/brz/remove-bazaar

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

« back to all changes in this revision

Viewing changes to bzrlib/workingtree_4.py

  • Committer: Aaron Bentley
  • Date: 2007-12-19 13:20:40 UTC
  • mto: This revision was merged to the branch mainline in revision 3135.
  • Revision ID: aaron.bentley@utoronto.ca-20071219132040-bzbczkkcjko742i1
Work around double-opening lock issue

Show diffs side-by-side

added added

removed removed

Lines of Context:
 
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
 
2
#
 
3
# This program is free software; you can redistribute it and/or modify
 
4
# it under the terms of the GNU General Public License as published by
 
5
# the Free Software Foundation; either version 2 of the License, or
 
6
# (at your option) any later version.
 
7
#
 
8
# This program is distributed in the hope that it will be useful,
 
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
 
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 
11
# GNU General Public License for more details.
 
12
#
 
13
# You should have received a copy of the GNU General Public License
 
14
# along with this program; if not, write to the Free Software
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
16
 
 
17
"""WorkingTree4 format and implementation.
 
18
 
 
19
WorkingTree4 provides the dirstate based working tree logic.
 
20
 
 
21
To get a WorkingTree, call bzrdir.open_workingtree() or
 
22
WorkingTree.open(dir).
 
23
"""
 
24
 
 
25
from cStringIO import StringIO
 
26
import os
 
27
import sys
 
28
 
 
29
from bzrlib.lazy_import import lazy_import
 
30
lazy_import(globals(), """
 
31
from bisect import bisect_left
 
32
import collections
 
33
from copy import deepcopy
 
34
import errno
 
35
import itertools
 
36
import operator
 
37
import stat
 
38
from time import time
 
39
import warnings
 
40
 
 
41
import bzrlib
 
42
from bzrlib import (
 
43
    bzrdir,
 
44
    cache_utf8,
 
45
    conflicts as _mod_conflicts,
 
46
    debug,
 
47
    delta,
 
48
    dirstate,
 
49
    errors,
 
50
    generate_ids,
 
51
    globbing,
 
52
    ignores,
 
53
    merge,
 
54
    osutils,
 
55
    revision as _mod_revision,
 
56
    revisiontree,
 
57
    textui,
 
58
    trace,
 
59
    transform,
 
60
    urlutils,
 
61
    xml5,
 
62
    xml6,
 
63
    )
 
64
import bzrlib.branch
 
65
from bzrlib.transport import get_transport
 
66
import bzrlib.ui
 
67
""")
 
68
 
 
69
from bzrlib import symbol_versioning
 
70
from bzrlib.decorators import needs_read_lock, needs_write_lock
 
71
from bzrlib.inventory import InventoryEntry, Inventory, ROOT_ID, entry_factory
 
72
from bzrlib.lockable_files import LockableFiles, TransportLock
 
73
from bzrlib.lockdir import LockDir
 
74
import bzrlib.mutabletree
 
75
from bzrlib.mutabletree import needs_tree_write_lock
 
76
from bzrlib.osutils import (
 
77
    file_kind,
 
78
    isdir,
 
79
    normpath,
 
80
    pathjoin,
 
81
    rand_chars,
 
82
    realpath,
 
83
    safe_unicode,
 
84
    splitpath,
 
85
    )
 
86
from bzrlib.trace import mutter, note
 
87
from bzrlib.transport.local import LocalTransport
 
88
from bzrlib.tree import InterTree
 
89
from bzrlib.progress import DummyProgress, ProgressPhase
 
90
from bzrlib.revision import NULL_REVISION, CURRENT_REVISION
 
91
from bzrlib.rio import RioReader, rio_file, Stanza
 
92
from bzrlib.symbol_versioning import (deprecated_passed,
 
93
        deprecated_method,
 
94
        deprecated_function,
 
95
        DEPRECATED_PARAMETER,
 
96
        )
 
97
from bzrlib.tree import Tree
 
98
from bzrlib.workingtree import WorkingTree, WorkingTree3, WorkingTreeFormat3
 
99
 
 
100
 
 
101
# This is the Windows equivalent of ENOTDIR
 
102
# It is defined in pywin32.winerror, but we don't want a strong dependency for
 
103
# just an error code.
 
104
ERROR_PATH_NOT_FOUND = 3
 
105
ERROR_DIRECTORY = 267
 
106
 
 
107
 
 
108
class WorkingTree4(WorkingTree3):
 
109
    """This is the Format 4 working tree.
 
110
 
 
111
    This differs from WorkingTree3 by:
 
112
     - Having a consolidated internal dirstate, stored in a
 
113
       randomly-accessible sorted file on disk.
 
114
     - Not having a regular inventory attribute.  One can be synthesized 
 
115
       on demand but this is expensive and should be avoided.
 
116
 
 
117
    This is new in bzr 0.15.
 
118
    """
 
119
 
 
120
    def __init__(self, basedir,
 
121
                 branch,
 
122
                 _control_files=None,
 
123
                 _format=None,
 
124
                 _bzrdir=None):
 
125
        """Construct a WorkingTree for basedir.
 
126
 
 
127
        If the branch is not supplied, it is opened automatically.
 
128
        If the branch is supplied, it must be the branch for this basedir.
 
129
        (branch.base is not cross checked, because for remote branches that
 
130
        would be meaningless).
 
131
        """
 
132
        self._format = _format
 
133
        self.bzrdir = _bzrdir
 
134
        assert isinstance(basedir, basestring), \
 
135
            "base directory %r is not a string" % basedir
 
136
        basedir = safe_unicode(basedir)
 
137
        mutter("opening working tree %r", basedir)
 
138
        self._branch = branch
 
139
        assert isinstance(self.branch, bzrlib.branch.Branch), \
 
140
            "branch %r is not a Branch" % self.branch
 
141
        self.basedir = realpath(basedir)
 
142
        # if branch is at our basedir and is a format 6 or less
 
143
        # assume all other formats have their own control files.
 
144
        assert isinstance(_control_files, LockableFiles), \
 
145
            "_control_files must be a LockableFiles, not %r" % _control_files
 
146
        self._control_files = _control_files
 
147
        self._dirty = None
 
148
        #-------------
 
149
        # during a read or write lock these objects are set, and are
 
150
        # None the rest of the time.
 
151
        self._dirstate = None
 
152
        self._inventory = None
 
153
        #-------------
 
154
        self._detect_case_handling()
 
155
 
 
156
    @needs_tree_write_lock
 
157
    def _add(self, files, ids, kinds):
 
158
        """See MutableTree._add."""
 
159
        state = self.current_dirstate()
 
160
        for f, file_id, kind in zip(files, ids, kinds):
 
161
            f = f.strip('/')
 
162
            if self.path2id(f):
 
163
                # special case tree root handling.
 
164
                if f == '' and self.path2id(f) == ROOT_ID:
 
165
                    state.set_path_id('', generate_ids.gen_file_id(f))
 
166
                continue
 
167
            if file_id is None:
 
168
                file_id = generate_ids.gen_file_id(f)
 
169
            # deliberately add the file with no cached stat or sha1
 
170
            # - on the first access it will be gathered, and we can
 
171
            # always change this once tests are all passing.
 
172
            state.add(f, file_id, kind, None, '')
 
173
        self._make_dirty(reset_inventory=True)
 
174
 
 
175
    def _make_dirty(self, reset_inventory):
 
176
        """Make the tree state dirty.
 
177
 
 
178
        :param reset_inventory: True if the cached inventory should be removed
 
179
            (presuming there is one).
 
180
        """
 
181
        self._dirty = True
 
182
        if reset_inventory and self._inventory is not None:
 
183
            self._inventory = None
 
184
 
 
185
    @needs_tree_write_lock
 
186
    def add_reference(self, sub_tree):
 
187
        # use standard implementation, which calls back to self._add
 
188
        # 
 
189
        # So we don't store the reference_revision in the working dirstate,
 
190
        # it's just recorded at the moment of commit. 
 
191
        self._add_reference(sub_tree)
 
192
 
 
193
    def break_lock(self):
 
194
        """Break a lock if one is present from another instance.
 
195
 
 
196
        Uses the ui factory to ask for confirmation if the lock may be from
 
197
        an active process.
 
198
 
 
199
        This will probe the repository for its lock as well.
 
200
        """
 
201
        # if the dirstate is locked by an active process, reject the break lock
 
202
        # call.
 
203
        try:
 
204
            if self._dirstate is None:
 
205
                clear = True
 
206
            else:
 
207
                clear = False
 
208
            state = self._current_dirstate()
 
209
            if state._lock_token is not None:
 
210
                # we already have it locked. sheese, cant break our own lock.
 
211
                raise errors.LockActive(self.basedir)
 
212
            else:
 
213
                try:
 
214
                    # try for a write lock - need permission to get one anyhow
 
215
                    # to break locks.
 
216
                    state.lock_write()
 
217
                except errors.LockContention:
 
218
                    # oslocks fail when a process is still live: fail.
 
219
                    # TODO: get the locked lockdir info and give to the user to
 
220
                    # assist in debugging.
 
221
                    raise errors.LockActive(self.basedir)
 
222
                else:
 
223
                    state.unlock()
 
224
        finally:
 
225
            if clear:
 
226
                self._dirstate = None
 
227
        self._control_files.break_lock()
 
228
        self.branch.break_lock()
 
229
 
 
230
    def _comparison_data(self, entry, path):
 
231
        kind, executable, stat_value = \
 
232
            WorkingTree3._comparison_data(self, entry, path)
 
233
        # it looks like a plain directory, but it's really a reference -- see
 
234
        # also kind()
 
235
        if (self._repo_supports_tree_reference and
 
236
            kind == 'directory' and
 
237
            self._directory_is_tree_reference(path)):
 
238
            kind = 'tree-reference'
 
239
        return kind, executable, stat_value
 
240
 
 
241
    @needs_write_lock
 
242
    def commit(self, message=None, revprops=None, *args, **kwargs):
 
243
        # mark the tree as dirty post commit - commit
 
244
        # can change the current versioned list by doing deletes.
 
245
        result = WorkingTree3.commit(self, message, revprops, *args, **kwargs)
 
246
        self._make_dirty(reset_inventory=True)
 
247
        return result
 
248
 
 
249
    def current_dirstate(self):
 
250
        """Return the current dirstate object.
 
251
 
 
252
        This is not part of the tree interface and only exposed for ease of
 
253
        testing.
 
254
 
 
255
        :raises errors.NotWriteLocked: when not in a lock.
 
256
        """
 
257
        self._must_be_locked()
 
258
        return self._current_dirstate()
 
259
 
 
260
    def _current_dirstate(self):
 
261
        """Internal function that does not check lock status.
 
262
 
 
263
        This is needed for break_lock which also needs the dirstate.
 
264
        """
 
265
        if self._dirstate is not None:
 
266
            return self._dirstate
 
267
        local_path = self.bzrdir.get_workingtree_transport(None
 
268
            ).local_abspath('dirstate')
 
269
        self._dirstate = dirstate.DirState.on_file(local_path)
 
270
        return self._dirstate
 
271
 
 
272
    def filter_unversioned_files(self, paths):
 
273
        """Filter out paths that are versioned.
 
274
 
 
275
        :return: set of paths.
 
276
        """
 
277
        # TODO: make a generic multi-bisect routine roughly that should list
 
278
        # the paths, then process one half at a time recursively, and feed the
 
279
        # results of each bisect in further still
 
280
        paths = sorted(paths)
 
281
        result = set()
 
282
        state = self.current_dirstate()
 
283
        # TODO we want a paths_to_dirblocks helper I think
 
284
        for path in paths:
 
285
            dirname, basename = os.path.split(path.encode('utf8'))
 
286
            _, _, _, path_is_versioned = state._get_block_entry_index(
 
287
                dirname, basename, 0)
 
288
            if not path_is_versioned:
 
289
                result.add(path)
 
290
        return result
 
291
 
 
292
    def flush(self):
 
293
        """Write all cached data to disk."""
 
294
        if self._control_files._lock_mode != 'w':
 
295
            raise errors.NotWriteLocked(self)
 
296
        self.current_dirstate().save()
 
297
        self._inventory = None
 
298
        self._dirty = False
 
299
 
 
300
    @needs_tree_write_lock
 
301
    def _gather_kinds(self, files, kinds):
 
302
        """See MutableTree._gather_kinds."""
 
303
        for pos, f in enumerate(files):
 
304
            if kinds[pos] is None:
 
305
                kinds[pos] = self._kind(f)
 
306
 
 
307
    def _generate_inventory(self):
 
308
        """Create and set self.inventory from the dirstate object.
 
309
        
 
310
        This is relatively expensive: we have to walk the entire dirstate.
 
311
        Ideally we would not, and can deprecate this function.
 
312
        """
 
313
        #: uncomment to trap on inventory requests.
 
314
        # import pdb;pdb.set_trace()
 
315
        state = self.current_dirstate()
 
316
        state._read_dirblocks_if_needed()
 
317
        root_key, current_entry = self._get_entry(path='')
 
318
        current_id = root_key[2]
 
319
        assert current_entry[0][0] == 'd' # directory
 
320
        inv = Inventory(root_id=current_id)
 
321
        # Turn some things into local variables
 
322
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
323
        factory = entry_factory
 
324
        utf8_decode = cache_utf8._utf8_decode
 
325
        inv_byid = inv._byid
 
326
        # we could do this straight out of the dirstate; it might be fast
 
327
        # and should be profiled - RBC 20070216
 
328
        parent_ies = {'' : inv.root}
 
329
        for block in state._dirblocks[1:]: # skip the root
 
330
            dirname = block[0]
 
331
            try:
 
332
                parent_ie = parent_ies[dirname]
 
333
            except KeyError:
 
334
                # all the paths in this block are not versioned in this tree
 
335
                continue
 
336
            for key, entry in block[1]:
 
337
                minikind, link_or_sha1, size, executable, stat = entry[0]
 
338
                if minikind in ('a', 'r'): # absent, relocated
 
339
                    # a parent tree only entry
 
340
                    continue
 
341
                name = key[1]
 
342
                name_unicode = utf8_decode(name)[0]
 
343
                file_id = key[2]
 
344
                kind = minikind_to_kind[minikind]
 
345
                inv_entry = factory[kind](file_id, name_unicode,
 
346
                                          parent_ie.file_id)
 
347
                if kind == 'file':
 
348
                    # This is only needed on win32, where this is the only way
 
349
                    # we know the executable bit.
 
350
                    inv_entry.executable = executable
 
351
                    # not strictly needed: working tree
 
352
                    #inv_entry.text_size = size
 
353
                    #inv_entry.text_sha1 = sha1
 
354
                elif kind == 'directory':
 
355
                    # add this entry to the parent map.
 
356
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
357
                elif kind == 'tree-reference':
 
358
                    assert self._repo_supports_tree_reference, \
 
359
                        "repository of %r " \
 
360
                        "doesn't support tree references " \
 
361
                        "required by entry %r" \
 
362
                        % (self, name)
 
363
                    inv_entry.reference_revision = link_or_sha1 or None
 
364
                elif kind != 'symlink':
 
365
                    raise AssertionError("unknown kind %r" % kind)
 
366
                # These checks cost us around 40ms on a 55k entry tree
 
367
                assert file_id not in inv_byid, ('file_id %s already in'
 
368
                    ' inventory as %s' % (file_id, inv_byid[file_id]))
 
369
                assert name_unicode not in parent_ie.children
 
370
                inv_byid[file_id] = inv_entry
 
371
                parent_ie.children[name_unicode] = inv_entry
 
372
        self._inventory = inv
 
373
 
 
374
    def _get_entry(self, file_id=None, path=None):
 
375
        """Get the dirstate row for file_id or path.
 
376
 
 
377
        If either file_id or path is supplied, it is used as the key to lookup.
 
378
        If both are supplied, the fastest lookup is used, and an error is
 
379
        raised if they do not both point at the same row.
 
380
        
 
381
        :param file_id: An optional unicode file_id to be looked up.
 
382
        :param path: An optional unicode path to be looked up.
 
383
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
384
        """
 
385
        if file_id is None and path is None:
 
386
            raise errors.BzrError('must supply file_id or path')
 
387
        state = self.current_dirstate()
 
388
        if path is not None:
 
389
            path = path.encode('utf8')
 
390
        return state._get_entry(0, fileid_utf8=file_id, path_utf8=path)
 
391
 
 
392
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
393
        # check file id is valid unconditionally.
 
394
        entry = self._get_entry(file_id=file_id, path=path)
 
395
        if entry[0] is None:
 
396
            raise errors.NoSuchId(self, file_id)
 
397
        if path is None:
 
398
            path = pathjoin(entry[0][0], entry[0][1]).decode('utf8')
 
399
 
 
400
        file_abspath = self.abspath(path)
 
401
        state = self.current_dirstate()
 
402
        if stat_value is None:
 
403
            try:
 
404
                stat_value = os.lstat(file_abspath)
 
405
            except OSError, e:
 
406
                if e.errno == errno.ENOENT:
 
407
                    return None
 
408
                else:
 
409
                    raise
 
410
        link_or_sha1 = state.update_entry(entry, file_abspath,
 
411
                                          stat_value=stat_value)
 
412
        if entry[1][0][0] == 'f':
 
413
            return link_or_sha1
 
414
        return None
 
415
 
 
416
    def _get_inventory(self):
 
417
        """Get the inventory for the tree. This is only valid within a lock."""
 
418
        if 'evil' in debug.debug_flags:
 
419
            trace.mutter_callsite(2,
 
420
                "accessing .inventory forces a size of tree translation.")
 
421
        if self._inventory is not None:
 
422
            return self._inventory
 
423
        self._must_be_locked()
 
424
        self._generate_inventory()
 
425
        return self._inventory
 
426
 
 
427
    inventory = property(_get_inventory,
 
428
                         doc="Inventory of this Tree")
 
429
 
 
430
    @needs_read_lock
 
431
    def get_parent_ids(self):
 
432
        """See Tree.get_parent_ids.
 
433
        
 
434
        This implementation requests the ids list from the dirstate file.
 
435
        """
 
436
        return self.current_dirstate().get_parent_ids()
 
437
 
 
438
    def get_reference_revision(self, file_id, path=None):
 
439
        # referenced tree's revision is whatever's currently there
 
440
        return self.get_nested_tree(file_id, path).last_revision()
 
441
 
 
442
    def get_nested_tree(self, file_id, path=None):
 
443
        if path is None:
 
444
            path = self.id2path(file_id)
 
445
        # else: check file_id is at path?
 
446
        return WorkingTree.open(self.abspath(path))
 
447
 
 
448
    @needs_read_lock
 
449
    def get_root_id(self):
 
450
        """Return the id of this trees root"""
 
451
        return self._get_entry(path='')[0][2]
 
452
 
 
453
    def has_id(self, file_id):
 
454
        state = self.current_dirstate()
 
455
        row, parents = self._get_entry(file_id=file_id)
 
456
        if row is None:
 
457
            return False
 
458
        return osutils.lexists(pathjoin(
 
459
                    self.basedir, row[0].decode('utf8'), row[1].decode('utf8')))
 
460
 
 
461
    @needs_read_lock
 
462
    def id2path(self, file_id):
 
463
        "Convert a file-id to a path."
 
464
        state = self.current_dirstate()
 
465
        entry = self._get_entry(file_id=file_id)
 
466
        if entry == (None, None):
 
467
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
468
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
469
        return path_utf8.decode('utf8')
 
470
 
 
471
    def _is_executable_from_path_and_stat_from_basis(self, path, stat_result):
 
472
        entry = self._get_entry(path=path)
 
473
        if entry == (None, None):
 
474
            return False # Missing entries are not executable
 
475
        return entry[1][0][3] # Executable?
 
476
 
 
477
    if not osutils.supports_executable():
 
478
        def is_executable(self, file_id, path=None):
 
479
            """Test if a file is executable or not.
 
480
 
 
481
            Note: The caller is expected to take a read-lock before calling this.
 
482
            """
 
483
            entry = self._get_entry(file_id=file_id, path=path)
 
484
            if entry == (None, None):
 
485
                return False
 
486
            return entry[1][0][3]
 
487
 
 
488
        _is_executable_from_path_and_stat = \
 
489
            _is_executable_from_path_and_stat_from_basis
 
490
    else:
 
491
        def is_executable(self, file_id, path=None):
 
492
            """Test if a file is executable or not.
 
493
 
 
494
            Note: The caller is expected to take a read-lock before calling this.
 
495
            """
 
496
            if not path:
 
497
                path = self.id2path(file_id)
 
498
            mode = os.lstat(self.abspath(path)).st_mode
 
499
            return bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
 
500
 
 
501
    @needs_read_lock
 
502
    def __iter__(self):
 
503
        """Iterate through file_ids for this tree.
 
504
 
 
505
        file_ids are in a WorkingTree if they are in the working inventory
 
506
        and the working file exists.
 
507
        """
 
508
        result = []
 
509
        for key, tree_details in self.current_dirstate()._iter_entries():
 
510
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
 
511
                # not relevant to the working tree
 
512
                continue
 
513
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
 
514
            if osutils.lexists(path):
 
515
                result.append(key[2])
 
516
        return iter(result)
 
517
 
 
518
    def iter_references(self):
 
519
        for key, tree_details in self.current_dirstate()._iter_entries():
 
520
            if tree_details[0][0] in ('a', 'r'): # absent, relocated
 
521
                # not relevant to the working tree
 
522
                continue
 
523
            if not key[1]:
 
524
                # the root is not a reference.
 
525
                continue
 
526
            path = pathjoin(self.basedir, key[0].decode('utf8'), key[1].decode('utf8'))
 
527
            try:
 
528
                if self._kind(path) == 'tree-reference':
 
529
                    yield path, key[2]
 
530
            except errors.NoSuchFile:
 
531
                # path is missing on disk.
 
532
                continue
 
533
 
 
534
    def kind(self, file_id):
 
535
        """Return the kind of a file.
 
536
 
 
537
        This is always the actual kind that's on disk, regardless of what it
 
538
        was added as.
 
539
 
 
540
        Note: The caller is expected to take a read-lock before calling this.
 
541
        """
 
542
        relpath = self.id2path(file_id)
 
543
        assert relpath != None, \
 
544
            "path for id {%s} is None!" % file_id
 
545
        return self._kind(relpath)
 
546
 
 
547
    def _kind(self, relpath):
 
548
        abspath = self.abspath(relpath)
 
549
        kind = file_kind(abspath)
 
550
        if (self._repo_supports_tree_reference and
 
551
            kind == 'directory' and
 
552
            self._directory_is_tree_reference(relpath)):
 
553
            kind = 'tree-reference'
 
554
        return kind
 
555
 
 
556
    @needs_read_lock
 
557
    def _last_revision(self):
 
558
        """See Mutable.last_revision."""
 
559
        parent_ids = self.current_dirstate().get_parent_ids()
 
560
        if parent_ids:
 
561
            return parent_ids[0]
 
562
        else:
 
563
            return _mod_revision.NULL_REVISION
 
564
 
 
565
    def lock_read(self):
 
566
        """See Branch.lock_read, and WorkingTree.unlock."""
 
567
        self.branch.lock_read()
 
568
        try:
 
569
            self._control_files.lock_read()
 
570
            try:
 
571
                state = self.current_dirstate()
 
572
                if not state._lock_token:
 
573
                    state.lock_read()
 
574
                # set our support for tree references from the repository in
 
575
                # use.
 
576
                self._repo_supports_tree_reference = getattr(
 
577
                    self.branch.repository._format, "supports_tree_reference",
 
578
                    False)
 
579
            except:
 
580
                self._control_files.unlock()
 
581
                raise
 
582
        except:
 
583
            self.branch.unlock()
 
584
            raise
 
585
 
 
586
    def _lock_self_write(self):
 
587
        """This should be called after the branch is locked."""
 
588
        try:
 
589
            self._control_files.lock_write()
 
590
            try:
 
591
                state = self.current_dirstate()
 
592
                if not state._lock_token:
 
593
                    state.lock_write()
 
594
                # set our support for tree references from the repository in
 
595
                # use.
 
596
                self._repo_supports_tree_reference = getattr(
 
597
                    self.branch.repository._format, "supports_tree_reference",
 
598
                    False)
 
599
            except:
 
600
                self._control_files.unlock()
 
601
                raise
 
602
        except:
 
603
            self.branch.unlock()
 
604
            raise
 
605
 
 
606
    def lock_tree_write(self):
 
607
        """See MutableTree.lock_tree_write, and WorkingTree.unlock."""
 
608
        self.branch.lock_read()
 
609
        self._lock_self_write()
 
610
 
 
611
    def lock_write(self):
 
612
        """See MutableTree.lock_write, and WorkingTree.unlock."""
 
613
        self.branch.lock_write()
 
614
        self._lock_self_write()
 
615
 
 
616
    @needs_tree_write_lock
 
617
    def move(self, from_paths, to_dir, after=False):
 
618
        """See WorkingTree.move()."""
 
619
        result = []
 
620
        if not from_paths:
 
621
            return result
 
622
 
 
623
        state = self.current_dirstate()
 
624
 
 
625
        assert not isinstance(from_paths, basestring)
 
626
        to_dir_utf8 = to_dir.encode('utf8')
 
627
        to_entry_dirname, to_basename = os.path.split(to_dir_utf8)
 
628
        id_index = state._get_id_index()
 
629
        # check destination directory
 
630
        # get the details for it
 
631
        to_entry_block_index, to_entry_entry_index, dir_present, entry_present = \
 
632
            state._get_block_entry_index(to_entry_dirname, to_basename, 0)
 
633
        if not entry_present:
 
634
            raise errors.BzrMoveFailedError('', to_dir,
 
635
                errors.NotVersionedError(to_dir))
 
636
        to_entry = state._dirblocks[to_entry_block_index][1][to_entry_entry_index]
 
637
        # get a handle on the block itself.
 
638
        to_block_index = state._ensure_block(
 
639
            to_entry_block_index, to_entry_entry_index, to_dir_utf8)
 
640
        to_block = state._dirblocks[to_block_index]
 
641
        to_abs = self.abspath(to_dir)
 
642
        if not isdir(to_abs):
 
643
            raise errors.BzrMoveFailedError('',to_dir,
 
644
                errors.NotADirectory(to_abs))
 
645
 
 
646
        if to_entry[1][0][0] != 'd':
 
647
            raise errors.BzrMoveFailedError('',to_dir,
 
648
                errors.NotADirectory(to_abs))
 
649
 
 
650
        if self._inventory is not None:
 
651
            update_inventory = True
 
652
            inv = self.inventory
 
653
            to_dir_ie = inv[to_dir_id]
 
654
            to_dir_id = to_entry[0][2]
 
655
        else:
 
656
            update_inventory = False
 
657
 
 
658
        rollbacks = []
 
659
        def move_one(old_entry, from_path_utf8, minikind, executable,
 
660
                     fingerprint, packed_stat, size,
 
661
                     to_block, to_key, to_path_utf8):
 
662
            state._make_absent(old_entry)
 
663
            from_key = old_entry[0]
 
664
            rollbacks.append(
 
665
                lambda:state.update_minimal(from_key,
 
666
                    minikind,
 
667
                    executable=executable,
 
668
                    fingerprint=fingerprint,
 
669
                    packed_stat=packed_stat,
 
670
                    size=size,
 
671
                    path_utf8=from_path_utf8))
 
672
            state.update_minimal(to_key,
 
673
                    minikind,
 
674
                    executable=executable,
 
675
                    fingerprint=fingerprint,
 
676
                    packed_stat=packed_stat,
 
677
                    size=size,
 
678
                    path_utf8=to_path_utf8)
 
679
            added_entry_index, _ = state._find_entry_index(to_key, to_block[1])
 
680
            new_entry = to_block[1][added_entry_index]
 
681
            rollbacks.append(lambda:state._make_absent(new_entry))
 
682
 
 
683
        for from_rel in from_paths:
 
684
            # from_rel is 'pathinroot/foo/bar'
 
685
            from_rel_utf8 = from_rel.encode('utf8')
 
686
            from_dirname, from_tail = osutils.split(from_rel)
 
687
            from_dirname, from_tail_utf8 = osutils.split(from_rel_utf8)
 
688
            from_entry = self._get_entry(path=from_rel)
 
689
            if from_entry == (None, None):
 
690
                raise errors.BzrMoveFailedError(from_rel,to_dir,
 
691
                    errors.NotVersionedError(path=str(from_rel)))
 
692
 
 
693
            from_id = from_entry[0][2]
 
694
            to_rel = pathjoin(to_dir, from_tail)
 
695
            to_rel_utf8 = pathjoin(to_dir_utf8, from_tail_utf8)
 
696
            item_to_entry = self._get_entry(path=to_rel)
 
697
            if item_to_entry != (None, None):
 
698
                raise errors.BzrMoveFailedError(from_rel, to_rel,
 
699
                    "Target is already versioned.")
 
700
 
 
701
            if from_rel == to_rel:
 
702
                raise errors.BzrMoveFailedError(from_rel, to_rel,
 
703
                    "Source and target are identical.")
 
704
 
 
705
            from_missing = not self.has_filename(from_rel)
 
706
            to_missing = not self.has_filename(to_rel)
 
707
            if after:
 
708
                move_file = False
 
709
            else:
 
710
                move_file = True
 
711
            if to_missing:
 
712
                if not move_file:
 
713
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
 
714
                        errors.NoSuchFile(path=to_rel,
 
715
                        extra="New file has not been created yet"))
 
716
                elif from_missing:
 
717
                    # neither path exists
 
718
                    raise errors.BzrRenameFailedError(from_rel, to_rel,
 
719
                        errors.PathsDoNotExist(paths=(from_rel, to_rel)))
 
720
            else:
 
721
                if from_missing: # implicitly just update our path mapping
 
722
                    move_file = False
 
723
                elif not after:
 
724
                    raise errors.RenameFailedFilesExist(from_rel, to_rel)
 
725
 
 
726
            rollbacks = []
 
727
            def rollback_rename():
 
728
                """A single rename has failed, roll it back."""
 
729
                # roll back everything, even if we encounter trouble doing one
 
730
                # of them.
 
731
                #
 
732
                # TODO: at least log the other exceptions rather than just
 
733
                # losing them mbp 20070307
 
734
                exc_info = None
 
735
                for rollback in reversed(rollbacks):
 
736
                    try:
 
737
                        rollback()
 
738
                    except Exception, e:
 
739
                        exc_info = sys.exc_info()
 
740
                if exc_info:
 
741
                    raise exc_info[0], exc_info[1], exc_info[2]
 
742
 
 
743
            # perform the disk move first - its the most likely failure point.
 
744
            if move_file:
 
745
                from_rel_abs = self.abspath(from_rel)
 
746
                to_rel_abs = self.abspath(to_rel)
 
747
                try:
 
748
                    osutils.rename(from_rel_abs, to_rel_abs)
 
749
                except OSError, e:
 
750
                    raise errors.BzrMoveFailedError(from_rel, to_rel, e[1])
 
751
                rollbacks.append(lambda: osutils.rename(to_rel_abs, from_rel_abs))
 
752
            try:
 
753
                # perform the rename in the inventory next if needed: its easy
 
754
                # to rollback
 
755
                if update_inventory:
 
756
                    # rename the entry
 
757
                    from_entry = inv[from_id]
 
758
                    current_parent = from_entry.parent_id
 
759
                    inv.rename(from_id, to_dir_id, from_tail)
 
760
                    rollbacks.append(
 
761
                        lambda: inv.rename(from_id, current_parent, from_tail))
 
762
                # finally do the rename in the dirstate, which is a little
 
763
                # tricky to rollback, but least likely to need it.
 
764
                old_block_index, old_entry_index, dir_present, file_present = \
 
765
                    state._get_block_entry_index(from_dirname, from_tail_utf8, 0)
 
766
                old_block = state._dirblocks[old_block_index][1]
 
767
                old_entry = old_block[old_entry_index]
 
768
                from_key, old_entry_details = old_entry
 
769
                cur_details = old_entry_details[0]
 
770
                # remove the old row
 
771
                to_key = ((to_block[0],) + from_key[1:3])
 
772
                minikind = cur_details[0]
 
773
                move_one(old_entry, from_path_utf8=from_rel_utf8,
 
774
                         minikind=minikind,
 
775
                         executable=cur_details[3],
 
776
                         fingerprint=cur_details[1],
 
777
                         packed_stat=cur_details[4],
 
778
                         size=cur_details[2],
 
779
                         to_block=to_block,
 
780
                         to_key=to_key,
 
781
                         to_path_utf8=to_rel_utf8)
 
782
 
 
783
                if minikind == 'd':
 
784
                    def update_dirblock(from_dir, to_key, to_dir_utf8):
 
785
                        """Recursively update all entries in this dirblock."""
 
786
                        assert from_dir != '', "renaming root not supported"
 
787
                        from_key = (from_dir, '')
 
788
                        from_block_idx, present = \
 
789
                            state._find_block_index_from_key(from_key)
 
790
                        if not present:
 
791
                            # This is the old record, if it isn't present, then
 
792
                            # there is theoretically nothing to update.
 
793
                            # (Unless it isn't present because of lazy loading,
 
794
                            # but we don't do that yet)
 
795
                            return
 
796
                        from_block = state._dirblocks[from_block_idx]
 
797
                        to_block_index, to_entry_index, _, _ = \
 
798
                            state._get_block_entry_index(to_key[0], to_key[1], 0)
 
799
                        to_block_index = state._ensure_block(
 
800
                            to_block_index, to_entry_index, to_dir_utf8)
 
801
                        to_block = state._dirblocks[to_block_index]
 
802
 
 
803
                        # Grab a copy since move_one may update the list.
 
804
                        for entry in from_block[1][:]:
 
805
                            assert entry[0][0] == from_dir
 
806
                            cur_details = entry[1][0]
 
807
                            to_key = (to_dir_utf8, entry[0][1], entry[0][2])
 
808
                            from_path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
809
                            to_path_utf8 = osutils.pathjoin(to_dir_utf8, entry[0][1])
 
810
                            minikind = cur_details[0]
 
811
                            if minikind in 'ar':
 
812
                                # Deleted children of a renamed directory
 
813
                                # Do not need to be updated.
 
814
                                # Children that have been renamed out of this
 
815
                                # directory should also not be updated
 
816
                                continue
 
817
                            move_one(entry, from_path_utf8=from_path_utf8,
 
818
                                     minikind=minikind,
 
819
                                     executable=cur_details[3],
 
820
                                     fingerprint=cur_details[1],
 
821
                                     packed_stat=cur_details[4],
 
822
                                     size=cur_details[2],
 
823
                                     to_block=to_block,
 
824
                                     to_key=to_key,
 
825
                                     to_path_utf8=to_path_utf8)
 
826
                            if minikind == 'd':
 
827
                                # We need to move all the children of this
 
828
                                # entry
 
829
                                update_dirblock(from_path_utf8, to_key,
 
830
                                                to_path_utf8)
 
831
                    update_dirblock(from_rel_utf8, to_key, to_rel_utf8)
 
832
            except:
 
833
                rollback_rename()
 
834
                raise
 
835
            result.append((from_rel, to_rel))
 
836
            state._dirblock_state = dirstate.DirState.IN_MEMORY_MODIFIED
 
837
            self._make_dirty(reset_inventory=False)
 
838
 
 
839
        return result
 
840
 
 
841
    def _must_be_locked(self):
 
842
        if not self._control_files._lock_count:
 
843
            raise errors.ObjectNotLocked(self)
 
844
 
 
845
    def _new_tree(self):
 
846
        """Initialize the state in this tree to be a new tree."""
 
847
        self._dirty = True
 
848
 
 
849
    @needs_read_lock
 
850
    def path2id(self, path):
 
851
        """Return the id for path in this tree."""
 
852
        path = path.strip('/')
 
853
        entry = self._get_entry(path=path)
 
854
        if entry == (None, None):
 
855
            return None
 
856
        return entry[0][2]
 
857
 
 
858
    def paths2ids(self, paths, trees=[], require_versioned=True):
 
859
        """See Tree.paths2ids().
 
860
 
 
861
        This specialisation fast-paths the case where all the trees are in the
 
862
        dirstate.
 
863
        """
 
864
        if paths is None:
 
865
            return None
 
866
        parents = self.get_parent_ids()
 
867
        for tree in trees:
 
868
            if not (isinstance(tree, DirStateRevisionTree) and tree._revision_id in
 
869
                parents):
 
870
                return super(WorkingTree4, self).paths2ids(paths, trees, require_versioned)
 
871
        search_indexes = [0] + [1 + parents.index(tree._revision_id) for tree in trees]
 
872
        # -- make all paths utf8 --
 
873
        paths_utf8 = set()
 
874
        for path in paths:
 
875
            paths_utf8.add(path.encode('utf8'))
 
876
        paths = paths_utf8
 
877
        # -- paths is now a utf8 path set --
 
878
        # -- get the state object and prepare it.
 
879
        state = self.current_dirstate()
 
880
        if False and (state._dirblock_state == dirstate.DirState.NOT_IN_MEMORY
 
881
            and '' not in paths):
 
882
            paths2ids = self._paths2ids_using_bisect
 
883
        else:
 
884
            paths2ids = self._paths2ids_in_memory
 
885
        return paths2ids(paths, search_indexes,
 
886
                         require_versioned=require_versioned)
 
887
 
 
888
    def _paths2ids_in_memory(self, paths, search_indexes,
 
889
                             require_versioned=True):
 
890
        state = self.current_dirstate()
 
891
        state._read_dirblocks_if_needed()
 
892
        def _entries_for_path(path):
 
893
            """Return a list with all the entries that match path for all ids.
 
894
            """
 
895
            dirname, basename = os.path.split(path)
 
896
            key = (dirname, basename, '')
 
897
            block_index, present = state._find_block_index_from_key(key)
 
898
            if not present:
 
899
                # the block which should contain path is absent.
 
900
                return []
 
901
            result = []
 
902
            block = state._dirblocks[block_index][1]
 
903
            entry_index, _ = state._find_entry_index(key, block)
 
904
            # we may need to look at multiple entries at this path: walk while the paths match.
 
905
            while (entry_index < len(block) and
 
906
                block[entry_index][0][0:2] == key[0:2]):
 
907
                result.append(block[entry_index])
 
908
                entry_index += 1
 
909
            return result
 
910
        if require_versioned:
 
911
            # -- check all supplied paths are versioned in a search tree. --
 
912
            all_versioned = True
 
913
            for path in paths:
 
914
                path_entries = _entries_for_path(path)
 
915
                if not path_entries:
 
916
                    # this specified path is not present at all: error
 
917
                    all_versioned = False
 
918
                    break
 
919
                found_versioned = False
 
920
                # for each id at this path
 
921
                for entry in path_entries:
 
922
                    # for each tree.
 
923
                    for index in search_indexes:
 
924
                        if entry[1][index][0] != 'a': # absent
 
925
                            found_versioned = True
 
926
                            # all good: found a versioned cell
 
927
                            break
 
928
                if not found_versioned:
 
929
                    # none of the indexes was not 'absent' at all ids for this
 
930
                    # path.
 
931
                    all_versioned = False
 
932
                    break
 
933
            if not all_versioned:
 
934
                raise errors.PathsNotVersionedError(paths)
 
935
        # -- remove redundancy in supplied paths to prevent over-scanning --
 
936
        search_paths = osutils.minimum_path_selection(paths)
 
937
        # sketch: 
 
938
        # for all search_indexs in each path at or under each element of
 
939
        # search_paths, if the detail is relocated: add the id, and add the
 
940
        # relocated path as one to search if its not searched already. If the
 
941
        # detail is not relocated, add the id.
 
942
        searched_paths = set()
 
943
        found_ids = set()
 
944
        def _process_entry(entry):
 
945
            """Look at search_indexes within entry.
 
946
 
 
947
            If a specific tree's details are relocated, add the relocation
 
948
            target to search_paths if not searched already. If it is absent, do
 
949
            nothing. Otherwise add the id to found_ids.
 
950
            """
 
951
            for index in search_indexes:
 
952
                if entry[1][index][0] == 'r': # relocated
 
953
                    if not osutils.is_inside_any(searched_paths, entry[1][index][1]):
 
954
                        search_paths.add(entry[1][index][1])
 
955
                elif entry[1][index][0] != 'a': # absent
 
956
                    found_ids.add(entry[0][2])
 
957
        while search_paths:
 
958
            current_root = search_paths.pop()
 
959
            searched_paths.add(current_root)
 
960
            # process the entries for this containing directory: the rest will be
 
961
            # found by their parents recursively.
 
962
            root_entries = _entries_for_path(current_root)
 
963
            if not root_entries:
 
964
                # this specified path is not present at all, skip it.
 
965
                continue
 
966
            for entry in root_entries:
 
967
                _process_entry(entry)
 
968
            initial_key = (current_root, '', '')
 
969
            block_index, _ = state._find_block_index_from_key(initial_key)
 
970
            while (block_index < len(state._dirblocks) and
 
971
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
972
                for entry in state._dirblocks[block_index][1]:
 
973
                    _process_entry(entry)
 
974
                block_index += 1
 
975
        return found_ids
 
976
 
 
977
    def _paths2ids_using_bisect(self, paths, search_indexes,
 
978
                                require_versioned=True):
 
979
        state = self.current_dirstate()
 
980
        found_ids = set()
 
981
 
 
982
        split_paths = sorted(osutils.split(p) for p in paths)
 
983
        found = state._bisect_recursive(split_paths)
 
984
 
 
985
        if require_versioned:
 
986
            found_dir_names = set(dir_name_id[:2] for dir_name_id in found)
 
987
            for dir_name in split_paths:
 
988
                if dir_name not in found_dir_names:
 
989
                    raise errors.PathsNotVersionedError(paths)
 
990
 
 
991
        for dir_name_id, trees_info in found.iteritems():
 
992
            for index in search_indexes:
 
993
                if trees_info[index][0] not in ('r', 'a'):
 
994
                    found_ids.add(dir_name_id[2])
 
995
        return found_ids
 
996
 
 
997
    def read_working_inventory(self):
 
998
        """Read the working inventory.
 
999
        
 
1000
        This is a meaningless operation for dirstate, but we obey it anyhow.
 
1001
        """
 
1002
        return self.inventory
 
1003
 
 
1004
    @needs_read_lock
 
1005
    def revision_tree(self, revision_id):
 
1006
        """See Tree.revision_tree.
 
1007
 
 
1008
        WorkingTree4 supplies revision_trees for any basis tree.
 
1009
        """
 
1010
        dirstate = self.current_dirstate()
 
1011
        parent_ids = dirstate.get_parent_ids()
 
1012
        if revision_id not in parent_ids:
 
1013
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1014
        if revision_id in dirstate.get_ghosts():
 
1015
            raise errors.NoSuchRevisionInTree(self, revision_id)
 
1016
        return DirStateRevisionTree(dirstate, revision_id,
 
1017
            self.branch.repository)
 
1018
 
 
1019
    @needs_tree_write_lock
 
1020
    def set_last_revision(self, new_revision):
 
1021
        """Change the last revision in the working tree."""
 
1022
        parents = self.get_parent_ids()
 
1023
        if new_revision in (NULL_REVISION, None):
 
1024
            assert len(parents) < 2, (
 
1025
                "setting the last parent to none with a pending merge is "
 
1026
                "unsupported.")
 
1027
            self.set_parent_ids([])
 
1028
        else:
 
1029
            self.set_parent_ids([new_revision] + parents[1:],
 
1030
                allow_leftmost_as_ghost=True)
 
1031
 
 
1032
    @needs_tree_write_lock
 
1033
    def set_parent_ids(self, revision_ids, allow_leftmost_as_ghost=False):
 
1034
        """Set the parent ids to revision_ids.
 
1035
        
 
1036
        See also set_parent_trees. This api will try to retrieve the tree data
 
1037
        for each element of revision_ids from the trees repository. If you have
 
1038
        tree data already available, it is more efficient to use
 
1039
        set_parent_trees rather than set_parent_ids. set_parent_ids is however
 
1040
        an easier API to use.
 
1041
 
 
1042
        :param revision_ids: The revision_ids to set as the parent ids of this
 
1043
            working tree. Any of these may be ghosts.
 
1044
        """
 
1045
        trees = []
 
1046
        for revision_id in revision_ids:
 
1047
            try:
 
1048
                revtree = self.branch.repository.revision_tree(revision_id)
 
1049
                # TODO: jam 20070213 KnitVersionedFile raises
 
1050
                #       RevisionNotPresent rather than NoSuchRevision if a
 
1051
                #       given revision_id is not present. Should Repository be
 
1052
                #       catching it and re-raising NoSuchRevision?
 
1053
            except (errors.NoSuchRevision, errors.RevisionNotPresent):
 
1054
                revtree = None
 
1055
            trees.append((revision_id, revtree))
 
1056
        self.set_parent_trees(trees,
 
1057
            allow_leftmost_as_ghost=allow_leftmost_as_ghost)
 
1058
 
 
1059
    @needs_tree_write_lock
 
1060
    def set_parent_trees(self, parents_list, allow_leftmost_as_ghost=False):
 
1061
        """Set the parents of the working tree.
 
1062
 
 
1063
        :param parents_list: A list of (revision_id, tree) tuples.
 
1064
            If tree is None, then that element is treated as an unreachable
 
1065
            parent tree - i.e. a ghost.
 
1066
        """
 
1067
        dirstate = self.current_dirstate()
 
1068
        if len(parents_list) > 0:
 
1069
            if not allow_leftmost_as_ghost and parents_list[0][1] is None:
 
1070
                raise errors.GhostRevisionUnusableHere(parents_list[0][0])
 
1071
        real_trees = []
 
1072
        ghosts = []
 
1073
        # convert absent trees to the null tree, which we convert back to
 
1074
        # missing on access.
 
1075
        for rev_id, tree in parents_list:
 
1076
            _mod_revision.check_not_reserved_id(rev_id)
 
1077
            if tree is not None:
 
1078
                real_trees.append((rev_id, tree))
 
1079
            else:
 
1080
                real_trees.append((rev_id,
 
1081
                    self.branch.repository.revision_tree(None)))
 
1082
                ghosts.append(rev_id)
 
1083
        dirstate.set_parent_trees(real_trees, ghosts=ghosts)
 
1084
        self._make_dirty(reset_inventory=False)
 
1085
 
 
1086
    def _set_root_id(self, file_id):
 
1087
        """See WorkingTree.set_root_id."""
 
1088
        state = self.current_dirstate()
 
1089
        state.set_path_id('', file_id)
 
1090
        if state._dirblock_state == dirstate.DirState.IN_MEMORY_MODIFIED:
 
1091
            self._make_dirty(reset_inventory=True)
 
1092
 
 
1093
    def _sha_from_stat(self, path, stat_result):
 
1094
        """Get a sha digest from the tree's stat cache.
 
1095
 
 
1096
        The default implementation assumes no stat cache is present.
 
1097
 
 
1098
        :param path: The path.
 
1099
        :param stat_result: The stat result being looked up.
 
1100
        """
 
1101
        return self.current_dirstate().sha1_from_stat(path, stat_result)
 
1102
 
 
1103
    @needs_read_lock
 
1104
    def supports_tree_reference(self):
 
1105
        return self._repo_supports_tree_reference
 
1106
 
 
1107
    def unlock(self):
 
1108
        """Unlock in format 4 trees needs to write the entire dirstate."""
 
1109
        # do non-implementation specific cleanup
 
1110
        self._cleanup()
 
1111
 
 
1112
        if self._control_files._lock_count == 1:
 
1113
            # eventually we should do signature checking during read locks for
 
1114
            # dirstate updates.
 
1115
            if self._control_files._lock_mode == 'w':
 
1116
                if self._dirty:
 
1117
                    self.flush()
 
1118
            if self._dirstate is not None:
 
1119
                # This is a no-op if there are no modifications.
 
1120
                self._dirstate.save()
 
1121
                self._dirstate.unlock()
 
1122
            # TODO: jam 20070301 We shouldn't have to wipe the dirstate at this
 
1123
            #       point. Instead, it could check if the header has been
 
1124
            #       modified when it is locked, and if not, it can hang on to
 
1125
            #       the data it has in memory.
 
1126
            self._dirstate = None
 
1127
            self._inventory = None
 
1128
        # reverse order of locking.
 
1129
        try:
 
1130
            return self._control_files.unlock()
 
1131
        finally:
 
1132
            self.branch.unlock()
 
1133
 
 
1134
    @needs_tree_write_lock
 
1135
    def unversion(self, file_ids):
 
1136
        """Remove the file ids in file_ids from the current versioned set.
 
1137
 
 
1138
        When a file_id is unversioned, all of its children are automatically
 
1139
        unversioned.
 
1140
 
 
1141
        :param file_ids: The file ids to stop versioning.
 
1142
        :raises: NoSuchId if any fileid is not currently versioned.
 
1143
        """
 
1144
        if not file_ids:
 
1145
            return
 
1146
        state = self.current_dirstate()
 
1147
        state._read_dirblocks_if_needed()
 
1148
        ids_to_unversion = set(file_ids)
 
1149
        paths_to_unversion = set()
 
1150
        # sketch:
 
1151
        # check if the root is to be unversioned, if so, assert for now.
 
1152
        # walk the state marking unversioned things as absent.
 
1153
        # if there are any un-unversioned ids at the end, raise
 
1154
        for key, details in state._dirblocks[0][1]:
 
1155
            if (details[0][0] not in ('a', 'r') and # absent or relocated
 
1156
                key[2] in ids_to_unversion):
 
1157
                # I haven't written the code to unversion / yet - it should be
 
1158
                # supported.
 
1159
                raise errors.BzrError('Unversioning the / is not currently supported')
 
1160
        block_index = 0
 
1161
        while block_index < len(state._dirblocks):
 
1162
            # process one directory at a time.
 
1163
            block = state._dirblocks[block_index]
 
1164
            # first check: is the path one to remove - it or its children
 
1165
            delete_block = False
 
1166
            for path in paths_to_unversion:
 
1167
                if (block[0].startswith(path) and
 
1168
                    (len(block[0]) == len(path) or
 
1169
                     block[0][len(path)] == '/')):
 
1170
                    # this entire block should be deleted - its the block for a
 
1171
                    # path to unversion; or the child of one
 
1172
                    delete_block = True
 
1173
                    break
 
1174
            # TODO: trim paths_to_unversion as we pass by paths
 
1175
            if delete_block:
 
1176
                # this block is to be deleted: process it.
 
1177
                # TODO: we can special case the no-parents case and
 
1178
                # just forget the whole block.
 
1179
                entry_index = 0
 
1180
                while entry_index < len(block[1]):
 
1181
                    # Mark this file id as having been removed
 
1182
                    entry = block[1][entry_index]
 
1183
                    ids_to_unversion.discard(entry[0][2])
 
1184
                    if (entry[1][0][0] in 'ar' # don't remove absent or renamed
 
1185
                                               # entries
 
1186
                        or not state._make_absent(entry)):
 
1187
                        entry_index += 1
 
1188
                # go to the next block. (At the moment we dont delete empty
 
1189
                # dirblocks)
 
1190
                block_index += 1
 
1191
                continue
 
1192
            entry_index = 0
 
1193
            while entry_index < len(block[1]):
 
1194
                entry = block[1][entry_index]
 
1195
                if (entry[1][0][0] in ('a', 'r') or # absent, relocated
 
1196
                    # ^ some parent row.
 
1197
                    entry[0][2] not in ids_to_unversion):
 
1198
                    # ^ not an id to unversion
 
1199
                    entry_index += 1
 
1200
                    continue
 
1201
                if entry[1][0][0] == 'd':
 
1202
                    paths_to_unversion.add(pathjoin(entry[0][0], entry[0][1]))
 
1203
                if not state._make_absent(entry):
 
1204
                    entry_index += 1
 
1205
                # we have unversioned this id
 
1206
                ids_to_unversion.remove(entry[0][2])
 
1207
            block_index += 1
 
1208
        if ids_to_unversion:
 
1209
            raise errors.NoSuchId(self, iter(ids_to_unversion).next())
 
1210
        self._make_dirty(reset_inventory=False)
 
1211
        # have to change the legacy inventory too.
 
1212
        if self._inventory is not None:
 
1213
            for file_id in file_ids:
 
1214
                self._inventory.remove_recursive_id(file_id)
 
1215
 
 
1216
    def update_basis_by_delta(self, new_revid, delta):
 
1217
        """See MutableTree.update_basis_by_delta."""
 
1218
        assert self.last_revision() != new_revid
 
1219
        self.current_dirstate().update_basis_by_delta(delta, new_revid)
 
1220
 
 
1221
    @needs_read_lock
 
1222
    def _validate(self):
 
1223
        self._dirstate._validate()
 
1224
 
 
1225
    @needs_tree_write_lock
 
1226
    def _write_inventory(self, inv):
 
1227
        """Write inventory as the current inventory."""
 
1228
        assert not self._dirty, ("attempting to write an inventory when the "
 
1229
            "dirstate is dirty will cause data loss")
 
1230
        self.current_dirstate().set_state_from_inventory(inv)
 
1231
        self._make_dirty(reset_inventory=False)
 
1232
        if self._inventory is not None:
 
1233
            self._inventory = inv
 
1234
        self.flush()
 
1235
 
 
1236
 
 
1237
class WorkingTreeFormat4(WorkingTreeFormat3):
 
1238
    """The first consolidated dirstate working tree format.
 
1239
 
 
1240
    This format:
 
1241
        - exists within a metadir controlling .bzr
 
1242
        - includes an explicit version marker for the workingtree control
 
1243
          files, separate from the BzrDir format
 
1244
        - modifies the hash cache format
 
1245
        - is new in bzr 0.15
 
1246
        - uses a LockDir to guard access to it.
 
1247
    """
 
1248
 
 
1249
    upgrade_recommended = False
 
1250
 
 
1251
    def get_format_string(self):
 
1252
        """See WorkingTreeFormat.get_format_string()."""
 
1253
        return "Bazaar Working Tree Format 4 (bzr 0.15)\n"
 
1254
 
 
1255
    def get_format_description(self):
 
1256
        """See WorkingTreeFormat.get_format_description()."""
 
1257
        return "Working tree format 4"
 
1258
 
 
1259
    def initialize(self, a_bzrdir, revision_id=None, from_branch=None,
 
1260
                   accelerator_tree=None):
 
1261
        """See WorkingTreeFormat.initialize().
 
1262
 
 
1263
        :param revision_id: allows creating a working tree at a different
 
1264
        revision than the branch is at.
 
1265
 
 
1266
        These trees get an initial random root id, if their repository supports
 
1267
        rich root data, TREE_ROOT otherwise.
 
1268
        """
 
1269
        if not isinstance(a_bzrdir.transport, LocalTransport):
 
1270
            raise errors.NotLocalUrl(a_bzrdir.transport.base)
 
1271
        transport = a_bzrdir.get_workingtree_transport(self)
 
1272
        control_files = self._open_control_files(a_bzrdir)
 
1273
        control_files.create_lock()
 
1274
        control_files.lock_write()
 
1275
        control_files.put_utf8('format', self.get_format_string())
 
1276
        if from_branch is not None:
 
1277
            branch = from_branch
 
1278
        else:
 
1279
            branch = a_bzrdir.open_branch()
 
1280
        if revision_id is None:
 
1281
            revision_id = branch.last_revision()
 
1282
        local_path = transport.local_abspath('dirstate')
 
1283
        # write out new dirstate (must exist when we create the tree)
 
1284
        state = dirstate.DirState.initialize(local_path)
 
1285
        state.unlock()
 
1286
        del state
 
1287
        wt = WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1288
                         branch,
 
1289
                         _format=self,
 
1290
                         _bzrdir=a_bzrdir,
 
1291
                         _control_files=control_files)
 
1292
        wt._new_tree()
 
1293
        wt.lock_tree_write()
 
1294
        try:
 
1295
            if revision_id in (None, NULL_REVISION):
 
1296
                if branch.repository.supports_rich_root():
 
1297
                    wt._set_root_id(generate_ids.gen_root_id())
 
1298
                else:
 
1299
                    wt._set_root_id(ROOT_ID)
 
1300
                wt.flush()
 
1301
            wt.set_last_revision(revision_id)
 
1302
            wt.flush()
 
1303
            basis = wt.basis_tree()
 
1304
            basis.lock_read()
 
1305
            # if the basis has a root id we have to use that; otherwise we use
 
1306
            # a new random one
 
1307
            basis_root_id = basis.get_root_id()
 
1308
            if basis_root_id is not None:
 
1309
                wt._set_root_id(basis_root_id)
 
1310
                wt.flush()
 
1311
            transform.build_tree(basis, wt, accelerator_tree)
 
1312
            basis.unlock()
 
1313
        finally:
 
1314
            control_files.unlock()
 
1315
            wt.unlock()
 
1316
        return wt
 
1317
 
 
1318
    def _open(self, a_bzrdir, control_files):
 
1319
        """Open the tree itself.
 
1320
 
 
1321
        :param a_bzrdir: the dir for the tree.
 
1322
        :param control_files: the control files for the tree.
 
1323
        """
 
1324
        return WorkingTree4(a_bzrdir.root_transport.local_abspath('.'),
 
1325
                           branch=a_bzrdir.open_branch(),
 
1326
                           _format=self,
 
1327
                           _bzrdir=a_bzrdir,
 
1328
                           _control_files=control_files)
 
1329
 
 
1330
    def __get_matchingbzrdir(self):
 
1331
        # please test against something that will let us do tree references
 
1332
        return bzrdir.format_registry.make_bzrdir(
 
1333
            'dirstate-with-subtree')
 
1334
 
 
1335
    _matchingbzrdir = property(__get_matchingbzrdir)
 
1336
 
 
1337
 
 
1338
class DirStateRevisionTree(Tree):
 
1339
    """A revision tree pulling the inventory from a dirstate."""
 
1340
 
 
1341
    def __init__(self, dirstate, revision_id, repository):
 
1342
        self._dirstate = dirstate
 
1343
        self._revision_id = revision_id
 
1344
        self._repository = repository
 
1345
        self._inventory = None
 
1346
        self._locked = 0
 
1347
        self._dirstate_locked = False
 
1348
 
 
1349
    def __repr__(self):
 
1350
        return "<%s of %s in %s>" % \
 
1351
            (self.__class__.__name__, self._revision_id, self._dirstate)
 
1352
 
 
1353
    def annotate_iter(self, file_id,
 
1354
                      default_revision=_mod_revision.CURRENT_REVISION):
 
1355
        """See Tree.annotate_iter"""
 
1356
        w = self._get_weave(file_id)
 
1357
        return w.annotate_iter(self.inventory[file_id].revision)
 
1358
 
 
1359
    def _get_ancestors(self, default_revision):
 
1360
        return set(self._repository.get_ancestry(self._revision_id,
 
1361
                                                 topo_sorted=False))
 
1362
    def _comparison_data(self, entry, path):
 
1363
        """See Tree._comparison_data."""
 
1364
        if entry is None:
 
1365
            return None, False, None
 
1366
        # trust the entry as RevisionTree does, but this may not be
 
1367
        # sensible: the entry might not have come from us?
 
1368
        return entry.kind, entry.executable, None
 
1369
 
 
1370
    def _file_size(self, entry, stat_value):
 
1371
        return entry.text_size
 
1372
 
 
1373
    def filter_unversioned_files(self, paths):
 
1374
        """Filter out paths that are not versioned.
 
1375
 
 
1376
        :return: set of paths.
 
1377
        """
 
1378
        pred = self.has_filename
 
1379
        return set((p for p in paths if not pred(p)))
 
1380
 
 
1381
    def get_root_id(self):
 
1382
        return self.path2id('')
 
1383
 
 
1384
    def id2path(self, file_id):
 
1385
        "Convert a file-id to a path."
 
1386
        entry = self._get_entry(file_id=file_id)
 
1387
        if entry == (None, None):
 
1388
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1389
        path_utf8 = osutils.pathjoin(entry[0][0], entry[0][1])
 
1390
        return path_utf8.decode('utf8')
 
1391
 
 
1392
    def _get_parent_index(self):
 
1393
        """Return the index in the dirstate referenced by this tree."""
 
1394
        return self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1395
 
 
1396
    def _get_entry(self, file_id=None, path=None):
 
1397
        """Get the dirstate row for file_id or path.
 
1398
 
 
1399
        If either file_id or path is supplied, it is used as the key to lookup.
 
1400
        If both are supplied, the fastest lookup is used, and an error is
 
1401
        raised if they do not both point at the same row.
 
1402
        
 
1403
        :param file_id: An optional unicode file_id to be looked up.
 
1404
        :param path: An optional unicode path to be looked up.
 
1405
        :return: The dirstate row tuple for path/file_id, or (None, None)
 
1406
        """
 
1407
        if file_id is None and path is None:
 
1408
            raise errors.BzrError('must supply file_id or path')
 
1409
        if path is not None:
 
1410
            path = path.encode('utf8')
 
1411
        parent_index = self._get_parent_index()
 
1412
        return self._dirstate._get_entry(parent_index, fileid_utf8=file_id, path_utf8=path)
 
1413
 
 
1414
    def _generate_inventory(self):
 
1415
        """Create and set self.inventory from the dirstate object.
 
1416
 
 
1417
        (So this is only called the first time the inventory is requested for
 
1418
        this tree; it then remains in memory until it's out of date.)
 
1419
 
 
1420
        This is relatively expensive: we have to walk the entire dirstate.
 
1421
        """
 
1422
        assert self._locked, 'cannot generate inventory of an unlocked '\
 
1423
            'dirstate revision tree'
 
1424
        # separate call for profiling - makes it clear where the costs are.
 
1425
        self._dirstate._read_dirblocks_if_needed()
 
1426
        assert self._revision_id in self._dirstate.get_parent_ids(), \
 
1427
            'parent %s has disappeared from %s' % (
 
1428
            self._revision_id, self._dirstate.get_parent_ids())
 
1429
        parent_index = self._dirstate.get_parent_ids().index(self._revision_id) + 1
 
1430
        # This is identical now to the WorkingTree _generate_inventory except
 
1431
        # for the tree index use.
 
1432
        root_key, current_entry = self._dirstate._get_entry(parent_index, path_utf8='')
 
1433
        current_id = root_key[2]
 
1434
        assert current_entry[parent_index][0] == 'd'
 
1435
        inv = Inventory(root_id=current_id, revision_id=self._revision_id)
 
1436
        inv.root.revision = current_entry[parent_index][4]
 
1437
        # Turn some things into local variables
 
1438
        minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1439
        factory = entry_factory
 
1440
        utf8_decode = cache_utf8._utf8_decode
 
1441
        inv_byid = inv._byid
 
1442
        # we could do this straight out of the dirstate; it might be fast
 
1443
        # and should be profiled - RBC 20070216
 
1444
        parent_ies = {'' : inv.root}
 
1445
        for block in self._dirstate._dirblocks[1:]: #skip root
 
1446
            dirname = block[0]
 
1447
            try:
 
1448
                parent_ie = parent_ies[dirname]
 
1449
            except KeyError:
 
1450
                # all the paths in this block are not versioned in this tree
 
1451
                continue
 
1452
            for key, entry in block[1]:
 
1453
                minikind, fingerprint, size, executable, revid = entry[parent_index]
 
1454
                if minikind in ('a', 'r'): # absent, relocated
 
1455
                    # not this tree
 
1456
                    continue
 
1457
                name = key[1]
 
1458
                name_unicode = utf8_decode(name)[0]
 
1459
                file_id = key[2]
 
1460
                kind = minikind_to_kind[minikind]
 
1461
                inv_entry = factory[kind](file_id, name_unicode,
 
1462
                                          parent_ie.file_id)
 
1463
                inv_entry.revision = revid
 
1464
                if kind == 'file':
 
1465
                    inv_entry.executable = executable
 
1466
                    inv_entry.text_size = size
 
1467
                    inv_entry.text_sha1 = fingerprint
 
1468
                elif kind == 'directory':
 
1469
                    parent_ies[(dirname + '/' + name).strip('/')] = inv_entry
 
1470
                elif kind == 'symlink':
 
1471
                    inv_entry.executable = False
 
1472
                    inv_entry.text_size = None
 
1473
                    inv_entry.symlink_target = utf8_decode(fingerprint)[0]
 
1474
                elif kind == 'tree-reference':
 
1475
                    inv_entry.reference_revision = fingerprint or None
 
1476
                else:
 
1477
                    raise AssertionError("cannot convert entry %r into an InventoryEntry"
 
1478
                            % entry)
 
1479
                # These checks cost us around 40ms on a 55k entry tree
 
1480
                assert file_id not in inv_byid
 
1481
                assert name_unicode not in parent_ie.children
 
1482
                inv_byid[file_id] = inv_entry
 
1483
                parent_ie.children[name_unicode] = inv_entry
 
1484
        self._inventory = inv
 
1485
 
 
1486
    def get_file_mtime(self, file_id, path=None):
 
1487
        """Return the modification time for this record.
 
1488
 
 
1489
        We return the timestamp of the last-changed revision.
 
1490
        """
 
1491
        # Make sure the file exists
 
1492
        entry = self._get_entry(file_id, path=path)
 
1493
        if entry == (None, None): # do we raise?
 
1494
            return None
 
1495
        parent_index = self._get_parent_index()
 
1496
        last_changed_revision = entry[1][parent_index][4]
 
1497
        return self._repository.get_revision(last_changed_revision).timestamp
 
1498
 
 
1499
    def get_file_sha1(self, file_id, path=None, stat_value=None):
 
1500
        entry = self._get_entry(file_id=file_id, path=path)
 
1501
        parent_index = self._get_parent_index()
 
1502
        parent_details = entry[1][parent_index]
 
1503
        if parent_details[0] == 'f':
 
1504
            return parent_details[1]
 
1505
        return None
 
1506
 
 
1507
    @symbol_versioning.deprecated_method(symbol_versioning.zero_ninety)
 
1508
    def get_weave(self, file_id):
 
1509
        return self._get_weave(file_id)
 
1510
 
 
1511
    def _get_weave(self, file_id):
 
1512
        return self._repository.weave_store.get_weave(file_id,
 
1513
                self._repository.get_transaction())
 
1514
 
 
1515
    def get_file(self, file_id, path=None):
 
1516
        return StringIO(self.get_file_text(file_id))
 
1517
 
 
1518
    def get_file_lines(self, file_id):
 
1519
        entry = self._get_entry(file_id=file_id)[1]
 
1520
        if entry == None:
 
1521
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1522
        return self._get_weave(file_id).get_lines(entry[1][4])
 
1523
 
 
1524
    def get_file_size(self, file_id):
 
1525
        return self.inventory[file_id].text_size
 
1526
 
 
1527
    def get_file_text(self, file_id):
 
1528
        return ''.join(self.get_file_lines(file_id))
 
1529
 
 
1530
    def get_reference_revision(self, file_id, path=None):
 
1531
        return self.inventory[file_id].reference_revision
 
1532
 
 
1533
    def iter_files_bytes(self, desired_files):
 
1534
        """See Tree.iter_files_bytes.
 
1535
 
 
1536
        This version is implemented on top of Repository.iter_files_bytes"""
 
1537
        parent_index = self._get_parent_index()
 
1538
        repo_desired_files = []
 
1539
        for file_id, identifier in desired_files:
 
1540
            entry = self._get_entry(file_id)
 
1541
            if entry == (None, None):
 
1542
                raise errors.NoSuchId(self, file_id)
 
1543
            repo_desired_files.append((file_id, entry[1][parent_index][4],
 
1544
                                       identifier))
 
1545
        return self._repository.iter_files_bytes(repo_desired_files)
 
1546
 
 
1547
    def get_symlink_target(self, file_id):
 
1548
        entry = self._get_entry(file_id=file_id)
 
1549
        parent_index = self._get_parent_index()
 
1550
        if entry[1][parent_index][0] != 'l':
 
1551
            return None
 
1552
        else:
 
1553
            # At present, none of the tree implementations supports non-ascii
 
1554
            # symlink targets. So we will just assume that the dirstate path is
 
1555
            # correct.
 
1556
            return entry[1][parent_index][1]
 
1557
 
 
1558
    def get_revision_id(self):
 
1559
        """Return the revision id for this tree."""
 
1560
        return self._revision_id
 
1561
 
 
1562
    def _get_inventory(self):
 
1563
        if self._inventory is not None:
 
1564
            return self._inventory
 
1565
        self._must_be_locked()
 
1566
        self._generate_inventory()
 
1567
        return self._inventory
 
1568
 
 
1569
    inventory = property(_get_inventory,
 
1570
                         doc="Inventory of this Tree")
 
1571
 
 
1572
    def get_parent_ids(self):
 
1573
        """The parents of a tree in the dirstate are not cached."""
 
1574
        return self._repository.get_revision(self._revision_id).parent_ids
 
1575
 
 
1576
    def has_filename(self, filename):
 
1577
        return bool(self.path2id(filename))
 
1578
 
 
1579
    def kind(self, file_id):
 
1580
        entry = self._get_entry(file_id=file_id)[1]
 
1581
        if entry == None:
 
1582
            raise errors.NoSuchId(tree=self, file_id=file_id)
 
1583
        return dirstate.DirState._minikind_to_kind[entry[1][0]]
 
1584
 
 
1585
    def path_content_summary(self, path):
 
1586
        """See Tree.path_content_summary."""
 
1587
        id = self.inventory.path2id(path)
 
1588
        if id is None:
 
1589
            return ('missing', None, None, None)
 
1590
        entry = self._inventory[id]
 
1591
        kind = entry.kind
 
1592
        if kind == 'file':
 
1593
            return (kind, entry.text_size, entry.executable, entry.text_sha1)
 
1594
        elif kind == 'symlink':
 
1595
            return (kind, None, None, entry.symlink_target)
 
1596
        else:
 
1597
            return (kind, None, None, None)
 
1598
 
 
1599
    def is_executable(self, file_id, path=None):
 
1600
        ie = self.inventory[file_id]
 
1601
        if ie.kind != "file":
 
1602
            return None
 
1603
        return ie.executable
 
1604
 
 
1605
    def list_files(self, include_root=False):
 
1606
        # We use a standard implementation, because DirStateRevisionTree is
 
1607
        # dealing with one of the parents of the current state
 
1608
        inv = self._get_inventory()
 
1609
        entries = inv.iter_entries()
 
1610
        if self.inventory.root is not None and not include_root:
 
1611
            entries.next()
 
1612
        for path, entry in entries:
 
1613
            yield path, 'V', entry.kind, entry.file_id, entry
 
1614
 
 
1615
    def lock_read(self):
 
1616
        """Lock the tree for a set of operations."""
 
1617
        if not self._locked:
 
1618
            self._repository.lock_read()
 
1619
            if self._dirstate._lock_token is None:
 
1620
                self._dirstate.lock_read()
 
1621
                self._dirstate_locked = True
 
1622
        self._locked += 1
 
1623
 
 
1624
    def _must_be_locked(self):
 
1625
        if not self._locked:
 
1626
            raise errors.ObjectNotLocked(self)
 
1627
 
 
1628
    @needs_read_lock
 
1629
    def path2id(self, path):
 
1630
        """Return the id for path in this tree."""
 
1631
        # lookup by path: faster than splitting and walking the ivnentory.
 
1632
        entry = self._get_entry(path=path)
 
1633
        if entry == (None, None):
 
1634
            return None
 
1635
        return entry[0][2]
 
1636
 
 
1637
    def unlock(self):
 
1638
        """Unlock, freeing any cache memory used during the lock."""
 
1639
        # outside of a lock, the inventory is suspect: release it.
 
1640
        self._locked -=1
 
1641
        if not self._locked:
 
1642
            self._inventory = None
 
1643
            self._locked = 0
 
1644
            if self._dirstate_locked:
 
1645
                self._dirstate.unlock()
 
1646
                self._dirstate_locked = False
 
1647
            self._repository.unlock()
 
1648
 
 
1649
    def walkdirs(self, prefix=""):
 
1650
        # TODO: jam 20070215 This is the lazy way by using the RevisionTree
 
1651
        # implementation based on an inventory.  
 
1652
        # This should be cleaned up to use the much faster Dirstate code
 
1653
        # So for now, we just build up the parent inventory, and extract
 
1654
        # it the same way RevisionTree does.
 
1655
        _directory = 'directory'
 
1656
        inv = self._get_inventory()
 
1657
        top_id = inv.path2id(prefix)
 
1658
        if top_id is None:
 
1659
            pending = []
 
1660
        else:
 
1661
            pending = [(prefix, top_id)]
 
1662
        while pending:
 
1663
            dirblock = []
 
1664
            relpath, file_id = pending.pop()
 
1665
            # 0 - relpath, 1- file-id
 
1666
            if relpath:
 
1667
                relroot = relpath + '/'
 
1668
            else:
 
1669
                relroot = ""
 
1670
            # FIXME: stash the node in pending
 
1671
            entry = inv[file_id]
 
1672
            for name, child in entry.sorted_children():
 
1673
                toppath = relroot + name
 
1674
                dirblock.append((toppath, name, child.kind, None,
 
1675
                    child.file_id, child.kind
 
1676
                    ))
 
1677
            yield (relpath, entry.file_id), dirblock
 
1678
            # push the user specified dirs from dirblock
 
1679
            for dir in reversed(dirblock):
 
1680
                if dir[2] == _directory:
 
1681
                    pending.append((dir[0], dir[4]))
 
1682
 
 
1683
 
 
1684
class InterDirStateTree(InterTree):
 
1685
    """Fast path optimiser for changes_from with dirstate trees.
 
1686
    
 
1687
    This is used only when both trees are in the dirstate working file, and 
 
1688
    the source is any parent within the dirstate, and the destination is 
 
1689
    the current working tree of the same dirstate.
 
1690
    """
 
1691
    # this could be generalized to allow comparisons between any trees in the
 
1692
    # dirstate, and possibly between trees stored in different dirstates.
 
1693
 
 
1694
    def __init__(self, source, target):
 
1695
        super(InterDirStateTree, self).__init__(source, target)
 
1696
        if not InterDirStateTree.is_compatible(source, target):
 
1697
            raise Exception, "invalid source %r and target %r" % (source, target)
 
1698
 
 
1699
    @staticmethod
 
1700
    def make_source_parent_tree(source, target):
 
1701
        """Change the source tree into a parent of the target."""
 
1702
        revid = source.commit('record tree')
 
1703
        target.branch.repository.fetch(source.branch.repository, revid)
 
1704
        target.set_parent_ids([revid])
 
1705
        return target.basis_tree(), target
 
1706
 
 
1707
    _matching_from_tree_format = WorkingTreeFormat4()
 
1708
    _matching_to_tree_format = WorkingTreeFormat4()
 
1709
    _test_mutable_trees_to_test_trees = make_source_parent_tree
 
1710
 
 
1711
    def _iter_changes(self, include_unchanged=False,
 
1712
                      specific_files=None, pb=None, extra_trees=[],
 
1713
                      require_versioned=True, want_unversioned=False):
 
1714
        """Return the changes from source to target.
 
1715
 
 
1716
        :return: An iterator that yields tuples. See InterTree._iter_changes
 
1717
            for details.
 
1718
        :param specific_files: An optional list of file paths to restrict the
 
1719
            comparison to. When mapping filenames to ids, all matches in all
 
1720
            trees (including optional extra_trees) are used, and all children of
 
1721
            matched directories are included.
 
1722
        :param include_unchanged: An optional boolean requesting the inclusion of
 
1723
            unchanged entries in the result.
 
1724
        :param extra_trees: An optional list of additional trees to use when
 
1725
            mapping the contents of specific_files (paths) to file_ids.
 
1726
        :param require_versioned: If True, all files in specific_files must be
 
1727
            versioned in one of source, target, extra_trees or
 
1728
            PathsNotVersionedError is raised.
 
1729
        :param want_unversioned: Should unversioned files be returned in the
 
1730
            output. An unversioned file is defined as one with (False, False)
 
1731
            for the versioned pair.
 
1732
        """
 
1733
        utf8_decode = cache_utf8._utf8_decode
 
1734
        _minikind_to_kind = dirstate.DirState._minikind_to_kind
 
1735
        cmp_by_dirs = dirstate.cmp_by_dirs
 
1736
        # NB: show_status depends on being able to pass in non-versioned files
 
1737
        # and report them as unknown
 
1738
        # TODO: handle extra trees in the dirstate.
 
1739
        # TODO: handle comparisons as an empty tree as a different special
 
1740
        # case? mbp 20070226
 
1741
        if (extra_trees or (self.source._revision_id == NULL_REVISION)
 
1742
            or specific_files == []):
 
1743
            # we can't fast-path these cases (yet)
 
1744
            for f in super(InterDirStateTree, self)._iter_changes(
 
1745
                include_unchanged, specific_files, pb, extra_trees,
 
1746
                require_versioned, want_unversioned=want_unversioned):
 
1747
                yield f
 
1748
            return
 
1749
        parent_ids = self.target.get_parent_ids()
 
1750
        assert (self.source._revision_id in parent_ids), \
 
1751
                "revision {%s} is not stored in {%s}, but %s " \
 
1752
                "can only be used for trees stored in the dirstate" \
 
1753
                % (self.source._revision_id, self.target, self._iter_changes)
 
1754
        target_index = 0
 
1755
        if self.source._revision_id == NULL_REVISION:
 
1756
            source_index = None
 
1757
            indices = (target_index,)
 
1758
        else:
 
1759
            assert (self.source._revision_id in parent_ids), \
 
1760
                "Failure: source._revision_id: %s not in target.parent_ids(%s)" % (
 
1761
                self.source._revision_id, parent_ids)
 
1762
            source_index = 1 + parent_ids.index(self.source._revision_id)
 
1763
            indices = (source_index,target_index)
 
1764
        # -- make all specific_files utf8 --
 
1765
        if specific_files:
 
1766
            specific_files_utf8 = set()
 
1767
            for path in specific_files:
 
1768
                specific_files_utf8.add(path.encode('utf8'))
 
1769
            specific_files = specific_files_utf8
 
1770
        else:
 
1771
            specific_files = set([''])
 
1772
        # -- specific_files is now a utf8 path set --
 
1773
        # -- get the state object and prepare it.
 
1774
        state = self.target.current_dirstate()
 
1775
        state._read_dirblocks_if_needed()
 
1776
        def _entries_for_path(path):
 
1777
            """Return a list with all the entries that match path for all ids.
 
1778
            """
 
1779
            dirname, basename = os.path.split(path)
 
1780
            key = (dirname, basename, '')
 
1781
            block_index, present = state._find_block_index_from_key(key)
 
1782
            if not present:
 
1783
                # the block which should contain path is absent.
 
1784
                return []
 
1785
            result = []
 
1786
            block = state._dirblocks[block_index][1]
 
1787
            entry_index, _ = state._find_entry_index(key, block)
 
1788
            # we may need to look at multiple entries at this path: walk while the specific_files match.
 
1789
            while (entry_index < len(block) and
 
1790
                block[entry_index][0][0:2] == key[0:2]):
 
1791
                result.append(block[entry_index])
 
1792
                entry_index += 1
 
1793
            return result
 
1794
        if require_versioned:
 
1795
            # -- check all supplied paths are versioned in a search tree. --
 
1796
            all_versioned = True
 
1797
            for path in specific_files:
 
1798
                path_entries = _entries_for_path(path)
 
1799
                if not path_entries:
 
1800
                    # this specified path is not present at all: error
 
1801
                    all_versioned = False
 
1802
                    break
 
1803
                found_versioned = False
 
1804
                # for each id at this path
 
1805
                for entry in path_entries:
 
1806
                    # for each tree.
 
1807
                    for index in indices:
 
1808
                        if entry[1][index][0] != 'a': # absent
 
1809
                            found_versioned = True
 
1810
                            # all good: found a versioned cell
 
1811
                            break
 
1812
                if not found_versioned:
 
1813
                    # none of the indexes was not 'absent' at all ids for this
 
1814
                    # path.
 
1815
                    all_versioned = False
 
1816
                    break
 
1817
            if not all_versioned:
 
1818
                raise errors.PathsNotVersionedError(specific_files)
 
1819
        # -- remove redundancy in supplied specific_files to prevent over-scanning --
 
1820
        search_specific_files = set()
 
1821
        for path in specific_files:
 
1822
            other_specific_files = specific_files.difference(set([path]))
 
1823
            if not osutils.is_inside_any(other_specific_files, path):
 
1824
                # this is a top level path, we must check it.
 
1825
                search_specific_files.add(path)
 
1826
        # sketch: 
 
1827
        # compare source_index and target_index at or under each element of search_specific_files.
 
1828
        # follow the following comparison table. Note that we only want to do diff operations when
 
1829
        # the target is fdl because thats when the walkdirs logic will have exposed the pathinfo 
 
1830
        # for the target.
 
1831
        # cases:
 
1832
        # 
 
1833
        # Source | Target | disk | action
 
1834
        #   r    | fdlt   |      | add source to search, add id path move and perform
 
1835
        #        |        |      | diff check on source-target
 
1836
        #   r    | fdlt   |  a   | dangling file that was present in the basis. 
 
1837
        #        |        |      | ???
 
1838
        #   r    |  a     |      | add source to search
 
1839
        #   r    |  a     |  a   | 
 
1840
        #   r    |  r     |      | this path is present in a non-examined tree, skip.
 
1841
        #   r    |  r     |  a   | this path is present in a non-examined tree, skip.
 
1842
        #   a    | fdlt   |      | add new id
 
1843
        #   a    | fdlt   |  a   | dangling locally added file, skip
 
1844
        #   a    |  a     |      | not present in either tree, skip
 
1845
        #   a    |  a     |  a   | not present in any tree, skip
 
1846
        #   a    |  r     |      | not present in either tree at this path, skip as it
 
1847
        #        |        |      | may not be selected by the users list of paths.
 
1848
        #   a    |  r     |  a   | not present in either tree at this path, skip as it
 
1849
        #        |        |      | may not be selected by the users list of paths.
 
1850
        #  fdlt  | fdlt   |      | content in both: diff them
 
1851
        #  fdlt  | fdlt   |  a   | deleted locally, but not unversioned - show as deleted ?
 
1852
        #  fdlt  |  a     |      | unversioned: output deleted id for now
 
1853
        #  fdlt  |  a     |  a   | unversioned and deleted: output deleted id
 
1854
        #  fdlt  |  r     |      | relocated in this tree, so add target to search.
 
1855
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1856
        #        |        |      | this id at the other path.
 
1857
        #  fdlt  |  r     |  a   | relocated in this tree, so add target to search.
 
1858
        #        |        |      | Dont diff, we will see an r,fd; pair when we reach
 
1859
        #        |        |      | this id at the other path.
 
1860
 
 
1861
        # for all search_indexs in each path at or under each element of
 
1862
        # search_specific_files, if the detail is relocated: add the id, and add the
 
1863
        # relocated path as one to search if its not searched already. If the
 
1864
        # detail is not relocated, add the id.
 
1865
        searched_specific_files = set()
 
1866
        NULL_PARENT_DETAILS = dirstate.DirState.NULL_PARENT_DETAILS
 
1867
        # Using a list so that we can access the values and change them in
 
1868
        # nested scope. Each one is [path, file_id, entry]
 
1869
        last_source_parent = [None, None]
 
1870
        last_target_parent = [None, None]
 
1871
 
 
1872
        use_filesystem_for_exec = (sys.platform != 'win32')
 
1873
 
 
1874
        # Just a sentry, so that _process_entry can say that this
 
1875
        # record is handled, but isn't interesting to process (unchanged)
 
1876
        uninteresting = object()
 
1877
 
 
1878
 
 
1879
        old_dirname_to_file_id = {}
 
1880
        new_dirname_to_file_id = {}
 
1881
        # TODO: jam 20070516 - Avoid the _get_entry lookup overhead by
 
1882
        #       keeping a cache of directories that we have seen.
 
1883
 
 
1884
        def _process_entry(entry, path_info):
 
1885
            """Compare an entry and real disk to generate delta information.
 
1886
 
 
1887
            :param path_info: top_relpath, basename, kind, lstat, abspath for
 
1888
                the path of entry. If None, then the path is considered absent.
 
1889
                (Perhaps we should pass in a concrete entry for this ?)
 
1890
                Basename is returned as a utf8 string because we expect this
 
1891
                tuple will be ignored, and don't want to take the time to
 
1892
                decode.
 
1893
            :return: None if these don't match
 
1894
                     A tuple of information about the change, or
 
1895
                     the object 'uninteresting' if these match, but are
 
1896
                     basically identical.
 
1897
            """
 
1898
            if source_index is None:
 
1899
                source_details = NULL_PARENT_DETAILS
 
1900
            else:
 
1901
                source_details = entry[1][source_index]
 
1902
            target_details = entry[1][target_index]
 
1903
            target_minikind = target_details[0]
 
1904
            if path_info is not None and target_minikind in 'fdlt':
 
1905
                assert target_index == 0
 
1906
                link_or_sha1 = state.update_entry(entry, abspath=path_info[4],
 
1907
                                                  stat_value=path_info[3])
 
1908
                # The entry may have been modified by update_entry
 
1909
                target_details = entry[1][target_index]
 
1910
                target_minikind = target_details[0]
 
1911
            else:
 
1912
                link_or_sha1 = None
 
1913
            file_id = entry[0][2]
 
1914
            source_minikind = source_details[0]
 
1915
            if source_minikind in 'fdltr' and target_minikind in 'fdlt':
 
1916
                # claimed content in both: diff
 
1917
                #   r    | fdlt   |      | add source to search, add id path move and perform
 
1918
                #        |        |      | diff check on source-target
 
1919
                #   r    | fdlt   |  a   | dangling file that was present in the basis.
 
1920
                #        |        |      | ???
 
1921
                if source_minikind in 'r':
 
1922
                    # add the source to the search path to find any children it
 
1923
                    # has.  TODO ? : only add if it is a container ?
 
1924
                    if not osutils.is_inside_any(searched_specific_files,
 
1925
                                                 source_details[1]):
 
1926
                        search_specific_files.add(source_details[1])
 
1927
                    # generate the old path; this is needed for stating later
 
1928
                    # as well.
 
1929
                    old_path = source_details[1]
 
1930
                    old_dirname, old_basename = os.path.split(old_path)
 
1931
                    path = pathjoin(entry[0][0], entry[0][1])
 
1932
                    old_entry = state._get_entry(source_index,
 
1933
                                                 path_utf8=old_path)
 
1934
                    # update the source details variable to be the real
 
1935
                    # location.
 
1936
                    source_details = old_entry[1][source_index]
 
1937
                    source_minikind = source_details[0]
 
1938
                else:
 
1939
                    old_dirname = entry[0][0]
 
1940
                    old_basename = entry[0][1]
 
1941
                    old_path = path = None
 
1942
                if path_info is None:
 
1943
                    # the file is missing on disk, show as removed.
 
1944
                    content_change = True
 
1945
                    target_kind = None
 
1946
                    target_exec = False
 
1947
                else:
 
1948
                    # source and target are both versioned and disk file is present.
 
1949
                    target_kind = path_info[2]
 
1950
                    if target_kind == 'directory':
 
1951
                        if path is None:
 
1952
                            old_path = path = pathjoin(old_dirname, old_basename)
 
1953
                        new_dirname_to_file_id[path] = file_id
 
1954
                        if source_minikind != 'd':
 
1955
                            content_change = True
 
1956
                        else:
 
1957
                            # directories have no fingerprint
 
1958
                            content_change = False
 
1959
                        target_exec = False
 
1960
                    elif target_kind == 'file':
 
1961
                        if source_minikind != 'f':
 
1962
                            content_change = True
 
1963
                        else:
 
1964
                            # We could check the size, but we already have the
 
1965
                            # sha1 hash.
 
1966
                            content_change = (link_or_sha1 != source_details[1])
 
1967
                        # Target details is updated at update_entry time
 
1968
                        if use_filesystem_for_exec:
 
1969
                            # We don't need S_ISREG here, because we are sure
 
1970
                            # we are dealing with a file.
 
1971
                            target_exec = bool(stat.S_IEXEC & path_info[3].st_mode)
 
1972
                        else:
 
1973
                            target_exec = target_details[3]
 
1974
                    elif target_kind == 'symlink':
 
1975
                        if source_minikind != 'l':
 
1976
                            content_change = True
 
1977
                        else:
 
1978
                            content_change = (link_or_sha1 != source_details[1])
 
1979
                        target_exec = False
 
1980
                    elif target_kind == 'tree-reference':
 
1981
                        if source_minikind != 't':
 
1982
                            content_change = True
 
1983
                        else:
 
1984
                            content_change = False
 
1985
                        target_exec = False
 
1986
                    else:
 
1987
                        raise Exception, "unknown kind %s" % path_info[2]
 
1988
                if source_minikind == 'd':
 
1989
                    if path is None:
 
1990
                        old_path = path = pathjoin(old_dirname, old_basename)
 
1991
                    old_dirname_to_file_id[old_path] = file_id
 
1992
                # parent id is the entry for the path in the target tree
 
1993
                if old_dirname == last_source_parent[0]:
 
1994
                    source_parent_id = last_source_parent[1]
 
1995
                else:
 
1996
                    try:
 
1997
                        source_parent_id = old_dirname_to_file_id[old_dirname]
 
1998
                    except KeyError:
 
1999
                        source_parent_entry = state._get_entry(source_index,
 
2000
                                                               path_utf8=old_dirname)
 
2001
                        source_parent_id = source_parent_entry[0][2]
 
2002
                    if source_parent_id == entry[0][2]:
 
2003
                        # This is the root, so the parent is None
 
2004
                        source_parent_id = None
 
2005
                    else:
 
2006
                        last_source_parent[0] = old_dirname
 
2007
                        last_source_parent[1] = source_parent_id
 
2008
                new_dirname = entry[0][0]
 
2009
                if new_dirname == last_target_parent[0]:
 
2010
                    target_parent_id = last_target_parent[1]
 
2011
                else:
 
2012
                    try:
 
2013
                        target_parent_id = new_dirname_to_file_id[new_dirname]
 
2014
                    except KeyError:
 
2015
                        # TODO: We don't always need to do the lookup, because the
 
2016
                        #       parent entry will be the same as the source entry.
 
2017
                        target_parent_entry = state._get_entry(target_index,
 
2018
                                                               path_utf8=new_dirname)
 
2019
                        assert target_parent_entry != (None, None), (
 
2020
                            "Could not find target parent in wt: %s\nparent of: %s"
 
2021
                            % (new_dirname, entry))
 
2022
                        target_parent_id = target_parent_entry[0][2]
 
2023
                    if target_parent_id == entry[0][2]:
 
2024
                        # This is the root, so the parent is None
 
2025
                        target_parent_id = None
 
2026
                    else:
 
2027
                        last_target_parent[0] = new_dirname
 
2028
                        last_target_parent[1] = target_parent_id
 
2029
 
 
2030
                source_exec = source_details[3]
 
2031
                if (include_unchanged
 
2032
                    or content_change
 
2033
                    or source_parent_id != target_parent_id
 
2034
                    or old_basename != entry[0][1]
 
2035
                    or source_exec != target_exec
 
2036
                    ):
 
2037
                    if old_path is None:
 
2038
                        old_path = path = pathjoin(old_dirname, old_basename)
 
2039
                        old_path_u = utf8_decode(old_path)[0]
 
2040
                        path_u = old_path_u
 
2041
                    else:
 
2042
                        old_path_u = utf8_decode(old_path)[0]
 
2043
                        if old_path == path:
 
2044
                            path_u = old_path_u
 
2045
                        else:
 
2046
                            path_u = utf8_decode(path)[0]
 
2047
                    source_kind = _minikind_to_kind[source_minikind]
 
2048
                    return (entry[0][2],
 
2049
                           (old_path_u, path_u),
 
2050
                           content_change,
 
2051
                           (True, True),
 
2052
                           (source_parent_id, target_parent_id),
 
2053
                           (utf8_decode(old_basename)[0], utf8_decode(entry[0][1])[0]),
 
2054
                           (source_kind, target_kind),
 
2055
                           (source_exec, target_exec))
 
2056
                else:
 
2057
                    return uninteresting
 
2058
            elif source_minikind in 'a' and target_minikind in 'fdlt':
 
2059
                # looks like a new file
 
2060
                if path_info is not None:
 
2061
                    path = pathjoin(entry[0][0], entry[0][1])
 
2062
                    # parent id is the entry for the path in the target tree
 
2063
                    # TODO: these are the same for an entire directory: cache em.
 
2064
                    parent_id = state._get_entry(target_index,
 
2065
                                                 path_utf8=entry[0][0])[0][2]
 
2066
                    if parent_id == entry[0][2]:
 
2067
                        parent_id = None
 
2068
                    if use_filesystem_for_exec:
 
2069
                        # We need S_ISREG here, because we aren't sure if this
 
2070
                        # is a file or not.
 
2071
                        target_exec = bool(
 
2072
                            stat.S_ISREG(path_info[3].st_mode)
 
2073
                            and stat.S_IEXEC & path_info[3].st_mode)
 
2074
                    else:
 
2075
                        target_exec = target_details[3]
 
2076
                    return (entry[0][2],
 
2077
                           (None, utf8_decode(path)[0]),
 
2078
                           True,
 
2079
                           (False, True),
 
2080
                           (None, parent_id),
 
2081
                           (None, utf8_decode(entry[0][1])[0]),
 
2082
                           (None, path_info[2]),
 
2083
                           (None, target_exec))
 
2084
                else:
 
2085
                    # but its not on disk: we deliberately treat this as just
 
2086
                    # never-present. (Why ?! - RBC 20070224)
 
2087
                    pass
 
2088
            elif source_minikind in 'fdlt' and target_minikind in 'a':
 
2089
                # unversioned, possibly, or possibly not deleted: we dont care.
 
2090
                # if its still on disk, *and* theres no other entry at this
 
2091
                # path [we dont know this in this routine at the moment -
 
2092
                # perhaps we should change this - then it would be an unknown.
 
2093
                old_path = pathjoin(entry[0][0], entry[0][1])
 
2094
                # parent id is the entry for the path in the target tree
 
2095
                parent_id = state._get_entry(source_index, path_utf8=entry[0][0])[0][2]
 
2096
                if parent_id == entry[0][2]:
 
2097
                    parent_id = None
 
2098
                return (entry[0][2],
 
2099
                       (utf8_decode(old_path)[0], None),
 
2100
                       True,
 
2101
                       (True, False),
 
2102
                       (parent_id, None),
 
2103
                       (utf8_decode(entry[0][1])[0], None),
 
2104
                       (_minikind_to_kind[source_minikind], None),
 
2105
                       (source_details[3], None))
 
2106
            elif source_minikind in 'fdlt' and target_minikind in 'r':
 
2107
                # a rename; could be a true rename, or a rename inherited from
 
2108
                # a renamed parent. TODO: handle this efficiently. Its not
 
2109
                # common case to rename dirs though, so a correct but slow
 
2110
                # implementation will do.
 
2111
                if not osutils.is_inside_any(searched_specific_files, target_details[1]):
 
2112
                    search_specific_files.add(target_details[1])
 
2113
            elif source_minikind in 'ra' and target_minikind in 'ra':
 
2114
                # neither of the selected trees contain this file,
 
2115
                # so skip over it. This is not currently directly tested, but
 
2116
                # is indirectly via test_too_much.TestCommands.test_conflicts.
 
2117
                pass
 
2118
            else:
 
2119
                raise AssertionError("don't know how to compare "
 
2120
                    "source_minikind=%r, target_minikind=%r"
 
2121
                    % (source_minikind, target_minikind))
 
2122
                ## import pdb;pdb.set_trace()
 
2123
            return None
 
2124
 
 
2125
        while search_specific_files:
 
2126
            # TODO: the pending list should be lexically sorted?  the
 
2127
            # interface doesn't require it.
 
2128
            current_root = search_specific_files.pop()
 
2129
            current_root_unicode = current_root.decode('utf8')
 
2130
            searched_specific_files.add(current_root)
 
2131
            # process the entries for this containing directory: the rest will be
 
2132
            # found by their parents recursively.
 
2133
            root_entries = _entries_for_path(current_root)
 
2134
            root_abspath = self.target.abspath(current_root_unicode)
 
2135
            try:
 
2136
                root_stat = os.lstat(root_abspath)
 
2137
            except OSError, e:
 
2138
                if e.errno == errno.ENOENT:
 
2139
                    # the path does not exist: let _process_entry know that.
 
2140
                    root_dir_info = None
 
2141
                else:
 
2142
                    # some other random error: hand it up.
 
2143
                    raise
 
2144
            else:
 
2145
                root_dir_info = ('', current_root,
 
2146
                    osutils.file_kind_from_stat_mode(root_stat.st_mode), root_stat,
 
2147
                    root_abspath)
 
2148
                if root_dir_info[2] == 'directory':
 
2149
                    if self.target._directory_is_tree_reference(
 
2150
                        current_root.decode('utf8')):
 
2151
                        root_dir_info = root_dir_info[:2] + \
 
2152
                            ('tree-reference',) + root_dir_info[3:]
 
2153
 
 
2154
            if not root_entries and not root_dir_info:
 
2155
                # this specified path is not present at all, skip it.
 
2156
                continue
 
2157
            path_handled = False
 
2158
            for entry in root_entries:
 
2159
                result = _process_entry(entry, root_dir_info)
 
2160
                if result is not None:
 
2161
                    path_handled = True
 
2162
                    if result is not uninteresting:
 
2163
                        yield result
 
2164
            if want_unversioned and not path_handled and root_dir_info:
 
2165
                new_executable = bool(
 
2166
                    stat.S_ISREG(root_dir_info[3].st_mode)
 
2167
                    and stat.S_IEXEC & root_dir_info[3].st_mode)
 
2168
                yield (None,
 
2169
                       (None, current_root_unicode),
 
2170
                       True,
 
2171
                       (False, False),
 
2172
                       (None, None),
 
2173
                       (None, splitpath(current_root_unicode)[-1]),
 
2174
                       (None, root_dir_info[2]),
 
2175
                       (None, new_executable)
 
2176
                      )
 
2177
            initial_key = (current_root, '', '')
 
2178
            block_index, _ = state._find_block_index_from_key(initial_key)
 
2179
            if block_index == 0:
 
2180
                # we have processed the total root already, but because the
 
2181
                # initial key matched it we should skip it here.
 
2182
                block_index +=1
 
2183
            if root_dir_info and root_dir_info[2] == 'tree-reference':
 
2184
                current_dir_info = None
 
2185
            else:
 
2186
                dir_iterator = osutils._walkdirs_utf8(root_abspath, prefix=current_root)
 
2187
                try:
 
2188
                    current_dir_info = dir_iterator.next()
 
2189
                except OSError, e:
 
2190
                    # on win32, python2.4 has e.errno == ERROR_DIRECTORY, but
 
2191
                    # python 2.5 has e.errno == EINVAL,
 
2192
                    #            and e.winerror == ERROR_DIRECTORY
 
2193
                    e_winerror = getattr(e, 'winerror', None)
 
2194
                    win_errors = (ERROR_DIRECTORY, ERROR_PATH_NOT_FOUND)
 
2195
                    # there may be directories in the inventory even though
 
2196
                    # this path is not a file on disk: so mark it as end of
 
2197
                    # iterator
 
2198
                    if e.errno in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
 
2199
                        current_dir_info = None
 
2200
                    elif (sys.platform == 'win32'
 
2201
                          and (e.errno in win_errors
 
2202
                               or e_winerror in win_errors)):
 
2203
                        current_dir_info = None
 
2204
                    else:
 
2205
                        raise
 
2206
                else:
 
2207
                    if current_dir_info[0][0] == '':
 
2208
                        # remove .bzr from iteration
 
2209
                        bzr_index = bisect_left(current_dir_info[1], ('.bzr',))
 
2210
                        assert current_dir_info[1][bzr_index][0] == '.bzr'
 
2211
                        del current_dir_info[1][bzr_index]
 
2212
            # walk until both the directory listing and the versioned metadata
 
2213
            # are exhausted. 
 
2214
            if (block_index < len(state._dirblocks) and
 
2215
                osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2216
                current_block = state._dirblocks[block_index]
 
2217
            else:
 
2218
                current_block = None
 
2219
            while (current_dir_info is not None or
 
2220
                   current_block is not None):
 
2221
                if (current_dir_info and current_block
 
2222
                    and current_dir_info[0][0] != current_block[0]):
 
2223
                    if cmp_by_dirs(current_dir_info[0][0], current_block[0]) < 0:
 
2224
                        # filesystem data refers to paths not covered by the dirblock.
 
2225
                        # this has two possibilities:
 
2226
                        # A) it is versioned but empty, so there is no block for it
 
2227
                        # B) it is not versioned.
 
2228
 
 
2229
                        # if (A) then we need to recurse into it to check for
 
2230
                        # new unknown files or directories.
 
2231
                        # if (B) then we should ignore it, because we don't
 
2232
                        # recurse into unknown directories.
 
2233
                        path_index = 0
 
2234
                        while path_index < len(current_dir_info[1]):
 
2235
                                current_path_info = current_dir_info[1][path_index]
 
2236
                                if want_unversioned:
 
2237
                                    if current_path_info[2] == 'directory':
 
2238
                                        if self.target._directory_is_tree_reference(
 
2239
                                            current_path_info[0].decode('utf8')):
 
2240
                                            current_path_info = current_path_info[:2] + \
 
2241
                                                ('tree-reference',) + current_path_info[3:]
 
2242
                                    new_executable = bool(
 
2243
                                        stat.S_ISREG(current_path_info[3].st_mode)
 
2244
                                        and stat.S_IEXEC & current_path_info[3].st_mode)
 
2245
                                    yield (None,
 
2246
                                        (None, utf8_decode(current_path_info[0])[0]),
 
2247
                                        True,
 
2248
                                        (False, False),
 
2249
                                        (None, None),
 
2250
                                        (None, utf8_decode(current_path_info[1])[0]),
 
2251
                                        (None, current_path_info[2]),
 
2252
                                        (None, new_executable))
 
2253
                                # dont descend into this unversioned path if it is
 
2254
                                # a dir
 
2255
                                if current_path_info[2] in ('directory',
 
2256
                                                            'tree-reference'):
 
2257
                                    del current_dir_info[1][path_index]
 
2258
                                    path_index -= 1
 
2259
                                path_index += 1
 
2260
 
 
2261
                        # This dir info has been handled, go to the next
 
2262
                        try:
 
2263
                            current_dir_info = dir_iterator.next()
 
2264
                        except StopIteration:
 
2265
                            current_dir_info = None
 
2266
                    else:
 
2267
                        # We have a dirblock entry for this location, but there
 
2268
                        # is no filesystem path for this. This is most likely
 
2269
                        # because a directory was removed from the disk.
 
2270
                        # We don't have to report the missing directory,
 
2271
                        # because that should have already been handled, but we
 
2272
                        # need to handle all of the files that are contained
 
2273
                        # within.
 
2274
                        for current_entry in current_block[1]:
 
2275
                            # entry referring to file not present on disk.
 
2276
                            # advance the entry only, after processing.
 
2277
                            result = _process_entry(current_entry, None)
 
2278
                            if result is not None:
 
2279
                                if result is not uninteresting:
 
2280
                                    yield result
 
2281
                        block_index +=1
 
2282
                        if (block_index < len(state._dirblocks) and
 
2283
                            osutils.is_inside(current_root,
 
2284
                                              state._dirblocks[block_index][0])):
 
2285
                            current_block = state._dirblocks[block_index]
 
2286
                        else:
 
2287
                            current_block = None
 
2288
                    continue
 
2289
                entry_index = 0
 
2290
                if current_block and entry_index < len(current_block[1]):
 
2291
                    current_entry = current_block[1][entry_index]
 
2292
                else:
 
2293
                    current_entry = None
 
2294
                advance_entry = True
 
2295
                path_index = 0
 
2296
                if current_dir_info and path_index < len(current_dir_info[1]):
 
2297
                    current_path_info = current_dir_info[1][path_index]
 
2298
                    if current_path_info[2] == 'directory':
 
2299
                        if self.target._directory_is_tree_reference(
 
2300
                            current_path_info[0].decode('utf8')):
 
2301
                            current_path_info = current_path_info[:2] + \
 
2302
                                ('tree-reference',) + current_path_info[3:]
 
2303
                else:
 
2304
                    current_path_info = None
 
2305
                advance_path = True
 
2306
                path_handled = False
 
2307
                while (current_entry is not None or
 
2308
                    current_path_info is not None):
 
2309
                    if current_entry is None:
 
2310
                        # the check for path_handled when the path is adnvaced
 
2311
                        # will yield this path if needed.
 
2312
                        pass
 
2313
                    elif current_path_info is None:
 
2314
                        # no path is fine: the per entry code will handle it.
 
2315
                        result = _process_entry(current_entry, current_path_info)
 
2316
                        if result is not None:
 
2317
                            if result is not uninteresting:
 
2318
                                yield result
 
2319
                    elif (current_entry[0][1] != current_path_info[1]
 
2320
                          or current_entry[1][target_index][0] in 'ar'):
 
2321
                        # The current path on disk doesn't match the dirblock
 
2322
                        # record. Either the dirblock is marked as absent, or
 
2323
                        # the file on disk is not present at all in the
 
2324
                        # dirblock. Either way, report about the dirblock
 
2325
                        # entry, and let other code handle the filesystem one.
 
2326
 
 
2327
                        # Compare the basename for these files to determine
 
2328
                        # which comes first
 
2329
                        if current_path_info[1] < current_entry[0][1]:
 
2330
                            # extra file on disk: pass for now, but only
 
2331
                            # increment the path, not the entry
 
2332
                            advance_entry = False
 
2333
                        else:
 
2334
                            # entry referring to file not present on disk.
 
2335
                            # advance the entry only, after processing.
 
2336
                            result = _process_entry(current_entry, None)
 
2337
                            if result is not None:
 
2338
                                if result is not uninteresting:
 
2339
                                    yield result
 
2340
                            advance_path = False
 
2341
                    else:
 
2342
                        result = _process_entry(current_entry, current_path_info)
 
2343
                        if result is not None:
 
2344
                            path_handled = True
 
2345
                            if result is not uninteresting:
 
2346
                                yield result
 
2347
                    if advance_entry and current_entry is not None:
 
2348
                        entry_index += 1
 
2349
                        if entry_index < len(current_block[1]):
 
2350
                            current_entry = current_block[1][entry_index]
 
2351
                        else:
 
2352
                            current_entry = None
 
2353
                    else:
 
2354
                        advance_entry = True # reset the advance flaga
 
2355
                    if advance_path and current_path_info is not None:
 
2356
                        if not path_handled:
 
2357
                            # unversioned in all regards
 
2358
                            if want_unversioned:
 
2359
                                new_executable = bool(
 
2360
                                    stat.S_ISREG(current_path_info[3].st_mode)
 
2361
                                    and stat.S_IEXEC & current_path_info[3].st_mode)
 
2362
                                yield (None,
 
2363
                                    (None, utf8_decode(current_path_info[0])[0]),
 
2364
                                    True,
 
2365
                                    (False, False),
 
2366
                                    (None, None),
 
2367
                                    (None, utf8_decode(current_path_info[1])[0]),
 
2368
                                    (None, current_path_info[2]),
 
2369
                                    (None, new_executable))
 
2370
                            # dont descend into this unversioned path if it is
 
2371
                            # a dir
 
2372
                            if current_path_info[2] in ('directory'):
 
2373
                                del current_dir_info[1][path_index]
 
2374
                                path_index -= 1
 
2375
                        # dont descend the disk iterator into any tree 
 
2376
                        # paths.
 
2377
                        if current_path_info[2] == 'tree-reference':
 
2378
                            del current_dir_info[1][path_index]
 
2379
                            path_index -= 1
 
2380
                        path_index += 1
 
2381
                        if path_index < len(current_dir_info[1]):
 
2382
                            current_path_info = current_dir_info[1][path_index]
 
2383
                            if current_path_info[2] == 'directory':
 
2384
                                if self.target._directory_is_tree_reference(
 
2385
                                    current_path_info[0].decode('utf8')):
 
2386
                                    current_path_info = current_path_info[:2] + \
 
2387
                                        ('tree-reference',) + current_path_info[3:]
 
2388
                        else:
 
2389
                            current_path_info = None
 
2390
                        path_handled = False
 
2391
                    else:
 
2392
                        advance_path = True # reset the advance flagg.
 
2393
                if current_block is not None:
 
2394
                    block_index += 1
 
2395
                    if (block_index < len(state._dirblocks) and
 
2396
                        osutils.is_inside(current_root, state._dirblocks[block_index][0])):
 
2397
                        current_block = state._dirblocks[block_index]
 
2398
                    else:
 
2399
                        current_block = None
 
2400
                if current_dir_info is not None:
 
2401
                    try:
 
2402
                        current_dir_info = dir_iterator.next()
 
2403
                    except StopIteration:
 
2404
                        current_dir_info = None
 
2405
 
 
2406
 
 
2407
    @staticmethod
 
2408
    def is_compatible(source, target):
 
2409
        # the target must be a dirstate working tree
 
2410
        if not isinstance(target, WorkingTree4):
 
2411
            return False
 
2412
        # the source must be a revtreee or dirstate rev tree.
 
2413
        if not isinstance(source,
 
2414
            (revisiontree.RevisionTree, DirStateRevisionTree)):
 
2415
            return False
 
2416
        # the source revid must be in the target dirstate
 
2417
        if not (source._revision_id == NULL_REVISION or
 
2418
            source._revision_id in target.get_parent_ids()):
 
2419
            # TODO: what about ghosts? it may well need to 
 
2420
            # check for them explicitly.
 
2421
            return False
 
2422
        return True
 
2423
 
 
2424
InterTree.register_optimiser(InterDirStateTree)
 
2425
 
 
2426
 
 
2427
class Converter3to4(object):
 
2428
    """Perform an in-place upgrade of format 3 to format 4 trees."""
 
2429
 
 
2430
    def __init__(self):
 
2431
        self.target_format = WorkingTreeFormat4()
 
2432
 
 
2433
    def convert(self, tree):
 
2434
        # lock the control files not the tree, so that we dont get tree
 
2435
        # on-unlock behaviours, and so that noone else diddles with the 
 
2436
        # tree during upgrade.
 
2437
        tree._control_files.lock_write()
 
2438
        try:
 
2439
            tree.read_working_inventory()
 
2440
            self.create_dirstate_data(tree)
 
2441
            self.update_format(tree)
 
2442
            self.remove_xml_files(tree)
 
2443
        finally:
 
2444
            tree._control_files.unlock()
 
2445
 
 
2446
    def create_dirstate_data(self, tree):
 
2447
        """Create the dirstate based data for tree."""
 
2448
        local_path = tree.bzrdir.get_workingtree_transport(None
 
2449
            ).local_abspath('dirstate')
 
2450
        state = dirstate.DirState.from_tree(tree, local_path)
 
2451
        state.save()
 
2452
        state.unlock()
 
2453
 
 
2454
    def remove_xml_files(self, tree):
 
2455
        """Remove the oldformat 3 data."""
 
2456
        transport = tree.bzrdir.get_workingtree_transport(None)
 
2457
        for path in ['basis-inventory-cache', 'inventory', 'last-revision',
 
2458
            'pending-merges', 'stat-cache']:
 
2459
            try:
 
2460
                transport.delete(path)
 
2461
            except errors.NoSuchFile:
 
2462
                # some files are optional - just deal.
 
2463
                pass
 
2464
 
 
2465
    def update_format(self, tree):
 
2466
        """Change the format marker."""
 
2467
        tree._control_files.put_utf8('format',
 
2468
            self.target_format.get_format_string())