
2112
                    if current_dir_info[0][0] < current_block[0] :
 
2113
                        # filesystem data refers to paths not covered by the dirblock.
 
2114
                        # this has two possibilities:
 
2115
                        # A) it is versioned but empty, so there is no block for it
 
2116
                        # B) it is not versioned.
 
2117
                        # in either case it was processed by the containing directories walk:
 
2118
                        # if it is root/foo, when we walked root we emitted it,
 
2119
                        # or if we ere given root/foo to walk specifically, we
 
2120
                        # emitted it when checking the walk-root entries
 
2121
                        # advance the iterator and loop - we dont need to emit it.
 
2122
                        try:
 
2123
                            current_dir_info = dir_iterator.next()
 
2124
                        except StopIteration:
 
2125
                            current_dir_info = None
 
2126
                    else:
 
2127
                        # We have a dirblock entry for this location, but there
 
2128
                        # is no filesystem path for this. This is most likely
 
2129
                        # because a directory was removed from the disk.
 
2130
                        # We don't have to report the missing directory,
 
2131
                        # because that should have already been handled, but we
 
2132
                        # need to handle all of the files that are contained
 
2133
                        # within.
 
2134
                        for current_entry in current_block[1]:
 
2135
                            # entry referring to file not present on disk.
 
2136
                            # advance the entry only, after processing.
 
2137
                            for result in _process_entry(current_entry, None):
 
2138
                                # this check should probably be outside the loop: one
 
2139
                                # 'iterate two trees' api, and then _iter_changes filters
 
2140
                                # unchanged pairs. - RBC 20070226
 
2141
                                if (include_unchanged
 
2142
                                    or result[2]                    # content change
 
2143
                                    or result[3][0] != result[3][1] # versioned status
 
2144
                                    or result[4][0] != result[4][1] # parent id
 
2145
                                    or result[5][0] != result[5][1] # name
 
2146
                                    or result[6][0] != result[6][1] # kind
 
2147
                                    or result[7][0] != result[7][1] # executable
 
2148
                                    ):
 
2149
                                    yield (result[0],
 
2150
                                           (utf8_decode_or_none(result[1][0]),
 
2151
                                            utf8_decode_or_none(result[1][1])),
 
2152
                                           result[2],
 
2153
                                           result[3],
 
2154
                                           result[4],
 
2155
                                           (utf8_decode_or_none(result[5][0]),
 
2156
                                            utf8_decode_or_none(result[5][1])),
 
2157
                                           result[6],
 
2158
                                           result[7],
 
2159
                                          )
 
2160
                        block_index +=1
 
2161
                        if (block_index < len(state._dirblocks) and
 
2162
                            osutils.is_inside(current_root,
 
2163
                                              state._dirblocks[block_index][0])):
 
2164
                            current_block = state._dirblocks[block_index]
 
2165
                        else:
 
2166
                            current_block = None
 
2167
                    continue
 
2168
                entry_index = 0
 
2169
                if current_block and entry_index < len(current_block[1]):
 
2170
                    current_entry = current_block[1][entry_index]
 
2171
                else:
 
2172
                    current_entry = None
 
2173
                advance_entry = True
 
2174
                path_index = 0
 
2175
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2176
                    current_path_info = current_dir_info[1][path_index]
 
2177
                    if current_path_info[2] == 'directory':
 
2178
                        if self.target._directory_is_tree_reference(
 
2179
                            current_path_info[0].decode('utf8')):
 
2180
                            current_path_info = current_path_info[:2] + \
 
2181
                                ('tree-reference',) + current_path_info[3:]
 
2182
                else:
 
2183
                    current_path_info = None
 
2184
                advance_path = True
 
2185
                path_handled = False
 
2186
                while (current_entry is not None or
 
2187
                    current_path_info is not None):
 
2188
                    if current_entry is None:
 
2189
                        # the check for path_handled when the path is adnvaced
 
2190
                        # will yield this path if needed.
 
2191
                        pass
 
2192
                    elif current_path_info is None:
 
2193
                        # no path is fine: the per entry code will handle it.
 
2194
                        for result in _process_entry(current_entry, current_path_info):
 
2195
                            # this check should probably be outside the loop: one
 
2196
                            # 'iterate two trees' api, and then _iter_changes filters
 
2197
                            # unchanged pairs. - RBC 20070226
 
2198
                            if (include_unchanged
 
2199
                                or result[2]                    # content change
 
2200
                                or result[3][0] != result[3][1] # versioned status
 
2201
                                or result[4][0] != result[4][1] # parent id
 
2202
                                or result[5][0] != result[5][1] # name
 
2203
                                or result[6][0] != result[6][1] # kind
 
2204
                                or result[7][0] != result[7][1] # executable
 
2205
                                ):
 
2206
                                yield (result[0],
 
2207
                                       (utf8_decode_or_none(result[1][0]),
 
2208
                                        utf8_decode_or_none(result[1][1])),
 
2209
                                       result[2],
 
2210
                                       result[3],
 
2211
                                       result[4],
 
2212
                                       (utf8_decode_or_none(result[5][0]),
 
2213
                                        utf8_decode_or_none(result[5][1])),
 
2214
                                       result[6],
 
2215
                                       result[7],
 
2216
                                      )
 
2217
                    elif current_entry[0][1] != current_path_info[1]:
 
2218
                        if current_path_info[1] < current_entry[0][1]:
 
2219
                            # extra file on disk: pass for now, but only
 
2220
                            # increment the path, not the entry
 
2221
                            advance_entry = False
 
2222
                        else:
 
2223
                            # entry referring to file not present on disk.
 
2224
                            # advance the entry only, after processing.
 
2225
                            for result in _process_entry(current_entry, None):
 
2226
                                # this check should probably be outside the loop: one
 
2227
                                # 'iterate two trees' api, and then _iter_changes filters
 
2228
                                # unchanged pairs. - RBC 20070226
 
2229
                                path_handled = True
 
2230
                                if (include_unchanged
 
2231
                                    or result[2]                    # content change
 
2232
                                    or result[3][0] != result[3][1] # versioned status
 
2233
                                    or result[4][0] != result[4][1] # parent id
 
2234
                                    or result[5][0] != result[5][1] # name
 
2235
                                    or result[6][0] != result[6][1] # kind
 
2236
                                    or result[7][0] != result[7][1] # executable
 
2237
                                    ):
 
2238
                                    yield (result[0],
 
2239
                                           (utf8_decode_or_none(result[1][0]),
 
2240
                                            utf8_decode_or_none(result[1][1])),
 
2241
                                           result[2],
 
2242
                                           result[3],
 
2243
                                           result[4],
 
2244
                                           (utf8_decode_or_none(result[5][0]),
 
2245
                                            utf8_decode_or_none(result[5][1])),
 
2246
                                           result[6],
 
2247
                                           result[7],
 
2248
                                          )
 
2249
                            advance_path = False
 
2250
                    else:
 
2251
                        for result in _process_entry(current_entry, current_path_info):
 
2252
                            # this check should probably be outside the loop: one
 
2253
                            # 'iterate two trees' api, and then _iter_changes filters
 
2254
                            # unchanged pairs. - RBC 20070226
 
2255
                            path_handled = True
 
2256
                            if (include_unchanged
 
2257
                                or result[2]                    # content change
 
2258
                                or result[3][0] != result[3][1] # versioned status
 
2259
                                or result[4][0] != result[4][1] # parent id
 
2260
                                or result[5][0] != result[5][1] # name
 
2261
                                or result[6][0] != result[6][1] # kind
 
2262
                                or result[7][0] != result[7][1] # executable
 
2263
                                ):
 
2264
                                yield (result[0],
 
2265
                                       (utf8_decode_or_none(result[1][0]),
 
2266
                                        utf8_decode_or_none(result[1][1])),
 
2267
                                       result[2],
 
2268
                                       result[3],
 
2269
                                       result[4],
 
2270
                                       (utf8_decode_or_none(result[5][0]),
 
2271
                                        utf8_decode_or_none(result[5][1])),
 
2272
                                       result[6],
 
2273
                                       result[7],
 
2274
                                      )
 
2275
                    if advance_entry and current_entry is not None:
 
2276
                        entry_index += 1
 
2277
                        if entry_index < len(current_block[1]):
 
2278
                            current_entry = current_block[1][entry_index]
 
2279
                        else:
 
2280
                            current_entry = None
 
2281
                    else:
 
2282
                        advance_entry = True # reset the advance flaga
 
2283
                    if advance_path and current_path_info is not None:
 
2284
                        if not path_handled:
 
2285
                            # unversioned in all regards
 
2286
                            if want_unversioned:
 
2287
                                new_executable = bool(
 
2288
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2289
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2290
                                if want_unversioned:
 
2291
                                    yield (None,
 
2292
                                        (None, utf8_decode_or_none(current_path_info[0])),
 
2293
                                        True,
 
2294
                                        (False, False),
 
2295
                                        (None, None),
 
2296
                                        (None, utf8_decode_or_none(current_path_info[1])),
 
2297
                                        (None, current_path_info[2]),
 
2298
                                        (None, new_executable))
 
2299
                            # dont descend into this unversioned path if it is
 
2300
                            # a dir
 
2301
                            if current_path_info[2] in ('directory'):
 
2302
                                del current_dir_info[1][path_index]
 
2303
                                path_index -= 1
 
2304
                        # dont descend the disk iterator into any tree 
 
2305
                        # paths.
 
2306
                        if current_path_info[2] == 'tree-reference':
 
2307
                            del current_dir_info[1][path_index]
 
2308
                            path_index -= 1
 
2309
                        path_index += 1
 
2310
                        if path_index < len(current_dir_info[1]):
 
2311
                            current_path_info = current_dir_info[1][path_index]
 
2312
                            if current_path_info[2] == 'directory':
 
2313
                                if self.target._directory_is_tree_reference(
 
2314
                                    current_path_info[0].decode('utf8')):
 
2315
                                    current_path_info = current_path_info[:2] + \
 
2316
                                        ('tree-reference',) + current_path_info[3:]
 
2317
                        else:
 
2318
                            current_path_info = None
 
2319
                        path_handled = False
 
2320
                    else:
 
2321
                        advance_path = True # reset the advance flagg.
 
2322
                if current_block is not None:
 
2323
                    block_index += 1
 
2324
                    if (block_index < len(state._dirblocks) and
 
2325
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2326
                        current_block = state._dirblocks[block_index]
 
2327
                    else:
 
2328
                        current_block = None
 
2329
                if current_dir_info is not None:
 
2330
                    try:
 
2331
                        current_dir_info = dir_iterator.next()
 
2332
                    except StopIteration:
 
2333
                        current_dir_info = None
 
2334
 
2214
2335
 
2215
2336
    @staticmethod
2216
2337
    def is_compatible(source, target):
2217
2338
        # the target must be a dirstate working tree
2218
 
        if not isinstance(target, DirStateWorkingTree):
 
2339
        if not isinstance(target, WorkingTree4):
2219
2340
            return False
2220
 
        # the source must be a revtree or dirstate rev tree.
 
2341
        # the source must be a revtreee or dirstate rev tree.
2221
2342
        if not isinstance(source,
2222
2343
            (revisiontree.RevisionTree, DirStateRevisionTree)):
2223
2344
            return False
2224
2345
        # the source revid must be in the target dirstate
2225
 
        if not (source._revision_id == _mod_revision.NULL_REVISION or
 
2346
        if not (source._revision_id == NULL_REVISION or
2226
2347
            source._revision_id in target.get_parent_ids()):
2227
 
            # TODO: what about ghosts? it may well need to
 
2348
            # TODO: what about ghosts? it may well need to 
2228
2349
            # check for them explicitly.
2229
2350
            return False
2230
2351
        return True
2240
2361
 
2241
2362
    def convert(self, tree):
2242
2363
        # lock the control files not the tree, so that we dont get tree
2243
 
        # on-unlock behaviours, and so that noone else diddles with the
 
2364
        # on-unlock behaviours, and so that noone else diddles with the 
2244
2365
        # tree during upgrade.
2245
2366
        tree._control_files.lock_write()
2246
2367
        try:
2272
2393
 
2273
2394
    def update_format(self, tree):
2274
2395
        """Change the format marker."""
2275
 
        tree._transport.put_bytes('format',
2276
 
            self.target_format.get_format_string(),
2277
 
            mode=tree.bzrdir._get_file_mode())
2278
 
 
2279
 
 
2280
 
class Converter4to5(object):
2281
 
    """Perform an in-place upgrade of format 4 to format 5 trees."""
2282
 
 
2283
 
    def __init__(self):
2284
 
        self.target_format = WorkingTreeFormat5()
2285
 
 
2286
 
    def convert(self, tree):
2287
 
        # lock the control files not the tree, so that we don't get tree
2288
 
        # on-unlock behaviours, and so that no-one else diddles with the
2289
 
        # tree during upgrade.
2290
 
        tree._control_files.lock_write()
2291
 
        try:
2292
 
            self.update_format(tree)
2293
 
        finally:
2294
 
            tree._control_files.unlock()
2295
 
 
2296
 
    def update_format(self, tree):
2297
 
        """Change the format marker."""
2298
 
        tree._transport.put_bytes('format',
2299
 
            self.target_format.get_format_string(),
2300
 
            mode=tree.bzrdir._get_file_mode())
2301
 
 
2302
 
 
2303
 
class Converter4or5to6(object):
2304
 
    """Perform an in-place upgrade of format 4 or 5 to format 6 trees."""
2305
 
 
2306
 
    def __init__(self):
2307
 
        self.target_format = WorkingTreeFormat6()
2308
 
 
2309
 
    def convert(self, tree):
2310
 
        # lock the control files not the tree, so that we don't get tree
2311
 
        # on-unlock behaviours, and so that no-one else diddles with the
2312
 
        # tree during upgrade.
2313
 
        tree._control_files.lock_write()
2314
 
        try:
2315
 
            self.init_custom_control_files(tree)
2316
 
            self.update_format(tree)
2317
 
        finally:
2318
 
            tree._control_files.unlock()
2319
 
 
2320
 
    def init_custom_control_files(self, tree):
2321
 
        """Initialize custom control files."""
2322
 
        tree._transport.put_bytes('views', '',
2323
 
            mode=tree.bzrdir._get_file_mode())
2324
 
 
2325
 
    def update_format(self, tree):
2326
 
        """Change the format marker."""
2327
 
        tree._transport.put_bytes('format',
2328
 
            self.target_format.get_format_string(),
2329
 
            mode=tree.bzrdir._get_file_mode())
 
2396
        tree._control_files.put_utf8('format',
 
2397
            self.target_format.get_format_string())